Letters

To the Editor: Nordin and colleagues raised important issues about prescribing for osteoporosis.1 We agree that the Pharmaceutical Benefits Schedule guidelines for therapy are imperfect, but they do not necessarily lead, as Nordin et al claim, to inappropriate prescribing. For historical reasons, osteoporosis is held to be synonymous with vertebral fractures, but this misrepresents the epidemiology of fractures. Non-vertebral fractures account for 80% of all fractures and 90% of the loss of quality of life and economic costs. Vertebral fractures contribute only 20% of the burden.2 Most fractures arise in the large population at moderate risk with osteopenia — the “bell” of the Gaussian bone mineral density (BMD) distribution, not its “tail”, which comprises those with osteoporosis (defined by a bone densitometry T-score less than – 2.5). Concentrating on vertebral fractures and screening for osteoporosis with bone densitometry, as recommended by Nordin et al, is no solution to this public health problem. Nutritional change and exercise are appealing because they are safe and cost-effective approaches for early intervention, but are supported only by level D evidence (expert opinion).3 Although these approaches are plausible, no trials demonstrate their antifracture efficacy. There are no means of early identification of individuals who will sustain a fracture. Densitometry is neither sensitive nor specific for fracture; most people with osteoporosis do not sustain a fracture, and most fractures arise in people without osteoporosis, who would, paradoxically, be excluded from treatment by screening.4 Bone densitometry should be more accessible for case finding, but its use for screening does not reduce the fracture burden because of this screening paradox. However, Medicare reimbursement for densitometry is available for high-risk individuals (those with premature menopause, other illnesses or who are taking corticosteroids), not just for those aged over 70 years or those with fractures. Restricting treatment on the basis of BMD results is not advocated by the Australian and New Zealand Bone and Mineral Society precisely because it excludes this moderate-risk group from treatment, particularly those with fractures and osteopenia. There is level A evidence (meta-analysis of multiple randomised trials)5 for the antifracture efficacy of bisphosphonates in patients with osteoporosis, and evidence based on single trials6 of their antifracture efficacy in those with osteopenia and prevalent fractures, whose fracture risk is similar to that of people with osteoporosis and no prevalent fracture. There is limited evidence of antifracture efficacy of bisphosphonates in individuals with osteopenia alone.6 Preventing the first fracture is important, and guidelines are deficient in this way. Case finding to estimate absolute risk is the best approach available at this time, using risk factors, remodelling markers and, more recently, microstructural analysis to improve sensitivity and specificity. Rather than inappropriate or overprescribing, evidence suggests underutilisation of drug therapy for osteoporosis.7,8 Osteoporosis remains underdiagnosed, underinvestigated and undertreated, and limiting access to bone densitometry is not supported by the Australian and New Zealand Bone and Mineral Society.

Ego Seeman · Mark A Kotowicz · Peter T Nash · Philip N Sambrook

In reply: Seeman and colleagues agree that most patients with minimal trauma fractures do not have osteoporosis. The figures are clear: only 13% of patients with a peripheral fracture have a hip bone mineral density (BMD) T-score less than or equal to − 2.5 and only 25% have a score less than or equal to − 1.5. The corresponding figures for vertebral fractures are 25% and 38%, respectively.1 We do not argue that the − 2.5 T-score threshold for defining osteoporosis is sacrosanct, but simply that some bone density threshold be defined for subsidised therapy, for which virtually all the supporting evidence is based on treatment of patients with established osteoporosis. Osteopenia is an artificial concept with an arbitrary definition, but we agree that the T-score threshold for subsidised therapy need not be as low in those with prevalent adult fracture as in those without — perhaps − 1.5, which is the threshold recently adopted for patients receiving corticosteroid therapy. We disagree about the predictive power of bone densitometry; it is comparable to that of blood cholesterol level for heart attacks and blood pressure for stroke.2 It therefore makes sense to measure BMD in all women at menopause and all men at age 60 years to identify those with osteoporosis before they sustain fractures, as well as those with normal but negative T-scores, who have a fracture risk twice that of those with positive T-scores.3 Those with proven osteoporosis could receive subsidised therapy, and those with low normal values could be advised on lifestyle measures, such as calcium supplementation (which significantly delays or prevents bone loss in postmenopausal women).4 People with positive T-scores can be reassured. To suggest that no trials have demonstrated the antifracture efficacy of nutritional measures is to argue against three large meta-analyses showing significant prevention of fractures with vitamin D and calcium supplementation.5-7 The additional cost of confirming low bone density before providing subsidised therapy in fracture cases is likely to be more than offset by the savings from reduced inappropriate therapy; bone densitometry costs about $80 per test, but bisphosphonate therapy costs about $50 a month for each patient. The extra cost of bone densitometry for every woman at menopause and every man at age 60 years could be $20 million a year, but even with subsidised therapy for those without fracture but proven osteoporosis (with a T-score less than or equal to − 2.5, for instance), the cost is also likely to be more than offset in the long term by reducing the enormous cost of osteoporotic fractures ($8 billion annually8). We find it hard to understand why any of our colleagues would not support proposals that would transfer treatment from those who do not need it to those who do.

B E Christopher Nordin · Michael Horowitz

To the Editor: The ASPREE (ASPirin in Reducing Events in the Elderly) study may provide useful data on the benefits and risks of aspirin therapy in patients aged ≥ 70 years, as described by Woods and colleagues.1 However, the decision to allow general practitioner co-investigators to “help decide whether the patient is a suitable candidate for the placebo-controlled trial” introduces a source of selection bias that may limit the generalisability of the results. Without pre-specified objective selection criteria, it is likely that primary-prevention patients assessed by GP co-investigators as being at high vascular risk will be excluded because the GPs believe they should be taking antiplatelet agents. Similarly, those at low risk may be thought inappropriate participants because the risks of random allocation to this therapy might outweigh the perceived benefits, as has been shown in previous meta-analyses.2,3 ASPREE may end up with a disproportionate number of intermediate-risk patients. In the case of diabetes, a recent observational study from our group highlighted patients with diabetes and retinopathy and those taking a sulfonylurea as being at increased risk of complicated peptic ulcer disease.4 By contrast, we did not find that aspirin use, positive serological results for Helicobacter pylori, or the interaction of these two factors predicted complicated peptic ulcer disease. If GP co-investigators were aware of these findings, they might also influence the screening and recruitment of patients with diabetes to ASPREE. According to the trial registration details (ISRCTN83772183), patients with diabetes were eligible for recruitment to ASPREE from late February 2009, even though the trial started 6 years ago.5 Given this delayed eligibility, the fact that a substantial proportion of patients with diabetes older than 70 years will already have vascular disease, and the expected total sample size of 19 000,1 the trial might include fewer than 1000 patients with diabetes and thus have insufficient statistical power to assess the risks and benefits of aspirin for primary prevention in this important subgroup. We question why subjective assessment forms part of patient selection for a potentially important study such as ASPREE, and also what steps the investigators are taking to determine whether the sample they recruit is representative. In addition, details of planned statistical analyses involving diabetic participants in this non-superiority trial would be reassuring.

Timothy M E Davis · Brett A Sillars · Wendy A Davis

In reply: Recruitment to clinical trials through general practice is representative of the population, as a high proportion of all Australians regularly attend their general practitioners.1 GP co-investigators are appropriate to decide whether their patients are suitable for the ASPREE (ASPirin in Reducing Events in the Elderly) study because their assessment includes objective inclusion and exclusion criteria that must be satisfied before enrolment in the study (clinical trial registration number ISRCTN83772183),2 as well as patient-specific potential risks with using aspirin, and known medical factors likely to influence patient survival during the trial. These include the risk of complicated peptic ulcer disease in patients with diabetes treated with a sulfonylurea.3 GP co-investigators support participation in ASPREE by eligible patients because of aspirin’s therapeutic equipoise for primary prevention in older patients4 and in those with diabetes.5 Because of age alone, ASPREE participants will be at least at intermediate risk of cardiovascular disease and also at increased risk of bleeding. Determining the aspirin balance underpins the importance of collecting more data in older people, who have been under-represented in previous primary prevention trials. ASPREE is a superiority trial with pre-specified subgroup analyses, including for the subgroup with diabetes.2 The study is powered to address the primary question reliably in the total cohort rather than subgroups. To date, fewer than 500 participants have been randomly allocated, with recruitment slowed subject to National Institutes of Health funding deliberations. Recruitment will be reinvigorated in late 2009, and will continue to include people with diabetes.

Robyn L Woods · Mark R Nelson · Andrew M Tonkin · Christopher M Reid

To the Editor: Wiseman and colleagues reported the 9-month virological follow-up of babies born to pregnant women attending urban antenatal clinics who tested positive for hepatitis B surface antigen (HBsAg).1 Evaluation of outcomes is vital to determine implementation and effectiveness of current policy. The authors state that hepatitis B immunoglobulin (HBIG) and hepatitis B virus (HBV) vaccine were delivered within 12 hours of birth to all infants of HBsAg-positive mothers. Nine months after birth, transmission was documented in only four infants, all of whom were born to mothers who were positive for hepatitis B “e” antigen (HBeAg) and had very high HBV DNA levels (> 108 copies/mL). Yet one of the four infected infants had inadvertently not received the “routinely offered” HBIG. Whether any other babies (even if not infected) also failed to receive timely active and passive immunisation was not explicitly reported. The other three infected infants completed HBIG injection and HBV vaccination according to the “recommended schedule”. Given that the Australian immunisation handbook states that [t]he first dose of monovalent hepatitis B vaccine should be given at the same time as HBIG ... as soon as possible — preferably within 24 hours of birth, and definitely within 7 days,2 the exact time is unclear. Potentially, a delay of many hours to many days could occur. Yet, for all infectious diseases for which post-exposure prophylaxis is given, administration as soon as possible after exposure is universally recommended to prevent transmission.3 For babies born to known HBeAg-positive mothers, particular effort is warranted to ensure that they receive HBIG and HBV vaccine immediately after delivery, and certainly within hours of birth. There are many difficulties in achieving high rates of follow-up in this group, and only 66% of babies in the study by Wiseman and colleagues had virological follow-up results available. Unfortunately, the authors did not make the important distinction between babies lost to follow-up and those who were not yet 9 months old at the time of follow-up. This is in spite of the fact that the study included many South-East Asian women, who have markers of high HBV replication and whose follow-up is suboptimal.4 Evaluation of HBV vaccination policy requires that studies report fully on outcomes, including loss to follow-up. Also, given that delay in administration of HBIG and HBV vaccine after delivery may be a critical source of variation in outcome, future studies should specifically report this information separately for all births to HBsAg- and HBeAg-positive mothers, and for any cases in which transmission does occur.

Sandra C Thompson

In reply: We thank Thompson for her comments. We endeavoured to contact women by phone, by letter, or through their local doctor (if known). With respect to the 69 infants lost to follow-up, five of the mothers declined assessment, four had moved interstate or overseas with their families, five had infants who had not reached 9 months of age, and the remaining mothers were uncontactable. Follow-up of a cohort of this kind is difficult, for reasons that include language barriers and high rates of mobility. Furthermore, at the time of the study, testing the babies was not standard practice and not enthusiastically received by all mothers. We are not a public health unit and are not resourced to pursue follow-up beyond the measures described. Nevertheless, through strong engagement with the mothers, who were referred on the basis of their hepatitis B virus (HBV) infection, we achieved follow-up for a large number of babies. The data stand as valuable but not perfect. As mentioned in our article,1 vaccination practice for babies born to infected mothers is for hepatitis B immunoglobulin and the first HBV vaccine dose to be given (in opposite thighs) within 12 hours of birth. The babies’ vaccination records were checked by consulting the “blue book”, a standardised medical record for infants, and, for those infected, by cross-checking with the hospital medication chart.

Miriam T Levy · Elke Wiseman

To the Editor: I was interested to read the results of Sweidan and colleagues’ study of drug interaction alerts in prescribing and dispensing software.1 I believe their use of the terms “sensitivity” and “specificity” differ from the standard definitions, which are usually: Sensitivity = true positives ÷ (true positives + false negatives) Specificity = true negatives ÷ (true negatives + false positives)2 To test sensitivity and specificity, one requires a dataset that includes positives and negatives. I do not view “minor interactions” as a complete set of negatives, because a complete set of negatives should include a statistically valid number of randomly chosen drug sets without interactions. Minor interactions do not meet my criteria for “negatives” because, to me, a minor interaction is still an interaction that may sometimes be clinically significant. A true negative should meet the test of “never clinically significant”. Some reported minor interactions would meet that test and some would not. I believe sensitivity and specificity data should be reported for both major and minor interaction alerts. I also believe there should be some alignment of definitions between “drug interaction” research and “adverse drug event” research.3 Bates and colleagues talked about “adverse drug events” and “preventable adverse drug events” in 1995.4 Of most clinical interest are the preventable adverse drug events, which could be minimised by the use of appropriate decision support.5,6 Certainly, there is a need for independent assessment of the quality of electronic prescribing decision support systems. A robust assessment methodology is required to permit potential government regulation of such resources, which are of national and community importance.

Ian R Cheong

To the Editor: As the former Clinical Information Specialist Manager for the MIMS DrugAlert knowledgebase (from 2003 to 2005), I write in response to the study by Sweidan and colleagues examining the quality of drug interaction alerts in prescribing and dispensing software.1 The authors point out that the success of any knowledgebase in providing clear, correct and specific alerts at the point of care is subject to the quality of its technical integration into decision support software. I would like to add that the sensitivity of drug interaction decision support is determined largely by the knowledgebase, while the specificity of the system is subject to the intelligence of the software in which it is employed. I would be interested to know if the low specificity that Sweidan et al found for the MIMS DrugAlert database was due to lack of use of the severity or level of evidence settings, or having these set at inappropriate levels. The MIMS DrugAlert knowledgebase was in some ways a unique decision support database, written by Australian staff for use in Australia and New Zealand. It soon became one of the largest commercially available databases of drug interaction information in the world, covering over 4600 drug-class and individual drug interactions. Its writing alone was a remarkable feat, being completed in a matter of months and further expanded over a subsequent 18-month period. There are significant variations in practical advice between American and European sources of drug interaction information. In writing the MIMS DrugAlert database, we sought to communicate “the right information, at the right time, in the right way” to local professionals. The foundations of MIMS DrugAlert were based on a clear understanding that we would be representing relatively simple pharmacological principles through the structure and content of a relational database. This meant creating interacting drug classes reflective of the pharmacological properties of groups of drugs, rather than simply grouping drugs based on their chemical families alone. Sweidan and colleagues should be congratulated on highlighting the need for comprehensive, accurate and useful information that can reduce medication error and save lives at the point of care. What is lacking is a clinical outcomes-based study focusing on the real-world benefits that can be achieved if the right system can be implemented in the right way, at the right price. Perhaps this type of research would then build on the excellent, basic foundational research carried out by Sweidan et al.

Bryan W Tan

In reply: Cheong notes that the terms “sensitivity” and “specificity” have a slightly different meaning in our study compared with the usual definitions. This was intentional, and the definitions we used are clearly described in our article.1 Our definitions for sensitivity and specificity were based on two important practical considerations. First, an electronic prescribing system should alert the clinician to potentially clinically significant drug interactions (“true positives” by our definition); and second, a system should not inundate clinicians with alerts containing irrelevant or unhelpful information about minor or clinically unimportant interactions (“false positives” by our definition). We know that these latter alerts can cause “alert fatigue” and are a subject of complaint for doctors and pharmacists. We are not aware of there being any problem with prescribing systems producing alerts for pairs of drugs that do not interact at all; hence, we did not investigate this group. Tan questions whether the low specificity we found for the MIMS DrugAlert database may have been due to inappropriate severity level settings. Although it might seem appealing to reduce the number of alerts by allowing users to “switch off” drug interaction alerts that are classified as low severity, there are difficulties in doing this because of a lack of evidence for the application of severity ratings to drug interactions. Severity ratings are subjective — studies have shown there is little consensus on such ratings between major reference sources.2,3 This is not surprising, given that there is little evidence available on adverse clinical outcomes resulting from drug interactions, and also because the clinical outcome is context-dependent according to variables such as patient characteristics and drug dosage. We believe that, rather than relying on software vendors or users to switch off some alerts, drug interaction knowledgebases should include only potentially clinically significant interactions. We tested all systems at the lowest severity setting (if available) for consistency, and to maximise detection of drug interaction alerts. For minor interactions, the rating was based on both presence of the alert and quality of the information — if a minor interaction was either not detected or was detected and provided appropriate information indicating it was minor, then it was a “pass”.

Michelle Sweidan · James F Reeve · Jo-anne E Brien · Pradeep Jayasuriya · Jennifer H Martin · Graeme M Vernon

To the Editor: Francis and colleagues revealed a significant 29-minute decrease in the median emergency department (ED) length of stay and reductions in mean turnaround times for full blood count requests following the redesign of pathology processes.1 Laboratory performance and clinician satisfaction are intrinsically bound up with the timeliness of test results, not least because of their effects on patient diagnosis and treatment. The association between test turnaround times and ED length of stay is difficult to decipher. The reasons for this, as pointed out by Francis et al, include the many potential variables that contribute to patient length of stay in EDs. Turnaround times can also vary dramatically according to laboratory operating procedures and work processes. Nevertheless, there is a strong imperative to monitor these indicators as contributors to designing effective interventions to improve the quality and outcomes of patient care. Our research has also revealed that pathology test turnaround times are a significant contributor to ED length of stay at a major teaching hospital in Sydney. Using regression analysis to account for a number of contributing length-of-stay variables (eg, triage category, patient age, number of tests ordered), we produced a model that accounted for 25.4% of the variance in ED length of stay, of which pathology test turnaround time was a significant contributor.2 Further, studies of the effects of electronic test-ordering systems on test turnaround times in five Australian hospitals have shown that introduction of these systems in each instance was associated with significant declines in test turnaround times.2-4 Thus, there appears to be converging evidence of a relationship between efficient test ordering and processing and reduced ED length of stay.5 In addition to the types of interventions described by Francis et al, electronic test ordering should be considered as a system-wide intervention that may contribute to improved efficiency and patient outcomes. Decision support within such systems provides an added ability to provide alerts and guidance in the test-ordering process.

Andrew Georgiou · Johanna I Westbrook

In reply: A recent report on the state of Australian public hospitals highlights figures of 320.6 emergency presentations per 1000 weighted population, with 88% of presentations being triage categories 1–4, and 66% of patients not admitted.1 In our experience, pathology tests are performed on more than 50% of patients in these triage categories who present to emergency departments, and decisions regarding patient care are delayed pending availability of test results. In addition to the initiatives that we described,2 Georgiou and Westbrook describe other initiatives that deliver significant, sustained improvements in health care efficiency. Many have emphasised the need to improve efficiency in the health care system by utilising evidence-based best practice. However, a recently published discussion paper highlights the challenges associated with ensuring that health care workers have access to, and actually use, evidence-based processes and protocols that have been shown to benefit patients.3 Similarly, delays in care may contribute to patient harm, and decisions about resource allocation and organisational systems do not always explicitly take this risk into consideration.3 These challenges may be partly related to the different funding arrangements in the Australian health care system. Apart from the obvious federal–state dichotomy, diagnostics, such as pathology services, and the associated information technology are often funded separately to the hospitals and health care professionals relying on these services. As a result, one of the challenges of implementing initiatives that improve efficiency relates to current and historical allocative funding arrangements, where costs may be borne by one cost centre, and benefits are delivered to patients and other elements of the health care system (ie, other health care providers and departments). It is essential that the Productivity Commission4 considers all of these elements in its review to identify significant efficiency improvements within the hospital system. Effectiveness, allocative efficiency and dynamic efficiency are just as important as simple economic efficiency. Unnecessary delays in the health care system are not only associated with capital costs — they are also associated with an array of direct and indirect costs related to caring for patients who endure an unnecessarily protracted journey through the health care system. It will take courage and commitment to implement and achieve appropriate use of evidence-based care (perhaps via mandates3) and appropriately fund all elements of patient care, so that improvements in care and overall cost savings are not hindered by historical funding constraints.

Andrew J Francis · Michael J Ray · Mary C Marshall

To the Editor: Grayson and Johnson’s editorial of 17 June1 and Eizenberg’s viewpoint article of 1 July2 both betray blind spots regarding Victoria’s laboratory response to pandemic influenza (H1N1) 2009 (“swine flu”) and the role of the Victorian Infectious Diseases Reference Laboratory (VIDRL). Grayson and Johnson incorrectly suggest that Victoria’s swine flu infection case definition represented a barrier to laboratory testing and detection of disease spread. Between 18 April 2009 (when the second case was identified in the United States) and 18 May 2009 (when the first case was identified in Victoria), VIDRL tested more than 500 specimens for respiratory viruses. All 16 influenza viruses detected were seasonal influenza strains. Swine flu testing was appropriately reserved for cases with a high pre-test probability of being positive, and even in this at-risk population, no positive cases were detected. The same authors highlight VIDRL’s prominent role in public hospital testing for viral diseases, but mistakenly conflate this with public health laboratory support of infectious disease outbreaks, as if concentration of diagnostic virology at VIDRL were a matter of policy. VIDRL’s diagnostic service is available to Victorian health care institutions for as long as they elect to refer specimens. There is no barrier to decentralising this capacity through access to appropriate new technology and scientific expertise. Nevertheless, centralisation of laboratory capability with optimised specimen transport and reporting offers an arguably more efficient model, and is widely used for this reason. On the other hand, VIDRL’s public health role in supporting the Victorian Department of Human Services’ (DHS’s) outbreak response capability represents a deliberate centralisation of capacity — for good reasons. Public health laboratory testing needs enormous surge capacity — on 1 June 2009, for example, we processed 1004 specimens requiring 1401 swine flu polymerase chain reaction (PCR) tests in a laboratory normally doing 100 respiratory tests daily. Public health laboratories also require expertise to develop, validate and perform essential tests during the early high-pressure phase of an outbreak, while also working closely with the DHS to manage the large patient data flows that underpin public health actions. There is nothing visionary about frittering away this crucial response capacity by devolving responsibility to a series of small nodes, each below critical mass. Both Grayson and Johnson and Eizenberg make misleading generalisations about test turnaround times and refer to non-existent backlogs. On 1 June, the most demanding day of the swine flu outbreak, the mean turnaround time for the 1401 PCR tests performed was 24 hours, with 90% of samples being tested within 32 hours of receipt. There are several reasons outside VIDRL’s control why occasional delays might have occurred. Firstly, transport from point of collection to VIDRL was often slow (1–3 days, with few if any public hospitals achieving faster times). Secondly, more than 10% of samples arrived with missing or incorrect information regarding addresses for reports. Eizenberg should note the high number of general practitioners completing a private pathology provider’s request form but sending the specimen directly to VIDRL, causing delay while results went to the apparent referring laboratory. More than 200 specimens arrived with no request forms at all. Finally, many organisations had no mechanism in place for receipt of large numbers of test results late in the evening, when VIDRL was continuing to work. With Victoria’s move to the “Sustain” phase of the pandemic influenza plan on 3 June, test capacity was directed to defined clinically at-risk patients.3 Many hundreds of samples not meeting the criteria for testing continued to arrive each day. Between 5 and 16 June, these were stored but not tested, and an immediate report went to the sender saying so, and explaining why. Although the capacity to test approved samples was never under threat, stocks of key PCR reagents were transiently sufficiently limited in Australia to preclude testing of samples classified as not meriting a laboratory test in the first place. It may be this scenario that Eizenberg tries inaccurately to describe. It is disappointing to be drawn into exchanging correspondence in these pages rather than having a constructive debriefing together at the end of this outbreak. No criticism by us of laboratory colleagues is intended, as we know there is a shared sense of the logistic challenges with which we have grappled, and can improve together. However, the record still needs to be put straight for a small number of physicians with a more limited grasp of the issues.

Michael G Catton · Julian D Druce · Chris J Birch

To the Editor: The increasing overcrowding on public transport, particularly trains, in many Australian cities makes for considerable discomfort. As the problem deteriorates, concerns about attributable illness and even death have been raised.1 Overcrowding on public transport also contributes to the spread of respiratory diseases such as influenza (pandemic or otherwise).2 The risk of contracting influenza is greatest for people who are within 1 m of an infectious person, through exposure to respiratory droplets, particularly for periods of more than 15 minutes.3,4 Seasonal influenza, which infects millions of Australians annually — resulting in an estimated 2000 deaths and 10 000 hospitalisations5 — is a major health concern. As a resident of Melbourne and regular train commuter, personal observation supports recent claims of severe overcrowding,1 with the average number of people sitting or standing within 1 m of another person on Melbourne trains during peak periods having increased markedly over the past few years. If an average peak commuter is now placed within 1 m of 10 people for a prolonged period twice a day (a conservative estimate), at least 100 infectious influenza contacts potentially occur each week. With annual attack rates for seasonal influenza of 5%–10%,5 the likelihood of contracting infection while crammed into under-resourced train networks is significant. In the context of an influenza pandemic, with higher attack rates, the risk is even greater. Substantial community resources have appropriately been invested in pandemic planning and mitigation strategies. Perhaps we should devote some of these resources to public transport services to reduce overcrowding. Our national pandemic plan advises that A very simple way of reducing the chances of being infected or passing on infection is to stand or sit back from other people in public or in the workplace. Where possible, you should try to maintain a distance of at least a metre, which is about a large step.4 Such advice is impossible to follow on peak-period train services, certainly in Melbourne. Investment in train services to reduce the spread of infections would not only help delay the onset of the next influenza pandemic, but would also reduce seasonal influenza and other respiratory virus transmission. Furthermore, it could avert injuries and illness due to crushing and overheating/dehydration, and could conceivably reduce the road toll by taking cars off roads during peak hours. Few pandemic influenza planning investments could deliver such diverse public health dividends while we ride out the latest influenza pandemic and await the inevitable next one.

Benjamin C Cowie

To the Editor: Nosocomial infections caused by Pseudomonas aeruginosa often prove difficult to treat because of their resistance to multiple drugs. Carbapenems play a pivotal role in the management of severe multidrug-resistant gram-negative Enterobacteriaceae and P. aeruginosa infections. However, reports in Australia of carbapenem resistance due to production of a variety of carbapenemases, including metallo-β-lactamases (MBLs), have been increasing alarmingly.1,2 We wish to report the first clinical case of a VIM-1-producing MBL in Sydney. To our knowledge, this is the first reported locally acquired case of a P. aeruginosa strain producing acquired VIM-1 MBL in Australia. The patient was an 81-year-old man with chronic rheumatoid arthritis, managed with prednisone. He had a prosthetic knee infection that was first diagnosed in 1997 and, because of multiple recurrences, had been managed with oral moxifloxacin since 2003. The patient had not travelled outside the Sydney area since before his knee surgery. He was hospitalised in November 2005, when he underwent repair of a colovesical fistula. Carbapenem-resistant P. aeruginosa was isolated repeatedly from both urine and sputum cultures from December 2005 until February 2008, when further testing was performed using newly available molecular real-time polymerase chain reaction (PCR) technology with VIM generic and specific primers and DNA sequencing. This testing detected a VIM-1 gene. The P. aeruginosa isolates were routinely screened for antibiotic susceptibility. Multiresistance to numerous antibiotic classes was detected, with high minimum inhibitory concentrations for meropenem, gentamicin, ciprofloxacin, ceftazidime, cefepime, piperacillin–tazobactam, and ticarcillin–clavulanic acid. The isolates were susceptible to polymyxin B and aztreonam, and had intermediate resistance to amikacin. Multiresistant P. aeruginosa is a therapeutic challenge when managing patients with such infections.3 In the context of facilities such as large burns units or intensive care units, the presence of plasmid-transmissible carbapenem resistance within the gram-negative bacterial population has serious infection control implications. Until novel molecular real-time PCR methods became available, the underlying mechanism of carbapenem resistance in these organisms could not be adequately delineated. The increasing availability of such molecular technology and its routine application in the diagnostic laboratory will support appropriate antibiotic prescribing practices and enhance infection control measures in the hospital setting. Identification of a plasmid-mediated carbapenem-resistant strain is a concern in our hospital and throughout Australia, as outbreaks of VIM-1 resistance have been reported in Europe and the United States.4 As demonstrated by our isolates, such strains have a broad spectrum of hydrolytic activity against amino-, carboxyl- and ureido-penicillins, cephalosporins, cephamycins and carbapenems, but not monobactams. Our isolate was susceptible only to polymyxin and aztreonam.

John Merlino · Harold W Stokes · Elaine Y-L Cheong · Thomas Gottlieb

To the Editor: A 79-year-old man from Gatton, 80 km west of Brisbane (Box 1), presented in July 2008 with an extensive area of cellulitis on the right knee surrounding a central ulcer 2 cm in diameter (Box 2). Gram staining of swabs taken from the ulcer showed polymorphs and gram-negative bacilli. An oxidase-positive, gentamicin-resistant, gram-negative bacillus was cultured. It had the biochemical profile and characteristic colonial morphology of Burkholderia pseudomallei, the causative organism of melioidosis. The identity of the organism was confirmed by polymerase chain reaction. Blood and urine cultures were negative, and a chest x-ray was normal. The patient was treated with intravenous ceftazidime 2 g four times a day for 10 days, as well as oral cotrimoxazole 320/1600 mg twice a day for 6 months. The infection appeared to be localised, and the patient made a successful recovery. Melioidosis can have a wide spectrum of clinical manifestations.1 Skin and soft tissue infections, as seen in our patient, may lead to fulminating systemic infections if treatment is inadequate.1 There were also several factors that predisposed our patient to melioidosis, including type 2 diabetes, renal impairment, and concurrent steroid treatment (for persistent sinusitis). The patient had a history of recent local exposure to floodwaters and soil: he had spent several hours kneeling in wet mud repairing a burst water pipe. Within a few days, an abrasion on the knee had developed into the presenting lesion. The patient denied visiting any areas where tropical melioidosis was endemic. In Australia, melioidosis is generally considered endemic in areas north of 20°S. In subtropical Australia, below 20°S, sporadic endemic infections in domestic animals and humans have occurred in south-eastern Queensland2-6 and south-western Western Australia.7 Three fatal human cases have been reported from the Brisbane River valley (two in 1996, near a reservoir 20 km north of Gatton;4 one in 1999, 5 km from Ipswich city centre5). All three patients were exposed to floodwaters, two had infected skin lesions, and all three progressed to fulminating pneumonia. All three also had alcohol-associated pathology, a recognised comorbidity.1 There were also two less well documented (but apparently local) human cases from the Brisbane region in 19676 and 1974.4 Recent molecular typing of five strains of B. pseudomallei isolated from south-eastern Queensland showed them to be genetically distinct from each other and from isolates obtained from tropical Australia, suggesting that this subtropical focus is most likely a natural phenomenon from ancient times.5 The reticulated water supply to our patient’s house was sourced from the Wivenhoe Dam. The source of the bacteria was most likely the local soil, which is of a heavy clay type suitable for this organism. The Brisbane River valley appears to be a subtropical endemic area for melioidosis, and further sporadic cases can be expected. 1 Location of melioidosis cases reported in Queensland 2 Area of cellulitis around a central ulcer on the patient’s right knee

Roger W Guard · Peter J Morero · Win Yi · Maureen J Mackay

To the Editor: Rickettsia australis is the causative organism of Queensland tick typhus, also known as Australian spotted fever. It is an obligate, intracellular organism that invades endothelial cells, causing vasculitis.1,2 Its transmission to humans is via Ixodes tick species, which can occur along the east coast of Australia, but predominantly occur in the north-east.3 R. australis was previously thought to have a low complication rate; however, severe sequelae such as multiorgan failure, severe pneumonia and digital necrosis have emerged in recent years.1 A 54-year-old woman with type 1 diabetes presented to a rural hospital in Queensland with a 1-week history of vomiting, diarrhoea, rigors, and fevers to 39.4°C. Her blood sugar level on arrival was 31 mmol/L (reference range [RR], 3.0–7.8 mmol/L) and diabetic ketoacidosis was diagnosed. Despite initial treatment, her condition continued to decline, and she was transferred to a tertiary referral centre. Her condition deteriorated into multiorgan failure, requiring ventilation and inotropic support. Triple antibiotic therapy comprising ciprofloxacin, meropenem and doxycycline was initiated. Results of blood cultures for anaerobic and aerobic bacteria and a vasculitic screen were negative. The patient underwent serial chest x-rays, which demonstrated a resolving left lower lobe collapse/consolidation and a right-sided pleural effusion. Paired sera from Day 1 and Day 11 of admission to the tertiary hospital were tested in parallel for antibodies to R. australis. A rise in R. australis antibody titre, from 256 to 1024 (RR, < 32), supported a diagnosis of rickettsial disease. Serum from Day 1 was negative for antibodies to Mycoplasma, Leptospira, and Brucella species. A family conference later suggested that the patient may have been bitten by an insect; however, no suspicious lesion was identified. Her treatment was changed to intravenous doxycycline monotherapy, and slow improvement was noted. Two days after admission, the patient developed widespread bullae and dermal necrosis with large areas of affected dermis sloughing off (Box), as well as digital and proximal foot ischaemia. A skin biopsy showed changes consistent with septic vasculitis. The patient was reviewed by a dermatologist, who concluded that it was unlikely to be from a drug reaction. Thirty-six days after admission, she underwent bilateral below-knee amputations. Formalisation of the stumps was delayed to ensure viable tissue for coverage. Necrotic areas of her lower limbs and arms were also debrided and grafted, and seven of her fingers were amputated at the level of the proximal interphalangeal joint. The patient was discharged to a smaller centre for ongoing rehabilitation and support. This case highlights the possible severity of R. australis infection, which can be complicated by septic shock, coagulopathy, multiorgan failure and digital gangrene. In addition, diabetes and the resultant ketoacidosis contributed to a compromised host and an unusually severe clinical course. Bullae and dermal necrosis on the legs of a patient with severe rickettsial disease

Theo F Birch · Michael Muller

To the Editor: In Australia, patients with type 1 diabetes and private health insurance are eligible for rebates on the purchase price of insulin pumps if deemed necessary for treatment. In contrast, hospital-funded or donated pumps are often used by non-insured patients. Hospitals may provide pumps to certain patients for various reasons — for example, to pregnant women (to improve their glycaemic control), to patients who want to try the pump to determine their preference or their ability to use it, or to patients waiting for private health insurance cover to be activated. Patient selection is important, as insulin pumps are cost-effective only if they reduce levels of glycated haemoglobin (HbA1c) and the frequency of hypoglycaemia1 — although quality of life may also be an important benefit. We conducted a study to compare outcomes for patients with public hospital-funded pumps (Group A) with outcomes for those with private health insurance-funded pumps (Group B). All pump starts between June 2000 and January 2008 at Fremantle Hospital and Rockingham General Hospital in Western Australia were assessed. HbA1c levels before and 6 months after pump initiation were recorded. Diabetes-related hospital admissions over a 1-year period before and a 1-year period after pump commencement were recorded using hospital software (TOPAS KEA! 340, version 5.106) that tracked admissions to all hospitals within the greater metropolitan area of Perth. Patients were excluded from our study if they had type 2 diabetes; had commenced pump use at a different hospital; had moved during the study period to a region not captured on the database; or had used a pump for less than a year (this last exclusion criterion was to ensure that admission rates for the subsequent 12 months were representative of the influence of pump therapy). We identified 109 patients (32 in Group A, 77 in Group B). There were no significant differences between the two groups in age, diabetes duration, initial HbA1c levels (9.2% v 8.7%; P = 0.29) or sex, although the proportion of females was higher in both groups (65.6% and 70.1%, respectively). Patients in Group A had more hospital admissions than those in Group B before and after commencement of pump therapy (0.7 v 0.2 admissions/year before [P = 0.02]; 0.7 v 0.2 admissions/year after [P = 0.04]). After commencing pump therapy, HbA1c levels fell significantly in Group B patients (8.7% v 8.0%; P < 0.005) but not in Group A patients (9.2% v 8.9%; P = 0.17). The mean interval between pump initiation and follow-up HbA1c readings was similar in both groups (10.3 months [Group A] v 10.8 months [Group B]; P = 0.70). There was no significant difference in diabetes-related admissions before and after commencement of pump therapy in either group.

Ken Y Thong · P Gerry Fegan · Bu B Yeap

To the Editor: I have read with interest your recent series of articles on access block.1-3 While access block is clearly bad for patients, it is a measure that is focused on emergency departments and not patients. Access block only measures the first part of the process of admission to hospital. Surely, the best measure of access into a hospital is not the rapidity with which patients get out of the emergency department but how long before patients get into the ward that is most suited for their care. At my hospital, patients are shuffled out of the emergency department into a variety of “holding pens”. By this I mean wards (often temporary) whose sole purpose is to act as an overflow area while the patient is waiting to obtain a bed in the home ward of the medical or surgical unit that is looking after the patient. The purpose is to allow the emergency department to function better, but also to allow the hospital to look good from an “access block” point of view. This is hardly optimal patient care. A recent survey at my institution of 136 medical patients at high risk of delirium showed that over 43% had three or more ward moves during their admission, and 60% had three or more bed moves. No wonder they become confused. Patients who are moved to holding pens have intrinsic disadvantages to their care. They are often seen late in the day by medical teams; the medical and nursing teams are not used to working together; allied health professionals may change from ward to ward, and holding pens often have no allied health staff; patient’s belongings, pathology and radiology requests get lost in the transition to different wards; patient meals can be substandard in holding pens (because they are ordered at short notice); and there are multiple handovers between many different groups of nurses. Medications are missed. So, hospitals should not be allowed to play games with this metric. The time for patients to get into the home ward (or the most appropriate ward) of the admitting medical or surgical team needs to be part of the equation.

Charles P Denaro

To the Editor: Duke and colleagues recently reported the excess mortality and extra bed-days caused by intensive care access block in metropolitan Melbourne.1 Access block is an important patient safety issue, and we report here additional data that support their results. The Australasian Clinical Indicator Report: 2001–2007, published by the Australian Council on Healthcare Standards (ACHS), reported that the national rate of intensive care access block was 5.9% in 2007, a statistically significant (P < 0.001) increase from 5.3% in 2001.2 There were large differences between states, with higher rates in Victoria in 2006 and 2007. In 2007, the rate of cancellation or postponement of elective major surgery due to lack of intensive care beds was 3.1%, and the rate of interhospital transfer was 1.3%.2 Other rates reported were: the rate of discharge from the intensive care unit delayed more then 12 hours (16.6%) and the rate of after-hours (between 18:00 and 06:00) discharge (17.5%).2 The Victorian rate of after-hours discharge from the intensive care unit, calculated from seven participating hospitals in 2007 (21.6%),2 is comparable with the rate of after-hours step-down to a low-acuity ward found in Duke et al’s study — 18.6% for the period July 2004 to June 2006.1 The ACHS Clinical Indicator Program provides national and peer-group benchmarking to health services participating in its accreditation program, the Evaluation and Quality Improvement Program (EQuIP). Through its annual Clinical Indicator Report, the data are analysed and allow identification of unsatisfactory rates and wide variations in practice. Such national data can help health policymakers identify areas for potential improvement in the standards of health care delivery, particularly in areas where indicators address access. Duke and colleagues’ salient article illustrates the importance of such data being utilised for this purpose.

Helen E Stark · Chris N Maxwell · Robert W Gibberd

To the Editor: Cameron and colleagues are to be congratulated on their article outlining strategies that do and do not help improve the access of emergency medical patients to public hospital ward beds.1 The authors fail to mention one strategy that is particularly relevant to rural hospitals, namely, referring privately insured medical patients who present to emergency departments directly to tertiary medical services at private hospitals. This strategy has the dual benefit of providing a hospital bed for a patient in a rural emergency department who requires hospital admission, and relieving some of the external pressures on metropolitan tertiary referral public hospitals to provide beds. Traditionally, private health insurance has been thought of as providing patients with the ability to obtain treatment from the doctor and hospital of their choice. Increasingly, private health insurance is giving patients the ability to choose between staying on an emergency department trolley and being able to access a hospital bed in a timely manner. It is worth noting that privately insured patients who require services that are not provided at our hospital (such as interventional cardiology, neurosurgery, cardiothoracic surgery and faciomaxillary surgery) can sometimes access a bed in a private hospital about 500 km away sooner than we can find them a bed in our hospital prior to transfer to a tertiary public hospital. A significant impediment to this process for some privately insured patients is the inability of private hospitals to perform insurance fund checks out-of-hours. While most private health insurance funds provide internet authorisations and confirmations of a patient’s insurance status to private hospitals, some do not. In one case, this meant the wife of a patient with unstable angina pectoris had to provide a cash guarantee of $28 000 on a Sunday so that her husband could have an angiogram the next day. Even though they had “top-level” private health insurance cover, the private hospital could not confirm their insurance status and obtain an authorisation from their health insurance fund on a weekend. Patients with private health insurance should be able to use the benefits of their insurance 24 hours a day, 7 days a week. Private health insurance funds that do not provide 24-hour authorisations to private hospitals for hospital admission should state this limitation clearly in their insurance product as a part of their statutory product disclosure statements. Such disclosure would provide relevant information to consumers to enable them to make an informed choice about their private health insurer.

Antony Nocera

To the Editor: In March 2008, nicotine patches (15 mg per 16 hours) were authority-listed for Aboriginal and Torres Strait Islander people by the Pharmaceutical Benefits Advisory Committee (PBAC) as part of efforts to improve access to medications.1,2 Although timely and welcome, the decision to list only 15 mg patches should be revisited, as it is not consistent with current smoking cessation clinical guidelines.3-5 Neither is it consistent with evidence from a 2008 tobacco survey that we conducted in remote Aboriginal communities in Arnhem Land, Northern Territory. Clinical guidelines advise that smokers will require more intensive support to quit smoking if they: (i) demonstrate a high level of dependency; (ii) have had more than one short quit attempt; (iii) still smoke or experience cravings when using nicotine replacement therapy (NRT); and/or (iv) are frequently exposed to other smokers.3,4 This support includes appropriate medication and counselling, which are more effective in combination.5 Under-dosing is a common problem, given that NRT products deliver nicotine plasma levels well below that delivered by a cigarette.3 Higher-dose NRT products (eg, 4 mg gum) or combination nicotine therapies (eg, patches with gums) are effective for highly dependent smokers.3,5 In the Arnhem Land survey, we interviewed 397 people (aged ≥ 16 years) about tobacco. Of these, 77% were current smokers. Among the current smokers, 17% were attempting to quit or had tried to quit, and 58% were contemplating quitting. Dependency was common, with 55% of smokers reporting they smoked first thing in the morning or during the night. With many dependent smokers, high rates of smoking and widespread “cue” exposure, there is a high need for intensive quit support. We also interviewed 24 smokers interested in quitting who were offered 21 mg patches, and 4 mg and 2 mg gums and lozenges. All used the gums; four combined gum with 21 mg patches, with one of these initially trying lozenges. Modest increases in periods of abstinence and reduction in daily consumption were documented in the 11 participants followed up so far. This evidence, albeit limited, challenges the PBAC rationale for listing patches but not gums for Aboriginal and Torres Strait Islander people, namely: “this population eschews oral aids for smoking cessation”.1 Some Aboriginal and Torres Strait Islander smokers wanting to quit may benefit from combination NRT or gums alone. A wider range of NRT products including gums should therefore be considered by the PBAC.

Jan A Robertson · David J MacLaren · Alan R Clough

To the Editor: The article by Calver and colleagues on the use of gastrostomy tubes in older Western Australians raises important issues regarding decision making for gastrostomy tube insertion and ongoing care of gastrostomy-fed patients. Calver et al report a high incidence of readmissions within 1 year for gastrostomy tube replacement or gastrostomy-related complication (25%) and a high 1-year mortality rate (54%)1 Extrapolation of New South Wales data suggests that about 11 000 Australians rely on gastrostomy feeding at home as their sole source of nutrition and hydration.2 In NSW, about 2300 gastrostomy and jejunostomy procedures are performed each year in public health care facilities, for which about 60% of patients are discharged home. Of these patients, 40% require tube feeding for 2 or more years, and 11% for 5 or more years. In the financial year 2004–05, there were about 700 reported presentations to emergency departments of patients requiring percutaneous endoscopic gastrostomy tube replacement or experiencing tube-associated feeding complications (eg, stoma site infection, tube blockage, buried bumper syndrome, and diarrhoea related to tube feeding); 15% of these presentations resulted in ward admission.2 Many complications can be prevented or treated effectively in the community, provided that patients, carers and health professionals are adequately trained and supported, and that formula, consumables and equipment are affordable. Of particular concern are situations where patients who have multiple comorbidities and disabilities and who cannot advocate for themselves are discharged to nursing homes and group homes. The NSW Ombudsman reported the deaths of two people in 2006 as a result of poor management of their gastrostomy tube feeding and recommended that minimum care standards be introduced.3 Enteral nutrition is an orphan therapy, with no single professional group taking ownership of it. Hospitals release their responsibility when they discharge a patient, as the patient is no longer admitted, and there are limited community services to take over care. Patients are left to fend for themselves, resulting in poor outcomes. How a patient will manage tube feeding at home should be an important part of the decision-making process that occurs before a tube is inserted, rather than an afterthought. As the use of therapies that can be performed at home increases (eg, dialysis and enteral nutrition), there needs to be increased investment in community-based health services to support patients in caring for themselves at home. This will bring social and economic benefits to both patients and the health care system.

William H Watt · Kate A Needham · Peter L Talbot · Janet P Bell · Glen J Pang

To the Editor: We applaud the attempt by Boyes and colleagues to ascertain the level of psychological distress experienced by patients over the years following diagnosis with cancer, through a retrospective, cross-sectional survey of New South Wales cancer registrants.1 However, we believe several methodological limitations ought to reduce the confidence with which the authors drew their conclusions. The authors’ comment that “life after cancer is not all doom and gloom” was, perhaps, intended to be a little facetious. From a scientific point of view, however, such a statement is also very difficult to ever disprove — of course it isn’t all doom and gloom. Further, the authors’ assertion that psychosocial wellbeing several years after cancer diagnosis is comparable with that of the general population cannot be substantiated by studies conducted by this method. As the survey was cross-sectional, we have no information about the level of distress experienced during the years since diagnosis. A longitudinal design is now de rigueur in this field for this reason. Certainly, clinical experience shows us that many patients actually do experience “doom and gloom” often, indeed, arising from “insidious and relentless disease” — this group may not, however, be well represented by a cancer registry survey sampling 5 years after diagnosis. Selection bias is a major problem. Of the sample of 2029 eligible, randomly selected people, 655 (32%) were deemed ineligible, with one of the exclusion criteria being not “mentally capable of participating”. Could this sizable subgroup have included those who were distressed? Also, of the eligible sample of 1374, 366 (27%) declined to participate. What were the reasons for refusal? Were some too distressed to participate? In summary, how were the 37% who did not participate in the study faring 5 years after diagnosis? The Hospital Anxiety and Depression Scale (HADS), used by Boyes et al to detect psychological distress in early-stage breast cancer, under-reports distress when recommended cut-off scores are used, compared with a structured clinical interview validated to provide Diagnostic and statistical manual of mental disorders, fourth edition (DSM-IV2) diagnoses.3 In other words, the HADS is known to lack sensitivity and positive predictive power in the cancer setting. Other cancer researchers found the same.4,5 Further, measuring distress only in terms of anxiety and depression 5 years into the adjustment process fails to capture the quality of continuing distress and the degree to which traumatic growth and other forms of meaning-based adjustment have been achieved. The limitations of the HADS should have been better acknowledged.

Jeremy W Couper · Anthony W Love · Annabel C Pollard · Sidney Bloch

In reply: We thank Couper and colleagues for their interest in our article,1 but reiterate that we focused on anxiety and depression experienced by long-term cancer survivors, specifically at 5–6 years after diagnosis. We agree that longitudinal studies are vital for understanding the level of psychological distress experienced during the years since diagnosis, and as discussed in our article, we are currently undertaking a longitudinal study with a diversity of cancer patients to assess a comprehensive range of physical, psychological, social and lifestyle effects of cancer. Although opinions on the performance of the Hospital Anxiety and Depression Scale (HADS) vary, it is one of the most popular measures of psychological distress, and has been used extensively across the cancer continuum.2 A recent review of the validity of the HADS concluded that it performs well in screening for caseness of anxiety disorders and depression in a range of patient populations, including patients with cancer, and in the general population.3 We acknowledge that our study had the strengths and limitations normally associated with recruiting through a population-based cancer registry.4 Nevertheless, our results are consistent with a growing body of evidence indicating that most cancer survivors are doing well 5 or more years after diagnosis.5 We believe our conclusions are justified.

Allison W Boyes · Afaf Girgis · Alison C Zucca · Christophe Lecathelinais

To the Editor: Sikic and colleagues state that the 1990 legislation making helmets compulsory for bicyclists in Victoria was associated with a decrease in non-fatal head injuries and fatalities.1 However, of the three citations given to support this statement (references 8–10 in Sikic et al),1 one is a study performed before the helmet law was introduced, another is an editorial, and the third makes the common error of attributing to helmet use the effects of economic recession and road safety campaigns. These together reduced all road deaths in Australia by about a third.2 Careful analysis takes account of such factors. Western Australia and New Zealand offer good datasets of injuries to cyclists and control groups through the period of increasing helmet use and enforcement: neither dataset shows evidence that mass helmet use reduced the occurrence of serious head injuries to cyclists.3 With fewer serious crashes, there were fewer serious head injuries, and a general reduction in severity of injury in road accidents for all road users. These observations make it hard to accept that cycle helmets reliably confer significant protection. Sikic et al further state that “Wearing an approved safety helmet substantially reduces the risk of serious head injury in cyclists who fall or are involved in collisions with motor vehicles”,1 citing case–control studies. Such studies are known to have serious weaknesses when applied to voluntary behaviour in a socially disparate population.4,5 Risk assessments do not justify helmet laws for cyclists alone. Analysis of Australian Government data (1988–1990) showed that cyclists faced a lower risk of death per hour than car occupants (0.41 v 0.46 fatalities per million hours of use).6 Wider risk assessment based on European data confirms that cycling risks are in the same range as for walking and driving.7 Sikic et al ask for further research to identify factors other than helmet wearing that contribute to preventing cycling injuries. One important factor is already well known: an increase in cyclists on the roads means less risk per cyclist.8 Considerable experience is now available to show that mass helmet use has not been effective in preventing serious head injuries in cycling populations.3 Enforced helmet laws in Australia may deter people from cycling9 and getting the major health benefits of moderate exercise.10

Malcolm J Wardlaw