Digestive system diseases

Autoimmune hepatitis is a rare but increasingly recognised serious complication of treatment with the tumour necrosis factor-α (TNF-α) blocking agent, infliximab. This report adds to the small but growing number of articles describing this significant adverse effect. Baseline liver function should be routinely tested in all patients receiving anti-TNF-α agents, and periodic monitoring for the development of hepatitis is important. Clinical recordA 56-year-old woman was referred with a 15-year history of ankylosing spondylitis, with clinical features of inflammatory back pain, fused sacroiliac joints, spondylosis and a positive test result for human leukocyte antigen (HLA)-B27. She was an ex-smoker and non-drinker, and was taking no regular medication. She had taken non-steroidal anti-inflammatory agents initially, but tolerated them poorly. Leflunomide therapy was trialled for 9 months in 2004, then ceased because of a poor clinical response; this therapy had no adverse effects (including hepatic complications) in this patient. Infliximab (infusion dose, 5 mg/kg) was introduced in 2005. Baseline findings included normal liver function test (LFT) results, and a negative anti-nuclear antibody (ANA) test result. During infliximab therapy, routine blood tests (including LFTs) were performed every 3 months. A minor elevation in transaminase level was noted within 3 months of commencing infliximab therapy. After a total of six infusions, infliximab therapy was ceased because the patient developed epigastric discomfort, jaundice and pale stools, and her LFT results showed gradual deterioration, with a mixed hepatitic/cholestatic picture. Results of her blood tests at cessation of infliximab therapy are shown in Box 1. Computed tomography of the abdomen showed diffusely non-homogeneous liver texture consistent with chronic liver disease, with no intra- or extrahepatic duct dilatation. A past cholecystectomy was also evident. A liver biopsy showed marked inflammatory infiltrate, particularly around the portal tract, with neutrophils, lymphoid, and plasma cells, bridging necrosis and piecemeal necrosis consistent with autoimmune chronic active hepatitis (Box 2). Prednisone therapy was commenced and, after 3 months, the patient’s clinical condition had improved with the only persisting liver function abnormality being an elevated γ-glutamyl transferase level (162 U/L; normal, 5–30 U/L). At this time the patient tested positive for ANA (titre, 1/160, homogeneous pattern); mildly positive for double-stranded DNA antibody (anti-dsDNA; titre, 5.7 IU/mL); and positive for smooth muscle antibody (titre, 1/2560). An attempt to introduce azathioprine therapy was unsuccessful because the patient developed a drug fever. DiscussionTumour necrosis factor-α blocking agents (anti-TNF-α agents) are used for treating conditions such as Crohn’s disease, ankylosing spondylitis, rheumatoid arthritis, and psoriatic arthritis. Commercially available anti-TNF-α agents include etanercept, infliximab, and adalimumab. Both infliximab and etanercept have been evaluated in double-blind placebo-controlled studies for the treatment of Crohn’s disease, rheumatoid arthritis, and ankylosing spondylitis. Minor side effects of treatment with anti-TNF-α agents are common, and include upper respiratory tract infections, rash, myalgias, asthenia, sinusitis, flushing, fever and headache.1 More serious side effects include the reactivation of latent Mycobacterium tuberculosis; reactivation of hepatitis B in chronic carriers; infusion reactions (related to induction of anti-chimeric antibodies against infliximab); severe neutropenia and thrombocytopenia; demyelinating disorders; autoimmune antibody formation; and hepatotoxicity. Minor abnormalities in LFT results are relatively common with the use of anti-TNF-α agents.1 Severe hepatic reactions are much less common. These may include jaundice, hepatitis and cholestasis, autoimmune hepatitis, and acute liver failure. It is noteworthy that a number of cases of liver failure resulting in liver transplantation or death have now been reported.2,3 According to the manufacturer of infliximab, at least 35 patients were reported to have had severe hepatic adverse events in the postmarketing period up to 2004.4 Autoimmune hepatitis (AIH) is a rare but nonetheless increasingly recognised complication of treatment with infliximab. Ten cases of AIH had been reported to the manufacturer of infliximab up to 2003, eight of which were considered either possibly or probably related to infliximab (H B, personal correspondence). However, events had been submitted voluntarily and thus may be under-reported. This report is important in highlighting a case of the de-novo development of autoantibodies and onset of AIH related to the use of infliximab in a patient not receiving concurrent confounding immunosuppressive therapy such as methotrexate. It adds to the small but increasing number of published cases of these adverse effects. Most cases published to date describe reversibility of the hepatitis and resolution of LFT abnormalities with immunosuppressive therapy. Box 3 shows a comparison of our case and several other published cases of postulated infliximab-related AIH, including a case of acute liver failure which led to liver transplantation.2 Neither that patient2 nor our patient received concurrent immunosuppressive therapy during treatment with infliximab. Autoimmune manifestations may possibly be more significant when infliximab is used alone, or in higher doses. This is often the situation with spondyloarthropathy or psoriasis, where additional immunosuppression may not necessarily be part of the protocol. This differs from the treatment of, for example, rheumatoid arthritis, for which patients often also take methotrexate. One published report describes the development of autoantibodies and liver inflammation following infliximab therapy in two women treated with infliximab for rheumatoid arthritis.2 Liver disease occurred 8 and 17 months, respectively, after infliximab therapy was initiated, and there was de-novo development of ANA in both patients. In the first and most striking case, a 39-year-old woman eventually required liver transplantation after a 45-day admission with hepatic failure and cirrhosis. She had previously been taking leflunomide (20 mg/day for 3 months) with no significant response. Before infliximab therapy, her LFT results were normal, and she tested negative for ANA. At last report, post-transplant, she was well on immunosuppressive therapy including cyclosporin and corticosteroids. In the second case, a 54-year-old woman with rheumatoid arthritis developed acute hepatitis after infliximab treatment. A complete recovery (based on symptoms and LFT results) was noted 1 month later, after treatment with prednisolone (0.5 mg/kg/day) and azathioprine (50 mg/day). Two major types of AIH have been described: type 1 classic AIH, as seen in our patient (related to ANA and/or smooth muscle antibody seropositivity; and type 2 (related to liver/kidney microsomal antibody seropositivity). The histological hallmark of AIH is interface hepatitis, characterised by mononuclear cells, lymphocytes, plasma cells and macrophages infiltrating the portal tract.7 The primary pathogenetic mechanism of AIH is thought to be a loss of tolerance against the patient’s own liver.8 This predominantly periportal hepatitis is thought to be initiated by CD4+ T cells, which recognise self-antigen. Hepatocytes in patients with AIH are postulated to aberrantly express HLA class II molecules on their surface, a process normally involved in the presentation of antigen to CD4+ T cells and other inflammatory cells. Autoimmune disease results from the resultant inflammatory cell activation, with subsequent hepatocyte destruction (cell-mediated autoimmunity).7,8 AIH may also involve the unmodulated production of autoantibodies directed against the hepatocyte membrane (antibody-dependent autoimmunity). This includes elevated immunoglobulin G levels, seen in up to 80% of patients with AIH.7 A genetic predisposition is also thought to be important, with the HLA DR3 and DR4 alleles being associated with the development of type 1 AIH.8 Infliximab is thought to contribute to development of AIH in predisposed patients by triggering development of autoantibodies. These include ANA and anti-dsDNA.9 One possible explanation for this is that TNF-α blockade interferes with the normal cytotoxic T lymphocyte suppression of auto (self)-reactive B cell production. Another is that anti-TNF-α agents interfere with the induced cell death of CD8 T cells, leading to an accentuated lymphocyte presence.9 Studies in patients with rheumatoid arthritis support infliximab-related autoantibody development,9,10 showing elevations in both ANA and anti-dsDNA titres after infliximab therapy. The occurrence of AIH, and other serious, though uncommon adverse drug reactions, may become more evident with increased use of anti-TNF-α agents. Monitoring of liver function is not routine for patients being treated with anti-TNF-α agents (or those receiving other disease modifying agents, such as methotrexate and leflunomide). Routine blood tests, including LFTs, should be considered for all patients receiving anti-TNF-α agents. 1 Results of blood tests at cessation of infliximab therapy Parameter Patient Reference range Electrolytes, renal function Normal Normal Bilirubin 97 pmol/L 2–21 pmol/L Albumin 25 g/L 35–53 g/L Alkaline phosphatase 521 U/L 30–115 U/L γ-Glutamyl transferase 614 U/L 5–30 U/L Alanine aminotransferase 621 U/L 7–33 U/L Aspartate aminotransferase 821 U/L 10–45 U/L Prothrombin time 13 s 11–17 s Anti-nuclear antibody titre 1/640 < 1/40 Anti-nuclear antibody pattern Homogeneous No pattern or speckled Anti-dsDNA titre 5.7 IU/mL 0–5 IU/mL Anti-smooth muscle antibody titre 1/2560 Negative Anti-mitochondrial antibodies Negative Negative Anti-liver/kidney antibodies Negative Negative Viral hepatitis serum tests Negative Negative Iron studies Normal Normal Caeruloplasmin Normal Normal Anti-dsDNA = double-stranded DNA antibody. 2 Thin sections of the core liver biopsy sample A: × 50 view, showing inflammation around portal tracts with architectural distortion. B: × 200 view, showing lymphocytes, plasma cells, neutrophils and occasional eosinophils, piecemeal necrosi and focal apoptotic cells. 3 Comparison of our case with other published cases of postulated infliximab-related autoimmune hepatitis Patient Sex Age (years) Disease Infliximab infusions Methotrexate use ANA titre Anti-dsDNA development Steroid response Reversible condition Dose (mg/kg) No. Concurrent Past Pre- infliximab Post- infliximab A2 F 39 RA 3 6 No No 0 1/640 Yes No No† B2 F 54 RA 3 12 Yes Yes 0 1/160 nd Yes Yes C5 F 36 RA 3 3 No Yes 1/20 1/640 Yes Yes Yes D6 F 53 PsA 3 8 Yes Yes 1/80 1/160 Yes Yes Yes E* F 56 AS 5 6 No No 0 1/640 Yes Yes Yes * Our patient. † Requiring liver transplantation. ANA = anti-nuclear antibodies. Anti-dsDNA = double-stranded DNA antibody. AS = ankylosing spondylitis. nd = not done. PsA = psoriatic arthritis. RA = rheumatoid arthritis.

Gerard Ozorio MB BS · Bruce McGarity MB BS, FRACP · Haesung Bak MB BS, FRACP · Andrew S Jordan MB BS, BSci(Med) · Henry Lau MB BS, FRCPA, FACTM · Cathy Marshall MB BS

An 80-year-old woman presented with abdominal pain, nausea and diarrhoea. She had undergone a laparoscopic cholecystectomy 10 years previously. The surgery was complicated by an umbilical hernia at the umbilical port site, which was repaired surgically 5 years later. The patient reported intermittent drainage of clear fluid from the umbilicus after the hernia repair. She had right lower quadrant tenderness without guarding or a palpable mass. Laboratory investigations showed a serum lipase concentration of 393 U/L (reference range, 114–286 U/L) with normal white blood cell count. Abdominal computed tomography revealed a 2.8 × 2.4 cm mass in the right lower quadrant, with a calcific rim and foci of calcium within the lesion (Figure, A). The mass appeared to be outside the small bowel, suggesting a mucocele or mucoid tumour. Laparoscopy revealed a 2.5 cm solid mass containing a gallstone (Figure, B and C). The drainage from the umbilicus continued after surgery and was apparently unrelated to the spilled gallstone. Gallstone spillage is more common during laparoscopic than open cholecystectomy. It occurs in 5%–40% of laparoscopic cholecystectomies, but complications are very rare,1 with a reported incidence of 0.08%–0.3% of cases.2 The reason that only a fraction of patients develop complications is uncertain.1 Figure A: Computed tomography scan of the abdomen showing a 2.8 × 2.4 cm mass (arrow) external to the small bowel. B, C: The gallstone recovered during exploratory laparoscopy.

Edgard Wehbe · Reggie J Voboril · Elisha J Brumfield

To the Editor: I read with great interest the recent updated review of amoebiasis by van Hal and colleagues1 and their previous letter2 describing three cases of locally acquired amoebiasis due to Entamoeba histolytica in Australian men who have sex with men (MSM). These articles should alert clinicians to the emergence of invasive amoebiasis and the possibility of person-to-person transmission of E. histolytica through oral–anal or oral–genital sex among MSM in developed countries. The same phenomenon has been reported in Taiwan3 and Japan.4 The prevalence or incidence of intestinal amoebiasis among people at risk may have been underestimated in the past, as microscopy of stool specimens has lower sensitivity and specificity than E. histolytica antigen detection methods for diagnosing the disease.1,5 Cases of amoebiasis may evade detection using the diagnostic algorithm proposed by van Hal and colleagues,1 which suggests using microscopy of stool specimens to detect E. histolytica complex followed by confirmation with specific antigen detection methods or molecular methods. To increase diagnostic sensitivity and specificity, I suggest revising the diagnostic algorithm for intestinal amoebiasis in developed countries to include more accurate first-line detection methods. For example, specific antigen detection methods or polymerase chain reactions, as proposed by Tanyuksel and Petri,5 could be incorporated.

Chien-Ching Hung

To the Editor: van Hal and colleagues deserve congratulations for their lucid, concise and timely review of the complex problem of human amoebic infection and its diagnosis.1 Not surprisingly, however, their article raises more questions than it answers. To me, the gist of their message was as follows: what was in the past diagnosed as Entamoeba histolytica infection, based on the microscopic identification of organisms from faeces, culture or histological sections, actually may have been caused by other species, viz. E. dispar (a recently described non-pathogen) or E. moshkovskii (known for a long time from sewage samples, but only recently found to infect humans). Because these species are all identical morphologically, they can be distinguished reliably only by sophisticated molecular techniques. To complicate matters further, despite E. dispar having been virtually defined as the “non-invasive form” of Entamoeba, most true E. histolytica infections are still asymptomatic.2 Not only the parasite, but also individual host factors (perhaps including genetics), determine pathogenicity. Thus, not all people infected with the same pathogenic strain will manifest symptoms or signs of invasive disease. Given their biology and evolution, it is conceivable that, eventually, invasive strains of even E. dispar will be discovered! Furthermore — and this seems not to have been investigated yet — mixed infections involving different species and strains of these parasites almost certainly occur (not to mention the “traditional” non-pathogenic amoebae, which frequently do occur in mixed infections). The authors advocate treatment of even asymptomatic E. histolytica infections, but how would these be detected outside epidemiological surveys or healthy population screening programs? Given the difficulty and expense of specifically identifying the infective organism even in symptomatic cases, and the relative cheapness of treatment, surely it would be sufficient simply to treat on the basis of clinical presentation plus the identification of E. histolytica-like parasites, with or without objective evidence of histopathology. Anything more could be justified only within the context of a well funded and carefully designed research program and/or epidemiological study.

Paul Prociv

In reply: We agree with Hung that molecular and antigen testing methods are more sensitive for Entamoeba histolytica detection than microscopy and that reliance on microscopy alone would result in under-detection. Our algorithm1 was presented the way it was for several reasons. Firstly, both molecular and antigen testing are significantly more expensive than microscopy. Secondly, as these tests can currently only detect a single pathogen, they would not replace microscopy. Most patients, especially men who have sex with men (MSM), have multiple intestinal parasites, so the more specific methods would remain an adjunct in parasite detection.2 Thirdly, the positive predictive value of any test is dependent on the prevalence of the disease. The prevalence of E. histolytica in Australia, based on current data, is less than 1% in high-risk populations, including MSM. Thus, at present, molecular and antigen tests would be more likely to give false positive than true positive results. However, we agree that our algorithm could be modified as suggested if prevalence rates were between 5% and 10%. Finally, as seen in the MSM population in Taiwan, this is not a static situation, and ongoing local surveillance is required.3 We agree with Prociv that, before the introduction of molecular techniques, E. histolytica prevalence would have been overestimated. We also agree that specific host factors and/or undefined parasitic virulence factors can lead to invasive disease. However, given the extensive molecular work that has been undertaken, we believe it unlikely that invasive strains of E. dispar will be discovered.4 Furthermore, recent studies show that mixed infections are common.2,5 In symptomatic patients, empirical amoebicidal therapy is warranted. However, to ensure that alternative diagnoses (eg, inflammatory bowel disease) that require different treatment are not overlooked, all attempts to accurately speciate Entamoeba complex should be undertaken. We acknowledge that speciation using the polymerase chain reaction is beyond the means of most laboratories, but this is not the case for enzyme immunoassay testing of stool samples, which is rapid, sensitive and relatively cheap. For asymptomatic patients who are carriers of E. histolytica cysts, the World Health Organization recommends treatment.5 However, in areas of low prevalence such as Australia, Entamoeba cysts are more likely to be non-pathogenic E. dispar or E. moshkovskii species than E. histolytica.3 Thus, in Australia, treatment would be unnecessary in a high proportion of patients. Furthermore, therapy requires a luminal agent (paramomycin), which is difficult to obtain. The most practical solution is to either give no treatment or to treat only those patients who have tested positive for E. histolytica.

Sebastiaan J van Hal · Damien J Stark · Debbie Marriott · Jock L Harkness

Objective: To evaluate the outcomes 10 years after a flexible sigmoidoscopy colorectal cancer (CRC) screening program in asymptomatic average-risk individuals.Design, setting and patients: In 1995, a program of flexible sigmoidoscopy-based screening of asymptomatic average-risk individuals aged 55–64 years was established at Fremantle Hospital, Western Australia. Insertion depths, pathological findings and subject-rated pain scores have been prospectively recorded. A follow-up flexible sigmoidoscopy examination was offered to attendees 5 years after the initial screening. Post-screening malignancies were determined by linkage with the Western Australian Cancer Registry in September 2006.Main outcome measures: Yield of neoplasia at initial and follow-up sigmoidoscopy, and the incidence of CRC detected after screening.Results: Between 1995 and 2005, 3402 people underwent an initial flexible sigmoidoscopy screening examination (mean age, 60 years; women, 41%) and 1025 had a 5-year recall examination. Mean insertion depth was greater in men than women (60 cm v 52 cm, P < 0.001). The insertion depth in women was more likely to be < 40 cm (17% v 6%, P < 0.001). Mean pain score was 2.9 for men and 4.0 for women (P < 0.001). Fourteen per cent of initial screenings detected at least one adenoma. Over a mean follow-up time of 8 years, invasive CRC was detected by flexible sigmoidoscopy screening in 0.4% of participants; 0.7% of those with a normal result of screening later developed CRC, with 75% of these found proximal to the splenic flexure.Conclusions: Flexible sigmoidoscopy is a viable screening method, with well defined utility and limitations, for CRC screening of asymptomatic people with average risk.

Charlie H Viiala MB BS, FRACP · John K Olynyk MB BS, FRACP, MD

To the Editor: It was interesting to read of the trial conducted by Manley et al1 using yoghurt containing Lactobacillus rhamnosus to clear vancomycin-resistant enterococci. I would like to add an historical note. The use of yoghurt in restoring bowel flora was practised by Dr J H Kellogg (of Corn Flakes fame) around the end of the 19th century. Kellogg was the chief physician at the Battle Creek Sanitarium in Michigan and was an advocate of high colonic irrigation, for he believed the colon was a sewer of toxic materials that were the causal factor in many diseases. Following this procedure, the patient was given a pint of yoghurt — half to be taken orally, and the remainder given by enema.2,3 By these measures, Kellogg claimed to have cured many conditions, from cancer of the stomach to psychiatric problems.

H Reginald Magee

To the Editor: The recent editorial by Macrae1 outlined the potential issues facing the rollout of the National Bowel Cancer Screening Program. It is estimated there will be about 5000 additional colonoscopies performed in New South Wales in the first year.2 To ascertain how the NSW public health system might absorb this increased demand, the Greater Metropolitan Clinical Taskforce (GMCT) Gastroenterology Network conducted a survey in 2006, featuring structured interviews with clinicians. Our aims were to estimate the current capacity of NSW public hospitals to perform colonoscopies and to identify perceived impediments to meeting future demand. From a total of 113 public hospitals where colonoscopies had been performed in the period 2001–2004,3 we selected a purposive sample of 32 where there had been at least 500 procedures, or which were regarded as major providers in their area health service. Responses were received from 26 hospitals, which represented 54% of colonoscopies performed in all public hospitals in NSW in 2001–20043 and included both metropolitan and rural hospitals. No information was collected from the private sector. Fifteen hospitals had dedicated endoscopy suites, 10 used operating theatres, and one unit used a day surgery centre. We found that for the majority of these hospitals (23/26), colonoscopy activity is currently at or near maximal capacity, with limited potential for expansion of services. The additional number of colonoscopies that could be accommodated ranged from one to four per week in six hospitals, up to a maximum of eight per week in a single hospital. Reasons for unbooked hours included insufficient funds, and a lack of nurses and anaesthetists. The Box summarises other key findings, which emphasise that the three major factors limiting activity are insufficient endoscopy nurses, insufficient nursing applicants, and the need for more equipment. Importantly, an absolute shortage of proceduralists was not found to be a restricting factor. Rather, insufficient available colonoscopy time for existing proceduralists was identified as a limitation. While our survey was successful in clarifying factors that impede optimal performance, the small number of participating hospitals is a relative limitation. Nevertheless, the survey suggests a requirement for additional nursing staff and equipment, and for the establishment of uniform data collection and reporting systems. Areas requiring further exploration include greater unit efficiency, opportunities for workplace redesign, and consideration of inequities in access to anaesthetic cover for patients in public hospitals. The GMCT Gastroenterology Network is currently working on these issues in collaboration with the NSW Department of Health. Factors affecting colonoscopy capacity in selected New South Wales public hospitals* Most responding units (22/26) regarded additional nursing staff as a medium to high priority requirement Increasing time available to existing proceduralists (18/25) and allocating time for new proceduralists (16/25) were regarded as medium to high priorities 11/15 hospitals with dedicated endoscopy suites had an unused endoscopy room. Most common reasons were: insufficient funds (5); lack of nurses (5); insufficient equipment, including anaesthetic machines (3); and lack of anaesthetists (2) No unit cited a shortage of proceduralists as a reason for the unit not running at full capacity “For” (11/24) and “against” (13/24) responses for the addition of procedure rooms were evenly spread across responding facilities Prioritisation of colonoscopies over other procedures was not favoured (14/24 ranked this of low importance, 5 as neither low nor high, 5 as medium to high) Most important factors impeding capacity were: Lack of approval to recruit nursing staff (18/23) Shortage of nursing staff applicants (16/25) Budgetary limitations (19/25) Insufficient endoscopy time for existing proceduralists (17/25) Insufficient budget to recruit new proceduralists (16/23) Only 2/26 units (both using operating theatres) had a data manager recording data and compiling statistics; others (13/26) in dedicated endoscopy suites used commercial reporting systems, such as Endoscribe, however the survey did not determine how systematically and completely data were compiled and reported from these systems * Some respondents did not answer all questions.

Shelanah A Fernando · Anne E Duggan · Owen F Dent · Maeve C Eikli

To the Editor: I note with interest the results of the study by Morris and colleagues comparing survival outcomes for patients with colorectal cancer treated in public and private hospitals.1 Stage of disease is a major determinant of survival in colorectal cancer, yet, as stated in their article, stage of disease was derived solely from pathology reports. I believe that this method would have very limited accuracy in diagnosing stage IV disease. Determining stage IV disease requires additional investigations such as computed tomography, which may either not have been performed before surgery, or the results of which may not have been noted on the pathology request form. Underdiagnosis of stage IV disease appears likely, as the incidence of stage IV disease of about 10% noted in this study is substantially less than the 20%–25% incidence observed in other epidemiological studies.2 Moreover, the limitations associated with the use of pathology reports to determine stage IV disease have been observed by other investigators.3 Although this study evaluates an interesting question in relation to the management of colorectal cancer, the significant possibility of an imbalance of important prognostic factors (such as tumour stage) between public and private patients creates major doubt about its conclusions.

Niall C Tebbutt

To the Editor: We suggest that marked differences in comorbidities could be one explanation for the superior outcomes for patients with colorectal cancer treated in private hospitals as reported in the study by Morris et al.1 We also have concerns about the quality of the data used in their study. We used a comprehensive prospective database to examine a cohort of Victorian patients treated at the Royal Melbourne Hospital (n = 260) and the adjacent Melbourne Private Hospital (n = 118) between 2003 and 2006. Specifically, we included data on important patient variables that were not considered in the Western Australian series (Box). These data showed a clear bias towards improved postoperative and long-term survival outcomes for patients in the private system, independent of cancer treatment. Significantly, diabetes, which affected a much larger proportion of public than private hospital patients, is also associated with inferior cancer-specific outcomes.2 Together, these results could explain the differences reported by Morris et al. In support of this, the 5-year cancer-specific survival rates they report for stage I cancer in the two hospital groups is similar (89% public versus 92% private), but the overall survival rates are markedly different (74% public versus 85% private) — consistent, we contend, with an excess of non-cancer deaths in public patients. Morris et al also report an imbalance in the receipt of adjuvant chemotherapy between public and private patients, possibly a marker of inequality of care. We analysed prospectively collected data for our patients with stage III colon cancer, where adjuvant chemotherapy has a proven impact on survival. As shown in the Box, a similar percentage were offered, accepted and completed adjuvant chemotherapy, suggesting, in Victoria at least, similar access to care and support for public and private patients. The pathology-based staging used by Morris et al may also be inaccurate, as recording of stage IV disease relies on the surgeon noting this on the pathology request form and patients having been fully evaluated before surgery (we contend that preoperative computed tomography scanning would not have been routine). This may explain the relatively low percentage of patients recorded as having stage IV disease (12% public; 9% private) compared with 17% in our combined series in which prospective clinicopathological staging was used. Finally, inaccurate data are suggested by the reported 5-year overall survival figures for patients with stage IV cancer (17% in private care).1 That this is superior to figures reported in recent clinical trials is an unexpected finding, particularly as clinical trials enrol only a select subset of patients and provide access to novel combination chemotherapy, and presentation with metastatic disease is an adverse prognostic factor. Comparison of public and private hospital patients with colorectal cancer Public hospital Private hospital P Number of patients 277 122 Smoker 141 (51%) 16 (13%) < 0.001 Diabetes 57 (21%) 8 (7%) 0.002 Emergency presentation 22 (8%) 2 (2%) 0.02 ASA score* 1 28 (10%) 36 (30%) 2 131 (47%) 32 (26%) 3 89 (32%) 22 (18%) 4 15 (5%) 0 Unknown 14 (5%) 32 (26%) Stage III colon cancer 64 38 Chemotherapy advised 50 (78%) 30 (79%) 0.22 Patient followed advice 44 (88%) 29 (97%) 0.12 Chemotherapy completed 28 (64%) 21 (72%) Chemotherapy not completed 16 (36%) 8 (28%) Toxicity 7 (44%) 3 (38%) Patient request 4 (25%) 0 Ongoing computed tomography 3 (19%) 4 (50%) Other 2 (12%) 1 (12%) * American Society of Anesthesiologists physical status score.

Suzanne Kosmider · Ian T Jones · Ian P Hayes · Peter Gibbs

In reply: The first issue is whether we accurately identified patients with stage IV disease, as the incidence in our study was only 11% and the anticipated incidence is normally 20%–25%. We agree that simply reviewing pathology reports would tend to underdiagnose stage IV disease. However, all patients were crosschecked to the linked database to see if they had had a computed tomography scan or ultrasound image showing metastatic disease. In addition, we only reviewed patients whose primary cancer was resected. This excluded about a third of all the patients with stage IV disease. In my (C P) own surgical prospective colorectal cancer database of 781 patients, (1996 to 2007), 24% of the 185 referred with colorectal cancer had stage IV disease (private hospital patients, 21% v public hospital patients, 25%). Resection of the primary cancer was only undertaken in 126 of the 185 patients with stage IV disease (68%). Therefore, only 16% of all patients with colorectal cancer who had resections had stage IV disease (126/781). Tebbutt argues that there could be a possibility of some imbalance between public and private patients based on this observation. We argue that the difficulty in diagnosing stage IV disease applies equally to public and private patients. Similarly, my own database does not show a significant difference in the incidence of stage IV disease between public and private hospital patients. Kosmider et al report that, at their own institution, completion rates for chemotherapy for stage III colon cancers are the same for public and private patients, “. . . suggesting, in Victoria at least, similar access to care and support for public and private patients”. These results are hardly comparable with our study. Their observations are limited to a select subgroup and only relevant to a small section of the Victorian population, where there is a collocated public and private hospital. In contrast, our study included the entire population of Western Australia, and all public and private hospital patients. Interestingly, we noted wide variations in the rates at which chemotherapy was used, not only between public and private hospitals, but also between individual hospitals. We also noted that there are few collocated public and private hospitals within WA. The fact remains that, at a population level, public hospital patients in WA were less likely to receive chemotherapy. Finally, Kosmider et al raise the issue that comorbidities can influence overall and cancer-specific survival (especially in relation to diabetes). We recognise this as a weakness of our study, and highlighted it in the discussion. Nonetheless, we did include a number of measures of social disadvantage in the analysis. Such measures can act as surrogate markers of common comorbidities. For example, there is a fairly clear association between type 2 diabetes and disadvantage.1 We therefore do not believe that comorbidities can account for all of the observations seen in our study. Our conclusion therefore remains — that patients with colon and rectal cancer treated in private hospitals in WA had superior outcomes. In reality, given the inequities between the two systems, is this really so surprising?

Cameron Platell · Melinda Morris · Barry Iacopetta

Until recently, liver transplantation was contraindicated in Jehovah’s Witness patients because of recipient-imposed restrictions on use of blood products. However, recent improvements in surgical and anaesthetic techniques and new procoagulant agents challenge this practice. We describe two Jehovah’s Witness patients who had successful liver transplantation without blood transfusion. To our knowledge, these are the first such cases in Australasia. The techniques used to minimise blood loss and transfusion requirements could potentially benefit all patients undergoing major surgery. Clinical recordsPatient 1A 48-year-old farmer with end-stage cirrhosis due to α-1 antitrypsin deficiency had evidence of moderate portal hypertension with splenomegaly and ascites (Child–Pugh score B), but no significant lung disease. He met minimum recipient suitability criteria for liver transplantation, according to the Transplantation Society of Australia and New Zealand (TSANZ) liver standing committee.1 As a Jehovah’s Witness, he would not accept transfusion of red blood cells, fresh frozen plasma or platelets. However, he indicated that he would accept blood fractions and recirculated autologous blood and cell-saved blood. A relative, also a Jehovah’s Witness with similar restrictions on use of blood products, offered to be a live liver donor, but this offer was rejected by the treating team on the basis of unacceptable donor risk. The patient was placed on the transplantation waiting list, and was treated with erythropoietin. Over 6 months, this increased the haemoglobin concentration from 112 g/L to 151 g/L (reference range [RR], 135–180 g/L). Results of other preoperative blood tests included: platelet count, 74 × 109/L (RR, 150–400 × 109/L); international normalised ratio (INR), 1.2 (RR, 0.9–1.3); serum concentration of bilirubin, 60 μmol/L (RR, < 20 μmol/L); albumin, 26 g/L (RR, 35–50 g/L); alanine aminotransferase (ALT), 102 U/L (RR, < 40 U/L); and creatinine, 143 μmol/L (60–110 μmol/L). Liver transplantation was performed using an organ from a 54-year-old deceased donor. The piggyback implantation technique, without venovenous bypass, was used. Coagulation was monitored intraoperatively using routine coagulation tests and thromboelastography (Haemoscope, Skopie, Ill, USA). The latter technique measures the kinetics and tensile strength of clot formation. Prophylactic aprotinin was administered as a bolus followed by a constant infusion. The patient also received cryoprecipitate, albumin, haemodilution, and autotransfusion of cell-saved and recirculated blood. On arrival in the intensive care unit, haemoglobin concentration was 118 g/L. The patient received recombinant factor VIIa to treat an INR of 3.4, and erythropoietin was continued. There was significant primary graft dysfunction, and ascites was slow to resolve. Three months after transplantation, the patient developed a pulmonary embolism and required anticoagulation. Currently, at 4 years after transplantation, the patient is well and works full time. Patient 2A 43-year-old woman with chronic hepatitis B infection was found to have an unresectable multifocal hepatocellular carcinoma at laparotomy. She had well compensated cirrhosis (Child–Pugh score A) with no evidence of portal hypertension, and met listing criteria for transplantation. A Jehovah’s Witness, she would not accept transfusion of red blood cells, fresh frozen plasma or platelets, but determined that she would accept blood fractions and recirculated autologous blood and cell-saved blood. Pretransplant laboratory results were: haemoglobin concentration, 129 g/L; platelets, 203 × 109/L; INR, 0.9; bilirubin, 7 μmol/L; albumin, 38 g/L; ALT, 58 U/L; and creatinine, 70 μmol/L. Liver transplantation was performed using an organ from a 42-year-old deceased donor. An inferior vena cava interposition technique was used without venovenous bypass because of the proximity of the tumour to this vessel. The central venous pressure was maintained below 5 cmH2O to minimise blood loss. The patient received cryoprecipitate, haemodilution, autotransfusion, and cell-saved and recirculated blood. Unfortunately, she had an allergic reaction to the colloidal plasma-volume substitute, gelofusine; coagulation studies and thromboelastography showed fibrinolysis, which was treated with aprotinin and recombinant factor VIIa. On arrival in the intensive care unit, haemoglobin concentration was 75 g/L, and INR was 1.3. Erythropoietin and iron supplements were started. Postoperative recovery was uncomplicated, and the patient remains well 3 years after the operation. DiscussionTo our knowledge, these are the first reported cases of liver transplantation in Jehovah’s Witness patients in Australasia. While the two patients filled accepted criteria for recipient suitability for liver transplantation,1 the likely need for blood transfusion would until recently have precluded this procedure. Liver transplantation is a well established and successful intervention for liver failure that results in long-term survival (70% at 10 years) in individuals who otherwise have minimal 1-year survival.2 The shortage of deceased donor livers remains the major factor limiting the number of liver transplantation operations in Australia and New Zealand. In 2000, the death rate while waiting for a donor liver in Australia and New Zealand was 40% for acute liver failure and 5%–8% for chronic liver disease.3 This donor shortfall creates an ethical dilemma in which the potential benefit to individual patients has to be balanced against the need to maximise the benefits of this scarce resource. Following a well publicised case in Edinburgh of a death due to acute liver failure, a recommendation was made for a colloquium to address the question of patient selection for liver transplantation and the need for a uniform code of practice in the United Kingdom. The colloquium, held in 1999, recommended that liver transplantation should be performed in patients when their expected survival is less than 12 months and the expected post-transplant survival is over 50% at 5 years.4 These recommendations have been incorporated into the minimum recipient listing criteria used by the TSANZ liver standing committee,1 and were met by both the reported patients. Over the past decade, improvements in surgical and anaesthetic techniques, combined with new procoagulant agents, have resulted in a dramatic reduction in the requirement for transfusion of blood and blood products during liver transplantation. In selected patients, the need for blood transfusions can be avoided completely.5,6 These advances have resulted in reassessment of the use of liver transplantation in Jehovah’s Witness patients. The first-ever reported liver transplantation in a Jehovah’s Witness patient was in 1994.7 Since then, transplantation has been successfully performed in selected individuals for acute and chronic liver failure without the need for blood products.8,9 Outcomes of liver transplantation in adult Jehovah’s Witness patients have been reported as 92% survival with a mean follow-up of 2.2 years (range, 0.3–5.6 years).10 Live-donor liver transplantation using Jehovah’s Witness donor/recipient pairs has more recently been reported.10 However, the risk to the potential live donor in the case of our first patient through refusing blood products, added to the known 0.5% mortality associated with donation of the right lobe of the liver, was thought to be excessive, and this option was rejected.11 As always, careful selection of the recipient is required. Two other Jehovah’s Witness patients referred to us for liver transplant assessment rejected, or were rejected for, transplantation: one, after lengthy consideration, refused to accept a donor liver; while the other had multiple hepatocellular cancer tumours which fell outside the minimal listing criteria. By way of comparison, in a previously reported series, only nine of 29 Jehovah’s Witness patients were found to be suitable for liver transplantation.9 The Jehovah’s Witness church teaches that blood transfusion (whole blood, red blood cells, white blood cells, platelets and plasma) should not be accepted, but individuals themselves are to decide whether to accept organ transplantation and blood fractions. Both our patients accepted the use of cryoprecipitate, albumin, recombinant factor VIIa, recirculated autologous blood and cell-saved blood, and signed a preoperative agreement to this effect. Consent to the use of these factors and techniques were minimum listing criteria required by the treating teams to proceed with liver transplantation. In selected Jehovah’s Witness patients with hypersplenism (not present in our patients), the use of partial splenic artery embolism increased platelet count, allowing transplantation to proceed.9,10 Transjugular intrahepatic portosystemic shunt formation has been less successful in reversing hypersplenism, and should not be used for this indication.12 Jehovah’s Witness patients with severe decompensated liver disease and coagulopathy (Child–Pugh score C), severe portal hypertension and renal failure are at high risk for perioperative mortality and should not receive liver transplantation. Preoperative use of erythropoietin to increase haemoglobin levels has a number of potential benefits. The most obvious is that the patient begins the procedure with a higher blood haemoglobin level. This also enables use of haemodilution to minimise red cell loss during the explant procedure, and autotransfusion to raise the haematocrit after haemostasis is secure. Maintaining a low central venous pressure also decreases blood transfusion requirements during liver transplantation.13 Although recombinant factor VIIa is expensive (average $6000 per patient), it reduces coagulopathy and transfusion requirements.14 Overall, use of these blood conservation techniques may result in a cost benefit, compared with use of large volumes of blood product.7,10 Successful liver transplantation is possible in selected Jehovah’s Witness patients, but early referral before the development of severe, decompensated liver disease is mandatory. Also, we believe that techniques that minimise blood loss and transfusion requirements for liver transplantation should be more widely practised to benefit all those undergoing major surgery.

Gary P Jeffrey FRACP, MD, MRCP · John McCall FRACS · Edward Gane MD, FRACP · Andrew W Mitchell FRACS · Neville M Gibbs MD, FANZCA · Vanessa Beavis FANZCA · Kerry Gunn FANZCA · Stephen Munn FRACS · Anthony K House MS, FRACS

A 45-year-old woman presented with a 6-month history of dysphagia. Pallor and koilonychia were present on examination. Laboratory tests revealed microcytic hypochromic anaemia (haemoglobin, 59 g/L; mean cell volume, 65 fL), with decreased iron stores (on serum iron studies and bone marrow examination). Barium swallow (Figure A) and oesophagoscopy (Figure B) revealed an upper oesophageal web, which was fractured using Savary–Gilliard dilators. A diagnosis of Plummer–Vinson syndrome (oesophageal web, dysphagia and sideropenic anaemia) was made. As further investigations gave negative results (oesophagoduodenoscopy, colonoscopy, duodenal biopsy, measurement of anti-tissue transglutaminase antibodies, and stool examination for occult blood, ova and cysts), we attributed the iron deficiency to inadequate iron intake. This syndrome is associated with upper alimentary tract cancer, and surveillance endoscopy is recommended. A: Barium swallow showing oesophageal web (arrow). B: Oesophageal web seen on upper gastrointestinal endoscopy (arrows).

Sandeep Chauhan MD(Med) · Atul Sachdev MD(Med), DM(Gastro) · Sanjay D’Cruz MD(Med), DNB(Med), DM(Nephrol) · Ram Singh MD(Med) · Sandeep Singla MD(Med)

We describe a previously unreported association between hyoscine hydrobromide and severe oesophagitis, with ulceration visible almost to the submucosa in the distal oesophagus. The condition resolved with cessation of tablets and 2 months’ treatment with a proton-pump inhibitor. Clinical recordA 31-year-old previously well man presented with a 10-day history of low retrosternal pain. The pain was intermittent, sharp and occasionally burning, with no initial response to antacid or proton-pump inhibitor therapy. There was severe odynophagia, sufficient to prevent the ingestion of solid food, although he was able to consume liquids with minimal discomfort. There had been a single episode of vomiting during this time. The patient denied any previous peptic ulcer disease, gastro-oesophageal reflux or dysphagia. He was otherwise well. These symptoms developed during a boating holiday, when he had taken hyoscine tablets (an over-the-counter preparation) for motion sickness. Over a period of 4 days, these had been chewed and swallowed twice daily without food or water. He had not ingested other medications or potentially corrosive agents. There was no other relevant medical history. The patient was a non-smoker and consumed only small amounts of alcohol. Results of a physical examination, electrocardiogram and chest x-ray were all normal. Upper gastrointestinal endoscopy revealed circumferential ulceration, with clearly visible submucosa of the distal 8 cm of the oesophagus (Box A). The stomach was normal. There was patchy erythema in the first part of the duodenum, but the second and third parts were normal. Histological examination of oesophageal biopsies demonstrated only granulation tissue. Duodenal biopsies were normal. The patient was commenced on 40 mg daily of esomeprazole (a proton-pump inhibitor). Over the following week, he experienced a gradual improvement in his odynophagia and was able to return to a normal diet. Esomeprazole treatment was continued for 8 weeks and he remained asymptomatic. A repeat upper gastrointestinal endoscopy a further 2 weeks later showed complete healing of the oesophagus (Box B). DiscussionA search of the Adverse Drug Reactions Advisory Committee online database and MEDLINE database (1966 to present) uncovered no previous reports of ulcerative oesophagitis secondary to consumption of hyoscine hydrobromide. An objective causality assessment was made in accordance with the Naranjo algorithm.1 This algorithm involves summing the numeric scores for 10 standard questions to provide an overall probability index, with possible outcomes of “highly probable”, “probable”, “possible” or “doubtful”. Our score of 6 indicated that the adverse drug reaction association between hyoscine and ulcerative oesophagitis was probable. Medication-induced oesophagitis may result from either local or systemic effects of a drug. Direct mucosal injury occurs after prolonged exposure of the oesophageal mucosa to the caustic contents of the medication.2 Examples of medications known to cause this type of injury include tetracycline, aspirin and non-steroidal anti-inflammatory drugs, as well as alendronate, potassium chloride, and iron compounds. Endoscopy often reveals a discrete ulcer with relatively normal surrounding mucosa.3 Risk factors for retention of pills in the oesophagus include a lack of adequate liquid bolus, lying down after swallowing, and older age.2,4 Less commonly, certain medications cause oesophageal injury via systemic mechanisms such as promotion of gastro-oesophageal reflux or by immune system compromise, which may result in complications such as cytomegalovirus infection. To our knowledge, this case represents the first report of oesophageal mucosal injury following ingestion of hyoscine. Hyoscine, also known as scopolamine, has several derivatives, including hyoscine butylbromide, hyoscine hydrobromide, hyoscine methobromide, and hyoscine methonitrate. In this case, hyoscine hydrobromide was taken as prescribed for motion sickness (which is thought to be caused by an imbalance in adrenergic and cholinergic input as a result of vestibular stimulation). Hyoscine acts as an anticholinergic, reducing cholinergic input into the vomiting centre in the brainstem. Reported adverse effects of the drug include cardiac dysrhythmia (usually tachyarrhythmia), hypotension, rashes, xerostomia, constipation, dizziness, restlessness, tremors, fatigue, anisocoria, psychosis and urinary retention. Although we could find no reports of hyoscine hydrobromide having a direct toxic effect on the oesophagus, the related compound hyoscine butylbromide has been shown to significantly increase the number of reflux episodes compared with placebo,5 in contrast with atropine, another anticholinergic, which has been shown to reduce reflux episodes.6 If hyoscine hydrobromide acts similarly to hyoscine butylbromide, it would be reasonable to hypothesise that the oesophagitis in our patient may be secondary to prolonged acid exposure. As the tablets were consumed without water, it is also possible that local mucosal injury may have occurred due to retention of tablets in the lower oesophagus, with subsequent pressure necrosis or caustic injury. Further studies would be required to elucidate the precise mechanism of injury. A: Endoscopy photograph showing severe ulcerative oesophagitis in the distal 8 cm of the oesophagus. Note the readily visible muscle fibres of the submucosa. B: Comparative photograph of the distal oesophagus 2 months later, after treatment with a proton-pump inhibitor.

Stephen Philcox MB BS(Hons), BMedSc · Andrew Keegan MB BS(Hons), PhD, FRACP

Clinical record A 45-year-old woman was referred to the Victorian Liver Transplant Unit with acute liver failure following the use of paracetamol at a therapeutic dose during prolonged fasting. Eight weeks previously, the patient had undergone total abdominal hysterectomy for symptomatic uterine fibroids. Seven weeks after surgery, she was readmitted to hospital with subacute bowel obstruction secondary to adhesions. She had lost 10 kg in weight since the surgery, which was attributed to poor food intake as a result of recurrent abdominal pain and intermittent vomiting. Conservative treatment failed, and the patient subsequently underwent laparoscopic division of adhesions on the fifth day after admission. She had remained nil-by-mouth throughout this period. From the time of admission, and for the following 8 days, the patient was prescribed paracetamol 1 g “QID strict” for her abdominal pain (total cumulative dose, 32 g). Four days after surgery (ninth day after admission), the patient developed haematuria and was noted to be displaying odd behaviour. The following morning, she became increasingly confused and drowsy. She was admitted to the intensive care unit, where her conscious state deteriorated rapidly and she required intubation. She was found to have acute hepatic failure, renal failure, and profound metabolic acidosis (Table). Paracetamol levels were in the toxic range (238 μmol/L; therapeutic range, 66–132 μmol/L). The patient was commenced on an N-acetylcysteine intravenous infusion and transferred to the Victorian Liver Transplant Unit for consideration for urgent liver transplantation. The patient was subsequently listed for liver transplantation, but died before a suitable donor became available. Postmortem examination confirmed that the cause of death was multisystem organ failure secondary to acute liver failure associated with paracetamol toxicity. The liver weighed 0.786 kg (normal weight in women, 1.2–1.4 kg), and the capsule appeared wrinkled. Histological assessment of the liver revealed a pattern of massive panlobular necrosis, with some mild fatty changes associated with preserved hepatocytes adjacent to portal tracts. The sinusoids contained increased numbers of inflammatory cells (Box 1). Blood test results for the patient Test Result Reference range Sodium 148 mmol/L 135–145 mmol/L Potassium 3.3 mmol/L 3.5–5.0 mmol/L Urea 4.3 mmol/L 2.5–7.7 mmol/L Creatinine 271 μmol/L 30–110 μmol/L Albumin 28 g/L 36–48 g/L Bilirubin 57 μmol/L < 18 μmol/L ALP 247 U/L 35–104 U/L ALT 5859 U/L < 55 U/L GGT 141 U/L < 45 U/L Lactate 14.6 mmol/L 0.5–2.2 mmol/L pH(arterial) 7.01 7.35–7.45 INR 8.4 0.9–1.2 Paracetamol 238 μmol/L 66–132 μmol/L Parameters in italic are of prognostic use in paracetamol overdose.1,2 ALP = alkaline phosphatase. ALT = alanine aminotransferase. GGT = γ-glutamyl transferase. INR = international normalised ratio. Paracetamol toxicity is the most common cause of acute liver failure in Australia.3 A recent study of 662 patients from 22 tertiary care centres in the United States found that 48% of patients presenting with severe paracetamol hepatotoxicity had not intended to poison themselves, and took the drug for therapeutic purposes only. These patients tended to be older than patients who intentionally poison themselves.4 Paracetamol hepatotoxicity in those older than 40 years is associated with an increased risk of fulminant hepatic failure, death or liver transplantation.4,5 An analysis of 67 cases of “therapeutic misadventure” with paracetamol found that 60% of patients with accidental paracetamol poisoning had taken no more than 6 g paracetamol per day, and 40% had not exceeded the recommended maximum dose of 4 g per day.6 Risk factors for developing hepatic toxicity under these circumstances include prolonged fasting,7 regular excessive alcohol intake,6 or concurrent use of drugs that induce cytochrome P450 (CYP450), especially CYP450 2E1. Elderly patients with renal and cardiopulmonary insufficiency may also be at increased risk.8 Interestingly, there is little evidence that pre-existing liver disease increases the risk.1 Lessons from practice Paracetamol poisoning at recommended therapeutic doses can occur. Accidental paracetamol poisoning should be suspected in any patient with acute liver failure. Prolonged starvation can increase susceptibility to paracetamol poisoning by depleting hepatic glutathione stores. Reduce the daily dose of paracetamol in people who are fasting, in heavy users of alcohol, or in those taking medications that induce cytochrome P450. Paracetamol poisoning is typically associated with large rises in serum transaminase levels. Patients with hepatic failure due to paracetamol poisoning should be referred early to a liver transplant unit for further management. The link between fasting or CYP450 induction and increased susceptibility to paracetamol toxicity can be readily explained by reviewing the metabolism of this drug. When paracetamol is taken in standard doses in healthy individuals, more than 90% is conjugated to form inactive metabolites, which are then excreted in the urine. A small proportion of the paracetamol is metabolised by the CYP450 system to N-acetyl-p-benzoquinone imine (NAPQI), which, if allowed to accumulate, is toxic to the liver. Normally, NAPQI is conjugated with glutathione and the harmless products excreted in the urine (Box 2). Prolonged starvation can severely deplete the cosubstrates required for paracetamol conjugation and reduce glutathione stores, so that even therapeutic doses of the drug can result in the accumulation of toxic amounts of NAPQI (Box 2). Regular and prolonged treatment with paracetamol may be particularly dangerous in fasting patients, as ongoing metabolism of the drug leads to further consumption of already depleted glutathione stores. Alcohol and other drugs capable of inducing CYP450 predispose patients to paracetamol toxicity by increasing the production of NAPQI. In addition, chronic alcohol misuse leads to depletion of glutathione. Between 1989 and 2004, 42 patients were referred to the Victorian Liver Transplant Unit with paracetamol-induced acute liver failure.9 In 11 of these (26%), the poisoning was clearly accidental. Eight patients recovered spontaneously, one survived after undergoing liver transplantation, and two died — including the patient described in this report. Thus, in Australia, severe life-threatening liver injury from accidental paracetamol poisoning appears to be quite uncommon. Nevertheless, this is a readily preventable syndrome of which both patients and the medical profession should be made more aware. In particular, clinicians should be cautious about prescribing regular doses of paracetamol for pain control in malnourished or fasting patients, and need to appropriately counsel patients who are regular users of the drug. Although many cases of accidental paracetamol poisoning at therapeutic dosage have been described, controversy surrounds the actual incidence and mechanism. Unfortunately, as with our patient, it is impossible to determine whether larger quantities of paracetamol had been ingested either accidentally or intentionally.2,10 1 Photomicrograph of the liver at autopsy Massive panlobular necrosis with relative preservation of portal tracts (4 × objective). Inset: Hepatic necrosis with mild fatty changes (20 × objective). 2 Metabolism of paracetamol A: In a healthy individual, about 95% of paracetamol is conjugated with glucuronide and excreted in the urine. Most of the remainder is conjugated with glutathione. B: In starvation, impaired conjugation and depletion of glutathione result in accumulation of hepatotoxic N-acetyl-p-benzoquinone imine (NAPQI).

John S Lubel MRCP, FRACP · Peter W Angus MD, FRACP · Paul J Gow MD, FRACP

A woman presenting with recurrent purpura fulminans was eventually found to have inflammatory bowel disease. We suggest the inflammatory state resulted in a deficiency of functional protein C. Clinical recordFirst admissionA 28-year-old woman presented to hospital with a 1-week history of diarrhoea (three soft, non-bloody stools daily), 2 days of vomiting, and a 12-hour history of a purpuric truncal rash. Five years previously, she had experienced a 1-month diarrhoeal illness, managed as an outpatient. At that time, the full blood count and erythrocyte sedimentation rate were within the normal range. Faecal microscopy had shown scanty leukocytes. Colonoscopy was scheduled; however, the symptoms disappeared and investigation did not proceed. The patient subsequently suffered from mild, intermittent diarrhoea. Her uncle had a history of inflammatory bowel disease, although the patient was not aware of this when she first presented to us. On presentation, her vital signs were normal. There was mild right upper quadrant abdominal tenderness, and multiple large, purpuric lesions over the trunk. Blood tests suggested a systemic inflammatory response and mild disseminated intravascular coagulation (Box 1). Faecal microscopy was positive for leukocytes and erythrocytes. The differential diagnosis was broad, and included systemic infection (such as meningococcaemia), Henoch Schönlein purpura, the antiphospholipid syndrome, thrombotic thrombocytopenic purpura, and purpura fulminans associated with protein C deficiency. Ceftriaxone and enoxaparin were commenced as empirical treatment for infection and/or purpura fulminans. Over the next day, the skin lesions became larger and confluent, with new lesions on the legs and perineum. A morphine infusion was required for analgesia. Four units of fresh frozen plasma were infused for suspected protein C deficiency, while awaiting laboratory test results. These confirmed a slightly reduced functional protein C level (63%). This was considered most likely an acquired deficiency, as seen in inflammatory states. Further fresh frozen plasma or protein C replacement was regarded as unnecessary. Additional tests for prothrombotic (Box 1) and autoimmune diseases (antinuclear, extractable nuclear antigen, and double-stranded DNA antibodies) revealed no abnormality. Thrombocytopenic purpura was excluded by the finding of a normal haptoglobin level. There was no significant bacterial growth from blood, urine or faeces. Ceftriaxone was ceased on Day 3. Purpura fulminans was confirmed on skin biopsy by the finding of dermal venous and capillary thrombosis, without evidence of vasculitis (Box 2). There was no further progression in the number or size of the skin lesions. However, on Day 5, the lesions formed bullae and ulcerated (Box 3). Metronidazole and dicloxacillin were added to treat secondary infection of the ulcers. The morphine infusion was ceased on Day 14, and oral oxycodone was substituted. The results of blood tests taken on Day 22 (Box 1) reflected an ongoing intense systemic inflammatory response. At the time of discharge on Day 23, the patient continued to pass three soft stools daily and had multiple ulcers over the trunk and perineum, which required daily nursing care for wound management. The discharge diagnosis was purpura fulminans, possibly precipitated by an enteric infection. After dischargeBy Day 36 of the illness, the diarrhoea had resolved. However, the patient remained fatigued, and the blood tests (Box 1) indicated persistent systemic inflammation and rising serum liver enzyme concentrations. Abdominal ultrasound was normal. The liver abnormalities were considered secondary to the oxycodone. Over the ensuing weeks, the analgesia was reduced and stopped, the blood parameters normalised, and the patient returned to her normal level of function. The skin lesions had healed by Week 15, but left significant scarring. Twenty-five weeks following the onset of the illness, the patient re-presented as an outpatient with a 1-week history of diarrhoea (six watery stools daily). There was one mouth ulcer, but no new skin lesions. A faecal specimen showed only erythrocytes. The erythrocyte sedimentation rate, C-reactive protein, and platelets were slightly elevated (Box 1). The symptoms and abnormal blood parameters resolved completely over the following week. Referral was made for colonoscopy to investigate for inflammatory bowel disease. Second admissionThe patient presented again, 46 weeks after her original hospitalisation, with a 1-week history of bloody diarrhoea and colicky abdominal pain. The colonoscopy was due the following week. Her vital signs were normal. She had left lower quadrant tenderness. Laboratory tests indicated inflammation and mild coagulation abnormalities (Box 1). She was admitted to hospital. Over the next 24 hours, a purpuric rash consistent with purpura fulminans developed on her abdomen, sacrum and vulva (Box 4). Colonoscopy revealed widespread inflammation involving the entire colon, with multiple deep ulcers, but relative rectal sparing. The histology showed active chronic pancolitis, most suggestive of ulcerative colitis (Box 5). The consensus diagnosis was inflammatory bowel disease (colitis, unspecified). Methylprednisolone 60 mg intravenously was administered daily for 1 week. The skin lesions regressed, without ulceration, and the diarrhoea ceased. She was discharged on a reducing dose of prednisolone. Five months after ceasing prednisolone, she has remained asymptomatic. DiscussionPurpura fulminans is an uncommon condition characterised by rapidly progressive dermal vascular thrombosis leading to haemorrhagic necrosis of the skin, which may extend to the muscle and bone, resulting in severe scarring.1,2 It has been described in three settings: (1) hereditary or acquired abnormalities in the protein C anticoagulant pathway; (2) infections, especially meningococcaemia; and (3) idiopathic.3 Purpura fulminans is associated with high mortality and morbidity rates, reflecting the seriousness of the underlying disease, such as bacterial sepsis, and the complications of tissue necrosis, such as secondary infection and limb loss.4,5 Therapy should be primarily aimed at treating the underlying cause and providing supportive care. Surgical intervention such as fasciotomy, skin grafting and amputation may be required.2,4,6 However, because of its rarity, there are few data to support specific treatments for purpura fulminans. Fresh frozen plasma is recommended for idiopathic purpura fulminans or protein C or S deficiencies.1,2,7 Idiopathic purpura fulminans has also been reported to respond to heparin.2 Protein C replacement has been most widely used for purpura fulminans as a result of severe sepsis or congenital protein C deficiency.2,5,7,8 The use of corticosteroids in treating purpura fulminans remains controversial.1,5 There has been one other report of purpura fulminans associated with ulcerative colitis, involving a 60-year-old man who was deficient in protein S, protein C and antithrombin III.9 There was no improvement with heparin or methylprednisolone. Fresh frozen plasma was administered and he underwent total colectomy. No further skin lesions developed, but he died as a consequence of nosocomial infections. In retrospect, our patient had symptoms of inflammatory bowel disease for several years. Purpura fulminans, a rare manifestation of inflammatory bowel disease, had dominated the initial clinical presentation, distracting clinicians from the underlying condition. We believe that a possible pathophysiological explanation for this case is that the inflammation associated with the colitis led to a deficiency in protein C, thus precipitating purpura fulminans. Methylprednisolone appeared most effective in halting the progression of purpura fulminans, although this may have been secondary to treatment of the colitis, rather than of the purpura fulminans per se. What remains to be seen is whether future exacerbations of her colitis will be associated with recurrent purpura fulminans. 1 Results of laboratory tests Test, unit (reference range) First admission Recovery period Recurrent symptoms Second admission Day 1 Day 2 Day 22 Day 36 Week 15 Week 25 Week 46 Haemoglobin, g/L (115–160) 129 123 84 93 126 122 126 White cell count, 109/L (4–11) 5.4 4.8 7.9 8.5 8.0 9.1 11.6 Platelet count, 109/L (150–400) 150 56 888 948 395 510 414 ESR, mm/h (1–20) — 16 106 112 14 44 36 Total bilirubin, μmol/L (2–20) 15 10 4 7 2 6 12 ALT, U/L (< 55) 52 29 7 67 16 31 51 ALP, U/L (20–110) 204 209 157 1039 99 108 129 GGT, U/L (12–43) 34 48 48 293 56 33 50 Albumin, g/L (33–50) 34 31 24 30 42 39 42 C-reactive protein, mg/L (0–10) — 197 97 31 <5 25 239 Haptoglobin, g/L (0.6–2.7) — 2.1 — — — — 2.9 Prothrombin time, s (9–15) 17 18 15 — — — 17 APTT, s (23–34) 31 33 40 — — — 32 Fibrinogen, g/L (1.5–4.0) 4.6 2.9 4.4 — — — 5.9 XDP, mg/L (0–0.19) 3.24 4.30 0.22 — — — 0.75 Functional protein C (70%–130%) — 63% — — — — 65% Free protein S Ag (50%–130%) — 106% — — — — 116% Antithrombin III (70%–145%) — 95% — — — — 63% Lupus anticoagulant, factor V Leiden, and prothrombin mutations — Not detected — — — — — ACA IgG (< 8) — 2 — — — — — Bold indicates abnormal results. — = not measured. ACA = anti-cardiolipin antibody. ALP = serum alkaline phosphatase level. ALT = serum alanine transaminase level. APTT = activated partial thromboplastin time. ESR = erythrocyte sedimentation rate. GGT = serum γ-glutamyl transferase level. XDP = cross-linked fibrin degradation products. 2 Skin lesions: histopathology A: Thrombosis of venules and capillaries is visible in both superficial and deep dermis (arrows) with intact overlying epidermis (haematoxylin and eosin [H&E]; ×50). B: Higher power view of thrombosed venule showing some apoptotic debris (arrows) but no inflammation (H&E; ×200). 3 Skin lesions: first hospital admission Three bullous lesions (arrows) involving the abdomen, with mild surrounding erythema and diffuse purpura. The largest lesion involves the umbilicus. 4 Skin lesions: second hospital admission Within 24 hours of presentation, multiple purpuric lesions developed over the patient’s abdomen and perineum. Two of these early lesions (black arrows) can be seen adjacent to the large scar involving the umbilicus and three smaller scars (white arrows) from the previous episode. 5 Colonic mucosa: histopathology A: There was some crypt shortening and architectural distortion, and the lamina propria contained a brisk mixed inflammatory infiltrate, which extended into the submucosa (haematoxylin and eosin [H&E]; ×100). B: Eosinophils and neutrophils were prominent in the lamina propria, and there was patchy cryptitis (arrow) (H&E; ×400).

Karina J Kennedy MB BS · Sarah Walker MB BS, BSc(Hons) · Paul Pavli MB BS(Hons), PhD, FRACP · Lavinia Hallam BSc(Hons), FRCPath, FRCPA · Chris Hemmings BHB, MB ChB, FRCPA

There are many ways of freeing up and making maximal use of existing colonoscopy resources The Australian health care system is a complex mix of service providers, variably responsive to the demands placed upon them, and sometimes with limited flexibility, given resource and bureaucratic constraints. So, since the Australian Government Department of Health and Ageing introduced national screening for bowel cancer, based on incontestable evidence of reduction of mortality from colorectal cancer, there have inevitably been some uncertainties about whether the health care system can cope. In particular, there is concern about the follow-up and management of participants with positive faecal occult blood test (FOBT) results. Follow-up and management will be delivered through “usual care”, without any program-specific funding. How will this affect the health care system? The crux of the problem is that the Australian Government’s program — initially, the offer of an FOBT to Australians turning 55 or 65 years of age between May 2006 and June 2008 — will generate a need for extra colonoscopies. In 2005, there were 423 510 Australians who turned 55 or 65 years of age.1 With a 45% participation rate (the participation rate in the pilot program), and a 6% FOBT positivity rate, 11 484 colonoscopies will be needed over the first year. In my own public hospital catchment, where about half of the population has private health insurance, we estimate that this will mean an extra three to four colonoscopies per week for the initial roll-out, rising to 50 when, in 2016, FOBTs will be offered to all people aged 55–74 years. This estimate includes follow-up colonoscopies for those with adenomas detected during screening. These figures equate to a need for one extra dedicated endoscopy theatre. Shifting the (outpatient) load to the federal government through service arrangements between state and federal governments will protect the state purse from some of the burden, but 30%–50% of the load will still fall on the public sector. Estimates of this proportion vary, depending on state or federal perspectives, and will likely change as such arrangements unfold across the states. Some states, particularly Queensland, Western Australia and New South Wales, are already well advanced in addressing the impending load through, for example, commitment of state resources for non-insured patients to service the non-medical facility costs of delivering services by private providers. Other states have proved less responsive and willing to commit resources in this or other ways. But what can be done at the local level, and why should state and federal health departments shoulder the entire burden? The article by Viiala et al in this issue of the Journal (→ Waiting times for colonoscopy and colorectal cancer diagnosis) and other Australian reports shed light on the problem and provide some guidelines.2-4 Direct management of waiting lists will free up colonoscopy resources. In particular, there is evidence that National Health and Medical Research Council guidelines are not being rigorously applied and colonoscopies are being ordered too frequently, both in the private and public systems, for surveillance of patients after adenoma and cancer detection and for screening of those with a family history of colon cancer.3-5 The report by Viiala et al2 is a timely reminder of the extra capacity, after careful management of waiting lists, that can be called on to meet the needs of the National Bowel Cancer Screening Program. Based on data from randomised controlled trials, a repeat colonoscopy after detection of an adenoma is rarely needed in less than 3 years, and most patients can be offered 5-yearly surveillance.6 Adequate staffing is critical. At present, most states are identifying any excess capacity in their endoscopy services, and some are encouraging endoscopy nurses back into the workforce, with appropriate consideration of their needs. In Australia, training of endoscopists is well developed and monitored, and now accredited by the Conjoint Committee (of the Royal Australasian College of Physicians, the Royal Australasian College of Surgeons and the Gastroenterological Society of Australia) for Recognition of Training in Gastrointestinal Endoscopy. However, there is room for increased efforts to monitor quality and competence, and to improve training standards. This has been strikingly effective in the United Kingdom.7 Recognition of the changing demography of disease, and a matching of resources to these changes, needs informed high-level hospital management decisions. Colorectal cancer continues to increase in incidence, while the prevalence of other common diseases is decreasing. As has already occurred in some settings, it should be possible to spread the load across both private and public facilities, with appropriate jurisdictional contractual agreements between federal and state governments, and optimal use of existing facilities and personnel. Any opportunities for funding extra services should be quickly identified by hospital administrators to enhance service capacity. Categorisation of waiting lists according to the likelihood of a diagnosis of cancer or advanced adenoma with respect to the indication for the colonoscopy is an important principle. The government’s choice of immunochemical testing for faecal occult blood, with a 6%–8% overall rate of positive test results, offers, on the evidence of receiver operating characteristics, the best balance between sensitivity and specificity. The National Bowel Cancer Screening Pilot Program, using the same test, identified cancer in 5.3% of patients with positive FOBT results and an additional 13.9% with advanced adenomas. In the report of Viiala and colleagues, two of 14 patients with a positive FOBT result had colorectal cancer, a similar proportion to those presenting with iron deficiency/anaemia.2 As Viiala et al point out, these figures must lead to a high prioritisation of patients with positive FOBT results, approaching or matching other categories with high probabilities, such as radiological or clinical findings indicative of colorectal cancer. Curiously, there appears to be, if anything, an inverse relationship between duration of symptoms, or time on a waiting list, and cancer stage at diagnosis.8,9 Indeed, in the study by Viiala et al, there was no evidence of a link between prolonged colonoscopy waiting times and stage of carcinoma at diagnosis.2 However, this statistic should not be misrepresented — it can be explained by duration bias. Tumours detected by screening are more likely to be indolent, or have long dwell times, making them, at any time point (eg, of screening), more prevalent and more likely to be at an early stage. Aggressive tumours, with short histories and rapid stage evolution, are not as easily detected by periodic screening, or will have short symptom durations, leading to a statistical distortion in the relationship between symptom duration, or waiting-list duration, and stage at diagnosis. Nevertheless, for the individual tumour, it is axiomatic that detection at an earlier time point must be associated with at least an earlier stage of disease and logically a better outcome for the patient. What other issues may arise in referral for colonoscopy within the National Bowel Cancer Screening Program? While it is obvious that the benefits of early detection through screening can only be enjoyed by participants, it should also be recognised that co-morbidities, and age itself, may limit any increments to survival otherwise implicit in screening.10 The Australian Government’s offer to Australians is irrespective of comorbidities. General practitioners have an important role — in assessing the comorbid status of their patients invited to participate in the national program; and in recommending non-participation if estimates of limited life expectancy or other factors would render follow-up of a positive test result by colonoscopy inappropriate. It may be a difficult call for a GP to advise a patient with three or four comorbidities who returns a positive FOBT result, but advice not to proceed with a colonoscopy may be appropriate. No monitoring, evaluation or planning can be done without data. The National Bowel Cancer Screening Pilot Program told us that transfer of data on the outcomes of colonoscopy to the central register was inadequate. The exception was the pilot site in Mackay, Queensland, which dealt with this issue very successfully through a variety of processes, but, most notably, by providing personnel to monitor the pathway to colonoscopy of screenees with positive results. The way forward is clear. There must be investment in supervision of this vital part of the program — data transfer — at least until electronic and integrated data systems are universally available to do this. Responsibility for this rests with both state and federal governments. The National Bowel Cancer Screening Program offers our best chance of having a significant impact on the outcome of colorectal cancer, our second commonest cancer killer. From ministers to minions, we all have a responsibility to make it work.

Finlay A Macrae MB BS, MD, FRACP

Objective: To evaluate whether prolonged waiting times for colonoscopy in public hospitals could result in delayed diagnosis of colorectal carcinoma.Design, setting and patients: Analysis of all outpatient colonoscopies performed at a Western Australian tertiary teaching hospital, 1 November 2003 – 31 October 2005. Colonoscopy data, corresponding pathological findings, category of urgency at referral for colonoscopy, and waiting time for colonoscopy were obtained. Patients were coded as having cancer if it was diagnosed by colonoscopy or if colonoscopy identified a lesion subsequently diagnosed as cancer.Main outcome measures: Colorectal carcinoma detected by outpatient colonoscopy and length of waiting time to colonoscopy.Results: 1632 outpatient colonoscopies were recorded. Category I patients received a colonoscopy within the recommended 30 days from referral. Median waiting times for Category II and Category III patients exceeded recommendations (observed, 113 days and 258 days; recommended, within 90 days and 180 days, respectively), although the number of cancers detected was low (2.4% and 0.6% of referrals, respectively in each category). Early- and late-stage cancers had similar median waiting times from referral to diagnosis. Age over 65 years and the blood-loss indications — a positive faecal occult blood test or iron deficiency/anaemia — were predictors of an increased risk of carcinoma at colonoscopy.Conclusions: Waiting time for colonoscopy was not associated with an increase in the proportion of late-stage cancers diagnosed. Age over 65 years and evidence of blood loss increased the likelihood of a cancer diagnosis.

Charlie H Viiala MB BS, FRACP · Kevin W Tang MB BS · Ian C Lawrance MB BS, FRACP, PhD · Kevin Murray BSc, MSc · John K Olynyk BMedSc, MB BS, FRACP, MD

Objective: To describe the process of colorectal cancer diagnosis in Queensland, and to determine factors associated with time to diagnosis.Design, setting and participants: Cross-sectional study of 1996 patients with colorectal cancer recruited through the Queensland Cancer Registry. Data were collected by computer-assisted telephone interview between May 2003 and August 2005.Main outcome measures: Time to diagnosis: pre-presentation time (time from first noticing a symptom to first presenting to a doctor); and post-presentation time (time between the first presentation and diagnosis).Results: Most patients (90%) had experienced symptoms before being diagnosed with colorectal cancer; only 2% of patients were diagnosed by faecal occult blood testing. Older participants and those who experienced abdominal pain had the shortest time from symptom onset to their first doctor consultation, while participants with a change in bowel habit, or rectal bleeding, and those without private health insurance tended to wait longer to see a doctor. Participants who experienced abdominal pain were diagnosed more quickly, whereas those who experienced a change in bowel habit, women, and those without private health insurance experienced a longer time to diagnosis.Conclusions: The strong association between not having health insurance and longer post-presentation times is concerning. The other hypothesised predictors of time to diagnosis were not as strongly associated as we anticipated.

Brigid M Lynch BSc, MScsComm · Peter Baade PhD · Lin Fritschi MB BS, PhD · Barbara Leggett MB BS, MD · Neville Owen PhD · Kenneth Pakenham PhD, MAP(Clin) · Beth Newman PhD · Joanne F Aitken PhD

Objective: To determine whether treatment in a private versus public hospital was an independent predictor of survival outcomes in patients with colorectal cancer.Design: Retrospective, population-based study.Setting: Tertiary care hospitals.Participants: All patients diagnosed with colorectal cancer in Western Australia between 1993 and 2003.Interventions: Management in private versus public hospitals.Main outcome measures: Overall survival and cancer-specific survival rates.Results: 5809 patients were treated for colorectal cancer. Of these, 1523 (26%) were managed in private hospitals. The 5-year overall survival rates for private and public hospital patients were 59.4% (95% CI, 56.9%–61.9%) and 48.6% (95% CI, 47.0%–50.2%), respectively. Significant independent predictors of overall survival were: treatment in a private hospital (P = 0.0001; relative risk [RR], 0.764; 95% CI, 0.696–0.839); younger age (P = 0.0001; RR, 1.032; 95% CI, 1.029–1.036); male sex (P = 0.001; RR, 1.148; 95% CI, 1.068–1.234); and cancer stage (eg, Stage II: P = 0.0001; RR, 1.508; 95% CI, 1.316–1.729).Conclusions: Treatment in a private hospital was a significant independent predictor of survival outcomes. Further validation of these results would have a significant bearing on how we approach health care delivery for patients with colorectal cancer.

Melinda Morris MB BS · Barry Iacopetta PhD · Cameron Platell MB BS, PhD, FRACS

Testing for Helicobacter pylori, and acid-suppression therapy are nearly always better strategies Most patients with upper gastrointestinal symptoms can be effectively managed without investigation. Recent long-term follow-up of patients with upper gastrointestinal symptoms shows that most have a benign course.1,2 A recent follow-up of 300 patients 9 years after investigations showed that 40% were asymptomatic; 70% of these without medication.2 Such a good outcome is the result of the decline of Helicobacter pylori3 (making peptic ulcer uncommon and gastric cancer rare in the absence of genetic or ethnic predisposition) and the easy availability of effective acid-suppression therapy (making gastro-oesophageal reflux disease easily treatable). For the vast majority of patients, upper gastrointestinal symptoms are now a dis-ease, not a disease. These changes in epidemiology and treatment simplify the management approach to upper gastrointestinal symptoms (Box). Gastroscopy now has a low diagnostic yield. A review of 22 studies investigating dyspepsia found that, overall, findings in 50% of gastroscopies were normal, 12% revealed reflux oesophagitis, 33% gastroduodenal ulceration, and 1.2% malignancy.5 International management guidelines recommend two alternatives to gastroscopy: empiric acid-suppression therapy; or H. pylori testing and treatment.4 Acid-suppression therapy is effective treatment for gastro-oesophageal reflux disease (GORD), and the “omeprazole test” (a simple trial of omeprazole [40 mg twice daily for a week]) diagnoses GORD more accurately than endoscopy, and with a sensitivity of around 80%.6 For population groups with a high prevalence of H. pylori infection, such as the elderly and some ethnic groups, H. pylori testing and treatment has advantages. For younger patients, H. pylori infection is unlikely, as childhood domestic hygiene has improved. If a test for H. pylori is positive, treatment provides: definitive treatment of peptic ulcer disease; no adverse outcome for non-ulcer disease; risk reduction for ulcer disease associated with non-steroidal anti-inflammatory drug (NSAID) treatment; and possible risk reduction for future H. pylori-associated gastric cancer. For those who test negative for H. pylori, the test provides reassurance. Population studies have shown that people who do and those who don’t consult general practitioners for their upper gastrointestinal symptoms have similar symptom severity, suggesting that reasons for consultation may include anxiety about the significance of symptoms.7 Patients should be asked why they present. For those who want information, data show that lifestyle contributes to GORD and adenocarcinoma development, in particular, obesity, smoking, poor fruit and vegetable intake and a sedentary lifestyle.8,9 This information is a useful tool in encouraging patients with GORD to modify their risk and improve their overall health. Overly rapid prescription of acid-suppression therapy may waste opportunities for lifestyle modification. H. pylori testing should be offered to those likely to be infected because of advanced age, ethnic background, or a past or family history of ulcer disease, and to those who take NSAIDs. After advice, and serological or urea breath testing and treatment of patients with a positive result, symptomatic patients can be offered empiric therapy. Whether this should be a step-up approach with initial antacids followed by H2-receptor antagonists, or a step-down approach with proton-pump inhibitors followed by H2-receptor antagonists remains controversial, as does on-demand versus ongoing treatment. Population data show that, after initial consultation, a substantial proportion of patients cease therapy or continue on an intermittent basis.2 There are no data to indicate that for most patients this is harmful. Non-responders often have non-ulcer or functional dyspepsia and treatment is unrewarding. Prescribed and complementary medications, a frequent cause, should be reviewed. Delayed gastric emptying affects up to 40% of non-ulcer dyspepsia patients and is particularly frequent in patients with long-standing diabetes. Previous gastroenteritis is a recently identified cause of non-ulcer dyspepsia. Prokinetic therapy (with drugs that increase the contactility of the smooth muscle of the upper gut, such as motilium) may benefit both groups. Endoscopy should be reserved for those with a familial or ethnic risk of upper gastrointestinal cancer, older patients with alarm symptoms such as dysphagia, haematemesis or weight loss10 and cancer-phobic patients, as its role in patient reassurance is small and not cost-effective compared with other strategies.11 Concern about missing a cancer diagnosis should be tempered by awareness of its low incidence and the limited value of “alarm symptoms”.5,10 In 2001, there were 10 gastric cancers and fewer than six oesophageal cancers per 100 000 population reported in Australia.12 A recent meta-analysis found alarm symptoms had sensitivity of 0–83% and specificity of 40%–98%, and a study of patients with newly diagnosed gastric cancer found poor correlation between symptoms and resectability with: 41% having symptoms; 81% of these having alarm symptoms; 12% of patients with early gastric cancer having prior symptoms; and no correlation between symptom duration and disease stage.10,13 In clinical practice, a third of people with upper gastrointestinal symptoms consult their GP within 6 months of symptom onset.2,6 As oesophageal adenocarcinoma is associated with severe, long-standing upper-gastrointestinal symptoms, “one-off” endoscopy is unlikely to improve early diagnosis.10,13 Today, most patients, particularly those aged under 55 years, can be reassured that: their symptoms are benign; organic disease, if present, is likely to be responsive to H. pylori eradication or acid-suppression therapy; and their lifetime risk of upper gastrointestinal cancer is exceedingly small and may best be reduced by primary prevention with lifestyle modification. If the temptation to refer a patient for endoscopy persists, then the question is, is this to meet an unmet need or is an un-need being met? Algorithm for the management of uninvestigated dyspepsia* * Adapted from Talley.4 OGD = Oesophago-gastro-duodenoscopy. PPI = Proton-pump inhibitor.

Anne E Duggan MHP, FRACP, PhD

Helicobacter pylori is established as a cause of peptic ulcer (PU). Less well recognised is that an epidemic of PU began around the middle of the 19th century, reached a peak at the turn of the century, and is now on the wane. As the epidemic developed, the risk of PU increased in successive generations throughout life. Then the epidemic diminished in successive generations. The risk of gastric ulcer (GU) was highest in people born around 1885, while the risk of duodenal ulcer (DU) was highest in those born about 10–30 years later. H. pylori infection offers an inadequate explanation of the PU epidemic. Although the epidemic coincided with a major rise in cigarette smoking, PU then declined in spite of an increased incidence of smoking. None of the other possible causes of ulcer (non-steroidal anti-inflammatory drugs, stress or diet) satisfactorily explains the epidemics of GU and DU and their asynchronicity. The best, but inadequate, explanation for the epidemic is the coincidence of the acquisition of a new potent strain of H. pylori in childhood and the uptake of smoking in adult life.

John M Duggan MD, FRACP, FRCP · Anne E Duggan MHP, FRACP, PhD

A 26-year-old woman accidentally swallowed a 15 cm long spoon during a “laughing attack”. The spoon was removed from the stomach with a single lumen gastroscope (with great difficulty).

Bernard R Beldholm MB BS · Alice U Lee MB BS, FRACP, PhD

Objective: To ascertain whether current practice in teaching hospitals in New South Wales and the Australian Capital Territory delivers adequate dietary protein in the management of malnutrition in adults with cirrhosis, in accordance with European Society for Clinical Nutrition and Metabolism (ESPEN) guidelines for nutrition in liver disease.Study design: Cross-sectional study of dietitians using a self-administered, mail-back survey.Setting: Teaching hospitals in NSW and the ACT treating patients with cirrhosis.Participants: Dietitians seeing patients with cirrhosis in the 12 months prior to completing the survey.Main outcome measures: Current dietary protein prescription practice for patients with cirrhosis (with and without hepatic encephalopathy); use of nutritional supplements and enteral feeding for malnourished patients with cirrhosis.Results: Dietitians following the ESPEN guidelines were in the minority: 36% of the dietitians recommended an adequate protein intake for patients with hepatic encephalopathy. Sixty-four per cent of the dietitians had received referrals from the medical team requesting inappropriate protein-restricted diets for patients without hepatic encephalopathy. Seventy-eight per cent of the dietitians requested clarification of the recommended nutritional management of patients with cirrhosis.Conclusion: Many medical and dietetic staff inappropriately restrict protein intake of patients with cirrhosis.

Joanne K Heyman BSc(Hons) · Carol J Whitfield RN · Kaye E Brock PhD · Geoffrey W McCaughan MB BS, PhD · Anthony J Donaghy MB BS, PhD

Objective: Exposure to non-steroidal anti-inflammatory drugs (NSAIDs) is associated with increased risk of serious gastrointestinal (GI) events compared with non-exposure. We investigated whether that risk is sustained over time.Data sources: Cochrane Controlled Trials Register (to 2002); MEDLINE, EMBASE, Derwent Drug File and Current Contents (1999–2002); manual searching of reviews (1999–2002).Study selection: From 479 search results reviewed and 221 articles retrieved, seven studies of patients exposed to prescription non-selective NSAIDs for more than 6 months and reporting time-dependent serious GI event rates were selected for quantitative data synthesis. These were stratified into two groups by study design.Data extraction: Incidence of GI events and number of patients at specific time points were extracted.Data synthesis: Meta-regression analyses were performed. Change in risk was evaluated by testing whether the slope of the regression line declined over time. Four randomised controlled trials (RCTs) provided evaluable data from five NSAID arms (aspirin, naproxen, two ibuprofen arms, and diclofenac). When the RCT data were combined, a small significant decline in annualised risk was seen: − 0.005% (95% CI, − 0.008% to − 0.001%) per month. Sensitivity analyses were conducted because there was disparity within the RCT data. The pooled estimate from three cohort studies showed no significant decline in annualised risk over periods up to 2 years: − 0.003% (95% CI, − 0.008% to 0.003%) per month.Conclusions: Small decreases in risk over time were observed; these were of negligible clinical importance. For patients who need long-term (> 6 months) treatment, precautionary measures should be considered to reduce the net probability of serious GI events over the anticipated treatment duration. The effect of intermittent versus regular daily therapy on long-term risk needs further investigation.

Delia Schaffer MMedSci, BSc · Timothy Florin FRACP, MB BS(Hons), MSc · Craig Eagle FRACP, FRCPA, MB BS · Ian Marschner PhD · Gurkirpal Singh MD · Mendel Grobler MBA, BPharm · Chris Fenn PhD, MB BS · Manjula Schou MSc · Kathleen M Curnow PhD