Letters

To the Editor: Jamrozik's editorial1 about our report of a failed randomised controlled trial (RCT)2 in an Aboriginal medical service helps to explain why researchers might be reluctant to submit articles describing unsuccessful trials, thus limiting potential for the scientific community to learn from such experiences. The main point of our article was to describe the manifest difficulties of implementing an RCT — the evidence "gold standard" — in this type of setting. Interestingly, Jamrozik largely attributes these difficulties to incompetence or naivety (or both) on the part of the researchers and funders, rather than to complexities inherent in the study design, the setting and the intervention. A separately funded pilot study is, in principle, a good idea, but extremely difficult to get funding for in today's environment. Of course, we did conduct a pilot — that, in fact, was what we reported on — but it is unclear how this would have helped us better estimate absolute prevalences and effect sizes for intervention and control groups, as a substantial number of participants, followed up for six months, would have been needed to do this. Nor is it clear how taking a population approach and distributing guidelines to all drinkers rather than offering personalised advice to hazardous drinkers would have helped — firstly, because we were specifically trialling the internationally validated brief intervention, and secondly, because the effect size of the alternative approach would have been so small that we would have needed very much larger numbers to test its effectiveness. We had no intention of "stumbling down something like this path". Nor do we agree that the blood tests were "medicalising a social problem". They were intended not only to provide robust outcome measures (a mark of a good trial), but also tangible evidence to clients of the health effects of alcohol, shown from previous research to be well received by Aboriginal people.3,4 They were not a requirement for participation. Further, that we should have got around the potentially off-putting business of seeking informed consent by bypassing this step almost defies comment. While trials of some therapeutic interventions can be undertaken blind with patient consent by using placebos, this does not mean that where blinding is not possible patient consent should be done away with in order to avoid a Hawthorne effect! However, we do agree with Jamrozik on one point — nothing about this study or our report could reasonably "compound any negative perceptions about Aboriginal Medical Services and Aboriginal patients".1

Beverly M Sibthorpe · Ross S Bailie · Maggie A Brady · Sandra A Ball · Polly Sumner-Dodd · Wayne D Hall · Alan Pettigrew · Tom Gavranic

To the Editor: I am responding to a recent editorial by Jamrozik1 commenting on a study proposed by Sibthorpe and colleagues to assess a brief intervention for hazardous use of alcohol by Indigenous people in an urban setting.2 After two unsuccessful attempts to recruit participants, the study was discontinued and funds returned to the National Health and Medical Research Council (NHMRC) in 1998. Sibthorpe et al identified their difficulties as primarily the result of having overestimated the number of suitable participants, for a number of complex reasons. Jamrozik's criticisms rest disproportionately with the NHMRC and are based on procedures and processes in effect in 1996 and 1997, yet they are informed by contemporary knowledge and wisdom. This seems somewhat anomalous. In 2000, the NHMRC revised its system for assessing research applications. This involved several developments which would have had a direct impact on the assessment of this application had they been instituted in 1996. Some of these include: the introduction of panels comprising 11 experts in the domain of the application; the introduction of the Indigenous Health Research Panel (IHRP), which provides advice on cultural appropriateness, community consultation and methods in applications with an Indigenous component (most members are Indigenous people); and the opportunity for IHRP to make stipulations upon which funding is contingent. Also of significance was the establishment of the Research Agenda Working Group (RAWG), which oversaw the formulation of intervention-based criteria. Colloquially known as the "Darwin criteria", these principles ensure that all Indigenous research design has: sufficient Indigenous community consultation and participation; transferability (of the methods to other settings); and sustainability (of resulting changes). The NHMRC was disappointed that the study by Sibthorpe et al did not proceed and did not result in usable data to inform a significant problem. However, it is also important to recognise that unanticipated outcomes, which can often lead to other, very positive results, are an integral part of the learning process. The NHMRC has supported Australian health and medical research since 1936. It has a strong commitment to ensuring the continuing evolution of its procedures and practices. The new systems implemented in 2000 were designed to ensure the continuing tradition of funding high quality, relevant and applicable research.

Beverly M Sibthorpe BA(Hons), PhD · Ross S Bailie MD, FAFPHM · Maggie A Brady MA, PhD · Sandra A Ball BCom, GradDip Public Administration · Polly Sumner-Dodd DipManagement · Wayne D Hall BSc, PhD · Alan Pettigrew BSc, PhD · Tom Gavranic MB BS, DPH, FRACGP

To the Editor: The article by Bellomo et al,1 with its alarmist conclusions, received a lot of media attention. However, the authors' methodology is flawed and their conclusions are unsupported by their data. They describe postoperative adverse events in a group of largely elderly patients (median age, 65.5 years) who stayed in hospital more than 48 hours after inpatient surgery. These selective criteria were used "to exclude patients having day surgery or minor procedures". Stratifying the severity of operations according to duration of stay is fundamentally flawed. This would have excluded many major, short-stay operations if the patient had had an uneventful postoperative course (eg, laparoscopic cholecystectomy and complex endoscopic procedures), and included others simply because a complication prolonged the patient's stay. The result is a selective bias towards a high complication rate. A more valid approach would have been to stratify all inpatient operations by severity and to include all major operations in the denominator for the study. This strategy would undoubtedly have shown significantly lower complication and mortality rates than those reported by Bellomo et al. The "silent epidemic" referred to in the study is neither silent nor an epidemic. An epidemic refers to a disease normally absent but liable to outbreaks. What the authors describe is an endemic situation (habitually present, of common occurrence); it is quite obvious and already extensively documented. Elderly patients undergoing major operations (especially in an emergency — "unscheduled surgery" in the authors' pejorative lexicon) are likely to have complications, and, when they do, need to stay in hospital longer. The authors did not analyse whether the adverse events were preventable, and therefore they cannot justify their conclusion that "there is much scope for improving postoperative care".

Thomas B Hugh · G Douglas Tracy · Rinaldo Bellomo · Donna Goldsmith · Sarah Russell · Shigehiko Uchino

To the Editor: The recent article by Sibthorpe et al1 and the accompanying editorial2 on the issue of the failure of an alcohol intervention trial in an Aboriginal Health Service deal with problems facing all primary care practitioners in the field of "alcohol misuse" and should not be seen as a peculiarly Aboriginal problem. Firstly, despite what the academics may tell us, administering an Alcohol Use Disorders Identification Test (AUDIT) questionnaire in general practice as a screening measure meets with huge resistance, no matter where you practice. Denial of the disease-inducing potential of alcohol is certainly not peculiar to Aboriginal society. Secondly, I find that the bulk of the medical profession reinforces this community denial by diagnosing conditions such as diabetes, hypertension, obesity, anxiety, depression and schizophrenia instead of seeing these problems as being a manifestation of alcoholism or other "alcohol misuse" until proven otherwise. Indeed, the denial is so extreme that they tend to avoid the term "alcoholism" altogether. Specialists are in even greater denial and are more often a hindrance than a help to general practitioners in this regard. As a consequence, community leaders and affected families are unable to develop effective strategies for dealing with their problems. What they get instead is increasing healthcare costs, hospital bed shortages, increasing domestic violence, more "drug problems" and more prisons. So "GP reluctance or inability to follow through . . ."2 is not surprising. Indeed, denial of alcohol is so strong in the medical profession that it is harder, in my experience, to get doctors and even medical students (let alone healthcare workers) to attend open meetings of Alcholics Anonymous and Al-Anon than it is to persuade affected people to do so. Thirdly, general practice throughout Australia has been organised for episodic, fast-throughput care. People have become so accustomed to this that they see any attempt at a comprehensive preventive approach to illness as odd, out of place, time-consuming and even intrusive, especially so where alcohol and family histories are concerned. That Aboriginal people are no different from the rest of us in this regard should cause no surprise.

Beverly M Sibthorpe · Ross S Bailie · Maggie A Brady · Sandra A Ball · Polly Sumner-Dodd · Wayne D Hall

To the Editor: The information in the article by Bellomo et al,1 which documents postoperative serious adverse events in a teaching hospital, contains no surprises; nor does it support some of the authors' conclusions. In 1995, the findings of the Quality in Australian Health Care Study2 were immediately sensationalised by the press with the headline "Hospital errors kill 18 000 a year".3 The article by Bellomo and colleagues provoked similar predictable media sensation. As acknowledged by the authors, the study addressed neither the causes of the serious adverse events, nor whether they were "preventable". Furthermore, the authors fail to show how their findings "suggest that there is much scope for improving perioperative care in our tertiary hospitals", or why "this is a 'silent' epidemic which requires urgent and systematic attention". However, in televised interviews, they made no effort to reduce the alarm aexpressed at the prevalence of errors. They have invented a new designation of "unscheduled surgery" (which presumably refers to acute, urgent or emergency admissions), preferring a title that suggests an avoidable lack of scheduling. It is hardly surprising that this group of patients required most of the admissions to the intensive care unit for which no prior booking had been made. It is unclear why the authors mention that "six of nine patients over 92 years of age having hip surgery died". Again, one presumes that these operations were for hip fracture, a condition with 100% mortality if untreated. And why leave out patients aged between 90 and 92 years? No amount of statistical manipulation conceals the bias that is obvious in their article. It might provide a media story, but it has minimal value for the critical reader.

Thomas B Hugh FRCS, FRACS · G Douglas Tracy · Rinaldo Bellomo MD, FRACP · Donna Goldsmith RN · Sarah Russell RN, PhD · Shigehiko Uchino MD

To the Editor: I read with interest the recent correspondence in the Journal from Del Mar and Glasziou.1 Their appeal to one of their critics was to "abandon throwing bricks from the sidelines and join us in trying to help clinicians assess research evidence in [a] timely fashion". More recently, their defence in relying on generalists, rather than experts, to assess the evidence was, somewhat curiously, that "a cat may look at a king".2 In response to these comments, I believe that one of their recent presentations relating to natural remedies for osteoporosis in postmenopausal women3 falls short of current evidence-based medicine requirements. I suspect that this may be because they have no expert in the area they are reviewing to assist them in assessing the data. An expert would have been able to tell them that their statement "Although no trials specifically compared 'just walking' with 'exercise in the gym', there was reasonable evidence supporting the beneficial role of walking in this patient group"3 was incorrect.4 Furthermore, an expert would have known that the fundamental biological basis of the effect of exercise on the skeleton relates to the induction of significant strains within the skeleton.5 Because these effects are site-specific and load-dependent, the physiological mechanisms differ from those involved in cardiovascular health. Secondly, an expert in the area may have pointed out that the principal constituent of codliver oil that affects the skeleton is vitamin D. To claim that there were no benefits on bone mineral density or fracture from vitamin D would not be supported by current data.6 Keeping up-to-date in medical practice has become more difficult as the information base expands. Performing an Internet search may be adequate for answering some patient-based specific questions, but, in general, reading an up-to-date review is probably better. Presenting the results of a rapidly performed, deficient analysis in an internationally renowned journal is another thing entirely.

Richard L Prince · Christopher B Del Mar · Paul P Glasziou

In reply: We thank Tracy and Hugh for the issues they raise. The goal of our study was to establish baseline information on the incidence of serious adverse events (SAEs) for use in subsequent intervention studies.1 The data were needed for statistical power calculations. Our inclusion criteria were predefined, as is scientifically orthodox for any study. We chose to study a population of clinical relevance to inpatient medicine. Most simple procedures at our hospital require day admission with no overnight inpatient stay, so these were not relevant to our goals. Others may wish to study different patient populations and are free to do so. In our opinion, there was no particular bias in our study, just accurate, prospective documentation of events. We used the term "unscheduled surgery" because it is verifiable and objective. An operational definition is necessary; otherwise, judgements about what is a true emergency (like judgements about what is preventable) are very dependent on observer bias. Nonetheless, according to our judgement, only 48 of 426 "unscheduled" operations were true emergencies. We wanted to identify groups that were at particular risk of death, hence the mention of patients over 92 years of age who had had hip surgery. Up to what level of expected postoperative mortality does it remain acceptable to perform major surgery in very elderly patients? We stand by our opinion that we are dealing with a silent epidemic. It is silent because we could find no previous prospective studies of SAEs for all major operations published (in English) in the medical literature, and there was no systematic plan to tackle them. We use the term epidemic because (in the absence of objective documentation of rates of SAEs in the past) our impression is that this is a growing phenomenon, related to the increased use of major surgery in the elderly. We also consider that only the absence of SAEs would offer no scope for improvement. A rate of SAEs of 16.9% should, logically, offer much scope for improvement. Whether such improvement can be realised remains a matter for future interventional investigations.

Thomas B Hugh

In reply: We agree with Prince that expertise is needed. The only question is, expertise in what? If we rely on experts in the content area, we are subject to error from ignoring studies that do not fit the expert's view or from an overemphasis on studies familiar to the expert.1 On the other hand, if we rely on experts in systematic reviews, we are likely make other errors, especially errors of omission, because we do not know the content area so well. Perhaps what we need is a marrying of the two. First, experts should learn the business of evidence-based medicine (EBM). It should be part of the training of specialists and general practitioners, as well as medical students. In addition, the Australasian Cochrane Centre runs Australia-wide courses in preparing and interpreting systematic reviews.2 Second, we should find a content expert to assist any review being undertaken (and, indeed, we try to do this when writing systematic Cochrane reviews). But when we need to respond rapidly to clinicians' and our own questions, there is not time to consult experts for each one. (The doctor in question3 asked five rather separate questions, of which Prince addresses only two.) Instead, we try to do exactly what Prince has suggested: we first look for an up-to-date review (a systematic review, to avoid bias),1 and, in the absence of one of those, proceed with weaker levels of evidence progressively down a cascade. And that is exactly what we did in this case.3 The purpose of our rapid search service is to provide, as quickly as possible, the best available evidence to assist a doctor help a patient — "EBM in action". The five meta-analyses we identified3 did not include your references. We do not, and do not claim to, offer the last word in a very complicated area of clinical practice.

Richard L Prince MB BS FRACP MD · Christopher B Del Mar MD FRACGP FAFPHM · Paul P Glasziou MB BS PhD

Buzz off! With summer fast approaching and the spectre of evening barbecues dogged by mosquitoes, it is interesting to note that a recent US study confirms that personal repellents containing N,N-diethyl-3-methylbenzamide (DEET) are still the best available in that country. Fifteen volunteers tested 16 products three times each (a total of 720 tests) using the “arm-in-cage” technique. DEET-based products provided complete protection for 88.4 to 301.5 minutes, depending on their concentration. Citronella-based products protected for 2.8 minutes (0.5%) to 18.9 minutes (10%, combined with other botanical oils) and a 2% soybean oil product (marketed for children) protected for 94.6 minutes. Wrist bands impregnated with either DEET or citronella were completely ineffective. For those of you who swear by a popular bath oil, this failed the test after 9.6 minutes. Interestingly, a product containing eucalyptus oil came onto the US market just as the study finished. Testing on six subjects ended in a skin reaction in one and a mean complete protection time of 120 minutes in the other five. N Engl J Med 2002; 347: 13-18 ... A thousand words Recent articles in the MJA highlighting a lack of psychosocial outcome data for cosmetic surgery did not include children, but a report from a US plastic surgeon suggests that we can learn much about children’s response to such surgery from their art. The surgeon collected 200 drawings from children, before and after they underwent surgery for a variety of congenital, aesthetic and traumatic deformities. A child psychiatrist and an art therapist evaluated the drawings, finding recurring themes of low self-esteem, isolation, unhappiness and fear in the “before” drawings, with improvements in all these parameters in the “after” drawings. Twelve examples are reproduced in the printed article, which is well worth a look. Plast Reconstr Surg 2002; 109: 1777-1786 Making the news In Other Journals image At the MJA we often wonder why some of our articles capture the imagination of the popular press, while others are met with a deafening silence. A study of press releases by medical journals and medical reporting in two British newspapers suggests that bias operates at both levels. Of 1193 original research articles published in the Lancet and BMJ in 1999 and 2000, 517 received a press release and 81 were reported in either the Times or the Sun newspapers. No studies without press releases made it into the papers, but newspapers were more likely to follow-up press releases reporting “bad” than “good” news, and preferred observational studies to randomised controlled trials. The hot topics were women’s health, reproduction and cancer. BMJ 2002; 325: 81-84 Don’t worry, be happy! Expectant mothers have yet another thing to worry about, as new research reveals that anxiety in pregnancy is associated with behavioural and emotional problems in their offspring. The Avon Longitudinal Study of Parents and Children followed 7448 women in Avon, UK, from early pregnancy until their children were four years old. Maternal anxiety and depression were measured via validated self-report instruments at 18 and 32 weeks’ gestation, and four times in the three years following birth. The children’s behavioural adjustment at age four was estimated by a validated parental report measure. High antenatal maternal anxiety scores in late pregnancy were associated with behavioural and emotional problems in both boys and girls (ORs 1.56 and 1.51, respectively, after controlling for antenatal, obstetric and psychosocial factors, and postnatal depression and anxiety). Br J Psychiatry 2002; 180: 502-508 A burning issue With the annual incidence of melanoma currently in excess of 45/100 000 for men and 35/100 000 for women, Australia remains the melanoma mecca of the world. Public health messages about sun exposure are starting to show benefit, however, with a recent slight fall in incidence in younger women. The same is not true for Scotland, a country better known for its bleak weather than its opportunities for sunbathing. The Scottish Melanoma Group collated data on the 8830 patients diagnosed with invasive melanoma in Scotland between 1979 and 1998. The incidence increased steadily during the study period (from 3.5 to 10.6/100 000 in men, and 7.0 to 13.1/100 000 in women). Five-year survival rates improved, however, due to a higher proportion of thinner detected tumours in later years, suggesting that the Scots are at least responding to secondary prevention messages. Lancet Volume 360 Issue 9328 Page 23 Published online June 25 2002-07-19

To the Editor: Thalidomide (N-α-phthalimidoglutarimide) was first marketed as a sedative-hypnotic in 1957. It was withdrawn from the market in 1961 as it was found to cause congenital malformations.1 Infant mortality statistics in Germany for the years 1959 to 1963 show that about 40% of thalidomide-affected babies died in the neonatal period.2 The main causes of death were atresia of the bowel, renal dysgenesis and heart malformations. As a result of extensive studies on the pathogenesis of the malformations, it was found that thalidomide is an immunosuppressant.3 The use of two different 14C-labelled thalidomide preparations showed that a portion or the whole of the glutarimide molecule binds to the DNA of rabbit embryos.4 In Britain and Ireland, 480 thalidomide-affected infants survived. Of these, 25 died before reaching the age of 40 years. The causes of death were cancer (4), heart disease (4), diabetes (3), hypertension and renal failure (3), motor accidents (3), and substance misuse or suicide (8) (M Johnson, Director, the Thalidomide Trust [United Kingdom], personal communication). Four deaths from cancer before the age of 40 years in a cohort of 480 is an incidence of 0.83%. The death rate from cancer in England and Wales before the age of 40 is 9.4 per 100 000 population, or 0.0084%.5 Thus, the thalidomide-affected individuals had a 99-fold increase in the age-related cancer death rate. Another of the cohort died aged 41 years of round-cell sarcoma. The high incidence of malignancy, together with the knowledge that a portion of the thalidomide molecule binds with the DNA of laboratory animals, suggest a possible mutational change in some of the cells of thalidomide-affected people. Thalidomide is currently being used to treat a variety of diseases, some because of its immunosuppressant properties. These diseases include graft-versus-host disease, leprosy, AIDS, Behçet's syndrome, tuberculosis, multiple myeloma and many dermatoses. It is also being used for treating some cancers. Its chemotherapeutic value probably results from the ability of the glutarimide component of the thalidomide molecule to bind with the DNA of rapidly dividing cells. However, if the genetic injury is not accurately repaired, it may result in mutations or even cell death. Although thalidomide is now proving to be a useful therapeutic agent, its ability to bind with DNA makes it dangerous, not only when taken by pregnant women, but also potentially when taken by men, whose sperm might be affected.6

William McBride

To the Editor: Bupropion hydrochloride was listed on the Pharmaceutical Benefits Scheme (PBS) on 1 February 2001 for use as short-term adjunctive therapy for high nicotine dependence, with the goal of maintaining abstinence. Supply is limited to one application per year, with no repeats, and a maximum quantity of 120 tablets at a dispensed cost of $238.95. From the beginning of February to the end of December 2001, 351 772 bupropion prescriptions had been processed by the Health Insurance Commission (HIC) at a cost of $83.14 million.1 This is equivalent to about 2% of total PBS-related drug expenditure in Australia.2 Given this high cost, it is reasonable to consider the extent to which this investment represents value for money. An important first step is to assess the penetration of bupropion into the population of regular smokers in Australia. One indicator of this is the proportion of regular smokers who have filled a bupropion prescription (Box). Estimates suggest that 22.8% of the population aged 20 years and over (25.2% males and 22.8% females) are current regular smokers.3 Seventy-three per cent (14 099 273) of the Australian population were aged 20 years or over in June 2001.4 Combining population and age-specific smoking prevalence estimates results in an estimated 3 198 738 current regular smokers aged 20 years and over. Given that the PBS guidelines allow for only one prescription of bupropion per smoker per year, an estimated 11% (351 772 prescriptions divided by 3 198 738 smokers) of current regular smokers filled a prescription for bupropion in 2001. Excluding smokers aged less than 20 years will have minimal effect on this estimate, as the incidence of high nicotine dependence in this group is likely to be low. Applying this method to each State and Territory reveals marked variation between them in the apparent proportions of smokers filling a prescription for bupropion. For example, an estimated 16.5% of all smokers in Tasmania had such a prescription filled, compared with 8% in Victoria. Although data from the HIC may be incomplete,5 the same method is applied to each jurisdiction in compiling them. Key unanswered questions relate to the extent to which (i) smokers complete a full course of bupropion, (ii) the field effectiveness of bupropion in aiding smoking cessation is comparable with abstinence rates achieved in clinical trials, and (iii) bupropion is used in conjunction with a comprehensive treatment program. We are currently conducting research to examine such questions. Characteristics of bupropion use in Australian States and Territories, February to December 2001, inclusive ACT WA Vic Tas SA Qld NT NSW Total Population (≥ 20 years) 225 797 1 372 378 3 550 348 338 108 1 110 897 2 602 539 131 542 4 767 664 14 099 273 Current regular smokers (≥ 20 years) 53 448 316 608 804 455 74 507 245 553 593 268 33 254 1 077 132 3 198 225 Bupropion prescriptions processed 4 470 46 186 64 482 12 279 32 455 81 619 3 803 106 478 351 772 Total cost of prescriptions processed $1 043 863 $10 854 828 $15 253 431 $2 920 481 $7 713 913 $19 237 576 $881 709 $25 238 851 $83 144 652 Proportion of current regular smokers who used bupropion 8.4% 14.6% 8.0% 16.5% 13.2% 13.8% 11.4% 9.9% 11.0%

Christopher M Doran · Anthony P Shakeshaft · Jennifer A Gates · Julia E Fawcett · Richard P Mattick

To the Editor: Cyanocobalamin and hydroxycobalamin are synthetically derived preparations of vitamin B12. Allergy to vitamin B12 injection is infrequent, but may be serious. We describe a patient with allergy to hydroxycobalamin, without cross-reaction to cyanocobalamin. Our patient was a 45-year-old woman with vitamin B12 deficiency. She had positive antiparietal cell antibodies and normal results of Schilling's test after addition of intrinsic factor. Otherwise she was in good health, with no other evidence of autoimmune disease. Her allergy commenced after an intramuscular injection of hydroxycobalamin, with onset of mild generalised pruritus. Subsequent monthly 1 mg injections of hydroxycobalamin were followed by incrementally worsening pruritus, and then frank urticaria. The last of nine injections was followed by urticaria, bronchospasm and oropharyngeal angioedema, which responded to administration of adrenalin. The patient underwent skinprick and intradermal testing with hydroxycobalamin and cyanocobalamin. Wheal-and-flare reactions occurred with injection of dilutions of hydroxycobalamin, suggesting an IgE-mediated response. No reactions were evident with dilutions of cyanocobalamin (Box). Subsequently, the patient had no reaction to a challenge of subcutaneously administered cyanocobalamin 0.1 mL (100 μg), and then intramuscularly administered cyanocobalamin 0.5 mL (500 μg). Her macrocytic anaemia resolved with ongoing monthly injections. After one year of treatment, the patient described an episode of delayed urticaria after a routine cyanocobalamin injection. The skinprick and intradermal tests were repeated, with negative reactions to cyanocobalamin, and wheal-and-flare reactions to hydroxycobalamin. She has since tolerated monthly intramuscular cyanocobalamin for over 12 months. Vitamin B12 allergy is rare, but has been reported.1-3 Positive results of basophil histamine release assay and skin testing suggest an IgE-mediated mechanism.2 Desensitisation is therefore theoretically possible; however, anaphylaxis during desensitisation has occurred.3 In Australia, both hydroxycobalamin (Neo-cytamen, David Bull) and cyanocobalamin (Cytamen, David Bull) are available. The excipients of each preparation are identical — sodium chloride, glacial acetic acid, and sterile water. Although the risk of severe allergy is low, adequate facilities for resuscitation should be available when parenteral vitamin B12 is administered. One approach to dealing with vitamin B12 allergy is to use the alternative compound after skin testing to exclude cross-reactivity. If cross-reactivity occurs, then desensitisation may be considered. Alternatively, oral administration of vitamin B12 may also be used.4 Results of skinprick and intradermal testing with hydroxycobalamin and cyanocobalamin* Dilution Saline (negative control) Morphine (positive control) Hydroxycobalamin Cyano-cobalamin Skinprick 1 : 1000 Negative Positive Positive Negative Intradermal 1 : 100 Negative Positive Positive Negative Intradermal 1 : 10 Negative Positive Positive Negative * A positive reaction was a wheal greater than that of the positive control.

David Heyworth-Smith · Patrick G Hogan

To the Editor: Safety is a rather subjective concept, so to use the word without definition, as Clarke et al did,1 is somewhat misleading. One possible definition is that the complication rate for general practitioners is no greater than for anaesthetists in the same circumstances. In the study by Clarke et al,1 the GPs were allocated the lower-risk cases and the anaesthetists were allocated the more difficult ones. Direct comparison was made without any adjustment for this difference. The data suggest that the GPs had similar or higher rates of adverse events or interventions despite handling lower-risk cases. The most recent data on anaesthesia-related mortality reports 20 deaths at endoscopy from 1994–1996, with a note that this is likely to be an underestimate.2 Using the authors' denominator of 430 000 endoscopies per year, an estimated risk of anaesthesia-related death is therefore about 1 in 64 000. The risk of anaesthesia-related death for all surgery is quoted as 1 in 63 000, which implies that anaesthesia for endoscopy is of average risk. The sample size of 28 000 lacks sufficient power to make any comment on safety as regards the risk of death. Although I applaud the clinical standards of the authors' institution and fully agree that propofol has many clinical benefits over other agents, Clarke et al do not prove safety in the use of propofol by non-anaesthetists.

Patricia Mackay · Patrick J Hughes

To the Editor: We read with interest the article by Clarke et al1 and the accompanying editorial by Knoblanche,2 and are concerned that they may be interpreted as endorsing the use of the anaesthetic agent propofol in sedation techniques by personnel inadequately trained in anaesthetic techniques. Cases reported to the Victorian Consultative Council on Anaesthetic Mortality and Morbidity confirm the risk of serious morbidity and mortality associated with these procedures. Our report for the triennium 1997–1999 will include two deaths at endoscopy where a non-specialist administered the sedation or anaesthetic. In both of these cases, propofol was used. The circumstances described by Clarke et al are exceptional. They combine a scrupulous adherence to the professional guidelines3 and a significant involvement by the administration of the endoscopy centre in the selection, education and on-going training of the general practitioner sedationists. They include incident reporting of adverse events and non-standard treatments as part of a quality assurance program. Although not specified, there is also, presumably, access to high-quality back-up. This level of attention to detail is by no means universal. We would like to endorse several observations made in the articles: For a procedure to be considered sedation, it is imperative that the drugs used are not intended to, and do not, cause loss of consciousness or the loss of protective reflexes or spontaneous ventilation; by definition, this would be anaesthesia. Proper selection and careful medical assessment of patients is very important, and training must enable the identification of patients at higher risk. People administering sedation must have knowledge of the pharmacology of the agents being administered and modifications necessary because of concurrent therapeutic regimens or disease states. They must also ensure adequate intraprocedure monitoring is provided, that they have experience in interpretation of abnormal indices, and that they can manage any complications arising from the procedure, with particular emphasis on airway management and cardiovascular resuscitation. There may be benefits with the use of propofol, but the guidelines, designed for patient safety, clearly state: "Intravenous anaesthetic agents such as propofol must only be used by an anaesthetist."3 Technological advances in non-invasive and minimally invasive procedures have led to an explosion in demand for sedation of increasing complexity in areas removed from the traditional operating room environment. Consequently, demand for sedation by non-anaesthetists is likely to grow. It is important that the standard of care and patient safety be maintained in all these circumstances. The concern is not whether the practitioner administering the sedation is a general practitioner or a specialist, but whether he or she has the training and skills necessary to function as an anaesthetist. The terms "general practitioner sedationist" and "non-anaesthetist" might give an impression of diminished risk, which is not supported by the experience of this committee.

Jon P Clarke · Anthony C Clarke FRCP, FRACP · Lybus C Hillman MD, FRACP

In reply: We thank Clarke and Mackay and Hughes for their interest and comments. We accept that a sample larger than the 28 000 endoscopies we reported1 would be required to establish the true incidence of death or other catastrophic complications of our sedation service. As the mortality rate is expected to be so low, it would take many years to achieve an adequate sample size. It is even difficult to determine the mortality from endoscopy in Australia, as quantifying all the endoscopies performed is problematic, and the Royal Australian and New Zealand College of Anaesthetists believes that not all deaths occurring from endoscopy are reported to anaesthetic mortality committees.2 It is essential that any sedation service is carefully planned, and that all doctors providing sedation receive adequate training and follow the protocols and guidelines of the endoscopy centre. However, we challenge the opinion that only anaesthetists should use propofol. We have not been able to find any clinical safety studies that demonstrate that only anaesthetists are able to use propofol safely. We believe the results of our study show that, when propofol is used in the manner described in the article, the rate of ventilatory and other complications is low (but clearly not zero). There is no reason to believe that the propofol component of the sedation regimen increased the rate of ventilatory problems. Indeed, it might, through its short duration of action, minimise such problems. To arbitrarily exclude the use of propofol by appropriately trained GP sedationists would deny many patients the manifest benefits of this drug. Importantly, our GPs have been shown capable of successfully managing the problems of airway obstruction and apnoea that were encountered — whatever the cause. We agree with Mackay and Hughes that the GP sedationists need to have the anaesthesia skills necessary to maintain patient ventilation, as well as an excellent understanding of all the drugs they use. Anaesthetists play a major role in improving safety standards in the provision of sedation for endoscopy through codifying the standards required,3 assisting the training of staff, and delivering sedation services to high-risk patients. But many patients can be successfully sedated without a specialist anaesthetist being present. We argue there is no evidence that these patients should receive a suboptimal regimen. Most specialist anaesthetists have a more valuable role to fill than providing sedation for straightforward endoscopies.

Jon P Clarke

To the Editor: The study by Douglas et al1 reports that the purposeful hastening of death in terminal illness is both widely practised and an acceptable method of palliative care for over a third of Australian surgeons. Yet the study is based on a questionnaire strongly favouring theoretical scenarios rather than actual practice. The framing of the questions maximises reporting bias towards "hastening death" by the use of absolute terms (eg, "Have you ever . . .?"; "[Are] there any circumstances . . .?"1). The study also fails to determine the stage in a terminal illness at which hastening of death had been, or might possibly be, acceptable to the surgeons, nor the reason for its implementation. If the patient's symptoms are already adequately controlled and the dying process is not prolonged, we do not see why it is necessary to administer doses in excess of those required to control symptoms. Whose symptoms are we treating?

Mathew Piercy · Gerald B Fogarty · Aubrey W Jansz · David M Gawler · Luis Vitetta · David Kenner · Avni Sali

To the Editor: We wish to comment on the article by Douglas et al about surgeons hastening death.1 Recently, Ray2 noted that, historically, surgeons have had to witness their patients' pain probably longer than any other medical specialty. Hence, it is not surprising to read Douglas and colleagues' report1 that more than a third of surgeons surveyed performed life-terminating events without explicit and persistent requests. The desire for death, as perceived by doctors and patients, has been correlated with ratings of pain, depression and poor family support.3,4 Surgeons, while intending "no harm", could be seen to be palliating themselves (relieving their own distress) as well as their patients' when performing life-terminating events. Helplessness, hopelessness and negative attitudes toward life-sustaining treatments are common and understandable in palliative care, and experienced by both doctors and patients.3,4 Surgeons are relatively new to palliative care and could have a significant contribution to make to the psychosocial support of cancer patients with terminal illness. Decisions by surgeons about hastening death may be modified by greater interaction with their patients' families and with other healthcare providers working with terminally ill patients. A clinical-outcomes study of 102 consecutive deaths of terminally ill patients admitted to hospice care during 1997–1999 highlights the importance of family support in the outcomes immediately before death.5 The study period coincided with a time of heightened awareness of euthanasia in Australia. Fifty-six per cent of the patients had received continuous family support and 44% sporadic support; these patients appeared to have less distress and better preterminal outcomes, despite 10% experiencing significant family conflict and distress. However, no spontaneous or other requests for euthanasia were recorded in patient files or notes by staff. This could reflect, in part, family support received in an environment conducive to positive interactions between patients, their families and health professionals. Involving such palliative care teams in surgeons' care for terminally ill patients may have a significant effect on both pre-terminal outcomes and hence requests for euthanasia.

Mathew Piercy MB BS · Gerald B Fogarty BSc, MB BS(Hons) · Aubrey W Jansz MB BS, FRACS · David M Gawler MB BS, FRACS, FRCS · Luis Vitetta · Avni Sali Professor and · David Kenner

To the Editor: We were interested to read the recent supplement on preventing osteoporosis.1 We could find only one reference to cigarette smoking, on page S13, where it is noted that "the role of lifestyle changes (including specific exercise regimens, changes in diet and quitting smoking) has not been evaluated adequately". The orthopaedic literature is replete with information and evidence on the adverse effects of cigarette smoking on bone density, healing of fractures, incorporation of bone grafts, etc. A meta-analysis2 has suggested that smokers have greater bone loss over time than non-smokers. Granted, there may as yet be no direct evidence that quitting smoking reduces "the fracture burden". However, to discuss osteoporosis and management of fractures without discussing the major adverse effects of cigarette smoking is akin to discussing the prevention of melanoma and the outcomes of treatment without discussing unprotected exposure to sunlight (for example). Could the "writing group" tell us what steps are being taken to inform Australians of the adverse effects of cigarette smoking on their bones, quite apart from the other public health issues surrounding this extraordinarily harmful habit?

Roy PL Carey · Walter E Plehwe · Peter R Ebeling

In reply: On behalf of the writing group, I wish to thank Carey and Plehwe for their perceptive comment on the effects of cigarette smoking on bone health. The focus of our supplement was the prevention of fragility fractures, and unfortunately there is no evidence that smoking cessation reduces fracture rate. However, the authors are correct to emphasise the negative impact of cigarette smoking on fracture healing, bone graft incorporation and bone density, the last factor being a strong predictor of fragility fracture. The meta-analysis that Carey and Plehwe refer to1 showed that hip bone mineral density (BMD) in current smokers was one-third of a standard deviation below that of people who had never smoked. This meta-analysis and another recent study2 showed that these effects are greatest in men and are dose-dependent. Prospective studies also show that smokers have higher rates of bone loss than non-smokers. Extrapolations from these BMD data suggest smoking increases the lifetime risk of vertebral fracture by 13% in women and 32% in men, while hip fractures are increased by 31% and 40%, respectively. In response to the question of what steps are being taken to publicise the effects of smoking on bone health, we would like emphasise that further prospective studies are urgently required to assess the effect of smoking cessation on fracture risk, BMD and bone turnover; and the message that smoking has a negative impact on BMD should be incorporated into public education campaigns run by government and non-government organisations for both osteoporosis prevention and smoking cessation. This area of bone health is eminently suited to successful intervention.

Roy PL Carey FRACS · Walter E Plehwe MB BS FRACP PhD · Peter R Ebeling MD FRACP

To the Editor: I applaud the ideology of the Journal in its pursuit of evidence-based excellence, drawn to our attention by Rosselli in a recent letter.1 Based on data that the MJA is at the top of the list of English-language journals publishing "evidence-based medicine" references, Rosselli asks "could it be that MJA readers are three times more interested in EBM-related topics than BMJ readers?". My response is that the editors (and perhaps also reviewers) of articles for the MJA may well be more interested in EBM than their BMJ counterparts, but this does not necessarily apply to readers. Without actual evidence of what interests MJA readers, it is not possible to draw any conclusions. Perhaps the readers should be asked?

C Ross Philpot

To the Editor: It was a most interesting idea to call for a list of favourites books.1 I have some of my own to add. The death of Ivan Ilyich, by Tolstoy — all the different responses to a dying man handled in the way only Tolstoy can. Interestingly, the most comforting presence was that of his illiterate man-servant, Gerassim. The doctor, his patient, and the illness, by Michael Balint — this was introduced to me by a very thoughtful medical student when I was a surgical registrar in England 40 years ago. A new edition edited by Balint's son has appeared in the past year or two. Contrary imaginations, by Liam Hudson — the advice in this book on selecting medical students and introducing first-year students to clinical medicine is now being carried out by many teaching schools 35 years after it was written! A long season in hell, by Gail Graham — the story of a mother's fight to rehabilitate her head-injured son. This book has all the characters that a doctor should be aware of — good doctors, bad doctors, unimaginative bureaucrats and politicians, helpful people and an inspiring physiotherapist who lost her husband, apparently through suicide, during the period she was fighting for her son. The book would be an ideal seminar topic for medical students and ethics discussion groups. Lastly, Intellectual impostures, by Sokal and Bricmont — a book with little medical content, but which should be read by every intelligent person. Sokal submitted pretentious gobbledygook to a postmodern journal, which was published. He then exposed the hoax, creating an enormous stir. This book is an expansion of that exposure.

Aubrey W Jansz

To the Editor: Recent astronomic awards for medical mishaps, and the follow-on medical indemnity crisis, bring back memories of an earlier editorial I wrote for the Journal dealing with some aspects of this problem. Its message is worth repeating. In 1990, a letter by Armstrong was published in the Journal,1 which enclosed a newspaper advertisement for a firm of solicitors, Messrs Stern, Stern & Tanner, inviting custom from anyone who may have been a victim of "obstetric negligence", "even if your child was born as long ago as 1965 or even earlier. Initial consultation free". The Journal's then editor invited me to write an accompanying editorial. The editorial, published simultaneously with Armstrong's letter and accompanying advertisement, was headed "No hawkers, canvassers or solicitors".2 It began with a quotation from United States Chief Justice Burger: "Never, never, never under any circumstances hire an advertising lawyer!".3 It noted that some solicitors claimed to possess special expertise in asbestos-related diseases, others claimed to know all about defoliants. I suggested, in those innocent days, that "obstetric negligence" was a new subspecialty which should henceforth be known as Stern–Tanner disease. (If Drs Guillain and Barré were to be forever enshrined in medicine's Hall of Fame, why not Messrs Stern and Tanner?) My editorial suggested a remedial response, stating that "the law permits parties to any contract to abrogate, limit or qualify their legal rights, duties, liabilities and remedies which might otherwise arise. Thus, while the courts — both here and in the United Kingdom — tend to lean against total exclusion from liability, they are, surprisingly, more benevolent towards clauses which limit liability, however contemptuous the specified amount." I adhere to that view. I can see no reason why doctors — save in an emergency — cannot demand that patients enter into a written "contract of treatment" which limits the treating doctor's liability to a specified amount. And while such contracts will not be valid against children born impaired as a result of "obstetric negligence", I can see no reason why the courts should deny the validity of such contracts as a matter of public policy. Why should "contracts of treatment", entered into between consenting adults, be any different from those entered into with lawyers or dry cleaners? It may seem a trifle offensive, but then the practice of medicine is treated in law as a business, much the same as touting lawyers or dry cleaners. If doctors were once put on a pedestal, this pedestal has been effectively knocked down by astronomic awards of damages freely awarded by the courts — whether by judge or jury — largely as a result of the freely expanded constituent elements of the tort of negligence. I researched the law relating to "limitation of liability" clauses, which was published in the Australian Law Journal4 and quoted in the MJA editorial: "Why the courts should be more benevolent to clauses of limitation than to clauses of exemption is not easy to comprehend. . . . The principle, it seems to the present writer, is much the same: Exclusion clauses save a party from having to pay anything, whilst a limitation saves him from having to pay as much as he would otherwise have to pay. [ . . . ] Alas, the last word on this seems to have been spoken for some time to come." As far as I am aware, the law has not changed since the publication of my earlier research.

Paul Gerber

To the Editor: The ME/Chronic Fatigue Syndrome Association of Australia Limited. has expressed its concern over the content of the Royal Australasian College of Physicians' clinical practice guidelines on chronic fatigue syndrome, published as a recent supplement to the Journal.1 Recognising a shared objective to overcome the challenges of chronic fatigue syndrome (CFS), neither the Association nor the College believes that conflict will provide a useful path to future answers. Accordingly, as the Chairman of the ME/Chronic Fatigue Syndrome Association of Australia and the President (at the time the guidelines were published) of the Royal Australasian College of Physicians, we would like to document the common ground we have identified. We acknowledge, as do the guidelines, that CFS is a serious, disabling illness. There is no evidence that the illness is primarily psychological in origin. There is significant evidence of a range of biological abnormalities occurring in people with CFS. It remains unclear whether these are primary or secondary. Treatment should be personalised according to the symptoms and circumstances of the individual patient. Treatment plans should be worked out by the patient together with a healthcare professional and designed to be within the capabilities of the patient. Scientific evidence on aetiology, pathophysiology and treatment is, at this stage, grossly deficient. More research is required to understand the biological mechanisms involved and to clarify the role that genetic, environmental and infectious agents might have in the aetiology and pathophysiology of this complex and debilitating illness. The medical community, other health professionals and patients and their families should work together to encourage increased funding and research into the epidemiology, aetiology and pathophysiology of CFS so that we may find more effective treatments for this condition (or these conditions). All clinical guidelines should be viewed as documents that will, in time, require refinement, rewriting and replacement. Doctors must be cognisant of the limitations of all such guidelines and be aware that the investigation and management of a patient's condition must be determined with the assistance of the best and latest information as it emerges and, in all instances, be tailored to the needs of the individual patient.

Richard G Larkins · Simon R Molesworth

To the Editor: I read with interest the article by Goldsworthy and McFarlane on Howard Florey, Alexander Fleming and penicillin.1 With regard to the cause of Florey's "famous pinched smile", which allegedly hid tooth erosion caused by his drinking dilute hydrochloric acid prescribed for achlorhydria, a more prosaic yet interesting explanation is found in the memoirs of Raymond Valentine Hennessy. Hennessy was Senior Ear, Nose And Throat Surgeon at St Vincent's Hospital, Melbourne, between 1928 and 1951.2 Howard Florey in the late 1930s In August 1936, Florey, who was then Professor of Pathology at Oxford, visited his dying mother in Melbourne. He and his family stayed with his sister, Dr Hilda Gardner. Florey had a supply of sulfanilamide, probably the first in Melbourne, to treat his daughter, who was convalescing from a recent mastoid operation. During his stay, Florey attended a local dentist for treatment of a painful lateral incisor tooth. Some days later (on a Saturday evening), his face had become swollen and he began having rigors. His sister, a medical graduate who was then working as a clinical pathologist and microbiologist at the Melbourne Hospital, appreciated the danger — an abscess of a lateral incisor tooth can produce a cavernous sinus thrombosis — and quickly contacted Raymond Hennessy, who lived nearby. Hennessy had graduated as a dentist before pursuing a career as an ear, nose and throat surgeon and had written about the dangers of a lateral incisor dental abscess.3 After examining Florey, Hennessy told him that the offending tooth required extraction that night. Initially, Florey refused to heed his advice, preferring to see his own dentist the following Monday. Fortunately for Florey, he was persuaded by his sister to have the extraction. Hennessy then telephoned a dentist colleague, and they all met at the latter's surgery in Collins Street, where Hennessy gave Florey a gas–oxygen anaesthetic, and the nervous dentist proceeded to extract the incisor. However, he extracted the normal central incisor, not the offending lateral! On realising his mistake, the dentist "went to water", but Hennessy immediately rose to the occasion and extracted the correct tooth. When Florey woke from the anaesthetic, as Hennessy well remembered, he was not amused. Later, he had a dental plate made but did not like wearing it. Whether he took the sulfanilamide is not known. This episode is not mentioned in Gwyn Macfarlane's biography of Florey.4 However, the photograph of Florey in the frontispiece of this book shows the gap in his upper incisors (pictured). I believe this is the explanation for Florey's "famous pinched smile".

Ivo D Vellar · Thomas B Hugh

To the Editor: The patronising article by Goldsworthy and McFarlane on the discovery of penicillin1 depicts the popular heroic view of Alexander Fleming as a myth, but also promulgates myths of its own. Their description of the Fleming saga is historically accurate. Fleming searched for an answer to the riddle of infection, and, to paraphrase Pasteur, chance in the form of a spore of a rare subtype of Penicillium favoured his prepared mind. Whether or not the spore came through an open window is irrelevant, but the windows in Fleming's laboratory — now preserved as a museum (pictured) — could be opened3 and probably were on occasion, as Fleming was a heavy smoker. Fleming perceived the significance of inhibition (or, more correctly, lysis) of staphylococcal colonies, named the active agent "penicillin" and studied its effect on animals. Goldsworthy and McFarlane are "astonished" that he failed to inject it into infected animals to investigate its therapeutic effect, but the reason is simple: Fleming discovered that penicillin was rapidly inactivated by serum, dashing his hopes for its use as a systemic agent.3 Although he met opposition from his chief, Almroth Wright (known to his students as "Almost Right"), who rejected the view that penicillin might be a useful therapeutic agent, it is absurd to say that Fleming was "a victim of the pessimistic mind-set against toxic chemical antimicrobials". His confidence in its lack of toxicity led him, in 1929, to use penicillin to treat pneumococcal conjunctivitis in one of his assistants, with dramatic success.2 Site of Fleming's laboratory The Clarence Wing, St Mary's Hospital, London, in 1910. Fleming's laboratory, where penicillin was discovered in 1928, was on the third floor of the tower on the right. The windows of the laboratory could be opened by an internal system of ropes and pulleys, but a more likely source of the Penicillium spore was a dumb-waiter shaft communicating with a mycology laboratory on the floor below.2 To say that "Fleming had little idea what to do with his mould apart from dabbing it on infected wounds" unfairly trivialises his actions after the discovery. In addition to clinical and animal studies, he had the mould identified, deposited a culture with the collection held by the Medical Research Council and published his observations. He set two researchers to work purifying the active principle of the mould broth, and established that penicillin was soluble in alcohol and that its stability was pH-dependent. He also developed an assay for penicillin and defined the range of organisms that were sensitive to it. He sent cultures of the penicillin-producing strain to many laboratories, including Oxford, where that very culture later provided the starting point for Chain and Florey's work. Fleming's subsequent work on penicillin was stalled by his lack of biochemical expertise; he was unable to overcome the difficulties of purification and stabilisation.2 Fleming is recorded as saying, "It's up to the chemists now, I'm no chemist".2 It is quite untrue that "he then effectively forgot about it for 13 years". Although Fleming ceased clinical work on penicillin in 1934, he continued with laboratory studies. A contemporary at St Mary's Hospital, Dr A G Cross, recalled that in the 1930s "penicillin was on his mind all the time and in the minds of those who worked with him".2 Fleming had his faults, but the genius of his prepared mind did indeed present humanity with a fairy tale come true. Perhaps Ernst Chain, who did not particularly like Fleming, should have the last word: "There is no doubt that this discovery, which changed the history of medicine, has justly earned [Fleming] a position of immortality."2

Ivo D Vellar · Thomas B Hugh · Peter D Goldsworthy MB BS · Alexander C McFarlane MD, Dip Psychother, FRANZCP