Letters

To the Editor: The decision by Wilkinson and colleagues at the University of Queensland to abandon interview selection methodology represents a regressive step in medical student selection.1 In particular, the problems with accepting past academic performance as an infallible “gold standard” criterion for student selection become evident when considering the less tangible but no less important issues of social equity, “fitness-to-task”, community expectations and corporate responsibility. While prior academic achievement is the best predictor of early medical student examination performance,2 non-cognitive variables appear to become more predictive as training progresses.3 The use of academic achievement as the main or sole criterion diminishes social equity by discriminating against students from under-resourced areas.2 Fitness-to-task is relevant because mistakes in medicine cause serious consequences, as in other occupations such as military personnel, air traffic controllers, and pilots. These “restricted” occupations require specialised training preceded by mandatory selection processes typically involving physical, psychological and skills assessments. Consider also the community’s high expectations for the personal integrity of doctors, and the issue of how medical schools meet community, professional and stakeholder expectations — their corporate social responsibility. Medical schools clearly have a “duty of care” to both students and the community at large in their selection of future doctors. Consequently, it is disappointing that Australia’s largest medical school at the University of Queensland has discontinued interviews, the study authors describing them as “inherently unreliable”.1 Unfortunately, this also reflects widespread imprecision when discussing the different interview methodologies of individual, panel or Multiple Mini-Interview formats. The Multiple Mini-Interview in particular has demonstrated promising reliability and validity.3,4 Stated bluntly, Wilkinson and colleagues’ conclusions are confusing. Despite their results suggesting the GAMSAT (Graduate Australian Medical School Admissions Test) has no predictive validity, it has been retained, while their interview procedure that demonstrated modest increasing predictive validity was dropped! In addition, their use of cognitive outcome measures to assess the predictive validity of non-cognitive variables is conceptually flawed. Finally, an emerging body of evidence is leading to more sophisticated medical student selection methods. Community confidence in neither doctors nor the medical profession itself is endeared or strengthened by the use of selection methods that do not encourage this process. World’s best practice requires the use of evidence-based methods — this is the work ahead, so let’s not throw the baby out with the bathwater.

David W Harding · Ian G Wilson

In reply: Harding and Wilson offer no new data or insights to the challenge of medical student selection. They also say they are confused. We will try to help. The GAMSAT (Graduate Australian Medical School Admissions Test) provided no additional predictive value in our study1 — effectively, it measures what grade point average (GPA) does, which is academic ability. We are continuing to use it because it is highly reliable, whereas GPA, being derived from multiple courses marked by a wide range of individuals in different universities, is less so. Simply put, the GAMSAT offers a highly reliable method of ranking students. Harding and Wilson acknowledge our finding that the interview score adds very little predictive value, but this value increases in the later years of the program. The problem is that, in our judgement, the absolute increase is so small as to be useless. Let us be clear here. We fully agree that good doctors need to be much more than smart. Our point is simply that there is no evidence that these additional characteristics can be selected for. Having dropped the interview, we can now focus even more on the quality of our teaching and professional development programs. Until Harding and Wilson, or others, can demonstrate a selection process that clearly works — in terms of delivering even more effective doctors to society — we will focus more on teaching, training and development. As noted in our article,1 we will carefully monitor and report on the impact of the changes we have instituted to our admissions process.

David Wilkinson · Jianzhen Zhang · Gerard J Byrne · Malcolm H Parker · Raymond F Peterson

To the Editor: The University of Queensland (UQ) study of medical student selection criteria and academic performance reported by Wilkinson and colleagues1 has shortcomings beyond those pointed out by Powis.2 The outcome measure of “academic performance” was assessed by student performance in exams. These exams vary in content from year to year and are, in Years 1 and 2 of the program, mostly multiple-choice and short-answer written exams. A finding that grade point average (GPA) in a previous degree correlates with academic performance in medical school may simply mean that the exams are written and marked in a way that rewards the competencies one acquires in getting a high GPA, and ignores the skills and personal qualities that generate a high interview score. There is thus a danger that the researchers have themselves created the phenomenon they are now discovering — that you can set exams in such a way that people who do well at sitting exams will do well. Those study authors who identify themselves as holding leadership and teaching positions in the UQ School of Medicine do not state whether they held those positions during the years in which they analysed student performance. If they did, then the researchers are also participants in the study by virtue of the fact that they set and marked exams. Even if they did not, their academic positions expose them to at least the risk of partiality, and potentially to the appearance of a conflict of interest — there are logistical and personal pressures on administrators to reduce the time, energy and expense of evaluating students, especially in a school with so many students. It is not clear from the article what, if any, measures were taken to control for these things, nor what, if any, caution this led the authors to exercise in drawing the policy conclusion that interviews should be abandoned. The proper research question is not “How do we select students who will do well on the sort of examinations we set?”, but “How do we select students who will create an atmosphere of excellence in the school, and who will carry that with them into the health system and into society as a whole?” The UQ study does not address the broader questions of the social purpose of schools of medicine in general, or the philosophy and ideals of the UQ School of Medicine in particular. Yet the action UQ has taken — to abandon assessment of the personal qualities of its students — risks a profound impact on the realisation of those ideals.

Nicholas Jefferson-Lenskyj

To the Editor: The article by Wilkinson and colleagues1 and the accompanying editorial by Powis2 remind me of the medieval debate about the number of angels who could dance on the head of a pin. While there is obvious merit in selecting students able to actually pass medical school examinations,3 preferably at their first attempt, where is the evidence that this correlates with their later performance as medical chemists, physicists, researchers, puzzle-solvers, mechanics, artists, analysts or “jacks of all trades”? What is the point of trying to select students on the basis that they would make good “doctors”, when medicine, perhaps the broadest of all churches, offers professional scope to people of almost every imaginable natural bent and talent? Selecting for interpersonal relationship skills is fine if selecting general practitioners and psychiatrists. How relevant is it, however, for someone whose talents and skills are intellectual curiosity or manual dexterity?4 Powis wants us to produce doctors who have “the required skills”. Is this not the role of postgraduate educational bodies, helping the undifferentiated graduate pursue a course relevant to their abilities and interests? If, faced with an almost infinite variety of doctoring, we cannot define “doctor” other than by possession of a medical degree, how can we possibly define the attributes needed to be one? There are many more important issues in health care that could benefit from the time and money being wasted on chasing this particular chimera.

Peter C Arnold

In reply: I am sorry Arnold considers that, in the context of selecting medical students based on their suitability to be a doctor, interpersonal relationship skills are qualities necessary only for general practitioners and psychiatrists. He is presumably unaware that the most frequent complaints made by patients about doctors of all kinds concern the very absence of such skills. I agree with his statement that medicine is a broad church, with many professional pathways to suit individual preferences and skills, but that doesn’t mean that anybody is suitable to fill the positions, or indeed fit to be any sort of doctor.1,2 In any country, medical boards and medical indemnity insurers could give many examples of inadequate practitioners. We should remember that all of these practitioners were admitted to medical school, passed their exams and graduated as fit to practise medicine. If there is any chance of identifying such individuals before they start their medical training, then it would be unethical not to do so.1 This means we have to select students based on more than their academic achievements at school, and a suitably structured interview has been shown to be a reasonably effective tool in this context.3,4

David A Powis

To the Editor: There is a significant shortage of general practitioners in Australia, and many of those in practice are nearing retirement. The government has belatedly realised the need to train more GPs and in the past few years has dramatically increased university places for medical students. Coupled with this increase in numbers is the need for medical students to gain more of their education in general practice, where patients with many of the common medical problems are now managed. This has increased the pressure on universities to place medical students in general practice, and they are having difficulty finding enough willing and capable GPs to train their students. Paying GPs more to teach medical students might help. However, students’ experience of general practice needs to be positive if they are to be persuaded to become GPs themselves in the future, and a positive experience is more likely when the supervising GP is not rushed and can provide students with hands-on practice. Therefore, apart from increasing financial incentives, it is important to reduce “red tape” for training medical students and to free GPs’ time for teaching. A positive step would be to provide an extra payment to the GP on top of the consultation fee for each patient seen by a student under the GP’s supervision, similar to the extra payment for a consultation with a pensioner. This would contrast with the current situation, where a GP who supervises students must fill out forms and wait for a PIP (practice incentive payment) many months later. Another step would be to make accreditation less onerous for GPs who regularly provide clinical experience for students. Adequate continuing medical education (CME) points should be given for providing quality training without the GP needing to “jump through hoops”. If the medical student training crisis is to be resolved, then these issues need urgent attention.

Andrew A Beveridge

To the Editor: The recent call by Robotin and colleagues for a national strategy to respond to the increasing incidence of hepatitis B and hepatocellular carcinoma (HCC) in Australia1 is timely. I would like to add the following comments. First, a comprehensive Australian hepatitis B strategy should include prisoners and Indigenous Australians. Among Australian prisoners, hepatitis B carrier prevalence is 3%–5% — more than three times the national average — and prevalence of hepatitis C, which independently and synergistically increases the risk of severe liver disease, exceeds 30%.2 In addition, of 526 acute hepatitis B notifications in Australia in 2000–2002, 57 were in Indigenous Australians, a notification rate more than four times that in non-Indigenous Australians. Indigenous people are 12 times more likely to die of liver cancer than the general Australian population.3 Second, in New South Wales, the median age of diagnosis of HCC was found to vary significantly by country of birth;4 it was 5 years younger in the Asian-born group than the Australian-born group overall (64 v 69 years), and 9 years younger in those who were hepatitis B carriers (57 v 66 years) (P < 0.001 for both differences). Early onset of HCC among Asian-born Australians may be a result of hepatitis B infection in the perinatal and early childhood period. However, other factors that promote progression to HCC, such as diabetes, alcoholism, and inadequate health care access, are amenable to targeted public health interventions. Third, hepatitis B e antigen (HbeAg) positivity is strongly associated with high hepatitis B virus DNA counts (≥ 100 000 copies/mL), which are in turn highly predictive of cirrhosis and HCC risk. It is thus counterintuitive that — as implied by Robotin et al — hepatitis B carriers who are positive for HbeAg are less likely to progress to cirrhosis and HCC than those who have undergone seroconversion and are positive for hepatitis B e antibody. In fact, HBeAg positivity is associated with increased risk of HCC and liver-related mortality.5,6 Finally, the omission of hepatitis B vaccine — the world’s first anticancer vaccine — from Robotin et al’s list of elements of a public health response to hepatitis B and liver cancer is unfortunate. Hepatitis B vaccination is essential to any credible medium- and long-term strategy to prevent hepatitis B infection and, by extension, HCC, both in Australia and globally.

Niyi Awofeso

In reply: We agree with Awofeso that prisoners and Indigenous people have an increased risk of developing chronic hepatitis B. However, as no large-scale population-based studies of hepatitis B prevalence have been published in Australia, estimates of the risk vary widely.1 A national hepatitis B strategy may provide additional impetus for obtaining high-quality data. We also concur that modifiable behavioural factors may play a role in the age of hepatocellular carcinoma diagnosis, but differences in clinical course between Asian and white Australians,2 and the specific viral genotypes prevalent in Asia,3 are likely to be more important. Although white populations who undergo hepatitis B e antigen (HbeAg) seroconversion and develop hepatitis B e antibodies have a good prognosis, this is not so for Asian populations,2 or for other populations who are mostly infected in childhood, such as Indigenous Australians and Māori in New Zealand. The median age of HBeAg seroconversion in Asian patients with chronic hepatitis B is 34.5 years,4 while the median age at diagnosis of hepatocellular carcinoma of Asian patients quoted by Awofeso is 57 years, by which age most would have seroconverted. Australia has been successful in primary prevention of hepatitis B through vaccination (albeit less so in migrants, some Indigenous communities and catch-up vaccination), and hence the omission of vaccination from our “wish list” for a public health response. However, Australia has been less successful in secondary and tertiary prevention. We hope that a national strategy would be a catalyst for these interventions to be given the priority they deserve.

Monica C Robotin · Jacob George · Rajah Supramaniam · Freddy Sitas · Andrew G Penman

To the Editor: We concur with Chan and Sullivan regarding the importance of targeting modifiable risk factors, such as smoking during pregnancy, to improve perinatal outcomes.1 Alcohol consumption during pregnancy is another important modifiable risk factor. Alcohol exposure in utero is associated with growth retardation, birth defects, and impaired development and neurological function. Individuals exposed to alcohol in utero may have lifelong medical and psychosocial problems.2 In our national survey of 1103 Australian women aged 18–45 years conducted in 2006, 34.2% of women reported that they had consumed alcohol during their most recent pregnancy and 16.2% had smoked. When asked whether they would consume alcohol or smoke if they were to become pregnant in the future, 23.7% said they would consume alcohol and 4.0% said they would smoke. Intention to smoke during a future pregnancy was significantly associated with intention to consume alcohol (odds ratio, 5.1 [95% CI, 2.7–9.4]; P < 0.001).3 This strong association suggests that strategies aimed at reducing smoking and alcohol consumption during pregnancy should target both behaviours. Chan and Sullivan also note the need for uniform national data on smoking during pregnancy. Although alcohol consumption is more common than smoking during pregnancy, fewer data have been collected and reported on alcohol consumption during pregnancy than on smoking. Five Australian states and territories collect data on maternal smoking status,1 and the data are published by the Australian Institute of Health and Welfare in their annual Australia’s mothers and babies report.4 In contrast, data on alcohol consumption during pregnancy are routinely collected by only three states and territories (Tasmania, the Australian Capital Territory and the Northern Territory). These data are minimal and inadequate for monitoring trends or evaluating the effectiveness of public health interventions. Smoking and alcohol can both cause preventable harm to the unborn child. Public health strategies, national uniform data collection and community education are required to address these issues in order to promote healthy pregnancies and healthy babies.

Elizabeth M Peadon · Carol I Bower · Elizabeth J Elliott

To the Editor: The authors of the recent supplement on clinical process redesign have shown that improvement can occur in the efficiency and quality of hospital care.1 They acknowledged that this process was accompanied by an investment in external consultants and a boost to the system of 1800 beds. These interventions were necessary, and have been successful in the short term. However, these measures alone may not be sustainable for hospitals in the long term with the projected needs for the health care of an ageing population.2 The table in the appendix to the supplement showed most hospitals continuing to perform poorly in their response to increasing demands on emergency departments (EDs), with corresponding increases in admissions to hospital.3 It was interesting to note that, of all the hospitals listed, Campbelltown Hospital, with the largest increase in demand (27%), also had the lowest increase in admissions through its ED (6%). A similar phenomenon was noted at Bankstown Hospital, which had a 27% increase in demand and a 13% increase in admissions. Could these have been the result of the community and ambulatory redesign that occurred in these hospitals from 2000 to 2004?4 The common feature of these two hospitals is the linkages developed between ED community nurses and general practitioners, creating a situation whereby increasing demand does not have to rely entirely on hospital beds as a solution.5 The case of an older woman with diabetes, sepsis and heart failure described in one of the supplement articles6 implicates a failure in community chronic disease management, as well as hospital care. Older people will continue coming to EDs, and will continue to be admitted to hospital in increasing numbers. To avert future crises in health care, the success of hospital redesign needs to be matched by an equally well resourced redesign of primary health and community care.

Stephen F Wilson · Nicholas Collins

To the Editor: The recent supplement to the Journal on clinical process redesign1 is a tepid attempt by NSW Health and their colleagues in South Australia to disguise their own shortcomings. Astute readers need more than fancy jargon, acronyms and pretty diagrams to be convinced that reform of the sort suggested by NSW Health cuts the mustard. Indeed, some of the language (“The process will proceed with or without you”2) provides disturbing insight into the mindset of those at the helm. Also, the ideas exemplified in another section are at clear odds with what I experience daily as a senior clinician. To illustrate, McGrath and colleagues write of “engaging clinical leaders” and that “solutions need to be evidence-based”.3 Regrettably, the opposite is the reality. Senior medical clinicians are sidelined and the decision-making process becomes the domain of a few select individuals, thus making it sclerotic, remote from the clinical interface and, at times, autocratic. Further, some management decisions are implemented without any of the supposed evidence base that McGrath and colleagues3 refer to. Perhaps of greater concern is the notion of “stretch targets”.3 These are considered “essential to stimulate real innovation”, but, translated into plain English, sound like asking staff to work harder with no additional resources. In the past 15–20 years I have witnessed several changes in senior hospital management, each bringing the “latest and greatest” ideas on public hospital reform. Sorry folks, we don’t need more of this nonsense. Rather than yet another futile cycle of reform per se, the community needs to have a debate on the level of health care that it wants, contrasted with how much of the “pie” should be consumed relative to other needs. Until that happens, we are merely pretending that “process redesign” is the answer to our problem.

George Larcos

To the Editor: In regard to your recent supplement dedicated to clinical process redesign in health care,1 we support the need to learn from other industries, but have concerns about an exclusive focus on process redesign to improve the quality and safety of health care for patients. In industry, unlike in health care, outcomes surveillance is almost always feasible. Clinical practice reminds us of the fallibility of surrogate measures of benefit, which, like “processes”, require validation by “hard” outcomes.2 Industry knows when its services or products meet minimum quality standards and satisfy “customer” needs. By contrast, a “lean approach” in health care, while taking a patient-centred perspective of care processes, makes several assumptions. The first is that better flow processes relate directly to improved outcomes. Process redesign focuses on measuring targets such as access block and elective surgery waiting lists, which are subject to confounding and manipulation.3,4 The impact on health care outcomes remains unclear. Furthermore, the studies cited in these supplement articles had no concurrent controls.3,4 The second assumption is the relative values of the process measures chosen. Is a relatively small reduction in time spent in the emergency department more important than use of an appropriate device for relieving pressure ulcers in an older woman with a fracture?5 The third assumption is that there are cause and effect links between process redesign and outcomes, such as fewer episodes of litigation at Flinders Medical Centre.6 Such links cannot be substantiated without a control group, as concurrent changes, such as open disclosure and guideline implementation, may have similar effects. Other literature pertaining to the benefits of lean thinking in process redesign emphasises opportunities to reallocate resources to implementing best practice as a result of the efficiencies and cost savings achieved.7 This assumes that process redesign occurs quickly, and that all cost savings are reallocated. These claims appeal to managers and directors constrained by external demands for meeting efficiency targets. The inevitable consequence is that organisations are likely to focus on simple “fixable” problems rather than more fundamental system-based problems requiring more resources and longer timeframes. Also, it can be argued that the process redesign examples are “micro-reforms” within a “macro-system” that remains unchanged. The hospital sector needs broader redesign wherein existing models and systems of care, and not just internal processes, are subject to critical review and improvement. Process redesign should be viewed as a useful tool, but the primary starting point must continue to be the delivery of evidence-based care, which is known to give patients the best chance of optimal outcomes.

Caroline A Brand · Peter A Cameron · Peter B Greenberg · Ian A Scott

In reply: We agree with Wilson and Collins. Community-based or ambulatory alternatives to admission to an acute facility are essential adjuncts to the redesign and increased bed capacity referred to in our article.1 The capital costs alone will be prohibitive if our only strategy is adding bed capacity. We see a significant shift in capacity from the acute to community sector as eminently amenable to redesign methods: to map current constraints (as the issue is not just inadequate community services), engage clinicians in changing their referral and treatment patterns, improve awareness of alternatives, identify new processes to facilitate use of the community as a viable alternative, and embed these new behaviours through easily accessible redesigned pathways. We chose the case study of the frail older patient deliberately, as it highlights how our current default option, hospitalisation, does not necessarily give these patients the best outcomes. In response to Larcos, we are concerned that the frustrating complexity of our current system for patients and frontline staff alike has produced so many clinicians who, like him, are cynical about improvement. Good redesign activity does engage clinicians, and our best improvements arising from redesigned processes are those that have incorporated clinician and patient input. That is the practical everyday “evidence” on which good redesign is based. Redesign does not ask staff to work harder. Significant leaps in performance can be achieved by redesigning to make an increased throughput easier to deliver. Good redesign eliminates the frustrating and wasteful steps in care that add no value to the staff or patient experience. Finally, in response to Brand and colleagues, both delivery of evidence-based care and process redesign are required to improve access to services and, hence, equity, patient flow, and patient and staff experience, and to reduce wasted effort. There is mounting evidence that better flow processes are associated with better outcomes. An Australian study has shown that delayed progress through Australian emergency departments (EDs) is associated with increased mortality.2 Our own article illustrates a 30% reduction in statewide mortality in New South Wales EDs as flow improved, with a concomitant reduction in statewide hospital standardised mortality rate.1 When patient flow improves because constraints and disconnects are eliminated, then system efficiency improves; it is only when patients are processed with indecent haste that one might expect a deterioration in quality. The improvements reported were not just measured in minutes from initial triage, but also in fewer hours spent in an ED before transfer to a ward, and in fewer days of waiting for tests and consultations as an inpatient before discharge. The results provided in the supplement3 are for an entire state health system, the largest in Australia, illustrating a turnaround in state performance, and they therefore warrant serious consideration.

Tony J O’Connell · David I Ben-Tovim · Brian C McCaughan · Michael G Szwarcbord · Katherine M McGrath

To the Editor: A 26-year-old woman presented with 24 hours of severe constant epigastric pain and vomiting. She had no history of similar pains, alcohol consumption or gallstones. Four days before presentation she had received her first dose of human papillomavirus (HPV) vaccine. Two days after vaccination she developed a fever and self-limiting rash of 3 days’ duration. Examination revealed marked epigastric tenderness and temperature of 40°C. Other physical parameters were within normal limits. Biochemical investigations showed normal liver function, moderate leukocytosis, a serum amylase level of 1900 U/L (reference range [RR], 23–85 U/L) and lipase level of 3400 U/L (RR, 0–160 U/L). An upper abdominal ultrasonography showed a non-dilated biliary tree and no evidence of gallstones. Computed tomography showed an oedematous pancreas with peripancreatic fat stranding and arterial enhancement of the pancreatic parenchyma, consistent with pancreatitis without necrosis (Box). Other investigations showed normal serum levels of calcium, triglycerides and parathyroid hormone. Serological tests were negative for acute infection with coxsackie A9, coxsackie B1–6, echo, mumps, herpes simplex, hepatitis and varicella zoster viruses. The patient was diagnosed with pancreatitis and treated conservatively with intravenous fluids and analgesia. Pain, symptoms and biochemical abnormalities settled after 10 days. She was discharged and remains well. Magnetic resonance cholangiopancreatography performed after discharge showed no pancreatic parenchymal or ductal abnormality. Acute pancreatitis is common, with an incidence of 5.4–80 per 100 000.1 Gallstones and alcohol use account for 70%–85% of cases; other causes include drugs, viral infections, tumours, hyperlipidaemia, hypercalcaemia, trauma, iatrogenic injury and pancreatic ductal anomalies. The cause is unidentified in up to 10% of cases.1,2 Viral pancreatitis is well recognised, with cytomegalovirus and mumps, coxsackie, hepatitis, herpes simplex, and varicella viruses all known causes.3 Vaccines have been implicated, with pancreatitis associated with measles–mumps–rubella and hepatitis A and B vaccines.4,5 To date, there has been no report linking HPV vaccination with pancreatitis. The pathophysiology linking vaccination with pancreatitis is unclear. It has been postulated that viral replication in immunodeficient hosts receiving live attenuated viral vaccines can cause pancreatitis. Alternatively, “molecular mimicry” could stimulate production of auto-antibodies, which react with host antigens and cause autoimmunity.5 The HPV vaccine is a quadrivalent, recombinant, non-infectious formulation, eliminating viral replication as a mechanism of pancreatitis. Therefore, an autoimmune mechanism is possible. Extensive clinical testing has demonstrated the safety of HPV vaccine in the general population. In our patient, intensive history taking and investigation failed to identify another cause for pancreatitis, and the close temporal relation of the HPV vaccination, the development of a prodromal illness, and fever without evidence of sepsis led us to postulate that pancreatitis was secondary to vaccination. A coincidental illness causing pancreatitis cannot be ruled out, but neither can HPV vaccination be excluded as a potential cause. We therefore suggest that pancreatitis be considered in cases of abdominal pain following HPV vaccination and if proven, notified to the Adverse Drug Reactions Advisory Committee. Computed tomography scan of the abdomen in a patient with pancreatitis Portal venous computed tomography images showing oedematous enlargement of the pancreas, with surrounding fat stranding and ascites. The pancreas (arrows) appears fully enhanced with contrast, suggesting there was no necrosis. A: Pancreatic head. B: Pancreatic body and tail.

Amitabha Das · David Chang · Andrew V Biankin · Neil D Merrett

To the Editor: Australian consensus guidelines for selecting formulas for infants with cows milk protein allergy (CMPA) have recently been published.1 We reviewed formula choices and outcomes for 51 children with immediate allergic reactions to cows milk protein who were referred to one of us (S S M) in a tertiary specialist clinic over a 2-year period before the guidelines were published. The formula was selected by the referring specialist medical practitioner in 44 cases (and by S S M in the other seven). Of the 51 children (mean age at initial reaction to cows milk protein, 7.8 months), 42 had skin and/or gastrointestinal features, and nine had an anaphylactic reaction with respiratory and/or cardiac features. Forty-six children had a positive skin prick test to cows milk protein, and one had a positive radioallergosorbent test. Four children with immediate (< 30 min) reactions of generalised erythema and/or angioedema (3) or vomiting (1), but a negative skin prick test, were also included. Soy was the most common formula used, followed by extensively hydrolysed formula (EHF) (Box). Three of eight children commenced on EHF had allergic reactions, with urticaria and angioedema, and one child also had a transient (60 s) cough. Three children were given partially hydrolysed formula (PHF), with one experiencing an immediate cutaneous reaction. These observations suggest that, in clinical practice, soy is frequently a satisfactory first choice for children with CMPA, as suggested in the guidelines.1 Some children with CMPA will also react to EHF, providing a rationale for choosing amino acid-based formula as a first-line treatment prior to allergy evaluation in children with anaphylaxis to cows milk protein. As about 5% of infants with CMPA also react to EHF,2 some allergists advocate the introduction of EHF under medical supervision in either all children with immediate CMPA3 or only those who have had severe life-threatening reactions.4 Although PHF is tolerated by a significant proportion of children (70%) with immediate CMPA,3 it is not recommended for the treatment of CMPA1 due to its high content of potentially allergenic cows milk protein. The fact that three children with CMPA were given PHF suggests there is confusion in the prescribing community, and that the availability of the new guidelines may help in achieving a more appropriate choice of formula. Feeding choice for 51 children referred with cows milk protein allergy Type of feeding selected No. of children Mean age at initial reaction to cows milk (months) No. who reacted to selected feeding Soy 29 9.5* 0 Extensively hydrolysed formula (EHF) 8 5.3 3 Amino acid-based formula (AAF) 6 4.2 0 Partially hydrolysed formula (PHF) 3 6.0 1 Breastfeeding 5 6.0 0 * P < 0.05 for soy versus PHF, EHF, AAF or continuing to breastfeed (t test).

Sam S Mehr · Andrew S Kemp

To the Editor: A 37-year-old man was referred to our emergency department with an acute 7 cm abscess of the buttock. The abscess was incised, and the patient was prescribed oral β-lactam antibiotics and discharged. After 48 hours, culture of samples from the abscess showed methicillin-resistant Staphylococcus aureus with a community-acquired antibiotic resistance pattern (CA-MRSA). The isolate was resistant to β-lactam antibiotics, but sensitive to trimethoprim, gentamicin, and tetracycline. Unusually for an Australian CA-MRSA strain,1 the isolate was also resistant to erythromycin and ciprofloxacin. On reviewing the patient’s history, it was noted that he was a previously well United States resident who had visited Australia and New Zealand as part of the support team for an international rock band. Further testing was undertaken, and the isolate tested positive for genes coding for the Panton–Valentine leukocidin toxin, associated with staphylococcal virulence (eg, recurrent furunculosis, abscess formation, and necrotising pneumonia).2 Pulsed-field gel electrophoresis (performed by the Gram-Positive Bacteria Typing and Research Unit, Department of Microbiology and Infectious Diseases, Royal Perth Hospital, WA) confirmed the isolate as the ST8-MRSA-IV strain, also known as USA300. Most CA-MRSA strains remain susceptible to a majority of non-β-lactam antibiotics, including clindamycin, trimethoprim–sulfamethoxazole, tetracyclines and fluoroquinolones. This helps distinguish CA-MRSA isolates from the typically multiresistant hospital strains, and facilitates oral outpatient therapy. USA300 is the dominant strain causing CA-MRSA infections in the US.3 Among 422 patients with soft-tissue infections presenting to 11 US emergency departments in 2004, 59% of cases were caused by CA-MRSA, of which 99% were USA300. Recent reports indicate that multiresistance is emerging within this strain, with acquisition of resistance to erythromycin, clindamycin, mupirocin and fluoroquinolones. An increasing association of USA300 infections with buttock and perineal infections is also reported, as well as potential sexual transmission, particularly among men who have sex with men. Our case highlights the ease of international spread of microorganisms. Arguably, a “one-night stand” tour could be an ideal vehicle for microbial dissemination. CA-MRSA was not considered in the patient’s initial assessment, and the patient was discharged with oral β-lactam antibiotics and no planned follow-up. Moreover, the isolate may not have been identified as the “epidemic” USA300 strain without more involved tests. A recent study documented a rising incidence of USA300 isolates in Western Australia between 2003 and 2007. Of 61 patient isolates, 35 were diagnosed in 2007 (Pearson J, Coombs G, Christiansen K, et al. USA300 MRSA identified in the Australian community [abstract PP3.2]. Abstract presented at the Australian Society for Antimicrobials 9th Annual Scientific Meeting; 2008; Feb 21–23; Sydney). Our case suggests we should be more alert to CA-MRSA infection presenting with furunculosis and soft tissue infections, not only in Indigenous communities and young people, but also in international travellers and patients whose infections fail to respond to usual therapy. It also reinforces the value of incision and drainage. As β-lactam susceptibility is no longer assured, such specimens should routinely undergo culture and susceptibility testing.

Thomas Gottlieb · Wei-Yuen Su · John Merlino · Elaine Y-L Cheong

To the Editor: We report significant sinus bradycardia in a patient presenting with an acute coronary syndrome shortly after beginning bupropion therapy to assist with smoking cessation. A 53-year-old man was attended by paramedics for typical ischaemic chest pain. He had sinus bradycardia (45 beats/min) and hypotension (blood pressure, 85/60 mmHg), and was found to have a serum troponin I concentration of 1.2 μg/L, but no diagnostic electrocardiographic changes. He was admitted to our hospital with an acute coronary syndrome. He reported his medications at the time of admission as including metoprolol 50 mg twice daily (for hypertension) and paroxetine 20 mg daily (for depression). The patient was given multiple doses of intravenous atropine (total, 1.2 mg) and adrenalin (total, 2 mg). After an adrenalin infusion was begun, he developed ventricular tachycardia (170 beats/min), but his cardiac rhythm spontaneously returned to sinus bradycardia. Two days after admission, two coronary stents were successfully deployed in a critically stenosed right coronary artery. Bradycardia (45–50 beats/min) persisted. The following day, it was discovered that 3 weeks previously, the patient’s general practitioner had prescribed bupropion 150 mg twice daily to assist with smoking cessation, which he had been taking up until the day of admission. Bradycardia continued until hospital discharge. One month after discharge, he was in sinus rhythm (60 beats/min) and was clinically well. Bupropion is a selective noradrenalin, dopamine and serotonin reuptake inhibitor. The mechanism by which it enhances the ability of patients to abstain from smoking is unknown.1 Bupropion inhibits the activity of the cytochrome P450 2D6 isoenzyme, which metabolises metoprolol.2 Concurrent use of bupropion and metoprolol can increase serum metoprolol levels, and clinically significant bradycardia has been reported.3 Further, paroxetine, a selective serotonin reuptake inhibitor (SSRI), is a potent cytochrome P450 2D6 inhibitor, which would have further increased serum metoprolol levels. Bradycardia associated with metoprolol and paroxetine dual therapy has been described.5 Additionally, there is the potential for serotonin syndrome to develop in a patient administered multiple SSRIs. In our patient, the administration of bupropion and paroxetine could have potentially led to serotonin syndrome.6 Our patient’s pharmacological profile was complex, with potential adverse pharmacodynamic effects. The most likely precipitant of the patient’s bradycardia was his acute coronary syndrome, although bupropion may have contributed. The case highlights the potential for significant drug interactions when new drug therapies are initiated. Bupropion and metoprolol (and other drugs metabolised by the cytochrome P450 2D6 isoenzyme pathway) should be co-administered with caution. The importance of common pathways of drug metabolism should be recognised to avoid potential adverse events, particularly when multiple medications are used.

Jacqueline Landau · Andrew E Ajani

To the Editor: I challenge the assertion made by Donohoo (Managing Editor of MIMS [the Monthly Index of Medical Specialities]) that “MIMS is held . . . in high regard” and that the “vast majority of MIMS subscribers recognise that the quality information provided by MIMS is essential in their daily encounters with their patients”.1 In fact, the most common MIMS annual to be seen around hospitals, in nursing homes and in doctors’ surgeries is an out-of-date one. Furthermore, as a general practitioner, when I do use MIMS, it is because it is packaged with our desktop software, rather than by choice or active decision. I have online access to the Australian medicines handbook (http://www.amh.net.au/), and various other references. I have no need to refer to MIMS, and I tire of the understandable bias MIMS has always had for proprietary prescribing.

Lilon G Bandler

To the Editor: Walters recently suggested that Australia should implement population control strategies as part of an approach to reduce global warming.1 As a father of four, I found his assertion that my decision to father more than two children is “arrogant” to be offensive. Walters’ arguments are, at best, poorly reasoned. As a “citizen of this world”, he clearly rejects the rights of other citizens to live on an equal footing and follow their religious, cultural or social beliefs if those beliefs oppose contraception. I would argue this is contrary to law.2,3 His mathematical calculations ignore all costs required to achieve his objective, such as those associated with “contraceptives, intrauterine devices, diaphragms, condoms and sterilisation procedures”. Further, he fails to consider costs associated with the supporting bureaucracies required to effect his policy, including material amendments to the Australian taxation system. Rather, and in my opinion strangely, he advocates issuing carbon credits for the additional consumption of contraceptive products. According to Walters, people should be judged by their anticipated rather than actual emissions. A logical extension would be to punish those who exceed a predetermined mean acceptable level of emissions. No doubt, meeting the medical and ancillary needs of many sick, older and disabled people often generates excess emissions. Perhaps we should adopt some of the practices used in China and India, including abandonment and neglect of disabled children and older people.4,5 How would we deter and punish those who cannot pay? Walters addresses the issue of overpopulation by comparing Australia to India and China. This is notwithstanding that Australia has one of the lowest population growth rates in the world6 and, with its ageing population and labour shortages,7,8 has significantly different population and social concerns to these countries. There are no grounds to support the comparison made. Environmental issues are among the greatest challenges facing society. As a father, I am deeply concerned for the world my children will inherit. We must deploy our limited resources efficiently and effectively to maximise their impact. To demand social controls in the manner Walters suggests, within a society heavily burdened with laws and struggling to meet labour and health system demands, would defeat this objective. Reading Walters’ views, which I consider fundamentally flawed, in a publication such as the Journal imparts to them a validity I believe is unjustified. I do not consider that Walters’ social engineering policies could benefit anyone in Australia, while his “moral” concerns are ill conceived.

Cathal A Smith

In reply: I thank Smith for the opportunity to clarify some scientific points. The science behind climate change is undeniable and was reviewed in February this year. Moreover, “there is a greater than 90 per cent probability that the warming observed since the 1950s is due to human activities”. Therefore, attempts to prevent environmental calamity will not succeed with boundless population growth. In this sense, the more people there are, the worse it is for our earth. In particular, no nation should encourage population growth. I do not argue for compulsory sterilisation. I do argue that we recognise the cost of every extra human being to our overburdened earth. Smith labelled my note of caution about limitless procreation as “offensive”. I believe such disparagement is founded on personal and cultural beliefs, not on science, which informs and guides medicine. There is only one atmosphere. Australians occupy this planet with no more rights than others do. Racism is anathema to us. If others must observe population restraint, then so must we. Contrary to Smith’s assertion, I plead that all should be able to “live on an equal footing”. Is this not the laudable basis of law? I share his concern for the world that his “children will inherit”; my concern embraces the children of others as well.

Barry N J Walters

To the Editor: We read Pham and colleagues’ recent article1 with interest, as evidence mounts against the use of probiotics in critically ill patients. Although a plausible and attractive theory, probiotics in the patient with acute illness now appear ineffective, if not positively harmful. A recent randomised trial of probiotics in 298 patients with severe acute pancreatitis showed a non-significant rise in infective complications,2 in keeping with results of previous studies of critically ill patients.3,4 Disturbingly, mortality in the probiotic group was more than double that in the placebo group (P < 0.01). Bowel ischaemia was a prominent feature of deaths in the probiotic group (eight patients), but was not associated with any deaths in the placebo group (P < 0.004). It may be that non-occlusive mesenteric ischaemia in critical illness is exacerbated by the added bacterial load itself, or through a pro-inflammatory response by gut epithelial cells. While probiotics may be a benign and beneficial adjunct to enteral feeding in certain clinical situations, there is persuasive evidence that probiotic therapy is associated with increased infective complications in critically ill patients and significant mortality in patients with severe acute pancreatitis. Until there is evidence to the contrary, we believe probiotics should not be administered to patients with severe acute illness.

Shimonti Chatterjee · John Fraser

To the Editor: Pham and colleagues commented on the effects of probiotics; however, not much is known about the impact of probiotics on weight gain and obesity. It is known that a predominance of certain bacteria, such as Lactobacillus, in the bowel can promote weight gain. Many of these bacteria are found in probiotic products. The human intestinal microbiota is predominantly colonised by the Firmicutes and Bacteroidetes phyla of bacteria. Lactobacillus and Bifidobacterium, found in a number of probiotic products, belong to the Firmicutes phylum. A study has shown that obese people carry a higher proportion of bacteria from the Firmicutes phylum and that there is a statistically significant decrease in the proportion of Firmicutes bacteria as they lose weight.3 A similar pattern of Firmicutes predominance has been found in obese mice. Furthermore, the microbiota of the obese mice were more likely than those of lean mice to break down otherwise indigestible polysaccharides from the diet.4 In other words, a higher proportion of Firmicutes bacteria was associated with increased and more efficient caloric uptake from food. These data did not necessarily imply causation, so the investigators performed another experiment. They transferred intestinal microbiota from obese and lean donor mice to germ-free mice, and found that the mice who received microbiota from the obese donors had a significant increase in body fat after 2 weeks compared with the recipients from the lean donors.4 It is therefore likely that the bacteria often found in probiotics can cause weight gain. Obesity in children and adults is a major health problem in developed nations. Given the increasing use of probiotic products in such countries, large studies should be performed to characterise the association between probiotics and obesity. Such studies may not find any association or may find that there is only a dose-related risk. If there is an association, probiotics may still prove useful in certain circumstances (eg, for weight gain in children failing to thrive).

Sanjaya N Senanayake

In reply: The comments by Chatterjee and Fraser regarding the danger of administering probiotics to patients with acute severe illnesses are important additions to the debate on the risks and benefits of probiotic administration. We also thank Senanayake for his interesting comments on the possible role of probiotics in weight gain. The recently published multicentre trial1 describing unexpected adverse events associated with probiotics in acutely unwell patients with severe pancreatitis is one example of an adverse outcome following probiotic administration. The use of probiotics in patients with severe comorbidities and in those who are immunocompromised is also contraindicated. There are reported cases of Lactobacillus GG sepsis in premature babies with short gut syndrome,2 and Saccharomyces boulardii fungaemia has been described in immunocompromised patients.3 It is interesting to note that two systematic reviews have assessed the efficacy of probiotics in prevention of necrotising enterocolitis in premature (< 33 weeks’ gestation) and very low birthweight (< 1500 g) infants.4,5 Both reviews concluded that probiotics may decrease the incidence of necrotising enterocolitis in preterm infants, and that severe adverse events were not associated with probiotics in these unwell and immunodeficient patients. However, there were insufficient data to comment definitively on the short-term or long-term safety of probiotics in these infants; this will require assessment in future large trials. The increased scrutiny of probiotics resulting from the publication of the adverse outcomes in adults with severe acute pancreatitis1 may, by necessity, slow the commencement and progression of these larger trials in infants in the neonatal intensive care setting.

Mimi Pham · Daniel A Lemberg · Andrew S Day

To the Editor: On 25 January 2008, the Australian Transport Council approved the National Transport Commission’s seventh amendment to the Australian Road Rules. This amendment provides for the mandatory use of forward-facing child restraints for children aged 6 months to 4 years, and the use of Australian Standards-approved booster seats for children aged 4–7 years and weighing up to 26 kg. It also recommends that children aged less than 7 years should not travel in the front passenger seat. These changes are welcome. They bring Australian rules on child restraints and seating position closer to (but still not on par with) restraint laws already implemented in the United Kingdom and other countries in the European Union, where booster seats are mandatory for all children aged under 12 years or less than 145 cm tall. Implementation of the amendment poses several challenges. A small proportion of children will exceed the 26 kg weight limit for booster seats by their seventh birthday; however, there is no evidence that these seats are not safe for slightly heavier children. In addition, the Australian Standard (AS 1754) is currently being revised and is likely to move towards recommending restraint selection based on seated height rather than weight, as well as developing new standards for booster seats for older children. Height is the most important determinant of adequate seatbelt fit, and children need to be about 145 cm tall before the lap portion of an adult seatbelt sits correctly over the iliac crests rather than on the soft abdomen. Community education campaigns will be pivotal in successfully implementing these new road rules. As misuse of restraints is high, education campaigns must emphasise correct use of recommended restraints.-4 Furthermore, the new recommendations may contribute to financial hardship, particularly for low-income families with several children under the age of 7 years. Subsidies or loan schemes may be required to assist such families. Fitting three restraints across the rear seat of small cars may also be difficult. Adequate enforcement will be required to maximise compliance. Research studies and injury surveillance will play an important part in maximising the effectiveness of these rule changes. Finally, the seventh amendment to the Australian Road Rules does not constitute law, and legislative changes will need to be enacted by each state and territory before these changes become law. We hope that the state and territory governments will take swift action to enact these laws, to prevent injuries and deaths in Australian children due to motor vehicle crashes.

Yvonne A Zurynski · Lynne Bilston · Elizabeth J Elliott

To the Editor: Court, Bergh and Södersten raise the issue of why and how some therapies with prima facie evidence for their efficacy have a significant take-up by medical practitioners, while others are allowed to languish, sometimes for decades.1 It is 6 years since Bergh and colleagues conducted their Swedish trial on eating disorders, with significant encouraging results.2 Again they report — albeit this time with a non-randomised but local sample — above-average outcomes.1 Again, the fact that their patients had had previous treatments that failed renders the results compelling. We have to ask why no one has found the time, money or inclination to attempt to reproduce their findings or examine which elements of their intervention are successful. It would be ironic if the answer is that medical researchers are afraid of the unusual. While Australian medical research and public health ignore this mandometer treatment, some private health funds have been prepared to contribute up to $60 000 per patient for it, suggesting that they view it as better value for money than alternative therapies.

Phillip Gray