Letters

Michael Thomsen,* Luis Vitetta,† Avni Sali,‡ Matthias Schmidt§ * Research Associate, † Deputy Head, and Director of Research, ‡ Head, Graduate School of Integrative Medicine, Swinburne University, 9 Frederick Street, Hawthorn, VIC 3122; § Research Head of Toxicology, Society of Nutritional Medicine and Dietetics, Harsewinkel, Germany. LVitettaATmedicine.swin.edu.au To the Editor: We wish to comment on the case report by Lontos and colleagues on the proposed causal relationship between the herb black cohosh (Cimicifuga racemosa) and acute hepatic failure.1 One other case has been reported in Australia,2 and the evidence linking black cohosh to liver toxicity was weak and contested.3 The medication in the case report presented by Lontos and colleagues1 included a herb (ground ivy) containing a known liver toxin (pulegone). Pulegone is considered a strong hepatotoxin and should not be dismissed, even though it was reported that there was less pulegone in ground ivy than in pennyroyal. The authors do not indicate the daily dose of the pulegone ingested. The Therapeutic Goods Administration (TGA) made only qualitative analyses of three of the five herbs in the mixture. In our opinion the most suspect ingredient was ground ivy, and it was not assayed. The argument that the TGA could not find a standard for pulegone or ground ivy is untenable given the level of expertise and the capacity of the TGA and its laboratories. Was the supply company asked to provide analytical evidence of the contents of the herbal extracts? Ground ivy is not known to be hepatotoxic, but it is possible that the extract could have contained ground ivy with pennyroyal, which might explain the hepatoxicity of the mixture. Although uncertain without thoroughly investigating all ingredients in the herbal mixture, this alternative is a possibility. Without thorough investigation of herbal preparations, adverse events attributed to certain herbs remain dubious at best. Black cohosh has a very good safety record. There is a large body of clinical evidence and research which suggests that this herb has no hepatotoxic effects. Indeed, a German manufacturer has sold more than 350 million daily doses of black cohosh preparations worldwide since the pharmacovigilance system was introduced, and no comparable cases have been reported until these two reports in Australia. An Ames test (salmonella microsomal assay) showed no in-vitro evidence of mutagenic potential of an extract of black cohosh,4 and no chemical or organ toxicities were observed in Wistar rats given up to 5000 mg of a Cimicifuga racemosa extract granulate per kilogram body weight for 26 weeks.5 What may also be of concern is that the TGA may not have the full capabilities to test ingredients used in Australian herbal products. That substitution of herbs with potentially toxic herbs may be a common event, and that neither the TGA nor the manufacturers may have the expertise to prevent deleterious contaminants, is of even greater concern.

Michael Thomsen · Luis Vitetta · Avni Sali · Matthias Schmidt

Fiona J Cumming,* Larry Kelly† * Director, Office of Complementary Medicines; † Acting Director, TGA Laboratories; Therapeutic Goods Administration, PO Box 100, Woden, ACT 2606. fiona.cummingAThealth.gov.au Comment: Thomsen and colleagues have pointed to what they believe were uncertainties in the case described by Lontos et al, associating the use of a herbal preparation containing black cohosh with acute liver failure.1 They assert that on the basis of these uncertainties, and other evidence, black cohosh was unlikely to have been the cause of the liver failure. There are indeed uncertainties about the cause of the illness. The first, and quite critical one, is that we cannot be certain that the patient was given the same herbal materials as those supplied to the Therapeutic Goods Administration (TGA) for analysis. The pharmacist who dispensed the herbal formula for the patient provided the TGA with samples for analysis of each of the five individual extracts in the formula. Documentary evidence supporting the correct identity of the five herbs was supplied in the form of certificates of analysis from the manufacturers of the individual herbal extracts. However, it was not possible to ascertain whether the exact batch of the exact formulation taken by the patient was that which was tested by the TGA. While the title of the article by Lontos et al suggests a “problem” with black cohosh, the body of the article makes clear that the causative agent(s) are unknown,1 which is in agreement with the letter by Thomsen et al.1 Thomsen and colleagues claim that the TGA did not test the ground ivy extract supplied by Lontos et al, and said that this was because the TGA did not have a standard for pulegone or ground ivy. This is not the case, and the letter by Lontos et al1 does not state that either. The reason the TGA did not test initially for pulegone was because it is a minor constituent of the essential oil of ground ivy. We did not expect there to be any significant levels of this compound. In fact, we have subsequently confirmed that pulegone was not detectable (limit of detection 5 ppm) in the sample of ground ivy extract provided by Lontos et al. Thomsen and colleagues refer to the good safety record of black cohosh in Australia and internationally, and claim there has been only one other case in Australia where black cohosh was linked to liver toxicity. Some caution is needed in drawing broad conclusions about the safety of herbal medicines. Most countries do not have adverse reaction reporting systems which include herbal remedies. In Australia, where we have a well-developed reporting system, there have been several reports of liver problems in patients taking various preparations containing black cohosh. However, causality has not been established beyond doubt in these cases. The TGA has state-of-the-art testing facilities and a team of internationally recognised scientific staff. The formulation supplied to the patient was extemporaneously dispensed by a pharmacist. Such medicines are not subject to the regulatory controls of the TGA and would not normally be included in the TGA’s testing program. However, the TGA offered to test the herbal formulation supplied by the pharmacist to assist the clinical team. The results of the TGA’s testing confirmed the absence of undeclared pharmaceuticals in the samples of herbal extracts provided. In view of these uncertainties, it is not possible to conclusively identify the cause of the patient’s liver failure. On the evidence available it cannot be concluded that black cohosh was a cause. It is simply not possible to rule in or rule out black cohosh, or indeed any of the other herbal extracts in the formulation taken by the patient, as a cause. Where practitioners suspect a complementary medicine is involved in an adverse reaction, providing the exact product and batch taken by the patient is essential if laboratory testing is to help in confirming causality.

Fiona J Cumming · Larry Kelly

Leon A Flicker,* Samuel D Vasikaran,† Jenny Thomas,‡ John G Acres,§ Paul E Norman,¶ Konrad Jamrozik,** Nicola T Lautenschlager,†† Peter J Leedman,‡‡ Osvaldo P Almeida§§ * Professor of Geriatric Medicine, ‡ Research Nurse, School of Medicine and Pharmacology, § Research Fellow, School of Medicine and Pharmacology and School of Psychiatry and Clinical Neurosciences, ¶ Associate Professor of Surgery, †† Senior Lecturer in Psychiatry of Old Age, ‡‡ Professor of Medicine, §§ Professor of Psychiatry of Old Age; University of Western Australia, Royal Perth Hospital, Box X2213, Perth, WA 6000. † Head, Department of Core Clinical Pathology and Biochemistry, Royal Perth Hospital, Perth, WA. ** Professor of Primary Care Epidemiology, Imperial College London, London, UK. leonflicATcyllene.uwa.edu.au To the Editor: Elevated levels of homocysteine (Hcy) have recently been associated with increased risk of vascular events1 and dementia.2 The clearance of Hcy is dependent on three vitamins — folate, B6, and B12. Vitamin B12 deficiency has been described in older people for over 40 years,3 and may have wide-ranging effects through this vitamin’s influence on Hcy. The aims of this study were to examine serum B12 and folate status, and their relationships with plasma Hcy concentrations, in community-dwelling healthy older people living in Perth. Older men and women were recruited from two different sources: 299 men aged 75 years and over were recruited from a large population-based study of screening for abdominal aortic aneurysm,4 where 70% of those invited joined the project; and we recruited 273 community-dwelling women aged 70 years and over through advertisements. Exclusion criteria for both groups included significant cognitive impairment, severe physical illness and current use of B-group vitamin supplements. The Human Research Ethics Committee at the University of Western Australia approved the study, and all participants provided informed consent. Fasting total plasma Hcy, serum B12 and folate concentrations were measured in all participants, and serum creatinine concentration was measured in the men only to calculate glomerular filtration rate (cGFR). For analyses, the variable plasma Hcy was heavily skewed to the right and natural logarithmic transformation was used. Pearson’s product moment correlations were calculated for univariate analyses of continuous variables. Descriptive statistics are presented in Box 1. Fourteen per cent and 1% of the men, and 6% and 1% of the women, were deficient in B12 and folate, respectively. Hcy concentrations above upper reference limits (15 μmol/L for men and 13 μmol/L for women) were found in 24% of both men and women. There were significant (P < 0.001) positive correlations between age and log Hcy concentration for men (r = 0.23; 95% CI, 0.12–0.33) and women (r = 0.25; 95% CI, 0.13–0.36), inverse correlations between B12 and log Hcy concentrations for men (r = – 0.25; 95% CI, – 0.14 to – 0.35) and women (r = – 0.30; 95% CI, – 0.19 to – 0.41), and inverse correlations between folate and Hcy concentrations for men (r = – 0.43; 95% CI, – 0.33 to – 0.52) and women (r = – 0.28; 95% CI, – 0.16 to – 0.39). Plots of log Hcy against B12 and folate concentrations for all participants are presented in Box 2. Under multiple regression, the association of B12 and folate concentrations with log Hcy concentration remained after adjustment for age and cGFR in men only; beta values (SE) were: – 0.00060 (0.00011) for B12 concentration; – 0.0155 (0.0017) for folate concentration; – 0.0029 (0.0008) for cGFR; and 0.012 (0.005) for age. In this sample there were high prevalences of B12 deficiency and hyperhomocysteinaemia. Although the prevalence of folate deficiency was substantially lower, there were still moderate inverse associations between serum folate and Hcy concentrations. Unfortunately, vitamin B12 deficiency of this kind may not be universally corrected with small doses of oral supplements,5 and this has intensified concerns about precipitating neurological complications by population-based folate supplementation.6 There is a need for intervention studies of B-group supplements to evaluate whether the risks associated with hyperhomocysteinaemia can be ameliorated. 1: Demographic characteristics, serum B12 and folate, and plasma homocysteine in 299 older men and 273 older women Men Women Mean (SD) Range Mean (SD) Range Age (years) 78.9 (2.8) 68–86 74.8 (4.4) 70–92 Weight (kg) 78.4 (1.2) 50.6–119.5 69.3 (1.3) 39.0–120.0 Height (cm) 171 (6.5) 150–197 159 (6.7) 132–176 Body mass index (kg/m2) 26.6 (3.5) 16–37 27.4 (5.3) 17–52 Ever smoked 66% 41% Ever drank alcohol 95% 66% Serum folate (nmol/L) 24.3 (7.6) 5.5–45.3 (RI, 7–34) 25.3 (7.6) 3.9–45.2 (RI, 7–34) Serum B12 (pmol/L) 254.5 (116.7) 57–890 (RI, 140–646) 313.5 (158.7) 59–1270 (RI, 140–646) Plasma Hcy (μmol/L) 13.50 (5.3) 6.7–70.5 (RI, 6.0–15.0) 11.46 (6.8) 3.8–96 (RI, 5.0–13.0) Glomerular filtration rate (mL/min) 78.3 (16.3) 35.8–142.4 SD = standard deviation. Hcy = homocysteine. RI = reference interval. 2: Plot of serum B12 and folate concentration against log homocysteine concentration (with regression line for B12) in 299 older men and 273 older women

Leon A Flicker · Samuel D Vasikaran · Jenny Thomas · John G Acres · Paul E Norman · Konrad Jamrozik · Nicola T Lautenschlager · Peter J Leedman · Osvaldo P Almeida

Malcolm H Parker Associate Professor of Medical Ethics, School of Medicine, University of Queensland, Herston Road, Herston, QLD 4006. m.parkerATuq.edu.au To the Editor: Gerber1 appears sympathetic to the following reasons for rejecting damages in cases of wrongful birth, quoted from the High Court minority in Cattanach v Melchior,2 and judges in similar cases: doctors do not owe a duty of care that protects the economic interests of patients; the birth of a healthy child should not be regarded as a legal harm, because the birth of a healthy child is a good thing, the cost of rearing a child does not exceed the value of parenthood, and parents ought not to enjoy the advantages of parenthood without the concomitant responsibilities; awarding damages would indicate to the child that he or she was unwanted; and doctors should not be liable for most of the costs of rearing a child, because parents have a choice of rearing the child or surrendering it for adoption. As to the first point, why should doctors be exempt from any loss that results from their negligence? Depicting this duty as one of “protecting economic interests” distorts the nature of the duty of care, which requires the duty holder to avoid foreseeable damage, which includes economic loss consequent on the negligence. Regarding the second point, balancing the costs of rearing the child against the value and advantages of parenthood is conceptually misleading. The implication is that the value of parenthood should render the costs of rearing the child relatively trivial. A more coherent interpretation is that, because the costs of rearing the child and the value of parenthood are incommensurable, the natural love for the child and the value of parenting that follow the birth ought not discount the damage. This interpretation also informs responses to the third and fourth points. Not wanting another child, and not wanting the particular child once it exists, are distinct concepts. The idea that Jordan Melchior was not wanted at the time his parents took steps to avoid further pregnancies is incoherent. Wanting to avoid further pregnancies and wanting to nurture the child who is born are perfectly consistent positions, whether the case involves forgotten contraception or negligent sterilisation. Finally, the idea that, because they have the choice to keep or surrender the child, the parents should bear the costs if they keep it suggests that, once born, the particular child can be regarded as a commodity — something that the general position against awarding damages for wrongful birth repeatedly disavows. One of the judges quoted by Gerber claimed that such children would come to think of themselves as unwanted, and that this was “obscene”. The same judge’s glib claim that the parents can choose to keep or surrender the child strikes me as the obscenity.

Malcolm H Parker

Paul Gerber Honorary Reader in Legal Medicine, University of Queensland, Brisbane, QLD. In reply: Parker’s letter does little more than repeat the arguments that found favour with the majority in the High Court. He approaches the issues raised by this controversial litigation from an ethical perspective. Alas, the parents’ claim for the cost of raising a healthy child, conceived as a result of the alleged negligence of the defendant gynaecologist, raises the legal issue of restitution: does the law of tort recognise this head of damages as a “loss” for which parents may be compensated? Courts have, in the past, answered this question by reference to general principles based upon legal values. In the Melchior case,1 the majority departed from that hallowed principle. So be it. Does that make Jordan Melchior a “commodity”, having a commercial value? Yes! The plaintiffs faced the choice of either keeping their son, or mitigating their “loss” by placing him for adoption. They chose the former. Priestley JA had, in an earlier case, put the issue succinctly: “After that decision was made, the defendant was not legally responsible for the parents’ financial cost of rearing the child”2 (this was restated by Kirby J in the Melchior case1). Parker may not like it, but the law of tort has — up till now — shown more caution in awarding damages for what is called pure economic loss (ie, loss affecting purely financial interests) than it has in relation to conduct that causes damage to person or property. Before the Melchior case, that distinction had been firmly embedded in the law of tort and formed the basis of established rules governing liability for damages.3 That distinction has now been blurred by placing a financial value on the parent–child relationship.

Paul Gerber

Bronwyn A L Crawford,* Channa Perera† * Endocrinologist and Clinical Senior Lecturer, Royal Prince Alfred Hospital and the University of Sydney, Sydney, NSW 2050; † Endocrinologist, Orange Base Hospital, Orange. brcrawfoATmail.usyd.edu.au To the Editor: Prolonged hypoglycaemia in patients taking a sulfonylurea may be refractory to intravenous glucose treatment with fatal consequences, as described by Veitch and Clifton-Bligh.1 Although these authors briefly mention the use of octreotide, we believe that an additional point in the “Lessons from practice” should have been: Octreotide may be an effective therapy in refractory sulfonylurea-induced hypoglycaemia. We describe the first two patients in whom we used this therapy. A 76-year-old man with type 2 diabetes was admitted after an acute myocardial infarction and cardiac arrest. He was successfully resuscitated and underwent emergency bypass surgery. His diabetes was controlled with gliclazide 80 mg twice a day. After surgery, he developed cardiac failure, renal impairment (serum creatinine level, 0.22 mmol/L) and frequent hypoglycaemic episodes. The gliclazide was stopped but, despite good oral dietary intake, hypoglycaemia worsened and failed to respond to vigorous intravenous glucose therapy. He suffered a hypoglycaemic seizure (blood sugar level, 0.8 mmol/L). Over the next day, he received more than 300 g of glucose in the form of a 10% glucose intravenous infusion, but, despite this, went into hypoglycaemic coma. Blood results were: insulin, 472 pmol/L (reference range [RR], 15–60 pmol/L); C-peptide, 7616 pmol/L (RR, 300–800 pmol/L); and gliclazide, 9.4 mg/L (steady state average, 2.5 mg/L). He was given an intravenous infusion (30 ng/kg per minute) of octreotide.2 Within an hour, his blood sugar level rose to 7.9 mmol/L and continued to rise. Dextrose and octreotide infusions were ceased within 13 hours, with no further episodes of hypoglycaemia. He was discharged home 2 days later. A 75-year-old man with type 2 diabetes was taking glibenclamide 2.5 mg each morning. He was transferred from a rural hospital with acute on chronic renal failure (serum creatinine level, 0.4 mmol/L), as well as recurrent hypoglycaemia. The glibenclamide was stopped, but blood sugar levels remained low, and he became comatose despite boluses of 50% dextrose and a continuous infusion of 10% dextrose. The high volume of intravenous fluid precipitated cardiac failure and pulmonary oedema, requiring inotropic support. Blood results were: blood sugar, 1.5 mmol/L; insulin, 1250 pmol/L; C-peptide, 20 949 pmol/L. As an alternative to the high-dose, continuous infusion of octreotide used in our first patient, we administered a single subcutaneous injection of octreotide 50 μg. One hour later, the patient’s blood sugar level had risen to 9.0 mmol/L. He had no further episodes of hypoglycaemia. Eight hours later, insulin and C-peptide levels had fallen markedly (insulin, 153 pmol/L; C-peptide, 5654 pmol/L). He made a full recovery. Octreotide is a somatostatin analogue that inhibits the secretion of a number of neuropeptides, including insulin. It is a safe and effective therapy for sulfonylurea-induced hypoglycaemia when initial therapy with oral or intravenous glucose fails.2-5 It is particularly useful in elderly patients with renal or cardiac complications, in whom fluid overload may be a limiting factor in intravenous dextrose therapy. Octreotide may also break the vicious circle that can occur in sulfonylurea-induced hypoglycaemia in which repeated dextrose boluses further stimulate insulin release.

Bronwyn A L Crawford · Channa Perera

Peter C Veitch,* Rory J Clifton-Bligh† *Specialist in Geriatric Medicine, Department of Aged Care and Rehabilitation Medicine; † Registrar in Endocrinology, Department of Endocrinology, Royal North Shore Hospital, Clinic 1, Level 3, Pacific Highway, St Leonards, NSW 2065. rcliftonATmed.usyd.edu.au In reply: Crawford and Perera highlight a very important point with respect to hospital-based treatment of sulfonylurea-induced hypoglycaemia. Whereas, in our article,1 we wished to emphasise the importance of recognising and preventing this condition in primary care settings, we agree that octreotide is an effective therapy in treating this condition. Octreotide inhibits the specific effect of sulfonylureas (glucose-stimulated β-cell insulin release) and prevents rebound hypoglycaemia, which, in this situation, may occur with the use of glucose.2 Case reports, including those elegantly presented and referenced by Crawford and Perera, clearly illustrate the safety and efficacy of octreotide in treating sulfonylurea toxicity when initial therapy with glucose fails. Octreotide may be administered either intravenously or subcutaneously, and its effect is maintained after a short course of therapy. At our own institution, we have now adopted guidelines for the treatment of refractory sulfonylurea-induced hypoglycaemia, which include the administration of octreotide (50 μg subcutaneously) every 8 hours for up to three doses, although, as Crawford and Perera note, some patients will have a sustained response after a single dose.

Peter C Veitch · Rory J Clifton-Bligh

Malvinder S Parmar Medical Director (Internal Medicine), Timmins and District Hospital, Suite 108, 707 Ross Ave East, Timmins, ON P4N 8R1, Canada. atbeatATntl.sympatico.ca To the Editor: I wish to add another cause of exogenous hyperthyroidism to those mentioned in the comprehensive review on hyper- and hypothyroidism by Topliss and Eastman.1 Inadvertent ingestion of animal thyroid (“hamburger” thyrotoxicosis), although rare, is worth mentioning. Meat may be inadvertently contaminated with thyroid tissue through the process of “gullet trimming” during butchering. While this process has been prohibited in most countries since the recognition of outbreaks of hamburger thyrotoxicosis,2,3 it may still occur when farm animals or wild game are prepared for consumption by farmers, hunters or local butchers unaware of the prohibition. I recently reported a case of a woman living on a farm in Canada who had five episodes of transient hyperthyroidism over a decade.4 These were initially diagnosed as episodes of “silent thyroiditis”, but were later attributed to consumption of meat patties contaminated with thyroid tissue, as the local butcher was not aware of the prohibition on gullet trimming. A history of eating wild game or locally prepared meat should be considered before a diagnosis of silent thyroiditis is made. Thyroid uptake of radioiodine is low in both conditions, but serum thyroglobulin level is raised in thyroiditis and decreased during the hyperthyroid phase of exogenous hyperthyroidism.

Malvinder S Parmar

Ngaire T Jones Medical Practitioner, Beaconsfield, WA. ngairejATbigpond.com To the Editor: The recent article on thyroid disorders by Topliss and Eastman notes that “around the world, iodine deficiency still remains the predominant cause of hypothyroidism” and furthermore that “mild iodine deficiency is re-emerging in Australia”.1 Indeed, the Journal has recently published at least two articles suggesting that the iodine status of the Australian population needs to be further explored.2,3 My question therefore is: when treating a patient who has results indicating clinical or subclinical hypothyroidism, would it be relevant and important to test for iodine deficiency (by 24-hour urine collection)? This seems analogous to undertaking iron studies in a patient with a falling haemoglobin level. In the same way that iron deficiency can exist and produce symptoms, even in the absence of anaemia, may not iodine deficiency affect health and well-being? Without the elemental “building blocks” of iron and iodine, the relevant systems are put into overdrive to no avail. It seems simple to test routinely for this possibility, correct any deficiency and then recheck thyroid function. There may be more “clinically significant iodine deficiency” than we realise. As it will no doubt be some time until further studies in the Australian population shed more light on this, is it not relevant meanwhile to at least check for this possibility in individual patients?

Ngaire T Jones

Duncan J Topliss,* Creswell J Eastman† * Director, Endocrinology and Diabetes, Alfred Hospital, Commercial Road, Melbourne, VIC 3004; † Director, Institute of Clinical Pathology and Medical Research, Westmead Hospital, Sydney, NSW. Duncan.toplissATmed.monash.edu.au In reply: Jones asks if iodine excretion should be measured routinely in all patients with hypothyroidism in Australia as part of the initial assessment. We do not advocate this for the following reasons. Urinary iodine estimations are unreliable for assessing individual patients, as urinary iodine levels can vary considerably from day to day with iodine intake. These measurements should be reserved for population studies to provide an overall assessment of iodine nutrition in that population. In Australia, it is probable that virtually all cases of primary hypothyroidism are caused by chronic autoimmune lymphocytic thyroiditis, ablative therapy for Graves’ disease, or inadequate thyroxine replacement therapy in these conditions, as is the case in the United States and the United Kingdom.1-3 This contention is supported by data from the Busselton (Western Australia) survey on thyroid peroxidase antibody levels,4 which suggest that these antibodies will be of great diagnostic assistance, in contrast to the dubious clinical value of individual measurement of iodine excretion. In support of this view, in a large survey of the US population, where iodine intake has probably fallen similarly but not to the same degree as in Australia, there was no association between low urinary iodine levels and increased serum levels of thyroid stimulating hormone (TSH).5 In that study, the significant association between raised TSH and female sex disappeared after controlling for the presence of thyroid peroxidase antibodies, while the prevalence of clinical hypothyroidism was strongly correlated with positive results for these antibodies. In Australia, current information indicates that iodine deficiency, where it exists, is mild. The prevalence and regional variation are currently the subject of the National Iodine Nutrition Survey, which is surveying iodine excretion and thyroid size in primary school children. There is no evidence that this mild deficiency is associated with an increased prevalence of hypothyroidism. Jones’s suggestion that iodine deficiency impairs health by a mechanism other than impairment of thyroid function is not supported by any scientific evidence. Any consideration of advocating routine assessment of iodine status should await the results of the ongoing national study. However, iodine nutrition would appear to be best addressed as a public health issue, by promoting use of iodised salt and ensuring adequate iodine nutrition in pregnant and breastfeeding women and their infants.

Duncan J Topliss · Creswell J Eastman

David S Elder Occupational Physician, 517 St Kilda Road, Melbourne, VIC 3004. delderATbigpond.net.au To the Editor: In Bogduk’s review of the management of low back pain,1 he cited several international guidelines but did not address the effect of returning the patient to work. Disappointingly, return to work was mentioned only as an outcome of multidisciplinary therapy, with no mention at all of a planned and purposeful return to work in the suggested approach. This is surprising, given the literature available2,3 and the significant adverse effects of being out of work.4 Further, the algorithm in Box 3 (general practice management of chronic low back pain) appears to have a never-ending loop: I am cautious of the adverse effects that the reductionist model can have,5 and it appears possible in this algorithm to be forever stuck in the investigative loop. An additional pathway from this loop to intensive therapy would allow progression in some cases. The inclusion of a return to work in management of low back pain has been extensively analysed in the Australian setting and shown to significantly reduce disability.6 This advice should be included in any clinical update on management of low back pain.

David S Elder

John Salmon,* Anna Hilyard† * Pain Management Specialist, Bethesda Hospital, 25 Queen Mary Drive, Claremont, WA 6010; † Director, Achieve Pain Control Group, Perth, WA. salmon8ATbigpond.com To the Editor: Bogduk’s article on management of chronic low back pain1 was disappointingly retrogressive as a guide for general practitioners. Compartmentalising back pain management as monotherapy, multidisciplinary therapy or “reductionism”, and favouring the last, reinforces the medical model which has singularly failed to stem the epidemic of low back pain disability affecting the developed world. The biopsychosocial model of chronic spinal pain is now widely accepted and rationally emphasises the multi- or interdisciplinary model of management.2,3 Bogduk’s preference for the reductionist approach may be reasonable in a specialised centre and as a basis for research, but must justify its practical relevance in the face of the following: Available data on the reductionism approach are meagre, conflicting and mainly derived from pain-clinic populations likely to differ from patients presenting to GPs. Diagnostic joint and disc injection procedures and radiofrequency treatment performed to the required standard are available in only a very few centres. Radiofrequency lesioning of the nerve supply to symptomatic joints has been shown to provide pain relief limited to 9–18 months.4 Repeat lesioning may be less effective and is impracticable in the long term. At best, these treatments could be considered palliative. Of course, patients can only benefit from accurate diagnosis and reduction of pain from identified peripheral generators. Unfortunately, for most people with chronic back pain, it is not that simple. Usually there are multiple pathologies and pain generators, multisegmental dysfunction, disrupted motor control and interacting peripheral and central neural sensitisation mechanisms. And that is just the “bio” of the biopsychosocial model. There is then the interplay with the individual’s psychological and social environment. It is often a challenge to communicate the diagnosis effectively in the face of conflicting input from other health providers, the media and patient preconceptions. Just “plonking” “the diagnosis” before a patient and dangling a seductive “techno fix” that does not deliver in the long term is precisely what renders patients with chronic pain increasingly bewildered, dysfunctional and desperate to try one passive treatment after another. The biopsychosocial model provides a basis for management in both general and specialist practice. Appropriate interventions to reduce pain-generator input are embedded in a cognitive behavioural management matrix that imbues patients with accurate, relevant knowledge of their conditions and commonsense self-management techniques to maintain appropriate activity levels, goal setting and psychological positivity. For a time-challenged GP, collaboration with an activation- and exercise-oriented physiotherapist can be effective. The GP’s role is to provide the “white coat authority” so vital in recruiting patient confidence.

John Salmon · Anna Hilyard

Nikolai Bogduk Director, Department of Clinical Research, Royal Newcastle Hospital, Newcastle, NSW 2300. mgillamATmail.newcastle.edu.au In reply: There is a difference between wishful thinking and evidence. Elder advocates a focus on return to work. Elsewhere, I have described how this should be pursued.1 However, the evidence supports success only in the context of acute and subacute pain. I was commissioned to write on low back pain. In that context, evidence is lacking. Even Waddell, whom Elder cites,2 conspicuously avoided the issue of chronic low back pain; his evidence pertains only to acute low back pain. Salmon and Hilyard promote the biopsychosocial model. Indeed, this model is now widely accepted. Even our own studies have shown how successful it can be to recognise and treat patients’ fears and mistaken beliefs.3 However, the evidence of success is limited to acute and subacute low back pain. The predictions of the biopsychosocial model have not been fulfilled in the context of chronic low back pain. Although better than no therapy, behavioural therapy is not more effective than other therapies, and does not “reduce pain generator input”, as Salmon and Hilyard contend. Insurers, who pay for this treatment, do not share their enthusiasm for it.4 Salmon and Hilyard also repeat the commonly held view that patients have multiple pain generators. There is no actual evidence for this assertion, while the available evidence indicates the opposite. When investigated comprehensively, fewer than 10% of patients have more than one simultaneous pain generator.5 Further, Salmon and Hilyard consider that complete relief of pain for 9–18 months amounts to palliative therapy. Yet the opposite is true. Not relieving pain by behavioural therapy is palliative. They also deprecate radiofrequency neurotomy with the accusation that it “may be less effective” when repeated, but fail to cite the literature showing that this is not the case. They are correct in stating that reductionist procedures performed to the required standard are available in only a few centres. However, this does not invalidate these procedures; it reflects only a political and ideological problem in healthcare delivery. They also fail to reveal that in many places where these procedures are available, they are not performed according to best-practice standards. It is not the procedures, but misguided and unscrupulous practitioners, who render patients bewildered and dysfunctional.

Nikolai Bogduk

Gordon Broderick Executive Director, Distilled Spirits Industry Council of Australia, 1st Floor, 117 Ferrars Street, South Melbourne, VIC 3205. gordonbATdsica.com.au To the Editor: In their article on risk-taking behaviour among young Australian women, Carr-Gregg and colleagues make a number of statements about alcohol consumption among young women.1 Unfortunately, these statements are not supported by the facts. The authors assert, citing a national study of 14 762 women aged 18–23 years,2 that “seventy percent of young women engage in ‘binge drinking’ (5 or more drinks on one occasion) at some time, with 19% doing so on a weekly basis”. The level and frequency of alcohol consumption that constitutes “binge drinking” is a matter of conjecture. The National Health and Medical Research Council (NHMRC), in guidelines released in 2001,3 state that “binge drinking” is “not a preferred term due to its lack of consistent and specific meaning”. The NHMRC guidelines on short term risk specify 5–6 alcoholic drinks for a female on any one day as being “risky” for health, and 7 drinks or more being “high risk”. For long term risk, 3–4 drinks on an average day, or 15–28 drinks a week, is considered “risky”, with any more constituting “high risk”. Applying these guidelines to the Women’s Health Australia dataset shows that 5.1% of young women engage in drinking that is “risky” or “high risk” in the long term. Of the remaining 94.9%, 14.4% drink 5 or more drinks weekly or more, and 51.9% drink five or more drinks monthly or less. This is a more revealing (and accurate) picture than the blanket statement that “70% of young women are ‘binge drinkers’ ”. Carr-Gregg and colleagues also claim that “22% of females aged 14–19 years drink between 9 and 30 alcoholic drinks a day”. The source for this statement is a survey conducted for the Salvation Army.4 The survey has several limitations, not least the small sample size. The survey sampled 614 respondents, of whom 70 were aged 14–19 years. The assertion that 22% of females in this age category were “binge drinkers” is based on just seven respondents. This number is well below what is required for any reliable statistical estimation. Encouraging responsible drinking among younger people is a major goal of health professionals and the alcohol industry. A constructive policy debate on this issue requires sound, objective evidence about alcohol consumption among younger people. The article by Carr-Gregg et al does not represent progress towards providing that evidence.

Gordon Broderick

Alex J Chamberlain,* John W Kelly† * Registrar, † Head, Victorian Melanoma Service, The Alfred, Prahran, VIC. alex_chamberlainAThotmail.com To the Editor: Janda and colleagues’ survey of skin screening by general practitioners in regional Queensland1 demonstrated that only 20% of participants had undergone a total-body skin examination in the previous 3 years and that examination rates were slightly lower for patients 50 years of age and older. They conclude that those at greatest risk (older people, especially men) will need to be targeted if screening programs are to reduce melanoma mortality. In the same issue of the Journal, a study by English and colleagues of excisional biopsies of pigmented lesions by Perth GPs showed that the excision rate for patients under the age of 50 years (62%) is nearly twice that for patients over 50 years (38%), and that in the younger cohort this is primarily harvesting benign naevi.2 They conclude by encouraging GPs to increase their suspicion and lower their biopsy threshold in older patients in order to detect more melanomas. These studies serve to remind us that it is predominantly older people (and especially men) who are at greatest risk of thick (and potentially lethal) melanoma.3 The evidence provided by these studies of mismatch between skin cancer risk and resource allocation to surveillance and excisional surgery might help explain why recent advances in early detection seem to have bypassed elderly men. It is of continuing concern that despite the fact that melanomas are currently diagnosed at an earlier stage when compared with 20 or 30 years ago, the background incidence of thick melanoma has remained stable, both in Australia and around the world. In a recent study conducted in Victoria,4 we found that thick melanomas (≥ 3 mm) were predominantly nodular melanomas, primarily affecting people over 50 years of age and especially men. This elusive subtype frequently fails to fulfil the “ABCD” (asymmetry, border, colour, diameter) diagnostic criteria in that they are more often uniform in colour, symmetric in shape and predominantly amelanotic.5 A possible aide-mémoire that we have suggested elsewhere for identifying nodular melanoma is the addition of “EFG” (“elevated, firm, growing for more than 1 month”) criteria.6 The observation that nodular melanomas grow quickly, leading to deep invasion within a few months, poses a significant obstacle to the potential success of community-based screening programs.1 It is likely that many nodular melanomas will escape early detection by such programs, as they will evolve significantly in the interval between screening examinations.4,5 It is important that all clinicians recognise this subtype if we hope to reduce melanoma mortality in Australia.

Alex J Chamberlain · John W Kelly

Stephanie L Jones,* Eugene Athan,† Daniel O’Brien,† Stephen R Graves,‡ Chelsea Nguyen,§ John Stenos¶ * Infectious Diseases Registrar, † Infectious Diseases Physician, Geelong Hospital, Ryrie St, Geelong, VIC 3220; ‡ Medical Microbiologist, § Scientist, ¶ Senior Scientist, Australian Rickettsial Reference Laboratory, Barwon Health, Geelong, VIC. StephljonesATyahoo.com To the Editor: Murine typhus (caused by Rickettsia typhi) has not been previously described in the state of Victoria, although it is well known in Western Australia, Queensland and South Australia. In 2002, a 49-year-old man presented to Geelong Hospital, Victoria, with a 10-day history of fever, myalgia, rigors, headache, rash, sore throat, dry cough and pleuritic chest pain. On examination, he had a fever (temperature, 39.2°C), hypoxia (oxygen saturation, 91% in room air), tachycardia, a central maculopapular rash and conjunctivitis. Blood tests revealed hyponatraemia, thrombocytopenia, white cell count in the reference range, with left-shifted neutrophil change (toxic granulation and increased immature forms) and a C-reactive protein level of 377 mg/L (reference range, < 10 mg/L). The patient lived on a hobby farm close to Geelong. Two weeks before becoming unwell, he had cleaned out the contents of a shearing shed, including two rotten sheepskins in which rats had been nesting. He reported generating a lot of dust and debris in the air. He had not noticed any tick, flea or other insect bites. Serological testing was performed for rickettsia. Baseline serum, taken 10 days after symptom onset, showed antibodies to the typhus group of rickettsiae, R. typhi (murine typhus) and R. prowazekii (epidemic typhus), with a titre of 2000. The titre rose over the following 4 days to 64 000, a fivefold increase, diagnostic of typhus group infection. Antibody titre to the spotted fever group of rickettsiae was significantly lower (peak titre, 8000). The patient was treated with oral doxycycline and recovered completely. Murine typhus was first described in Adelaide in 19221 and is now considered endemic in parts of Western Australia and Queensland.2,3 A possible case reported from Melbourne4 was, in retrospect, probably Brill–Zinsser disease (relapsed epidemic typhus). Murine typhus has an incubation period of 8–16 days and is generally self-limiting, although fatalities have occurred.1 The disease typically presents with fevers, prominent myalgia, a central rash, nausea, conjunctivitis, and often significant pulmonary involvement. Unlike the tick-borne spotted fever group of rickettsiae, R. typhi is transmitted by rodent fleas. Transmission occurs either by aerosolisation and inhalation of infected flea faeces, often during demolition or cleaning of rat-infested environments, or, less commonly, by inoculation of faeces into a fleabite. Murine typhus is usually diagnosed retrospectively by serological testing using microimmunofluorescence. Antibodies are usually detectable 7 to 9 days after disease onset, and IgG may persist for years. Cross-reactivity is seen between R. typhi and R. prowazekii; it is not possible to identify the pathogen by serological testing alone.5 Specific diagnosis is based on known local epidemiology and, as epidemic typhus does not occur in Australia, we believe this was a case of murine typhus, the first described in Victoria.

Stephanie L Jones · Eugene Athan · Daniel O’Brien · Stephen R Graves · Chelsea Nguyen · John Stenos

Mark Little,* Peter L Pereira,† Richard Mulcahy,‡ Teresa Carrette,§ Jamie Seymour¶ * Emergency Physician, Sir Charles Gairdner Hospital, Hospital Avenue, Nedlands, WA 6009. † Director, ‡ Emergency Physician, Cairns Base Hospital, Cairns, QLD. § Research Officer, ¶ Senior Lecturer, Department of Tropical Biology, James Cook University, Cairns, QLD. Mark.littleAThealth.wa.gov.au To the Editor: We are concerned that a recent report announced that the Queensland Ambulance Service now treats patients with hypertension caused by Irukandji syndrome with sublingual glyceryl trinitrate, based on an uncontrolled, unrandomised, unblinded “trial” of three patients.1 We believe the authors have made the justification for such treatment on questionable assumptions. The authors claim that venom from Carukia barnesi is a sodium channel agonist that causes massive noradrenaline release. This claim is based on a letter which reported whole C. barnesi being crudely blended and injected into piglets.2 There is no mention in this letter of any proven sodium channel agonist action.2 We are unconvinced that it was solely the venom of the jellyfish that resulted in a rise in noradrenaline levels in the piglets, especially since there has been a recent report of toxicity from homogenised jellyfish with the nematocysts (ie, venom) removed.3 While we believe that magnesium may be a promising treatment for patients with Irukandji syndrome, there has only been one case reported,4 and we are aware of several anecdotal failures, including the first patient in the letter by Fenner and Lewin.1 This does not constitute the justification for “proven effective treatment”. Fenner and Lewin claim that patients were envenomed by Carukia barnesi and yet provide no evidence. Carukia barnesi has yet to be discovered or reported south of Townsville. Two of us have reported that more than one jellyfish is responsible for Irukandji syndrome, and a second jellyfish has recently been identified.5 Accurate identification of the envenoming animal is essential in toxinology. Patients who present with Irukandji syndrome are in severe pain, and pain is a well known cause of hypertension. In a series of 116 patients with Irukandji syndrome, the mean dose of morphine administered was 31 mg.6 In the three patients reported by Fenner and Lewin, all were in severe pain, but only received 10 mg of morphine. In light of subtherapeutic analgesia, attributing the persistent hypertension solely to an unmanageable disease process is therefore premature and possibly incorrect. As a further confounder, might the reduction of blood pressure in two patients have resulted from the delayed onset of action of intramuscular morphine and promethazine? Without the addition of control patients to this trial, no conclusions can be drawn on the causative factors associated with this treatment. Too often in the past, unsupported treatments have been adopted, only to be abandoned several years later when shown to be ineffectual, or even dangerous to patients. There are too many assumptions, unsupported claims and confounders in Fenner and Lewin’s letter. Toxinology research needs good science and not poor anecdotes.

Mark Little · Peter L Pereira · Richard Mulcahy · Teresa Carrette · Jamie Seymour

Geoffrey K Isbister Toxicologist, and Consultant Clinical Toxicologist, NSW Poisons Information Centre, Newcastle Mater Misericordiae Hospital, Locked Bag 7, Hunter Region Mail Centre, NSW 2310. gsbiteATferntree.com To the Editor: Fenner and Lewin describe three cases of Irukandji syndrome in which glyceryl trinitrate (GTN) was used to transiently lower blood pressure.1 However, they provide no evidence that this relieved the patients’ pain or improved their ultimate outcome (development of pulmonary oedema or myocardial injury); they therefore do not show sufficient evidence for instituting such therapy. It is worrying that, with little evidence, sublingual GTN is now recommended as a prehospital treatment in Irukandji syndrome. Although GTN has been used safely in other conditions, its use without adverse effects in three patients is not evidence that it can be used safely in treating jellyfish envenoming. It is of concern that the treatment of bites and stings remains based on anecdotal evidence, and that clinical toxinology has not moved with the rest of medicine to developing evidence-based approaches. Fenner and Lewin state, in their second paragraph, that intravenous magnesium has proven to be effective in treating symptoms of Irukandji syndrome. This is based on a single case report and no controlled trials. It simply provides a starting point so that properly designed studies can be done to determine if magnesium is effective and safe in Irukandji syndrome. It is not appropriate to suggest that magnesium is a proven therapy without further investigation. Are we not learning from previous problems in clinical toxinology and prehospital care? Both are areas of medicine where treatment protocols are rarely based on substantial evidence.2,3 The pressure immobilisation bandage was introduced for the first-aid treatment of Chironex fleckeri stings with no supporting evidence, and it has taken two well designed animal studies to show that this is dangerous.4,5 A recent well designed study in the United States showed that prehospital intubation by paramedics of patients with head injury resulted in transient desaturation in 57% of patients and significant bradycardia (< 50 bpm) in 19%.6 This occurred despite the paramedics describing the intubation as “easy” in 84% of the patients who desaturated. This shows that it is essential to properly evaluate the safety and efficacy of treatments before recommending them for prehospital use. Although the infrequency of many envenoming syndromes makes controlled studies difficult to do, it is essential that clinical investigators collaborate to institute such studies. This is now happening in Australia with the commencement of randomised controlled trials on redback spider antivenom, magnesium in Irukandji syndrome and hot-water first-aid in jellyfish stings. Hopefully we will move from anecdote to evidence in clinical toxinology with collaborative studies such as these.

Geoffrey K Isbister

Peter J Fenner National Medical Officer, Surf Life Saving Australia, Sydney, NSW, and Associate Professor, School of Medicine, James Cook University, Townsville, QLD; PO Box 3080, North Mackay, QLD 4740. pjfennerATozemail.com.au In reply: We did not provide evidence of any effect on pulmonary oedema, myocardial injury or any effect in stopping pain, as that was not our intention. Also, even intravenous nitrates that have been used for many years in hospital are ineffective in reducing the muscle-cramping pains of Irukandji envenomation, and, although they reduce hypertension and ischaemic myocardial pain, there is no correlation or evidence that they prevent pulmonary oedema or myocardial injury. The assumption that the Queensland Ambulance Service (QAS) introduced this treatment after our trial is incorrect. The QAS Medical Director will confirm that it was introduced independently of our trial and actually before we treated our first case. While pain causes hypertension, in our Patient 1 effective analgesia had been attained without any effect on his hypertension. Also, many patients with Irukandji syndrome have pain without hypertension, while others have severe hypertension even with effective pain control.1 Further evidence that the venom from Carukia barnesi is a sodium channel modulator has been submitted for publication (Ken Winkel, Director, Australian Venom Research Unit, Melbourne, personal communication). Carukia barnesi has now been caught in the Mackay region, 400 kilometres south of Townsville (identified by me, and confirmed by L A Gershwin, PhD student in cubozoan taxonomy, James Cook University, 2003, personal communication). A number of other species can cause the Irukandji syndrome, but the jellyfish causing the sting is rarely caught, so cause and effect are hard to establish at present. Our aim was to present early evidence that sublingual nitrates may be effective in reducing the hypertension that occurs in some cases of Irukandji syndrome; our suggestion was that they be further trialled.2 It seems worthwhile to do so, especially as sublingual nitrates have also been shown to be effective in dysreflexia from spinal injury,3 where hypertension also occurs from the massive release of similar catecholamines. At least two deaths have occurred after Irukandji envenomation, both as a result of cerebrovascular accidents from severe hypertension.4 When treating a patient with severe hypertension from Irukandji envenomation in a prehospital situation, it is reasonable to try to reduce the hypertension. Doing so might reduce the risk of a cerebrovascular accident and possible death, and so further trials of this prehospital treatment must take place. Other prehospital treatments must also be trialled, especially those for pain. Early promising leads need to be published early for all to evaluate. The Irukandji syndrome is a dreadful experience for the victim; all possible prehospital relief measures must be tried and evaluated.

Peter J Fenner

Richard J G Bonham Medical Director, Queensland Ambulance Service, PO Box 1425, Brisbane, QLD 4001. rbonhamATemergency.qld.gov.au Comment: I would like to clarify a few points. Firstly, all Queensland Ambulance Service (QAS) paramedics use glyceryl trinitrate (GTN) frequently to treat cardiac chest pain, and have done so for many years. They are very familiar with its effects and interactions. Secondly, a small number of intensive care paramedics (ICPs), the top tier of officer, with many years of training and experience and tertiary qualifications, have, for a number of years, been authorised to consult with the senior doctor at the receiving hospital for use of GTN to treat acute severe hypertension. In Queensland, GTN has been used for this indication in one or two patients per year, compared with its use in 25 000 patients with cardiac chest pain. Finally, QAS reviewed its protocol for managing marine envenomation after two people died from Irukandji syndrome two years ago. The current protocol for suspected Irukandji stings is now standard care, including oxygen, and vinegar if there is a visible sting, followed by aggressive pain management, including up to 30 mg of morphine intravenously. If the patient has significant hypertension (> 200/120 mmHg) after effective analgesia, ICP officers are able to consult with a senior doctor to administer GTN according to the above protocol. Obviously, the concern of the QAS Medical Advisory Council (comprising representatives of all medical specialty colleges and other medical groups) was to prevent avoidable cerebral haemorrhage or acute heart failure, as occurred in the two deaths from Irukandji syndrome, but, to date, such use of GTN has not been invoked.

Richard J G Bonham

Peter A Leggat,* Robin Griffiths† * Associate Professor, School of Public Health and Tropical Medicine, James Cook University, Townsville, QLD 4811; † Senior Lecturer in Occupational and Aviation Medicine, Wellington School of Medicine, University of Otago, Wellington, New Zealand. Peter. LeggatATjcu.edu.au To the Editor: Grace and Penny present some fascinating “travellers tales” concerning travel insurance and medical evacuation.1 Although I recognise that there may have been difficulties with individual cases, the article probably paints an unnecessarily bleak picture of travel insurance and medical assistance companies. About a fifth of travel insurance claims involve the successful use of the emergency assistance service, which mostly does not involve aeromedical evacuation.2 Evacuation by a dedicated air ambulance is uncommon among travellers.2 Almost all travellers in need of medical and dental treatment source treatment locally, for which they are generally reimbursed. Aeromedical evacuation, where needed, is more likely by scheduled airlines with or without an escort. The article may raise unrealistic expectations among travellers for aeromedical evacuation. Air ambulances do not operate and respond in the same way as ground ambulances. Time is required to assess and prioritise cases, select suitable aircraft, obtain flight plan clearance, check and equip aircraft, brief retrieval and receiving hospital personnel, and develop contingency plans. Aircraft may be required to refuel en route, as air ambulances tend to be based where they can be maintained and staffed adequately and safely — this may be far from the retrieval site. It is useful to raise travellers’ awareness of the possible difficulties in accessing adequate medical facilities in many developing countries, especially where tourism is promoted, and create a sense of travellers’ responsibility for their own health, safety and welfare. The International Society of Travel Medicine has articulated this in a recently released policy statement.3 Readers should be wary of generalising from a small number of case studies from a “popular tropical island holiday destination”.1 It would be useful to collect data on medical retrievals and emergency assistance provided from various sources and consult all interested parties before attempting to establish guidelines for medical evacuation. Grace and Penny do raise the important issue of the need for appropriate travel insurance for all travellers. This message needs to be conveyed by the travel industry and by travel health advisers. The article referred to a study of the travel health advice provided by general practitioners in New Zealand, but did not mention that only about half the GPs in that study routinely discussed travel insurance.4 A similar study in Australia indicated that less than 40% of GPs routinely give advice on travel insurance.5

Peter A Leggat · Robin Griffiths

Fred Gilligan,* Peter Sharley,† Andrew Berry‡ * Emeritus Director of Retrieval and Resuscitation, † Director, Retrieval Services, RAH Mediflight, Royal Adelaide Hospital, North Terrace, Adelaide SA 5000; ‡ Director, NSW Newborn & Paediatric Emergency Transport Services, Sydney NSW. To the Editor: As consultants in intensive care, experienced in transporting critically ill patients within Australia and internationally, we have also received complaints from clients about some travel insurance organisations.1 Poor service appears due partly to economic restrictions and partly to the paucity of experienced staff and specialised aircraft available. Assistance companies implement travel insurance policies for underwriters, quoting 24-hour emergency call centres. Some companies economise by subcontracting (eg, episodic diversion of calls to another organisation). This can result in coordinators lacking an understanding of regional geography, population and medical services and omission of the early, vital input of senior medical advisers. Information relayed between several people can be lost or distorted. Failure to ask key questions can result in inappropriate clinical planning. Furthermore, time zone differences can result in calls being received late at night, further reducing availability of immediate expert opinion. Only a well-organised (and thus expensive) control centre with a critical care focus can manage all these variables. Furthermore, it is difficult for a company to permanently employ current, high-grade healthcare staff in adequate numbers. Current critical care retrieval staff in Australia are confined to a few stand-alone aeromedical organisations or public hospitals which run aeromedical services for state governments and other organisations, using their regular anaesthesia, intensive care or emergency medicine staff. Reliable assistance companies tend to contract with these retrieval organisations or their off-duty staff. In Australia, ambulance aircraft able to travel offshore, with adequate oxygen systems, stretcher attachments, electrical power, and so on, are uncommon and expensive to equip. Portable equipment to care for a critically ill patient represents a capital investment of over $100 000, and some services try to achieve results with inadequate tools or by borrowing. ISAS (the International Society of Aeromedical Services, Australasian Chapter) promotes standards on staffing and equipment,2 including the Australian and New Zealand College of Anaesthetists/Australian College for Emergency Medicine standard for transporting the critically ill.3 Based on critical care practice, the standards are not legally binding, but one suspects they would be quoted in any litigation. It behoves all travellers to scrutinise their travel insurance policies closely — many think of them only in terms of lost or stolen baggage. Following serious injury or illness, the policy may dictate what kind of care is offered.

Fred Gilligan · Peter Sharley · Andrew Berry

Howard Roby Specialist in Anaesthesia and Intensive Care; and Medical Director, Customer Care Medical Assistance, Private Bag 913, North Sydney, NSW 2059. macrobyATozemail.com.au To the Editor: As the medical director of Customer Care Medical Assistance, which manages the travel insurance policies of most travelling Australians, I wish to reassure readers that none of the experiences chronicled by Grace and Penny1 related to our company. I have previously described the activities of Customer Care in the Journal.2 I would welcome any enquiries from colleagues about the way our company functions. I note that there was no declaration by Grace and Penny of their competing interests.

Howard Roby

Kirsty L Buising,* Karin A Thursky,† James F Black,‡ Graham V Brown§ * Clinical Research Fellow, † Infectious Diseases Physician, ‡ Head of Epidemiology, § Head, Victorian Infectious Diseases Service, Royal Melbourne Hospital, Grattan Street, Parkville, VIC 3050. Kirsty.buisingATmh.org.au To the Editor: The 2003 Australian guidelines on antibiotic therapy suggest that the Pneumonia Severity Index (PSI) may be used to triage site of care and antibiotic selection for patients with community-acquired pneumonia.1,2 The PSI was developed as a mortality prediction tool, using data from over 14 000 patients with community-acquired pneumonia.3 The antibiotic guidelines suggest specifically that PSI classes I and II represent patients suitable for outpatient therapy, and that class V can identify patients likely to require intensive care and broad-spectrum antibiotic therapy. We believe this is beyond the previously recommended applications of the PSI and advise caution about its use to identify patients with severe pneumonia. In the cohort used to validate the PSI, only 32% of patients with severe pneumonia (requiring intensive care) were in class V, indicating that the PSI has poor sensitivity for severe pneumonia.3 This finding has been reflected in other studies.4 The strength of the PSI lies in its ability to identify low-risk patients, as the title of the validating article suggests.3 The PSI is so heavily weighted by age and comorbidities that younger patients needing intensive care are unlikely to accumulate enough points to reach class V. This is important, as early identification of patients with severe pneumonia and initiation of broad-spectrum antibiotic therapy and intensive-care support improves outcomes. We are concerned that the PSI may be widely accepted for a purpose for which it was not intended and has not been validated. In underestimating the severity of illness in two-thirds of patients with “severe pneumonia”, the guidelines may provide false reassurance, while clinicians may lose confidence in the PSI if they find it “misses” most patients requiring intensive care. Current evidence does support use of the PSI to guide decisions about inpatient or outpatient therapy. However, the modified British Thoracic Society (BTS) Severity Score is a simpler, better-validated tool to identify patients with “severe pneumonia” who are likely to need intensive care assessment5 (Box). This tool is useful for junior staff to “flag” patients with potentially severe pneumonia and ensure that they are discussed with a senior clinician. As always, the final management and antibiotic selection should be guided by clinical judgement. We believe that the antibiotic guidelines are valuable to encourage appropriate antibiotic use; our aim is to promote discussion of their content relating to this particular condition. Suggested alternative approach to assessing patients with community-acquired pneumonia Step 1: Does the patient need admission to hospital? Assess with the Pneumonia Severity Index (PSI). Class I or II: consider outpatient management (but also need to consider comorbidities, social supports, likelihood of compliance). Class III-V: likely to need inpatient management. Step 2: Does the patient need admission to the intensive care unit? Assess with the modified British Thoracic Society (BTS) Severity Score. Class as severe if two or more of the following features are present on initial assessment or within 24 hours of presentation (and are not attributable to another cause): Confusion (acute onset) Serum urea level > 7 mmol/L Respiratory rate ≥ 30 breaths/minute Systolic blood pressure < 90 mmHg or diastolic blood pressure ≤ 60 mmHg If severe, discuss the case with a senior clinician and consider intensive-care review and aggressive broad-spectrum antibiotics.

Kirsty L Buising · Karin A Thursky · James F Black · Graham V Brown

Wenbiao Hu,* Anthony J McMichael,† Shilu Tong‡ * PhD candidate, Centre for Health Research, Queensland University of Technology; † Director, National Centre for Epidemiology and Population Health, Australian National University, Canberra; ‡ NHMRC Senior Research Fellow, School of Public Health, and Centre for Health Research, Queensland University of Technology, Kelvin Grove, QLD 4059. s.tongATqut.edu.au To the Editor: We examined the possible association between the Southern Oscillation Index (SOI) and the occurrence of hepatitis A in Australia using a Seasonal AutoRegressive Integrated Moving Average (SARIMA) regression model.1 Our results indicate that the SOI is statistically significantly associated with the transmission of hepatitis A. We obtained data on the monthly counts of hepatitis A cases in Australia and the monthly SOI between 1 January 1991 and 31 December 2000 from the Commonwealth Department of Health and Ageing and the Australian Bureau of Meteorology, respectively. Data on population sizes were obtained from the Australian Bureau of Statistics. Cross-correlations were used to compute a series of correlations between SOI and the incidence of hepatitis A over a range of time lags (defined as the time span between the SOI and the incidence of hepatitis A). A SARIMA model was used to estimate the independent contribution of SOI in our study. We adjusted for seasonality by “seasonally differencing” (ie, replacing each observation by the difference between it and the observation from the previous year). In the modelling process, attention was paid to observations well outside the main body of the data (outliers) and the only outlier was excluded in the final SARIMA model. We used the SARIMA [1,0,0] [1,1,0]12 model (ie, first-order autoregressive combined, first-order seasonal autoregressive, after adjustment for first-order seasonal integration) to assess the association between SOI and the incidence of hepatitis A. We found that a decrease in the SOI (ie, warmer and drier conditions) was statistically significantly associated, at a lag of 1 month, with an increase in the monthly incidence of hepatitis A (β = − 0.01; P = 0.001). Two El Niño events (1991–92 and 1997–98) were also clearly associated with an increased incidence of hepatitis A (Box). The results suggest that there was an increase of about 360 cases per year in Australia for an, on average, interquartile range decrease in the SOI. The residuals in the model fluctuated randomly around zero, and there was no apparent autocorrelation between residuals at different lag times (data are available from the corresponding author). These results indicate that the model fitted the data well, with no violation of assumptions. The significant association between SOI and the incidence of hepatitis A remained when the outlier was included in the model (β = − 0.013; P = 0.001). El Niño Southern Oscillation (ENSO) has been found to be related to various health outcomes, including waterborne disease, vectorborne disease, and natural disaster-related deaths (eg, floods, bushfires and cyclones).2-4 Our study adds further evidence of ENSO-related health effects. Infectious diseases are, in general, sensitive to climate variability, as climate can influence the development and transmissibility of pathogens, and can also affect people’s behaviour.3,5 If the relationship between ENSO and hepatitis A is confirmed by other studies, these findings may facilitate the development of early warning systems for controlling and preventing this widespread communicable disease. Incidence of hepatitis A in Australia and the Southern Oscillation Index between January 1991 and December 2000 Negative values of the SOI are associated with El Niño conditions (dry and warm in Australia), and positive values with La Niña conditions (wet and less warm in Australia).

Wenbiao Hu · Anthony J McMichael · Shilu Tong