Letters

John H T Ellard Psychiatrist, Medical Specialist Centre, 710 Military Road, Mosman, NSW. To the Editor: I read with interest the article in the Journal on ranking medical journals and the fallacies to be found therein.1 I have a simpler method. I subscribe to two classes of journals: those specialising in psychiatry, and more general journals. The psychiatry journals I keep entire. However, as my house is of modest size I cannot do that with the general journals, so I tear out and file the articles that I find interesting and informative. You will be interested to know that in the past month I have filed away one article from the Lancet, one from the New England Journal of Medicine and three from the Medical Journal of Australia. What better measure of merit could there be?

John H T Ellard

William Lancashire*, Craig Hore† and Jennifer A Law‡ *Conjoint Senior Lecturer, †Senior Lecturer, ‡Lecturer, School of Rural Health, University of New South Wales, PO Box 2466, Port Macquarie, NSW 2444. Bill.lancashireATmaynegroup.com To the Editor: We read with interest Hillman's editorial on the hospitalist movement.1 As our group includes a couple of recent expatriates from the Canadian healthcare system,* we can give some historical perspective on the evolution of the hospitalist in Canada, some of which parallels what is happening in Australia. Traditionally, family physicians (general practitioners) in Canada were able to manage their patients in hospital, either as the primary care doctor or in consultation with a specialist. Not infrequently, the specialist would assume primary care and consult with the family doctor. Continuity of care was assured, and both the family doctor and the specialist benefited socially and professionally from the interaction. The "corridor consultation" thrived and the doctor's lounge was a source of medical education and social interaction as GPs and specialists met over a morning coffee before rounds. Around 10 years ago, the family doctor became increasingly unwelcome in the hospital, particularly in teaching centres. As there was never a financial incentive to be involved in hospital practice, this atmosphere persuaded most family doctors to resign their hospital privileges. However, it soon became apparent that a visiting-consultant-based service could not cope with the numbers of patients being admitted to hospitals. Patients with no apparent "teaching value" were becoming difficult to admit into teaching units. Consequently, those few GPs who had retained hospital privileges were increasingly being asked to accept patients primarily under their care. As the system became more stressed, they found that they were managing more and more acutely ill patients. These experienced GPs evolved to become hospitalists — essentially, primary care doctors who were prepared to look after acutely ill inpatients, often in consultation with a specialist. Unfortunately, attempts to encourage GPs back into the hospital system have generally proved unsuccessful. The College of Family Physicians of Canada, recognising that there may no longer be ready access to specialist services or hospital beds, is starting to train its residents accordingly. We agree with Hillman that the complexities of acute medicine require specialists (such as emergency physicians, intensive care specialists and general physicians) with training and skills in acute medicine, resuscitation and multisystem problems. Indeed, our experience in rural Australia suggests that hospital-based multidisciplinary critical care physicians are already undertaking some of the hospitalist roles that Hillman describes. Perhaps we are witnessing the emergence of hospitalists in Australia.

William Lancashire · Craig Hore · Jennifer A Law

F Frank Pyefinch Director of MD Development, Health Communication Network, 2 Santa Fe Drive, Bundaberg, QLD 4670 frank.pyefinchAThcn.com.au To the Editor: I would like to comment on the recent article by Newby et al.1 They conclude that the default settings in computerised prescription packages result in a significant increase in the use of antibiotics. I do not believe this is a valid conclusion. As the authors state that 85% of general practitioners generating computerised prescriptions are using Medical Director (MD), it is reasonable to assume that the default settings in MD would contribute significantly to this effect if their conclusion is correct. I have installed and tested MD v.2.3 from February 2000, MDW v.1.85 from February 2000 and MD v.2.4 from May 2000. These were the versions that would have been in use at the time of this study. All versions default to printing "once-only" prescriptions without repeats. In fact, when a "once-only" prescription has been selected, MD's default behaviour is to display a prompt for the quantity and repeats with the default repeats field set to "0". This is very easy to verify simply by installing a copy of MD onto a "clean" computer and printing some scripts. As this was evidently not done, it casts doubt on the quality of the whole study. How can the authors reach a conclusion about the effect of the default settings in computerised prescription packages without first ascertaining what those default settings are? They appear to have assumed that the default behaviour of all computer prescription packages is to print the maximum number of repeats allowed by the Pharmaceutical Benefits Scheme. No attempt appears to have been made to verify whether this is the case. Whatever the reason for the observed increase in repeat antibiotic prescriptions, it is incorrect to conclude that it is due to the default settings in computerised prescribing packages. No discussion of other possible explanations for the observed increase is presented and it appears as though the data have been used to support a conclusion that had been decided before the study was commenced.

F Frank Pyefinch

David A Newby,* Jayne L Fryer,† David A Henry‡ * Lecturer, † Statistical Analyst, ‡ Professor, Department of Clinical Pharmacology, University of Newcastle, Newcastle Mater Misericordiae Hospital, Newcastle, NSW 2298 mddanATalinga.newcastle.edu.au In reply: As Pyefinch notes, if the "once only" option in Medical Director (MD) is chosen during prescribing, the doctor must enter the quantity and number of repeats that he or she wishes to order. However, if the doctor chooses the "regular" medicine option (both options are offered during prescribing), then the maximum Pharmaceutical Benefits Schedule quantities and repeats are inserted. There are various reasons why doctors may be using the "regular" option rather than the "once only" option when prescribing antibiotics using MD. Some of these have been discussed on the General Practice Computing Group Listserv,1 and include factors such as confusion regarding the terms "regular" and "once only" and difficulties recalling patient medication histories if the "once only" option is used. Another explanation is that doctors commonly prescribe chronic medications, and therefore use of the "regular" option may become a habit. Whatever the cause, there is no obvious explanation for the differences observed, except for the use of prescribing software. Our recommendation that prescribing software be altered to avoid these shortcuts was made because it represents the most immediate way of resolving the problem.

David A Newby · Jayne L Fryer · David A Henry

Jennifer A O'Dea Senior Lecturer, Faculty of Education, University of Sydney, Building A35, Sydney, NSW 2006. j.o'deaATedfac.usyd.edu.au To the Editor: As part of a large, national nutrition study, height and weight were measured among 4441 students from 38 schools randomly selected from lists of all state and territory schools in Australia in 2000. Public, private and Catholic schools, in both rural and urban areas, were represented. Schools were categorised as being of low or middle/high socioeconomic status (SES),1 based on direct measurement of parental income. Parental consent was obtained, and the study was approved by the University of Sydney Ethics Committee and all state departments of education. Overweight and obesity, as defined by an international standard definition,2 were identified in 17.3% and 6.4% of participants, respectively. These characteristics showed a trend towards greater prevalence among students from low-SES backgrounds compared with those from middle/high-SES backgrounds for the total group (19% v 16.8% overweight [P = 0.09]; 8.9% v 5.8% obese [P = 0.02]), females (19.7% v 17.2% overweight [P = 0.2]; 6.9% v 6.2% obese [P = 0.56]), and males (18.5% v 16.3% overweight [P = 0.23]; 9% v 5.5% obese [P = 0.003]), although not all differences were statistically significant. After controlling for SES differences in age and height, mean body mass index (BMI) was significantly higher among low-SES than middle/high-SES participants for the total group (20.3 kg/m2 [95% CI, 20.1–20.5 kg/m2] v 19.7 kg/m2 [95% CI, 19.6–19.9 kg/m2]; P < 0.001), females (20.4 kg/m2 [95% CI, 20.1–20.7 kg/m2] v 19.8 kg/m2 [95% CI, 19.6–19.9 kg/m2]; P < 0.001), and males (20.2 kg/m2 [95% CI, 20.0–20.5 kg/m2] v 19.6 kg/m2 [95% CI, 19.5–19.8 kg/m2]; P < 0.001). A breakdown of results by SES, sex and school level is shown in the Box. Low-SES primary school children were also 1–2 cm shorter, on average, than middle/high-SES primary school children (boys: mean 141.5 cm [95% CI, 140.6–142.5 cm] v 143.5 cm [95% CI, 143.0–144.0], P < 0.001; girls: mean 141.0 cm [95% CI, 140.8–142.6 cm] v 143.3 cm [95% CI, 142.5–143.6 cm], P = 0.01). The average proportions of overweight and obese children and adolescents in the study were similar to those found in other Australian studies.3-5 The results suggest that SES is a factor in the development of overweight and obesity among Australian school children. This may be a relatively recent trend, as these data were obtained in late 2000. Low SES in children may also be associated with nutritional deprivation and height retardation. Further research should clarify these relationships among children from low, middle and high SES backgrounds, as well as examining the combined impact of both SES and ethnicity. School students classified as overweight or obese* according to socioeconomic status (SES), school level and sex Males (n = 2232) Females (n = 2209) Low SES (n = 574) Middle/high SES (n = 1658) Low SES (n = 508) Middle/high SES (n = 1701) Primary school students (grades 1–6; ages 6–13 years) Overweight students 19.4% (42/216) 16.2% (110/680) 23.2% (51/220) 17.8% (136/766) Obese students 6.9% (15/216) 5.3% (36/680) 6.4% (14/220) 5.7% (44/766) High school students (grades 7–12; ages 13–18 years) Overweight students 17.6% (63/358) 16.4% (160/978) 17.0% (49/288) 16.8% (157/935) Obese students 10.1% (36/358) 5.6% (55/978) 7.3% (21/288) 6.5% (61/935) * Overweight and obesity are classified according to the international standard definition.2

Jennifer A O'Dea

Tony D Merritt,* Carolyn Herlihy† * Epidemiologist/Biostatistitian, † Environmental Health Officer, Hunter Public Health Unit, Hunter Area Health Service, LMB 119, Wallsend, NSW 2287. tmerrittATdoh.health.nsw.gov.au To the Editor: Travelling animal shows, with animals such as young poultry, rabbits and reptiles, commonly visit childcare centres in Australia. Transmission of Salmonella infection to children from ducklings and chickens is well documented in the United States1,2 and United Kingdom,3 but not in Australia. We investigated a cluster of Salmonella agona cases in children, identified through routine laboratory surveillance. Initial investigations identified a potential association with visits to childcare centres by a single local hatchery. At each show, children saw an egg hatching, watched day-old ducklings swim and had the opportunity to hold a day-old chick or duckling, and to touch an adult chicken. Details of gastrointestinal illness were sought from childcare centres that had hosted visits from the hatchery. This identified laboratory-confirmed cases with onsets between 14 April and 6 May 2002 in seven people. All attended one of four childcare centres; six were children and one was a staff member. Onset of illness occurred 2–12 days after the hatchery visit. A total of 316 children attended shows by the hatchery between 9 April and 8 May. Environmental sampling at the hatchery recovered Salmonella agona from multiple sites over repeated visits. These included the egg incubator, faeces from day-old hatchlings returning from a show with children, faeces from young chicks and ducklings in the brooder cage, duck faeces from their yard, corn meal used in the preparation of feed for all poultry, and rat droppings in the feed preparation and storage areas. No Salmonella were detected in samples collected from the corn meal supplier. New procedures based on guidelines from South Australia4 and the Centers for Disease Control and Prevention5 were introduced for all shows from 8 May. These included supervised handwashing after handling animals, avoiding carpeted areas for the show, adequate cleaning of macroscopic faecal contamination and disposal of the water used by swimming ducklings into the sewerage system. No further cases were identified among the 251 children who attended shows over the subsequent two weeks until the hatchery cancelled all further visits. We conclude that infection was acquired through contact with chicks and ducklings from a single hatchery, and that the outbreak was halted by the introduction of measures emphasising handwashing after animal contact. Contact between young children and young animals, both of whom have naïve immune systems, represents a special ecological niche. Young poultry are particularly vulnerable to salmonella infection and subsequent high level excretion, and young children are similarly vulnerable. As animal visits to childcare centres are highly valued by children and carers, appropriate guidelines should be widely promoted to the childcare sector and the petting zoo industry to reduce the risks these visits pose. The petting zoo guidelines developed in South Australia4 are an excellent resource and NSW Health is currently developing a fact sheet on the topic.

Tony D Merritt · Carolyn Herlihy

To the Editor: Metformin is a commonly prescribed oral hypoglycaemic agent used to treat type 2 diabetes mellitus. Its major effect is on hepatic glucose production and thereby fasting blood glucose level (f-BGL). Metformin does not commonly cause weight gain, and may be associated with significant weight loss.1 Over the past year, we have prescribed metformin as adjunctive therapy for five patients with type 1 diabetes (T1DM). ...

Jenny E Gunton · Stephen M Twigg

To the Editor: I concluded my recent article with a hope that the "recommendations of the Dawson Committee will provide much needed amendment to the Trade Practices Act".1 In the article, I argued that medical rosters ran the risk of illegality as exclusionary provisions under the Trade Practices Act 1974 (Cwlth) because of either a drafting deficiency in the Act or a policy non-appreciation of ...

Warren Pengilley

To the Editor: Chlamydial infection is the most common bacterial sexually transmitted infection in Victoria and Australia, and notifications are increasing significantly. Most chlamydial infections are asymptomatic and, if untreated, lead to significant morbidity.1 The direct costs of these infections to the Australian healthcare system have been estimated at $90–$160 million annually.2 Chlamydial infection has been implicated in as many as 50% of cases of infertility.3 Widespread screening is cost effective and reduces both the prevalence of infection and the rate of complications.4 Screening is recommended in a number of countries and in the recently released Victorian Chlamydia Strategy.2 To determine if there has been an increase in the proportion of women tested for chlamydial infection who are asymptomatic in Victoria, we analysed notification and enhanced surveillance data. Since 1997, the Department of Human Services has collected enhanced surveillance data using a standard questionnaire distributed by laboratories to clinicians. The forms record information on risk factors and the reason for testing. Data from the Health Insurance Commission were obtained to provide an indication of trends in testing through Medicare. The number of notifications almost doubled between 1997 and 2001, from 2059 to 3977 notifications (Box). In 2001, enhanced surveillance data were available on 2300 notifications (58%). Symptomatic presentation remained the major reason for testing (53%), with no significant change in the proportion tested for this reason since 1997. However, when data for the two sexes were analysed separately, the proportion of men who were symptomatic fell significantly between 1997 and 2001, from 77% to 65% (P = 0.02), while there was no change in the proportion of women who were symptomatic — 42% in 1997 and 41% in 2001 (P = 0.35) (Box). From 1997 to 2001, the number of positive chlamydial tests notified relative to the number of tests conducted has remained constant (9.7%, 8.8%, 11.8%, 11.5% and 9.8% in consecutive years, respectively), despite the increase in number of tests performed. However, even if all tests had been performed in the 20–30-year-old age group, they would represent only 10% of the population in that age group tested each year. Our data suggest that the reasons for testing over the past 5 years have not changed, and, in particular, that there has been no change in the proportion of women tested who are asymptomatic. Chlamydial infection meets the World Health Organization criteria for a screening program,5 and screening for this infection is cost effective.4 Although chlamydial infections are a significant public health concern for women, targeted screening is not occurring widely. Chlamydia notifications and enhanced surveillance data for Victoria, 1997–2001 1997 1998 1999 2000 2001 Male Female Male Female Male Female Male Female Male Female Number of notifications 788 1271 948 1542 1181 1761 1328 1915 1631 2346 Number with enhanced surveillance data 484 782 619 954 735 968 858 1037 992 1308 Reasons for testing* Symptomatic 76.9% 42.3% 75.8% 46.0% 73.9% 44.1% 68.6% 39.4% 64.9% 41.1% STI Screen 7.2% 25.4% 6.5% 25.6% 9.0% 26.3% 9.8% 27.2% 12.3% 27.2% Asymptomatic contact 13.8% 15.5% 16.2% 17.5% 15.8% 18.7% 18.1% 16.9% 18.9% 17.4% Abnormal examination 0.2% 3.2% 0.6% 3.8% 0.7% 4.0% 0.2% 1.6% 0.8% 2.3% Pre-termination screen 0 9.3% 0 5.5% 0 5.6% 0 5.6% 0 3.7% Other/not stated 1.9% 4.3% 0.9% 1.7% 0.6% 1.2% 3.2% 9.3% 3.1% 8.4% Total notifications 2059 2490 2942 3243 3977 STI = sexually transmitted infection. * Mutually exclusive choices presented to clinicians in the surveillance questionnaire.

Megan L Counahan · Jane S Hocking · Christopher K Fairley

To the Editor: We describe six cases of amoebic appendicitis, encountered during a 15-month period in the Pathology Department of the Royal Darwin Hospital. The patients were all Indigenous Australians and four of them came from remote communities. The average age was 24 years and five of the six were men. They all presented with abdominal pain, fever and right iliac fossa tenderness for up to 6 days. The clinical notes did not indicate the presence of pre-existing dysentery, and no faecal examination was undertaken. Recovery following appendicectomy was uneventful in all the cases. No further follow-up is available. The aetiology was established histo-logically, as no distinctive clinical or macroscopic features were noted. The appendices showed extensive coagulative necrosis of the mucosa, submucosa and muscularis propria and invasion of the wall by variable numbers of amoebae with ingested blood cells. Inflammation secondary to perforation was evident in the serosa and the mesoappendix, but no significant inflammation was seen in the inner layers of the wall (Box). Coagulative necrosis of the appendiceal wall was seen in all the cases. In fact, in some of the cases the diagnosis was suspected on seeing this distinctive coagulative necrosis and the amoebae were only found later. Thus, we believe this type of necrosis, which has not been previously emphasised in the literature, to be characteristic of the condition. Amoebiasis is rare in a developed country like Australia. The infection is generally acquired during travel to endemic parts of the world. However, case reports of invasive amoebiasis in Indigenous Australians who have not travelled outside Australia have been previously reported.1,2 Immigration, immunosuppression and poor sanitation are other settings in which amoebiasis can be seen.3 The finding of six cases of amoebic appendicitis in Indigenous Australians in the Northern Territory in a 15-month period is significant, as appendicitis is a very rare manifestation of the disease. McCarthy et al, in a recent MJA article,2 highlight the potential public health significance of endemic invasive amoebiasis because of its high transmissibility in settings where hygiene may be suboptimal. We agree with this and believe in selective screening and appropriate treatment of at-risk contacts of the patients, as prolonged latency between infection and disease is well documented.3 Cross-section of appendix Cross-section of appendix with coagulative mural necrosis and secondary serosal inflammation (haematoxylin and eosin [H&E] original 1×). Numerous amoebae in clear spaces, some with ingested blood cells, are seen in the insert (H&E original 40×).

Ibrahim M Zardawi · Joseph S Kattampallil · Jurgen W Rode

To the Editor: Murray et al recently suggested that after an initial screening urinalysis, routine urinalysis could be eliminated from antenatal care without adverse outcomes for women.1 Their conclusions are based on a prospective observational study of 1000 women, 26 of whom developed pre-eclampsia, with 6/24 (25%) developing new proteinuria before the onset of hypertension. We have concerns about the authors' claims, which, we believe, are not justified by their findings. Pre-eclampsia remains a major cause of maternal and perinatal mortality.2 A study of 26 cases of pre-eclampsia is clearly underpowered to evaluate important morbidity and mortality outcomes. Therefore, one must place great emphasis on potentially undetected cases of pre-eclampsia, such as those that repeatedly appear as examples of substandard care in the Report on confidential inquiries into maternal deaths in the United Kingdom.3 That the initial screening urinalysis detected only 2/26 (8%) women who subsequently developed pre-eclampsia is no surprise. In contrast, routine urinalysis at antenatal visits detected 25% of cases of pre-eclampsia before the onset of hypertension. In the UK, this would amount to 4500 women per annum. The detection of proteinuria provokes further antenatal visits earlier than would normally be planned, facilitating early detection and management of pre-eclampsia. Intervals of 2–4 weeks between visits are usual in the early part of the third trimester when the risks to mother and fetus of severe early onset pre-eclampsia are greatest.4 By eliminating routine urinalysis from antenatal care, a quarter of affected women are potentially exposed to the risk of ongoing undetected pre-eclampsia for up to four weeks. Murray et al suggest that most women would be detected by risk assessment at the initial visit. While there are known risk factors for pre-eclampsia, the largest group who develop pre-eclampsia are primigravid women with no previous history of note. In addition, there are no reliable studies to justify this claim of Murray et al. Recent randomised controlled trials have suggested that the frequency of antenatal visits could be reduced without demonstrating harmful effects.5 None of these trials, however, have the power to demonstrate safety. Even the recent large World Health Organization trial involving 24 526 women6 would have required 490 000 women for 80% power, and 649 910 women for 90% power, to exclude the 40% increase in maternal mortality actually observed with reduced antenatal care. Changes in well-tried methods of antenatal care need to be assessed with more stringency before it is concluded that they are safe.

Deirdre J Murphy · Christopher W Redman

In reply: We agree with Murphy and Redman that pre-eclampsia remains an important disorder and a major cause of maternal and perinatal mortality. However, we disagree with their interpretation of our data. Murphy and Redman allege that, by eliminating routine urinalysis, a quarter of cases of pre-eclampsia may remain undetected for up to four weeks. As we pointed out in our article,1 three of the six women who developed dipstick proteinuria before they developed pre-eclampsia were already considered "at risk" for pre-eclampsia — two because of multiple pregnancies and one with a history of prior pre-eclampsia. These women would, in our practice, continue to have routine urine tests during their pregnancies. The argument is then whether it is justifiable to undertake repeated urinalysis in almost 1000 women to detect three who have dipstick proteinuria, but no other warning signs before the onset of their hypertension. In practice, we would have had to increase antenatal clinic visits for 338 women with dipstick proteinuria (most of whom will have had false-positive results2) to detect these three women with proteinuria before they developed pre-eclampsia. It is already our normal practice for women to have antenatal visits every second week in their third trimester. Therefore, it is just as likely that their blood pressure changes would have been detected by our routine surveillance as by the knowledge that they had dipstick proteinuria. We did acknowledge clearly in our discussion that our study had a potential for type II error and that the best approach is a randomised controlled trial of outcomes between those who do and do not have continued urinalyses during pregnancy. None of this belittles the importance of pre-eclampsia, nor the need for us to separate women considered at "low risk" from those considered "at risk" for pre-eclampsia on the basis of well recognised risk factors. The latter group should never be considered among those in whom routine urinalysis can be omitted.

Mark A Brown · Caroline S E Homer · Gregory K Davis · George Mangos

To the Editor: As practitioners in a community-controlled Aboriginal and Torres Strait Islander primary healthcare centre, striving to better meet the healthcare needs of our community, we would like to offer some thoughts on a recent article by Coory et al.1 In reporting on cervical cancer screening participation rates in Indigenous communities in Queensland, Coory et al identified the communities but did not inform them that the study was being conducted. We feel that such an approach is unhelpful and reflects a paternalistic attitude. The accompanying editorial2 rightly drew attention to the lack of direct consultation with the communities involved and the consequences of this in terms of future interventions. Without a true partnership with the women and their communities, the authors were unable to do more than imply that participation rates were better in centres with a primary-healthcare approach to screening. Information about the types of services and choice of Pap smear providers available in each community, in addition to the Pap smear screening participation rate, would be of great interest. It would, in part, answer the question the authors themselves posed about the role of primary healthcare in improving Pap smear participation rates. Already scarce funds could then be committed to improving access to quality primary healthcare rather than to conducting further trials. A further factor, which was not explored in either the study or the editorial, was health promotion. The cornerstone of health promotion in the Indigenous community remains the "kit-video" model, comprising posters, leaflets and a video. In contrast, mainstream health promotion often involves expensive multimedia campaigns promoted by national celebrities. Such campaigns have been shown to increase Pap smear screening in the wider community.3,4 To date, these campaigns have rarely had an Indigenous focus, and it might be argued that a consequence of this is the low Pap smear participation rate documented by Coory et al. In our experience,5 the Indigenous community does respond to culturally appropriate holistic primary healthcare. The study by Coory et al demonstrates a lack of commitment, collaboration and innovation — essential features for the delivery of quality primary healthcare — that is all too common in the current approach to Indigenous health.

Katie S Panaretto · Sarah Larkins · Vivienne Manessis

In reply: Black has argued that the role of evidence in policy development is to create concern and set agendas.1 This was the aim of our article. Although there has been a national, organised approach to preventing cervical cancer since 1991, our study found that screening rates for Indigenous women still lag well behind those of non-Indigenous women. Workers in Indigenous health might have suspected as much, but valid data have not previously been available for a wide geographical area. As Murray and Lopez point out, the lack of good data on a health issue is often taken to mean that the problem is not important.2 We agree that the development of effective programs should be done in consultation and partnership with Indigenous communities. However, there is an additional need to influence decision-makers and budget-holders at national, State and regional levels. An evidence base that identified effective interventions would facilitate this process. Such evidence need not come from randomised controlled trials. On the other hand, case reports of successful interventions in a single community that cannot be sustained when key personnel leave are not very persuasive. It is also useful to demonstrate that the situation is not hopeless. The encouraging finding from our study was that the participation rate in cervical cancer screening was more than 50% in three of the 13 communities studied.

Michael D Coory · Patricia S Fagan · Jennifer M Muller · Nathan AM Dunn

To the Editor: The beyondblue guidelines for treating depression in primary care by Ellis and Smith1 are intended to assist both healthcare professionals and consumers. While they provide several helpful indications, they also include some misleading suggestions. The authors state that drug treatment of depression should continue for at least one year for a first episode of depression, and at least two years for repeated episodes or when there are other risk factors for relapse. However, no background literature is cited in support of this statement, and indeed would be difficult to find. Maintenance pharmacotherapy has been advocated as an effective tool for reducing relapses and recurrences in major depression.2 A number of studies have shown the superiority of antidepressant drugs (mostly tricyclics) compared with placebo in protecting the patient from relapse. Duration of drug treatment, however, did not seem to affect long-term prognosis once treatment with the drug was discontinued. In clinical terms this means that, whether you treat a depressed patient for three months or three years, it does not matter when you stop therapy with the drug. In fact, after recovery from an index episode of major depression, risk of postdiscontinuation relapse was nearly significantly greater after longer treatment (ρ = 0.37; P = 0.052).3 Further, Ellis and Smith1 fail to mention a vexing clinical problem in maintenance antidepressant treatment: the return of depressive symptoms.3 Dose increase is likely to entail only a temporary solution to the problem, which may occur in up to 57% of patients. However, there is a promising alternative. Treatment of depression by pharmacological means is likely to leave substantial residual symptoms.4 Residual symptoms hinder lasting recovery and are one of the strongest risk factors for relapse. In randomised controlled trials, cognitive behavioural treatment of residual symptoms was found to significantly improve long-term outcome of recurrent depression and to allow discontinuation of drug therapy.4 Preventing recurrence in major depression cannot simply be based on prolonging ongoing pharmacological treatment. The belief that a longer course of treatment will result in a more favourable outcome after discontinuation of antidepressant drug therapy is not supported by research evidence.5 Active collaboration with the patient (in choosing a treatment option, in lifestyle modification, in seeking treatment again when needed) is a crucial, and yet neglected, variable. It can lead to a more rational use of antidepressant drugs and to therapeutic efforts of more enduring quality than those prevailing today.

Giovanni A Fava · Chiara Ruini · Eliana Tossani

In reply: Fava observes that duration of treatment with an antidepressant does not affect the subsequent rate of relapse. Indeed, it would be unexpected if it did; medication only works while it is being taken. He then states that the duration of antidepressant treatment is immaterial. However, Figure 1 of the reference he quotes1 indicates that, for at least 54 months after the index episode, continuing medication provides greater protection against relapse than early discontinuation. This argues strongly for the beneficial effects of continuing antidepressant therapy for a significant period after recovery. The duration of this period depends on the number of previous episodes of depression and is a compromise between the benefits of effective prophylaxis and the burden of treatment. He also refers to a study of "tolerance" to 20 mg fluoxetine.2 Of patients who responded to treatment, 31.4% relapsed while on maintenance treatment. Of these, 57% responded to an increase to 40 mg fluoxetine and remained well for six months. He appears to have interpreted these data differently. I agree that treating depression involves more than prescribing. The beyondblue guidelines promote active collaboration with the depressed person in addressing key issues in their lives and a focus on relapse prevention.3 Cognitive behavioural therapy is one strategy for achieving this.

Pete M Ellis

To the Editor: Van Der Weyden has commented on the World Health Organization's Utopian definition of "health", first promulgated in 1948.1 In 1973, I addressed this matter in a speech to the All Nations Club in Sydney, and again in 1980 when presenting a paper to The Hope Foundation (a large charitable health organisation in the United States). For what it is worth, my suggested definition of "health" was: Health is a high level of physical, mental and social comfort appropriate to the age of the person concerned and attainable within the economic constraints of the particular environment. It seems we have some agreement, and I would like to see the WHO again consider definition!

Keith S Jones

In reply: I thank Sir Keith for his pragmatic proposal. The concept of health has individual and societal connotations. At its most basic level, it is the avoidance of pain and suffering. More broadly, it is a basic human resource for the pursuit of life's goals.1 But, as argued by Lewis and Leeder, "health is a good to be pursued, but not an absolute one" for the "functioning of social institutions does not require perfectly healthy citizenry, nor does the individual have to be perfectly healthy to take part in social life."2 As such the perfect definition of health espoused by the WHO is Utopian and removed from reality. It is, as poignantly captured by René Dubos, the "mirage of health", as "complete freedom . . . from disease is but a dream remembered from imaginings of a Garden of Eden."3

Martin B Van Der Weyden

To the Editor: The editorial by Cameron and Campbell on access block is an excellent summary of the causes and potential solutions to access block.1 The effects of overcrowding in the emergency department (ED) have been previously reported, including risks to patient safety, prolonged pain and suffering, and decreased clinical productivity and effectiveness.2 Access to emergency care is impaired. The fundamental problem is that while demand has increased, the capacity of the system has decreased.3 This is best exemplified by the significant reduction in hospital bed numbers. Healthcare in Western societies has been through a period of severe economic rationalisation, resulting in closure of thousands of beds in the acute hospital system. For example, in the United States, the number of medical and surgical beds declined by 18% in the period 1994–1999. From 1990 to 1999, attendances at EDs increased 15%. There have also been many aged care beds closed or changed to community-based facilities.4 It has been suggested that the problem is "a badly flawed approach to financing health care that values profits over patients."5 Spare bed capacity is essential for the effective management of emergency admissions. At least one study has found that if hospital bed occupancy rates exceed 85%, then bed crises occur.6 In fact, the Guinness Book of World Records now has a category for longest wait on a hospital trolley!7 The official record currently stands at 77 hours 30 minutes, although anecdotes report longer times. It is paradoxical that other departments within a hospital cannot exceed 100% occupancy, and yet the ED, which may contain some of the most seriously ill or injured, is allowed to exceed the safe level of 100% occupancy. The ED has always been available to help if all else fails in the healthcare system. That basic tenet is now being challenged, and the general public may no longer be able to rely on EDs for quality and timely emergency care, placing the safety of people at risk.8 In addition, Derlet has stated that should there be a major infectious disease epidemic or national catastrophe, EDs and hospitals could not accommodate the demand, undoubtedly leading to increased suffering and excess mortality.3 For all patients, increasing the capacity of the hospitals across the system (viz beds) would really make a difference.

Daniel M Fatovich

To the Editor: The series of articles concerning access block1 indicates that access block is a major health issue in this country. It is remarkable that, despite all these efforts to avoid admissions and reduce inpatient length of stay, only a few occasions of brief success at reducing ambulance diversions were described, and only one case of reducing access block (Royal Melbourne Hospital). We contend that it is now time to increase available beds. In several cases, the association between worsening access block and closure of beds was noted (Australian Capital Territory, Queen Elizabeth Hospital, Royal Perth Hospital). Available data show a steady reduction in hospital beds per thousand population in Australia, from 3.3 in 1995–96 to 2.8 in 1999–2000,2 a decrease of more than 11%. Keeping inpatient occupancy below a threshold percentage is important to controlling access block.3,4 Reducing occupancy by increasing available beds is the logical recommendation.

Peter A Roberts · Paul A Cunningham

To the Editor: I was interested in the letter by Marks et al,1 indicating that the efforts of over-worked medical staff in emergency departments to introduce patients to local general practitioners had been largely unsuccessful. A few years ago I noted the success with which this problem was handled by the emergency department management at Huddinge University Hospital in Stockholm. All patients were charged 60 krone at triage. Those who sat in the waiting room were confronted by two large electronic signs. The first listed the waiting time for the 10 most common GP-type ailments. The second listed 10 local GPs, where the consultation fee was then 50 krone, with the offer to refund their initial payment if they chose to take their business elsewhere. I was told that this was the very successful first of eight "barriers" between the emergency department door and the intensive care unit. Since the middle of last century, Sweden has been held up as a model provider of an egalitarian and "free" healthcare service. Perhaps our country could benefit from the revisions and improvements that the Swedes have made over recent decades.

Peter J Burke

To the Editor: An 8-month-old girl, born in Perth, Western Australia, who received only breast milk feeds for her first six months of life, was referred urgently to Princess Margaret Hospital for Children (the tertiary paediatric hospital in Western Australia) for investigations of seizures. Her parents described the seizures as episodic, involving all limbs, lasting less than five minutes and occurring over the 10 days before presentation. At presentation, the infant had carpopedal spasm. She was afebrile, and a septic screen gave negative results. Venous blood gas analysis showed a low ionised serum calcium level of 0.71 mmol/L (reference range [RR], 1.13–1.32 mmol/L). An x-ray film of her wrists showed signs of rickets, with cupping and fraying of the distal metaphyses of both radius and ulna. Other blood tests revealed elevated parathyroid hormone levels of 8.6 pmol/L (RR, 0.80–8.00 pmol/L), an elevated alkaline phosphatase level of 523 U/L (RR, 100–350 U/L), and an extremely low level of serum 25-hydroxyvitamin D3 (25OHD3) of 5 nmol/L (RR, 30–150 nmol/L). The infant responded to treatment with an infusion of 0.5 mg/kg of calcium gluconate and a 4-month oral course of 0.2 μg calcitriol daily. After treatment commenced, she did not have any further seizures. Radiogaphy performed 4 months later showed increasing calcium deposition at the distal metaphyses of the radius and ulna. Seizures are described as a presenting feature of hypocalcaemia in vitamin D deficiency rickets.1,2 Ultraviolet radiation and/or dietary vitamin D are required to prevent rickets in children. Perth, Western Australia, located at latitude 32° South has an average daily sunshine duration of at least five hours per day. However, even with this amount of sunshine, vitamin D deficiency can still occur in people who, for various reasons, receive little or no sun exposure, especially if their skin is darkly pigmented, and who have an inadequate dietary intake of vitamin D. The girl's mother, who wore a veil and clothing which protected her body from exposure for religious reasons, had a serum 25OHD3 level of 7 nmol/L. Breast milk is a poor source of dietary vitamin D, especially when the lactating woman is vitamin D deficient.3 In Australia, vitamin D deficiency rickets in infants of parents who have migrated from Mediterranean, African, Middle Eastern and southern Asian regions has been reported since the 1960s.4,5 Education of healthcare providers and their patients about the requirement for sunlight exposure or dietary supplementation to prevent vitamin D deficiency rickets needs to continue.

Graeme H Johnson · Francis Willis

Comment: Nutritional rickets is highly prevalent in countries such as Mongolia, Tibet and China1 where winter sunlight is reduced and there is no universal vitamin D supplementation. Paradoxically, rickets is also prevalent in developing countries in the tropics and subtropics where sunlight is unlikely to be a limiting factor. Low calcium intakes (including vegetarian diets), prolonged breast feeding, and covering of the skin may all contribute.2,3 Published reports and clinical experience in Sydney suggest an increase in prevalence of rickets, especially in infants and mothers in immigrant populations2,4-7 Other Western countries have reported similar findings. The vitamin D deficiency described by Johnson and Willis in an infant in Perth highlights a high-risk group — infants of mothers who are veiled. Treatment of associated nutritional deficiencies, especially iron deficiency,4,5 and giving a minimum of 300 000 IU of vitamin D over 6–8 weeks, should resolve the rickets. Data on the epidemiology of vitamin D deficiency in these high-risk groups in Australia and other Western societies are lacking and should be the subject of future research. The major source of vitamin D and its circulating form, 25-hydroxyvitamin D3 (25OHD3), in children and adults is the skin. It is estimated that exposure to sunlight for 15 minutes three times per week normalises 25OHD3 levels.8 Dark skin, increasing age, sun protection agents, and the angle of the sun in winter will attenuate this increase in 25OHD3.8 Neonates acquire their vitamin D3 stores from their mothers via the placenta, with only a small amount transferred in breast milk.9 By screening high-risk pregnant women, specifically veiled women and those with dark skin,10,11 prevention of most cases of infant rickets is possible. Levels of 25OHD3 should be measured and, if low, the mother should receive 4000 IU of vitamin D daily until 25OHD3 levels are normal. There is currently no recommendation for routine supplementation of vitamin D in infants, and most cereals and foods are not fortified with vitamin D. However, infant formulas are supplemented with 200 IU of vitamin D per litre. It is estimated that sufficient vitamin D levels to prevent rickets could be achieved if 400 IU of vitamin D were provided daily as part of a multivitamin supplement to high-risk infants.

Christopher T Cowell

To the Editor: It has been well established that improved compliance in hand washing significantly reduces hospital-acquired infections and cross transmission of methicillin-resistant Staphylococcus aureus.1 Although there have been no reports of infections resulting from cross-contamination via stethoscopes, studies have demonstrated that 80%–100% of these appliances are colonised by bacteria.2-5 However, most of the organisms isolated are considered non-pathogenic. The most common potentially pathogenic organism isolated from stethoscopes is S. aureus, with a prevalence of 4.2%–27.5%.2-5 Cleaning with either 70% isopropyl alcohol or benzalkonium chloride wipes can reduce the bacterial count on stethoscopes by 94%–100%.3,5 We undertook a study to assess the prevalence of S. aureus carriage on stethoscopes and hands of staff of the Canberra Hospital, and to measure the effectiveness of cleaning and hand washing in reducing colonisation. A convenience sample of healthcare workers from various areas of the hospital was obtained over a period of 6 months. The diaphragm of each participant's stethoscope was directly impressed on to mannitol salt agar, before and after being cleaned with a 70% isopropyl alcohol wipe. The dominant hand of each participant was also tested before and after washing with triclosan (1%) antimicrobial handwash and water. S. aureus was identified by standard laboratory methods. There were 134 participants: 69 doctors, 50 nurses, 10 medical and nursing students and five physiotherapists. Most doctors and physiotherapists used their own stethoscopes, whereas most nurses and students used ward stethoscopes. S. aureus was isolated from five stethoscopes before cleaning (4%), but from none after cleaning. The organism was also isolated from the hands of 11 people before hand washing (8%) and of one after hand washing (0.7%). Two people had S. aureus isolated from both sites. There was no statistically significant difference between the prevalence of S. aureus on hands and stethoscopes (P = 0.15, McNemar's test). Hand washing is the best recognised means of preventing cross-contamination in hospitals. However, the simple intervention of cleaning stethoscopes with an alcohol wipe was highly effective, and we believe that this practice should be more widely promoted.

Karina J Kennedy · Dianne E Dreimanis · Wendy D Beckingham · Francis J Bowden

To the Editor: At the Australasian Schizophrenia Conference in Sydney in October 2002, Professor Patrick McGorry of the Orygen Research Centre, University of Melbourne, presented draft guidelines on the management of schizophrenia and early psychoses.1 One of the recommendations was that atypical antipsychotic drugs should be used as the first-line pharmacological treatment in preference to typical antipsychotics and depot antipsychotics. With a shift in management of schizophrenia to community-based care, the number of patients with schizophrenia managed by general practitioners has increased over the past decade (from 36 per 10 000 encounters in 1990–91 to 45 per 10 000 in 2000–02).2 With the pending introduction of the guidelines, a baseline measure of GP prescribing rates of antipsychotics, both typical and atypical, will allow future measurement of the impact of the guidelines. We analysed the 1998–2002 data from the Bettering the Evaluation and Care of Health (BEACH) program, a continuous national cross-sectional survey of general practice.3 About 1000 GPs participate in this program every year, each providing details (on structured forms) about 100 consecutive patient encounters. Data collected include GP and patient characteristics, problems managed and treatment provided. We examined 401 300 encounters from 4013 GPs, with 431 537 medications recorded. Prescription rates were calculated and regression analyses performed using SAS software4 to adjust for the cluster effect of the study design. There were 1988 schizophrenia or psychosis problems managed in the four years of data collection (a rate of 49.5 per 10 000 encounters); 1883 medications were prescribed (94.7 per 100 contacts), of which 926 (49.2%) were typical antipsychotic drugs and 484 (25.7%) were atypical antipsychotic drugs. In 1998–99, the prescription rate of atypical antipsychotics was 15.7 per 100 contacts with patients with schizophrenia or psychosis (95% CI, 11.8–19.7). This rate increased to 31.1 per 100 contacts (95% CI, 26.3–36.0) in 2001–02. In the same period, the prescription rate for typical antipsychotics fell from 51.3 per 100 contacts (95% CI, 45.6–56.9) in 1998–99 to 40.6 per 100 contacts (95% CI, 35.4–45.8) in 2001–02. Linear regression showed that the prescription rate of atypical antipsychotics had increased by an average of 5.1 per 100 contacts per year over the four years (P < 0.0001), while the prescription rate of typical antipsychotics had decreased by an average of 3.8 per 100 per year (P < 0.005). Over the four-year period, there was no significant increase in the rate of overall prescriptions for people with schizophrenia (92.0 prescriptions per 100 contacts [95% CI, 83.4–100.7] in 1998–99 v 96.3 [95% CI, 89.4–103.3] in 2001–02). Between 1998 and 2002, the relative prescribing rate of atypical antipsychotics for schizophrenia and other psychoses in general practice nearly doubled. These results show that, even before the introduction of the guidelines, there has been a shift towards prescribing atypical antipsychotics in preference to typical antipsychotics. This change may reflect a change in specialist behaviour, as specialists have a direct effect on GP prescribing.5 The BEACH study will be able to assess the effect of the guidelines on the prescription rate of atypical antipsychotics by GPs.

Christopher M Harrison · Helena C Britt