Research

Research Reliability of sentinel node status in predicting axillary lymph node involvement in breast cancer James Kollias, P Grantley Gill, Barry E Chatterton, Vivian E Hall, Melissa A Bochner, Brendon J Coventry and Gelareh Farshid MJA 1999; 171: 461-465 For editorial comment, see Ung & Wetzig Abstract - Introduction - Methods - Results - Discussion - References - Authors' details - - More articles on Oncology Abstract Objectives: To assess the reliability of determining sentinel node status in staging regional lymph nodes in breast cancer. Design and setting: Prospective validation study in a major public teaching hospital, comparing histological sentinel node status with that of remaining axillary nodes. Patients: 117 women who underwent sentinel node biopsy and axillary dissection for primary breast cancer between 1995 and 1998. Main outcome measures: Intraoperative success rate in sentinel node identification; false negative rate; predictive value of negative sentinel node status; overall accuracy of sentinel node status. Results: The sentinel node was identified at operation in 95 patients (81.2%). Tumour involvement of the sentinel node was demonstrated in 29 of 31 women (93.5%; 95% CI, 79%-99%). Sixty-four of the 66 women in whom the sentinel node was negative for tumour showed no further involvement of remaining axillary nodes (standard haematoxylin-eosin histological assessment), giving a predictive value of negative sentinel node status of 97% (95% CI, 89%-100%). The overall accuracy in 95 women in whom sentinel node status was compared with axillary node status was 97.9%. Conclusions: Histopathological examination of the sentinel node is an accurate method of assessing axillary lymph node status in primary breast cancer and is likely to be incorporated into future surgical management of women with primary breast cancer. Introduction Axillary lymph node status is the most important prognostic indicator in early breast cancer, and the detection of nodal metastases is a key factor in recommending adjuvant systemic therapy after surgery.1,2 Surgical removal and histopathological assessment of these nodes remains the only accurate way of determining their involvement with tumour. Axillary dissection also reduces the risk of regional recurrence of breast cancer in the axilla,3 as the risk is inversely related to the number of axillary nodes removed.4However, axillary lymph node dissection is not without morbidity: seroma formation, wound infection, damage to nerves, and reduced shoulder mobility. Of particular importance is lymphoedema, which occurs in 15% and 30% of women.5-8 As a consequence, other, less invasive methods of assessing axillary node status have been investigated (eg, mammography, ultrasound and colour doppler imaging, magnetic resonance imaging [MRI] and positron emission tomography [PET] scanning), but have yet to achieve the accuracy of surgical staging. Axillary node sampling -- removal of a small number of Level 1 nodes (those below the lower border of the pectoralis minor muscle) -- is associated with fewer complications, and has been proposed as an alternative to complete axillary dissection for staging of the axilla.9,10 However, its efficacy has been questioned.11 With the advent of population-based mammographic screening programs, there has been a dramatic decrease in tumour size and lymph node involvement in women diagnosed with early breast cancer.12,13 Thus, an increasing proportion of women will undergo axillary dissection only to find that their lymph glands are free of disease. Ideally, there should be a method of providing accurate assessment of axillary lymph node status without the need for axillary dissection. The sentinel lymph node (the first draining node within a lymph node basin) is the first to receive lymphatic drainage from a tumour site. Selective biopsy of this node allows the detection of metastases in clinically normal nodes with a low false negative rate, and has been used in patients with operable breast cancer by several groups.14-19 Their findings indicate that the status of the sentinel node(s) can accurately predict that of the fully dissected axilla. We report our experience of lymphoscintigraphy, intraoperative sentinel node mapping and sentinel node biopsy in 117 women with primary operable breast cancer. Our aims were: To assess the success rate of lymphoscintigraphy and intraoperative lymph node mapping in identifying the sentinel node; and To assess the accuracy of sentinel node biopsy in staging the axillary nodes. Methods Patients A consecutive series of 117 women treated for primary breast cancer at the Royal Adelaide Hospital Breast Unit between June 1995 and August 1998 entered a prospective evaluation of the technique of sentinel lymph node biopsy in breast cancer. Ethical approval for the study was provided by the Human Ethics Committee of the Royal Adelaide Hospital. All women gave written informed consent to participate in the study. Eligibility criteria were: Operable primary breast cancer (tumour, < 5 cm in diameter), detected clinically and by imaging, and confirmed by cytology, core biopsy or open biopsy; Clinically impalpable axillary lymph nodes; and The usual surgical indications for axillary dissection (ie, invasive, operable cancer). Patients were excluded if their condition did not fulfil these criteria; if they were pregnant or currently breastfeeding; if there was a high clinical suspicion or preoperative verification of axillary nodal involvement; or if they had metastatic breast carcinoma or a preoperative diagnosis of ductal carcinoma-in-situ. The women's ages ranged from 31 to 82 years (median, 60 years). Their clinical characteristics are summarised in Table 1. During the period of study, no eligible women refused entry to the study. Isotope injection technique The radiopharmaceutical used was 99mTc-labelled antimony sulfide colloid ("Lymph-Flo", Royal Adelaide Hospital Radiopharmacy). The colloid underwent filtration through a 0.2-µ sterile filter, ensuring more than 80% of the filtered particles were smaller than 20 nm. A 32-mm, 25-gauge needle was used to inject 40 MBq of tracer to four sites surrounding the palpable margin of the breast lesion. If the lesion was not palpable, ultrasound localisation was performed, and the injection was given in a similar manner under ultrasound guidance. In the initial stages of the study, 0.5 mL of tracer was injected in each of 82 patients. For the remaining 35 women, the injected volume was increased to 4 mL in four divided doses. In these latter women, the injection site was lightly massaged, and they were instructed to move their arms to encourage lymphatic movement. All radioisotope injections were given on the morning of the day of surgery. Lymphoscintigraphy and lymph node mapping After injection, serial anterior and appropriate lateral images were obtained with a large-field-of-view gamma camera (GE XRT, General Electric) at about 15-minute intervals until the initial draining node (or nodes) was visualised (Figure 1). The surface projection of the sentinel node was then marked on the skin with a radioactive marker. Orthogonal projections were made by the established technique of "triangulation"; the marks were joined by a straight line to indicate the base of a right-angled triangle with the node at the apex. Body outline was marked with a radioactive marker, or a transmission image was performed by holding a "flood" source behind the patient. The intraoperative probe (RMD CTC 4 with audible guidance system, Gammasonics, Melbourne) was calibrated in the Nuclear Medicine Department to the counts detected at the skin surface. Surgical technique After completion of lymphoscintigraphy and sentinel node mapping, the patient and hand-held gamma probe were transferred to the operating theatre. In 66 patients, 1-2 mL of 2.5% Patent Blue V dye (Guerbet Laboratories, France; distributed by Fauldings Australia, Adelaide) was injected into the breast parenchyma or subdermal fat overlying the tumour to facilitate intraoperative identification of the sentinel node. Blue dye alone was used in 19 patients before a gamma probe was available. At operation, a 2-cm transverse axillary incision was made in accordance with the planned axillary lymph node dissection, but taking into account the preoperative skin markings indicating the location of the sentinel node at lymphoscintigraphy. An attempt was made to identify the node in vivo before commencement of axillary dissection. The node was identified by its blue colour (if dye was used) and/or by the hand-held gamma probe (in a sterile sheath). The probe enabled detection of individual nodes with radioactivity levels significantly greater than those of the axillary fat (Figure 2). Sometimes more than one sentinel node was identified. If both dye and radioisotope were used for lymphatic mapping, the blue node corresponded to the most radioactive node. Once the sentinel node was removed, its activity was reassessed ex vivo and it was sent for histological examination separately from the main axillary nodal specimen. The axillary fat was then examined with the gamma probe in vivo to exclude any residual activity suggesting further sentinel nodes. The axillary skin incision was then lengthened and a level I and II axillary lymph node dissection was performed. The resected axillary tissue was examined ex vivo using the probe to identify any further radioactive or blue lymph nodes not identified during in-vivo examination. Histopathological examination All specimens were examined by duty histopathologists at the Institute of Medical and Veterinary Science. The histological tumour features were classified according to tumour size and grade,20 and presence or absence of vascular invasion.21 Generally, sentinel nodes were submitted in their entirety for histological evaluation. Those larger than 1.5 cm were sliced before paraffin embedding. Each node was placed in an individual cassette. At least one section of each node was stained with haematoxylin-eosin (H&E) and examined with light microscopy. Immunohistochemical analysis (antikeratin antibody CAM 5.2, Becton Dickinson) was performed in H&E-stained sections suspected of having metastatic tumour deposits. The axillary fat was fixed in formalin and the nodes were later isolated from the fat after clearance in Carnoy's solution. Each node was placed in an individual cassette and larger nodes were sliced before being embedded in paraffin. At least one H&E-stained section of each node was examined. Statistical analysis A false negative sentinel node was defined as an excised sentinel lymph node which contained no microscopically detectable tumour, but which was associated with at least one tumour-positive node in the remaining resected axillary tissue. The false negative rate and the predictive value of negative sentinel node status were calculated together with 95% confidence intervals. The kappa (κ) statistic for paired data was used to assess the level of agreement between sentinel node status and axillary node status.22 A score of -1 indicates perfect disagreement and + 1 indicates perfect agreement. The corresponding z and P values were calculated. Univariate analysis was used to assess clinical and histological factors that predicted intraoperative sentinel node localisation. Fisher's exact and χ2 tests were used for other analyses between groups. Results Lymphoscintigraphy The sentinel node was identified on preoperative lymphoscintigraphy in 74 of 117 women (63.2%). One sentinel node was identified in 52 women, two were identified in 20 women, and in two further women three and four sentinel nodes were identified, respectively. The sentinel node was identified outside the lower axilla in nine patients (Table 2). A significant increase in sentinel node identification at lymphoscintigraphy was noted after the injection of larger isotope volumes into the breast (77% v 57%; χ2 = 4.15; P = 0.04), while rates of intraoperative detection of the sentinel node also increased (91% v 76%; χ2 = 3.4; P = 0.06). Intraoperative sentinel node identification The sentinel node was identified in 95 patients (81.2%) at operation. In 66 women, one sentinel node was identified, two were identified in 20 women, three in eight women, and in one four sentinel nodes were identified. The sentinel node was identified in 35 of the 51 women in whom radioisotope alone was used (68.6%), compared with 18 of 19 women in whom blue dye alone was used (94.7%) and 42 of 47 women in whom both isotope and blue dye were used (89.4%) (χ2 = 9.6; P = 0.008). Of the clinical and histological factors assessed for predicting intraoperative sentinel node identification, only a positive preoperative lymphoscintigram was significant (χ2 = 28.7; P < 0.001) (Table 3). Predictive value of sentinel node(s) In 95 patients in whom the sentinel node was identified, 31 had metastatic tumour involvement of axillary nodes (32.6%). Tumour involvement of the sentinel node was demonstrated in 29 of these 31 women (93.5%; 95% CI, 79%-99%), giving a false negative rate of 6.5%. The sentinel node was the only positive node in 13 of 31 women (41.9%). Of 66 women with a negative sentinel node, 64 had no tumour involvement in the remainder of the axillary nodes (by standard H&E histological assessment), giving a predictive value of negative sentinel node status of 97% (95% CI, 89%-100%). The overall accuracy in 95 patients in whom sentinel node status was compared with axillary node status was 97.9% (κ, 0.95; z = 9.3; P < 0.001) (Table 4). Of the 22 women in whom the sentinel node was not identified at operation, six had nodal metastases on histological examination of the dissected axillary nodes. Discussion The concept of the sentinel lymph node is based on the premise that the first lymph node to receive lymphatic drainage from a tumour site should be the first site of lymphatic spread; that "skip metastases" do not occur; and that the absence of tumour metastases in the sentinel node implies the absence of lymph node metastases in the entire lymphatic basin. This concept was first described in penile carcinoma in 197723 and was later studied in patients with cutaneous melanoma.24 Previous detailed pathological studies of axillary nodes in women with breast cancer have demonstrated a skip metastasis rate of less than 5%.25,26Our results confirm that the status of the sentinel lymph node(s) predicts the overall axillary lymph node status with a high degree of accuracy, and can thus be used to limit the morbidity associated with axillary surgery. More importantly, the predictive value of a tumour-free sentinel node was 97%. As such, women identified with a sentinel node free of metastatic tumour can be reassured that further axillary lymph node involvement is highly unlikely. Other studies of sentinel node biopsy in breast cancer (using blue dye and radioactive isotope techniques) have shown sentinel node status to accurately determine axillary lymph node status in more than 95% of women.14-19 We still need to deal with the problem that 3% of patients exhibited tumour-positive axillary nodes when the biopsied sentinel node was negative. The optimal method of pathological assessment of the sentinel node remains unresolved and was not addressed in our study. This issue was discussed at the Adelaide Workshop on Sentinel Node Biopsy in Breast Cancer27 and is the subject of further studies by one of us (G F). Giuliano et al28 have found that immunohistochemical studies of sentinel nodes showed micrometastases in an additional 11% of women whose sentinel node was tumour negative on light microscopy. However, similar assessment of two women with false negative results in our study did not reveal metastases. The implications of micrometastases detected by sensitive immunohistochemical and polymerase chain reaction (PCR) techniques for multidisciplinary care are unknown. They are currently being investigated in trials in the United States (Merrick Ross, Associate Professor of Surgical Oncology, M D Anderson Hospital, Texas, USA, personal communication). Until the answers to this question are available, a large UK trial (ALMANAC) is assessing sentinel node status by conventional microscopy (R Mansell, Professor of Surgery, Cardiff University, UK, personal communication), as this is the current method on which treatment planning is based. These uncertainties emphasise the need for Australian studies to incorporate detailed protocols for pathology assessment of the sentinel node. The prognostic implications of a false negative sentinel node are uncertain, but should be compared with the considerable physical morbidity associated with axillary dissection in lymph node negative women. There is a definite error rate in routine pathological assessment of axillary dissection specimens which may underestimate metastatic disease by 11%-30%,27,29 while unselective sampling of the axilla fails to remove involved nodes in many women.11 The false negative rate must ultimately be minimised by maximal detection of the sentinel node by scintigraphy, careful operative technique and optimal pathological assessment, which requires an experienced multidisciplinary team. The concomitant intraoperative use of both blue dye and radionuclide methods for lymphatic mapping was particularly useful for sentinel node biopsy. Preoperative lymphoscintigraphy permits identification of the sentinel node and subsequent planning of the site of skin incision. Several radiolabelled colloids are currently in use around the world, but the recent workshop in Adelaide27 identified antimony colloids as having excellent properties for lymphoscintigraphy. This is the only agent available for this purpose in Australia and is able to visualise sentinel nodes in the internal mammary chain as well as in the axillary node group. The blue dye technique facilitated visualisation of the sentinel node at the time of surgery and was supplemented by the use of an intraoperative gamma probe. In all patients in whom both blue dye and radionuclide were used, the blue node corresponded to the "hot" node previously identified on lymphoscintigraphy and identified intraoperatively with the hand-held gamma probe. Furthermore, the identification of a sentinel node at preoperative lymphoscintigraphy was the only factor significantly associated with the intraoperative identification of the sentinel node. Lymphoscintigraphy also demonstrates the number and location of potential sentinel nodes requiring biopsy. The initial rate of preoperative identification of the sentinel node by lymphoscintigraphy in our series was lower than that in published reports. However, this was overcome by increasing the volume of the isotope injection and presumably increasing tissue oncotic pressure, lymphatic uptake and drainage. The importance of isotope volume in achieving successful scintigraphic identification of the sentinel node has also been suggested by others.30 Sentinel lymph node mapping and biopsy are likely to be incorporated into clinical practice, provided they can be successfully performed in most patients, and it can be shown that women with negative sentinel nodes who undergo no further treatment to the axilla are not adversely compromised in terms of disease-free and overall survival. This will be best established by randomised controlled studies comparing sentinel node biopsy with standard axillary surgical management. In addition, these studies should address the implied assumption of lower short and long term morbidity associated with this procedure, the optimal methods of pathological assessment, and allow analysis and comparison with clinicopathological variables in predicting sentinel node status. Studies are currently being undertaken in Europe, the United Kingdom and the United States and it is hoped that Australian women can soon participate in similar trials in Australia. References Carter CL, Allen C, Henson DE. Relation of tumour size, lymph nodes status and survival in 24,740 breast cancer cases. Cancer 1989; 63: 181-187. Fisher ER, Anderson S, Redmond C, Fisher B. Pathologic findings from the National Surgical Adjuvant Breast Project Protocol B-06: 10 year pathological and clinical prognostic discriminants. Cancer 1993; 71: 2507-2514. Fisher D, Woolmark N, Bauer M, et al. The accuracy of clinical nodes staging and of limited axillary dissection as a determinant of histological nodal status in carcinoma of the breast. Surg Gynecol Obstet 1991; 152: 765-772. Axellsson CK, Mouridsen HT, Zedeler K. Axillary dissection of Level I and II lymph nodes is important in breast cancer classification: The Danish Breast Cancer Cooperative Group (DBCG). Eur J Cancer 1992; 28: 1415-1418. Kissin MW, Querci-Della-Rovere G, Easton D, Westbury G. Risk of lymphoedema following the treatment of breast cancer. Br J Surg 1986; 73: 580-584. Aitken RJ, Gayes MN, Rodger A, et al. Arm morbidity within a trial of mastectomy and either node sample with selective radiotherapy or axillary clearance. Br J Surg 1989; 76: 568-571. Larson D, Weinstein M, Goldburg I, et al. Oedema of the arm as a function of the extent of axillary surgery in patients with Stage 1-2 carcinoma of the breast treated with primary radiotherapy. Int J Radiat Oncol Biol Phys 1986; 12: 1575-1582. Liljegren G, Holmburg L. Arm morbidity after sector resection and axillary dissection with or without postoperative radiotherapy in breast cancer. Stage 1: Results from a randomised trial. Uppsala Orebro Breast Cancer Study Group. Eur J Cancer 1997; 33: 193-199. Steel RJC, Forrest APM, Gibson T, et al. The efficacy of lower axillary sampling in obtaining lymph node status in breast cancer: a controlled randomised trial. Br J Surg 1985; 72: 368-369. Dixon JM, Dillon P, Anderson TJ, Chetty U. Axillary node sampling in breast cancer: an assessment of its efficacy. Breast 1998; 7: 206-208. Kissin MW, Thompson PH, Price AB, et al. The inadequacy of axillary sampling in breast cancer. Lancet 1982; 1: 1210-1212. Tabar L, Fagerberg G, Duffy SW, et al. Update of the Swedish two-county program of mammographic screening for breast cancer. Radiol Clin North Am 1992; 30: 187-210. Cady B, Stone MD, Schuler JG, et al. The new era in breast cancer: invasion, size and lymph node involvement dramatically decreased as a result of mammographic screening. Arch Surg 1996; 131: 301-308. Giuliano AE, Kirgan DM, Guenther JM, Morton DL. Lymphatic mapping and sentinel lymphadenectomy in breast cancer. Ann Surg 1994; 220: 391-401. Albertini JJ, Lyman GH, Cox C, et al. Lymphatic mapping and sentinel node biopsy in the patient with breast cancer. JAMA 1996; 276: 1818-1822. Veronesi U, Paganelli G, Galimberti V, et al. Sentinel node biopsy to avoid axillary dissection in breast cancer with clinically negative lymph nodes. Lancet 1997; 349: 1864-1867. Borgstein PJ, Pijpers R, Comans EF, et al. Sentinel lymph node biopsy in breast cancer: guidelines and pitfalls of lymphoscintigraphy and gamma probe detection. J Am Coll Surg 1998; 186: 275-283. Cox CE, Pendas S, Cox JM, et al. Guidelines for sentinel node biopsy and lymphatic mapping of patients with breast cancer. Ann Surg 1998; 227: 645-653. O'Hea BJ, Hill ADK, El-Shirbiny AM, et al. Sentinel lymph node biopsy in breast cancer: initial experience at Memorial Sloan-Kettering Cancer Center. J Am Coll Surg 1998; 186: 423-427. Elston CW, Ellis IO. Pathological prognostic factors in breast cancer. The value of histological grade in breast cancer: experience from a large study with long-term follow-up. Histopathology 1991; 19: 403-410. Pinder SE, Ellis IO, Galea M, et al. Pathological prognostic factors in breast cancer. Vascular invasion: relationship with recurrence and survival in a large study with long-term follow-up. Histopathology 1994; 24: 41-47. Fliess JL. Statistical methods for rates and proportions. 2nd edition. New York, NY: John Wiley and Sons, 1981. Cabanas RM. An approach for the treatment of penile carcinoma. Cancer 1977; 39: 456-466. Morton DL, Wen D-R, Wong JH, et al. Technical details of intraoperative lymphatic mapping for early stage melanoma. Arch Surg 1992; 127: 392-399. Berg JW. The significance of axillary node levels in the study of breast cancer. Cancer 1955; 8: 776-778. Veronesi U, Rilke F, Luimi A, et al. Distribution of axillary node metastases by level of invasion: an analysis of 539 cases. Cancer 1987; 59: 682-687. Kollias J, Gill PG, Chatterton B, et al. Sentinel node biopsy in breast cancer: recommendations for surgeons, pathologists, nuclear physicians and radiologists in Australia and New Zealand. Aust N Z J Surg 1999. In press. Giuliano AE, Dale PS, Turner RR, et al. Improved axillary staging of breast cancer with sentinel lymphadenectomy. Ann Surg 1995; 222: 387-399. Hainsworth PJ, Tjandra JJ, Stillwell RG, et al. Detection and significance of occult metastases in node negative breast cancer. Br J Surg 1993; 80: 459-463. Krag DN, Ashikaga T, Harlow SH, Weaver DL. Development of sentinel node targeting technique in breast cancer patients. Breast J 1998; 4: 67-74. (Received 22 Apr, accepted 9 Sep, 1999) Authors' details Royal Adelaide Hospital and Women's Health Centre, Adelaide, SA. James Kollias, MB BS, FRACS, Staff Surgeon, Breast-Endocrine and Surgical Oncology Unit. P Grantley Gill, FRACS, MD, Head, Breast-Endocrine and Surgical Oncology Unit; and Associate Professor, University of Adelaide. Barry E Chatterton, MB BS, FRACP, Director, Department of Nuclear Medicine. Vivian E Hall, MB BS, FRACR, Radiologist, Department of Radiology. Melissa A Bochner, MB BS, FRACS, Senior Registrar, Breast-Endocrine and Surgical Oncology Unit. Brendon J Coventry, FRACS, PhD, Senior Surgeon, Breast-Endocrine and Surgical Oncology Unit; and Senior Lecturer, University of Adelaide. Department of Tissue Pathology, Institute of Medical and Veterinary Science, Adelaide, SA. Gelareh Farshid, MB BS, FRCPA, Senior Lecturer, University of Adelaide. Reprints will not be available from the authors. Correspondence: Associate Professor P G Gill, Breast-Endocrine Surgical Oncology Unit, Royal Adelaide Hospital, North Terrace, Adelaide, SA 5000. cbatesbrownswordATmedicine.adelaide.edu.au Back to textBack to textBack to textBack to text 3: Clinical and histological features predicting success in sentinel node identification at operationVariableNo. of womenSentinel node identified (%)χ2 (P)Age (years)< 503330 (91%)2.84> 508465 (77%)(0.09)Tumour site (quadrant)Upper/outer86 69 (80%)0.2Lower/inner31 26 (84%)(0.66)Tumour detectionScreening56 42 (75%)2.7Symptomatic61 53 (87%)(0.1)Previous core biopsyYes1210 (83%)0.04No10585 (79%)(0.84)Previous open biopsyYes11089 (81%)0.1No76 (86%)(0.75)Scintiscan resultPositive74 71 (96%)28.7Negative43 24 (56%)(< 0.001)OperationMastectomy31 24 (77%)1.23Wide local excision5547 (85%)(0.54)Localised wide local excision3124 (77%)Tumour size*< 2cm79 64 (81%)0.03≥ 2cm36 29 (81%)(0.95)Tumour grade*12721 (78%)2.425346 (87%)(0.31)33526 (74%)Lymphatic/vascular invasionNegative101 83 (82%)0.44Positive16 12 (75%)(0.5)Lymph node statusNegative80 64 (80%)0.2Positive37 31 (84%)(0.6) * Excludes two cases of ductal carcinoma-in-situ, diagnosed after excision. Back to textBack to text

James Kollias · Barry E Chatterton · Vivian E Hall · Melissa A Bochner · Brendon J Coventry · Gelareh Farshid

Research Iodine deficiency in ambulatory participants at a Sydney teaching hospital: is Australia truly iodine replete? Jenny E Gunton, Graham Hams, Marcelle Fiegert and Aidan McElduff MJA 1999; 171: 467-470 For editorial comment, see Eastman Abstract - Introduction - Methods - Results - Discussion - References - Authors' details - - More articles on Endocrinology Abstract Objective: To assess iodine status in four separate groups -- pregnant women, postpartum women, patients with diabetes mellitus and volunteers. Design and setting: Prospective cross-sectional study at a tertiary referral hospital in Sydney. Participants: 81 pregnant women attending a "high risk" obstetric clinic; 26 of these same women who attended three months postpartum; 135 consecutive patients with diabetes mellitus attending the diabetes clinic for an annual complications screen; and 19 volunteers. There were no exclusion criteria. Methods: Spot urine samples were obtained, and urinary iodine was measured by inductively coupled plasma mass spectrometer. Outcome measures: Iodine status based on urinary iodine concentration categorised as normal (> 100 µg/L), mild deficiency (51-100 µg/L) and moderate to severe deficiency (< 50 µg/L). Results: Moderate to severe iodine deficiency was found in 16 pregnant women (19.8%), five postpartum women (19.2%), 46 patients with diabetes (34.1%) and five volunteers (26.3%). Mild iodine deficiency was found in an additional 24 pregnant women (29.6%), nine postpartum women (34.6%), 51 patients with diabetes (37.8%) and 9 normal volunteers (47.4%). Median urinary iodine concentration was 104 µg/L in pregnant women, 79 µg/L in postpartum women, 65 µg/L in patients with diabetes mellitus and 64 µg/L in volunteers. Conclusions: The high frequency of iodine deficiency found in our participants suggests that dietary sources of iodine in this country may no longer be sufficient. Further population studies are required. Introduction It is currently believed that iodine deficiency does not exist in Australia.1,2 However, iodine status is seldom, if ever, measured in routine clinical care, and iodine deficiency may have significant adverse consequences, particularly during pregnancy (Box 1). Box 2 shows some of the reasons why iodine intake in Australia may be inadequate. The recommended daily intake (RDI) of iodine is 100 µg daily for the general population and 150-200 µg daily for women who are pregnant or breastfeeding6-8,10 (iodine demand increases during pregnancy because of increased renal clearance and fetal iodine transfer). Approximately 90% of iodine is excreted in the urine,1,11 and iodine status is usually assessed by measuring urinary iodine concentration. The accepted minimum adequate level of urinary iodine is 100 µg/L, and levels above this are considered normal.1,6-9,12 Urinary iodine concentrations below 25 µg/L are classified as severe deficiency, and are associated with an increased risk of cretinism; 26-50 µg/L is classified as moderate deficiency, and 51-100 µg/L is regarded as mild iodine deficiency.1,6-9,12 The World Health Organization (WHO) recommends that the median urinary iodine concentration for populations as a whole should be more than 100 µg/L, that less than 20% of the population should have a urinary iodine concentration below 50 µg/L, and that no cretinism occurs.12 Having previously found low levels of free thyroxine in pregnant women,13 and in light of the adverse consequences of iodine deficiency during pregnancy, we initially set out to test the iodine status of a group of pregnant women. We subsequently included other groups to widen our investigation of iodine status. Methods Study participants Our study was conducted at a tertiary referral hospital in Sydney. Participants in the study included women who attended a specialist "high risk" obstetric clinic, patients of both sexes with diabetes who attended the hospital's diabetes clinic, and healthy, non-pregnant volunteers recruited after a presentation about iodine. Participants thus comprised 81 consecutive pregnant women who attended the obstetric clinic between 1 August 1998 and 1 April 1999, 26 of these same women who were reassessed at three months postpartum, 135 consecutive patients who attended the diabetes clinic for an annual complications screen between 1 November 1998 and 1 February 1999, and 19 volunteers recruited between 1 February and 1 July 1999. All participants provided a routine urine sample. There were no exclusion criteria. One of the 81 pregnant women had thyrotoxicosis as a result of Graves' disease -- she particpated before commencing therapy. Twenty-two of the patients attending the diabetes clinic (16.1%) had type 1 diabetes, 103 (76.3%) had type 2 diabetes and 10 (7.5%) had impaired glucose tolerance. One of the patients with diabetes had recently received iodine-containing intravenous contrast medium during a coronary angiogram, and one was taking amiodarone. No other participant was known to have received contrast medium, or to be taking amiodarone or iodine supplements. Urinary iodine measurement Urinary iodine concentrations were determined by means of a Varian UltraMass inductively coupled plasma mass spectrometer with SPS-5 autosampler (Varian Inc., Palo Alto, California, USA). The measurement was calibrated over a range of 0-1000 µg iodine per litre. The lower limit of detection for the assay was 2 µg/L. The reproducibility of the assay as represented by the 100 µg/L calibrator assessed over three months was ± 6 µg/L (± 2 SD). Comparison with the colorimetric/Sandell-Koltkoff reaction method showed a highly significant correlation (P < 0.001; see Box 3). Other authors have also compared the methods and found high correlation.14 In particular, no systematic biases were found at low iodine concentrations. Some investigators use the urinary iodine/creatinine ratio to determine iodine status.1,14 We thus measured urinary creatinine by the Creatinine Jaffa method (Boehringer Mannheim Systems, Mannheim, Germany) and calculated iodine/creatinine ratios (µg iodine/g creatinine) for each participant. The correlation between urinary iodine and iodine/creatinine ratio was high for non-pregnant participants (r = 0.969; P < 0.001) and lower for the pregnant group (r = 0.419; P < 0.001). Twenty-four-hour urinary iodine measurement may be used to assess iodine status,6 but this method can be unreliable because of incorrect or incomplete collection,15 and is less practical than spot samples for population surveys.9,12 To compare this method with our spot sampling, we selected six pregnant women (on the basis of their spot urine concentrations to cover a range of values) who collected 24-hour urine samples for iodine content measurement. The correlation was highly significant (r2 = 0.82), thus confirming that spot urine samples were a reliable way of measuring iodine status. As part of routine care, 70 of the 81 pregnant women and 121 of the 135 patients with diabetes had thyroid function tests. Free thyroxine (FT4) and thyroid-stimulating hormone (TSH) levels were measured by means of an automated chemiluminescence system (Chiron Diagnostics, Scoresby, Vic.). We did not seek ethical approval for this study as it involved no deviation from usual care, except in the case of the 19 volunteers who agreed to provide a urine sample. Statistical analysis We used SPSS for statistical analysis.16 Means are expressed with ± 2 standard deviations, and medians with 95% confidence intervals (CI) are shown where data were not normally distributed. The results of non-parametric variables (including iodine results) were compared by means of the Mann-Whitney Wilcoxon rank sum test. Results Box 4 shows the mean and median ages and the results of spot urinary iodine concentration for the four groups. The three non-pregnant groups had similar urinary iodine concentration results, with a slightly higher median in the postpartum group compared with the group with diabetes. As expected,17-19 the pregnant women had higher urinary iodine concentrations than the other groups as a whole (P = 0.004). The iodine/creatinine ratios also show a high proportion of abnormal results (Box 4). The median iodine concentration in the 26 postpartum women (79 µg/L) who provided repeat urine samples for iodine measurement three months after delivery was considerably lower than that in the 81 pregnant women (104 µg/L). However, this difference was not statistically significant (P = 0.249). The patient with diabetes who had received iodine-containing intravenous contrast medium during a coronary angiogram in the month before the urinary spot test had a urinary iodine concentration of 2170 µg/L. Box 5 shows TSH levels and FT4 levels versus iodine status in pregnant and non-pregnant participants. There was no significant relationship between iodine status and FT4 or TSH levels in either the pregnant group or non-pregnant group. Separate analysis of patients with diabetes and postpartum women did not significantly alter these results. However, there was a weak correlation between FT4 and urinary iodine levels when examined as a continuous variable (Pearson correlation coefficient, 0.26; P = 0.016). Discussion By WHO criteria,12 the median iodine levels in our pregnant participants were only just adequate, while those in postpartum women, patients with diabetes and normal volunteers were inadequate. The slightly higher median iodine level in the postpartum group compared with that in the group with diabetes may have been the result of this concentration not having returned to baseline after pregnancy, although further study is required to document the rate of change post partum. We believe the low values in patients with diabetes was not a problem specific to diabetes, but merely a reflection of low urinary iodine levels in the general population. The similarity between the patients with diabetes and our small group of volunteers supports this view. Our data are consistent with generally low iodine intake. Our findings mirror recent reports from other countries.9,11,20 A United States study showed that the median urinary iodine concentration in 1988-1994 had decreased by more than 50% from that in 1971-1974.9,11 The 1988-1994 results showed 11.7% of the US population to be iodine deficient (a 4.5-fold increase since 1971-1974). The mean urinary iodine concentration in that population was 265 µg/L, and people from higher socioeconomic groups were more likely to be iodine deficient. Our data may also reflect this effect, as, although our patients were attending a public clinic, the hospital catchment area is a relatively high socioeconomic group. Our data suggest that Australia may be experiencing a similar trend to that seen in the US. Iodine deficiency during pregnancy can affect the thyroid glands of both the mother and baby,10,17-19 and may have many adverse health consequences (Box 1). Some, but not all, researchers have found an increase in urinary iodine levels during pregnancy.10,17-19 Smyth et al studied urinary iodine concentration in a group of pregnant women in an area of Ireland with known borderline iodine deficiency.10 In the third trimester, they found a mean urinary iodine concentration of 132 µg/L (standard error of the mean, 6.8), and found that urinary iodine concentration increased during pregnancy. In a more iodine-deficient area, Glinoer et al found that urinary iodine concentration did not increase during pregnancy (median iodine concentration 45 µg/L after 20 weeks' gestation, no mean given).19 So, it is not clear whether the apparently higher levels in our pregnant women were pregnancy related or, in fact, masked iodine deficiency in pregnancy. We found that a considerable percentage of pregnant women (4.9%) were severely iodine deficient, with spot urine results of < 25 µg/L. While this is the threshold below which cretinism may occur, other factors, such as selenium deficiency and the presence of dietary goitrogens, also play a part in determining cretinism,21 and these two factors are not usually seen in Sydney. Therefore, we would not expect to see an increased incidence of cretinism in Sydney on the basis of these results alone. However, more subtle adverse fetal outcomes may occur. Our findings suggest that we should no longer automatically consider Australia an iodine-replete country. We found that iodine deficiency was common among 235 people attending a Sydney teaching hospital and speculate that these data are applicable to the general population, although this will require independent confirmation. The frequency of iodine deficiency in our pregnant population (18.8%) approaches the maximum acceptable level recommended by WHO (20%); this recommendation was exceeded in our group with diabetes (34.1%) and the normal volunteers (26.3%). The postpartum women had a median iodine concentration of 79 µg/L, which is lower than the WHO recommendation of 100 µg/L. This has important public health implications. The weaknesses of this study include the small group of normal volunteers, and perhaps the use of a sample from a teaching hospital rather than the community. The normal volunteers had results which are equivalent to those seen in postpartum women and non-pregnant patients with diabetes. Our subjects were all ambulatory, not inpatients at the time of testing, and generally well. Although 24-hour urinary iodine excretion studies may be the ideal method of assessing iodine status, these are not generally performed in large numbers for a variety of technical and practical reasons. A weak correlation between urinary iodine and free thyroxine was observed for all non-pregnant participants in total, and for the participants with diabetes mellitus. Because of the large number of other factors which influence thyroid function (including pregnancy),13 the relatively loose correlations are an expected finding. Further studies are needed, and these include (i) population surveys in Sydney and elsewhere in Australia; (ii) assessment of thyroid size (eg, by ultrasound) in relation to iodine status; and (iii) detailed assessment of neonates, including thyroid size, neonatal TSH levels, and detailed neurological outcomes. References Hetzel BS. Iodine deficiency disorders. In: Garrow JS, James WPT, editors. Human nutrition and dietetics. Edinburgh: Churchill Livingstone, 1993: 534-555. Mortimer RH. Thyroid disease and pregnancy. Aust N Z J Med 1998; 28: 647-653. Tasmanian Thyroid Advisory Committee. Study in disease surveillance. 1950-1979. Med J Aust 1981; 2: 234-238. Clements FW. Goitre studies. 1. The incidence of endemic goitre in three areas in Australia. Med J Aust 1948; 21: 637-639. Hales I. Studies in diseases of the thyroid gland [MD thesis] Sydney: University of Sydney, 1971. Boyages S. Iodine deficiency disorders. J Clin Endocrinol Metab 1993; 77: 587-591. Clugston GA, Hetzel BS. Iodine. In: Shils ME, Olson JA, Shike M, editors. Modern nutrition in health and disease. 8th ed. Vol. 1. Philadelphia: Lea and Febiger, 1994; 252-263. Delange F. The disorders induced by iodine deficiency. Thyroid 1994; 4: 107-128. Hollowell JG, Staehling NW, Hannon WH, et al. Iodine nutrition in the United States. Trends and public health implications: iodine excretion data from the National Health and Nutrition Examination Surveys I and III (1971-1974 and 1988-1994). J Clin Endocrinol Metab 1998; 83: 3401-3408. Smyth PPA, Hetherton AMT, Smith DF, et al. Maternal iodine status and thyroid volume during pregnancy: correlation with neonatal iodine intake. J Clin Endocrinol Metab 1997; 82: 2840-2843. Dunn JT. What's happening to our iodine? [editorial]. J Clin Endocrinol Metab 1998; 83: 3398-3400. World Health Organization Nutrition Unit. Indicators for assessing iodine deficiency disorders and their control through salt iodization. Document No. WHO/NUT 94.6. Geneva: WHO, 1994: 36. McElduff A. Measurement of free thyroxine levels (fT4) in pregnancy. Aust N Z J Obstet Gynaecol 1999; 39: 158-161. May SL, May WA, Bourdoux PP, et al. Validation of a simple, manual urinary iodine method for estimating the prevalence of iodine-deficiency disorders, and interlaboratory comparison with other methods. Am J Clin Nutr 1997; 65: 1441-1445. McElduff A, Shuter B, Cooper R, et al. Measuring renal function in patients with diabetes mellitus. J Diabetes Complications 1997; 11: 225-229. SPSS [computer program], version 6.0. Chicago, Ill: SPSS Inc, 1996. Silva JE, Silva S. Interrelationships among serum thyroxine, triiodothyronine, reverse triiodothyronine, and thyroid-stimulating hormone in iodine-deficient pregnant women and their offspring: effects of iodine supplementation. J Clin Endocrinol Metab 1981; 52: 671-677. Glinoer D, De Nayer P, Bourdoux et al. Regulation of maternal thyroid during pregnancy. J Clin Endocrinol Metab 1990; 71: 276-287. Glinoer D, Delange F, Laboureur I, et al. Maternal and neonatal thyroid function at birth in an area of marginally low iodine intake. J Clin Endocrinol Metab 1992; 75: 800-805. Valiex P, Zarabska M, Preziosi P, et al. Iodine deficiency in France [letter]. Lancet 1999; 353: 1766-1767. Moreno-Reyes R, Suetens C, Mathieu F, et al. Kashin-Beck osteoarthropathy in rural Tibet in relation to selenium and iodine status. N Engl J Med 1998; 339: 1112-1120. Received 13 Apr, accepted 21 Aug, 1999 Authors' details Royal North Shore Hospital, St Leonards, NSW. Jenny E Gunton, MB BS, Endocrine Fellow, Department of Endocrinology. Graham Hams, MAppSc, Senior Staff Scientist, Pacific Laboratory Medicine Services. Marcelle Fiegert, BEd, MNutri Diet, Dietitian, Department of Nutrition. Aidan McElduff, FRACP, PhD, Senior Staff Specialist in Endocrinology, Department of Endocrinology. Reprints: Dr J E Gunton, C/- Clinic 1, Royal North Shore Hospital, St Leonards, NSW 2065. jennyegAThotmail.com. 1: Iodine deficiency disorders Maternal Goitre Hypothyroidism Decreased fertility Miscarriage Fetal Stillbirth Neonatal Cretinism Increased mortality Goitre Hypothyroidism Back to text 2: The iodine situation in AustraliaIn the past, an increased incidence of goitre and iodine deficiency was documented in certain parts of Australia.3-5 Prevention of iodine deficiency in industrialised countries most commonly relies on iodised salt, iodine in milk, or iodine-supplemented bread.1,6-8 The upper limit of the recommended daily intake of salt (NaCl) is 100 mmol, or 6 g (a heaped teaspoon); 100 mmol of iodised salt per day would provide 175-240 µg of iodine. However, most salt is incorporated into foods before purchase, and the three major Australian manufacturers of processed food we contacted all reported using non-iodised salt only. Non-iodised table salt is readily available, and may be used more frequently than in the past as campaigns to use iodised salt are forgotten. (We reviewed supermarket shelves in our local area, and found that the space allocated for display suggests that more non-iodised than iodised salt is purchased.) In the United States, only 50%-60% of salt currently consumed is iodised.9 Milk products, which used to contain significant concentrations of iodine (up to 300 µg/100 mL) by virtue of iodine-containing solutions used to clean the milk vats, now contain low levels of iodine because volatile cleaning solutions are used (Dairy Farmers Association, Nutrition Panel for Milks, personal communication). While the incidence of iodine deficiency and goitre was decreased by legislation requiring iodine supplementation of bread in 1966,3 this is no longer a requirement (because of concerns about an increased incidence of thyrotoxicosis). Marine fish, shellfish, seaweed and kelp contain high amounts of iodine,1,7 and such ocean seafood, as well as added iodised salt, provide most of the iodine in the Australian diet. However, many people may consume these products rarely, if at all. Back to text Back to text 4: Iodine status resultsGroupPregnant womenPostpartum womenPatients with diabetesVolunteersNumber of participants8126135 19Age (years)Mean (± 2 SD)32.9 ± 9.835.3 ± 11.350.1 ± 35.3*49.5 ± 17.4*Median (95% CI)34 (24-42)35 (25-42)50 (25.7-83.0)49 (45.3-53.8)Spot iodine concentration (µg/L)Median10479‡65† 64(95% CI)(89-129)(44-229)(58-89)(54-75)No. of participants (%) withSevere to moderate deficiency < 50 µg/L16 (19.8%)5 (19.2%)46 (34.1%)5 (26.3%)Mild deficiency 51-100 µg/L24 (29.6%)9 (34.6%)51 (37.8%)9 (47.4%)Normal iodine status > 100 µg/L41 (50.6%)12 (46.1%)38 (28.1%)5 (26.3%)Iodine/creatinine ratio (µg iodine/g creatinine)Median159131114 108(95% CI)(169-232)(106-218)(93-505)(84-209)No. of participants (%) withSevere to moderate deficiency< 50 µg iodine/g creatinine6 (7.4%)2 (7.7%)7 (5.2%)1 (5.3%)Mild deficiency 51-100 µg iodine/g creatinine 22 (27.2%)9 (34.6%) 50 (37.0%)5 (26.3%)Normal iodine status > 100 µg iodine/g creatinine53 (65.4%)15 (57.7%)78 (57.8%)13 (68.4%)* P < 0.001 for comparison with pregnant women. †P < 0.01 for comparison with pregnant women. ‡P < 0.05 for comparison with patients with diabetes. Back to text 5: Levels of thyroid-stimulating hormone and free thyroxine compared with iodine status in pregnant women and patients with diabetesThyroid-stimulating hormone (µIU/mL)Free thyroxine (pmol/L)GroupNo. Mean (± 2 SD)Mean (± 2 SD)Pregnant women70Normal iodine status*411.56 ± 0.8012.9 ± 3.70Mild deficiency†231.67 ± 0.9012.5 ± 3.00Severe to moderate deficiency‡16 1.56 ± 0.7712.1 ± 2.25Patients with diabetes121Normal iodine status*342.1 ± 3.1015.0 ± 2.60Mild deficiency†451.9 ± 1.2015.0 ± 2.60Severe to moderate deficiency‡42 2.6 ± 2.4014.5 ± 2.90 * > 100µg/L. † 51-100 µg/L. ‡ < 50 µg/L. Back to text

Jenny E Gunton · Graham Hams · Marcelle Fiegert · Aidan McElduff

Research End-stage renal disease in Aboriginals in New South Wales: a very different picture to the Northern Territory Alan Cass, Adrian G Gillin and John S Horvath MJA 1999; 171: 407-410 Abstract - Introduction - Methods - Results - Discussion - Acknowledgements - References - Authors' details - - More articles on Aboriginal health Abstract Objectives: To compare the incidence of end-stage renal disease (ESRD) among Aboriginals in New South Wales with the incidence among Aboriginals in the Northern Territory, and to compare the patterns of ESRD among Aboriginals and non-Aboriginals in NSW. Design: Secondary data analysis of information from unpublished and published Australia and New Zealand Dialysis and Transplant Registry reports. Main outcome measures: Average annual incidence of ESRD (persons per million); form of renal replacement therapy; mortality at 31 March 1998; patient and graft survival one and five years after transplant. Results: Each year in NSW, 5-17 new Aboriginal patients are treated for ESRD. There was no increase in the average annual incidence of ESRD among NSW Aboriginals (118 per million in 1988-1989 and 111 per million in 1996-1997), whereas incidence in the NT increased from 255 per million to 800 per million. In NSW, ESRD was attributed to diabetes in 32% of Aboriginal patients, compared with 13% of non-Aboriginal patients (P < 0.001). In NSW, Aboriginal patients were younger and more likely to be female, a pattern similar to that in the NT. The outcome of ESRD treatment is not significantly different between Aboriginals and non-Aboriginals in NSW. Conclusion: There is a different pattern of incidence of ESRD and of outcomes with treatment among Aboriginals in NSW compared with those in the NT. A possible explanation is that the lower incidence in NSW reflects less profound socioeconomic disadvantage and better access to primary and specialist care. Introduction Indigenous Australians experience high morbidity and mortality due to end-stage renal disease (ESRD). In the Northern Territory, the average annual incidence of ESRD for Aboriginals in 1988-1993 was 17.4 times that for non-Aboriginals,1 a disparity made more apparent by age adjustment.2 The number of dialysis treatments in the NT is doubling every two years.3The 30 June 1996 estimate of the Indigenous population (386 049) represented 2.1% of the total Australian population,4 but Aboriginals constitute 5% of the Australian members of the Australia and New Zealand Dialysis and Transplant Registry (ANZDATA).5 Whether the increasing incidence and prevalence of renal disease is occurring in NSW as well as nationally and in the NT has not been well documented. The aims of our study were to document the number of new Aboriginal patients with ESRD in NSW during 1987 to 1998 and compare recent trends in incidence with national and NT data, and to compare for Aboriginal and non-Aboriginal patients in NSW: the patterns of aetiology of ESRD; the demographic characteristics at the time of being entered into ANZDATA; the outcomes for individuals who were notified to the Registry during the period 1987 to 1998; and patient and graft survival for those who received transplants during the period 1987 to 1998. Methods Information was obtained from ANZDATA Annual Reports (12 and 18-21)6-10 and unpublished data from ANZDATA. All nephrology units in Australia and New Zealand that provide dialysis or transplant services submit detailed six-monthly reports to ANZDATA. The reports give information regarding new patients accepted onto treatment programs, deaths that have occurred, and any alteration in treatment for current patients, including changing the mode of dialysis or receiving a transplant. For NSW, the NT and across Australia, we analysed: the number of Aboriginal patients entered into ANZDATA from 1987 to 1998 in NSW, the NT and across Australia; and the average annual incidence of ESRD. For NSW only, we analysed: primary renal disease diagnostic category; mean age; outcome data at 31 March 1998 for people who had been entered into ANZDATA since 1 January 1987. Outcomes were categorised as death, functioning transplant, haemodialysis, continuous ambulatory peritoneal dialysis (CAPD), and loss to follow-up or having moved interstate; causes of death, categorised into cardiac, vascular, infection, social, malignancy and other; patient survival and graft survival for transplants performed between 1 January 1987 to 31 March 1998. The average annual incidence of ESRD was calculated using Australian Bureau of Statistics (ABS) estimates and projections of the Aboriginal and Torres Strait Islander population for the years between actual Census counts. These estimates are based upon current trends in fertility and mortality and take into account an increasing propensity for people to identify themselves as being of Indigenous origin. The definition of "Aboriginality" from both data sources (ANZDATA and ABS) relies upon self-identification. Annual incidence was calculated as an average for each two-year period, as there is marked variability in the number of new patients per year, and small absolute numbers. Statistical analysis of demographic and outcome data was performed using STATA 5.0.11 A t test of means and χ2 test or Fisher's exact test of proportions were performed. Survival analysis was performed at the ANZDATA Registry. Data were provided as actuarial life-table estimates and the log rank test was performed. The data were entered into STATA 5.0 and graphed. Results Incidence Each year in NSW, 5-17 new Aboriginal patients are treated for ESRD (Table 1). The average annual incidence of ESRD among Aboriginals in NSW for the two years 1988-1989 was 118 per million. This remained substantially unchanged at 111 per million in 1996-1997. During the same period the average annual incidence of ESRD in Aboriginals across Australia increased significantly (Table 1); in the NT, the rise in incidence was more marked. The crude incidence for non-Aboriginals in NSW was 78 per million for 1993-1997, a slight rise from the 1980s due to increased acceptance for dialysis of patients over the age of 65 years. Diagnostic categories Diabetes, glomerulonephritis and hypertension are the most common primary renal diseases among Aboriginals with ESRD in NSW (Table 2). Diabetes is listed as the primary renal disease in 32% of Aboriginals, compared with 13% of non-Aboriginals (P < 0.001). Analgesic nephropathy affects a significantly greater proportion of the non-Aboriginal population, mainly among females: 28% in non-Aboriginal females, compared with 10% in Aboriginal females. No definite diagnosis was recorded for 11% of Aboriginals, compared with 5% of non-Aboriginals (P = 0.004). Demographics Significantly more Aboriginal females than males entered the ESRD program in NSW (Table 3), the reverse pattern to non-Aboriginals (P = 0.03). The Aboriginal population was younger at entry to the program and there were significantly fewer people over the age of 65 years (P < 0.001). Outcome, or patient status at 31 March 1998, was not significantly different between the groups (P = 0.59). However, no attempt at age or sex standardisation has been made in this analysis. Causes of death The differences between Aboriginals and non-Aboriginals in NSW in causes of death approach statistical significance (P = 0.07). A significantly larger proportion of Aboriginals died from cardiovascular diseases (P = 0.01). A significantly smaller proportion of deaths were due to social reasons (P = 0.02). Patient and graft survival after transplant Aboriginal patients receiving transplants in NSW during the study period were younger (Figure 1), but experienced lower patient survival and graft survival rates (Figures 2a and 2b), although these differences were not significant (Table 4). Discussion Our data show no evidence of an epidemic of renal failure among Aboriginals in NSW, although the incidence remains higher than among the non-Aboriginal population. Aboriginal patients in NSW with ESRD are on average 10 years younger than non-Aboriginal patients, more likely to be female, and more likely to have diabetes and to die of cardiovascular disease. These features are similar to those reported in the NT.2 However, in NSW, there is no significant difference in outcome between Aboriginal and non-Aboriginal patients who have been entered into the Registry since January 1987, whereas, in the NT, survival is significantly worse in Aboriginal than in non-Aboriginal patients.2 The persistently high rate of withdrawal up to 1997 of NT Aboriginal people from ESRD treatment, about 25%,3 is not present in NSW. Our results show that diabetes, glomerulonephritis and hypertension are the prominent primary causes of ESRD among NSW Aboriginals. The rise in renal failure attributed to diabetes follows a similar pattern to that noted in Aboriginals across Australia.10 The pattern of primary causes of renal disease is consistent between Aboriginal populations in different States.10 Impediments to effective and culturally appropriate service delivery to Aboriginal patients have been postulated as reasons for poor survival and high withdrawal rates from treatment.12 Therapeutic programs have typically removed people from their cultural and social support networks by requiring patients to leave their land, families and communities.13 Unlike in the NT, South Australia and Western Australia, there are few remote, non-urbanised communities in NSW; Aboriginals in NSW reside predominantly in cities and rural towns. The higher proportion of Aboriginal patients with uncertain aetiology of their ESRD is consistent with a lower renal biopsy rate, which may relate to late referral and lack of access to renal specialist services. In NSW, specialist renal services are increasingly being provided where Aboriginals live. The Statistical Local Areas with the highest proportion of Indigenous people are Brewarrina (53.1%), Central Darling (25.3%), Bourke (24.5%) and Walgett (20.4%).14 CAPD training is now occurring in some larger rural centres, and haemodialysis facilities are provided in Bourke and Brewarrina. These initiatives, which significantly reduce the dislocation of patients from their community and remove impediments to the delivery of appropriate ESRD services, may facilitate improved survival. There is a tendency towards lower patient survival and graft survival among NSW Aboriginals compared with non-Aboriginals, despite the Aboriginal graft recipients' being younger. In the NT, graft and patient survival among Aboriginals are significantly worse at one and five years than among non-Aboriginals.2 A significant difference in survival is not evident in the NSW data. The larger 95% confidence intervals in the NSW Aboriginal group are a consequence of fewer transplants being performed: 36 among Aboriginal patients, compared with 1755 among non-Aboriginal patients. These numbers do not provide sufficient power to detect a significant difference between the groups. The pattern of rapidly increasing incidence of ESRD among Aboriginals across Australia, especially in the NT, is not seen in NSW. Spencer et al argue that the increase in the NT is real, not due to ageing of the Aboriginal population or improved ascertainment.2 The reason for this difference in incidence is not clear. It may be due to differences between the populations in apparent predisposition to renal disease or to differences in the prevalence of primary causes and promoters of chronic renal disease. The epidemic of disease in the NT is not only due to an increased prevalence of diabetes. Community screening studies show a prevalence of significant proteinuria in marked excess of the prevalence of diabetes or impaired glucose tolerance.2,15 In the NT, from 1988 to 1993, the average annual incidence of ESRD not attributable to diabetes was 350 per million per year among Aboriginals.1 Lower incidence of ESRD in NSW Aboriginals may reflect less profound socioeconomic disadvantage and readier access to effective primary and specialist care. However, there may be poor ascertainment, particularly in rural areas of NSW. Further study is indicated to analyse this question. Acknowledgements The data reported here have been supplied by the Australia and New Zealand Dialysis and Transplant Registry. The interpretation of these data is the responsibility of the authors and in no way should be seen as an official policy or interpretation of the Australia and New Zealand Dialysis and Transplant Registry. Dr Alan Cass is the recipient of a postgraduate research scholarship from the Centre for Kidney Research, New Children's Hospital, Sydney. We thank Dr Wendy Hoy, who critically reviewed the manuscript, and Dr Zhiqiang Wang, who provided statistical advice. References Hoy WE, Mathews JD, Pugsley DJ. Treatment of end-stage renal disease in the Top End of the Northern Territory: 1978-93. Nephrology 1995; 1: 307-313. Hoy WE, McFarlane R, Pugsley DJ, et al. Markers for cardiovascular and renal morbidity: expectations for an intervention programme in an Australian aboriginal community. Clin Exp Pharmacol Physiol 1996; 23: S33-S37. Spencer JL, Silva DT, Snelling P, Hoy WE. An epidemic of renal failure among Australian Aboriginals. Med J Aust 1998; 168: 537-541. Australian Bureau of Statistics. Experimental estimates of the Aboriginal and Torres Strait Islander population. Canberra: ABS, 1998. (Catalogue No. 3230.0.) Disney AP. Demography and survival of patients receiving treatment for chronic renal failure in Australia and New Zealand: report on dialysis and renal transplantation treatment from the Australia and New Zealand Dialysis and Transplant Registry. Am J Kidney Dis 1995; 25: 165-175. Disney APS, Collins J, Russ GR, et al. ANZDATA Registry Report 1989. Adelaide: Australia and New Zealand Dialysis and Transplant Registry, 1989. Disney APS, Collins J, Russ GR, et al. ANZDATA Registry Report 1995. Adelaide: Australia and New Zealand Dialysis and Transplant Registry, 1995. Disney APS, Collins J, Russ GR, et al. ANZDATA Registry Report 1996. Adelaide: Australia and New Zealand Dialysis and Transplant Registry, 1996. Disney APS, Collins J, Russ GR, et al. ANZDATA Registry Report 1997. Adelaide: Australia and New Zealand Dialysis and Transplant Registry, 1997. Disney APS, Collins J, Russ GR, et al. ANZDATA Registry Report 1998. Adelaide: Australia and New Zealand Dialysis and Transplant Registry, 1998. STATA statistical software [computer program]. Version 5.0. College Station, Texas: Stata Corporation, 1997. Bennett E, Manderson L, Kelly B, Hardie I. Cultural factors in dialysis and renal transplantation among aborigines and Torres Strait Islanders in north Queensland. Aust J Public Health 1995; 19: 610-615. Willis J. Fatal attraction: do high technology treatments for end-stage renal disease benefit aboriginal patients in central Australia? Aust J Public Health 1995; 19: 603-609. Australian Bureau of Statistics. Census of population and housing -- selected social and housing characteristics for statistical local areas, New South Wales and Jervis Bay. Canberra: ABS, 1996. (Catalogue no. 2015.1.) Van Buynder PG. The epidemiology of renal disease in Aboriginal Australians [Master of Public Health thesis]. Sydney: University of Sydney, 1991. (Received 12 Feb, accepted 19 Jul, 1999) Authors' details Menzies School of Health Research, Darwin, NT. Alan Cass, MB BS, FRACP, PhD student. Department of Renal Medicine, Royal Prince Alfred Hospital, Sydney, NSW. Adrian G Gillin, FRACP, PhD, Staff Specialist; John S Horvath, MB BS, FRACP, Professor. Reprints will not be available from the authors. Correspondence: Dr A Cass, Menzies School of Health Research, PO Box 41096, Casuarina, NT 0811. alancassATmenzies.edu.au Back to textBack to textBack to textBack to textBack to textBack to text

Alan Cass · Adrian G Gillin · John S Horvath

Research Is sudden infant death syndrome still more common in very low birthweight infants in the 1990s? Beverley Sowter, Lex W Doyle, Colin J Morley, Anne Altmann and Jane Halliday MJA 1999; 171: 411-413 Abstract - Introduction - Methods - Results - Discussion - References - Authors' details - - More articles on Paediatrics Abstract Objective: To determine the rate of sudden infant death syndrome (SIDS) in very low birthweight children (VLBW) relative to children with low (LBW) and normal birthweights. Design, setting and subjects: Cohort study of consecutive live births in Victoria, 1993-1997 inclusive. Main outcome measures: All sudden unexpected deaths in early childhood over this five-year period; all deaths from SIDS (defined as a sudden unexpected death without a definite pathological explanation); and the proportion of SIDS in live births in three birthweight subgroups (VLBW, 500-1499 g; LBW, 1500-2499 g; and normal birthweight, > 2499 g). Results: There were 316 028 live births (with known birthweight) in Victoria over the five-year period; 224 (0.71 per 1000 live births) died unexpectedly. In 10 of these deaths there was a definite pathological explanation, giving a rate of SIDS of 0.68 per 1000 live births. The rate of SIDS in VLBW children was 2.52 per 1000 live births, lower than the rate reported before the 1990s. The rate of SIDS in VLBW children was not significantly different from the rate in LBW children of 1.98 per 1000 live births (difference per 1000 live births, 0.53; 95% CI, 21.45 to 2.52), but was significantly higher than the rate in normal birthweight children of 0.59 per 1000 live births (difference per 1000 live births, 1.93; 95% CI, 0.06-3.79). Conclusions: The rate of SIDS in VLBW children has fallen in the 1990s, along with the overall fall in the rate of SIDS, but remains higher than that in normal birthweight children. Introduction Most parents learn to live with the fear that their baby may die of sudden infant death syndrome (SIDS). However, the fear of SIDS can be even greater for parents of babies with low birthweight, or those with babies who have been in intensive care or who have had apnoea. Before discharge, many parents attend education sessions on reducing the risks of SIDS and on infant resuscitation. At one of these sessions the parents may ask the difficult question: "Is my baby more likely to die of SIDS because he (or she) was so tiny when born?". Before the 1990s, children with very low birthweight (VLBW, 500-1499 g) were known to have a higher rate of SIDS than those with a normal birthweight (> 2499 g).1-3 With the advent of preventive measures, the overall rate of SIDS in Australia has fallen dramatically in the 1990s (from 1.87 per 1000 live births in 1990 to 0.78 per 1000 live births in 1995).4 However, it is unclear whether the rate of SIDS has also fallen in VLBW children. We aimed to determine the rate of SIDS in the 1990s for VLBW children relative to children of other birthweights (low birthweight [LBW], 1500-2499 g; and normal birthweight, > 2499 g). Methods We studied a cohort of all consecutive live births in Victoria during the five-year period from 1993 to 1997, inclusive, and recorded all sudden deaths in early childhood (divided into the first 28 days [the neonatal period], postneonatal infancy [29-365 days], and early childhood [more than 1 year]). SIDS was defined as a sudden unexpected death without definite pathological features to explain the death. In the SIDS group, some children were considered by the pathologist to have pathological features, but these were insufficient to explain the death. Some had only a minor condition, and in the remainder no pathological features were found. All births with unknown birthweight were excluded. Data sources Data on deaths were obtained from the annual reports of the Consultative Council on Obstetric and Paediatric Mortality and Morbidity,5-9 a government-legislated surveillance body. The Council collects data on all perinatal deaths from 20 weeks' gestation, and all infant and child deaths up to 14 years of age. Death registrations are forwarded directly from the Registry of Births, Deaths and Marriages, and information on all sudden unexpected deaths is supplemented by the Victorian State Coroner. These cases are then all reviewed and classified by an expert pathologist working in the field. Data on births in Victoria were supplied by the Perinatal Data Collection Unit of the Public Health and Development Division of the Department of Human Services. The Unit collects data on all births in Victoria from 20 weeks' gestation under a legislated notification system. Statistical analysis The proportions of deaths from SIDS (and 95% confidence intervals) in each of the birthweight subgroups (VLBW, 500-1499 g; LBW, 1500-2499 g; and normal birthweight, > 2499 g) were calculated,10 and comparisons made between the groups.10 Results Over the five-year period 1993-1997, there were 316 028 live births in Victoria for which birthweight was known. (Birthweight was not known for a total of 90 live births.) Over the same period, there were 224 sudden unexpected deaths with known birthweight, of which 10 (4.5%) had a definite pathological explanation, leaving 214 deaths from SIDS (0.68 per 1000 live births). Of these 214 children with SIDS, 23 (10.7%) died in the neonatal period (including one who died during the primary hospitalisation), 171 (79.9%) died in postneonatal infancy, and 20 (9.3%) died after the age of 1 year. VLBW children made up less than 1% of all live births over this period; the rate of SIDS in VLBW children was 2.52 per 1000 live births (Table). This was not significantly higher than the rate in LBW children of 1.98 per 1000 live births (difference per 1000 live births, 0.53; 95% CI, 21.45 to 2.52), but was significantly higher than the rate in normal birthweight children of 0.59 per 1000 live births (difference per 1000 live births, 1.93; 95% CI, 0.06-3.79). LBW children had a significantly higher rate of SIDS than normal birthweight children (difference per 1000 live births, 1.39; 95% CI, 0.69-2.09). Of the seven deaths in VLBW infants, definite pathological features were found in five, but these were insufficient to explain the death; three of these infants had respiratory disease (pneumonia, bronchiolitis, or tracheobronchitis). Discussion Our study showed no significant difference in the rate of SIDS between VLBW and LBW infants; however, the rate in each of these subgroups was significantly higher than in normal birthweight children. Before recommendations for reducing the risk of SIDS were introduced in 1991, the SIDS rate was higher in VLBW infants compared with those with normal birthweight, for both hospital and regional cohorts. In a VLBW hospital cohort, 1977-1978, the rate of SIDS before 2 years of age was 30.3 per 1000 live births (7/231);1 and in a regional cohort in New Zealand (children born in 1986), the incidence of SIDS in the VLBW group was 13 per 1000 live births, more than three times the rate of 4.0 per 1000 live births for all NZ children born in 1986.2 In a Californian study of 2962 children dying of SIDS between 28 days and 1 year of age in 1978-1982, the overall incidence was found to be 1.5 per 1000 live births. The highest incidence was in the VLBW group (7.5 per 1000 live births), decreasing to 1.3 per 1000 live births for the normal birthweight group.3 Studies have not shown evidence of a fall in the rate of SIDS in VLBW infants just before the recommendations were introduced. For example, in 1985-1991, the rate of SIDS in VLBW infants in the first year of life for singleton births in the United States remained relatively constant (average, 3.66 per 1000 live births) compared with rates for normal birthweight singleton infants (average, 1.07 per 1000 livebirths).11 Studies comparing the period before and after the recommendations have shown a change in rate of SIDS in VLBW children around the beginning of the 1990s. One study cited by l'Hoir et al12 estimated the rate of SIDS in VLBW children in the Netherlands to have decreased from 10 per 1000 in 1983 to 1 per 1000 (presumably live births) in 1995-96, a change in the rate of SIDS over time similar to that comparing our results with rates in the late 1980s. A more recent US report described a smaller reduction in the rate of SIDS in VLBW children before and after the recommendations about sleeping position, and the reduction was similar across birthweight subgroups (between 1991 and 1995 reductions of 37%, 36% and 30% were found for birthweight subgroups 500-1499 g, 1500-2499 g, and > 2499 g, respectively).13 Avoidance of risk factors probably explains the fall in the rate of SIDS in VLBW infants in the 1990s, the same reason that it has fallen for infants overall. The reason SIDS remains more prevalent in VLBW infants compared with normal birthweight infants may relate to the underlying pathological features, which were more common in VLBW infants with SIDS. So, what should we be telling parents when they ask if their tiny baby is more likely to die of SIDS? Firstly, VLBW and LBW children are at increased risk of SIDS. However, 399 out of 400 VLBW children, and 499 out of 500 LBW children, do not die of SIDS. Secondly, parents can help to decrease the risk of SIDS by following the recommendations: putting their baby to sleep supine, not smoking, and not allowing the baby to become overheated or covered over by bedding.4 Thirdly, as most VLBW children who died of SIDS had a definite pathological condition at autopsy, parents should seek medical advice early if the baby appears unwell in any way. Parents often recognise that their baby is unwell, but it is not always easy for them to decide what is a minor illness and when they need to seek medical advice. A scoring system such as "Baby Check" 14 can be used both by parents and general practitioners to help them to determine whether or not a baby is seriously ill. In a recent review of 37 sudden unexpected infant deaths, 3 (8%) scored very highly for serious illness on a retrospective score with Baby Check, suggesting that such a scoring system could have identified serious illness before death and led to appropriate treatment.15 The three VLBW children with significant respiratory disease in our study probably would have shown signs of illness before they died. In conclusion, the rate of SIDS in VLBW children has fallen from over 10 per 1000 live births before the 1990s to 2.5 per 1000 live births at the end of the 1990s, but remains higher than the rate in normal birthweight children. LBW children are also at greater risk of SIDS at the end of the 1990s. References Kitchen WH, Yu VYH, Lissenden JV, Bajuk B. Collaborative study of very-low-birthweight infants: techniques of perinatal care and mortality. Lancet 1982; 1: 1454-1457. Darlow BA, Horwood LJ, Mogridge N, Clemett RS. Prospective study of New Zealand very low birthweight infants: outcome at 7-8 years. J Paediatr Child Health 1997; 33: 47-51. Grether JK, Schulman J. Sudden infant death syndrome and birth weight. J Pediatr 1989; 114: 561-567. Henderson-Smart DJ, Ponsonby AL, Murphy E. Reducing the risk of sudden infant death syndrome: a review of the scientific literature. J Paediatr Child Health 1998; 34: 213-219. The Consultative Council on Obstetric and Paediatric Mortality and Morbidity. Annual Report for the Year 1993, incorporating the 32nd Survey of Perinatal Deaths in Victoria. Melbourne, 1994. The Consultative Council on Obstetric and Paediatric Mortality and Morbidity. Annual Report for the Year 1994, incorporating the 33rd Survey of Perinatal Deaths in Victoria. Melbourne, 1995. The Consultative Council on Obstetric and Paediatric Mortality and Morbidity. Annual Report for the Year 1995, incorporating the 34th Survey of Perinatal Deaths in Victoria. Melbourne, 1996. The Consultative Council on Obstetric and Paediatric Mortality and Morbidity. Annual Report for the Year 1996, incorporating the 35th Survey of Perinatal Deaths in Victoria. Melbourne, 1997. The Consultative Council on Obstetric and Paediatric Mortality and Morbidity. Annual Report for the Year 1997, incorporating the 36th Survey of Perinatal Deaths in Victoria. Melbourne, 1998. Gardner MJ, Altman DG. Statistics with confidence - confidence intervals and statistical guidelines. London: BMJ, 1989. Bigger HR, Silvestri JM, Shott S, Weese-Mayer DE. Influence of increased survival in very low birth weight, low birth weight, and normal birth weight infants on the incidence of sudden infant death syndrome in the United States: 1985-1991. J Pediatr 1998; 133: 73-78. l'Hoir MP, Engelberts AC, van Well GT, et al. Case-control study of current validity of previously described risk factors for SIDS in the Netherlands. Arch Dis Child 1998; 79: 386-393. Malloy MH. Birth weight and gestational age specific sudden infant death syndrome (SIDS) mortality: 1991 vs 1995. Pediatr Res 1999; 45: 249A. Morley CJ, Thornton AJ, Cole TJ, et al. Baby Check: a scoring system to grade the severity of acute systemic illness in babies under 6 months old. Arch Dis Child 1991; 66: 100-105. Cole TJ, Gilbert RE, Fleming PJ, et al. Baby Check and the Avon infant mortality study. Arch Dis Child 1991; 66: 1077-1078. (Received 21 Jun, accepted 6 Sep, 1999) Authors' details Division of Paediatrics, Royal Women's Hospital, Melbourne, VIC. Beverley Sowter, RN, Case Manager. Lex W Doyle, MD, FRACP, Paediatrician; and Associate Professor, Department of Obstetrics and Gynaecology, and Department of Paediatrics, University of Melbourne. Colin J Morley, MD, FRACP, Paediatrician. Consultative Council on Obstetric and Paediatric Mortality and Morbidity, Melbourne, VIC. Anne Altmann, MB BS(Hons), MPH, FAFPHM, Epidemiologist. Victorian Perinatal Data Collection Unit, Melbourne, VIC. Jane Halliday, PhD, Epidemiologist. Reprints will not be available from the authors. Correspondence: Associate Professor L W Doyle, Division of Paediatrics, The Royal Women's Hospital, 132 Grattan Street, Carlton, VIC 3053. l.doyleATobgyn-rwh.unimelb.edu.au Sudden unexpected deaths in early childhood in Victoria, 1993-1997Birthweight subgroup 500-1499 g1500-2499 g> 2499 gTotalLive births278115 630297 617316 028Sudden unexpected deaths (≤ 28 days)07*1623Sudden unexpected deaths (> 28 days)724170201Total sudden unexpected deaths731186224Sudden unexpected deaths explained at autopsy001010Total SIDS731176214Pathological featuresDefinite586174Minor21887107None052833Rate of SIDS (per 1000 livebirths) 2.521.98 0.590.68(95% CI)(1.04-5.19)(1.29-2.68)(0.50-0.68)(0.59-0.77)Data are numbers of infants, unless indicated otherwise. * One infant died while still in hospital after birth. SIDS = Sudden infant death syndrome (sudden unexpected deaths, excluding those explained at autopsy). Back to text

Beverley Sowter · Lex W Doyle · Colin J Morley · Anne Altmann · Jane Halliday

Research A South Australian population survey of the ownership of asthma action plans Richard E Ruffin, David Wilson, Anne Marie Southcott, Brian Smith and Robert J Adams MJA 1999; 171: 348-351 Abstract - Introduction - Methods - Results - Discussion - References - Authors' details - - More articles on Respiratory medicine Abstract Objective: To examine the relationships between ownership of written asthma action plans, asthma morbidity, use of devices, and patients' perceptions of their asthma management. Design and setting: A random population survey (in 1996) of the South Australian population aged 15 years or over, using interviewers to administer a questionnaire. Participants: People who reported that they had current, doctor-diagnosed asthma. Main outcome measures: Prevalence of written asthma action plans; night-time awakenings from asthma; ownership of peak flow meters; and people's perceptions of their asthma management. Results: The ownership of asthma action plans by people with self-reported asthma was 33% and has declined since 1995 (42%; P < 0.001). Fifteen per cent were wakened weekly or more frequently by asthma symptoms. These people were more likely to have a peak flow meter and a written action plan, but less likely to consider they had been provided with enough information about their asthma, to feel comfortable managing their asthma, or to find it easy to see their doctor. Having a written asthma action plan was associated with regular corticosteroid use, understanding asthma, having enough information and owning a peak flow meter. Conclusions: Ownership of asthma action plans in South Australia is suboptimal. Before we develop new strategies to improve asthma outcomes, we must determine whether there is a need to target people with less severe asthma and/or improve the use of guidelines by health professionals. Introduction The Australian and New Zealand Asthma Guidelines, developed in 1989, provide a list of objectives that would be desirable to achieve for every person with asthma,1 and studies of the use of asthma management plans have shown improved health outcomes for people with asthma.2-4 In Australia there is evidence that the promotion of asthma plan guidelines by the Thoracic Society of Australia and New Zealand and the National Asthma Campaign has led to increased uptake of plans.1 In South Australia the prevalence of adults with asthma reporting that they had a written action plan almost doubled between 1992 and 1995.5 However, there is evidence that asthma management is not ideal.6-8In one study in Victoria, 45% of people who died of their asthma had been assessed as having only a history of mild or moderate asthma.6 In another study in Victoria that examined the asthma knowledge of asthma patients, the median score obtained was less than 50%.7The effective implementation of asthma management plans in Australia to date has been seriously questioned by some investigators,8 and Bauman et al have concluded that the treatment and management of asthma is suboptimal.9 Our study aimed to provide representative population information on the ownership of written asthma action plans and the relationship to asthma morbidity and management factors. Methods Survey The data for this study were collected in the 1996 South Australian Health Omnibus Survey,10 a representative survey of people aged 15 years or older (n = 3010; response, 71%) . The survey was a multistage, systematic, clustered area sample of people who live in metropolitan Adelaide and major country centres with a population of over 1000. The survey was selected from a random sample of Australian Bureau of Statistics collector districts. Within each collector's district a random starting point was selected and from this point 10 households were selected using a fixed skip interval. Hotels, motels, hospitals, nursing homes and other institutions were excluded. The person whose birthday was next in each selected household was interviewed in their home by trained health interviewers. There was no replacement for non-respondents. Up to five call-backs were made in an attempt to interview the selected person. The data were weighted by age, sex, and geographic region to the estimated resident population data so that the analysis would be representative of the South Australian population. The part of the survey form dealing with asthma is shown in the Box. A person was classified as having current asthma if they answered yes to the first three questions. Social class was determined by referring to the gradation of occupational prestige given in the Australian Standard Classification of Occupations.11 An asthma action plan was defined as "written instructions of what to do if your asthma is out of control." Data analysis Possession of an asthma management plan and frequency of wakening at night with asthma were used as the two dependent variables for univariate analyses,12 which examined the associations between these variables and reported asthma management, knowledge and attitudes to management. Before conducting multiple logistic regression analyses, the explanatory variables were examined for collinearity or interactions. Stratified analyses were used to check homogeneity of associations across different levels of predictor variables. Smoking status and the information that people with asthma perceive they have for dealing with worsening asthma were found to interact, with an effect of these variables on worsening asthma. An interaction term for the two independent variables was included in the logistic regression analysis for frequency of wakening at night. This interaction term proved significant (P = 0.03), indicating the need to split the model and conduct separate logistic regression analyses of smokers and non-smokers. Therefore, we conducted three logistic regression analyses, using "asthma plan", "waken weekly-non-smokers" and "waken weekly-smokers" as the three response variables. All variables found to be significant at the univariate stage (ie, P = 0.25)13 were entered into each logistic regression. Insignificant variables were progressively omitted until satisfactory models were found that explained possession of an asthma plan and frequency of wakening at night. Results The prevalence of asthma was 11.6% (95% CI, 10.3%-12.9%). Of the 349 survey respondents with asthma, 33% (95% CI, 30.8%-35.2%) had a written asthma action plan and 15.2% (95% CI, 13.7%-15.7%) were awakened by asthma weekly or more frequently. Age, sex, migrant status, education level and socioeconomic status made no significant difference to the rate of possession of an asthma action plan or the rate of wakening with asthma weekly or more frequently. Variables significantly associated with ownership of an asthma action plan at the univariate level are shown in Table 1; those significantly associated with wakening with asthma at night are shown in Table 2. In the multivariate analysis (Table 3), the variables that best described those who had an asthma action plan were: using corticosteroids, understanding the effects of worsening asthma, having a peak flow meter, and believing they have enough information to deal with worsening asthma. The variables that best described non-smokers who waken weekly or more often were: having a peak flow meter, having an asthma action plan, not believing they have enough information to deal with worsening asthma, and not feeling comfortable taking care of their asthma. Only one variable -- not finding it easy or convenient to access their doctor about asthma -- explained wakening weekly or more often for smokers. Discussion The data obtained in this representative population study paint a bleak picture of the effectiveness of asthma management in Australia. As even people with mild asthma can die of the disease,6 every person with asthma may need an action plan. Yet, seven years after the promulgation of Australian guidelines on the implementation of asthma action plans,1 only 33% of people with diagnosed asthma had a written plan. The current level of plan ownership is significantly lower than the 42.1% (P < 0.001) reported 12 months earlier using the same survey methods.5 This may mean that vigilance regarding asthma management is declining. Our study has some limitations. We have no objective data to identify levels of asthma severity and asthma control in the respondents. The validity of the perception questions as repeatable measures has not been verified. Because we focused on ownership of written action plans, our study does not tell us whether patients are making appropriate use of these plans or of verbal instructions. In the multivariate analyses, only use of preventer medication, ownership of a peak flow meter and self-reported understanding of asthma were associated with plan ownership. This association could suggest that the more severe cases have better asthma management. Proof for this requires prospective measures (eg, lung function, medication doses) to assess the two critical factors of asthma severity and asthma control. Such knowledge could inform us of the potential need to target people with less severe asthma. People with asthma with nocturnal symptoms were more likely than those without nocturnal symptoms to report possession of a peak flow meter, and to have asthma action plans, but were less likely to consider they had been provided with enough information about their asthma (non-smokers) or to find it easy to see their doctor (smokers) (Table 2). Thus, although asthmatics with a higher level of morbidity are more likely to receive physical materials to assist in self-care, they continue to have greater unmet needs for general practitioner access and information about asthma self-management. What is the way forward? Randomised studies of the implementation of asthma plans show that good educational and skill objectives can be achieved.2,14-16 However, the complexity of the asthma management problem makes it impossible to provide for every contingency the patient might face in dealing with asthma. Asthma management decisions can be difficult, because the patient, the daily situation, the science base and the disease are constantly changing.2 The objectives of patient asthma management are the development of skills and positive attitudes to problem-solving, accompanied by sufficient knowledge to make sense of changing morbidity and symptoms. A controlled trial evaluation of a brief asthma education program (2.5 to 3 hours group work) demonstrated substantial changes in illness behaviour.2 These programs show that giving the patient the necessary skills cannot be achieved within the normal constraints of general practice. Randomised controlled trials of asthma clinics, where there is an emphasis on self-management, have demonstrated improvements in a range of morbidity and other health-related outcomes in a community-based setting.17 We must identify other ways of training the patient and focus the clinician on that part of the education program that can be managed in general practice. There is evidence that regular review improves asthma outcomes.18 It is pertinent to consider the possibility that inadequate use of guidelines by health professionals may be contributing to the fall in action plan ownership. Ways to improve use of guidelines need to include developments in information technology. Trostle has suggested that the inability of some people to comply with a treatment regimen is an unavoidable byproduct of collisions between the clinical world and other competing worlds of work, family, friends and recreation.19 Often, the process required to inform and empower the patient is more than an educational task. Some of the traditional models of patient education based on health beliefs or compliance frameworks have been seriously questioned.20,21 Patients have to fit their medical problems and medical regimens into the context of their daily lives. In doing so they will vary their compliance with advice and instructions to accommodate the social, psychological, economic and physical influences which are part of their lives.21 We need research that clearly articulates the complexity and variability of how asthma management fits into the context of individual patients' lives. The National Asthma Campaign has provided a guide for health professionals to assist with effective patient communication, which is a starting point for corrective strategies.22 We therefore know what optimal management of asthma is, but its attainment is elusive. Future strategies must be guided by studies defining asthma severity and asthma control, studies identifying factors that influence the use of guidelines by professionals, and studies elucidating the context of asthma management for the individual. References Woolcock A, Rubinfeld AR, Seale P, et al. Asthma management plan, 1989. Med J Aust 1989; 151: 650-653. Yoon R, McKenzie DK, Bauman A, Miles DA. Controlled trial evaluation of an asthma program for adults. Thorax 1993; 48: 1110-1116. Comino EJ, Mitchell CA, Bauman A, et al. Asthma management in eastern Australia. Med J Aust 1996; 164: 403-406. Beasley R, Cushley M, Holgate ST. A self-management plan in the treatment of adult asthma. Thorax 1989; 44: 200-204. Adams R, Ruffin R, Wakefield M, et al. Asthma prevalence, morbidity and management practices in South Australia, 1992-1995. Aust N Z J Med 1997; 27: 672-679. Robertson C, Rubinfeld AR, Bowes G. Deaths from asthma in Victoria: a twelve-month survey. Med J Aust 1990; 152: 511-517. Rubinfeld AR, Dunt DR, McLure BG. Do patients understand asthma? A community survey of asthma knowledge. Med J Aust 1988; 149: 526-530. Bauman A, Young L, Peat JK, et al. Asthma under-recognition and under-treatment in an Australian community. Aust N Z J Med 1992; 22: 36-40. Bauman A, Mitchell CA, Henry RL, et al. Asthma morbidity in Australia: an epidemiological study. Med J Aust 1992; 156: 827-831. Wilson D, Wakefield M, Taylor A. The South Australian Health Omnibus Survey. Health Promotional J Aust 1992; 2: 47-49. Kelley JL, Evans MDR. Using ASCO for socio-economic analysis: assessment and conversion into status and prestige indices. Canberra: Research School of Social Sciences, Australian National University, 1988. SPSS for Windows. Release 8.0 [computer program]. Chicago, IL: SPSS Inc, 1998. Hosmer DW, Lemeshow S. Applied logistic regression. New York: John Wiley, 1989. Wilson-Pessano SR, McNabb WL. The role of patient education in the management of childhood asthma. Prev Med 1985; 14: 670-687. Clark NM, Feldman CH, Evans D, et al. Managing better: children, parents and asthma. Patient Educ Counsell 1986; 8: 27-38. D'Sousa WD, Crane J, Burgess C, et al. Community-based asthma care: trial of a "credit card" asthma self-management plan. Eur Respir J 1994; 7: 1260-1265. Lahdenso A, Haajtela T, Herrala J, et al. Randomised comparison of guided self-management and traditional treatment of asthma over one year. BMJ 1996; 312: 748-752. Asthma management handbook 1998. Melbourne: National Asthma Campaign, 1998. Trostle JA. Medical compliance as an ideology. Soc Sci Med 1988; 18: 1299-1308. Carter WB. Psychology and decision making: modelling health behaviour with multiattribute theory. J Dental Educ 1992, December: 800-807. Morris SL, Schulz RM. Medication compliance: the patients' perspective. Clin Ther 1993; 15: 593-606. Asthma adherence: a guide for health professionals. Melbourne: National Asthma Campaign, 1999. (Received 19 Apr, accepted 26 Aug, 1999) Authors' details The Queen Elizabeth Hospital, Adelaide, SA. Richard E Ruffin, FRACP, MD, Head, Division of Medicine, and Michell Professor of Medicine, University of Adelaide; Anne Marie Southcott, MB BS(Hons), FRACP, Acting Director, Respiratory Medicine; Brian Smith, PhD, FRACP, Director, Clinical Epidemiology and Health Outcomes Unit, and Senior Lecturer, University of Adelaide. Centre for Population Studies in Epidemiology, Department of Human Services, Adelaide, SA. David Wilson, PhD, MPH, Head. Channing Laboratory, Brigham and Women's Hospital, Harvard Medical School, Boston, MA, USA. Robert J Adams, MB BS, FRACP, Research Fellow. Reprints: Professor R E Ruffin, Department of Medicine, University of Adelaide, The Queen Elizabeth Hospital Campus, 28 Woodville Road, Woodville, SA 5011. rruffinATmedicine.adelaide.edu.au Survey questions Have you ever had asthma? Was your asthma confirmed by a doctor? Do you still have asthma? If answer yes to these questions, then: Do you have an asthma action plan (written instructions of what to do if your asthma is out of control)? Do you have a home nebuliser for asthma treatment? Interviewer: if "yes" prompt "have you used it in the last 12 months?" What preventive (not reliever) medicine do you use regularly for your asthma? None - Intal - Becotide - Becloforte Pulmicort - Tilade - Other (specify) Which of the statements shown on this card do you feel are true of your preventer medication? Works by relieving narrowed breathing tubes quickly Needs to be used when you feel unwell Works by slowly reducing inflammation in the breathing tubes Must not have the dose changed Don't know How often do you awaken during the night with asthma? Nightly - Most nights - About twice a week - Weekly - Monthly - Less often than monthly Only at certain times of the year (ie seasonal) Never In the last 12 months have you had any hospital admissions for asthma where you stayed at least one night in hospital? In the last 12 months have you had any days lost from work, school or home duties from asthma? How many days would you estimate? What would you do if you had a bad attack of asthma and six puffs of your reliever (eg, ventolin, respolin) had not helped? Which of these statements shown on this card most closely matches what you would be likely to do? Wait another two hours and take more reliever medication Seek medical advice Take another six puffs of reliever medication and see what happens Call an ambulance Get someone to take you to hospital Do something else (specify) What feelings would you have if you had to get help for a bad attack of asthma? Which of the statements shown on this card most closely match how you would be feeling? You feel that you have failed You would feel embarrassed You do not want to bother others It is the right thing to do You know you will be OK because of past experience Something else (specify) I am now going to read out a number of statements and show you a card for each of them. Could you please tell me which number from 1 to 5 best reflects the way you feel. I am the sort of person who understands all about my asthma Always - Often - Sometimes - Rarely - Never - Don't know/other If I took care of my asthma myself, most of the time, I would . . . Manage well Manage sometimes Not manage at all Don't know/other If I were having a severe attack of asthma I would feel comfortable about going to a doctor or hospital Very comfortable Comfortable Don't mind Uncomfortable Not comfortable at all Don't know/other Going to see a doctor for help with my asthma is Easy and convenient Not easy nor convenient Don't know/other I have information to use (such as "Asthma Action Plan" or other instructions) to deal with worsening asthma. Yes, all I need Some No, none at all Don't know/other Other questions were asked about smoking status, educational level and migrant status. Back to text 1: Variables associated with ownership of an action plan*VariableSubjects (n = 349)%Odds ratioWrong about CS effect10332.01.0Correct about CS effect162†44.41.7 (0.98-2.94) P = 0.04Don't use CS regularly14721.81.0Use CS regularly20241.62.56 (1.54-4.26) P < 0.01No home nebuliser28229.11.0Have home nebuliser6750.72.51 (1.41-4.48) P < 0.01No peak flow meter29626.61.0Have peak flow meter5371.76.99 (3.50-14.14) P <0.01Don't always understand asthma8516.51.0Understand asthma26438.63.19 (1.65-6.27) P < 0.01Not enough information11715.51.0Enough information23255.56.77 (3.98-11.56) P <0.01Feel bad getting help4721.31.0Getting help OK30235.12.26 (1.04-4.90) P = 0.04No hospital admission within 12 months 33432.1 1.0Hospital admission1560.03.17 (1.00-10.32) P = 0.05No days lost from work/school30231.11.0Days lost from work/school4746.81.95 (1.00-3.79) P = 0.05CS = corticosteroids. * Variables tested but not found to be significant were: sex, age, migrant status, educational level, socioeconomic status, weight, access to doctor, uncomfortableness dealing with asthma, perception of dealing with asthma, comfortableness in going to hospital if required, exercise, smoking status, and smoking bans at home. †Only those who regularly used preventive medicine for their asthma (n = 265) were asked about its effects. Back to text 2: Variables associated with the likelihood of wakening with asthma on a weekly basis or more frequently*VariableSubjects (n = 349)%Odds ratioWrong about CS effect1038.71.0Correct about CS effect162†25.83.63(1.60-8.45) P < 0.01No home nebuliser28212.81.0Have home nebuliser6730.81.96 (0.96-3.95) P = 0.04No peak flow meter29612.81.0Have peak flow meter5330.83.02 (1.45-6.27) P < 0.01Not easy to see doctor6033.31.0Easy access to doctor28911.80.27 (0.13-0.53) P < 0.01Uncomfortable taking care of asthma 4836.2 1.0Comfortable taking care30112.30.25 (0.12-0.52) P < 0.01Perception of self-management good 3813.7 1.0Perception poor31129.42.62 (1.09-6.25) P = 0.0 3No days lost from work/school30213.61.0Days lost from work/school4727.72.42 (1.11-5.25) P = 0.02CS = Corticosteroids. * Variables tested but not found to be significant were: sex, age, migrant status, educational level, socioeconomic status, weight, regular use of corticosteroids, not always understanding asthma, not having enough information, feeling bad getting help, comfortableness in going to hospital if required, exercise, smoking status, smoking bans at home, and having an asthma action plan. †Only those who regularly used preventive medicine for their asthma (n = 265) were asked about its effects. Back to text 3: Logistic regression analyses of variables associated with ownership of an asthma management plan and frequency of wakening at night with asthma (n = 349)VariableOdds ratioAsthma planNo peak flow meter1.0Have peak flow meter4.32 (2.91-8.43)Don't always understand asthma 1.0Understand asthma2.01 (1.01-4.02)Not enough information1.0Enough information4.32 (2.11-8.85)Don't use corticosteroid regularly1.0Use corticosteroid regularly2.08 (1.21-3.58)Waken weekly (non-smokers)No peak flow meter1.0Have a peak flow meter7.32 (2.59-20.07)Not enough information1.0Enough information0.12 (0.04-0.39)Uncomfortable taking care of asthma1.0Comfortable taking care0.30 (0.14-0.77)No asthma action plan1.0Asthma action plan2.79 (1.09-7.15)Waken weekly (smokers)Not easy to see doctor1.0Easy to see doctor0.28 (0.10-0.79)All results significant, P < 0.05. Back to text

Richard E Ruffin · David Wilson · Brian Smith · Robert J Adams

Research Erectile dysfunction in the community: a prevalence study Carole B Pinnock, Alan M F Stapleton and Villis R Marshall MJA 1999; 171: 353-357 For editorial comment see Lowy Abstract - Introduction - Methods - Results - Discussion - Acknowledgements - References - Authors' details - - More articles on Urology Abstract Objective: To investigate the prevalence of erectile dysfunction (ED) in the South Australian community, and the influence of demographic and other risk factors. Design: Survey by mailed questionnaire (based on the University of California, Los Angeles prostate cancer index) of a subset (men who agreed to participate) of a probability sample of the South Australian community who completed a multiuser interview survey. Participants and setting: Men over the age of 40 in South Australia. Main outcome measures: Sexual desire, orgasm, ability to have an erection, adequacy (firmness) of erections for intercourse, frequency of erections when wanted, frequency of intercourse, nocturnal or morning erections, and history of prostate surgery; total sexual function score based on these. Results: 612 men (86.7%) agreed to answer the sexual function survey; 427 (69.8%) returned questionnaires. ED was strongly correlated with age in all seven domains of sexual function. Erections inadequate for intercourse affected 3% of 40-49-year-olds, increasing to 64% of 70-79-year-olds. The frequency of intercourse considered normal for age by men 50-69 years was 1-6 times weekly; the disparity between this and reported frequency increased in men over 60 years, as did the difference between sexual desire and potency. A history of vigorous exercise was protective across all ages. High triglyceride levels, blood pressure medication and non-cancer surgery for prostate disease were independent predictors of poor sexual function at older ages. High cholesterol level was an independent predictor of impotence. Conclusions: We found similar or higher levels of ED than in comparable overseas studies. Disparity between potency and desire was greatest, and hence the age group in whom demand for treatment may be highest, in those 60 years and older. Cardiovascular risk factors were predictors of ED in these older men, suggesting that prevention may benefit sexual function. Non-cancer prostate surgery may be a greater contributor to ED than previously realised. Introduction Establishing the prevalence of erectile dysfunction (ED) in the community is difficult,1 but important for understanding the need for services, establishing functional expectations of men as they age and for determining the influence of demographic and preventable risk factors. An understanding of community prevalence of ED is useful when evaluating treatments for prostate disease which carry high risks of ED. Community prevalence has been examined in international, but not Australian, studies.1-4 One Australian study of ED prevalence in a Perth general practice population has been reported in conference proceedings.5 We undertook this study to establish the prevalence of erectile dysfunction among Australian men, and to investigate the influence of risk factors. Methods Questions were included in the spring 1997 Omnibus survey, a multiple-user household interview survey which provides a representative sample of the South Australian population. We have reported the methods of and results from this annual survey previously.6-8 The Omnibus survey included questions about age, educational attainment, marital status, household income, area of residence, blood pressure group (low, normal, borderline, high), cholesterol group (low, normal, high), alcohol intake by frequency (days per week) and volume (standard drinks per day when drinking), doctor's report of high triglyceride levels, body mass index (BMI), whether vigorous exercise was undertaken in the past two weeks, current and previous smoking status, and number of cigarettes per day usually smoked. Also included were questions about the presence of lower urinary tract symptoms (LUTS) such as nocturia, frequency, urgency, and a visit to a doctor for LUTS, as reported previously.6 In addition, men aged 40 years and over were asked if they would be prepared to answer a further, mailed questionnaire on urological issues. This comprised an introductory letter explaining the purpose of the survey,7 questions on sexual function and one on history of prostate surgery. To provide an indication of men's expectations of intercourse frequency across age groups, the first question asked what frequency of intercourse the respondent considered normal for a man of his age. Respondents were then offered the opportunity to return the questionnaire without answering further questions if they considered these too intrusive. Sexual function questions were derived from the UCLA (University of California, Los Angeles) prostate cancer index developed by Litwin et al.9,10 These applied to the previous three months and covered seven domains: (i) sexual desire, (ii) orgasm, (iii) ability to have an erection, (iv) frequency of erections when wanted, (v) frequency of intercourse, (vi) frequency of morning or nocturnal erections, and (vii) firmness of erections. The first six of these had five response options ("nil", "poor", "fair", "good", "very good" for domains i-iii; "never had an erection when I wanted one", "less than half the time when I wanted one", "about half the time I wanted one", "whenever I wanted one" for domain iv; "once or more daily", "1-6 times weekly", "1-3 times monthly", "less than once a month", "not at all" for domain v; "never", "seldom [less than a quarter of the time]", "not often [less than half the time]", "often [more than half the time]", "very often [more than 75% of the time]" for domain vi), while the seventh had four ("no erections at all", "not firm enough for any sexual activity", "firm enough for masturbation and foreplay only", "firm enough for intercourse"). Additional questions involved history of prostate surgery (for cancer and for non-cancerous conditions). Ethical consideration for the study was by Repatriation General Hospital's Research Ethics Committee. Analysis We defined sexual dysfunction as a response of one of the two lowest response categories (eg, "nil" or "poor") in each of the first six domains. Confidence intervals were adjusted for a design effect of 1.1 to allow for clustering.11 Dysfunction in the seventh domain (firmness of erections) was termed "impotence", defined as "usual quality of erections during the past three months not firm enough for intercourse" (corresponds with National Institutes of Health definition)12 -- one of the three lowest response options for this domain. A total erectile function score was derived by adding scores for each of the seven domains. The resulting score out of a possible 34 was expressed as a percentage; a score of 0 indicated worst possible function and 100% indicated best possible function in all domains. The reliability coefficient (Cronbach's alpha) of this seven-item score was 0.940. The association of demographic and cardiovascular risk factors with total sexual function (expressed as a continuous variable) and with impotence (a dichotomous variable) was examined after allowing for the effects of age by means of analysis of variance and logistic regression. For analysis, data were weighted by household size, age, sex and geographic region to benchmarks derived from the resident South Australian population in August 1996. We used SPSS for Windows13 for statistical analysis. Results The numbers and age breakdown of respondents are shown in Box 1. The mean age of respondents was 57.5 years (standard error, 0.58). Of the 745 men aged 40 years and over who responded to the Omnibus survey (representing a response rate of 70.8%), 612 (82.1%) agreed to complete a further, mailed questionnaire on urological issues. These questionnaires were returned by 427 respondents (69.8%), 371 (86.9%) of whom agreed to answer the full questionnaire. Box 1 shows that the age distribution was the same in all of these groups. We weighted further analyses to reflect the age distribution in the South Australian community; marginal totals thus do not correspond to the totals in Box 1. A comparison of respondents who agreed to answer the sexual function questions with those who did not agree to a further survey, did not respond to the mailed form or did not agree to answer the sexual function questions showed no significant differences in age, marital status, blood pressure, cholesterol and triglyceride levels, blood pressure medication or visit to a doctor for LUTS. Men who answered the sexual function questions were more likely to report vigorous exercise in the past two weeks (25.0% v 15.4%; P = 0.001), and more likely to drink alcohol five or more times per week (34.2% v 25.3%; P = 0.026) than those who did not answer these questions. Frequency of intercourse The frequency of intercourse considered by respondents as normal for their age, shown in Box 2(a), and actual reported frequency of intercourse were strongly correlated (Spearman's correlation coefficient, 0.62; P < 0.0001). Most 40-60-year-olds considered that intercourse 1-6 times per week was normal for their age and most reported this frequency of intercourse. In older men, however, a discrepancy emerged: most 70-79-year-olds considered 1-3 times per month to be normal for their age, but most reported an actual frequency of less than once per month. Box 2(b) illustrates this more clearly. In younger age groups, the reported frequency approximated that considered normal for age, but in older age groups it was substantially lower. Similarly, Box 2(c) shows that, while potency exceeds desire in men aged 40-59 years, after age 60 years desire exceeds potency. Prevalence of erectile dysfunction Erectile dysfunction was strongly correlated with age across all seven domains (Box 3). The prevalence of impotence (defined above) increased sharply from 3% in 40-49-year-olds to 42% in 60-69-year-olds and 64% in 70-79-year-olds. Other domains of sexual function, including morning or nocturnal erections, followed this pattern of steep decline in the fifth and sixth decades. Total sexual function scores ranged from 17% to 97%, with a mean of 68.1% (standard error, 1.1); 21% of respondents had a score of 50% or less. The sexual function score was also strongly correlated with age (Pearson's correlation coefficient, 0.63; P < 0.0001). Demographic risk factors Age, income, education, marital status, area of residence, occupation and country of birth were examined for their association with total sexual function and impotence. By far the strongest contributor to total sexual function score was age, and so the contribution of other factors was examined singly after first accounting for the effects of age in analysis of variance models, including first-order interactions (sexual function) and logistic regression (impotence). Where factors or their first-order interactions were significant, a combined model was examined. A summary of these analyses is given in Box 4. Men with lower socioeconomic status (income, education, occupation) tended to have lower sexual function, and this association was seen particularly in older men. In the 70-79 years age group, the total sexual function score varied from 47.1 in the lowest education category to 61.7 in the highest. Sexual function tended to be lower in Australian-born than overseas-born respondents. When tested in a model containing cardiovascular risk factors, this association remained. No demographic factors, apart from age, were significant contributors to impotence. Cardiovascular and other risk factors The cardiovascular risk factors shown in Box 4, as well as presence of LUTS, doctor visit for LUTS and surgery for prostate disease (cancer or non-cancer), were examined for their effects on total sexual function and impotence, as described for demographic factors. A number of factors were significantly (P < 0.05) associated with decreased sexual function after controlling for the effects of age. For total sexual function, these were ever having smoked, taking blood pressure medication and having had surgery for non-cancer prostate disease. A history of vigorous exercise was protective. Risk factors for impotence were high body mass index, high levels of triglycerides and cholesterol, and surgery for non-cancer prostate disease. Again, a history of vigorous exercise was protective. When combined models were constructed with these factors, the number that remained independently significant was reduced (Box 4). For total sexual function, these were vigorous exercise, high triglyceride levels, blood pressure medication, and surgery for non-cancer prostate disease. The combined model for total sexual function explained 55% of the variance. The observed power was low for all non-significant factors. The effects of cardiovascular risk factors were stronger in older age groups. For example, men in their 60s with high triglyceride levels had a total sexual function score of 41% (95% CI, 31%-52%), compared with 64% (95% CI, 59%-70%) for men of the same age who did not have this risk factor. Discussion Few international or Australian studies of the prevalence of sexual dysfunction have surveyed the general community.1 Our study was based on a representative sample of the South Australian community, and employed rigorous methods which have been previously used in other prevalence studies.1,8 The study was undertaken before the recent introduction of the drug oral sildenafil for erectile dysfunction. The subsample who answered questions on sexual function were similar in terms of most demographic and risk factor variables tested, but did report undertaking more vigorous exercise, and more frequent alcohol consumption. They may therefore reflect a more physically active subgroup, and our estimates of erectile dysfunction are likely to be conservative. Relative to other studies with comparable methods and measures of sexual dysfunction, the prevalences we found were similar or higher. For example, a study of frequency of sexual activity reported that 34.7% of married 60-65-year-old people did not have sexual intercourse within the preceding month,14 while we found that 34.3% of 60-69-year-old men reported having intercourse "less than once a month". A survey of 1240 men attending general practices in Perth found a 45% prevalence of "complete ED [erectile dysfunction]" in 70-79-year-old men,5 compared with our finding of impotence in 64.2%, and 64.3% in a study by Diokno et al.15 The frequently quoted Massachusetts Male Aging Study reports a prevalence of complete impotence of 9.6% for a 40-70-year-old population;2 the corresponding figure in our study was 16%. As expected, frequency of intercourse considered normal for age was strongly correlated with the frequency of intercourse reported (Figure 2). Solstad and Hertoft found that, while 40% of interviewed Danish men reported some kind of sexual dysfunction, only 7% considered their problems abnormal for their age,16 implying that men tend to see their own functional level as normal. Nevertheless, in our study, the discrepancy between reported frequency of intercourse and that considered normal for age increased with age. It was greatest in men 70 years and older, suggesting that it is older men who may be most concerned about their sexual function. Consistent with this, potency exceeded desire in younger age groups, but the relationship was reversed in men aged 60 years and over. About one in five men over the age of 50 experience good to very good sexual desire, but nil to poor erectile function; these men may be more likely to seek treatment. If the goal of treatment is to achieve perceived "normal function for this age", this may be a frequency of intercourse of 1-3 times monthly for 70-79-year-olds and weekly in the case of 40-69-year-olds. The effects of demographic risk factors such as income and education were more apparent in older age groups and were consistent with lower sexual function in lower socioeconomic groups. No reason is immediately apparent for the lower sexual function of Australian-born men compared with their overseas-born counterparts. Cardiovascular and medical risk factors reflected a conventional pattern.1,17 Predictors of low total sexual function scores after controlling for age only were smoking, blood pressure medication, high triglyceride levels and non-cancer prostate surgery, while a history of vigorous exercise provided a protective effect. In a combined model, smoking was no longer significant. Predictors of impotence after controlling for age only were high body mass index, high triglyceride levels, high cholesterol level, surgery for non-cancer prostate disease, while vigorous exercise was again protective. In a combined model, only high cholesterol remained significant. However, it is likely that small numbers of respondents limited the power of our study to investigate all these factors simultaneously. Our findings suggest that men experience poor sexual function as a deficit in the same age ranges in which cardiovascular risk factors are major determinants of that function, raising the possibility of prevention. It would be interesting to know whether, for men 60 years and over, improving cardiovascular risk factor profile would also reduce the prevalence of sexual dysfunction. Also of interest is the consistent and independent effect of non-cancer prostate surgery on sexual function. While its effects are small relative to age, it may point to a higher impotence rate than is commonly believed to result from this type of surgery,18 and this warrants further investigation. Acknowledgements We thank Living Health and Quitline for generously sharing their data on cardiovascular risk factors, and the AntiCancer Foundation for funding the prostate cancer questions. We would also like to thank the Behavioural Epidemiology Unit, South Australian Department of Human Services, and Lynne Giles, Flinders University, for assistance with aspects of the statistical analyses. References Bortolotti A, Parazzini F, Colli E, Landoni M. The epidemiology of erectile dysfunction and its risk factors. Int J Androl 1997; 20: 323-334. Feldman HA, Goldstein I, Hatzichristou DG, et al. Impotence and its medical and psychosocial correlates: results of the Massachusetts Male Aging Study. J Urol 1994; 151: 54-61. Helgason AR, Adolfsson J, Dickman P, et al. Factors associated with waning sexual function among elderly men and prostate cancer patients. J Urol 1997; 158: 155-159. Jonler M, Moon T, Brannan W, et al. The effect of age, ethnicity and geographical location on impotence and quality of life. Br J Urol 1995; 75: 651-655. Chew K, Burio C, Stuckey B, Jamrozik K. Erectile dysfunction in general medical practice. A study in Perth, Australia. Int J Impotence Res 1997; 9 (Suppl 1): S36. Pinnock C, Marshall V. Troublesome urinary symptoms in the community: a prevalence study. Med J Aust 1997; 167: 72-75. Weller D, Pinnock C, Silagy C, et al. Prostate cancer testing in South Australian men: influence of sociodemographic factors, health beliefs and lower urinary tract symptoms. Aust N Z J Public Health 1998; 22: 400-402. Pinnock C, Weller D, Marshall V. Self-reported prevalence of prostate specific antigen (PSA) testing in South Australia: a community study. Med J Aust 1998; 169: 25-28. Litwin MS, Hays RD, Fink A, et al. Quality-of-life outcomes in men treated for localized prostate cancer. JAMA 1995; 273(2): 129-35. [See comments.] Litwin MS, Nied RJ, Dhanani N. Health-related quality of life in men with erectile dysfunction. J General Intern Med 1998; 13: 159-166. Kish L. Estimates of unit variance: design effect. Survey sampling. New York: John Wiley and Sons, 1965: 257-263. NIH Consensus Conference. Impotence. NIH Consensus Development Panel on Impotence. JAMA 1993; 270: 83-88. SPSS for Windows [computer program], version 6.1. Chicago, Ill: SPSS Inc, 1996. Marsiglio W, Donnelly D. Sexual relations in later life: a national study of married persons. J Gerontol 1991; 46(6): S338-S344. Diokno AC, Brown MB, Herzog AR. Sexual function in the elderly. Arch Intern Med 1990; 150: 197-200. Solstad K, Hertoft P. Frequency of sexual problems and sexual dysfunction in middle-aged Danish men. Arch Sexual Behavior 1993; 22: 51-58. Helgason AR, Adolfsson J, Dickman P, et al. Waning sexual function -- the most important disease-specific distress for patients with prostate cancer. Br J Cancer 1996; 73: 1417-1421. Clinical practice guidelines for the management of uncomplicated lower urinary tract symptoms in men. Canberra: National Health and Medical Research Council, 1997. (Received 30 Nov 1998, accepted 28 Jun 1999) Authors' details Repatriation General Hospital, Daw Park, Adelaide, SA. Carole B Pinnock, PhD, Principal Research Scientist, Division of Surgery; Alan M F Stapleton, PhD, FRACS, Director, Urology Unit; Villis R Marshall, MD, FRACS, Head, Division of Surgery, and Head, Department of Surgery, Flinders Medical Centre. Reprints will not be available from the authors. Correspondence: Dr C B Pinnock, Division of Surgery, Repatriation General Hospital, Daws Road, Daw Park, SA 5041. carole.pinnockAThealth.sa.gov.au Back to text Back to text Back to text Back to text

Carole B Pinnock · Villis R Marshall

Editorial Breaking bad news: explaining cancer diagnosis and prognosis Doctors and nurses need training in communicating information about cancer and responding to patients' concerns MJA 1999; 171: 288-289 For related articles see Prince, Lobb et al & Naganathan et al Most patients, if they have cancer, want to be told about it, and they want to know what the likely treatments are, the side effects of treatment and their prognosis.1 A clear understanding of prognosis can be particularly important in conditions such as breast cancer, because patients need prognostic information to make informed decisions about systemic treatment. So, how can this information best be communicated? Based on a literature review and recommendations of a consensus panel of doctors (with input from patients with cancer), Girgis and Sanson-Fisher2 published some useful guidelines on conveying information to patients about serious disease or death. Their guidelines included ensuring privacy and allowing adequate time, assessing patients' understanding, giving information about diagnosis and prognosis simply and honestly, avoiding euphemisms, encouraging patients to express feelings, being empathic, giving a broad but realistic time-frame concerning prognosis, and arranging a review. The crucial question is how well these recommendations are followed in clinical practice, as discussing prognosis should be part of the process of breaking bad news. Audiotape recordings of consultations have shown that doctors break the bad news of a cancer diagnosis to patients in a predictable and routine way regardless of patients' individual information needs.3 Although the doctors gave reassurance that something could be done, few attempted to elicit patients' thoughts and feelings about the symptoms and their cause. They gave the information in a consistent order -- diagnosis, the relevant evidence, the need for further investigations, the treatments being considered and the probable outcome -- with no heed to which issues patients wished to address first. Obvious verbal and non-verbal cues of distress were not acknowledged and patients' immediate concerns were not explored. This consultation structure led patients to believe they were not entitled to talk about their feelings or their major concerns. Consequently, their preoccupation with these feelings and concerns meant that they did not assimilate the information and advice given. On being interviewed at the end of the consultation, patients reported that they were left with important, but undisclosed, concerns and also felt that the information given had been inadequate for their needs. The important article by Lobb and colleagues4 published in this issue of the Journal looks at one aspect of breaking bad news to cancer patients -- explaining prognosis to women with breast cancer. It was clear from the women's answers to a questionnaire and a clinical vignette that many had problems understanding prognostic information in the form it is usually presented. The women also varied considerably in what prognostic information they would like to receive and how they preferred it to be presented. The study's findings showed that not all women will desire or understand standard methods of giving prognostic information. It has been suggested that giving patients the opportunity to talk with nurses after consultations in which they have been told they have cancer, or have been given complex information about cancer prognosis, would result in their disclosing concerns and misunderstandings and these could then be fed back to the treating clinicians.5 However, this solution ignores important evidence that nurses are just as reluctant as doctors to acknowledge patients' distress and elicit their underlying concerns.6 Like doctors, nurses in these situations have been found to adopt behaviours designed to prevent further disclosure.7 These "blocking behaviours" include telling patients that any distress is normal, switching the subject to neutral topics, giving information and advice before patients' concerns have been identified, focusing only on physical aspects of the condition, and using leading, closed and multiple questions.7 Doctors and nurses avoid exploring patients' feelings and concerns because they fear that it will provoke too much emotion, which could be harmful to patients.8 They feel that their training has not equipped them with the necessary skills to explore these issues and respond appropriately.9,10 Feeling that they are not being supported emotionally and practically by colleagues and supervisors has also been linked to a greater use of these "blocking behaviours".6,11 The lack of adequate training of doctors and nurses in communicating information to patients with cancer can have important negative psychological consequences for the patients. The development of clinical anxiety and depression is more likely when patients have unresolved concerns and perceive that they have been given inadequate information.12,13 So, how can these communication deficiencies be remedied? The data collected by Lobb et al4 could be used as a basis for ongoing research to identify how best to structure and describe prognostic information so that concepts such as "median survival" and "relative risk" are made understandable. It would then be possible to develop training programs in communication skills that teach doctors how to elicit patients' preferences for information about prognosis. The use of intradisciplinary10 and multidisciplinary workshops14 has been advocated. While objective evidence of the value of intradisciplinary workshops is awaited, multidisciplinary workshops have proved successful in helping doctors and nurses acquire key communication skills.14 Despite workshops being effective in changing key communication behaviours, it is not certain how much of what is learnt is applied to clinical practice. Two randomised trials are being conducted in the United Kingdom in an effort to determine this. The Cancer Research Campaign (CRC) Psychosocial Oncology Group is studying whether training doctors in small groups enables them to be more effective in communicating with patients and better able to cope with breaking bad news and dealing with patient concerns. The CRC Psychological Medicine Group is assessing whether senior doctors benefit from six sessions of individual feedback on their "bad news" consultations and whether this results in better patient recall and less patient distress in future consultations (as measured by a patient interview and the Hospital Anxiety and Depression Scale), as well as a reduction in the level of burnout in the doctors themselves. Without systematic training, the breaking of bad news and discussions of cancer prognosis are likely to fall short of existing guidelines and patients' needs and expectations. Consultations of this type can be difficult and painful. Yet, for too long, we have expected doctors and nurses to undertake these difficult tasks without the necessary training and support. G Peter Maguire Consultant Psychiatrist; and Director Cancer Research Campaign Psychological Medicine Group, Manchester, UK Meredith C, Symonds P, Webster L, et al. Informational needs of cancer patients in West Scotland: cross sectional survey of patients' views. BMJ 1996; 313: 724-726. Girgis A, Sanson-Fisher RW. Breaking bad news: consensus guidelines for medical practitioners. J Clin Oncol 1995; 13: 2449-2456. Maguire P. Breaking bad news. Cambridge: Cambridge Handbook of Psychology, Health and Medicine, 1998: 273-275. Lobb EA, Butow PN, Kenny DT, Tattersall MHN. Communicating prognosis in early breast cancer: do women understand the language used? Med J Aust 1999; 171: 290-294. Watson M, Denton S, Baum M, Greer S. Counselling breast cancer patients: a specialist nurse service. Counselling Psychol Q 1988; 1(i): 23-31. Wilkinson SM. Factors which influence how nurses communicate with cancer patients. J Adv Nurs 1991; 16: 677-688. Maguire P. Barriers to psychological care of the dying. BMJ 1985; 291: 1711-1713. Maguire P, Faulkner A, Booth K, et al. Helping cancer patients disclose their concerns. Eur J Cancer 1996; 32A: 78-81. Maguire P, Faulkner A. How to improve the counselling skills of doctors and nurses in cancer care. BMJ 1988; 297: 847-849. Fallowfield L, Lipkin M, Hall A. Teaching senior oncologists communication skills: Results from phase 1 of a comprehensive longitudinal programme in the United Kingdom. J Clin Oncol 1998; 16: 1961-1968. Booth K, Maguire P, Butterworth T, Hillier VT. Perceived professional support and the use of blocking behaviours by hospice nurses. J Adv Nurs 1996; 24: 622-527. Parle M, Jones B, Maguire P. Maladaptive coping and affective disorders in cancer patients. Psychol Med 1996; 26: 735-744. Fallowfield LJ, Hall A, Maguire GP, Baum M. Psychological outcomes of different treatment policies in women with early breast cancer outside a clinical trial. BMJ 1990; 301: 575-580. Maguire P, Booth K, Elliott C, Jones B. Helping health professionals involved in cancer care acquire key skills -- the impact of workshops. Eur J Cancer 1996; 32A: 1486-1489.

Research Communicating prognosis in early breast cancer: do women understand the language used? Elizabeth A Lobb, Phyllis N Butow, Dianna T Kenny and Martin H N Tattersall MJA 1999; 171: 290-294 For related articles see Maguire, Prince & Naganathan et al Abstract - Introduction - Methods - Results - Discussion - Acknowledgements - References - Authors' details - - More articles on Oncology Abstract Objectives: To determine the degree to which women with early breast cancer understand the prognostic information communicated by clinicians after breast cancer diagnosis, and their preferences for how this information is presented. Design: Cross-sectional survey conducted within two months of breast cancer diagnosis, using a self-administered written questionnaire. Participants and setting: One hundred women attending five Sydney teaching hospitals and one country hospital, who were diagnosed with early stage breast cancer between January and December 1997. Results: The 100 respondents represented 70% of the 143 women originally approached to participate. Many respondents did not fully understand the language typically used by surgeons and cancer specialists to describe prognosis: 53% could not calculate risk reduction (with adjuvant therapy) relative to absolute risk; 73% did not understand the term "median" survival; and 33% believed a cancer specialist could predict an individual patient's outcome. Women in professional/ paraprofessional occupations understood more prognostic information than non-professional women. There was no agreement on the descriptive equivalent of a "30%" risk, nor the numerical interpretation of a "good" chance of survival. Forty-three per cent of women preferred positively framed messages (eg, "chance of cure"), and 33% negatively framed messages (eg, "chance of relapse"). The information women most wanted was that relating to probability of cure, staging of their cancer, chances of treatment being successful, and 10-year survival figures with and without adjuvant therapy. Conclusions: Our results suggest that misunderstanding is responsible for women's confusion about breast cancer prognosis. Clinicians should use a variety of techniques to communicate prognosis and risk, and need to verify that the information has been understood. Introduction To make informed decisions women with breast cancer must understand what their prognosis is without systemic treatment, and the likely advantages and disadvantages of treatment. While most Australian doctors now tell cancer patients their diagnosis,1 prognosis is less commonly discussed.2 Reticence to provide prognostic information is often based on concerns that the information will be overwhelming, not understood or will destroy hope.3,4 However, it is not clear whether the information itself, or the language used, is the critical feature. Many patients have a poor understanding of their disease and their prognosis,5,6 or have difficulty recalling the information they have been given about their disease.7,8 Similarly, women at risk of developing breast cancer commonly misreport individual and population risk.9 Denial and minimisation of risk are also common psychological reactions to cancer risk notification after screening procedures.10 If it were possible to determine whether patients cannot understand the terminology or mathematics of risk information, or, rather, prefer not to be told or do not absorb the information, clearer directions for best clinical practice in discussing prognosis could be established. Most previous studies on risk communication in cancer have not dealt specifically with issues pertinent to women with breast cancer. In two studies that did, the women surveyed were well down the treatment path and their experience of learning their prognosis was long past.5,11 To our knowledge, there are no reports of women's understanding of specific prognostic information in early breast cancer. We investigated women's understanding of prognostic information and their preferences for the way the information on the risk of their breast cancer recurring after surgery is presented to them. Methods Survey subjects Women were recruited through their treating physician. To ensure input from a range of women, five urban centres attracting referrals from populations with different socioeconomic profiles (Royal Prince Alfred, Royal North Shore, Prince of Wales, St George, and Westmead hospitals, all in Sydney, New South Wales) and one rural centre (Tamworth Hospital, Tamworth, NSW) were approached to participate in the study. Thirteen breast surgeons and 13 medical oncologists from these centres were invited to participate in the study and all agreed. One hundred and forty-three consecutive women newly diagnosed with stage I or II breast cancer at any of the six treatment centres between January and December 1997 were contacted by letter to request their participation. The women received the letter within 2-4 weeks of making their own decisions about adjuvant treatment and within 2 months of their initial diagnosis. (The timing of questionnaire administration was carefully considered to maximise the saliency of the issues while avoiding distressing women making their own treatment decisions.) The letter was followed up by a phone call from the research coordinator, who obtained verbal consent for participation and then sent out the questionnaire by mail. One centre opted to send women a letter signed by their oncologist inviting them to participate in the study. The survey sample included patients of surgeons and medical oncologists in both private and public practice. Women from a non-English-speaking background with insufficient knowledge of English to complete the questionnaire, and women presenting with a second cancer, were excluded. Questionnaire We gathered the data using a self-administered written 17-item questionnaire, designed on the basis of a review of the literature; an analysis of 20 audiotapes of initial oncology consultations with breast cancer patients (collected from two centres -- Royal Prince Alfred Hospital and Westmead Hospital -- during another study undertaken between 1995 and 199712); and expert consultation (a working party set up by the National Breast Cancer Centre). The 20 audiotapes were transcribed and the contents analysed to identify the range of ways in which prognosis was conveyed to patients (eg, absolute and relative risk, cumulative risk, numerical or non-numerical probability, and individual versus population risk). The questionnaire investigated women's understanding of and preferences for these different formats used by doctors for disclosing prognosis and risk information. In addition, a standard hypothetical scenario of adjuvant therapy in early stage breast cancer was included, and women responded to questions applying to that scenario (Box 1). Six of the questions explored women's understanding of different ways in which the risk of breast cancer recurring after surgery could be presented. Two sample questions are shown in Box 2. A "don't know" option was not offered in the items relating to "understanding" in order to force a choice and allow an analysis of common errors in interpretation. The remaining questions focused on the importance of different prognostic information to women's decision making, and on their preferences for presentation of risk - for example: percentages versus numbers (eg, "70%" v. "7 in 10"); numerical versus verbal descriptions of risk (eg, "30%" v. "small"); and positively framed versus negatively framed statements (eg, "70% chance of remaining free of cancer" v. "30% chance of the cancer coming back"). The questionnaire also elicited the women's demographic data and details of their breast cancer diagnosis and treatment (Box 3). Statistical analysis Appropriate sample sizes were calculated using the SAM sample size software package.13 Sample size calculations were based on effect sizes from related studies in patients' level of recall after a variety of interventions. In an Australian study measuring understanding of information presented in an oncology consultation, a sample size of 47 per group was sufficient to detect statistically significant differences of 7% (P < 0.005) in recall between groups. Thus, in a comparison of two patient subgroups (eg, young v. old), a total sample size of 100 would allow us to detect a similar difference in responses in the two groups. A sample size of 100 would also allow detection of a difference of 30% or more (felt to be clinically significant) between subgroups in the proportion of women preferring one presentation of risk versus another, with a power of 0.8 and a significance level of 0.05. A "total understanding" score was calculated by summing correct responses to the six items assessing "understanding". The summary score was normally distributed (K-S Lilliefors .0514). Descriptive statistics were used to identify the percentage of patients understanding and preferring different risk information. Analysis of variance (ANOVA), Student's t tests and χ2 tests of association were used to examine the relationship between demographic variables and outcomes, as appropriate; two-sided tests were used.15 Ethical approval Approval was granted for this study by the Ethics Committee of the University of Sydney, the Central Sydney Area Health Service, the Southern Sydney Area Health Service, and individual hospital ethics committees at Westmead, Royal North Shore and Tamworth hospitals. Results Of the 118 women who agreed to participate in the survey, 100 returned questionnaires (70% of the original 143 women contacted). Demographic data Demographic characteristics of the participants are presented in Box 3. Their mean age was 56 years and most were city dwellers. Just over half had completed the Higher School Certificate, university or some form of tertiary training. The percentage of women with tertiary qualifications was 42% (compared with 37% in the general Australian population16).. Nearly two-thirds worked (or had worked) in professional or paraprofessional occupations, and 22% were working in occupations related to medicine (eg, doctor, nurse, medical receptionist, technician). Questionnaire responses A summary of the women's responses to the questionnaire is given in Box 4. Discussion We have identified some of the problems women with breast cancer experience when trying to interpret prognostic information presented by their doctors. Our results support the hypothesis that it is misunderstanding, not denial, that causes confusion. A considerable number of women in our study did not clearly understand some of the language used to describe the risk of breast cancer recurrence after surgery or how additional treatment might benefit them. Moreover, the response from this group of relatively highly educated women probably represents a "best case" scenario, and, if anything, one might expect understanding to be poorer in the general population of women with breast cancer. These findings have implications for informed consent. Clinicians need to explain what type of prognostic information can be given, and enquire how much of this information women want to hear. They should check very carefully how women have interpreted the information presented to them, and must not assume that, because a woman has already consulted a number of specialists, her prognosis has been conveyed to her and clearly understood. This applies to all patients with breast cancer, but especially those who work in unskilled occupations. It might be argued that our sample was not truly representative, as (i) the women surveyed, having recently been told their diagnosis, may not have been in the best frame of mind to answer the questionnaire clearly and impartially; and (ii) we did not include a similar group of women who had never had breast cancer. Furthermore, patient responses may have been different had the questions concerned personal experience rather than a hypothetical case scenario.5,9,10,19 However, data from Degner et al20 suggest that views expressed by people diagnosed with cancer differ considerably from those of the well population, which underscores the importance of surveying those who have actually been diagnosed with cancer. In addition, we felt that a typical case vignette was the most appropriate tool to control for the influence of individual disease variables and treatment protocols; to reduce the positive bias associated with evaluating one's own treatment team; and to examine all aspects of risk communication in adjuvant therapy. Creative measures to assist women in understanding risk statistics are needed. Bunker et al have recently proposed that a life table constructed from published statistics on national morbidity and mortality may be used to display the likelihood of developing or dying of a disease at any given moment.21 A similar approach could be used to display the likelihood of disease recurrence and premature death after cancer diagnosis. We believe that these and other information aids may contribute to informed patients' involvement in treatment decisions, and a more realistic understanding of prognosis. Acknowledgements We thank Dr Afaf Girgis, Dr Lyn Mann, Ms Kate White, Ms Joan Wilson and Ms Kim Hobbs for their assistance and advice; also the 26 clinicians who participated in this project, and the women who so willingly filled out the questionnaire. The research was funded by the National Health and Medical Research Council National Breast Cancer Centre of Australia. References Charlton RC. Breaking bad news. Med J Aust 1992; 157: 615-621. Butow PN, Kazemi J, Beeney LJ, et al. When the diagnosis is cancer: patient communication experiences and preferences. Cancer 1996; 77: 2630-2637. Oken D. What to tell cancer patients: a study of medical attitudes. JAMA 1961; 175: 1120-1128. Beisecker AE, Helmig I, Graham D, et al. Attitudes of oncologists, oncology nurses and patients from a women's clinic regarding medical decision making with older and younger breast cancer patients. Gerontologist 1994; 34: 505-512. Siminoff LA, Fetting JH, Abeloff MD. Doctor-patient communication about breast cancer adjuvant therapy. J Clin Oncol 1989; 7: 1192-1200. Sheldon JM, Fetting JH, Siminoff LA. Offering the option of randomized clinical trials to cancer patients who overestimate their prognoses with standard therapies. Cancer Invest 1993, 11: 57-62. Dunn SM, Butow PN, Tattersall MHN, et al. General information tapes inhibit recall of the cancer consultation. J Clin Oncol 1993; 11: 2279-2285. Mackillop WJ, Stewart WE, Ginsburg AD, Stewart SS. Cancer patients' perceptions of their disease and its treatment. Br J Cancer 1988; 58: 355-358. Evans DR, Blair V, Greenhalgh R, et al. The impact of genetic counselling on risk perceptions in women with a family history of breast cancer. Br J Cancer 1994; 70: 934-938. Lerman C, Rimer BK, Engstrom PF. Cancer risk notification: psychosocial and ethical implications. J Clin Oncol 1991; 9: 1275-1282. Hughes KK. Decision making by patients with breast cancer: the role of information in treatment decision selection. Oncol Nurs Forum 1993; 20: 623-628. Brown R, Dunn S, Butow P. Meeting patient expectations in the cancer consultation. Ann Oncol 1997; 8: 877-882. Glasziou P. SAM 2.1: a sample size calculator [computer program]. Sydney: NHMRC Clinical Trials Centre, University of Sydney, 1992. Armitage P, Berry G. Statistical methods in medical research. Oxford: Blackwell Scientific, 1994: 397. SPSS Advanced Statistics, TM6.1. Chicago; SPSS Inc, 1994. Australian women's year book. Canberra: Australian Bureau of Statistics, 1997. Degner LF, Kristjanson LJ, Bowman D, et al. Information needs and decisional preferences in women with breast cancer. JAMA 1997; 277: 1485-1492. Bilodeau BA, Degner LF. Informational needs, sources of information, and decisional roles in women with breast cancer. Oncol Nurs Forum 1996; 23: 691-696. Marteau TM. Framing of information: its influence upon decisions of doctors and patients. Br J Soc Psychol 1989; 28: 89-94. Degner LF, Sloan JA. Decision making during serious illness: what role do patients really want to play? J Clin Epidemiol 1992; 45: 941-950. Bunker JP, Houghton J, Baum M. Putting the risk of breast cancer in perspective. BMJ 1998; 317: 1307-1309. (Received 4 Jan, accepted 11 May, 1999) Authors' details University of Sydney, Sydney, NSW. Elizabeth A Lobb, BAdEd, MAppSci, Associate Lecturer, Medical Psychology Unit. Phyllis N Butow, PhD, MPH, Executive Director, Medical Psychology Unit; and Research Co-ordinator, Department of Psychological Medicine, Royal North Shore Hospital. Dianna T Kenny, PhD, MA, Associate Professor of Psychology, Faculty of Health Sciences. Martin H N Tattersall, MD, FRACP, Professor of Cancer Medicine, Department of Medicine. Reprints: Ms E A Lobb, Associate Lecturer, Medical Psychology Unit, Department of Psychological Medicine, University of Sydney, NSW 2006. Email: lizlobbATblackburn.med.usyd.edu.au 1: Hypothetical breast cancer scenario used in the questionnaire Sheila is a 54-year-old woman with breast cancer. Sheila has gone through the menopause. Sheila chose to have her breast cancer (tumour) removed by a lumpectomy, but she also had some of the lymph glands in her armpit removed. Sheila was advised to have radiotherapy after her lumpectomy. Sheila's breast cancer was small, and the lymph nodes under her arm were not affected with cancer. Her tumour contained receptors to oestrogen, suggesting that it may be sensitive to the effects of hormones. Sheila's doctor uses this information to decide if she would benefit from additional treatment. Sheila understands that any additional treatment other than surgery is called "adjuvant" therapy. Adjuvant therapy means giving treatment now after her surgery to try to prevent the cancer returning in the future. Following her breast cancer surgery, Sheila was told of her risk of having her cancer return (her prognosis) if she has no further treatment. This risk can be expressed in different ways. Back to text 2: Examples of questions to assess women's understanding of relative risk reduction Example A: Sheila's doctor told her that 30% of women with a cancer similar to hers will have their cancer come back within 5 years. If Sheila has additional treatment, the risk of her cancer coming back will be reduced by 30%. Tick one only If Sheila has additional treatment this means the risk of her cancer coming back within 5 years is zero. If Sheila has additional treatment this means the risk of her cancer coming back within 5 years is 21%. If Sheila has additional treatment this means the risk of her cancer coming back within 5 years is 30%. Example B: If Sheila's doctor says the median time for her breast cancer to return without further treatment is about 5 years, he/she means: The average time for Sheila's cancer to return is 5 years That 50% of women with breast cancer like Sheila's will have their cancer return within 5 years That the women whose cancer will come back will have it come back within 5 years I don't understand the word "median" Back to text 3: Demographic characteristics of the respondents (n = 100) to a questionnaire about provision of information on breast cancer prognosis*CategoryNumber of participantsAge (mean, 56 years; range, 35-88 years)Postcode City82 Country18Educational level Non-tertiary58 Tertiary42Occupation Professional/paraprofessional63 Non-professional36Marital status Married58 Other41English as first language84Working in medicine-related occupation22Time since diagnosis 1-2 months71 ≥3 months26Treatment for breast cancer Lumpectomy only14 Mastectomy only17 Lumpectomy + R38 Lumpectomy +R + C15 Mastectomy + C15Family member/friend with breast cancer Yes61 No38* Not all categories sum to 100 because of missing data. R = radiotherapy; C = chemotherapy. Back to text 4: Summary of questionnaire responses Understanding of risk Risk concepts tested Absolute risk of relapse: -- 86% of respondents gave a correct response. 30% relative risk reduction, with therapy, of an absolute risk of 30% (sample question A, Box 2): -- 47% gave a correct response; -- 28% thought additional treatment would reduce the risk of relapse to zero; and -- 25% thought the risk would remain at 30%. Median 5-year survival (sample question B, Box 2): -- 27% answered correctly; -- 43% thought it meant "average" survival; -- 10% thought that half the women not having adjuvant therapy would have their breast cancer return within 5 years; and -- 20% did not understand the term "median". Interpretation of a graphical representation of risk: -- 80% of respondents answered correctly. Distinguishing individual risk from population risk: -- 66% gave correct response. -- The remaining women believed that their cancer specialist knew whether or not they would respond to treatment. Association with demographic variables Mean number of correct responses, 3.4 (95% CI, 3.08-3.6); only one woman answered all six questions correctly. Women in professional employment (mean number of correct responses, 3.6 [95% CI, 3.2-4.0]) or paraprofessional employment (mean number of correct responses, 3.4 [95% CI, 3.0-3.8]) understood more prognostic information than women in non-professional employment (mean number of correct responses, 2.5 [95% CI, 1.7-3.4] [F2,87 = 4.24, P = 0.02]). No other variables were found to be associated with understanding of risk (eg, working in a medically related field; having tertiary qualifications; having had surgery, radiotherapy and/or chemotherapy for breast cancer; or time elapsed since consultation in which prognosis was discussed). Interpretation of words versus statistics There was no consistency in respondents' interpretation of "a good chance of remaining free of cancer" in statistical terms, nor agreement on the non-numerical interpretation of "a 30% risk" (17% thought it was a very high or high risk, 34% that it was a medium risk, and 49% that it was a low risk). Preferences for language 44% of respondents preferred "a 70% chance of cure", 13% preferred "a 7 in 10 chance", and the remainder had no preference. 53% of women preferred "a 30% chance of cancer coming back", 38% preferred "a small chance of cancer coming back", and the remainder had no preference. 49% of non-tertiary-educated women versus 24% of tertiary-educated women preferred the descriptive option (a "small" chance) (χ22 = 8.17, P = 0.02). 43% of women preferred the wording "70% chance of cure", 33% preferred "30% chance of the cancer coming back", and 25% had no preference. Preferences for framing of information Reasons given for choosing positively framed prognostic information: "a more positive/optimistic statement" and "encourages determination to manage treatment positively". Reasons given for choosing negatively framed prognostic information: "it emphasises the importance of additional treatment" and "more specific/precise". Importance attributed to prognostic information (Table) Over 90% of respondents regarded information about their chances of being cured, the staging of their cancer, and the chances that the recommended treatment would work as very important to their decision making. Nearly two-thirds of women regarded the 10-year survival rate with adjuvant therapy as very important information, and 45% wanted to know this rate without adjuvant therapy. These percentages are considerably higher than documented in previous studies.17,18 Women's ratings of the importance of different types of prognostic informationPrognostic informationVery importantSomewhat importantNot importantMy chances of being cured94%2%4%What things about my cancer influence my chances of being cured (eg, size of my cancer, whether lymph nodes are involved, etc)92%6%2%The chances that the recommended treatment will work91%7%2%How many women in my situation choosing to have the recommended treatment would be alive in 10 years60%29%11%Statistics about long term outcome of breast cancer50%32%18%How many women not choosing to have the recommended treatment are alive in 10 years45%35%20%The longest anyone in my situation has lived34%19%47%The shortest anyone in my situation has lived30%14%56%The risk of my cancer shortening my life compared with other life events (eg, heart disease, old age)45%23%32%The average time people in my situation have lived44%28%28%Back to text

Elizabeth A Lobb · Phyllis N Butow · Dianna T Kenny

Editorial Diagnosing osteoporosis: the value of quantitative ultrasound Currently, screening by quantitative ultrasound does not appear to be a good deal MJA 1999; 171: 295-296 For related articles see Maguire, Lobb et al & Naganathan et al Towards the end of the 20th century, the problems of diagnosis have not really changed. These remain how to evaluate the risk of disease, and then how to explain risk reduction clearly to patients. What has changed is our ability to predict risk, and to reduce that risk by the powerful public health, pharmacological or surgical interventions now available. What has this to do with osteoporosis, a microarchitectural disorder leading to fragility fracture? Osteoporosis is a classic chronic disorder in which the actual individual risk of clinical disease (in this case future fracture) is often difficult to quantify. The situation is similar for hypertension and hypercholesterolaemia. A few statistics may aid risk evaluation. A risk of 5%-10% or greater over five years of any osteoporotic fracture (arms, legs, pelvis, spine or rib) is generally regarded as requiring intervention. Interventions can reduce fracture risk by about 15%-50%.1,2 To prevent one fracture in such a population, between 20 and 133 patients need to be treated for five years. The risk of osteoporotic fracture depends on age. In women over the age of 65 years, the five-year risk rises dramatically from 5%, and to 20% in women over the age of 90 years.3 The presence of a previous osteoporotic fracture at least doubles the risk of future fracture.4-6 What does bone densitometry have to do with risk evaluation? It is this: when bone density is measured by dual energy x-ray absorptiometry (DEXA), then, for each standard deviation that the result falls below the mean for the individual's age (the Z score), the future risk of fracture doubles. Thus, an individual with a Z score of 22 has a fourfold greater risk of fracture than the average person of the same age. In a 65-year-old, this would give an actual five-year risk of fracture of about 30%. Because the bone density of fracture populations is independent of age, the concept of a standard deviation unit with respect to a fixed, low-risk population -- healthy 20-30-year-olds -- has been introduced (the T score) (see Figure). Significant microarchitectural deterioration (osteoporosis) is defined in bone density terms as a T score of -2.5 or less. Patients with a T score in this range have at least 5.65 times the risk of fracture relative to normal young individuals. However, their absolute five-year risk of fracture is related to the actual population risk at their age; for women aged 65-70 years this is about 18%.2 Factors other than age and bone density contribute to calculation of absolute fracture risk. Particularly important is a previous history of osteoporotic fracture. Currently, bone density risk evaluation by DEXA or quantitative computed tomography is supported by rebates from the Health Insurance Commission (HIC) for patients at high risk based on clinical information. Rebates are for evaluation of individuals who have had an osteoporotic fracture; who have clinical risk factors, such as corticosteroid treatment or premature menopause; or who have had osteoporosis diagnosed on a previous bone density test. The HIC will not fund the first bone density test in unselected individuals, otherwise known as population screening. This is because it is currently considered that bone density testing and the interventions consequent on finding high-risk individuals do not fulfil Australian cost-effectiveness criteria. However, individuals who do not meet current HIC criteria for bone density testing often decide to pay for the test themselves. How should we advise the HIC or the individual patient about the most effective screening for osteoporosis? Based on the epidemiology of fracture in Australia, it could be argued that, as well as the categories of high risk patients already outlined, all women aged 65-75 years should have bone density testing. This is because the population risk of future fracture rises dramatically in this age group. If screening is performed at an earlier age, the benefits are diluted by the small number of patients with detectable osteoporosis and the lack of controlled trial evidence that treatment prevents fracture at these ages when event rates are low. Could ultrasound become a "front end" to DEXA testing? In this issue of the Journal, the article by Naganathan et al7 is a useful contribution to the debate, providing a framework for considering the value of ultrasound screening. Naganathan et al report a simple method for calculating the benefits of testing bone structure by ultrasound compared with the current "gold standard" of DEXA. They achieved this by comparing the pre- and post-test probabilities of DEXA-defined osteoporosis (T score ≤ -2.5 at spine or hip sites) after ultrasound testing. Interestingly, they showed that 37% of their selected population had a normal combined quantitative ultrasound score, and that this finding completely excluded DEXA-defined osteoporosis. However, based on these data, an ultrasound test does not seem to be a good deal either for the individual patient or for the HIC, should it fund screening. The cost of screening 100 patients with ultrasound ($40 each) plus DEXA for those with abnormal ultrasound results ($80 x 63) would be $9040; the cost of screening with DEXA alone would be only $8000. Thus, both patients and the HIC should be advised not to pay for commercial ultrasound testing as a "front end" to DEXA at present. What about the future -- could ultrasound replace DEXA as the "gold standard" for predicting fracture? The answer is yes, possibly. The evidence-based approach would demand large prospective studies showing that ultrasound is better and cheaper than DEXA in predicting fracture, and that patients treated on the basis of ultrasound testing have a reduced risk of fracture compared with those who are not treated. To date, a couple of studies have taken the first steps to show effective fracture prediction in elderly women.8,9 It has taken 20 years to validate DEXA as a clinically useful predictor of patients who should be treated to prevent fracture, so don't hold your breath over ultrasound! Richard L Prince Associate Professor, University Department of Medicine Sir Charles Gairdner Hospital, Perth, WA Reprints: Associate Professor R L Prince, University Department of Medicine, Sir Charles Gairdner Hospital, Nedlands, WA 6009. Eddy DM, Johnston CC, Cummings SR, et al. Osteoporosis: review of the evidence for prevention, diagnosis and treatment and cost-effectiveness analysis. Osteoporos Int 1998; 8 Suppl 4: S7-S80. Cummings SR. Effect of alendronate on risk of fracture in women with low bone density but without vertebral fracture: Results from the Fracture Intervention Trial. JAMA 1998; 280: 2077-2082. Sanders KM, Seeman E, Ugoni AM, et al. The age- and gender-specific rate of fractures in Australia: a population based study. Osteoporos Int 1999. In press. Ross PD, Genant HK, Davis JW, et al. Predicting vertebral fracture incidence from prevalent fractures and bone density among non-black, osteoporotic women. Osteoporos Int 1993; 3: 120-126. Wasnich RD, Davis JW, Ross PD. Spine fracture risk is predicted by non-spine fractures. Osteoporos Int 1994; 4: 1-5. Cummings SR, Nevitt MC, Browner WS, et al. Risk factors for hip fracture in white women. N Engl J Med 1995; 332: 767-773. Naganathan V, March L, Hunter D, et al. Quantitative heel ultrasound as a predictor of osteoporosis. Med J Aust 1999; 171: 297-300. Bauer DC, Gluer CC, Cauley JA, et al. Broadband ultrasound attenuation predicts fractures strongly and independently of densitometry in older women. Arch Intern Med 1997; 157: 629-634. Porter RW, Miller C, Grainger D, Palmer SB. Prediction of hip fracture in elderly women: a prospective study. BMJ 1990; 301: 638-641. Back to text

Richard L Prince

Research Quantitative heel ultrasound as a predictor for osteoporosis Vasi Naganathan, Lyn March, David Hunter, Nick A Pocock, Joanna Markovey and Philip N Sambrook MJA 1999; 171: 297-300 For related articles, see Prince, Maguire & Lobb et al Abstract - Introduction - Methods - Results - Discussion - References - Authors' details - - More articles on Rheumatology Abstract Objective: To determine the diagnostic value of quantitative ultrasound (QUS) to predict bone mineral density (BMD) categories as defined by dual-energy x-ray absorptiometry. Design: Cross-sectional survey. Setting: Rheumatology department of a tertiary care hospital (Royal North Shore Hospital, Sydney, NSW), 1997-1998. Subjects: 326 healthy women aged 45-80 years who had volunteered for a twin study. Our study included both members of non-identical twin pairs but only one randomly selected member of identical twin pairs. Main outcome measures: BMD categories as defined by dual-energy x-ray absorptiometry of lumbar spine and left hip, and QUS of calcaneus; sensitivity, specificity and likelihood ratios (LRs) of QUS parameters to diagnose osteoporosis as defined by BMD. Results: The sensitivity of QUS to diagnose BMD osteoporosis varied between 9% and 47%, depending on the QUS parameter. The specificity of QUS was high (88%-100%). If all QUS parameters were normal, osteoporosis was unlikely (LR, 0-0.2). One QUS parameter, broadband ultrasound attenuation (BUA), was highly predictive of osteoporosis by BMD when in the osteoporotic range (LR, Infinity), but had low sensitivity (9%). QUS results in the osteoporotic range for other parameters and all QUS results in the osteopenic range were less predictive (LR, 1.0-5.2) of osteoporotic BMD. Conclusion: These results suggest that, for most of those tested for osteoporosis by QUS in the community, uncertainty remains about expected BMD. Introduction Quantitative heel ultrasound has recently been introduced in Australian pharmacies as a "screening" tool for osteoporosis. The technology is relatively cheap, radiation-free and portable, but its accuracy in diagnosing osteoporosis is unclear. Bone mineral density (BMD), measured by dual energy x-ray absorptiometry (DEXA), is the best predictor of fracture risk and is currently considered the "gold standard" for diagnosing osteoporosis. Although prospective studies have shown that quantitative ultrasound (QUS) predicts future fracture risk independently of BMD,1,2 most women with abnormal results will proceed to formal BMD measurement to assess the need for therapeutic intervention; women with normal QUS results may be reassured they do not need BMD measurement. It is unclear how many women have unnecessary further investigations or are falsely reassured. When used in this way, QUS has diagnostic value only if it can accurately predict BMD categories as determined by DEXA. Our aim was to examine the role of QUS in predicting BMD diagnostic categories. We determined conventional sensitivity and specificity, as well as likelihood ratios (LRs). These have the advantage of allowing test results to be assessed for several diagnostic categories rather than only at a single cut-off between "normal" and "abnormal".3 Methods Subjects and setting The study was a cross-sectional survey of healthy women aged 45-80 years who had volunteered to take part in a twin study. They were recruited from the Australian Twin Registry and media advertising. Our ultrasound study included both members of each non-identical twin pair and one randomly selected member of each identical twin pair. The study was conducted in the Rheumatology Department of the Royal North Shore Hospital, Sydney, NSW (a tertiary care hospital), in 1997 and 1998. It was approved by the hospital's Human Research Ethics Committee. Assessment Subjects had BMD measurements of their lumbar spine (L1-L4) and left hip (neck of femur and total hip) by DEXA using a Hologic QDR450 instrument (Hologic Inc, Waltham, Mass, USA). The same machine was used on all patients. QUS of the left calcaneus was performed on the same day using a CUBA Mark II ultrasound instrument (McCue Ultrasonics, London, UK). The two most commonly used QUS parameters were measured: broadband ultrasound attenuation (BUA), which is thought to reflect bone mass and architecture, and velocity of sound (VOS), which reflects mass and elasticity of bone.4 Analyses Each BMD and QUS value was converted to a T score (number of standard deviations from the population mean for young, healthy, sex-matched adults). This population mean was estimated from measurements in 50 women aged 20-30 years who also took part in the twin study. T scores were used to categorise BMD values as normal (T > -1) or indicating osteopenia (T, -2.5 to -1) or osteoporosis (T < -2.5), as proposed by a working party of the World Health Organization.5 QUS values were classified in the same way. Although no consensus has been reached on what T-score cut-offs and diagnostic categories to use with QUS, the instrument used commonly in Australian pharmacies uses the WHO criteria and cut-off values. Subjects were classified as having osteoporosis if at least one of the three BMD measurements (lumbar spine, neck of left femur or total left hip) indicated osteoporosis, and as having osteopenia if at least one measurement indicated osteopenia but none indicated osteoporosis. BUA and VOS results were combined as a cQUS category: this was defined as normal if both results were normal, as osteopenic if either indicated osteopenia but neither indicated osteoporosis, and as osteoporotic if either indicated osteoporosis. Some QUS scanners calculate a stiffness parameter (unrelated to mechanical stiffness) from a linear combination of normalised BUA and VOS values. We calculated stiffness in an analogous manner,6 and categorised it as normal, osteopenic or osteoporotic based on T scores in the same way as other QUS values. We calculated the sensitivity and specificity of QUS parameters in predicting BMD-defined osteoporosis and osteopenia. We also calculated the likelihood ratio (LR) for each QUS result (sensitivity/1 - specificity), defined as the ratio of the probability of the particular QUS result (normal, osteopenic or osteoporotic) in women with BMD-defined osteoporosis or osteopenia to the probability of the same result in women with normal BMD.7 As there is no consensus on what QUS cut-off values should be used to diagnose osteoporosis, the statistical analyses were repeated using a range of QUS T-score cut-off values for osteoporosis between -2.5 and -1.0. Results Subjects and osteoporosis There were 326 subjects, with mean age 58.5 years; 255 (78%) were postmenopausal. Of the 326, 47 (14%) had a BMD measurement indicating osteoporosis at one or more of the three sites where BMD was measured, and a further 160 (49%) had a measurement indicating osteopenia. QUS results are compared with BMD results in Box 1. The percentage of women with values in the osteoporotic range varied between QUS parameters (1% for BUA, 17% for VOS and the combined BUA-VOS category, and 14% for stiffness). Sensitivity and specificity of QUS Sensitivity and specificity of QUS for predicting BMD diagnostic categories are shown in Box 2. Sensitivity and specificity varied between QUS parameters. A BUA result in the osteoporotic range (T < -2.5) had very low sensitivity for predicting BMD-defined osteoporosis (9%), but high specificity (100%). In contrast, VOS, cQUS and stiffness results in the osteoporotic range had sensitivities of almost 50%, and specificities that were again high. For predicting either osteoporosis or osteopenia, stiffness had the best combination of sensitivity (77%) and specificity (81%). Positive and negative predictive values are also shown in Box 2. Negative predictive values were high (87%-91%) for QUS as a predictor of BMD-defined osteoporosis versus osteopenia/normal BMD. This indicated that a woman with BMD-defined osteoporosis was unlikely to have a QUS result in the normal-osteopenic range. Likelihood ratios of QUS LRs for different QUS results to predict BMD-defined osteoporosis are summarised in Box 2 and interpreted in Box 3. Normal QUS result: A BUA, VOS, cQUS or stiffness result in the normal range had a low LR (0-0.2) (ie, a normal result significantly lowered the odds or probability of the woman's having BMD-defined osteoporosis). Osteoporotic QUS result: A BUA result in the osteoporotic range had an LR approaching infinity and so was highly predictive of BMD-defined osteoporosis. In contrast, a VOS, cQUS or stiffness result in the osteoporotic range had a much lower LR (4.0-5.2), increasing the odds of BMD-defined osteoporosis, but to a lesser extent than a BUA result in the osteoporotic range. Osteopenic QUS result: QUS results in the osteopenic range were less predictive, as LR ranged from 1.0 to 2.4. Between 37% and 50% of subjects (depending on the QUS parameter) had results in this range (Box 1). LRs for predicting low BMD (ie, osteoporosis or osteopenia; BMD T score < -1) are also shown in Box 2, and followed a similar pattern to LRs for predicting BMD-defined osteoporosis. When QUS T-score cut-off values for osteoporosis were increased from -2.5 to -1.0, sensitivity increased, but at the expense of decreasing specificity and LR (data not shown). For example, a BUA cut-off of -1.0 gave 83% sensitivity, 69% specificity and LR, 2.6. Corresponding values for a VOS cut-off of -1.0 were 96%, 41% and 1.6. Discussion We found that QUS had variable usefulness in predicting BMD categories. Specificity for predicting BMD-defined osteoporosis was high for all QUS parameters (88%-100%), but sensitivity was low and variable (9%-47%). A BUA result in the osteoporotic range was highly predictive of BMD-defined osteoporosis (LR, Infinity), but had low sensitivity (9%). Results in the osteoporotic range for other QUS parameters and in the osteopenic range for all QUS parameters were less predictive of BMD-defined osteoporosis (LRs, 1.0-5.2). In the light of our study, how can we interpret QUS results? If a BUA result is in the osteoporotic range (LR, Infinity), then BMD-defined osteoporosis is almost certain (predictive value, 100%). However, the low sensitivity of BUA (9%) means that many women with osteoporosis would be missed if BUA alone were used. If results are normal for both BUA and VOS (cQUS normal; LR, 0), we can confidently rule out BMD-defined osteoporosis. If LR is 0, then, no matter what the pre-test probability of osteoporosis, the post-test probability will be < 5% (Box 3). All results in the osteopenic range, and VOS and stiffness results in the osteoporotic range, are less predictive of BMD category. Therefore, if QUS were performed on a population similar to ours (14% prevalence of osteoporosis), then (from Box 1) 1% would have a BUA result in the osteoporotic range (likely to have osteoporosis) and 33% would have both BUA and VOS results in the normal range (osteoporosis could fairly confidently be ruled out, with a post-test probability < 5%). However, there would be a degree of uncertainty about the remaining 66%, who would then need a DEXA scan to identify those with osteoporosis. Previous studies of QUS as a predictor of BMD have generally used conventional sensitivity and specificity analyses only, not LRs, and have not used the WHO BMD definitions. For example, two community-based cross-sectional studies on 700 postmenopausal10 and 1000 perimenopausal women,11 respectively, found that there was a 40%-50% overlap in the number of women in the lowest quartile of both DEXA and QUS measurements. Two other studies found QUS parameters to have a sensitivity of 65%-70% for BMD in the lowest quartile.6,12 Only one study other than ours has evaluated QUS in terms of WHO BMD definitions. It found BUA and VOS to have higher sensitivities, of 77% and 69%, respectively, for diagnosing osteoporosis in 100 women aged 60-69 years.13These higher sensitivities may have been due to use of higher BUA and VOS cut-off values. As expected, specificities were lower than in our study. The ultrasound instrument used in this study, the McCue Cuba Mark II, is not identical to the Achilles ultrasound instrument (Lunar, Madison, Wis, USA) used in Australian pharmacies. Nevertheless, a comparison of the two machines found that BUA measurements on a Cuba Mark II instrument were highly correlated with "stiffness" measurements on a Lunar Achilles instrument (r = 0.906; 95% CI, 0.873-0.931).14 Another study compared measurements of 30 women between the Cuba Mark II used in our study and an Achilles, finding an r value of 0.8.15 Therefore, it is unlikely that the Achilles instrument would be a significantly better predictor of BMD than our Cuba Mark II. There is no consensus on what cut-off values to use with QUS to diagnose osteoporosis. We found that changing the cut-off could achieve higher sensitivity, but only by accepting higher rates of false positives (lower specificity) and less discriminating LRs. Although there is enough evidence to support the use of QUS as an independent predictor of fracture risk,1,2 our study shows that QUS should not been seen as a substitute for BMD measurement. The results of our study suggest that, when women in the community are "screened" for osteoporosis using QUS, a few women will be confidently identified with BMD-defined osteoporosis. Another small group will be able to be reassured that they are unlikely to have osteoporosis. However, for the great majority, the presence or absence of osteoporosis will remain uncertain. References Hans D, Dargent MP, Schott AM, et al. Ultrasonographic heel measurements to predict hip fracture in elderly women: the EPIDOS prospective study. Lancet 1996; 348 (9026): 511-514. Bauer DC, Gluer CC, Cauley JA, et al. Broadband ultrasound attenuation predicts fractures strongly and independently of densitometry in older women. A prospective study. Study of Osteoporotic Fractures Research Group. Arch Intern Med 1997; 157: 629-634. Sackett DL, Haynes BR, Guyatt GH, et al. Clinical epidemiology. A basic science for clinical medicine. 2nd ed. Boston: Little, Brown and Company, 1991. Gluer CC, Wu CY, Jergas M, et al. Three quantitative ultrasound parameters reflect bone structure. Calcif Tissue Int 1994; 55: 46-52. World Health Organization Study Group. Assessment of fracture risk and its application to screening for prostmenopausal osteoporosis. World Health Organ Tech Rep Ser 1994; 843: 1-129. Herd RJ, Blake GM, Miller CG, et al. The ultrasonic assessment of osteopenia as defined by dual X-ray absorptiometry. Br J Radiol 1994; 67: 631-635. Fletcher RH, Fletcher SW, Wagner EH. Clinical epidemiology: the essentials. 3rd ed. Baltimore: Williams & Wilkins, 1988: 65. Fagan TJ. Normogram for Bayes theorem [letter]. N Engl J Med 1975: 293; 257. Australian National Consensus Conference 1996. The prevention and management of osteoporosis. Consensus statement. Med J Aust 1997; 167 Suppl: S1-S15. van Daele Burger H, Algra D, et al. Age-associated changes in ultrasound measurements of the calcaneus in men and women: the Rotterdam Study. J Bone Miner Res 1994; 9: 1751-1757. Massie A, Reid DM, Porter RW. Screening for osteoporosis: comparison between dual energy X-ray absorptiometry and broadband ultrasound attenuation in 1000 perimenopausal women. Osteoporos Int 1993; 3: 107-110. Young H, Howey S, Purdie DW. Broadband ultrasound attenuation compared with dual-energy X-ray absorptiometry in screening for postmenopausal low bone density. Osteoporos Int 1993; 3: 160-164. Langton CM, Ballard PA, Bennett DK, Purdie DW. A comparison of the sensitivity and specificity of calcaneal ultrasound measurements with clinical criteria for bone densitometry (DEXA) referral. Clin Rheumatol 1997; 16: 117-118. Greenspan SL, Bouxsein ML, Melton ME, et al. Precision and discriminatory ability of calcaneal bone assessment technologies. J Bone Miner Res 1997; 12: 1303-1313. Harris ND, Griffiths MR, Nguyen TV, et al. Quantitative ultrasound of the heel: a comparison of Lunar and McCue instruments [abstract]. Proceedings of the Australian and New Zealand Bone and Mineral Society 7th Annual Scientific Meeting. 1997. 29 Sept-1 Oct; Canberra, ACT: 62. (Received 24 Dec 1998, accepted 9 Jul 1999) Authors' details Department of Rheumatology, Royal North Shore Hospital, Sydney, NSW. Vasi Naganathan, FRACP, Research Scholar; Lyn March, FRACP, FAFPHM, Staff Specialist; David Hunter, MB BS, Advanced Physician Trainee in Rheumatology; Joanna Markovey, MSc, Research Bone Densitometry Technician; Philip N Sambrook, FRACP, MD, Head of Department. Department of Nuclear Medicine, St Vincent's Hospital, Sydney, NSW. Nick A Pocock, FRACP, MD, Senior Staff Specialist. Reprints will not be available from the authors. Correspondence: Dr V Naganathan, Department of Rheumatology, Royal North Shore Hospital, St Leonards, NSW 2065. Email: vasinATmed.usyd.edu.au 1: Association between quantitative heel ultrasound (QUS) results and bone mineral density* in 326 women aged 45-80 yearsBone mineral densityQUS resultNormal (n = 119)Osteopenia (n = 160)Osteoporosis (n = 47)Total (n = 326)Broadband ultrasound (BUA) Normal103918202 (62%) Osteopenia166935120(37%) Osteoporosis0044 (1%)Velocity of sound (VOS) Normal81332116 (36%) Osteopenia369623155 (47%) Osteoporosis2312255 (17%)Combined category (cQUS) Normal79280107 (33%) Osteopenia3810125164 (50%) Osteoporosis2312255 (17%)Stiffness Normal96444144 (44%) Osteopenia209421135 (41%) Osteoporosis3222247 (14%) * Measured by dual energy x-ray absorptiometry. Combined result for broadband ultrasound and velocity of sound: normal if both normal; osteopenic if either osteopenic and neither osteoporotic; and osteoporotic if either osteoporotic. Back to text 2: Use of ultrasound parameters to predict osteoporosis or osteopenia defined by dual energy x-ray absorptiometry (DEXA) measurement of bone mineral densityPredictive valuesLikelihood ratio for ultrasound measurement (95% confidence interval)Sensitivity SpecificityPositiveNegativeNormalOsteopeniaOsteoporosisTo predict osteoporosisBroadband ultrasound (BUA)9%100%100%87%0.2 (0.11-0.38)2.4 (1.9-3.1)InfinityVelocity of sound (VOS)46%88%40%91%0.1 (0.03-0.4)1.0 (0.7-1.4)4.0 (2.6-6.2)Combined category* (cQUS)47%88%40%91%0(1.1 (0.8-1.5)4.0 (2.6-6.2)Stiffness47%91%46%91%0.2 (0.08-0.5)1.1 (0.8-1.6)5.2 (3.2-8.4)To predict osteoporosis or osteopeniaBroadband ultrasound (BUA)52%87%87%51%0.6 (0.5-0.7)3.7 (2.3-6.0)InfinityVelocity of sound (VOS)83%68%82%70%0.25 (0.18-0.35)1.9 (1.4-2.6)15.0 (3.7-60.5)Combined category* (cQUS)86%66%82%74%0.20 (0.14-0.29)1.9 (1.4-2.5)15.2 (3.8-61.3)Stiffness77%81%87%67%0.3 (0.2-0.4)3.3 (2.2-5.0)8.4 (2.7-26.5) DEXA = dual energy x-ray absorptiometry. * Combined broadband ultrasound attenuation and velocity of sound category. Back to text 3: Interpretation of likelihood ratios The likelihood ratio (LR) indicates how much a test result raises or lowers the probability of an individual's having "disease". It can be used to determine post-test probability of a disease from the estimated pre-test probability using a normogram (Figure). Thus, if an early postmenopausal woman (aged 60-64 years) had a pre-test probability of osteoporosis of 15%,9 then, from the normogram, a cQUS result in the osteoporot ic range (LR, 4.0) would increase her probability to 40%. If she had a very high pre-test probability of osteoporosis, for example 50%, because of multiple risk factors (eg, including family history of osteoporosis and recent wrist fracture after a fall), then a cQUS result in the osteoporotic range would increase her probability to 80%. A BUA result in the osteoporotic range (LR, Infinity) would make BMD-defined osteoporosis highly likely, no matter what the pre-test risk. However, a cQUS result in the normal range (LR, 0) would make BMD-defined osteoporosis unlikely. QUS results in the osteopenic range, with LRs closer to unity (LR, 1.0-2.4), would make BMD category far less certain. Normogram for applying likelihood ratios, adapted from Fagan.8 Lines show post-test probabilities when LR = 4.0 and pre-test probabilities are 15% and 50%, respectively. Back to text

Vasi Naganathan · Lyn March · David Hunter · Nick A Pocock · Joanna Markovey · Philip N Sambrook

Research Hypertension in pregnancy: maternal and fetal outcomes according to laboratory and clinical features Mark A Brown and Megan L Buddle MJA 1996; 165: 360-365 Abstract - Introduction - Methods - Results - Discussion - Acknowledgements - References - Authors' details - - More articles on Sexual health Abstract Objectives: To determine the predictive value of clinical and laboratory parameters for maternal and fetal complications in pregnant women with hypertension. Design: Prospective data collection. Setting: Two primary referral hospitals in the southern suburbs of Sydney between March 1987 and July 1994. Subjects: 1183 pregnant women with hypertension managed conjointly by a physician and obstetrician. Intervention: Uniform management protocol, plus antihypertensive medications if systolic blood pressure was persistently ≥ 160 mmHg and/or diastolic blood pressure ≥ 90 mmHg. Main outcome measures: Maternal and fetal complications, as defined by the Australasian Society for the Study of Hypertension in Pregnancy Consensus Statement. Results: Of 825 women with pre-eclampsia (502 mild; 323 severe), univariate analysis showed that hyperuricaemia, proteinuria and severe hypertension were significantly associated with a higher rate of maternal and fetal complications. In multivariate analyses without confounders, only primiparity, low serum albumin levels and absence of diabetes were significantly associated with severe pre-eclampsia. Severe pre-eclampsia, high haemoglobin levels and low platelet count were associated with higher rates of small-for-gestational-age babies, but only low serum albumin levels were associated with increased perinatal mortality rates. Low birthweight was associated with severe hypertension and severe pre-eclampsia. Conclusions: Simple clinical and laboratory parameters are useful predictors for maternal and fetal outcomes in pregnancies complicated by hypertension. Introduction Hypertensive disorders in pregnancy are common and their incidence appears to be increasing.1 These disorders comprise hypertension de novo in pregnancy (variably called pre-eclampsia, gestational or transient hypertension), and chronic forms of hypertension (essential and secondary hypertension), and have multiple modes of clinical presentations. With the current emphasis on outpatient management of pregnant women with non-proteinuric hypertension,2 clinicians require readily available clinical and laboratory parameters as indicators of the likely outcome of pregnancies complicated by hypertension. Such information is often difficult to obtain because of the selective nature of cases and complications reported by high-risk tertiary-care obstetric units, and is confounded by the varying treatments given to pregnant women before referral.3,4 Our study examined prospectively the maternal and fetal outcomes in women with hypertension in pregnancy, and the associated clinical and laboratory features. In particular, we sought to determine whether complications were greater if the woman had hyperuricaemia, severe hypertension and/or proteinuria, was nulliparous, or presented with hypertension early in the pregnancy. We also sought to determine which clinical and laboratory features were associated with severe pre-eclampsia. Methods The study group consisted of all pregnant women with hypertension who were referred by their obstetrician to one physician (M A B) at primary referral hospitals (St George or Hurstville community hospitals) in Sydney between March 1987 and July 1994. Indications for referral were: Hypertension failing to settle after overnight rest in hospital, or repeated measurement in a day-only unit; The presence of proteinuria or other abnormal urinalysis, abnormal biochemistry (serum electrolytes and creatinine levels, liver function tests) or thrombocytopenia; Recurrent admissions for hypertension; or A suspected secondary cause for hypertension. The definitions, classifications and complications of hypertension in pregnancy used in our study were those of the Australasian Society for the Study of Hypertension in Pregnancy (ASSHP) (Boxes 1 and 2).5 The maternal biochemical and haematological data used in our analysis were those recorded immediately before delivery. Complications were those having occurred by the time of discharge from hospital or by the postpartum check-up (after three months). Each patient's management was overseen by one physician (M A B) in conjunction with the patient's obstetrician. A standardised management protocol was followed by all staff during the course of this study (Box 3). Statistical analysis All data were stored in a database program (iBANK) and analysed using Minitab statistical package6 and BMDP statistical software.7For initial bivariate analyses, the frequency of adverse events within groups was tested by χ2 analysis and, where significant, by 2 x 2 contingency tables. Biochemical and haematological indices and birthweights were compared using Student's t tests, and significance assessed using Hochberg's method for interpreting multiple comparisons. The maternal-outcome variable was severity of pre-eclampsia. Logistic regression analysis was used to analyse the following variables as predictors of severe pre-eclampsia: parity; gestation at presentation; age; plasma uric acid level; serum albumin, haemoglobin and haematocrit levels; and diabetes. Proteinuria, platelet count and serum creatinine levels were not analysed because abnormalities of these form part of the diagnosis of severe pre-eclampsia. Fetal-outcome variables were birthweight, small-for-gestational-age (SGA) and perinatal mortality (PNM). Each of the following variables was analysed as a predictor of poor fetal outcome: maternal complications, age, parity, diabetes, gestation at presentation and laboratory data. Multiple regression analysis was used for birthweight and logistic regression analysis for SGA and PNM. For regressions, backward elimination was used until all remaining predictors were statistically significant (P < 0.05). Results Clinical details From March 1987 to July 1994, 1183 pregnant women with hypertension fulfilled the criteria for referral to a physician (M A B) and were entered into the study, representing 6.7% of all confinements at the two hospitals. During this period, 2121 women were documented in hospital records as having hypertension in pregnancy, representing 12% of all confinements. There were 825 women (70%) with pre-eclampsia (mild, 502; severe, 323), 82 (7%) with superimposed pre-eclampsia, 223 (19%) with essential hypertension, and 53 (4%) with secondary hypertension (predominantly renal disease). Oral antihypertensive medications were given to 563 women (48%), additional parenteral antihypertensives were required by 281 (24%), and 339 (29%) received no antihypertensive medications. Thirteen women (1%) received "prophylactic" low-dose aspirin. Five women (0.4%) required dialysis during or after pregnancy and four women (0.3%) had convulsions. Seventy-three women (6%) had pre-existing or gestational diabetes mellitus. Of the 1242 babies born to the women in the study, there were 15 perinatal deaths (12 per 1000 hypertensive pregnancies). There was one maternal death (0.08%) in a 29-year-old woman who presented at 29 weeks' gestation with severe pre-eclampsia and who delivered immediately and died unexpectedly the next day. Despite autopsy, the cause of death was not established. Predictors of outcomes in women with pre-eclampsia Univariate analysis Hyperuricaemia was significantly associated with higher rates of all maternal complications and SGA babies and babies with lower average birthweights than normal plasma uric acid levels (Box 4). Severe hypertension was significantly associated with higher maternal complication rates and earlier gestation at delivery than mild hypertension (Box 4). This group of women had babies of a lower average birthweight, a higher proportion of SGA babies and a higher perinatal mortality rate. Proteinuria was significantly associated with similar maternal and fetal complication rates as severe hypertension, and the proportion of SGA babies was similar to non-proteinuric hypertension (Box 4). Parity did not affect maternal complication rates, except for a higher prevalence of severe hypertension and liver disease in nulliparae (Box 5). Early gestation at presentation with hypertension (at or before 32 weeks' gestation) was significantly associated with a higher prevalence of proteinuria than later presentation (Box 5). Multivariate analysis Severe pre-eclampsia was associated with low serum albumin (odds ratio [OR], 0.93; 95% confidence interval [CI], 0.90-0.97; P = 0.0001), primiparity (OR, 0.51; 95% CI, 0.33-0.79; P = 0.002) and the absence of diabetes (OR, 0.21; 95% CI, 0.05-0.93; P = 0.04). SGA was associated with severe pre-eclampsia (OR, 2.5; 95% CI, 1.6-3.9; P = 0.0001), high maternal haemoglobin levels (OR, 3.7; 95% CI, 1.1-12.8; P = 0.04) and low platelet counts (OR, 1.4; 95% CI, 1.1-1.8; P = 0.02). Birthweight was positively associated with gestation at presentation (regression coefficient [β], 41; standard error [SE], 5; P < 0.0001) and serum albumin levels (β, 28; SE, 5; P < 0.0001) and inversely associated with severe hypertension (β, 2193; SE, 69; P < 0.005) and severe pre-eclampsia (β, 2311; SE, 60; P < 0.0001). Perinatal mortality was associated only with low serum albumin levels (OR, 0.88; 95% CI, 0.82-0.95; P < 0.0003), but there were too few perinatal deaths to support a useful formal analysis. Outcomes in women with all forms of hypertension in pregnancy Box 6 shows the maternal and fetal outcomes in women with the different forms of hypertension in pregnancy. Maternal and fetal outcomes were better for women with essential hypertension than for the other groups. Twenty-seven per cent (or 10% if proteinuria was included in the diagnosis) of women with essential hypertension developed superimposed pre-eclampsia. Women with superimposed pre-eclampsia delivered their babies earlier, had babies with lower average birthweights, significantly more SGA babies and significantly higher rates of all maternal complications than women with essential hypertension alone. Women with secondary (predominantly renal) hypertension had similar maternal and fetal outcomes to women with superimposed pre-eclampsia (data available from author). Discussion In our prospective study, there was uniform management of all women with hypertensive disorders in pregnancy. As the database excluded very mild cases of hypertension in pregnancy, an inherent bias exists, but our study provides a general estimate of fetal and maternal outcomes for women with hypertension managed in primary referral hospitals. During the study period, 12% of all confinements were complicated by hypertension, which is either similar to or higher than previous reports.8-11 Hjertberg et al. noted a 5% incidence of hypertension in pregnancy, of whom 83% had hypertension de novo in pregnancy.12 In our study, 70% of women had pre-eclampsia (hypertension de novo in pregnancy), 19% had essential hypertension alone, 7% had superimposed pre-eclampsia, and 4% had a renal or other secondary cause. These results are similar to those from a smaller study in a tertiary referral unit.13 When analysed separately, hyperuricaemia, proteinuria and severe hypertension were associated with higher rates of all maternal and fetal complications, while parity was associated with severe hypertension and liver disease. Presentation before 32 weeks' gestation impacted negatively on fetal outcome. We have reported previously on the adverse impact of proteinuria on maternal and fetal outcomes.14 Hyperuricaemia is an established marker of severe pre-eclampsia, correlating histologically with the severity of renal lesions,15 and clinically with adverse fetal outcomes.16 We found that hyperuricaemia, when considered alone, was associated with higher maternal complication rates and fetal growth retardation, but, with multivariate analysis, was not a significant predictor of adverse maternal or fetal outcomes. Hyperuricaemia remains a useful marker of pre-eclampsia, but women with normal serum uric acid levels and pre-eclampsia may still develop complications. Severe hypertension was confirmed by multivariate analysis to be associated with low birthweight babies, and, in univariate analysis, with a higher incidence of all fetal complications and a higher incidence of SGA babies. Therefore, the more severe the hypertension, the more the likelihood of fetal as well as maternal complications. Pre-eclampsia has long been considered a disorder of primigravidae.17 However, it is clear that pre-eclampsia does occur in second or subsequent pregnancies (particularly following a change of partner); in our study, approximately two-fifths of the women with pre-eclampsia were parous women. Parous women had a lower incidence of severe hypertension and liver disease when compared with nulliparae, but there were no other differences in outcomes between these groups. Therefore, equal risk to the fetus should be assumed in both groups of women once hypertension has developed in the second half of pregnancy. The gestation at presentation in pre-eclampsia influences fetal outcome because of the likelihood of early delivery, but we are unaware of any prospective studies that relate gestation at presentation to maternal complication rates. Apart from proteinuria, maternal complication rates for women presenting before 33 weeks' gestation in our study were similar to those who presented later. Two retrospective studies have reported a higher incidence of maternal complications in women with pre-eclampsia of early onset.18,19 Our patients presenting early with pre-eclampsia had milder disease than that reported by Sibai et al.20 As expected, the earlier the gestation with pre-eclampsia the lower the birthweight (because gestation did not proceed as far as those presenting later), but the likelihood of having an SGA baby was not increased. Although maternal diabetes was associated with a lower risk of severe pre-eclampsia, only 6% of our study population had diabetes. Therefore, it should not be assumed that diabetes protects pregnant women against severe pre-eclampsia. The consistent relationship between low serum albumin levels and adverse maternal and fetal outcomes was independent of proteinuria and implies that hepatic albumin synthesis is altered in severe pre-eclampsia. The association between high maternal haemoglobin levels and SGA babies presumably reflects a reduced plasma volume and haemoconcentration, which is associated with poor fetal growth.21 In conclusion, for women with pre-eclampsia, the traditional clinical markers of proteinuria, parity and severe hypertension remain useful clinical indices of maternal risk, while severe pre-eclampsia, severe hypertension, elevated haemoglobin and low serum albumin levels portend increased fetal risk. Acknowledgements We wish to thank the obstetricians of St George and Hurstville community hospitals, Professor J A Whitworth, Ms Jodie Wilkinson and Dr M Jones (Intstat Australia Pty Ltd). References Crawford JS. Epidemic pre-eclampsia. Lancet 1987; 1: 329-330. Tuffnell DJ, Lilford RJ, Buchan PC, et al. Randomised controlled trial of day care for hypertension in pregnancy. Lancet 1992; 339: 224-227. Horvath JS, Korda A, Child A, et al. Hypertension in pregnancy. A study of 142 women presenting before 32 weeks' gestation. Med J Aust 1985; 143: 19-21. Clarke M, Mason ES, Macvicar J, Clayton DG. Evaluating perinatal mortality rates: effects of referral and case mix. BMJ 1993; 306: 824-827. Australasian Society for the Study of Hypertension in Pregnancy. Consensus Statement on Management of Hypertension in Pregnancy: Executive Summary. Med J Aust 1993; 158: 700-702. Minitab [computer program], version 9.1. Pennsylvania State College: Minitab Inc, 1993. BMDP [computer program]. Berkeley, Calif: University of California, 1990. Martikainen AM, Heinonen KM, Saarikoski SV. The effect of hypertension in pregnancy on fetal and neonatal condition. Int J Gynaecol Obstet 1989; 30: 213-220. Andrews WW, Cox SM, Sherman ML, Leveno KJ. Maternal and perinatal effects of hypertension at term. J Reprod Med 1992; 37: 73-76. Safflas AF, Oldson DR, Franks AL, et al. Epidemiology of pre-eclampsia and eclampsia in the United States, 1979-1986. Am J Obstet Gynecol 1990; 163: 460-465. Pietrantoni M, O'Brien WF. The current impact of the hypertensive disorders of pregnancy. Clin Exp Hypertens 1994; 16: 479-492. Hjertberg R, Bellrage P, Hanson U. Conservative treatment of mild and moderate hypertension in pregnancy. Acta Obstet Gynecol Scand 1992; 71: 439-446. Horvath JS, Phippard A, Henderson-Smart D, et al. High risk hypertensive pregnancies: maternal and foetal outcome. Clin Exp Hypertens Pregnancy 1983; B2: 21-28. Brown MA, Buddle ML. The importance of non proteinuric hypertension in pregnancy. Hypertens Pregnancy 1995; 14: 57-65. Pollak VE, Nettles JB. The kidney in toxemia of pregnancy: a clinical and pathologic study based on renal biopsies. Medicine (Baltimore) 1960; 39: 469-526. Redman CWG, Beilin LJ, Bonnar J, Wilkinson PH. Plasma urate measurements in predicting fetal death in hypertensive pregnancy. Lancet 1976; 2: 1370-1373. Chesley LC. Diagnosis of pre-eclampsia. Obstet Gynecol 1985; 65: 423-425. Sibai BM, Taslimi M, Abdella TN, et al. Maternal and perinatal outcome of conservative management of severe pre-eclampsia in midtrimester. Am J Obstet Gynecol 1985; 152: 32-37. Railton A, Allen DG. Management and outcome of pregnancy complicated by severe pre-eclampsia of early onset. S Afr Med J 1987; 72: 608-610. Sibai BM, Mercer BM, Schiff E, Friedman SA. Aggressive versus expectant management of severe pre-eclampsia at 28 to 32 weeks' gestation: a randomized controlled trial. Am J Obstet Gynecol 1994; 171: 818-822. Brown MA, Zammit VC, Mitar DM. Extracellular fluid volumes in pregnancy-induced hypertension. J Hypertens 1992; 10: 821-829. (Received 9 Oct 1995, accepted 13 May 1996) Authors' details Departments of Medicine, Renal Medicine and Obstetrics, St George and Hurstville community hospitals, University of NSW. Mark A Brown, MB BS, FRACP, MD, Associate Professor of Medicine; Megan L Buddle, SRN, Hypertension Unit Registered Nurse. No reprints will be available. Correspondence: Associate Professor M A Brown, Department of Renal Medicine, St George Hospital, Kogarah, NSW 2217. E-mail: "nest::brown"ATdodo.ssahs.unsw.edu.au 1: ASSHP classifications and definitions of hypertension in pregnancy Hypertension: A sitting systolic blood pressure (BP) ≥ 140 mmHg and/or diastolic BP (phase IV Korotkoff sound) ≥ 90 mmHg, or a rise in systolic BP of ≥ 25 mmHg and/or diastolic BP ≥ 15 mmHg from first-trimester BP (two readings taken six hours apart). Proteinuria: ≥ 300 mg/day (24-hour urine collection) or persistently (over two or more days) ≥ 2+ protein (1 g/L) on urinalysis (dipstick testing). As urinalysis was performed daily, non-proteinuric hypertension that progressed to proteinuric hypertension was always detected. Hyperuricaemia: Plasma uric acid level > 0.35 mmol/L. Pre-eclampsia: The development of hypertension after 20 weeks' gestation in a woman with no known history of hypertension or renal disease and whose blood pressure (BP) was normal in the first half of the pregnancy and returned to normal after delivery: (i) mild pre-eclampsia: hypertension only; (ii) severe pre-eclampsia: hypertension and evidence of maternal organ dysfunction (see Maternal complications, Box 2). Essential hypertension: Hypertension in the first half of pregnancy (often occurring before pregnancy) without a demonstrable secondary cause, or the appearance of hypertension in the second half of pregnancy in a woman whose BP failed to return to normal within three months after delivery. Superimposed pre-eclampsia: The development of proteinuria and/or hyperuricaemia in the second half of pregnancy in a woman with essential hypertension. Secondary hypertension: To exclude a secondary cause of hypertension, all women had serum potassium levels measured and had urinalysis, microscopy and a clinical examination; urinary catecholamines and renal ultrasound were performed in selected cases. Within this group, superimposed pre-eclampsia was not diagnosed as a separate entity as it was too difficult to ascertain whether changes (e.g., increasing proteinuria) reflected superimposed pre-eclampsia or progression of the underlying (usually renal) disease. ASSHP = Australasian Society for the Study of Hypertension in Pregnancy.5 Back to text 2: ASSHP complications of hypertension in pregnancyMaternal complications:(i)Severe hypertension: Blood pressure (BP) ≥ 170 mmHg systolic and/or ≥ 110 mmHg diastolic;(ii)Proteinuria (Box 1);(iii)Haematological abnormalities: haemolysis (haemoglobin < 110g/L, with reticulocytosis, fragments) or thrombocytopenia (< 150 x 109/L platelets);(iv)Renal impairment: serum creatinine level ≥ 0.10 mmol/L;(v)Liver disease: elevated serum aspartate aminotransferase level > 40 IU/L, with or without severe epigastric pain;(vi)Neurological disturbances: visual scotoma; severe headaches with hyperreflexia; hyperreflexia with sustained clonus (> 3 beats). Fetal complications:(i)Small-for-gestational-age (SGA): birthweight below the 10th centile for gestation, corrected for sex;(ii)Perinatal mortality (PNM): stillbirths and neonatal deaths per 1000 hypertensive pregnancies. ASSHP = Australasian Society for the Study of Hypertension in Pregnancy.5Back to text 3: Management protocol during pregnancy Maternal complications: Blood pressure (BP) measurements six times per day; Fetal cardiotocograph recordings on alternate days (or more often if clinically indicated); Daily urinalysis; Biochemistry and haematological testing at least twice weekly; Bedrest if there is evidence of intrauterine fetal growth retardation, or maternal neurological features or severe hypertension. Antihypertensive therapy: Oral antihypertensive therapy (single agent or combinations) of oxprenolol, hydralazine, methyldopa, prazosin or nifedipine was instituted if the systolic BP was persistently > 160 mmHg or diastolic BP persistently > 90 mmHg after overnight bedrest in hospital. Parenteral antihypertensive therapy (sublingual nifedipine or intravenous hydralazine) was given for severe hypertension (Box 2). Convulsion prophylaxis (intravenous phenytoin followed by oral phenytoin) was used for women with neurological disturbances (Box 2). Indications for delivery. Delivery (by induction) was scheduled after 38 weeks' gestation if the BP remained stable and the cervix was favourable. Indications for delivery before 38 weeks' gestation were:5 Fetal compromise: premorbid (cardiotocographic) tracing of fetal heart rate or failure of fetal growth (assessed by doppler ultrasound). Inability to control maternal BP; Persistent neurological disturbances (see Box 2); Progressively rising liver enzyme or serum creatinine levels, or worsening thrombocytopenia. Women whose pregnancies could not be prolonged beyond 30 weeks' (and occasionally 32 weeks') gestation were sometimes transferred before delivery to a tertiary referral unit with neonatal intensive care facilities. Outcomes of this group were included in our analysis. Back to textFor box 4 click here 5: Maternal and fetal outcomes in women with pre-eclampsia (mild and severe) according to parity and gestation at presentation (see text for definitions). Data are percentages or means (standard deviation). Gestation at presentationNulliparae ParousP ≤ 32 wks> 32 wks Pn508317129696Age (years)27 (5)30 (5) < 0.0001*29 (5)28 (5)0.30Nulliparae (%)10000.0001*57630.20Gestation at presentation (weeks)36 (4)35 (6)0.00927 (7)37 (2)< 0.0001*Gestation at delivery (weeks)38 (3)38 (2) 0.8935 (4)38 (2)< 0.0001*Hyperuricaemia (%)61540.0456590.53Maternal complicationsSevere hypertension24%15%0.001*29%19%0.008Thrombocytopenia11%8%0.1011%10%0.75Proteinuria21%16%0.0931%17%0.003*Renal insufficiency8%5%0.089%7%0.27Liver disease12%6%0.005*15%9%0.04Neurological13%10%0.4116%11%0.09Laboratory dataPlasma creatinine (mmol/L).07 (.04).07 (.02).18.08 (.08)0.07 (.02) 0.51Plasma uric acid (mmol/L).40 (.20).37 (.04) .01.38 (.09)0.39 (0.17)0.66Plasma albumin (g/L)33 (5)34 (5) .3431 (8)34 (4)0.002*Haemoglobin (g/L)119 (23)122 (16) .03117 (37)121 (15)0.20Haematocrit (%)35.1 (5.5)35.9 (5.6) .1033.7 (8.7)35.7 (4.7)0.02Platelets (x 109/L)223 (82)232 (82) .11206 (95)230 (79)0.009*Fetal complicationsSGA (%)17210.1724180.10PNM (per 1000)1730.06544< 0.0001*Birthweight (g)2984 (701)2985 (694) 0.982312 (960)3105 (561)< 0.0001* * Statistically significant . SGA = small-for-gestational-age. PNM = perinatal mortality. Back to text 6: Maternal and fetal outcomes in women with different forms of hypertension in pregnancy (ASSHP classifications). Data are percentages or means (standard deviation). Pre-eclampsiaEssential hypertension Superimposed pre-eclampsiaP*n82523382Age (years)28 (5)30 (530 (5)> 0.0001†Nulliparae (%)6237560.290.002†Gestation at presentation (weeks)35 (5)26 (10)28 (10) < 0.0001†0.17Gestation at delivery (weeks)38 (3)38 (2)36 (1)0.0007†< 0.0001†Hyperuricaemia (%)581683< 0.0001†0.0001†Maternal complicationsSevere hypertension21%13%38%0.0005†< 0.0001†Thrombocytopenia10%1%16%0.11< 0.0001†Proteinuria19%035%0.0005†< 0.0001†Renal insufficiency7%09%0.62< 0.0001†Liver disease10%1%11%0.79< 0.0001†Neurological12%3%20%0.06< 0.0001†Laboratory dataPlasma creatinine (mmol/L)0.07 (0.04)0.06 (0.02)0.07 (0.02)0.78< 0.0001†Plasma uric acid (mmol/L)0.39 (0.16)0.32 (0.25)0.46 (0.35)0.090.002†Plasma albumin (g/L)33 (5)35 (6)33 (5)0.810.04Haemoglobin (g/L)120 (20)119 (14) 117 (17)0.140.52Haematocrit (%)35.4 (5.5)35.1 (4.0) 34.7 (5.4)0.290.64Platelets (x 109/L)226 (82)256 (68) 235 (78)0.360.03Fetal complicationsSGA (%)199280.20< 0.0001†PNM (per 1000)129120.930.80Birthweight (g)2984 (701)3260 (652) 2795 (935)0.080.0001†* Significance of difference between outcomes for women with pre-eclampsia and superimposed pre-eclampsia (and for women with essential hypertension and superimposed pre-eclampsia). †Statistically significant. SGA = small-for-gestational-age. PNM = perinatal mortality. ASSHP = Australasian Society for the Study of Hypertension in Pregnancy.5 Back to text

Mark A Brown · Megan L Buddle

Research Evaluation of the PAPNET system in a general pathology service Annabelle Farnsworth, Fay M Chambers and Colin S Goldschmidt MJA 1996; 165: 429-431 Abstract - Introduction - Methods - Results - Discussion - References - Disclaimer of conflict of interest - Authors' details - - More articles on Obstetrics & gynaecology and women's health Abstract Objective: To compare the results of an automated rescreening device (PAPNET) with manual screening of Papanicolaou (Pap) smears. Design: All normal or technically unsatisfactory smears and a random sample of abnormal smears on manual screening were submitted for PAPNET rescreening. Setting: Large general pathology laboratory in Sydney between January and September 1995. Results: Of 54 658 PAP smears classified on manual screening as normal, 266 were reclassified as abnormal after PAPNET screening (32 atypical squamous cells of uncertain significance, 217 low-grade squamous intraepithelial lesions and 17 high-grade intraepithelial lesions). Of the random sample of 1022 smears classified on manual screening as abnormal, all high-grade intraepithelial lesions (122 smears) and squamous cell carcinomas (2 smears) were also detected by PAPNET, and 112 were reclassified as normal by PAPNET (14 atypical squamous cells of uncertain significance and 98 low-grade squamous intraepithelial lesions). Histological follow-up confirmed 15 of the 17 smears classified as high-grade intraepithelial lesions on PAPNET screening and detected a further seven that had been classified by PAPNET as either atypical squamous cells of uncertain significance or low-grade squamous intraepithelial lesions. Conclusions: When used as a quality-control measure in a general pathology laboratory, the PAPNET automated screening system detects higher numbers of abnormal PAP smears than manual screening. Introduction Papanicolaou (Pap)-smear examination has been shown to lower mortality and morbidity from cervical cancer,1 but the screening test itself is imperfect. False negative results (when a smear is reported as normal but has been taken from a woman with a high-grade lesion in her cervix) have been reported to be between 10%-20%.2The failure of Pap smears to detect some high-grade abnormalities is the result of either sampling or laboratory error.2 Sampling error occurs when the person taking the smear fails to collect the abnormal cervical cells because of either poor technique or the small size or localised nature of the lesion.3 Laboratory error is usually caused by the screening cytologist failing to detect the abnormal cells on the slide. Screening of Pap smears is a labour-intensive task requiring a high degree of skill and concentration. Even in laboratories using well-recognised quality control measures and staffed by highly trained cytologists, many high-grade abnormalities remain undetected by routine manual screening. These may also be missed on a manual rescreen.3 PAPNET is an automated interactive system for analysis of Pap smears which has been shown to detect abnormalities that were repeatedly missed on manual screening.4,5 We introduced the PAPNET system into our laboratory in January 1995 as a quality-control measure for rescreening of all negative (normal) smears. Here, we compare the detection of abnormalities by PAPNET with those detected by manual screening in our laboratory, and report on the histological follow-up of the PAPNET cytopredictions and the work practices of cytologists using this system. Methods Our laboratory is a large general practice in Sydney with a referral base covering metropolitan and country areas of New South Wales. During the period of our survey (23 January to 30 September 1995), all Pap smears were screened manually using routine methods. Quality assurance measures used in the study (e.g., rescreening and follow-up of abnormal smears) were routine procedures used in our laboratory. All smears considered "normal" (negative) and "technically unsatisfactory" on routine manual screening, together with a random sample of smears considered "abnormal" on routine screening during the same period, were submitted for PAPNET analysis. Smears were considered technically unsatisfactory when a confident report could not be given because of scanty material, or obscuring of cellular detail by blood, inflammation, cellular degeneration or air drying. The abnormal smears were randomly selected from the recent file of atypical smears. The PAPNET rescreening was undertaken blind to the results of manual screening. (See Box 1 for a description of the PAPNET technology.) Interim reports were generated for smears considered normal and technically unsatisfactory, and, after the slide review was completed in the laboratory, a final report was issued incorporating the PAPNET findings. Reporting terminology was based on the Bethesda system of reporting cervical/vaginal cytological diagnoses.6 A record was kept of the number of cases each cytologist was able to review per day with PAPNET. The number of slides that required manual rechecking and complete rescreening was also recorded. This involved the cytologists using the light microscope to recheck individual abnormal cells that were tagged on the PAPNET monitor. If these cells were considered to be significantly atypical, the whole slide was rescreened. If any degree of abnormality was found after rescreening, the slide was submitted to the cytopathologist for checking. To test the specificity of PAPNET, histological follow-up was performed for all patients who had an additional lesion predicted by PAPNET and were recommended to have colposcopy for this lesion. As follow-up of cytological predictions can take many months, the process was expedited and, where possible, histology results are given. In some cases, follow-up was still pending or the referring doctor had decided to follow-up by repeated observation and Pap smears. These are designated as "follow-up to come". In some cases, follow-up was unavailable; the referring doctor had been unable to contact the patient, the patient had moved or had decided to proceed no further with management. These cases are designated as "lost to follow-up". Results During the 10-month period of this survey, 60 317 Pap smears were submitted to the laboratory for examination. These smears are categorised in Box 2 (cytopredictions). The results fall within suggested National Cervical Screening Programme standards for reporting categories of technically satisfactory smears. PAPNET screening was performed on 54 658 smears classified on manual screening as negative (or within normal limits) and 1819 smears classified as technically unsatisfactory. Of the 3840 smears classified on manual screening as abnormal, 1022 smears (27%) were randomly selected and submitted for PAPNET screening: 807 low-grade squamous intraepithelial lesions, 122 high-grade squamous intraepithelial lesions, 91 atypical squamous cells of uncertain significance, and two cases of squamous cell carcinoma. Manual screening compared to PAPNET screening Of the smears classified as negative on manual screening, 266 were reclassified as abnormal after PAPNET rescreening, including 17 reported as high-grade squamous intraepithelial lesions (Box 3). This represents a 7% increase in abnormal smears after PAPNET rescreening. All of the 122 smears reported as high-grade squamous intraepithelial lesions and both squamous cell carcinomas detected on manual screening were also detected by PAPNET (Box 3). Fourteen (15%) of the 91 smears reported as atypical squamous cells of uncertain significance and 98 (12%) of the 807 reported as low-grade squamous intraepithelial lesions were reclassified as negative on PAPNET rescreening (Box 3). Histological follow-up of PAPNET cytopredictions Histology confirmed an abnormality in 28% of smears reported as atypical squamous cells of uncertain significance and 61% reported as low-grade squamous intraepithelial lesions (Box 4). Of the 17 smears reported as high-grade squamous intraepithelial lesions on PAPNET rescreening, 15 (88%) were confirmed on histology (Box 4). Three of the 32 cases reported as atypical squamous cells of uncertain significance and four of the 31 reported as low-grade squamous intraepithelial lesions (where colposcopy was recommended) were also shown to have a high-grade lesion on histological biopsy. In total, there were an extra 22 histologically confirmed high-grade lesions not detected on manual screening, of which 15 were detected by PAPNET. Cytologists' work practices In our laboratory, cytologists manually screened between 25 and 50 slides (mean, 42; standard deviation [SD], 6) per day. The large SD reflected the range of experience among the laboratory's cytologists, who alternate between gynaecological and non-gynaecological cytology. PAPNET rescreening was included in this rotation. After training, cytologists performing PAPNET rescreening were able to screen between 100 and 160 slides (mean, 126; SD, 15) per day. This included loading and unloading the tapes, viewing and assessing the images on the PAPNET screen, documentation and generating patient reports. The number of cases submitted for limited rechecking varied between 10%-90%, depending on the cytologist. The number requiring complete rescreening by a senior cytologist varied between 0 to 10 per cytologist per day and averaged five per day. The number of cases submitted to a cytopathologist by each cytologist averaged one per day. Discussion Pap smears remain the highest-volume pathology test that is not automated. Attempts have been made to automate this complex process for nearly 40 years, but it is only recently that advances in the development of automated systems have occurred which offer practical advantages in a laboratory. Although the detection rate of abnormalities on manual screening was well within normal standards in our study, PAPNET rescreening was able to detect an increased number of abnormal Pap smears. This increase was not associated with a loss of specificity, as often occurs with increased detection by manual screening.3 Many studies have shown the ability of PAPNET rescreening to detect a significant number of abnormal Pap smears when compared with manual screening, but these have been predominantly smaller studies using archived material.8,9 Our study was undertaken as part of the routine workflow in a laboratory processing a large number of smears and is the first Australian study to compare the two screening systems. The reported increase in detection of abnormalities in other studies has been as high as 16%,10 and the company which developed PAPNET (Neuromedical Systems Inc., New York, USA) claims increases of up to 30%. The increase in our study was much smaller (7%), with most detected abnormalities being low-grade lesions. However, we believe this increase was worthwhile given the reasonably high rate of detection on manual screening and the large volume of routine smears processed. Boon et al. reported the results of a large series of smears that were not double-screened, but had been either manually screened or PAPNET screened. Higher detection rates of abnormalities were found in the PAPNET group. Our study is the first report of a large series in which all negative smears were screened both manually and by PAPNET. Our study showed that the PAPNET system could be feasibly integrated into a routine working laboratory situation. It was a valuable educational and quality assurance tool. Cytologists found it easy to use and were able to review slides at three to four times their manual-screening rate. From our experience, cytologists did not become complacent in their manual screening, but rather became more cautious knowing that all the negative smears would be examined by a computerised device. The additional work for cytopathologists and senior cytologists was not great. PAPNET is only one of a number of automated cytology systems currently available as an additional test for routine Pap-smear screening. We did not undertake a comparison of other automated systems or their cost-effectiveness. Further studies on the cost-effectiveness of PAPNET testing and its efficacy for use in public health screening programs are needed. It may be that automated testing technology will replace current screening methods in the future. Disclaimer of conflict of interest This work was conducted independently. Neither Douglass Hanly Moir Pathology nor any of the individual authors has any financial or other association with the suppliers of the PAPNET service, Neuromedical Systems Inc. References Mitchell HS, Giles GG. Cancer diagnosis after a report of negative cervical cytology. Med J Aust 1996; 164: 270-273. Koss LG. The Papanicolaou test for cervical cancer detection. A triumph and a tragedy. JAMA 1989; 261: 737-743. Boscha MC, Rietveld-Scheffers PEM, Boon ME. Characteristics of false negative smears tested in the normal screening situation. Acta Cytol 1992; 36: 711-716. Koss LG, Hin E, Schreiber K, et al. Evaluation of the PAPNET cytologic screening system for quality control of cervical smears. Am J Clin Pathol 1994; 101: 220-229. Boon M, Kok CP. Neural network processing can provide means to catch errors that slip through human screening of Pap smears. Diagn Cytopathol 1993; 9: 411-416. The Bethesda system of reporting cervical/vaginal cytologic diagnoses: revised after the second National Cancer Institute Workshop, April 29-30, 1991. Acta Cytol 1993; 37: 115-124. Royal College of Pathologists of Australasia. Performance Standards for Australian Laboratories Reporting Cervical Cytology. Sydney: RCPA Quality Assurance Programs, 1995. Rosenthal DL, Acosta D, Peters RK. Computer-assisted re-screening of clinically important false negative cervical smears using the PAPNET testing system. Acta Cytol 1996; 40: 120-126. Sherman ME, Mango LJ, Kelly D, et al. PAPNET analysis of reportedly negative smears preceding the diagnosis of a high grade squamous intraepithelial lesion or carcinoma. Mod Pathol 1994; 7: 578-581. Elgert PA, Suhrland MJ, Re E, et al. Prospective quality control of cervical smears using the PAPNET system [abstract No. 5]. In: Abstracts of the Scientific Session of the 42nd Annual Scientific Meeting of the American Society of Cytology; 1994 Nov 1-6; Chicago. Acta Cytol 1994; 38: 793-805. Boon ME, Kok LP, Beck S. Histologic validation of neural network assisted cervical screening: comparison with the conventional procedure. Cell Vision 1995; 2: 23-27. (Received 10 Jan, accepted 14 Aug 1996) Authors' details Douglass Hanly Moir Pathology, 95 Epping Road, North Ryde, NSW. Annabelle Farnsworth, MB BS, FRCPA, Cytopathologist. Fay M Chambers, MB BS, FRCPA, Cytopathologist. Colin S Goldschmidt, MB BCh, FRCPA, Cytopathologist. Reprints: Dr A Farnsworth, Douglass Hanly Moir Pathology, 95 Epping Road, North Ryde 2113. Journalists are welcome to write news stories based on what they read here, but should acknowledge their source as "an article published on the Internet by The Medical Journal of Australia <http://www.mja.com.au>". <URL: http://www.mja.com.au/> We appreciate your comments. 1: The PAPNET system. PAPNET consists of a scanning apparatus and a review station.4 In Australia, the cervical slides are sent to Hong Kong for the initial scanning. The scanning apparatus includes an automated microscope with an attached video camera which transmits images to a primary classifier. Selected images are then passed into a neural net computing unit which chooses the 128 highest-ranking images for storage on an optical disk or digital tape, which is sent back to Australia for display at the review station. The review station is located in the cytology laboratory where manual screening takes place. The 128 recorded images are displayed in panels on a high-resolution colour monitor and are reviewed by a cytologist. If an abnormality is suspected, the original slide is manually checked and, if necessary, manually rescreened. Back to text 2: Categories of Pap smears on manual screening (23 January to 30 September 1995). CytopredictionNumber of smearsNegative*54 658 (90.6%)Low-grade squamous intraepithelial lesions 2974 (4.9%)High-grade squamous intraepithelial lesions 432 (0.7%)Atypical squamous cells of uncertain significance434 (0.7%)Technically unsatisfactory1819 (3%)Total60 317 (100%)* Cellular appearance was within normal limits. Back to text 3: PAPNET rescreening of PAP smears classified as normal and abnormal on manual screening. CytopredictionNormal smears: manual screeningPAPNET rescreeningAbnormal smears: manual screeningPAPNET rescreeningNegative*54 65854 359--112Atypical squamous cells of uncertain significance--329177Low-grade squamous intraepithelial lesions--217807709High-grade squamous intraepithelial lesions--17122122Squamous cell carcinoma--022Technically unsatisfactory18191852---- * Cellular appearance was within normal limits. The 112 negative smears on PAPNET screening had been classified on manual screening as atypical squamous cells of uncertain significance (14 smears) and low-grade squamous intraepithelial lesions (98 smears). Back to text 4: Histological follow-up of PAPNET cytopredictions. Histological follow-up PAPNET-detected abnormalitiesNegative/inflammation*Low-grade squamous intraepithelial lesionsHigh-grade squamous intraepithelial lesionsFollow-up to comeLost to follow-upTotalAtypical squamous cells of unknown significance66371032Low-grade squamous intraepithelial lesions21544631High-grade squamous intraepithelial lesions 01150117 *Cellular appearance was within normal limits. Back to text

Annabelle Farnsworth · Fay M Chambers · Colin S Goldschmidt

Research Urinary urge incontinence: randomised crossover trials of penthienate versus placebo and propantheline Graham M Coombes and Richard J Millard MJA 1996; 165: 473-476 Abstract - Introduction - Methods - Results - Discussion - Acknowledgements - References - Authors' details - - More articles on Obstetrics & gynaecology and women's health Abstract Objective: To compare the efficacy of penthienate with that of propantheline and placebo for treatment of primary idiopathic detrusor instability. Design: Two prospective, randomised, crossover trials (double-blind for penthienate versus placebo and non-blinded for penthienate versus propantheline). Setting: Urology Clinic of Prince Henry Hospital, Sydney, NSW (an outpatient clinic of a tertiary referral hospital), in 1993-1994. Participants: Neurologically intact patients with urodynamically proven detrusor instability, urgency and urge incontinence, but no stress incontinence (20 participated in the penthienate/placebo trial and 23 in the penthienate/propantheline trial). Outcome measures: Cystometrography results before and after treatment; frequency and volumes of urine voided in weeks 1 and 4 of treatment; and patient scores for degree of continence, side effects, efficacy and acceptability of treatment. Interventions: Penthienate (5 mg), propantheline (15 mg) or placebo (all three times a day) for 4 weeks. Results: Penthienate produced significantly greater improvements than placebo in frequency (daytime, P = 0.002; and night-time, P = 0.02), incontinence scores (P = 0.002) and amplitude of unstable detrusor contractions, when present (P = 0.01), and significantly increased diurnal and nocturnal bladder capacity, both on cystometrography (P = 0.003) and by voiding-diary records (P < 0.001). It also increased residual urine volume over the baseline level, but not significantly. Side effects, especially dry mouth, were common with penthienate, and one patient developed urinary retention. Penthienate was significantly better than propantheline in improving cystometric capacity (P = 0.03), and reducing the amplitude of unstable detrusor contractions (P = 0.01), and was perceived as more effective by patients for frequency, nocturia and incontinence. Conclusions: Penthienate (5 mg three times a day) was objectively and subjectively significantly better than both placebo and propantheline (15 mg three times a day) for treatment of primary idiopathic detrusor instability. Introduction Detrusor instability is a common urologic problem, causing symptoms of urgency, frequency and urge incontinence. It is defined by the International Continence Society as the presence of detrusor contractions with pressure greater than 15 centimetres of water during the filling phase of cystometrography while the patient is trying to inhibit micturition.1 Although detrusor instability may result from neuropathy (e.g., after spinal injury and in multiple sclerosis, where it is best termed detrusor hyperreflexia), in most patients it is primary and idiopathic.2Many treatments have been used, including bladder training, biofeedback, anticholinergics, analgesics, muscle relaxants, tricyclic antidepressants and acupuncture.2-7 More invasive, surgical procedures, such as subtrigonal phenol injections, neuromodulation and bladder augmentation, are less commonly required. Success rates are variable, and depend partly on the cause of the instability, but are usually no higher than 70%.8,9 The anticholinergic propantheline bromide is used worldwide for detrusor instability and, along with oxybutynin, is among the most often studied anticholinergic agents. However, penthienate bromide, which was originally marketed in Australia for gastrointestinal spasm, has been in clinical use for detrusor instability at Prince Henry Hospital, Sydney, NSW, and, in our clinical experience, may be more effective than propantheline. It is a quaternary ammonium antimuscarinic that inhibits parasympathetic, postganglionic bladder smooth muscle receptors, has a short half-life and is optimally absorbed from the gastrointestinal tract on an empty stomach. However, its comparative efficacy has not been established. We therefore undertook randomised trials of penthienate versus placebo and penthienate versus propantheline in primary idiopathic detrusor instability. Methods Approval for the trials was given by the Eastern Sydney Area Health Service Ethics Committee. The trials were independent of drug company participation or sponsorship, apart from the manufacture and supply of placebo and penthienate tablets (for the first trial). Penthienate versus placebo The first trial was a prospective, randomised, double-blind crossover trial of penthienate versus placebo. Participants were 20 neurologically intact patients recruited from the outpatient Urology Clinic and Urodynamic Service of Prince Henry Hospital, Sydney, NSW. Inclusion criteria. These were: Age over 18 years; Presence of symptomatic urgency or urge incontinence; Detrusor instability (shown on medium-fill, room-temperature, videourodynamic studies); No history of neurologic disease, narrow-angle glaucoma or severe cardiac disease (American Heart Association Grade 3 or 4); No stress incontinence, bladder-outlet obstruction, significant postvoid residual urine (residuum greater than 10% of voided volume), current urinary tract infection or interstitial cystitis; No current gastrointestinal obstruction; No allergy or sensitivity to anticholinergic agents; Not pregnant; and Patient gave informed consent to active and placebo treatment. First phase. The trial began with a one-week run-in period in which patients kept charts of frequency and volume of urine voided. Informed consent was then obtained and patients were randomised in a double-blind fashion to placebo or penthienate. Codes were generated from randomisation tables in blocks of ten, and provided in sealed envelopes by the Pharmacy Department; neither patients nor investigators knew the code until after collection of all follow-up data. Patients received either placebo or penthienate (5 mg three times a day before meals) for four weeks. Dosage reduction was ethically necessary if the patient developed blurred vision, severe constipation, or excessive mouth dryness that interfered with nutrition. Placebo and penthienate tablets were made to the same size, shape and colour specifications by the manufacturer and stored in the refrigerator. Assessment. All patients were asked to keep records of frequency and volume of urine voided for two 72-hour periods in the first and fourth weeks of treatment. A visual analogue score for continence was also kept (from 1 [nil leakage], to 5 [wet all the time]) and episodes of incontinence were noted. Apart from keeping voiding diaries at the set times, patients were taught no formal bladder- training techniques. In the final (fourth) week of treatment, patients attended the outpatient clinic for assessment. History was taken and time/volume charts collected. Medium-fill room-temperature saline-filling cystometrography was performed by the urology registrar, and residual urine volume, flow rate, cystometric capacity and incidence and amplitude of any unstable detrusor contractions were recorded. Urodynamic methods, definitions and units conformed to the standards recommended by the International Continence Society.1In addition, patients scored side effects, efficacy of treatment and willingness to have that treatment in the future on a scale of 1-100. Second phase. After a five-day washout period, patients crossed to the other arm of the trial. As it was not considered ethical to subject patients to a further urodynamic study during this period (to prove stable disease course), treatment-period interaction could not be totally excluded. Again, treatment was for four weeks, with the same assessment as previously. Penthienate versus propantheline The second trial was an unblinded, prospective, randomised crossover trial of penthienate and propantheline in 23 patients with detrusor instability. Inclusion criteria, run-in period and randomisation were the same as those in the first trial, except that neither patients nor observers were blinded to the treatment. Patients received either penthienate (5 mg three times a day) or propantheline (15 mg three times a day) for four weeks. There was no placebo arm. All patients were then assessed as outpatients, with the same outcome measures as those in the first trial. After a five-day drug washout period, patients crossed over to receive the other medication for a further four weeks before reassessment. Statistical analyses Analysis of paired data was by the Wilcoxon matched-pairs signed-rank test. Unpaired group data were analysed with the Mann-Whitney U test. Results Penthienate versus placebo Patients comprised 16 women and four men, with median age 63.5 years (range, 36-85). Urological baseline characteristics are shown in Box 1. There were no significant differences in baseline data between those patients receiving placebo first and those receiving penthienate first. To investigate the possibility of a carryover effect of penthienate in patients who received placebo second, the data were analysed initially in groups according to which drug had been given first. In the placebo arm of the trial there were no significant differences in cystometric capacity, residual postvoid volume, amplitude of unstable detrusor contractions, flow rate, frequency, nocturia or subjective incontinence scores between patients who took placebo first and those who took it second. Thus, five days was adequate for washout of penthienate. Thereafter, data were analysed without regard to which drug was taken first. Continence and urodynamic characteristics after penthienate and placebo treatment are shown in Box 1. After penthienate treatment, there were increases over the baseline in daytime voiding interval, mean voided volume (daytime and night-time) and cystometric capacity, and decreases in night-time frequency and subjective incontinence score. Values for all these parameters were significantly "better" after penthienate than after placebo. The decrease in median incontinence score with penthienate was dramatic, from a baseline of 5 ("wet all the time") to 1 ("nil leakage"), compared with 4 for placebo. Ten of 20 patients taking penthienate became dry (incontinence score of 1, 50% cure rate), compared with only two patients taking placebo. The same 10 patients taking penthienate were also completely stable (detrusor pressure < 15 cm H2O) to a capacity of 500 mL on cystometrography. Among the other 10, the median amplitude of unstable detrusor contractions was reduced from 40 to 27 cm H2O. However, with placebo only four patients (20%) tested stable and the other 16 had unstable detrusor contractions, with a median pressure of 30 cm H2O. The difference between the penthienate and placebo groups in detrusor pressures in those patients who remained unstable was significant (P = 0.05). Residual urine volume increased from the baseline level in the penthienate group and decreased in the placebo group, but neither change was significant. However, the resulting difference between penthienate and placebo residual volumes was significant (P = 0.02). Urine flow rate increased significantly in both the penthienate and placebo group over the baseline level (P = 0.01), but did not differ significantly between the two groups. The increase in flow rate appeared to be related to the increase in voided volume, as there was no change in percentile performance on the Haylen Liverpool flow nomogram.10 Side effects of treatment were common with penthienate (Box 2). All patients mentioned a dry mouth, and in one elderly patient this led to difficulty swallowing, which necessitated reducing the penthienate dose. Although 11 of 20 patients taking placebo also mentioned mouth dryness, the scores for degree were significantly less than for penthienate. A patient taking penthienate complained of diarrhoea, but there was no evidence of faecal impaction to suggest spurious diarrhoea caused by decreased gastrointestinal motility. Another patient taking penthienate in the second phase of the trial developed chronic urinary retention (with a residual postvoid volume of 550 mL). This patient's results were excluded from the analysis. Patients' subjective impressions of treatment are also shown in Box 2. There was a significant preference for penthienate compared with placebo in both perceived efficacy and willingness for retreatment (P = 0.002). Penthienate was considered beneficial by 17 (85%), but placebo by only five (25%). Two patients preferred placebo to penthienate; their cystometric capacities and filling pressures on testing differed little between penthienate and placebo, but their voiding diaries showed greater daytime and night-time voided volumes during the penthienate arm. Penthienate versus propantheline Patients comprised 18 women and five men, with median age 63 years (range, 34-85). Propantheline was given first to 12 patients and penthienate to 11. Baseline characteristics are compared with those after propantheline and penthienate treatment in Box 3. Both penthienate and propantheline significantly increased daytime voiding interval, cystometric capacity and flow rate over baseline figures. However, the improvements in cystometric capacity were significantly greater with penthienate than with propantheline. Furthermore, penthienate, but not propantheline, significantly increased voided volume (both daytime and night-time) and decreased night-time frequency and the incidence of incontinence (by patient score). Consequently, daytime voided volume and incontinence score were significantly "better" for penthienate than for propantheline. The difference in incontinence score was dramatic -- 1.5 after penthienate (where 1 = "nil leakage"), but 5 ("wet all the time") after propantheline. Stable cystometrograms were seen in 13 of the 23 patients (56%) after penthienate, but in only 10 (43%) after propantheline. In the patients with residual unstable detrusor contractions, the amplitude of these contractions decreased with penthienate but increased with propantheline, which led to a significant difference between the two treatments (P = 0.01). Although residual urine volume increased over the baseline value with penthienate, the change was not significant. Similarly, although detrusor compliance (detrusor pressure at a given bladder volume, which measures bladder-wall stiffness) increased after penthienate but decreased after propantheline, none of the changes were significant and did not lead to significant differences between the two drug treatments. Side effects and subjective response to treatment are shown in Box 4. Mouth dryness was significantly worse with penthienate, but patients believed that it had significantly greater efficacy and were more willing to repeat penthienate treatment. Discussion Our trials showed a significantly greater response to penthienate than to placebo or propantheline treatment, both objectively and subjectively, in primary idiopathic detrusor instability. For drug therapy to be useful in this condition, it needs to be more effective than placebo, which can produce significant improvement11 (up to 43%, or as high as that of some active drug therapies).12 In addition, results should be at least as good as those that can be achieved with non-invasive therapies such as bladder training, without intolerable side effects. The Urology Clinic at Prince Henry Hospital has previously shown that bladder training and/or biofeedback gives up to 74% cure or improvement in symptoms of detrusor instability.5 In this study, we found that after penthienate there were objective improvements in bladder capacity, stability and amplitude of unstable detrusor contractions and 50%-56% of patients became objectively dry. There were also improvements in subjective criteria for continence after penthienate, with 85% feeling subjectively cured or improved. These subjective criteria were included to assess clinical worth from the patients' perspective, as statistical significance may not necessarily equate with clinical significance.4There have been no previous comparative trials of the relative efficacy of penthienate and propantheline for detrusor instability. We found that, in the doses studied (manufacturer's recommended starting doses), both produced some urodynamic and symptomatic improvements, but that penthienate was significantly better than propantheline in improving cystometric capacity and reducing the amplitude of unstable detrusor contractions, and was perceived as more effective by patients for frequency, nocturia and incontinence. Most recent studies on propantheline in detrusor instability have compared it with another anticholinergic, oxybutynin, which has been found to produce response rates of up to 60% and a median increase in cystometric capacity of 104 mL;13 it was more effective than propantheline in some studies,12,14 but not others.15 Similarly as for penthienate, the price of its therapeutic effect was a higher residual urine volume and higher rate of side effects.12 However, as we found for penthienate, the anticholinergic side effects were generally tolerable and, if the treatment was discontinued, short-lived. Nevertheless, in detrusor hyperreflexia caused by multiple sclerosis oxybutynin and propantheline treatment had to be discontinued because of side effects in 22% and 27% of patients, respectively.15 We conclude that penthienate bromide has a significant place in the management of detrusor overactivity. It has been our practice to use penthienate as first-line drug treatment in those patients with idiopathic detrusor instability not responsive to conservative measures such as bladder training and biofeedback. Unfortunately, it is currently unavailable in Australia as it is no longer manufactured locally. The side effects of anticholinergic treatment are well recognised and occurred with both penthienate and propantheline. With appropriate modifications of dosage (e.g., in the elderly, who are more likely to have side effects, and in those with neurogenic bladders and detrusor hyperreflexia caused by multiple sclerosis, who are more sensitive to penthienate), most patients can tolerate these side effects and still benefit from therapy. Treatment should start with the lowest therapeutic dose and be monitored closely to maximise patient compliance if long term use is considered. Acknowledgements Winthrop Laboratories manufactured and supplied the penthienate (Monodral) and placebo tablets for the first crossover trial. We thank Dorothy Wheeler, RN, and Marilynne Berg, RN, of the Urology Training Unit, Prince Henry Hospital, for their help in the follow-up studies. References Abrams P, Blaivas JG, Stanton SL, Andersen JT. The standardisation of terminology of lower urinary tract function. Scand J Urol Nephrol 1988; Suppl 114: 5-19. Tapp AJS, Cardozo LD, Versi E, Cooper D. The treatment of detrusor instability in controlled study postmenopausal women with oxybutynin chloride: a double blind placebo. Br J Obstet Gynaecol 1990; 97: 521-526. Blaivas JG, Labib KB, Michalik SK, Zayed AAH. Cystometric response to propantheline in detrusor hyperreflexia: therapeutic implications. J Urol 1980; 124: 259. Wein AJ. Pharmacologic treatment of incontinence. J Am Geriatr Soc 1990; 38: 317-328. Millard RJ, Oldenburg BF. The symptomatic, urodynamic and psychodynamic results of bladder re-education programs. J Urol 1983; 53: 565-566. Chang PL. Urodynamic studies in acupuncture for women with frequency, urgency and dysuria. J Urol 1988; 140: 563-566. Burgio KL, Engel BT. Biofeedback-assisted behavioural training for elderly men and women. J Am Geriatr Soc 1990; 38: 338-340. Koldewijn EL, Rijkhoff NJM, van Kerrebroeck PE, et al. Selective stimulation of sacral nerves for bladder control: a computer model. Neurourol Urodyn 1992; 11: 328-330. Mundy AR. Detrusor instability. Br J Urol 1988; 62: 393-397. Haylen BT, Ashby D, Sutherst JR, et al. Maximum and average urine flow rates in normal male and female populations -- the Liverpool nomograms. Br J Urol 1989; 64: 30-38. Benson H, Epstein MD. The placebo effect. JAMA 1975; 232: 1225. Thuroff JW, Bunke B, Ebner A, et al. Randomized, double blind, multicenter trial on treatment of frequency, urgency and incontinence related to detrusor hyperactivity: oxybutynin versus propantheline versus placebo. J Urol 1991; 145: 813-817. Moore KH, Hay DM, Imrie AE, et al. Oxybutynin Hydrochloride (3 mg) in the treatment of women with idiopathic detrusor instability. Br J Urol 1990; 66: 479-485. Gajewski JB, Awad SA. Oxybutynin versus propantheline in patients with multiple sclerosis and detrusor hyperreflexia. J Urol 1986; 135: 966-968. Holmes DM, Montz FJ, Stanton SL. Oxybutynin versus propantheline in the management of detrusor instability. A patient-regulated variable dose trial. Br J Obstet Gynaecol 1989; 96: 607-612. (Received 22 Jun, accepted 20 Jul 1996) Authors' details Department of Urology, Prince Henry Hospital, Sydney, NSW. Graham M Coombes, FRACS, Urology Registrar. Currently, Royal North Shore Hospital, Sydney, NSW; Richard J Millard, FRACS, FRCS, Associate Professor of Urology, University of NSW, Sydney, NSW. Reprints: Dr G M Coombes, 74 Osborne Road, Lane Cove, NSW 2066. 1: Continence and urodynamic characteristics before and after treatment with penthienate and placebo (median and range)*Baseline (n = 20)Penthienate (n = 19)†Placebo (n = 19)†P‡Daytime voiding interval (hours)2 (1-5)3 (2-6)2 (1-4)0.002Night-time frequency (times)2.5 (1-6)1 (0-5)2 (0-4)0.02Mean daytime voided volume (mL)200 (75-400)270 (125-400)168 (50-300)0.0002Mean night-time voided volume (mL)250 (50-650)300 (195-740)275 (50-700)0.02Incontinence score§5 (2-5)1 (1-5)4 (1-5)0.002Urodynamic testingCystometric capacity (mL)283 (150-450)500 (225-600)400 (200-500)0.003Amplitude of unstable detrusor contractions (cm H2O)¶40 (16-80)27 (16-40)30 (20-45)0.05Residual postvoid volume (mL)30 (0-160)40 (0-150)15 (5-50)0.02Flow rate (mL/s)14 (6-25)17 (10-30)20 (15-45)0.68Number of patients With stable cystometrogram**0104NA"Dry" (incontinence score of 1)0102NANA=not assessed. *Unless otherwise stated. †One patient was excluded because of chronic urinary retention. ‡ For difference between penthienate and placebo. § On analogue scale, with range 1 ("nil leakage") to 5 ("wet all the time"). ¶ n = 10 for penthienate and n = 16 for placebo, after exclusion of subjects with stable cystometrogram. ** Detrusor pressure < 15 cm H2O on filling cystometrography to a capacity of 500 mL. Back to text2: Side effects and subjective patient impressions of penthienate and placebo PenthienatePlaceboDry mouthNumber of patients20 (100%)11 (55%)Mean score (range)*65 (25-95)25 (0-80)Nasal dryness2 (10%)0Constipation/lower abdominal pain1 (5%)0Upper abdominal pain1 (5%)0Blurred vision1 (5%)0Chronic urinary retention1 (5%)0Mean subjective efficacy (range)*80 (10-95)23 (0-95)Number of patients Improved17 (85%)5 (25%) Willing to repeat treatment12 (60%)5 (25%)* Subjective score on scale of 0-100. Patient had diagnosed hiatus hernia, but no documented peptic ulcer or reflux oesophagitis. Back to text3: Continence and urodynamic characteristics before and after treatment with penthienate and propantheline (median and range)*Baseline (n = 23)Penthienate (n = 23)Propantheline (n = 23)P †Daytime voiding interval (hours)2 (1-5)3 (2-6)3 (2-4)0.53Night-time frequency (times)2 (1-6)1 (0-5)2 (1-6)0.07Mean daytime voided volume (mL)200 (75-400)260 (125-400)200 (125-450)0.001Mean night-time voided volume (mL)250 (50-650)300 (195-740)325 (230-550)0.17Incontinence score‡5 (2-5)1.5 (1-5)5 (1-5)0.001Urodynamic testingCystometric capacity (mL)280 (150-450)500 (225-600)470 (275-550)0.03Amplitude of unstable detrusor contractions (cm H2O)40 (15-80)27 (16-40)50 (20-75)0.01Residual postvoid volume (mL)25 (0-160)37 (0-150)17.5 (0-100)0.1Flow rate (mL/s)15 (6-29)20 (10-50)20 (15-35)0.48Compliance (mL/cmH2O)70 (12-210)100 (23-600)55 (14-166)0.08Number with stable cystometrogram§01310NANA=not assessed. * Unless otherwise stated. † For difference between penthienate and propantheline. ‡ On analogue scale, with range 1 ("nil leakage") to 5 ("wet all the time"). § Detrusor pressure < 15 cm H2O on filling cystometrography to a capacity of 500 mL. Back to text 4: Side effects and subjective patient impressions of penthienate and propanthelinePenthienate PropanthelineP*Mean score for subjective efficacy† (range)83 (10-95)38 (5-90)0.002Mean score for mouth dryness† (range)73 (25-45)55 (30-90)0.05Mean score for constipation† (range)5 (0-50)7 (0-75)0.73Number of patients "dry" (incontinence score, 1)1310 NANumber of patients improved1913NANumber of patients willing to repeat treatment1211 NANA=not assessed. * By Wilcoxon matched-pairs signed-rank test. † Subjective score on scale of 0-100. Back to text

Graham M Coombes · Richard J Millard

Research Outcomes after hip or knee replacement surgery for osteoarthritis A prospective cohort study comparing patients' quality of life before and after surgery with age-related population norms Lyn M March, Marita J Cross, Helen Lapsley, Alan J M Brnabic Katherine L Tribe, Clarissa J M Bachmeier, Brett G Courtenay and Peter M Brooks* MJA 1999; 171: 235-238 For editorial comment, see Williamson Abstract - Introduction - Methods - Results - Discussion - Acknowledgements - References - Authors' details - - More articles on Orthopaedic surgery Abstract Objective: To compare the health-related quality of life of people with osteoarthritis before and after primary total hip and knee replacement surgery with that of the general Australian population. Design: A prospective cohort study. Setting: Three Sydney hospitals, public and private. Participants: Patients with osteoarthritis undergoing primary total hip (n = 59) and knee (n = 92) joint replacement surgery. Main outcome measure: Medical Outcomes Study Short Form (SF-36) scores before and 12 months after joint replacement surgery (compared with population norms). Results: Patients in each age group showed a significant improvement in health-related quality of life after joint replacement surgery in most scales of the SF-36, particularly physical function, role physical and bodily pain. SF-36 scores for the 42 hip-replacement patients aged 55-74 years improved to equal or exceed the population norm on all scales. SF-36 scores of the 52 knee replacement patients aged 55-74 years improved, but physical function and bodily pain scores remained significantly worse than the population norm. SF-36 scores for both hip (n = 17) and knee (n = 40) replacement patients aged 75 years and over improved significantly, becoming similar to population norms for this age group. Conclusions: Total hip or knee replacement for osteoarthritis significantly improves patient health and well-being at 12 months after surgery. Age alone should not be a barrier to surgery. Introduction Osteoarthritis is one of the leading causes of pain and disability in the Australian population.1,2 Total joint replacement is the most common treatment for advanced osteoarthritis of the hip or knee, with the primary goal of the procedure being to improve the patient's quality of life.3In Australia, in the financial year 1997/98, Medicare benefits of $13 500 000 were paid for 17 000 hip and knee replacements. This number represents only some of the total surgery performed, as the number of operations on Veterans' Affairs patients and on public patients in public hospitals are not available. Despite joint replacement surgery being one of the most common operations performed in this country, very limited Australian outcomes data have been published.4 The Medical Outcomes Study Short-Form 36 (SF-36) has been used extensively to assess the effect on quality of life of several procedures, including total joint replacement,5,6 and it has been suggested that it should be the focus of preoperative and postoperative outcome evaluation for total hip arthroplasty.7 As a generic measure of health-related quality of life with standardised scoring, it enables comparisons between diseases, treatments and published population norms. It measures health on eight 100-point scales: physical function, role physical (ie, role limitations due to physical problems), bodily pain, general health, vitality, social function, role emotional (ie, role limitations due to emotional problems), and mental health. It is one of the most widely used health-related quality-of-life instruments and has been shown to be reliable and valid. Studies in the US have shown that the SF-36 detects a significant improvement in health-related quality of life in patients undergoing total hip or knee replacement when preoperative scores are compared with postoperative scores.8-11 The SF-36 also demonstrates a difference in outcome between total hip and total knee replacement patients,8,10 with patients undergoing total hip replacement achieving significantly better outcomes than the knee replacement patients. We aimed to compare the health-related quality of life (as measured by the SF-36) of people with osteoarthritis undergoing primary total hip and knee replacement surgery with that of the normal Australian population.2 Putting changes in the health-related quality of life of these patients in the context of population norms for the same age group gives us a meaningful measure of the effectiveness of this surgery. It also provides information in a way that allows comparisons with other diseases and treatments, assisting rational choices to be made about the use of the limited healthcare dollar. Methods The information on the patients undergoing total joint replacement collected for this analysis was part of a long-term follow-up of patients in a cohort study aiming to assess the costs of arthritis and the effectiveness of its treatment. The cohort includes patients from both public and private hospitals and surgeons operating in both sectors are involved. Patients with osteoarthritis booked for primary total hip or knee replacement surgery at three Sydney hospitals (St Vincent's public and private hospitals and the Centre for Bone and Joint Diseases, North Ryde) between March 1994 and December 1995 were approached to participate in the study. Recruitment was through regular contact with the orthopaedic surgeons and their practice staff. Baseline information was collected from patients through a series of questionnaires, including the SF-36. Questionnaires were administered between one week and three months before surgery. After surgery, patients completed the SF-36 questionnaire at the end of each three months for their first postoperative year. Most questionnaires were self-administered, with less than 10% requiring face-to-face interview. This is consistent with the Australian Bureau of Statistics methodology for collecting general population data.2 Annual follow-up of these patients by mail is continuing. This article presents SF-36 information from patients at baseline and 12 months after surgery. Ethics committee and Medical Board approval was obtained from the St Vincent's campus of the University of New South Wales Medical School, the Centre for Bone and Joint Diseases and the Royal North Shore Hospital. All patients in the study gave written informed consent. Analysis Mean scores for each of the SF-36 dimensions were calculated for patients in the age groups 55-64, 65-74 and 75 years and over. Scores for men and women were combined, as, apart from older males undergoing joint replacement having lower preoperative scores for role emotional, there was no significant difference in scores between the sexes. Scores for each dimension were transformed according to the SF-36 user's guide12 to a scale of 0-100 (100 = best possible score). We used one-sample t tests to compare the transformed scores with general population norms derived by the Australian Bureau of Statistics from the 1995 National Health Survey. The t test was chosen on the assumption that the population value was the real value (given that published standard errors were so low) and that we were observing the variance of the study data from the population value. A P value of 0.05 was considered to be significant and no adjustment was made for multiple comparisons. SF-36 on the web More information about the SF-36 health survey can be found on the SF-36 website: http://www.sf-36.com An online demonstration of the SF-36 survey (score yourself and read an explanation of the result in comparison with US population norms) is available at: http://www.qmetric.com/demo/sf-36v1.shtml Results Two-thirds of the eligible patients were recruited (226 of 343). Reasons for exclusion included being unable to be contacted before surgery (50% of those who did not participate in the study), being non-English-speaking (10%) and refusal or inability to complete the questionnaires (40%). Eligible patients who did not participate in the study did not differ significantly from the cohort in terms of age, sex or type of joint replacement (ie, hip or knee) (data not shown). At the time of analysis, we had complete information for 151 patients at 12 months' follow-up (59 with hip replacement and 92 knee replacement; 52% female; median age, 72 years). An additional 75 patients (60% female; median age, 74 years) had incomplete follow-up information. Their last recorded SF-36 scores showed no significant differences from scores at the same stage of follow-up among those with complete information (data not shown). Patients' SF-36 scores before and 12 months after surgery are shown in the Figure, in comparison with scores for the general population. Knee replacements 55-64 years age group: Eight patients undergoing knee replacement were recruited. Six of these patients reported having other illnesses, most commonly cardiovascular disease (reported by five patients). At 12 months' follow-up, these eight patients showed improvement from baseline according to mean SF-36 scores for physical function, role physical, bodily pain and vitality, but these scores remained significantly lower than the population norms. The power to detect the observed difference to be statistically significant at the 5% level ranged from 83% to 100% for these four comparisons. Their scores on the other four scales (general health, social function, role emotional, mental health) remained lower than population norms, but this difference was not statistically significant (power to detect a significant difference ranged from 7% to 55%). 65-74 years age group: Twenty-eight of the 44 patients (64%) reported having a comorbid illness, with the most commonly reported again being cardiovascular disease (61% of those with comorbidity). Mean SF-36 scores improved significantly on all scales except general health. The mean general health score was significantly higher than the population norm at 12 months' follow-up, but had been higher to begin with at baseline. Mean physical function, role physical and bodily pain scores remained significantly lower than the population norms at 12 months' follow-up. The power to detect these differences ranged from 85% to 99%. Mean vitality, social function, role emotional and mental health scores improved from baseline to be similar to population norms. The power to detect the significance of these differences ranged from 8% to 36%. 75 years and over: Thirty-two of the 40 patients (80%) reported suffering from another illness, most commonly cardiovascular disease (60% of those with comorbidities). Mean SF-36 scores improved significantly on most scales (particularly role physical, physical function and bodily pain), but not on general health and mental health, which were fairly high before surgery. The mean general health score was significantly higher than the population norm at 12 months' follow-up; the other scores improved to population levels. Given the small difference seen between patient and population scores, large numbers would be required to show statistically significant differences (power ranged from 7% to 33%). Hip replacements 55-64 years age group: Comorbidities were reported by 5 of the 14 patients (36%). All five had cardiovascular disease among other comorbidities. Mean scores for vitality and social function improved to become significantly higher than the population norms (power to detect significant difference, 90% and 96%, respectively), while the other scores improved to be similar to the population norms, with particular improvement in physical function, role physical and bodily pain. 65-74 years age group: Thirteen of the 28 patients reported suffering from a comorbid illness, most commonly cardiovascular disease (46% of patients with comorbidities). Mean SF-36 scores improved on all scales except general health, which had a baseline value above the population norm. Mean scores for general health and mental health were significantly higher than the population norms at 12 months' follow-up (power to detect difference, 85% and 75%, respectively) and the other scores improved to be similar to the population norms (power to detect difference ranged from 7% to 58%). The improvement from baseline was particularly evident in role physical, physical function and bodily pain scores. 75 years and over: Comorbidities were reported by 12 of the 17 patients (71%), most commonly cardiovascular disease (reported by 58% of those with comorbidities). Mean SF-36 scores improved on all scales except general health and mental health. At 12 months' follow-up, mean scores on none of the scales were significantly different from the population norms, possibly due to the small numbers in this group (power to detect a difference less than 30% for most scales). Again, there was improvement to population norms in role physical, physical function and bodily pain. Discussion We found that knee or hip replacement surgery significantly improved the health-related quality of life of patients with osteoarthritis. Before surgery, the patients had poor SF-36 scores for bodily pain, physical function and role physical, a clear reflection of the impact of chronic osteoarthritis on health-related quality of life. After surgery, the biggest improvements were in these scores. Improvement in these physical dimensions of health is likely to lead to improvements in social function, mental health and vitality. We found no improvement in general health despite gains in the other scales. This has been shown in other studies,11 and has been suggested to be due to the patients' previous medical history and pre-existing and general health-related conditions. In our study it was apparent that the general health of patients was already higher at baseline than the population norm, suggesting that relatively healthy patients are being selected for total joint replacement. A US study that compared patients undergoing total hip replacement with age-matched and sex-matched population norms found that age and sex made important differences in SF-36 scores.9 Men younger than 65 years had scores lower than the norms in the physical scales, but were comparable in the mental scales, and women scored lower than the norm in all scales. No sex differences were found in our study, but similar age differences were observed, particularly for patients undergoing total knee replacement, among whom the youngest age group had the poorest outcome relative to the age-matched population. We hypothesise that this may be related to the higher population norms for this age group, unrealistic expectations for outcomes among patients, and the presence of existing comorbidities in younger patients requiring knee replacement. One implication of these findings is that older age should not be a barrier to joint replacement surgery, as the outcome is likely to be relatively successful for older patients. Potential limitations that need to be considered when interpreting these results include the non-randomised cohort design, the power of the analysis given the small sample size of the subgroups, and the overall response rate. However, the inclusion of several groups of surgeons operating at different sites and across the public and private sectors contributed significantly to the generalisability of the results. This study showed the improvement of SF-36 scores of patients undergoing total hip or knee replacement up to and in some cases beyond population norms. Whether these statistically significant differences are clinically important remains to be seen. Longer follow-up is required to determine whether the dramatic improvement from baseline is maintained. Nonetheless, at one year follow-up, it would appear that total knee or hip joint replacements are successful at restoring health and well-being, and that older age alone should not be a barrier to surgery. Acknowledgements This study was funded by grants from the National Health and Medical Research Council. We are grateful to the secretarial staff of the orthopaedic surgeons for their assistance with recruitment to this study and to the patients who have been so cooperative. References Australian Bureau of Statistics. Disability and handicap, Australia 1988. Canberra: ABS, 1996 (Catalogue No. 4120.0). Australian Bureau of Statistics National Health Survey: SF-36 population norms, Australia, 1995. Canberra: ABS, 1997 (Catalogue No. 4399.0). Bombardier C, Melfi CA, Paul J, et al. Comparison of a generic and a disease-specific measure of pain and physical function after knee replacement surgery. Med Care 1995; 33 Suppl 4: AS131-AS144. Van Essen GJ, Chipchase LS, O'Connor D, Krishnan J. Primary total knee replacement: short-term outcomes in an Australian population. J Quality Clin Practice 1998; 18: 135-142. Bayley KB, London MR, Grunkemeier GL, Lansky DJ, Measuring the success of treatment in patient terms. Med Care 1995; 33 Suppl 4: AS226-AS235. Stucki G, Liang MH, Phillips C, Katz JN. The Short-Form 36 is preferable to the SIP as a generic health status measure in patients undergoing elective total hip arthroplasty. Arthritis Care Res 1995; 8: 174-181. Ritter MA, Albohm MJ, Overview: maintaining outcomes for total hip arthroplasty. The past, present and future. Clin Orthop 1997; 344: 81-87. Hozack J, Rothman RH, Albert TJ, et al. Relationship of total hip arthroplasty outcomes to other orthopaedic procedures. Clinical Orthop 1997; 344: 88-93. Lieberman JR, Dorey F, Shekelle P, et al. Outcome after total hip arthroplasty. Comparison of a traditional disease-specific and a quality of life measurement of outcome. J Arthroplasty 1997; 12: 639-645. Kiebzak GM, Vain PA, Gregory AM, et al. SF-36 general health status survey to determine patient satisfaction at short-term follow-up after total hip and knee arthroplasty. J Southern Orthop Assoc 1997; 6: 169-172. Ritter MA, Albohm MJ, Keating EM, et al. Comparative outcomes of total joint arthroplasty. J Arthroplasty 1995; 10: 737-741. Medical Outcomes Trust. How to score the SF-36 health survey. Boston: The Trust, 1994. (Received 16 Oct 1998, accepted 12 Jun 1999) Authors' details University of Sydney, Department of Rheumatology, Royal North Shore Hospital, Sydney, NSW. Lyn M March, MB BS, PhD, FRACP, FAFPHM, Associate Professor. Department of Medicine, University of New South Wales, Sydney, NSW. Marita J Cross, BSc(Hons), Research Assistant; Katherine L Tribe, BSc(Hons), Research Assistant; Clarissa J M Bachmeier, MD, MMed(ClinEpidem), Research Fellow. School of Health Services Management, University of New South Wales. Helen Lapsley, BA, MEc, Senior Lecturer. Northern Sydney Public Health Unit, Hornsby Ku-ring-gai Hospital, Sydney, NSW. Alan J M Brnabic, MSc, Statistician. Department of Orthopaedics, St Vincent's Hospital, Sydney, NSW. Brett G Courtenay, MB BS, FRACS, Orthopaedic Surgeon. Faculty of Health Sciences, University of Queensland, Brisbane, Qld. Peter M Brooks, MB BS, FRACP, FAFPHM, Executive Dean. No reprints will be available from the authors. Correspondence: Associate Professor L M March, Department of Rheumatology, Royal North Shore Hospital, St Leonards, NSW 2065. Email: lmarcATdoh.health.nsw.gov.au Click in figure for a larger version Back to text

Lyn M March · Marita J Cross · Helen Lapsley · Katherine L Tribe · Brett G Courtenay · Peter M Brooks

Research Impact of written information on knowledge and preferences for cardiopulmonary resuscitation Ian H Kerridge, Sallie-Anne Pearson, Isobel E Rolfe, Michael Lowe and John R McPhee MJA 1999; 171: 239-242 Abstract - Introduction - Methods - Results - Discussion - Acknowledgements - References - Authors' details - - More articles on Ethics Abstract Aim: To investigate knowledge about and attitudes to cardiopulmonary resuscitation (CPR), and to determine whether written information about CPR alters knowledge and choices made. Design: Questionnaire-based survey before and immediately after provision of written information describing CPR and its risks and benefits. Subjects and setting: All health professionals (803) and competent inpatients (260) in a tertiary care hospital (John Hunter Hospital, Newcastle, New South Wales, Australia) in June 1994. Main outcome measures: CPR knowledge scores and choice scores (number of hypothetical clinical scenarios in which CPR would be chosen) before and after provision of information about CPR. Results: Response rates were 64% (health professionals) and 58% (patients). Patients had limited awareness of procedures involved in CPR, while both patients and health professionals overestimated its success rates. Mean knowledge scores increased after provision of information: for patients, from 6.4 out of 18 (95% confidence interval [CI], 6.0-6.9) to 10.4 (95% CI, 9.9-11.1); and for health professionals, from 11.9 (95% CI, 11.7-12.1) to 13.9 (95% CI, 13.7-14.2). In contrast, mean choice scores decreased after provision of information: for patients, from 5.3 out of 12 (95% CI, 4.7-5.7) to 4.4 (95% CI, 3.9-4.8); and for health professionals, from 4.1 (95% CI, 3.9-4.2) to 3.5 (95% CI, 3.3-3.7). Conclusion: Our results imply that people understand and use prognostic information to make decisions about CPR. To make autonomous judgements, patients and health professionals need better education on CPR outcomes. Introduction While cardiopulmonary resuscitation (CPR) can be lifesaving, success rates (survival to discharge) are less than 5% in some types of patients, particularly those with chronic illness or multiple comorbidities.1 Among those who survive, quality of life is often poor and life expectancy often short.2 In the mid-1970s, growing concern about inappropriate application of CPR and increased awareness of patient rights led to the development of "do-not-resuscitate" (DNR) orders.3 Health professionals are now increasingly encouraged to discuss DNR decisions with patients and their families.4 However, for decision-making about CPR to be appropriate, patients and their surrogates must have some understanding of the likelihood of surviving CPR and the possible adverse effects. Previous research has found that 50%-80% of patients claimed awareness of CPR, but their knowledge was very limited and derived mainly from television dramas. Most patients believed erroneously that CPR is generally successful,5,6 and both patients and health professionals were found to overestimate its success by up to 300%.7 We aimed to investigate knowledge of and attitudes to CPR among hospital inpatients and health professionals and to determine whether written information about CPR alters knowledge and preference for CPR. We have previously reported the opinions of patients and health professionals on the process of decision-making about CPR.8 Methods Subjects and setting The study was conducted at the John Hunter Hospital, Newcastle, New South Wales (a 530-bed tertiary care hospital), over a four-day period in June 1994. Subjects comprised all health professionals working in the hospital (doctors, nurses and allied health professionals) and all eligible (competent) inpatients, as described previously.8 Informed consent was obtained by trained interviewers, and a Mini Mental State Examination (MMSE) performed on patients. Those with MMSE scores < 24 were excluded. A questionnaire for self-completion was administered before and immediately after provision of written information about CPR. Questionnaire The questionnaire asked about: sociodemographic characteristics; sources of information on CPR (respondents could nominate as many as applied from a list of 11); whether each of 10 procedures is part of CPR (possible answers: yes, no or don't know); how successful CPR is in eight clinical scenarios, using a five-point scale: rarely (< 5%), seldom (5%-30%), sometimes (31%-60%), mostly (61%-90%) or almost always (> 91%); and whether respondents would like CPR performed on themselves in 12 specific clinical scenarios (possible answers: yes or no). On completing the questionnaire, respondents were given a one-page information sheet that explained various aspects of CPR, including its definition, procedures that may or may not be considered part of CPR, risks and benefits, success rates, and prognostic indicators (good prognosis: after acute myocardial infarction; poor prognosis: cancer, severe infection, organ failure). Respondents were then asked to repeat the questions on knowledge and choices. Statistical analyses Data were analysed with SPSS version 6.0 for Windows.9 Sociodemographic characteristics of patients and health professionals were compared using continuity-corrected χ2 analyses. All other analyses were performed for patients and health professionals separately. Knowledge of CPR was scored by allocating a point for each correct response. We assessed the relationship between this score and sociodemographic characteristics by standard multiple linear regression, with knowledge score as the dependent variable. For this analysis, we converted the discrete sociodemographic variable (health professional background) into a set of dichotomous independent variables using dummy variable coding (ie, doctors versus all other health professionals, and nurses versus all other health professionals). Significance of change in knowledge scores after provision of information was examined by one-way repeated-measures analysis-of-variance (ANOVA). As answers to some of the knowledge questions may be contentious (whether CPR involves intravenous drugs, intubation and defibrillation), analyses were repeated using scores with these questions omitted. Composite "choice scores" about CPR were calculated by allocating a point for each condition in which the subject would choose to have CPR. The relationship between choice score, sociodemographic characteristics and knowledge score after provision of information was assessed by standard multiple linear regression, with choice score as the dependent variable and knowledge score and sociodemographic factors as independent variables. Significance of change in choice scores after provision of information was examined by ANOVA. Ethical approval The study was approved by the Hunter Area Health Service Research Ethics Committee and the University of Newcastle Human Research Ethics Committee. Results Subjects Of the 803 questionnaires delivered to health professionals, 511 (64%) were returned (148 from doctors, 312 from nurses and 51 from allied health professionals). Of the 443 adult patients in the hospital during the study, 183 were excluded (because of dementia, neurological impairment or delirium [110], incompetence as assessed by MMSE [35], visual or language problems [24], psychiatric illness [9], or as they were undergoing procedures [5]); 153 (58%) of the remaining 260 completed the questionnaire. Sociodemographic characteristics of respondents are shown in Box 1. Knowledge about CPR Sources of information about CPR most commonly identified by health professionals were first aid or in-service courses (33% of responses); school, college or university (27%); and other health professionals (17%). Sources most commonly identified by patients were television (28% of responses), books or magazines (15%), first aid classes (15%), and school (10%). Knowledge before provision of information is shown in Box 2. Patients' knowledge scores were low (mean, 6.4 out of 18; 95% confidence interval [CI], 6.0-6.9), mainly because of lack of awareness of success rates of CPR, with 90% overestimating success for the "all patients" category. Health professionals' knowledge scores were higher than patients' (mean, 11.9; 95% CI, 11.7-12.1), but health professionals also overestimated success of CPR, with 65% overestimating success for "all patients". ANOVA showed that knowledge scores improved significantly after provision of written information for both patients and health professionals (patients: mean, 10.4; 95% CI, 9.9-11.1; health professionals: mean, 13.9; 95% CI, 13.7-14.2). The only sociodemographic characteristics that significantly predicted knowledge scores before provision of information were age (for patients) and professional background (for health professionals): younger patients achieved higher scores than older patients, while doctors and nurses achieved higher scores than allied health professionals. Repeat analysis of the data after omission of contentious questions did not affect the relationship between sociodemographic variables and knowledge scores or the change in knowledge scores after provision of information. Choices about CPR Percentages of patients and health professionals who chose CPR are shown in Box 3. Percentages varied between clinical scenarios, but decreased after provision of information for almost all. ANOVA showed that choice scores also decreased significantly for both patients and health professionals. Mean choice scores decreased for patients from 5.3 out of 12 (95% CI, 4.7-5.7) to 4.4 (95% CI, 3.9-4.8), and for health professionals, from 4.1 (95% CI, 3.9-4.2) to 3.5 (95% CI, 3.3-3.7). CPR was chosen for more scenarios by patients who were younger or had lower knowledge scores and by health professionals who were younger, male, tertiary educated or had better self-reported health status. Discussion We found that patients in an Australian teaching hospital had poor knowledge of CPR, and that both they and, to a lesser extent, hospital staff had unrealistic expectations about its success rates. We also found that provision of written information about CPR risks, benefits and success rates had a clear impact on whether patients and health professionals reported wanting CPR performed on themselves in hypothetical clinical scenarios, decreasing their preference for CPR. Other studies have also found that patients are unaware of the procedures involved in CPR and, along with their relatives, generally overestimate success rates.10 This is not surprising, as television, books and magazines are often their most common sources of information. In addition, community education tends to be positive about CPR, often failing to describe its real success rates. Several previous studies have investigated the relationship between choices about CPR and estimated probability of survival.5,11 They found, similarly to us, that many patients change their minds about wanting CPR when they learn the true probability of survival. Furthermore, formal processes of information disclosure, including discussion of the likely outcome of resuscitation in specific clinical situations, has been shown to modify preferences for CPR.12 In our study, the change in preference followed provision of a one-page information sheet. It is possible that a more optimal form of education (eg, repeated explanations tailored to the individual, with time for reflection and questions) would have produced a greater change in preferences. Perceived morbidity after CPR also strongly influences many people's choices. A survey of 200 medical inpatients found that choice of CPR or DNR status was strongly influenced by anticipated outcome; 90% of patients desired CPR if they were to be restored to their normal health, 30% if the likely outcome after recovery was dependence, 15% if it was perceived as "hopeless", and 6% if it was coma.13 Other studies from the United States14 and United Kingdom15 have found that senile dementia, more than any other condition, is associated with a preference for DNR status. We found similarly that patients and health professionals would be unlikely to opt for CPR in the presence of brain injury or severe dementia. Nevertheless, as found by others,6,10 some patients continued to opt for CPR even if they were likely to have a serious disability, such as coma or terminal illness. We also found, in common with others overseas, that health professionals vastly overestimate the success rates of CPR. These rates (3%-30% in general hospitals) have not changed significantly in the past 30 years.1,16 However, they are well below the rates perceived by physicians and nurses, who are responsible for making decisions about resuscitation status and informing patients and their surrogates.17,18 There are several limitations to our study. Subjects included hospitalised patients with acute illness, and the findings may not be generalisable to other patient populations. The study instrument was a self-report questionnaire using hypothetical clinical scenarios; answers may not accurately reflect what individuals would choose in reality. However, the study has strengths; it assessed both knowledge of and attitudes to CPR in variable clinical contexts and included a formal assessment of competence (the MMSE). The study also used trained interviewers rather than clinicians to distribute questionnaires and so was less likely to introduce bias and perhaps more likely to elicit patients' true preferences. As factors such as likelihood of survival and functional status after resuscitation may contribute to patients' wishes for CPR, it is ethically and clinically desirable that patients be provided with this information. Furthermore, the High Court of Australia has reaffirmed that patients must be given adequate information when making decisions concerning their healthcare.19,20 It is also essential that health professionals who advise patients are aware of the real success rates of CPR. Otherwise, they have failed to meet the standards required both for ethical medical care and by law.21 Respect for autonomy demands that the views of patients or their surrogates should be sought in decision-making about CPR. Our study implies that patients understand and use prognostic information in their decision-making. For truly autonomous judgements, patients and health professionals clearly need better education on risks and benefits of CPR. Acknowledgements We would like to thank the interviewers and patients and staff of the John Hunter Hospital for their generous participation in this study. References Robinson GR, Hess D. Post-discharge survival and functional status following in-hospital cardiopulmonary resuscitation. Chest 1994; 105: 991-994. Landry FJ, Parker JM, Phillips YY. Outcome of cardiopulmonary resuscitation in the intensive care setting. Arch Intern Med 1992; 152: 2305-2308. Rabkin MT, Gillerman JD, Rice NR. Orders not to resuscitate. N Engl J Med 1976; 295: 364-366. Decisions relating to cardiopulmonary resuscitation. A statement from the British Medical Association and the Royal College of Nursing in association with the Resuscitation Council (UK). London: BMA, 1993. Miller DL, Jahnigen DW, Gorbien MJ, Simbarti L. Cardiopulmonary resuscitation: how useful? Attitudes and knowledge of an elderly population. Arch Intern Med 1992; 152: 578-582. Schonwefter RS, Walker RM, Kramer DR, Robinson BE. Resuscitation decision-making in the elderly: the value of outcome data. J Gen Intern Med 1993; 8: 295-300. Potter JM, Stewart D, Duncan G. Living wills: would sick people change their minds? Postgrad Med J 1994; 70: 818-820. Kerridge IH, Pearson SA, Rolfe IE, Lowe M. Decision making in CPR: attitudes of hospital patients and health care professionals. Med J Aust 1998; 169: 128-131. SPSS. Statistical package for the social sciences. Version 6.0 for Windows. Chicago, Ill: SPSS Inc, 1990. Schmerling RH, Bedell SE, Lilienfeld A, Delbanco TL. Discussing cardiopulmonary resuscitation: a study of elderly outpatients. J Gen lntern Med 1988; 3: 317-321. Murphy DJ, Burrows MD, Santilli S, et al. The influence of the probability of survival on patients' preferences regarding cardiopulmonary resuscitation. N Engl J Med 1994; 330: 545-549. Schonwetter RS, Teasdale TA, Taffet G, et al. Educating the elderly: cardiopulmonary resuscitation decisions before and after intervention. J Am Geriatr Soc 1991; 39: 372-377. Frankl D, Oye RK, Bellamy PE. Attitudes of hospitalised patients toward life support: a survey of 200 medical inpatients. Am J Med 1989; 6: 645-648. Ebell MH, Doukas DJ, Smith MA. The do-not-resuscitate order: a comparison of physician and patient preferences and decision-making. Am J Med 1991; 91: 255-260. Robertson GS. Resuscitation and senility: a study of patients' opinions. J Med Ethics 1993; 19: 104-107. Bedell SE, Delbanco TL, Cook EF, Epstein FH. Survival after cardiopulmonary resuscitation. Crit Care Med 1983; 309: 569-576. Miller DL, Gorbien MJ, Simbarti LA, Jahnigen DW. Factors influencing physicians in recommending in-hospital cardiopulmonary resuscitation. Arch Intern Med 1993; 153: 1999-2003. Wagg A, Kinirons M, Stewart K. Cardiopulmonary resuscitation: doctors and nurses expect too much. J R Coll Physicians Lond 1995; 29: 20-24. Rogers v Whitaker (1992) 175 CLR 479. Chappel v Hart [1998] HCA 55. Wyong Shire Council v Shirf (1980) 146 CLR 40. (Received 13 Jan, accepted 7 Jul, 1999) Authors' details Faculty of Medicine and Health Sciences, University of Newcastle, Newcastle, NSW. Ian H Kerridge, MPhil, Lecturer in Clinical Ethics; Sallie-Anne Pearson, PhD, Research Academic; Isobel E Rolfe, MMedEd, Senior Lecturer in Medical Education; Michael Lowe, FRACP, Tutor in Clinical Ethics; John R McPhee, BCom(Hons) (Legal Studies), Consultant in Health Law. Reprints will not be available from the authors. Correspondence: Dr I H Kerridge, Clinical Unit in Ethics and Health Law, Locked Bag 1, Hunter Region Mail Centre, Newcastle, NSW 2310. Email: ikerridgATmail.newcastle.edu.au 1: Sociodemographic characteristics of patients and health professionals surveyed about cardiopulmonary resuscitationPatients (n =153)Healthcare professionals (n =511)PAge > 35 years*90 (59%)235 (46%) 0.007Female98 (65%)373 (73%)< 0.001Marital status† Married/de facto97 (64%)328 (64%) Single/divorced/widowed55 (36%)183 (36%) 0.99Education† Secondary or less117 (77%)72 (14%) Tertiary35 (23%)438 (86%) < 0.001Religion† Religious‡127 (84%)401 (79%) Not religious25 (16%)109 (21%) 0.18Ethnic background† Australian/British144 (95%)463 (91%) Other7 (5%)45 (9%) 0.09Self-reported health status Good68 (45%)493 (97%) Fair/poor/very poor84 (55%)17 (3%)< 0.001CI = confidence interval. * Median split. †Data were missing for some respondents. ‡Any belief system: Christian, Jewish, Islamic, Hindu or Buddhist. Back to text 2: Percentages of 153 patients and 511 health professionals who replied correctly to questions about cardiopulmonary resuscitation (CPR) before provision of written information% Correct (95% confidence interval)QuestionCorrect responsePatientsHealthcare professionalsWhat is involved in CPR?Chest compression (external cardiac massage)Yes*84% (77%-90%)100%Kidney machine (dialysis)No62% (54%-70%)99% (98%-100%)Intravenous drugsYes*25% (18%-33%)84% (80%-87%)Mouth-to-mouth (artificial respiration) Yes86% (80%-91%)99% (98%-100%)Feeding tube into the nose (nasogastric tube)No48% (40%-56%)96% (94%-98%)AntibioticsNo56% (48%-64%)98% (93%-97%)Tube into the throat to assist breathing (intubation)Yes*35% (27%-43%)85% (82%-88%)External electric shock to the heart (defibrillation)Yes*60% (51%-68%)91% (88%-93%)SurgeryNo48% (40%-56%)98% (96%-99%)Breathing machine (ventilator)No27% (20%-35%)57% (53%-61%)How successful is CPR?All patients5%-30%1% (0-4%)26% (22%-30%)Nursing home patients< 5%8% (4%-13%)62% (58%-66%)Patients who have had recent heart attack31%-60%40% (32%-48%)40% (36%-44%)Patients with widespread cancer< 5%42% (34%-50%)61% (57%-65%)Patients with severe infections (eg, pneumonia)< 5%17% (11%-24%)28% (24%-32%)Patients with kidney failure< 5%25% (18%-33%) 39% (35%-43%)Patients under 60 years5%-30%5% (2%-10%)22% (18%-26%)Patients over 70 years5%-30%15% (10%-22%)34% (30%-38%)* As these answers are contentious, analyses were performed using both responses. Back to text

Ian H Kerridge · Sallie-Anne Pearson · Isobel E Rolfe · Michael Lowe · John R McPhee

Research Investigation of a cluster of leukaemia in the Illawarra region of New South Wales, 1989-1996 Victoria J Westley-Wise, Bernard W Stewart, Irene Kreis, Paolo F Ricci, Anthony Hogan, Chris Darling, Steve Corbett, John Kaldor, Neill H Stacey, and Pauline Warburton MJA 1999; 171: 178-183 For editorial comment, see Cartwright Abstract - Introduction - Methods - Results - Discussion - Conclusions - Follow-up - Acknowledgements - References - Authors' details - - More articles on Haematology Abstract Objectives: To investigate a cluster of leukaemia among young people and assess the plausibility of a disease-exposure relationship. Design: Descriptive analysis of population-based leukaemia incidence data, review of evidence related to the causation of leukaemia, assessment of environmental exposures to known leukaemogens, and resulting risks of leukaemia. Setting: Illawarra region of New South Wales, Australia, focusing on suburbs between the Port Kembla industrial complex and Lake Illawarra (the Warrawong area). Main outcome measures: Standardised incidence ratios (SIRs) for leukaemia; current measured and past estimated ambient air benzene concentrations; and expected leukaemia cases attributable to estimates of ambient air benzene concentrations. Results: In 1989-1996, 12 leukaemia cases among Warrawong residents aged less than 50 years were observed, more than the 3.49 cases expected from the rate in the rest of the Illawarra region (SIR, 343.8; 99% CI, 141.6-691.7). These people lived in suburbs immediately to the south-southwest of a coke byproducts plant (a major industrial source of benzene, one of the few known leukaemogens). The greatest excess was among 15-24-year-olds (SIR, 1085.6; 99% CI, 234.1-3072.4). In 1996, ambient air concentrations of benzene averaged less than 1 part per billion (ppb). Since 1970, ambient air concentrations of benzene were estimated to have averaged up to 3 ppb, about one-thousandth of the level at which leukaemia risk has been identified in occupational epidemiological studies. Using the risk assessment model developed by the US Environmental Protection Agency, we estimate that past benzene levels in the Warrawong area could have resulted in 0.4 additional cases of leukaemia in 1989-1996. Conclusions: The excess occurrence of leukaemia in the Warrawong area in 1989-1996 is highly unusual. Current environmental benzene exposure and the reconstructed past environmental benzene exposure level are too low to explain the large excess of leukaemia. The cause of the cluster is uncertain. Introduction In July 1996 the Illawarra Public Health Unit (located in the Wollongong/Port Kembla region, New South Wales) was notified that four former students of a local high school had been diagnosed with leukaemia since 1989. Preliminary investigations established that a cluster of at least 11 people aged less than 40 years who had lived in suburbs near the school had been diagnosed with leukaemia since 1989. On the basis of New South Wales cancer registration data, only about 2 to 3 cases would have been expected. Established causes of leukaemia include occupational benzene exposure, ionising radiation, chemotherapeutic agents, and some inherited and congenital conditions.1-5 Coke byproduct plants are a recognised source of occupational, and potentially of environmental, benzene exposures.1,6,7 The people in the cluster lived in suburbs adjacent to the Port Kembla industrial complex, which includes coke ovens and an associated byproducts plant. We report the investigation of the Illawarra region leukaemia cluster and discuss the plausibility of a disease-exposure relationship. Methods Our investigation followed published guidelines for cancer cluster investigations.8-10 The main components were evaluations of: the pattern of leukaemia incidence in the Illawarra region, with specific attention to residential areas near the Port Kembla industrial complex; environmental exposure to known and putative leukaemogens; and the plausibility of a disease-exposure relationship (whether past environmental exposures to known leukaemogens could explain the excess leukaemia occurrence). The NSW Cancer Council Ethics Committee approved the study. Case finding and investigation Active and passive case-finding methods were used to identify all people resident in the Illawarra region (Wollongong, Shellharbour and Kiama Local Government Areas) aged less than 50 years who had been diagnosed with leukaemia in 1989-1996. The cut-off at age 50 years was chosen as the index cases were young and the age interval 40-50 years represents a natural change in leukaemia occurrence, when the leukaemia risk begins to rise steeply. We actively identified cases from bone marrow aspirate reports, hospital discharge and day-only admission data, and discussions with clinicians in Sydney and Wollongong, community members and organisations. The population-based New South Wales Central Cancer Registry provided the passive case-finding data. For each leukaemia case, we sought to review the medical record and interview the patient and/or a relative to obtain or confirm information about dates of birth and diagnosis, leukaemia cell-type, genetic and medical risk factors, residential and school histories, and personal and/or parental occupational histories. Occurrence evaluation Leukaemia incidence was analysed in eight areas within the Illawarra region with a similar population size (about 20 000-30 000 in 1986), including the area close to the Port Kembla industrial complex (Area 1) (Figure 1). Leukaemia incidence rates in the whole Illawarra region, and each of the eight areas within it, were compared with rates in a reference population by calculating standardised incidence ratios (SIRs) as a means of indirect age standardisation.11 The reference population used was "Urban NSW" (Sydney, Wentworth, Central Coast, Hunter and Illawarra administrative health areas). The calculation of rates was based on place of residence at diagnosis. Using Central Cancer Registry data, we calculated leukaemia SIRs for males and females and for people aged less than 50 years for four five-year periods which had Census years as their mid-points: 1974-1978, 1979-1983, 1984-1988 and 1989-1993. The SIR is the ratio of the number of cancer cases in a study population to the number of cases expected according to the age-specific rate in the reference population (multiplied by 100). For the Illawarra region, leukaemia rates could be calculated to 1996. Thus, for 1989-1996, leukaemia SIRs were calculated for each of the Illawarra areas. In Area 1, they were also classified by age group and cell-type (according to ICD-9),12 using the rest of the Illawarra region as the reference population. Australian Bureau of Statistics (ABS) census data for 30 June 1976, 1981, 1986 and 1991 were used for urban NSW reference populations.13 For the Illawarra region, the ABS provided population data by postcode. Exact Poisson confidence intervals (CI) around the SIRs were estimated.14 CIs were set at 99%, rather than 95%, to reduce the possibility of identifying a chance excess of cancer as statistically significant. SIRs and CIs were calculated with SAS for Windows version 6.11. NSW Central Cancer Registry data and urban NSW population data were accessed from NSW Health's Health Outcomes Information and Statistical Toolbox, a repository of health-related databases for New South Wales. Environmental monitoring and historical exposure reconstruction We undertook an extensive review of the literature on risk factors for and causes of leukaemia, and on carcinogenic effects of occupational and environmental exposures. The environmental assessment focused on exposure to known leukaemogens from the 1970s to 1996. It involved interviewing representatives from industry, government agencies, local residents and workers; inspecting relevant sites; reviewing government, industry and press reports; and collecting and reviewing information on environmental and occupational exposure for residents and workers. The NSW Environment Protection Authority (EPA) and BHP Steel began daily ambient air benzene monitoring in September 1996 in the residential areas nearest the plant (Figure 2). Monitoring was also conducted at three control sites. The EPA used the standard protocol developed by the US Environmental Protection Agency (US EPA) for assessing toxic organic compounds in ambient air.15 BHP used personal samplers, adapted to a stationary role, which collected organic vapours onto an active adsorbent medium by drawing air through the sampler. Both the EPA and BHP analysed the samples with gas chromatography at laboratories registered with the National Association of Testing Authorities. We estimated environmental benzene exposure from the main local sources before September 1996. Using methods developed by the US EPA,16 we estimated levels of emissions from the coke production facilities for each year since 1970 (based on levels in 1996, adjusted for changes in plant equipment and processes and changes in coke and benzene production). Information related to benzene emissions from motor vehicles and other petroleum sources, including Roads and Traffic Authority data on local traffic volumes, was used to provide an upper estimate of the extent to which emissions from these other major local sources may have differed in previous years relative to 1996. Risk estimation The US EPA's benzene risk assessment model of dose-response17 was used to estimate the number of excess leukaemia cases expected in Area 1 between 1989-1996. It was assumed that the Area 1 population had breathed ambient air (70 kg person breathing 20 m3 of air daily) with benzene concentrations equivalent to the estimated maximum annual average concentration since 1970 for the maximally exposed site, continuously over a lifetime (24 hours per day for 70 years). Results Case finding and investigation The same cases were identified by both active and passive case-finding methods. We identified 44 Illawarra residents aged less than 50 years who were diagnosed with leukaemia in 1989-1996. These included 12 people resident in Area 1 at diagnosis, and a 13th person who had moved out of Area 1 a few months before diagnosis (Table 1). None of these 13 people from Area 1 were found to have genetic or medical risk factors for leukaemia. Nor had they ever worked in the production of coke or its byproducts. Six were diagnosed with acute lymphoblastic leukaemia (ALL), four with chronic myeloid leukaemia (CML), and three with acute myeloid leukaemia (AML). Immunophenotypic features and leukaemic classifications revealed no unusual patterns. Six people have died. Of the nine people from Area 1 aged 20 years or less, seven had lived there all their lives, and two for about 11 years. Four attended the same high school in the late 1980s, with three being in the same school year; these three people had leukaemia of different cell-types. Occurrence evaluation For the four five-year periods between 1974-1993, leukaemia incidence in the Illawarra region, and each of its areas, was not significantly different to that throughout urban New South Wales.18 The incidence of total cancers and other specific cancers (including lymphoma and multiple myeloma) was also not significantly higher in Area 1.18However, leukaemia incidence among Area 1 residents aged less than 50 years in 1989-1996 was more than three times higher than in the rest of the Illawarra region (12 cases observed, versus 3.49 expected; SIR, 344; P = 0.0003) (Table 2). The SIR was more than 200 for all leukaemia cell-types, and was significantly increased for ALL (Table 3). The greatest excess of leukaemia was among teenagers and young adults. Among 15-24-year-olds, five cases were observed, versus 0.46 expected (SIR, 1086; 99% CI, 234.1-3072; P = 0.0001). Environmental monitoring and reconstruction of historical exposure The Port Kembla industrial complex contains heavy industries such as copper smelting, sulphuric acid and superphosphate manufacture, petroleum depots, and Australia's largest steelworks, which includes coke ovens and their byproducts plant. In spring and summer, the predominant wind direction is from the northeast; hence, the residential area where the people with leukaemia lived received the greatest exposure from industrial emissions in the region. Benzene was the only known leukaemogen for which local environmental exposures may have been relevant. While ionising radiation is also an established leukaemogen, and there is evidence that exposure to occupational ethylene oxide or 1,3-butadiene can cause leukaemia,3-5 no specific local environmental sources of these agents were identified. At the byproducts plant, coke oven gases are distilled into a benzene- toluene-xylene commercial product (which is 80% benzene), while past practices (until 1977) separated them. The closest residences are more than 1 km from the plant, with all but two Area 1 cases residing 1-3 km from the plant. The other major benzene sources in Area 1 are motor vehicles and petroleum storage tanks. Both EPA and BHP monitoring found that ambient air benzene concentrations in Area 1 averaged less than one part per billion (ppb) in 1996 (Table 4), typical of urban sites in Sydney.19 Analysis of benzene, toluene and xylene ratios in the EPA's samples indicated that about 50% of the benzene was from petroleum. Annual average ambient air benzene concentrations at the most exposed site within Area 1, since 1970, are estimated to have been up to about 3 ppb. Roads and Traffic Authority data showed that traffic volumes in Area 1 have not changed appreciably since the 1970s.20 Local petroleum storage tanks had a greater storage capacity in previous years, which may have been associated with higher emissions.20 Risk estimation The estimated maximum annual average ambient air benzene concentration in Area 1 of 3 ppb is only about one-thousandth of concentrations at which leukaemia risk has been detected in occupational studies.1 The World Health Organization (WHO) has concluded that occupational exposure to an average of 1 part per million (1 ppm, ie 1000 ppb) over a working lifetime has been associated with no statistical increase in leukaemia deaths.1Using the US EPA benzene risk assessment model,17 if a population of 17 500 people (the number of people in Area 1 aged less than 50 years in 1991) had all breathed air with an average benzene concentration of 3 ppb over a lifetime, at the most 0.2 excess leukaemia deaths (or 0.4 cases) would have been expected in 1989-1996. Discussion This investigation into the reported cluster found a highly significant excess leukaemia occurrence in the Warrawong area, particularly among teenagers, in 1989-1996. Estimated past environmental benzene levels are too low to explain this excess. Cluster studies Cancer cluster studies have rarely provided insights into aetiology.3,8-10,21,22 Studies of very rare diseases with well-defined, high exposures are the most likely to yield conclusive results.22 However, these circumstances, in the context of geographic (spatial or spatiotemporal) clusters, are uncommon.10,22 Disease clusters occur continually in any population, and as such represent "expectedly unexpected" events.9,22Drawing boundaries tightly around people observed in clusters inadvertently identifies and overestimates disease excesses.21,22 However, in this study, the geographic, age and time criteria setting the boundaries were not defined or varied to influence the magnitude of the observed excess. Area 1 was a natural geographic grouping of postcodes bounded by industrial zones, Lake Illawarra and the Pacific Ocean. It included suburbs and a postcode area in which no cases were resident. The age range 0-49 years was broad given that the reported cluster was among teenagers. On the other hand, the most recent period analysed (1989-1996) was preceded by a period in which few leukaemia cases were diagnosed.18 Possible leukaemogens Several known leukaemia causes and risk factors were excluded as explanations for the cluster: ionising radiation, ethylene oxide, 1,3-butadiene, and genetic and medical risk factors. For other agents considered, such as dioxins, pesticides, and heavy metals, evidence is lacking of a causal relationship between these agents and leukaemia, despite numerous studies conducted worldwide.5 Although benzene is structurally related to carcinogenic polycyclic hydrocarbons, these agents are generally associated with lung and some other cancers, but not leukaemia.23 Similarly, the influence of genetic polymorphisms, specifically within the cytochrome P450 family, has been associated with lung cancer rather than leukaemia.24Viruses have been causally associated with the rare hairy cell leukaemia and adult T-cell leukaemia.4,5 They have also been suspected to cause leukaemia in childhood and adolescence, but there is still little convincing evidence that they play an important role.4,5,25 Several studies have associated parental smoking with childhood leukaemia.3,26 While many other parental occupational and/or personal exposures (including pesticides, benzene, solvents, petroleum products, and spray paints) have also been implicated in childhood leukaemia, most relevant studies have used poor exposure measures, with inconsistent results.2,3 To our knowledge no leukaemia "clusters" have been reported and investigated in close proximity to industrial facilities similar to those in the Warrawong area. Occupational exposure during steel and coke production has been causally associated with an increased risk of lung and certain other cancers, but not leukaemia.23,27-29 Benzene Benzene was the only known human leukaemogen to which people in Area 1 had potentially significant environmental exposures. Despite the inevitable focus on the steelworks and coke byproducts plant as a benzene source, petrol exhaust and tobacco smoke (the primary source of benzene for smokers, and relevant to non-smokers through passive smoking) are the most significant sources in urbanised populations.1 Food and water are not major sources of benzene exposure,30 and the possibility of contamination of the water supply in Area 1 with benzene from local sources was examined and excluded.20While other haematological malignancies have been associated with occupational benzene exposure,31-36 the evidence has been considered strongest for AML,1 which affected only three people in this cluster. However, a recent review concluded that the few available studies of leukaemia cell-types do not indicate larger or more consistent elevations in risk for AML than for other cell-types.36 Study strengths and limitations Studies of geographic cancer clusters must typically deal with poor information about environmental exposures.22 In this study we had to rely on estimates of past environmental benzene exposure, but several factors suggest that the emission estimates for the coke production facilities are accurate. An independent audit of the emissions estimates concluded that the underlying assumptions were robust and that the emissions inventory was calculated as accurately as possible without an onsite testing program.37 The estimated past annual average ambient air benzene concentration in the maximally exposed part of the Warrawong area, 3 ppb, was similar to concentrations measured38-40 and modelled41 at similar distances from byproducts plants overseas. Measured occupational benzene exposures for Port Kembla byproducts plant workers between 1979 and 199642 were similar to occupational monitoring results for byproducts plant workers from the United States43 and Britain in the 1980s.29,44Using the US EPA benzene risk assessment model,17 even if we assume that average ambient air benzene concentration in Area 1 was 30 ppb (rather than the estimated 3 ppb), this level of exposure would still only explain up to four excess leukaemia cases. Given that the observed leukaemia excess was primarily among young people, the default assumption of 70 kg adults used in our risk estimate is also likely to have slightly overestimated the risk, and hence the number of expected cases. Consistent with public health principles, the US EPA benzene risk assessment model, which uses data from studies of US rubber and chemical workers,36,45,46 is itself based on assumptions that would exaggerate rather than minimise risk, including a linear relationship between exposure to genotoxic carcinogens and leukaemia risk. However, results from some animal studies suggest that a non-linear response may be more biologically plausible.47,48 If this applies to benzene-induced human leukaemia, application of the US EPA model may considerably overestimate risks in the low exposure range. More plausible are biologically based multistage stochastic models for chemical carcinogenesis, which account for cellular birth, death, initiation, promotion and other biological processes, each stage being linked by a stochastic transition probability that accounts for exposure (or dose).47,48 Such models are likely to produce lower estimates of leukaemia risk for low levels of environmental exposure. However, large uncertainties are inherent in any conversion of risk estimates from animal studies or occupational studies to risk estimates for benzene exposure in the general community. Many people -- children, people of reproductive age, those with other risk factors -- may have susceptibilities to leukaemia quite different from those of the male workers studied in occupational studies. Animal and human studies have begun to clarify the potential risk associated with relatively high transient and/or intermittent benzene exposure49 versus cumulative exposure, but the relationship is still poorly understood for low dose extrapolations. In addition, people in the community are exposed to a variety of agents, which may have as yet unidentified additive and possibly synergistic effects. Conclusions On current knowledge, the recent ambient air benzene concentrations in the Warrawong area represent a negligible leukaemia risk, and the estimated past benzene concentrations are too low to explain the large excess of leukaemia cases that occurred in 1989-1996. However, factors such as variation in susceptibilities of individuals and population groups such as children, the possible effects of intermittent and high transient benzene exposures, and interactions between different agents, mean that we cannot exclude a causal association between leukaemia occurrence among young people in the Warrawong area and chemical exposures. Follow-up A feasibility study is being undertaken to examine the potential for relating disease to chemical exposures through a case-control study. Broadly, chemical exposures of interest are personal exposures (environmental and individual, to benzene and industrial emissions in general) and parental exposures (focusing on those with prior evidence of an association with leukaemia). In addition, routine surveillance of leukaemia and lymphoma is continuing, as is ambient air monitoring for benzene and other hazardous pollutants in the Warrawong area. Acknowledgements We gratefully acknowledge the assistance given by the individuals with leukaemia and their relatives, and other community members who participated in the investigation's Community Reference Group. We give special thanks to members of the Illawarra Leukaemia Investigation Steering Committee for their commitment and contributions: Giovanna Crocco and David Gilmour (Community Reference Group), Richard Willison and Trevor Dunn (Illawarra Public Health Unit), Joe Woodward and Craig Lamberton (NSW Environmental Protection Authority), Ron Hales (Wollongong City Council), and Christine Ewan (University of Wollongong). Many individuals and organisations assisted in and supported the study, but we would particularly like to thank the following organisations: NSW Health; Illawarra Area Health Service; New South Wales Cancer Council; BHP Port Kembla; NSW Environmental Protection Authority; and the University of Wollongong. We also thank John Marthick (University of Wollongong) and Paddy Ranasinghe (Illawarra Public Health Unit) for preparing the maps. References World Health Organization. Benzene (Environmental Health Criteria No. 150). Geneva: International Programme on Chemical Safety, 1993. Ross JA, Davies SM, Potter JD, Robison LL. Epidemiology of childhood leukaemia, with a focus on infants. Epidemiol Rev 1994; 16: 243-272. Cartwright RA, Staines A. Acute leukaemias. Clin Haematol 1992; 5: 1-26. Finch SC, Linet MS. Chronic leukaemias. Clin Haematol 1992; 5: 27-56. Tomatis L, editor. Cancer: causes, occurrence and control (IARC Scientific Publications No. 100). Lyon: International Agency for Research on Cancer, 1990. Fishbein L, O'Neill IK, editors. Benzene and alkylated benzenes. Environmental carcinogens: methods of analysis and exposure measurement, vol. 10 (IARC Scientific Publications No. 85). Lyon: International Agency for Research on Cancer, 1988. United States Environmental Protection Agency. National emission standards for hazardous air pollutants; regulation of benzene; response to public comments. US Federal Register 1984, 40 CFR Part 61, AD-FRL-2523-7. Centers for Disease Control. Guidelines for investigating clusters of health events. MMWR Morb Mortal Wkly Rep 1990, 39 (R-11): 1-23. Bender AP, Williams AN, Johnson RA, Jagger HG. Appropriate public health responses to clusters: the art of being responsibly responsive. Am J Epidemiol 1990; 132 Suppl 1: S48-S52. Fiore BJ, Hanrahan LP, Anderson HA. State health department response to disease cluster reports: a protocol for investigation. Am J Epidemiol 1990; 132 Suppl 1: S14-S22. Breslow NE, Day NE. Statistical methods in cancer research. Vol II. Lyon: International Agency for Research on Cancer, 1987. World Health Organization. Manual of the International Statistical Classification of Diseases, Injuries and Causes of Death. 9th Revision. Geneva: WHO, 1977. Australian Bureau of Statistics. Estimated resident populations by age and sex in statistical local areas, New South Wales (Catalogue. No. 3209.1, 30 June 1976, 1981, 1986, 1991). Canberra: ABS, 1978, 1983, 1988, 1993. Daly L. Simple SAS macros for the calculation of exact binomial and Poisson confidence limits. Comput Biol Med 1992; 22: 351-361. United States Environmental Protection Agency. Compendium of methods for the determination of toxic organic compounds in ambient air. 2nd edition. Compendium method TO-14A. Determination of volatile organic compounds (VOCs) in ambient air using specially prepared canisters with subsequent analysis by gas chromatography (EPA/625/R-96/010b). Cincinnati, OH: Center for Environmental Research Information, 1999. <http://www.epa.gov/ttn/amtic/files/ambient/airtox/to-14ar.pdf> Accessed 5 July 1999. United States Environmental Protection Agency. AP-42, 5th ed., vol. I. Chapter 12: metallurgical industry. 12.2 Coke production [draft]. Research Triangle Park, NC: Emission Factor And Inventory Group, US EPA, 1995. <http://www.epa.gov/ttn/chief/ap42pdf/c12s02.pdf>. Accessed 5 July 1999. [No longer available, but see final version at <http://www.epa.gov/ttn/chief/ap42/ch12/final/c12s02.pdf> Accessed 10 May 2001. United States Environmental Protection Agency. Integrated risk information system (IRIS). Benzene (CASRN 71-43-2). 16 October 1998. Cincinnati, OH: Environmental Criteria and Assessment Office, Office of Health and Environmental Assessment, Office of Research and Development, Cincinnati, 1996. <http://www.epa.gov/ngispgm3/iris/subst/0276.htm> Accessed 5 July 1999. Westley-Wise V, Hogan A. Report on the occurrence of leukaemia (1974-96) and other cancers (1974-93) in the Illawarra. Illawarra Area Health Service. Wollongong 1997. Wadge A, Salisbury J. Benzene. National Environmental Health Forum Monographs, Air Series 2. National Environmental Health Forum, South Australian Health Commission, Adelaide,1997. Kreis I, Willison R. Environmental assessment related to a leukaemia cluster. University of Wollongong. Wollongong 1997. Olsen SF, Martuzzi M, Elliott P. Cluster analysis and disease mapping -- why, when and how? A step by step guide. BMJ 1996; 313: 863-866. Rothman KJ. A sobering start for the cluster busters' conference. Am J Epidemiol 1990; 132 Suppl 1: S6-S13. International Agency for Research on Cancer. Polynuclear aromatic compounds. Part 3, Industrial exposures in aluminium production, coal gasification, coke production, and iron and steel founding. IARC Monographs on the evaluation of the carcinogenic risk of chemicals to humans, vol. 34. Lyon: IARC, 1984. Wormhoudt LW, Commandeur JNM, Vermeulen NPE. Genetic polymorphisms of human N-acetyltransferase, cytochrome P450, glutathione-S-transferase, and epoxide hydrolase enzymes: relevance to xenobiotic metabolism and toxicity. Crit Rev Toxicol 1999; 29: 59-124. Alexander FE. Viruses, clusters and clustering in childhood leukaemia: a new perspective? Eur J Cancer 1993; 29A: 1424-1443. Sorahan T, Lancashire RJ, Hulten MA, Stewart AM. Childhood cancer and parental use of tobacco: Deaths from 1953 to 1955. Br J Cancer 1997; 75: 134-138. Redmond CKA, Strobino BR, Cypress RH. Cancer experience among coke byproduct workers. Ann N Y Acad Sci 1976; 217: 102-115. Swaen GM, Slanged JAM, Volovics A, et al. Mortality of coke plant workers in the Netherlands. Br J Indust Med 1991; 48: 130-135. Hurley JF, Cherrie JW, Maclaren W. Exposure to benzene and mortality from leukaemia: results from coke oven and other coal product workers. Br J Indust Med 1991; 48: 502-504. Wallace L. Environmental exposure to benzene: an update. Environ Health Perspect 1997; 104: 1129-1136. Yin SN, Hayes RB, Linet MS, Li GL, et al. A cohort study of cancer among benzene-exposed workers in China: overall results. Am J Indust Med 1996; 29: 227-235. Hayes RB, Yin S-N, Dosemeci M, Li G-L, et al. Benzene and the dose-related incidence of hematologic neoplasms in China. J Natl Cancer Inst 1997; 89: 1065-1071. Rinsky RA, Alexander B, Smith MD, Hornung R, et al. Benzene and leukaemia: an epidemiological risk assessment. N Engl J Med 1987; 136: 1044-1050. Christie D, Robinson K, Gordon I, Bisby J. A prospective study in the Australian petroleum industry. I. Mortality. Br J Indust Med 1991; 48: 507-510. Wong O. An industry wide mortality study of chemical workers occupationally exposed to benzene. II. Dose response analyses. Br J Indust Med 1987; 44: 382-395. Savitz DA, Andrews KW. Review of epidemiologic evidence on benzene and lymphatic and hematopoietic cancers. Am J Indust Med 1997; 31: 287-295. Holmes Air Sciences 1997. Air quality report: review of BHP report to the Illawarra Area Health Service Leukaemia Task Force. Report prepared for the NSW Environment Protection Authority. Sydney: Holmes Air Sciences, 1997. Harkov R, Olsakovsky AC, Fillo JP. Determining the ambient impacts of coke and coke by-products manufacturing on selected pollutant levels in neighboring communities: I -- results from a six-month ambient air benzene monitoring study. Air toxics and volatile organic compounds: papers from the 84th Annual Meeting and Exhibition of the Air and Waste Management Association, Vancouver, 1991. Vol 6. Pittsburgh: Air and Waste Management Association, 1991. GDCh (Society of German Chemists) Advisory Committee on Existing Chemicals of Environmental Relevance. Benzene. Weinhmam, VCH Verlagsgesellschaft, 1988. In: World Health Organization. Benzene (Environmental Health Criteria No. 150). Geneva: WHO, 1993: 36. Fentiman AF, Neher MB, Kinzer GW, et al. Environmental monitoring benzene (PB-295 641). Prepared for US EPA. Springfield, VA: Battelle Columbus Laboratories, National Technical Information Service, 1979. In: IARC. Some industrial chemicals and dyestuffs (IARC monographs on the evaluation of the carcinogenic risk of chemicals to humans, vol. 29). Lyon: IARC, 1982. United States Environment Protection Agency. Benzene emissions from coke byproduct recovery plants -- background information for proposed standards (EPA-450/3-83-016a). Research Triangle Park, NC: Office of Air Quality Planning and Standards, 1984. BHP Flat Products Division. Report to the Illawarra Area Health Service. Port Kembla: BHP Flat Products Division, 1997. Runion HE, Scott LM. Benzene exposure in the United States, 1978-1983: an overview. Am J Indust Med 1985; 7: 385-393. Drummond L, Luck R, Afacan AS, Wilson HK. Biological monitoring of workers exposed to benzene in the coke oven industry. Br J Indust Med 1988; 45: 256-261. Rinsky RA, Young RJ, Smith AB. Leukemia in benzene workers. Am J Indust Med 1981; 2: 217-245. Ott MG, Townsend DT, Fishbeck WA, Langner RA. Mortality among workers occupationally exposed to benzene. Arch Environ Health 1978; 33: 3-10. Cox LA Jr, Ricci PF. Reassessing benzene cancer risks using internal doses. Risk Analysis 1992; 12: 401-409. Cox LA Jr, Ricci PF. Dose-response non-linearities for benzene revisited: A reply to C Crump. Risk Analysis 1993; 14: 485-486. (Received 2 July 1998, accepted 18 May 1999) Authors' details Illawarra Public Health Unit, Illawarra Area Health Service, Wollongong, NSW. Victoria J Westley-Wise, MPH, FAFPHM, Director; Anthony Hogan, MSc(Hons), PhD, Public Health Officer. Children's Cancer Research Institute, Sydney Children's Hospital, Sydney, NSW. Bernard W Stewart, PhD, FRACI, Research Director; now Head of Cancer Control Program, South Eastern Sydney Area Health Service. University of Wollongong, Wollongong, NSW. Irene Kreis, PhD, FAFPHM, Senior Lecturer; Paolo F Ricci, MSc, PhD, Professorial Fellow. BHP Steel Flat Products Division, Wollongong, NSW. Chris Darling, MSc(Occup Med), FAFOM, Occupational Health Advisor. NSW Health Department, Sydney, NSW. Steve Corbett, MPH, FAFPHM, Manager. National Centre for HIV Epidemiology and Clinical Research, University of New South Wales, Sydney, NSW. John Kaldor, PhD, Deputy Director, and Professor of Epidemiology. University of Sydney, Sydney, NSW. Neill H Stacey, BSc(Hons), PhD, Associate Professor. Illawarra Regional Hospital, Illawarra Area Health Service, Wollongong, NSW. Pauline Warburton, MB BS, FRACP, Director. Reprints: Dr V J Westley-Wise, Illawarra Public Health Unit, PO Box 66, Keiraville, NSW 2500. Email: vwestATdoh.health.nsw.gov.au Figure 1 (above): Map of the Illawarra region, showing areas used for comparing leukamia incidence. Back to textFigure 2 (below): Map of the Warrawong area (central portion of area 1), showing the location of the four ambient air monitoring stations. Back to textBack to textBack to textBack to textBack to text

Victoria J Westley-Wise · Bernard W Stewart · Irene Kreis · Paolo F Ricci · Anthony Hogan · Chris Darling · Steve Corbett · John Kaldor · Neill H Stacey · Pauline Warburton

Research Vancomycin and teicoplanin use in Victorian hospitals Marion B Robertson, Jonathan G A Dartnell and Tony M Korman, on behalf of the Victorian Drug Usage Evaluation Group MJA 1999; 171: 127-131 See also Ferguson, Grayson et al & Collignon Abstract - Introduction - Methods - Results - Discussion - Acknowledgements - References - Authors' details - - More articles on Pharmacology Abstract Objective: To determine patterns of prescribing of glycopeptide antibiotics (vancomycin and teicoplanin) in Victorian hospitals and identify areas for targeted intervention. Design: A concurrent, observational, multisite evaluation of drug use. Setting: Thirty-five Victorian hospitals, 1-14 September 1997. Study population: Patients commencing a glycopeptide antibiotic course. Main outcome measures: Rate of glycopeptide antibiotic use; indications; duration of use; main hospitals using glycopeptide antibiotics. Results: 293 patients (269 adults and 24 neonates) commenced on 302 glycopeptide antibiotic courses: 296 intravenous (IV) vancomycin courses and three each of oral vancomycin and parenteral teicoplanin. The overall rate of use was 10.3 courses per 1000 inpatient separations. Of 271 IV vancomycin courses for adults, 176 (65%) were for treatment -- 120 empirically. The median duration of treatment courses was 4.7 days (interquartile range, 2.0-8.2 days). A flucloxacillin-resistant organism was confirmed for 44% of treatment courses. Ninety-five IV vancomycin courses were for prophylaxis, including for cardiac (54%) and vascular surgery (21%); 82% of prophylactic courses were administered for less than 24 hours. Of all the glycopeptide antibiotic courses, 69% were administered at five major metropolitan hospitals. Conclusions: Glycopeptide antibiotic use in Victoria is concentrated in the major metropolitan hospitals. Prolonged durations of vancomycin therapy, including for surgical prophylaxis and empirical therapy not subsequently confirmed by microbiology findings, would be suitable targets for interventional strategies. Introduction The emergence of resistant strains of Staphylococcus aureus and coagulase-negative staphylococci has resulted in increased use of the glycopeptide antibiotics vancomycin and teicoplanin.1 These antibiotics are the only effective treatments for infections with these pathogens, but the emergence of vancomycin-resistant enterococci (VRE) threatens their utility.1,2 VRE can cause serious life-threatening infections, and can transfer their resistance in vitro to other pathogens, such as Staphylococcus, rendering the bacteria resistant to currently available antimicrobials.3,4The emergence of VRE has been linked to both overuse and inappropriate use of antibiotics such as vancomycin, teicoplanin and extended-spectrum cephalosporins.2,4 There is also a strong relationship with the use of glycopeptide antibiotics in animals.5 Australian studies have reported inappropriate use of vancomycin ranging from 42% to 65% in individual hospitals.6,7 Responding to these concerns, consensus guidelines have been disseminated to all Victorian hospitals.8 The aim of this study was to examine patterns of prescribing vancomycin and teicoplanin in Victorian hospitals in order to identify potential areas for targeted intervention to improve use of these antibiotics. Methods This study was conducted by the Victorian Drug Usage Evaluation Group, a multidisciplinary group that aims to promote and improve coordination of drug use evaluation activities as a means to improve drug use. All Victorian public and private hospital pharmacy departments listed in the Society of Hospital Pharmacists of Australia directory9 were invited to participate. All patients at participating hospitals who commenced a course (Box 1) of oral or intravenous (IV) vancomycin or teicoplanin between 1-14 September 1997 inclusive were enrolled. Pharmacists at each hospital collected data concerning: demographic details, the indication for antibiotic therapy as determined from the medical record or by consultation with the prescriber, site and source of infection, beta-lactam hypersensitivity, and past history of methicillin-resistant S. aureus (MRSA) or methicillin-resistant S. epidermidis (MRSE), microbiology results of specimens taken up to seven days before or concurrent with vancomycin or teicoplanin therapy, administration of antibiotics before, during and immediately after vancomycin or teicoplanin therapy, and specialist consultation and advice. Prescription data were collected until the end of the course, until the day of discharge or death, or until 28 September 1997. Each hospital provided details of the number of inpatient separations between 1-14 September 1997. This was used to estimate the number of glycopeptide antibiotic courses commenced per 1000 inpatient separations. Data were evaluated by the Mann-Whitney rank sum test of the equivalence of medians of samples not drawn from a normally distributed population. Proportions were compared using the χ2 test. Data are presented as proportions, medians and interquartile ranges. Results Hospitals: Thirty-five hospitals participated in the study: 33 of 58 public and two of 14 private hospitals invited to participate. Twenty hospitals were in the Melbourne metropolitan area and 15 in regional areas. Patients: In the study period, 293 patients (mean age, 54 years; range, 0-90 years; 112 females) commenced a course of vancomycin or teicoplanin. The 293 patients received 302 courses of glycopeptide antibiotics -- three teicoplanin, three oral vancomycin, and 296 intravenous vancomycin (Box 2). Five of Melbourne's six major metropolitan hospitals participated and administered 209 (69%) of the 302 courses. Twenty hospitals enrolled between one and 11 patients and contributed the remaining 93 courses. Glycopeptide antibiotic use in neonates: Four hospitals enrolled 24 neonates who were prescribed 25 courses of IV vancomycin, with a median duration of 2.3 days (interquartile range, 2.0-4.6 days). One course was for prophylaxis for abdominal surgery; the other 24 courses were for empirical treatment. One treatment course was for respiratory infection and the others were for an unknown site of infection. Flucloxacillin-resistant organisms were isolated for six courses (two MRSA, four coagulase-negative staphylococci). Glycopeptide antibiotic use in adults: Twenty-five hospitals enrolled 269 adults who commenced 277 glycopeptide antibiotic courses; 143 (53%) patients were treated by a medical unit and 126 (47%) by a surgical unit; 73 (27%) of the patients were in an intensive care unit at some stage during the glycopeptide antibiotic course. Of the 269 patients, 235 (87%) were discharged, 30 (11%) died, and 4 (1%) were still in hospital three months after the study. For 94 courses (34%), specialist consultation for the use of vancomycin and teicoplanin was noted in the records. More than 80% of these consultations were with infectious disease/microbiology specialists. There were 263 patients who received 271 courses of IV vancomycin -- 176 (65%) for treatment and 95 for prophylaxis (Box 3). Of the 176 IV vancomycin treatment courses, 120 (68%) were for empirical treatment and 56 (32%) as specific treatment. Empirical courses were shorter than specific courses (P < 0.02). Patients being treated empirically had been in hospital for fewer days before the course commenced than patients receiving specific treatment (P < 0.001). The duration of empirical courses for which a flucloxacillin-resistant organism was subsequently identified was significantly greater than that of the unconfirmed courses (P < 0.05). Of the other six patients, three received a course of oral vancomycin for the treatment of confirmed (two cases) or suspected (one case) Clostridium difficile diarrhoea, and three received IV teicoplanin for treatment of wound infections (2 patients) or cellulitis (1 patient), commenced on the advice of infectious diseases clinicians. Other antibiotic use in adults: In the seven days before IV vancomycin treatment courses, the most frequently prescribed antibiotics were ceftriaxone and cefotaxime (28% of courses), metronidazole (16%), flucloxacillin (15%), and gentamicin (15%). The most frequently prescribed concurrent antibiotics were ceftriaxone and cefotaxime (11% of vancomycin courses), gentamicin (11%), ceftazidime (9%), imipenem (7%), ciprofloxacin (6%) and metronidazole (6%). For 14 treatment courses, vancomycin was continued beyond the close of the study. Immediately following the 162 completed courses (in 162 patients), other antibiotic therapy was commenced for 57 patients (35%), concurrent antibiotic therapy continued for 35 patients (22%), and there was no antibiotic therapy for 50 patients (31%). On-going therapy was not recorded for nine patients (6%), and the remaining 11 patients (7%) died. The antibiotics most frequently commenced immediately after vancomycin were oral flucloxacillin (12 courses), oral fusidic acid and oral rifampicin (12 courses), and oral ciprofloxacin (eight courses). For the 95 IV vancomycin prophylaxis courses, other antibiotics were given concurrently for 46 (48%) courses, most frequently gentamicin (14% of vancomycin courses), ceftriaxone (14%), cephazolin (6%) and ticarcillin/clavulanate (5%). Comparison of five main vancomycin users: The five hospitals that used most vancomycin were major metropolitan university teaching hospitals where use was restricted by protocols and consultations. Use by number of courses was greatest in Hospital 1 but the total quantity used was less than half that used at Hospital 2 (Box 4). Rate of use was significantly lower in Hospital 4 than in the other four hospitals. Surgical use of vancomycin predominated in Hospital 1, while medical use predominated in Hospital 4. Hospitals 2, 3 and 5 had a significantly smaller proportion of single doses for prophylaxis than Hospitals 1 and 4. Discussion We have examined the pattern of use of vancomycin in a large sample of Victorian hospitals. We found that use of teicoplanin and oral vancomycin was low, and that intravenous vancomycin was used predominantly for empirical treatment or prophylaxis. Our study is the first to capture a statewide picture of hospital drug use linked to indication, and we are not aware of any published comparable multihospital pharmacoepidemiological data. Based on the number of inpatient separations for Victorian public hospitals in 1996-97 (about 890 000),10 we reviewed an estimated 82% of public hospital inpatient separations in the two-week study period. We estimated the overall rate of glycopeptide antibiotic use to be 10.3 courses per 1000 inpatient separations, which suggests about 9160 courses are used annually in Victorian public hospitals. The main limitation of our study was that vancomycin courses were not individually compared against explicit criteria to determine the proportion of appropriate use on the basis of indication, dosage and duration. We chose not to compare against criteria because of the variety of prescribing restrictions in participating hospitals. The most frequent indications for empirical IV vancomycin were febrile neutropenia, pneumonia and wound infections. The indications for and duration of empirical therapy could be targets for intervention. For pneumonia, IV vancomycin use may be reasonable for hospital-acquired infections in institutions with a high prevalence of MRSA. For febrile neutropenia, IV vancomycin may be best restricted to patients with suspected associated IV line sepsis. For wound infections, unless there is a high prevalence of MRSA, IV vancomycin treatment should wait until after microbiological confirmation. Oral vancomycin should be restricted to the treatment of antibiotic-associated colitis due to toxigenic C. difficile unresponsive to or relapsing after an adequate course of metronidazole (or bacitracin), or for patients with severe colitis.11 Clinicians should be encouraged to regularly review the need for ongoing drug therapy; however, to improve the current situation, we need effective decision support tools to facilitate timely attention to important test findings.12 Almost half the treatment courses were sanctioned or recommended by infectious diseases/microbiology specialists, who should be familiar with relevant guidelines for vancomycin use. Surgical prophylaxis consumed 35% of IV vancomycin courses, with cardiac and vascular surgery accounting for 75% of prophylactic courses. The duration of vancomycin prophylaxis was generally according to recommendations, with 72% of courses given as single doses and 82% of courses given for less than 24 hours. However, there was considerable interhospital variation in the duration of surgical prophylaxis. The implementation of hospital policies on duration of surgical prophylaxis would be an important target for intervention. Glycopeptide antibiotic use was concentrated in five major metropolitan teaching hospitals. Although these five hospitals all had policies in place, vancomycin use varied substantially. It is of critical importance for patient care and resource management that there is a conscious effort to preserve the utility of vancomycin and teicoplanin. This should be founded on good infection control practice, but there is also a need for all hospitals to implement effective interventional strategies to improve the use of the glycopeptide antibiotics. Acknowledgements Financial support for the employment of the Project Coordinator was provided by the Victorian Drug Usage Advisory Committee and the Victorian Standing Committee on Infection Control. The project was also reliant on the voluntary work of pharmacists at the participating hospitals who undertook all the data collection. We acknowledge the assistance of the other members of the Victorian Drug Usage Evaluation Group in the planning and execution of the study and the preparation of this manuscript: Stephanie J Alvarez, Drug Utilisation Evaluation Pharmacist, Monash Medical Centre; Dr Jo-anne Brien, Senior Lecturer, Department of Pharmacy Practice, Monash University (Parkville Campus); Dr Lisa L Ioannides-Demos, Senior Research Fellow, Department of Epidemiology and Preventative Medicine, Monash University, and Senior Research Officer, Victorian Centre for Ambulatory Care Innovation, Alfred Hospital; Sam Koroneos, Senior Drug Utilisation Pharmacist, Pharmacy Department, Austin and Repatriation Medical Centre; Anne Leversha, Senior Lecturer, Monash University, Faculty of Medicine, and Victorian College of Pharmacy, and Deputy Manager Pharmacy Services, Latrobe Regional Hospital, Traralgon West; Julie A V Lord, Senior Drug Information and Clinical Research Pharmacist, St Vincent's Hospital, Melbourne; Heather J Lyall, Deputy Director of Pharmacy, Geelong Hospital; Roslyn I McKinnon, Executive Officer, Victorian Drug Usage Advisory Committee; Associate Professor R Moulds, Director, Department of Clinical Pharmacology and Therapeutics, Royal Melbourne Hospital; Susan G Poole, Deputy Director, Peter MacCallum Cancer Institute; Dr Gail J Ware, Drug Usage Evaluation Pharmacist, Alfred Hospital. References Ena J, Dick RW, Jones RN, Wenzel RP. The epidemiology of intravenous vancomycin usage in a university hospital: a 10 year study. JAMA 1993; 269: 598-602. Heath CH, Blackmore TK, Gordon DL. Emerging resistance in Enterococcus spp. Med J Aust 1996; 164: 116-120. Reduced susceptibility of Staphylococcus aureus to vancomycin -- Japan, 1996. MMWR Morb Mortal Wkly Rep 1997; 46: 624-626. Quale J, Landman D, Atwood E, et al. Experience with a hospital-wide outbreak of vancomycin-resistant enterococci. Am J Infect Control 1996; 24: 372-379. Witte W. Medical consequences of antibiotic use in agriculture. Science 1998; 279: 996-997. Misan GMH, Martin ED, Smith ER, et al. Drug utilisation review in a teaching hospital: experience with vancomycin. Eur J Clin Pharmacol 1990; 39: 457-461. Radford JM, Whitby RM, Looke DFM, Coombes JA. Vancomycin usage review in the era of vancomycin-resistant enterococci (VRE). Aust J Hosp Pharm 1997; 27: 1410-1413. Guidelines for preventing emergence of vancomycin-resistant enterococci. Melbourne: Victorian Drug Usage Advisory Committee and the Standing Committee on Infection Control in collaboration with the Writing Group for the Therapeutic Guidelines: Antibiotic, 1997. Vernon G, Thomson W, editors. Directory of hospital pharmacy and pharmaceutical organisations. Melbourne: Society of Hospital Pharmacists of Australia, 1996. Department of Human Services Annual Report 1996-97. Melbourne: Department of Human Services, Victoria, 1997. Therapeutic Guidelines: Antibiotic, 10th ed. Melbourne: Therapeutic Guidelines Limited, 1998. Schiff GD, Rucker D. Computerized prescribing. Building the electronic infrastructure for better medication usage. JAMA 1998; 279: 1024-1029. (Received 21 Dec 1998, accepted 17 May 1999) Authors' details Victorian Drug Usage Evaluation Group, Melbourne, VIC. Marion B Robertson, BPharm, MSc, Project Coordinator. Royal Melbourne Hospital, Melbourne, VIC. Jonathan G A Dartnell, BPharm, MPS, Senior Pharmacist, Department of Clinical Pharmacology and Therapeutics. Monash Medical Centre, Melbourne, VIC. Tony M Korman, FRACP, Infectious Diseases Physician. Reprints will not be available from the authors. Correspondence: Mr J G A Dartnell, Department of Clinical Pharmacology and Therapeutics, c/- Post Office, Royal Melbourne Hospital, VIC 3050. Email: Jonathan. DartnellATnwhcn.org.au 1: Definitions Course: The administration to a patient of at least one dose of vancomycin or teicoplanin. If a dose was administered more than 24 hours after a previous dose (and the drug was re-prescribed), this was considered a new course. Patients with renal impairment on regimens with dosing intervals longer than 24 hours were considered to have received a continuous course. Prophylaxis: Antibiotic administration commenced perioperatively to prevent postoperative infection. Empirical treatment: Antibiotic administration commenced before or without identification of flucloxacillin-resistant bacterial pathogens. Specific treatment: Antibiotic administration commenced after identification of flucloxacillin-resistant bacterial pathogens. Duration: Durations of courses were calculated by subtracting the date and time of the first dose from the date and time of the last dose. The duration of a single dose was considered to be 0 hours. Back to text 2: Glycopeptide antibiotic courses commenced in participating hospitals during study period*Hospital locationNumber of courses † (patients)Number of inpatient separationsCourses per 1000 separationsMelbourne52 (51)‡235622.0Melbourne49 (48)245520.0Melbourne38 (37)194319.6Melbourne37 (37)299712.3Melbourne33 (30)§141623.3Regional11 (11)14777.4Melbourne10 (9)53818.6Melbourne10 (10)14776.7Melbourne9 (9)10718.4Melbourne8 (8)10897.3Melbourne8 (7)13575.9Melbourne7 (7)42316.5Regional5 (5)8086.2Melbourne5 (5)10554.7Melbourne3 (3)15002.0Melbourne3 (2)27111.1Regional3 (3)3748.0Regional2 (2)§14613.7Melbourne2 (2)8772.3Melbourne2 (2)4694.3Melbourne1 (1)6681.5Regional1 (1)7251.4Regional1 (1)§2663.8Melbourne1 (1)6361.6Melbourne1 (1)5102.0Overall*302 (293)2944510.3*Another 9 regional and 1 Melbourne hospital accounted for 2785 inpatient separations but did not use glycopeptide antibiotics during the study period. †Intravenous vancomycin except where indicated. ‡Includes 3 courses of teicoplanin. §Includes 1 course of oral vancomycin. Private hospital. Back to text 3: Intravenous vancomycin courses commenced in adultsAll treatment courses 176 treatment courses median duration, 4.7 days (interquartile range, 2.0-8.2 days) indications: wound infections33 (18.8%)pneumonia29 (16.5%)febrile neutropenia21 (11.9%)septicaemia12 (6.8%)intravenous catheter-associated sepsis11 (6.3%)cellulitis9 (5.1%)infected prosthesis9 (5.1%)peritonitis/CAPD8 (4.5%)meningitis3 (1.7%)other16 (9.1%)not recorded25 (14.2%)Gram-positive bacteria, including Staphylococcus, Streptococcus, Enterococcus and Bacillus species isolated for 113 (64%) courses Isolated organism confirmed flucloxacillin-resistant in 78 courses; organism was MRSA in 67 coursesEmpirical treatment courses 120 empirical treatment courses (68% of all treatment courses) median duration, 4.4 days (interquartile range, 1.5-8.0 days) patients in hospital for median 4.0 days (interquartile range, 1-12 days) before course commenced most frequent indications: febrile neutropenia (18%), pneumonia (16%) and wound infections (12%) 32 (26%) of patients had history of beta-lactam antibiotic hypersensitivity or a previous infection with MRSA or MRSE. Two patients had a history of severe hypersensitivity that may have necessitated use of vancomycin rather than beta-lactam antibiotics. flucloxacillin-resistant organism subsequently identified in 24 (20%) empirical courses duration of courses with confirmed resistant organism was 6.9 days (interquartile range, 2.0-13.8 days) compared with 3.9 days (interquartile range, 1.3-6.7 days) for unconfirmed courses (P<0.05) Specific treatment courses 56 specific treatment courses (32% of all treatment courses) median duration, 6.1 days (interquartile range, 3.0-11.6 days) patients in hospital for median 12 days (interquartile range, 4-22 days) before course commenced Surgical prophylaxis courses95 surgical prophylaxis courses 68 (72%) single-dose courses, 78 (82%) less than one day, 12 (12%) one to three days, 5 (5%) more than three days 12 (13%) were for patients with history of beta-lactam hypersensitivity, including four severe cases, and 5 (5%) were for patients with history of MRSA or MRSE infection 51 (54%) used in cardiac surgery; 31 single doses, 12 (24%) lasted more than 36 hours 20 (21%) used in vascular surgery; all were single dosesCAPD=chronic ambulatory peritoneal dialysis. MRSA=methicillin-resistant Staphylococcus aureus. MRSE=methicillin-resistant S. epidermidis. Back to text 4: Comparison of intravenous vancomycin in adults in the major teaching hospitalsHospital 1Hospital 2Hospital 3Number of courses494938Number of patients484837Courses per 1000 separations*20.820.019.6Vancomycin used (g)177371280Median length of stay (days)121816Courses given in intensive care9 (18%)9 (18%) 17 (45%)Courses prescribed by medical unit12 (24%)28 (57%)20 (53%)Indication: prophylaxisNumber of courses (%)28 (57%)15 (31%)15 (39%)Number (%) as single doses †28 (100%)7 (47%)9 (60%)Surgery type (number of courses) cardiac2476 vascular241 other248Indication: treatmentNumber (%) empirical13 (62%)30 (88%) 19 (83%)Median duration of courses (days) empirical3.73.64.8 specific3.04.511.0Indications (number of courses) pneumonia724 wound infection436 febrile neutropenia181 other92112 Hospital 4Hospital 5Number of courses3332Number of patients3229Courses per 1000 separations*12.322.6Vancomycin used (g)299118Median length of stay (days)209Courses given in intensive care6 (18%)9 (28%)Courses prescribed bymedical unit30 (91%)17 (53%)Indication: prophylaxisNumber of courses (%)8 (24%) 16 (50%)Number (%) as single doses †8 (100%)11 (69%)Surgery type (number of courses) cardiac18 vascular63 other15Indication: treatmentNumber (%) empirical18 (72%)10 (62%)Median duration of courses (days) empirical5.51.5 specific3.02.0Indications (number of courses) pneumonia60 wound infection33 febrile neutropenia24 other149* Hospital 4 significantly less than Hospitals 1, 2, 3 and 5 (P<0.02). †Hospitals 2, 3, and 5 significantly less than Hospital 1 (P<0.01) and Hospital 4 (P<0.1). Back to text

Marion B Robertson

Research Outcome of a screening program for vancomycin-resistant enterococci in a hospital in Victoria M Lindsay Grayson, Elizabeth A Grabsch, Paul D R Johnson, Dianne Olden, Melissa Aberline, H Y Li, Geoffrey Hogg, Marguerite Abbott and Peter G Kerr MJA 1999; 171: 133-136 See also Ferguson, Robertson et al & Collignon Abstract - Introduction - Methods - Results - Discussion - Acknowledgements - References - Authors' details - - More articles on Infectious diseases and parasitology Abstract Objective: To screen for faecal colonisation with vancomycin-resistant enterococci (VRE) among potentially at-risk patients. Design: Infection control screening program. Setting: Monash Medical Centre (a tertiary care hospital), Melbourne, Victoria, in the seven months from June 1997. Patients: Patients in the Renal, Oncology and Intensive Care (ICU) Units. Main outcome measures: Presence of VRE in a rectal swab or faecal specimen taken at admission and at regular intervals during inpatient stay; presence of vancomycin-resistance genes (vanA, vanB and vanC) assessed by polymerase chain reaction (PCR); genetic clonality of isolates assessed by pulsed-field gel electrophoresis (PFGE). Results: 574 patients (356 renal, 134 ICU and 84 oncology) were screened; 12 were colonised with VRE -- nine renal inpatients, two having peritoneal dialysis or in-centre haemodialysis, and one ICU patient. Nine isolates were Enterococcus faecalis (seven positive for vanB and two negative for all three resistance genes) and three were Enterococcus faecium (all positive for vanB). Eight were high-level gentamicin resistant. PFGE suggested genetic clonality between the index isolate and five other isolates from renal patients. No specific clinical practice was associated with VRE colonisation. Attempts to clear rectal carriage with oral ampicillin/amoxycillin or bacitracin were of limited success. Although antibiotic prescribing in the Renal Unit was generally consistent with defined protocols, use of vancomycin and third-generation cephalosporins has been further restricted. Conclusions: Renal inpatients in our institution appear most at risk of VRE colonisation (4.6% overall) and therefore of VRE infection. Routine screening, especially of potentially high-risk patients, should be considered in major Australian hospitals. Introduction Vancomycin-resistant enterococci (VRE) were first reported in the United Kingdom, Europe and the United States in 1988,1-6 and in Australia in 1994.7 In the United States, VRE have become common and potentially fatal nosocomial pathogens; they account for 14% of enterococcal bacteraemias in intensive care patients, while VRE bacteraemia has an attributable mortality of 30%-46%.4,8 In Australia, cases of VRE infection have now been reported from hospitals in all States and Territories except the Australian Capital Territory and Tasmania.9,10In May 1997, a patient receiving renal haemodialysis at Monash Medical Centre, Melbourne, developed a VRE urinary tract infection (see Box 1). This was the first identified case of VRE infection at this institution. As the patient had had close contact with other patients and had suffered diarrhoea, we were concerned about a significant nosocomial outbreak of VRE infection. Clinical infection is almost always associated with faecal colonisation with this organism. We therefore assessed the extent of faecal VRE colonisation among renal and other high-risk patients by active screening, and isolated infected or colonised patients. We describe the results of this screening program and our infection control measures, and highlight some of the clinical issues. Methods An infection control screening program for faecal colonisation with VRE was implemented in the Renal, Oncology and Intensive Care Units at Monash Medical Centre, Melbourne, Victoria, for the seven months from June 1997. These patients were chosen as experience (local, US and European) suggested they were at greatest risk of VRE infection.4-6,8,10 Patients In the Renal Unit, screening was planned for all patients receiving care in the renal ward and in-centre haemodialysis or continuous ambulatory peritoneal dialysis and, when possible, for dialysis patients managed at home. In the Oncology and Intensive Care Units (ICU), screening was planned for an arbitrary number of 100 consecutive patients admitted to each unit. All inpatients were screened on admission and discharge. In the renal ward this was later modified to on admission and a regular day of the week (Tuesday). Patients managed by the in-centre haemodialysis unit were screened every three months, and other outpatients were screened at least once. Screening Rectal swabs were obtained using standard cotton-alginate-tipped sterile swabs from all patients except neutropenic oncology patients. In these, a perianal swab or faecal specimen was substituted for a rectal swab because of their increased risk of septicaemia after rectal trauma.13 Specimens were plated on media specifically selective for vancomycin-resistant enterococci (bile esculin azide agar with 6 µg/mL vancomycin) and cultured for up to 72 hours. Esculin-positive isolates with possible resistance to vancomycin were identified, and single colonies of each morphology type were assessed to identify Enterococcus faecalis and Enterococcus faecium (the pathogenic enterococcal species most commonly associated with vancomycin resistance). Assessment included Gram stain, tests of motility and pigment production, the pyrrolidonyl arylamidase test (Murex Diagnostics Ltd, Dartford, UK) and streptococcal latex grouping.14 VRE assessment VRE isolates were tested further for antibiotic susceptibility. Minimum inhibitory concentrations (MICs) were determined for vancomycin, teicoplanin and ampicillin and high-level gentamicin (MIC > 500 µg/mL) using the E test (AB Biodisk, Dalvagen, Sweden). Production of β-lactamase was assesssed by nitrocephin disc (Becton Dickinson Microbiology Systems, Cockeysville, MD, USA). Enterococcal species was confirmed and presence of vancomycin-resistance genes vanA, vanB or vanC was assessed by polymerase chain reaction (PCR) genetic probe using a modification of techniques described previously.10,15 Genetic similarity (ie, potential clonality) of VRE isolates was assessed by pulsed-field gel electrophoresis (PFGE) using a method modified after Miranda et al.16 Factors that were potentially associated with VRE colonisation were assessed retrospectively.6,8 Statistical analyses were by χ2 or t test. Results Patients Screening was undertaken on 574 patients -- 356 renal, 134 ICU and 84 oncology patients. Renal patients comprised: 194 of 238 inpatients in the renal ward (82%); 66 of 82 peritoneal dialysis and in-centre haemodialysis patients (80%); 94 outpatients (mostly satellite and home haemodialysis patients); and 2 of 180 renal transplantation patients. In general, consecutive oncology and ICU patients were screened; none refused screening. In the Oncology Unit, the target number (100 patients) was not attained because of a protocol lapse. VRE colonisation Faecal colonisation with VRE was found in 12 patients, including the index patient -- 11 renal patients (3% of renal patients tested) and one ICU patient (0.7% of ICU patients tested). Their characteristics are shown in Box 2. No VRE colonisation was found in oncology patients. The 11 renal patients with VRE colonisation comprised nine inpatients (9/194 [5%]) and two having peritoneal dialysis or in-centre haemodialysis (2/66 [3%]). No non-dialysis renal outpatients were colonised. VRE colonisation was found on the initial rectal swab for seven patients and after a series of negative cultures for the remaining five patients. Characteristics of VRE isolates Characteristics of the 12 VRE isolates are shown in Box 2. Nine of the 12 were E. faecalis -- seven testing positive for vanB, including the index isolate (vancomycin MICs, 12-32 µg/mL), and two testing negative for vanA, vanB and vanC (vancomycin MICs, 6 and 8 µg/mL, respectively). The other three isolates were E. faecium -- all testing positive for vanB (vancomycin MICs, 16, 64 and > 256 µg/mL, respectively). All isolates were susceptible to teicoplanin. While all E. faecalis isolates were susceptible to ampicillin, all E. faecium isolates were resistant (MICs > 256 µg/mL). Six of the nine E. faecalis isolates and two of the three E. faecium isolates had high-level resistance to gentamicin. None of the 12 produced detectable β-lactamase. Eleven isolates were assessed by PFGE. Results are shown in the Figure. Six of the seven vanB E. faecalis isolates, including the index isolate, appeared genetically similar (lanes 4-8 and index isolate in lane 9). Three of these similar isolates, plus one dissimilar vanB E. faecalis isolate (not shown), were isolated from patients who had been nursed together in a four-bed area. One of these patients (with vanB E. faecalis) had been nursed with the index patient six months before screening positive. The three E. faecium isolates and the non-ABC E. faecalis isolate (lanes 1-3 and 10, respectively) showed a variety of electrophoretic patterns on PFGE, suggesting they were genetically dissimilar from each other. Factors potentially associated with VRE Among the 12 patients with VRE colonisation, nine had received vancomycin in the previous month, four of whom had also received a third-generation cephalosporin (ceftriaxone). Information on previous antibiotics was not available for patients without VRE colonisation. Among the six renal patients found to have VRE colonisation on their initial swab (including the index case), five had been inpatients during the previous three months, compared with 127 of the 350 non-colonised patients (36%). Mean duration of preswab inpatient stay for these six colonised patients was 17.5 days (range, 0-62), compared with 2.6 days (range, 0-43) for non-colonised patients (P < 0.001). VRE control measures All patients identified with faecal VRE colonisation were nursed in a single room according to infection control guidelines.11 In January 1998, an eight-bed VRE isolation facility was established. Continued, less rigorous screening of renal inpatients identified four new cases of VRE colonisation in the following three months (not described here), but these VRE strains were dissimilar on PFGE to the previous 12 strains. Antibiotic usage patterns were also reviewed. Most antibiotic use in the Renal Unit was found to be consistent with the unit's protocols. These were amended to further restrict use of third-generation cephalosporins and glycopeptides to specific situations, such as nosocomial pneumonia and serious staphylococcal infections. In seven patients, an attempt was made to "clear" faecal VRE colonisation with either ampicillin or amoxycillin (variable doses, depending on renal function) or oral bacitracin (25 000 units four times a day for 7-14 days).17,18 In four of these patients, follow-up rectal swabs were taken, and in two (one taking ampicillin and one bacitracin) VRE was no longer detected 18 and 13 days, respectively, after therapy. At completion of the study, nine of the 12 patients with VRE colonisation had died, although, other than the index patient, none had developed VRE infection. Discussion A screening program introduced at Monash Medical Centre after identification of VRE infection in a renal patient found faecal VRE colonisation in another 10 renal patients (3% of renal patients overall) and one ICU patient (0.7%). Isolates from six of the renal patients, including the index patient, were genetically similar and probably clonal. No patients except the index patient developed VRE infection. VRE is now a major nosocomial pathogen in many US and European centres, but until recently relatively few clinical VRE infections had been reported in Australia.6,8-10 To our knowledge, this is the first Australian report of a systematic screening program for VRE among potentially at-risk patients. Our results were consistent with those of previous studies, which suggested that 10-20 patients are likely to have faecal colonisation for every case of clinical VRE infection.6,8,19,20 Although risk factors for VRE infection have been identified by US and European investigators,6,8the factors associated with VRE colonisation are less clear and may vary depending on the epidemiology of VRE in different countries. Antibiotic prescribing patterns, nosocomial transmission and use of antibiotics (eg, avoparcin) in the veterinary industry appear of varying importance in different regions.8,10 Our study did not allow valid assessment of all factors potentially associated with VRE colonisation in our patients. Nevertheless, the fact that renal patients with VRE colonisation spent significantly more days in hospital in the previous three months than patients without colonisation raises the possibility that the hospital environment or illness-related factors influenced the likelihood of VRE colonisation. Resistance to vancomycin among enterococci is generally due to presence of one of four resistance genes -- vanA, vanB, vanC and vanD. These genes result in synthesis of abnormal precursors in the peptidoglycan layer of the bacterial cell wall, thereby reducing the affinity with which vancomycin binds to this target site.8VanA is associated phenotypically with resistance to vancomycin (MIC > 64 µg/mL) and teicoplanin (MIC > 16 µg/mL), and is the most common genotype found in Europe and some centres in the US.8VanB is associated with medium-level resistance to vancomycin (MIC > 4 µg/mL) but susceptibility to teicoplanin. Consistent with our findings, it is the predominant genotype noted in Australia.9,10VanC is associated with naturally occurring low-level resistance to vancomycin and susceptibility to teicoplanin among less pathogenic enterococcal species, while vanD, which is phenotypically similar, has been occasionally noted in some E. faecium isolates.8,21,22 It is possible that our two non-ABC E. faecalis isolates contain vanD, but we are currently unable to test for this gene. Presence of faecal VRE colonisation among 5% of renal inpatients at our institution (3% of renal patients overall) was higher than expected, but suggested that nosocomial transmission of VRE was not yet a widespread problem. Nevertheless, our PFGE data suggested that six of the seven vanB E. faecalis strains were clonal, raising infection control issues for the Renal Unit. As reported previously,17,18 we found attempts to clear faecal VRE carriage with antibiotic therapy were unsuccessful and not worthwhile. The screening program and establishment of VRE isolation facilities to readily cohort and barrier-nurse patients with VRE colonisation appeared to assist in limiting nosocomial VRE transmission, while continuing to provide medical care for patients in a compassionate manner. The incidence of faecal VRE colonisation that we found among high-risk patients at our institution suggests that routine screening for faecal VRE colonisation should now be considered by other similar Australian hospitals. Acknowledgements We wish to acknowledge the contribution of the nursing staff of the Renal, Oncology and Intensive Care Units and the Outpatient Department in obtaining rectal cultures. References Leclercq R, Derlot E, Duval J, Courvalin P. Plasmid-mediated resistance to vancomycin and teicoplanin in Enterococcus faecium. N Engl J Med 1988; 319: 157-161. Uttley AHC, Collins CH, Naidoo J, George RC. Vancomycin-resistant enterococci. Lancet 1988; 1: 57-58. Clark NC, Cooksey RC, Hill BC, et al. Characterization of glycopeptide-resistant enterococci from U. S. hospitals. Antimicrob Agents Chemother 1993; 37: 2311-2317. Centers for Disease Control. Nosocomial enterococci resistant to vancomycin -- United States, 1989-1993. MMWR Morb Mortal Wkly Rep 1993; 42: 597-599. Frieden TR, Munsiff SS, Low DE, et al. Emergence of vancomycin-resistant enterococci in New York City. Lancet 1993; 342: 76-79. Boyce JM. Vancomycin-resistant enterococcus. Detection, epidemiology, and control measures. Infect Dis Clin North Am 1997; 11: 367-384. Kamarulzaman A, Tosolini FA, Boquest AL, et al. Vancomycin resistant Enterococcus faecium infection in a liver transplant recipient [abstract]. Aust N Z J Med 1995; 25: 560. Eliopoulos GM. Vancomycin-resistant enterococci. Mechanism and clinical relevance. Infect Dis Clin North Am 1997; 11: 851-865. Bell J, Turnidge J, Coombs G, O'Brien F. Emergence and epidemiology of vancomycin-resistant enterococci in Australia. Commun Dis Intell 1998; 22: 249-252. Bell JM, Paton JC, Turnidge J. Emergence of vancomycin-resistant enterococci in Australia: phenotypic and genotypic characteristics of isolates. J Clin Microbiol 1998; 36: 2187-2190. Standing Committee on Infection Control (SCIC), Department of Human Services, Victoria. Guidelines for the management of patients with confirmed vancomycin-resistant enterococci (VRE) infection/colonisation. Melbourne: Department of Human Services, 1996. Moellering RC Jr. The Garrod lecture. The enterococcus: a classic example of the impact of antimicrobial resistance on therapeutic options. J Antimicrob Chemother 1991; 28: 1-12. Weinstein JW, Tallapragada S, Farrel P, Dembry L-M. Comparison of rectal and perirectal swabs for detection of colonisation with vancomycin-resistant enterococci. J Clin Microbiol 1996; 34: 210-212. Facklam RR, Sahm DF. Enterococcus. In: Murray PR, Baron EJ, Pfaller MA, et al, editors. Manual of clinical microbiology. 6th ed. Washington: ASM Press, 1995: 308-314. Dutka-Malen A, Evers S, Courvalin P. Detection of glycopeptide resistance genotypes and identification to the species level of clinically relevant enterococci by PCR. J Clin Microbiol 1995; 33: 24-27. Miranda AG, Singh KV, Murray BE. A fingerprinting of Enterococcus faecium by pulsed-field gel electrophoresis may be a useful epidemiologic tool. J Clin Microbiol 1991; 29: 2752-2757. O'Donovan CA, Fan-Havard P, Tecson-Tumang FT, et al. Enteric eradication of vancomycin-resistant Enterococcus faecium with oral bacitracin. Diagn Microbiol Infect Dis 1994; 18: 105-109. Chia JKS, Nakata MM, Park SS, et al. Use of bacitracin therapy for infection due to vancomycin-resistant Enterococcus faecium. Clin Infect Dis 1995; 21: 1520. Jordens JZ, Bates J, Griffith DT. Faecal carriage and nosocomial spread of vancomycin-resistant Enterococcus faecium. J Antmicrob Chemother 1994; 34: 515-528. Montecalvo MA, deLaencastre H, Carraher M, et al. Natural history of colonization with vancomycin-resistant Enterococcus faecium. Infect Control Hosp Epidemiol 1995; 16: 680-685. Leclercq R, Courvalin P. Resistance to glycopeptides in enterococci. Clin Infect Dis 1997; 24: 545-556. Perichon B, Reynolds P, Courvalin P. VanD-type glycopeptide-resistant Enterococcus faecium BM4339. Antimicrob Agents Chemother 1997; 41: 2016-2018. (Received 15 Dec 1998, accepted 23 Apr 1999) Authors' details Infectious Disease and Clinical Epidemiology Department, Monash Medical Centre, Melbourne, VIC. M Lindsay Grayson, MD, FRACP, Director, and Professor of Epidemiology and Preventive Medicine, Monash University, Melbourne, VIC; Elizabeth A Grabsch, BSc, GradDipClinEpid, Infection Control Scientist; Paul D R Johnson, PhD, FRACP, Infectious Disease Physician, and Senior Lecturer, Microbiology Department, Monash University, Melbourne, VIC; Dianne Olden, PhD, Research Scientist. Infection Control Unit, Monash Medical Centre, Melbourne, VIC. Melissa Aberline, RN, BSc, Infection Control Nurse. Microbiological Diagnostic Unit, Melbourne University, Melbourne, VIC. H Y Li, MMed, Scientist; Geoffrey Hogg, FRACP, FRCPA, Director. Nephrology Department, Monash Medical Centre, Melbourne, VIC. Marguerite Abbott, RN, BAppSci, Nurse Director; Peter G Kerr, PhD, FRACP, Deputy Director. Reprints will not be available from the authors. Correspondence: Professor M L Grayson, Infectious Diseases and Clinical Epidemiology Department, Monash Medical Centre, 246 Clayton Road, Clayton, VIC 3168. Email: Lindsay. GraysonATmed.monash.edu.au Pulsed-field gel electrophoresis of vancomycin-resistant enterococcal isolates. Lanes 1-3: E. faecium (vanB) isolates from renal patients. Lanes 4-9: E. faecalis (vanB) isolates from renal patients, including index isolate (lane 9). Lane M: Molecular weight markers. Lane 10: E. faecalis (non-ABC) isolate from renal patient. Back to text 1: Case history of the index patient In May 1997, a 26-year-old woman presented to Monash Medical Centre with a fever and urinary tract infection caused by vancomycin-resistant enterococci (VRE). She had faecal colonisation with the same VRE strain. She had endstage renal failure requiring in-centre haemodialysis three times a week. Ten days later, the patient presented in status epilepticus with faecal incontinence that led to substantial faecal contamination of the Emergency Department and the in-centre Haemodialysis Unit. Appropriate cleaning protocols11 were implemented in each area, and limited environmental cultures suggested no contamination. Five months later, the patient developed symptomatic VRE bacteraemia after surgical insertion of a femoral Goretex arteriovenous fistula. The VRE isolate was phenotypically identical to the initial urinary isolate. As it had high-level gentamicin resistance, she was treated with continuous-infusion ampicillin, continuing for 10 weeks because of the possibility of graft sepsis.12 The patient continued to show faecal VRE colonisation until her death (not related to VRE) in September 1998. Back to text

Elizabeth A Grabsch · Dianne Olden · Melissa Aberline · H Y Li · Geoffrey Hogg · Marguerite Abbott · Peter G Kerr

Research Australian suicide trends 1964-1997: youth and beyond? Christopher H Cantor, Kerryn Neulinger and Diego De Leo MJA 1999; 171: 137-141 Abstract - Introduction - Methods - Results - Discussion - Acknowledgements - References - Authors' details - - More articles on Psychiatry Abstract Objective: To examine Australian suicide rates across all ages, and compare Australian rates with those of other Western nations. Design: Australian Bureau of Statistics data were used to examine Australian suicide trends, 1964-1997, by age and sex. For comparison, suicide rates of 22 other Western nations, 1990-1994, were obtained from the World Health Organization. Results: Australian suicide rates for males 15-24 years and 25-34 years rose from 1964-1997. Comparable rates for females showed no significant change. Suicide rates for several of the older age and sex groups declined over this period. Comparison with suicide rates of other Western nations showed that, while Australian youth suicide rates are relatively high, this is not the case for older age groups. Australian suicide rates are higher than those in the European nations of origin of our major migrant groups, but similar to those of other Western nations also recently colonised by Europeans (Canada, the United States and New Zealand). Conclusions: Priorities for suicide prevention in Australia are correctly concentrated on youth, but the targeted age range should be extended to include men aged 25-34 years. A comprehensive policy should also not neglect the needs of other age groups. Further epidemiological study of national and international data may suggest new approaches to suicide prevention. Introduction It is well known that there is a high suicide rate in Australian males aged 15-24 years, and previous and present governments have instituted committees to address youth suicide rates.1 There have been only sporadic reports of suicide rates in other age groups,2 including suggestions of high rates in elderly people.2,3We present suicide rates and trends across all ages, placing them in context by comparing them with rates from other Western nations. The national focus on youth suicide to the relative exclusion of older age groups is examined in the light of these data. Methods Australian suicide and population data for 1964-1997 were obtained from the Australian Bureau of Statistics (ABS) and annual suicide rates (per 100 000 estimated mean resident population) were calculated (see Appendix). These were plotted by 10-year age groups and sex over time. Change over time was examined by comparing the suicide rate in 1964-1966 (rates were averaged to smooth out random fluctuations within one-year periods) with the suicide rate in 1995-1997; 95% confidence intervals were calculated for these two time periods to assess the significance of change. Variations in suicide rate trends over this 34-year period were assessed by calculating the series autocorrelation suicide rates and the values of Pearson's correlation coefficient. Years of potential life lost (YPLL) were calculated for 10-year age and sex groups for the most recent year (1997). The formula used was adapted from that used by the Australian Bureau of Statistics4 in order to incorporate age groups (Box 1). The Australian population at 30 June 1991 was chosen as the standard population, as suggested by the Australian Bureau of Statistics. International data were obtained from the World Health Organization (WHO)5-8 and Lester and Yang.9 Mean suicide rates were calculated for 23 Western countries (including Australia), averaged over 1990-1994, for 10-year age and sex groups. These 10-year age and sex rates were then ranked to compare Australia with other nations. Results Figures 1-8 show suicide rates over time for sex and 10-year age groups, as well as all ages combined. Males Suicide rates overall and for those aged 15-24 and 25-34 years increased significantly over time (1964-1966 to 1995-1997). The rise in suicide rates from 1973-1997 for men aged 25-34 years was of similar magnitude to the rise for those aged 15-24 years (Figures 2 and 3). Suicide rates in the older age groups (35 to 75 years and over) declined (for men aged 35-44 years and 75 years and over these declines were not statistically significant). Significant autocorrelations (P < 0.01) were found in each age group and for all ages combined. Females Suicide rates overall and for each age group from 25 to 75 years and over declined significantly (1964-1966 to 1995-1997). There was no change in suicide rates for females aged 15-24 years. Significant autocorrelations (P < 0.01) were found for all ages combined and for most age groups (the exceptions were those aged 15-24 years and 75 and over). Years of potential life lost by suicide in 1997 are shown in Box 2. Boxes 3a and 3b list the mean suicide rates (per 100 000 population) by age group and sex for 23 Western nations, including Australia. Discussion The priority accorded to youth suicide prevention by successive Australian governments is supported by our data. In the last three decades, when overall suicide rates were falling, the suicide rate of males 15-24 years more than doubled. Suggestions of a plateauing of this rate since 1989 are not supported by the latest figures for 1997, although it is premature to draw conclusions either way. (It should also be noted that at the time of submission of this article there were significant concerns about the validity of the 1997 suicide rate figures. For example, major rises in rates in New South Wales and Victoria were absent from Queensland -- the next most populous State -- suggesting the possibility of interstate data collection influences.) From a policy perspective, the most important finding is that suicide rate increases since 1973 for men aged 25-34 years have paralleled those of 15-24 year olds, with even less evidence of a recent plateau. Despite having shorter life expectancies, the toll in potential years of life lost (in 1997) was greatest for 25-34 year olds. These years are also the peak years for early parenthood, making the likelihood of there being bereaved offspring greater than in the younger group. These findings suggest that priority initiatives for youth suicide prevention should be extended to those aged 15-34 years, if not further (the high level of potential years of life lost continues to 44 years). The Federal Government's recently announced National Suicide Prevention Policy provides for ages beyond youth, by recognising that rising rates are not confined to the age group 15-24 years and that high suicide rates in the elderly should not be overlooked. It is unclear whether or to what extent the suicide phenomenon in boys and young men might be a cohort effect -- a "damaged" generation. The later commencement of the rise in suicide rate in men aged 25-34 years and the, as yet, absence in this age group of the possible 1990s plateau could be seen as consistent with this. Rates in females aged 15-24 years showed no convincing rise. It has previously been noted that suicide rates in females aged 15-19 years rose modestly between 1960 and 1989, while rates for those aged 20-24 years showed no overall change over this period.10 However, much of the rise in females aged 15-19 years was in the early 1960s, predating the rise in rates of 15-19-year-old males, but coinciding with a general, transient rise in suicide rates in females of all ages. Perhaps the most important observation regarding young females in the present study is that their suicide rates remained relatively static at a time when rates for most other female age groups declined. Suicide rates in children under 15 years of age have not significantly altered over time, although the small numbers involved do not lend confidence to rate estimates. The recording of suicide in children under the age of 10 years is affected by the different criteria used by the Australian Bureau of Statistics for determining suicide: it is considered that children have a limited ability to form an intent to suicide. Hence, more explicit evidence of suicidal intent is required,11 usually a coroner having stated explicitly that suicide was the cause of death. As most State and Territory Coroners' Acts discourage if not prohibit such findings (for all ages), such explicit statements are unlikely.11 Beyond 35 years in women, and 45 years in men, and up to 75 years and over for both sexes, Australian suicide trends are striking. In both sexes there have been major declines, and for much of the period these declining male and female rates have been parallel, quite unlike the trends for 15-24 year olds. It is unclear whether these declining rates have as yet plateaued. Even if this were not the case, it is inevitable that these declining rates will shortly level out and increase again (from the trends over the complete period a number of these age rates would reach zero in the early decades of the next century). In the early 1960s, suicide rates, particularly in women and in older age groups, rose in association with an epidemic of barbiturate poisoning, subsequently to fall in the late 1960s, reaching rates in the early 1970s similar to those before this epidemic.12,13 The causal factors determining the subsequent decline in rates of those between 35 and 74 years are yet to be determined. It is possible that such determination might yield valuable clues as to the causes and possible methods of preventing suicide in young people. International comparisons should be interpreted conservatively, as variations in data collection systems will account for some of the variations. With these reservations in mind, among 23 Western nations, Australian suicide rates for 15-24 year olds ranked fourth for males and eighth for females (Boxes 3a and 3b). Australian rates for both sexes between the ages of 35 and 74 years, as well as declining, compared favourably internationally (ranking in the middle of or below those of other Western nations). Although Australian suicide rates for those over 75 years were not declining, they nevertheless again compared favourably from an international perspective. Whereas the Australian suicide rate for 15-24-year-old males was over half that of the highest-placed nation (Finland), the corresponding ratio for males 75 years and over was about one-sixth. For females, these comparisons were even more marked. Nevertheless, a comprehensive policy should not overlook the needs of men over 75 years, especially as, in Australia, suicide rates in this age group surpass those of all other age groups. The Western nations with lower suicide rates include those that provided most of the early immigrants to Australia: Greece, Italy (except in the elderly), the United Kingdom and Ireland. The highest suicide rates were in mainland Western Europe (especially Hungary) and Scandinavia (especially Finland). Some of these nations, for example Denmark, have profiles radically different from that of Australia, with lower youth suicide rates but much higher rates in those over 35 years of age. Canada and, to a lesser extent, the United States and New Zealand are the nations with suicide rate profiles most similar to Australia's. All four of these nations have common characteristics of European migration, a comparatively short history (apart from their indigenous populations), geographical isolation, climatic extremes and more.10 They serve as potentially valuable nations to study shared characteristics that may affect suicide rates. Studies have yet to provide explanations for these similarities and differences which have often been overlooked, largely as a consequence of the worldwide lack of epidemiological mental health data. While differences in prevalence of psychiatric disorders are potentially relevant, the magnitude of the different international rates suggests other factors may operate. Hungary and Finland share common cultural origins and high suicide rates. It has been suggested that Finland's high suicide rates in young men may relate to cultural expectations that men should be tough and resilient14 -- a suggestion that might be equally relevant in Australia, New Zealand, the United States and Canada. It is possible that a better understanding of cultural influences and how to positively modify them might be relevant to suicide prevention. Acknowledgements The study was funded by Griffith University and Queensland Health through their co-funding of the Australian Institute for Suicide Research and Prevention. We thank the Australian Bureau of Statistics for providing information on suicide and population numbers. References Commonwealth Department of Health and Family Services. Youth suicide in Australia: a background monograph. 2nd edition. Canberra: AGPS, 1997. Snowdon J. Suicide rates and methods in different age groups: Australian data and perceptions. Int J Geriat Psychiatry 1997; 12: 253-258. Hassan R, Carr J. Changing patterns of suicide in Australia. Aust N Z J Psychiatry 1989; 23: 226-234. Australian Bureau of Statistics. Causes of death, Australia, 1996: 73. (Catalogue No. 3303.0.) World Health Organization. World Health Statistics Annual 1992. Geneva: WHO, 1993. World Health Organization. World Health Statistics Annual 1993. Geneva: WHO, 1994. World Health Organization. World Health Statistics Annual 1994. Geneva: WHO, 1995. World Health Organization. World Health Statistics Annual 1995. Geneva: WHO, 1996. Lester D, Yang B. Suicide and homicide in the twentieth century: Changes over time. Commack, NY: Nova Science, 1998: 165-204. Cantor CH, Leenaars AA, Lester D, et al. Suicide trends in eight predominantly English-speaking countries 1960-1989. Soc Psychiatry Psychiatr Epidemiol 1996; 31: 364-373. Cantor CH, Neulinger K, Roth J, Spinks D. The epidemiology of suicide and attempted suicide among young Australians: a report to the National Health and Medical Research Council: Australian Institute for Suicide Research and Prevention. In press. Oliver G, Hetzel BS. An analysis of recent trends in suicide rates in Australia. Int J Epidemiol 1973; 2: 91-101. Whitlock FA. Suicide in Brisbane, 1956-1973: The drug-death epidemic. Med J Aust 1975; 1: 737-743. Retterstol N. Suicide in the Nordic countries. Psychopathology 1992; 25: 254-265. (Received 14 Jan, accepted 7 Jun, 1999) Authors' details Australian Institute for Suicide Research and Prevention, Griffith University, Nathan, QLD. Christopher H Cantor, FRANZCP, MRCPsych, Senior Research Psychiatrist; Kerryn Neulinger, BBehSc, GradDipPsych, Research Assistant; Diego De Leo, MD, PhD, Director. Reprints will not be available from the authors. Correspondence: Dr C H Cantor, Australian Institute for Suicide Research and Prevention, Griffith University, Nathan, QLD 4111. Email: C. CantorATmailbox.gu.edu.au Back to textBack to text Back to text 3a: Mean male suicide rates (per 100000 population) for 23 Western countries by 10-year age groups, 1990-199415-24 yearsFinland 41.4 New Zealand 39.0 Switzerland 25.8 Australia 25.7 Canada 25.2 Norway 24.9 Austria 24.3 USA 21.9 Hungary 20.1 Scotland 19.0 Ireland 18.3 N Ireland 17.6Belgium 15.7 France 15.3 Germany 14.0 Sweden 13.4 Denmark 13.0 Engl/Wales 11.1 Netherlands 9.3 Spain 7.0 Italy 6.1 Portugal 5.8 Greece 4.0 25-34 yearsFinland 60.7 Hungary 54.4 Switzerland 32.7 New Zealand 32.0 France 32.0 Belgium 30.5 Austria 30.3 Australia 29.0 Canada 29.0 Ireland 27.1 Denmark 26.4 Norway 26.1Scotland 26.1 USA 24.6 Sweden 23.9 N Ireland 22.4 Germany 21.3 Engl/Wales 16.3 Netherlands 15.9 Portugal 13.2 Spain 10.6 Italy 10.3 Greece 5.635-44 yearsHungary 82.0 Finland 67.8 France 40.1 Denmark 38.1 Austria 37.2 Belgium 35.6 Switzerland 33.0 Sweden 29.3 Canada 27.3 Norway 26.9 Scotland 26.2 Germany 26.0Australia 25.2 New Zealand 23.9 USA 23.5 Ireland 22.9 Netherlands 17.7 Engl/Wales 17.4 N Ireland 15.5 Portugal 11.8 Italy 10.6 Spain 9.4 Greece 5.945-54 yearsHungary 95.1 Finland 64.1 Denmark 47.5 Austria 41.5 France 40.1 Switzerland 39.8 Belgium 36.2 Sweden 31.9 Germany 31.1 Norway 28.8 Canada 25.6 Scotland 24.2Australia 24.2 New Zealand 24.2 USA 23.1 Ireland 19.6 Netherlands 16.7 Engl/Wales 16.2 N Ireland 15.1 Portugal 14.6 Italy 12.6 Spain 11.9 Greece 6.755-64 yearsHungary 84.6 Finland 57.3 Austria 46.7 Denmark 42.6 Switzerland 41.9 belgium 38.9 France 38.1 Germany 32.2 Sweden 30.7 Norway 28.8 Ireland 25.9 USA 25.0 Canada 24.2 New Zealand 23.2 Australia 22.9 Portugal 21.5 Netherlands 18.6 Scotland 18.1 N Ireland 17.4 Spain 17.4 Italy 17.1 Engl/Wales 12.8 Greece 7.865-74 yearsHungary 92.5 Austria 61.1 Belgium 50.4 Switzerland 47.4 France 47.1 Denmark 46.4 Finland 45.9 Germany 35.9 Sweden 33.7 USA 30.9 Norway 30.7 Portugal 30.1 Australia 24.4 Spain 23.2 Italy 22.9 Canada 22.1 New Zealand 21.2 Netherlands 19.7 Ireland 18.3 Scotland 14.3 N Ireland 12.8 Engl/Wales 11.9 Greece 10.175+ yearsHungary 183.0 Austria 118.0 France 103.0 Belgium 98.6 Switzerland 89.8Germany 86.1 Denmark 74.9 Finland 71.9 Portugal 59.1 USA 55.4Sweden 51.9 Spain 47.8 Italy 44.3 Netherlands 35.4 Australia 32.8 Norway 31.8 New Zealand 29.8 Canada 28.9 Engl/Wales 17.1 Scotland 16.0Greece 15.8 Ireland 13.8 N Ireland 13.3Back to text Appendix: Identifying suicide deaths in Australia11 Coding deaths as suicide relies on an interaction between the Australian Bureau of Statistics (ABS) and State and Territory coroners and government medical officers. The six States and two Territories of Australia each have different Coroners' Acts. Most States and Territories discourage coroners from making formal pronouncements about suicide. Western Australia and New South Wales are the only States that routinely use suicide verdicts. ABS receives information from coroners that is generally sufficient for coding in most cases. Deaths that were in reality suicides, but were not deemed as such by coroners or the ABS, will most likely be coded as undetermined deaths or accidents, and involve similar causes as suicide deaths (eg, poisoning). There has been a rise in both undetermined and accidental deaths in recent years. However, the impact of these uncertainties is unlikely to be sufficient to greatly alter our overall study findings. Back to text 3b: Mean female suicide rates (per 100000 population) for 23 Western countries by 10-year age groups, 1990-199415-24 yearsFinland 7.5 Austria 6.2 Hungary 6.2 New Zealand 6.2 Sweden 5.9 Switzerland 5.8 Norway 5.5 Australia 5.1 Belgium 5.1 Canada 4.9 France 4.5 USA 3.8Netherlands 3.7 Scotland 3.7 Germany 3.5 Denmark 3.3 Ireland 2.5 N Ireland 2.4 Portugal 2.2 Engl/Wales 2.1 Italy 1.8 Spain 1.7 Greece 0.725-34 yearsFinland 12.0 Belgium 11.8 Hungary 11.6 Sweden 10.1 Switzerland 9.0 France 9.0 Scotland 8.3 Austria 8.0 Denmark 7.7 New Zealand 7.3 Netherlands 7.2 Norway 7.1Ireland 6.7 Australia 6.6 Canada 6.4 Germany 5.7 Germany 5.7 N Ireland 3.9 Portugal 3.5 Engl/Wales 3.5 Italy 2.9 Spain 2.6 Greece 1.435-44 yearsHungary 20.3 Finland 17.4 Denmark 15.7 Belgium 14.5 Switzerland 13.5 France 13.0 Austria 12.1 Sweden 11.8 Norway 9.8 Netherlands 9.6 Canada 8.1 Germany 7.7New Zealand 6.9 Scotland 6.8 N Ireland 6.8 Australia 6.6 USA 6.6 Ireland 4.8 Engl/Wales 3.9 Italy 3.9 Portugal 3.8 Spain 3.0 Greece 1.3 45-54 yearsHungary 26.5 Denmark 25.5 Finland 20.4 Belgium 18.2 Austria 17.1 Switzerland 16.7 France 16.5 Sweden 15.0 Germany 12.1 Norway 11.5 Netherlands 9.5 N Ireland 9.4New Zealand 8.9 Canada 8.1 USA 7.3 Scotland 7.2 Australia 7.0 Ireland 6.8 Portugal 5.0 Italy 4.9 Engl/Wales 4.7 Spain 3.9 Greece 2.355-64 yearsDenmark 28.5 Hungary 28.0 Belgium 17.9 France 17.6 Finland 17.5 Austria 17.4 Switzerland 17.0 Sweden 15.4 Germany 12.9 Norway 12.0 Netherlands 10.9 New Zealand 7.7Ireland 7.7 Australia 6.9 USA 6.8 Italy 6.8 Scotland 6.7 Canada 6.4 Portugal 6.2 Spain 6.0 N Ireland 4.8 Engl/Wales 4.7 Greece 2.4 65-74 yearsHungary 37.6 Denmark 31.5 Belgium 23.5 Switzerland 19.8 Austria 18.5 France 17.9 Germany 16.7 Sweden 13.5 Finland 13.3 Norway 12.4 Netherlands 10.4 Spain 8.8Portugal 8.1 Italy 8.0 New Zealand 6.6 Australia 6.6 Ireland 6.4 Scotland 6.4 USA 6.2 Canada 6.1 Engl/Wales 5.2 N Ireland 3.9 Greece 2.875+ yearsHungary 67.3 Denmark 30.2 Austria 28.5 Germany 26.4 France 25.3Belgium 24.2 Switzerland 23.0 Sweden 14.2 Portugal 12.2 Netherlands 12.1Spain 11.9 Finland 9.6 Italy 9.3 Norway 9.2 Australia 8.0Scotland 6.0 USA 6.0 Engl/Wales 5.9 Canada 4.7 New Zealand 4.3 Greece 3.4 Ireland 3.0 N Ireland 2.5Back to text

Christopher H Cantor · Kerryn Neulinger

Editorial Asthma in general practice: action plans or planned actions There's more than one way to implement effective asthma management in general practice Peter G Gibson MJA 1999; 171: 67 Good general practice is integral to effective asthma management. Asthma prevalence and hospitalisations are high, and, although we don't yet know how to reverse the rise in asthma prevalence, we know that severe exacerbations can be reduced by inhaled corticosteroid therapy and by effective education that involves an individualised, written action plan and regular medical review.1The burden of illness from asthma is concentrated in general practice, but studies of asthma education have mainly been conducted in hospitals. We need to transfer the improvements seen in these randomised trials to Australian general practice. Several approaches have been tried, including public health initiatives conducted by the National Asthma Campaign (NAC), practice audit,2,3 and nurse-run asthma clinics.4 In this issue of the Journal, Abdulwadud et al provide data showing just how different asthma can be in general practice to asthma in a specialist setting.5 Predictably, asthma is less severe in general practice.5 Medication use and understanding about asthma are similar, but fewer general practice patients have a written action plan and understand how to respond to an asthma emergency. Thus, for Australia, where inhaled corticosteroid use is already high, educational interventions in general practice should focus on the early management of exacerbations using written action plans and on smoking cessation. Also in this issue, Heard et al report their trial of a combined nurse educator/general practitioner asthma clinic.6 As part of the intervention, patients attending the asthma clinic received action plans and education about asthma. Patients in the control group also received action plans from their doctors, so that by the end of the study about 70% of patients had a written action plan and 80% were taking inhaled preventer medication. Asthma morbidity was reduced in both groups, and the asthma clinic did not prove to be any better than regular review of asthma by a general practitioner. Presumably outcomes in the control group improved because the participating general practitioners, who were caring for patients in both intervention and control groups, modified their behaviour by appropriately following proven guidelines. This problem is common in studies of asthma education, where blinding is seldom used and participation in the study is enough to improve management. Thus, this article compares asthma clinics to enhanced medical care, and a direct comparison with usual medical practice is lacking. What can we learn from this? Firstly, that the improvements in asthma mordibity that follow education and medical review can be achieved in general practice in Australia, and can be achieved in several ways. Regular review of asthma control and issuing an action plan by a general practitioner can be as effective as a special asthma clinic. General practitioners can now be confident that modifying their practice will improve asthma control. We know that this is needed, as surveys indicate that, although more people use action plans since the NAC, their use is still unacceptably low.7 General practitioners can choose between setting up an asthma clinic or implementing a structured program of regular review in their practices. The NAC and asthma liaison officers within the divisions of general practice are implementing suggested schemes at present. Regular review could turn into complacency, where a general practitioner feels that his or her current practice is satisfactory and that people with asthma are well managed. The best way to avoid this is to conduct regular quality control of asthma management. There are some simple interventions with asthma audit that do reap rewards for patients and doctors alike.2,3 Special asthma clinics also need to conduct regular quality control of their services, as when they provide asthma education without general practitioner consultations and action plans there is no benefit to patients.8 Structured asthma management programs, whether by systematic regular review or a special asthma clinic, will need to be adapted to the organisational structure of the practice. This can be done fairly simply in practices with several doctors. It is more difficult to do so in solo practices, but here it is even more important, as patients from small practices have a higher admission rate for asthma than those from larger practices.9 General practitioners need to adapt the National Asthma Campaign's six-point asthma management plan to their practices and ensure that all patients with asthma understand their disease, have a set of written instructions about when and how to treat exacerbations (all people with asthma are at risk of exacerbations), and are prescribed optimal therapy to control symptoms and exacerbations. Special asthma clinics or a system of regular education and review will suffice, provided that the processes and outcomes are monitored to ensure success. Organising the process of care for asthma in general practice and defining how to do it may be just as important as educating the doctor about what to do. Most general practitioners understand asthma management. Effective implementation is the issue. Peter G Gibson Staff Specialist, Respiratory Medicine John Hunter Hospital, Newcastle, NSW Gibson PG, Wilson AJ, Coughlan J, et al. The effects of self-management asthma education and regular practitioner review in adults with asthma. In: Cates C, DuCharme F, Gibson PG, et al. Airways module. Cochrane database of systematic reviews, issue 4. Oxford, UK: Update Software, 1998. Bryce FP, Neville RG, Crombie IK, et al. Controlled trial of an audit facilitator in diagnosis and treatment of childhood asthma in general practice. BMJ 1995; 310: 838-842. Feder G, Griffiths C, Highton C, et al. Do clinical guidelines introduced with practice based education improve care of asthmatic and diabetic patients? A randomised controlled trial in general practices in east London. BMJ 1995; 311: 1473-1478. Charlton I, Charlton G, Bloomfield J, et al. Audit of the effect of a nurse-run asthma clinic on workload and patient morbidity in general practice. Br J Gen Pract 1991; 41: 227-231. Abdulwadud OA, Abramson MJ, Light L, et al. Comparison of patients with asthma managed in general practice and in a hospital clinic. Med J Aust 1999; 171: 72-75. Heard AR, Richards IJ, Alpers JH, et al. Randomised controlled trial of general practice based asthma clinics. Med J Aust 1999; 171: 68-71. Comino EJ, Mitchell CA, Bauman A, et al. Asthma management in eastern Australia, 1990 and 1993. Med J Aust 1996; 164: 403-406. Premaratne UN, Sterne JAC, Marks GB, et al. Clustered randomised trial of an intervention to improve the management of asthma: Greenwich asthma study. BMJ 1999; 318: 1251-1255. Griffiths C, Sturdy P, Naish J, et al. Hospital admissions for asthma in East London: associations with characteristics of local general practices, prescribing, and population. BMJ 1997; 314: 482-486.

Peter G Gibson

Research Randomised controlled trial of general practice based asthma clinics Adrian R Heard, Ian J Richards, John H Alpers, Louis S Pilotto, Brian J Smith and Julie A Black MJA 1999; 171: 68-71 See also Gibson & Abdulwadud et al Abstract - Introduction - Methods - Results - Discussion - Acknowledgements - References - Authors' details - - More articles on General practice and primary care Abstract Objective: To compare the effects on asthma morbidity of asthma clinics based in general practice with standard general practice care. Design and setting: A randomised controlled trial in eight general practices. Patients, general practitioners and outcomes assessors were not blinded to treatment allocation. Participants: 195 patients with asthma aged 5-64 years; 191 completed the trial. Intervention: Three asthma clinic sessions over six months involving nurse counselling, education about asthma management, spirometry and consultation with the general practitioner. Main outcome measures: Patients reporting days lost from work or school, number of days lost, the presence of morning or nocturnal asthma symptoms, use of an action plan, medication use, current smoking, hospitalisation, and emergency visits. Results: Asthma clinics were associated with a greater reduction in nocturnal symptoms, an increase in the ownership of peak flow meters and an increase in the number of patients commencing or resuming smoking. Both control and intervention groups showed similar improvement in days lost from work or school, the presence of symptoms, use of an action plan and taking reliever medication. Conclusion: Our study does not show that asthma clinics are more effective than standard general practice care in reducing asthma morbidity. It is uncertain how much of the improvement in outcomes was due to the asthma clinic, the influence of the study itself upon patients and practitioners, or other factors, such as the tendency for a patient's asthma management to improve over time. Introduction Asthma is the most common chronic disease in children and a leading cause of morbidity in adults in Australia.1,2 Asthma clinics may be a way of improving comanagement practices between health professionals and people with asthma. They involve trained nurses conducting asthma education, peak flow readings and spirometry with asthma patients, combined with a brief general practitioner review. Asthma clinics are an integral part of general practice in the United Kingdom, but have yet to gain acceptance in Australia.3 One reason for this might be that studies have not provided conclusive evidence that these clinics reduce morbidity.4-9 This may be because regular general practitioner review10 and/or asthma education11 is the key element in reducing asthma morbidity, and therefore only studies comparing asthma clinics with patients receiving irregular reviews are likely to show significant differences in outcomes. In this study, we used a randomised controlled trial design to test whether asthma clinics (intervention) were more effective in reducing morbidity from asthma than standard medical treatment (control). The clinics used the Patient Management System, a recognised model for operating asthma clinics.12 Methods Ethical approval Ethics approval for the study was obtained through the Royal Australian College of General Practitioners' ethics committee at the RACGP national office. Recruitment A previous major study of asthma self-management found a 50% reduction in the number of people reporting at least one day lost from work.6 We calculated that to detect a similar reduction in a study with 80% power and a 95% confidence interval a sample size of 170 was required. Patients were recruited from eight general practices staffed by 42 general practitioners. Consent was sought from all people with asthma aged 5-64 years attending these practices during three months of 1997. Subjects were individually randomised within practices into intervention and control groups. This meant that each participating general practitioner potentially saw both intervention and control patients. A randomisation chart was set up for each participating practice at Asthma South Australia, and general practitioners and the asthma educators were informed of a subject's treatment allocation before the baseline interview. Thus, patients, doctors and outcome assessors were not blinded as to who had received the intervention and who were controls. Intervention Each general practice operated one three-hour asthma clinic session per week. The asthma educators in the trial were practising registered nurses with extensive experience in respiratory care. Their component of the clinic session involved education in asthma management strategies, including a written asthma management plan, spirometry and instruction on using peak flow meters, inhalers and an asthma diary card. The session ended with a consultation by the general practitioner. Each asthma clinic patient was asked to attend three asthma clinic sessions within the six months of the study. Gathering outcomes data We conducted a telephone interview using adapted questions from the Southampton Morbidity Index13 and questions relating to clinical practice14 at the beginning and end of the study. Our outcome measures included number of patients reporting days lost from work or school, number of days lost, use of an action plan, medication use, current smoking, the level of morning or nocturnal asthma symptoms, hospitalisation, emergency visits, and number of home visits by the general practitioner. Data analysis All data were analysed on an intention-to-treat basis. For analysis, we categorised morning and nocturnal symptoms as "at least weekly" or "less frequently/never" to distinguish regular from irregular symptoms and to maintain adequate numbers of participants in each of the two categories. We analysed the number of days lost from work with both linear and logistic regression. Linear regression was conducted on the difference in the number of days lost from work between baseline and six months, and was adjusted for clustering by doctor. For logistic regression, also adjusted for clustering by doctor, the variable was divided into two categories (two or fewer days lost and three or more days lost), representing better-controlled and less well controlled asthma, respectively. The statistical analyses for all other study factors used logistic regression, adjusted for baseline measurements and clustering by treating doctor. We also conducted a longitudinal analysis using χ2 tests to examine changes in the intervention and control groups over the six months of the trial. Odds ratios and 95% confidence intervals (95% CIs) were calculated for all outcome measures. All tests were conducted using the Stata statistical software package.15 Results Patients Over the three-month recruitment period, 195 people entered the study, and 191 (97 in the clinic group and 94 controls) completed both interviews (Box 1). The demographic profile of the control group (mean age, 26.3 years, 95% CI, 22.3-30.2; proportion of males, 45%, 95% CI, 34%-55%) was similar to the intervention group (mean age, 27.5 years, 95% CI, 23.6-31.4; proportion of males, 42%, 95% CI 32%-52%). Although there were no records of the number of eligible subjects across all participating practices, an audit of all patients seeing general practitioners at two of the eight practices showed a participation rate of 45.1% of all asthmatics. The main reasons for non-participation were either the general practitioner's failing to ask the patient for consent or the patient's refusing consent. The average age of non-participants (mean age, 20.8 years, 95% CI 17.1-24.5) was not significantly different from that of participants (mean age, 21.4 years, 95% CI 17.1-25.7) in the two audited practices. Of the 97 people in the clinic group, 67 completed the three planned clinic sessions, 17 attended only one or two sessions, and 13 failed to attend any sessions. Outcomes There was no difference between the clinic and control groups at baseline for dichotomous study variables (Box 2), which provided evidence for effective randomisation. There was a low incidence of hospitalisation and emergency department attendance in both groups. At six months, however, waking at night at least weekly due to asthma and current smoking were significantly different between the two groups, as was ownership of a peak flow meter (Box 3). There were no differences reported at six months in patients' having discussed trigger factors with their general practitioner or receiving an action plan from their general practitioner. Unlike the dichotomous outcome variables, the mean number of days lost from work was significantly different between the treatment groups at baseline, with a high level of variability in the control data (intervention group mean, 2.62, 95% CI 1.84-3.40; control group mean, 5.37; 95% CI 3.46-7.29). There was no difference in the number of days lost from work at six months in the intervention group compared with the control group (intervention group mean, 2.09, 95% CI 0.91-3.27; control group mean, 2.66; 95% CI 1.66-3.66). Linear regression to compare the number of days lost at six months minus those at baseline between intervention and control participants showed a significant reduction in number of days lost in the control group compared with the intervention group (P = 0.04). Logistic regression comparing two or fewer days lost with three or more days lost showed a non-significant trend towards fewer days lost from work at six months in the intervention group (odds ratio, 0.50; 95% CI 0.24-1.03). Longitudinal analyses within the clinic and control groups showed a trend towards decreasing morbidity over the six-month period, with a 13% decrease in the number of people reporting days lost from work, school or usual activities in both groups (intervention group odds ratio, 0.57, 95% CI 0.31-1.06; control group odds ratio, 0.57, 95% CI 0.31-1.06). Longitudinal analyses also showed that the reported issuing of action plans increased significantly in both groups (intervention group odds ratio, 11.25, 95% CI 3.07-41.21; control group odds ratio, 3.94, 95% CI 1.53-10.10). An increase in the discussion of trigger factors was only significant for the control group (intervention group odds ratio, 2.6, 95% CI 0.84-8.02; control group odds ratio, 3.72, 95% CI 1.36-10.21). Discussion Our study does not show that asthma clinics are more effective than standard general practice care in reducing asthma morbidity. We observed an improvement in most of the outcome measures in both the intervention and control groups. Outcomes for patients who attended an asthma clinic had few differences from those in the control group. The intervention group was less likely to be woken "at least weekly" at night due to asthma. This may signal better home management of asthma in the intervention group, an outcome that is anticipated by guidelines16 but challenged in recent studies.17 Ownership of peak flow meters increased among the intervention group during the course of the study, probably because the asthma educators promoted the use of peak flow readings for asthma self-management. More people in the intervention group either adopted or recommenced smoking during the study. Smoking cessation advice was not included as part of the study intervention. In intensive interventions such as asthma clinics, smokers should be identified to ensure that smoking cessation is achieved or maintained if possible. In this study there were few reports of hospitalisations and emergency department visits at baseline, reflecting our broad inclusion criteria, which did not focus on severe asthmatics. Outcomes which were infrequent in this study may be significant in much larger studies. In relation to days lost from work or school, the significant difference between the intervention and control groups at baseline, the high variability at baseline in the control group and the different findings between the logistic and linear regressions at six months do not allow an adequate understanding of the effect of the intervention. The lack of difference between groups in the proportion of participants who had discussed trigger factors with their doctor and who had received an asthma action plan raises one of the possible explanations for the limited differences observed in this study. Baseline levels of both these clinical practice indicators are usually around 40%-60%,14 and were within that range at baseline in our study. Yet the frequency of both indicators increased substantially over the six months for both groups, which might suggest that clinical practice for the control subjects was contaminated as a result of general practitioners' seeing both intervention and control subjects. Other explanations may be: The asthma clinic in itself makes no difference to the outcomes being measured, and that a simpler intervention, such as regular general practitioner review, is the critical factor. The results of a review of asthma education support this.10 Whether the presence of an asthma clinic in a general practice increases the rate of general practitioner review is unknown. A possible Hawthorne effect.18 With general practitioners and patients being unblinded to the study, it is possible that both increased behaviours which led to better asthma self-management outcomes. People with asthma tend to show improved outcomes over time. This effect is likely to be enhanced after a visit to a general practitioner where management decisions are made. The data showing that the number of people reporting days lost from work or school decreased in both intervention and control groups support this explanation. Baseline data may have been biased because subjects were randomised into treatment groups before the baseline interview, and some subjects may have been aware of their treatment status at the baseline interview. Whatever reasons led to the apparent improvement in outcomes, a factor that may have contributed to the lack of difference between intervention and control groups may be the poor compliance with the intervention regimen. Nearly one-third of patients from the intervention group did not comply with the required three clinic visits, and 13% did not attend any visits. The outcome of this study holds some lessons for future randomised controlled trials in general practice. The major limitation of the study was the randomisation by patient rather than practice, which may lead to modification of general practitioner behaviour towards the control group as well as the intervention group, thus increasing the chance of contamination of controls and the Hawthorne effect. The inability to blind general practitioners and patients to the patient's treatment allocation also limited the study. The participation rate of less than 50% is lower than the rate considered desirable for randomised controlled trials, but compares favourably with other community-based studies.6,19 This study also demonstrates that, in general practice, larger, longer-term or more targeted studies (such as to more severe asthmatics) are required to show differences, especially for variables such as hospitalisation for asthma, which have a relatively low incidence among most general practice patients. Conducting this study for a longer period would also have allowed it to run over the full range of seasons, which is preferable for studies of seasonal diseases such as asthma. Acknowledgements We would like to thank the two nurses who conducted the asthma clinics, and Kieran McCaul from the Department of Human Services, South Australia, for statistical advice on the final drafts of the article. Thanks also go to the doctors and staff at the following participating practices: Angaston Medical Centre, Brooker Clinic, Greenwith Medical Centre, Golden Grove Medical Centre, Flinders Clinic, Nuriootpa Medical Centre, North Plympton Medical Centre and Tanunda Medical Centre. This study was made possible through a grant from the Department of Human Services, Adelaide, South Australia. References Abramson M, Kutin J, Czarny D, et al. The prevalence of asthma and respiratory symptoms among young adults: Is it increasing in Australia? J Asthma 1996; 33: 189-196. Peat JK, Toelle BG, Gray EJ, et al. Prevalence and severity of childhood asthma and allergic sensitisation in seven climatic regions of New South Wales. Med J Aust 1995; 163: 22-26. Barnes G, Partridge MR. Community asthma clinics: 1993 survey of primary care by the National Asthma Task Force. Qual Health Care 1994; 3: 133-136. Jones KP, Mullee MA. Proactive, nurse-run asthma care in general practice reduces asthma morbidity: scientific fact or medical assumption? Br J Gen Pract 1995; 45: 497-499. Lahdensuo A, Haahtela T, Herrala J, et al. Randomised comparison of guided self-management and traditional treatment of asthma over one year. BMJ 1996; 312: 748-752. Mayo PH, Richman J, Harris HW. Results of a program to reduce admissions for adult asthma. Ann Intern Med 1990; 112: 864-871. Martys C. Asthma care in Darley Dale: general practitioner audit. BMJ 1992; 304: 758-760. Evans D, Mellins R, Lobach K, et al. Improving care for minority children with asthma: Professional education in public health clinics. Paediatrics 1997; 99: 157-164. Charlton I, Charlton G, Broomfield J, et al. Audit of the effect of a nurse run asthma clinic on workload and patient morbidity in a general practice. Br J Gen Pract 1991; 41: 227-231. Gibson P, Coughlan J, Abramson M, et al. The effects of self-management education and regular practitioner review in adults with asthma. (Cochrane Review). In: The Cochrane Library, Issue 2. Oxford: Update Software; 1998. Bauman A. Effects of asthma patient education upon psychological and behavioural outcomes. Int Rev Health Psych 1993; 2: 199-212. Charlton I, Antoniou AG, Atkinson J, et al. Asthma at the interface: bridging the gap between general practice and a district general hospital. Arch Dis Child 1994; 70: 313-318. Jones K, Charlton I, Middleton M, et al. Targeting asthma care in general practice using a morbidity index. BMJ 1992; 304: 1353-1356. Beilby J, Wakefield M, Ruffin R. Reported use of asthma management plans in South Australia. Med J Aust 1997; 166: 298-301. StataCorp. Stata Statistical Software: Release 5.0. [computer program]. College Station, Texas: Stata Corporation, 1997. The Thoracic Society of Australia and New Zealand. Peak flow meter use in asthma management. Med J Aust 1996; 164: 727-730. Garrett J, Fenwick J, Taylor G, et al. Peak expiratory flow meters (PEFMs): who uses them and how does education affect the pattern of utilisation? Aust N Z J Med 1994; 24: 521-528. Rice B. The Hawthorne effect: persistence of a flawed theory. Psychol Today 1982; 16: 71-74. Garrett J, Fenwick J, Taylor G, et al. Prospective controlled evaluation of the effect of a community based asthma education centre in a multiracial working class neighbourhood. Thorax 1994; 49: 976-983. (Received 4 May 1998, accepted 31 May 1999) Authors' details SA HealthPlus, Department of Human Services, Adelaide, SA. Adrian R Heard, BSocAdmin, MPH, Senior Project Officer. Centre for Health Care Evaluation, Bedford Park, SA. Ian J Richards, BApplSc, MPH, Project Officer. Flinders Medical Centre, Bedford Park, SA. John H Alpers, MD, FRACP, Director, Respiratory Unit. North Western Adelaide Health Service, Woodville South, SA. Louis S Pilotto, FAFPHM, PhD, Senior Consultant. Brian J Smith, FRACP, PhD, Senior Lecturer, Department of Medicine. Asthma SA, Royston Park, SA. Julie A Black, BNG, Chief Executive Officer. Reprints will not be available from the authors. Correspondence: Mr A R Heard, SA HealthPlus, PO Box 65, Rundle Mall, SA 5000. Email: Adrian. HeardaATdhs.sa.gov.au Back to textBack to textBack to text

Adrian R Heard · Ian J Richards · John H Alpers · Louis S Pilotto · Brian J Smith · Julie A Black

Research Comparison of patients with asthma managed in general practice and in a hospital clinic Omar A Abdulwadud, Michael J Abramson, Larry Light, Francis C K Thien and E Haydn Walters MJA 1999; 171: 72-75 See also Heard et al & Gibson Abstract - Introduction - Methods - Results - Discussion - Acknowledgements - References - Authors' details - - More articles on Respiratory medicine Abstract Objectives: To compare knowledge and attitudes about asthma, self-management skills and impact of asthma on quality of life between patients managed in general practice (GP) and in a hospital clinic. Design: Cross-sectional survey with six months' follow-up. Patients and setting: 105 adults with asthma: 61 from the Alfred Hospital Asthma and Allergy Clinic, Melbourne, and 44 from nearby general practices, in 1994-1995. Main outcome measures: Patient sociodemographic and clinical characteristics; patient knowledge, attitudes and beliefs about asthma; self-management skills; and impact of asthma on quality of life. Results: GP patients were more educated (P = 0.04) and more likely to smoke (P = 0.04) and to have mild asthma (P = 0.04) than hospital patients; they were less likely to use theophylline (P = 0.006) and to have exercise limitation (P = 0.03), and had fewer previous hospital admissions (P = 0.01). Impact of asthma on quality of life was greater in the hospital group than in the GP group. At baseline, the GP group were less likely to have written asthma action plans (P = 0.018), and were less able to manage rapid onset attacks than the hospital group (P = 0.02). More subjects in the hospital group than the GP group felt their asthma was severe (P = 0.02) and were optimistic about their asthma improving (P = 0.03). GP patients increased their knowledge about asthma significantly (P = 0.002) over six months. Conclusions: Patients with asthma managed in general practice and in hospital differ in clinical parameters, quality of life and attitudes to asthma. Future educational initiatives should take such differences into account. Introduction Asthma is the third most common reason for consultations with general practitioners (GPs) in Australia,1 but is both under-recognised and undertreated.2 The finding that GPs were the usual source of routine asthma treatment for 89% of the people who died of asthma3 demonstrates that GPs have the primary management role even in severe asthma.4To guide both doctors and patients, in 1989 the Thoracic Society of Australia and New Zealand developed the "six-point" Australian Asthma Management Plan (AMP).5 This was adopted by the National Asthma Campaign (NAC) at its inception in 1990. While the Royal Australian College of General Practitioners is a partner in the NAC, the plan was originally drawn up by hospital-based specialists. There is concern among GPs that they do not have sufficient time and knowledge to devise action plans and counsel patients about asthma management,4 as suggested in the AMP. The effectiveness of patient education about asthma as proposed in the AMP has not been widely evaluated in general practice. Only one such study has been reported,6 and only one nationwide survey of selected GPs was conducted before the launch of the NAC intervention.7 We aimed to examine patient knowledge about asthma and its management, self-management skills, impact of asthma on quality of life, and attitudes to asthma. We compared these between patients managed in general practice and in a hospital clinic over a six-month period that coincided with widespread advertising and dissemination of the AMP. Methods Design and setting The study was a cross-sectional survey with six months' longitudinal follow-up during 1994-1995. There was no intervention, and patients received their usual care from their GPs or hospital specialists. The study was conducted at the Alfred Hospital Asthma and Allergy Clinic, Melbourne, Victoria, and at 13 general practices near the hospital. Fourteen GPs were recruited by a practising GP (L L). The study was approved by the Ethics Review Committee at the Alfred Hospital. Subjects Consecutive patients being seen for asthma at the participating general practices were nominated by their GPs. Of 54 patients invited to participate, 44 were recruited (nine refused or did not respond and one died). Sixty-one patients were recruited from the Alfred Clinic. Inclusion and exclusion criteria were reported previously.8 All patients gave written informed consent. Diagnosis of asthma was based on American Thoracic Society criteria.9 Assessments Clinical and demographic characteristics of patients were determined by questionnaire. The term "priority asthmatic" was given to asthmatics who had severe and precipitately acute attacks of asthma; they had usually (but not necessarily) been admitted to an intensive care unit. Forced expiratory volume in 1 second (FEV1) was obtained from the records of the lung function laboratory, and predicted FEV1 at baseline was calculated for males and females separately, as suggested by Gibson et al.10 The mean daily peak expiratory flow (PEF) variability was estimated from patient diaries.5 Severity of asthma was classified as mild, moderate or severe based on medication use, as reported previously.8 Medication was grouped into generic categories. Outcome measures Patients completed the Asthma General Knowledge,11 Quality of Life,12 Self-Management Skills,13 and Attitudes and Beliefs14 questionnaires on entry and six months later. Selection of these outcomes was based on the asthma education targets and outcome measures proposed by the National Asthma Campaign.15 Statistical analysis Data were analysed with the SAS for Windows statistical package.16 Categorical variables were summarised as percentages, and associations were tested in contingency tables by χ2 tests. For continuous variables that were normally distributed, differences in mean scores were assessed by Student's t test. For continuous variables that were not normally distributed, the Wilcoxon signed rank and rank sum tests were used. Results Participation Follow-up questionnaires were returned by 39 of the 44 (87%) GP patients (with five of those with the most severe asthma lost to follow-up) and by 47 of the 61 (77%) hospital patients (with the remaining 14 either lost to follow-up or not responding). Subject characteristics Characteristics of the two patient groups are shown in Box 1. They were similar in age and sociodemographic characteristics, including occupation (data not shown), except that the GP group had a significantly higher proportion of tertiary-educated people and of current smokers than the hospital group. GP patients had less exercise limitation and milder asthma than the hospital patients, and were less likely to have been admitted to hospital and to have priority asthma. There was no significant difference in lung function or other clinical parameters (data not shown). GP patients were significantly less likely to use theophylline and to have a written asthma action plan than hospital patients. Proportions using β-agonists, anticholinergics, and inhaled or oral steroids and owning peak flow meters were not significantly different in the two groups. Asthma outcome measures Asthma knowledge: Asthma General Knowledge scores did not differ significantly between the GP and hospital groups either at baseline (means, 20.2 and 20.5 out of 31, respectively) or six months later. Scores increased in both groups over the six months, but the increase was significant only in the GP group (P = 0.002), which had a mean score increase of 1.66 (95% confidence interval [CI], 0.58 to 2.76) compared with 0.83 (95% CI, 20.39 to 2.03) in the hospital group. Quality of life: Scores for impact of asthma on quality of life are shown in Box 2. Impact at baseline was significantly greater in the hospital group than in the GP group for total quality of life and the subcategories of breathlessness, social disruption, and concern for health, but not mood disturbance. After six months, impact had significantly decreased in the hospital group for all these parameters except mood disturbance, but there was no significant change in the GP group. Despite the significant decrease in the hospital group, impact was still significantly greater than in the GP group for total quality of life (P = 0.03), social disruption (P = 0.008) and concern for health (P = 0.04). None of the improvements in quality of life differed significantly between the two groups over six months. Self-management skills: At baseline, the hospital group had a significantly higher median score for knowledge about self-management of a rapid onset asthma attack than the GP group (P = 0.02), but the two groups had similar scores for a slow onset attack (Box 3). After six months, the hospital group had significantly improved its slow onset scenario score (P = 0.02), but the GP group had not. Overall, there was no difference in the median change over six months in scores for either scenario between the two groups. Attitudes and beliefs about asthma: Patients' attitudes and beliefs about their asthma are shown in Box 4. At baseline, hospital patients were significantly more likely than GP patients to believe that their asthma was severe, but also that it would improve in the future. After six months, hospital patients were significantly less optimistic (P = 0.03), and the difference in this parameter between the two groups was no longer significant. The proportion of patients who said they could do everything they wanted regardless of the effect it might have on their asthma did not differ significantly between the two groups at baseline, but at follow-up hospital patients were significantly less likely to say this than GP patients. More hospital than GP patients wished their doctor would talk more to them about their asthma at both baseline and six-month follow-up, but the difference was significant only at follow-up. Concomitantly, fewer hospital patients than GP patients felt that their doctor had told them everything they wanted to know about their asthma; the difference was significant at both baseline and follow-up. Discussion We found that our samples of GP and hospital patients with asthma differed in sociodemographic and clinical parameters, quality of life, self-management skills and attitudes and knowledge about asthma. Some of our findings about the GP sample were more positive than reported previously. Use of inhaled steroids was significantly higher than found in random community samples,17,18 possibly because our sample of GP patients had more severe asthma, or because of industry promotion of these drugs.18 It is also possible that this form of treatment has become more acceptable in our group of GPs, many of whom had a particular interest in asthma. Use of peak flow meters and prevalence of written action plans in the GP patients were also higher than previously reported,17,18 but, once again, this finding may be confounded by the group of GPs studied. Actual prevalence of peak flow meter use and written action plans may be substantially lower in more typical general practices. There was some evidence from our study that implementation of the AMP may have improved, especially in general practice, over the period of our study. However, despite NAC recommendations, half of the hospital patients and three-quarters of the GP patients had no written asthma action plans; and about a third of the hospital patients and half of the GP patients did not have access to a peak flow meter. While our study sheds no light on the reasons for these findings, it demonstrates that there was room for improvement in asthma management in both settings, but particularly in general practice. Asthma knowledge among the GP patients was similar to that among the hospital patients at baseline, but improved more over the six months to follow-up. This was possibly due to the GP patients' higher educational level, our interaction with them during the study or better than average care from their doctors. As doctors had recruited the patients to the study, they may have put more effort into educating them before follow-up, although we have no evidence for this. The absence of any significant improvement in quality of life among the GP patients after six months suggests that insufficient attention may be given to this aspect of asthma. Furthermore, the potential selection bias towards milder disease and the loss to follow-up of five of those with more severe asthma from the GP group may have biased the results away from detecting improvement. Overall, the general practice group had better quality of life than the hospital group. This would be expected, as the GP group had milder disease with fewer hospital admissions. It is surprising that mood disturbance was the only category which did not differ between the two groups at baseline. Possibly moods such as sadness, depression and frustration are similar in all patients regardless of the severity of asthma. At baseline, the GP group was less able to manage the rapid onset attack, perhaps because they were likely to have experienced fewer such attacks. The hospital group improved their ability to manage slow onset attacks significantly over six months. Maybe their doctors educated them informally, although they did not have access to the results of the questionnaires. Alternatively, participants may have improved their answers with practice, but as this did not happen in the GP group it is unlikely. There was little difference in attitudes and beliefs about asthma between groups. However, at six months, the hospital group was significantly less optimistic about their asthma improving in the future than the GP group. This could result from the difference in asthma severity between the two groups, which resulted in hospital patients having a more realistic appreciation of their poor prognosis after education. At six months, more of the hospital patients than GP patients also wished that their doctors would tell them more about their asthma. Perhaps this reflects a low level of communication between doctors and patients in the hospital asthma clinic compared with general practice and requires further study. This study has some limitations. The GP patients may not be representative of the total general practice population, and recruiting interested GPs may have biased results. In the absence of any viable alternative strategy, evaluation focused on patients rather than GPs. Despite potential sampling and selection biases, the results highlight the impact of patient education among patients attending general practices and the differences between GP and hospital patients. Future educational initiatives should take such differences into account. The type and scope of patient education in general practice and hospitals should be thoroughly evaluated to identify areas in which GPs and specialists could be trained and supported more effectively. Continued dissemination and implementation of the Australian AMP is required to improve the self-management skills of patients in general practice. Acknowledgements We acknowledge a public health postgraduate research scholarship from the National Health and Medical Research Council. Dr Andrew Forbes, Jan Driver and Michael Bailey provided statistical support. We are grateful to the general practitioners who participated in the study. Drs Guy Marks, Rae Allen and Bonnie Sibbald gave permission to use their questionnaires. Dr John Kolbe gave permission to use his scoring system for the asthma attack scenarios. References Bridges-Webb C, Britt H, Miles DA, et al. Morbidity and treatment in general practices in Australia 1990-1991. Med J Aust 1992; 157 Suppl: 1-56. Tse M, Cooper C, Bridges-Webb C, Bauman A. Asthma in general practice. Opportunities for recognition and management. Aust Fam Phys 1993; 22: 736-741. Robertson CF, Rubinfeld AR, Bowes G. Deaths from asthma in Victoria: a 12 month survey. Med J Aust 1990; 152: 511-517. Antic R. Asthma in Australia: the current understanding. Overview of a national series of interactive meetings for general practitioners. Sydney: Excerpta Medica, 1993. Woolcock A, Rubinfeld AR, Seale JP, et al. Asthma management plan, 1989. Med J Aust 1989; 151: 650-653. Byrne DM, Drury J, Mackay RC, et al. Evaluation of the efficacy of an instructional program in the self-management of patients with asthma. J Adv Nursing 1993; 18: 637-646. Tse M, Bauman A, Bridges-Webb C. Asthma management in general practice. Aust Fam Phys 1991; 20: 1085-1092. Abdulwadud O, Abramson M, Forbes A, et al. Evaluation of a randomized controlled trial of adult asthma education in a hospital setting. Thorax 1999; 54: 493-500. American Thoracic Society. Standards for the diagnosis and care of patients with chronic obstructive pulmonary disease (COPD) and asthma. Am Rev Respir Dis 1987; 136: 225-244. Gibson J, Gallagher H, Johansen A, Webster I. Lung function in an Australian population: spirometric standards for non-smoking adults. Med J Aust 1979; 1: 292-295. Allen RM, Jones MP. The validity and reliability of an asthma knowledge questionnaire used in the evaluation of a group asthma education self-management program for adults with asthma. J Asthma 1998; 35: 537-545. Marks GB, Dunn SM, Woolcock AJ. A scale for the measurement of quality of life in adults with asthma. J Clin Epidemiol 1992; 45: 461-472. Sibbald B. Patient self care in acute asthma. Thorax 1989; 44: 97-101. Sibbald B, Collier J, D'Souza M. Questionnaire assessment of patients' attitudes and beliefs about asthma. Fam Pract 1986; 3: 37-40. National Asthma Campaign (NAC) National Asthma Strategy. Goals and targets. Melbourne: National Asthma Campaign, 1994. SAS Institute Inc. SAS language guide for personal computers, the SAS system for Microsoft Windows. Release 6.10. Cary, NC: SAS Institute Inc, 1994. Abramson MJ, Kutin JJ, Rosier MJ, Bowes G. Morbidity, medication and trigger factors in a community sample of adults with asthma. Med J Aust 1995; 162: 78-81. Comino EJ, Mitchell CA, Bauman A, et al. Asthma management in eastern Australia, 1990 and 1993. Med J Aust 1996; 164: 403-406. (Received 28 Sep 1998, accepted 7 Jun 1999) Authors' details Department of Epidemiology and Preventive Medicine, Monash Medical School, The Alfred Hospital, Melbourne, VIC. Omar A Abdulwadud, PhD, Postdoctoral Fellow; Michael J Abramson, PhD, FRACP, Associate Professor. Department of Allergy and Clinical Immunology, Monash Medical School, The Alfred Hospital, Melbourne, VIC. Larry Light, MB BS, Clinical Assistant; Francis C K Thien, MD, FRACP, Staff Physician. Department of Respiratory Medicine, Monash Medical School, The Alfred Hospital, Melbourne, VIC. E Haydn Walters, DM, FRACP, Professor, and Director of Respiratory Medicine. Reprints: Associate Professor M J Abramson, Department of Epidemiology and Preventive Medicine, Monash Medical School, The Alfred Hospital, Prahran, VIC 3181. Email: Michael. AbramsonATmed.monash.edu.au Back to textBack to textBack to text Back to text

Omar A Abdulwadud · Michael J Abramson · Larry Light

Abstract Objective: To investigate possible HIV transmission among prison inmates. Setting: A prison system in an Australian State. Participants: 13 ex-prisoners and their prison contacts. Methods: Ex-prisoners who claimed to have been infected with HIV in prison and their prison contacts were interviewed about HIV risk behaviour. Entries in prison and community medical records were used by a three-member expert panel to establish the likelihood of primary HIV infection and its possible timing and location. Main outcome measures: Determination of whether HIV infection probably occurred in prison. Results: There was a very high probability that at least four of 13 ex-prisoners investigated acquired HIV in prison from shared injection equipment. Another two ex-prisoners most probably acquired HIV infection outside prison. The location of infection for the remaining seven could not be determined. Conclusions: HIV transmission in prison has substantial public health implications as most drug-using prisoners soon return to the community. HIV prevention strategies known to be effective in community settings, such as methadone maintenance treatment and syringe exchange schemes, should be considered for prisoners. Introduction HIV transmission in prison has been reported in the United States,1 Scotland2 and Australia.3 The infrequency of these reports has led to a belief that HIV transmission occurs rarely among inmates. A more likely explanation is that confirmation of HIV transmission is more difficult in prisons than community settings.4 Multiple and powerful factors conducive to high HIV incidence are found in prisons. These include that: HIV prevalence is generally several times higher in prisons than in surrounding communities because of the considerable over-representation of injecting drug users (IDUs) among prisoners;5 reports of syringe sharing with multiple injectors are still common in prisons but now rare in community settings;6 HIV infection has been associated with imprisonment in France7 and Spain;8 and incidence of hepatitis C among IDUs incarcerated twice within a 12-month period was double that among IDUs who remained at liberty.9 Furthermore, HIV prevention measures, such as provision of sterile injecting equipment, condoms and methadone maintenance, are uncommon in prisons.10 Although several estimates of HIV prevalence have been conducted in correctional institutions,1,11,12 assessing the incidence of HIV transmission within a prison system poses considerably greater challenges.4 Most drug users serve short, repeated sentences. This hampers the investigation of infection outbreaks and identification of transmission location. In an earlier Australian study,13 several IDUs claimed to have become infected with HIV in prison. Some reported symptoms indicative of primary HIV infection while incarcerated. The aim of this study was to assess, using epidemiological data, whether these IDUs or their contacts had become infected in prison. A similar approach has been used to investigate an HIV outbreak in a Scottish prison.2 Methods Index cases The investigation took place between 1993 and 1994. Seven IDUs from an earlier study13 who claimed to have acquired HIV infection in prison were recontacted. In the earlier study, respondents who had injected drugs and had recently been released from prison were recruited from methadone units, hostels for ex-prisoners or drug injectors, probation offices, syringe exchange schemes, local media advertisements, AIDS organisations and via street networking.13 Prison contacts We traced prison contacts nominated by these seven people through the state methadone registry, AIDS services, drug users' organisations, HIV physicians and the State Registry of Deaths. The contacts were inmates with whom the seven index inmates had engaged in syringe sharing, anal sex or tattooing while in prison. Assessment of HIV infection HIV infection was assessed by an expert panel of three HIV physicians. The experts were provided with dates of entry to and exit from prison, HIV test results, symptoms recorded at the time the prisoner believed infection occurred, self-reported symptoms and self-reported risk behaviour. All dates were referenced in months from the year before the first detected case had last tested HIV negative. Each expert independently assessed whether the recorded symptoms indicated an HIV seroconversion illness and whether the infection occurred in prison, in the community, or if the location was indeterminate. We then accepted the majority decision in each assessment. Ethical approval Relevant ethics committees approved the study on condition that study participants' contacts in the community following their release from prison were not traced. We were also required to alert potential study participants to the possibility of legal or other consequences of admitting drug use in prison or transmitting HIV to another person. Participation in the study required signed, informed consent. Information which would identify the exact time and location of these possible infections has not been included, in accordance with requirements of one ethics committee. Results Contact tracing Between 1993 and 1994, seven male IDUs described in a previous study13 were recontacted (subjects A, B, C, D, E, F, G). They identified 20 prison contacts: six of these contacts could not be located, six had died of AIDS (according to death certificates), and two declined to participate for fear of repercussions for transmitting HIV. The six remaining contacts (H, I, J, K, L, M) plus the seven index cases made a total of 13 ex-prisoners available for investigation (Figure). Prison clusters Prison records revealed two clusters (C1, C2) of six subjects in one or two prison wings. Subjects B, E, I, K and M were held in Prison 1 (population about 250 inmates) in months 22 and 23, during which time index subject B seroconverted. Subjects B, D, I, K and M were held in Prison 2 (population about 300 inmates) in months 29 and 30, during which time contact subject K seroconverted. Index participant A was not part of either cluster, but another participant reported sharing syringes with him. Of the six deceased potential respondents, two had been part of C1 and another two had been part of C2. According to death certificates, two deceased potential respondents became infected with HIV in the year when they were held with the clusters. Cluster analysis was not possible because of the lack of records being kept on the total number of inmates held in the prison wings during the crucial times. Self-reported risk behaviour Eleven participants (A, B, C, E, I, K, M, D, F, G, L) reported syringe sharing in prison, with the first seven nominating another person in this series as a sharing partner. Contact participants I and M also reported receiving a tattoo in prison, and index subject C reported unprotected anal sex. All six deceased prison contacts wereidentified by one or more participants as having shared syringes with them around the crucial periods of months 22 and 23 and months 29 and 30. Assessment of HIV infection Medical file entries for 10 subjects (A, B, C, E, F, G, I, K, L, M) were reviewed by the expert panel. The experts concluded that five participants (A, B, C, G, K) had experienced primary HIV infection and that the most likely location of transmission for individuals A, B and C was in prison (Box 1). Overall, it was concluded that infection occurred in prison for four subjects and in the community for two (Box 2). Discussion The epidemiological evidence that individual A was infected in prison is beyond doubt: he tested negative and then positive for HIV infection after years of confinement in prison. There was strong epidemiological evidence that individuals B, C and J were also infected in prison. Individuals D and G were infected with HIV in the community. The location of infection for the remaining seven subjects (and the six deceased potential respondents) could not be determined. Thus, on epidemiological grounds, at least four of the 13 people investigated were infected with HIV while in an Australian prison system. The most likely route of HIV transmission was shared injection equipment. These are both conservative and probabilistic assessments. It is likely that a prospective investigation of these 13 people, or even a retrospective investigation closer to the events, would have yielded a larger number of confirmed HIV transmissions in prison. However, the strength of evidence for this network and the multitude of factors conducive to HIV infection in prisons suggest that the extent of HIV transmission occurring in prisons through shared injection equipment is underestimated. A mathematical model of HIV transmission in an Australian prison system14 using values derived from empirical studies also suggests that transmission is occurring within correctional centres. This study illustrates some of the difficulties of confirming HIV transmission in prison. We became aware of a possible outbreak by chance. Obtaining ethical approval for the study was an extremely protracted process requiring the assistance of a legal expert. Inmates were understandably wary of admitting risk behaviour because of the potentially serious consequences (as outlined in the consent form). Apart from the logistical and ethical problems of this type of research, the incubation period for HIV infection is almost as long as the average duration of a prison sentence served for drug-related offences in Australia. Consequently, many HIV infections occurring in prisons will not be detected by conventional surveillance. These factors may help to explain why so few cases of HIV transmission among inmates have been reported. Our study differed from previous reports10 in the extent of transmission detected and the type of prisoners studied. Previous studies have investigated long term, high security prisoners, who have less opportunity to associate with other inmates and visitors to obtain drugs and consequently become infected with HIV.15 Our study found a relatively large number of incident cases considering the small sample size, the extremely low prevalence of HIV infection in the Australian prison population16 and the rapid turnover of inmates. All these factors militate against detection of HIV transmission in prison. A limitation of the study was reliance on self-reported risk behaviour. However, we accepted self-reported data only if corroborated by another external source. Moreover, symptoms reported by informants coincided closely with medical records, supporting the validity of self-reported data. We were precluded by ethics committee requirements from determining whether any sexual partners (and their children) were infected with HIV by participants following release from prison. This restriction prevented investigation of possible HIV transmission beyond prison to the community. However, medical files indicated that four subjects had each had an HIV-positive female sexual partner following release from prison. Two of these women, and an additional HIV-negative partner, became pregnant, with at least two pregnancies reaching full term. Medical files also indicated that at the time of the investigation two former inmates had been engaging in unprotected sex with two HIV-negative women, against the advice of their counsellors. Existing evidence of HIV transmission among prisoners has persuaded prison authorities in few countries to implement effective prevention strategies for inmates. Confirmation of HIV infection from prison to the community may be more persuasive for authorities. Syringe exchange, methadone and bleach programs reduce the spread of HIV in community settings,16-18 and preliminary results from these programs in prison appear promising.19-21 The paucity of data confirming HIV transmission in prison should not be regarded as adequate justification for the lack of effective HIV prevention measures within prisons. There is already sufficient information on HIV transmission between prisoners to justify rapid implementation in correctional institutions of prevention measures shown to be effective in community settings. Improved methods of monitoring the spread of HIV within prisons and from inmates to community members following release are required urgently. Acknowledgements We are grateful to the NSW Department of Health for funding this study. We also wish to thank David Buchanan, Andrew Carr, Ying Chun Ge, David Cooper, Anthony Cunningham, Basil Donovan, John Dwyer, Tania Sorrell and Dominic Dwyer. Results of tracing 20 prison contacts of seven index cases for recruitment into the study. References Brewer TF, Vlahov D, Taylor E, et al. Transmission of HIV-1 within a statewide prison system. AIDS 1988; 2: 363-367. Taylor A, Goldberg D, Emslie J, et al. Outbreak of HIV infection in a Scottish prison. BMJ 1995; 310: 289-292. Dolan K. AIDS, drugs and risk behaviour in prison: state of the art. Int J Drug Policy 1997; 8: 5-17. Dolan K. Why is there conflicting evidence of HIV transmission in prison? In: O'Brien O, editor. Report of the 3rd European Conference on Drug and HIV/AIDS Services in Prison. London: Cranstoun Drug Services, 1997; 19-21. Gaughwin MD, Douglas RM, Wodak AD. Behind bars -- risk behaviours for HIV transmission in prisons, a review. In: Norberry J, Gerull SA, Gaughwin MD, editors. HIV/AIDS and prisons conference proceedings. Canberra: Australian Institute of Criminology, 1991; 89-107. Crofts N, Webb-Pullman J, Dolan K. An analysis of trends over time in social and behavioural factors related to the transmission of HIV among IDUs and prison inmates. Evaluation of the National HIV/AIDS Strategy. Technical Appendix 4. Canberra: AGPS, 1996. Richardson C, Ancelle-Park R, Papaevangelou G. Factors associated with HIV seropositivity in European injecting drug users. AIDS 1993; 7: 1485-1491. Granados A, Miranda MJ, Martin L. HIV seropositivity in Spanish prisons. Presented at the VIth International AIDS Conference, San Francisco. Abstract no Th. D.116, 1990. Crofts N, Stewart T, Hearne P, et al. Spread of blood-borne viruses among Australian prison entrants. BMJ 1995; 310: 285-288. Dolan K, Wodak A, Penny R. AIDS behind bars: preventing HIV spread among incarcerated drug injectors. AIDS 1995; 9: 825-832. Vlahov D, Brewer TF, Castro KG, et al. Prevalence of antibody to HIV-1 among entrants to US correctional facilities. JAMA 1991; 265: 1129-1132. Bird AG, Gore SM, Jolliffe DW, Burns SM. Anonymous HIV surveillance in Saughton Prison, Edinburgh. AIDS 1992; 6: 725-733. Dolan K, Wodak A, Hall W, et al. Risk behaviour of IDUs before, during and after imprisonment. Addict Res 1996; 4: 151-160. Dolan K, Wodak A, Hall W, Kaplan E. A mathematical model of HIV transmission in NSW prisons. Drug Alcohol Depend 1998; 50: 197-202. Dye S, Isaacs C. Intravenous drug misuse among prison inmates: implications for spread of HIV. BMJ 1991; 302: 1506. Feachem R. Valuing the past . . . investing in the future. Evaluation of the National HIV/AIDS Strategy. 1993-94 to 1995-96. Canberra: AGPS, 1996. Ward J, Mattick R, Hall W. Methadone maintenance treatment and other opioid replacement therapies. Amsterdam: Harwood Academic Publishers, 1998. Normand J, Vlahov D, Moses LE. Preventing HIV transmission: the role of sterile needles and bleach. Washington: National Academy Press, 1995. Nelles J, Harding T. Preventing HIV transmission in prison: a tale of medical disobedience and Swiss pragmatism. Lancet 1995; 346: 1507-1508. Dolan K, Wodak A, Hall W. Methadone maintenance treatment reduces heroin injection in NSW prisons. Drug Alcohol Rev 1998; 17: 153-158. Dolan K, Wodak A, Hall W. A bleach program for inmates in NSW: an HIV prevention strategy. Aust N Z J Public Health 1998; 22: 838-840. Authors' details National Drug and Alcohol Research Centre, University of New South Wales, Sydney, NSW. Kate A Dolan, BSc, PhD, Research Fellow. Alcohol and Drug Services, St Vincent's Hospital, Sydney, NSW. Alex Wodak, FRACP, FAFPHM, Director. Reprints: Dr K A Dolan, Research Fellow, National Drug and Alcohol Research Centre, University of New South Wales, Sydney, NSW 2052. Email: ndarc8@unsw.edu.au

Kate A Dolan · Alex Wodak

Research Women's role and satisfaction in the decision to have a caesarean section Deborah A Turnbull, Chris Wilkinson, Anisa Yaser, Vanessa Carty, John M Svigos and Jeffrey S Robinson MJA 1999; 170: 580-583 For editorial comment, see "de Costa" Rapid reader response (with authors' reply) added 28/6/99: see Chung Abstract - Introduction - Methods - Results - Discussion - Acknowledgements - References - Authors' details - - More articles on Obstetrics & gynaecology and women's health Abstract Objective: To examine women's role in the decision to perform caesarean section (CS). Design: Cross-sectional survey. Written questionnaires were completed seven weeks after giving birth by CS. Setting: An obstetric tertiary referral hospital (Women's and Children's Hospital, Adelaide, South Australia), July to December 1996. Participants: A consecutive sample of women who underwent CS over a six-month period. To be eligible, women had to be at least 18 years old, able to complete a questionnaire in English and well enough to consent to study participation. Main outcome measures: Women's involvement in decision making, stated preference for CS, and satisfaction with obstetric care. Results: 278 women (76.4%) returned questionnaires: 171 women (61.5%; 95% confidence interval [CI], 55.8%-67.2%) reported being involved in the decision to have a CS. Factors influencing their decision were physical duress and partner's reaction during labour (emergency CS), considerations about recovery, planning for the event and pain (elective CS), and information from the doctor (both groups). Half the women "strongly agreed" that they were satisfied with the decision to have a CS, but 40.9% only "agreed" and 4.7% were "not sure". About 20% reported they needed more information on other options, and only 28.8% "strongly agreed" that they had been given good information to prepare for the possibility of CS. 27.9% of women (95% CI, 22.5%-33.2%) "agreed" or "strongly agreed" that they had "insisted on a CS" and 21.3% (95% CI, 16.4%-26.2%) that they had told the staff they were "keen to have a CS". Given the option of a vaginal delivery, 37.8% of women (95% CI, 22.5%-55.2%) with a breech presentation, and 34% of women (95% CI, 21.2%-48.8%) who had had a previous CS, chose a CS. Conclusions: It is of concern that over a third of women felt they had not been involved in the decision to have a CS; others were very positive about CS, but an appreciable proportion may not have received sufficient information. A broad-based strategy of providing more information to women and their partners could be one way of ensuring appropriate CS rates and should be tested in a randomised controlled trial. Introduction Despite national policy documents calling for a reduction in medical interventions in birth,1 Australia now has one of the highest caesarean section (CS) rates among First World countries.2 At the same time, it is recognised that the "ideal" or "correct" rate is difficult to derive,3 given the varying risk profiles of women attending different institutions. A range of strategies aimed at obstetricians4,5 have had limited success in reducing CS rates. The assumption is often made that the decision to perform a CS is made on clinical grounds only.6 The role of maternal request in decision-making has been raised,7 but so far studies from the woman's perspective have tended to use small samples with unique characteristics.8,9Our study examines the role of women in the decision to have a CS in an effort to identify an alternative approach to lowering the rates. Methods Setting The study was conducted at the Women's and Children's Hospital in Adelaide, South Australia. In 1996, this State had the highest CS rate of any State in Australia (23.1% v. national average of 19.5%).10 The Women's and Children's Hospital is a major obstetric tertiary referral centre for South Australia and much of the Northern Territory. In 1996, 23% (811/3536) of the women giving birth at the hospital were considered to have very high risk pregnancies and the CS rate for high risk pregnancies at the Women's and Children's Hospital was 43.9% (356/811).11 During that year, the overall CS rate was 25.4% (898/3536). Sample Over a six-month period (July to December 1996) questionnaires were sent to consecutive women who underwent CS. To be eligible for our study, the women had to be at least 18 years old, able to complete a questionnaire in English, and well enough to give consent to participate in the study (as decided by the senior midwife on the postnatal ward). We aimed to receive responses from about 288 women having a CS. This would provide a true population proportion of 60% (to within 5%) of women stating they had a say in the decision. To obtain this sample, we allowed for a non-eligibility, non-response rate of about 25% and thus sampled from a consecutive population of 375 women. Questionnaire The questionnaire was developed on the basis of a review of published reports, a review of existing questionnaires, and interviews with women, obstetricians, midwives and perinatal epidemiologists. It consisted of a combination of questions with forced-choice responses (mainly presented as a complete statement with a five-point response scale ranging from "strongly agree" to "strongly disagree"), open-ended questions and questions on demographic characteristics. (The questionnaire is available from the authors.) Procedure Ethical approval was obtained from the Hospital Ethics Committee, and, with the Committee's permission, a member of the research team (A Y), not involved in providing care, approached women on the postnatal ward between the second and fourth day and sought written consent for participation in the study. Questionnaires were sent to the participating women's homes seven weeks after the birth, to give women time to recover and to minimise biases such as the "halo" effect and social desirability (ie, the effect on respondents of the intensity of labour; and their tendency soon after the event to be uncritical of healthcare providers, particularly if they think their comments can be attributed to them).12 Follow-up of non-respondents included the mailing of a second questionnaire and a telephone call. A summary of the results was sent to participating women at the conclusion of the study. Data analysis The data were analysed using SPSS for Windows13 and Epi Info 6.14 We assessed whether variables such as age, language spoken at home, age left school, type of care, and type of CS distinguished between those who reported involvement in the decision and those who reported no involvement. Categorical data were analysed by χ2 tests. Point estimates and odds ratios are presented with 95% confidence intervals (CI). Results Sample Our sample consisted of 375 women having a CS over a six-month period. Of these, 11 (2.9%) women were not eligible, and 16 (4.3%) were "missed" by the researcher (mainly because they were discharged before contact). Thus, there was a defined sample of 364 women (taking the conservative view that all of the 16 missed women were in fact eligible and thus could be included in the sample). Completed questionnaires were obtained from 278 women (76.4% of our defined sample). The sample included women from a range of sociodemographic backgrounds receiving a diversity of care, including private obstetric care, hospital clinic care and different models of publicly funded midwifery care. About a third of the sample had an elective CS. Women's involvement in decision to have a caesarean section 171 women (61.5% of the study group; 95% CI, 55.8%-67.2%) reported that they were involved in the decision to have a CS. 81.4% (70/86) of those having an elective CS reported involvement in the decision, compared with 53.2% (100/188) of those having an emergency CS (odds ratio [OR] = 3.85; 95% CI, 2.01-7.47). Women cared for by a private obstetrician were less likely to report involvement in the decision (51.9%; 40/77) compared with women receiving other models of care (67.0%; 120/179) (OR, 0.53; 95% CI, 0.30-0.95). Reasons for the decision to have a caesarean section The 171 women who reported involvement in the decision were asked an open-ended question: "What led you to make the decision to have a CS?". A content analysis of the 164 responses to this question indicated that women framed their responses in terms of medical risks and benefits and not in terms of non-clinical personal issues. The questions with forced-choice responses gave women the opportunity to reveal that other factors were also important (Table 1). All women were very likely to have been influenced in their decision by information from their doctor. Women who had an elective CS were influenced by factors such as their recovery and the ability to plan. In contrast, women who had an emergency CS were influenced by the physical stress of labour ("I just couldn't go on any longer"), as well their partner's reaction in the labour ward. Other issues which distinguished the two groups were considerations about pain and previous negative experiences of childbirth: and in each of these about twice as many women with an elective CS agreed or strongly agreed that these issues had influenced their decision. In addition, about four times as many women with an elective CS reported that they had been influenced by family and friends during their pregnancy. Rejecting alternative options to caesarean section We examined women's responses to: Vaginal delivery after previous CS (women who had had one previous CS) Fifty of 209 women (23.9%) for whom these data were available reported one previous CS; 17 of these women (34.0%; 95% CI, 21.2%-48.8%) reported that their doctor talked to them about the possibility of having a vaginal delivery, but they did not subsequently try to have the baby vaginally. Vaginal breech delivery Of the total sample, 62 women (22.3%) reported that they had a breech presentation, and were asked "Did the doctor suggest that it would be possible to have the baby vaginally even though it was positioned bottom first?". In 14 of 37 women (37.8%; 95% CI, 22.5%-55.2%) with whom the doctor had discussed this option, the women reported "I decided against it". External cephalic version The women with a breech presentation were also asked "Did the doctor talk to you about the possibility of turning your baby while it was still in the womb?". In 12 of 39 women (30.8%; 95% CI, 17.0%-47.6%) with whom the doctor had discussed this option, the woman reported "I decided against it". Reporting preference for caesarean section "I insisted on a CS" Women having an emergency CS were less likely to agree with this statement. Demographic factors did not affect the response. Agreement: 27.9% of 269 women (95% CI, 22.5%-33.2%) (including 13.8% who strongly agreed); 18.8% of 181 women having emergency CS v. 46.4% of 84 women having elective CS (OR, 0.27; 95% CI, 0.15-0.49). "I told the staff at the very start that I was keen to have a CS" Women having an emergency CS were less likely to agree with this statement, and the only demographic factor affecting the response was educational level, with women in an intermediate category for education being more likely to agree. Agreement: 21.3% of 268 women (95% CI, 16.4%-26.2%) (including 11.6% who strongly agreed); 13.3% of 181 women having an emergency CS v. 38.5% of 83 women having an elective CS (OR, 0.24; 95% CI, 0.13-0.47), and 29.2% of 106 women in an intermediate education category v. 16% of 159 women in lower and higher education categories (χ2 = 6.31, df = 2, P = 0.04). "At the beginning of my pregnancy, I really wanted to have this baby vaginally" Agreement was not affected by demographic factors, but fewer women attending a private obstetrician agreed (non-significant difference), and fewer women having an elective CS agreed. Agreement: 64.6% of 274 women (95% CI, 58.9%-70.3%) (agreed or strongly agreed); 57.3% of 75 women in private obstetric care v. 68.0% of 178 women in other care (OR, 0.63; 95% CI, 0.35-1.15), and 52.4% of 84 women having elective CS v. 71.0% of 186 women having emergency CS (OR, 0.45; 95% CI, 0.26-0.79). Satisfaction Interpretation of the results for satisfaction is based on the assumption that women who answered anything other than "strongly agree" to positively worded items were not entirely satisfied or may have experienced problems with their care. This approach has been recommended in previous research15 and has been applied to other Australian studies examining satisfaction in childbirth.16 An appreciable number of women were only able to "agree" to such items or were "not sure" about their care (Table 2). For example, while half the women were satisfied with the decision to have a CS (as judged by indicating "strongly agree"), a further 40.9% only "agreed" and 4.7% were "not sure". Similarly, about a third and a half of women, respectively, were unable to "strongly agree" that they were "confident in the final decision" and "believe that caesarean section was the only alternative". More than two-thirds of women were unable to "strongly agree" that they had been "given good information to prepare for the possibility of a CS". Using the related assumption that we should pay attention to any level of affirmative responses (ie, "strongly agree", "agree", or "not sure") to negatively worded items, it can be seen that about 20% of women reported that they "needed more information on other options". Similarly, between about one in 10 and one in four women expressed some degree of dissatisfaction with the decision-making process. Discussion Over a third of the women in our study felt that they were not involved in the decision to have a CS, and an appreciable proportion expressed some degree of dissatisfaction with the decision, or may not have been given sufficient information. The strengths of our study include the good response rate and the fact that we used a hospital-based sample of women, including those with private health insurance. This is an important consideration given that about one-third of childbearing women are privately insured.10 The study also avoided some of the pitfalls of recently published research which relied on women giving feedback to the clinicians involved in their care.17 A shortcoming of our study was that it was conducted in only one hospital and may not be representative of women attending other Australian centres. For example, the average age of women in our study was 30 years, compared with a mean age for childbearing women in Australia of 28.6 years.10 The broader implications of the findings need to be considered, and should be the subject of further research. Clearly, there would be public health benefits in increasing the proportion of women who feel that they have been involved in the decision to have a CS, to avoid the possibility of psychological sequelae.18 While this may not be possible for the 50% of women who felt uninvolved in the decision to have an emergency CS, there may be scope for addressing the one in five women who felt uninvolved in the decision to have an elective CS. There are, of course, some absolute indications for CS (eg, major placenta praevia, massive placental abruption where the baby is still alive, transverse lie with shoulder presentation, and cord prolapse before full dilatation), and alternatives in these instances would not be discussed. These defined reasons may comprise up to 5%-8% of all births,19 and may account for a proportion of those who felt uninvolved in the decision-making process. Our study findings suggest that an informational package would be worth testing in a randomised controlled trial. This package should be broad based, rather than targeted to specific sociodemographic groups. In addition, it should be aimed not only at women, but also their partners, and families and friends. The package, which would include videos, posters and pamphlets, should address issues such as recovery after a CS and the risks and benefits of alternatives to CS. Evaluation of outcomes could be stratified for type of CS (elective or emergency), as it is possible that education would have a more substantial impact on elective CS rates, which appear to be influenced by non-clinical issues outside the labour ward experience. Any such intervention should aim to provide women with balanced information for making a truly informed choice. It is imperative that health messages are not framed in a way that makes women feel inadequate, or that they have failed in some aspect of their pregnancy. To this end, any evaluation of interventions should include outcome data examining not only intervention rates and associated morbidity, but also looking at psychosocial outcomes, such as maternal well-being and satisfaction. The need for information is well recognised in countries such as Britain, where there has been a major drive to develop evidence-based information pamphlets for pregnant women on issues such as ultrasound scans.20 This is not to suggest that information alone will address the high rates of CS,21 which relate to a multiplicity of factors. However, this strategy may help to create a different clinical climate and would at least partially meet the needs of the almost one in five women who agreed that they needed (or were unsure whether they needed) more information on other options. Acknowledgements We wish to thank the women who participated in the study. We would also like to thank the obstetricians and the postnatal ward midwives of the Women's and Children's Hospital for their support. This study was funded by a grant from the Women's and Children's Hospital Foundation and was supported by the Department of Public Health, University of Adelaide. Dr Yaser's position was funded by the Commonwealth Government Jobskills Participation Scheme. References Commonwealth Department of Health, Housing and Community Services. Goals and targets for Australia's health in the year 2000 and beyond. Canberra: AGPS, 1993. National Health and Medical Research Council. Options for effective care in childbirth. Canberra: AGPS, 1996. What is the right number of caesarean sections [editorial]? Lancet 1997; 349: 815. Wilkinson C, McIlwaine G, Boulton-Jones C, Cole S. Is a rising caesarean section rate inevitable? Br J Obstet Gynaecol 1998; 105: 45-52. Molloy D, Richardson P. Caesarean section -- public versus private. Why the differential? A reply to "Obstetric intervention and the economic imperative". Br J Obstet Gynaecol 1993; 101: 88-90. Spurrett B, Cook CM. Why we choose caesarean section: a prospective study. Aust N Z J Obstet Gynaecol 1997; 37: 297-300. Controversies in management. Should doctors perform an elective caesarean on request? BMJ 1998; 317: 462-465. Shepherd McCC. Patent decision making: the case of delivery method after a previous caesarean section. Culture Med Soc 1987; 11: 495-508. Ryding EL. Investigation of 33 women who demand a caesarean section for personal reasons. Acta Obstet Gynaecol Scand 1993; 72: 280-285. Day P, Sullivan EA, Lancaster P. Australia's mothers and babies 1996. Sydney: Australian Institute of Health and Welfare National Perinatal Statistics Unit, 1999. (Perinatal Statistics Series No. 7.) Computerised Maternity Database. Adelaide: Women's and Children's Hospital, Clinical Information Services Database, 1998. Lumley J. Assessing satisfaction with childbirth. Birth 1985; 12 (3): 141-145. SPSS (Statistical package for the social sciences) [computer program], standard version. Chicago, Ill: SPSS Inc, 1997. Epi Info (Word processing, database and statistical program for public health) [computer program], version 6.03. Atlanta, Ga: Centers for Disease Control and Prevention, 1996. Carey RG, Posavac EJ. Using patient information to identify areas for service improvement. Health Care Manage Rev 1982; 7(2): 43-48. Brown S, Lumley J. Satisfaction with care in labour and birth. A survey of 790 Australian women. Birth 1994; 21(1): 4-13. Mould TA, Chong S, Spencer JAD, Gallivan S. Women's involvement with the decision preceding their caesarean section and their degree of satisfaction. Br J Obstet Gynaecol 1996; 103: 1074-1077. Astbury J, Brown S, Lumley J, Small R. Birth events, birth experiences and social differences in postnatal depression. Aust J Public Health 1994; 18: 176-184. Francome C, Savage W. Caesarean section in Britain and the United States -- 12% or 24%: is either the right rate? Soc Sci Med 1993; 37: 1199-218. Oliver S, Rajan L, Turner H, et al. Informed choice for users of health services: views of ultrasonography leaflets on women in early pregnancy, midwives, and ultrasonographers. BMJ 1996; 313: 1251-1253. Fraser W, Maunsell E, Hodnett E, Moutquin J. Randomised controlled trial of a prenatal vaginal birth after cesarean education and support programme. Am J Obstet Gynecol 1997; 176: 419-425. (Received 21 Sep 1998, accepted 30 Apr 1999) Authors' details University of Adelaide, Adelaide, SA. Deborah A Turnbull, MPsych(Clin), PhD, Senior Lecturer in Public Health (Epidemiology), Department of General Practice; Anisa Yaser, MB BS, Research Officer, Department of Public Health; Vanessa Carty, MPsych, Research Student, Department of Public Health; Jeffrey S Robinson, FRCOG, FRACOG, Professor, and Head, Department of Obstetrics and Gynaecology. Department of Perinatal Medicine, Women's and Children's Hospital, Adelaide, SA. Chris Wilkinson, MPH, FRACOG, Staff Specialist Obstetrician Gynaecologist. John M Svigos, FRCOG, FRACOG, Senior Visiting Medical Officer. Reprints will not be available from the authors. Correspondence: Dr D A Turnbull, Department of General Practice, University of Adelaide, SA 5005. Email: dturnbull@medicine.adelaide.edu.au 1: Reasons women give for deciding to have a caesarean section (CS)* Elective Emergency†OR (95% CI)Information from doctor 75.7% (53/70)77.4% (79/102)0.91 (0.42-1.98)Considerations about recovery60.0% (42/70)36.3% (37/102)2.64 (1.35-5.18)Partner's attitude in general44.3% (31/70)35.3% (36/102)1.46 (0.75-2.85)Couldn't go on any longer21.2% (14/66)44.5% (45/101)0.34 (0.15-0.72)Could plan what was going to happen38.6% (27/70)16.0% (16/100)3.30 (1.52-7.22)Considerations about pain30.4% (21/69)15.7% (16/102)2.35 (1.06-5.26)Previous pregnancy/labour awful26.5% (18/68) 12.0% (12/100)2.64 (1.10-6.41)Partner's reaction in the labour room8.8% (6/68)23.8% (24/101)0.31 (0.10-0.85)Influence of family/friends during pregnancy15.7% (11/70)3.9% (4/102)4.57 (1.27-20.39)Felt too old4.5% (3/67)2.0% (2/102)2.34 (0.26-28.63)Previous infertility4.3% (3/69)2.0% (2/101) 2.25 (0.25-27.48)Medical risks/benefits not that important21.4% (15/70)27.0% (27/100) 0.74 (0.34-1.61)*Missing data are excluded. †Some totals exceed 100 (= number of women with emergency CS who indicated that they were involved in the decision-making process), because two additional women answered some of these questions, but denied involvement in the decision. Back to text 2: Women's satisfaction with decision to have a caesarean section (CS)*StronglyNotStronglyItemagreeAgreesureDisagreedisagreeSatisfaction with the decision itselfSatisfied with the decision (n = 274)50.4%40.9%4.7%2.9%1.1%Confident in the final decision (n = 276)62.3%31.2%2.9%2.9%0.7%Believe that a CS was the only alternative (n = 276)53.3%26.8%12.3%5.4%2.2%Now feel a bit cheated (n = 274)5.1%10.9%9.1%28.5%46.4%Believe that CS may not have been necessary (n = 275)2.2%2.5%14.2%31.3%49.8%Satisfaction with information provisionGiven good information about why CS was necessary (n = 275)58.2%33.1%4.7%3.3%0.7%Needed more information on otheroptions (n = 272)2.9%4.4%11.8%48.5%32.4%Given good information to prepare for possibility of CS (n = 274)28.8%40.1%10.2%12.8%8.0%Satisfaction with decision-making processFelt pressured to try to have the baby vaginally (n = 272)4.0%7.4%5.1%34.9%48.5%Unhappy about the amount of say in the decision (n = 275)5.5%6.5%5.8%45.1%37.1%Took too long to make decision (n = 274)7.7%9.9%8.4%38.7%35.4%Felt pressured by staff to have a CS (n = 272)1.8%3.3%3.3%36.8%54.8%* Missing data are excluded. Back to text Reader response to: Women's role and satisfaction in the decision to have a caesarean section I commend the authors for this paper. However I wonder if the contention that unless the respondents "strongly agree" or "strongly disagree", the response is mixed is debatable. Patient satisfaction is multidimensional and to focus on patient's response to assess whether for example the information given has been adequate may be unnecessarily harsh on the care givers. Almost all major decisions are accompanied by a degree of regret and doubt by a proportion of people. To infer, unless the patient is entirely satisfied with something, that the care given was somehow less than satisfactory is to demand too much of a complex human relationship, that of a technically dependent patient and a care giver. I note that only 61.5% felt that they were involved in the decision to have a caesarean section, and only 81.4% in the elective cases where presumably there was ample time to ask for further information. I presume that they had in fact legally consented to the decision. What would have been interesting is a personality study on all the respondents. It may be that information are not the issue but the inherent psychological characteristics of the study subjects. Information packs may be useful but the decision to consent to a caesarean section rests with the woman alone. Packs aimed at her partner, family and friends may do more harm than good where there are conflicting opinions. One only has to look at the diverse response in this study itself to confidently predict that there will be such conflict amongst any group and an already potentially emotional decision may be further complicated. Tony Chung Associate Professor Department of Obstetrics & Gynaecology The Prince of Wales Hospital Shatin, NT Hong Kong SAR, China Submitted by email 24/6/99, posted online with author response 28/6/99. Reply by the authors We would like to thank Dr Chung for his thoughtful reply to our paper. The use of 'strongly agree' and 'strongly disagree' as markers of satisfaction has a precedence in the literature on satisfaction with hospital care.1 Importantly, it has been used in relation to maternity service provision.2 Although this approach may seem unnecessarily critical, it avoids over-simplistic interpretations, as demonstrated in a British report into this issue which concluded (after using disputed methodology) that attempts to reduce caeasrean section rates may be futile, because of women's preferences.3-5 Dr Chung's second point relates to the issue of legal consent. In contrast, our paper examines participation in decision making, for which the signing of informed consent is only one part of a broader process. The paper does not intend to infer that the requirements for informed consent for a procedure have been neglected. The point about measuring women's personalities is an interesting one. There is research which finds increased psychological morbidity in women demanding caesarean section in the absolute absence of clinical indications.6 In contrast to this somewhat rare individual,6 our study examines a cross-section representing about one quarter of the women at the study hospital. Dr Chung's point implies that the problem rests with these women. This seems an unreasonable assumption to make of such a large proportion of childbearing women. It is imperative that we avoid 'blaming' the women. Finally Dr Chung is concerned that the provision of information to the broader community may lead to conflict. Our research suggests that there is already an involvement of family and friends in the decision making process. It is for this very reason that we hypothesize that providing evidenced based information may lead to a more beneficial outcome. To this end any randomized controlled trial evaluating such a strategy should include not only clinical outcomes but also psychosocial variables such as anxiety, feelings of failure and reassurance. Deborah Turnbull Chris Wilkinson Submitted by email 28/6/99, posted online 28/6/99. 1. Carey RG, Posavac EJ. Using patient information to identify areas for service improvement. Health Care Manage Rev 1982; 7(2): 43-48. 2. Brown S, Lumley J. Satisfaction with care in labour and birth. A survey of 790 Australian women. Birth 1994; 21(1): 4-13. 3. Mould TA, Chong S, Spencer JAD, Gallivan S. Women's involvement with the decision preceding their caesarean section and their degree of involvement. Br J Obstet Gynaecol 1996; 103: 1074-1077. 4. Grant JM, Women are satisfied with caesarean section [editorial; comment]. Br J Obstet Gynaecol 1996; 103 (11): 7-8. 5. Hemminki E. Cesarean Sections: Women's choice at giving birth? Birth 1997; 24: 124-125. 6. Ryding EL. Investigation of 33 women who demanded a cesarean section for personal reasons. Acta Obstet Gynecol Scand 1993; 72: 280-285. Back to top of article Back to reader response

Deborah A Turnbull · Chris Wilkinson · Anisa Yaser · Vanessa Carty · John M Svigos · Jeffrey S Robinson

Research Why do preterm infants die in the 1990s? Lex W Doyle, Sheryle Rogerson, Shu-Ling Chuang, Matthew James, Ellen D Bowman and Peter G Davis MJA 1999; 170: 528-532 Abstract - Introduction - Methods - Results - Discussion - References - Authors' details - - Articles on similar material Abstract Objectives: To describe the mortality rate for preterm infants (born 23-36 completed weeks' gestational age) and to determine the causes of death, focusing on avoidable causes. Design and setting: Prospective cohort study of preterm infants born at Royal Women's Hospital, Melbourne (a tertiary referral hospital with a neonatal intensive care unit and a special care nursery) from January 1994 to December 1996. Subjects: 2475 consecutive liveborn infants with gestational ages from 23 to 36 weeks. Main outcome measures: Mortality rate during the primary hospitalisation, and causes of death. Results: The total mortality rate was 4.8% (118/2475). The mortality rate declined with increasing maturity. The decrease in mortality was rapid between 23 and 28 weeks' gestational age, from 64.5% at 23 weeks to 4.0% at 28 weeks, then slower, falling to 0.4% at 36 weeks. Fifty of the 118 infants who died had lethal congenital anomalies. Lethal anomalies accounted for three-quarters of deaths in infants aged 28-36 weeks. The mortality rate in infants free of lethal anomalies was 2.8% (68/2425) and only 0.2% (4/1759) for infants aged 32-36 weeks. In the 68 infants without lethal anomalies who died, few obvious preventable causes were identified. Conclusions: Mortality rates fell rapidly between 23 and 28 weeks' gestational age. Survival rates for preterm infants born after 31 weeks' gestational age approached the survival rates of term infants. Lethal congenital anomalies were the most common cause of death; preventable causes of death were rare. Introduction The survival rate for very preterm infants (born at 23-27 weeks' gestational age) has improved dramatically with advances in perinatal care,1 particularly efforts to reduce neonatal respiratory distress syndrome (hyaline membrane disease), such as administering corticosteroids to the mother before delivery,2 administering exogenous surfactant to preterm infants,3 and assisted ventilation.4 However, mortality rates and causes of death for more mature but still preterm infants (28-36 weeks' gestational age) have received little attention, especially since the advent of exogenous surfactant in Australia in 1991. As hyaline membrane disease has diminished as a cause of death of preterm infants, other causes of death, some of which may be preventable, have assumed more prominence. The aim of this study of preterm infants 23-36 weeks' gestational age born in 1994-1996 in a hospital with neonatal intensive care facilities was to describe the variation with gestational age in the mortality rate and the causes of death, focusing on avoidable causes. Methods Setting This was a prospective cohort study of consecutive livebirths between 23 and 36 completed weeks of gestational age in the Royal Women's Hospital, Melbourne, over the three years from 1 January 1994 to 31 December 1996. Data collection Data have been collected prospectively since 1977 on all infants admitted to the neonatal intensive care unit (NICU), all infants of birthweight below 1500 g, and all infants below 32 weeks' gestational age, whether admitted to the NICU or not. Data on infants 32 to 36 weeks were collected from the special care nursery (SCN) admissions book, and the hospital's main computer database was checked to obtain data on those not admitted to either the NICU or SCN. Data collected included gestational age, birthweight, and mortality during the primary hospitalisation, whether during the neonatal period (first 28 days after birth) or later. Data were included for infants transferred from our hospital to another. Gestational age in completed weeks was assigned according to the first antenatal ultrasound scan, or maternal dates if no ultrasound report was available. Birthweight ratio was calculated by dividing the infant's birthweight by the expected gender-specific median birthweight for that gestational age.5 Causes of death For infants who died, the major causes of death were determined after the regular monthly clinicopathological conference at which the deaths of all livebirths within the hospital were discussed, and which was chaired by one of the authors ( LW D), who retained all records from the meeting. Causes of death included lethal congenital anomalies, complications of prematurity, perinatal asphyxia, or sepsis (Box 1). Preventable causes of death For each infant who died, avoidable factors included whether antenatal corticosteroids or surfactant had been given to infants who died from respiratory causes, whether intramuscular vitamin K had been given to those who died from pulmonary haemorrhage, whether appropriate anti-infective therapies were given to those who died from infection, and whether antenatal corticosteroids had been given to those who died from cerebroventricular haemorrhage or cystic periventricular leukomalacia. Data analysis Mortality data for preterm births at the Royal Women's Hospital were compared with published data for livebirths in Victoria in 1994,7 1995,8 and 1996,9 and for admissions to neonatal intensive care units in Australia and New Zealand in 1994.10Data were edited and analysed with SPSS for Windows.11 Dichotomous variables were contrasted by 2 analysis, and continuous variables were compared by Mann-Whitney U test,12 as most data were skewed. Results Causes of death There were 2475 infants of 23-36 weeks' gestational age born during the study period, of whom 118 died (4.8%). The mortality rate diminished rapidly between 23 and 28 weeks' gestational age (from 64.5% at 23 weeks to 4.0% at 28 weeks), then more slowly, to reach 0.4% at 36 weeks (Box 2). The autopsy rate in the infants who died was 51.8% (59/114), with no data on autopsies for four infants who died after transfer to other hospitals. Fifty infants died of lethal anomalies, all but two within 28 days of birth (Box 3). Lethal anomalies accounted for 11 of 67 deaths (16.4%) in infants born at 23-27 weeks' gestational age, 16 of 24 deaths (66.7%) in infants born at 28-31 weeks' gestational age, and 23 of 27 deaths (85.2%) in those born at 32-36 weeks' gestational age. Sixty-eight infants died who were without lethal anomalies: 2.8% of the 2425 livebirths, but only 0.2% of infants of 32-36 weeks' gestational age (4/1759) (Box 4). Sixty-one died within 28 days of birth and seven (10.3%) died later during the primary hospitalisation. In the Australian and New Zealand Neonatal Network, 35 of 256 deaths (13.7%) of infants of less than 32 weeks' gestational age without lethal anomalies died during the primary hospitalisation but after 28 days of age.10 Preventable causes of death Respiratory problems: The mothers of 23 of the 30 infants who died from HMD or BPD had received antenatal corticosteroids. In the remaining seven cases, the gestational ages were 23 or 24 weeks. In six cases there was not enough time to give corticosteroids, five mothers presenting in advanced preterm labour and one with eclampsia. In the seventh case, the gestational age was 23 weeks and before birth the infant was considered incapable of surviving, yet was given full intensive care postnatally. All 30 infants who died from HMD or BPD received exogenous surfactant after birth. The one infant who died from pulmonary haemorrhage without HMD received antenatal corticosteroids and did not have respiratory distress after birth and therefore did not re- ceive exogenous surfactant; the fatal pulmonary haemorrhage occurred at three days of age. The two infants who had HMD and pulmonary haemorrhage also received antenatal corticosteroids. All three infants who died from pulmonary haemorrhage had received intramuscular vitamin K at birth. None had clinical signs of heart failure or a patent ductus arteriosus before the haemorrhage. None had clinical indications for an echocardiogram, hence subclinical cardiac dysfunction cannot be excluded. Considering the 12 infants in the gestational age range 28-36 weeks who died without lethal anomalies, six died from sepsis, two from asphyxia, two from BPD, one from pulmonary haemorrhage, and one from SIDS. None died acutely from HMD. Infections: In the 22 infants who died from sepsis without NEC, there were only two in whom the management might have been different. In one infant of 30 weeks' gestational age the mother ignored the antenatal signs of sepsis for several days before presenting to hospital and delivering a moribund infant who died at 11 hours of age from E. coli pneumonia, despite full treatment, including appropriate antibiotics. In the other infant, of 33 weeks' gestation, who died of herpes simplex, the clinical presentation was that of sepsis several days before death, but, in the absence of other features of herpes in the mother or infant, no antiviral therapy was given. CVH and CPVL: Antenatal corticosteroids were given to the mothers of all nine infants in whom CVH or CPVL were major contributing causes of death. In addition to the five infants in whom CVH was a major cause of death, another 12 infants who died had a grade 3 or 4 CVH that was thought not to contribute substantially to their deaths. Comparison with regional outcomes The neonatal survival rate for infants of 24-31 weeks' gestational age at our hospital was similar to that reported in regional data from Victoria for the same three years7-9 (Box 5). In livebirths of less than 2500 g birthweight in Victoria in 1994-1996, 377 infants died within 28 days; 33.4% with lethal anomalies (compared with 44.0% in our hospital cohort; 2 = 3.7, not significant) and 5.6% died from infections (compared with 24.8% in our hospital; 2 = 32.9, P < 0.0001). The neonatal mortality rate in livebirths free of lethal malformations with birthweights greater than 2499 g was 0.45 per 1000 (81/178834) in Victoria in 1994-1996,7-9 compared with 2.27 per 1000 (4/1735) in infants of 32-36 weeks' gestational age in our hospital. The survival rate to hospital discharge in 1994 for infants free of lethal malformations cared for in neonatal nurseries at 23-31 weeks' gestational age was higher in our hospital than in reported data from the Australian and New Zealand Neonatal Network (Box 6). Discussion Comparing our data with regional data in Australia is difficult. States and territories provide data to the Australian Institute of Health and Welfare (AIHW), which then produces an annual report.13 However, the denominator in the AIHW report is predominantly determined by birthweight; gestational age is reported for confinements (ie, mothers), or for births, but not for all livebirths. Regional reports from Victoria provide data for births, including stillbirths and livebirths, by gestational age, but in two-week intervals, and only up to 31 weeks' gestational age.7-9 Moreover, the numerator in both of these regional data sets is usually limited to the neonatal period, rather than the primary hospitalisation. The Australian and New Zealand Neonatal Network collects data from all the Australian and New Zealand Neonatal Intensive Care Units.10 In this data set, the denominator is limited to admissions to neonatal units, eliminating livebirths who die outside the neonatal unit, and hence augmenting the reported survival rates. Moreover, this data set is limited to infants up to 31 weeks' gestation and excludes those with lethal malformations. The numerator, however, includes deaths in the neonatal period and those beyond 28 days of age that occur during the primary hospitalisation. Comparing data on cause of death is also difficult. Most government data collection sources, such as those in Victoria7-9 and other States, rely on confidential reports. They may not obtain enough data to classify the cause of death, and rarely can they confirm the gestational age. The final report includes an individual infant under only one cause of death, even though there may be several equally contributing causes of death, such as the common combination of HMD and airleak, which caused 25% of deaths from non-lethal causes in our study. Our system of clinicopathological conferences for each death, although not perfect, provided more detailed information about causes of death. As regional data sources do not provide complete data on deaths during the primary hospitalisation at all gestational ages, we cannot compare survival rates for preterm infants (32-36 weeks) with those of term infants (37-42 weeks). However, we have calculated rates of neonatal survival in Victoria for livebirths free of lethal malformations with birthweight greater than 2499 g, assuming that the results would be similar for infants of 37-42 weeks' gestation. In our hospital, the neonatal survival rate for infants of 32-36 weeks' gestational age (99.77%) approached the neonatal survival rate expected of term infants in Victoria (99.95%). The commonest causes of death were lethal anomalies, sepsis, and HMD. Deaths from lethal anomalies are over-represented in our hospital compared with Victoria as a whole as most have been diagnosed antenatally, and, in many cases, the mother has been transferred before birth from another hospital for management because of the fetal anomaly. The death rate from infection is higher in our hospital than for Victoria, but this is probably because infants in our hospital live longer, and hence acquire infections, rather than dying soon after birth from other problems related to prematurity. E. coli and Group B streptococci remain the chief cause of septic deaths, particularly soon after birth, consistent with the observations of Isaacs et al.14 Staphylococcus species are an increasingly frequent cause of late infections,15 and of late deaths in our study. HMD remains a leading cause of death for very preterm infants despite the fact that most mothers receive antenatal corticosteroid therapy and all infants receive exogenous surfactant. Improvements in exogenous surfactant offer hope of reducing mortality further, but other antenatal interventions, such as thyrotropin stimulating hormone,16 have been disappointing in large randomised controlled trials. HMD did not cause any deaths in our hospital in more mature preterm infants (28-36 weeks). Within the infants who died from HMD, the observation that those who also had an airleak were not as growth restricted as those who had no airleak might reflect some structural change within the lung or with use of pulmonary surfactant that is associated with growth restriction. The problem of extreme intrauterine growth restriction is highlighted by the deaths from pulmonary haemorrhage in three infants, all with birthweight ratios below 0.55. The pulmonary haemorrhage was not caused by failure to give intramuscular vitamin K at birth.17 The mechanism for fatal pulmonary haemorrhage is unclear, but may represent acute left heart failure, contributed to by a patent ductus arteriosus. The association of pulmonary haemorrhage and fetal growth restriction is well known and may be related to histological changes in the ductus arteriosus in growth-restricted fetuses.18 Pulmonary haemorrhage is also seen more frequently with exogenous surfactant therapy,19 which was given to two of the three infants in our study who died of pulmonary haemorrhage. Preventing prematurity, an obvious solution to the problem of higher death rates in preterm infants, remains an elusive goal. In all infants who died, few obvious preventable factors, apart from avoiding preterm birth, were evident. The Annual Report from the Victorian Consultative Council on Obstetric and Paediatric Mortality and Morbidity consistently identifies more avoidable factors in stillbirths than neonatal deaths.7-9 In summary, mortality rates fell sharply between 23 and 28 weeks' gestational age, and few infants of more than 28 weeks' gestational age without lethal anomalies died. Survival rates for preterm infants of more than 31 weeks' gestational age approached the survival rates expected of term infants. There were few obvious avoidable factors in the deaths of any of the infants who died, either in the very preterm infants of 23-27 weeks' gestational age, or the more numerous preterm infants of 28-36 weeks' gestational age. References The Victorian Infant Collaborative Study Group. Outcome at 2 years of children 23-27 weeks' gestation born in Victoria in 1991-92. J Paediatr Child Health 1997; 33: 161-165. Crowley P. Corticosteroids before preterm delivery (Cochrane Review). In: The Cochrane Library, Issue 2. Oxford: Update Software, 1998. [Updated quarterly.] Soll RF. Natural surfactant extract vs synthetic surfactant in the treatment of established respiratory distress syndrome (Cochrane Review). In: The Cochrane Library, Issue 2. Oxford: Update Software, 1998. [Updated quarterly.] Doyle LW, Davis P, Dharmalingam A, Bowman E. Assisted ventilation and survival of extremely low birthweight infants. J Paediatr Child Health 1996; 32:138-142. Beeby PJ, Bhutap (sic) T, Taylor LK. New South Wales population-based birthweight percentile charts. J Paediatr Child Health 1996; 32: 512-518. Northway WH Jr, Rosan RC, Porter DY. Pulmonary disease following respirator therapy of hyaline- membrane disease: bronchopulmonary dysplasia. N Engl J Med 1967; 276: 357-368. The Consultative Council on Obstetric and Paediatric Mortality and Morbidity. Annual report for the year 1994. Melbourne: Department of Human Services, 1995. The Consultative Council on Obstetric and Paediatric Mortality and Morbidity. Annual Report for the Year 1995: Incorporating the 34th Survey of Perinatal Deaths in Victoria. Melbourne: Department of Human Services, 1996. <http://hna.ffh.vic.gov.au/phb/hce/peri/ob95/index.htm> Accessed 11 May 1999. The Consultative Council on Obstetric and Paediatric Mortality and Morbidity. Annual report for the year 1996. Melbourne: Department of Human Services, 1997. <http://hna.ffh.vic.gov.au/phb/hce/peri/rep_96/contents.htm> Accessed 11 May 1999. Donoghue DA. Australian and New Zealand Neonatal Network, 1994. Sydney: AIHW National Perinatal Statistics Unit, 1996. [Neonatal Network Series no. 1.] SPSS for Windows version 6.1 [computer program]. Chicago: SPSS Inc, 1994. Moses LE, Emerson JD, Hosseini H. Analyzing data from ordered categories. N Engl J Med 1984; 311: 442-448. Lancaster P, Huang J, Lin M. Australia's mothers and babies 1993. Sydney: AIHW National Perinatal Statistics Unit, 1996. [Perinatal Statistics Series No. 3.] Isaacs D, Barfield C, Clothier T, et al. Early-onset group B streptococcal infections in Aboriginal and non- Aboriginal infants. Med J Aust 1995; 163: 302-306. Isaacs D, Barfield C, Clothier T, et al. Late-onset infections of infants in neonatal units. J Paediatr Child Health 1996; 32:158-161. Actobat Study Group. Australian collaborative trial of antenatal thyrotropin-releasing hormone (ACTOBAT) for prevention of neonatal respiratory disease. Lancet 1995; 345: 877-882. Loughnan PM, McDougall PN, Balvin H, et al. Late onset haemorrhagic disease in premature infants who received intravenous vitamin K1. J Paediatr Child Health 1996; 32: 268-269. Ibara S, Tokunaga M, Ikenoue T, et al. Histologic observation of the ductus arteriosus in premature infants with intrauterine growth retardation. J Perinatol 1994; 14: 411-416. Raju TN, Langenberg P. Pulmonary hemorrhage and exogenous surfactant therapy: a metaanalysis. J Pediatr 1996; 123: 603-610. (Received 29 Jun 1998, accepted 25 Mar 1999) Authors' details Division of Paediatrics, The Royal Women's Hospital, Melbourne, VIC. Lex W Doyle, MD, MSc, FRACP, Paediatrician, and Associate Professor, Department of Obstetrics and Gynaecology, University of Melbourne; Sheryle Rogerson, MB BS, Paediatric Fellow; Shu-Ling Chuang, MB BCh, MRCP, Paediatric Fellow; Matthew James, MB ChB, MRCP, Paediatric Fellow; Ellen D Bowman, MB BS, FRACP, Paediatrician; Peter G Davis, MD, BS, FRACP, Paediatrician. Reprints will not be available from the authors. Correspondence: Associate Professor Lex Doyle, Department of Obstetrics and Gynaecology, University of Melbourne, Parkville, VIC 3052. Email: l.doyle@obgyn-rwh.unimelb.edu.au ©MJA 1999 Journalists are welcome to write news stories based on what they read here, but should acknowledge their source as "an article published on the Internet by The Medical Journal of Australia <http://www.mja.com.au>". <URL: http://www.mja.com.au/> © 1999 Medical Journal of Australia. 1: Causes of death defined Lethal congenital anomalies: Any infants with major malformations that were considered untreatable or who died despite attempts at surgical correction, and infants with untreatable inborn errors of metabolism, chromosomal abnormalities, or overwhelming congenital infection acquired early in pregnancy. Complications of prematurity: Hyaline membrane disease (HMD), with or without an airleak: clinical signs of respiratory distress in the first days after birth, supported by radiological and/or autopsy evidence. Airleak comprised any air not confined to normal airspaces, such as pulmonary interstitial emphysema or pneumothorax. Bronchopulmonary dysplasia (BPD): respiratory distress beyond four weeks of age requiring oxygen therapy, and radiographic changes typical of bronchopulmonary dysplasia,6 or a pathological diagnosis of bronchopulmonary dysplasia at autopsy. Necrotising enterocolitis (NEC): confirmed by definite radiological, operative, or autopsy evidence. Cerebroventricular haemorrhage (CVH): included any evidence of haemorrhage of any degree identified by ultrasound examination or autopsy. Cerebroventricular haemorrhage was considered a substantial contributor to death when knowledge of an intracerebral haemorrhage (grade 4) led to withdrawal of intensive care and the infant subsequently died. Cystic periventricular leukomalacia (CPVL): included any cystic degeneration within the cerebral cortex. Cystic periventricular leukomalacia was considered the cause of death when it was severe enough to lead to the withdrawal of intensive care. Ultrasound scanning was routine within the first three days after birth, at the end of the first week, then monthly until discharge as a minimum. Pulmonary haemorrhage was diagnosed in infants with frothy, blood-stained tracheal fluid and clinical deterioration in respiratory function, or at autopsy. Perinatal asphyxia: Infants who were liveborn but failed to respond adequately to resuscitation at birth. Neonatal sepsis: Infants who died with clinical signs of sepsis supported by positive blood or cerebrospinal fluid (CSF) culture, but also included the occasional infant in whom sepsis was considered likely, but in whom blood or CSF cultures were sterile because of prior treatment with antibiotics, most commonly via the mother before birth. Pneumonia was diagnosed in infants with respiratory distress, signs of sepsis and a chest x-ray consistent with pneumonia, or at autopsy. Back to textBack to textBack to text 4: Causes of death in 68 preterm infants without lethal anomalies born at Royal Women's Hospital, Melbourne, January 1994-December 1996Median gestationalMedianMajorNumberage in weeksbirthweightcause*(%)(range)(range)Not offered intensive care7 (10.3%)23 (23-23)580 (545-725)Perinatal asphyxia5 (7.6%)26 (24-28.5)725 (630-1322)Sepsis27 (39.7%)25 (23-26)751 (606-896)NEC5 (7.4%)24 (23-29.5)756 (610-916)HMD23 (33.8%)24 (23-26)600 (522-750)HMD with airleak17 (25.0%)24 (23-25.5)670 (561-853)HMD without airleak6 (8.8%)25.5 (23-27) 509 (371-582)BPD7 (10.3%)26 (24-28)711 (640-751)CPVL4 (5.9%)25 (24-27)765 (568-858)Pulmonary haemorrhage3 (4.4%)27 (27-30)¶390 (315-756)¶Other**5 (7.4%)Total6825 (23-27)721 (590-856)MedianMedian days ofMajorbirthweight ratioage at deathcause*(range)(range)Not offered intensive care1.03 (0.94-1.23)1 (1-1)Perinatal asphyxia0.96 (0.83-1.13)1 (1-1)Sepsis1.00 (0.90-1.09)8 (1-15)NEC1.02 (0.68-1.16)14 (10-22.5)HMD0.92 (0.82-1.07)2 (1-4)HMD with airleak0.99 (0.89-1.13)2 (1-3.5)HMD without airleak0.66 (0.36-0.92)1.5 (1-5)BPH0.80 (0.66-0.92)38 (19-141)CPVL0.82 (0.76-1.13)49 (22-55)Pulmonary haemorrhage0.37 (0.30-0.54)¶5 (3-5)¶Other**Total0.95 (0.82-1.07)* Some infants had more than one major cause of death. Range = interquartile range. BPD = bronchopulmonary dysplasia Includes five infants who died with NEC. Six also had HMD, two had both HMD and CVH, and two had CVH. CPVL = cystic periventricular leukomalacia Three infants also had CVH and two had pulmonary haemorrhage. CVH = cerebroventricular haemorrhage¶ Range = complete range. HMD = Hyaline membrane disease** Two with cardiomyopathy following twin-twin transfusion syndrome, one spontaneous gut perforation, one neuroblastoma, one sudden infant death syndrome. NEC = necrotising enterocolitisInfants not offered intensive care were significantly less mature than those who died after intensive care (median gestational age 23, interquartile range 23-23 v. median gestational age 25, interquartile range 24-27; z = 3.0, P < 0.01). Excluding those with lethal anomalies and those not offered intensive care, children dying with HMD were less mature than those dying without HMD (median gestational age 24, interquartile range 23-26 v. median gestational age 26, interquartile range 24-28; z = 2.9, P < 0.01) and lighter at birth (median birthweight 600 g, interquartile range 522-750 g v. median birthweight 758 g, interquartile range 683-888 g; z = 3.0, P < 0.01), but had similar birthweight ratios (median 0.92, interquartile range 0.82-1.07 v. median 0.96, interquartile range 0.81-1.05, not significant). Within the group of infants who died of HMD, those who also had airleak were not significantly different in maturity, but were significantly heavier (z = 2.7, P < 0.01) and had higher birthweight ratios (z = 2.5, P < 0.02). The three infants who died of pulmonary haemorrhage were particularly growth restricted, with birthweight ratios of 0.30, 0.37, and 0.54, respectively; their median birthweight ratio of 0.37 was substantially below that of the infants who died of other causes (median birthweight ratio 0.96, interquartile range 0.85-1.07; z = 2.8, P < 0.01). Most infants who died did so soon after birth, except those who died of sepsis, BPD, or CPVL. Infants who died of sepsis were generally quite immature and small at birth, but were not particularly growth restricted. In infants without NEC who died of sepsis, the infectious organisms were Escherichia coli (8), Group B streptococci (3), Staphylococcus aureus(5), S epidermidis (1), Haemophilus influenzae (1), Klebsiella and Pseudomonas spp. combined (1), Streptococcus viridans (1), Candida albicans (1) and herpes simplex (1). E. coli and Group B streptococci were the predominant organisms causing early infections, and Staphylococcus species, later infections. Back to textBack to textBack to text

Lex W Doyle · Sheryle Rogerson · Shu-Ling Chuang · Matthew James · Ellen D Bowman · Peter G Davis