Volume 187 Issue 10

Australian health systems must rise to the challenge of providing thrombolysis to more stroke patients The substantial benefits and relative safety of tissue plasminogen activator (tPA) for acute ischaemic stroke within 3 hours of symptom onset have been accepted by stroke clinicians around the world.1 It is one of the most effective treatments in acute medicine, with a 30% increase in excellent outcomes and a “number needed to treat” for clinical improvement as low as three patients.2 A European register of 6483 patients (SITS-MOST, Safe Implementation of Thrombolysis in Stroke Monitoring Study) indicated that tPA is safe and effective, even when used in relatively inexperienced centres.3 The clinical benefits overwhelm a small rate of bleeding complications, chiefly symptomatic haemorrhagic transformation of the infarct. The rate of symptomatic intracerebral haemorrhage was actually lower in this large register than in the earlier randomised clinical trials.3 Small Australian tPA audits have confirmed these conclusions.4,5 Based on level 1 evidence, the therapy was licensed in Australia in 2003, and is recommended in Australian, North American and European stroke guidelines.6 However, despite this overwhelming information, probably some thousands of Australian patients are effectively denied tPA each year.7 This is a major challenge for our health system. Batmanian and her colleagues have made an important contribution,8 demonstrating that a 24-hour comprehensive protocol delivered by a team involving emergency physicians and stroke neurologists can deliver thrombolysis to 14% of their acute stroke patients (→ A protocol-driven model for the rapid initiation of stroke thrombolysis in the emergency department). This figure is in line with the best international stroke centres. The involvement of emergency physicians is essential to optimise the delivery of tPA, particularly given the worldwide shortage of stroke specialists. With a protocol aimed at rapid triage, assessment and investigation of stroke patients, Batmanian et al have shown that nearly all eligible patients can receive thrombolytic therapy. Their centre at St Vincent’s Hospital (Sydney) and other Australian centres are participating in the ongoing SITS international registry, aimed at auditing the efficacy and safety of tPA for acute stroke. Nearly 400 Australian patients have now been entered on this registry, with safety of tPA consistent with the benchmark European figures (Dr M Parsons, neurologist, John Hunter Hospital, Newcastle, personal communication). A key barrier to the use of tPA is delayed arrival to hospital after stroke onset. In the St Vincent’s study, 40% of patients arrived within 3 hours, a higher rate than in many centres. Improved access to tPA and other acute stroke therapies can be enhanced by public education about stroke symptoms (such as the FAST: Face, Arm, Speech, Time campaign of the National Stroke Foundation in Australia), followed by rapid ambulance transport to a hospital with an organised acute stroke team. The public must be educated to call an ambulance, not a general practitioner, when stroke symptoms occur. Australian studies have confirmed that ambulance officers can accurately diagnose most strokes.9 Ideally, paramedics inform the emergency department of their impending arrival, facilitating the 60-minute “door to needle” target time for tPA therapy. Implicit in achieving good outcomes after therapy is rapid access to a stroke care unit (SCU), the benchmark for optimising outcomes in a condition with a high mortality rate and the commonest cause of disability in our society.10 Stroke strategies in most Australian states have been developed to increase the access of acute stroke patients to SCUs, which are expert multidisciplinary and geographically localised units, using evidence-based protocols for acute stroke therapy. In Australia, these are generally not intensive care units (ICUs), although cardiac monitoring is desirable during the acute period. In the St Vincent’s study, patients were monitored in the ICU for up to 24 hours, because the protocol mandated a 1 : 2 nursing care ratio. We generally admit tPA patients to our SCU during or after the initial 1-hour infusion in the emergency department. Our SCUs use a 1 : 4 nursing ratio. We do not consider that ICU management is needed for most patients, but acknowledge that differing models will suit different institutions. Acute reperfusion therapy is the most promising approach for acute stroke. The evidence for efficacy of tPA within 3 hours is overwhelming, although there is still some debate about treatment in the very elderly, patients with very severe neurological impairment, or those with extensive early ischaemic changes on computed tomography (CT) scan. Meta-analysis of the tPA trials shows that there are treatment responders beyond 3 hours. A number of trials, such as the ECASS 3 and the IST 3 trials, are aimed at extending the current time window for intravenous tPA. Intra-arterial thrombolysis or mechanical clot retrieval may be alternative approaches for selected patients who do not respond to tPA within 3 hours, or within later time windows.11 Thrombolysis is based on the recanalisation of occluded arteries and reperfusion of the ischaemic penumbra, a region of injured brain that is potentially salvageable with rapid restoration of blood flow. Identification of the penumbra using magnetic resonance imaging or perfusion CT may also allow individualised therapy at longer windows.12 In 2007, we do not recommend routine use of intravenous tPA beyond 3 hours, but strongly encourage Australian clinicians to enrol patients into trials addressing these hypotheses. Intravenous tPA is relatively simple to use, particularly with an effective partnership between emergency physicians and stroke clinicians, using appropriate protocols and stroke unit care. It is highly effective and relatively safe. In the 12 years since it was proven, the major systems failure in most countries has been delivering it to more patients. In Australia, we should not shirk from this challenge.

Stephen M Davis MD, FRACP · Peter J Hand MD, FRACP · Geoffrey A Donnan MD, FRACP

The Queensland approach may deliver empirical evidence on whether performance monitoring leads to improved quality of care Several high profile medical disasters have occurred recently in Australia and England and have raised concern about quality of care and patient safety. In Australia, the major disasters have included those at the Bundaberg Hospital in Queensland,1 and the Campbelltown and Camden Hospitals in New South Wales; and in England, the Bristol Royal Infirmary scandal2 and the Shipman Affair.3 Although the specifics in each instance are unique, a common outcome from the subsequent inquiries has been a recommendation for some form of centralised performance monitoring using routinely collected data. Why? Because retrospective desktop analysis of routinely collected data signalled the medical disaster before the “whistle blew”. What is interesting about Queensland Health’s response to the Bundaberg scandal, as described by Duckett et al in this issue of the Journal,4 is that these workers have combined their preferred casemix-adjusted statistical analysis (variable life-adjusted display [VLAD] plots) with an explicit protocol for action (→ Identifying variations in quality of care in Queensland hospitals). In so doing, Duckett et al have recognised that the ultimate purpose of data (and analysis) is to guide action5 (for improvement) while avoiding the commonly held casemix adjustment fallacy.6 The casemix adjustment fallacy begins with a deceptively simple equation that informally relates the variance in outcomes (eg, mortality) to a combination of three variables — chance, patient casemix factors, and quality of care. Statistical techniques are used to account for chance and patient casemix factors. The residual unexplained variance in outcome is then prematurely assigned to quality of care.6 This is naïve, because it assumes too much and is an oversimplification. For example, it assumes that the outcome variance equation is of a closed form (three variables only, when we have no basis to conclude this); it assumes perfect data (no errors in data, no measurement error) and perfect casemix adjustment (when we know that important casemix variables are not as yet measured or measurable); and, finally, it fails to recognise that all statistical models are but simplified approximations subject to error. This casemix fallacy — where the residual unexplained variance in outcome is automatically attributed to quality of care — is to be diligently avoided7 if the process of monitoring is to remain credible and useful in improving quality of care. The protocol for action developed by Duckett et al has two key characteristics: use of a pyramid model of investigation,6,7 which recognises that, in searching for assignable causes of variation, a useful strategy (which mitigates the casemix adjustment fallacy) is to systematically check five variables (data, patient casemix, structure, process of care, and carers), underpinned by an explicit clinical governance framework. A key feature of the pyramid model of investigation is the prior degree of belief, for which there is empirical evidence,8 that the vast majority of explanations for unusual variation will be located at the base (data, patient casemix) and the least likely explanations will be located at the apex (quality of care and carers). Indeed, as part of the Shipman Inquiry, two general practitioners were flagged as having “unacceptably” (the casemix adjustment fallacy) high casemix-adjusted death rates. Application of the pyramid model of investigation aided the discovery that this excess mortality was credibly explained by the numbers of patients in nursing homes — place of death was not included in the casemix adjustment method. Furthermore, the inclusion of data (at the base of the pyramid) ensures that data improvement is an integral part of any monitoring process, as this is usually a highly cost-effective way of reducing variation, to the benefit of all concerned. So it is a matter of concern that Duckett et al found that data improvement was frequent and “frustrating to hospitals and clinicians”. One reason for this frustration is because concepts such as “true positives” and “false positives” from diagnostic testing are applied to performance monitoring — but this is unhelpful and misleading. Consider, for example, a signal of high mortality which is due to a data error. If we predicate the definition of signals of special cause variation on malpractice or quality of care,9 then we are required to concede that a false-positive signal has occurred. But the notion of a false positive in the evaluation of a diagnostic test is based on a comparison with a reference “gold standard” test, by which the “true” disease state of each patient is known (albeit using the gold standard). In contrast, in performance monitoring, there is no gold standard and the true state of each data point is unknown, so the notion of false-positive signals makes little sense. Furthermore, in diagnostic testing, a false-positive result can mislead us into taking the wrong action, whereas finding and fixing a data error actually moves us in the right direction — towards understanding and eliminating special causes of variation.10 The possible confusion with diagnostic testing should be remedied through clarifying the purpose of monitoring and providing training on the method and its limitations. Duckett et al have adopted VLAD plots, which are simple visual tools for displaying variation. They show the indicator of interest one patient at a time, while accounting for the prior risk (casemix adjustment) and the play of chance as indicated by upper and lower limits. It is important to appreciate that, wherever the limits are set (except for 0 or infinity), we will occasionally commit an error of commission (when we wrongly identify a signal from the noise) or an error of omission (when we fail to detect a true signal from the noise). The choice of limits is, as Duckett et al state, a matter of judgement that is informed by statistical theory, economic consideration, and empirical evidence. What is exciting about the Queensland approach is that it is well poised to deliver the oft-lacking empirical evidence (because most monitoring systems do not include a specific action protocol) on the use of a performance monitoring system. This empirical evidence is the ultimate criterion by which to judge the usefulness of the performance monitoring system itself. As Duckett et al recognise, performance monitoring is the search for signals of variation from background noise in the presence of imperfect data, imperfect casemix adjustment, and imperfect methods of action. Although the performance monitoring system must be alert to “bad apples” (and this is noted in the pyramid model for investigation), care needs to be taken to ensure that the search for “bad apples” does not become the primary aim, as this will (most likely) increase fear and thereby hinder genuine continual improvement.9 Ultimately, performance monitoring needs to embrace the wider aspects of the science of improvement,9 in which statistical methods for analysing variation are but one component. The other major components involve “appreciation of a system”, “theory of knowledge” and “psychology of humans”, in which all stakeholders in the performance monitoring system cooperate towards the aim of continual improvement. The challenge for performance monitoring in general6,10 has been that it has itself been at variance with key principles of the science of improvement, and so the extent to which the Queensland model delivers continual improvement remains to be seen.

Mohammed A Mohammed PhD · Andrew J Stevens

Objective: To evaluate a hospital-wide surveillance and intervention program introduced to reduce the incidence of bloodstream infections (BSIs) caused by intravascular (IV) catheters.Design, setting and participants: Prospective surveillance of all inpatients and outpatient attendees with positive blood cultures (both hospital-onset and community-onset) at a 500-bed tertiary referral hospital from 1998 to 2005.Interventions: Prompt review of all positive blood cultures with identification of BSIs due to IV catheters and associated preventable factors; weekly team meetings and regular reports to clinical areas, with assistance to implement targeted interventions.Main outcome measure: Number of BSI episodes due to IV catheters per year.Results: There were 491 BSI episodes due to IV catheters, mainly central venous catheters. Episodes per year fell from 110 in 1998 to 48 in 2005 (from 32% of all BSI episodes to 14%; a > 50% reduction). From 1998 to 2005, the rate per 1000 discharges fell from 2.3 to 0.9 (P for trend < 0.0005) and the rate per 1000 patient-days fell from 0.6 to 0.3 (P for trend < 0.0005).Conclusions: Our program was associated with a profound drop in the number of IV catheter-related BSIs per year. Active surveillance and intervention programs can lead to substantial and sustained reductions in these common life-threatening infections.

Peter J Collignon FASM, FRCPA, FRACP · Dianne E Dreimanis BHSc(Nursing), MN · Wendy D Beckingham RN, BHSc(Nursing) · Jan L Roberts RN, BHSc(Nursing) · Anne Gardner BA, PhD

Objective: To determine the intentions that young people have for seeking help if they were to develop a mental disorder.Design, participants and setting: National telephone survey of 3746 Australians aged 12–25 years and 2005 co-resident parents, which asked questions about vignettes portraying either depression, depression with alcohol misuse, social phobia or psychosis.Main outcome measures: Where participating young people or co-resident parents would seek help if they (or their child) had one of the problems portrayed in the vignettes; barriers to seeking help.Results: For adolescents, family was the main source of intended help, mentioned by 45%–60% (depending on the vignette), while general practitioners were mentioned by only a small minority (4%–13%). For young adults, family was relatively less important (21%–31%) and GPs relatively more so (19%–34%). By contrast, parents frequently mentioned GPs as an intended source of help for their children (by 40%–72% of parents of adolescents and 61%–76% of parents of young adults). For young people, the main barriers to seeking help were embarrassment or concern about what others might think, while the main barrier for parents was resistance from the child.Conclusions: Recent initiatives to extend the uptake of treatment for mental disorders have been centred around GPs as the initial point of help-seeking. Few young people see GPs as a preferred source of help, and action is needed to alter this perception or to reform mental health services to be more attractive to this age group.

Anthony F Jorm MPsychol, PhD, DSc · Annemarie Wright BAppSc(OT), MMedSc(HProm) · Amy J Morgan BASc, BAppSci(Psychol)(Hons)

Objective: To document the epidemiology, clinical characteristics and diagnosis of an outbreak of Mycobacterium ulcerans infection (Bairnsdale or Buruli ulcer [BU]) during the period 1998–2006, and compare delays in diagnosis between residents of endemic and non-endemic regions.Design and setting: Retrospective case study of patients identified through infectious disease physicians on the Bellarine Peninsula and the Victorian Department of Human Services notifiable diseases database.Main outcome measures: Description of events leading to diagnosis of BU.Results: Eighty-five BU patients recalled their experience. Fifty-three patients were older than 60 years, and 61 permanently resided on the Bellarine Peninsula. The onset of symptoms occurred most frequently in mid winter. Twenty-eight patients had lesions on the arm and 51 on the leg. The median time between onset of symptoms and first medical contact was shorter for those living in the endemic area (3.0 weeks; interquartile range [IQR], 1.0–5.0 weeks) compared with non-endemic areas (5.3 weeks; IQR, 2.0–9.5 weeks) (P = 0.05). Patients who resided in the endemic area had a shorter median time from their first medical appointment to diagnosis (1.0 week; IQR, 0.0–3.9 weeks) than those who resided in non-endemic areas (5.0 weeks; IQR, 1.3–8.0 weeks) (P = 0.001).Conclusion: Delay in presentation and time to diagnosis of BU are longer in non-endemic than endemic areas. Measures should be taken to raise awareness of the disease in non-endemic areas.

Tricia Y J Quek BSc(Hons) · Margaret J Henry BSc(Hons), PhD · Julie A Pasco BSc(Hons), PhD · Daniel P O’Brien MB BS, FRACP · Paul D R Johnson PhD, FRACP · Andrew Hughes MB BS(Hons), FRACP · Allen C Cheng FRACP, PhD · Jane Redden-Hoare BN(Hons), PhD · Eugene Athan MB BS, FRACP

Objective: To compare demographic and clinical characteristics of methamphetamine users and patients with other toxicology-related problems requiring medical intervention in a hospital emergency department (ED).Design and setting: Prospective observational study of toxicology-related presentations to the ED of St Vincent’s Hospital (SVH), Sydney, an inner-city tertiary hospital, between 1 October and 31 December 2006.Main outcome measures: Differences between methamphetamine-related and other toxicology-related presentations to the ED in relation to behaviour, mode of arrival, accompaniment, need for scheduling, location of drug use, intravenous drug use history, psychiatric history and demographic characteristics.Results: During the study period there were 10 305 patient presentations to SVH ED; 449 (4%) were toxicology-related presentations, of which 100 (1% of total) were methamphetamine-related. Methamphetamine users were significantly more agitated, violent and aggressive than patients with other toxicology-related presentations and significantly less alert, communicative and cooperative (P < 0.001); 24% of methamphetamine users (24/100) arrived with police accompaniment versus 9% of other toxicology patients (33/349) (P < 0.001). Methamphetamine users were more likely to have a history of intravenous drug use and mental health problems (P < 0.001); 39% of methamphetamine presentations (39/100) required scheduling under the Mental Health Act 1990 (NSW) compared with 19% of other toxicology-related presentations (67/349) (P < 0.001); 43% of methamphetamine-related presentations (43/100) involved drug use on the street compared with 24% of other toxicology-related presentations (83/349) (P < 0.001). Two-thirds of all methamphetamine users were male, and the most common age group for both male and female users was 26–30 years. The mean age and sex distribution of patients with other toxicology-related presentations were not significantly different. Among methamphetamine users, 27% of women (9/33) were in the 21–25-year age group compared with 10% (7/67) of men (P < 0.001).Conclusion: There were significant differences between methamphetamine-related and other toxicology-related presentations to SVH ED. Methamphetamine users were more aggressive, violent and dangerous, and thus more likely to pose a risk to health personnel and others. Methamphetamine appeared to be used consistently, rather than as an episodic “party drug”.

Philippa J Bunting BBus, CA · Gordian W O Fulde MB BS, FRACS, FACEM · S Lesley Forster MB BS, FRACMA, FAFPHM

Objective: To assess efficacy and safety of a 24-hour comprehensive protocol-driven model for rapid assessment and thrombolysis of stroke patients in the emergency department.Design: Prospective open observational study.Participants and setting: All patients with acute stroke presenting within 3 hours to the St Vincent’s Hospital (Sydney) emergency department between 1 December 2004 and 30 July 2005.Main outcome measures: Proportion of patients treated, patient demographics, clinical outcome, adverse events and time to treatment parameters.Results: 134 patients (100 stroke; 34 transient ischaemic attack) were admitted to the stroke unit during the study period. Of the 100 stroke patients, 40 presented within 3 hours of symptom onset. Fifteen patients had no contraindications and received intravenous thrombolysis. At 3 months, 10 patients (67%) were independent (modified Rankin score [mRS], 0–2) and seven (47%) had an excellent functional outcome (mRS ≤ 1). Symptomatic intracranial haemorrhage was not observed. The median time from symptom onset to tissue plasminogen activator treatment was 155 minutes (range, 105–197 min). Median onset-to-door, door-to-computed tomography, and door-to-needle times were 48, 25, and 87 minutes, respectively.Conclusion: Rapid assessment of stroke in the emergency department according to a comprehensive protocol allows identification and treatment of acute ischaemic stroke patients eligible for thrombolysis.

Julia J Batmanian BSc(Med), MB BS(Hons) · Meeyin Lam BAppSc(Physio), MIPH · Caitlin Matthews BSc(Hons), MB BS(Hons) · Andrew Finckh MB BS, FACEM · Martin Duffy MB BS, MMed(ClinEdi), FACEM · Robert Wright FRACP, FFARACS, FJFICM · Bruce J Brew MB BS, MD, FRACP · Romesh Markus PhD, FRACP, MB ChB

Identifying and acting on variations from good practice is one of the critical tasks of clinical governance. We describe one aspect of Queensland’s post-Bundaberg clinical governance arrangements: the use of variable life-adjusted displays (VLADs) to monitor outcomes of care in the 87 largest public and private hospitals in Queensland, which together account for 83% of all hospital activity. VLAD control charts were created for 31 clinical indicators using routinely collected data, and are disseminated monthly. About a third of hospitals had a run of cases in the 3-year period that flagged at the 30% level (local level investigation). For three indicators, about one in five hospitals had sufficiently cumulatively more deaths than statistically expected that the hospital was highlighted for state-wide review. VLADs do not provide definitive answers about the quality of care. They are used to develop ideas about why variations in reported outcomes occur and suggest possible solutions, be they ways of improving data quality, improving casemix adjustment, or implementing system changes to improve quality of care. Critical to the approach is that there is not just monitoring — the monitoring is tied in with systems that ensure that investigation, learning and action occur as a result of a flag.

Stephen J Duckett PhD, DSc, FASSA · Michael Coory MB BS, PhD, AStat · Kirstine Sketcher-Baker BAppSc

In April 2004, Melbourne’s Peter MacCallum Cancer Centre, Australia’s only stand-alone dedicated cancer hospital, became the first Australian site to offer digital mammography (DM). In the first year of DM operation, 1208 mammograms were performed on 1157 women; 17 new cases of invasive carcinoma and six new cases of ductal carcinoma-in-situ (DCIS) were detected; and 30 hook-wire needle localisations were conducted in 29 patients. We developed a unit policy to manage indeterminate microcalcifications newly demonstrated on DM that were not previously detected by conventional screen-film mammography (CM): those believed to have malignant morphology were recommended for biopsy, and those without were recommended for 6-month DM follow-up to confirm microcalcification stability. DM detected 56 new stand-alone microcalcifications (18 suspicious and 38 indeterminate). Tissue diagnosis of 12 suspicious microcalcifications yielded four cases of DCIS and one of atypical ductal hyperplasia. Of the indeterminate microcalcifications, 35 have demonstrated stability at DM follow-up to date, over a mean period of 23.6 months. From our experience, we believe DM’s superior demonstration ability uncovered microcalcifications previously undetected by CM, rather than microcalcification progression. We suggest that routine review with DM, rather than biopsy, is appropriate management when new indeterminate microcalcifications without malignant characteristics are identified by DM.

Emma Pun MMed, FRANZCR · W F Eddie Lau BPharm, FRANZCR · Robin Cassumbhoy FRANZCR · Anthony J Taranto FRANZCR · Alexander G Pitman BMedSci, FRANZCR

We treat patients, not levels The goal of therapy with antiepileptic drugs (AEDs) is to prevent all seizures, using drugs that produce minimal side effects. Success in achieving this goal must be assessed clinically. Routine ordering of tests to measure AED levels is misguided, given that the tests are costly and often unnecessary.1 Therapeutic drug level monitoring was introduced because of a belief that drug levels in the blood were a more reliable indicator of efficacy than drug dose alone. Although this relationship applied to phenytoin, the main drug in use three decades ago, it does not apply to many AEDs. The concept of a “therapeutic range” is a statistical concept, applying to populations rather than individuals. There are many patients whose condition is perfectly well controlled with so-called subtherapeutic doses, while other patients may require more than the recommended therapeutic dose. For most patients, management need not involve serum drug measurements. They are certainly unnecessary for the so-called second generation AEDs or for benzodiazepines, which are selectively taken up by the brain and for which serum level measurements are not meaningful for management.2 Sodium valproate (SVP) is widely used. The relationship between SVP serum levels and efficacy has not been confirmed, and serum testing is often inappropriate, as reported in this issue of the Journal (→ Serum sodium valproate testing: is it appropriate?).3 On conventional doses of SVP (which should be kept under 1000 mg per day in women of childbearing age), most compliant patients achieve a serum level within the so-called therapeutic range, which is wide (350–700 μmol/L). Previous studies have shown that results of SVP assays do not influence the doses used in management.4 If the results are not acted on, the tests are a waste of time and money. On the other hand, patients can be destabilised if the results of SVP assays directly influence doctors to alter doses regardless of clinical indications.5-7 For the AEDs topiramate, vigabatrin, oxcarbazepine, gabapentin and levetiracetam, insufficient data are available to justify serum level monitoring. In fact, for vigabatrin, which acts by permanent inhibition of an enzyme related to neurotransmitters, measuring levels is totally unnecessary.4 Lamotrigine may represent a special category. The metabolism of this drug is induced by sex hormones in the oral contraceptive pill and is also altered during pregnancy. In these circumstances, the drug is rapidly eliminated. It may be helpful to measure lamotrigine concentrations in each trimester to keep them at pre-pregnancy levels by adjusting the dose. At delivery, the dose must be reduced. There is no accepted therapeutic range for lamotrigine, and the concept of individual ranges has been proposed. Once a level is established that is associated with good control of the patient’s seizures, this level may be a target to be aimed at when changes are made to AED doses and co-medications in that particular patient. However, routine measurements are not recommended.8 Thus, with the exception of phenytoin (and possibly lamotrigine), none of the AEDs are suitable candidates for therapeutic drug monitoring assays in routine practice, as such tests put a high cost burden on the health service without significant patient benefit. The International League Against Epilepsy guidelines state that AED serum concentrations must not be measured indiscriminately and that adjusting doses merely to achieve cosmetic correction of levels is inadvisable. However, compliance problems can be an indication for performing an assay.9 In conclusion, we treat patients, not levels. The rationale for performing drug assays should be based on clear indications for why they are to be ordered. Clinical symptoms, inadequate response and possible toxicity may be valid indications, but the most important one appears to be suspicion of non-compliance.

Frank J E Vajda MD, FRCP, FRACP

Objective: To assess whether serum sodium valproate (SVP) testing in the hospital setting is being performed according to evidence-based criteria.Design and setting: Retrospective audit of serum SVP concentration measurements performed on inpatients and emergency department patients at The Canberra Hospital from May to July 2005.Main outcome measures: Indication for performing the test, assessed against evidence-based criteria; timing of blood sample collection; whether the test result altered patient management; whether the request form allowed laboratory staff to assess the appropriateness of the test; cost of performing inappropriate tests.Results: We retrieved 211 test results performed on a total of 95 patients. Notes on 89 patients were available for analysis. Based on evidence-based criteria, 15% of tests were done for an appropriate indication and 29% of the samples were taken at an appropriate time. At most (using generous criteria), 57% of test results made a difference to patient management. Forty-four per cent of request forms contained sufficient detail to allow the pathology department to assess the appropriateness of the test. An estimated $13 236 would be spent unnecessarily on SVP testing at our hospital over a 1-year period.Conclusions: Most serum SVP level measurements were requested inappropriately, and many were not taken at the correct time, thereby rendering the results uninterpretable. Better education of requesting clinicians could significantly reduce the number of unnecessary tests and thus reduce the cost to the health service.

Chamishani Rathmalgoda · Julia M Potter MB BS, PhD, FRCPA · Christian J Lueck PhD, FRACP, FRCP

A number of recent studies have allayed fears about antidepressants precipitating suicidal behaviour in young people. Indeed, antidepressants appear to be conspicuous by their absence of use in young people who die by suicide. Furthermore, there is concern that the reduced prescribing of antidepressants to young people may be associated with an increase in youth suicide in the United States. Although not first-line treatment, antidepressants should not be denied young people if psychosocial and cognitive behavioural therapies are not effective for major depression.

Robert D Goldney MD, FRANZCP, FRCPsych

Clinical recordA 35-year-old British man was admitted with a 2-week history of abdominal discomfort, fatigue and intermittent high fevers associated with drenching night sweats. He reported having returned to Australia 2 months before presentation from a 12-year around-the-world trip, having travelled extensively for 6 months through South-East Asia and northern Australia. Before that, he had been to Argentina, the Maldives, Egypt, India and Cambodia, where he had worked as a professional kitesurfing instructor. His past medical history was unremarkable and he was not on regular medication. His social history revealed intravenous heroin addiction in his early 20s. He had consumed more than 80 g of alcohol daily for a number of years, with occasional binges, until recently, when drinking alcohol provoked nausea and vomiting. The patient appeared unwell with a tympanic temperature of 40°C. He was fair-skinned and of muscular build. Physical examination showed conjunctival jaundice but no evidence of needle tracks, rashes, finger clubbing, lymphadenopathy or suspicious cutaneous lesions. Tender hepatomegaly was noted, but no ascites or splenomegaly. Findings of cardiac, respiratory and neurological examinations were normal. Abnormal laboratory findings on admission were: albumin concentration, 29 g/L (reference range [RR], 33–47 g/L); alkaline phosphatase titre, 164 U/L (RR, 30–115 U/L); γ-glutamyl transferase titre, 207 U/L (RR, 0–45 U/L); alanine aminotransferase titre, 76 U/L (RR, 0–40 U/L); aspartate aminotransferase titre, 279 U/L (RR, 0–40 U/L); lactate dehydrogenase titre, 4268 U/L (RR, 100–225 U/L); white cell count, 13.4 × 109/L (RR, 4.0–11.0 × 109/L); absolute neutrophil count, 9.7 × 109/L (RR, 1.5–6.0 × 109/L). There was no evidence of thrombocytopenia, renal dysfunction or coagulopathy. Findings on abdominal ultrasonography the day before admission were reported to be normal. Despite there being few clinical features to suggest a source of sepsis apart from the hepatomegaly, because of his travel history and previous history of injecting drug use, a wide variety of infectious diseases, including subacute bacterial endocarditis, typhoid, malaria, and viral hepatitis, were all initially considered in the differential diagnosis. Multiple sets of blood cultures, urine and stool samples were sent for microscopy, culture and sensitivity. Three sets of thick and thin films for malaria as well as a test for Plasmodium falciparum antigen were negative. In view of the clinical hepatomegaly, a computed tomography (CT) scan of the abdomen was performed, which was reported as strongly suggestive of microabscesses (Box). The presumptive diagnosis at this stage was a tropical pyogenic liver abscess, although military tuberculosis and candidiasis were also considered. Intravenous therapy with flucloxacillin, ceftriaxone, ciprofloxacin and metronidazole was initiated. The ceftriaxone was subsequently changed to meropenem, and the ciprofloxacin ceased 5 days after there had been no growth on any cultures. Serological tests for a wide variety of pathogens and diseases were also performed, including: HIV 1 and 2; hepatitis viruses A, B and C; cytomegalovirus; Epstein–Barr virus; Q fever; brucellosis; cryptococcosis; Entamoeba histolytica; Ross River virus; Dengue virus; flavivirus; leptospirosis; schistosomiasis; cysticercosis; and melioidosis. Results for all of these eventually returned negative. Despite the antibiotics and fluids given intravenously, the patient’s condition deteriorated over the course of 6 days and he developed hepatic encephalopathy, ascites, pleural effusions and peripheral oedema. He also continued to spike high temperatures daily, but repeated blood cultures were sterile. A CT scan performed a week after the initial CT scan suggested enlargement of the liver lesions. Although hepatic abscesses were still considered most likely, given the markedly elevated lactate dehydrogenase titre and lack of clinical improvement with broad-spectrum antibiotic therapy, alternative diagnoses, in particular malignancy, were also considered. We therefore performed a core biopsy of the liver under ultrasound guidance. There was no evidence of pus, and the lesional tissue showing strong monoclonal antibody staining against Melan-A and HMB-45 confirmed liver infiltration by a poorly differentiated malignant melanoma. Repeat physical examination included dilated pupil fundoscopy, which showed a brown, dome-shaped subretinal lesion just below the left optic disc, most suggestive of a primary choroidal melanoma. The patient said that he never used sunglasses while kitesurfing. DiscussionExposure of the unprotected eye to sunlight or sunlamps is an important risk factor for the development of intraocular melanoma.1 The incidence of ocular melanoma in dark-eyed individuals is lower, probably because they are less sensitive to solar radiation, or less is transmitted to the choroids.2 Uveal melanoma (affecting the iris, ciliary body, and choroids) is the most common primary intraocular malignancy in the Western world, affecting six to eight adults per million each year. Although fewer than 2% of patients show evidence of metastatic spread at presentation, over 40% will eventually die from widespread disease.3 Most intraocular melanomas are initially asymptomatic. Tumour enlargement may then cause distortion of the pupil (iris melanoma), blurred vision (ciliary body melanoma), or decreased visual acuity caused by either central growth close to the macula or secondary retinal detachment (choroidal melanoma). Because the uveal tract is a vascular structure without lymphatic channels, tumour spread occurs primarily by either local extension or by haematogenous dissemination. The first site of systemic metastases is the liver, although spread to other organs such as lung, bone, and subcutaneous sites have been described.3 Metastasis of melanoma to the liver, although rare, can produce a dramatic initial presentation with fulminant hepatitis, shock, and multisystem organ failure.4 An elevated lactate dehydrogenase titre is one of the most predictive factors for metastatic spread and decreased survival in patients with malignant melanoma, with a sensitivity of 79% and specificity of 92% in detecting disease progression to stage IV melanoma.5 Extraocular extension and metastatic spread are associated with an extremely poor prognosis, and response rates with contemporary single-agent chemotherapy are generally below 10%. A recently published study investigated the use of chemotherapy with intra-arterial hepatic fotemustine.6 Median survival rates were among the longest reported, with an overall response rate of 36%, a median overall survival of 15 months, and a 2-year survival rate of 29%. Our patient was scheduled for three cycles of intravenous fotemustine therapy, but developed significant tumour lysis syndrome with intractable hyperkalaemia. He died shortly after the second cycle, only 28 days after being admitted to hospital. The final diagnosis was hepatic failure secondary to metastatic melanoma from an intraocular primary melanoma. An autopsy was not performed. Coronal reconstruction computed tomography image of the abdomen showing hepatomegaly with numerous low-density lesions scattered throughout both lobes of the liver

Stefan Buchholz MD, MRCP(UK) · George Rudan MB BS, FRACP

A 64-year-old man was referred for assessment of a symmetric violaceous rash of the dorsal aspect of the fingers (Figure, A), trunk, olecranon processes and malleoli. An oedematous, blue-purple discolouration of the eyelids was also apparent. The patient complained of severe weakness of the proximal muscles and dysphagia. T2-weighted magnetic resonance imaging showed muscular inflammation and a pulmonary mass (Figure, B) that was histologically identified as squamous cell carcinoma and staged as T3 N0 M0. The dermatomyositis, confirmed by elevation of serum muscle enzymes and electromyography, improved clinically with oral prednisone plus hydroxychloroquine, and did not recur after surgical excision of the tumour. A: Symmetrical violaceous rash with papules on the dorsal aspect of the interphalangeal and metacarpophalangeal joints. Cuticle telangiectasias can also be seen. B: T2-weighted magnetic resonance image showing marked swelling and diffuse dyshomogeneity of skeletal muscles (thick arrows), a 4 cm peripheral mass in the upper lobe of the right lung (thin arrow) and bilateral pleural effusion (arrowheads).

Daniele Torchia · Emiliano Antiga · Letizia Ricupero · Marzia Caproni · Paolo Fabbri

A 45-year-old man presented with frequent palpitations. Clinical examination and electrocardiogram were unremarkable. Transthoracic echocardiography suggested asymmetrical septal hypertrophy, although the acoustic windows were poor. Transoesophageal echocardiography revealed a broad false tendon within the left ventricle, extending from the basal septum to the apical lateral wall (Figure). Holter monitoring showed frequent premature ventricular complexes, indicating right bundle branch block morphology. The incidence of false tendons — fibromuscular intracavitary bands anatomically distinct from the valvular cusps — is 0.4% to 3.0%.1 They may be associated with malignant ventricular arrhythmias, which should be excluded before making a diagnosis of benign premature ventricular complexes in a healthy patient.1 Transoesophageal echocardiogram: the arrow shows the false tendon from the basal septum to the apical lateral wall of the left ventricle (LV).

Robin A P Weir · Henry J Dargie · Iain N Findlay

Re: “Overweight and obesity from childhood to adulthood: a follow-up of participants in the 1985 Australian Schools Health and Fitness Survey”, the letter by Alison J Venn, Russell J Thomson, Michael D Schmidt, Verity J Cleland, Beverley A Curry, Hanni C Gennat and Terence Dwyer, in the 3 September issue of the Journal (Med J Aust 2007; 187: 314-315). The Box showing the distribution of body mass index values for men and women in three different age groups showed two identical graphs of the data for men. The graphs were clearly different in the original submission, but the data supplied were incorrect. The correct graphs are shown here. Distribution of body mass index values for men and women in three different age groups* * 24–27 years, 757 men and 854 women; 28–30 years, 767 men and 807 women; and 31–34 years, 673 men and 691 women in the 20-year follow-up of the 1985 Australian School Health and Fitness Survey.

Alison J Venn · Russell J Thomson · Michael D Schmidt · Verity J Cleland · Beverley A Curry · Hanni C Gennat · Terence Dwyer

To the Editor: Cannabis use is implicated in serious social disruption in many Northern Territory Aboriginal communities.1 Rising levels of cannabis use were first reported in Aboriginal communities in Arnhem Land in 2002, along with associated concerns about escalating social impacts and mental health effects compounded by other substance use.2 A random sample of 164 people in Arnhem Land initially interviewed and assessed in 2004 were followed up between October 2005 and June 2006. Their cannabis use was measured using health worker assessments and self-reports from interviews. Ethical approval was granted by the NT Health Department, Menzies School of Health Research, and James Cook University. Despite a modest decline in cannabis use in this population between 2002 and 2004,3 the 2005–2006 data indicate persisting high rates, with 61% of males and 58% of females (aged 13–34 years) using cannabis at least weekly. In a subsample of 60 cannabis users opportunistically recruited for in-depth interviews in 2005–2006 (37 male and 23 female, aged 13–42 years), 92% of males and 78% of females used cannabis daily; 88% reported cannabis dependence symptoms. These figures appear to be far higher than national rates, although national data for similar age groups are not available.4,5 Research has found that, nationally, 6% of males and 3% of females (aged ≥ 14 years) reported using cannabis in the past week; 18% of males and 13% of females smoked cannabis daily;4 and 21% of adults (aged ≥ 18 years) using cannabis were dependent.5 Beyond high rates of cannabis use in Arnhem Land communities, we also found local characteristics and perceptions that illustrate the drug’s distinctive context of use (Box). Quantities of cannabis used appear to be higher than in the general population; unemployment among users is higher; and violence related to diminished supply is common. One Indigenous community leader described attitudes to cannabis use: “... if there’s a bowl of it on the table, it is smoked until gone, morning to night”. Interestingly, some respondents reported that using cannabis prevents them from engaging in criminal activity (Box). While key community members may believe that cannabis is a tool for social control — “good for calming down people” — they are increasingly recognising the significant social and mental health problems it causes: People get chained by [cannabis], they don’t go hunting with family ... lots of fights when they can’t get any ... [Cannabis] becomes the boss. Continued concerns about adverse mental health consequences for Aboriginal people in Arnhem Land who use cannabis seem to be warranted. Cannabis appears to be firmly entwined in these isolated communities in a manner not seen nationally. High levels of concurrent drug use, particularly tobacco, raise additional health concerns. Resources are urgently needed for prevention programs and targeted interventions for chronic cannabis users and those with psychiatric comorbidity. If these patterns of use continue, the implications for compounding of pre-existing mental illness and the potential mental health burden are disturbing. Characteristics and perceptions of cannabis use in Arnhem Land Aboriginal communities (57 males and 49 females, aged 13–42 years*) in 2005–2006 compared with available national data from 1997† and 2004‡ National4,5 Arnhem Land* National4,5 Arnhem Land* Number of cones smoked Per cent unemployed current users / daily users 3.2 (average per day)‡ 7.4 (average per occasion) 25.6%† / nd 60% / Males, 41%; females, 94% Concurrent drug use Motivations for use Alcohol (86.2%); stimulants§ (8.2%–27.9%); none (10.8%); analgesics (6.6%); antidepressants (5.7%); tranquillisers/sleeping pills (4.4%); other (3.9%)‡ Tobacco (100%); alcohol, restricted access (40%); kava (15%); petrol (5%)¶ nd Socialisation (tempted, lonely, copying friends); mood altering (“calms me down”, “gets me going in the morning”, “makes my mind straight”); drug substitution (from alcohol or petrol); prevents criminal activity (stealing or other trouble) Drug substitution (when cannabis unavailable) Motivations for ceasing/moderating use Alcohol (60.4%); no substitution (34.2%); ecstasy/designer drugs (1.3%); painkillers/ analgesics (0.8%); tranquillisers/sleeping pills (0.5%); heroin (0.3%); antidepressants (0.2%); cocaine/crack (0.1%); other (1.1%)‡ No substitution (83%); kava (7%); alcohol (5%); petrol (5%)¶ nd Limited supply; starting a family (females); “sick of fighting when cannabis runs out”; “made me sick”; “mind not straight”; expenses and time spent looking for cannabis; employment (males) * Self-report interview data from an opportunistically recruited sample (using age and sex quotas) of respondents, including people who had never used cannabis as well as current and former cannabis users. † People aged ≥ 18 years.5 ‡ People aged ≥ 14 years.4 § Including ecstasy. ¶ There have been no reliable reports of stimulant, benzodiazepine or barbiturate use in these communities. nd = data not available.

K S Kylie Lee · Alan R Clough · Katherine M Conigrave

To the Editor: I was not surprised by the results published by Fassett et al1 regarding clinical handover. Despite being a crucial part of health care, clinical handover has only recently become a topical issue in the clinical governance arena. In May 2007, the World Health Organization launched the “Nine patient safety solutions” to “help reduce the toll of health care-related harm affecting millions of patients worldwide”.2 Solution number three relates to “communication during patient hand-overs”. Australia is in fact leading the international collaboration on clinical handover through the National Clinical Handover Initiative. This was recently launched by the Australian Commission on Safety and Quality in Health Care to develop and implement standardised solutions to patient safety problems associated with clinical handover.3 At a state level, the Victorian Quality Council conducted a clinical handover survey of Victorian health services in May 2006 to provide an overview of areas of concern relating to clinical handover.4 The Council launched a pilot project in 2007 to look at morning handover processes between junior doctors. Despite these efforts, hospitals are still struggling with the issue of clinical handover at the local level. All health professionals know that clinical handover is good clinical practice, but they need to be provided with the tools to carry it out properly. Clinical handover, depending on your definition, includes referral letters from general practitioners, discharge summaries from hospitals, as well as handover of clinical information between different shifts, treating teams, wards, health professionals and health services. It makes sense that the process of handover of clinical information be carried out in a standardised format, as the minimum dataset required is consistent for most circumstances. While there is a need for standardisation, health organisations must also be able to devise local innovative solutions that work for them. Information technology can assist in developing a solution to this issue. For example, different local health services have already developed in-house systems for electronic discharge summaries that are integrated with an electronic health record. While national quality bodies endeavour to develop national standards and tools for clinical handover, local health services must not sit idly and wait for these, but must continue to innovate and provide workable local solutions for their own health professionals. Otherwise, future surveys of clinical handover will continue to show that a problem still exists in relation to this issue.

Erwin Loh

In reply: Our national survey was specifically confined to medical morning handover report, one of many forms of clinical handover.1 The clinical governance area may have only recently recognised the importance of clinical handover, but it has been a topical issue for many years in the clinical arena, particularly in the United States.2 While Australia may be leading the way in international collaborations, our survey suggests that this is not translating into clinical practice. Rather than commissions and councils providing tools and guidelines to carry out clinical handover, clinical leaders need to participate and conduct clinical handover themselves at the local level. From our previous experience with morning report, we provided simple tips on how to implement clinical handover at the local level,3 and these have been incorporated into Australian Medical Association guidelines.4 Information technology can be employed to help with the clinical handover process. However, when using the all-inclusive definition of clinical handover suggested by Loh, this task is complex, as research has shown that information management needs vary significantly between different clinical environments and would require multiple end-user-defined outputs from a standardised data repository.5,6 Clinical handover needs to be implemented from the bottom up (by clinicians) rather than from the top down (by commissions).

Matthew J Fassett · Terry J Hannan · Iain K Robertson · Steven J Bollipo · Robert G Fassett

To the Editor: I was part of a review of the Therapeutic Goods Administration (TGA)-approved product information (PI) monographs contained in MIMS (Monthly index of medical specialties) annual with respect to their poisoning management advice.1 We looked at the 10 most common poisonings presenting to Westmead Hospital and another 15 clinically important poisonings as determined by two of the authors, and compared the poisoning management advice given in MIMS to a “gold standard” derived from a consensus of five pharmacological resources. For the 25 drugs examined, 14 monographs contained inaccurate information, one contained a recommendation for ineffective treatments, and 14 omitted specific treatments or antidotes. Many of these errors could delay or even prevent patients receiving currently accepted and effective therapies for life-threatening poisonings if MIMS were used as the primary resource. The omission of sodium bicarbonate for ventricular conduction delay and hypotension in amitriptyline, quinine and thioridazine poisonings is particularly problematic. Ventricular conduction delay and hypotension is often refractory to other therapies, and delay in bicarbonate administration could result in avoidable deaths. The recommendation of sodium bicarbonate for ventricular conduction delay and hypotension was included in the TGA-approved monograph for amitriptyline in 1984, but was subsequently removed without qualification in 1990 and remains absent. Cyproheptadine, an important therapy for serotonin syndrome, is not included in the TGA-approved monograph for sertraline, and its delay could result in avoidable morbidity. Atropine for verapamil-induced bradycardia is a simple and intuitive therapy; however, its omission from the TGA-approved monograph could again result in avoidable morbidity and mortality. We also found potentially dangerous treatments recommended in the TGA-approved monographs not covered by the consensus opinion. These included intravenous amphetamine or intramuscular ephedrine to counter the sedative effects of promethazine poisoning, administration of enteric-coated ammonium chloride tablets to increase urinary excretion in chloroquine poisoning, and forced osmotic diuresis using a urea or mannitol infusion for lithium poisoning. These treatments are out of the Dark Ages, and their use should be considered negligent. Based on this and other reports in previous editions of the Journal, the TGA-approved PI monographs contain inaccurate, inadequate, out-of-date or potentially dangerous information relating to poisoning management advice, paediatric drug dosages,2 drug interactions,3 breastfeeding mothers,4 and various thyroid medications.5 How can we use MIMS for anything other than simple drug formulation information? Surely, to anyone who wishes to practise up-to-date, safe, evidence-based medicine, the answer must be that we cannot. It is time for the TGA to take up its role as regulator and insist on updated and accurate PI monographs from the pharmaceutical companies.

James L Mallows

In reply: An article by Stockigt1 and follow-up correspondence from Mallows raised concerns about the limitations of approved product information (PI). The purpose of PI needs to be realised. The document is not intended to act as a textbook of medicine or general management of patients, but is intended to contain sufficient information to allow a health professional, in average circumstances, to use the specified medicine safely and to refer to other sources of information and expertise should they be required. The PI covers uses of the medicine evaluated and approved by the Therapeutic Goods Administration (TGA). The sponsor of the medicine is responsible for maintaining the PI, and the TGA has procedures in place to support their timely updating. On occasions, the TGA initiates reviews of PI when a need is identified. Health professionals, especially specialists, are important in identifying possible improvements in PI, based on their experience, knowledge or awareness of current medical practice. The TGA encourages physicians who have suggestions to approach the sponsor of the medicine or the TGA. It is not possible to address in detail the article by Stockigt1 or the letter by Mallows. The thyroid PI documents have been reviewed by the sponsor companies, and changes have been made where evidence supports this. However, as indicated above, there are limitations on the role of PI. Mallows raises the issue of complex management instructions on overdosage, including the specifics of bicarbonate administration for potential metabolic acidosis. The PI documents for the named products do mention that overdose patients are likely to develop such complications, and that they require admission to hospital and management by appropriate specialists; intensive care admission is recommended in several of the documents. It is arguable how much further detail is required. The PI cannot replace careful consideration of the individual circumstances of the patient combined with expert knowledge of patient management. PIs are complex documents. The TGA is currently considering the format of the PI and whether it can be rearranged to better balance provision of basic messages and more complex material in separate presentations.

David T Graham

To the Editor: The recent conference report by Knight and Taruscio highlights the need for a coordinated effort to fill knowledge gaps and improve service provision for Australians with rare diseases.1 Although, by definition, individual rare diseases occur infrequently, there are about 6000 rare diseases affecting 6%–10% of the population.2,3 This equates to 1.2 million Australians, 30 million people in Europe and 25 million in the United States. By comparison, diabetes affects 1.4 million Australians.4 It is increasingly acknowledged that low prevalence does not equate to low impact. Rare diseases often have their onset in childhood, continue throughout life, are difficult to diagnose, are disabling, and have significant impact on patients, their families, the community and health services.2,5 However, rare diseases receive such scant attention that they have been dubbed “orphan” diseases. Lack of epidemiological and scientific data has hindered development of evidence-based practice, policy and services. To tackle the problem of rare diseases, the European Union, the US, Canada and New Zealand have established policies and agencies to foster research, and develop resources for clinicians, community information services, and appropriate health facilities. The National Institutes of Health in the US established the Office of Rare Diseases because: . . . rare disease research requires the collaboration of scientists from multiple disciplines and the capacity to share access to geographically distributed national research resources and patient populations . . . knowledge about rare diseases may offer leads for scientific advancement in other rare diseases and in more common diseases.6 There is no coordinated national effort or policy in Australia. Currently, there are 14 national paediatric surveillance units, including the Australian Paediatric Surveillance Unit (APSU), to which paediatricians contribute epi-demiological, clinical and outcome data on rare conditions of childhood.7 These data inform development of health policy and improved diagnosis and clinical management, and result in the establishment of cohorts, thereby enabling further research.7 The APSU is developing information resources for clinicians and the community on rare infections, genetic disorders, mental health conditions and injuries in children. In Australia, the APSU is the only provider of prospective national data on up to 16 rare childhood diseases concurrently, but it receives no ongoing core funding. Sound evidence is needed to underpin development of policy and services. Sound evidence requires sound research into the causes, management and effects of rare diseases. Australian clinicians, researchers and, most importantly, patients and their families deserve the benefits of a coordinated national plan to address the common burden of rare diseases.

Yvonne A Zurynski · Katie N Reeve · Elizabeth J Elliott

To the Editor: In recent years, there has been an increase in the use of highly mobile three-wheel strollers that facilitate parental activities such as jogging. Unfortunately, the very design feature that enables fast transit over uneven ground also makes it possible for unsupervised strollers to move rapidly into situations that may be highly dangerous. Within the past year in South Australia there have been two separate incidents where infants, one aged 5 months and the other aged 10 months, died after being immersed in the Torrens River. They had both been strapped into three-wheel strollers. In both instances, carers, who had been walking or jogging in the park along the banks of the river, were momentarily distracted — one by a mobile phone call and the other while attempting to use a plastic bag dispenser. The strollers had not had their front wheels locked or their brakes engaged, or been attached to the carers by wrist straps and so were unrestrained, enabling them to roll rapidly forwards into the water. Police re-enactments confirmed the scenarios described by the carers. While three-wheel strollers have safety features, such as brakes and sometimes wrist straps, these are not always used. Although consumer organisations have listed a series of recommendations for users of these strollers, this advice is not always being followed. The recommendations include never leaving a child unattended in one of these strollers, always using a wrist safety strap, always engaging the brake when stationary, and locking the front wheel when jogging to prevent swivelling. In addition, specific warnings are issued about stopping on slopes, being distracted by mobile phone calls, and being particularly careful near water, roads and railway lines.1 Drowning of infants and toddlers in rivers is an uncommon event, with only two cases documented of a total 32 drowning deaths of children under the age of 2 years in South Australia over the 35 years from 1963 to 1998 (rate, 6.25%).2 Thus, the occurrence of two drowning deaths within 4 months associated with three-wheel stroller use in parks next to a river is of concern. While mandatory requirements for safety devices such as parking brakes and tether straps will take effect on 1 July 2008,3 this legislation will have little effect if the devices are not used. Parents and carers must be made aware that infants or toddlers in three-wheel strollers near water are at risk of immersion and drowning. Such warnings should be clearly specified on these products.

Roger W Byard · Neil Matthews

To the Editor: A 30-year-old man presented with pain, swelling and discharge from lesions on his left arm. He had undergone extensive tattooing on the arm 3 days earlier at a commercial tattoo operation, where single-use needles were used, with initial drawing of lines followed by additional shading. The patient complained of severe neuropathic pain in the arm, which was greater than would be expected from uncomplicated bacterial cellulitis. He did not give a history of oral or genital herpes and had previously been tattooed without complication. On presentation, the patient had a low-grade fever (37.9°C), a heart rate of 80 beats/min and blood pressure of 130/80 mmHg. He had no neurological deficit. Vesicular lesions were visible in the region of the tattoo marks, predominantly affecting areas of shading and with minimal spread outside tattooed areas (Box). A bacterial swab of the lesions grew methicillin-sensitive Staphylococcus aureus, and a polymerase chain reaction test of vesicular fluid was positive for herpes simplex virus type 1 (HSV-1). The patient was commenced on intravenous flucloxacillin (1 g four times daily) and oral famciclovir (250 mg three times daily). He required ongoing inpatient management for pain relief. Five days after development of the lesions, HSV-1 serology demonstrated positive results for IgM and IgG. An HIV test was negative. The patient’s lesions slowly resolved, and he was discharged 7 days after admission. Most concerns regarding infectious complications of tattooing have focused on transmission of blood-borne viruses, but superficial infections with other pathogens have also been described.1 The personal care and body art industries are regulated in Australia to minimise the transmission of blood-borne infection,2 and most state and territory authorities also publish infection control guidelines. Herpes dermatitis is often confused with bacterial infection, although co-infection may occur. This distinction is clinically important, as antibiotics and surgical debridement are not usually required for herpetic infections, and herpetic lesions may recur. Secondary herpetic infection complicating skin disease is most commonly associated with eczema (eczema herpeticum) or other skin diseases (Kaposi’s varicelliform eruption), and minor skin trauma, such as in herpetic whitlow or herpes gladiatorum.3,4 We are not aware of any previous reports of herpetic infection complicating tattoo placement. The distribution of herpetic lesions in our patient suggested that the needle used for tattoo shading became contaminated with HSV-1 during the course of tattoo placement, but it is also possible that superinfection occurred through damaged skin after the procedure. We propose the term “herpes compunctorum” to describe this condition. Vesicular lesions on patient’s tattooed forearm

Catherine S Marshall · Felicity Murphy · Shannon E McCarthy · Allen C Cheng

To the Editor: Helping patients who have had a stroke return to driving when possible should be an important focus in rehabilitation wards. The National Stroke Foundation supports a three-stage approach to assessing ability to drive, comprising physical and cognitive assessment, an off-road driving test and an on-road driving test.1 Austroads is the association of Australian and New Zealand road transport and traffic authorities, which aims to improve road and road transport outcomes. It provides clear guidelines on criteria for licence and assessment after stroke, but the implementation of these guidelines varies in practice.2 We conducted a review of 53 consecutive patients with a primary diagnosis of stroke admitted to a specialised rehabilitation ward over a 6-month period between January and July 2007. The mean age of the sample population was 77.0 years (SD, 10.7 years), and cognition score (Functional Independence Measure) on discharge was 28.3 (SD, 7.1) (maximum possible score, 35). Each patient completed a survey on driving history. Case notes were reviewed for medical factors associated with admission, any notations about driving, and actions taken regarding driving. Patients were telephoned 6 months after the date of the stroke to determine whether they had resumed driving and, if not, to explore the reasons. At the time of admission, 26 of the 53 patients held a current drivers licence, a proportion higher than the South Australian state rate of less than 10% for people aged 75 years and over. At the time of discharge, 12 of the 26 patients had their licences cancelled, 11 were referred for medical review after discharge without formal suspension, two were referred for occupational therapy driving assessment, and one was advised not to drive for 6 weeks. At 6 months, only five of the 26 patients (19%) had resumed driving, with one having regained a cancelled licence; six patients cited “lack of confidence” as the reason for not resuming driving. Reasons for the doctors’ decisions regarding driving were poorly documented in the case notes, and the time frame proposed for medical review ranged from 1 to 4 months. Austroads requires a minimum time of 4 weeks post-stroke before patients can resume driving, but does not specify a time frame for medical or on-road reassessment. Commonly, patients undergoing acute rehabilitation are still within the 4-week period, and assessments regarding return to driving are premature. The low rate of referral to available occupational therapy on-road driving assessment may reflect poor awareness of available hospital resources and online guidelines. While overseas studies of post-stroke populations indicate a return-to-driving rate between 30% and 58%,3-5 our rate was 19%. This low rate may represent a lack of formal assessment and driver rehabilitation opportunities. The benefits of formal driving assessment and training are supported by recent studies which found that licensed drivers post-stroke did not have an increased incidence of either car accidents or driving violations.6 While most doctors in the rehabilitation ward seemed to have understood the need to address the issue of driving, more formal training in this field is required for doctors.

Zoe A Allen · Julie Halbert · Lydia Huang

HPV DNA v Pap smears Human papillomavirus (HPV) testing appears to have a greater sensitivity for the detection of cervical intraepithelial neoplasia when compared with conventional Pap testing, according to a Canadian randomised controlled trial. Both tests were performed in a randomly assigned sequence on 10 154 women aged 30 to 69 years presenting for cervical screening. Participants with a positive Pap or HPV screening test were referred for colposcopy, as were a proportion of those with negative tests. HPV testing was more sensitive by 39.2% and only 2.7% less specific than Pap testing for the detection of cervical intraepithelial neoplasia grades 2 or 3. The authors note that testing for HPV has entered a new era, with the introduction of HPV vaccines changing how women will be tested in the future. They suggest that the HPV DNA test may prove useful in screening those women who undergo screening less frequently. N Engl J Med 2007; 357: 1579-1588 Maternal mortality — sad statistics Maternal mortality rates remain high in sub-Saharan Africa despite a worldwide trend of decreasing maternal deaths, a global study has found. Although research efforts have been hampered by lack of records and varying data collection methods between areas, the multinational Maternal Mortality Working Group has summarised available data on maternal deaths by country. The researchers estimated that there were 535 900 maternal deaths worldwide in 2005, with most deaths concentrated in sub-Saharan Africa and Asia. This corresponds to a ratio of 402 deaths per 100 000 live births. All countries with data showed a decrease of 2.5% per year between 1990 and 2005, but no such reduction was seen in the African nations. To put these figures in perspective, the estimated maternal mortality ratio for Australia in 2003 was 4 per 100 000 live births. In Niger in 2005, it was 1800. As the authors note, “progress is slow”. Lancet 2007; 370: 1311-1319 Antioxidants disappointing for AMD Dietary antioxidants may not be as effective in the prevention of age-related macular degeneration (AMD) as previously thought, a systematic review and meta-analysis has suggested. Data from 12 studies involving over 149 000 participants were pooled, evaluated and analysed by the Australian research team in an effort to judge the effect of antioxidants such as vitamins A, C, and E, zinc, and β-carotene. Preventive therapy with antioxidants has been based on the susceptibility of the retina to oxidative damage. Whereas previous studies have suggested a protective effect of antioxidants on the progression of intermediate to advanced AMD, the evidence for the usefulness of such therapies in the prevention of early AMD is not conclusive. Although the current review indicated that a range of antioxidants have little or no effect on the development of early AMD, vitamin E had a borderline protective effect. BMJ Online, 8 Oct 2007 Battling Bell’s palsy Early treatment of idiopathic facial paralysis (Bell’s palsy) with corticosteroids appears to significantly improve the outcome for affected patients, but the usefulness of anti-viral agents is questionable, according to Scottish researchers. A total of 496 patients were assessed in the double-blind, placebo-controlled, randomised trial. All participants were recruited within 72 hours of developing Bell’s palsy, and were randomly assigned into four groups treated with prednisolone, aciclovir, both agents, or placebo. Recovery of facial function at 3 months was the primary outcome. Early treatment with prednisolone appeared to significantly improve the chance of recovery of facial function (83%, compared with 64% in those not receiving prednisolone). Aciclovir alone or in combination with a corticosteroid did not show any benefit. N Eng J Med 2007; 357: 1598-1607 Stateside superbugs Invasive methicillin-resistant Staphylococcus aureus (MRSA) infections in the United States are no longer confined to a health care setting, and their incidence is increasing, according to a population-based study. Almost 9 000 observed cases of invasive MRSA in nine geographic sites were reported over the 18-month surveillance period. The authors extrapolate these findings to estimate that close to 95 000 such infections occurred in the US in 2005, with almost 19 000 deaths due to the infection. Although the majority of invasive infections occurred outside the hospital setting, most of these individuals had established risk factors for MRSA, such as a relatively recent (within the past 12 months) history of hospitalisation. Molecular analysis of isolates from these patients suggests a health care origin for the pathogen responsible. JAMA 2007; 298: 1763-1771

Tanya Grassi