Research

Australian patients' delay in response to heart attack symptoms Kathleen Dracup, Sharon M McKinley and Debra K Moser MJA 1997; 166: 233 For comment see Langton and Thompson Readers may print a single copy for personal use. No further reproduction or distribution of the articles should proceed without the permission of the publisher. For permission, contact the Australasian Medical Publishing Company 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/>". Abstract - Introduction - Methods - Statistical analyses - Results - Associations with delay time - Discussion - Acknowledgements - References - Authors' details - - ©MJA1997 Abstract Objectives: To examine delay in seeking treatment among patients with an evolving acute myocardial infarction (MI), and to identify factors which contributed to this delay. Design: Patient interview combined with medical record review. Participants and setting: 317 patients with confirmed diagnosis of acute MI interviewed within 72 hours of admission to three hospitals. Main outcome measures: Delay from onset of symptoms to arrival at hospital, and cognitive, emotional, sociodemographic and clinical factors which contributed to increased prehospital delay. Results: Median prehospital delay was 6.4 hours; 41% of patients delayed less than four hours, while 28% delayed less than two hours. Prehospital delay was increased in patients with fewer years of education (P = 0.001), lower income (P = 0.003) and transportation to the hospital by private car rather than ambulance (P = 0.02). Delay time was increased by several cognitive and emotional processes (P < 0.001), such as waiting to see if symptoms would go away, being too embarrassed to ask for assistance, and not recognising the importance of symptoms. Delay time was increased with heartburn, breathlessness or intermittent symptoms and decreased with sweating and dizziness (P < 0.05). Independent predictors of increased prehospital time (P < 0.01) were fewer than 10 years of education, not wanting to trouble anyone, failing to recognise the symptoms of delay, and the intermittent nature of symptoms. Conclusion: Over 50% of acute MI patients delay seeking treatment by six hours or more. Many factors related to cognitive and social processes that contribute to this delay may be remediable with appropriate patient and community education. MJA 1997; 166: 233-236 Introduction Every year in Australia approximately 22 000 people suffer a heart attack and approximately half of these die. Most of these deaths occur out of hospital.1 The high mortality associated with acute myocardial infarction (MI) is dismaying, given that major large-scale clinical trials have shown that thrombolytic therapy can significantly reduce mortality from MI.2 With this therapy, the shorter the interval between the onset of symptoms and when the drug is given, the better the outcome.2 If therapy is started within three hours of symptoms, mortality is reduced by 23%;3 this becomes 45% if started within one hour of symptoms.3 Similarly, the earlier the treatment, the better the resulting cardiac function and the lower the incidence of morbidity after MI.4 Prehospital delay is known to be a substantial problem. Median delay times ranging from just under two hours5 to six-and-a-half hours have been reported.6 Overall, approximately 25%-50% of MI patients delay seeking medical care for cardiac symptoms for more than six hours.7,8 In other countries, the factors associated with increased prehospital delay are being older; being female; having a clinical history of hypertension, diabetes, angina or congestive heart failure; having low emotional or somatic awareness; consultation with a spouse or other relative; consultation with a doctor; and self-treatment. Decreased prehospital delay occurs when there is haemodynamic instability, large infarct size, sudden onset of severe chest pain, recognition of symptoms as heart-related and consultation with a non-relative.8,9 We do not know if the characteristics that contribute to delay in seeking treatment for acute MI symptoms are similar for Australian patients. Such information is important for designing appropriate patient and community education programs. Therefore our aims were, firstly, to describe prehospital delay time in Australian patients experiencing symptoms of evolving acute MI and, secondly, to identify the cognitive, emotional, sociodemographic and clinical factors which contribute to this delay. Methods We gained appropriate institutional ethics committee review and approval for the study from the study hospitals and the University of Technology, Sydney. Eligible patients admitted to three hospitals between 1 March and 31 July 1995 were invited to participate in the study and gave written informed consent. Patients were eligible if they were able to read and write English, lived independently, were free from malignancy or other complicating illness, alert and orientated, were within 24 and 72 hours of hospital admission, and had a diagnosis of acute MI, determined by:Chest pain lasting longer than 20 minutes; Serial electrocardiographic evidence of acute MI (i.e., appearance of new Q waves in at least two leads of a 12-lead standard electrocardiogram); and Elevations in serum creatine kinase-MB consistent with evolving MI. Only patients with Q-wave MI were studied. Patients were excluded if they were being mechanically ventilated. The hospitals were two metropolitan university-affiliated city teaching hospitals and a district hospital, chosen because their geographical catchment areas were relatively small, so transportation time would not add significantly to the time between symptom onset and arrival at the hospital. We reviewed medical records to obtain sociodemographic and clinical data. We assessed information about the cognitive, emotional and social processes surrounding the decision to seek treatment using the Response to symptoms questionnaire, designed to obtain information about patient delay and factors contributing to delay in six domains: (1) context in which MI symptoms appeared; (2) antecedents of symptoms; (3) affective response to symptoms; (4) behavioural response to symptoms; (5) cognitive response to symptoms; and (6) the response of others to patient symptoms.10 Based on growing evidence that cognitive, symptom-appraisal and social factors may be more important determinants of delay than knowledge about appropriate re sponse to symptoms,10 we modified the instrument to further assess cognitive, symptom-appraisal and social factors surrounding a patient's decision to seek care for acute MI symptoms. The resulting instrument assesses factors surrounding the onset of acute MI symptoms, the response of others to symptoms, and the patient's response to symptoms. We established the time from symptom onset to hospital admission by reviewing the medical record for the time stated in the admission record by the admitting physician and the time of admission to the hospital. Also, each patient was asked about the time of first symptoms. If symptoms had come and gone, patients were carefully interviewed to determine the nature of those symptoms and the exact time that cardiac symptoms began. Where the patient's account differed from the time recorded by the admitting doctor, cardiac enzymes and serial electrocardiograms were reviewed to identify the most likely time of symptom onset. The value that most closely matched the expected clinical course reflected in the tests was chosen. Patients were asked if they wanted to complete the questionnaire themselves or if they wanted the researcher to read the questions; 96% chose the latter. Statistical analyses Independent t tests or analysis of variance (with post-hoc t tests adjusted for multiple comparisons) were used to assess the impact of sociodemographic, clinical, cognitive and social factors. Delay time was markedly skewed, so we used square root transformations to obtain a normal distribution, and all analyses were performed on transformed values. Multiple logistic regression and χ2 tests were used to determine odds ratios for delays of more than two hours among several sociodemographic, clinical, social and cognitive variables. Variables that were significantly different (P < 0.05) on delay time were entered into the forward stepwise logistic regression. Criteria for entry and removal of variables were based on the likelihood ratio test, with enter and remove limits set at P < 0.05 and P > > 0.05. Results Of 321 eligible patients approached, 317 agreed to participate in the study; four declined, giving fatigue as a reason. Patients were predominantly older (mean, 63 years; standard deviation, 12 years), male, married and white. There was diversity in years of formal education and socioeconomic status. Most patients experienced their symptoms at home and most went to hospital by ambulance (Box 1). Many patients were at greater than average risk for suffering an acute MI, with previous medical histories of angina pectoris (43%), MI (28%), hypertension (53%) and/or diabetes mellitus (19%). Most MIs were inferior (46%) or anterior (32%). Median prehospital delay time was 6.4 hours (interquartile range, 1.9-19.9 h; mean, 17.3 h). Only 14% of patients arrived within one hour of the onset of symptoms, 28.5% within two hours and 41% within four hours. However, 54% arrived at the hospital more than six hours after first experiencing cardiac symptoms. One hundred and fifty-three patients (48%) received thrombolytic therapy; 29 within one hour of the onset of symptoms. Associations with delay time Sociodemographic factors: Box 1 shows that there were differences in delay time according to education, annual income and use of ambulance. There was no significant association with sex, age, ethnicity, the location where symptoms occurred or whether or not a spouse was present when they occurred. Symptoms and pain: We examined breathlessness, fatigue, weakness, vague chest discomfort, chest pain, neck pain, arm numbness, pain between shoulder blades, back pain, nausea, feeling that something was wrong, dia phoresis, feeling clammy, heartburn, and dizziness. Delay time was less if patients were dia phoretic (P = 0.001) or experienced dizziness (P = 0.03) and increased if they were breathless (P = 0.03) or experienced heartburn (P = 0.03). The remaining symptoms made no difference to delay times. If symptoms were intermittent, patients took longer to seek treatment (P = 0.0001), while those who believed their symptoms to be cardiac in nature delayed for shorter times (P = 0.001). However, 60% originally attributed their symptoms to another cause (e.g., muscle pain, fatigue, indigestion). Patients who appraised their pain as very serious delayed a shorter time than patients who appraised their pain as not serious (P = 0.001). However, patients who rated their initial chest pain as severe had similar delay times to patients who rated their pain as mild (P = 0.644). Knowledge of coronary heart disease: There were no differences in the average time to arrival at hospital related to previous myocardial infarction, previous cardiac surgery, diabetes mellitus, hypertension or angina. Forty per cent of patients reported that their source of information about MI symptoms was television or reading; 27% received their information from a friend; 23% from a doctor; and 10% from a nurse. Sixteen per cent of patients knew about thrombolysis therapy for acute MI and these, on average, came to the hos pital earlier than patients who did not know about this treatment (P = 0.001). Cognitive and psychological factors: Box 2 (below) shows significant associations between patients' responses and delays in seeking medical treatment. Independent predictors of delay: We tested factors that were significant on univariate analyses in a multiple regression model. The five independent predictors of delay are shown in Box 3. Discussion In 317 patients admitted to hospital for evolving acute MI, we found that median prehospital delay was 6.4 hours. Almost 60% of patients delayed more than four hours, while more than 70% delayed longer than two hours. These times are substantially longer than those recorded in most studies from other countries,4-6,8,11 as well as those documented in previous Australian studies.12,13 The disparity between this and previous Australian studies is un doubtedly related to a difference in the definitions used to characterise prehospital delay. In the first Australian study, time from symptom onset to the first attempt to get help was reported.12 In the second, the definition of total delay time was the period from the onset of chest pain to arrival at hospital; where there were multiple episodes of chest pain, the time from the last episode of pain was used.13 Both variations in definitions would shorten reported delay time relative to the definition we used. It is unlikely that longer transportation times to the hospital contributed to longer delay times in our study because of the geographic situation of the participating hospitals and the high rate of ambulance use. Moreover, all studies on this topic have found that home-to-hospital time, as well as in-hospital time, play a minor role in the delay that occurs before definitive treatment.8,14,15 We included only those patients who survived to hospital admission and who met our inclusion criteria, and hence our patients were clearly not representative of all patients with acute MI. Unfortunately, approximately 30% of such patients die of sudden cardiac death before reaching the hospital,1 and we excluded patients because they were unable to respond to questions within the designated 72 hours of hospital admission (e.g., because they required mechanical ventilation or had compromised mental status). In the past, most researchers have focused on identifying patients who are more likely to delay a decision to come to the hospital because of various sociodemographic and clinical characteristics. In our study, the most important factor predicting delay was years of formal education, a characteristic that health professionals cannot alter. However, it may be linked with other characteristics (e.g., income, age, coping styles or other underlying cognitive mechanisms affecting delay), some of which may be amenable to change. Many of our patients had difficulty identifying their symptoms as cardiac in nature, initially thinking they were gastrointestinal, musculoskeletal, or fatigue. Many patients knew the classic signs of acute MI but did not know the diversity of its presentation, notably its often intermittent nature. Patients should be told that the symptoms may come on gradually or may not be constant, as many believe that MI is accompanied by sudden, crushing chest pain and unconsciousness. As in other studies,5,12,16,17 experience of a previous MI did not result in patients coming to the hospital sooner. Thus, the public, particularly patients at high risk for a future MI, need to be informed that the manifestations of an evolving MI may vary, even within the same individual, and should be clear about the actions to take if and when symptoms occur. In most studies in the United States, less than half of MI patients arrive by ambulance,18 whereas in our study most patients in all three hospitals used this means of transportation. Ambulance is preferable to other types of transportation because of the early treatment available and because delay times are usually reduced.18 From the accounts of patients in this study, doctors and nurses played minimal roles in providing information about the symptoms of MI, compared with knowledge gained from television, reading and friends. Counselling strategies tested in other clinical populations support the assumption that individuals who are prepared for certain signs and symptoms will delay less in seeking care than individuals who have no such preparation.19-21 They also suggest that health professionals should take a more proactive and intensive approach to educating patients about the nature of MI signs and symptoms and the steps to take to receive early treatment. These data also indicate the opportunity for professional organisations to enhance the knowledge and beliefs of patients by community educational programs via television, radio and newspapers. Very few patients knew about thrombolysis and its effectiveness in treating acute MI, but those who did presented sooner than those who did not. Based on these findings, health care professionals should emphasise the rationale for early treatment to patients at high risk for future MI, and public campaigns should advertise the benefit of seeking early treatment. Patients need to know that their decision to seek immediate treatment has important future rewards as this may override whatever trepidation they experience in acknowledging that they might be having a heart attack. In summary, almost all the significant predictors of delay we identified related to social and cognitive processes that might be amenable to change. Health care professionals should recognise the reluctance of patients to recognise MI symptoms as cardiac and provide them with the necessary guidance, so that they will seek care as soon as possible. The availability of newer forms of treatment, particularly thrombolytic therapy, provides a compelling reason for physicians and nurses to encourage patients and their families to seek care early to preserve ischaemic myocardium and reduce infarct size in the face of an evolving MI. Acknowledgements Funding for this study was provided by a Fulbright Foundation Senior Scholars grant and from the L W Hassenplug Chair funds. We thank the following nurses who assisted with data collection: Patricia Cameron, Marie Droulers, Cameron Goodear, Jennifer Kidd, Jacqueline Padley, Rebecca Scammell and Jennifer Williams. References Heart and stroke facts. Canberra: National Heart Foundation of Australia, 1995. Fibrinolytic Therapy Trialists' (FTT) Collaborative Group. Indications for fibrinolytic therapy in suspected acute myocardial infarction: collaborative overview of early mortality and major morbidity results from all randomised trials of more than 1000 patients. Lancet 1994; 343: 311-322. Gruppo Italiano per lo Studio della Stretochinasi nell'Infarto Miocardico (GISSI). Effectiveness of intravenous thrombolytic treatment in acute myocardial infarction. Lancet 1986; 1: 397-401. Simoons ML, Serruys PW, Brand MVD, et al. Early thrombolysis in acute myocardial infarction: Limitation of infarct size and improved survival. J Am Coll Cardiol 1986; 7: 717-721. Maynard C, Althouse R, Olsufka M, et al. Early versus late hospital arrival for acute myocardial infarction in the western Washington thrombolytic therapy trials. Am J Cardiol 1989; 63: 1296-1300. Cooper RS, Simmons B, Castaner A, et al. Survival rates and prehospital delay during myocardial infarction among black persons. Am J Cardiol 1986; 57: 208-211. Dracup K, Moser D. Treatment seeking behavior among those with symptoms and signs of acute myocardial infarction. Heart Lung 1991; 20: 570-575. GISSI-Avoidable Delay Study Group. Epidemiology of avoidable delay in the care of patients with acute myocardial infarction in Italy. Arch Intern Med 1995; 155: 1481-1488. Dracup K, Moser D, Eisenberg M, et al. Causes of delay in seeking treatment for symptoms of acute myocardial infarction. Soc Sci Med 1995; 40: 379-392. Burnett RE, Blumenthal JA, Mark DB, et al. Distinguishing between early and late responders to symptoms of acute myocardial infarction. Am J Cardiol 1995; 75: 1019-1022. Schwarz B, Schoberberger R, Rieder A, Dunze M. Factors delaying treatment of acute myocardial infarction. Eur Heart J 1994; 15: 1595-1598. Bett N, Aroney G, Thompson P. Impact of a national educational program to reduce patient delay in possible heart attack. Aust N Z J Med 1993; 23: 157-161. Leitch JW, Birbara T, Freedman B, et al. Factors influencing the time from onset of chest pain to arrival at the hospital. Med J Aust 1989; 150; 6-8. Schroeder JS, Lamb IH, Hu M. The pre-hospital course of patients with chest pain: analysis of the prodromal, symptomatic, decision-making, transportation and emergency room periods. Am J Med 1978; 64: 742-748. Schmidt SB, Borsch MA. The prehospital phase of acute myocardial infarction in the era of thrombolysis. Am J Cardiol 1990; 65: 1411-1415. Turi ZG, Stone PH, Muller JE, et al. Implications for acute intervention related to time of hospital arrival in acute myocardial infarction. Am J Cardiol 1986; 58: 203-209. Weilgosz ATJ, Nolan RP, Earp JA, Biro E. Reasons for patients' delay in response to symptoms of acute myocardial infarction. Can Med Assoc J 1988; 139: 853-857. Meischke H, Ho MT, Eisenberg MS, et al. Reasons patients with chest pain delay or do not call 911. Ann Emerg Med 1995; 25: 193-197. Bandura A. Self-efficacy; toward a unifying theory of behavior change. Psychol Bull 1977; 84: 191-198. Bailey WC, Richards JM Jr, Brooks CM, et al. A randomized trial to improve self-management practices of adults with asthma. Arch Intern Med 1990; 150: 1664-1667. Avis NE, Smith DW, McKinlay JB. Accuracy of perceptions of heart attack risks: what influences perceptions and can they be changed? Am J Public Health 1989; 17: 1608-1611. (Received 28 Mar, accepted 20 Nov, 1996) Authors' details University of California, Los Angeles, CA, USA. Kathleen Dracup, RN, DNSc, L W Hassenplug Professor of Nursing. University of Technology, Sydney, and Royal North Shore Hospital, Sydney, NSW. Sharon M McKinley, RN, PhD, Professor of Critical Care Nursing. College of Nursing, Ohio State University, Columbus, OH, USA. Debra K Moser, RN, DNSc, Assistant Professor. Reprints: Professor S M McKinley, Level 6, Royal North Shore Hospital, St Leonards, NSW 2065. E-mail: smckinle AT doh.health.nsw.gov.au - - To top of article - ©MJA 1997 <URL: http://www.mja.com.au/> © 1997 Medical Journal of Australia.

Kathleen Dracup · Sharon M McKinley · Debra K Moser

"Mental health literacy": a survey of the public's ability to recognise mental disorders and their beliefs about the effectiveness of treatment Anthony F Jorm, Ailsa E Korten, Patricia A Jacomb, Helen Christensen, Bryan Rodgers and Penelope Pollitt MJA 1997; 166: 182 Abstract - Introduction - Methods - Sample - Interview - Ethical approval - Results - Recognition - Choice and rating of available help - Prognosis - Discussion - References - Authors' details - ©MJA1997 Abstract Objectives: To assess the public's recognition of mental disorders and their beliefs about the effectiveness of various treatments ("mental health literacy"). Design: A cross-sectional survey, in 1995, with structured interviews using vignettes of a person with either depression or schizophrenia. Participants: A representative national sample of 2031 individuals aged 18 -74 years; 1010 participants were questioned about the depression vignette and 1021 about the schizophrenia vignette. Results: Most of the participants recognised the presence of some sort of mental disorder: 72% for the depression vignette (correctly labelled as depression by 39%) and 84% for the schizophrenia vignette (correctly labelled by 27%). When various people were rated as likely to be helpful or harmful for the person described in the vignette for depression, general practitioners (83%) and counsellors (74%) were most often rated as helpful, with psychiatrists (51%) and psychologists (49%) less so. Corresponding data for the schizophrenia vignette were: counsellors (81%), GPs (74%), psychiatrists (71%) and psychologists (62%). Many standard psychiatric treatments (antidepressants, antipsychotics, electroconvulsive therapy, admission to a psychiatric ward) were more often rated as harmful than helpful, and some non-standard treatments were rated highly (increased physical or social activity, relaxation and stress management, reading about people with similar problems). Vitamins and special diets were more often rated as helpful than were antidepressants and antipsychotics. Conclusion: If mental disorders are to be recognised early in the community and appropriate intervention sought, the level of mental health literacy needs to be raised. Further, public understanding of psychiatric treatments can be considerably improved. MJA 1997; 166: 182-186 Introduction "Health literacy" has been defined as the ability to gain access to, understand, and use information in ways which promote and maintain good health.1 By extension, we have coined the term "mental health literacy" to refer to knowledge and beliefs about mental disorders which aid their recognition, management or prevention. Mental health literacy includes the ability to recognise specific disorders; knowing how to seek mental health information; knowledge of risk factors and causes, of self-treatments, and of professional help available; and attitudes that promote recognition and appropriate help-seeking. The lifetime risk of developing a mental disorder is so high (nearly 50%)2 that almost the whole population will at some time have direct experience of such a disorder, either in themselves or in someone close. A high public level of mental health literacy would make early recognition of and appropriate intervention in these disorders more likely. Previous information on this topic is limited and is derived from national surveys on depression alone,3-5 or on depression and schizophrenia.6 Although these surveys found that most people believed depression to be treatable,3-5 most respondents had negative views about the effectiveness of medication for mental disorders. In contrast, counselling and psychotherapy were generally viewed more favourably.3,4,6 To assess the mental health literacy of the Australian population, we surveyed a representative national sample of adults on their knowledge of and beliefs about schizophrenia and depression. We report our findings on the ability of this population to recognise these disorders and their beliefs about the effectiveness of various treatments. Methods Sample The survey was carried out by the Australian Bureau of Statistics in August 1995 as part of its Population Survey Monitor.7 This is a household survey covering all private dwellings in urban and rural areas (excluding the sparsely settled areas) across all States and Territories. Selected households were initially sent a letter explaining that their dwelling had been selected for the survey. The letters gave advance notice that an interviewer would call to make an appointment. Interviewers made at least three call-backs in rural areas and at least five in urban areas before a dwelling was classified as "non-contact". Contact was made with a sample of 2531 households, with one person randomly sampled per household for a personal interview; 2164 persons agreed to participate (85%). Because a pilot study showed that people aged more than 75 years often had trouble understanding the interview, this age group was excluded, leaving a sample of 2031 respondents, aged 18-74. Fifty-six per cent of the sample was female and 74% Australian-born. The age distribution was: 21% aged 18-29, 25% aged 30-39, 22% aged 40-49, 16% aged 50-59, 11% aged 60-69 and 5% aged 70-74. The highest educational qualification was: secondary school certificate (51%), trade certificate/apprenticeship (11%), other certificate (17%), associate or undergraduate diploma (7%), bachelor's degree or higher (13%), still at school (1%). Weights were provided for each respondent, based on complex ratio estimation procedures, to adjust for probabilities of selection and to reduce non-response bias.7 Weighted percentages, which represent estimates of the whole of the Australian population aged 18-74, are presented here. Interview The interview was based on a vignette of a person suffering from a mental disorder. Half the sample were shown a vignette describing a person who met ICD-108 and DSM-IV9 criteria for major depression ( Box 1) and the others were shown a vignette of a person who met ICD-108 and DSM-IV9 criteria for schizophrenia ( Box 2). The sex of the person described was randomly assigned to be male (John) or female (Mary). After being shown the vignette and having it read out to them, respondents were asked two open-ended questions: "What would you say, if anything, is wrong with John/Mary?" and "How do you think John/Mary could best be helped?" The rest of the interview consisted of questions to determine the respondents' knowledge of and views about: Various people who could help (whether each category of person was likely to be helpful, harmful, or neither, for the person described); A range of possible treatments (whether each treatment was likely to be helpful, harmful, or neither, for the person described); Knowledge of likely prognosis; Knowledge of risk factors; and Beliefs associated with stigma and discrimination. Ethical approval Approval was obtained from the Ethics in Human Experimentation Committee of the Australian National University. Statistical analysis Using the chi-squared test, all estimates were compared according to recognition of a mental health problem. Only differences significant at the 0.01 level (P < 0.01) are reported below. Results Of the 2031 persons interviewed, 1010 were shown the depression vignette (508, John and 502, Mary) and 1021 were shown the schizophrenia vignette (514, John and 507, Mary). Recognition Figure 1 summarises responses to the question "What would you say, if anything, is wrong with John/Mary?", and shows those categories mentioned by at least 5% of the respondents (all responses were later categorised by the researchers). Multiple responses were allowed, and 30% of respondents gave at least two answers. For the depression vignette, 39% correctly identified depression and 22% mentioned stress. In all, 72% mentioned a category that could be regarded as being within the sphere of mental health. Eleven per cent mentioned items that we categorised as physical disorders (e.g., viruses, nutritional deficiencies, cancer), and half of these respondents did not mention a mental problem. A further 17% gave only answers that were extremely variable, but which we grouped as "personal or employment-related problems", "problems with not being active or sociable enough", and "other". Seven per cent of the sample responded with "don't know". For the schizophrenia vignette, although 84% mentioned at least one category in the sphere of mental health, only 27% recognised schizophrenia and a further 26% mentioned depression. Physical disorders were the only suggestion from 2% of the respondents, while 13% gave responses that described neither physical nor mental disorders (e.g., "has a problem"). There was less uncertainty with the schizophrenia vignette, however, in that only 4% responded with "don't know". Choice and rating of available help For the second open-ended question -- "How do you think John/Mary could best be helped?" -- 34% of the respondents (across both vignettes) made more than one suggestion. For the depression vignette, the most frequent response was "see a doctor" (44%), followed by "see a counsellor" (23%) and "talk over with family or friends" (20%). A psychiatrist was mentioned by 8%, while 5% answered "don't know". Responses for the schizophrenia vignette were: counsellor (31%), psychiatrist (28%), doctor (27%), family or friends (20%) and "don't know" (4%). The respondents were given a list of people who might potentially provide help and were asked to rate the various helpers by saying whether each would be helpful or harmful (Figure 2a). For the depression vignette, most of the respondents regarded GPs (83%), counsellors (74%), close friends (73%) and close family (70%) as helpful; around half the population rated telephone counselling services (53%), psychiatrists (51%) and psychologists (49%) as helpful. Fewer than 10% felt that any of the above groups would be harmful, although 43% believed it would be harmful for someone with depression to deal with it on their own. For the schizo phrenia vignette, most respondents regarded counsellors (81%), GPs (74%) and psychiatrists (71%) as helpful; a larger proportion of the population than for the depression vignette believed it would be harmful to try and deal with such problems alone (55%). Rating of pharmacological treatments Respondents were given a list of pharmacological treatments (Figure 2b) to rate as helpful or harmful. For the depression vignette, more of the respondents regarded each of the medications as harmful than helpful. The exception was the category vitamins, minerals, tonics or herbal medicines, which were regarded as helpful by 57% of respondents, and as harmful by 3%. Antidepressant medication was recognised as helpful by 29% and as harmful by 42% of respondents. For the schizophrenia vignette, antidepressants were regarded as helpful by 38% of respondents, followed by vitamins and minerals (34%) and antipsychotics (23%). The greatest percentage of "don't know" responses was for antipsychotics (about one-fifth of the respondents for both vignettes). Rating of non-pharmacological treatments When respondents were asked to rate non-pharmacological treatments (Figure 2c), most (for both the depression and the schizophrenia vignettes) regarded non-standard interventions (more physical or social activity; learn relaxation [including stress management, meditation or yoga courses]; reading about people with similar problems) as helpful and not harmful. On the other hand, most regarded admission to a psychiatric ward as harmful (depression, 62%; schizophrenia, 51%) and most regarded having electroconvulsive therapy (ECT) as harmful (depression, 72%; schizophrenia, 66%). For the depression vignette, psychotherapy was seen as helpful by 34% and harmful by 13%, compared with 55% helpful and 7% harmful for the schizophrenia vignette. The highest number of "don't know" responses was elicited for psychotherapy (16% for depression, 15% for schizophrenia) and for ECT (10% for depression and 14% for schizophrenia) (data not shown). As opinions about treatment might vary according to whether or not the respondent thought the person in the vignette had a mental health problem, the respondents were divided accordingly. The major difference in findings was that those who did not perceive a mental health problem were more likely to rate treatments as "neither helpful nor harmful" or to respond "don't know" . However, the rank ordering of treatments in terms of helpfulness was generally similar. Spearman rank correlation coefficients for the depression vignette were 0.82 (people), 0.90 (medicines) and 0.98 (treatments), and for the schizophrenia vignette they were 0.87 (people), 0.71 (medicines) and 0.98 (treatments). Prognosis All respondents were asked to give their views on prognosis with and without the professional help they thought most appropriate. For the depression vignette, 80% thought that there would be full recovery with help. If there was no help, 56% believed the person would get worse, and 5% that there would be full recovery. For the schizophrenia vignette, 69% believed that help would result in full recovery; if there was no help, 75% believed that the person would get worse, and 3% that there would be full recovery. Discussion Recognition of the presence of a mental disorder was high in our population sample, although only a minority gave the correct psychiatric label to their vignette. While it is not known whether there is any benefit to the public in being able to apply the correct psychiatric label, misidentifying a mental disorder as a physical one or as a problem unrelated to health may lead to inappropriate use or avoidance of health services. The major limitation in recognition is therefore seen in the 28% who thought the person described in the depression vignette did not have a mental disorder and the 16% who had the same opinion about the person in the schizophrenia vignette. When respondents were asked about the helpfulness of various people, GPs were rated very highly for both vignettes. Only half the respondents thought that a psychiatrist or psychologist would be helpful for the person in the depression vignette, a proportion less than that cited for GPs, counsellors, close friends, family, and telephone counselling. While psychiatrists and psychologists were rated as relatively more helpful for the person in the schizophrenia vignette, they were nevertheless less likely to be rated as helpful than counsellors or GPs. This suggests that public perceptions of mental health specialists need to be changed. Ratings given for the helpfulness of various treatments for depression are not consistent with the evidence of controlled trials, which have indicated that both antidepressant medication and psychotherapy are effective treatments. 10,11 Antidepressants were rated as helpful by 29% of our sample and as harmful by 42%, while psychotherapy was rated as helpful by 34% and harmful by 13%. Both were regarded as less helpful than treatments such as vitamins and minerals and special diets. The treatment with the highest negative rating was ECT. Although the patient described to the respondents could not be regarded as severely depressed enough to warrant ECT, 11 there is clearly a public perception that this treatment is harmful. The treatments that the public rated most highly were all non-standard in nature. These views may not be entirely misguided; there is evidence (e.g., from controlled trials) that physical exercise may have a positive effect on depression. 12 The findings were similar for the schizophrenia vignette. Although controlled trials show that antipsychotic medication is an effective treatment, 13 this was rated as helpful by 23% of the respondents and harmful by 34%; 20% did not offer an opinion. Similarly, admission to a psychiatric ward, which can be useful in the management of schizophrenia, 13 was rated as harmful by half the respondents. As with depression, non-standard interventions were the most likely to be rated as helpful. Despite these negative opinions of, or ignorance about, the helpfulness of many standard treatments, the public clearly sees the conditions described in the vignettes as treatable. The predominant belief that mental disorders are treatable has also been found in overseas surveys, 3,4 although a United States survey found that most respondents believed it possible to get better through one's own efforts. 5 There were some marked differences in responses to the depression and schizophrenia vignettes in terms of recognition, perceived helpfulness of treatments and prognosis. These differences show that the respondents did not see all mental disorders as the same and recognised that the condition described in the schizophrenia vignette required more vigorous intervention. Our results also indicate that the views of many members of the public diverge from those of health professionals, particularly mental health specialists. Such differences may lead to unwillingness to accept help from mental health professionals, or to a lack of adherence to advice given. Clearly, if mental disorders are to be recognised early and appropriate action taken, the level of mental health literacy in the population should be raised. There has been considerable interest in trying to improve the recognition and management of mental disorders in primary care, 10,14 but this knowledge needs to reach the consumers of services so that they can play a more effective role in the management of their own mental health. References Nutbeam D, Wise M, Bauman A, et al. Goals and targets for Australia's health in the year 2000 and beyond. Canberra: Australian Government Publishing Service, 1993. Kessler RC, McGonagle KA, Zhao S, et al. Lifetime and 12-month prevalence of DSM-III-R psychiatric disorders in the United States: Results from the National Comorbidity Survey. Arch Gen Psychiatry 1994; 51: 8-19. McKeon P, Carrick S. Public attitudes to depression: a national survey. Ir J Psychol Med 1991; 8: 116-21. Sims A. The scar that is more than skin deep: the stigma of depression. Br J Gen Pract 1993; 43: 30-31. Regier DA, Hirschfeld RM, Goodwin FK, et al. The NIMH depression awareness, recognition, and treatment program: structure, aims, and scientific basis. Am J Psychiatry 1988; 145: 1351-1357. Angermeyer MC, Matschinger H. Public attitude towards psychiatric treatment. Acta Psychiatr Scand 1996; 94: 326-336. Australian Bureau of Statistics. Population survey monitor, August 1995 (No. 4103.0). Adelaide: ABS, 1995. World Health Organization. The ICD-10 classification of mental and behavioural disorders. Diagnostic criteria for research. Geneva: WHO, 1993. American Psychiatric Association. Diagnostic and statistical manual of mental disorders (4th ed) (DSM-IV). Washington DC: APA, 1994. Depression Guideline Panel. Depression in primary care: Volume 2. Treatment of major depression. Clinical practice guideline, number 5. Rockville, MD: US Department of Health and Human Services, Public Health Service, Agency for Health Care Policy and Research, 1993. The Quality Assurance Project. A treatment outline for depressive disorders. Aust N Z J Psychiatry 1983; 17: 129-146. Byrne A, Byrne DG. The effect of exercise on depression, anxiety and other mood states: a review. J Psychosom Res 1993; 37: 565-574. The Quality Assurance Project. A treatment outline for the management of schizophrenia. Aust N Z J Psychiatry 1984; 18: 19-38. Ustun TB, Goldberg DP, Cooper JE, et al. A new classification for mental disorders with management guidelines for use in primary care: The ICD-10 PHC. Br J Gen Pract 1995; 45: 211-215. (Received 22 Feb 1996, accepted 4 Nov, 1996) Authors' details NHMRC Social Psychiatry Research Unit, The Australian National University, Canberra, ACT. Anthony F Jorm, PhD, DSc, Deputy Director; Ailsa E Korten, BSc, Research Officer; Patricia A Jacomb, MSc, Research Assistant; Helen Christensen, PhD, Fellow; Bryan Rodgers, PhD, Fellow; Penelope Pollitt, PhD, Research Fellow. No reprints will be available from the author. Correspondence: Dr A F Jorm, NHMRC Social Psychiatry Research Unit, The Australian National University, Canberra, ACT 0200. E-mail: Anthony. Jorm AT anu.edu.au - - - To top of article - ©MJA 1997 <URL: http://www.mja.com.au/> © 1997 Medical Journal of Australia.

Anthony F Jorm · Ailsa E Korten · Patricia A Jacomb · Helen Christensen · Bryan Rodgers · Penelope Pollitt

End-of-life decisions in Australian medical practice Helga Kuhse, Peter Singer, Peter Baume, Malcolm Clark and Maurice Rickard MJA 1997; 166: 191 For editorial comment see Van Der Weyden and Ashby Readers may print a single copy for personal use. No further reproduction or distribution of the articles should proceed without the permission of the publisher. For permission, contact the Australasian Medical Publishing Company 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/>". Abstract - Introduction - Methods - Results - Discussion - Acknowledgement - References - Authors' details - - - ©MJA1997 Abstract Objective: To estimate the proportion of medical end-of-life decisions in Australia, describe the characteristics of such decisions and compare these data with medical end-of-life decisions in the Netherlands, where euthanasia is openly practised. Design: Postal survey, conducted between May and July 1996, using a self-administered questionnaire based on the questionnaire used to determine medical end-of-life decisions in the Netherlands in 1995. Participants: A random sample of active medical practitioners from all Australian States and Territories selected from medical disciplines in which there were opportunities to be the attending doctor at non-acute patient deaths, and hence to make medical end-of-life decisions. Main outcome measure: Proportion of Australian deaths that involved a medical end-of-life decision, using ratio-to-size estimation based on the sampled doctors' responses to the questionnaire. The response rate was 64%. Results: The proportion of all Australian deaths that involved a medical end-of-life decision were: euthanasia, 1.8% (including physician-assisted suicide, 0.1%); ending of patient's life without patient's concurrent explicit request, 3.5%; withholding or withdrawing of potentially life-prolonging treatment, 28.6%; alleviation of pain with opioids in doses large enough that there was a probable life-shortening effect, 30.9%. In 30% of all Australian deaths, a medical end-of-life decision was made with the explicit intention of ending the patient's life, of which 4% were in response to a direct request from the patient. Overall, Australia had a higher rate of intentional ending of life without the patient's request than the Netherlands. Conclusions: Australian law has not prevented doctors from practising euthanasia or making medical end-of-life decisions explicitly intended to hasten the patient's death without the patient's request. MJA 1997; 166: 191-196 Introduction In Australia, the practice of euthanasia or physician-assisted suicide (Box 1) is unlawful, except in the Northern Territory. It is, however, widely recognised in common law (in some Australian States and Territories in statute law) that mentally competent patients have a right to refuse medical treatment, and that doctors who comply with a patient's request for non-treatment are not generally deemed to have acted unlawfully or to have practised euthanasia or physician-assisted suicide, even if their action or omission was foreseen to lead to the patient's death. Before the Northern Territory Rights of the Terminally Ill Act 1995, the only country where doctors could openly practise euthanasia was the Netherlands. Although this practice is subject to criminal law in the Netherlands, it has been long accepted by strong public support, and the Dutch Parliament has recognised that doctors practising it should not be prosecuted if they comply with certain guidelines.1,2 In 1990, the Dutch Government appointed the Remmelink Commission to review the nationwide practice of euthanasia and other end-of-life medical practices. The investigators used three different methods to ascertain the incidence of these practices. In order to facilitate comparisons with our study, we summarise the findings of their questionnaire study.3 Of all deaths in the Netherlands in 1990 (approximately 129 000), 1.7% were the result of euthanasia and 0.2% were the result of physician-assisted suicide; these deaths complied with Royal Dutch Medical Association guidelines on euthanasia. However, in a clear violation of the guidelines, 1000 deaths (0.8%) involved the administration of drugs with the explicit intention of ending the patient's life, but without the patient's explicit request. In a second nationwide study of medical end-of-life practices in the Netherlands in 1995, by van der Maas et al.,4 the corresponding findings were: euthanasia, 2.3%; physician-assisted suicide, 0.2%; and ending of life without the explicit concurrent request of the patient, 0.7%. Although recent Australian surveys have reported on doctors' attitudes about medical end-of-life decisions,5 and doctors' attitudes to and practice of voluntary euthanasia,6,7 no study has examined the number of deaths in Australia involving medical end-of-life decisions. In this study, we sought to estimate the number of deaths involving a medical end-of-life decision in Australia, to describe the characteristics of such decisions, and to compare these data with those of the Dutch studies. A comparison of medical end-of-life decisions in the Netherlands with a country where euthanasia cannot be practised openly, such as Australia, would indicate whether the Netherlands has a higher rate of ending of life without the patient's explicit request. Australia and the Netherlands have a similar population size (18 million and 15.3 million, respectively) and comparable annual death rates (7.1 and 8.7 per thousand, respectively).8 In the Netherlands in 1995 there were 135 546 deaths,3 while in Australia there were 125 771 deaths from July 1994 to June 1995.9 Methods Ethical approval Ethical approval for this study was obtained from the Monash University Standing Committee on Ethics in Research on Humans. Questionnaire We used an English version of the retrospective questionnaire used by van der Maas et al.4 Our questionnaire was based on an authoritative translation provided by the Dutch researchers, so the terminology and meanings of the questions were comparable, and contained the original 24 items used in their survey. We considered anonymity to be essential because we were seeking information about conduct for which doctors would be liable to prosecution. Therefore, unlike the study by van der Maas et al.,4 we did not conduct interviews. Grouping euthanasia and physician-assisted suicide together, our questionnaire addressed all medical end-of-life decisions, as defined in Box 1. The questionnaire examined the medical action that preceded the death, the doctor's intention, whether an explicit request from the patient was involved, whether a wish had been expressed to hasten death, whether discussion took place with the patient or others, whether the patient was competent when making the request or when participating in discussions, and the estimated length of life shortened. Most questionnaires were completed carefully and consistently. (Copies of the questionnaire are available from the authors.) Medical practitioner sample We received the addresses of 3000 doctors taken at random from a list of 27000 Australian doctors that had been extracted from the Australian Medical Masterfile Database (Australasian Medical Publishing Company, Sydney). Each doctor on the list belonged to one of 27 medical disciplines where there would be the possibility of making a medical end-of-life decision (i.e., being the attending doctor at a non-acute death) (Box 2). The sample of doctors in the study by van der Maas et al.4 consisted of those who had attended deaths where the cause of death was highly likely to have involved making a medical end-of-life decision (as determined from death certificates). Australia does not have a national Death Certificate Register for identifying causes of death and attending doctors, as in the Netherlands. Therefore, the 27 medical disciplines (also extracted from the Australian Medical Masterfile Database) in our study were chosen to be as comparable as possible with the broader categories of doctors (cardiology, surgery, internal medicine, respiratory medicine [pulmonology], neurology, general practitioners and nursing home physicians) who were attendant to 87% of hospital deaths and nearly all deaths outside hospitals in the Netherlands.4 Between May 1996 and July 1996, we sent questionnaires to the random sample of 3000 doctors in the selected medical disciplines. The sampling method was chosen to reflect the wider distribution of doctors in Australia with regard to sex, age, medical discipline and State or Territory of residence. The response rate, after a second mailing to non-respondents, was 1918 questionnaires (64%). The initial questions on the questionnaire narrowed the field of respondents to include the 1361 doctors who had attended a death within the last 12 months. These doctors were asked to recall the most recent death they had attended during that period. The field was further narrowed by excluding doctors who, in respect of that death, had had no contact with the patient until after death or where the death had been sudden and totally unexpected. Of the remaining field of 1112 doctors who had attended a death where there was an opportunity to make a medical end-of-life decision (i.e., a non-acute death), 800 doctors made a decision intended either to shorten life or foreseen as probably or certainly shortening life, and 312 doctors did not make such a decision. Statistical analysis The percentages and confidence intervals regarding doctors were computed using standard formulas for the estimation of proportions under simple random sampling.10 The sampling scheme regarding deaths was equivalent to two-stage cluster sampling. Each selected doctor was asked for details regarding only the most recent death he or she had attended in the past 12 months. Thus, the percentages of deaths and corresponding confidence intervals were computed using the usual ratio-to-size estimator for such sampling.10,11 The virtue of the ratio-to-size estimator was that it could take into account differences in the rates of death that the different doctors in our sample attended in the normal course of their duties (each doctor was asked to indicate the number of deaths they would attend on average within a certain period). For example, oncologists and general practitioners in our sample were each asked about only one death, but the differences in the number of deaths they normally attend were taken into account by the ratio-to-size method when extra polated to percentages of Australian deaths. Surveyed doctors' responses were weighted in these statistical extrapolations in proportion to the number of deaths they normally attend. A consequence of the differential weightings involved in the ratio-to-size method, however, is that there is no simple way of deriving the proportion of deaths in Australia that were preceded by a medical end-of-life decision from our sample of doctors who reported on their decision with regard to the most recent death they had attended. Furthermore, the confidence intervals take into account possible differences in attitudes towards end-of-life decisions between doctors and the fact that each respondent was asked about only one death. Results Survey response rates From the 3000 doctors who were sent questionnaires, we received responses from 1918 doctors (64%). There was no significant difference in response rate for men versus women (chi-squared test, 3.03; 1 degree of freedom [df]), nor between the States and Territories (chi-squared test, 1.94; 7 df). The response rate did, however, differ significantly with age, ranging from 74.9% for the under-35 years age group to 50.5% for the 56-65 years age group (chi-squared test, 26.73; 4 df). The response rate also differed significantly between different categories of medical discipline: there was a higher response rate from doctors in Groups A and B (116 and 340 doctors, respectively), and a lower response rate from doctors in Group C (1265 doctors) (chi-squared test, 160.4; 2 df). One hundred and ninety-seven doctors either did not state their medical discipline or indicated that it did not fall into one of these groups. (Further data on respondents' medical discipline, age, sex and State or Territory of residence are available from the authors.) Medical end-of-life decisions: medical practitioner sample Box 3 shows the characteristics of the deaths reported by the 800 doctors in our study who made medical end-of-life decisions. There were no significant differences with regard to sex (chi-squared test, 0.94; 1 df), State or Territory of residence (chi-squared test, 1.07; 7 df) or medical discipline (chi-squared test, 4.06; 3 df) between the 800 doctors who made medical end-of-life decisions and the 312 who did not make such a decision. There was, however, a marginally significant difference in the age distribution (chi-squared test, 11.50; 4 df; P=0.02). There was a higher than expected number of doctors aged between 56-65 years who did not make such a decision. The medical end-of-life decisions reported by the 800 doctors were as follows: 26 doctors (3.2%) reported euthanasia; 51 doctors (6.4%) reported ending the patient's life without the patient's explicit request; 289 doctors (36.1%) reported making a decision not to treat, of which 55 doctors (19%) reported no intention to hasten death, and 234 doctors (81%) reported an explicit intention to hasten death; and 434 doctors (54.2%) reported alleviating the patient's pain with opioids in large doses, of which 335 doctors (77.2%) reported no intention to hasten death, and 99 doctors (22.8%) reported a partial intention to hasten death (Box 3). We are planning to publish further analysis of the data at a later date. Medical end-of-life decisions: proportions of Australian deaths Box 4 shows the proportion of Australian deaths (acute and non-acute) that involved a medical end-of-life decision. The results in Box 4 and the results discussed below are based on an extrapolation of the data from Box 3 by means of ratio-to-size estimation (see Methods). A medical decision either partly or explicitly intended to hasten death or not prolong life occurred in 36.5% (±3.5%) of all Australian deaths. In almost half (17.8% [±2.9%]), there had been no explicit request from the patient, nor had there been any discussion of ending the patient's life, and the doctor did not believe that the patient had expressed a wish for death to be hastened. Euthanasia and physician-assisted suicide An estimated 1.8% (±0.6%) of all Australian deaths were the result of euthanasia or physician-assisted suicide (of which an estimated 0.1% were cases of physician-assisted suicide). Ending of life without an explicit request from the patient An estimated 3.5% (±0.8%) of all Australian deaths involved termination of the patient's life without the patient's explicit request. There was some discussion with the patient -- although not an explicit request for death to be hastened -- in 38% of these cases. In virtually all of the remaining cases, the doctor did not consider the patient competent (i.e., capable of assessing his or her situation and making a decision about it). In 0.7% of all Australian deaths, attending doctors (who were asked to give as many reasons as applied) stated that they did not discuss with the patient the possible hastening of death by prescribing, supplying or administering drugs because the patient was unconscious, demented or mentally handicapped. In 0.5% of all Australian deaths, attending doctors felt that the act of hastening death was "clearly the best one for the patient", or that "discussion would have done more harm than good". Decisions not to treat An estimated 24.7% (±3.1%) of all Australian deaths involved a decision not to treat with the explicit intention of not prolonging life or of hastening death. Of these, less than one-tenth (or 2.2% of all Australian deaths) were in response to an explicit request from the patient, and in most cases (22.5% of all Australian deaths) there was no explicit request from the patient. In 3.2% of all deaths, the doctor believed that the patient had expressed a wish for death to be hastened, but there had been no discussion with the patient, and in 5% respondents did not indicate whether the patient was involved in the decision. Therefore, in 14.3% (±2.8%) of all Australian deaths, the death was preceded by a medical decision to withhold or withdraw treatment with the explicit intention of not prolonging life or of hastening death, despite the fact that the decision was not based on an explicit request from a patient, there was no discussion with the patient about whether to hasten death by the omission of treatment, and the doctor did not believe that the patient had expressed a wish for death to be hastened. In almost all of these cases (14.1% of all deaths) the doctor did not consider the patient capable of assessing his or her situation and making a decision about it. In 0.16% of all Australian deaths, the doctor did consider the patient capable of assessing his or her situation and making a decision about it, but intentionally hastened death by forgoing treatment, without discussing this with the patient. In 10.5% (±2.7%) of all Australian deaths, attending doctors (who were asked to give as many reasons as applied) stated that they did not discuss with the patient the possible hastening of death by withholding or withdrawing treatment because the patient was unconscious. Dementia or mental handicap was offered as the reason in 2.6% of all Australian deaths. In 1.6% of all Australian deaths, the attending doctor felt that the decision "was clearly the best one for the patient" or "discussion would have done more harm than good". Alleviation of pain and/or symptoms with opioids In 6.5% (±1.1%) of all Australian deaths, doctors prescribed opioids (morphine or a comparable drug) with a dual intention: in part, to alleviate pain and symptoms and, in part, to hasten death. Drug doses were large enough to have, in the judgement of the respondent and with regard to the particular death, a life-shortening effect. In 0.9% of all Australian deaths, doctors had partly intended to hasten death by alleviating pain and/or symptoms using opioids on the basis of an explicit request; and in 3.2% there was either some discussion with the patient, or the doctor believed that the patient had expressed a wish (but not an explicit request) for death to be hastened. In 2.4% of all deaths, doctors had partly intended to hasten death by this practice, despite the fact that there was no explicit request, no discussion with the patient, and the doctor did not believe that the patient had expressed a wish for death to be hastened. Amount of time by which life was shortened Medical decisions that were explicitly or partly intended to hasten death or not prolong life generally brought death forward by relatively short periods, according to doctors' estimates. Of the 24.7% (±3.1%) of all Australian deaths that involved a decision to forgo treatment with the explicit intention of hastening death or not prolonging life, the decision was estimated to hasten death by more than a week in 3.6% of all Australian deaths. Of the 5.3% of all Australian deaths that were caused by the prescription, supply or administration of drugs explicitly intended to hasten death, life was estimated to have been shortened by more than a week in 1.1% of all Australian deaths. Discussion One of the main findings of our study was that 30% (±3.3%) of all Australian deaths were preceded by a medical decision explicitly intended to hasten the patient's death: doctors prescribed, supplied or administered drugs with the explicit intention of ending the patient's life in 5.3% (±1%) of these deaths, and withdrew or withheld life-prolonging treatment with the explicit intention of not prolonging life or of hastening death in 24.7% (±3.1%) of these deaths. Our study shows that Australia had a significantly higher rate of intentional ending of life without the patient's consent, both through the administration of drugs and by withholding or withdrawing treatment, than the Netherlands. Overall, 36.5% (±3.5%) of all Australian deaths involved a medical decision either partly or explicitly intended to hasten death or not prolong life, compared with 19.5% of deaths in the Netherlands.4 Further comparisons with the data from the Netherlands are shown in Box 5. The difference in the rates of euthanasia between the two countries was not statistically significant. However, the rates of intentionally ending life without an explicit request from the patient were significantly higher in Australia than in the Netherlands (P < 0.0001). In 22.5% (±3.1%) of all Australian deaths, doctors withheld or withdrew treatment from patients, without the patient's explicit request, with the explicit intention of ending life. No comparable 1995 figure for the Netherlands was available (the 1991 figure was 5.3%), but the 1995 figure for all decisions to forgo treatment with an explicit intention of hastening death or not prolonging life in the Netherlands was 13.3%.4 If we classify euthanasia as all cases in which death is "intentionally accelerated by a doctor,"12 30% of all Australian deaths (or 37000 cases) would be cases of euthanasia. The comparable 1995 figure for the Netherlands was 16.6%. The response rate of 64% to our questionnaire was adequate to give a broad picture of medical end-of-life decisions in Australia, but all postal surveys are subject to some degree of non-response bias. The validity of our observations and comparisons is limited by the degree to which our study can be compared with the studies from the Netherlands. Although we strove to obtain a sample population of Australian doctors as comparable as possible to the sample in the study by van der Maas et al.,4 the 27 medical disciplines we selected may not have matched exactly the groups of doctors in their study. Furthermore, in our study there was a higher response rate from doctors whose medical discipline belonged to Groups A and B, and a lower response rate from those in Group C (Box 2). An analysis of the responses of doctors in each of these categories indicates that, in so far as this response bias had any effect on our results, this effect would have been to reduce the reported rate of euthanasia and ending of a patient's life without an explicit request. Cultural differences between Australia and the Netherlands could have accounted for some of the differing responses to the questionnaire. In the Netherlands, the community has a different relationship with primary care doctors compared with Australians, which may account for some of the differences in results. In the Netherlands, most primary care is provided by the huisarts ("home doctor"), the general practitioner or family doctor, either in the patient's home or in the doctor's home surgery. All Dutch patients have a personal huisart, who usually serves a practice of about 2300 people, lives in the area and makes frequent house calls when a patient is ill. This allows for close and enduring contact between the doctor and the patient, and distinguishes the delivery of primary care in the Netherlands from the increasingly institutionalised delivery of primary care in Australia. However, our study undermines suggestions that the rate at which doctors intentionally end patients' lives without an explicit request is higher in a country where euthanasia is practised openly (the Netherlands) than in a comparable country which has not allowed euthanasia to be practised openly, such as Australia. Although Australian law recognises a right to refuse treatment, it generally prohibits the intentional termination of life, whether by act or omission.13 Our findings, together with the previously cited studies of doctors in Victoria and New South Wales,6,7 suggest that Australian law has not prevented the practice of euthanasia or the intentional ending of life without the patient's consent. Our study shows that, while 30% (±3.3%) of all Australian deaths were preceded by an action or omission explicitly intended to end the patient's life, in only 4% was the decision taken in response to an explicit request from the patient. Far more doctors in our sample intentionally hastened death by forgoing (withholding or withdrawing) treatment than by prescribing, supplying or administering drugs with the explicit intention of ending life. For the purposes of this comparative study, we used the same narrow definition of euthanasia (the administration of drugs with the explicit intention of ending the patient's life, at the patient's request) that was used by van der Maas et al.4 Traditionally, "euthanasia" has been defined more broadly: "an action or an omission which of itself or by intention causes death, in order that all suffering may in this way be eliminated."14 This study suggests that the contemporary focus on euthanasia in the narrow sense may be too limited, at least if the aim is to prevent patients having their lives shortened by medical end-of-life decisions being made without their consent. Our study raises the question of why some Australian doctors choose intentionally to end the lives of some of their patients without the patients' consent, especially in situations where the patient is competent and could be consulted. While this issue remains the subject of further empirical research, it may be that, because existing laws prohibit the intentional termination of life, doctors are reluctant to discuss medical end-of-life decisions with their patients lest these decisions be construed as collaboration in euthanasia or in the intentional termination of life. Acknowledgement This study was supported by a grant from the National Health and Medical Research Council. Responsibility for the views expressed is that of the authors alone. References The Royal Dutch Medical Association issues revised guidelines on euthanasia. Int Digest Health Legislation 1996; 47: 401-405. van der Wal G, Dillmann RJM. Euthanasia in the Netherlands. BMJ 1994; 308: 1346-1349. van der Maas PJ, van Delden JJM, Pijnenborg L. Euthanasia and other medical decisions concerning the end of life. Lancet 1991; 338: 669-674. van der Maas PJ, van der Wal, Haverkate I, et al. Euthanasia, physician-assisted suicide, and other medical practices involvingthe end of life in the Netherlands, 1990-1995. N Engl J Med 1996; 335: 1699-1705. Waddell C, Clarnette RM, Smith M, et al. Treatment decision-making at the end of life: a survey of Australian doctors' attitudes towards patients' wishes and euthanasia. Med J Aust 1996; 165: 540-544. Kuhse H, Singer P. Doctors' practices and attitudes regarding voluntary euthanasia. Med J Aust 1988; 148: 623-627. Baume P, O'Malley E. Euthanasia: attitudes and practices of medical practitioners. Med J Aust 1994; 161: 137-144. Encyclopaedia Britannica 1996 Year Book. Chicago: Encyclopaedia Britannica, Inc. 1996: 554, 679. Australian Bureau of Statistics. Australian Demographic Statistics, June 1996. Canberra: AGPS, Cat. No. 3101.0. Cochran WG. Sampling techniques. 2nd edition. New York: Wiley, 1963: section 3.2, 11.6: 49-52, 300-303. Scheaffer RL, Mendenhall W, Ott L (editors). Elementary survey sampling. 4th edition. Boston: PWS-Kent, 1990, section 9.5: 294-296. Keown J. Euthanasia in the Netherlands: sliding down the slippery slope? In: Keown J, editor. Euthanasia examined: ethical, clinical and legal perspectives. Cambridge: Cambridge University Press, 1995: 261-296. Crimes Act 1900 (NSW), s. 19(1)(a). Sacred Congregation for the Doctrine of the Faith: Declaration on Euthanasia, Vatican City, 1980: 6. (Received 20 Dec 1996, accepted 22 Jan 1997) Authors' details Centre for Human Bioethics, Monash University, Clayton, VIC. Helga Kuhse, PhD, Director; Peter Singer, FAHA, FASSA, Deputy Director; Maurice Rickard, PhD, NHMRC Research Officer. School of Community Medicine, University of New South Wales, Sydney, NSW. Peter Baume, MD, FAFPHN, Professor. Department of Mathematics, Monash University, Clayton, VIC. Malcolm Clark, PhD, Senior Lecturer. No reprints will be available from the author. Correspondence: Associate Professor H Kuhse, Centre for Human Bioethics, Monash University, Clayton, VIC 3168. E-mail: helga.kuhse AT arts.monash.edu.au - - - To top of article - ©MJA 1997 <URL: http://www.mja.com.au/> © 1997 Medical Journal of Australia. 1: Definitions of medical end-of-life decisions4 Euthanasia:The administration of drugs with the explicit intention of ending the patient's life, at the patient's request. Physician-assisted suicide:The prescription or supplying of drugs with the explicit intention of enabling the patient to end his or her own life. Ending of life without the patient's explicit request:The administration of drugs with the explicit intention of ending the patient's life, without a concurrent, explicit request by the patient. Alleviation of pain and symptoms with opioids: The administration of doses large enough so that there was a probable life-shortening effect. A decision not to treat:The withholding or withdrawing of potentially life-prolonging treatment. 2: Medical disciplines in Australia that involve the possibility of doctors making medical end-of-life decisions* Group A: Emergency medicine, intensive care, neonatology, medical oncology, palliative care, renal medicine, cardiothoracic surgery. Group B: Cardiology, clinical haematology, geriatric medicine, general medicine/general physician, immunology, infectious diseases, radiation oncology, respiratory medicine, abdominal surgery, neurosurgery, vascular surgery, gynaecological oncology, otorhinolaryngology. Group C: Breast surgery, endocrinology, gastroenterology, general practice, general surgery, neurology, urological surgery * Participating doctors were asked to nominate the group containing the medical discipline in which they practise. Medical disciplines were extracted from the Australasian Medical Masterfile Database. 3: Characteristics of medical end-of-life decisions reported by 800 Australian doctors 4: Estimated proportion of medical end-of-life decisions in Australia 5: Comparison of medical end-of-life decisions in Australia and the Netherlands*

Helga Kuhse · Peter Singer · Peter Baume · Malcolm Clark · Maurice Rickard

Hepatitis B and C in New South Wales prisons: prevalence and risk factors Tony G Butler, Kate A Dolan, Mark J Ferson, Linda M McGuinness, Phillip R Brown and Peter W Robertson MJA 1997; 166: 127 For comment see Crofts Readers may print a single copy for personal use. No further reproduction or distribution of the articles should proceed without the permission of the publisher. For permission, contact the Australasian Medical Publishing Company 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/>". Abstract - Introduction - Methods - Laboratory methods - Statistical methods - Results - Participants - Serology and risk factors - Discussion - Acknowledgements - References - Authors' details - ©MJA1997 Abstract Objectives: To determine the prevalence of hepatitis B virus (HBV) and hepatitis C virus (HCV) infection among inmates entering the New South Wales correctional system and to examine risk factors for infection. Design: Cross-sectional survey. Setting: Reception Centre at Long Bay Correctional Centre, Sydney, New South Wales, June to December 1994. Participants: 408 adult male inmates received at the reception centre (28% of the 1450 new inmates eligible for compulsory HIV testing). Outcome measures: Presence of HBV core and surface antibody and surface antigen; HCV antibody; risk factors; inmates' knowledge about risk factors. Results: 37% of inmates tested positive for HCV antibody, 31% for HBV core antibody and 3.2% for HBV surface antigen (indicating recent infection or carrier status). Among those who reported a history of injecting illegal drugs, rates rose to 66% for HCV antibody and 43% for HBV core antibody. Prevalence of HBV and HCV antibodies was similar in Aboriginal and non-Aboriginal inmates, but HBV antigen carrier rate was significantly higher among Aboriginals (12% versus 2.2%). Knowledge about hepatitis risk factors was poor (only 20% named injecting drug use), although recidivists were significantly better informed than those new to the correctional system. Multivariate analysis identified injecting drug use, past exposure to hepatitis B virus and previous imprisonments as significant predictors for HCV infection, and age over 25 years and HCV antibodies for HBV infection. Conclusions: Results suggest that about a third of adult male prisoners entering the NSW correctional system may have been infected with HBV or HCV. Measures such as education about hepatitis risk factors and HBV vaccination are needed to reduce hepatitis transmission in this population. MJA 1997; 166: 127-130 Introduction The reported high prevalence of hepatitis B and C in prison populations is attributed to the disproportionate number of people in prisons who engage in risk behaviours, particularly injecting illegal drugs.1-8 It is estimated that up to 60% of inmates are committed for drug-related offences.9 Further, an Australian study estimated that during their incarceration 25%-44% of inmates occasionally injected illegal drugs, 14%-34% engaged in occasional anal intercourse and 5%-18% did both.10 Among inmates entering the Victorian prison system in 1991 and 1992, 39% were positive for hepatitis C antibody (including 64% of those reporting a history of injecting illegal drugs), 33% were positive for hepatitis B antibody and 2.5% for hepatitis B surface antigen.11 There are no recent studies of the prevalence of infection with hepatitis B and C viruses (HBV and HCV) in New South Wales prisons. Our study aimed to determine the prevalence of HBV and HCV infection among inmates entering the NSW correctional system and to examine the risk factors associated with these infections. Methods The study was performed at the Reception Centre at Long Bay Correctional Centre, Sydney, NSW, from June to December 1994. The Reception Centre receives about 51% of adult males entering the NSW correctional system, most of whom are then transferred to other prisons. At the time of our survey, all inmates entering NSW prisons were routinely screened for the human immunodeficiency virus (HIV) by public health nurses of the Corrections Health Service. Inmates were also invited to participate in the hepatitis survey. Recruitment depended on the nurses' availability to enrol inmates during HIV screening, and in busy periods it was not always possible to enrol all new inmates. No information was sought from those who did not agree to participate. Inmates who agreed to participate were briefed on the project and informed that a consent form had to be signed, a blood specimen was needed, and a risk factor questionnaire would be administered by the nurse. Nurses also asked inmates to describe ways in which hepatitis can be transmitted. All inmates who tested positive for HBV and HCV antibodies were counselled by public health nurses. HCV-positive inmates routinely receive follow-up liver function tests. Inmates who reported having had HBV vaccination were given booster vaccination if antibody tests indicated they were not immune. All other inmates who tested negative for HBV core antibody were offered hepatitis B vaccination. Ethics approval for the study was granted by the Eastern Sydney Area Health Service Ethics Committee and the NSW Department of Corrective Services. Laboratory methods HBV core antibody (indicating past exposure) was tested with an anti-HBc enzyme immunoassay kit (General Biologicals, Taiwan). Samples positive for HBV core antibody were tested for HBV surface antigen (indicating carrier status) with the HBsAg-enzyme immunoassay (Murex, Dartford); positive results were confirmed by HBs reverse passive haemagglutination assay (Serodia, Tokyo). Inmates who reported having been vaccinated against hepatitis B were tested for HBV surface antibody with the HBsAb-enzyme immunoassay (General Biologicals, Taiwan). A level of 30 IU/mL was considered the minimum necessary for immunity. HCV antibody was detected with the Innotest HCVAbIII assay (Innogenetics, Belgium). Reactive samples were retested in duplicate and, if again reactive, were tested with anti-HCV (Murex, Dartford). Samples that were reactive in each of the two types of assay were classified as positive, and those with discrepant results as equivocal. Statistical methods Relative risks were calculated with the statistical software Epi Info-6,12 and logistic regression was performed with the software SPSS-6. 13 The c 2 test was used to test for association between Aboriginality and serostatus. Results Participants About 1450 inmates were tested for HIV at the Long Bay Reception Centre between June and December 1994. Of these, 410 adult males (28%) consented to be screened for HBV and HCV antibodies. It is not known how many of the remainder were not invited to participate and how many refused. Two inmates were counted twice as they were released and reincarcerated during the study period. Results of their second tests were excluded from the analysis. The 408 subjects were aged 17-73 years (mean, 30.6 years; standard deviation, 10.1), and 296 (73%) had been imprisoned previously. Country of origin was: Australia, 80%; Europe, 8.6%; New Zealand and the Pacific Islands, 3.5%; Asia, 2.7%; the Middle East, 2.7%; and elsewhere or unknown, 2.5%. Forty-one inmates (10%) identified themselves as Aboriginal. Serology and risk factors Prevalence of HBV and HCV antibodies and HBV antigens is shown in Box 1 (below). About a third of inmates were positive for HBV core antibody (31%) or for HCV antibody (37%), with 21% positive for both. Among those who reported having received HBV vaccination, only 34% had HBV surface antibody levels (indicating immunity). No significant differences were found between Aboriginal and non-Aboriginal inmates in prevalence of HBV core and HCV antibodies. However, the HBV carrier rate was significantly higher among Aboriginals (12% versus 2.2%; c2 =11.8; P < 0.001). Among those reporting hepatitis B vaccination, more Aboriginals (50%) than non-Aboriginals (32%) had developed immunity, but the difference was not significant. The association between risk behaviours and HBV and HCV infection is shown in Box 2. The risk of each infection was significantly increased by a history of injecting drug use, previous imprisonment, sharing of injecting equipment, age over 25 years, injecting drug use during previous imprisonment, and tattooing (P< 0.05 or less). Risk also increased significantly with increasing numbers of previous imprisonments (P< 0.01 or less). Risk of HCV infection was also significantly increased by tattooing, sex with an injecting drug user and presence of HBV antibodies (P< 0.001). Among the 67 inmates who reported injecting drug use during a previous imprisonment, 77% were positive for HCV antibody and 56% for HBV antibody. In addition, among the 150 inmates positive for HCV antibody, only 14 (9%) did not report a history of injecting drug use. The following independent variables were entered into a logistic regression model, with HBV and HCV infection as separate outcome variables: injecting drug use, sex with an injecting drug user, tattoos, Aboriginality, age group (over 25 years versus 25 years and under), previous imprisonment, and HBV or HCV infection. For HCV infection, significant predictors were injecting drug use (odds ratio [OR], 19.9; P< 0.001), presence of HBV antibody (OR, 5.6; P< 0.001), and previous imprisonment (OR, 3.9; P< 0.001). For HBV infection, significant predictors were presence of HCV antibody (OR, 6.2; P< 0.001), and age over 25 years (OR, 3.4; P< 0.001). Among the 85 inmates positive for both HBV and HCV antibodies, 83 (98%) reported previous imprisonment, 78 (92%) reported injecting drug use, 40 (47%) reported sharing injecting equipment, 64 (75%) had tattoos, and 31 (36%) reported being tattooed in prison. Inmates' knowledge of risk factors for hepatitis B and C transmission is shown in Box 3 (below). Few inmates were knowledgeable about risk factors, with injecting drug use nominated by only 20% and tattooing by only 2%. However, those who had been imprisoned previously were significantly more likely to identify injecting drug use as a risk factor than those new to the correctional system, and significantly less likely to answer "no idea" about risk factors. Discussion Our results agree with those of a 1991-1992 study of Victorian prison inmates.11 We found that 37% of inmates were positive for HCV antibodies (39% in Victoria), 31% for HBV antibodies (33% in Victoria), and 3.2% for HBV surface antigen B (2.5% in Victoria). Our results also agree with those of two Victorian studies, which found rates of 62% and 68%, respectively, for HCV infection among injecting drug users.8,14 Extrapolating our results to the current NSW male prison population of over 6000 implies that almost 2000 inmates are likely to have been exposed to HBV, about 200 are HBV carriers, and over 2000 are HCV-antibody-positive. In contrast, there were 25 HIV-positive inmates in NSW correctional centres at the time of the study (0.4%; 23 male and 2 female) (unpublished data). NSW prison inmates are offered hepatitis B vaccination if their sentences exceed six months and they are considered "at risk". However, inmates with shorter sentences may also be at risk. We believe that all inmates have the right to be protected from possible infection and that all should start a course of hepatitis B vaccination on entry to the correctional system. It may be appropriate to use recently described accelerated vaccination schedules, which provide protective levels of anti-hepatitis B surface antibody relatively quickly.15 In addition, if hepatitis B vaccination courses are not completed in prison, inmates should be educated about the need to complete them after release. We found that among the 108 inmates (26%) who reported having had hepatitis B vaccination only a third were immune, possibly because of failure to complete the recommended vaccination schedule. Vaccination history could not be verified. We found that previous imprisonment was a significant risk factor for HCV infection, suggesting that measures to minimise the spread of hepatitis within prisons are essential. In addition, as many injecting drug users spend time in prison, it is an appropriate point for intervention to break the cycle of infection by educating them about risks for hepatitis transmission and providing vaccination. Prison education programs seem to have improved awareness about transmission of hepatitis, as inmates with a prison history were more likely to know the role of injecting drug use and less likely to have "no idea" about hepatitis transmission. However, knowledge was still poor, and only 2% of inmates identified tattooing as a risk factor for hepatitis transmission, which is of concern given the popularity of tattooing in this population. The link between tattooing and HCV infection has been identified elsewhere,16,17 and should receive more emphasis in hepatitis education programs. The recent decision by the NSW Department of Corrective Services to make condoms available in prisons may reduce hepatitis transmission. Other measures, such as removing obstacles for accessing bleach, and needle exchange, should also be considered in the fight to curb the spread of hepatitis. Our findings suggest that chronic hepatitis may become one of the prison system's major health concerns over the next two decades. Acknowledgements The study was funded in part by a grant from the NSW Health Department AIDS/Infectious Diseases Branch. We wish to thank the nursing staff at the NSW Corrections Health Service for assistance in data collection (Amanda Christensen, Sheryl Frewin, Marion Grey, Cherylyn Jesson, Linda Kemp, Jodie Lee, and Sandra Parsons). References Acedo A, Campos A, Bauza J, et al. HIV Infection, hepatitis, and syphilis in Spanish prison [letter]. Lancet 1989; 2: 226. Hull HF, Lyons LH, Mann JM, et al. Incidence of hepatitis B in the penitentiary of New Mexico. Am J Public Health 1985; 75: 1213-1214. Koplan JP, Walker JA, Bryan JA. Prevalence of hepatitis B surface antigen and antibody at a state prison in Kansas. J Infect Dis 1978; 137: 505-506. Melico-Silvestre A, Pombo V, Pereira A, et al. Seroepidemiological survey of transmissible infections in Portuguese prisoners [letter]. AIDS 1991; 5: 780-781. Decker M, Vaughn W, Brodie J, et al. Seroepidemiology of hepatitis B in Tennessee prisoners. J Infect Dis 1984; 150: 450-458. Chiaramonte M, Trivello R, Renzulli G, et al. Hepatitis B virus infection in prisons. J Hyg Camb 1982; 89: 53-58. Vlahov D, Nelson K, Quinn T, Kendig N. Prevalence and incidence of hepatitis C virus infection among male prison inmates in Maryland. Eur J Epidemiol 1993; 9: 566-569. Fairley CK, Leslie DE, Nicholson S, et al. Epidemiology and hepatitis C virus in Victoria. Med J Aust 1990; 153: 271-273. Corrections Health Service. Strategic Plan 1993-1998. Sydney: NSW Health Department, 1994. Douglas RM, Gaughwin MD, Ali RL, et al. Risk of transmission of the human immunodeficiency virus in the prison setting [letter]. Med J Aust 1989; 150: 722. Crofts N, Stewart T, Hearne P, et al. Spread of blood borne viruses among Australian prison entrants. BMJ 1995; 310: 285-288. Epi Info [computer program], version 6.0. Atlanta, Ga: Centers for Disease Control, 1994. SPSS-6: Statistical package for the social sciences [computer program], version 6. Chicago, Ill: SPSS Inc, 1994. Crofts N, Hopper J, Bowden S, et al. Hepatitis C virus infection among a cohort of Victorian injecting drug users. Med J Aust 1993; 159: 237-241. Bayas J, Bruguera M, Martin V, et al. Hepatitis B vaccination in prisons: the Catalonian experience. Vaccine 1993; 11: 1441-1444. Kaldor J, Archer, G, Buring M, et al. Risk factors for hepatitis C virus infection in blood donors: a case-control study. Med J Aust 1992; 157: 227-230. Holsen DS. Prevalence of antibodies to hepatitis C virus and association with intravenous drug abuse and tattooing in a national prison in Norway. Eur J Clin Microbiol Infect Dis 1993; 12: 673-676. (Received 10 Jan, accepted 21 Oct, 1996) Authors' details New South Wales Health Department, AIDS/Infectious Diseases Branch, Sydney, NSW. Tony G Butler, MSc, Public Health Officer. National Drug and Alcohol Research Centre, University of New South Wales, Sydney, NSW. Kate A Dolan, BSc, Research Officer. South Eastern Sydney Area Health Service, Sydney, NSW. Mark J Ferson, FRACP, FAFPHM, Director of Public Health. NSW Corrections Health Service, Long Bay Correctional Centre, Sydney, NSW. Linda M McGuinness, Grad Dip HSc, RN, Assistant Director of Nursing; Phillip R Brown, MBA, FAFPHM, Chief Executive Officer. Microbiology Department, Prince of Wales Hospital, Sydney, NSW. Peter W Robertson, PhD, Serologist. No reprints will be available. Correspondence: Mr T G Butler, NSW Health Department, AIDS/Infectious Diseases Branch, Locked Bag 961, North Sydney, NSW 2059. - To top of article - ©MJA 1997 <URL: http://www.mja.com.au/> © 1997 Medical Journal of Australia.

Tony G Butler · Kate A Dolan · Mark J Ferson · Linda M McGuinness · Phillip R Brown · Peter W Robertson

Readers may print a single copy for personal use. No further reproduction or distribution of the articles should proceed without the permission of the publisher. For permission, contact the Australasian Medical Publishing Company 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". Abstract - Introduction - Methods - Results - Discussion - Acknowledgements - References - Author's details Abstract Objective: To ascertain the prevalence of childhood sexual abuse (CSA) in a community sample of Australian women. Design: Retrospective study, done in 1994, of cross-sectional data on the prevalence of CSA, collected as part of a larger two-stage case-control study of the possible relationship between CSA and alcohol abuse. Data were appropriately weighted to adjust for the different selection probabilities of cases and controls. Participants: 710 Women randomly selected from Australian federal electoral rolls. Results: One hundred and forty-four women (20%) had experienced CSA. In 14 of these 144 women (10%), the abuse involved either vaginal or anal intercourse (i.e., 2% of the sample population experienced such abuse). The mean age at first episode of CSA was 10 years, and most (71%) of the women were aged under 12 years at the time. Perpetrators of the abuse were usually male (98%) and usually known to the child; 41% were relatives. The mean age of abusers was 34 years, with a median age difference of 24 years from that of the abused individual. Only 10% of CSA experiences were ever reported to the police, a doctor or a helping agency (e.g., community organisations, such as sexual assault services). Conclusion: The high rates of CSA (estimated to be 20% of all women) and low rates of reporting (10%) indicate the need for general practitioners and other health professionals to be aware that a history of such abuse may be common in women in the general population. MJA 1997; 166: 65-68 Introduction Community-based studies in North America, the United Kingdom and New Zealand have shown that, for women, the prevalence of at least one incident of sexual contact with an adult before age 18 years ranges from 6% to 62%. 1-8 In an Australian study examining the extent of childhood sexual abuse (CSA) among social science students, 28% of the female students and 9% of the male students reported at least one such incident before the age of 16 years. 9 Official figures on the incidence of child abuse (including physical and sexual abuse and neglect) show that, in 1992-93, 59 122 cases were reported and investigated throughout Australia. Of these cases, 27 196 (46%) were officially substantiated, and the number of child sexual abuse cases was 5979 (22%). 10 Because of the secrecy and shame surrounding sexual abuse, the criminal sanctions against it, and the young age and dependent status of the child, it is likely that official statistics seriously underestimate its true extent. 11 To determine the true extent of CSA, prevalence studies that rely on retrospective reports of adult subjects are needed; this study presents epidemiological data on the prevalence of CSA in a community sample of Australian women. Methods Participants recruited The current study was part of a larger two-stage case-control investigation (conducted in 1994) into the relationship between CSA and alcohol abuse in women. Stage 1 of the study involved a postal questionnaire sent to 6000 women randomly selected from Australian federal electoral rolls. Approval for the study was provided by the ethics committee of the Australian National University. A total of 3958 (66%) valid responses were received. Women who had an alcohol problem were identified as cases from their answers to this questionnaire. Stage 2 of the survey involved sending all women identified as having an alcohol problem ( n= 184) and a random selection of controls (i.e., women identified as not having an alcohol problem [ n= 910]) a second questionnaire, which explored childhood sexual experiences. This paper reports the results of analysis of the responses to this second questionnaire. Questionnaire The questions used to ascertain the prevalence of CSA were based on those developed by Wyatt 2 . In addition to the questions on sexual experiences shown, questions were also asked about other areas, including family background and physical abuse. Women who said they had had sexual experiences before age 16 were asked about the sexual experience, including information on the type of abuse; age at the time of abuse; relationship to the abuser; age of the abuser; frequency and period of the abuse (the period being obtained by subtracting the age at first episode from age when the abuse stopped); whether coercion was used; whether the abuse was disclosed to anyone; and the self-perceived short- and long-term effects of the abuse. The effects of the abuse are dealt with in Fleming et al. (unpublished data) and Fleming et al. 12 If the women had told anyone about the abuse, they were asked whom and when they had told. Definition of childhood sexual abuse CSA was defined as all experiences of sexual contact occurring before the age of 12 with a person five or more years older, irrespective of consent , and all experiences of sexual contact occurring between age 12 and 16 years with a person five or more years older that were not wanted or were distressing. Sexual contact was defined as: touching or fondling the child's body; attempts to have the child arouse the adult, or touch his/her body in a sexual way; the adult rubbing his/her genitals against the child's body in a sexual way; touching the child's genitals with the mouth, or having the child touch the adult's genitals with the mouth; attempts to have anal or vaginal intercourse with the child; and completed anal or vaginal intercourse. For each episode, there was a progressive range of possible experiences, such as touching of the genitals or breasts to intercourse. Sample and statistical analysis A response rate of 65% ( n = 710) was obtained, with the final sample comprising 124 of 184 women identified as cases (67%) and 586 of 910 women (64%) as controls. To estimate the underlying prevalence of CSA in the population, the data were weighted to account for the different probabilities of selection for cases and controls. The controls received higher weighting than the cases (who were more likely to be selected in the sample because of the original design). The HLOGIT and HREG procedures in STATA Release 4 for Windows 13 were used to perform logistic and ordinary regressions on the weighted data and to obtain Huber standard errors, which were then used to obtain Wald statistics and corresponding P values. The significance testing thus allowed for individuals in the sample having different probabilities of selection. Results Respondents to the survey showed a demographic distribution similar to that recorded in the Australian Bureau of Statistics (ABS) census data. 14 Childhood sexual experiences Two hundred and ninety four of the 710 women (41%) responded that they had had at least one of the sexual experiences at least once before the age of 16 years. For 45 (6%) of these women, the experiences were categorised as consensual sexual experiences with peers (a peer being defined as a person less than five years older than the subject). Accordingly, 249 (35%) women reported some sexual abuse or experience that was unwanted or distressing during childhood. The sexual experiences were divided into four categories: non-contact (exposure and masturbation) (n = 87, 12%), consensual with peers (n = 45, 6%), unwanted with peers (n = 18, 3%), and unwanted sexual experiences with an adult that involved at least genital contact (which is defined as CSA and is the group on which the analyses are based) (n = 144, 20.3%; 95% CI = 17.3%-23.5%). A total of 13 women (2%) did not answer any of the questions on childhood sexual experiences. There was no significant difference (t = 1.63; P = 0.11) in mean current age between women who stated that they hadexperienced CSA (38.6 years; 95% CI, 36.7-40.4; range, 18 to 74 years) and women who had had no experience of CSA (40.3 years; 95% CI, 39.2-41.5; range, 17 to 79 years). To gain a measure of severity of abuse in the women (n =144) with CSA, sexual experiences were classified according to the most invasive act perpetrated. As shown in Box 2, 10% of these 144 women stated that intercourse had occurred and a further 17% stated that the abuser had attempted intercourse. From the sample group of 710 women, this extrapolates to 2% of the total Australian population of women having experienced CSA involving intercourse. Frequency and period of CSA The Figure shows the number of abuse episodes and Box 3 shows the frequency in women abused more than once. The period of abuse was less than one year for 57% of episodes, less than two years for 14% of episodes, and more than two years for 29% of episodes. Characteristics of individuals who experienced CSA CSA was rarely stated to have occurred before age five years. Most abuse (102/144 [71%]) occurred up to age 12 years, with the mean age at first abuse being 10 years (SD, 3.5 years). No significant differences were found for age at first abuse and relationship to the abuser. The rate of CSA reported in this study was two to four times higher for women who worked part-time (OR, 2.15; 95% CI, 1.24-3.71) or were students (OR, 4.44; 95% CI, 1.56-12.44) than for women who were either unemployed or at home. Women who were separated or divorced were more likely to have experienced CSA than women who were not separated or divorced (OR, 2.27; 95% CI, 1.12-4.64). There was evidence for a quadratic relationship with age (chi-squared = 4.1, df=1, P = 0.04), with prevalence increasing to approximately age 38 and then decreasing. There was no significant relationship for country of birth (Australia versus other), highest qualification, whether or not they had children, or number of children. Characteristics of the abuser Men were by far the most frequent abusers -- women accounted for only two reports. The mean age of abusers was 33.5 years (95% CI, 31.0-36.0; range, 11-70 years). Almost two-thirds of the abusers (83/144 [58%]) were more than 20 years older than the abused individual at the time of the abuse. The median age difference between the abused and the abuser was 24 years (range, 5-64 years). Most respondents (120/144 [83%]) stated their relationship to the abuser. Only 8% (11/143) of abusers were strangers. ( Note: Discrepancies in total group numbers arise from missing data.) Forty-one per cent of abusers (58/143) were family members (biological relatives, such as fathers, grandparents, uncles, siblings, and cousins, as well as stepfathers and adoptive fathers). Although biological fathers were more often stated to be abusers than step fathers, more women grew up with a biological father. Children who grew up without any father were the most likely to have experienced CSA (OR, 3.8; 95% CI, 1.1-13.8), followed by children with stepfathers (OR, 1.9; 95% CI, 1.0-3.6) and children who grew up with their biological father (reference group). The frequency of CSA was significantly different in women abused by relatives and those abused by non-relatives. For the 55 women who were abused once, 37 (70%) were abused by someone outside the family. When the abuser was a relative, the abuse was significantly more likely to have occurred regularly (daily or weekly, 14/25 [56%], and fortnightly, monthly or annually, 27/40 [68%]) (chi-squared = 13.7, df = 2, P = 0.001). Those abused by a relative were significantly more likely to have been abused more often (two to five times, 15/27 [56%]; six to 10 times, 6/12 [50%]; more than 10 times, 21/24 [88%]) than those abused by non-relatives (chi-squared = 16.6, df = 3, P = 0.001). Coercion Most women (102/143 [72%]) stated that some form of coercion was used; most commonly, they were frightened into compliance. The data were re-coded to provide a measure of the severity of force used: 92/143 (64%) said there had been verbal threats and threats of violence; 10/143 (7%) said there had been actual violence; and 41/143 (29%) said no form of coercion was used. Disclosure More than half (75/144 [52%]) of the women had disclosed the abuse. A further five women had tried to disclose the abuse, but their attempts to do so were unsuccessful. Of these 80 women, almost one-third (23 [28%]) disclosed, or tried to disclose, at the time of the CSA, seven (9%) within the first year, and 14 (18%) between one and 10 years later; almost half (36 [45%]) did not disclose until at least 10 years after the first abuse episode. There were no significant differences between age at first abuse experience and whether or not the women disclosed the abuse (chi-squared = 1.0, df = 1, P = 0.3). However, there were significant differences in the timing of disclosure (within one year, one to 10 years, more than 10 years) by age at time of abuse. Girls aged under 12 years at the time of the abuse were less likely to tell someone within a year of the abuse than were girls aged over 12 years (chi-squared = 7.8, df = 2, P = 0.02). Mothers were the most frequently told (39/80 [49%]), followed by friends (25/80 [32%]), and siblings (23/80 [29%]). Fewer than 10% of women (7/80) reported or had the abuse reported to either the police, a doctor or a helping agency (e.g., community organisations, such as sexual abuse services). Twenty-three per cent (17/75) stated that disclosure had stopped the abuse, compared with 22% (16/75) who said it had not. ( Note: More than one answer was possible for the questions about who was told and the reasons for prevention of disclosure.) When the women were asked what prevented disclosure, by far the most common reason given was embarrassment or shame (47/80 [46%]), followed by the belief that the other person would not be able to help them (23/80 [23%]), or would somehow blame or punish them for the abuse (19/80 [18%]). No significant differences in disclosure rates were found by relationship to abuser. Rates of disclosing abuse, examined by current age of the women, showed a significant decrease with age (chi-squared = 6.9, df = 1, P = 0.008), with 10/12 (83%) of young women (aged 17-24 years) having disclosed the abuse, compared with 26/44 (59%) of women aged 25-35 years, 23/45 (51%) aged 35-44 and 16/42 (38%) aged 45 or more. The timing of disclosure after abuse (within compared with more than one year) was not significantly different between the current 10-year age groups (chi-squared = 0.9, df = 4, P = 0.9). Discussion Although Australians have become increasingly aware of the existence and frequency of CSA, its prevalence in the community is not known. This study, the first national survey of Australian women on the prevalence of CSA, demonstrates a high rate of CSA. A rate of 20% for sexual abuse by an adult involving at least genital contact closely parallels the levels in a New Zealand study that used a comparable method of ascertainment. 7 The figure of 35% for any sexual abuse or unwanted sexual experience during childhood found in the present study is comparable to other reports which suggest that more than 50% of women had experienced sexual abuse or an unwanted sexual experience during childhood, but nearer to 10% to 15% had experienced CSA involving genital contact or intercourse. 2,4,7,8,15,16 A consistent finding from studies of CSA is that most of the reported abuse occurs in prepubescent girls. In this study, the mean age at first abuse experience was 10 years. This is within the range of ages of greatest risk, 8-12 years, reported by other studies. 3,7,15,17,18 It is important to note that, although few of the respondents stated that they had experienced sexual abuse before the age of five, it may be that such victims cannot remember the abuse. Australian Institute of Health and Welfare figures for sexual abuse by age group show that the rate of sexual abuse per 1000 children is similar in the two- to five- and six- to 10-years age groups (2.7 and 2.6 per 1000 children). 10 This suggests that abuse in this survey may have been under-reported because abuse that occurred before the age of five was not remembered. Another finding in this study is that only 10% of abuse experiences were ever reported to authorities. This supports claims that official figures may seriously underestimate the extent of the problem. Relatively high rates of abuse and low rates of reporting suggest that general practitioners, teachers, police and health and welfare personnel need to be aware that sexual abuse may be behind a range of interpersonal, behavioural and psychological problems in children. 19 Children will rarely disclose the abuse, but may have other symptoms which should arouse suspicion. This study indicates that young girls aged eight to 10 years are most at risk. Because of the usual closeness of the perpetrator to the victim and the victim's fear of being blamed or not being believed, our community as a whole needs to be mindful of the possibility of childhood sexual abuse. Acknowledgements This paper is based on data gathered for a PhD dissertation. The results presented represent stage 2 of the project. I am especially grateful to all the women who responded to this survey. I would like to thank my supervisors, Dr Gabriele Bammer, Dr Bev Sibthorpe and Dr Aileen Plant, for their advice and assistance. I would also like to thank Professor Paul Mullen for his editorial comments and suggestions and for his ongoing advice in the area of CSA, and Ms Robyn Attewell for her statistical advice. Financial support for this project was provided by the National Campaign Against Drug Abuse and by the National Centre for Epidemiology and Population Health, Australian National University. References Russell D. The incidence and prevalence of intrafamilial and extrafamilial sexual abuse of female children. Child Abuse Negl 1983; 7: 133-146. Wyatt GE. The sexual abuse of Afro-American and White American women in childhood. Child Abuse Negl 1985; 9: 507-519. Baker AW, Duncan SP. Child sexual abuse: a study of prevalence in Great Britain. Child Abuse Negl 1985; 9: 457-467. Bagley C, Ramsay R. Sexual abuse in childhood: psychosocial outcomes and implications for social work practice. J Soc Work Hum Sexuality 1986; 4: 33-46. Finkelhor D, Hotaling G, Lewis IA, Smith C. Sexual abuse in a national survey of adult men and women: prevalence, characteristics and risk factors. Child Abuse Negl 1990; 14: 19-28. Leventhal JM. Epidemiology of child sexual abuse. In: Oates RK, editor. Understanding and managing child sexual abuse. Sydney: Harcourt Brace Jovanovich Group, 1990: 18-42. Anderson JC, Martin JL, Mullen PE, et al. The prevalence of childhood sexual abuse. Experiences in a community sample of women. J Am Acad Child Adolesc Psychiatry 1993; 32: 911-919. Mullen PE, Romans-Clarkson SE, Walton VA, Herbison GP. Impact of sexual and physical abuse on women's mental health. Lancet 1988; I: 841-845. Goldman RJ, Goldman JDG. The prevalence and nature of child sexual abuse in Australia. Aust J Sex Marriage Family 1988; 9: 94-106. Angus G, Wilkinson K, Zabar P. Child abuse and neglect. Australia 1992-1993. In: Australian Institute of Health and Welfare, Child Welfare Series, No. 9; 1995 Canberra: AGPS. Finkelhor D. The international epidemiology of child sexual abuse. Child Abuse Negl 1994; 18: 409-417. Fleming J, Mullen P, Bammer G. A study of potential risk-factors for sexual abuse in children. Child Abuse Negl 1997; 21: 49-58. Stata C. Stata Statistical Software: Release 4.0. Stata Corporation, 1995 (Station C, ed. TX). Australian Bureau of Statistics. 1991 Census of population and housing -- Australia profile. 1993; Canberra: AGPS. Seigel JM, Sorenson SB, Golding JM, et al. The prevalence of childhood sexual assault. The Los Angeles Epidemiologic Catchment Area Project. Am J Epidemiol 1987; 126: 1141-1153. Russell DEH, Schurman RA, Trocki K. The long-term effects of incestuous abuse. A comparison of Afro-American and White American victims. In: Wyatt GE, Powell GJ, editors. Lasting effects of child sexual assault. Newbury Park: Sage Publications, 1988: 119-134. Sedney MA, Brooks B. Factors associated with a history of childhood sexual experience in a nonclinical female population. J Am Acad Child Adolesc Psychiatry 1984; 23: 215-218. Briere J, Runtz M. Symptomatology associated with childhood sexual victimization in a nonclinical adult sample. Child Abuse Negl 1988; 12: 51-59. Mullen PE, Martin JL, Anderson JC, et al. The long-term impact of the physical, emotional and sexual abuse of children. A community study. Child Abuse Negl 1996; 20: 7-21. (Received 15 Jul, accepted 20 Sep 1996) Author's details National Centre for Epidemiology and Population Health, Australian National University, Canberra, ACT. Jillian M Fleming, BSc, PhD. Current position: Coordinator, ACT Alcohol and Drug Service, Canberra, ACT. No reprints will be available. Correspondence: Dr J M Fleming, ACT Community Care, GPO Box 825, Canberra, ACT 2601. E-mail: jmf868 AT nceph.anu.edu.au ©MJA 1997 <URL: http://www.mja.com.au/> © 1996 Medical Journal of Australia.

Jillian M Fleming

Treatment decision-making at the end of life: a survey of Australian doctors' attitudes towards patients' wishes and euthanasia Charles Waddell, Roger M Clarnette, Michael Smith, Lynn Oldham and Allan Kellehear MJA 1996; 165: 540 For editorial comment, see Hassan Readers may print a single copy for personal use. No further reproduction or distribution of the articles in whole or in part should proceed without the permission of the publisher. For copyright permission, contact the Australasian Medical Publishing Company 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/>". Abstract - Introduction - Methods - Survey participants - Questionnaire - Statistical analysis - Results - Case scenarios 1 and 2 - Case scenario 3 - Case scenario 4 - Responses to request for assisted death - Factors influencing choice of treatment - Factors influencing doctors' responses to patient's request for assisted death - Discussion - Acknowledgements - References - Authors' details - - ©MJA1996 Abstract Objective: To examine factors that influence medical practitioners' treatment decisions for patients with life-threatening or terminal illnesses. Design: Postal survey, conducted between September and November 1995, of a self-administered questionnaire, describing four clinical case scenarios, to a random sample of 2172 Australian doctors in all States and Territories. Respondents were asked to prescribe treatment for the patients described in the scenarios. Patients' characteristics varied in terms of mental competence, illness severity, prognosis, the presence of advance directives, request for assisted death, and sociodemographic factors. The respondents' sociodemographic and medical training characteristics were also obtained. Setting: Random national sample of all active medical practitioners. Participants: Hospital trainees, general practitioners, physicians, palliative care practitioners and surgeons were surveyed. A response rate of 73% was achieved. Main outcome measures: Frequency of prescription of supportive, acute or intensive treatment for patients in the four clinical scenarios based on respondents' sex, religion, medical training and country of medical degree. Results: Three main findings were: (i) doctors did not make consistent decisions, but their decisions varied systematically by sociodemographic and medical training factors; (ii) doctors generally adhered to patient and family wishes when these were known; (iii) doctors did not generally adhere to a patient's request for assisted death. Conclusion: Treatment provided is significantly determined by the individual characteristics of the doctor and not solely by the nature of the medical problem. Participation in the informed-consent process and in the preparation of advance health care directives would enable practitioners to be familiar with patient and family wishes and could reduce variations of treatment related to sociodemographic and medical training factors. Stronger empirical data on the way that treatment decisions are made could provide the basis for an informed euthanasia policy. MJA 1996; 165: 540-544 Introduction Advances in medical technology have allowed the medical profession to maintain life to an extent that is beyond the limits of what many would consider reasonable quality. 1-5 The use of advanced technology, particularly for terminal illnesses, has created a complex array of ethical, moral and legal issues, 6-8 and has been accompanied by opportunities for greater patient participation in treatment decision-making and individual involvement in choosing care. 1,3,4,8 The most contentious aspect of these developments seems to be the promotion of requests for physician-assisted death -- euthanasia. 1,6,9 However, the emphasis on euthanasia has been tempered by a greater focus on palliative-care principles in the management of patients with terminal and life-threatening illnesses. 10-12 A major impediment to the consideration of these issues is the lack of strong empirical data to indicate the way in which treatment decisions are reached. The process that leads to decision-making and action in clinical practice is unclear, and the factors that determine who ultimately makes a treatment decision, what influences treatment choice, and how the treatment is carried out have yet to be established. Although the idea of informed consent implies significant patient participation in decision-making, there is no evidence that such participation actually occurs. Indeed, many patients are incapable of giving informed consent. Doctors and family members usually make decisions in such circumstances, but the path for this process is unknown. Some authors have advocated the use of advance health care directives that can advise doctors of a patient's wishes regarding care should he or she become incompetent. 2 Several studies, surveying both the general public and health professionals, have examined areas relevant to these issues, including reactions of family members to the death of loved ones; 8,10,13 attitudes of doctors to death, dying 1-5 and palliative care; 12,14 doctors' experience with requests for euthanasia; 1,9,15 and factors that influence doctors' choice of treatment. 1-5,7 However, while many of these reports discuss the legal, ethical, medical and philosophical problems confronting health professionals, they do not address the factors that influence the health professionals to act as they do. To examine the factors that influence doctors' decisions on treatment for patients with life-threatening or terminal illnesses, we presented doctors with actual comprehensive clinical scenarios described in a questionnaire and asked them to state what they would do. Methods Survey participants A postal survey was conducted between September and November 1995. An initial mailing of the questionnaire was sent to a national sample of 2016 doctors with a workload of 50 or more services (i.e., items billed to Medicare) per quarter, randomly selected by the Health Benefits Division of the then Commonwealth Department of Human Services and Health. The questionnaire was also posted to all Australian palliative care practitioners registered with the Australian and New Zealand Society of Palliative Medicine ( n = 156). Each practitioner was asked to complete and return the questionnaire anonymously. After two weeks, a reminder card was sent to all in the sample. Of the 2172 posted, 1588 usable questionnaires were returned (3% trainees [interns, residents, registrars]; 41% general practitioners; 10% palliative care practitioners; 46% specialists). The response rate, 73.1%, compared favourably with that of Baume and O'Malley (76.1%). 9 Eighty-seven questionnaires were returned because they were incorrectly addressed. Questionnaire The questionnaire was based upon that used by Molloy and colleagues 2 in Canada. We adapted their case scenarios to the 1995 Australian context, particularly including the issue of physician-assisted death. The questionnaire presented scenarios based on clinical cases (see Box 1). The doctors were asked to choose one of three treatment options in each scenario: 1. Supportive care -- keep warm, dry and pain-free; use only those measures that enhance comfort or minimise pain; start intravenous line only if it improves comfort; perform no x-rays or blood tests and give no antibiotics unless such measures are intended only to improve comfort; 2. Acute care -- may involve transfer to acute-care facility; perform emergency surgery if necessary; do not ventilate (except during and after surgery); carry out all acute-care interventions, except for admission to the intensive care unit or the coronary care unit and performing organ transplantation; 3. Intensive care -- all possible interventions necessary to maintain life. The doctors were also asked how they would respond to a patient's request for assisted death (scenario 4). Three possible responses were listed: 1. Refuse request for assisted death, and explain and provide palliative (supportive) care only; 2. Agree to assist death by providing the means (e.g., by prescribing lethal medication); and 3. Agree to assist death by active intervention (e.g., by administering a fatal dose of narcotic analgesic). To assess the degree to which six independent variables -- legal, ethical, religious and economic factors and patient's and family's wishes -- influenced the choice of treatment, the doctors were asked to complete a five-point Likert-type scale following each scenario. Finally, the questionnaire addressed the sociodemographic characteristics (sex; religious background) and the training of the participating doctors (country of basic medical degree; number of years qualified; level of medical training). The study was approved by the Human Rights Committee, the University of Western Australia. Statistical analysis Our analysis cross-tabulated the aggregated responses to the treatment options of the four clinical scenarios with the sociodemographic and medical training factors to examine whether medical decision-making varied significantly depending on these factors. Statistical significance was tested by chi-squared tests, with exact probability levels being reported. 16 We examined the strength and direction of the relationships among treatment options for the four clinical scenarios and the six independent variables. Somers' d correlation matrix is presented with chi-squared levels of statistical significance of 0.01 and 0.001. 16 Stepwise regression analysis was used to assess the degree of variance explained by the six independent variables in the doctors' responses to the request for physician-assisted death. Results Case scenarios 1 and 2 In the 80-year-old incompetent woman with life-threatening and acute illnesses without the medical directive (Box 2, scenario 1), there were significant differences among doctors' choices by all sociodemographic and medical training factors, except for religion. In the 80-year-old incompetent woman with life-threatening and acute illnesses with a medical directive for acute treatment (Box 2, scenario 2), there were significant differences by religion and medical training factors, but no significant difference between male and female doctors. This woman was twice as likely to receive acute care than was the woman without a medical directive. Case scenario 3 In their treatment choices for the 33-year-old competent woman with terminal and acute illnesses insisting on intensive care, there were significant differences among the doctors by medical training factors but not by sociodemographic factors (see Box 2). Doctors from non-Western countries were more likely to honour the patient's request for intensive care than were Western-trained practitioners; however, most doctors chose acute care for this patient. Older doctors and those not trained in palliative care were also more likely to honour the patient's request for intensive care. Case scenario 4 Nearly all doctors (more than 93%) chose supportive care for the treatment of the 56-year-old competent man with motor neurone disease who requested physician-assisted death. The only significant difference was between doctors trained in Western and non-Western countries, with a greater percentage of the latter choosing more interventionist medical treatment. Responses to request for assisted death Box 2 shows three notable findings on the doctors' responses to the request for assisted death: 1. Most doctors reported that they would provide palliative care only to this patient (explaining the grounds for the decision); 2. The percentage of doctors who were willing to provide only the means to assist death was larger than the percentage who were willing to intervene actively to assist death (frequently referred to as active euthanasia); 3. Female, Catholic, Western-trained and older doctors, as well as palliative care practitioners, were the least willing to engage in any form of euthanasia; agnostic/atheist doctors were the most willing to engage in some form of euthanasia; and doctors trained in non-Western countries were more likely than Western-trained doctors to be willing to provide the means to assist death, but were less likely than Western-trained doctors to actively intervene to assist death. Factors influencing choice of treatment Box 3 presents the Somers' d correlation matrix for scenario treatments with factors influencing treatment choice. For the incompetent 80-year-old woman without the medical directive (scenario 1), legal and ethical factors showed a significant positive correlation with the treatment choice, whereas the patient's and the family's wishes had a significant negative correlation with treatment choice. For the incompetent 80-year-old woman with a medical directive (scenario 2), cost and the patient's and family's wishes were significantly related to the treatment choice, the latter two of these three independent variables being positively associated with treatment choice. For the 33-year-old competent woman insisting on intensive treatment (scenario 3), ethics and the patient's and family's wishes had a significant relation to treatment choice. For the 56-year-old man with motor neurone disease who requested assisted death (scenario 4), none of the factors examined were significantly related to treatment choice. Across the matrix, patient's wishes seemed to have a significant positive relationship with doctors' treatment choice in scenarios 2 and 3, but not in scenario 1 (patient's wishes not known) or in scenario 4 (the patient who requested assisted death). This pattern of correlation suggests face validity for this set of questions. It should be noted that, while some coefficients are statistically significant, they do not suggest strong relationships between the variables. Factors influencing doctors' responses to patient's request for assisted death Stepwise regression analysis of the six independent variables that influenced the doctors' responses to the patient's request for assisted death showed three of these variables to be significant: patient's wishes (F = 67.0; P < 0.001), ethical factors (F = 38.8; P < 0.001) and religious beliefs (F = 27.2; P < 0.001). However, together, these three variables still explained only seven per cent of the variance in doctors' responses to the patient's request for assisted death. Legal (F = 0.028; P = 0.87) and cost factors (F = 2.618; P = 0.11) and family wishes (F = 0.017; P = 0.90) were not significant in predicting doctors' responses to the patient's request for assisted death in this analysis. Discussion A major finding of this study was that doctors did not make consistent decisions on the treatment of severely ill and terminally ill patients at the end of life. The data suggest that there are no uniform criteria to guide doctors in managing such clinical situations. If these decisions are left solely to doctors then the outcome will vary systematically according to medical training and sociodemographic factors. The data also show that doctors generally adhered to patients' wishes, both those expressed contemporaneously and those expressed via advance directives. The doctors also considered the wishes of family members to be important in making their treatment choices. Doctors' capacity to treat patients according to their wishes (and, if necessary, according to the wishes of their families) could be enhanced in two ways: 1. By engaging competent patients (and their families, if required) in comprehensive discussions of treatment options and likely outcomes -- in short, by participating in the informed-consent process. 2. By developing and implementing methods for advanced care planning (advance directives) so that patients can inform doctors of the care they would want if they were to suffer severe and terminal illness and become incompetent. The first practice, informed consent, may be required by law. However, the concept of informed consent presents problems in a multicultural society, not only because of potential communication problems, but also because of cultural differences in such matters as "truth telling" about the seriousness of an illness. 17 The second practice, the use of advance care planning, is accepted in principle (although its manner of implementation is debated) in North America and the United Kingdom. The practice is rarely even discussed in Australia. In multicultural Australia, research that could inform policy on the acceptability and the implementation of these two practices is lacking. Such research would seem to be imperative. With respect to euthanasia, this study showed that few doctors would have complied with the wish of the patient who requested assisted death. Baume and O'Malley 9 speculated that such reluctance was a function of the illegality of the action. Our data do not support this speculation. While patients' wishes, and ethical and religious factors for doctors (as with Baume et al. 18 ), seem to be more efficient predictors than legal factors, even the three former variables explain little of the variance in doctors' response to this request for assisted death (only seven per cent for the three variables together). As an alternative to the speculation of Baume and O'Malley, 9 it may be that, in Australia today, doctors believe that assisting a patient to die is an act that negates what they perceive to be the very essence of their profession. The different ways that doctors perceive this indefinable essential quality (that is, what it means to each of them to be a doctor) may explain the variations in their decision-making. As Lickiss 19 writes: There will always be differences of opinion on profound matters in a free society, but being put to death with one's consent is not a private matter, for it strikes at the foundations of what we are, and affects not only the one put to death but the one who carries it out. Our acts shape us, and the act of putting another person to death must change us. If we are doctors, it strikes at the core of what we should be in society: bringers of life, of hope, of healing, of comfort, sometimes bringers of bad news, companions on the way. But not bringers of death. Hunt, 20 an apparent proponent of some forms of euthanasia, notes that palliative-care practitioners are firmly opposed to physician-assisted death. Other medical practitioners are also opposed. 6,15 Nevertheless, despite the finest symptom management, some terminally ill patients make serious requests for assisted death. 21,22 Perhaps, if some form of euthanasia is to be available to these patients, medical practitioners may not be the most appropriate group to carry out the request. Certainly, it would be prudent to investigate doctors' perceptions of what it means to them to be a doctor in Australia today. Furthermore, as stated at the beginning of the paper, a major impediment to informed consideration of the issue of euthanasia in Australia is the lack of strong empirical data. While we have added to and improved the data by this study, in which we questioned a national sample of doctors about their end-of-life treatment decisions in the specific clinical scenarios described, we did not attempt the difficult task of obtaining empirical data related to clinical decision-making at the bedside. Data drawn from strong research into this decision-making process could provide important information before policy on euthanasia is drafted and implemented. Acknowledgements This research was funded by the Commonwealth Department of Health and Family Services, Canberra. References Stevens CA, Hassan R. Management of death, dying and euthanasia: attitudes and practices of medical practitioners in South Australia. J Med Ethics 1994; 20: 41-46. Molloy W, Guyatt GH, Alemayehu E, et al. Factors affecting physicians' decisions on caring for an incompetent elderly patient: an international study. Can Med Assoc J 1991; 145: 947-952. Kelner MJ, Bourgealt IL. Patient control over dying: responses of health care professionals. Soc Sci Med 1993; 36: 757-765. Solomon MZ, O'Donnell L, Jennings B, et al. Decisions near the end of life: Professional views on life- sustaining treatments. Am J Public Health ; 1993; 83: 14-22. Christakis NA. Physician characteristics associated with decisions to withdraw life support. Am J Public Health 1995; 85: 367-372. Komesaroff PA, Lickiss JN, Parker M, Ashby MA. The euthanasia controversy: decision-making in extreme cases. Med J Aust 1995; 162: 594-597. Kaufman SR. Decision making, responsibility, and advocacy in geriatric medicine: physician dilemmas with elderly in the community. Gerontologist 1995; 35: 481-488. Ashby M, Wakefield M. Attitudes to some aspects of death and dying, living wills and substituted health care decision-making in South Australia: public opinion survey for a parliamentary select committee. Palliat Med 1993; 7: 273-282. Baume P, O'Malley E. Euthanasia: attitudes and practices of medical practitioners. Med J Aust 1994; 161: 137-145. Maddocks I. Good palliative care orders. Palliat Med 1993; 7: 35-37. Baume P. Living and dying: a paradox of medical progress. Med J Aust 1993; 59: 792-794. Wakefield MA, Beilby J, Ashby MA. General practitioners and palliative care. Palliat Med 1993; 7: 117-126. Seale C, Addington-Hall J. Euthanasia: the role of good care. Soc Sci Med 1995; 40: 581-587. Allbrook D. Palliative care in the 1990s? Med J Aust 1991; 155: 286-287. Anderson JG, Caddell DP. Attitudes of medical professionals toward euthanasia. Soc Sci Med 1993; 37: 105-114. Siegel S. Nonparametric statistics for the behavioural scientist . London: McGraw-Hill 1956. Waddell C, McNamara B. The stereotypical fallacy: a comparison of Chinese- and Anglo-Australians' thoughts about death-related matters. Presented at the Annual Australian Sociology Association Meeting, Newcastle, NSW. December 1995. Baume P, O'Malley E, Bauman A. Professed religious affiliation and the practice of euthanasia. J Med Ethics 1995; 21: 49-54. Lickiss, N. Chapter in: Chapman S, Leeder S, editors The last right? Australians take sides on the right to die. Melbourne: Mandarin, 1995: 98. Hunt RW. The hospice movement matures. Med J Aust 1996; 164: 452-453. Stephany TM. Assisted suicide: how hospice fails. Am J Hospice Palliat Care 1994; 11: 4-5. Peteet JR. Treating patients who request suicide: a closer look at the physician's role. Arch Fam Med 1994; 3: 723-727. (Received 15 May, accepted 29 Jul 1996) Authors' details The University of Western Australia, Nedlands, WA. Charles Waddell, PhD, Senior Lecturer in Anthropology. Department of Geriatric Medicine, Osborne Park Hospital, Perth, WA. Roger M Clarnette, MB BS, FRACP, Consultant Physician. Silver Chain Hospice Care Services, Perth, WA. Michael Smith, MB BS, MRACMA, Director of Clinical Services. The Cancer Foundation Cottage Hospice, Perth, WA. Lynn Oldham , BN(Hons), Clinical Nurse Specialist. Turning Point Alcohol and Drug Centre, Fitzroy, VIC. Allan Kellehear, PhD, Professor, and Head of Research Development. No reprints will be available. Correspondence: Dr Charles Waddell, Department of Anthropology, The University of Western Australia, Nedlands, WA 6907. E-mail: cwaddell AT uniwa.uwa.edu.au - - To top of article - ©MJA 1996 <URL: http://www.mja.com.au/> © 1996 Medical Journal of Australia.

Charles Waddell · Rodger M Clarnette · Michael Smith · Lynn Oldham · Allan Kellehear

Patterns of alternative medicine use by cancer patients Stephen D Begbie, Zoltan L Kerestes and David R Bell MJA 1996; 165: 540 For editorial comment, see Lowenthal Readers may print a single copy for personal use. No further reproduction or distribution of the articles in whole or in part should proceed without the permission of the publisher. For copyright permission, contact the Australasian Medical Publishing Company 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/>". Abstract - Introduction - Methods - Results - Experiences of conventional treatment - Alternative treatment - Patient characteristics and alternative therapy use - Discussion - Acknowledgements - References - Authors' details - - ©MJA1996 Abstract Objective: To assess the patterns of alternative medicine use in patients of a public hospital oncology unit, and to compare patients' experience of alternative with conventional medicine. Design and setting: Self-administered questionnaire survey of cancer patients attending specialist consulting rooms at the Royal North Shore Hospital and the Oncology Outpatient Clinic at Port Macquarie Base Hospital during August 1995. Participants: 507 patients attended the clinics; 335 (66%) returned questionnaires, of which 319 (62%) were sufficiently complete for analysis. Main outcome measures: Expectations of and satisfaction with both conventional and alternative treatment, use of alternative treatment, and patient characteristics associated with this use. Results: Expectations of and satisfaction with both conventional and alternative treatment were very high. Alternative treatments (most commonly dietary and psychological methods) were used by 21.9% of patients. Median annual cost of alternative therapy was $530, with most patients reporting "value for money". Younger age and being married were positively associated, and satisfaction with conventional treatment was negatively associated, with alternative medicine use; 40% of patients did not discuss alternative medicine with their physician. Conclusions: A significant proportion of cancer patients use one or more forms of alternative therapy. The use of alternative therapy may reflect on deficiencies in the current standard of care. MJA 1996; 165: 545-548 Introduction Rapid developments in the conventional management of cancer have been accompanied by an increased consumer-driven move toward choice and individual control. While alternative (or complementary) medicine has always existed, its availability and variety are growing and its use increasing. 1 Cancer patients use alternative therapies despite arguments such as lack of scientific validity, proven ineffectiveness of some treatments and reliance on the placebo effect. It may be helpful for Australian doctors to know the range of alternative therapies used by cancer patients, and to have an understanding of the reasons why they use such therapies, as this may cast some light on deficiencies in the current standard of conventional care. We aimed to investigate the range of alternative therapies used by cancer patients attending outpatient clinics conducted by a teaching hospital clinical oncology unit, and to assess the level of satisfaction with these therapies and with conventional therapy. Methods Our subjects were all patients who attended the specialist consulting rooms of three medical oncologists at Royal North Shore Hospital (RNSH) in Sydney, and the Oncology Outpatient Clinic at Port Macquarie Base Hospital (PMBH), New South Wales, during August 1995. Reception staff offered each patient a self-administered questionnaire, with a covering letter from the chief investigator encouraging participation, and assuring confidentiality. Response to the questionnaire was voluntary, and patients were not asked to give explanations for not responding. The questionnaire sought demographic data, and the following information about their conventional treatment: understanding of diagnosis and treatment, expectations before treatment, impressions at the conclusion of treatment, and overall satisfaction. All questions were multiple choice, with the exception of diagnosis. Those who had used alternative medicine were asked (in multiple-choice questions, but with space for comments) to outline their reasons for doing so, as well as the treatments chosen. The questionnaire did not seek details of the chosen therapies. The same questions that were used to assess their conventional medicine experience were used to assess their alternative medicine experiences. In addition, they were asked about the reaction of family and friends to their use of alternative therapy, whether they had discussed alternative treatment with their medical oncologist, to estimate the financial cost, and whether they felt they had received value for money. Data were analysed using the Statistical Package for the Social Sciences 2 to determine associations between patient characteristics and alternative medicine use; significance was measured by the odds ratio. Logistic regression analysis was used to analyse all variables which may affect use of alternative therapy. Approval for this study was received from the Royal North Shore Hospital Ethics Committee, and from Port Macquarie Base Hospital. Results Five hundred and seven patients attended, of whom 335 (66%) responded (271 from RNSH, 64 from PMBH). Sixteen questionnaires were excluded from assessment (the alternative medicine question was unanswered in 12, no demographic details were given in 3, and one was returned blank), leaving 319 (62%) to be assessed. Demographic details for the 319 patients are shown in Box 1. It is noteworthy that in our study women, married people, and those with high levels of education and private health insurance predominate. Only 22.9% of the 319 patients were in paid employment, and median household income was in the $20 000-$40 000 range. A wide variety of diagnoses were represented; breast cancer was the single most common diagnosis. Experiences of conventional treatment Most respondents had received chemotherapy, and many had undergone other kinds of cancer therapy (Box 1). When asked about their expectations at the beginning of treatment, 85.6% believed that it would cure them or prolong their lives. By contrast, at the end of conventional treatment 63.0% felt that they had been cured or that their life had been prolonged, while 8.2% felt that their treatment had been of no benefit, or had made them worse. Most respondents (75.3%) were either satisfied or very satisfied with their experience of conventional therapy, while only 3.1% were either unsatisfied or very unsatisfied. Alternative treatment Seventy patients of the 319 assessed (21.9%) indicated that they were using alternative therapy. The most frequently given reasons were a preference for natural therapy, and seeing the alternative therapy as another source of hope (Box 2). Box 3 shows the alternative therapies chosen; dietary and psychological methods were most prevalent, followed by herbalism. Seventy-five per cent of patients tried more than one therapy (median, 3; range, 1-8). Most of the patients had learnt of alternative therapies through friend or family recommendation (37; 52.9%), and personal research (32; 45.7%), while doctor's recommendation (11; 15.7%) and media reports (6; 8.6%) were less common (some patients cited more than one source). In addition, family and friends were overwhelmingly supportive, with those of 64 patients (91.4%) being either encouraging or tolerant of the use of alternative therapies. At the beginning of alternative treatment 51 patients (72.9%) expected it would cure them or prolong their lives, while at the completion 34 (48.6%) felt that they had been cured, or that their lives had been prolonged. Only seven (10%) had the impression that alternative treatment was of no benefit, or had made them worse. Overall, 49 patients (70%) were satisfied or very satisfied with alternative treatment, and only one (1.4%) was unsatisfied. Forty-five patients gave an estimate of the annual cost of their alternative therapy; the median annual cost was $530 (range, 0-$20 000). Of the 70 patients who had alternative therapy, 45 (64.3%) felt they were getting value for money, five (7.1%) did not, and the remainder did not answer this question. Importantly, 37 (52.9%) felt that they could discuss their alternative treatment with their physician, while 28 (40%) felt they could not. Patient characteristics and alternative therapy use Being young and being married were significantly associated with use of alternative medicine, with marital status significant at P = 0.02. Logistic regression analysis showed that age and satisfaction with conventional therapy are the key predictors for a decision to use alternative therapy. Overall, younger patients used more alternative therapy (odds ratio, 1.67; P < 0.0001; 95% CI, 1.32-2.13), and those who were very satisfied with their conventional treatment used less alternative therapy (odds ratio, 0.55; P < 0.0001; 95% CI, 0.38-0.79). Sex, level of education, employment status, income, private health insurance, diagnosis, expectations at beginning of treatment or impressions at the end of treatment were not significantly associated with use of alternative therapy. Discussion Any discussion of alternative medicine is complicated by arguments about terminology. The most commonly used terms in the literature are "alternative", "complementary" or "unproven". Some have suggested that the term "alternative" is spurious, because it suggests equally valid options, 3 but we have chosen to use it because it is in common use in describing treatments outside conventional medicine. "Complementary" may seem an appropriate term for a study of patients attending a hospital, but it suggests therapies that are valuable when combined with conventional medicine. "Unproven" is the American Cancer Society's preferred term and relates to methods that are "on the basis of careful review by scientists and/or clinicians, not deemed proven, nor recommended for current use". 4 It has also been suggested that "There are no unproven treatments for universally curable diseases" 5 -- the very fact that this issue is so controversial in cancer medicine suggests that conventional treatment has a long way to go in optimising management of malignant disease. Previous studies have shown that between 9% and 54% of adult cancer patients use some form of alternative medicine, 6-10 while a recent Australian study reported a 46% rate of alternative therapy use in a paediatric oncology population. 11 The rate of alternative medicine use in our study (21.9%) is within this range. It may not be an accurate reflection of alternative therapy use in cancer patients as our patients' response to the questionnaire was voluntary, creating a potential selection bias. Nevertheless it does indicate that a significant proportion of cancer patients use some form of alternative therapy, and gives some idea of the current range of therapies. Several features of our study population should be considered when relating our results to other populations. Royal North Shore Hospital services the northern Sydney area, where the population is characterised by being relatively wealthy, and having high levels of education and private health insurance. Port Macquarie Base Hospital serves a community with a large retired population and relative wealth (by rural standards). The high prevalence of breast cancer managed by this unit explains the predominance of women in the group studied. There was a high level of optimism at the outset of both conventional treatment and alternative therapy (85.6% v. 63.0% belief in cure/life-prolongation), which fell modestly by completion of treatment (72.9% v. 48.6%). As most patients were receiving palliative chemotherapy, it seems that many patients had an unrealistic expectation about the goals of treatment. This reinforces the need for clear communication with patients. However, most patients were highly satisfied with both conventional and alternative treatments (75.3% v. 70.0%), and dissatisfaction was almost non-existent (3.1% v. 1.4%). This high level of satisfaction, often in the absence of objective benefit, has been described in studies of conventional 12 and alternative 6,11 medicine, and may be related to a hope engendered by the intervention of a clinician. 13,14 As in other studies which report that patients will frequently try multiple alternative therapies, 6,11 we found that over 75% of patients used two or more therapies. Alternative therapies available in Australia appear to be a mixture of the British and United States experience. Psychological methods such as relaxation, imagery and healing, which are prevalent in the United Kingdom, 6 and interventional methods, such as diet and megavitamins, more common in the US, 7 are used with similar frequency. The predominance of friend and family recommendation (52.9%) as the source of information on alternative medicine may partially explain the significantly increased use by married patients. The contribution of recommendation by doctors is comparatively low, but it must be remembered that these patients were attending an oncology service. Of course, some cancer patients may not seek conventional care, and consultations with alternative practitioners, or with medical practitioners who combine conventional and alternative therapies, are becoming more common. 15,16 In the US there are currently more referrals to alternative practitioners than to primary care physicians. 17 Many studies report the significant financial costs of alternative medicine. 5,9,18-20 There is a wide variation in the cost of various therapies, with psychological methods usually far less expensive than interventions such as metabolic therapy, Laetrile and high-dose vitamin C. In one US study, the median annual outlay was US$262, and the mean was US$3492, suggesting enormous expense in some cases. 7 Our findings are similar, with expenditure ranging from nothing to $20 000 per year. However, nearly all of those using alternative medicine felt they were getting value for money. Previous studies have reported that users of alternative therapy tend to be young, female, of higher social class and more highly educated, while income, increased time since cancer diagnosis, increased number of people in a household, and tumour site may be associated with alternative therapy use. 6,7,9 We found that being young and being married were significantly associated with increased use, and that those who were very satisfied with conventional therapy used less alternative therapy. The finding that being married is associated with increased use of alternative therapy may be related to having increased numbers in a household, 9 and perhaps greater access to ideas and support for trying new things. Many patients in our study did not discuss their use of alternative therapy with their physicians, possibly because they were afraid of getting a negative response. This may reflect a need to improve doctor-patient communication. If a proportion of patients are going to use alternative therapies, it is important that their doctors are informed: firstly, so that possible interactions and complications can be avoided, and, secondly, to enable doctors to provide informed opinion on the alternative choices available. Acknowledgements We thank Professor John A Levi and Dr Helen R Wheeler for contributing their patients to the study. References MacLennan AH, Wilson DH, Taylor AW. Prevalence and cost of alternative medicine in Australia. Lancet 1996; 347: 569-573. SPSS: Statistical package for the social sciences [computer program], version 5. Chicago, Ill: SPSS inc, 1990. McGinnis LS. Alternative therapies, 1990. An overview. Cancer 1991; 67: 1788-1792. Pamphlet number 3028-REV. Atlanta, Ga.: American Cancer Society, 1990. Cassileth BR. The social implications of questionable cancer therapies. CA Cancer J Clin 1989; 39: 311-316. Downer SM, Cody MM, McCluskey P, et al. Pursuit and practice of complementary therapies by cancer patients receiving conventional treatment. BMJ 1994; 309: 86-89. Cassileth BR, Lusk EJ, Strouse TB, Bodenheimer BA. Contemporary unorthodox treatments in cancer medicine. A study of patients, treatments and practitioners. Ann Intern Med 1984; 101: 105-112. Harris L and associates. Health information and the use of questionable treatments: a study of the American Public, 1987. Washington DC: Department of Health and Human Services, 1987. Lerner IJ, Kennedy BJ. The prevalence of questionable methods of cancer treatments in the United States. CA Cancer J Clin 1992; 42: 181-191. Miller MJ, Boyer MJ, Dunn SM, et al. Why do Australian cancer patients use unproven therapies? Proc Clin Oncol Soc Aust 1995; 22: 78. Sawyer MG, Gannom AF, Toogood IR, et al. The use of alternative therapies by children with cancer. Med J Aust 1994; 160: 320-322. Coates A, Gebski V, Bishop JF, et al. Improving the quality of life during chemotherapy for advanced breast cancer. N Engl J Med 1987; 317: 1490-1495. Stoll BA. Can unorthodox cancer therapy improve quality of life? Ann Oncol 1993; 4: 121-123. Kodish E, Post SG. Oncology and hope. J Clin Oncol 1995; 13: 1817-1822. Borkan J, Neher JO, Anson O, Smoker B. Referrals for Alternative therapies. J Fam Prac 1994; 39: 545-550. Paterson C, Peacock W. Complementary practitioners as part of the primary health care team: evaluation of one model. Br J Gen Pract 1995; 45: 255-258. Burke C, Sikora K. Complementary and conventional cancer care: the integration of two cultures. Clin Oncol 1993; 5: 220-227. Guzley GJ. Alternative cancer treatments: impact of unorthodox therapy on the patient with cancer. South Med J 1992; 85: 519-523. Cassileth BR, Lusk EJ, Guerry D, et al. Survival and quality of life among patients receiving unproven as compared with conventional cancer therapy. N Eng J Med 1991; 324: 1180-1185. Ernst E. Complementary medicine: common misconceptions. J R Soc Med 1995; 88: 244-247. (Received 24 Jan, accepted 5 Aug 1996) Authors' details Department of Clinical Oncology, Royal North Shore Hospital, Sydney, NSW. Stephen D Begbie, MB BS, Oncology Registrar; Zoltan L Kerestes, PhD, Computer Scientist; David R Bell, MB BS, FRACP, Medical Oncologist. Reprints: Dr D R Bell, Department of Medical Oncology, Royal North Shore Hospital, St Leonards, NSW 2065. - To top of article - ©MJA 1996 <URL: http://www.mja.com.au/> © 1996 Medical Journal of Australia.

Stephen D Begbie · Zoltan L Kerestes · David R Bell

Routine pulmonary function tests in young adolescents with asthma in general practice Peter H Hewson, Elizabeth A Tippett, Danny M Jones, Justin P Madden and Peter Higgs For editorial comment, see Sawyer and Bowes Readers may print a single copy for personal use. No further reproduction or distribution of the articles in whole or in part should proceed without the permission of the publisher. For copyright permission, contact the Australasian Medical Publishing Company 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/>". Abstract - Introduction - Methods - Results - Discussion - Acknowledgements - References - Authors' details - - ©MJA1996 Abstract Objective: To assess the value of performing routine pulmonary function tests by flow-loop spirometry in young adolescents with asthma. Design: A prospective clinical study comparing clinical assessment and patients' self- reporting of asthma severity with the results of pulmonary function tests. Setting: General practice in a small rural community of about 30 000 people. Patients: Young adolescents with asthma, aged 10-15 years, were enrolled in the study over a two-year period from July 1993 to June 1995 when they presented for either elective, interval assessments or with an acute exacerbation of asthma. Main outcome measures: Discrepancy between (i) the doctor's and the patient's perception of asthma control (six scale measures) and the consequent management plans, and (ii) the results of pulmonary function tests that indicated less than adequate airway function (i.e., forced expiratory volume in one second as a percentage of predicted vital capacity for height and sex [FEV 1 %] less than 65% or average flow rate over the middle 50% of forced vital capacity as a percentage of predicted normal value [FEF 25%-75% ] less than 65%). Results: Twenty-seven adolescents with asthma were assessed on a total of 37 occasions. The results of pulmonary function tests did not correlate with asthma symptoms and treatment in 11 of the 37 assessments (30%; 95% confidence interval [CI], 16%-47%). The 11 assessments were performed on eight patients. Conclusions: This small community-based study of adolescents with asthma supports the view that pulmonary function testing by flow-loop spirometry should be part of the routine assessment of acute and chronic asthmatics. Further study in a larger community is needed to clarify the frequency of over- and underestimation of asthma severity in this difficult age group. MJA 1996; 165: 469-472 Introduction Older children and adolescents often have difficulty assessing the severity of their asthma, 1 and consequently pulmonary function tests have been recommended for those with frequent or chronic asthma who are able to perform spirometry adequately. 2-4 Pulmonary function testing by flow-loop spirometry allows measurement of forced expiratory volume in one second as a percentage of the predicted vital capacity for height and sex (FEV 1 %), and forced mid expiratory flow (FEF 25%-75% : the average flow rate over the middle 50% of forced vital capacity). While peak flow monitoring and FEV 1 % have been used for decades as a measure of pulmonary function, forced mid expiratory flow is not as well known a measure, but is more sensitive than FEV 1 in detecting small airways obstruction. 2 Pulmonary function tests have been shown to be useful in adults, but there are few studies of their useful ness in children and young adolescents with asthma. One study found that 17% of apparently well children with asthma had a low forced expiratory volume in one second / vital capacity ratio (FEV 1 /FVC) and 54% had low forced mid expiratory flow (FEF 25%-75% ), suggesting that, without these tests, both patients and doctors were unable to detect mild degrees of broncho constriction, especially of the smaller airways. 5 To date, no study has described the use of pulmonary function tests as part of routine community medical practice in young adolescents. In fact, in some centres, spirometry tests are not yet done routinely in chronic asthmatics in this age group. We performed pulmon ary function tests in adolescents with asthma in an Australian country setting, comparing patients' and doctors' clinical assessment of asthma severity with the results of routinely performed pulmonary function tests. Methods This two-year study took place in Colac, Victoria (population, 14 000; regional population, 30 000) between July 1993 and June 1995. All young adolescents with asthma (recurrent wheeze responsive to bronchodilators) between the ages of 10 and 15 years pre senting to their general practitioner Monday to Friday either for an elective asthma management review, or because of an acute exacerbation of their asthma, were assessed by the general practitioner and included in the study. No adolescents were excluded from the study or refused to participate. Appointments for elective management reviews (interval assessments) had been made weeks in advance. An acute exacerbation of asthma was defined as an appreciably more severe wheeze and cough than usual (as assessed by the general practitioner). The severity of symptoms and the perception of asthma control were assessed by the general practitioner, giving day and night symptom scores (0-5), and a wheeze description, based on questioning of the patient. The doctor and the patient then separately graded asthma control as excellent, very good, fair, poor, or very poor. (The scoring system is given in the footnote to the Table). Forced expir atory volume in one second (FEV 1 ) and forced mid expiratory flow (FEF 25%-75% ) were measured -- on the same day if it was a daytime presentation (usually immediately after the clinical assessment) or the morning after if it was an evening presentation. An Alpha Vitalograph spirometer (Fisher & Paykel, Melbourne, Vic.) was used and the tests were conducted by the senior physiotherapist at Colac Hospital (E A T), who had attended the spirometry technician's course run by the Department of Respiratory Medicine, Alfred Hospital, Melbourne. The tests were repeated if the assessment was judged to be inadequate by the physiotherapist. The effect of bronchodilators on pulmonary function tests was determined in all patients, with recordings being taken 20 minutes (in one patient 25 minutes) after bronchodilator therapy. The pulmonary function tests were done without knowledge of the clinical status of the patient and the results were made available to the treating doctor if requested. The patients' previous home peak expiratory flow measurements were also recorded, as was peak flow measured with the spirometer. The results of the pulmonary function tests were considered not to correlate with asthma symptoms and treatment, and to be likely to lead to a change in asthma management, if: During an interval assessment: The patient complained of no, few or only moderate symptoms, the doctor felt asthma control was excellent, very good or reasonable, respectively, and did not change treatment, but FEV 1 % (normal, > >80%) or FEF 25%-75% (normal, > >65%) was less than 65%. The patient and the doctor thought control was poor, the dose of maintenance asthma therapy was increased, but pulmonary function tests revealed no evidence of bronchospasm. During assessment of an acute exacerbation: The patient and the doctor felt current asthma control was excellent or very good, no increase in dose of inhaled steroids was suggested and no oral corticosteroids were prescribed, but FEV 1 % or FEF 25%-75% was less than 65%. Results Twenty-seven adolescents with asthma (male : female ratio, 2 : 1) were assessed on a total of 37 occasions. On 22 occasions they were elective interval assessments and 15 were for an acute asthma exacerbation. Nineteen patients (70%) were monitoring peak flow at home. Twenty-nine (78%) of the pulmonary function tests were within two hours, four (11%) between two and four hours and four (11%) between four and 14 hours after the clinical assessment. Satisfactory spirometry measurements were achieved in all patients without difficulty (fewer than four repeats). Of the 22 interval assessments, 16 (73%) were in patients taking inhaled corticosteroids (in 11 of these the patients were taking more than 700 µg inhaled corticosteroids per day, and in five FEF 25%-75% values were less than 65%). In two (9%) interval assessments peak expiratory flows were less than 300 L/min (in only one of the assessments with FEF 25%-75% less than 65% were peak flows less than 300 L/min). Of the 15 acute exacerbation assessments, eight (53%) were in patients taking inhaled corticosteroids (in three of these the patients were taking more than 700 µg inhaled corticosteroids per day). In five (45%) acute exacerbation assessments peak expiratory flows were less than 300 L/min (four were unavailable). Details of the patients whose pulmonary function tests did not correlate with asthma symptoms and treatment are given in the Table. Overall, pulmonary function tests not correlating with clinical and patient assessment of asthma severity were found in 11 of 37 assessments (30%; 95% confidence interval [CI], 16%-47%). If only those tests performed within four hours of the clinical assessment are included, pulmonary function tests in 10 of 33 assessments (30%; 95% CI, 15%-46%) did not correlate with asthma severity. Pulmonary function tests in six of the 22 interval assessments (27%; 95% CI, 11%-50%) showed FEV 1 % or FEF 25%-75% to be less than 65% when management had not been changed after clinical assessment (Cases 1-6). During acute exacerbations, four of 15 assessments (27%; 95% CI, 8%-55%) showed FEV 1 % or FEF 25%-75% values to be less than 65% when no change in treatment had been made on clinical grounds (Cases 8-11). Tests in one patient (Case 7) showed normal pulmonary function after clinical assessment had suggested poor asthma control. Her pulmonary function tests were carried out within two hours of the clinical evaluation; her inhaled cortico steroid dose had been doubled. Discussion This community-based study of all young adolescent asthmatics presenting to their general practitioners in a country town found that in 30% of assessment opportunities the results of pulmonary function tests were likely to change management. This proportion was maintained even if slightly delayed pulmonary function tests (4-14 hours) were excluded. This result is comparable with previously published findings that 54% of apparently well asthmatics had lower than expected FEF 25%-75% values at follow-up, despite being asymptomatic. 5 At interval assessments in which management had not been changed and the results of pulmonary function tests were low, three of five adolescents (Table: assessments 1, 3-6) reported frequent wheeze. The other two thought their asthma was well controlled; however, their FEF 25%-75% values were 62% and 42%, respectively. In these patients peak flows, measured at home and by spirometry, were all above 300 L/min and this may have accounted for ALIGN=TOP the reluctance to change therapy. However, it has been shown previously that peak flow results can be misleading, and widely varying optimal values can be expected. 3 A low expectation of what can be achieved in frequently symptomatic patients may contribute to a less aggressive approach in adolescents with asthma. The rather frequent use of inhaled corticosteroids in relatively high doses in this small sample of asthmatics suggests that a more objective measure of asthma status should be used. Potential overuse of inhaled corticosteroids in these growing young people may not always safeguard those at risk of more severe asthma. Six of 11 patients (55%) taking more than 700 µg inhaled corticosteroids had FEF 25%-75% values less than 60%, suggesting that in these patients an even higher dose of inhaled corticosteroid may be required. More specific alterations to long term inhaled steroid use would be possible if pulmonary function tests were performed regularly. During assessments for acute exacerbation the rate of pulmonary function tests not correlating with symptoms was still high (27%) and the FEF 25%-75% was worryingly low in three patients in whom management was not altered (Table: assessments 8, 9, 10, and 11). These three adolescents all had frequent wheeze, but presumably were not distressed, with peak flows (measured at home and by spirometry) above 320 L/min. Pulmonary function tests gave an indication of small airways disease which they were not aware of or was not revealed by their peak flow measurements. Our study involved a small number of patients and the confidence intervals calculated suggest a larger study is necessary. Nevertheless, data on mortality in asthma indicate that underestimation of asthma severity can be extremely important. Robertson et al. found that about 35% of possibly preventable asthma deaths may have been related to medical practitioner underestimation of asthma severity. 6 The full implications of FEF 25%-75% values in the 55%-65% range is not yet certain. A long term follow-up of asymptomatic patients with values in this range has not yet been done. Most of our patients with low FEF 25%-75% values not correlating with symptoms had values below 55% (Table: 7 of 10 assessments). Thus, even if the implications of an FEF 25%-75% value between 55% and 65% is disputed, the frequency of the lower results suggests more aggressive treatment is required. This relatively small study of asthmatic adolescents in a small country town strongly supports the view that pulmonary function tests need to be part of the routine assessment of acute and chronic asthmatics. Acknowledgements We thank Professor P D Phelan, Department of Paediatrics, University of Melbourne, for his provocation, encouragement and support; and Mr Ross Gollan, Senior Lecturer, Department Mathematics and Statistics, Deakin University. References Sly PD, Landau LI, Weymouth R. Home recording of peak expiratory flow rates and perception of asthma. Am J Dis Child 1985; 139: 479-482. Landau LI. The value of lung function in guiding drug therapy in childhood asthma. Eur Respir Rev 1994; 4: 10-14. Phelan PD, Olinsky A, Robertson CF. Respiratory illness in children, 4th edition. Cambridge: Blackwell Scientific Publications, 1994; 152-156. Milner AD. Childhood asthma: diagnosis, treatment and management. London: Martin Dunitz, 1987; 18-38. Bye MR, Kerstein D, Barsh E. The importance of spirometry in the assessment of childhood asthma. Am J Dis Child 1992; 146: 977-981. Robertson CF, Rubinfeld AR, Bowes G. Deaths from asthma in Victoria: a 12-month survey. Med J Aust 1990; 152: 511-517.(Received 19 Sep 1995, accepted 4 Jul 1996) Authors' details 106 McKillop Street, Geelong, VIC 3220. Peter H Hewson, MD, FRACP, Consultant Paediatrician. Colac Hospital, Corangamite Street, Colac, VIC. Elizabeth A Tippett, DipPhys, Senior Physiotherapist. 2 Connor Street, Colac, VIC. Danny M Jones, MB BS, DA, DipRACOG, General Practitioner; Justin P Madden, MB BS, DA, DRCOG, General Practitioner; Peter Higgs, MB BS, General Practitioner. No reprints will be available. Correspondence: Dr P H Hewson. - - To top of article - ©MJA 1996 <URL: http://www.mja.com.au/> © 1996 Medical Journal of Australia.

Peter H Hewson · Elizabeth A Tippett · Danny M Jones · Justin P Madden · Peter Higgs

Major depression and refusal of life-sustaining medical treatment in the elderly Stuart C Hooper, Kevin J Vaughan, Christopher C Tennant and Janette M Perz MJA 1996; 165: 416-419 Abstract - Introduction - Methods - Patients and assessment - Follow-up - Statistical analyses - Results - Patients - Intensity of depression - Desire for life-sustaining treatment - Discussion - Acknowledgements - References - Authors' details Abstract Objective: To examine the effect of improvement or recovery from major depression in elderly patients on their desire for life-sustaining treatments. Design: Prospective case survey. Setting: Psychogeriatric Service of Hornsby Ku-ring-gai Hospital & Community Health Services and Ryde Hospital & Community Health Services (a community-based service serving an urban population with over 29 000 elderly people). Subjects: All patients referred with major depression (diagnosed by DSM-IV criteria) and no significant cognitive impairment between October 1994 and January 1995. Outcome measures: Number of life-sustaining treatments desired before and after treatment of depression in two hypothetical acute life-threatening illnesses (one with a good and the other with an uncertain prognosis), and intensity of depression on the Geriatric Depression Scale (GDS). Results: 22 subjects completed both stages of the study. Initial depression was severe in five, moderate in 16 and mild in one. GDS scores decreased in 18 subjects after treatment. Mean number of life-sustaining treatments desired by these patients increased significantly from 4.0 to 6.1 (out of seven possible treatments) in the "good prognosis" illness and from 6.4 to to 9.6 (out of 14) overall. The increase in the "uncertain prognosis" illness (2.3 to 3.4) was not significant. Conclusions: Moderate or severe major depression in the elderly is associated with a high degree of refusal of life-sustaining treatments. Treatment of the depression leads to increased acceptance of these treatments, particularly if prognosis is good. MJA 1996; 165: 416-419 Introduction Decisions about withdrawing or withholding life-sustaining medical treatments have become problematic in recent decades. Dramatic advances in medical technology enable many severely ill patients to remain alive, yet the decision to prolong life is tempered by the finite nature of resources, economic rationalism and the patient's quality of life. In the United States, it is estimated that decisions to limit medical treatment are made for about 70% of patients who die in hospital.1 In many cases, this is due to patient refusal of further medical treatment and is part of the strong shift towards respecting patients' autonomy and right to self-determination. In all Australian States, there is a common-law right for competent patients to refuse medical treatment, including life-sustaining treatment. In Victoria, South Australia and the Northern Territory, this right has received statutory recognition. In these States, legislation also permits competent individuals over 18 years to make advance directives ("living wills") specifying that in the event of a terminal illness the individual does not wish to be subjected to extraordinary life-prolonging measures.2 However, the potential influence of depressive illness on patients' refusal of life-sustaining treatment is often neglected. In the elderly (who are disproportionately represented among those who have life-sustaining treatment withheld or withdrawn), depression is often not recognised by primary care physicians.3-7 Misdiagnosis is especially likely in the elderly if symptoms are atypical (e.g., cognitive deficits [pseudo dementia], somatisation or anxiety). Further, with depression impairments in decision-making may be under-recognised, as the cognitive aspects of competence (which are emphasised by doctors and courts) may remain intact, while more subtle abilities (such as the ability to assign realistic values or meanings to the risks and benefits of prospective treatments)8,9 may be impaired. Depressed patients may undervalue potential positive outcomes and predict negative outcomes.9 These "affective" aspects of clinical competence are generally not considered as important as the "cognitive" aspects and can be more difficult to detect.10 The negative outlook disappears after recovery from depression, suggesting that it is specific to the illness phase and not a trait of depression-prone individuals.11,12 Case reports have highlighted patients who have initially refused, but later accepted, life-sustaining treatment after recovery from depression.9,15-18 Longitudinal studies of male Veterans Administration patients in the United States found that desire for life-sustaining treatment was not increased by recovery from major depression of mild to moderate severity,13,14 but was significantly increased if the major depression was severe.14 However, the extent to which results of these case reports and longitudinal studies can be generalised is uncertain. We therefore investigated the effect of recovery from major depression on preferences for life-sustaining treatments in a group of patients referred to an Australian community psychogeriatric service. Methods Our study was approved by the Ethics Committee of the Hornsby Ku-ring-gai Hospital & Community Health Services and Ryde Hospital & Community Health Services. Patients and assessment Subjects for the study were drawn from consecutive patients referred to the Psychogeriatric Service of Hornsby Ku-ring-gai Hospital between October 1994 and January 1995. The Psycho geriatric Service is a community-based service with inpatient facilities, serving an area with over 29 000 elderly people. Inclusion criteria: Diagnosis of DSM-IV major depression,19 made by clinical interview with the consultant psychiatrist and senior psychiatric registrar of the Psychogeriatric Service, according to DSM-IV criteria (Box 1). Severity of depression (mild, moderate, severe) was also assessed clinically with DSM-IV criteria; Mini-Mental State Examination score (MMSE) > 17.20 This was used to screen out patients with significant cognitive impairment; and Informed consent given. Intensity of depression was assessed with the Geriatric Depression Scale (GDS). This is a rating scale (not a diagnostic instrument) that provides a numerical index of intensity of depression and was designed exclusively for use in elderly patients. It contains 30 questions requiring a yes or no answer, which can be either self- or observer-administered. One-week test-retest reliability of 0.85 and criterion validity of 0.82 have been recorded.21,22 Desire for life-sustaining treatment was assessed by a questionnaire adapted from Lee and Ganzini,13 with simplification of the hypothetical scenarios. Patients were asked to "imagine that you developed a sudden life-threatening illness with an uncertain chance of recovery". They were then asked, "Would you want the following procedures should they be needed?". These comprised intravenous fluids with medication, nasogastric tube, blood transfusions, intensive care, kidney dialysis, mechanical ventilatory support, and cardiopulmonary resuscitation. The question was repeated for a similar illness with a good chance of complete recovery. Desire for life-sustaining treatment was quantified by assigning a point for each treatment desired. Follow-up After standard treatment for major depression (including pharmacological agents in all cases and electroconvulsive therapy in two), clinical assessments, GDS and questionnaires were repeated. At least six weeks was allowed before follow-up as the elderly may take longer to respond to antidepressant treatment. Recovery (remission) from major depression was diagnosed clinically at follow-up interview if patients no longer met the DSM-IV criteria for major depression. Statistical analyses The paired t test was used to compare initial and follow-up GDS scores. The Wilcoxon signed-rank test (a non-parametric test) was used to compare initial and follow-up scores for the life-sustaining treatment questionnaire, scoring 1 for any increase and 0 for any decrease and ignoring patients who did not change. The binomial distribution was consulted, with parameters 0.5 and the number of people who changed preferences. Results Patients Of 25 consecutive patients referred with major depression to the Psychogeriatric Service during the study period, 22 were included in the study. Two failed to meet the inclusion criteria (one did not give informed consent and the other did not score sufficiently on the MMSE) and a third died before follow-up. Demographic characteristics of the 22 who completed the study are shown in Box 2. A notable feature was the very high proportion of women in the group (82%). All but four of the patients were new referrals to the service and 10 (46%) were reporting their first depressive disorder. None had a history of previous prolonged psychiatric hospitalisations or previous suicide attempts. Twelve patients (55%) were treated in their homes, eight (36%) were treated predominantly in an inpatient psychiatric unit and two (9%) predominantly in a medical ward. Follow-up after a mean of 88.5 days (SD, 26.3) showed no significant changes in the number of medical illnesses or in demographic variables. Intensity of depression Severity of the depression was classed as moderate for 16 patients (73%), severe for five (23%) and mild for one (5%). Mean GDS score for the group was 24.4 (SD, 3.3) and decreased significantly on follow-up to 16.7 (SD, 8.0) ( P < 0.0005). Individual GDS scores decreased ("improved") in 18 patients, and 12 of these (55% of the total number) were diagnosed clinically as recovered from major depression, according to DSM-IV criteria. Desire for life-sustaining treatment Changes in desire for life-sustaining treatment among those whose depression improved or recovered are shown in Box 3. There were significant increases in the number of interventions desired overall and for the "good prognosis" illness among both the 12 patients who recovered from depression (according to DSM-IV criteria for remission) and the 18 whose GDS scores improved (including six who were not classed as recovered by DSM-IV criteria). There were also increases in the number of interventions desired for the "uncertain prognosis" illness, but the change was significant only in the "recovered" group. Of the 12 patients who recovered, nine had had major depression of moderate severity. The number of interventions desired among these nine also increased significantly overall ( P = 0.05) and for the "good prognosis" illness ( P = 0.04), but not for the "uncertain prognosis" illness. Discussion We found that remission of major depression in the elderly was associated with a significant increase in acceptance of life-sustaining treatments. This increase occurred for both moderate and severe forms of major depression. However, our study has limitations which should caution against drawing firm general conclusions. The main limitations are the relatively small sample size and the possibility of a gender bias caused by the high proportion of female patients. We also do not know the correlation between responses to hypothetical questionnaires and actual decisions when confronted with an "end of life" situation. However, indicating preferences to hypothetical scenarios is similar to executing a "living will" (also, in effect, hypothetical). The hypothetical scenarios examined only acute, not chronic, illnesses, and the findings of this study cannot necessarily be generalised to patients who have coexisting life-threatening medical conditions. In addition, the word "uncertain", used to describe prognosis in one of the scenarios, may have been ambiguous, as all prognoses are uncertain. It was chosen to enable comparison with Lee and Ganzini's studies,13,14 but "poor" or "unfavourable" would have been less ambiguous. A further limitation involved the diagnostic assessments, which did not use structured diagnostic instruments. Our results vary from those of other longitudinal studies.13,14 Ganzini et al.14 found that recovery from major depression was associated with change in preferences for life-sustaining treatments only if the depression was severe. Lee and Ganzini13,14 concluded that, in major depression of mild to moderate severity, "patients should not be discouraged from completing advance treatment directives and that choices by these patients to limit treatment should be respected". However, our finding of a significant increase in the number of life-sustaining treatments desired by nine patients who recovered from major depression of moderate severity suggests that doctors should be cautious about adopting such an approach in the elderly with moderate major depression. Differences between the results of our study and those of Ganzini et al.14 may have been caused by differences between samples and methods. Their sample had a higher proportion of men (81% versus 18% in our study) and their questionnaire about life-sustaining treatments was more complex, possibly affecting subject comprehension and reducing the sensitivity of the instrument. They used DSM-III-R criteria to diagnose major depression, but the differences between these criteria and the DSM-IV criteria that we used are minor. However, neither set of criteria precisely defines the boundaries between mild, moderate and severe forms of major depression, allowing the possibility of diagnostic bias in classifying severity. Further, in analysing their data, Ganzini et al.14 defined a clinically evident increase in preference for medical therapy as an increase in desire for a mean of three or more of the 14 possible interventions. A different cut-off (e.g., two or more of the 14 interventions) would have led to different results. In addition, we allowed a longer time before follow-up than Ganzini et al.14 (mean, 88.5 days versus 24.5 days), and, although we found that degree of recovery measured by GDS was no greater, it is possible that preferences for life-sustaining treatments may take longer to improve than depression. Another possibility is that the longer follow-up period in our study allowed factors other than changes in level of depression to affect preferences. These studies highlight the potential for depression to influence patient desire for life-sustaining treatments, suggesting that clinicians would be wise to take into account patients' mental state when assessing refusal of life-sustaining treatments. If doubt exists about a patient's decisional capacity, or whether major depression is present, psychiatric consultation is indicated. In the absence of an advance directive (completed before the onset of depression), severely depressed patients' wishes to forgo life-sustaining treatments should not be respected until an attempt is made to treat the depression. In contrast, it appears reasonable to respect the wishes of mildly depressed patients, whereas in moderate major depression consensus is lacking. It would be prudent (until further studies clarify this question) to err on the side of preserving life and to treat moderate major depression of moderate severity before respecting a refusal of life-sustaining treatments. Furthermore, patients with moderate or severe major depression who are planning to write advance directives should be encouraged not to do so until their depression has been treated. In our patients whose depression "improved", we found no significant increase in number of life-sustaining treatments desired for an illness with uncertain prognosis. Although the lack of significance may have been due to the small sample size, it suggests that the prognosis of the illness should also be considered in clinical settings. When prognosis is so poor that treatment becomes futile, refusal of life-sustaining treatments should be respected regardless of the presence of major depression, in accord with the ethical and legal principle that doctors are not required to administer futile treatments. The ethical dilemma presented by a depressed patient with dubious decisional capacity who refuses life-sustaining treatments has received little attention. The physician who complies faces the prospect that the patient's decision was biased by the depression and would have reversed on recovery. The physician who does not comply faces the possibility that the request was authentic and the patient's life has been prolonged against his or her wishes. In resolving the dilemma, careful consideration should be given to the severity of the depression, the prognosis of the illness and whether treatment would be deemed futile, and also to any previous directives made by the patient when their decisional capacity was clearly intact. Acknowledgements We acknowledge the generous support of Dr R Russell (Psychogeriatrician, Royal North Shore Hospital), and Dr W Jenneke (Staff Specialist Psychiatrist, Hornsby Ku-ring-gai Hospital). References Greco P, Shulman K, Lavizzo-Mourey R. The patient self-determination Act and the future of advance directives. Ann Intern Med 1991; 115: 639-643. CCH Australia. Death with dignity. Australian Health and Medical Law Reporter. Sydney: CCH Australia Ltd, 1995: paragraph 22-360. Regier D, Hirschfeld R, Goodwin F. The NIMH depression awareness, recognition, treatment program. Am J Psychiatry 1988; 145: 1351-1357. NIH Consensus Development Panel on Depression in Late Life. Diagnosis and treatment of depression in late life. JAMA 1992; 268: 1018-1024. Eisenberg L. Treating depression and anxiety in primary care -- closing the gap between knowledge and practice. N Engl J Med 1992; 16: 1080-1084. Nielson C, Williams T. Depression in ambulatory medical patients: prevalence by self report questionnaire and recognition by nonpsychiatric physicians. Arch Gen Psychiatry 1980; 37: 999-1004. Rapp S, Walsh D, Parisi S. Detecting depression in elderly medical inpatients. J Consult Clin Psychol 1988; 56: 509-513. Appelbaum P, Grisso T. Assessing patients' capacities to consent to treatment. N Engl J Med 1988; 319: 1635-1638. Gutheil T, Bursztajn H. Clinicians' guidelines for assessing and presenting subtle forms of patient incompetence in legal settings. Am J Psychiatry 1986; 143: 1020-1023. Bursztajn HJ, Harding HP, Gutheil TG, Brodsky A. Beyond cognition: the role of disordered affective states in impairing competence to consent to treatment. Bull Am Acad Psychiatry Law 1991; 19: 383-388. Chochinov HM, Wilson KG, Enns M, et al. Desire for death in the terminally ill. Am J Psychiatry 1995; 152: 1185-1191. Wilkinson IM, Blackburn I. Cognitive style in depressed and recovered depressed patients. Br J Clin Psychol 1981; 20: 283-292. Lee M, Ganzini L. The effect of recovery from depression on preferences for life-sustaining therapy in older patients. J Gerontol 1994; 49: M15-M21. Ganzini L, Lee M, Heintz R, et al. The effect of depression treatment on elderly patients' preferences for life-sustaining medical therapy. Am J Psychiatry 1994; 151: 1631-1636. Salzman C. ECT and ethical psychiatry. Am J Psychiatry 1977; 134: 1006-1009. Weitzel W, Purtilo R. Aggressive treatment of geriatric depression: what limits on intervention? Psychiatr Opin 1979; 160: 9-14. Swartz C, Stewart C. Melancholia and orders to restrict resuscitation. Hosp Community Psychiatry 1991; 42: 189-191. Baile F, DiMaggio J, Schapira D. The request for assistance in dying. Cancer 1993; 72: 2786-2791. American Psychiatric Association diagnostic and statistical manual of mental disorders (DSM-IV). 4th ed. Washington DC: APA, 1994. Folstein M, Folstein S, McHugh P. "Mini-Mental State": a practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975; 12: 189-198. Yesavage J, Brink T, Rose T. Development and validation of a geriatric depression screening scale: a preliminary result. J Psychiatr Res 1983; 17: 37-49. Koenig HG, Meador KG, Cohen HF, Blazer DG. Self-rated depression scales and screening for major depression in older hospitalized patient with medical illness. J Am Geriatr Soc 1988; 36: 699-796. (Received 30 Oct 1995, accepted 3 Jun 1996) Authors' details Department of Mental Health, Hornsby Ku-Ring-Gai Hospital, Sydney, NSW. Stuart C Hooper, FRANZCP, Senior Psychiatric Registrar in Psychogeriatrics; now Consultant Psychiatrist, Sydney, NSW; Kevin J Vaughan, FRANZCP, Staff Specialist. Department of Academic Psychiatry, Royal North Shore Hospital, Sydney, NSW. Christopher C Tennant, FRANZCP, MD, Professor. Department of Psychology, Faculty of Arts and Social Sciences, University of Western Sydney, NSW. Janette M Perz, BA(Hons), Research Psychologist. No reprints will be available. Correspondence: Dr S C Hooper, 11 Clanalpine Street, Eastwood, NSW 2122.

Stuart C Hooper · Kevin J Vaughan · Christoper C Tennant · Janette M Perz

Prevalence of hepatitis G virus in Queensland blood donors Len D Moaven, Catherine A Hyland, Ian F Young, D Scott Bowden, Rhonda McCaw, Leigh Mison and Stephen A Locarnini MJA 1996; 165: 369 Subsequently cited in Moaven L. Should we be screening blood donors for hepatitis G virus? The case for screening. MJA 1998; 169: 373-374 and Wong et al. Should we be screening blood donors for hepatitis G virus? The case against screening. MJA 1998; 169: 375-377 Readers may print a single copy for personal use. No further reproduction or distribution of the articles in whole or in part should proceed without the permission of the publisher. For copyright permission, contact the Australasian Medical Publishing Company 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/>". Abstract - Introduction - Methods - Blood donors - Serological and biochemical testing - Reverse transcription PCR testing - Results - Discussion - Box - References - Authors' details - - ©MJA1996 Abstract Objective: To determine the prevalence of hepatitis G virus (HGV) carriage in Queensland blood donors. Design: Cross-sectional survey with retrospective longitudinal study of HGV-positive donors. Setting: Brisbane Red Cross Blood Bank, 1995. Subjects: 100 consecutive blood donors attending the Blood Bank on two days in October 1995 and 20 blood donors with a raised plasma alanine aminotransferase (ALT) level on their last donation. Outcome measures: Presence of HGV RNA by reverse transcription polymerase chain reaction (RT-PCR) in currently donated blood and in blood samples archived for up to 34 months. RT-PCR used two different reverse transcription methods and three different specific sets of primers and probes. Results: Five of the 120 blood donors were positive for HGV RNA by all RT-PCR methods (four of the 100 with normal ALT levels [4%] and one of the 20 with raised ALT levels [5%]). Retrospective testing of archived samples showed that four of these five had been persistently HGV RNA-positive for at least two years, while the fifth had been HGV RNA-negative on two donations before becoming HGV RNA-positive. No risk factors were identified for this donor. Conclusions: A relatively large number of Queensland blood donors (4%) are persistently HGV RNA-positive. MJA 1996; 165: 369-371 Introduction S ince the introduction of improved diagnostic assays for hepatitis C virus 1 and increased screening of blood donors, 2 the incidence of post-transfusion hepatitis has been reduced to low levels. However, there are still cases of post-transfusion hepatitis where an aetiological agent is not identified. 3 While searching for novel agents that may be associated with post-transfusion hepatitis, two independent groups recently discovered an RNA virus that causes acute and chronic infection in man. 4,5 The two isolates, designated hepatitis G virus (HGV) and GB virus C, respectively, belong to the same species, 6 and findings for one seem applicable to the other. Currently, the only reliable method for determining the prevalence of infection with these viruses is specific reverse transcription polymerase chain reaction (RT-PCR) assay to detect viral RNA. A serological assay to detect carriers of HGV is unlikely to be available in the immediate future, and much of the epidemiology of HGV will rely on RT-PCR-based assays. From the available RT-PCR data, it appears that HGV is transmitted parenterally, with increased prevalence in injecting drug users and patients who have received multiple transfusions. 4,7,8 HGV has a high prevalence in developed countries, with about 1.5% of volunteer blood donors from the United States and Europe being positive for HGV RNA. 4,8 It is not known whether these positive donors are transiently infected or chronic carriers of HGV. We examined the prevalence of HGV in Queensland blood donors. As most donors had plasma stored from previous donations, we were also able to collect unique longitudinal data. Methods Blood donors Consecutive blood donors at the Brisbane Red Cross Blood Bank over two days in October 1995 were asked to participate. The first 100 to sign a consent form agreeing to be tested for HGV were studied (compliance was 100%). These donors all had normal plasma alanine aminotransferase (ALT) levels. Blood was also collected over a three-month period from 20 donors with a raised ALT level on their most recent donation. Six of the 20 had raised ALT levels on two or more previous donations. All donors had signed the donor declaration form indicating they did not fall into an "at risk" group. All testing was performed with the approval of the Queensland Red Cross Blood Transfusion Service Ethics Committee. Serological and biochemical testing All donors were tested for antibodies to hepatitis C virus (HCV), HIV 1, HIV 2 and human T-cell lymphotropic virus (HTLV) and for hepatitis B surface antigen (HBsAg) (Abbott Diagnostics, Chicago, Illinois, USA). They were also screened with a syphilis antibody assay, Sysmic RPR (Diagast, Lille, France). Plasma ALT levels were measured by a microtitre tray assay. 9 Samples positive for HGV RNA were also tested for hepatitis B core specific antibody (Abbott Diagnostics). Reverse transcription PCR testing For RT-PCR testing, the recommendations of Kwok and Higuchi 10 were followed to prevent sample contamination. Blood samples were collected in acid-citrate-dextrose and plasma was stored at 2 70 o C within two hours of collection. RNA was subsequently extracted from these plasma samples with a guanidine thiocyanate and phenol-chloroform extraction procedure. 11 RT-PCR was performed with random hexamers and two sets of PCR primers from the 5 9 untranslated region and the region encoding the non-structural protein 5a of the HGV genome. These primers are part of an HGV primer and capture probe set commercially available from Boehringer Mannheim (Germany). In addition, RT-PCR was performed with primers from the region encoding the protein 5b, as described previously. 12 If a plasma sample was positive for HGV RNA, a separate aliquot of plasma was retrieved from frozen storage and RNA was extracted and tested with all three primer sets. Samples positive for HGV RNA were also tested for HCV RNA (Amplicor, Roche Diagnostic Systems, Branchberg, New Jersey, USA). Donors positive for HGV RNA were tested retrospectively by recovering archived frozen plasma samples from previous donations. These samples had been frozen within two hours of collection and stored at 2 30 o C. As controls, archived plasma samples were also recovered from donors negative for HGV RNA. Archived samples were coded and tested blind. In addition, four of the five donors found to be positive for HGV RNA returned subsequently to donate blood and were retested for HGV RNA. These retrospective and prospective samples were tested with the NS5b primers alone. Results Five of the 120 blood donors were HGV-positive by RT-PCR with all three of the primer sets. These positive donors comprised four of the 100 with normal ALT levels (4%; 95% confidence interval [CI], 1%-10%) and one of the 20 with raised ALT levels (5%; 95% CI, 0.1%-25%). This donor had not had raised ALT levels previously. The remaining 115 donors were negative by all three primer sets. Repeat PCR testing of a separate aliquot of each HGV-positive sample revealed complete concordance of results. The five HGV-positive donors were negative for hepatitis B core-specific antibody and HCV RNA. All donors were negative for antibodies to HCV, HIV 1, HIV 2 and HTLV, for HBsAg and for syphilis RPR antibody. Plasma samples from 22 previous donations given by the five HGV-positive donors were also tested for HGV RNA (see Box). Four donors were HGV-positive for all previous donations ( n = 18), made up to 22-34 months previously. The fifth, a 38-year-old man, was HGV-positive from August 1994, but two previous donations (May 1994 and November 1993) were HGV-negative. His ALT level was normal on each occasion and no risk factor or illness could be identified for this donor despite specific questioning about the period May 1994 to August 1994. Potential risk factors were noted for three of the other HGV carriers (see Box). All HGV-positive donors were well. Four returned for further testing, including the donor who appeared to acquire HGV infection in 1994; they remained HGV-positive on all occasions (up to six months after initial tests). The control archived plasma samples were tested blindly and found to be HGV-negative. Discussion This is the first study to describe the prevalence of HGV in an Australian blood donor population and to provide longitudinal data on a cohort of HGV-infected donors. Five of 120 blood donors (4.2%) were HGV-positive, a higher prevalence than seen in other developed countries. 4,8 All five appeared to be chronic carriers of HGV (assuming infection does not resolve after HGV carriage for longer than a year). It has yet to be determined why this infection is so prevalent in a group normally assumed to be at low risk for exposure to blood-borne viruses. Mother-to-baby (and early horizontal) transmission may account for the pattern of distribution (analogously to HBV in developing countries). 12 One donor had grown up in a developing country and another had potential risk factors (e.g., tattooing). Even so, one of the donors appeared to acquire HGV infection recently, without evident risk factors. This suggests that there are multiple modes of transmission. We found that only one of the HGV-positive donors had a raised ALT level, and on only one occasion. Although HGV infection is associated with an early and mild rise in results of liver function tests in some recipients of HGV-infected blood, 4 this is not unusual in acute viral infections. There is currently little evidence that HGV commonly causes chronic liver disease, but this does not preclude association with another disease. Indeed, we would argue that it is unlikely that a positive-strand RNA virus could cause chronic infection and not be associated with a disease syndrome. It is worth noting that with such a high prevalence of HGV in the general population the demonstration of a disease association will have to be particularly rigorous. The strategy we developed for HGV testing is time consuming and expensive, but, without a reliable serological or alternative virological marker, nucleic acid amplification by PCR remains the method of choice. An alternative to our PCR strategy could be use of a single reliable primer set, with automation of RNA extraction and RT-PCR. Blood donations may have to be tested for HGV RNA when suitable assays become available for mass screening, especially if evidence emerges that it causes significant disease. A National Health and Medical Research Council working party on novel viral hepatides has been set up in collaboration with the Red Cross Blood Transfusion Service to determine the significance of HGV in blood donations. In the meantime, it is important that recipients of donated blood continue to be advised of the potential complications of heterologous blood transfusion and, specifically, that they may acquire HGV infection. 2 However, currently the clinical significance of HGV infection is unclear and further clinical and epidemiological studies are needed on this new and emerging virus. References Aach RD, Stevens CE, Hollinger B, et al. Hepatitis C virus infection in post-transfusion hepatitis: an analysis with first and second generation assays. N Engl J Med 1991; 325: 1325-1329. Sloand EM, Pitt E, Klein HG. Safety of the blood supply. JAMA 1995; 274: 1368-1373. Alter HJ. Transfusion transmitted hepatitis C and non-A, non-B, non-C. Vox Sang 1994; 67: 19-24. Linnen J, Wages J, Zhen-Yong ZK, et al. Molecular cloning and disease association of hepatitis G virus: a transfusion-transmissible agent. Science 1996; 271: 505-508. Simons JN, Leary TP, Dawson JG, et al. Isolation of novel virus-like sequences associated with human hepatitis. Nat Med 1995; 1: 564-569. Zuckerman AJ. Alphabet of hepatitis viruses [editorial]. Lancet 1996; 347: 558-559. Bowden DS, Moaven LD, Locarnini SA. New hepatitis viruses: are there enough letters in the alphabet? Med J Aust 1996; 164: 87-89. Dawson GJ, Schlauder GG, Coleman P, et al. Prevalence and clinical significance of GBV-C [abstract]. Proceedings of the Ninth Triennial International Symposium on Viral Hepatitis and Liver Disease; 1996 April 21-25; Rome (Italy). Rome: CpA (Viale delle Medaglie d'Oro, 342100136), 33: 116. Hyland CA, Douglas R, Mazzocchi R, Young IF. Surrogate testing for non-A, non-B hepatitis in Queensland, Australia, an ALT microtitre tray method for screening blood donors. Pathology 1988; 20: 271-274. Kwok S, Higuchi R. Avoiding false positives with PCR. Nature 1989; 339: 237-238. Chemczynski P, Sacchi N. Single-step method of RNA isolation by acid guanidinium thiocyanate phenol-chloroform extraction. Anal Biochem 1987; 167: 156-159. Moaven LD, Tennakoon PS, Bowden DS, Locarnini SA. Mother-to-baby transmission of hepatitis G virus. Med J Aust 1996; 165: 84-85. (Received 15 May, accepted 9 Aug 1996) Authors details Victorian Infectious Diseases Reference Laboratory, Fairfield Hospital, Melbourne, VIC. Len D Moaven, FRCPA, Senior Registrar in Virology; D Scott Bowden, PhD, Senior Scientist (Molecular Virology); Rhonda McCaw, BSc, Scientist (Molecular Virology); Stephen A Locarnini, PhD, MRC(Path), Director. Red Cross Blood Transfusion Service, Brisbane, QLD. Catherine A Hyland, PhD, Scientist-in-Charge (Viral Serology); Ian F Young, FRCPA, Director; Leigh Mison, BSc, Scientist. Reprints: Dr S A Locarnini, Victorian Infectious Diseases Reference Laboratory, Fairfield Hospital, PO Box 65, Fairfield, VIC 3078. E-mail: stephenL AT hna.ffh.vic.gov.au <URL: http://www.mja.com.au/> © 1997 Medical Journal of Australia. We appreciate your comments.

Len D Moaven · Catherine A Hyland · Ian F Young · Rhonda McCaw · Leigh Mison · Stephen A Locarnini

An outbreak of Japanese encephalitis in the Torres Strait, Australia, 1995 Jeffrey N Hanna, Scott A Ritchie, Debra A Phillips, Jack Shield, M Clare Bailey, John S Mackenzie, Michael Poidinger, Bradley J McCall and Phillip J Mills MJA 1996; 165: 256-260 Readers may print a single copy for personal use. No further reproduction or distribution of the articles in whole or in part should proceed without the permission of the publisher. For copyright permission, contact the Australasian Medical Publishing Company 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/>". Abstract - Introduction - Methods - Results - Discussion - Acknowledgements - References - Authors' details - Register to be notified of new articles by email - - ©MJA1996 Abstract Objectives: To determine the distribution of virus infection during an outbreak of Japanese encephalitis (JE) in the Torres Strait, and to describe the environmental factors facilitating the outbreak. Design: Human and porcine serological surveys for JE virus activity throughout the Torres Strait, and mosquito and household surveys on the island of Badu. Setting: The island of Badu (where the clinical cases occurred) and the other islands of the Torres Strait, Australia, during April-May 1995. Results: The serological surveys identified recent JE virus infection among residents or domestic pigs on at least nine outer Torres Strait islands. A JE virus, confirmed by nucleotide sequencing, was isolated from two asymptomatic Badu residents. Virus isolations and mosquito surveys implicated Culex annulirostris as the major vector involved in the outbreak. There was prolific Cx. annulirostris breeding in a variety of water bodies close to and within the Badu community. Over half (53%) of the households kept pigs in pens, and many (63%) of the pigpens were situated near standing water; in 56% of these "wet" pigpens Cx. annulirostris was breeding. Conclusions: There was evidence of widespread JE virus activity throughout the outer islands of the Torres Strait. We suggest that migratory birds and/or wind-blown mosquitoes could have imported the virus into the Torres Strait from a focus of viral activity, possibly in Papua New Guinea, thereby initiating the outbreak. A combination of environmental factors, with large numbers of domestic pigs in close proximity to human dwellings and mosquito breeding sites, undoubtedly facilitated the outbreak on Badu. MJA 1996; 165: 256-260 Introduction Over a two-week period in March-April 1995, three cases of Japanese encephalitis (JE) occurred among residents of the island of Badu in the Torres Strait, Australia 1 ( Box 1). Although JE is widespread throughout Asia, 3 the Torres Strait outbreak is the first time that it has been recognised in Australia. J E is caused by a mosquito-borne flavivirus, and results in an acute illness characterised by headache, fever, convulsions, depressed level of consciousness and coma. It has a high case-fatality rate and there is a high prevalence of neurological sequelae in those who survive the acute illness. However, infection with the JE virus does not invariably cause disease; there is a high ratio of asymptomatic to symptomatic infections. 3 The JE virus is maintained in a natural cycle involving water birds such as egrets and herons. Pigs are very efficient amplifying hosts for the virus and therefore almost always contribute to outbreaks of JE in human populations. However, humans and other large verte brates, such as horses, are not efficient amplifying hosts, and are therefore "dead-end" hosts for the JE virus. 3 The outbreak was assumed to have been caused by a mosquito-borne virus, and Murray Valley encephalitis (MVE) virus (the major cause of arboviral encephalitis in Australia) was first suspected. Although the current status of flavivirus infections in the Torres Strait is uncertain, MVE and other Australian flaviviruses (Alfuy, dengue, Kokobera, Kunjin, Stratford and Sepik viruses) have been reported from either Cape York or Papua New Guinea (PNG). 4 Once laboratory findings confirmed the outbreak as JE, further investigations were undertaken. Methods Serological survey In April 1995, a serological survey was undertaken among residents of Badu. A non-random convenience sample was used. Human sera were screened for the presence of (i) flavivirus IgG and IgM by enzyme-linked immunosorbent assay (ELISA) 5 and haemagglutination inhibition assay, 6 and (ii) antibodies to JE, MVE and Kunjin viruses by plaque reduction neutralisation assay. 7 Sera testing positive for flavivirus IgM by ELISA were fractionated by ultracentrifugation over a sucrose gradient. IgM fractions were then tested for haemagglutination-inhibiting antibodies against JE, MVE and Kunjin viruses. 8 A diagnosis of a recent JE virus infection was made if there was a fourfold or greater rise in haemagglutination-inhibiting antibody titres in paired sera tested in parallel, or if JE virus IgM antibody and JE neutralising antibody were detected in serum at significantly higher titres than antibody to MVE and Kunjin viruses. Blood samples for serological studies were also collected from domestic animals, in particular pigs, on Badu. The animal sera were screened for the presence of total (i.e., IgM and IgG) antibody to JE, MVE and Kunjin viruses by haemagglutination inhibition assay. A portion of the sera positive for total JE antibody by haemagglutination inhibition assay was then tested by plaque reduction neutralisation assay. The presence of neutralising antibody at significantly higher titres to JE virus than to MVE and Kunjin viruses was considered evidence of exposure of an animal to JE virus. Isolation of virus Virus isolations were attempted from human sera obtained from residents of Badu: 100 m L of serum was inoculated onto confluent monolayers of C6/36 ( Aedes albopictus ) cells. 9 Viral growth was monitored by testing the culture super natant for haemagglutinating ability. Box 2 details the method used for identification of virus isolates. Mosquito survey Because the outbreak was initially suspected to be MVE, mosquito surveillance focused on the primary vector of MVE in Australia, Culex annulirostris . 4 Commencing on 7 April, larval and adult mosquito surveys were conducted within a 1-km radius of the centre of the Badu community. Mosquito larvae were sampled with a 350 mL dipper and adult mosquitoes were trapped with Centers for Disease Control light traps -- eight, each baited with 1 kg of dry ice (carbon dioxide) and 1-octen-3-ol (octenol; release rate 10 mg/h), 16 were run overnight at locations throughout the community on four occasions throughout April 1995. Household survey In May 1995, environmental health staff undertook a house-to-house survey to interview one adult resident of each household on Badu about household characteristics that may have contributed to the outbreak of JE. These included the presence of backyard pigs and horses and the state of repair of waste disposal (drainage and sewerage) systems. Survey of other communities To determine the extent of recent JE infection, blood was taken from convenience samples of people residing on 12 other outer islands, three inner islands and seven Cape York communities. Blood was also taken from pigs from eight other outer islands, three inner islands and two Cape York communities. Results Serological survey Two hundred and fifteen Badu residents were tested for JE infection: 35 people had serological evidence of recent JE infection (Box 3), and 21 of these had a fourfold or greater rise in JE-specific antibody titres in paired sera. All 11 pigs from Badu (tested by plaque reduction neutralisation assay) showed serological evidence of JE infection (Box 3). Most of the horses (7/10) and dogs (10/16) tested also showed serological evidence of JE infection, but not one of six chickens was positive. Isolation of virus JE virus was isolated from the sera of two Badu residents, both of whom remained asymptomatic. Nucleotide sequences obtained by reverse transcriptase-polymerase chain reaction (RT-PCR) amplification showed that the two viruses (designated FU and NO) were definitely strains of JE, with a 90% nucleotide homology with other strains of JE virus but with less than 70% homology with other flaviviruses. Both Badu isolates were very closely related, with 99% homology with each other. Detailed sequence comparison of the prM region of the viruses with that of other JE virus isolates showed that the Badu viruses were most closely related (92% homology) to other viruses of genotype III, 12,13 especially WTP-70/22 (from Malaysia) and B1065 (from southern Thailand) (Figure 1). The nucleotide sequences of JE-FU and JE-NO have been deposited in GenBank (the international computerised depository of genomic-sequence information), with accession numbers L43565 ( prM ) and L48968 ( NS5 ) for JE-FU, and L43566 ( prM ) and L48967 ( NS5 ) for JE-NO. Mosquito survey Numerous water bodies contained large numbers of Cx. annulirostris larvae. There were extensive swampy areas close to the community, and those contaminated by horse faeces had high larval densities ( >= 10 larvae/dip). A waterhole on the inland side of the community contained a large amount of grass clippings, rubbish and horse faeces; high larval densities were found along the margins. Many of the concrete-lined drains running through the community were overgrown with vegetation and contained mosquito-infested accumulations of water. High densities were also found in some of the defective household waste disposal systems (see below), and in pools in horse hoofprints. A total of 22 190 adult mosquitoes were trapped in the Badu community in April. Aedes kochi , Aedes culiciformis and Cx. annulirostris comprised 99% of the collection, with means of 364, 243 and 127 adult mosquitoes per trap, respectively. While the largest collections of Cx. annulirostris (up to 607 per trap) were taken from swampy areas near the community, collections of more than 100 per trap were taken within the community and close to houses. Eight JE viruses were isolated from 2871 Cx. annulirostris mosquitoes collected at Badu, but not from any other species (S A Ritchie, D A Phillips and A K Broom, unpublished data). Six of the isolates were from mosquitoes collected from within the community. Household survey An adult resident from 97 of the 102 houses in the community was interviewed for the household survey. Fifty-one (53%) of the households kept pigs; 35 (69%) of the pigpens were within 50 metres of the house. There were 179 domestic pigs in the community, an average of 3.5 (range, 1-11) per pig-rearing household. On inspection, 32 (63%) of the pigpens were situated either over or surrounded by standing water, and in 18 (56%) of these "wet" pigpens Cx. annulirostris was breeding. The householders of 15 (15%) of the houses owned a total of 18 horses; they were all kept in paddocks within 500 metres of the houses. The interviewees reported that 62 (64%) of the 97 houses had defective waste disposal systems: 49 of these houses had either waste water or raw sewage overflowing from septic tanks either into the house or into the yard, with 10% of the defective septic tanks containing Culex larvae. Survey of other communities A total of 1242 human serum samples were collected from the other communities. Twenty people from three other outer islands had serological evidence of recent JE virus infection. There was no evidence of JE infection in those tested from the inner islands or from the Cape York communities (Box 3). There was no evidence of any prior infection (i.e., only IgG JE antibodies) in any individual. Of the 182 pig sera collected from the other communities, 121 (66%) were tested by plaque reduction neutralisation assay. There was serological evidence of JE infection in pigs from all eight other outer islands, but no evidence of infection in pigs from the inner islands and Cape York Peninsula (Box 3). Discussion The human serological survey indicated that people from four outer islands had been recently infected with the JE virus. The porcine serological survey found evidence of JE activity in pigs from another five outer islands. Therefore, the two surveys indicated widespread, and presumably recent, JE virus activity in at least nine of the outer Torres Strait islands. There was no evidence of either human or porcine infection in the communities surveyed on the inner islands or on Cape York. The lack of evidence of prior infection in those tested suggests that the 1995 outbreak was the first incursion of the JE virus into the Torres Strait. Nucleotide sequencing studies clearly defined the virus strains isolated from the two people from Badu (and presumably the virus that infected people and pigs throughout the outer Torres Strait islands) as JE virus, and that the isolates were distinct from, but related to, virus strains that were circulating in southern Thailand, Malaysia and Indonesia between 1968 and 1983. 12 The virus differed, however, from a newly recognised genotype known to be circulating in Indonesia in 1980-1981. A strain (JKT-6468, Figure 1) of this latter genotype has been isolated from culicine mosquitoes collected from Flores, east of Bali, and is therefore the closest known isolate to Australia. 13 JE virus has not been isolated or reported as causing human disease in either PNG or Irian Jaya. Although no evidence of JE virus has been demonstrated conclusively in a number of serological surveys, there was possible JE seropositivity in a few single-serum specimens collected in 1956-1957 in the Western Province of PNG. 17 More recently, antibody to JE virus was detected by competitive ELISA in at least 23% of human sera collected in the Western Province in 1989 (R A Hall and J S Mackenzie, unpublished data), and 49% of porcine sera collected in the Western Province in 1995 were positive for neutralising antibody to JE virus (J Shield and R A Lunt, unpublished data). Thus, it seems possible that JE virus has become enzootic in parts of southwestern PNG. The northwestern islands (Boigu, Dauan and Saibai), because of their closeness to PNG, receive Papuan visitors virtually every day, raising the question of whether a viraemic visitor could have brought the JE virus across from PNG to the Torres Strait. However, humans, as "dead-end" hosts of the JE virus, have a low level of JE viraemia of short duration. 3 On the other hand, a viraemic pig imported from PNG, being a very efficient amplifying host, might have initiated the Torres Strait outbreak. However, the Australian Quarantine Act 1908 (Cwlth) prohibits the movement of live animals from PNG to the Torres Strait. The ban is actively enforced and respected; since the appointment of the first indigenous Quarantine Officers in 1982 they have not had occasion to seize even one illegally imported pig (P Stephen, Australian Quarantine and Inspection Service, personal communication). Therefore, we believe it most unlikely that an imported pig initiated the outbreak. Over 100 species of bird migrate annually between Australia and New Guinea, usually in a predictable seasonal pattern. 18 A further 63 species cross the Torres Strait at irregular intervals; these include numerous aquatic species, including wading birds. 18 The rufous night heron ( Nycticorax caledonicus ), for example, is a common nomad found throughout the Torres Strait; 18 it is closely related to the black-crowned night heron ( N. nycticorax ), a principal bird species implicated in the natural JE virus bird-mosquito cycle in Asia. 19 Experimental JE virus infection of the rufous night heron produces levels of viraemia that are quite adequate to infect the most efficient JE vector in Asia ( Culex tritaeniorhynchus ). 20 From December to April the prevailing wind in the region is from the northwest, raising the possibility that the dispersal of "wind-blown" mosquitoes could have carried the JE virus from New Guinea to the Torres Strait. Indeed, it has been reported that female Cx. annulirostris mosquitoes "can disperse at least 12 km and probably further"; 21 the northernmost outer islands are all less than 10 km from the PNG coastline (Box 1). We therefore suggest that two natural phenomena -- viraemic migratory birds and/or infectious wind-blown mosquitoes -- are plausible mechanisms for the importation of the JE virus from New Guinea to the Torres Strait, thereby initiating the outbreak. Cx. annulirostris was undoubtedly the major vector at Badu, and presumably at the other outer islands. Multiple JE virus isolations were made from Cx. annulirostris but not from any other mosquito species. The very small (fewer than one adult mosquito per trap) collections of Culex quinquefasciatus and Culex bitaeniorhynchus (both recognised as being either "marginal" or "occasional" vectors of the JE virus in Asia) 3 indicate that these species were of no importance in this outbreak. Cx. annulirostris was breeding in abundance in a variety of sites close to the community. However, the natural surface waters were extensive, and produced the most mosquitoes. The waterhole with run-off watercourses passing through the community meant that there was extensive mosquito breeding close to the $omestic pigs and to the people. The close proximity of the blocked drains to many of the houses also made them a significant risk. Defective waste disposal systems may also have contributed to the outbreak. The most striking combination of environmental factors contributing to the outbreak was the large number of domestic pigs adjacent to human dwellings and prolific mosquito breeding sites (Figure 2). The density of pigs, the abundance of the vector species and the human population density are all critical factors in determining the risk of human infection. 3,22 Horses, although "dead-end" hosts for the JE virus, 3 nevertheless contributed to the outbreak by providing numerous hoofprint breeding sites, nutrient (i.e., faeces) to the larvae and bloodmeals for female mosquitoes. 23 We need to determine the likelihood of future incursions of the JE virus into the Torres Strait. At the same time there is a clear need not only to reduce the mosquito breeding potential but also to improve the environmental conditions on the islands. Meanwhile, an inactivated JE vaccine has been offered to the inhabitants of the outer islands to confer protection while these risk assessment studies and risk reduction interventions are being implemented. 24 Acknowledgements Many people assisted with the investigation of the outbreak. We thank the staff of the Community Health Centers throughout the Torres Strait, Torres Strait Public Health Program, Tropical Public Health Unit, Australian Quarantine and Inspection Service, Queensland Department of Primary Industries, Laboratory of Microbiology and Pathology and the Australian Animal Health Laboratory. We are particularly grateful to Dr Ted Tsai (Division of Vector-Borne Viral Diseases, Centres for Disease Control and Prevention, USA) for his expert advice and support. References Hanna J, Ritchie S, Loewenthal M, et al. Probable Japanese encephalitis acquired in the Torres Strait. Commun Dis Intell 1995; 19: 206-208. The Torres Strait Health Workshop Working Party. Torres Strait Health Strategy. Thursday Island: Torres Strait Health Council, 1993. Vaughn DW, Hoke CH Jr. The epidemiology of Japanese encephalitis: prospects for prevention. Epidemiol Rev 1992; 14: 197-221. Mackenzie JS, Lindsay MD, Coelen RJ, et al. Arboviruses causing human disease in the Australasian zoogeographic region. Arch Virol 1994; 136: 447-467. Burke D, Nisalak A, Ussery M. Antibody capture immunoassay detection of Japanese encephalitis virus immunoglobulin M and G antibodies in cerebrospinal fluid. J Clin Microbiol 1982; 16: 1034-1042. Clarke DH, Cassals J. Techniques for haemagglutination and haemagglutination inhibition with arthropod borne viruses. Am J Trop Med Hyg 1958; 7: 561-573. Gorman BM, Leer JR, Filippich C, et al. Plaquing and neutralization of arboviruses in the PS-EK line of cells. Aust J Med Technol 1975; 6: 65-71. Field PR, Murphy AM. The role of specific IgM globulin estimations in the diagnosis of acquired rubella. Med J Aust 1972; 2: 1244-1248. Igarashi A. Isolation of a Singh's Aedes albopictus cell clone sensitive to dengue and chikungunya viruses. J Gen Virol 1978; 40: 531-544. Pierre V, Drout M-T, Deubel V. Identification of mosquito-borne flavivirus sequences using universal primers and reverse-transcriptase-polymerase chain reaction. Res Virol 1994; 145: 93-104. Sellner LN, Coelen RJ, Mackenzie JS. A one-tube, one manipulation RT-PCR reaction for detection of Ross River virus. J Virol Methods 1992; 40: 255-264. Chen W-R, Tesh RB, Rico-Hesse R. Genetic variation of Japanese encephalitis virus in nature. J Gen Virol 1990; 71: 2915-2922. Chen W-R, Rico-Hesse R, Tesh RB. A new genotype of Japanese encephalitis virus from Indonesia. Am J Trop Med Hyg 1992; 47: 61-69. Ni H, Barrett ADT. Nucleotide and deduced amino acid sequence of the structural protein genes of Japanese encephalitis viruses from different geographical locations. J Gen Virol 1995; 76: 401-407. Sumiyoshi H, Mori C, Fuke I, et al. Complete nucleotide sequence of the Japanese encephalitis virus genome RNA. Virology 1987; 161: 497-510. Ritchie SA, Kline DL. Comparison of CDC and EVS light traps baited with carbon dioxide and octenol for trapping mosquitoes in Brisbane, Queensland (Diptera: Culicidae). J Aust Entomol Soc 1995; 34: 215-218. Anderson SG, Price AVG, Nanadai-Koia, Slater K. Murray Valley encephalitis in Papua and New Guinea: II. Serological survey, 1956-1957. Med J Aust 1960; 2: 410-413. Draffan RDW, Garnett ST, Malone GJ. Birds of the Torres Strait: an annotated list and biogeographical analysis. The Emu 1983; 83: 207-234. Buescher EL, Scherer WF, McClure HE, et al. Ecologic studies of Japanese encephalitis virus in Japan. IV. Avian infection. Am J Trop Med Hyg 1959; 8: 678-688. Boyle DB, Dickerman RW, Marshall ID. Primary viraemia responses of herons to experimental infection with Murray Valley encephalitis, Kunjin and Japanese encephalitis viruses. Aust J Exp Biol Med Sci 1983; 61: 655-664. Bryan JH, O'Donnell MS, Berry G, Carvan T. Dispersal of adult female Culex annulirostris in Griffith, New South Wales, Australia: a further study. J Am Mosq Control Assoc 1992; 8: 398-403. Gingrich JB, Nisalak A, Latendresse JR, et al. Japanese encephalitis virus in Bangkok: factors influencing vector infections in three suburban communities. J Med Entomol 1992; 29: 436-444. Kay BH, Boreham PFL, Fanning ID. Host-feeding patterns of Culex annulirostris and other mosquitoes (Diptera: Culicidae) at Charleville, southwestern Queensland, Australia. J Med Entomol 1985; 22: 529-535. Hanna J, Barnett D, Ewald D. Vaccination against Japanese encephalitis in the Torres Strait. Commun Dis Intell 1996; 20: 188-190. (Received 9 Feb, accepted 27 May 1996) o Authors' details Tropical Public Health Unit, Queensland Health, Cairns, QLD. Jeffrey N Hanna, MPH, FAFPHM, Public Health Physician. Scott A Ritchie, PhD, Medical Entomologist. World Health Organization Collaborating Centre for Arbovirus Reference and Research, Laboratory of Microbiology and Pathology, Queensland Health, Brisbane, QLD. Debra A Phillips, BSc, MASM, Supervising Scientist. Queensland Department of Primary Industries, Cairns, QLD. Jack Shield, BVSc, Veterinary Officer. Torres Strait Public Health Program, Queensland Health, Thursday Island, QLD. M Clare Bailey, MAIEH, Environmental Health Officer. Department of Microbiology, The University of Queensland, Brisbane, QLD. John S Mackenzie, PhD, FASM, Professor of Microbiology. Michael Poidinger, PhD, Research Officer (NHMRC). Southern Zone Public Health Unit, Queensland Health, Upper Mount Gravatt, QLD. Bradley J McCall, MPH, FAFPHM, Public Health Physician. Queensland Health, Thursday Island, QLD. Phillip J Mills, Executive Officer, Torres Sector. Reprints: Dr J Hanna, Tropical Public Health Unit, Queensland Health, PO Box 1103, Cairns, QLD 4870. Email: troppubATcitec.qld.gov.au - Register to be notified of new articles by email - - To top of article - ©MJA1996 < URL: http://www.mja.com.au/> © 1996 Medical Journal of Australia.

Jeffery H Hanna · Scott A Ritchie · Debra A Phillip · Jack Shield · John S Mackenzie · Michael Poidinger · Bradley J McCall · Phillip J Mills

Error rates in Australian chemical pathology laboratories Mounira Khoury, Leslie Burnett and Mark A Mackay MJA 1996; 165: 128-130 For editorial comment, see Bryant Readers may print a single copy for personal use. No further reproduction or distribution of the articles in whole or in part should proceed without the permission of the publisher. For copyright permission, contact the Australasian Medical Publishing Company 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/>". Abstract - Introduction - Methods - Measurement criteria - Participants - Survey - Statistical analysis - Results - Participation - Transcription errors - Analytical performance - Combined error rates - Discussion - Acknowledgements - References - Authors' details Register to be notified of new articles by email - - ©MJA1996 Abstract Objective: To measure transcription and analytical errors made by Australian chemical pathology laboratories. Design: Retrospective data collection covering the period 1 November 1993 to 1 April 1994. Setting and participants: Fourteen pathology laboratories in five Australian States (seven in the public sector, and seven in the private sector). Main outcome measures: Error rates in transcribing information from request forms to computer record systems, and laboratory performance on chemical analysis. Results: Pathology laboratories had a transcription-error rate of up to 39% and an error rate of up to 26% for analytical results. The worst-performing laboratory had errors (of patient identification or results of analysis) in 46% of requests. The three best-performing laboratories achieved 85% error-free reporting, with one achieving 95%. Conclusions: Error rates in Australian pathology laboratories vary widely, but may be as high as 46% for all specimens in some laboratories. The types of errors reported were under the control of the laboratory, and would affect the accuracy of reported pathology test results, with potential adverse outcomes for patient care and inefficient use of health-care resources. There is a need to establish broader quality assurance programs and performance requirements to reduce these types of error. MJA 1996; 165: 128-130 Introduction T he Quality in Australian Health Care Study 1 described an unacceptable rate of preventable adverse events involving patients which occur in Australian hospitals. Errors introduced into the process of requesting pathology tests can adversely affect patient management and will detract from quality of care and services. 2 At best, a sample may need to be collected again to repeat the analysis, inconveniencing the patient, delaying treatment and increasing the cost. At worst, a doctor may act on incorrect results and the patient may be treated inappropriately. In its most common diagnostic setting, the process of requesting pathology tests seems simple: the doctor writes a request, the pathology laboratory interprets the request, transcribes information from the request form to its computer system, performs the analysis, and the pathologist produces a printed report for return to the doctor. This process assumes that the patient has been correctly identified, an appropriate specimen has been obtained, analysis has been correctly performed, and the results of analysis have been communicated to the original requesting practitioner. We recently examined the accuracy of one part of the process of obtaining pathology test results in a hospital environment and found that the address of the referring doctor had been incorrectly recorded on 3% of all printed reports issued, delaying or preventing delivery of the final pathology report. 3 If this rate of error were typical for the many other steps of the requesting process for pathology tests, then the total error rate in obtaining results would be extremely high. To address this question, we surveyed clinical pathology laboratories from both public and private sectors of five Australian States to determine whether error rates in pathology are as widespread and as clinically significant as our earlier study suggested. Methods Measurement criteria We identified potential indicators using national Quality Award criteria for benchmarking indicators, 4 focusing on factors affecting medical outcomes which arose from the pathology laboratory. In this paper, we report on transcription errors and analytical performance in chemical pathology quality assurance programs. Details of these criteria and how they were measured are shown in Box 1. We did not examine how laboratories functioned to deliver these outcomes. Participants We approached 18 large National Association of Testing Authorities-registered laboratories, well recognised in the industry, representing both public and private sectors, in five Australian States (Queensland, New South Wales, Victoria, Tasmania and South Australia). Survey Survey forms, which included detailed instructions and operational defini tions of the information required, were completed by the laboratories between 1 May and 31 July 1994. One hundred hand-written pathology request forms dated between 1 November 1993 and 1 April 1994 were randomly selected by the participating laboratories. We excluded request forms which also contained requests for blood transfusion or tissue pathology because some laboratories adopt more stringent identification criteria for these types of specimens. Laboratories scored the number of transcription errors (defined as any instances where the data on individual request forms were not identical to the data entered into the laboratory's computer system 5 ). Chemical pathology quality assurance (QA) data were from one cycle of analysis within the QA program (there are two or three cycles annually, depending on the program) during 1993 and 1994. Laboratories scored the total number of QA samples analysed, and the proportion of results of these analyses that lay outside the allowable limits of error of the QA program. All data, as well as a summary of the relative performance of all other participants (each assigned a number so that they were not identifiable), were returned to each laboratory for validation before final analysis. Statistical analysis Data analysis was based on the binomial distribution. 5 Calculation of combined error rates was based on the product of constituent error-free components, with the standard deviation calculated from the sum of variances of individual component errors. Results Participation Fourteen laboratories (seven public and seven private, numbered 1-14 to preserve anonymity) completed the entire benchmarking study; all five States were represented. Transcription errors Box 2 shows our findings for particular types of transcription errors. Most laboratories lay within a tight distribution clustered around median error rates of 1% to 3%. For each transcription error listed, a single outlier -- a different laboratory in each case -- performed particularly badly. The Figure shows that, while two laboratories were able to achieve overall transcription-error rates of less than 3%, four had error rates more than two standard deviations above the mean error rate (13%) of participating laboratories ( P < 0.05). The laboratory with the poorest performance reported a transcription-error rate of 39%. Analytical performance Laboratories recorded errors outside the allowable limits of error in up to 26% of analytical results. As shown in the Figure, four laboratories reported unacceptable performance more than two standard deviations above the mean rate (11.4%) of participating laboratories ( P < 0.05). Combined error rates By matching the transcription-error rate with analytical performance, we were able to measure the proportion of all pathology requests that would have been free of either type of error (Figure). Three laboratories showed superior performance compared with peers, achieving 85% error-free reporting, with one achieving 95% error-free reporting (a total error rate of less than 5%). These three laboratories represented both the public and private sectors, were all from different States, and each used a different computer system and clinical chemistry analyser. The two laboratories which showed the poorest performance reported transcription errors or unacceptable ana lytical performance in more than a third of all requests. They were both hospital laboratories from the same State. Their high total error rates were particularly influenced by transcription errors. Discussion This pilot survey of laboratory performance is the first of its kind to have been done in Australia. Previous studies have reported particular types of laboratory errors (e.g., patient ward location or address) in 0.08% 6 to 3% 3,6,7 of requests. We have confirmed that median error rates for individual types of errors in routine pathology requests are of the order of 3% 3 (see Box 2), and overall error rates can be as high as 46%. While some of the laboratories operated with few or no errors, eight of the fourteen reported significantly inferior performance (Figure). Further, our findings indicate that some laboratories may be making transcription errors in up to 39% of all request forms, and can have error rates of 9% in patient identification data and 17% in data such as sex and date-of-birth. Such error rates should be a cause of clinical concern. Risk management in pathology laboratories recognises the relationship between proper identification of specimens and the prevention of adverse patient-care-related incidents. 2 Our study was not designed to trace particular errors to final clinical outcomes, or to distinguish between errors on the basis of severity. However, it would be surprising if transcription-error rates of the magnitude we encountered did not contribute to adverse clinical outcomes. Are the errors caused by the requesting doctor, or by the laboratory? Several laboratories used direct electronic capture of patient data from a database, which should minimise errors in patient identification data arising from illegible handwriting. These laboratories did not have lower rates of transcription error for patient identification data than those relying solely on handwritten pathology requests. Indeed, the two laboratories with the highest overall rates of transcription error both obtained their patient identification data from a central database. This leads us to conclude that poor handwriting by the requesting doctor was not the major factor in high rates of transcription errors. Errors in pathological tests come at a considerable cost to the community. Expenditure on pathology investigations in Australia amounts to some $1.4 billion -- approximately 4% of total health services expenditure (Australian Association of Pathology Practices, personal communication). Incorrect tests being performed in up to 15% of cases wastes medical resources. Requesting doctors being incorrectly identified on up to 17% of requests will result in laboratories incurring unnecessary delivery costs and doctors incurring additional costs, as they will need to obtain duplicate copies of reports not delivered, and will request repeat testing for results unable to be located. We believe that most of these errors could have been prevented by the laboratories. The three best-performing laboratories in our study used different equipment and computer systems, and operated in both public and private sector environments. We therefore suggest that superior overall performance can be achieved by controlling error rates at each step in the process, and that this can be done in diverse ways. All recognised medical testing laboratories within Australia are required to be accredited, and to participate in QA programs for analytical quality. However, there are no minimum standards of performance which laboratories are required to maintain within these QA programs. There are also no Australian QA programs which monitor non-analytical aspects of pathology laboratories, such as transcription errors. We recommend that QA programs should be introduced for a range of clinically important areas of laboratory performance, such as request transcription accuracy, transport and analysis times for key analytes (e.g., potassium), rapid reporting of critical patient data, and client satisfaction, in addition to existing programs for analytical quality. We also recommend that summaries of actual and achievable performance in these QA programs be published. Without laboratory participation and satisfactory performance in such programs, doctors cannot be confident that test results are correct. Indeed, they cannot even be confident that the result was obtained for the correct patient. Lower error rates in pathology tests should lead to improved patient outcomes and less wastage of health resources. Acknowledgements We thank the staff of the participating laboratories for their assistance with this study. We are grateful to Elizabeth Benson, Jon Currie, Jeremy Chapman, Doug Chesher and Ruth Pojer for their critical comments on the manuscript. References Wilson RM, Runciman WB, Gibberd RW, et al. The Quality in Australian Health Care Study. Med J Aust 1995; 163: 458-471. Lord JT. Risk management in pathology and laboratory medicine. Arch Pathol Lab Med 1990; 114: 1164-1167. Banning J, Brown J, Hooper L, et al. Reduction of errors in laboratory test reports using continuous quality improvement (CQI) techniques. Clin Lab Management Rev 1993; 7: 424-437. Australian Quality Awards Foundation. Assessment criteria and application guidelines. Sydney: Australian Quality Council, 1994. Grant EL, Leavenworth RS. Statistical quality control. 6th ed. Singapore: McGraw-Hill. 1988. Burnett L, Banning J. Reduction of errors in laboratory test reports: comparison of continuous quality improvement techniques with laboratory information system techniques. In: Parkany M, editor. Quality assurance and TQM for analytical laboratories. Cambridge, UK: The Royal Society of Chemistry, 1995: 97-101. Chambers AM, Elder J, O'Reilly DStJ. Blunder-rate in a clinical biochemistry service. Ann Clin Biochem 1986; 23: 470-473. (Received 20 Nov 1995, accepted 22 May 1996) Authors' details Faculty of Clinical Chemistry, School of Life Sciences, University of Technology (Gore Hill Campus), Sydney, NSW. Mounira Khoury, MSc(Clin Biochem), Graduate Student. Institute of Clinical Pathology and Medical Research, Westmead Hospital, Sydney, NSW. Leslie Burnett, PhD, FRCPA, Director of Clinical Chemistry. Mark A Mackay, MSc, Hospital Scientist and Quality Facilitator. Reprints: Associate Professor Leslie Burnett, Assistant Director, Institute of Clinical Pathology and Medical Research, Westmead Hospital, Westmead, NSW 2145. < URL: http://www.mja.com.au/> © 1996 Medical Journal of Australia.

Mounira Khoury · Leslie Burnett · Mark A Mackay

Flexible sigmoidoscopy screening for colorectal cancer in average-risk subjects: a community-based pilot project John K Olynyk, Sina Aquilia, David R Fletcher and Jim A Dickinson ©MJA1996; 165: 74-76. This article has been cited in Sladden MJ, Ward JE. Australian general practitioners' views and use of colorectal cancer screening tests. MJA 1999; 170: 110-113 Readers may print a single copy for personal use. No further reproduction or distribution of the articles in whole or in part should proceed without the permission of the publisher. For copyright permission, contact the Australasian Medical Publishing Company Abstract - Introduction - Methods - Subjects - Data management - Acceptability survey - Results - Subject recruitment - Screening results - Reasons for non-compliance with screening - Discussion - Acknowledgements - References - Author's Details - Register to be notified of new articles by email - - To top of article - ©MJA; 1996 Abstract Objective: To test a pilot screening program for colorectal cancer. Design: Subjects, chosen at random and recruited by mail, were examined by flexible sigmoidoscopy. Participants and setting: Normal-risk, asymptomatic men and women aged 55-59 years recruited from the community, July to December, 1995. Main outcome measures: Number of polyps detected and cancers diagnosed, and compliance with screening. Results: Letters of invitation were sent to 3500 subjects; of these, 2881 were eligible for inclusion in the study and 342 (12%) consented to participate. A further 3.5% of non-compliant subjects attended the screening program after a telephone survey assessing reasons for non-attendance. Common reasons for non-attendance were a lack of interest (30%) or a lack of time, mainly due to work commitments (28%). A third of subjects had polyps and 46% of these were adenomas. Three subjects were found to have adenocarcinoma: in two the cancer was confined to a polyp and treated with polypectomy, and one subject underwent anterior resection (overall prevalence of cancer, 0.9%). The median depth of insertion achieved with flexible sigmoidoscopy was 55 cm (range, 25-100 cm). Median pain level (on a scale of 0 = no pain to 10 = worst pain imaginable) was 2 (range, 0-8.5), and 99% of the subjects would have the test again if required. Conclusions: Flexible sigmoidoscopy was well tolerated and had an acceptable detection rate of adenomatous polyps and early cancer. Subject compliance emerged as a major issue which requires further evaluation to maximise participation in future programs. MJA 1996; 165: 74-76 Introduction C olorectal cancer is the commonest cancer affecting both sexes and isnow the second-commonest cause of cancer-related death in both men and women. 1 Several studies have shown that screening asymptomatic populations may reduce mortality from colon cancer by up to 30%, 2-4 and recently the World Health Organization recommended screening of asymptomatic subjects beginning at age 50. 5 However, in Australia there are no uniformly agreed methods for screening asymptomatic subjects for colorectal cancer. 6 Factors such as screening efficacy, cost, subject compliance and strategies employed to evaluate positive results all influence the choice of screening test. Faecal occult blood testing for colo rectal cancer is relatively cheap, but is limited by a high rate of false positive results and poor sensitivity. 7-9 A recent report also suggests that up to half the mortality reduction observed with faecal occult blood screening may be due to chance selection for colonoscopy. 10 Flexible sigmoidoscopy has been proposed as an alternative screening test. It is more expensive but would prevent more cancer deaths than faecal occult blood testing alone. 8,11 Apart from determining the best screening methods, rates of compliance with screening programs range widely (8%-80%). 8,12 These issues, together with implications for health service resource allocation, have prompted recommendations for pilot programs to determine the efficacy of screening and compliance rates before more generalised screening is introduced. 6,8,13 The Western Australian Department of Health provided funds for such a pilot project at Fremantle Hospital. Thus, we were able to conduct a pilot community-based flexible sigmoid oscopy screening program, between July and December 1995, for colorectal cancer in normal-risk asymptomatic individuals, and in this setting determine (i) yield of screening, and (ii) subject compliance with screening and factors which influence compliance. Methods A flexible sigmoidoscopy facility dedicated to colorectal cancer screening was established at Fremantle Hospital. All procedures were performed on an outpatient basis after informed written consent. No sedation was used. After administration of a phosphate enema, procedures were performed either by a qualified gastroenterologist, or a general practitioner who was undergoing supervised training in flexible sigmoidoscopy. An Olympus TI100 colonoscope (Olympus Optical, Japan) was used, allowing insertion up to 100 cm. Biopsies were taken from polyps seen during the procedure. Subjects with biopsy-proven adenomas were advised to have a follow-up colonoscopy. A computer-generated report was issued to all subjects after the procedure, with instructions to return to their general practitioner for ongoing care. The study was approved by the Ethics Committee of Fremantle Hospital. Subjects We recruited male and female subjects aged 55 to 59 years of age using a computerised database derived from the Western Australian Electoral Commission. Letters were sent to 3500 randomly selected subjects inviting their participation. Exclusion criteria (apart from age less than 55 or greater than 59 years) were: symptoms of recent alteration in bowel habit, constipation, diarrhoea, or passage of blood with bowel motions; previous history of colonic polyps or colorectal cancer; and family history of colorectal cancer. A telephone survey of 200 subjects who did not respond to the initial letter showed that the reason for non-attendance in 14% of these subjects was that they met the exclusion criteria. Data management All data were recorded on a customised database from which reports were generated. We recorded basic demographic data, presence or absence of exclusion criteria, operator information, date of procedure, adequacy of bowel preparation, insertion depth, polyp data (number, size, pedunculated or sessile, biopsy report and date), and follow-up advice. Acceptability survey After the procedure, 77 consecutive subjects filled in a brief questionnaire, which included a standard 10-cm linear analogue pain scale. They marked a 10-cm line (labelled "0 = no pain" and "10 = worst pain imaginable") at a point which corresponded to their level of pain. Results Subject recruitment One hundred and fifty letters were returned, indicating that the subject was no longer resident at the mailing address. As 14% of non-attending subjects were found by the telephone survey to meet exclusion criteria, we estimate there were 2881 eligible subjects. Of those, 342 (12%) consented to participate in the study. There were 200 men and 142 women participants; 95% were born in Australia. Screening results Findings at flexible sigmoidoscopy are summarised in Box 1. The median depth of insertion achieved was 55 cm (range, 25-100 cm; 70% of subjects had at least 50 cm of bowel examined by flexible sigmoidoscopy. Thirty-five subjects (10%) were incompletely prepared with one enema and required a repeat bowel preparation before the procedure. Thirty-five per cent of subjects had polyps -- 46% of these polyps were adenomas, the remainder being hyperplastic or metaplastic polyps. The median adenoma size was 4 mm (range, 1-60 mm). Nineteen of the adenomatous polyps (5.6%) were at least 1 cm in size. One subject with a 6-cm sessile villous adenoma underwent anterior resection. Three subjects were found to have adenocarcinoma -- in two the cancer was confined to a polyp and treated with polypectomy, while the third required anterior resection for a cancer detected at follow-up colonoscopy. The overall prevalence of cancer in this group of subjects was 0.9%. The median pain score attributed to the procedure was 2 (range, 0-8.5). Seventy-six (99%) of these subjects would have the test again if required. Most subjects found the procedure interesting (they were able to watch on a video screen) and worthwhile. Reasons for non-compliance with screening The commonest reasons given for non-compliance were a lack of interest (30%) or a lack of time, mainly due to work commitments (28%) (Box 2). Interestingly, 16% of subjects who did not attend stated that they had either discussed the issue of bowel cancer screening to their satisfaction with their local general practitioner or had been screened for bowel cancer. Seventeen (8%) of the 200 subjects who did not attend as a result of the initial mail-out indicated that they would like to participate as a direct result of the telephone interview and were offered appointments; seven (3.5%) of these subjects have since attended for screening. Discussion The cancers detected by flexible sigmoidoscopy in our study were early cancers, but our three cases represent a detection rate of 8.8 cancers per 1000 asymptomatic subjects aged 55-59. Western Australian data for this age group predict an annual incidence of about 1 per 1000. 1 Our yield of polyps at least 1 cm in size (5.6%) is consistent with the previously reported prevalence of polyps in persons aged 55-59 14 and in asymptomatic Australian men of mean age 66 (3.8%). 15 However, the depth of insertion and number of polyps found in our study are both greater than those reported in several recent studies of flexible sigmoidoscopy: their average depth of insertion ranged from 30 to 50 cm 16-19 and polyp yield ranged from 1% to 4%. 18,19 There is little doubt that population response to screening will be a major determinant of the utility of flexible sigmoidoscopy as a screening tool for colo rectal cancer. The reasons given for non-compliance are similar to those reported for non-attendance in faecal occult blood screening programs. 20 Although the initial response in the pilot program was 12%, the survey of non-compliant subjects demonstrated a significant opportunity for improvement. An additional 3.5% of subjects could immediately be given an appointment at the time of the telephone survey of non-compliant subjects, bringing the compliance rate to 15.5%. Of the remaining reasons determined for non-compliance, most could be addressed by better education of subjects, and rescheduling of screening times to suit the work practices of potential candidates. It remains to be seen how much this would actually improve screening rates. Sixteen per cent of non-compliant subjects reported that they had either been screened for colorectal cancer or had discussed the issue to their satisfaction with their local general practitioner. This may reflect the growing public and general practitioner awareness of colon cancer screening; this has recently been reported from South Australia. 21 There is no doubt that the highest possible compliance should be aimed for, but in a society where currently only 41% of subjects with rectal bleeding seek medical attention 22 it would seem that much education needs to occur to facilitate this process. We have reported the first Australian pilot study of flexible sigmoidoscopy screening of asymptomatic subjects aged 55-59 years in which a high yield of adenomatous polyps and early cancers was found. The procedure was well tolerated. Subject compliance emerged as a major issue which requires further evaluation in order to maximise participation in future programs. This study strongly supports a more broad-scale evaluation of flexible sigmoidoscopy screening of asymptomatic subjects for colorectal cancer and polyps. Acknowledgements We wish to thank Dr Digby Cullen, Mr Graham Cullingford and the Gastroenterology Departments of Fremantle Hospital, Sir Charles Gairdner Hospital and Royal Perth Hospital; Dr Andrew Penman, formerly Chief Health Officer of the Health Department of Western Australia, and the Fremantle Division of General Practice, for their support of this study. References Dobson SK, Penman AG, and 82 others. Clinical health goals and targets for Western Australia, Vol. 2. First report of the Western Australian Task Force on State Health Goals and Targets. Perth: Health Department of Western Australia, July 1994. Mandel JS, Bond JH, Church TH, et al. Reducing mortality from colorectal cancer by screening for faecal occult blood. N Engl J Med 1993; 328: 1365-1371. Selby JV, Friedman GD, Quesenberry CP Jr, et al. A case-control study of screening sigmoidoscopy and mortality from colorectal cancer. N Engl J Med 1992; 326: 653-657 . Newcomb PA, Norfleet RG, Storer BE, et al. Screening sigmoidoscopy and colorectal cancer mortality. J Natl Cancer Inst 1992; 84: 1572-1575. Winawer SJ, St John DJ, Bond JH, et al. Prevention of colorectal cancer: guidelines based on new data. Bull World Health Organ 1995; 73: 7-10. Goulston K, St John J, Bokey L, et al. Guidelines for early detection, screening and surveillance for colo rectal cancer. 2nd edition. Sydney: Australian Gastroenterology Institute and Gastroenterological Society of Australia, 1994. Austoker J. Screening for colorectal cancer. BMJ 1994; 309: 382-386. Lieberman D. Screening/early detection model for colorectal cancer. Cancer 1994; 74: 2023-2027. Solomon MJ, McLeod RS and Canadian task force on periodic health examination. Can Med Assoc J 1994; 150: 1961-1970. Lang CA, Ransohoff DF. Faecal occult blood screening for colorectal cancer: is mortality reduced by chance selection for screening colonoscopy? JAMA 1994; 271: 1011-1013. Lieberman DA. Cost-effectiveness model for colon cancer screening. Gastroenterology 1995; 109: 1781-1790. Blalock SJ, De-Vellis BM, Sandler RS. Participation in faecal occult blood screening: a critical review. Prev Med 1987; 16: 9-18. St John J. Screening for colorectal cancer: "On your marks. . ." Med J Aust 1994; 160: 596-597. Arminski TC, McLean DW. Incidence and distribution of adenomatous polyps of the colon and rectum based on 1000 autopsy examinations. Dis Col Rectum 1964; 7: 249-261. Chapuis PH, Goulston KJ, Dent OF, Tait AD. Predictive value of rectal bleeding in screening for rectal and sigmoid polyps. BMJ 1985; 290: 1546-1548. Krevsky B, Fisher RS. Yield of rescreening for colonic polyps using flexible sigmoidoscopy. Am J Gastroenterol 1994; 89: 1165-1168. Maule WF. Screening for colorectal cancer by nurse endoscopists. N Engl J Med 1994; 330: 183-187. Renneker M, Saner H. Low-cost flexible sigmoid oscopy screening: a community demonstration and education project. J Cancer Educ 1995; 10: 25-30. Sakamoto MS, Hara JH, Schlumpberger JM. Screening flexible sigmoidoscopy in a low-risk, highly screened population. J Fam Pract 1994; 38: 245-248. Lindholm E, Berglund B, Haglind E, et al. Factors associated with participation in screening for colorectal cancer with faecal occult blood testing. Scand J Gastroenterol 1995; 30: 171-176. Weller D, Hiller J, Beilby J, et al. Screening for colorectal cancer: knowledge, attitudes and practices o f South Australian GPs. Med J Aust 1994; 160: 620-624. Crossland A, Jones R. Rectal bleeding: prevalence and consultation behaviour. BMJ 1995; 311: 486-488. (Received 6 Feb, accepted 22 Apr 1996) For editorial comment, see Solomon, page 68; see also Macrae, page 102. Authors' details University Department of Medicine and Department of Gastroenterology, Fremantle Hospital, Fremantle, WA. John K Olynyk , FRACP, MD, Senior Lecturer and Head of Department of Gastroenterology. Sina Aquilia , BA(Psych), Research Assistant. University Department of Surgery and Department of Gastroenterology, Fremantle Hospital, Fremantle, WA. David R Fletcher , FRACS, MD, Professor of Surgery, Head of Department of Surgery. University Department of General Practice, Fremantle Hospital, Fremantle, WA. Jim A Dickinson , FRACGP, PhD, Professor of General Practice. Reprints: Dr J Olynyk, University Department of Medicine, Fremantle Hospital, PO Box 480, Fremantle, WA 6160. E-mail: jolynyk AT uniwa.uwa.edu.au < URL: http://www.mja.com.au/> © 1996 Medical Journal of Australia.

John K Olynyk · Sina Aquilia · David R Fletcher · Jim A Dickinson

Prevalence of latex allergy in a dental school Constance H Katelaris, Richard P Widmer and Ross M Lazarus Readers may print a single copy for personal use. No further reproduction or distribution of the articles should proceed without the permission of the publisher. For permission, contact the Australasian Medical Publishing Company 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/>". Abstract - Introduction - Methods - Results - Discussion - Acknowledgements - References - Authors' details - ©MJA1997 Abstract Objective: To determine the prevalence of latex allergy in dental workers. Design: Questionnaire survey of staff of a dental school. Setting: The Westmead Dental School, a large dental facility in western Sydney. Participants: 230 staff members of the Westmead Dental School (consisting of general and specialist dentists, chairside assistants and registered nurses, laboratory technicians, dental therapists and hygienists) received questionnaires. Main outcome measures: The prevalence of latex allergy, defined by prompt onset of hand urticaria with or without generalised symptoms, and the prevalence of hand dermatitis and other glove-related symptoms. Also, the relationship between latex allergy and associated atopic status. Results: 177 staff (77%) responded by the set collection date; 33% reported symptoms related to wearing gloves and 22% satisfied the criteria for glove dermatitis. Sixteen respondents (9%) reported characteristics suggestive of latex- glove allergy. Conclusions: Confirmation of the 9% prevalence of latex allergy among dental workers will require further studies incorporating an objective measure of IgE-mediated hypersensitivity. MJA 1996; 164: 711-714 Introduction Adverse reactions to rubber products have been recognised for many years, and irritant reactions and cell-mediated hypersensitivity reactions such as contact dermatitis are well described. The latter are thought to be caused by one or more low-molecular-weight chemical compounds added in the rubber-making process.1 It is now recognised that latex allergy is a particular problem for certain "at risk" groups, including health care workers, patients with spina bifida or other spinal cord abnormalities and patients who have had multiple operations. In addition, atopic individuals are more likely to develop IgE antibodies to latex protein with increased exposure.2 The clinical manifestations of type I hypersensitivity reactions to latex protein are wide ranging, from contact urticaria to life-threatening anaphylaxis and death. There is now little doubt that at least some episodes of intraoperative anaphylaxis have occurred because the patient has been sensitised by latex and reacts after mucosal contact with the operating staff's gloves. This has led to some institutions providing latex-free operating suites. No studies of latex allergy in Australia have been published. In this study, we examined the prevalence of latex glove-associated symptoms in the staff of a dental institution. Dental workers are an excellent group for studying the question of latex glove-related symptoms as they may wear gloves for 8-10 hours daily, 4-5 days a week, giving them a much greater degree of exposure to latex than most other health care workers. Methods All 230 staff of the Westmead Dental School, a large dental facility in western Sydney, New South Wales, were surveyed by anonymous questionnaire distributed during Feburary 1995, with a three-week collection time allocated. Personnel surveyed included general and specialist dentists, chairside assistants, dental technicians and dental hygienists. Questionnaire details The questionnaire was divided into four sections. Section A dealt with dental practice and contained questions on demographic characteristics, including age, gender, job description and years in present occupation. Section B sought information about glove use and working habits, including the number of patients seen per day, duration that each pair of gloves was worn, information about hand washing and glove changing and the type of gloves worn. Section C sought information about symptoms or problems on contact with surgical gloves. Respondents who acknowledged glove-related problems then answered detailed questions about symptoms, the time course of symptom development, non-hand-related symptoms such as rhinoconjunctivitis, asthma, angioedema, hypotension, symptoms on contact with other dipped-rubber products such as balloons, domestic gloves and condoms, and specific interventions sought by the respondent because of the symptoms. Section D was directed at all respondents and dealt with history of other allergies, including other contact dermatitis, food allergies (with specific foods listed), animal and drug allergies. Also included were questions for eliciting a history of asthma, allergic rhinitis and eczema, both directly, by symptoms, and by requesting information about medications used. Definitions Glove dermatitis: Report of the presence of two or more symptoms of itching, redness and rash (excluding hives) which had been present long term. Latex hypersensitivity: Report of the prompt onset (within 30 minutes of wearing gloves) of hand urticaria with or without the occurrence of allergic rhinoconjunctivitis, lower respiratory tract symptoms such as cough, wheeze or chest tightness, generalised cutaneous itch or urticaria, or upper-airway oedema, hypotension, dizziness or collapse. Atopy: Present if the respondent gave positive answers to any two or more questions on a history of asthma, allergic rhinitis or atopic dermatitis. Statistical analysis All statistical analysis was performed using the SAS package.3 Fisher's exact test was used to test for marginal association in all two-way tables. For other contingency tables, the chi-squared statistic was used to test for marginal association unless otherwise specified. The distribution of demographic characteristics of the staff of the Dental School was used to estimate response rates in order to test for sampling (volunteer) bias. Ethical approval Ethics approval was obtained from the Westmead Hospital Human Research Ethics Committee. Results Completed questionnaires were received from 177 of the 230 staff (77%) by the closing date. Table 1 (below) shows the distribution of non-response from the population, sampled by age group, by sex and by job category. Not all respondents answered all the demographic questions and the response rates are slightly underestimated because 2.3% of age, 8.5% of sex and 14.1% of job category data were missing. It was not possible to compare non-respondents with respondents as the questionnaires contained no personal identifying data. Year of graduation was the question most frequently not answered (32 of 177 questionnaires [18%]). Almost all of the youngest staff responded, whereas among older individuals response rates were as low as 32% for age 51 or 60. As three cells of this Table had expected counts of less than five, we used Fisher's exact test, which showed a significant association (P < 0.00001) between non-response and age group. Men were significantly more likely to respond than women (chi-squared = 26.1; P < 0.00001). The job categories with the highest non-response rates were the small group of registered nurses (58.3%) and the general dentists (43.9%). Evidence that response rate was associated with job category was weak (chi-squared = 11.1; P = 0.05), and this question was not answered by a substantial proportion of respondents. Two-thirds of the study group spent more than half their working day wearing gloves. Symptoms attributed to glove use were reported by one-third of subjects, with itch and redness being the commonest symptoms reported. Table 2 (above) compares selected demographic and allergic characteristics of the 39 subjects (22%) who satisfied the criteria for glove dermatitis with the 16 subjects (9%) fulfilling the criteria set for latex hypersensitivity. Other symptomatic and allergic characteristics of the 16 dental workers with latex hypersensitivity are compared with those of the rest of the respondents in Table 3 (below). Almost all of the individuals reporting latex-allergic symptoms also met the criteria for glove dermatitis. Nineteen per cent of the latex-allergic group reported food allergy, although none reported symptoms with foods such as avocado, banana, nuts or soy products, foods which have been described previously as cross-reacting with latex protein. Fisher's exact test was used to test the hypothesis that atopic diathesis was associated with latex hypersensitivity and glove dermatitis. Atopy was associated with both latex hypersensitivity (P = 0.014) and with glove dermatitis (P = 0.006). Subjects who reported experiencing symptoms of either glove dermatitis or latex-allergic symptoms were asked to indicate any action they had taken; of those with glove dermatitis 59% had never sought professional help and one-third of the latex hypersensitivity group had not sought medical attention. Discussion Our findings indicate a significant occupational health problem among dental personnel wearing latex gloves. In our study, 33% of those surveyed reported some symptoms with glove use; 22% met criteria for glove dermatitis and 9% for latex allergy. The distribution of non-response rates by demographic characteristics shown in Table 1 indicates that the volunteer sample was not fully representative of the study population. In particular, female staff, older staff, general dentists and registered nurses were less likely to respond. As a result of these sampling biases, our findings may not be fully generalisable to dental workers. For example, if staff experiencing symptoms were more likely to volunteer for the survey, then our rates will be overestimates of true prevalences. Our results should be interpreted with caution, and future surveys should aim to obtain more representative samples. Relying on questionnaire data alone to determine hypersensitivity reactions has obvious shortcomings. Contact urticaria with glove use may be caused by non-IgE-mediated mechanisms, so our findings may overestimate the prevalence of latex allergy. However, the occurrence of symptoms in other organ systems soon after gloves are used is highly suggestive of an IgE-mediated allergic reaction. Confirmation of this prevalence rate will require some objective measure of IgE-mediated hypersensitivity such as skin testing with appropriate extracts, or in-vitro assay of specific IgE antibodies to the latex allergens. Our prevalence rate of 9% for latex allergy among dental personnel is comparable with, although lower than, rates reported elsewhere. In a similar questionnaire survey without objective testing, Berky et al.4 reported "symptoms of an allergic nature" in 13.7% of 1043 United States Army dental officers. However, this figure included those reporting delayed symptoms as well as those reporting localised contact urticaria or generalised urticarial reactions. Furthermore, a third of those reporting symptoms were unable to be classified as having either contact dermatitis or latex allergy from their questionnaire responses. Only 25% of those reporting latex allergy in the Berky study had consulted a physician. Rankin et al.5 also conducted a questionnaire survey in a large dental facility in the US, obtaining a 15% prevalence of adverse reactions to latex gloves (Table 4). A number of studies4-6 have confirmed a relationship between atopy and latex allergy (Table 4). We also found a significant association when atopy was defined by our criteria (reported presence of any two of asthma, allergic rhinitis or atopic dermatitis). Walsh et al.7 examined factors influencing the wearing of protective gloves in general dental practice in Brisbane. Of 201 dentists who replied to a survey, 84.6% (170) reported that they routinely used gloves in their practice, compared with 13.9% (28) who reported that they did not. Reasons given for not wearing gloves included reduced sensation, reduced movement, low infection risk, skin reaction, patient acceptance and cost. Our results imply that there is a high rate of glove-related symptoms, particularly of glove dermatitis, in this dental workforce. This indicates a need to educate this occupational group about proper hand care as repeated washing, scrubbing and numerous glove changes contribute to hand irritation. Latex allergy has many implications for health care workers which extend beyond the occupational setting. Most reports of life-threatening anaphylactic reactions during surgery and other procedures have involved sensitised health care workers.8 These individuals need to know the risk they run when they themselves are patients. Mucosal contact with latex appears to be a far more potent trigger for anaphylactic reactions than cutaneous contact. People with latex allergy must be given anaesthesia and surgery in a totally latex-free environment. Atopic individuals appear to be especially at risk for sensitisation to the latex protein. Adequate counselling about occupational exposure and advice about methods of minimising the risk need to be considered in view of this and other studies. Latex gloves from different sources vary in the amount of latex protein detectable on the surface.9 Presumably, those with less cause less sensitisation and should be preferred in a workplace where glove use is mandatory. Acknowledgements We would like to thank Maggie Melink for valuable assistance and the staff of the Westmead Dental School for their cooperation. References Toner JS. Rubber. In: Fisher AA, editor. Contact dermatitis. 3rd ed. Philadelphia, Pa: Lea & Febiger, 1986: 603-643. Slater JE. Rubber anaphylaxis. N Engl J Med 1989; 320: 1126-1130. PC SAS [computer program], version 6.04. Cary, NC: SAS Institute Inc, 1991. Berky ZT, Luciano J, James WD. Latex glove allergy. A survey of the US Army Dental Corps. JAMA 1992; 268: 2695-2697. Rankin KV, Jones DL, Rees TD. Latex glove reactions found in a dental school. J Am Dent Assoc 1993; 124: 67-71. Turjanmaa K. Incidence of immediate allergy to latex gloves in hospital personnel. Contact Derm 1987; 17: 270-275. Walsh LJ, Lange P, Savage MW. Factors influencing the wearing of protective gloves in general dental practice. Quintessence Int 1995; 26: 203-209. Leynadier S, Pecquet C, Dry J. Anaphylaxis to latex during surgery. Anesthesia 1989; 44: 547-550. Yunginger JW, Jones RT, Fransway AS, et al. Extractable latex allergens and proteins in disposable medical gloves and other rubber products. J Allergy Clin Immunol 1994; 93: 836-842. (Received 19 Sep 1995, accepted 6 Apr 1996) Authors' details Department of Immunology, Westmead Hospital, Sydney, NSW. Constance H Katelaris, PhD, FRACP, Senior Consultant. Department of Paediatric Dentistry, Westmead Dental School, Sydney, NSW. Richard P Widmer, MDSc, FRACDS, Associate Professor. Department of Public Health and Community Medicine, University of Sydney, Sydney, NSW. Ross M Lazarus, MB BS, FAFPHM, Lecturer. Reprints: Dr C H Katelaris, Department of Immunology, Westmead Hospital, Westmead, NSW 2145. ©MJA 1997 <URL: http://www.mja.com.au/> © 1997 Medical Journal of Australia.

Age-specific HIV incidence among homosexually active men in Australia Matthew G Law, Philip S Rosenberg, Ann McDonald and John M Kaldor MJA 1996; 164: 715-718. This article has been published in the Medical Journal of Australia. Readers may print a single copy for personal use. No further reproduction or distribution of the articles in whole or in part should proceed without the permission of the publisher. For copyright permission, contact the Australasian Medical Publishing Company Abstract - Introduction - Methods - Results - Discussion - Acknowledgements - References - Author's Details - Table - Figure 1 - Figure 2 Register to be notified of new articles by email - To Current Issues list - ©MJA1996 Abstract Objective: To estimate age at HIV infection among homosexually active men in Australia. Design: Age-specific back-projection estimates of HIV incidence. Methods: Monthly counts of AIDS among homosexually active men diagnosed by 30 June 1994 and reported by 31 March 1995 were obtained from the National AIDS Registry and were adjusted for reporting delays. The progression rate to AIDS was estimated from a large cohort study of HIV-infected homosexual men, with adjustment for the effect of age at HIV infection and the effect of antiretroviral and prophylactic treatments. Results: The median age at HIV infection was estimated to have decreased from 31 years of age between 1982 and 1984 to between 23 and 27 years in the periods 1987 to 1989 and 1990 to 1994. Despite the trend to a younger median age at HIV infection during the current epidemic, HIV incidence was estimated to have declined in all age groups from a peak in the mid-1980s. This decline was more pronounced in the older age groups, with more modest reductions in age groups under 30 years. Conclusion: Most HIV infections among homosexually active men since 1987 appear to have occurred in men aged under 30 years. This has implications for education programs aimed at preventing HIV infection among homosexually active men. MJA 1996; 164: 715-718 Introduction A rapid increase in HIV incidence in Australia in the early 1980s (with a peak of around 3300 cases per year in the mid-1980s, followed by a rapid decline) was reported by both the National Working Group on HIV Projections in 19921 and the Evaluation of the National HIV/AIDS Strategy in 1996.2Cumulative HIV incidence to the end of 1993 was estimated to be between 13 000 to 19 000 cases, with a preferred estimate of 15 200 cases. The estimated annual incidences of HIV infection from 1981 to 1993 for homosexually active men, for heterosexual women, and for women and heterosexual men who were intravenous drug users, are shown in the Box.3 About 85% of both AIDS cases and diagnosed HIV infections in Australia have been in men who have had homosexual contact.4 Over the period of the current AIDS epidemic, it could be expected that a new generation of homosexual men has become sexually active and at risk of HIV infection. An important question, therefore, is whether and how HIV incidence among younger homosexually active men has changed over time. The method of back-projection has become the preferred method for estimating past HIV incidence, and for projecting future AIDS incidence, in most developed countries.5-7 Here, using an extension of the back-projection method,8 we estimate the age at HIV infection among homosexually active men in Australia. Methods Our analyses were based on AIDS cases diagnosed by 30 June 1994 and reported to the National AIDS Registry (NAR) at the National Centre in HIV Epidemiology and Clinical Research by 31 March 1995 in men with a history of homosexual contact, including those who were intravenous drug users. These AIDS incidence data were adjusted for the reporting delay between the time of diagnosis to the time of entry of the case on the NAR.9 All estimates of HIV incidence were adjusted for underreporting of AIDS cases to the NAR, which is currently estimated to be about 10%.3 Age-specific HIV incidence was estimated using the back-projection method suggested by Rosenberg,8 in which the rate of progression to AIDS is assumed to be associated with age.10 To reflect the uncertainty in the rate of progression to AIDS, particularly beyond 10 years following HIV infection, we used two different progression-rate distributions derived from a large-cohort study of HIV-infected homosexual men.11 A faster progression-rate distribution was based on a median "time-to-AIDS" of nine years for people aged 30 at diagnosis of HIV infection, and the rate of progression to AIDS was taken to increase by 1.042 for each year the age at HIV infection was above 30, and to decrease by 1/1.042 for each year the age was below 30. A slower progression rate was based on a median "time-to-AIDS" of 9.8 years and an age-effect of 1.037. As the effect of age on modifying the progression rate to AIDS is uncertain and the assumption of an age-effect would lower the back-projection estimate of median age at HIV infection, we also analysed the progression rates with the age-effects at both zero and half of the above values. The extended definition of AIDS adopted in Australia in January 1988 was assumed to result in a 10% increase in the rate of progression to AIDS.12 The effect of antiretroviral and prophylactic treatments on the progression rate was incorporated into the analysis by assuming that 20% of HIV-infected people were first diagnosed as HIV positive at diagnosis of AIDS and so would not have received treatment before the diagnosis of AIDS,13 and that zidovudine became available in Australia from mid-1987 and was accepted and tolerated by up to 90% of people with diagnosed HIV infection. The effect of treatment was assumed to be a modest 20% reduction in the rate of progression to AIDS and to gradually diminish during ex- tended use, consistent with randomised clinical trials of zidovudine.14-16 We also did further analyses assuming a stronger treatment effect which corresponded to a 50% reduction in the progression rate. Back-projections were conducted for five time periods (early 1977 to mid-1981, mid-1981 to mid-1984, mid-1984 to mid-1986, mid-1986 to mid-1989 and mid-1989 to mid-1994) and for five age groups (17 to 20, 21 to 24, 25 to 29, 30 to 39 and 40 to 59 years). To allow direct comparison of estimates of HIV incidence in different age groups and time periods, estimates of HIV incidence in a given age group and time period were standardised by dividing the estimated total number of new HIV infections by the product of the number of age-years and calendar-years. Results There were 4661 cases of AIDS diagnosed by 30 June 1994 and reported to the NAR by 31 March 1995 among homosexually active men. After adjusting for reporting delays, analyses were based on 4769 cases of AIDS. We did not analyse other exposure categories as there were too few AIDS cases in these categories. The patterns of HIV incidence were similar for both the faster and the slower progression-rate distributions, with a rapid increase in the annual number of HIV infections in the early 1980s to a peak of between 2600 and 3100 per year in the mid-1980s, followed by a steep decline. The faster progression rate gave a smaller estimate of the cumulative HIV incidence to 30 June 1994 than the slower progression rate: 11 800 HIV infections, compared with 14 700, respectively. Both the pattern of HIV incidence and the cumulative HIV incidence estimated by the age-specific analysis are consistent with the non-age-specific back-projections for homosexually active men shown in the Box. The median age at HIV infection was similar for both the faster and slower progression-rate distributions (Figures 1a and 1b). As the epidemic developed, there was a decline in the median age at HIV infection, from 31 years in 1982 to 1984, to 23 and 25 years for both the faster and slower progression-rate distributions, respectively, in 1987 to 1989, and to 25 and 27 years in 1990 to 1994. The age-effect of zero on the progression rate to AIDS gave similar results, showing a decline in median age at HIV infection from 34 and 33 years for both the faster and slower progression-rate distributions, respectively, in 1982 to 1984, to 24 and 26 years, respectively, in 1987 to 1989, and to 27 and 28 years, respectively, in 1990 to 1994. Analyses which assumed half the age-effect also gave similar results. Figure 1: Estimated median age at HIV infection by calendar year (a) faster progression-rate distribution, (b) slower progression-rate ditribution (see text) Although the median age at HIV infection has declined during the epidemic, HIV incidence has declined in all age groups from a peak in the mid-1980s (Figures 2a and 2b). However, this decline has been more pronounced in the older age groups; most HIV infections in recent years appear to have been in men under 30 years. HIV incidence was similar in the age groups 21 to 24, 25 to 29 and 30 to 35 years in 1982 to 1984, but by 1990 to 1994 HIV incidence was substantially higher in the 21 to 24 years age group. Figure 2: Estimated HIV incidence by age group at HIV infection and by calendar year (a) faster progression-rate distribution, (b) slower progression-rate ditribution (see text) Sensitivity analyses which assumed a stronger treatment effect (a 50% reduction in the progression rate to AIDS) gave similar results and are not shown. Discussion The decrease in the estimated median age at HIV infection among homosexually active men, as shown by our age-specific back-projection analysis, has previously been reported using similar methods in the United States,17 and has also been observed in open cohorts of homosexual men attending genitourinary medicine clinics in London between 1988 and 1994.18 Since 1991, another Australian source of information about new HIV infections has been reported cases to the NAR of newly acquired HIV infection. Diagnosed HIV infections are considered to be newly acquired if the person had a negative HIV test within a year of their positive HIV test or if a HIV seroconversion illness was diagnosed. About 200 newly acquired HIV infections are reported each year,4 and the median age of these people has been around 30 to 31 years since 1991 (more than the median age of between 25 and 27 years estimated by the age-specific back-projection analyses in our study). However, HIV incidence since 1991 has been estimated to be around 500 cases per year,3 so only about 40% of these cases are diagnosed each year as newly acquired. It is possible that this 40% is a biased sample of all new HIV infections and, in particular, slightly older than new HIV infections which are not diagnosed as newly acquired. Back-projection analyses are based on assumptions that are subject to uncertainty, in particular the assumed progression-rate distributions. The available Australian data were part of a large- cohort study of HIV-infected homosexual men,11 and indicate that the overall progression rate in homosexually active men in Australia is broadly consistent with both the faster and slower progression-rate distributions that were assumed in the age-specific back-projections. Other major sources of uncertainty are in the modifying effect of age at HIV infection and the effect of antiretroviral and prophylactic treatments on the progression rate to AIDS. The general trend of decreasing age at HIV infection was, however, a robust feature of sensitivity analyses which investigated these uncertainties, suggesting that this trend is at least qualitatively correct. Examining the history of the HIV epidemic among homosexually active men in Australia lends further support for a decreasing age at HIV infection. When the HIV epidemic began in Australia around 1980, homosexually active men of all ages were at risk of infection. As the epidemic progressed, new homosexual men were at risk of HIV infection as they became sexually active, generally at a younger age. Thus, the age distribution of homosexually active men at risk of HIV infection (and hence of those already infected) would be expected to have decreased as the epidemic progressed. The estimates of HIV incidence were standardised for the number of calendar-years and age-years in each time period and age group. This allows direct comparison of HIV incidence if the number of homosexually active men can be assumed to be constant across different age groups over the time period of the epidemic. Based on the 1991 census,19 the resident male population in Australia was fairly uniformly distributed across the age range 15 to 44 years (at about 140 000 men per age-year) but there were fewer men in older age groups. In the age range 17 to 39 years, however, direct comparison of the standardised estimates of HIV incidence appears valid, provided that the proportion of homosexual men is constant over the age range. One feature of our analyses which requires cautious interpretation is the apparent small increase in HIV infection in some age groups between 1987 to 1989 and 1990 to 1994. As few people infected with HIV in the most recent period would be expected to have progressed to AIDS, these figures are uncertain. The most appropriate interpretation is that the number of new HIV infections has been reasonably stable in all age groups between 1987 and 1994. In conclusion, age-specific back-projections suggest that the median age at HIV infection among homosexually active men has decreased over the period 1982 to 1994, and that most HIV infections in more recent years have occurred in men aged under 30. These results suggest that education programs aimed at preventing HIV infections in homosexually active men should be targeted towards younger age groups, in particular homosexually active men in their early twenties. Register to be notified of new articles by email - To Current Issues list - ©MJA1996 Acknowledgements The National Centre in HIV Epidemiology and Clinical Research is funded by the Australian National Council on AIDS through the Commonwealth AIDS Research Grants Committee. The authors thank the National HIV Surveillance Committee and the doctors who reported AIDS cases under national surveillance procedures. References National Working Group on HIV Projections. Estimates and projections of the HIV epidemic in Australia, 1981-1994. Internal technical report 1. Sydney: National Centre in HIV Epidemiology and Clinical Research, April 1992. Valuing the past . . . investing in the future. Evaluation of the National HIV/AIDS Strategy 1993-94 to 1995-96. Canberra: Commonwealth Department of Human Services and Health, 1995. National Centre in HIV Epidemiology and Clinical Research. Estimates and projections of the HIV epidemic in Australia, 1981-1997. In: Technical appendix no. 1: An epidemiological assessment of the HIV epidemic in Australia. Evaluation of the National HIV/AIDS Strategy 1993-94 to 1995-96. Canberra: Commonwealth Department of Human Services and Health, 1996. National Centre in HIV Epidemiology and Clinical Research. Australian HIV Surveillance Report, Vol. 12, No. 1. Sydney: National Centre in HIV Epidemiology and Clinical Research, 1996. Brookmeyer R, Gail MH. A method for obtaining short-term projections and lower bounds on the size of the AIDS epidemic. J Am Stat Soc 1988; 3: 301-308. Taylor JM. Models for the HIV infections and AIDS epidemic in the United States. Stat Med 1989; 8: 45-58. Becker NG, Watson LF, Carlin JB. A method of non-parametric back-projection and its application to AIDS data. Stat Med 1991; 10: 1527-1542. Rosenberg PS. Backcalculation models of age-specific HIV incidence rates. Stat Med 1994; 13: 1975-1990. Brookmeyer R, Liao J. The analysis of delays in disease reporting: methods and results for the acquired immunodeficiency syndrome. Am J Epidemiol 1990; 132: 355-365. Rosenberg PS, Goedert JJ, Biggar RJ, for the multicenter Hemophilia Cohort Study and the International Registry of Seroconverters. Effect of age-at-seroconversion on the natural AIDS incubation distribution. AIDS 1994; 8: 803-810. Biggar RJ and the International Registry of Seroconverters. AIDS incubation in 1891 HIV seroconverters from different exposure groups. AIDS 1990; 4: 1059-1066. Selik RM, Buelher JW, Karon JM, et al. Impact of the 1987 revision of the case definition of Acquired Immune Deficiency Syndrome in the United States. J AIDS 1990; 3: 73-82. Kaldor JM, French MAK. When do patients present with HIV infection? Med J Aust 1993; 158: 37-38. Volberding PA, Lagakos SW, Koch MA, et al. Zidovudine in asymptomatic human immunodeficiency virus infection: A controlled trial in persons with fewer than 500 CD4-positive cells per cubic millimeter. N Engl J Med 1990; 322: 941-949. Cooper DA, Gatell JM, Kroon S, et al. Zidovudine in persons with asymptomatic HIV infection and CD4+ cell counts greater than 400 per cubic millimeter. N Engl J Med 1993; 329: 297-303. Concorde Coordinating Commitee. Concorde: MRC/ANRS randomised double-blind controlled trial of immediate and deferred zidovudine in symptom-free HIV infection. Lancet 1994; 343: 871-881. Rosenberg PS, Biggar RJ, Goedert JJ. Declining age at HIV infection in the United States [letter]. N Engl J Med 1994; 330: 789-790. Miller E, Waight PA, Tedder RS, et al. Incidence of HIV infection in homosexual men in London, 1988-94. BMJ 1995; 311: 545. Australian Bureau of Statistics. Australia in profile. Canberra: ABS, 1993. (Catalogue No. 2821.0.) (Received 13 Nov 1995, accepted 2 April 1996) Authors' details National Centre in HIV Epidemiology and Clinical Research, The University of New South Wales, Sydney, NSW. Matthew G Law, MSc, Statistician. Ann McDonald, MPH, Senior Research Assistant. John M Kaldor, PhD, Professor of Epidemiology. National Cancer Institute, Rockville, USA. Philip S Rosenberg, PhD, Biostatistician. No reprints will be available. Correspondence: Mr M G Law, National Centre in HIV Epidemiology and Clinical Research, The University of New South Wales, 376 Victoria Street, NSW 2010. E-mail: mlaw@nchecr.unsw.edu.au Readers may print a single copy for personal use. No further reproduction or distribution of the articles should proceed without the permission of the publisher. For permission, contact the Australasian Medical Publishing Company 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/> © 1996 Medical Journal of Australia.

Matthew G Law · Philip S Rosenberg · Ann McDonald · John M Kaldor

Limited adverse occurrence screening: using medical record review to reduce hospital adverse patient events Alan M Wolff Readers may print a single copy for personal use. No further reproduction or distribution of the articles should proceed without the permission of the publisher. For permission, contact the Australasian Medical Publishing Company 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/>". Abstract - Introduction - Methods - Results - Discussion - Acknowledgements - References - Authors' details - - ©MJA1997 Abstract Objectives: To determine whether continuous detection of adverse patient occurrences followed by analysis and medical intervention can alter the rate of adverse occurrences. Design and participants: 15 912 patients discharged from one hospital were reviewed in two stages. Medical records were screened retrospectively by medical records staff for one or more of eight general patient outcome criteria. Those that screened positive for the criteria were reviewed by one of four doctors. If an adverse occurrence was confirmed, further analysis and recommendations for action to prevent its recurrence were made at meetings of the four doctors, and forwarded to a committee of visiting medical officers who decided on the appropriate course of action. Setting: A rural base hospital in Horsham, Victoria, between July 1991 and June 1994. Main outcome measures: The rate and severity of adverse patient occurrences in each year. Results: 1465 records were screened positive for one or more criteria, and an adverse patient occurrence was confirmed in 155. 88 cases were determined to be minor or not preventable and further action (mostly by changes to hospital policies) was recommended for the remaining 67. Over the three years, the number of adverse occurrences fell from 69 (1.35% of all patient discharges in the first year) to 33 (0.58% of all patient discharges in the third year) ( P < 0.0001) and there was no significant change in severity. Conclusions: The rate of adverse patient occurrences can be significantly reduced by their continuous detection using retrospective screening in conjunction with review, analysis and action to prevent recurrences. MJA 1996; 164: 458-461 Introduction The objective of medical quality assurance programs is to improve the quality of care provided to patients. However, "quality medical care" has proved difficult to define, leading some programs to attempt to detect "disquality", or events that should not happen under optimal conditions.1 Quality, then, may be defined as the absence or lowering of the rate of adverse patient events or occurrences. Many preventable adverse patient events occur in hospitals. The Quality in Australian Health Care Study, which reviewed over 14 000 patient admissions in 28 hospitals in New South Wales and South Australia, found 16.6% involved an adverse event; half of these were assessed as highly preventable.2 The Harvard Medical Practice Study, a review of over 30 000 inpatient medical records from 51 acute-care hospitals in the United States, showed that adverse events occurred in 3.7% of hospitalisations and that many of these were the result of substandard care.3 Both these studies used methods that required substantial resources in time and money to detect adverse patient events, and neither looked at the effect of intervention in preventing recurrence of an adverse event. Limited adverse occurrence screening is a continuous process of retrospective screening and review of inpatient medical records to detect adverse patient occurrences.4 It is referred to as "limited" because only eight general patient outcome criteria are used. When an adverse occurrence is found, appropriate action is taken to prevent its recurrence and the success of the action is determined by ongoing monitoring.4 Thus, the quality assurance feedback loop is securely closed. This study was designed to determine whether limited adverse occurrence screening could reduce the number and rate of adverse occurrences and therefore improve the quality of patient care. Methods The study was undertaken at Wimmera Base Hospital in Horsham, 300 km north-west of Melbourne. There are 11 specialists and 14 general practitioners on the medical staff who treat between 5000 and 6000 inpatients per year. Four doctors were chosen by the Hospital Medical Staff Group to be medical reviewers and to form a Patient Care Committee. All doctors held dual positions in the hospital and had postgraduate clinical qualifications. They were a physician (also Director of Intensive Care), a surgeon (also Chairman of the Medical Staff Group), a general practitioner (also Director of Postgraduate Education) and the Medical Director (also Director of the Accident and Emergency Department). The medical records for all inpatients discharged between 1 July 1991 and 30 June 1994 were screened by medical records staff using eight general patient outcome criteria (Box 1, below). An adverse patient occurrence analysis form was attached to any medical record that met one or more of the criteria. The record was reviewed by the doctor allocated to that criterion, who evaluated the care given to determine if an adverse patient occurrence had arisen before or during admission, and then completed the form. Adverse occurrences for which patients had had previous admissions were excluded. 1: General patient outcome criteria used for screening of medical records Death. Return to operating theatre within 7 days. Transfer from general ward to intensive care unit. Unplanned readmission within 28 days of discharge. Cardiac arrest. Transfer to another acute-care facility. Length of stay greater than 35 days (reduced to 21 days in 1993-1994). Theatre booking cancelled. An adverse patient occurrence was defined as "an untoward patient event which, under optimal conditions, is not a natural consequence of the patient's disease or treatment".1 I have described the limited adverse occurrence screening method in detail in an earlier paper.4 Medical care was evaluated using a six-point scale,5 to help the reviewer decide whether an adverse patient occurrence was caused by medical management: 1 = little or no evidence, 2 = slight evidence, 3 = not likely (less than 50:50 odds but a close call), 4 = more likely than not (greater than 50:50 odds but a close call), 5 = strong evidence, 6 = virtually certain. A score of four or more on this scale was regarded as an adverse patient occurrence. Severity of adverse occurrences was graded on a seven-point scale:1 0 = minor severity, 1 = minor temporary, 2 = minor permanent, 3 = major temporary, 4 = major permanent, 5 = potential major or major continuing, 6 = death. The medical reviewer wrote brief notes about the case for the Patient Care Committee. Adverse occurrences were discussed at the Committee's bimonthly meetings and recommendations for further action relating to patient care were made and forwarded to the Medical Staff Group, which consisted of all visiting medical officers in the hospital and the Director of Medical Services. All data from the adverse patient occurrence analysis forms were entered into a database program developed from the Clipper Dbase Compiler software package.6 A chi-squared test was used for statistical analysis, and confidence intervals were calculated for odds ratios using standard methods.7 Results A total of 15 912 inpatients were discharged between 1 July 1991 and 30 June 1994. Using the eight general patient outcome criteria, 1465 records (9.21%) were screened positive for one or more criteria by medical records staff. The commonest criteria among these were unplanned readmission of a patient within 28 days of discharge (3.4% of all records screened) and transfer of a patient to another acute-care facility (2.8% of all records screened) (Box 2). A medical record was most likely to be subsequently confirmed as containing an adverse patient occurrence when it was selected by the criterion "return to operating theatre within seven days" (odds ratio = 3.98; 95% confidence interval, 2.15-7.37) (Box 2). After medical review, 155 records (0.97% of all records screened) were found to contain an adverse patient occurrence using the six-point scale (Box 3). Of these adverse occurrences, 45 were minor and 110 were major (Box 4). Of the 155 adverse patient occurrences confirmed on medical review, no further action was recommended by the Patient Care Committee in 88 cases (56.8%) (i.e., the adverse occurrence was considered to be not preventable or of minor significance). Recommended action for the other cases included changing the relevant hospital policy; presenting the case at a postgraduate meeting; undertaking a quality assurance program to investigate the adverse occurrence in detail; discussion with, or counselling of, the doctor involved; and, rarely, review of the doctor's clinical privileges or reporting the case to the hospital's insurers. Over the three-year period of the study, 66 recommendations were made by the Patient Care Committee and almost all were accepted by the Medical Staff Group and became hospital policy. Changes in hospital policies were both clinical and administrative. Clinical policy changes included restricting some drug prescribing; revised protocols for reporting vital signs; eliminating use of multidose drug vials; formulating guidelines regarding fitness for general anaesthesia; and developing protocols for managing patients with alcohol withdrawal, haematemesis and melaena, and for patients who have had a cerebrovascular accident or who require analgesia. Administrative policy changes included sending copies of referral letters when patients were transferred to other hospitals; sending private antenatal notes to the obstetric ward before the patient's confinement; redesigning the preadmission form to allow consent for procedures to be obtained in the consultant's room; and routinely including a copy of the death certificate in the medical record (if the patient had died in hospital). (Examples of the types of adverse occurrences, more detailed recommendations for further action and their subsequent reduction in incidence are given in an earlier paper.4) The adverse patient occurrences were grouped by year to determine whether the actions instigated had any effect on the subsequent rate of adverse occurrences. The rate of adverse patient occurrences fell from 69 in the first year (1.35% of all patient discharges in that year) to 33 in the third year (0.58% of all patient discharges in that year) (Box 4). The proportion of adverse occurrences fell significantly with time (chi-squared = 17.11; df = 2; P = 0.0002) and this trend was linear (chi-squared = 16.87; df = 1; P< 0.0001). When the severity rating of adverse events was divided into minor (severity score 0-2) and major (severity score 3-6), there was no significant change in the severity of adverse patient occurrences between the first, second and third years of the study (Box 4). Also, the proportion of adverse patient occurrences detected in each year by screening did not alter significantly over time for patient age, sex or insurance status (Box 5). The commonest major diagnostic categories, using the ICD-9 classification,8 in which patients had adverse occurrences were injury and poisoning, followed by digestive and circulatory disorders (Box 6, below). Patient characteristics such as sex and age did not change significantly during the three years of the study, but the proportion of inpatients privately insured fell from 26.6% to 24.3% (chi-squared = 10.24; df = 2; P< 0.01). Discussion Limited adverse occurrence screening has been shown to be effective in detecting approximately 50% of adverse patient occurrences,9 a much higher proportion than found by traditional medical quality assurance programs. This study has shown that it is also possible to subsequently reduce the number of adverse occurrences by more than 50% over a three-year period. These results indicate that use of a more complex and expensive screening system is not required to achieve a significant reduction in adverse occurrences. Adverse occurrence screening has been shown to be valid and reliable, with a high level of agreement between reviewers as to whether a patient has had an adverse occurrence.10 Although in this study each medical record was reviewed by only one doctor, other studies using a similar methodology and rating scale have shown high agreement.5 Limited adverse occurrence screening is efficient as it requires the review of the medical records of slightly less than 10% of all patients discharged, costs only 0.1% of an acute-care hospital's total budget, and is fast and accurate (false positive rate, 2.0%; false negative rate, 0.4%).9 In this study, 9.21% of medical records were screened positive for one or more of the eight general patient outcome criteria and an adverse occurrence was confirmed in 0.97%. Little or no clinical judgement was required in the screening process. This compares with a rate of 8.4% (confirmed in 0.73%) in a study using similar screening criteria in a medium-sized New South Wales hospital.11 In the Quality in Australian Health Care Study, potential adverse events were detected in 43.7% of the medical records on initial screening (a much higher rate than in other studies) using 18 criteria, some of which required clinical judgement.2 Adverse events were confirmed in 16.6%, although, unlike in this study, 49% of these occurred before the sample admission. An average rate of potential adverse events of 13.2% was found in a study of 146 Veterans Affairs Medical Centers in the United States that used nine screening criteria.12 There was considerable variation between centres, with a 25th percentile of 8.6% (confirmed in 1.7% of records) and a 75th percentile of 27.2% (confirmed in 7.9% of records). In the Harvard Medical Practice Study, potential adverse events were detected on initial screening in 25.9% of medical records using 18 screening criteria.13 Adverse events were confirmed in 0.2% to 7.9%, depending on hospital complexity and location, with the rate in rural hospitals being 1%.13 Limited adverse occurrence screening can be adapted to all sizes of hospital. It has recently been introduced into 10 small hospitals in central western Victoria by the Division of General Practice in that area working with the Monash University Centre for Rural Health in Moe, Victoria. In larger hospitals, limited adverse occurrence screening could be introduced on a departmental basis. The program could be enhanced by combining its retrospective medical record review with prospective critical incident reporting. In the United States, adverse occurrence screening and incident reporting both detected a substantial number of adverse events; however, there was less than a 50% overlap in the adverse events detected by each method when used concurrently.14 Thus, a program that simultaneously uses both methods may detect the largest pool of adverse occurrences for analysis and development of preventive strategies. Acknowledgements I wish to thank Mr Ian Campbell, Dr David Leembruggen and Dr Grant Phelps for their enthusiastic participation in the Patient Care Committee; Mrs Cathy Dooling, Chief Medical Records Administrator, and her staff, for their support with the screening program; Mr Kieran Loughran, Computer Systems Officer, for assistance with data collection; and Mrs Naomi Uytdehaag for preparing the manuscript. References Craddick JW, Bader B. Medical management analysis: a systematic approach to quality assurance and risk management. Auburn (CA): Joyce W. Craddick, 1983. Wilson RM, Runciman WB, Gibberd RW, et al. The Quality in Australian Health Care Study. Med J Aust 1995; 163: 458-471. Brennan TA, Leape LL, Laird NM, et al. Incidence of adverse events and negligence in hospitalized patients. N Engl J Med 1991; 324: 370-376. Wolff A. Limited adverse occurrence screening: a medical quality control system for medium sized hospitals. Med J Aust 1992; 156: 449-452. Hiatt HH, Barnes BA, Brennan TA, et al. A study of medical injury and medical malpractice: an overview. N Engl J Med 1989; 321: 480-484. Clipper [computer program]. Version 5.0. Los Angeles: Nantucket Corporation, 1990. Daly LE, Bourke GJ, McGilbray J. Interpretation and uses of medical statistics. Oxford: Blackwell Scientific Publications, 1991: 398-424. US Department of Health and Human Services. The international classification of diseases. 9th revision. Clinical modification (ICD-9-CM). 3rd ed. Bethesda, Md: DHHS, 1989. Wolff A. Limited adverse occurrence screening: an effective and efficient method of medical quality control. J Qual Clin Pract 1995; 15: 221-233. Panniers TL, Newland J. The adverse patient occurrence inventory: validity, reliability and implications. Qual Rev Bull 1986; 12: 311-315. Britton S. A hospitalwide outcome study. Aust Clin Rev 1991; 11: 132-135. Goldman R, Walder DJ. An initial assessment of the Veterans Affairs Occurrence Screening Program. Qual Rev Bull 1992; 18: 327-332. 13. Brennan TA, Herbert LE, Laird NM, et al. Hospital characteristics associated with adverse events and substandard care. JAMA 1991; 265: 3265-3269. 14. O'Neil AC, Petersen LA, Cook EF, et al. Physician reporting compared with medical record review to identify adverse medical events. Ann Intern Med 1993; 119: 370-376. (Received 2 Sep 1995, accepted 2 Feb 1996) Authors' details Wimmera Base Hospital, Horsham, VIC. Alan M Wolff, FRACGP, MBA, Director of Medical Services, and Director of the Accident and Emergency Department. Reprints: Dr A M Wolff, Medical Administration, Wimmera Base Hospital, Baillie Street, Horsham, VIC 3400. ©MJA 1997 <URL: http://www.mja.com.au/> © 1997 Medical Journal of Australia.

Alan M wolff

Suspected snakebite in children: a study of 156 patients over 10 years Helen J Mead and George A Jelinek Readers may print a single copy for personal use. No further reproduction or distribution of the articles should proceed without the permission of the publisher. For permission, contact the Australasian Medical Publishing Company 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/>". Abstract - Introduction - Methods - Results - Discusssion - References - Authors' details - - ©MJA1997 Abstract Objective: To describe the epidemiology and clinical features of children presenting to an emergency department with suspected snakebite. Design: A retrospective study of patient records. Setting: An emergency department of a children's teaching hospital (Princess Margaret Hospital) in Perth, Western Australia. Participants: All children attending the emergency department from 1984 to 1993 with suspected snakebite. Main outcome measure: Clinical and laboratory evidence of envenomation. Results: Over the decade studied, 156 children (mean age, six years and eight months) presented with suspected snakebite; over two-thirds (68%) were boys. In at least 31% of cases, no appropriate first aid had been applied. Only 14 children were envenomed according to clinical and laboratory criteria: 10 of these had coagulopathy; one of the 10 also had rhabdomyolysis. A Venom Detection Kit was used in 117 children. The test gave a positive result in 21 children (13%). Antivenom was given to 18 children, 14 of whom were definitely envenomed. Four of the envenomed children returned a negative result of Venom Detection Kit testing at all sites tested, and in five patients not clinically envenomed the urine specimen tested positive with the Venom Detection Kit (presumably a false positive result or subclinical envenomation). Of the 156 children, 130 were admitted to hospital, and 26 were discharged directly from the emergency department. All children recovered completely. Conclusions: (i) Many children did not receive appropriate first aid for snakebite; (ii) Most children with suspected snakebite presenting to the emergency department were not envenomed; (iii) Envenomation was best diagnosed by clinical features and laboratory investigations, with the Venom Detection Kit being used to determine the appropriate antivenom; (iv) Discharging children directly from the emergency department is not recommended. MJA 1996; 164: 467 Introduction Three major Australian studies of snakebite in children have been published.1-3 Two were from south-east Queensland1,2 and the third from Victoria.3 No similar studies have been reported from Western Australia, but one has been published on patterns of envenomation in 193 adults admitted to Perth teaching hospitals.4 There are also no published data on children presenting to Australian emergency departments with snakebite. Our aim was to describe the presenting features, treatment and outcome in children with suspected snakebite attending an emergency department. We also examined aspects of the use of the Venom Detection Kit, which was widely used as an adjunct in the diagnosis and management of these children. Methods We studied the records of all children presenting to Perth's Princess Margaret Hospital for Children with suspected or definite snakebite between 1 January 1984 and 31 December 1993. If parents or other carers had a reasonable suspicion that a child may have been bitten by a snake, this was classified as "suspected snakebite", and in "definite snakebite" a reliable history was available of a snake striking the child (whether or not envenomation occurred), or there was clinical and/or laboratory evidence of envenomation in circumstances consistent with snakebite. To ensure all records were found, multiple data sources were accessed, including hospital morbidity coding for inpatient admissions; intensive care unit records; and emergency department records. Data extracted from the records were entered onto a Microsoft Access 2.0 database to assist with analysis.5 To facilitate a comparison with the study in adults from Perth hospitals,4 data on patients' snakebite history and envenomation status were extracted and patients were grouped in the categories listed in the Perth study4 and shown in Box 1. Systemic envenomation was deemed present if there was clinical (vomiting, abdominal pain, or neurotoxic effects -- ptosis, convulsions, or difficulty with breathing or swallowing) or laboratory evidence (coagulopathy, haemolysis, rhabdomyolysis or renal failure). Results Over the 10 years, 160 children presented to the Emergency Department at Princess Margaret Hospital with suspected snakebite. Patient records were available for 156 children. The snake was brought in with the patient for identification on 16 occasions (on one occasion the snake was still alive!). Thirty-one children were admitted to the Intensive Care Unit, 87 to a general ward, 12 to the Emergency Department observation ward and 26 were discharged directly from the Emergency Department. Age and sex distribution The mean age of the children was six years and eight months (range, 15 months to 14 years 4 months), and over two-thirds (68%) were boys. Athough there was an even age distribution among children who were envenomed, there were 46 (29%) toddlers (less than three years) among the children presenting. Location Most snakebites occurred in the victims' garden (59 cases; 38%) or surrounding suburban bushland (51 cases; 33%). Eight (5%) occurred in a house or outbuilding and 26 (17%) children were bitten in open country. The location of the remaining 12 (7%) was not specified. Seasonal distribution Most children presented in the summer months, with very few presentations during winter; only three of the 14 children who were envenomed were bitten outside the months of October to April. Bite site Most children were bitten on a limb (lower limb: 103 cases, 66%; upper limb: 43 cases, 28%). There was one bite to the torso and one to the head, with the site not recorded in eight cases. Puncture marks were noted in 81 cases (52%). First aid It was often difficult to determine from the patient records whether first aid had been applied. However, in 39 cases it was clearly documented that no first aid was used, and inappropriate first aid (such as washing the site or use of an arterial tourniquet) was recorded in 10 cases (i.e., at least 49 children [31%] did not have effective first aid). The pressure- immobilisation first aid technique was used in 75 children (48%), although in some it was noted that the bandage had been applied only loosely or the limb was not fully immobilised. There was no record of whether first aid was applied in 32 children (21%). In some cases first aid measures were inappropriately left in place for a prolonged period after reaching hospital. Envenomation Fourteen children had systemic envenomation and 17 possible systemic envenomation according to the criteria used in this study (Box 1). The details of the clinical features, management and outcome of the 14 envenomed children are summarised in Box 2. In two children with possible envenomation, laboratory coagulation tests gave values just outside the normal range; both had an uneventful course and were not given antivenom. Four other child ren classified with possible envenomation received antivenom early for non-specific symptoms, without confirmatory laboratory evidence of envenomation. Venom detection Of the 117 cases in which the Venom Detection Kit was used, a positive result was obtained in 21 children (12 from wound swabs, 10 from urine, and four from blood). In nine of these child ren with no evidence of systemic envenomation, venom was detected in urine in five, and in skin swabs in the other four. In a child with borderline coagulopathy, the Venom Detection Kit gave a positive result in blood, but a negative one in urine. Another child treated for possible envenomation tested positive for venom at the "bite site", but tests of urine and blood gave a negative result with the Venom Detection Kit (Case D under "Use of antivenom"). The venom type detected by the Venom Detection Kit in envenomed patients is shown in Box 2. Use of antivenom Antivenom was given to 18 children (polyvalent in eight cases, brown snake in seven and tiger snake in five; two patients received more than one type of antivenom). Four children who were given antivenom may not have been envenomed. Case A: An eight-year-old boy with a witnessed bite received antivenom for regional lymphaden o pathy and headache. No venom was detected at the bite site, nor in blood or urine. Case B: A 10-year-old girl (no snake was seen) had a negative result of a Venom Detection Kit test of a bite site, and urine and coagulation studies showed no abnormality, but she was given antivenom because of symptoms of tiredness and non-specific weakness. Case C: A nine-month-old girl who was crying and vomiting and noted to have a mark on her arm was given polyvalent antivenom before transfer to Princess Margaret Hospital. No venom detection tests were undertaken. Her symptoms of fever and intermittent vomiting were subsequently thought to be due to a viral illness. Case D: A six-year-old girl had tiger snake venom detected at a possible bite site (but no venom detected in blood or urine). She was given tiger snake antivenom for symptoms of headache, fever, abdominal pain and nausea. A groin abscess was noted the next day, which may possibly have been the cause of her illness. Other supportive treatment Three children were administered fresh frozen plasma. No child required artificial ventilation. Outcome All children were discharged well from hospital; no permanent morbidity was recorded. Discusssion As in the previous epidemiological studies of snakebite in children in Australia,1-3 most of the children who presented to hospital with suspected snakebite were not envenomed (Box 3). In our study the over-representation of toddlers in the children presenting, despite an even age distribution of children with envenomation, probably reflected heightened parental concern and the difficulty obtaining a history from this age group. Patterns of envenomation and treatment for snakebite in Perth children were similar to those reported previously in adults.4 The sex distribution of victims, the seasonal pattern of envenomation and the site of bites were also similar. The venoms of the two snake genera responsible for most bites in Perth, Pseudonaja (brown snake) and Notechis (tiger snake), have powerful procoagulants, and coagulopathy was present in most (71%) of the envenomed patients. Among Perth snakes, only the tiger snake produces myolysis, and only one patient in this series had rhabdomyo lysis. Despite local differences in fauna, these findings are very similar to those in children with snakebite in Victoria.3 Despite continuing confirmation of the great efficacy of the pressure-immobilisation first aid technique in delaying venom spread,6,7 it is disturbing that this method is still not being used widely. In many cases it is also being applied incorrectly. More public education on first aid in snakebite is needed. The role of the Venom Detection Kit in patients with suspected snakebite attending emergency departments is discussed in Box 4. Although it appeared that no child came to harm after discharge from the Emergency Department, this practice is fraught with danger, as a case from 1973 illustrates. A two-year-old girl with restlessness and dry retching presented to hospital with an unclear history; the possibility of snakebite was raised by the parents. The examining medical officer concluded that she had been bitten by an insect, no investigations were ordered and she was discharged. The next morning she was found dead in bed, and it was later confirmed that tiger snake envenomation had been the cause of death. The Coroner found that there had been a lack of care on the part of the hospital, contributing to her death.10 In most children presenting to an emergency department there is no way of determining absolutely whether a venomous snakebite has occurred. There are special difficulties in obtaining a reliable history from young children and it may be tempting to use a negative result of a Venom Detection Kit to facilitate early discharge. Indeed, a number of children in this study were discharged from the Emergency Department without laboratory investigations, apart from use of the Venom Detection Kit. However, all children in whom snakebite is suspected, whether a snake was seen or not, should be admitted to hospital for observation and investigation (including coagulation profiles). In a number of cases of envenomation in our series, no snake bite was observed, and in two cases no snake was seen. Discharging children with suspected snakebite directly from emergency departments without a period of observation is not recommended. References Munro JGC, Pearn JH. Snake bite in children. A five year population study from South-East Queensland. Aust Paediatr J 1978; 14: 248-253. Jamieson R, Pearn J. An epidemiological and clinical study of snake-bites in childhood. Med J Aust 1989; 150: 698-702. Tibballs J. Diagnosis and treatment of confirmed and suspected snake bite. Implications from an analysis of 46 paediatric cases. Med J Aust 1992; 156: 270-274. Jelinek GA, Hamilton T, Hirsch RL. Admissions for suspected snake bite to the Perth adult teaching hospitals, 1979 to 1988. Med J Aust 1991; 155: 761-764. Microsoft Access [database]. Version 2.0. Redmond, WA: Microsoft Corporation, 1989-1994. Howarth DM, Southee AE, Whyte IM. Lymphatic flow rates and first-aid in simulated peripheral snake or spider envenomation. Med J Aust 1994; 161: 695-700. Sutherland SK. The pressure immobilisation technique. Med J Aust 1994; 161: 700-701. CSL Diagnostics. Snake venom detection kit. Product information leaflet. Melbourne: CSL Diagnostics, 1992. Wentworth B, Moisidis A, Jones S. Performance of the new snake venom detection kit at high and low concentrations [abstract]. Proceedings of the 22nd Annual Scientific Meeting of the Australasian Society of Immunology; 1992 Dec 2-4; Auckland. Parkville: Australasian Society of Immunology, 1992. Snake and lack of care killed girl. The West Australian 1973 19 Jun: 3 (col. 3, 4 and 5). (Received 6 Jul, accepted 20 Dec 1995) Authors' details Department of Emergency Medicine, Fremantle Hospital, Fremantle, WA. Helen J Mead, FRACP, FACEM, Staff Specialist; formerly, Senior Registrar, Intensive Care Unit, Princess Margaret Hospital for Children, Perth, WA; George A Jelinek, MD, FACEM, Assistant Director and Staff Specialist. No reprints will be available. Correspondence: Dr George A Jelinek, Department of Emergency Medicine, Fremantle Hospital, PO Box 480, Fremantle, WA 6160. ©MJA 1997 <URL: http://www.mja.com.au/> © 1997 Medical Journal of Australia.

Helen J Mead · George A Jelinek

Teaching hospital medical staff to handwashJames Tibballs Electronically published Tuesday April 30 1996 This article was published in the Medical Journal of Australia on 1 April 1996. Readers may print a single copy for personal use. No further reproduction or distribution of the article in whole or in part should proceed without the permission of the publisher. For copyright permission, contact the Australasian Medical Publishing Company Abstract - Introduction - Methods - Results - Group performance - Cohort performance - Discussion - Author's Details - Figure - Table 1 - Table 2 - - ©MJA1996 AbstractObjective: To increase the frequency of handwashing by medical staff. Design: A prospective study of handwashing before and after patient contact. Setting: A paediatric intensive care unit in a tertiary hospital. Participants: 61 intensive care unit medical staff and visiting medical staff. Interventions: A five-phase behaviour modification program: (i) unobtrusive observation for four weeks to obtain a baseline handwashing rate; (ii) overt observation for five weeks (preceded by written advice); (iii) overt observation continued for four weeks with performance feedback; (iv) all observation and feedback discontinued for seven weeks; and (v) unobtrusive observation for five weeks to obtain a residual rate. Results: 939 patient contacts were observed. The baseline handwashing rates before and after patient contact were 12.4% and 10.6%, respectively. During overt observation, the respective rates increased and plateaued at 32.7% and 33.3%, but increased further (to 68.3% and 64.8%) during the period of performance feedback. The residual handwashing rates, observed unobtrusively seven weeks after the cessation of performance feedback, were 54.6% before and 54.9% after patient contact. Conclusions: Performance feedback is moderately effective in training hospital medical staff to handwash. MJA 1996; 164: 395-398 IntroductionHandwashing is one of the most important measures for preventing and controlling nosocomial infection.1-4 Despite this, handwashing practices before and after patient contact by all hospital staff, and especially by physicians, are poor.5-9 Educational programs to improve handwashing by increasing awareness of nosocomial infection have had limited success,10-12 suggesting that an innovative approach is needed.13 A training program for nurses with observation and performance feedback as a behaviour modification technique was successful in the short term, but the relative effects of observation and performance feedback were inseparable.14 My study assessed the frequency of handwashing by medical officers in association with patient contact and determined the effects of observation and performance feedback. MethodsThe study was conducted over a five-month period in the Intensive Care Unit (ICU), Royal Children's Hospital, Melbourne, and was approved by the hospital ethics committee. The ICU is an 18-bed multidisciplinary unit treating critically ill infants and children. During the study, 807 patients were admitted: 37% had undergone cardiac surgery, 27% had respiratory illnesses, 9% had general medical illnesses, 8% had newborn medical and surgical illnesses, 8% had trauma, and 8% had undergone neurosurgery and 3% general surgery. Mechanical ventilation was provided for 84% of patients. Most had invasive monitoring or therapeutic devices inserted into peripheral arteries, peripheral or central veins, the thoracic cavity, the urinary bladder or the extradural space. The mortality rate was 5.9%. The subjects of the study were any medical officers who made hand contact with any patient in the ICU. The ICU is newly constructed and comprises four rooms adjacent to each other along a corridor. Windows enable vision between rooms and into each room from the corridor. Handwashing basins are located in each room within two to three metres of each bed and in the corridor. A choice of handwashing solutions are provided at each basin. During the study these were: (i) a mixture of 1% chlorhexidine gluconate, 70% ethyl alcohol and emollients; (ii) 2% aqueous chlorhexidine gluconate; (iii) 7.5% aqueous povidone-iodine; and (iv) domestic bar soap. Paper-towel dispensers are located at each basin. I observed patient contact and handwashing practices of medical officers during ward rounds; at other times this observation was by instructed members of the nursing staff. Observations were made between the hours of 7 am and 6 pm on weekdays. Handwashing was regarded as any action to cleanse the hands with water and one of the available handwashing agents, or with a handwashing agent alone. No judgement was made about the duration or efficacy of the handwashing technique. Observations were recorded with the names of subjects and the patients involved. However, anonymity was preserved for data analysis and feedback purposes. Patient contact was defined as any contact by the hands of a medical officer with a patient's skin or an indwelling invasive device during a procedure in which the hands could be contaminated with the patient's blood, secretions or excretions. Contact with bed linen, bed frames, medical records or monitoring equipment was not regarded as patient contact. Typical patient contact involved procedures such as examining wounds, inserting intravenous or arterial catheters, aspirating endotracheal tubes, performing echocardiographic examinations, auscultating the chest, palpating the chest or abdomen, manipulating chest drains or urinary catheters, inserting tracheal or gastric tubes, and withdrawing blood from intravascular catheters. Handwashing was assessed before and after contact with a patient. Handwashing was not recorded if the subject was not observed immediately before patient contact, or could not be observed after contact until handwashing had been performed, or contact with another patient had occurred. Departure from the ICU area without handwashing was regarded as failure to handwash. Contact by the subject with his or her own skin, secretions or a mucous membrane (e.g., nasal mucosa) after handwashing but before patient contact was registered as failure to handwash. Medical officers were not requested to handwash, and the ICU did not have any information about handwashing displayed (other than the performance feedback when this part of the study was reached). The behaviour modification program was divided into five phases1. Weeks 1-4: unobtrusive observation (only hospital personnel involved in the study knew this was happening). At the end of this phase, 19 medical officers who worked in the ICU or visited frequently were requested verbally to estimate their own handwashing rates before patient contact. 2. Weeks 5-9: overt observation; this was preceded by written advice to all medical staff of the intention to study handwashing in ICU. Subjects were not informed which aspect of handwashing was to be studied, although, if requested, information was not to be withheld (no medical staff requested clarification). 3. Weeks 10-13: overt observation with performance feedback; at the beginning of this phase all results collected thus far were displayed as brightly coloured histograms on 40 x 60 cm sheets of paper showing the group percentage handwashing rate before and after patient contact. These were placed in strategic locations used by the medical staff in ICU, but were out of sight of patients and visitors. Data were updated weekly for the four weeks. After two weeks of performance feedback, results of group data collected up to that time were mailed to the medical officers. 4. Weeks 14-20: observations and feedback were discontinued. 5. Weeks 21-25: unobtrusive observation, after which these data were added to the histograms displayed in the ICU. Handwashing by members of the cohort who were requested to estimate their own handwashing rate and who participated in all phases of the program was analysed separately and is also reported here. ResultsGroup performanceSixty-one medical officers were observed making 939 patient contacts. Twelve were full-time senior or junior ICU staff, and the remainder full-time or part-time senior staff from other disciplines or junior full-time staff. Half the observations (470/939) were made by the investigator, and half by nursing staff. The aggregate handwashing rates before and after patient contact during phases of the program are presented in The aggregate handwashing rates before and after patient contact during phases of the program are presented in Table 1. The week-to-week variation within each phase and between each phase is shown in the Figure. During overt observation the handwashing rate rose from the baseline level (determined by unobtrusive observation), with the peak rate occurring in Week 3 of this five-week period. It rose again during performance feedback, peaking in the last week of this phase. During the second period of unobtrusive observation, seven weeks after performance feedback, the handwashing rate was lower but still showed a more than fourfold increase over baseline levels. There was a gradual decline from week to week during this last phase of unobtrusive observation. Cohort performanceNineteen medical officers estimated their own handwashing rate before patient contact after the first phase of covert observation. Their mean estimate was 66% (range, 30%-95%). Several commented that other medical officers washed their hands infrequently. Fourteen of these 19 subjects subsequently participated in the remain- ing phases of the program. Among this cohort, the mean self-estimated handwashing rate was 73% (range, 50%-95%), compared with the covertly observed rate for this cohort of 8.6% before and 10.8% after patient contact, with an individual mean rate of 10% (range, 0-33%). The performance of the cohort during the other phases of the program is shown in Table 2. Significant improvement in handwashing before and after patient contact was observed during overt observation and performance feedback. The improved performance of these 14 subjects was maintained after the period without observation. DiscussionPatients in intensive care units are at greater risk of nosocomial infection than those elsewhere in hospitals: they are already critically ill; are subject to multiple hand contacts from a wide variety of staff; have indwelling invasive therapeutic and monitoring devices; and often receive non-specific antimicrobial therapy. Reports of nosocomial infection in intensive care units show infection rates varying from 13% to 26%,15-19 whereas hospital-wide nosocomial infection rates are about 5%-10%.20-23 Physicians freely acknowledge that handwashing is important in the control of nosocomial infection, and suggest pressure of work, a lack of patient contact, detrimental effects of handwashing agents on skin and the more important needs of the patient as reasons for not handwashing.4 Infection control is rarely taught in medical school,24 and poor handwashing practices may be learnt from peers at the bedside.25 In the first phase of this study, I found the handwashing rate among medical staff to be lower than rates reported previously (13%-28%,5 21%6 and 15%7). Physicians were observed unobtrusively as it is known that observation alone may alter behaviour, a phenomenon otherwise known as the "Hawthorne effect" (after the Hawthorne works of the Western Electric Company, where job performance was improved by the presence of observers).26 This effect was confirmed firstly by the abrupt increase in handwashing which accompanied notification of my intention to observe handwashing and again during the performance feedback phase. Despite the relative lack of handwashing in relation to patient contact by hospital staff and its possible detrimental consequences, few reports of attempts to improve handwashing have been published. "Educational programs" which aim to increase medical staffs' awareness of nosocomial infection and the importance of handwashing have not been successful. In an intensive care unit the handwashing rate among physicians increased to only 17% after an educational program (in-service teaching rounds, poster displays, specific requests to handwash and unspecified performance feedback to all staff).10 Another study, in a paediatric ambulatory setting, found that physicians did not respond to written and verbal requests to handwash before patient contact.27 Moreover, my study showed that medical staff have an inflated impression of their own handwashing performance, which may possibly prejudice learning. Educational programs to teach nurses to handwash have also been unsuccessful. The failure of one consisting of poster displays and the distribution of booklets on isolation policy was attributed to a lack of motivation.11 Another strategy to improve handwashing is the installation of handwashing machines. This was associated with improvement over a short period in a small group of nurses and physicians, but it was abandoned when there was an outbreak of methicillin-resistant Staphylococcus aureus.28 Several "training programs" to increase handwashing have been more successful than educational awareness programs. A program for nurses consisted of observation and daily feedback of handwashing performance from the previous day; the handwashing rate after patient contact rose from 59% to 92% during a three-week period.14 However, the relative effects of observation and performance feedback could not be separated. Among food handlers in a university cafeteria, performance feedback alone was successful in training to handwash.29 Performance feedback is an intervention frequently used in organisational behaviour management to improve performance.30 In my study, it resulted in an increase in handwashing among medical officers which was sustained at a rate six times the baseline in about two-thirds of contacts with patients. Regular programs are required to maintain an acceptable standard of performance among permanent staff10 and to compensate for departure and entry of staff members. In my study the difference in handwashing performance between the whole group of medical officers and a cohort within the group may be attributed to staff turnover outside the cohort or, alternatively, to greater exposure to the program, or may be a consequence of being asked to estimate their own handwashing rate. The cohort achieved and maintained a higher handwashing rate than the group as a whole. For reasons of economy and efficacy, behaviour intervention rather than educational awareness programs appears preferable in educating medical staff to handwash. Acknowledgements I thank Dr Gerald Elsworth, formerly of the Centre for the Study of Higher Education, University of Melbourne, for helpful suggestions concerning the design of the study. I thank the ICU nurses, particularly Sharon Kinney. (©MJA 1996; 164: 14-17) To beginning of this article ReferencesDaschner F. Useful and useless hygienic techniques in intensive care units. Intensive Care Med 1985; 11: 280-283. Davies PA. Please wash your hands. Arch Dis Child 1982; 57: 647-648. Steere AC, Mallison GF. Handwashing practices for the prevention of nosocomial infections. Ann Intern Med 1975; 83: 683-690. Larson E, Killien M. Factors influencing handwashing behaviour of patient care personnel. Am J Infect Control 1982; 10: 93-99. Albert RK, Condie F. Hand-washing patterns in medical intensive care units. N Engl J Med 1981; 304: 1465-1466. Donowitz LG. Handwashing technique in a pediatric intensive care unit. Am J Dis Child 1987; 141: 683-685. De Carvalho M, Lopes JMA, Pellitteri M. Frequency and duration of handwashing in a neonatal intensive care unit. Pediatr Infect Dis J 1989; 8: 179-180. Larson E. Compliance with isolation technique. Am J Infect Control 1983; 11: 221-225. Meengs MR, Giles BK, Chisholm CD, et al. Hand washing frequency in an emergency department. Ann Emerg Med 1994; 23: 1307-1312. Conly JM, Hill S, Ross J, et al. Handwashing practices in an intensive care unit: the effects of an educational program and its relationship to infection rates. Am J Infect Control 1989; 17: 330-339. Williams E, Buckles A. A lack of motivation. Nursing Times 1988; 84: 60-64. Doebbeling BN, Stanley GL, Sheetz CT, et al. Comparative efficacy of alternative hand-washing agents in reducing nosocomial infections in intensive care units. N Engl J Med 1992; 327: 88-93. Goldmann D, Larson E. Hand-washing and nosocomial infections. N Engl J Med 1992; 327: 120-122. Mayer JA, Dubbert PM, Miller M, et al. Increasing handwashing in an intensive care unit. Infect Control 1986; 7: 259-262. Daschner FD, Frey P, Wolff G, et al. Nosocomial infections in intensive care wards: a multicenter prospective study. Intensive Care Med 1982; 8: 5-9. Donowitz LG, Wenzel RP, Hoyt JW. High risk of hospital-acquired infection in the ICU patient. Crit Care Med 1982; 10: 355-357. Potgieter PD, Linton DM, Oliver SO, Forder AA. Nosocomial infections in a respiratory intensive care unit. Crit Care Med 1987; 15: 495-498. Donowitz LG. High risk of nosocomial infection in the pediatric critical care patient. Crit Care Med 1986; 14: 26-28. Craig CB, Connelly S. Effect of intensive care unit nosocomial pneumonia on duration of stay and mortality. Am J Infect Control 1984; 12: 233-238. McLaws M-L, Gold J, King K, et al. The prevalence of nosocomial and community-acquired infections in Australian hospitals. Med J Aust 1988; 149: 582-590. Meers PD, Ayliffe GAJ, Emmerson AM, et al. Report on the national survey of infection in hospitals, 1980. J Hosp Infect 1981; 2(Suppl): 13-17. Haley RW, Culver DH, White JW, et al. The nationwide nosocomial infection rate. A new need for vital statistics. Am J Epidemiol 1985; 121: 159-167. Mayon-White RT, Ducel G, Kereselidze T, Tikomirov E. An international survey of the prevalence of hospital-acquired infection. J Hosp Infect 1988; 11(Suppl A): 43-48. Moss F, Cochrane JPS, Yudkin JS. Introducing medicine to tomorrow's doctors. Lancet 1987; 1: 203-205. Larson E, McGinley KJ, Grove GL, et al. Physiologic, microbiologic and seasonal effects on the skin of health care personnel. Am J Infect Control 1986; 14: 51-59. Roethlisberger FJ, Dickson WJ. Management and the worker. Cambridge, Mass: Harvard University Press, 1942. Lohr JA, Ingram DL, Dudley SM, et al. Handwashing in paediatric ambulatory settings. Am J Dis Child 1991; 145: 1198-1199. Wurtz R, Moye G, Jovanovic B. Handwashing machines, handwashing compliance and potential for cross-contamination. Am J Infect Control 1994; 22: 228-230. Geller ES, Eason SL, Phillips JA, Pierson MD. Interventions to improve sanitation during food preparation. J Organizational Behav Manage 1980; 2: 229-240. Prue DM, Fairbank JA. Performance feedback in organizational behavior management: a review. J Organizational Behav Manage 1981; 3: 1-16. (Received 26 Jun, accepted 5 Dec 1995) Author's detailsIntensive Care Unit, Royal Children's Hospital, Melbourne, VIC. James Tibballs, MEd, FFICANZCA, Deputy Director. Reprints: Dr J Tibballs, Intensive Care Unit, Royal Children's Hospital, Flemington Road, Parkville, VIC 3052. E-mail: tibballjATcryptic.rch.unimelb.edu.au (©MJA 1996; 164: 14-17) < URL: http://www.mja.com.au/> © 1996 Medical Journal of Australia.

James Tibballs

AbstractObjective: To assess the effectiveness of the flashlamp-pumped pulsed dye laser in the treatment of port-wine stains. Design: A retrospective review of medical records and patients. Setting: Royal Perth Hospital (a tertiary referral hospital), August 1989 to December 1992. Subjects: 186 consecutive patients with port-wine stains treated with a flashlamp-pumped pulsed dye laser. Outcome measures: Degree of lesion fading; adverse reactions. Results: Of 131 patients who completed treatment, 78% had better than 50% fading of the lesion and only 9% had less than 25% fading. An average 3.4 treatments were needed to achieve more than 50% fading. The response was better in children than in adults, although the difference was not significant. Anaesthesia was needed for 44% of patients. Pigmentary change (usually transient) occurred in 6.1% of patients and permanent and significant adverse effects in only 4.6%. Conclusion: This study confirms the efficacy of the flashlamp-pumped pulsed dye laser in the treatment of port-wine stains in children and adults. Early treatment of port-wine stains should be encouraged to reduce the physical and psychological morbidity of disfiguring lesions. IntroductionPort-wine stains (now termed capillary malformations) are congenital vascular malformations characterised by ectatic vessels within the cutaneous superficial vascular plexus.1 They affect 0.3% of children at birth, with an equal sex distribution.2 Most port-wine stains are found on the head and neck, with 85% occurring in a unilateral, dermatomal distribution.3 Their natural history is to progress from a pink, macular lesion at birth to a dark red (or even purple), nodular, proportionately larger lesion in adulthood.1,4 Cosmetic disfigurement can cause considerable psychological morbidity, and be socially disadvantageous.5,6 Hypertrophic, nodular lesions may bleed, either spontaneously or after trauma, encroach on essential facial structures or even develop into pyogenic granulomas.7 All port-wine stains should be treated, preferably early in life, to prevent or reduce the potential physical and psychological complications. Before the development of laser technology, treatment for port-wine stains was often unsatisfactory.8 Earlier lasers, including ruby, carbon dioxide and argon lasers, improved the lesions in most patients, but, as they were non-selective in their effects on tissue, the frequency of side effects, such as scarring and pigmentary changes, was unacceptably high.9 The flashlamp-pumped pulsed dye laser (PDL) was the first laser to be based on selective photothermolysis;10 it produces vascular- specific damage without affecting surrounding dermal structures or the epidermis.11-13 Studies have confirmed its effectiveness in the treatment of port-wine stains in adults, children and infants, with an extraordinarily low incidence of side effects.14-19 The PDL is now regarded as the first-line treatment for port-wine stains whenever possible.8,9,20-22 In Australia, the PDL has been in use since the late 1980s for the treatment of cutaneous vascular lesions, especially port-wine stains. However, no Australian study of its effects has been published. Therefore, we undertook a retrospective clinical review of all patients with port-wine stains treated with PDL at Royal Perth Hospital between August 1989 and December 1992. MethodsThe study was a retrospective review of patient medical records. If response to treatment was not recorded, attempts were made to review the patient between January and December 1993. PatientsAll patients (adult and paediatric) with a port-wine stain treated with PDL at Royal Perth Hospital between August 1989 and December 1992 were eligible. Patients attended outpatient clinics, where the site and size of the port-wine stain and demographic data were recorded and the treatment procedure and its risks and benefits were explained. Patients were photographed before treatment by a professional photographer in a studio dedicated to medical photography, with efforts to use the same magnification, lighting and exposure. Laser and techniqueA flashlamp-pumped pulsed dye laser (Candela SPTL-1, Candela Corp, Wayland, Mass, United States) was used. It emitted yellow light at a wavelength of 585 nm, with a pulse duration of 450 µs and a 3 s pause between pulses. The laser beam was transmitted down a 1 mm fibre by a planoconvex lens and focused as a 5 mm spot beam. Energy densities were measured by an energy meter (Ophir, Jerusalem, Israel), calibrated to 10% accuracy. Both the physicians and the patients eyes were protected from laser light during treatment. Some patients had a small initial test patch treated, depending on patient anxiety and time of presentation (before 1991, most had a patch test). Otherwise, the entire lesion was treated at once, unless it involved a large surface area (> 100 cm2). Treatments were repeated at intervals of 23 months. The energy density used varied with the age of the patient and colour, nodularity and location of the lesion and was adjusted according to the degree of purpura produced and the patients response to the previous treatment. Pulses were overlapped by a maximum of 10% across the affected area. The anaesthetic varied according to the site and area to be treated and the level of patient cooperation. EMLA cream (eutectic mixture of 2.5% lignocaine and 2.5% prilocaine cream, Astra Pharmaceuticals, North Ryde, NSW) was used for topical anaesthesia, applied under occlusion for 60120 minutes before treatment. Local anaesthesia involved an injection of 1% lignocaine, either locally or as a regional nerve block. General anaesthesia was given to children who had extensive lesions or were uncooperative with topical or local anaesthesia. The treated area developed purpura within a few minutes, usually persisting for 710 days. No immediate postoperative care was necessary, except for an occasional ice pack to reduce oedema in those with large treatment areas. Postoperative instructions were to protect the area from trauma, avoid excessive exposure to sunlight and use a topical antiseptic cream for any scaling or crusting. Treatment evaluationEach port-wine stain was evaluated, either before the next treatment or 34 months after the final treatment. Lesional lightening was assessed as the percentage reduction in colour compared with the pretreatment photo (fading < 25%, poor; 25% to 50%, fair; > 50% to 75%, good; and > 75%, excellent) (Figures 1-4). Adverse effects, such as scarring and textural or pigmentary changes, were also noted. All patients were individually assessed by one or both investigators. The endpoint of treatment was assessed clinically. Data analysisData were analysed with the Statistical Analysis Systems software package.23 The chi-squared statistic was used to assess the difference in response between age groups. ResultsThere were 186 patients treated by PDL: 131 completed treatment (55 either did not complete treatment or were having ongoing treatment). Patients were either Caucasian or Asian and aged 8 months to 66 years (mean, 25.6 years). There were 59 males (32%) and 127 females (68%). Most of the treated lesions were present from birth (97%). Acquired lesions appeared most commonly between the ages of six and 12. Most lesions were on the face and neck (87%), with the rest distributed unilaterally on the arms (4%), legs (5%), back (2%) and chest (2%). The size of treated lesions ranged from 1 cm2 to 280 cm2 (mean, 42 cm2). All responded to energy fluences between 5 and 10 joules/cm2(mean, 6.7 joules/cm2). Sixty-two per cent of the patients had a patch test before treatment. Anaesthesia was used for 44% of patients (general anaesthesia by 20%, topical by 19% and local or regional block by 5%). Responses of patients who completed treatment are shown in the Box. A good-to-excellent response was achieved in 78% and a poor response in only 9%. An average 3.4 treatments per lesion were required to achieve a good-to-excellent response. Adverse side effects occurred in 11% of patients who completed treatment; all had some fading of the lesion. The most common adverse effect was pigmentary change (6.1%), which was usually transient and resolved in 23 months. Only 4.6% had significant permanent adverse effects; two had scarring (in both the port-wine stain was on the face and neck region). More children than adults had a good or excellent response, but the difference was not significant when compared with a 2 x 2 contingency table and chi-squared test (r = 0.60). Similarly, fewer children than adults had a poor response. DiscussionOur results compare favourably with those of other studies. A good-to-excellent response (more than 50% fading) was obtained in 78% of our patients (including both adults and children, with lesions on sites including the trunk and lower limbs), with an average 3.4 treatments required. Others have found more than 50% fading in 73%-95% of patients after 2.4-2.8 treatments.14-17,24,25 Response to treatment varies between sites: the periorbital area, temple, lateral aspect of the cheek, neck and chin have been observed to be more responsive18,25 and the centrofacial area and lower leg to be less responsive.18,19 We found that a poor response was more common in adults than in children (although the difference was not significant), possibly because port-wine stains become progressively hypertrophied and nodular in adults. We found a higher rate of adverse effects (11%) than in other studies. The most common (usually transient) adverse effect was pigmentary change (increase or decrease), possibly because of excessive sunlight exposure after treatment. This transient change may not have been recorded in other studies; when it was excluded from our figures, the rate of adverse reactions was reduced to less than 5%, which is comparable with that found in other studies. Scarring was seen in two of our patients, with lesions on the face and neck, where damage to dermal structures with fibrosis occurs when excessive energy fluence is used. A low energy fluence should be used initially when treating port-wine stains on the neck and anterior chest. The PDL is the first laser specifically designed for cutaneous vascular malformations. It is based on the theory of selective photothermolysis, which predicts selective destruction of blood vessels without damage to the surrounding tissues.10 Laser light emitted by the PDL is absorbed by oxyhaemoglobin in the dilated vessels of the lesion, producing agglutination of erythrocytes, thrombus formation and eventual destruction of the vessels.11 They are replaced by non-dilated superficial dermal blood vessels with a normal appearance.12 A recent comparison of PDL and the copper vapour laser showed that PDL produced significantly better fading of port-wine stains.22 The characteristics and degree of pain associated with PDL treatment have been well described.26 Initially, there is a sharp stinging pain, very similar to the snap of a rubber band against the skin. Accompanying this is a second distinct heat sensation that can be at least as unpleasant as the initial sting. Pain rapidly subsides but seems to build up if successive pulses are used for a moderately sized lesion. Our current practice is to give general anaesthesia to all children from four weeks of age, until they are able to co-operate with topical or local anaesthesia, usually at eight to 10 years. Young children undergoing multiple painful treatments with inadequate anaesthesia under restraint may develop phobic responses. Furthermore, a struggling child may compromise the clinicians ability to perform the procedure optimally. In conclusion, this study supports the contention that all port-wine stains should be treated with PDL, as it has a high therapeutic index with a low incidence of adverse effects. Patients should preferably be treated in infancy or childhood, under general anaesthesia, to minimise the potential psychological morbidity of disfiguring lesions. In addition, the response to treatment seems better in children than in adults, although the difference was not significant, possibly because of the relatively small sample size. Laser treatment of port-wine stains should no longer be considered just cosmetic, but a medical necessity for a problem that can cause psychological and physical morbidity. References Mulliken JB. Capillary (port-wine) and other telangiectatic stains. In: Mulliken JB, Young AE, editors. Vascular birthmarks -- haemangiomas and malformations. Philadelphia: W B Saunders, 1988: 179-195. Jacobs AH, Walton RG. The incidence of birthmarks in the neonate. Pediatrics 1976; 58: 218-222. Tallman B, Tan OT, Morelli JG, et al. Location of port-wine stains and the likelihood of ophthalmic and/or central nervous system complications. Pediatrics 1991; 87: 323-327. Barsky SH, Rosen S, Geer DE, Noe JM. The nature and evolution of port-wine stains: a computer-assisted study. J Invest Dermatol 1980; 74: 154-157. Lanigan SW, Cotterill JA. Psychological disabilities amongst patients with port wine stains. Br J Dermatol 1989; 121: 209-215. Pickering JW, Butler PH, Ring BJ, Walker EP. Copper vapour laser treatment of port wine stains: a patient questionnaire. Lasers Med Sci 1990; 5: 43-49. Geronemus RG, Ashinoff R. The medical necessity of evaluation and treatment of port-wine stains. J Dermatol Surg Oncol 1991; 17: 76-79. Wheeland RG. Treatment of port-wine stains for the 1990s. J Dermatol Surg Oncol 1993; 19: 348-356. Van Gemert MJ, Carruth JA, Shakespeare PG. Laser treatment of the port-wine stains. BMJ 1993; 306: 4-5. Anderson RR, Parrish JA. Selective photothermolysis: precise microsurgery by selective absorption of pulsed radiation. Science 1983; 220: 524-527. Nakagawa H, Tan OT, Parrish JA. Ultrastructural changes in human skin after exposure to a pulsed laser. J Invest Dermatol 1985; 84: 396-400. Morelli JG, Tan OT, Garden J, et al. Tunable dye laser (577 nm) treatment of port wine stains. Lasers Surg Med 1986; 6: 94-96. Tan OT, Morrison P, Kurban AK. 585 nm for the treatment of port-wine stains. Plast Reconstr Surg 1990; 86: 1112-1117. Garden JM, Polla LL, Tan OT. The treatment of port-wine stains by the pulsed dye laser: analysis of pulse duration and long-term therapy. Arch Dermatol 1988; 124: 889-896. Tan OT, Sherwood K, Gilchrest BA. Treatment of children with port-wine stains using the flashlamp-pulsed tunable dye laser. N Engl J Med 1989; 320: 416-421. Reyes BA, Geronemus RG. Treatment of port-wine stains during childhood with the flashlamp-pumped pulsed dye laser. J Am Acad Dermatol 1990; 23: 1142-1148. Ashinoff R, Geronemus RG. Flashlamp-pumped pulsed dye laser for port-wine stains in infancy: earlier versus later treatment. J Am Acad Dermatol 1991; 24: 467-472. Renfro L, Geronemus RG. Anatomical differences of port-wine stains in response to treatment with the pulsed dye laser. Arch Dermatol 1993; 129: 182-188. Garden JM, Bakus AD. Clinical efficacy of pulsed dye laser in the treatment of vascular lesions. J Dermatol Surg Oncol 1993; 19: 321-326. Geronemus RG. Pulsed dye laser treatment of vascular lesions for children. J Dermatol Surg Oncol 1993; 19: 303-310. Hruza GJ, Geronemus RG, Dover JS, Arndt KA. Lasers in dermatology Ñ 1993. Arch Dermatol 1993; 129: 1026-1035. Sheehan-Dare RA, Cotterill JA. Copper vapour laser (578 nm) and flashlamp-pumped pulsed tunable dye laser (585 nm) treatment of port-wine stains: results of a comparative study using test sites. Br J Dermatol 1994; 130: 478-482. Statistical Analysis Systems [computer program], version 6.08. Cary (NC): SAS Institute Inc, 1994. Goldman MP, Fitzpatrick RE, Ruiz-Esparza J. Treatment of port-wine stains (capillary malformation) with the flashlamp-pumped pulsed dye laser. J Pediatr 1993; 122: 71-77. Holy A, Geronemus RG. Treatment of periorbital port-wine stains with the flashlamp-pumped pulsed dye laser. Arch Ophthalmol 1992; 110: 793-797. Rabinowitz LG, Esterly NB, Frieden IJ, et al. Anesthesia and/or sedation for pulsed dye laser therapy. Pediatr Dermatol 1992; 9: 132-153. Authors detailsDepartment of Dermatology, Royal Perth Hospital, Perth, WA. Ernest Tan, MB BS, Dermatology Registrar; Carl Vinciullo, FACD, Visiting Dermatologist. No reprints will be available. Correspondence: Dr C Vinciullo, Department of Dermatology, Royal Perth Hospital, Wellington Street, WA 6000. E-mail: carlATdermlaser.com.au

Ernest Tan · Carl Vinciullo

Passive smoking and respiratory function in very low birthweight children Lex W Doyle, Geoffrey W Ford, Anthony Olinsky, Annette M L Knoches and Catherine Callanan For editorial comment see Woodward & Jamrozik Readers may print a single copy for personal use. No further reproduction or distribution of the articles should proceed without the permission of the publisher. For permission, contact the Australasian Medical Publishing Company 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/>". Abstract - Introduction - Methods - Results - Discussion - Acknowledgement - References - Authors' details - ©MJA1997 Abstract Aim: To determine if an adverse relationship exists between passive smoking and respiratory function in very low birthweight (VLBW) children at 11 years of age. Setting: The Royal Women's Hospital, Melbourne. Patients: 154 consecutive surviving children of less than 1501 g birthweight born during the 18 months from 1 October 1980. Methods: Respiratory function of 120 of the 154 children (77.9%) at 11 years of age was measured. Exposure to passive smoking was established by history; no children were known to be actively smoking. The relationships between various respiratory function variables and the estimated number of cigarettes smoked by household members per day were analysed by linear regression. Results: Most respiratory function variables reflecting airflow were significantly diminished with increasing exposure to passive smoking. In addition, variables indicative of air-trapping rose significantly with increasing exposure to passive smoking. Conclusions: Passive smoking is associated with adverse respiratory function in surviving VLBW children at 11 years of age. Continued exposure to passive smoking, or active smoking, beyond 11 years may lead to further deterioration in respiratory function in these children. MJA 1996; 164: 266-269 Introduction Passive smoking is associated with several adverse health outcomes in children, including higher rates of asthma,1 and infections of the upper2 and lower3 respiratory tract. Further, respiratory function is reduced with passive smoking in children who have no lung disease,4 as well as those with lung diseases such as asthma5 and cystic fibrosis.6 To survive the neonatal period, many very low birthweight (VLBW) children (less than 1500 g at birth) require prolonged periods of assisted ventilation, and some may develop bronchopulmonary dysplasia (BPD) and suffer from ongoing respiratory problems as a consequence. We have previously reported the respiratory health to eight years of age of cohorts of children of birthweight 500-999 g (n = 83), 1000-1500 g (n = 114) and > > 2500 g (n = 51).7 Passive smoking was significantly related to the duration of hospitalisation for respiratory problems up to two years of age for all children in that study, but was not associated with changes in respiratory function at eight years of age. In contrast, in another recent cohort study of respiratory function at seven years of age in children of birthweight less than 2000 g Chan et al.8 reported reduced air-flow rates with maternal smoking, but not with smoking by other household members. Because the effects of passive smoking could increase with increasing duration of exposure, the aim of this study was to determine if an adverse relationship exists between passive smoking and respiratory function at 11 years of age in VLBW children. Methods We studied 154 consecutive surviving children of less than 1501 g birthweight born during the 18 months from 1 October 1980 at the Royal Women's Hospital, Melbourne, the largest of the three tertiary-level perinatal centres in Victoria. Details of the survival rate and early neonatal care of this cohort have been described.9,10 Bronchopulmonary dysplasia (BPD) was diagnosed in children who had required intermittent positive pressure ventilation in the neonatal period, who had respiratory distress and were still having oxygen therapy at 28 days of age, and who had an abnormal chest x-ray consistent with stage III or IV disease (as defined by Northway et al.11 ) at or after 28 days. A previous report of the respiratory function of this cohort at eight years of age7 included data for some children with birthweights of less than 1000 g born before 1 October 1980. We did not have the resources to measure respiratory function at 11 years of age of the children born before October 1980. Respiratory health was determined by history and examination, and measurement of respiratory function. Children who had required bronchodilators within the previous year for attacks of wheezing were considered to have asthma. Data on passive smoking were obtained by asking the parents about the daily consumption of cigarettes by members of the household. We did not distinguish between mothers and other smokers, or between smoking inside or outside the home. Some data on maternal smoking in pregnancy had been collected in the perinatal period, but were obtained for only one-third of mothers. Children were questioned about active smoking in their parents' absence. As some children had changed households in their lifetimes, we considered those who had lived in any household with smokers over the 11-year period to have been passively smoking during childhood. Two categories of social class were determined -- unskilled or unemployed, and other (professional, skilled or semi-skilled) -- based on the occupation of the family breadwinner. Respiratory function was measured in the Department of Thoracic Medicine at the Royal Children's Hospital, Melbourne, as described previously,7 by personnel blinded to the exposure of individual children to passive smoking. Maximum expiratory flow rates were recorded with a pneumotachograph (Fleisch No. 3, Switzerland) and plotted against volume by integrating flow on an X-Y recorder to obtain flow-volume loops. Maximum flow rates at 75% (VEmax75%), 50% (VEmax50%) and 25% (VEmax25%) of forced vital capacity (FVC), and forced expiratory flow between 25% and 75% of FVC (FEF25%-75%), were measured from the loops. Flow rates were corrected for body size by dividing by vital capacity (VC). Vital capacity, FVC and forced expiratory volume in one second (FEV1) were measured with a water-filled spirometer (Godart Expirograph, Bilthoven, Netherlands) in accordance with standard guidelines, and results at body temperature and pressure saturated with water vapour were expressed as a percentage of the predicted value for age, height and sex.12 Total lung capacity (TLC) and residual volume (RV) were measured in a body plethysmograph (Jaeger Bodyscreen 2, Wurzburg, Germany). Children were not subjected to bronchial provocation tests as these are poorly tolerated, and we were eager to maintain a high degree of cooperation with these and with future respiratory function tests. Not all children could complete all respiratory function tests, either because of poor cooperation, or unavailability or malfunction of equipment on the day of testing. Data were edited and analysed using SPSS.13 Dichotomous variables were contrasted by chi-squared analysis, and continuous variables by t test, or Mann-Whitney U test if the data were skewed. The dose-response relationship between the estimated daily number of cigarettes consumed by members of the household and various respiratory function variables was established by linear regression; linear and quadratic relationships were tested. Data were then analysed by linear regression to adjust for the potentially confounding variables of birthweight, gestational age, birthweight ratio (child's birthweight divided by median birthweight for gestational age14 ), sex, BPD, and asthma; all variables were entered simultaneously, even if they were not statistically significant. Durations of intermittent positive pressure ventilation and oxygen therapy were not included as they were strongly related to BPD. For all analyses, P values of less than 0.05 for any test were regarded as statistically significant. Results We measured the respiratory function of 120 of the 154 (77.9%) children at 11 years of age. Of the 34 children not tested, 15 lived in another State, four lived in another country, 10 refused the tests, three were lost to follow-up, and two were too disabled to complete the tests. There were no substantial differences in perinatal variables between children who did and did not have respiratory function tests at 11 years of age. Eighty of the 120 children (66.7%) had been exposed to passive smoking in the household. The only substantial differences in perinatal or subsequent variables between children who were and were not exposed to passive smoking were a significantly longer duration of oxygen therapy and a lower proportion of unskilled or unemployed families in the group not exposed (Table 1). For children exposed to passive smoking, the median number of cigarettes consumed in the household per day was 21 (interquartile range, 15-25). Of the 15 children tested who had developed BPD in the newborn period, three (20%) had asthma at 11 years of age; this proportion was similar for children with asthma at 11 who did not have BPD (22 of 105; 21%). For variables expressed as a percentage of predicted values (FEV1, FVC, RV, TLC), the means of the measured values were all close to their expected values of 100% (Table 2). For all respiratory function variables significantly associated with the dose of passive smoking, a quadratic relationship was more significant than a linear relationship (Table 2, Figures 1 and 2). Most respiratory function variables reflecting airflow (VEmax75%/VC, VEmax50%/VC, FEF25%-75%/VC, FEV1 and FEV1/FVC) were significantly diminished by increasing exposure to passive smoking (Table 2 [below], Figures 1a and 1b). In addition, RV, TLC and RV/TLC rose significantly (consistent with progressive air trapping) with increasing exposure to passive smoking (Table 2 [below], Figure 1c). One child was exposed to 115 cigarettes per day, and the next highest exposure was only 70 cigarettes per day. When the child exposed to 115 cigarettes per day was excluded, most of the statistically significant relationships disappeared, except for the increases in RV and RV/TLC (Figure 2). From the multiple linear regression analyses, some variables reflecting flow (VEmax75%/VC, VEmax50%/VC, VEmax25%/VC, FEF25%-75%/VC and FEV1/FVC) were significantly higher in girls. BPD was significantly associated with reductions in some variables reflecting air-flow (VEmax50%/VC, FEF25%-75%/VC, FEV1 and FEV1/FVC), as was asthma (with significant reductions in FEF25%-75%/VC, FEV1 and FEV1/FVC). FVC was significantly lower and VEmax25%/VC significantly higher with lower social class. Birthweight ratio, birthweight and gestational age were not significantly associated with any lung function variable. None of the statistical conclusions relating respiratory function variables with passive smoking were altered by adjusting for all potentially confounding variables, except that the reduction in VEmax50%/VC was no longer statistically significant. Discussion Passive smoking was associated with reduced airflow and air-trapping in VLBW children at 11 years of age, which is consistent with observations in non-preterm children free of lung disease.4 However, this finding was different from our observations of these children at eight years of age,7 when passive smoking was unassociated with any lung function variable. Chan et al.8 reported reduced flow rates with smoking by mothers in children of less than 2000 g birthweight at seven years of age, but they did not measure variables reflecting air-trapping. We did not distinguish between mothers and other smokers in the household. The association between passive smoking and adverse respiratory function in our VLBW children at 11, but not at eight, years of age suggests that the harmful effects of passive smoking take time to become obvious in VLBW children. Moreover, the adverse response seems to accelerate with increasing dose of passive smoking (Figures 1 and 2). We are concerned that continued exposure to passive smoking, or, even worse, active smoking, beyond 11 years will lead to not only further, but also to an accelerating rate of, deterioration in respiratory function. Our results should not be overinterpreted. They were not substantially altered by adjusting for potentially confounding perinatal or other variables, but we did not have data on a wide range of confounding variables. Moreover, they were heavily influenced by one child who lived in a household whose members consumed 115 cigarettes per day. Excluding this child from the analysis, the only remaining statistically significant associations indicated air-trapping with increasing exposure to passive smoking. However, we consider that this child's data should not be excluded just on the basis of heavier-than-average exposure to passive smoking. To remove any doubt about the association between passive smoking and adverse lung function in VLBW children, lung function could be measured in another cohort of VLBW children, or the same cohort when they are older. Parents of VLBW children, particularly those of children who have received assisted ventilation, frequently ask about long-term lung problems. Many variables, such as family history or duration of assisted ventilation and oxygen therapy, may be related to long-term lung problems, but most cannot be altered by the parents. Exposure to passive smoking is one variable associated with poorer respiratory function in VLBW children they can influence. Until there is evidence to the contrary, families of VLBW children should be encouraged to stop exposing their children to cigarette smoke in the household. As there appears to be a dose-response relationship, those who cannot stop smoking should at least reduce their children's exposure to passive smoking. Acknowledgement This study was supported in part by a grant from the Royal Women's Hospital-3AW Community Services Trust. References Landau L. Smoking and childhood asthma. Med J Aust 1991; 154: 715-716. Wright AL, Holberg C, Martinez FD, et al. Relationship of parental smoking to wheezing and nonwheezing lower respiratory tract illnesses in infancy. J Pediatr 1991; 118: 207-214. Etzel RA, Pattishall EN, Haley NJ, et al. Passive smoking and middle ear effusion among children in day care. Pediatrics 1992; 90: 228-232. Cook DG, Whincup PH, Papacosta O, et al. Relation of passive smoking as assessed by salivary cotinine concentration and questionnaire to spirometric indices in children. Thorax 1993; 48: 14-20. Chilmonczyk BA, Salmon LM, Megathlin KN, et al. Association between exposure to environmental tobacco smoke and exacerbations of asthma in children. N Engl J Med 1993; 328: 1665-1669. Smyth A, O'Hea U, Williams G, et al. Passive smoking and impaired lung function in cystic fibrosis. Arch Dis Child 1994; 71: 353-354. Kitchen WH, Olinsky A, Doyle LW, et al. Respiratory health and lung function in 8-year-old children of very low birth weight: a cohort study. Pediatrics 1992; 89: 1151-1158. Chan KN, Noble-Jamieson CM, Elliman A, et al. Lung function in children of low birth weight. Arch Dis Child 1989; 64: 1284-1293. Kitchen WH, Ford GW, Murton LJ, et al. Mortality and two year outcome of infants of birthweight 500-1500 g: relationship with neonatal cerebral ultrasound data. Aust Paediatr J 1985; 21: 253-259. Kitchen WH, Yu VYH, Lissenden JV, Bajuk B. Collaborative study of very-low-birthweight infants: techniques of perinatal care and mortality. Lancet 1982; i: 1 454-1457. 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. Hibbert ME, Lanigan A, Landau LI, Phelan PD. Lung function values from a longitudinal study of healthy children and adolescents. Pediatr Pulmonol 1989; 7: 101-109. SPSS for Windows [computer program]. Version 6.1. Chicago, Ill: SPSS Inc, 1994. Kitchen WH, Robinson H, Dickinson AJ. Revised intrauterine growth curves for an Australian hospital population. Aust Paediatr J 1983; 19: 157-161. (Received 8 Jun, accepted 18 Nov 1995) Authors' details Division of Paediatrics, the Royal Women's Hospital, Melbourne, VIC. Lex W Doyle, MD, FRACP, Paediatrician; and Associate Professor, Departments of Obstetrics and Gynaecology, and Paediatrics, the University of Melbourne. Geoffrey W Ford, MB BS, FRACP, Paediatrician. Annette M L Knoches, MB BS, FRCP(C), Paediatrician. Catherine Callanan, RN, Research Nurse. Department of Thoracic Medicine, the Royal Children's Hospital, Melbourne, VIC. Anthony Olinsky, FRACP, Respiratory Physician. No reprints will be available. Correspondence: Associate Professor L W Doyle, Department of Obstetrics and Gynaecology, University of Melbourne, Parkville, VIC 3052. ©MJA 1997 <URL: http://www.mja.com.au/> © 1997 Medical Journal of Australia.

Lex W Doyle · Geoffery W Ford · Anthony Olinsky · Catherine Callanan

Heroin-related deaths in New South Wales, 1992: toxicological findings and circumstances Deborah Zador, Sandra Sunjic and Shane Darke For editorial comment see Hall Readers may print a single copy for personal use. No further reproduction or distribution of the articles should proceed without the permission of the publisher. For permission, contact the Australasian Medical Publishing Company 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/>". Abstract - Introduction - Methods - Results - Discussion - Acknowledgements - References - Authors' details - - ©MJA1997 Abstract Objective: To describe the circumstances of death and toxicological findings in all heroin-related deaths in New South Wales in 1992. Design: Coronial files of all cases of heroin-related deaths were reviewed. A standardised form was used to collect information on sociodemographics, history of drug use, circumstances of death, and results of toxicological analysis for each case. Results: 152 heroin-related deaths were identified. Subjects had a mean age of 29.7 years, 82% were male, and 98% were not enrolled in a methadone treatment program at the time of their deaths. Deaths occurred in the home environment in 68% of cases and in the company of at least one other person in 58%. There was intervention before the subject's death in only 21% of cases. Two or more drug classes were detected in 71% of subjects; alcohol was detected in 45%, with a mean blood alcohol concentration of 0.14 g/100mL. Conclusions: Fatal heroin overdose is potentially preventable. Educating users about the risks of co-administering alcohol and other depressant drugs with heroin, the comparative safety of injecting heroin in the company of others and the need to call for intervention sooner may reduce the frequency of heroin-related deaths. MJA 1996; 164: 204 Introduction Over the decade to 1990, mortality from opiate drugs increased by 170% while that from other drugs declined by 8%.1 There is little information available on the circumstances or characteristics of Australian heroin-related deaths. Swensen reviewed 108 opioid drug-caused deaths in Western Australia from 1974 to 1984 and found that propoxyphene was the most common cause, accounting for 36 deaths (33%), followed by methadone in 19 cases (18%), heroin in 12 cases (11%), and morphine in another 12 cases (which may include deaths from heroin).2 Walsh analysed 23 accidental opioid drug deaths in the Newcastle area of New South Wales from 1970 to 1987 and found that most (83%) were caused by heroin/morphine.3 Police reports of cases in this study indicated that in 61% help or intervention by observers was delayed. Both studies found a higher proportion of deaths at weekends, leading both authors to suggest that these deaths involved recreational users rather than "hard-core" addicts. The findings of these two studies cannot be generalised to heroin deaths in Australia because of their relatively small numbers of subjects studied over prolonged periods, the inclusion of all opioid (not just heroin) drug deaths, and the absence of data for Sydney (where most heroin deaths in this country occur). If interventions to reduce the rising mortality rate from opioid drug use are to be developed, then more needs to be known about the circumstances of fatal heroin overdose. Hence, the major aims of this study were to describe the sociodemographic characteristics, circumstances of death and the toxicological findings at autopsy of all cases of fatal heroin overdose in NSW for the year for which the most recent data on drug-caused mortality were available. Methods The approval of the South Western Sydney Area Health Service's Research Ethics Committee was obtained. Permission of the NSW Registry of Births, Deaths and Marriages was obtained to enable the Australian Bureau of Statistics (ABS) to release a list of all heroin-caused deaths in NSW from 1 January to 31 December 1992. ABS records were searched according to the following International Classification of Diseases version 9 (ICD-9) codes: 304.0/304.7 (opiate dependence), E850.0 (accidental opiate poisoning) and E950.0 (opiate-caused suicide).4 Permission was obtained from the Department of Courts Administration to inspect coronial files. A case was identified as a heroin-related death based on the coroner's conclusion alone or in conjunction with results of toxicological analysis. Deaths due to other opiates were excluded from the study. Coronial files contained police reports, ambulance officers' statements, other witnesses' statements, autopsy reports, and results of toxicological analysis. A standardised data collection form was developed to record information on sociodemographic characteristics, history of drug use, circumstances of death, and toxicological findings. Subjects were designated "believed to be a frequent user" if their coronial files provided evidence of two or more criteria indicative of regular heroin use (e.g., known history of heroin use, partner or friends known to be heroin users, history of heroin overdose or treatment for dependence, criminal record and/or unemployment). Subjects were designated "believed to be an infrequent user" if they were known to be primarily dependent on another drug (such as alcohol), were not known to be regular heroin users, and were employed full-time. A "novice" was a person who died from using heroin for the first time. Information on history of admission to methadone maintenance treatment was obtained from data at the Pharmaceutical Services Section, NSW Department of Health. Information on history of admission to other treatment programs for opioid dependence (e.g., detoxification units or residential re habilitation programs) could not be obtained for this study. Data for circumstances of death included the day and date, suburb or town, type of location (e.g., hotel room), time of death, time between injection of heroin and death, presence of other persons, and intervention. A death was considered "instant" if photographs showed a needle and syringe still in situ, if the posture of the body was consistent with a sudden collapse (e.g., slumped against toilet cubicle wall), or if witnesses provided evidence of instant death. "Intervention" was treatment received by the subject while still alive, and did not include treatment by ambulance officers or after admission to hospital if the subject was clinically dead. Information on results of toxicological analysis was obtained from reports of laboratory analyses (by the Division of Analytical Laboratories, NSW Department of Health, Lidcombe) of blood, urine and other tissue specimens taken at autopsy. Statistical analysis For continuous variables, t tests were used, except for highly skewed variables, for which the Mann-Whitney U test (a non-parametric analogue of the t test) was used. Categorical variables were analysed with the chi-squared test and corresponding odds ratios (OR) and 95% confidence intervals (CI) were calculated. To determine which factors were independently associated with blood morphine concentration, we performed multiple linear regressions. SYSTAT was used for all analyses.5 Results Sociodemographic characteristics For the 200 opiate-caused deaths identified by ABS in 1992, one file was unable to be located, one subject was noted to have died on 31 December 1991, one case was found not to be a coroner's case, and seven were found to be misclassifications (i.e., death due to drugs other than opiates). These 10 cases were excluded from the study. The study population thus comprised 190 cases. Of these, 152 (80%) were classified as heroin-related deaths, and the demographic characteristics of the people involved are shown in the Table. They were predominantly male (82%), with a mean age of 29.7 years (SD, 6.7; range, 18-59), and most were unemployed (76%). History of drug use Most (90%) were well known to be heroin users. Eighty per cent were judged to have been frequent heroin users, 17% were considered infrequent heroin users, and two were novice users (1%). Both cases involving novice users were classified as suicides by the coroner. In three cases (2%) it was not possible to estimate the frequency of heroin use from the coronial files. Only 2% were enrolled in a methadone maintenance program at the time of their deaths, with 72% never having been enrolled in methadone treatment in NSW. Some subjects may have been previously enrolled in an interstate program, but this information was not available. Circumstances of death Twenty-six per cent of deaths occurred on Saturdays and Sundays, compared with an expected 29%, assuming a uniform distribution ofdeaths over the seven days of the week. There was also no significant variation between seasons -- summer (18%), autumn (25%), winter (31%) and spring (26%). The times of deaths were distributed as follows: midnight-6am, 23%; 6am- midday, 14%; midday-6pm, 28%; and 6pm-midnight, 35%. Eighty-five per cent of deaths occurred in the Sydney metropolitan region. Fatal heroin overdoses occurred throughout the Sydney area, with the highest frequency (19%) in the East Sydney region (including Kings Cross and Darlinghurst), and the second-highest (16%) in South-Western Sydney (including Cabramatta and Liverpool). Fifty-three per cent of deaths occurred in the subject's usual place of residence or home, and a further 16% in the home of a friend or family member. Thus most (69%) occurred in a home environment. Only 14% of deaths appeared to be "instant" events. The presence of other persons at some time during the interval between the injection of heroin and death was noted in 58% of cases, while 41% of subjects died alone. The presence of others was unable to be determined in 1% of cases. In 79% of cases there was no intervention while the subject was still alive. An ambulance was called while the subject was alive in only 10% of cases and cardiopulmonary resuscitation was attempted by an onlooker in 11%. Toxicological findings Two subjects were deleted from toxicological analyses as they had been maintained on life support systems for two or three days in hospital intensive care units. Morphine was detected in 95% of subjects at autopsy. The second most frequently detected drug was alcohol (45%), with benzodiazepines in 26%. Methadone and amphetamines were each detected in 6% of subjects, and cocaine, antidepressant drugs and barbiturates were detected in 5%, 4% and nil, respectively. A single drug was found in only 27% of subjects -- morphine in 39 of 41, ethanol in one and oxazepam in one. In 71%, two or more different drugs were found at autopsy, and in 18% three or more drugs were detected. Men were more likely to have alcohol detected at autopsy than women (OR, 4.07; 95% CI, 1.44-11.47), while women were more likely to have benzodiazepines detected (OR, 2.82; 95% CI, 1.15-6.93). The distribution of blood morphine concentrations is shown in Figure 1. Twelve subjects were deleted from this analysis because no morphine was found in their blood (although in seven morphine was detected elsewhere, e.g., bile, liver or urine). The median blood morphine level was 0.24 mg/L. Alcohol was detected in 68 subjects (45%); Figure 2 shows the distribution of their blood alcohol concentrations (BACs). The mean BAC was 0.14 g/100 mL (SD, 0.08; range, 0.01-0.35 g/100 mL). Eighty-seven per cent had BACs of 0.05 or higher and 22% had BACs of 0.2 or higher. There were no statistically significant differences between men and women in either median blood morphine concentration (0.25 v. 0.19 mg/L; P < 0.4) or BAC (0.14 v. 0.1 g/100mL; P < 0.25). The median blood morphine concentration in the subjects in whom alcohol was detected was significantly lower than in the group without alcohol detected (0.17 v. 0.34 mg/L; U = 3152; P < 0.001). There was also a statistically significant negative correlation between blood morphine and alcohol concentrations (rs = - 0.28; P < 0.01) The presence of benzo diazepines at autopsy did not significantly affect blood morphine concentrations (0.25 v. 0.24 mg/L; U = 1871; P < 0.7). To determine which individual variables were related to blood morphine concentration, a simultaneous multiple linear regression was performed on log morphine level (because of the skewed distribution of blood morphine levels). Variables used were age, sex, the presence of alcohol and the presence of benzodiazepines. The presence of alcohol at autopsy was the only variable independently associated with lower blood morphine levels (b = - 0.28; P < 0.001); age (P < 0.96), sex (P < 0.2) and the presence of benzodiazepines (P < 0.7) did not significantly predict log blood morphine level. Diagnostic tests of the residuals indicated that the assumptions of the model were met. Classification of deaths Fifty-seven per cent of all death certificates completed by the forensic pathologists reported the cause of death as either "narcotism", "acute narcotism" or "acute intravenous narcotism". Although alcohol was detected in 45% of subjects at time of death, it was noted on the death certificate in only 7% of cases. Most deaths (80%) were classified according to ICD-94 coding as cases of "dependence" (code 304.0), 10% as cases of accidental poisoning (E850.0), 9% as suicides (E950.0) and 1% as opiate drug dependence with contributions from other drugs (304.7). Discussion We found that heroin-related deaths occurred overwhelmingly in people who were male, of an average age of 30 years, and frequent users. Few occurred in those considered to be infrequent users. A recent survey has shown that two-thirds of a large sample of heroin users have had at least one non-fatal overdose and that 86% have been present at another person's overdose.6 This suggests that campaigns designed to reduce the rising mortality rate caused by heroin need to target regular users. The finding that only a minority of cases had ever been in methadone treatment is consistent with other evidence that methadone treatment reduces mortality among the heroin-dependent.7,8 Some of our subjects might be alive today if they had entered and remained in methadone treatment. A striking finding from the toxicological data was the relatively small number of subjects in whom morphine only was detected. Most died with more drugs than heroin alone "on board", with alcohol detected in 45% of subjects and benzodiazepines in just over a quarter. Both of these drugs act as central nervous system depressants and can enhance and prolong the depressant effects of heroin. Our finding (and that of other studies9,10 ) of a significantly lower blood morphine level in subjects who tested positive for alcohol suggests that in the presence of alcohol less heroin is required to exert a fatal effect. Alternatively, alcohol-induced liver enzymes may increase the metabolism of heroin. The very low blood levels of morphine we found may also support these hypotheses. Some pathologists have been reluctant to attribute a heroin-caused death to true overdose because of the frequent finding of a low blood morphine concentration, and prefer to report the death as a case of "narcotism" (Johan Duflou, Deputy Director, Insti tute of Forensic Medicine, Sydney, personal communication, 1993). Further, Monforte showed that in 75% of a group of people who had died from fatal heroin "overdoses" blood morphine concentrations were not higher than in a group of heroin addicts who were victims of homicide.11 Bammer and Sengoz have also challenged the assumption that heroin deaths occur as a result of increased purity of heroin. In a recent paper they identified concomitant consumption of other drugs and reduced tolerance as alternative mechanisms for the cause of death in heroin fatalities.12 The disadvantage of continuing to describe heroin-related fatalities as "overdoses" is that it attributes the cause of death solely to heroin and detracts attention from the contribution of other drugs to the cause of death. Heroin users need to be educated about the potentially dangerous practice of concurrent polydrug and heroin use. Our findings that an ambulance was called while the subject was still alive in only 10% of cases, and that a substantial minority of heroin users died alone, strongly suggest that education campaigns should also emphasise that it is safer to inject heroin in the company of others, and important to call for an ambulance early in the event of an overdose. Consideration should also be given to trialling the distribution of the opioid antagonist naloxone to users to reduce mortality from heroin use. In conclusion, fatal heroin overdose is potentially preventable and research should now focus on reducing its occurrence. The introduction or intensification of education campaigns explaining the risk of polydrug use to heroin users may help to reverse the national trend in heroin-related deaths. Acknowledgements This research was funded by South Western Sydney Area Health Service and the National Drug and Alcohol Research Centre. We would like to thank the following organisations for their assistance: Department of Courts Administration, in particular Glebe Coroners Court and Westmead Coroners Court, and the NSW Registry of Births, Deaths and Marriages. We would also like to thank Mr Stephen Bull of the Australian Bureau of Statistics (NSW office), Dr Johan Duflou, Professor Wayne Hall and Dr Alex Wodak. References Drug Caused Deaths in Australia 1990. Drugs of Dependence Branch statistical update. Number 18. Canberra: Commonwealth Department of Health, Housing and Community Services, August 1992. Swensen G. Opioid deaths in Western Australia: 1974-1984. Aust Drug Alcohol Rev 1988; 7: 181-185. Walsh RA. Opioid drug accidental deaths in the Newcastle area of New South Wales, 1970-1987. Drug Alcohol Rev 1991; 10: 79-83. World Health Organization. International statistical classification of diseases, injuries and causes of death. Vol. 1. Geneva: WHO, 1977. Wilkinson L. SYSTAT: the system for statistics [computer program]. Evanston, Il.: SYSTAT Inc, 1990. Darke S, Ross J, Hall W. Overdose among heroin users in Sydney, Australia: I. Prevalence and correlates of non-fatal overdose. Addiction 1996; 91: 405-411. Gronbladh L, Ohland LS, Gunne LM. Mortality in heroin addiction: impact of methadone treatment. Acta Psychiatr Scand 1990; 82: 223-227. Caplehorn JRM, Dalton MSYM, Cluff MC, Petrenas AM. Retention in methadone maintenance and heroin addicts' risk of death. Addiction 1994; 89: 203-207. Ruttenber AJ, Kalter HD, Santinga P. The role of ethanol abuse in the etiology of heroin related deaths. J Forensic Sci 1990; 35: 891-900. Chan L, Prolov T, Verma S. Morphine tissue concentrations in fatal overdoses in New South Wales, 1986-1987. Proceedings of the 10th Australian International Forensic Science Symposium; 1988 May 23-27; Brisbane. Brisbane: Australian Forensic Science Society, 1988. (Abstract No. 2333.) Monforte JR. Some observations concerning blood morphine concentrations in narcotic addicts. J Forensic Sci 1977; 22: 718-724. Bammer G, Sengoz A. How would the controlled availability of heroin affect the illicit market in the Australian Capital Territory? An examination of the structure of the illicit heroin market and methods to measure changes in price, purity, availability, including heroin-related overdoses. Canberra: National Centre for Epidemiology and Public Health, 1994. (Working Paper No. 10.) (Received 13 Jun, accepted 19 Oct 1995) Authors' details Drug and Alcohol Services, Division of Public Health, South Western Sydney Area Health Service, Sydney, NSW. Deborah Zador, MB BS, FAFPHM, Staff Specialist. Currently, Staff Specialist, Drug and Alcohol Services, Central Sydney Area Health Service, NSW. Sandra Sunjic, BAppSc, MA, Clinical Nurse Specialist. National Drug and Alcohol Research Centre, University of New South Wales, Sydney, NSW. Shane Darke, PhD, Lecturer. No reprints will be available. Correspondence: Dr D Zador, Gladstone Hall Health Services, 114 Ewart Street, Dulwich Hill, NSW 2203. ©MJA 1997 <URL: http://www.mja.com.au/> © 1997 Medical Journal of Australia.

Deborah Zador · Sandra Sunjic · Shane Darke

Research Consumption of oily fish and childhood asthma risk Linda Hodge, Cheryl M Salome, Jennifer K Peat, Michelle M Haby, Wei Xuan and Ann J Woolcock For editorial comment, see Thien et al. Abstract - Authors' details - Introduction - Methods - Results - Discussion - Appendix - Acknowledgements - References - Box 1 - Box 2 - Box 3 - Figure - © MJA 1996 - Abstract Objective: To investigate the association between diet and airway disease in children in the light of epidemiological studies suggesting that consumption of fish more than once a week reduces the risk of developing airway hyperresponsiveness (AHR). Design: Diet was assessed by a detailed food frequency questionnaire and airway disease by respiratory symptoms or airway responsiveness to exercise. Methods: A questionnaire, containing questions about the frequency of eating more than 200 foods, was sent to the parents of 574 children in whom we had measured recent wheeze (by questionnaire), AHR (by exercise) and atopy (by skin prick tests) six months before this study. We defined current asthma as the presence of both recent wheeze and AHR. Results: Response rate to the questionnaire was 81.5% (n = 468). After adjusting for confounders such as sex, ethnicity, country of birth, atopy, respiratory infection in the first two years of life and a parental history of asthma or smoking, children who ate fresh, oily fish (> 2% fat) had a significantly reduced risk of current asthma (odds ratio, 0.26; 95% confidence interval, 0.09-0.72; P < 0.01). No other food groups or nutrients were significantly associated with either an increased or reduced risk of current asthma. Conclusion: These data suggest that consumption of oily fish may protect against asthma in childhood. MJA 1996; 164: 137-140 Introduction The substantial increase in the prevalence of childhood asthma in the past 20 years has affected both rural and urban communities of westernised countries,1,2 suggesting that local environmental factors, such as exposure to allergens or industrial air pollutants, are not the cause. However, the widespread changes in diet may be responsible. Seaton et al.3 have postulated that increases in the prevalence of asthma may be due to a reduced intake of antioxidant vitamins (beta-carotene, vitamins A, C and E) and mineral cofactors essential for antioxidant defence mechanisms (selenium, zinc and copper) as a result of reduced consumption of meat, fresh fish, fruit and vegetables in Western diets. Reduced consumption of magnesium4 and increased consumption of salt5 have been implicated as risk factors for airway hyperresponsiveness (AHR). Our own epidemiological studies of Australian schoolchildren have shown that children who eat fish more than once a week have a third the risk of AHR of children who do not eat fish regularly.6 However, these studies did not include other dietary questions, so that fish consumption may have been a marker for another dietary characteristic. Here, we investigate the association between diet, as assessed by a detailed dietary questionnaire, and airway disease, assessed by respiratory symptoms or airway responsiveness to exercise. Methods Subjects In June 1993, a cross-section of 808 children aged 8-11 years from schools randomly selected from all schools within a 10 km radius of Sydney General Post Office had airway responsiveness to exercise, respiratory symptoms and atopy measured and frequency of fish consumption assessed.7 In October 1993, 584 children were selected from this group in a stratified case-control design and their parents were asked to complete a detailed food frequency questionnaire about their child's eating habits. The selection criteria included all children with AHR, all children who had had wheeze in the last 12 months (recent wheeze) and a three-in-five sample of children with normal airways (no AHR or recent wheeze), who were chosen by excluding two children after every three from numerically ordered lists of children identified by number. The study coordinator who collected the food frequency questionnaires was blind to the respiratory symptom and AHR status of the subjects. Ethical approval for the study was obtained from the Ethics Review Committee of the University of Sydney. Permission to approach schools was obtained from the New South Wales Department of School Education and the Catholic Education Office. Respiratory questionnaire In June 1993 the parents or guardians of the children completed a standard respiratory questionnaire, with questions on age, sex, ethnicity, country of birth, history of asthma or wheeze in the last 12 months, medication use, and also parents' occupations, history of asthma and smoking. The questionnaire included the question used in previous studies about the dietary consumption of fish - "How often does your child eat a meal that contains fish?" - with the options of replying "never or rarely", "once a week", or "more than once a week". Dietary questionnaire In October 1993, a food frequency questionnaire (adapted from that developed and validated by the Commonwealth Scientific and Industrial Research Organisation [CSIRO], Division of Human Nutrition, South Australia8,9) was distributed to the selected children, whose parents were asked to complete this for their child's usual eating habits over the last year. The questionnaire identified consumption patterns (daily, weekly, monthly, rarely or never) of more than two hundred foods commonly consumed in Australia. Additional questions on the type of fresh fish consumed and regular consumption of vitamin, mineral or herbal supplements were included. Estimates of sodium intake included naturally occurring sodium in foods, salt added in cooking, at the table and from processed foods. If questionnaires were not returned after one month the parents were contacted by telephone and offers were made to replace the questionnaires, or to provide assistance. In 11 cases, where neither parent spoke fluent English, an interpreter was commissioned to complete the questionnaire with the parents over the telephone. Returned dietary questionnaires were checked for missing or obviously erroneous information. Parents were contacted by telephone to complete omitted sections or to clarify erroneous information. Each food in the dietary questionnaire was allocated to one of 23 different food groups (see Appendix). Diets were analysed for energy, fibre and 39 nutrients (see Appendix). Definitions of respiratory categories, atopy, and categories of fish, plus a list of fish with more than 2% fat, are given in Box 1. Statistical analyses The questionnaires were analysed by the Division of Human Nutrition, CSIRO, South Australia, using Australian tables of nutrient composition12 for energy, protein, fat, carbohydrates, vitamins and minerals. The total quantity of food in each food group for every child was converted to a common base of weekly serves with Clinical Reporting Systems software.13 Data were analysed with the statistical package SAS.14The association between fish, food or nutrient intake and respiratory category was analysed categorically using chi-squared tests, and continuously using Student's t tests and analysis of variance. Some values obtained from the nutrient analysis were well outside what could reasonably be expected in children of this age group. These outliers were excluded from the statistical analysis. The number of exclusions never exceeded nine subjects in any analysis and were not significantly associated with any of the respiratory groups. Logistic regression was used to adjust estimates for the effects of known confounders for the effect of fish consumption on AHR and symptoms of asthma (e.g., sex, race, country of birth, atopy, early respiratory infection, parental smoking and parental asthma). Only those confounding factors found to be significant or approaching significance (P < 0.1) (atopy, parental asthma, early respiratory infection, country of birth) were included in the model. Results Of the 584 children selected 574 received the dietary questionnaire and 468 completed questionnaires were returned (81.5%). Non-responders were not significantly different from responders in the prevalence of AHR (26.0% v. 27.1%) or fish consumption (46.2% v. 52.1%). Box 2 shows details of the children studied. Children with current asthma did not differ significantly from children with normal airways in the consumption of any nutrient or food group. (Tables showing mean weekly intake in standard serves of food groups and mean daily intake [SD] of nutrients for children with normal airways and children with current asthma are available from the authors.) Children with wheeze only had a significantly higher intake of red meat (P < 0.05), offal meat (P < 0.001) and vitamin B12 (P < 0.03) and a significantly lower intake of mixed vegetables (P < 0.05) than children with normal airways. Children with AHR only consumed significantly more offal meat (P = 0.001) and high fat/high sugar foods (P < 0.001) than children with normal airways. They also had higher intakes of nitrogen, protein, total sugar, cholesterol, potassium nicotinamide, total nicotinamide, calcium, copper, zinc, vitamin B12 (P < 0.05) and refined sugar (P < 0.01). Total fish intake per week did not differ significantly between children with normal airways (1.2 serves per week; 95% confidence interval [CI], 1.0-1.3), AHR only (1.2 serves; CI, 0.9-1.5), wheeze only (1.2 serves; CI, 0.8-1.5) and current asthma (1.0 serve; CI, 0.8-1.2). Fresh fish was eaten by 84% (CI, 79.6%-88.4%) of children with normal airways, and by 72% (CI, 61.6%-82.4%) of children with current asthma. When fresh fish was divided into oily and non-oily types, significantly fewer children with current asthma (15.5%; CI, 7.1%-23.9%) included oily fish in their diet than did children with normal airways (30.8%; CI, 25.2%-36.4%; P < 0.05). There were no significant differences in the proportions of children with current asthma (56.3%; CI, 44.8%-67.8%) and normal children (52.9%; CI, 46.9%-58.9%) who ate exclusively non-oily fish. Neither fresh fish consumption nor respiratory disease was significantly associated with socioeconomic status, as defined by the father's occupation, or with the consumption of vitamin, mineral or other dietary supplements (including fish oil). The unadjusted risk (odds ratio) for children having current asthma was significantly lower in those who consumed any fresh fish or oily fresh fish (Box 3). Current asthma was found in only 8.8% (CI, 3.8%-13.8%) of children who ate oily fish, but in 15.6% (CI, 11.2%-20.0%) of those who ate non-oily fish only and 23% (CI, 14.2%-31.8%) of those who never ate fresh fish (Figure). When the results were adjusted for the effects of other known risk factors such as atopy, parental asthma, parental smoking, ethnicity, country of birth, early respiratory illness and sex, only children who ate oily fresh fish had a significantly reduced risk of current asthma. In these children, the risk was almost a quarter that of children who did not eat oily fish (odds ratio, 0.26; CI, 0.09-0.72) (Box 3). Consumption of any fresh fish, whether or not it was separated into oily fresh fish and non-oily fresh fish, did not significantly reduce the risk of AHR only or wheeze only either before or after adjustment for other risk factors. Discussion Our study shows that regular consumption of fresh, oily fish is associated with a reduced risk of current asthma. This reduced risk remained significant after adjustment for other known risk factors for asthma, including sex, atopy, parental asthma, parental smoking, early respiratory infection, ethnicity and place of birth. The subjects were selected from a random cross-sectional sample of children which was stratified (on the basis of recent respiratory symptoms and AHR to exercise) to increase the proportion of cases in the study group. The response rate was high (81.5%) and non-responders were not different from responders with respect to AHR or fish consumption. Socioeconomic status was not a confounder for either respiratory illness or fish consumption. Current asthma was defined as recent wheeze plus AHR to exercise. We have shown previously that current asthma, defined as recent wheeze plus AHR to histamine, identifies a group with severe, ongoing respiratory impairment, while those with AHR only and wheeze only have a milder condition which differs only slightly from the normal group.15 The diets of children with current asthma differed from those of the normal group only in the consumption of fresh, oily fish. In our previous study, more than one serve of fish per week was associated with a reduced risk of asthma,6 but in that study it was not possible to distinguish the effects of oily and non-oily fish. In the study reported here we were unable to detect differences in total fish consumption, possibly because of the smaller sample size. There were no significant differences between respiratory groups in the consumption of non-oily fish, suggesting that parents had not selectively withheld fish from the diets of asthmatic children. It is unclear why consumption of canned and processed fish was not associated with reduced risk of asthma. Processing may alter the integrity or activity of the fatty acids in fish oils. Several foods and nutrients in the diets of children with AHR only and wheeze only differed significantly from those of the normal group. However, none of these factors differed between the asthmatic and normal groups, suggesting that they are unlikely to have an aetiological role. Intake of offal meats was higher in both the AHR-only and wheeze-only groups, but, as offal meats are eaten by very few children, this may be a type I error. Vitamin B12 intake was also higher in both the AHR-only and wheeze-only groups, but the mechanism by which this could affect respiratory symptoms or AHR is unclear. There were no significant differences between any of the respiratory groups in consumption of sodium, vitamin C, vitamin E, selenium or magnesium. These findings do not support previous evidence that these dietary factors are important in the aetiology of asthma.4,5,16 Reduced risk of current asthma was associated with the consumption of oily fish, but not with non-oily fish. Fish oil contains the omega-3 fatty acids eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA), which have anti-inflammatory effects.17 Theoretically, EPA could either prevent the development of asthma or reduce its severity by altering two of the cardinal features of asthma, namely airway inflammation and AHR. Supplementation with EPA reduces production of leukotriene B4,18 a chemical mediator responsible for the recruitment of inflammatory cells, such as neutrophils, into the airways. It also reduces production of the cytokine tumour necrosis factor (TNF),19 which increases airway responsiveness.20 Fish oil supplements given over 6-10 weeks cause a substantial uptake of EPA in neutrophil membrane phospholipids.18,21 In asthmatics, this may reduce the allergen-induced late asthmatic response,22 but does not change severity of asthma.21,22 However, a recent study suggests that a longer period of supplementation may be required to reduce asthma severity.23 Data of recent fish consumption (during the last 12 months) were used in our study, but may also reflect lifetime dietary habits. In conclusion, we have shown that consumption of oily fish is associated with a reduced risk of asthma in childhood. Although further studies are required to confirm these benefits, public health interventions to increase the consumption of oily fish may reduce the morbidity and prevalence of asthma in children. Appendix The 23 food groups were: cereals; dairy products; eggs; red, white, preserved meat and offal; seafood, fried, steamed, canned fish and fish fingers; red, green, white, mixed and other vegetables; legumes; high and low vitamin C fruit and other fruit; high sugar or fat content; and other. The 39 nutrients were: nitrogen; protein; starch; refined, natural and total sugar; total carbohydrate; saturated, monounsaturated, polyunsaturated and total fat; cholesterol; carotene; retinol; vitamin A; thiamine; riboflavin; potassium nicotinic acid and total nicotinic acid; niacin; vitamins B6 and B12; pantothenic acid; biotin; free and total folate; vitamins C, D, E; calcium; copper; iron; magnesium; manganese; phosphorus; potassium; selenium; sodium; and zinc. Acknowledgements This study was supported by the Fisheries Research and Development Corporation, Australia. The authors thank Dr Katrine Baghurst for allowing us to use the dietary questionnaire, Sally Record and Kay Pender for their help with the nutritional analyses, Elena Belooussova for data organisation and Suzanne Gray for her assistance with collecting the questionnaires. We are grateful for the support of the New South Wales Department of School Education, the Catholic Education Office and the Principals and teachers of all the schools involved. We are especially grateful to the parents and the children who participated in the survey. (©MJA 1996; 164: 137-140) References Robertson CF, Bishop J, Sennhauser FH, Mallol J. International comparison of asthma prevalence in children: Australia, Switzerland, Chile. Pediatr Pulmonol 1993; 16: 219-226. Burney P, Chinn S, Rona RJ. Has the prevalence of asthma increased in children? Evidence from the national study of health and growth 1973-86. BMJ 1990; 300: 1306-1310. Seaton A, Godden DJ, Brown K. Increase in asthma: a more toxic environment or a more susceptible population? Thorax 1994; 49: 171-174. Britton J, Pavord I, Richards K, et al. Dietary magnesium, lung function, wheezing, and airway hyperreactivity in a random adult population sample. Lancet 1994; 344: 357-362. Burney PG, Neild JE, Twort CHC, et al. Effect of changing dietary sodium on the airway response to histamine. Thorax 1989; 44: 36-41. Peat JK, Salome CM, Woolcock AJ. Factors associated with bronchial hyperresponsiveness in Australian adults and children. Eur Respir J 1992; 5: 921-929. Haby MM, Peat JK, Mellis CM, et al. An exercise challenge for epidemiological studies of childhood asthma: validity and repeatability. Eur Respir J 1995; 8: 729-736. Baghurst KI, Record SJ. Intake and sources in selected Australian subpopulations of dietary constituents implicated in the etiology of chronic diseases. J Food Nutr 1983; 40: 1-15. Rohan TE, Record SJ, Cook MG. Repeatability of estimates of nutrient and energy intake: the quantitative food frequency approach. Nutr Res 1987; 7: 125-137. Analyses of NSW fish and shellfish. Sydney: Australian Government Analytical Laboratory, 1989. Sinclair A, Dunstan GA, Naughton JM, et al. The lipid content and fatty acid composition of commercial marine and freshwater fish and molluscs from temperate Australian waters. Aust J Nutr Diet 1992; 49: 77-83. English R, Lewis J. Composition of foods Australia. 1st ed. Vols 1-5. Canberra: AGPS, 1989-1990. Clinical Reporting Systems [computer program], version 3.0. Sydney: Clinical Reporting Systems Pty Ltd. 1992. SAS [computer program], version 5. Cary, NC: SAS Institute, 1984. Toelle BG, Peat JK, Salome CM, et al. Toward a definition of asthma for epidemiology. Am Rev Respir Dis 1992; 146: 633-637. Stone J, Hinks LJ, Beasley R, et al. Reduced selenium status of patients with asthma. Clin Sci 1989; 77: 495-500. Kremer JM, Jubiz W, Michalek A, et al. Fish-oil fatty acid supplementation in active rheumatoid arthritis. Ann Intern Med 1987; 106: 497-503. Lee TH, Hoover RL, Williams JD, et al. Effect of dietary enrichment with eicosapentaenoic and docosahexaenoic acids on in vitro neutrophil and monocyte leukotriene generation and neutrophil function. N Engl J Med 1985; 312: 1217-1224. Endres S, Ghorbani R, Kelley VE, et al. The effect of dietary supplementation with n-3 polyunsaturated fatty acids on the synthesis of interleukin-1 and tumor necrosis factor by mononuclear cells. N Engl J Med 1989; 320: 265-271. Thomas PS, Yates DH, Barnes PJ. Tumor necrosis factor-alpha increases airway responsiveness and sputum neutrophilia in normal human subjects. Am J Respir Crit Care Med 1995; 152: 76-80. Arm JP, Horton CE, Mencia-Huerta J-M, et al. Effect of dietary supplementation with fish oil lipids on mild asthma. Thorax 1988; 43: 84-92. Arm JP, Horton CE, Spur BW, et al. The effects of dietary supplementation with fish oil lipids on the airways response to inhaled allergen in bronchial asthma. Am Rev Respir Dis 1989; 139: 1395-1400. Dry J, Vincent D. Effects of fish oil diet on asthma: results of a 1-year double blind study. Int Arch Appl Immunol 1991; 95: 156-157. (Received 31 May, accepted 28 Nov 1995) Authors' details Institute of Respiratory Medicine, Royal Prince Alfred Hospital, Sydney, NSW. Linda Hodge, MSc(Med), GradDipNutr&Diet, Dietitian. Department of Medicine, University of Sydney, Sydney, NSW. Cheryl M Salome, BSc, Senior Research Officer. Jennifer K Peat, PhD, Senior Research Officer. Michelle M Haby, MSc, Research Assistant. Wei Xuan, MSc, MApplStat, Statistician. Ann J Woolcock, MD, FRACP, Professor in Respiratory Medicine. Reprints: Professor A J Woolcock, Institute of Respiratory Medicine, Royal Prince Alfred Hospital, Camperdown, NSW 2050. (©MJA 1996; 164: 137-140)

Linda Hodge · Cheryl M Salome · Jennifer K Peat · Michelle M Haby · Wei Xuan · Ann J Woolcock

Drug-resistant Streptococcus pneumoniae: the beginning of the end for many antibiotics? Peter J Collignon and Jan M Bell, on behalf of the Australian Group on Antimicrobial Resistance (AGAR)* Abstract - Introduction - Methods - Antibiotic sensitivity testing - Statistical analysis - Results - Discussion - Acknowledgements - References - Authors' details - ©MJA1997 Abstract Objective: To determine the levels of antibiotic resistance in Streptococcus pneumoniae in Australia. Design: Prospective, Australia-wide, laboratory-based survey. Setting: 27 hospital and private laboratories around Australia, from January 1994 to August 1995. Subjects: First 100 patients with clinically significant isolates of S. pneumoniae at each laboratory. Outcome measures: Resistance to penicillin (determined from penicillin minimum inhibitory concentration [MIC] measured by the Etest), erythromycin, trimethoprim-sulfamethoxazole, tetracycline, chloramphenicol, cefotaxime and ceftriaxone. Results: A total of 2396 isolates were tested (including 537 invasive isolates and 740 from children). Penicillin resistance was seen in 161 isolates (6.7%), including 17 with high level resistance. Penicillin resistance rates were significantly lower in invasive than in non-invasive strains (3.7% versus 7.6%; odds ratio [OR], 0.47; 95% confidence interval [CI], 0.28-0.77; P = 0.001). There was no significant difference in penicillin resistance rates between children ( < 15 years) and adults (7.3% versus 6.5%; OR, 1.14; 95% CI, 0.80-1.63; P = 0.47). Resistance rates were higher for most other antibiotics than for penicillin (chloramphenicol, 6%; erythromycin, 11%; tetracycline, 15%; and trimethoprim-sulfamethoxazole, 42%). No high level resistance was seen to third generation cephalosporins, but 17 of 109 penicillin-resistant isolates tested (16%) displayed intermediate resistance to cefotaxime. Rates of antibiotic resistance varied between States, with the lowest rates in Tasmania. Conclusions: Antibiotic resistance levels in S. pneumoniae are increasing in Australia and high level penicillin resistance is being encountered for the first time (including in invasive strains). This will lead to an increasing number of therapeutic dilemmas and possible therapeutic failures, especially important in meningitis. MJA 1996; 164: 64 Introduction The pneumococcus (Streptococcus pneumoniae) continues to be a common cause of serious and life-threatening infections, including pneumonia, bacteraemia and meningitis. It is also a frequent cause of respiratory tract infections, such as otitis media and sinusitis.1-4 A major advance was made in the treatment of these infections with the introduction of penicillin 50 years ago. Until relatively recently, pneumococci were considered so uniformly sensitive to penicillin (with minimum inhibitory concentrations [MICs] < 0.02 mg/L) that sensitivity tests were usually not performed. It was from Australia in 1967 that the first clinically significant isolate of a penicillin-resistant pneumococcus was reported.5 However, penicillin resistance was not a major clinical problem in this country, although it caused major problems elsewhere, particularly in Papua New Guinea and South Africa.1-3 In the late 1970s and the 1980s, rates of resistance (including multiple resistance) increased in Western countries, particularly in Spain (with resistance levels of 50%).1,2,3 A recent United States study found 25% of invasive S. pneumoniae isolates were penicillin-resistant.6 Resistance rates are usually higher in children, and the distribution of resistance varies within countries and population groups.1,2,3,6 In an Australia-wide study of over 1800 isolates of S. pneumoniae in 1989, we found that only 1% were penicillin-resistant,7 a lower rate than in most other Western countries. However, some communities (especially Australian Aboriginals) have relatively high rates of resistance.8 Because of the worldwide increase in resistance to many antibiotics and the implications of penicillin resistance in S. pneumoniae for treatment of life-threatening conditions (particularly meningitis), we undertook a further study of resistance to penicillin and other commonly used antibiotics in clinically significant S. pneumoniae isolates from both the community and hospitals. Methods Twenty-seven hospital and private laboratories from around Australia parti cipated. From January 1994, each laboratory tested the first 100 consecutive clinically significant isolates. The rate of collection varied from 5 to 20 months, with all laboratories filling their quota in August 1995. Patients' sex, age, specimen site and inpatient or outpatient status were recorded prospectively. Clinically significant isolates were defined as those isolated either from normally sterile sites (e.g., cerebrospinal fluid and blood [invasive isolates]) or from specimens that made contact with mucosal surfaces (e.g., sputum) if they were associated with an increased white cell count on gram staining and would normally have been reported as clinically significant. Throat or surveillance swabs were excluded, as were duplicates of clinically significant isolates. S. pneumoniae was identified by colonial morphology, a -haemolysis on blood agar plates, susceptibility to optochin and/or bile solubility. Antibiotic sensitivity testing Isolates were tested for susceptibility to penicillin, erythromycin, trimethoprim-sulfamethoxazole, tetracycline and chloramphenicol by the standardised routine method of each laboratory. Methods included disc diffusion with either National Committee for Clinical Laboratory Standards (NCCLS)9 (14 laboratories) or Calibrated Dichotomous Sensitivity (CDS)10,11 (7 laboratories); agar dilution12 with either Isosensitest agar (Oxoid) (2 labora tories) or Mueller-Hinton agar (3 lab oratories); and the ATB system (BioMerieux sa , Marcy-l'Etoile, France) (1 laboratory). The MIC of penicillin was also determined for each isolate by the Etest on Mueller-Hinton agar supplemented with 5% blood;13,14 plates were incubated at 35¡C in 5% CO 2 for 20-24 hours. 14 The interpretive criteria of the NCCLS15 were used for susceptibility categorisation of Etest values (susceptible, MIC < 0.06 mg/L; intermediate resistance, MIC = 0.125-1 mg/L; and high level resistance, MIC > > 2 mg/L). For the study, antibiotic resistance was defined as decreased susceptibility (both intermediate and high level resistance),6,12 and multidrug resistance as decreased susceptibility to two or more of the antibiotic agents tested. Isolates from normally sterile sites and those that appeared resistant to penicillin or chloramphenicol by routine susceptibility testing or had penicillin MICs > > 0.047 mg/L were forwarded to Monash Medical Centre for further susceptibility testing: Etest strips were used to determine cefotaxime and ceftriaxone MICs. Statistical analysis Fisher's two-tailed exact test was used to calculate P values. Calculations were performed with True Epistat software.16 Results A total of 2396 isolates from different patients were tested. The average age of the patients was 41.6 years (range, < 1 day to 98 years); 32% were children ( < 15 years) and 60% were male. The percentage of penicillin-resistant isolates from each specimen site is shown in Box 1; the overall rate of penicillin resistance was 6.7%, with rates in individual laboratories ranging from 0 to 13%. High level resistance was seen in 17 isolates, including two from normally sterile sites. The rate of penicillin resistance was significantly lower among invasive isolates than among non-invasive isolates (3.7% versus 7.6%; odds ratio [OR], 0.47; 95% confidence interval [CI], 0.28-0.77; P = 0.001) (Box 2). The rate of penicillin resistance was slightly higher among children ( < 15 years) than among adults, but the difference was not statistically significant (7.3% versus 6.5%; OR, 1.14; 95% CI, 0.8-1.63; P = 0.47). Resistance to antibiotics other than penicillin was common (Boxes 2 and 3). Rates varied around Australia, with the lowest rates for nearly all antibiotics in Tasmania and the highest in the eastern States, particularly Queensland and New South Wales. The rate of penicillin resistance was highest in South Australia. Very high levels of resistance were seen for trimethoprim-sulfamethoxazole (29%-52%). All five antibiotics were tested on 1895 isolates; 267 (14%) were multi resistant, with 159 (8%) resistant to three or more antibiotics and 31 (1.6%) to all five (Box 3). Of 124 penicillin-resistant isolates tested, 72 (58%) were resistant to three or more non--lactam agents; 40% were resistant to chloramphenicol; 52% to erythromycin; 64% to tetracycline; and 78% to trimethoprim-sulfamethoxazole. Of the 1771 pencillin-susceptible isolates, 101 (6%) were resistant to three or more non--lactam agents; 3% to chloramphenicol; 8% to erythromycin; 12% to tetracycline; and 39% to trimethoprim- sulfamethoxazole . Of 109 penicillin-resistant isolates tested with cefotaxime, 17 (16%) had intermediate resistance (MIC, 1 mg/L). These comprised 12 of 13 isolates with high-level penicillin resistance and 5 of 96 with intermediate penicillin resistance. Only three of the 109 isolates had intermediate resistance to ceftriaxone (all with high-level penicillin resistance). Discussion We found that the level of penicillin resistance among S. pneumoniae isolates was six times higher than that found in 1989 in the only other large multicentre Australian study,7 but fortunately it was still lower than in most other countries. Penicillin resistance rates are very high in Third World countries, and in some areas of Western Europe and the USA.1-3 However, the rate of rise in resistance in Australia appears very similar to that seen in the early 1980s in countries such as Spain1-3 and in the early 1990s in the United States;6 there, only 0.02% of isolates nationally were penicillin-resistant in the early 1980s and still only 1.3% in 1992,6 but a recent study found a rate of 25%, with much higher rates in some subgroups (e.g., 40% in white children). Of equal concern was that 3% of isolates had high level resistance to both penicillin and third generation cephalosporins.6 Over the next few years, we are likely to see similar rates of resistance developing in Australia. The finding of high level penicillin resistance among S. pneumoniae isolates in Australia is of particular concern; in meningitis caused by organisms with any level of penicillin resistance, penicillin treatment is likely to fail.1-3,17,18 Penicillin resistance has consequences for other related drugs, as in S. pneumoniae it is not due to -lactamase production (as in Staphylococcus aureus ), but to changes in the target for penicillin (the penicillin-binding proteins).1,2,19 This change increases the MICs for all -lactams, including the third generation cephalosporins.1,2,19 However, the levels of third generation cephalosporins achieved in cerebrospinal fluid are still high enough to eradicate organisms with intermediate penicillin resistance.1-3,18 Alternative regimens include combination therapy with vancomycin, third generation cephalosporins and rifampicin, as well as newer agents such as meropenem, teicoplanin and quinolones (under investigation),1 but none has been adequately evaluated. Of even greater concern are the implications of the rapid rise in resistance to third generation cephalosporins noted in the United States. Primary resistance to these drugs is less frequent than to penicillin, but requires less genetic change.1,19 In some areas up to 27% of penicillin-resistant pneumococci have high level resistance to cefotaxime.1 This leads to therapeutic failure of these agents, yet they are the main treatment for the increasingly common intermediate penicillin-resistant strains. No high level cefotaxime-resistant strains were seen in our study or have been reported in Australia, to our knowledge. However, given the worldwide spread of resistant pneumococci in the recent past, they will inevitably be seen soon in Australia and leave us with major therapeutic dilemmas in the treatment of meningitis. In life-threatening situations other than meningitis (e.g., bacteraemia), high dose intravenous penicillin appears sufficient to eradicate organisms with intermediate resistance, as drug levels achievable in serum are still much higher than the MIC.1,2,3 There is, however, controversy, and many recommend use of either cefotaxime or ceftriaxone.1,3 For organisms with high level resistance, the most appropriate agent is unclear. However, we would favour vancomycin. In non-life-threatening infections with penicillin-resistant pneumococci, the most appropriate antibiotics are less clear. In otitis media, amoxycillin still appears the best choice,20,21 as drug levels achieved in the middle ear can still exceed the MICs of strains with intermediate resistance (although higher doses may be needed). Other oral agents available in Australia for use in children (cefaclor, trimethoprim, erythromycin and cefpodoxime) do not reach adequate levels to eradicate resistant isolates.20,21 Third generation cephalosporins, such as ceftriaxone, are active, but their parenteral route is likely to preclude their use. Combining clavulanic acid with amoxycillin is no advantage, as the resistance is not due to -lactamase. For high level penicillin-resistant isolates there does not appear to be a satisfactory oral agent. The reasons for the increasing resistance in S. pneumoniae worldwide are not completely understood, although antibiotic pressure appears to be a major factor.1 A few resistant clones were shown to have spread from one continent to others (e.g., from Spain to the United States and Iceland) and then through the local population, undergoing minor genetic changes in the process.1,3,19 The pneumococcus can acquire DNA molecules from other bacteria that probably include viridans group streptococci (e.g., Streptococcus mitis), which form part of the normal flora of the nasopharynx.1,19 While it would probably be impossible to eradicate carriage of these resistant organisms from the population, it may be possible to reduce the rate of increase in resistance by minimising the prescription of unnecessary antibiotics. Other strategies, such as vaccination, may be necessary. Unfortunately, the currently available vaccine is a polysaccharide and therefore a poor immunogen, especially in young children. Studies are under way to assess a conjugated pneumococcal vaccine (i.e., a carbohydrate with protein carrier), but vaccine development is difficult as there are over 80 serotypes of pneumococci (compared with only one commonly invasive serotype of Haemophilus influenzae -- type b). However, at present most of the resistant organisms belong to relatively few serotypes.1,2,3 A vaccine containing most of these might not only decrease life-threatening disease, but might also decrease carriage of the organisms, as was found for the H. influenzae type b (Hib) vaccine.1 However, as the pneumococcus can acquire DNA from other organisms,1,19 the number of resistant serotypes is likely to increase. Our study was one of the largest in the world where all organisms were clinically significant and all were assessed for MIC for penicillin. It is valuable not only for showing the rate of resistance (both intermediate and high level), but also for providing a baseline to assess future changes in resistance and to differentiate subtle shifts in resistance in the whole population of pneumococci from the introduction of resistant clones. In the past, determining MICs was time-consuming, laborious and not routine. The recent development of the Etest (which consists of a strip of paper impregnated with increasing concentrations of antibiotic from one end to the other) has simplified the procedure. This technological advance, along with the willingness of so many laboratories around Australia to participate in the project, has allowed us to obtain information essential for guiding us in making appropriate antibiotic choices and designing empiric therapy for these emerging threats. Acknowledgements We wish to thank the many doctors, scientists and technicians at the participating laboratories who donated their time and resources to carry out this project. The penicillin Etest strips were supplied by Australian Laboratory Services Pty Ltd at cost price. Eli Lilly provided funding for many of the participants to meet at the twice-yearly AGAR meeting. Cefotaxime and ceftriaxone Etest strips were donated by AB Biodisk (Sweden). References Lister PD. Multiply-resistant pneumococcus: therapeutic problems in the management of serious infection. Eur J Clin Microbiol Infect Dis 1995; 14 Supp 1: 18-25. Klugman K. Pneumococcal resistance to antibiotics. Clin Microbiol Rev 1990; 3: 171-196. Schreiber J, Jacobs M. Antibiotic-resistant pneumococci. Pediatr Clin North Am 1995; 42: 519-537. Collignon P. Penicillin-resistant pneumococci: will the recent Olympics bring back to Australia more than gold? Med J Aust 1992; 157: 655-657. Hansman D, Bullen M. A resistant pneumococcus. Lancet 1967; 2: 264-265. Hofmann J, Cetron M, Farley M, et al. The prevalence of drug resistant Streptococcus pneumoniae in Atlanta. N Engl J Med 1995; 333: 481-486. Collignon P, Bell J, on behalf of AGAR. Streptococcus pneumoniae : how common is penicillin resistance in Australia? Aust N Z J Med 1992; 22: 473-476. Hansman D, Morris S, Gregory M, McDonald B. Pneumococcal carriage amongst Australian aborigines in Alice Springs, Northern Territory. J Hyg (Camb) 1985; 95: 677-684. National Committee for Clinical Laboratory Standards. Performance standards for antimicrobial disk susceptibility tests. 5th ed. Approved standard. Document M2-A5. Villanova, Pa: NCCLS, 1993. Bell SM, Gatus BJ, Pham JN, et al. CDS users group newsletter No. 6. Sydney: The Prince of Wales Hospital, 1993. Bell SM. Additions and modifications to the range of antibiotics tested by the CDS method of antibiotic sensitivity testing. Pathology 1988; 20: 303-304. National Committee for Clinical Laboratory Standards. Methods for dilution antimicrobial susceptibility tests for bacteria that grow aerobically. 3rd ed. Approved standard. Document M7-A3. Villanova, Pa: NCCLS, 1993. Jorgensen JH, Ferraro MJ, McElmeel ML, et al. Detection of penicillin and extended spectrum cephalosporin resistance among Streptococcus pneumoniae clinical isolates by use of the Etest. J Clin Microbiol 1994; 32: 159-163. AB Biodisk . Etest Technical Guide 5B. Solna, Sweden: AB Biodisk , 1995. National Committee for Clinical Laboratory Standards. Performance standards for antimicrobial susceptibility testing. 5th informational supplement. Document M100-S5. Villanova, Pa: NCCLS, 1994. True Epistat [computer program]. Version 5.0. Richardson, TX: Epistat Services, 1994. Collignon P, Bell J, Hufton I, Mitchell D. Meningitis caused by a penicillin- and chloramphenicol-resistant Streptococcus pneumoniae . Med J Aust 1988; 149: 497-498. Friedland I, McCracken G. Management of infections caused by antibiotic-resistant Streptococcus pneumoniae . N Engl J Med 1994; 331: 377-382. Tomasz A. The pneumococcus at the gates. N Engl J Med 1995; 333: 514-515. Barnett E, Klein J. The problem of resistant bacteria for the management of acute otitis media. Pediatr Clin North Am 1995; 42: 509-517. Nelson C, Mason E, Kaplan S. Activity of oral antibiotics in middle ear and sinus infections caused by penicillin-resistant Streptococcus pneumoniae : implications for treatment. Pediatr Infect Dis J 1994; 13: 585-589. (Accepted 30 Nov 1995) Authors' details Infectious Diseases Unit, Woden Valley Hospital, Canberra, ACT. Peter J Collignon, FRACP, FRCPA, Head of Unit, Microbiologist and Infectious Diseases Physician. Department of Microbiology and Infectious Diseases, Monash Medical Centre, Melbourne, VIC. Jan M Bell, BSc(Hons), BA, Scientist. * Australian Group on Antimicrobial Resistance (AGAR). For this study AGAR consisted of the microbiology laboratories at: ACT: Woden Valley Hospital (Peter Collignon, Linda Halliday). NSW: Concord Hospital (Joan Yap, Tom Gottlieb, Glenn Funnell); Illawarra Regional Hospital (Keith Wise, Rodney Jones); Liverpool Hospital (Denise Daley, Rosemary Munro); Prince of Wales Hospital (Jeanette Pham, Barrie Gatus, Sydney Bell); Royal North Shore Hospital (Clarence Fernandes); Royal Prince Alfred Hospital (Richard Benn, Barbara Yan, Alison Vickery). QLD: Mater Misericordiae Hospital, Brisbane (Martyn Tilse, Janet Montgomery); Princess Alexandra Hospital (Graeme Nimmo, Jacqueline Schooneveldt); Royal Brisbane Hospital (Narelle George, Joan Faoagali); Sullivan, Nicolaides and Partners (Jenny Robson, Sylvia van der Valk); Toowoomba Base Hospital (David Farrell). SA: Flinders Medical Centre (Hendrik Pruul); Institute of Medical and Veterinary Science (Irene Lim, Richard Lumb); Queen Elizabeth Hospital (Peter Lawson, David Grove). TAS: Diagnostic Pathology (which includes Hobart Pathology and Launceston Pathology) (Barbara Henderson, Danny McColl, Gary Fenton); Launceston General Hospital (Erika Cox, Veronica Lyons); Royal Hobart Hospital (Keith Ott, Rob Peterson). VIC: Alfred Hospital (John Spicer, J Clare Franklin); Dorevitch Pathology (Liz Snashall); Heidelberg Repatriation Hospital (Barrie Mayall, Angie Chan, Vicki Moritz); Melbourne Pathology (Christine Hargreaves); Monash Medical Centre (Dianne Olden, Jan Bell, John Turnidge); Royal Children's Hospital and Microbiological Diagnostic Unit (Geoff Hogg, Marion Easton, Janet Strachan). WA: Fremantle Hospital (David McGechie, Neil Stingemore, Graham Francis); Royal Perth Hospital (Keryn Christiansen, Claire Khinsoe, Geoff Coombs). Reprints: Dr P J Collignon, Infectious Diseases Unit, Woden Valley Hospital, PO Box 11, Woden, ACT 2606. ©MJA 1997 <URL: http://www.mja.com.au/> © 1997 Medical Journal of Australia.

Peter J Collignon · Jan M Bell · the AGAR

Research Physical, sexual and emotional violence against women: a general practice-based prevalence study Danielle Mazza, Lorraine Dennerstein and Vicky Ryan Abstract - Authors' details - Introduction - Methods - Results - Discussion - Acknowledgement - References - Box 1 - Box 2 - Box 3 - Box 4 - Box 5 - ©MJA1996 - Abstract Objective: To determine the prevalence of domestic violence, childhood abuse and sexual assult experienced by women attending general practitioners. Design: A cross-sectional, questionnaire-based prevalence survey. Setting: 15 general practices in metropolitan Melbourne between November 1993 and February 1994. Subjects: 3026 women over the age of 18 attending for a consultation. Results: The response rate was 72%. Over a quarter of women in relationships had been victims of physical or emotional partner abuse in the previous year, one in 10 having experienced severe physical violence. Thirteen percent of women had experienced rape or attempted rape, 10% had been severely beaten during childhood and 28% had experienced childhood sexual abuse involving physical contact. The abuse had been disclosed to the woman's doctor by only 27% of those who had experienced partner or childhood physical abuse (mostly because the doctor had never asked) and 9% of those who had experienced sexual abuse (mostly because the woman did not see it as relevant to the consultation). Conclusion: There is a high prevalence of physical, sexual and emotional violence against women as well as poor communication about this violence to their general practitioners. Recommendation: Medical practitioners should be more proactive in questioning women about violence. (MJA 1996; 164: 14-17) Introduction Violence experienced by women results in significant morbidity and, in some cases, mortality. The experience of sexual abuse as a child has been linked to later development of psychological disorders1 and drug abuse and dependence.2,3 Domestic violence has health effects beyond the acute injuries; battered women are more likely to suffer from somatic complaints, anxiety and depression,4 pelvic pain,5 and sexual and gynaecological problems.6 They also use health services more often than women not subjected to domestic violence.7 Because of these associations and because victims of violence are more likely to turn to doctors for help than to any other person,8 they are highly likely to frequent a doctor's surgery. However, doctors are not skilled at recognising them and have been estimated to diagnose only one battered woman in 25.7 This may occur because doctors are unaware of the extent of physical and sexual abuse experienced by their female patients; while the prevalence has been studied in the United States and Europe, Australian data are sparse. The single authoritative study on the prevalence of childhood sexual abuse in Australia was conducted by Goldman and Goldman on a population of university students: 28% of female students and 9% of male students had experienced sexual abuse.9 Two recent studies have looked at the prevalence of domestic violence in Australia, but only in select populations: 23.3% of women attending an emergency department disclosed histories of domestic violence,10 while 8.9% of women in a hospital antenatal clinic stated that they had experienced physical abuse during their pregnancy.11 The aim of our study was to determine the prevalence of domestic violence, childhood abuse and sexual assault experienced by women attending general practitioners and to provide doctors with accurate information on the extent of the problem. Methods The study was a questionnaire-based prevalence survey carried out between November 1993 and February 1994. It was approved by the ethics committee of Monash University. Study population The study population comprised women over the age of 18 attending their general practitioner for a consultation. Fifteen general practices in metropolitan Melbourne were selected in a two-stage random sampling design described previously.12 The design incorporated practices that were broadly representative of all social classes, from all regions of metropolitan Melbourne. A total sample size of about 3000 women was calculated to be necessary to ensure adequate power, based on results of a pilot study; 220 questionnaires were therefore distributed to each practice. Questionnaire Consecutive women attending the practice for a consultation were invited to participate by the practice receptionist. The questionnaire was introduced by a covering letter, which explained the nature of the study, that it was voluntary and confidential, and that the information disclosed would not be entered in their medical file or given to their doctor. It was acknowledged that some questions might cause distress, and the questionnaire could therefore be completed either in the waiting room or at home and returned in a reply-paid envelope. Respondents were also given contact phone numbers of support services for the different forms of abuse, and informed that their doctor was happy to discuss with them any issues that might arise as a result of the survey. Before the study, doctors were given an information package with details of local support services for abused women. The self-administered questionnaire asked first for demographic details. Respondents then completed the Conflict Tactics Scale,13 with the modification that they were asked whether the tactic had occurred never, once or more than once in the last year, and with the addition of questions on emotional abuse. In accordance with the Conflict Tactics Scale, physical violence was classified as minor or severe (see Box 1). Questions about sexual abuse were derived from the studies of Wyatt14 and Russell,both of 15 which used multiple screening questions to allow time for the respondent to become accustomed to the nature of the questions. Childhood sexual abuse was classified as contact or non-contact (Box 1). Data were entered into a Microsoft Access database. Frequency tables were generated and prevalences calculated. Because the data came from 15 different general practices and not a simple random sample, confidence intervals (CIs) were adjusted for the effects of clustering.16 Results Of 3026 questionnaires distributed, 2181 were returned (response rate, 72%). Most questionnaires were completed in the waiting rooms of the practices, with only 18% of those answered returned by post. Domestic violence Prevalences for the different categories of domestic violence in the previous year are shown in Box 2. Only those women in a current relationship were asked to complete the section about domestic violence, so that the sample size was smaller than for other parts of the study. A total of 28% of these women had experienced either physical or emotional partner abuse, or both, in the previous year, and (notably) almost one in 10 had been victims of severe physical violence in that year. Among respondents in a current relationship 6% had been kicked, bitten or hit with a fist; 7% had been hit or their partner had tried to hit them with an object; 4% had been beaten up; 4% had been choked; 2% had been threatened with a knife or a gun; and 1% had actually had a knife or gun used against them. Twenty per cent of those in a current relationship had experienced emotional abuse in the previous year; 4% had their partner threaten or try to kill them; 8% had money withheld; 7% were prevented from leaving their home; 6% were stopped from seeing their friends and family or speaking to them on the phone; and 17% were constantly called names or humiliated. Adult sexual abuse Overall, 30% of women had been victim to some form of sexual abuse since the age of 16. Prevalences of different forms of abuse are shown in Box 3. Just over half of those who reported adult sexual abuse (356/626) had experienced more than one kind. Childhood abuse Overall, 10% of women experienced childhood physical abuse (95% CI, 8%-12%); 3% were severely beaten on one occasion and 7% repeatedly. Almost 40% of women had experienced some form of sexual abuse before the age of 16 (Box 3). Communication in general practice about violence The question about whether respondents had ever discussed the issues of domestic violence or childhood physical abuse with their doctors was answered by 1177 women. Among these only 27% had done so, although it was more likely among victims of domestic violence than among non-victims (Box 4). Most women (73%) said that their doctor had never asked them about these things, although this was less likely for victims than for non-victims. If respondents had experienced either childhood or adult sexual abuse, they were asked if they had ever disclosed this to their doctor; 1009 responded, only 87 (9%) in the affirmative. Reasons given for not disclosing are shown in Box 5. Discussion We believe that this study is the first to show the prevalence of physical, sexual and emotional abuse of women in an Australian general practice population. While we recognise that domestic violence and sexual abuse are not exclusively directed by men against women, the study found a high level of violence against women. Over a quarter of women in relationships had been victims of partner abuse in the previous year and one in 10 had experienced severe physical violence; 13% of women had experienced rape or attempted rape; 10% of women had been severely beaten during childhood; and 28% had experienced contact childhood sexual abuse. Despite these levels, few women disclose these events to their doctors. Our findings are similar to those of a family practice-based study in the United States, which found that 23% of women had been physically assaulted by their partners in the last year.17 In contrast,population-based studies in the United States and Canada estimate that between 10% and 14% of women in relationships experience physical abuse over a one-year period; for 3%-5% of women, the abuse is severe.18-22 Because domestic violence is associated with injury and illness, women sampled in medical environments would be expected to have a higher prevalence than women sampled in community settings. In fact, our findings may underestimate the true prevalence of domestic violence, as women who were separated or divorced were not questioned about their experience of it. We found a slightly higher prevalence of sexual abuse than that in a recent New Zealand community-based study, which found the prevalence of childhood sexual abuse overall to be 32%, with contact abuse in 25% and penetrative abuse in 4%.23 The differences are again probably due to the different populations sampled. Potential sources of bias exist. Because of the secret nature of physical and sexual abuse and the stigma attached, there is much controversy over whether self-disclosure by victims can give a true indication of prevalence. A self-administered questionnaire gives no opportunity for clarification of responses by an interviewer. It is also argued that victims will not respond because they fear that disclosure will cause further trauma. These factors would lower apparent prevalence.24 An alternative argument is that victims respond preferentially to surveys when given the opportunity to disclose; non-victims fail to respond as they feel they have no valid information to contribute.24 This would have the reverse effect on apparent prevalence; the two effects could simultaneously counterbalance each other. Another possible source of bias is that the self-administered questionnaire format may have precluded the participation of women from non-English-speaking backgrounds. However, with a sample size of over 3000 and a response rate of 72% (with over 80% of the surveys completed in the general practitioner's surgery), the questionnaire appears to have been well accepted by the subjects. Importantly, about three-quarters of respondents had never been asked by their doctors about domestic violence or childhood physical abuse. This is consistent with results of studies in other countries, which show physician enquiry rates into spouse abuse to be suboptimal.22,25 Contrary to prevailing belief,26 the main reason women did not discuss these issues with their doctor was not because they were afraid, embarrassed or untrusting, but because they were never asked. In addition, 53% of women had not disclosed their experiences of sexual abuse to their doctor because they had never found it relevant to the consultation. Either women are failing to make the connection between sexual abuse and their symptoms or their doctors lack knowledge of the short and long term health effects of sexual abuse. Doctors are crucially placed to deal with the problem of abuse. Our study suggests that medical practitioners must be more proactive in questioning women about violence. While many may hesitate to identify something they may be unable to directly treat,27 it is important to consider that when a diagnosis of abuse is missed treatment is likely to be inappropriate and potentially harmful.28 Detection is the first step in successful management to deal with both the immediate and long term effects of violence against women. This requires that medical practitioners not only develop the skills to diagnose violence perpetrated against women, but have knowledge of local agencies for referral as well as the legal and criminal options available to the woman. Acknowledgement This work was supported by a grant from the Shepherd Foundation in the Department of Community Medicine at Monash University. (©MJA 1996; 164: 14-17) References Mullen PE, Romans-Clarkson SE, Walton VA, Herbison GP. Impact of sexual and physical abuse on women-s mental health. Lancet 1988; 1: 841-845. Burnam MA, Stein JA, Golding JM, et al. Sexual assault and mental disorders in a community population. J Consult Clin Psychol 1988; 56: 843-850. Winfield I, George LK, Swartz M, Blazer DG. Sexual assault and psychiatric disorders among a community sample of women. Am J Psychiatry 1990; 147: 335-341. Jaffe P, Wolfe DA, Wilson S, Zak L. Emotional and physical health problems of battered women. Can J Psychiatry 1986; 31: 625-629. Schei B. Psycho-social factors in pelvic pain: a controlled study of women living in physically abusive relationships. Acta Obstet Gynecol Scand 1990; 69: 67-71. Schei B, Bakketeig LS. Gynaecological impact of sexual and physical abuse by spouse. A study of a random sample of Norwegian women. Br J Obstet Gynaecol 1989; 96: 1379-1383. Stark E, Flitcraft A, Zuckerman D, et al. Wife abuse in the medical setting. An introduction for health personnel. Domestic Violence Monograph Series No. 7. Washington, DC: US Government Printing Office, 1981. Dobash RE, Dobash RP. Violence against wives - a case against the patriarchy. New York: Free Press, 1979. Goldman R, Goldman J. The prevalence and nature of child sexual abuse in Australia. Aust J Sex Marriage Fam 1988; 9: 94-106. Roberts GL, O-Toole BI, Lawrence JM, Raphael B. Domestic violence victims in a hospital emergency department. Med J Aust 1993; 159: 307-310. Webster J, Sweett S, Stolz TA. Domestic violence in pregnancy. A prevalence study. Med J Aust 1994; 161: 446-470. Mazza D, Dennerstein L, Ryan V. Psychotropic drug use by women: current prevalence and associations. Med J Aust 1995; 163: 86-89. Straus MA. Measuring intrafamily conflict and violence: the Conflict Tactics (CT) Scales. J Marriage Fam 1979; 41: 75-88. Wyatt GE. The sexual abuse of Afro-American and white-American women in childhood. Child Abuse Negl 1985; 9: 507-519. Russell DEH. The incidence and prevalence of intrafamilial and extrafamilial sexual abuse of female children. Child Abuse Negl 1983; 7: 133-146. Rao JNK, Scott AJ. A simple method for the analysis of clustered binary data. Biometrics 1992; 48: 577-585. Hamberger LK, Saunders DG, Hovey M. Prevalence of domestic violence in community practice and rate of physician inquiry. Fam Med 1992; 24: 283-287. Rollins BC, Oheneba-Sakyi Y. Physical violence in Utah households. J Fam Violence 1990; 5: 301-309. Schulman MA. A survey of spousal violence against women in Kentucky. Study No. 792701 conducted for the Kentucky Commission on Women. Washington, DC: US Government Printing Office, 1979. Smith MD. The incidence and prevalence of woman abuse in Toronto. Violence Vict 1987; 2: 173-187. Straus MA, Gelles RJ, Steinmetz SK. Behind closed doors: violence in the American family. New York: Anchor, 1980. Straus MA, Gelles RJ. Societal change and change in family violence rates from 1975 to 1985 as revealed by two national surveys. J Marriage Fam 1986: 48; 465-479. Anderson J, Martin J, Mullen P, et al. Prevalence of childhood sexual abuse in a community sample of women. J Am Acad Child Adolesc Psychiatry 1993; 32: 911-919. Finkelhor D. A sourcebook on child sexual abuse. Beverley Hills: Sage Publications, 1986. Martins R, Holzapfel S, Baker P. Wife abuse: are we detecting it? J Wom Health 1992; 1: 77-80. Queensland Domestic Violence Task Force. Beyond these walls. Brisbane: Queensland Government, 1988. Brown JB, Sas G. Focus groups in family practice research: an example study of family physicians- approach to wife abuse. Fam Pract Res J 1994; 14: 19-28. Council on Ethical and Judicial Affairs, American Medical Association. Physicians and domestic violence. Ethical considerations. JAMA 1992; 267: 3190-3193. (Received 19 May, accepted 6 Oct 1995) Authors' details University of Melbourne, Melbourne, VIC. Danielle Mazza, FRACGP, DRACOG, Lecturer, Key Centre for Women's Health; formerly Assistant Lecturer, Department of Community Medicine, Monash University, Melbourne, VIC. Lorraine Dennerstein, AO, PhD, FRANZCP, Director, Key Centre for Women's Health. Vicky Ryan, MSc, Statistician, Statistical Consulting Centre. (©MJA 1996; 164: 14-17)

Danielle Mazza · Lorraine Dennerstein · Vicky Ryan

Human hydatidosis in New South Wales and the Australian Capital Territory, 1987-1992 David J Jenkins and Karen Power Abstract - Authors' details - Introduction - Methods - Results - Discussion - Acknowledgement - References - Box 1 - Box 2 - Figure 1 - Figure 2 - Figure 3 - ©MJA1996 - For editorial comment, see McCullagh Objective: To determine the prevalence of human hydatidosis in New South Wales and the Australian Capital Territory. Methods: Data on human hydatid infection occurring between 1987 and 1992 were collected retrospectively from 25 hospitals and 13 health services in New South Wales and four hospitals in the Australian Capital Territory. Mean annual prevalences of human hydatidosis were determined for shires in eastern New South Wales and data on infection in immigrants and Australian-born patients were compared. Results: 321 patients were diagnosed with hydatid disease, 1987-1992; 195 were new cases and 117 readmissions (nine cases were not identified as new or recurrent). Most patients lived in the eastern half of New South Wales (which includes the Australian Capital Territory), half in rural areas and half in the major coastal cities. Most Australian-born rural patients lived in 39 shires in the north-eastern and south-eastern Tablelands. Sixty per cent of the patients in major cities were born overseas. Conclusions: Hydatid infection occurs more commonly in south-eastern Australia than the official figures suggest. In rural areas of the north-eastern and south-eastern Tablelands hydatid infection is of public health importance. The national notification system must be improved and control campaigns alerting the public to the dangers of hydatid infection promoted. (MJA 1996; 164: 18-21) Introduction The tapeworm genus Echinococcus is an important zoonosis which is endemic in many parts of the world. The only species occurring in Australia is Echinococcus granulosus. It was probably introduced with infected domestic livestock during European settlement and is now widespread in domestic livestock and wildlife, with wildlife acting as an important reservoir.1 Dogs (domestic and wild) and foxes are the definitive hosts (Figure 1). Humans become infected by the ingestion of eggs passed in faeces of dogs. Oncospheres released from the eggs penetrate the intestinal mucosa and, via the portal system, lodge in the liver, lungs, muscle or other organs, where the hydatid cysts form. Because of inadequate reporting, the prevalence of human hydatid infection in Australia is unknown. From the earliest published studies human hydatidosis in New South Wales has occurred mainly in people living in rural areas in the eastern half of the State associated with the Great Dividing Range.2-4 Dew, in 1928, reported a relatively even distribution of patients with hydatidosis in eastern New South Wales, but subsequent reports showed an increasing trend for patients to be concentrated in the north-eastern and south-eastern Tablelands.3-5 To assess the health risk associated with E. granulosus, a retrospective survey of hydatid infection was conducted between 1987 and 1992 by examining records of patients with hydatidosis from hospitals and health services in New South Wales (NSW) and the Australian Capital Territory (ACT). Methods All the public and private hospitals and area and district health services in NSW and the ACT were asked to supply data of patients with confirmed hydatidosis who were admitted between January 1987 and December 1992. After approval of our written request, all institutions ( four hospitals in the ACT and 25 hospitals and 13 health services in NSW ( supplied data comprising: - Sex; - Date of admission; - Date and country of birth; - Place of residence at admission; - Cyst location; and - Whether the infection was new or recurrent. We maintained patient confidentiality by using initials only for individual identification. Multiple admissions for the same patient were identified from admission dates, initials, age, sex and general location of residence at the time of admission. We calculated mean annual prevalences of infection using 1991 Census data.6 Results Three hundred and twenty-one patients with confirmed hydatidosis were treated between 1987 and 1992. These comprised 195 new cases (107 males and 88 females), 117 recurrent cases (27 cases had their first treatment before this survey began) and nine cases not identified as new or recurrent. Two hundred and eighty-two patients (172 new cases and 110 recurrent cases [including those not classified as new or recurrent]) were treated in NSW and 39 patients (23 new cases and 16 recurrent and unclassified cases) were treated in the ACT (16 and 14, respectively, of those treated in the ACT lived in NSW). Rural patients Except in three cases, hydatid infection in rural patients, most of whom were Australia-born, occurred in the eastern half of NSW at higher altitudes, mainly associated with the Great Dividing Range (Figure 2). There were concentrations of patients in the north-eastern and south-eastern Tablelands, and these two areas were connected by a corridor parallel to the coast where further cases occurred. The mean annual prevalence of human hydatidosis in rural NSW was 2.6 cases per 100 000 rural population. Cases occurred in 15 shires [counties] in the north-east and 24 shires in the south-east. On a shire-to-shire basis, the mean annual prevalence of infection ranged from 0.3 to 17.7 and 0.5 to 23.5 (cases per 100 000 population), respectively, in these two areas (Box 1). Four of the cases (three new and one recurrent) were in Aboriginal people. These four cases represented a mean annual prevalence of hydatid infection of 1.1 cases per 100 000 in the Aboriginal population of NSW. Urban patients There were 152 cases diagnosed from the three major metropolitan centres of NSW (Sydney, Newcastle and Wollongong), which included 98 new cases, mostly in patients born overseas (60%); in rural areas the reverse was evident (85% of rural patients were born in Australia). Of the patients born overseas, all were living in NSW, except one ACT resident. The mean annual prevalence of infection was calculated for 25 ethnic groups (each with a population of over 1000) resident in NSW (Box 2). The highest mean annual prevalence occurred in the Iranian population (6.6 cases/100 000) and the lowest in the German population (0.5 cases/100 000). Most cases came from the Greek and Lebanese communities and communities of people from the former Yugoslavia (13, 10 and 8, respectively), but because of their relatively large populations in NSW, these cases represented only a mean annual prevalence of 4.8, 3.2 and 2.2, respectively. Age distribution Age-group distribution profiles of the patients born in Australia and those born overseas are compared in Figure 3. All age groups are represented among Australian-born patients with hydatidosis, especially the older age groups, whereas the immigrant population, being generally younger, has fewer cases in the older age groups. Cyst location Infection in the liver occurred most commonly (157 of the 195 new cases); 13 cases involved infection in the lungs and four had infection in both liver and lungs. Infection in other less common sites were two each in the spleen, pancreas and leg muscle and one each in the diaphragm, pelvic area, arm muscle, brain, adrenal gland and gallbladder. Discussion Retrospective survey data on human hydatidosis cannot give an accurate picture of the prevalence of infection. A number of cases are not seen in hospitals because the infection is asymptomatic, or does not require surgical intervention, and mistakes in coding may occur. However, these data remain a useful indication of infection prevalence. Our study confirmed the concentration of hydatidosis in the north-eastern and south-eastern Tablelands reported previously.3-5 The narrow corridor parallel to the coast in the central part of the State is an area where there have been few cases reported previously, but where considerable urban development has occurred over the last decade, and previously undiagnosed patients may have moved to this region. Population movement from country areas to cities may also account for many of the Australian-born patients diagnosed in urban areas. However, it is also possible for urban residents to be exposed to eggs of E. granulosus. Recent studies have identified infection with E. granulosus in dogs of a recreational pig hunter living in suburban Perth,7 and in dogs of Perth residents living in uncleared areas on the outskirts of the city.8 Foxes infected with E. granulosus have been found in the suburbs of Canberra.9 The older age of the Australian-born compared with the immigrant patients largely reflects the different age-group structures of the two groups. In 1990, 88.3% of immigrants were aged less than 45 years when they arrived in Australia and 27.4% were less than 14 years of age.10 Migrants may be already infected when they arrive as children, but the long latent period of hydatid disease means it is first detected in adulthood. It is difficult to explain why most Australian-born patients were detected in the age group 31 to 40 years whereas most immigrants were not detected until 41 to 50 years. Infections in immigrants may be caused by a different strain type of E. granulosus with a slower cystic growth rate. Alternatively, there may be a reluctance among newer immigrants to consult local doctors. The migrants infected with hydatid disease origin(Box 2)ated in countries where E. granulosus is endemic. The order of ranking of countries according to the mean annual prevalence of hydatid infection in their migrant populations in NSW closely reflected the relative importance of human hydatidosis in their countries of origin. The range of prevalences in the migrant populations were no higher than those recorded in shire populations in north-eastern and south-eastern NSW. In at least three of these shires the prevalence of human hydatidosis was two to three times higher than the highest level recorded in a migrant population. Three of the urban patients and one of the rural patients born in Australia were Aboriginals. The three urban Aboriginal patients are likely to be from a rural background, but as their place of birth was unknown it was not possible to calculate a prevalence of hydatid infection for rural Aboriginal people. The mean annual prevalence of 1.1 cases per 100 000 for the total Aboriginal population of NSW is about a third of the prevalence in the rural non-Aboriginal population. Few cases of hydatid disease in Aboriginal people have been reported. All the reports are from studies in Western Australia during the 1970s, where Aboriginals were always highly represented: 15/57 cases11 and 13/31 cases.12 Our data represent the first report of hydatid infection in Aboriginal people in NSW; a previous study reported E. granulosus infection in a dog from a NSW Aboriginal community.13 The reason for human hydatidosis not being perceived as a problem in Australia can be attributed largely to under-reporting of this notifiable disease;5 this has been a problem for many years,,5,12,14,15 with no signs of improvement. Only 17 of the 321 new and recurrent cases identified in this study had been notified, and in a retrospective study in Victoria for the 12 months up to July 199116 only two of the 50 new or recurrent cases had been notified. Disease recurrence after operative treatment is an important aspect of human hydatid infection. A carefully conducted follow-up study of 39 patients treated surgically in Australia first drew attention to this problem;17 22% had had recurrent infection by 30 months, mainly caused by cyst rupture before surgery. In our study, 37.5% of patients (for whom information on new or recurrent infection status was supplied) were treated for recurrent infection. Effective chemotherapy of patients with hydatid infection would substantially reduce the cost of treatment. This topic has been reviewed,18 and the most promising drug studied was albendazole. Data from studies on 253 patients indicated that albendazole was an effective cure in 28%, 51% showed improvement, 18% were unchanged and in 2% the cysts continued to grow.19 The most appropriate use of albendazole may be as an adjunct to surgery. Rupture of cysts and spilling of protoscoleces (which can form new cysts) into the body cavity during surgery is a constant risk, but an immediate postoperative course of albendazole will greatly reduce the chance of new cysts developing.20 Hydatid disease is preventable, and education is one of the most effective tools to achieve this. It is important that State and Federal governments take a responsible attitude towards increasing community awareness, and implement control strategies through education, either by themselves or by funding organisations such as the Australian Hydatid Control and Epidemiology Program. Control of this parasite in Australia requires a long term commitment; the alternative is that hydatid disease will continue to incapacitate individuals and be an additional financial drain on an already overstretched health service. Acknowledgements The authors gratefully acknowledge the assistance of the staff of the medical records departments of the following hospitals: Albury, Armidale and New England, Bathurst, Broken Hill, Calvary (ACT), Camperdown, Casino and District, Cooma, Dubbo, Goulburn, Grafton, Grenfell, Griffith, Inverell, John James Memorial (ACT), Lismore, Orange, Parkes District, Prince Henry, Prince of Wales Children's, Queanbeyan, Royal Canberra (ACT, now closed), Royal North Shore, Royal Prince Alfred, St Vincent's, St Vincent's Private, Tamworth, Wagga Wagga, Westmead and Woden Valley (ACT); also the following area and district health services: Brunswick-Byron, Central Coast, Cooma, Hunter, Illawarra, Macleay Valley, Manning Valley, Queanbeyan, South Western Sydney, Southern Sydney and Tumut, Wentworth. We also thank Ms Celia Moss (Australian Bureau of Statistics) and Ms Irene Pasaris (ACT Health Department) for their help and advice and Dr M W Lightowlers, Dr E Bennet, Dr P McCullagh and Professor R C A Thompson for their suggestions in the preparation of the manuscript. This study was funded partly through contributions from the Shire Councils of Bega Valley, Boorowa, Cooma-Monaro, Crookwell, Gunning, Harden, Snowy River, Tumbarumba, Yarrowlumla, Yass and Young, Queanbeyan City Council and the ACT Health Authority. Dr Jenkins' salary was funded by the National Health and Medical Research Council of Australia. (©MJA 1996; 164: 14-17) References Schantz PM, Chai J, Craig PS, et al. Epidemiology and control of hydatid disease. In: Thompson RCA and Lymbery AJ, editors. The biology of Echinococcus and hydatid disease. Wallingford, UK: C A B International, 1995: 233-302. Dew HR. Hydatid disease. Its pathology, diagnosis and treatment. Sydney: The Australasian Medical Publishing Company Ltd, 1928. Christopher PJ, Lopez WA. Hydatid disease notifications in New South Wales. Med J Aust 1970; 1: 54-56. Little JM. Hydatid disease at Royal Prince Alfred Hospital, 1964 to 1974. Med J Aust 1976; 1: 903-908. Schreuder S. Survey of hospital admissions for hydatidosis in New South Wales and the Australian Capital Territory, 1982-1987. Aust Vet J 1990; 67: 149-151. Australian Bureau of Statistics. 1991 Census of population and housing. State comparisons. Canberra: ABS, 1993. (Catalogue No. 2731.0.) Thompson RCA, Lymbery AJ, Hobbs RP, Elliot AD. Hydatid disease in urban areas of Western Australia: an unusual cycle involving western grey kangaroos (Macropus fuliginosus), feral pigs and domestic dogs. Aust Vet J 1988; 65: 188-190. Thompson RCA, Robertson ID, Gasser RB, Constantine CC. Hydatid disease in Western Australia: a novel approach to education and surveillance. Parasitol Today 1993; 9: 431-433. Jenkins DJ, Craig NA. The role of foxes Vulpes vulpes in the epidemiology of Echinococcus granulosus in urban environments. Med J Aust 1992; 157: 754-756. Australian Bureau of Statistics. Migration, Australia. Canberra: ABS, 1994. (Catalogue No. 3412.0.) Joske RA. The changing pattern of hydatid disease, with special reference to hydatid of the liver. Med J Aust 1974; 1: 129-132. Stein GR, McCully DJ. Hydatid disease in Western Australia (1957-1967). Med J Aust 1970; 1: 848-850. Jenkins DJ, Andrew PL. Intestinal parasites in dogs from an Aboriginal community in New South Wales. Aust Vet J 1993; 70: 115-116. Davies P, Nicholas WL, Beard TC. Hospital records of hydatid disease in Victoria for 1970 to 1974. Med J Aust 1977; 2: 493-495. Beard TC. Hydatids in Australia ( the present position in man. Aust Vet J 1979; 55: 131-135. Taylor K. Hydatids in 1992: public health lessons. Update. A quarterly bulletin of infectious diseases. Victoria: Department of Health and Community Services, 1993; 2: 63-64. Little JM, Hollands MJ, Ekberg H. Recurrence of hydatid disease. World J Surg 1988; 12: 700-704. Morris DL, Richards KS. Hydatid disease current medical and surgical management. Oxford: Butterworth-Heinemann Ltd, 1992: 94-118. Horton RJ. Chemotherapy of Echinococcus infection in man with albendazole. Trans R Soc Trop Med Hyg 1989; 83: 97-102. Morris DL, Taylor DH. Optimal timing of postoperative albendazole prophylaxis in E. granulosus. Ann Trop Med Parasitol 1988; 82: 65-66. (Received 11 Apr, accepted 27 Sep 1995) Authors' details Australian Hydatid Control and Epidemiology Program, Canberra, ACT. David J Jenkins, MSc, PhD, Research Officer. Karen Power, BApplSci, Field Officer.

David J Jenkins · Karen Power