Letters

In reply: It is with great pleasure that we resume our ongoing correspondence with Phillips concerning the medical and psychological status of patients who present with non-specific somatic complaints such as chronic fatigue.1 As we have reported previously,2 we have been strong advocates of both the need to develop appropriate instruments for measuring neuropsychiatric states characterised by non-specific somatic symptoms and to promote effective medical and psychological management of patients with these disabling conditions.3 In their letter, Phillips and colleagues fail to grasp the essential issue. To describe a condition as a neuropsychiatric state (or mental disorder) does not necessarily lead to simplistic and entirely unhelpful assumptions about "medical" versus "psychological" causes or treatments. Phillips et al attempt to promote once again the notion of "biological" (ie, acceptable) versus "psychological" (ie, unacceptable) theories of the causation of chronic fatigue syndrome. Such an approach is not only intellectually sterile and inconsistent with the past decade of intensive research by a wide range of medical and psychological research teams,4 but also profoundly unhelpful to people affected by these disabling disorders.3 In recent years, the very significant health burden of common mental disorders such as depression, anxiety, alcohol or other substance misuse, and neurasthenia (prolonged fatigue states lasting longer than three months) has been well documented in the Australian community5 and in the primary care setting.6 Phillips and colleagues appear to have no knowledge of the basic epidemiological fact that mental disorders are two to three times more common in primary and other medical care settings than in community studies (hence the total rate of disorder in our study is about twice that detected in the Australian National Survey of Mental Health and Well-being). Contrary to their implications, the total rates reported in our general practice study are entirely consistent with the largest multinational study of primary care ever conducted.7 That study indicated that a third of all primary care patients have mental disorders and another third have mental health difficulties (with or without concurrent medical disorders) requiring specific psychological assessment. The significance of our study is that it has brought the extent of common mental health needs (including depression, anxiety, alcohol or other substance-misuse and somatoform disorders) to the attention of the Australian medical profession. What is now required is a concerted and integrated response — not a return to dualistic notions of illness that have for so long hampered the provision of effective pharmacological and non-pharmacological treatments to patients with mental disorders who present for medical care.

Ian B Hickie MD, FRANZCP · Tracey A Davenport BA(Hons) · Elizabeth M Scott FRANZCP · Sharon L Naismith BA(Hons)

To the Editor: The EBM in Action article on laser treatment by Del Mar et al1 raises my anxiety about the reliability of evidence-based medicine (EBM) in general and, at the very least, the authors' assessment of the question they set out to answer. One is seldom, if ever, in a position to understand the breadth and depth of an issue unless it is the subject of particular study. Most of us can not challenge statements made in such articles without an intimate knowledge of the literature. As laser therapy is the topic of my PhD thesis, I am in a unique position to have much of the literature on the subject at my fingertips. The authors state in their conclusions that "low power laser therapy appears to be no more efficacious than placebo in relieving musculoskeletal pain". Several aspects of the analysis on which they base this conclusion cause me great concern. Firstly, they state that "The search report highlighted the high quality of evidence supporting the refined question". Quite the contrary. One of the two systematic reviews they cite2 has been criticised in the literature for its many inadequacies, not the least being that laser acupuncture and laser therapy are included in that review as if they were the same, which they are certainly not.3 In addition, the review by Beckerman et al concludes, with regard to musculoskeletal pain, that "the efficacy of laser therapy for musculoskeletal disorders seems, on average, to be larger than the efficacy of placebo treatment. More specifically, for rheumatoid arthritis, post-traumatic joint disorders and myofascial pain, laser therapy seems to have a substantial specific therapeutic effect".4 How can this fit with the conclusion of Del Mar et al? Furthermore, Gross et al state in their review, which consisted of three trials of laser therapy, that "In general, all therapies have not been studied in enough detail to adequately assess either efficacy or effectiveness".5 The choice of other articles by Del Mar et al is also somewhat mystifying in that, out of the nine references cited, one is in Russian and two in Danish. There are many other relevant articles in English that they have not cited.6,7 There is no doubt, as I myself have found, that it is difficult to search for this topic in the literature, as there are many terms used for laser therapy and the information comes from a broad range of sources. However, this is no excuse for a group that holds itself out to be "expert" in a field. It is very disappointing to see such an incomplete review with such an inappropriate conclusion based on such a poor sample of the literature. If this is an example of EBM in action, then I think we should be very concerned.

Roberta Chow MB BS (Hons), FRACGP, FAMAC, MApplSci(MedAcu)

In reply: Chow criticises us for not identifying all relevant trials. But is she willing to help others in the process of reviewing? We could not identify a systematic review of the area with her involvement, nor does she appear to have registered an appropriate protocol with the Cochrane Collaboration (instructions for which are available at <http://www.cochrane.de>). Systematic reviews are important to help clinicians make sense of a diversity of trials. If experts such as she, devoting years to the area, do not help us, who should? We were not attempting such a systematic review (which would probably take several months of work). Rather, we were trying to provide clinicians with the best obtainable answer in a 24–48-hour turnaround time.2 Given that clinicians might have one question per patient,1 attempting a systematic review with each question would give us a lifetime of work after only a fortnight of clinical work! We needed to balance the timely requirements of clinicians with the quality of the evidence obtained. As Chow points out, tracking down every last trial is difficult. Therefore, we first aimed to identify systematic reviews rather than attempting to find all trials ourselves. These reviews missed some relevant material — as indeed did Chow, who did not cite several recent studies,3,4 which makes us wonder if she has a strong prior belief that may bias her views. Would these missed trials have made any difference to the conclusions we reached previously? A systematic review published since our original literature report (January 2000) suggests not, although there may be some specific subgroups of patients and conditions for which laser therapy is effective.5 All of this highlights the need for a collective effort to sort out the mess of medical information. In seven years the Cochrane Collaboration has systematically reviewed less than 5% of more than 300 000 trials on the clinical trials registry. They are difficult to perform and maintain. Millions of dollars continue to pour into primary research that still remains inaccessible to us at the clinical frontline, the "great criticism" with which Archie Cochrane pricked the profession into action.6 Please, Dr Chow, abandon throwing bricks from the sidelines, and join us in trying to help clinicians access research evidence in the timely fashion needed for day-to-day practice!

Chris B Del Mar MD, FRACGP · Paul P Glasziou MB BS, PhD, FAFPHM

To the Editor: With respect to the article by Ferson in the Christmas issue of the Journal, a personal experience of an unwanted side effect which occurred in 1927 may be of interest. I was a boarder at school, and four weeks before sitting for the Leaving Certificate examination I contracted a violent sore throat associated with a bodywide erythema similar to sunburn, but without any associated burning sensation. The doctor had no hesitation in diagnosing scarlet fever, and I was transferred to the Coast (now Prince Henry) Hospital, which was the infectious diseases hospital for leprosy, scarlet fever, diphtheria and the like. I had asked if I could take my textbooks to the hospital, but was told that if I did they would be destroyed when I was discharged. I left the books at school, spent four weeks in isolation and returned to school with one weekend to prepare for the examinations — the results were quite disappointing!

Sir Keith Jones

To the Editor: In their recent article on the use of modelling in pharmacoeconomics to estimate the potential benefits, risks and costs of preventive drugs, Liew and colleagues highlighted important strengths and limitations of this technique.1 One limitation is that modelling is discretionary: different analysts elect different models and get different answers. We argue that modelling is the first step. The next step is testing the predictive validity of the model by systematically collecting cost and effectiveness data over a period of time. Then the predictions of the original analysis could be compared to what actually transpired. The goal of pharmacoeconomics is the most accurate estimation of costs and benefits. The more these variables are truly study outcomes (that is, experimental and not constrained by assumptions of the economic model), the more valid the process. This will help reduce model discretion, improve data quality and increase the likelihood that the experiment could be replicated independently, the acid test of validity. In reducing the discretion inherent in pharmacoeconomics, we can allay the concerns of those who have questioned its underlying theory2 and validity.3 If one does not know the benefit of a drug, one conducts a study. Equally, if one does not know the cost of a drug, one needs a study. This has been the impetus for randomisation in design of pharmacoeconomic studies, thereby decreasing reliance on economic modelling.4 Furthermore, there are well-tested methods for quantifying the uncertainty of estimates obtained with randomised studies. In contrast, economic modelling assesses the robustness of the model assumptions, as reflected in the estimate, using sensitivity analysis. However, this analysis cannot separate uncertainties attributable to the model assumptions, uncertainty inherent in the data put into the model, and uncertainty of outcome estimates. One is left with nostalgia for the simplicity of the null hypothesis. How can we move forward? To reduce reliance on modelling and to collect better data, we propose that Australia, with its "culture of evaluation",5 again take the lead by creating a new "conditional listing" category on the Pharmaceutical Benefits Scheme for all drugs, not just preventive therapy. This would be available for selected products with strong biological rationale but inadequate current evidence on cost-effectiveness. By necessity, these would include only high-volume/low-cost and low-volume/high-cost products, as high-volume/high-cost products are rarely developed, and low-volume/low-cost products are not problematic. The sellers would then collect prospective data to substantiate cost-effectiveness of the products or would have them delisted. This would be truly innovative and, like the Pharmaceutical Benefits Advisory Committee itself, a first for the Commonwealth. Certainly, no other jurisdiction is even considering this, let alone proposing systematic study. With better data, we would learn which models and model assumptions yield accurate predictions. Of course, many challenges and difficulties will need to be addressed regarding this proposal, but in our opinion none are insurmountable, and the benefits of a program of this type clearly exceed the risks.

Kent R Johnson · Marissa N Lassere

To the Editor: Carotid stenting is a new application of endovascular therapy. Its efficacy in preventing strokes is yet to be established, by contrast with the proven Level 1 evidence of benefit from carotid endarterectomy. The risks of implanting carotid stents at present appear greater than the risks of carotid endarterectomy. An overview of carotid endarterectomies in Australia is maintained by vascular surgeons, through audits such as the ongoing Melbourne Vascular Surgeons Association Audit and the New South Wales Carotid Endarterectomy Audit. The technique of carotid stenting, the stents themselves and the brain-protective devices used during the implanting of stents are expensive and still evolving. The long-term durability of stents is unknown. Australian vascular surgeons, neuroradiologists and neurologists are awaiting the outcome of two major international randomised trials of carotid stenting versus endarterectomy (the US Carotid Revascularization Endarterectomy versus Stent Trial and the European International Carotid Stenting Study). These seek Level 1 evidence of the comparative risks and success of the new stenting procedures in stroke prevention and aim to document the late outcome of stenting, particularly the incidence of restenosis, which is a significant problem in other arteries after stenting. While these definitive trials are in progress, vascular surgeons of the Royal Australasian College of Surgeons wish to add their note of caution to the reservations expressed in the NHMRC guidelines on stroke prevention1 and the recommendations of the Australian Association of Neurologists.2 A recent commentary by Spence and Eliasziw3 illustrates the disparate nature and the limitations of existing studies of carotid stenting. We consider carotid stenting is not yet appropriate for widespread use in Australia. Experienced endovascular and neurology teams should continue to evaluate the new procedure. Stenting of symptomatic carotid atheroma should only be conducted with the consent of patients who are fully informed about stenting's known hazards and unproven status and who understand that the established treatment is carotid endarterectomy.4 Clinicians should audit closely the immediate outcome and long-term complications of any carotid stenting they perform.

Peter L Field

To the Editor: Preparations containing isoflavone phytoestrogens are widely used as an alternative therapy for treating symptoms of the menopause. Although Australian government regulations strictly control the components of alternative therapies, adherence to the stated amounts of the components in alternative therapies is not routinely assessed. Isoflavones exist in two forms — aglycone (the free form) and glycosylated or glycone (the conjugated form) — the relative proportions of which vary between preparations. As glycosylation contributes considerably to the mass of isoflavone molecules, it is relevant to consider the total amount of potentially available isoflavone in alternative therapy preparations. Isoflavone-containing preparations which had a recommended daily dose on their labels were purchased at random from pharmacies around Sydney during September 1999. Where possible, products from more than one manufacturing batch were purchased and all products were well within their stated shelf life. The tablets, capsules or powder were removed from their packaging to conceal their identity and randomly allocated to numbered plastic bags by the hospital pharmacy department. The samples were then sent to PhytoChem Technologies Inc (Chelmsford, Mass, USA), an independent reference laboratory for the assay of isoflavones. There, isoflavones were extracted within four months of purchase (and before their stated use-by date) from 500 mg of each specimen after dissolution in 70% methanol. Glycosylated and free isoflavones were assayed in duplicate by gradient high-pressure liquid chromatography, with detection of isoflavones at 254 nm using a Waters 996 series photodiode detector with a limit of detection of 0.2 µg/mL. The identity of chromatogram peaks was confirmed by UV–V spectral analysis, and by comparison with standards. The mobile phase was acetonitrile, and adequate peak separation, linearity, accuracy and reproducibility were demonstrated. The total amount of available aglycone isoflavones in each sample was estimated (see Table). Only two products (Phytolife and Promensil) had total isoflavone contents close to the stated amount, and the content of the Phytolife product was variable. Estimated aglycone contents of preparations demonstrated that glycosylated isoflavones contributed substantially to the stated content of the product. A previous study of isoflavone-containing preparations marketed in the United States produced similar results to ours.1 Consumers may wish to consider not only whether an alternative therapy is of use, but also whether the product they purchase contains what they expect. Actual and stated isoflavone content of commercially available preparations Manufacturer Product No. of batches assayed Stated isoflavone content (mg) in recommended daily maximum dose of product Actual total isoflavone content per daily dose (mg) Estimated aglycone isoflavone content per daily dose (mg) Blackmores Phytolife one a day 5 40 41.02 ± 6.12 25.75 ± 6.04 Bioglan Soy powder plus 4 68 48.75 ± 1.42 30.44 ± 0.86 Earths Own Soy + calcium 1 68 42.52 25.67 Health Direction Femme phase 1 235 mg soy protein* 0.29 0.20 Herron Phyto source 1 22.5 16.27 9.93 Natural Nutrition Menopause 1 60 0.56 0.51 Natural Nutrition Phytobalance 3 90 58.12 ± 6.26 34.96 ± 3.79 Novogen Promensil 4 40 40.12 ± 1.98 38.38 ± 1.20 Pretorius Maxi soy plus red clover wild yam and calcium 4 68 50.36 ± 1.64 31.24 ± 1.13 Wagner Probiotics Femme soy plus with red clover 2 27 30.76 ± 0.12 19.65 ± 0.05 * Soy protein has a high isoflavone content. Values are the mean ± standard deviation. Total isoflavones = glycone plus aglycone. Estimated available aglycone isoflavones = weight of aglycone isoflavones plus weight of glycone isoflavones corrected for glycone content.

Jan B Howes · Laurence G Howes

To the Editor: The 50 g glucose challenge test (GCT) is widely recommended as a screening test for gestational diabetes (GD).1 The test consists of a 50 g oral glucose load given at any time of the day, followed one hour later by the measurement of the plasma glucose concentration.2 This test is recognised as imperfect for screening, as sensitivity and specificity are not 100%.2,3 It is known that glucose tolerance deteriorates in the afternoon,4 which raises the question of whether time of day influences the response to the 50 g GCT. At Royal North Shore Hospital, screening for GD is performed at the 26–28-week visit by means of the 50 g GCT. In 2000, screening for GD was introduced into a morning midwives antenatal clinic, whereas previously it had only been performed in the afternoon. The population attending the clinic at the 26–28-week visit includes many women receiving shared care, and is regarded as being at low obstetric risk. The Table shows the results of screening at the morning clinic compared with screening in the afternoon over the same time period. The two groups were identical in terms of age, weight, ethnicity, and family history of diabetes or past history of GD. The percentage of women with a positive screening test result during the morning clinic (17.0%) was significantly lower than that during the afternoon clinic (31.1%). Positive screening results were followed up with a diagnostic 75 g glucose tolerance test, and GD was diagnosed according to the Australian Diabetes in Pregnancy Society criteria.5 Women with a positive screening test result confirmed with a 75 g glucose tolerance test in the afternoon were less likely to have GD than those with a positive test in the morning (31.5% v 40.0%). Despite the fact that a smaller percentage of women who screened positive in the afternoon had GD, a greater percentage of the total number screened in the afternoon had GD than in the morning group. In this cohort, the difference (9.8% v 6.8%) was not significant (Table; P = 0.15). These results are consistent with the hypothesis that a 50 g GCT test performed in the afternoon results in a greater number of positive results, a greater number of women undergoing diagnostic testing and a greater number of women identified with GD. The morning GCT appears to increase specificity, with an associated decrease in sensitivity. These results need to be taken into consideration when designing or implementing a screening program. Screening for gestational diabetes (GD): the effect of screening time Time Morning (0930–1200) Afternoon (1205–1710) Number screened 176 470 Age in years (mean ± SD) 31.2 ± 4.7 31.7 ± 5.0 Weight (mean ± SD) 59.4 kg ± 10.5 kg 60.8 kg ± 12.9 kg Family history of diabetes 27 24 Past history of gestational diabetes 1 3 % White/Asian/Middle Eastern 62.6/28.0/9.0 67.5/25.9/5.8 Positive result, 50 g glucose challenge test 30 (17.0%) 146* (31.1%) Abnormal result, 75 g glucose tolerance test 12 (6.8%)† 46‡ (9.8%)† *P < 0.001, χ2. † % Of number screened. ‡ P = 0.15, χ2.

Aidan McElduff · Rosemary Hitchman

To the Editor: We have recently noticed changes in the incidence of invasive cervical carcinoma in the Gippsland Health Region and would like to know whether other regions have noticed similar demographic changes. During 24 months in 1999–2000, 19 women (median age, 59 years; range, 33–88 years) with squamous carcinoma were registered in our pathology practice. Based on information from the Victorian Cervical Cytology Register, almost half (10 women) had no previous cervical smear history whatsoever, while three had had smears, but at irregular intervals up to 14 years apart. The remaining six women had had regular Pap smears, with 1–3 negative smears preceding the diagnosis of cancer. Sixteen of the women had consulted their general practitioner for some other ailment before the cervical cancer was discovered (median interval, 14 months), but no cervical smear had been obtained. Of particular interest is the fact that six of the 19 patients are in their seventh decade or older, with a median age of 80 years (range, 79–88 years). The Victorian Department of Human Services reports that the cervical-smear participation rate for eligible women in the Gippsland region is 68%, a rate not much different from the other regions.1 The two-yearly participation rate for the 60–69-years age group is 56%, and, although no official figure is available for women in their seventh or eighth decades, it is likely to be considerably lower. The Cancer Epidemiology Centre has recorded that, over a 16-year period, the incidence of cervical carcinoma in Victorian women aged over 70 years fell by 50%, and simultaneously there has been a shift in the peak incidence from the 70–74-years age group to one a decade older (Vicky Thursfield, Information Manager, personal communication). Accordingly, we suspect that women over 70 years of age still have a significant incidence of invasive cervical carcinoma, but are not being offered cervical smears, even when the National Health and Medical Research Council guidelines indicate the necessity.

Nicholas J Mulvany · Norman R Sonenberg

To the Editor: In response to Leeder and Rychetnik's article,1 evidence-based medicine (EBM) also has significant potential to reduce the quality of patient care, with obvious ethical implications. I refer to two specific issues of concern. The first relates to the increased expectation that clinicians, and especially trainees, not only understand the role of EBM in clinical practice, but actively contribute to its underlying database. Indeed, some of the professional colleges (eg, the Faculty of the Australian and New Zealand College of Anaesthetists) now include formal projects (which are often, but not necessarily, clinical trials) in their final assessment of trainees.2 This is leading to increasing numbers of poorly designed trials that are unlikely to make useful contributions to the clinical database. These commonly take two forms: studies which lack sufficient power to confirm the absence of a true difference between groups,3,4 or studies which use a placebo when effective therapeutic alternatives exist.5 Yet such studies are frequently published in reputable, peer-reviewed journals.6 Both of these types of studies are unethical, and both impact adversely on patient care. Increasing the evidence base of clinical medicine is important, but our primary responsibility remains the maintenance of quality of care of all patients, especially those involved in clinical trials. Therefore, education of clinicians, and especially trainees, must emphasise the role and importance of statistics, epidemiology and study design in all areas of medicine to prevent unnecessary reductions in the quality of care of this subset of patients. The second concern relates to the increase in "quality improvement" projects that are rarely submitted to ethics committees for approval. These activities also contribute to the evidence base, primarily at a local level, but patients are usually unaware that they are involved in these projects, and that these activities may have significant quality-of-care implications for them. It is therefore important that internal hospital quality assurance activities undergo a similar level of scrutiny by ethics committees to that of clinical trials. Patients involved in any audit or project that has the potential to influence their care should be required to give informed consent. Only then can we reassure a patient that, while we are continually striving to improve the care we provide by developing the evidence base for clinical medicine, the care of each individual remains our primary concern.

Simon R Tomlinson · Kerry J Breen · Malcolm H Parker · Chris B Del Mar · Paul P Glasziou · Lucie Rychetnik and · Stephen R Leeder

Comment: Tomlinson raises two important ethical issues in clinical research. While there are several ethical issues which may arise in acquiring and applying evidence to clinical practice and health administration, as outlined previously by Kerridge et al1 and picked up well by Leeder and Rychetnik,2 it is misleading if Tomlinson wishes to imply that the two issues he has raised can be attributed to the evidence-based medicine (EBM) movement. His first issue could be summarised as "badly designed clinical research should be seen as unethical", for reasons which may include wasting scarce resources or placing participants at discomfort or risk when the likelihood of benefit is slight. He alleges that poorly designed studies are "frequently published in reputable peer-reviewed journals", a statement which would surprise most editors. Peer review prior to publication, and the earlier independent prospective scrutiny of research proposals by human research ethics committees (HRECs) (according to updated national guidelines published by the National Health and Medical Research Council [NHMRC] in 19993), are two processes designed to prevent this. Assessment of the quality of design and execution of clinical research protocols is to some extent subjective, so these processes will never be perfect. His second issue relates to the definition of quality improvement/assurance and clinical audit, and whether such studies should be regarded as clinical research, and thus subject to prospective ethical review by an HREC. His own conclusion "that internal hospital quality assurance activities should undergo a similar level of scrutiny by ethics committees" is not consistent with the NHMRC guidelines,3 which, in the context of outlining the difficulties in defining research, state (on page 6) "such lists risk including activity that would not normally be included, like quality assurance activities or audits". Nevertheless, there are definitional uncertainties. The Australian Health Ethics Committee (AHEC) of the NHMRC has recognised that clinicians, HRECs and their hospitals need clear advice on how quality assurance and audit activities, which may not need ethical review, are to be separated from clinical research, which does need review. We are not the only country considering this matter.4 AHEC has established a working party to prepare such advice. The working party includes members drawn from AHEC, HRECs, consumer groups, medical colleges and health administrators. Draft advice will be subject to wide stakeholder consultation. The working party commenced its task in November 2001 and its final report is expected by mid-2002.

Simon R Tomlinson · Kerry J Breen · Malcolm H Parker MB BS, MLitt · Chris B Del Mar MD, FRACGP, FAFPHM · Paul P Glasziou MB BS, PhD · Lucie Rychetnik MPH, PhD · Stephen R Leeder FRACP, FFPHM, FAFPHM

To the Editor: We consider that Leeder and Rychetnik make several mistakes in their exploration of the relationships between ethics and evidence-based medicine (EBM).1 We share some of their ethical concerns about the determinants of the research agenda — lack of consumer input, emphasis on the benefits of interventions rather than harms, and funding structures favouring commercially promising interventions or biased by the status of the methodology to be used. However, these are criticisms that relate to producing new research, not using available research. The definition of EBM used by Leeder and Rychetnik2 values evidence that is non-quantitative, and explicitly demands the inclusion of patient preferences in clinical decision-making. Evidence about effects comes from research, while evidence about concerns and values comes from individual patients. To incorporate a patient's pre-ferences in the consultation is crucial to ethical practice, but quite independent of any particular hierarchy of evidence. Similarly, in claiming that treatment may be denied those of low social utility if patient autonomy is not valued, they mistakenly confuse preference or value with the quality of the evidence. The sin of old-fashioned paternalism is falsely attributed to EBM. The suggestion that EBM can exclude the importance of patient narrative is also at odds with the authors' chosen definition, which emphasises the "identification and compassionate use of patients' predicaments, rights, and preferences". Next, they worry that EBM might be misused in public health policy by neglecting areas where evidence is difficult to obtain, offering mental illness as an example. In fact, mental health attracts considerable attention3 and funding as one of the Commonwealth's current priority health areas.4 It has also been an area of considerable activity in EBM, with the Cochrane Mental Health groups and the BMJ's evidence-based summary journal Evidence-Based Mental Health. It may be that EBM has done the opposite of their prediction by highlighting an area of "evidence need". We agree that patients require support when they confront ambiguity and uncertainty. Nevertheless, doctors are ethically and legally obliged5 to provide full disclosure. Patients should be given correct information — warts, uncertainty and all — as often as possible. To suggest that the time spent seeking evidence threatens other elements of clinical practice is misleading. Clinical practice requires judgement to balance all its competing demands. We suggest that, by increasing the efficiency of continuing education, EBM should actually release more time for other requirements. Leeder and Rychetnik also misinterpret the relation between EBM and the law. They suggest that some practitioners who consult evidence, but practise against published guidelines, may be compromised. The point of EBM is to find the best available evidence. If that is clinical experience, consensus or narrative, rather than quantitative data, then so be it. Similarly, it is wrong to imply that those who practise EBM may be sued because they fail "to try everything". EBM, which is simply the getting of the best available information, changes nothing in the formal relationship between clinical evidence and the legal standard of care. Moreover, we believe that, by fostering patient involvement in decision-making, EBM should help protect clinicians from medicolegal dispute. We are always at risk of using new tools overzealously. But the champions of EBM temper its promotion with words like "judgement", "incorporating patient preference" and "conscientiousness". EBM involves a sensible and systematic search for the best information to include in the decision-making process. Great care should be exercised in issuing warnings about how it might be misused, in case EBM becomes unfairly caricatured, which may reduce the motivation of health professionals to find and apply the best treatment.

Simon R Tomlinson

In reply: Much of the literature about evidence-based medicine (EBM) has focused on the science of generating evidence, or the technical process of critically appraising and interpreting evidence for individual patients. In our article1 we opted to discuss EBM as a social activity with inherent potential for multiple manifestations. To describe EBM as a social activity is to emphasise how its definition, interpretation and application, and the ethical implications of those factors, are dependent on societal values and priorities — be they explicit or implicit. Debate about what EBM means, or should mean, in the context of Australian policy and practice does not degrade or negate the clearly ethical practice of consulting the best available research when making clinical or policy decisions. We challenge the view of Parker et al that by identifying and discussing how the concepts or language associated with EBM could be misused or misappropriated we will somehow reduce the motivation of health professionals to find and apply the best treatments. Indeed, our proposition is quite the opposite. Few of today's readers of the MJA will be unfamiliar with the benefits of systematic reviews of the best available research in their clinical area, and few are likely to be dissuaded of that view by our article. Our exploration of the relationship between ethics and EBM does not "misinterpret" EBM, but, rather, purposefully describes scenarios or social consequences about which there may be ethical concerns. If we can articulate clearly what we do not want EBM to mean, and describe the processes and consequences that we would consider unethical, it can only strengthen the development of an ethical and acceptable notion of what we do want from evidence-based policy and practice.

Simon R Tomlinson

To the Editor: Chronic hepatitis C virus (HCV) infection affects almost 200 000 Australians.1 It is monitored clinically by serial liver function tests (LFTs) and HCV RNA detection by polymerase chain reaction (PCR). HCV RNA is a marker of chronic infection and levels reflect response to antiviral therapy. However, testing for the presence of HCV RNA is expensive and, under the current Medicare Benefits Schedule, is not available to people with HCV antibodies and abnormal LFTs unless they are undergoing antiviral therapy. Using an in-house PCR assay, it has been shown that abnormal LFTs largely predict the presence of HCV RNA.2 We aimed to confirm this finding using a more reproducible PCR assay (Roche Amplicor HCV test) and to further investigate the relationship between LFTs and HCV RNA. We studied 323 HCV antibody-positive patients seen at the Fairfield Infectious Diseases Hospital, Melbourne, between May 1995 and September 1996. The Victorian Infectious Diseases Reference Laboratory performed all PCR assays and LFTs on these patients. Approval for the use of de-identified data was obtained from the Ethics Committee of the Royal Melbourne Hospital Research Foundation. Normal serum alanine aminotransferase (ALT) levels from at least two tests over a period of at least six months were considered to demonstrate normal liver function. In order to determine improved predictors of the presence of HCV RNA, the proportion of patients who were HCV RNA-positive and had an initially normal ALT level and the proportion with a normal ALT level persisting over six months were examined for each 10-IU/mL subdivision within the normal ALT range (0–50 IU/mL). Of the 323 patients, 88% were aged between 20 and 49 years and 68% were men. At initial testing, 251 (78%) were HCV RNA-positive by PCR, 206 (64%) had an abnormal ALT result and 183 (57%) had both a positive PCR result and an abnormal ALT level. An abnormal ALT level predicted the detection of HCV RNA in 89% (183/206) of patients and in 82% (14/17) if an abnormal ALT result was found within six months of an initial normal result. Of the 117 patients with a normal initial ALT level, only 49 (42%) had a negative PCR result. However, an initial ALT level of ≤ 20 IU/mL was more likely to be associated with a negative PCR result than an initial normal ALT level > 20 IU/mL (78% v 23%, respectively; P < 0.001). The probability of a negative PCR result was highest if the initial ALT level was ≤ 20 IU/mL and remained normal for at least six months (see Box). We concluded that, while an abnormal ALT level in a patient with HCV antibody generally predicted the presence of HCV RNA, the absence of HCV RNA was best predicted by an initially low ALT level that remained within the normal range for at least six months. Proportion of patients with normal initial serum ALT level and persistently normal ALT level who were positive for HCV RNA by PCR, divided into 10-IU/mL subdivisions of the normal ALT range Initial ALT range (IU/mL) Number (%) PCR positive, all patients Number (%) PCR positive, patients with persistently normal ALT over six months ≤ 10 2/11 (18%) 2/3 (67%) 11–20 7/29 (24%) 0/11 (0) 21–30 18/30 (60%) 11/12 (92%) 31–40 23/27 (85%) 8/8 (100%) 41–50 18/20 (90%) 2/2 (100%) ALT = Alanine aminotransferase. HCV = Hepatitis C virus. PCR = Polymerase chain reaction.

Heath A Kelly · William J Maskill · William Sievert · D Scott Bowden

To the Editor: The letter "Prejudice against mental illness" in the 20 August issue of the Journal1 resonates with my own experiences when attempting to obtain income protection insurance. Many colleagues have a similar story. In 1994, I suffered an episode of major depression requiring hospitalisation. I was a first-year resident medical officer, had relocated to a different city, and had just moved out of home for the first time, at the age of 25. Subsequently, I have progressed well in my career, and will soon complete specialist training. I have dealt successfully with a number of substantial personal and career hurdles, including postgraduate examinations and training, difficult property transactions, engagement and subsequent break-up of the engagement, the death of a flatmate, and illness and personal difficulties within my family and close friends. I have invested considerable time and effort in cognitive–behavioural therapy, and changed many attitudes and behaviours contributing to the initial breakdown. Wary of the implications for insurance, I have been reluctant to have any contact with my psychiatrist. My last appointment was a courtesy visit two years ago, with another visit 12 months before that. I have continued to self-prescribe an antidepressant tablet in the belief that it is probably doing more good than harm. Without any reference to medical reports, an examination, or an appraisal of my achievements and performance, several experienced brokers have confirmed that my history prevents me from being able to obtain income protection insurance, even with an exclusion clause for mental health problems. The one insurer that will consider covering people with a history of depression, with such an exclusion clause, requires me to be off any treatment for 12 months. It is understandable that an income protection insurer would refuse cover for depression-related claims. However, I fail to see why I am denied the opportunity to purchase insurance against the myriad other accidents and illnesses that can befall anyone, irrespective of their history. It appears the underwriters' methods are based on actuarial data that are overly generalised, and undoubtedly many years out of date in terms of diagnosis and prognosis of mental illness. Ironically, private health insurers cannot refuse to cover any patient, regardless of pre-existing conditions. Such discrimination reflects endemic ignorance and prejudice about mental health and illness. It is time for a review of the criteria by which insurers may reject applications.

Name and address withheld

To the Editor: The recent creation of the National Institute for Clinical Studies (NICS) is an exciting new opportunity for bridging the gap between evidence and practice.1 In carrying out this task, NICS will be directed by the members of its Board. Balanced stakeholder representation on the Board is required for NICS to produce optimal results. At present the Board consists of nine members, of whom eight are medical practitioners. The importance of doctors in the process and implementation of quality improvement initiatives is indisputable. However, other healthcare professionals also play a central role in achieving quality health outcomes for patients.2 Board membership more representative of its stakeholders would provide NICS with a broader range of perspectives, which could only be seen as beneficial. Given the current debate surrounding ethics and evidence-based healthcare, the values and expectations of healthcare consumers also need to be taken into account.3 One of the definitions of quality in healthcare is "consistently meeting or exceeding informed customers' opinion".4 It is crucial that the consumer's voice be heard in matters relating to healthcare research and in the implementation of quality initiatives. As the relevance and acceptability of quality initiatives undertaken by NICS will have an impact on health outcomes for consumers, it is important that such initiatives take into account the preferences of consumers. For this reason, we believe it is imperative that NICS include a consumer on its Board. An example of successful integration of a wide range of stakeholders onto a board is the Federal Government-funded National Health Priority Action Council, with representation from State/Territory, Indigenous and consumer groups and a balanced gender mix. We hope that NICS has strategies in place to enhance stakeholder representation on its Board, as this may be a factor in determining whether or not NICS becomes another forgettable acronym.

Louise V Hall

In reply: The Board of the National Institute of Clinical Studies (NICS) agrees strongly with Hall and Lauder that closing the gap between evidence and practice involves input from consumers. We also agree that the Board of Directors should seek to incorporate input from consumers in its strategic and operational activities. Of equal concern to the Board is ensuring the input of other stakeholder groups also currently not reflected in the composition of Board membership. For example, nursing and allied health professions comprise about 80% of the healthcare workforce and have shown strong leadership in relation to evidence-based practice. We are keen to see such groups actively involved in all aspects of the Institute's work. As a Federal Government-owned company, the selection and appointment process for Board members is the responsibility of government and our constitution does not allow the Board to change its own membership. However, the Board is seeking input from both consumers and other key stakeholder groups, both through its initial consultation processes and through establishment of Board advisory groups specifically focused on consumer issues and nursing and allied health. These groups will provide direct and valued input into the strategic and operational activities of the NICS. Our first round of consultation, with over 300 organisations, highlighted a number of areas where there are currently major gaps between evidence and practice, such as cardiac failure, various forms of cancer treatment, prevention of deep vein thrombosis in hospitalised patients, prevention of bedsores, and prescribing of psychotropic drugs for children. We are now examining ways in which the NICS might usefully help in some of these areas to identify barriers and possible solutions that can be rolled out across the healthcare system and sustained. The success of the NICS in achieving this will depend on the willingness of all stakeholders (including health professionals, consumers and managers) to work together in a constructive way.

Louise V Hall BPhty · Allison E Lauder BSc(Nut), MND · Chris A Silagy

To the Editor: I am concerned that several statements in the article on cyclooxygenase-2 (COX-2) inhibition by Cleland and colleagues1 do not accurately reflect the clinical data. The authors postulate a prothrombotic tendency of celecoxib on the basis of the CLASS study (comparing celecoxib with ibuprofen or diclofenac)2 and four case reports. The authors concede that celecoxib has no effect on the rate of myocardial infarction (MI) in the CLASS study (a conclusion also reached by the United States Food and Drug Administration [FDA] review of CLASS3), which would seem to contradict their hypothesis that celecoxib is prothrombotic. Cleland and colleagues speculate that the differences between the CLASS study and the VIGOR study (which compared rofecoxib with naproxen)4 may be explained by low-dose aspirin use in CLASS and failure to use aspirin in 4% of patients in VIGOR with "CV [cardiovacular] risk factors". This speculation is unfounded. In the CLASS study patients in all treatment groups who used aspirin had higher MI rates than non-aspirin users, and presumably this higher rate would have been observed in VIGOR if aspirin users had been enrolled. This higher rate is probably because aspirin use serves as a marker for increased CV risk. In patients in CLASS similar to the 4% with "CV risk factors" in VIGOR, MI rates were similar in the celecoxib and non-steroidal anti-inflammatory drug (NSAID) groups (data on file, Pharmacia) and numerically much lower than in the VIGOR study subgroup. On the basis of these two flawed arguments, Cleland and colleagues apparently extrapolate the high rate of MI seen with the use of rofecoxib to celecoxib and suggest that high MI rates are a "class" effect. This proposal is scientifically unsound and is not supported by other clinical data, including over 12 000 patients in the celecoxib registration program (data on file, Pharmacia). No celecoxib study has shown an increased risk of MI compared with traditional NSAIDs. The authors correctly assert there is "little clinical evidence from community use to suggest that selective COX-2 inhibition has serious unwanted effects other than those seen with standard NSAIDs", but imply there are few community data. In fact, community use of celecoxib in Australia (at least 1.5 million patients exposed) and worldwide (more than 20 million) has been extensive, and with this degree of exposure one would expect significant adverse event patterns to emerge. Reference to the Adverse Drug Reactions Advisory Committee and FDA database does not indicate a prothrombotic tendency of celecoxib. Further, we at Pharmacia do not consider that the four case studies presented by Cleland et al provide strong support for a prothrombotic tendency for celecoxib, especially as all patients described had diseases with high risk for thrombosis. On the basis of a large body of controlled trial data (including CLASS) and extensive community exposure, the evidence does not show any more thrombosis with celecoxib than with NSAIDs. Results of the CLASS and VIGOR studies clearly differ. It is clinically unjustified and scientifically unsound to suggest that rates of MI seen with rofecoxib can be ascribed to celecoxib and described as a "class effect".

Christopher G Fenn

In reply: The response from the Medical Director of Pharmacia to our article highlights some problems for all clinicians and independent scientists seeking to evaluate the balance of risks and benefits of pharmaceuticals and to validate the marketing messages of pharmaceutical companies. On the one hand, we lack the time and statistical resources to trawl through all data related to all trials with a test drug. On the other, our efforts to evaluate data are confounded by the publication and reporting biases associated with company-sponsored studies. In this regard, it is notable that the definitive results of CLASS1 have not been published, although the Food and Drug Administration (FDA) review of the data is available through an FDA website,2 as indicated by Fenn. While this document places data in the public domain, its location is neither within the pathway of MEDLINE search engines, nor is it known to the general body of clinicians. As reported in the FDA presentation, CLASS was a very large, double-blind safety study of at least six months' treatment that failed to achieve its primary endpoint of reduced complicated upper gastrointestinal events with celecoxib relative to the comparator, non-steroidal anti-inflammatory drugs (NSAIDs). While an interim analysis at six months was published, with extrapolation of event rates to 12 months,3 failure to publish the final results has withheld important results from wider scrutiny. In essence, the FDA document shows no overall long-term safety advantage of celecoxib over standard NSAIDs.2 The FDA analysis4 of the VIGOR study5 also shows no overall safety advantage for rofecoxib compared with NSAID, with fewer complicated upper gastrointestinal events being offset by a highly statistically significant (P = 0.0016) increase in serious thrombotic cardiovascular events. Collectively, these FDA analyses invalidate the promotion of selective cyclooxygenase-2 (COX-2) inhibitors as a safe alternative to NSAIDs, notwithstanding encouraging results from short-term trials. Further, although an increase in serious cardiovascular events was not seen in the CLASS study, its design was not optimal for detecting increased cardiovascular risk, and it is unlikely that CLASS was sufficiently powered to detect the degree of increased risk seen with rofecoxib in VIGOR. As explained in our article,6 unbalanced prothrombotic eicosanoid production associated with selective COX-2 inhibition (ie, a class effect) appears the most likely explanation for the increased cardiovascular events seen in VIGOR. Finally, we wish to reassert that, for effective postmarketing surveillance, it is essential that prescribers be adequately informed about safety concerns associated with new drugs, particularly when they involve events that are common and not usually seen as unwanted drug effects.

Leslie G Cleland · Michael J James

To the Editor: Chronic hepatitis C (HCV) infection affects more than 200 000 Australians.1 As the degree of hepatic fibrosis is the best predictor of morbidity, liver biopsy has a central role in management. Biopsy is also carried out to exclude additional pathology. However, because liver biopsy carries real risks and is expensive,2,3 debate exists as to whether liver biopsy should be performed routinely.3,4 Despite controversy surrounding the need to treat patients with minor histological changes,4 our impression is that many informed patients request treatment irrespective of liver histology. In Australia, liver biopsy is a prerequisite for antiviral therapy under the Pharmaceutical Benefits Scheme Highly Specialised Drugs Program (Box).5 To assess the impact of liver biopsy on management, we performed a retrospective study of patients with chronic HCV infection who underwent liver biopsy from March 1998 to December 2000. We identified 76 patients (51 men, 25 women), with a mean age of 29 years (range, 20–52 years). The biopsy was performed to stage and grade hepatitis C in all patients, and additionally to investigate a second pathology in seven patients. No alternative diagnoses were raised. Additional diagnoses (all suspected before biopsy) were confirmed in three patients and refuted in four patients. Biopsy findings were all consistent with chronic HCV infection, with some degree of fibrosis in 69 patients. There were five patients with histologically confirmed cirrhosis (including incomplete cirrhosis in three), and this was clinically evident in two patients. When S100 criteria at the time of biopsy were applied, after exclusions on clinical grounds, only one patient would have been ineligible for interferon monotherapy based on liver histology. Under current S100 criteria, nine patients would be ineligible for combination therapy, but all nine would remain eligible for monotherapy. Of our patients who attended follow-up and were HCV RNA positive, 62 of 64 patients received or are awaiting therapy. Our results confirm the finding that liver biopsy in patients with chronic HCV infection rarely identifies alternative diagnoses.3 These data reflect the fact that most chronic liver diseases can be diagnosed before biopsy. Biopsy remains an important tool for histological diagnosis of cirrhosis. However, as unexpected findings are uncommon and some patients will be treated irrespective of liver histology, there is an emerging argument not to perform liver biopsy routinely. This argument will strengthen if valid biochemical markers of fibrosis are confirmed.6 We believe each patient should be assessed individually, and treatment could be offered without biopsy to patients who: meet all criteria for treatment under current guidelines other than known liver histology; have no alternative or additional diagnoses after thorough work-up; strongly desire treatment regardless of histology, and there is sound indication for treatment (eg, extrahepatic symptoms, concerns of vertical or occupational transmission); have a high chance of sustained viral response (eg, favourable genotype); have no clinical, biochemical or haematological suggestion of cirrhosis; and with the physician, accept the implications of treatment without biopsy. An alternative strategy could be to consider a biopsy in patients who do not have a sustained virological response, to allow prognostication. Clearly, such changes would greatly affect biopsy practices in Australia. Section 1005 criteria for use of combination therapy with interferon alfa 2b and ribavirin Patients with chronic hepatitis C who satisfy the following criteria are eligible for interferon alfa 2b and ribavirin: On liver biopsy, are staged as METAVIR stage 2 or greater, or METAVIR stage 1 with grade A2 or A3 inflammation (except patients with coagulation disorders); Have abnormal alanine aminotransferase levels in conjunction with demonstration of viral infection (HCV RNA positive);* Do not have other liver disease;* Are not pregnant, not lactating, and are using two reliable methods of contraception;* Have no history of significant psychiatric illness; Would be likely to attend regularly for treatment and follow-up; and Take no more than seven standard alcoholic drinks a week. Genotype of virus should be assessed before treatment Treatment course is 24 weeks except: With Genotype 1 and patients with cirrhosis or bridging fibrosis (regardless of genotype), where treatment course is 48 weeks; and Treatment will be continued for 48 weeks only if HCV RNA qualitative assay is negative at 24 weeks. OR Patients who have relapsed after treatment with interferon 2a/2b monotherapy supplied as a Section 100 medication. This course is limited to 24 weeks. Section 1005 criteria for use of monotherapy with interferon alfa 2b Patients with chronic hepatitis C confirmed on liver biopsy (except patients with coagulation disorders) are eligible for interferon alfa 2b monotherapy if they satisfy the criteria marked with asterisks above. (When monotherapy fails, patients become eligible for combination therapy.) Treatment is to cease if plasma HCV RNA remains detectable by HCV RNA qualitative assay after 12 weeks of treatment. The course must be continuous and excludes retreatment of non-responders or patients who relapse.

Ian F Yusoff · Lindsay Mollison · Leanne Totten · John Olynyk

To the Editor: With the recognition that supplements of folate given early in pregnancy can reduce the incidence of neural tube defects,1,2 Australian manufacturers were allowed voluntary food fortification with folate from 1995, and these foods subsequently became available from August 1996. We assessed the impact of this fortification by comparing the results of assays of serum folate, a sensitive index of folate intake, before and after the introduction of folate-fortified foods. Data were available for serum folate samples assayed by the chemiluminescence method (Chiron Healthcare Pty Ltd, Scoresby, Victoria) at our laboratory in Melbourne. Quality assurance data indicated no analytical drift, and external proficiency testing yielded satisfactory results throughout the period under study. A total of 20 506 samples from women aged 14–45 years, the target group for supplementation, and 5528 samples from men of the same age group were assayed during 1993–2000. The results for the years 1993–1996, before fortification, where sample numbers were relatively small, were pooled. The results were analysed by Bhattacharya plot, eliminating the effect of outlier values,3 mindful of the limitations of extrapolating data derived from clinical material to the community. The mean values are shown in the Box. In both groups there was a small incremental rise in mean serum folate concentrations, and a fall in the prevalence of low values, after the introduction of fortification. The mean value of 14.0 nmol/L for women in 1993–1996 increased by about 19% to 16.7 nmol/L in 2000. The percentage of low values decreased from 8.5% to 4.1% over that period. Although it is possible that the increase in serum folate concentration reflects dietary education and use of folate supplements in women of reproductive age rather than food fortification specifically, parallel changes were also observed in men. In the United States, folate fortification of all enriched cereal grain products was mandated from January 1998. This led to a dramatic increase of 250% in mean folate levels in women aged 15–44 years,4 and an increase of 50% in the median values (uncorrected for outliers) for men, women and children of all ages submitted for clinical evaluation.5 By comparison, the increase in folate levels in Australia has been very small. We conclude that to obtain a significant increase in folate intake in the community by food fortification, a policy of mandatory rather than voluntary fortification is required. Serum folate concentrations in Victorian women and men aged 15–45 years before (1993–1996) and after (1997–2000) the introduction of voluntary food folate fortification Year 1993 to 1996 1997 1998 1999 2000 Women Number 3865 2989 4168 4385 5099 Mean folate concentration (nmol/L)* 14.0 14.5 15.3 16.4 16.7 95% confidence limits 6.7–28.3 5.0–38.6 5.7–38.1 5.7–41.0 5.6–45.5 % Low values† 8.5 7.1 5.7 4.3 4.1 Men Number 1077 849 1130 1117 1355 Mean folate concentration (nmol/L)* 14.0 14.9 15.7 16.5 16.2 95% confidence limits 6.4–28.7 4.7–31.0 5.5–35.6 6.2–41.6 5.9–42.1 % Low values† 7.9 8.1 6.5 4.1 5.1 * Derived from log normal distribution of community values by Bhattacharya method. † 8.0 nmol/L.

Jack Metz · Ken A Sikaris · Ellen L Maxwell · Mark D Levin

To the Editor: In late 1998, a clinical audit in the Thoracic Division of the Prince Charles Hospital found the autopsy rate was 7% of all patients who died in the Division (excluding Palliative Care) for the 12 months to September 1998. Following discussions and acknowledgement of the importance of hospital autopsy as a clinical audit tool, the Division's policy to consider an autopsy in all patients who died was reinforced. Registrars were educated in seeking approval and in counselling relatives. As a result of these interventions and ongoing audit, a decision in relation to autopsy is now recorded in more than 90% of charts following a patient's death, compared with 40% initially. The autopsy rate progressively increased, and, from March 2000 to January 2001, it was 35%, five times the baseline rate, and the refusal rate was 11% (Box). The rate of limited autopsies (generally only excluding the brain) increased from 20% to 50%. However, from early 2001, coinciding with the ongoing negative Australian press coverage related to aspects of autopsies, there has been a marked decrease in relatives' agreement to allow autopsy and extent of autopsy. The refusal rate for autopsy increased to 30% for the four months to May 2001, and was 25% to September 2001. The autopsy rate fell dramatically to 27% and 13% for the same periods. Nine of the 10 autopsies were limited, usually to a single organ or body cavity. Data from death certificates are vital for education, research and public health purposes.1,2 Autopsies remain the only way to audit the accuracy of death certificates. A review found that the rate of clinical diagnostic inaccuracy for major findings at autopsy is about a third, and this rate has not changed since 1912.3 This unavoidable baseline of diagnostic error4 does not necessarily indicate incompetence or malpractice. It is essential that the public understand that medicine is not an exact science, that we do misdiagnose conditions, and that identification of these "errors" is of value to relatives, to future patients and to society. Legislative changes are being proposed in Australia that will make obtaining consent for autopsies more complex and potentially distressing for relatives. Education of medical staff and the general public must accompany these changes if they are not to be the final "nail in the coffin" of the hospital autopsy and remove an important facet of continuing improvement of medical practice. Autopsy rates in the Thoracic Division of the Prince Charles Hospital

Helen E Ward · Belinda E Clarke · Paul V Zimmerman · Michael I Cleary

To the Editor: The National Health and Medical Research Council has launched new Australian alcohol guidelines1 to help reduce alcohol-caused deaths in Australia, estimated to have been 3290 in 1997.2 Male drinkers are advised to drink no more than an average of 40 g alcohol per day and females no more than an average of 20 g of alcohol per day to prevent chronic health problems (eg, alcoholic liver cirrhosis). Furthermore, it is recommended that, provided there are no other situational or individual risk factors (such as driving or being pregnant), men drink no more than 60 g on any day and women no more than 40 g to prevent acute conditions associated with bouts of intoxication (eg, alcohol-related injuries). Volumes of alcohol intake reported at each NHMRC risk level by 10 030 respondents to the 1998 National Drug Strategy Household (NDSH) survey were calculated.3 The data were weighted for age and sex. Volumes of alcohol reported to have been consumed at different risk levels were summed for all subjects and expressed as percentages of the total reported consumption of the sample. It was found that 39% of total consumption was categorised as being drunk by people who exceeded low-risk limits for chronic harm (36% for men, 45% for women). It was found that 51% of total consumption occurred on days when the drinker exceeded low-risk limits for acute harm (53% for men and 47% for women). Drinking that was risky for either acute or chronic harm was found to comprise 67% of total consumption (see Table). For young men aged 18–24 years this figure was 93% of all alcohol consumed. These estimates are conservative, as the level of drinking reported in the 1998 NDSH survey is consistent with an adult per capita consumption of only 46.5% of that estimated by the Australian Bureau of Statistics for that year on the basis of import, export and production data.3 The high proportion of all alcohol consumed that places drinkers at risk of serious harm helps explain why per capita alcohol consumption correlates so closely with levels of suicide, road death, homicide, liver cirrhosis and other causes of death.4 It follows that policies which reduce the total consumption of alcohol in Australia will reduce the associated health and economic costs by reducing levels of risky drinking. Further, policies which successfully reduce high-risk drinking will also reduce total population consumption of alcohol. Percentage of alcohol consumed at risk levels for acute and/or chronic harm, as specified in the new Australian alcohol guidelines,1 by age and sex (n = 10 030, weighted data) Age (years) Females (%) Males (%) Total (%) 14–17 71.3 77.4 75.1 18–24 82.3 92.9 89.9 25–39 68.1 66.4 66.9 40–64 67.7 62.0 63.6 65 + 46.5 39.5 41.4 All ages 68.4 66.5 67.0

Tim R Stockwell · Penny Heale · Tanya N Chikritzhs · Paul Dietze · Paul Catalano

To the Editor: From 1 February to 30 June 2001, 277 602 prescriptions for the smoking cessation drug bupropion hydrochloride (Zyban, GlaxoSmithKline) were processed. The Health Insurance Commission approved 343 737 prescriptions for bupropion between 1 February and 30 June.1 Comparing this figure with the 277 602 processed scripts, some 66 135 (19.2%) scripts went unfilled. One reason for this may have been extensive publicity given to reports of deaths and numerous adverse reactions following bupropion use. The website of the Australian Drug Reactions Advisory Committee (ADRAC) reports that, as at 22 June, there had been 18 reports of deaths in patients aged from 30 years to 69 years who were using or who had recently stopped using bupropion.2 ADRAC summarised intelligence on these deaths thus: ... there were a variety of reported causes of death and not a single consistent mode of death. In addition to being smokers, several patients had other existing risk factors for unexpected death such as alcohol abuse, diabetes or cardiomyopathy. Eleven of the 18 patients had an alternative explanation for death that was at least as plausible as a possible effect of bupropion. In four reports, the available information was very limited and it was not possible to assess the cause of death. Further information is being sought on three cases to aid assessment of the cause of death.2 Smokers are at 3.1 times greater risk of dying (from any cause) than non-smokers and twice as likely to die from coronary disease and stroke.3 People with depression are three times as likely to be daily smokers4 and have double the suicide rate of non-smokers.5 In Australia, sudden coronary fatalities occur at a rate of about 450 per million people aged under 65,6 perhaps at a rate of 355 per million in non-smokers and about double that in smokers. In three months (the period of recommended bupropion use), one would expect 180 deaths per million smoker-users. Thus, among 277 602 Australian smokers, 50 might die during any given three-month period without any added risk from bupropion. This estimate helps to place the 18 fatalities reported to ADRAC in context. The 277 602 scripts represent about 9.5% of Australia's 2.9 million regular smokers. These people, their families and doctors deserve to have their anxieties about the risks of using bupropion addressed. We would urge the government to commission urgently a case–control study of morbidity and mortality among smokers and their relationships to use or non-use of bupropion.

Simon C Chapman · Konrad Jamrozik

To the Editor: I read with interest the article by Hender and colleagues recently published in the Journal.1 I agree with the authors' conclusion that there is presently insufficient evidence to suggest the widespread use of endoluminal treatment for carotid artery disease. However, there are a number of problems with the authors' interpretation of our recent article comparing the outcome of surgical and endoluminal treatment of symptomatic carotid stenosis.2 Firstly, the figures in the Box are completely misleading. The percentages of adverse outcome quoted for patients undergoing endoluminal treatment are those that were found for cases receiving endarterectomy, while the figures quoted for endarterectomy are the findings for endoluminal treatment. [A correction of this error was published in the 3/17 December 2001 issue of the Journal, page 672.] Hence, any reader simply looking at the Box would be left with the false conclusion that the outcome of endoluminal treatment is superior. Secondly, the authors refer to our article2 as a "meta-analysis". In our article we went to some trouble to explain that a meta-analysis was not possible, as only one small randomised trial had been published at that time. Instead, we had to use reports from single centres and we discussed the difficulties of comparing the results when patients had not been randomised. Thirdly, the results of the CAVATAS trial were published in June 2001.3 A surprising finding was that the perioperative stroke rate (defined as a neurological deficit lasting seven days or more) for patients undergoing either carotid angioplasty (with or without stenting) or conventional endarterectomy was the same (around 10%). In fact, the disabling stroke rate of around 6% after either endovascular treatment or endarterectomy was three times higher than that found in the North American randomised trial of endarterectomy.4 Finally, the authors refer to our patients undergoing carotid stenting, whereas the majority of the patients referred to in fact received angioplasty alone.

Jonathan Golledge