Endocrinology

To the Editor: The excellent article by Jung and Inder1 in a recent issue of the Journal contains a detailed discussion of different regimens proposed for glucocorticoid supplementation in the perioperative period and makes recommendations for the use of hydrocortisone therapy according to the degree of “surgical stress”. It is worth noting that, in many cases, these recommendations and the detailed advice of endocrinologists regarding individual patients are rendered moot by the changes in routine perioperative antiemetic therapy that have occurred in the past decade. The use of intravenous dexamethasone as an antiemetic has been the subject of much clinical research. The IMPACT study2 showed that it has an antiemetic efficacy similar to that of ondansetron or droperidol when given prophylactically. Dexamethasone is less expensive than either of these drugs and is ineffective as rescue therapy in the setting of postoperative nausea and vomiting (PONV), unlike the alternative drugs. As a result, it is used routinely on induction of anaesthesia in many cases of surgery associated with an increased risk of PONV or where PONV would pose a risk of injury or delayed discharge. A range of doses of dexamethasone for antiemetic prophylaxis has been investigated without finding superior efficacy from higher doses (of up to 1.0 mg/kg).3 The dose typically used in clinical anaesthesia practice is 0.05–0.1 mg/kg. This is equivalent in glucocorticoid activity to more than the highest dose of hydrocortisone described in the guidelines of Jung and Inder1 and should provide a self-tapering effect over 2–3 days, consistent with their recommendations for hydrocortisone dosing.

James A Mitchell

In reply: We thank Woodforth for his interest in our article.1 The cited case report2 involved a patient with panhypopituitarism who had septic arthritis following a total knee replacement, requiring knee washout. As stated by Woodforth, the patient was without glucocorticoid replacement for 5 days, during which time he underwent two surgical procedures. Symptoms of cortisol deficiency were described on Days 1 and 2 postoperatively, with overt sepsis not manifesting until Day 3. The absence of adequate glucocorticoid replacement while the patient was under a “nil oral” instruction and suffering sepsis was undoubtedly a contributory factor in his decline, given that his condition improved significantly after he had received 24 hours of intravenous hydrocortisone treatment and other supportive care. It appears that the cortisone acetate was withheld because of concerns about administering it without food, as other medications were in fact given. We stand by our assertion that patients with proven or suspected cortisol deficiency should receive adequate glucocorticoid replacement before and after surgery, according to the likely stress of the procedure. Often, for minor procedures, an oral route of administration will suffice. Doses of oral glucocorticoids given under these circumstances do not need to be taken with food. If there are sound clinical reasons for the patient not to have any medications orally, then parenteral administration is appropriate.1 Maguire and colleagues correctly point out that the glucocorticoid dosage recommendations in our article are suitable only for adults. They have made a significant contribution to the literature on the investigation and management of paediatric adrenal insufficiency and we would like to thank them for providing the appropriate glucocorticoid doses for paediatric patients under stress. We are aware of the use of dexamethasone as a perioperative antiemetic, as outlined by Mitchell, although it is not clear how widespread this practice is. He is correct in stating that in cases in which dexamethasone is used for this purpose, the glucocorticoid dose thereby provided is likely to be more than adequate for adrenal replacement. However, dexamethasone has no mineralocorticoid activity, and this must be taken into account when treating patients with primary adrenal insufficiency. Doses of hydrocortisone greater than 50–75 mg per 24 hours provide adequate mineralocorticoid replacement. If dexamethasone is used for patients with primary adrenal insufficiency undergoing surgery, it is imperative that the patient continue to take oral fludrocortisone throughout the perioperative period to provide mineralocorticoid replacement. This highlights the need for good communication between the patient’s general practitioner, endocrinologist, surgeon and anaesthetist to ensure the best patient outcome.

Caroline Jung · Warrick J Inder

Recent trials and meta-analyses have raised questions about choice of therapy and use of strict glycaemic targets In August 2006, representatives of the American Diabetes Association (ADA) and European Association for the Study of Diabetes (EASD) published a consensus algorithm for glycaemic management of type 2 diabetes.1 The algorithm provoked debate, but the authors defended their recommendations,2 including the introduction of metformin at diagnosis and the addition of insulin, sulfonylureas or glitazones as second-line therapy if satisfactory glycaemic control is not achieved. The glycaemic target at this and later stages of therapeutic intensification was a glycated haemoglobin (HbA1c) level less than 7.0%. However, based on epidemiological data from studies suggesting no threshold for microvascular or macrovascular benefit, including pooled results from the United Kingdom Prospective Diabetes Study (UKPDS),3 organisations such as the ADA suggested more stringent goals — including a normal HbA1c level (< 6.0%) — if the clinical situation allowed this.4 The first challenge to the applicability of the algorithm came when the results of A Diabetes Outcome Progression Trial (ADOPT) were published at the end of 2006.5 These confirmed the durable glycaemic efficacy of rosiglitazone monotherapy but showed an unexpected increase in distal fractures in women, which was subsequently also reported for pioglitazone,6 the other agent in the class. Then, in May 2007, a meta-analysis of published and unpublished trial data revealed a significant 43% increase in risk of myocardial infarction with rosiglitazone relative to other therapies for type 2 diabetes.7 Although the validity of this analysis is debated,8 pioglitazone appears to have no such deleterious cardiovascular effects and may reduce all-cause mortality.9 The ADA/EASD consensus algorithm was updated in early 2008 to include warnings about the association between rosiglitazone and myocardial infarction, as well as the risk of fracture with both glitazones (Box 1).10 Three recent large-scale randomised controlled trials investigated whether the 7.0% HbA1c threshold recommended in the ADA/EASD algorithm should be lowered. In the glycaemic control arms of the Action to Control Cardiovascular Risk in Diabetes (ACCORD)11 and Action in Diabetes and Vascular disease: preterAx and diamicroN modified release Controlled Evaluation (ADVANCE)12 trials, and in the Veterans Affairs Diabetes Trial (VADT),13 type 2 participants who were at high vascular risk were randomly assigned to either conventional or intensive therapy. The target HbA1c level in the intensive arm was < 6.0% in both ACCORD and VADT, and ≤ 6.5% in ADVANCE. Median HbA1c levels achieved at close of the trials were 6.4%, 6.9%, and 6.5%, respectively, compared with a median ≥ 0.7% greater HbA1c in the respective conventional therapy groups.11-13 Although ADVANCE used sulfonylurea-based intensive treatment,12 other therapies could be added as needed. In ACCORD and VADT, there was no uniform management strategy, but most intensively treated patients were prescribed rosiglitazone and insulin.11,13 Despite differences in the components of the primary endpoints between these trials, intensive therapy was not associated with significant macrovascular benefit. In addition, a 22% increase in all-cause mortality in the ACCORD intensive group, detected 18 months before close of the trial, resulted in all patients being transferred to less intensive glycaemic control for the remainder of the ongoing blood pressure and lipid treatment arms.11 Microvascular endpoint data are not yet available for ACCORD and VADT,11,13 but ADVANCE data show significant reduction in new or worsening nephropathy but no effect on retinopathy with intensive therapy.12 The lack of effect of intensive glycaemic control on cardiovascular disease in these trials appears to be at odds with epidemiological data, especially from the UKPDS.3 It is possible that better contemporary management of non-glycaemic cardiovascular risk factors, with increased use of statins, angiotensin-converting enzyme inhibitors, β-blockers and antiplatelet agents, attenuates the benefits of improved glycaemic control. For example, fewer than 2% of UKPDS patients took lipid-lowering therapy during the trial — which closed in 1997 — compared with more than 50% of ACCORD and ADVANCE patients at the end of these studies.11,12 Although overt hypoglycaemia was not implicated in the increased mortality in ACCORD,11 it was linked to later cardiovascular events in VADT13 and it is also possible that unrecognised low blood glucose concentrations contribute silently to a higher than expected cardiovascular event rate in intensively treated patients. In addition, as suggested by ACCORD subgroup analyses,11 established atherosclerosis may be refractory to glycaemic intervention. A further possibility is that, in contrast to statin studies, in which benefit is evident early, the cardiovascular effects of glycaemic improvement only manifest many years later. This “legacy effect” has been reported in type 1 diabetes14 and may not have been detected in ACCORD, ADVANCE and VADT as average duration of follow-up was ≤ 6 years. What are the lessons from what some might regard as an “annus horribilis” for research into type 2 diabetes management? One is the clear need for carefully designed and adequately powered studies that assess the long-term efficacy and safety of new treatments from both metabolic and cardiovascular standpoints. The Rosiglitazone Evaluated for Cardiac Outcomes and Regulation of Glycaemia in Diabetes (RECORD) study might provide endpoint data that help overcome the limitations of rosiglitazone meta-analyses. However, despite an interim unblinded review of data from the study,15 and post-hoc ACCORD11 and VADT13 analyses that did not suggest increased risk in rosiglitazone-treated patients, it is likely that larger studies will be needed for a valid assessment of the cardiovascular effects of this drug. With new therapies now available in Australia — sitagliptin (a dipeptidyl peptidase-4 inhibitor) and exenatide (a glucagon-like peptide-1 mimetic) — it would be reassuring for prescribers to know that long-term, large-scale safety surveillance of these agents was in progress. It is hoped that such trials will not be viewed as too expensive or logistically difficult, and that the government, pharmaceutical industry, hospitals, academic institutions and consumer organisations will work together to ensure their viability. In the case of glycaemic targets, ACCORD, ADVANCE and VADT were necessary because the relative cost, inconvenience and rates of side effects (including hypoglycaemia) associated with intensive treatment needed to be weighed against cardiovascular and mortality benefit using “gold standard” randomised trial methodology. Their overall findings support the conservative glycaemic target (HbA1c < 7.0%) used in the consensus algorithm10 (Box 2). However, further analyses of their data — especially those relating to microangiopathy, together with post-trial follow-up — could allow a more complex treatment algorithm to be developed, with some well defined patient groups benefiting more from achieving HbA1c levels close to or below 6.0% than others. In the meantime, evidence is emerging of the importance of optimal management of non-glycaemic vascular risk factors in patients with diabetes. For example, the recently published Steno-2 study found short- and long-term morbidity and mortality benefits from multifactorial interventions including multiple drug combinations of renin-angiotensin system blockers, aspirin and lipid-lowering agents in addition to appropriate lifestyle modification.16 Indeed, at present the ADA recommends placing less emphasis on concerted efforts to achieve normal HbA1c levels, and more on optimal management of non-glycaemic vascular risk factors.17 1 Updated algorithm for the metabolic management of type 2 diabetes from the ADA and EASD (2008)10 Reinforce lifestyle intervention at every visit. a Check HbA1c every 3 months until HbA1c is < 7%, and then at least every 6 months. b Associated with increased risk of fluid retention, congestive heart failure and fractures. Rosiglitazone, but probably not pioglitazone, may be associated with an increased risk of myocardial infarction. c Although three oral agents can be used, initiation and intensification of insulin therapy is preferred based on effectiveness and lower expense. ADA = American Diabetes Association. EASD = European Association for the Study of Diabetes. HbA1c = glycated haemoglobin. Source: Nathan DM, Buse JB, Davidson MB, et al. Management of hyperglycaemia in type 2 diabetes mellitus: a consensus algorithm for the initiation and adjustment of therapy: update regarding the thiazolidinediones. Diabetologia 2008; 51: 9 (Figure 1). Reproduced with kind permission of Springer Science+Business Media. 2 Recommendations for glycaemic management of type 2 diabetes based on consensus guidelines and recent trial data A glycated haemoglobin (HbA1c) level < 7.0% remains an appropriate target when using established blood glucose-lowering therapies There are possible deleterious effects of attempting to achieve normoglycaemia (HbA1c level < 6.0%), which may include death due to cardiovascular causes associated with hypoglycaemia Prescribe metformin at diagnosis Add insulin, a sulfonylurea or a glitazone* as second-line therapy (see algorithm to aid choice, Box 1) Use triple therapy if necessary (ie, initiate insulin or add a second oral agent)* Intensify insulin therapy (continue metformin ± glitazone) as the final therapeutic option Use glitazones only when increased risks of osteoporosis, cardiac failure and myocardial infarction (rosiglitazone) have been considered carefully * Pharmaceutical Benefits Scheme restrictions to subsidised glitazone therapy may apply.

Timothy M E Davis DPhil, FRCP, FRACP

Objective: To establish first-trimester-specific reference intervals for thyroid function tests in pregnant Australian women.Design, setting and participants: Serum samples were collected from 2159 pregnant women (9–13 weeks’ gestation) attending a private pathology practice for first-trimester screening during October and November 2006. Levels of serum thyrotropin (TSH), free thyroxine (fT4), free triiodothyronine (fT3), thyroid peroxidase antibodies (TPOAb), and thyroglobulin antibodies (TgAb) were measured by chemiluminescent immunoassay (Abbott ARCHITECT analyser).Main outcome measures: Reference intervals based on 2.5th and 97.5th percentiles for TSH, fT4 and fT3, after exclusion of 338 women with positive TPOAb or TgAb tests; comparison with reference intervals for non-pregnant women (TSH, 0.40–4.0 mU/L; fT4, 9–19 pmol/L; fT3, 3.0–5.5 pmol/L).Results: Derived reference intervals for thyroid function tests during the first trimester of pregnancy were: TSH, 0.02–2.15 mU/L; fT4, 10.4–17.8 pmol/L; and fT3, 3.3–5.7 pmol/L. If the non-pregnant TSH reference range was applied to the study participants, 344 women (16.0%) whose serum TSH concentration was within the first-trimester-specific reference range would be misclassified as having subclinical hyperthyroidism, and 98 women (4.5%) with a TSH concentration above the first-trimester-specific upper reference limit would not be identified.Conclusions: The reference interval for TSH during the first trimester of pregnancy differs substantially from that for non-pregnant women, and applying the general laboratory reference range to pregnant women results in misclassification of thyroid status for 20.5% of women. Australian pathology laboratories should adopt pregnancy-specific reference intervals for thyroid function tests.

Rhonda M Gilbert MB BS(Hons), BPharm · Narelle C Hadlow MB BS, MAACB, FRCPA · John P Walsh MB BS, FRACP, PhD · Stephen J Fletcher DipCB, MSc · Suzanne J Brown BSc(Hons) · Bronwyn G Stuckey MB BS, FRACP · Ee Mun Lim FRACP, FRCPA

Determining risk is the first step in deciding on appropriate management Osteoporosis, most simply and elegantly defined as “too little bone in the bone”, is generally the result of progressive bone loss which, for all practical purposes, starts at menopause in women and at about the age of 50 years in men. Because women have a lower bone organ density than men and then lose bone more rapidly,1 and also because women live longer, osteoporotic fractures, particularly at the hip, affect more women than men in Western countries — there are 20 000 hip fractures per year in Australia, with women outnumbering men by a ratio of two to one. The immediate cost of osteoporosis in Australia has been estimated at nearly $2 billion per year, with a further $5–6 billion in indirect costs.2 Doctors are in a difficult position when it comes to managing osteoporosis and preventing fractures. As with most disorders, they need to know the risk of an event such as fracture before reaching a treatment decision. They know that bone mineral density (BMD), measured by dual energy x-ray absorptiometry (DXA), is a major determinant of fracture risk, and they may have read that the risk goes up by a factor of 1.5–2 for every standard deviation fall in BMD3 (which is actually incorrect, as shown below), but they have no means of converting this information into absolute numbers. This is partly because of long-standing confusion between odds and risk, exemplified by the fact that the relative risks quoted in the literature3 are generally odds ratios or hazard ratios. The difference between odds and risk, well understood by professional statisticians, is not well understood by most clinicians. If 30 women out of 100 develop a fracture over a given period, the fracture risk is 30/100 or 0.30, but the fracture odds are 30/70 or 0.43 — a very different figure. At low levels of risk, say below 0.10, the difference between odds and risk is very small (one in 10 is close to one to nine) and can legitimately be ignored. However, as the risk increases, or the period over which it is calculated is extended, odds rise in a multiplicative fashion with fall in BMD, but risk does not. Odds have no upper limit, whereas risk can never rise above unity or rise by a multiplicative factor. A recent article sought to dispel this confusion by explaining the difference between fracture odds and fracture risk by reference to published data.4 A follow-up article,5 based on a prospective study carried out in Perth,6 contained a graph representing true fracture risk as a function of age and BMD in women without prevalent fracture. For those who would like to calculate the 6-year risk, the formula is: Odds = 0.025 × 1.08age > 55 × 1.49 –ve T-score Risk is then derived from odds as: odds/(1 + odds) We now wish to make this graph more readily available to doctors in Australia by reproducing it (Box). In women with any symptomatic prevalent fracture after the age of 50 years, 5 years should be added to the patient’s age because, in the above study, the effect of prevalent fracture on fracture risk was equivalent to a 5-year increase in age.5 (This, incidentally, shows the fallacy of using fracture as a substitute for densitometry in the diagnosis of osteoporosis, as is increasingly happening in Australia). Moreover, as men and women experience fractures at about the same BMD,7 it is probably safe to use the same graph to calculate approximate fracture risk in men by adding one to the T-score. Needless to say, it is for the individual clinician, in consultation with the patient, to decide the level of risk at which any particular intervention is called for. An arguable policy is to use calcium supplementation (with vitamin D if indicated) to prevent bone loss in patients at low risk, and to reserve more expensive remedies for patients at high risk in whom osteoporosis is already established, especially as the pivotal studies for these remedies have been performed in patients with T-scores of − 2 or lower. We are aware that our fracture risks are somewhat higher than those derived from the Garvan Institute algorithm8 which are in turn higher than those from the World Health Organization algorithm,9 but the former appears to underestimate the effect of age, and the latter has already been criticised as being too low.10 Only time will show which model is nearest the truth. Six-year fracture risk in women aged over 50 years without prevalent fractures

B E Christopher Nordin MD, PhD, FRACP · Richard L Prince MD, ChB, FRACP · Graeme R R Tucker BSc

To the Editor: We believe that Tang and Nordin1 misunderstood the findings of our recent study of calcium supplementation.2 We disagree with their claim that the increase in the number of women with self- or family-reported myocardial infarction, stroke or sudden death became non-significant after adjustment for covariables. They correctly noted that the increased number of women experiencing the composite endpoint of cardiovascular events (after adjudication of events and inclusion of unreported events from hospital records) was not statistically significant. However, the increased event rate for this composite endpoint with calcium was statistically significant (rate ratio, 1.43; 95% CI, 1.01–2.04; P = 0.043). Thus, in our study, the number of women needed to treat with calcium for 5 years to cause one cardiovascular event was 29, and the corresponding number to prevent one fracture was 50.2 Tang and Nordin then meta-analysed data from five studies of calcium and vitamin D supplementation to conclude that calcium supplementation does not increase mortality.1 We disagree. For one of the studies, they classified a subgroup of participants who received annual vitamin D but no calcium supplements as having received “calcium supplementation”.3 Further, for the RECORD (Randomised Evaluation of Calcium Or vitamin D) study, they compared the number of deaths between people receiving and not receiving vitamin D (16.5% v 17.4%) rather than between those receiving and not receiving calcium (17.7% v 16.2%).4 The trend for increased deaths with calcium supplementation in RECORD was greater when analysis was restricted to those treated with calcium monotherapy (18.5%) and placebo (16.3%). As our study was of calcium monotherapy, the results of Tang and Nordin’s meta-analysis are of questionable relevance to our findings. In addition, ours was a 5-year study, and the differences in vascular events between the groups only emerged after 2 years.2 Only one study in Tang and Nordin’s meta-analysis had an average follow-up duration of more than 25 months.4 Further, there is evidence from other studies of trends towards vascular events occurring more frequently in people who take calcium monotherapy.2,5,6 In three out of four studies that reported mortality, there were trends towards increased death rates in people receiving calcium.2,4-6 As we concluded,2 these data are not definitive, but flag cardiac health as an area of concern in relation to calcium use. Finally, we did not suggest that calcium supplementation should not be given to older women. However, in view of the evidence that any fracture risk reduction with calcium is small (< 10%),7,8 and the suggestions that calcium supplementation might increase the risk of hip fractures9-11 and vascular events, it seems reasonable and timely to reassess the role of calcium supplementation.

Mark J Bolland · Andrew B Grey · Ian R Reid

In reply: In Table 5 of Bolland and colleagues’ study, the P value after allowing for covariables was 0.08,1 which is not significant. This was without including smoking, which would undoubtedly have reduced the significance further as there were more smokers in the calcium group. Based on Bolland and colleagues’ suggestion, we reanalysed the data by removing the group receiving vitamin D but no calcium supplements in the NoNOF (Nottingham Neck of Femur) study,2 and using data for those treated with calcium monotherapy (18.5%) compared to placebo (16.3%) in the RECORD (Randomised Evaluation of Calcium Or vitamin D) study.3 The reanalysis still failed to show any evidence of an increase in mortality (relative risk, 1.05; 95% CI, 0.88–1.26; P = 0.56).

Benjamin M P Tang · B E Christopher Nordin

Objective: To determine the causes of death in Australians with type 1 diabetes mellitus who died aged 40 years or younger.Design and setting: Retrospective review of autopsy reports at the Department of Forensic Medicine, Sydney, New South Wales, 1 January 1994 – 31 December 2006.Main outcome measure: Causes of mortality in people with type 1 diabetes aged ≤ 40 years.Results: Of the 26 682 autopsy reports, 1914 were for individuals with diabetes (type 1, 400; type 2, 1514). Cardiovascular disease accounted for 51% of deaths (169/333) in people with type 1 diabetes aged > 40 years, versus 13% among those aged ≤ 40 years (9/67; P = 0.001). Acute complications of diabetes (27%; 18/67), unnatural deaths (28%; 19/67), and sudden unexpected deaths (22%; 15/67) were the predominant causes of death in young individuals with diabetes. Sudden unexpected death was more common in those with type 1 diabetes compared with a sex-matched control population in the same age range (22% v 5%; χ2 P < 0.001). Of the sudden unexpected deaths, 10 people were found dead in an undisturbed bed with no cause of death found at autopsy (“dead-in-bed” syndrome; mean age [SD], 30.2 [9.4] years; males : females = 4 :1).Conclusions: In deceased young people with type 1 diabetes examined by the Coroner, acute diabetic complications, unnatural causes, and sudden unexpected deaths were the predominant causes of death. The relatively high frequency of sudden unexpected deaths, such as dead-in-bed syndrome, requires further investigation.

Emily Tu BSc · Stephen M Twigg PhD, FRACP · Johan Duflou FRCPA · Christopher Semsarian PhD, FRACP

To the Editor: Calcium and vitamin D play a central role in preventing osteoporosis and fractures,1 so a recent study published in the BMJ claiming that calcium supplements increased the risk of heart attacks and strokes in postmenopausal women2 naturally received widespread media attention — so much so that many patients are already stopping calcium treatment. The study, based on a previously published randomised controlled trial of calcium supplementation in 1471 healthy women,3 showed that self- or family-reported heart attack, stroke or sudden death was significantly more common in those taking calcium than in the placebo group (P = 0.008). This conflicted with the findings of a much larger study.4 Further, the difference became non-significant when the analysis was corrected for covariables (P = 0.08), or when the analysis was repeated using data on cardiovascular events obtained from medical records (P = 0.08). Yet, it still gained a place in a leading medical journal. The small excess of cardiovascular events in the women taking calcium could be due to chance and needs to be tested further; one way of doing this is to examine available data for evidence of mortality in patients taking calcium. We have done this. In the 29 randomised trials in a recent meta-analysis of the effect of calcium and vitamin D in fracture risk,1 five trials comprising 12 609 subjects provided crude mortality data.5-9 When these mortality data were pooled using a random effects model, there was no evidence that calcium supplementation increased mortality (Box). We find it hard to believe that calcium can have a significant adverse effect on cardiovascular disease without increasing mortality. Our reservations about this study are further strengthened by the weak theoretical basis of the case against calcium. Metastatic calcification in renal failure, which the authors quote as an analogy,2 is due to the high serum calcium–phosphorus (CaxP) product levels caused by hyperphosphataemia, which may be aggravated by calcium supplementation. In women without this condition, this degree of oversaturation cannot be reached by the 5% rise in plasma calcium10 resulting from the recommended dose of calcium citrate used for supplementation. Moreover, coronary blockage is not due to calcification of atheromatous vessels, which is a dystrophic calcification secondary to tissue damage, but rather to ruptured atheromatous plaques and the thrombi which form upon them. Thus, it is premature to conclude that calcium supplementation should not be given to older women. Effect of calcium supplementation on mortality, data pooled by a random effects model

Benjamin M P Tang · Christopher Nordin

Objective: To determine the incidence of lactic acidosis in community-based patients with type 2 diabetes, with special reference to metformin therapy.Design: Substudy within a longitudinal observational study, the Fremantle Diabetes Study (FDS).Participants and setting: 1279 patients from a postcode-defined population of 120 097 people in Western Australia.Main outcome measures: Confirmed hospitalisation with lactic acidosis identified through the WA Data Linkage System during two periods: (1) from study entry, between 1993 and 1996, and study close in November 2001; and (2) from study entry to 30 June 2006.Results: At entry, 33.3% of patients were metformin-treated, and 23.1% of these had one or more contraindications to metformin (55.1% and 38.0%, respectively, after 5 years’ follow-up). Five confirmed cases of lactic acidosis were identified during 12 466 patient-years of observation; all had at least one other potential cause, such as cardiogenic shock or renal failure. From study entry to close, the incidence was 0/100 000 patient-years in both metformin-treated and non-metformin-treated patients. Between study entry and 30 June 2006, incidence was 57/100 000 patient-years (95% CI, 12–168) in metformin-treated patients and 28/100 000 patient-years (95% CI, 3–100) in the non-metformin-treated group, an incidence rate difference of – 30 (– 105 to 46) (P = 0.4).Conclusion: The incidence of lactic acidosis in patients with type 2 diabetes is low but increases with age and duration of diabetes, as cardiovascular and renal causes become more prevalent. Metformin does not increase the risk of lactic acidosis, even when other recognised precipitants are present.

Niklaus Kamber MD · Wendy A Davis MPH, PhD · David G Bruce MD, FRACP · Timothy M E Davis MRCP, DPhil, FRACP

Objective: Minimal trauma fractures may be the first indication of osteoporosis. Our aim was to determine the proportion of patients who underwent bone density testing for osteoporosis of those with a minimal trauma wrist fracture treated in the emergency department (ED).Design: This observational retrospective cohort study used explicit medical record review and scripted telephone interviews.Setting: EDs of three metropolitan hospitals in Melbourne in 2006.Participants: Patients aged 50 years and over who were treated for wrist fracture due to minimal trauma. Data collected included demographic details, fracture details, causes of injury, any bone density testing and any osteoporosis-related medication change.Main outcome measure: The proportion of patients who underwent bone density testing in the follow-up period.Results: 131 patients were studied; 83% were female, and the median age was 71 years. No patient was referred by an ED or fracture clinic for bone density testing (95% CI, 0–3.5%). Telephone follow-up data were obtained from 91 patients. Of these, 28 reported having bone density testing after their fracture, of whom 14 (50%; 95% CI, 32%–67%) were found to have osteoporosis. Seven were treated with a bisphosphonate and one with a selective oestrogen-receptor modulator.Conclusion: Follow-up of patients suffering minimal trauma wrist fractures treated in the ED is poor. Systems to improve the identification and treatment of osteoporosis in this group are needed if future osteoporotic fractures and their consequences are to be avoided.

Anne-Maree Kelly MD, MClinEd, FACEM · Megan Clooney RN · Debra Kerr BN, MBL · Peter R Ebeling MD, FRACP

Patients with adrenal insufficiency (AI) require additional glucocorticoid doses during surgery or medical illness, but there is no universally accepted regimen for glucocorticoid supplementation therapy. The high doses and long duration of glucocorticoid coverage that have traditionally been used do not reflect the hypothalamic–pituitary–adrenal response to surgical stress and medical illness in normal people. While the optimal dose and duration of supplementation therapy have not been established, our recommendations are based on extrapolation from what constitutes a normal cortisol response to stress, on expert opinion derived from the medical literature, and on clinical experience. The recommended use of lower doses of glucocorticoids during surgical and medical stress should not de-emphasise the importance of additional supplementation during such events. Our recommendations do not replace clinical judgement, but their use will ensure that patients with AI are safely managed during illness or surgery without the risk of an adrenal crisis or excessive steroid dosing.

Caroline Jung MB BS, FRACP · Warrick J Inder MB ChB, MD, FRACP

We report the case of a 59-year-old severely obese man with coexisting type 2 diabetes, hypertension and dyslipidaemia who was treated with a very-low-calorie diet (VLCD) program for 12 months. He continued to lose weight during the treatment, with marked improvement in his comorbidities and no adverse effects. This case demonstrates that prolonged use of a VLCD under close medical supervision is safe and effective in certain obese patients. Clinical recordA 59-year-old severely obese man was admitted to hospital with a 3-day history of gradually progressive paraesthesia and mild weakness affecting all four limbs, with a background of polydipsia, polyuria and fatigue over the previous 4 months. On examination, his weight was 169.5 kg and his height was 1.90 m (body mass index [BMI], 47.0 kg/m2). He was hypertensive (167/80 mmHg) and afebrile. He had mild weakness in a pyramidal pattern in the legs. Upper and lower limb reflexes were brisk, ankle jerks were absent, plantar responses were upgoing bilaterally, and sensation was reduced in a stocking distribution. The cranial nerves and speech were unaffected and visual fields were normal. Cardiovascular, respiratory and abdominal examinations were unremarkable. The presence of acanthosis nigricans and a fungal toenail infection were noted. Initial laboratory investigations revealed a random blood glucose level of 22.2 mmol/L (reference range, 3.3–5.5 mmol/L) and a glycated haemoglobin (HbA1c) level of 12.2% (reference range, 4.3%–6.0%). Urinary ketones were not detected. Levels of electrolytes, inflammatory markers and vitamin B12 were normal. Liver function tests showed a mild generalised elevation, and lipid levels were raised (Box 1). An electrocardiogram (ECG), non-contrast computed tomography brain scan, lumbar puncture and carotid Doppler ultrasound were normal. Nerve conduction studies showed an axonal sensorimotor neuropathy, with the lower limbs more severely affected than the upper limbs. Neurological findings remained stable during the hospital stay, and a provisional diagnosis of a cervical myelopathy with peripheral neuropathy was made. Magnetic resonance imaging of the spine could not be performed because of the patient’s large size. The patient was treated with aspirin 100 mg daily, simvastatin 20 mg daily and metformin 500 mg twice daily. On the day of discharge (Day 6), he commenced a very-low-calorie diet (VLCD) (Optifast VLCD, Novartis Consumer Health Australasia, Melbourne, Vic) three times daily. Blood glucose readings at discharge were 6–8 mmol/L. After discharge, he commenced a daily exercise program. Over the following 12 months, the patient attended the weight control clinic at our hospital. He continued on an intensive regimen of Optifast VLCD three times a day for 5 months, and then began a transition phase (two Optifast VLCD meals plus one low-fat meal per day). Over the 12 months, the patient’s weight fell steadily (Box 2, Box 3). At 1-year follow-up, he had lost 43% of his initial weight and was down to 96.5 kg (BMI, 26.7 kg/m2). He was normotensive (125/80 mmHg), and his fasting blood glucose and HbA1c levels were in the non-diabetic range. Full blood examination, electrocardiographic readings, and levels of electrolytes, calcium, magnesium, folate, vitamin B12, vitamin D and uric acid were normal. Liver function tests had normalised, and the lipid profile had improved significantly (Box 1). Bone mineral density (measured by dual energy x-ray absorptiometry) was normal at 6 and 12 months after commencement of the VLCD. Neurological symptoms improved, with persistence of limb weakness and sensory loss. Planned management was to gradually reduce the Optifast diet and enter a supervised weight maintenance phase, which would include an education program about lifestyle changes, frequent clinic visits and close observation of weight. DiscussionVLCDs involve replacement of meals with foods or formulas providing 1675–3350 kJ/day. They are commonly used in medically supervised weight reduction programs for patients with BMI > 30 kg/m2 (or > 27 kg/m2 with obesity-related comorbidities), or for whom rapid weight loss is necessary. Several formulations are available without prescription (Box 4). VLCDs should provide at least 0.8 g protein per kilogram of ideal bodyweight per day (to preserve lean body mass1), along with the recommended daily allowances of minerals, vitamins, trace elements and essential fatty acids. Around 10 g/day fat is recommended to stimulate gallbladder contraction.2 Although optimal caloric and carbohydrate intakes are unknown, 50 g carbohydrate daily has been suggested as a suitable quantity to maintain normoglycaemia and to prevent loss of electrolytes and protein.2 There is no advantage in reducing energy intake below 3350 kJ/day.3 Treatment duration varies but is usually 8–16 weeks.1 To our knowledge, our report is the first to document the use of a VLCD for 12 months. There is concern about the safety of VLCDs, most of which stems from the occurrence of over 60 deaths associated with a popular liquid protein diet, available in the 1970s, that consisted entirely of solutions of collagen or gelatine hydrolysates.4 Many of the deaths occurred in people with pre-existing comorbidities.4 However, 17 deaths (16 in women) occurred in relatively young (median age, 35 years) and otherwise reasonably healthy people. These 17 people, who were severely obese (mean BMI, 40.6 kg/m2), remained on the diet for an average of 5 months and lost a large amount of weight (mean, 35%). Postmortem findings were consistent with the cardiac effects of protein-calorie malnutrition (which include myocardial atrophy, QT prolongation and ventricular arrhythmias).5 In contrast, VLCDs in current use contain high-quality protein and appear to be safe.1 In 1998, Rössner described the case of a patient who had lived solely on a VLCD preparation for 46 weeks without medical supervision, in whom no adverse events occurred other than cholelithiasis and one episode of palpitations, after which ECG monitoring over a 24-hour period showed no abnormality.6 Common minor adverse effects associated with VLCDs, which include cold intolerance, dry skin, hair loss, constipation, headaches, fatigue and dizziness, are generally self-limiting and transient. Other potential effects are gallstones, increased serum uric acid levels, precipitation of gout and reduced bone mineral density. When beginning a VLCD, liver function tests, lipid profile measurements, a full blood count and iron studies should be carried out, and levels of electrolytes, creatinine and uric acid should be measured. Testing of electrolyte and creatinine levels should be repeated about 6 weeks after commencement, or earlier if more careful monitoring is required (eg, in patients who have renal impairment or are using diuretics). If the VLCD is to continue for more than 12 weeks, it is advisable to repeat the baseline tests at least every 2 months.7 Obese people typically achieve a mean weight loss of 1.5–2.5 kg per week using a VLCD. A 2001 meta-analysis concluded that, after using a VLCD, subjects maintained a significantly greater weight loss at 4.5 years (mean loss, 6.6% of initial weight) than after a hypocaloric balanced diet (mean loss, 2.1% of initial weight).8 Taking meal replacements once or twice daily in conjunction with a reduced-calorie diet is also effective for long-term weight maintenance.9 Advantages of VLCDs include the motivating effect of rapid weight loss, the convenience of meal replacement (which improves adherence compared with an ad-libitum low-fat diet)10 and a mild ketosis, which may suppress hunger.7 VLCDs also help with several obesity-related comorbidities by reducing levels of total cholesterol, low-density lipoprotein, triglycerides and blood glucose and by reducing blood pressure, insulin resistance1 and hepatic steatosis.11 All of these benefits were obtained by our patient. To date, bariatric surgery has been the most effective long-term treatment option for obesity. The Swedish Obese Subjects Study, a large prospective controlled study of severely obese patients, found significantly greater weight loss in the surgically treated group than the control group (23.4% bodyweight lost [surgical group] v 0.1% gained [control group] at 2 years, and 16.1% lost [surgical group] v 1.6% gained [control group] at 10 years).12 Similar weight reductions after surgery were reported in a study of mildly to moderately obese subjects13 and in a meta-analysis of bariatric surgery trials.14 Pharmacotherapy for weight loss has only modest benefits, with loss of around 3%–5% bodyweight over 1–4 years reported in a recent meta-analysis.15 The 43% weight loss achieved by our patient at 12 months is comparable with results reported for surgery. Weight maintenance will be his next major challenge. Increasing evidence indicates that physiological changes occur after diet-induced weight loss, including decreased levels of leptin, glucose, insulin, free fatty acids, cholecystokinin and triiodothyronine (T3), and increased levels of reverse T3 and ghrelin. Many of these changes would be expected to reduce satiety and increase hunger, possibly contributing to the weight regain frequently seen after weight loss. Our patient may continue to replace one meal a day with Optifast in the longer term to assist with weight maintenance. As there is growing evidence of physiological adaptations after weight loss that encourage weight regain, pharmacotherapy will be introduced if lifestyle changes prove insufficient to maintain his current weight. Bariatric surgery could also be considered if his weight loss cannot be maintained. Our report shows that, under close medical supervision in certain obese patients, a VLCD may be used safely and effectively for a period of at least 12 months. Prevention of weight regain remains difficult. 1 Blood biochemistry measurements Biochemical levels At baseline At 6 months At 12 months Reference range Total cholesterol (mmol/L) 7.9 3.9 4.8 < 5.5 Triglycerides (mmol/L) 2.4 1.6 0.5 < 2.0 High-density lipoprotein (mmol/L) 1.0 0.8 1.4 > 1.0 Low-density lipoprotein (mmol/L) 5.8 2.4 3.2 < 2.5 Fasting glucose (mmol/L) 16.5 4.5 4.5 3.3–5.5 Glycated haemoglobin (%) 12.2 4.8 4.7 4.3–6.0 Potassium (mmol/L) 4.1 3.9 4.0 3.5–5.0 Creatinine (μmol/L) 107 85 89 30–110 Albumin (g/L) 34 39 40 36–48 Bilirubin (μmol/L) 31 22 20 < 18 Alkaline phosphatase (U/L) 100 70 68 32–91 Alanine aminotransferase (U/L) 88 23 22 < 45 γ-Glutamyltransferase (U/L) 113 21 27 < 55 2 Patient’s weight loss during maintenance of a very-low-calorie diet over a 12-month period 3 Patient at baseline and after 12 months on a very-low-calorie diet 4 Comparison of available very-low-calorie diet (VLCD) formulations* Product† Energy (kJ) Carbohydrate (g) Protein (g) Fat (g) Optifast VLCD‡ 1908 45.0 51.9 6.9 OptiSlim 2000§ 1958 48.1 48.4 4.7 KicStart¶ 2636 58.2 74.7 8.4 * Values given are per 3 × 40 g sachets of chocolate shake prepared as directed. Figures in the table are daily amounts, but do not take into account the additional vegetables recommended along with the VLCD. † All three products are available without prescription. ‡ Novartis Consumer Health Australasia, Melbourne, Vic. § OptiPharm, Melbourne, Vic. ¶ Pharmacy Health Solutions, Sydney, NSW.

Priya Sumithran MB BS, FRACP · Joseph Proietto MB BS, FRACP, PhD

To the Editor: We report the serendipitous finding of non-Hodgkin’s lymphoma in a patient with adrenal Cushing’s syndrome. A 62-year-old previously well man (body mass index, 22 kg/m2) was referred to our institution with newly diagnosed type 2 diabetes, hypertension and dyslipidaemia. Clinical findings included oral thrush, bilateral severe pitting lower limb oedema, lower limb proximal myopathy, kyphosis, and increased abdominal girth (waist circumference, 92 cm), raising suspicion of Cushing’s syndrome (Box 1). Biochemical assessment revealed normal electrolytes, an unsuppressed early morning cortisol (following 1 mg dexamethasone), urinary free cortisol 6475 nmol/day (reference range, 0 – 250 nmol/day), and undetectable adrenocorticotropic hormone levels. Twenty-four-hour urinary catecholamine was normal. His testosterone level was 3.2 nmol/L, and dehydroepiandrosterone sulfate level was normal. Abdominal computed tomography showed a right adrenal mass that measured 3.1 × 2.8 × 3.4 cm (density, 36 Hounsfield units). Thoracic spine x-rays revealed wedge compression fractures at T-10 and T-11. Bone densitometry showed T-scores of − 3.3 at L2 – 4 and − 2.3 at the right femoral neck. Total body fat (18.5 kg; 33%) was higher than the recommended range for age and sex (13%–25%). The patient had a laparoscopic right adrenalectomy. Surgical excision was complete. Post-operatively, blood glucose and blood pressure returned to normal. Histopathology revealed an adrenal cortical tumour with atypical features, including a preponderance of eosinophilic cells, small numbers of clear cells, prominent nuclear pleomorphism, large nucleoli and occasional mitoses (Box 2A). However, the proliferation fraction (Ki67) was low and there was no necrosis. There was no large vessel invasion, although a single area of small vessel invasion was present (Box 2B). Unexpectedly, the adipose tissue adjacent to the adrenal gland was infiltrated by a diffuse large B-cell non-Hodgkin’s lymphoma (Box 2B). This was confirmed by positive CD20 immunohistochemistry. Bone marrow biopsy was normal. [18F]Fluorodeoxyglucose positron emission tomography (FDG-PET) scan showed increased uptake in the right adrenal bed only. The patient was treated with six courses of CHOP chemotherapy (cyclophosphamide, doxorubicin, vincristine and prednisolone) in combination with rituximab. A repeat FDG-PET scan 1 month after chemotherapy was clear. We speculate that lymphoma progression was suppressed by the coexistent steroid-producing adrenal tumour. The decision to treat the non-Hodgkin’s lymphoma was, in part, based on reports of progression of haematological disease following treatment of Cushing’s syndrome.1,2 Although histopathological examination of the tumour revealed some features suggestive of adrenocortical carcinoma, the distinction between adenoma and carcinoma can be difficult. In patients with recurrent or metastatic adrenocortical carcinoma, partial response has been reported using a combination of cylophosphamide, vincristine, cisplatin and teniposide.3 Two of these agents were used to treat our patient’s lymphoma. 1 Patient appearance at presentation, with obvious kyphosis and abdominal swelling 2 Immunohistochemistry A: Right adrenal tumour composed of enlarged pleomorphic cells with prominent nucleoli. B: Adrenal cortex (on right) with tumour showing focal vascular invasion (centre), plus adjacent non-Hodgkin’s lymphoma (on left).

Lai Y Wong · John Moore · Debbie Hill · Phil Brenner · Warick Delprado · Jennifer Turner · Joanne Taylor · Lesley Campbell · Jerry R Greenfield

To the Editor: Senna is widely used in laxatives, but the results of its misuse are not inconsequential. We describe a 36-year-old woman admitted with hypercalcaemia and renal failure. She had a 6-year history of anorexia nervosa and ingestion of 50–100 senna tablets daily for weight loss. Examination revealed clubbing of the fingers and toes, a body mass index (BMI) of 17.7 kg/m2 and postural hypotension. Laboratory findings on admission are shown in Box 1. Results of autoimmune studies and protein electrophoresis, and the serum angiotensin-converting enzyme level were normal. Parathyroid hormone-related peptide was absent. She had a bland urinary sediment, trace proteinuria (150 mg/24 h; reference range, < 150 mg/24 h) and a urine pH of 5.0 (physiological range, 4.5–8.0). Computed tomography scans did not detect malignancy or infection, but showed bilateral medullary renal calcifications. Renal biopsy confirmed extensive nephrocalcinosis and the absence of primary glomerular disease. A skeletal survey showed prominent periosteal reaction and new bone formation at the ends of long bones (Box 2, A). A bone scan revealed increased tracer uptake in a pattern consistent with hypertrophic osteoarthropathy (HOA; Box 2, B). Interestingly, bone mineral density (BMD) scans showed increased lumbar and femoral T scores (1.2 and 1.3, respectively). Four years later, the renal failure, clubbing and HOA persisted despite a reduction in senna intake. Low urine volume is a prerequisite for urolithiasis, but hypercalcaemia is the key requirement for nephrocalcinosis.3 The suppressed parathyroid hormone level and elevated serum calcium level excludes primary hyperparathyroidism and made familial hypocalciuric hypercalcaemia unlikely. The serum and urine biochemistry was inconsistent with thiazide diuretic use or renal tubular acidosis. Hence, exogenous calcium is the likely cause of hypercalcaemia. Each of the senna (calcium sennosides) tablets the patient ingested contained 12.5 mg of calcium. Chronic ingestion, in addition to dehydration (with low calcium excretion) and a low BMI may contribute to a vicious circle of calcium phosphate retention, renal failure and nephrocalcinosis. Indeed, hydration increased her calcium excretion to 6.23 mmol/day and normalised her serum calcium and phosphate levels after a week. The association between finger clubbing and senna misuse, and the reversibility of finger clubbing, were reported in 1975.4 Several reports have followed, but only one noted concurrent HOA on plain x-rays.5 We believe that our report is the first to show the extent and distribution of HOA related to this disorder. It remains unknown whether HOA is reversible with abstinence from senna. Patients with anorexia are also more likely to have a low BMD, and the increased BMD scores seen in our patient might be the result of metastatic calcification or periosteal new bone formation. Neither could be conclusively proven. The development of nephrocalcinosis in anorexic patients is more common than is generally appreciated. Clubbing and HOA are useful clues to senna misuse, and BMD measurements should be interpreted with caution in this setting. 1 Biochemical parameters on admission before treatment Parameter Patient value Reference range Serum concentrations of: Sodium (mmol/L) 139 135–145 Potassium (mmol/L) 3.2 3.5–5.0 Magnesium (mmol/L) 0.74 0.80–1.50 Chloride (mmol/L) 102 95–107 Bicarbonate (mmol/L) 24 21–30 Urea (mmol/L) 8.6 2.5–7.8 Creatinine (μmol/L) 166 40–120 Albumin (g/L) 28 35–45 Corrected calcium (mmol/L) 2.93 2.20–2.60 Ionised calcium — pH adjusted (mmol/L) 1.51 1.14–1.29 Phosphate (mmol/L) 2.58 0.80–1.20 25-hyroxyvitamin D (nmol/L) 23 55–108 Parathyroid hormone (pmol/L) 1.0 1.1–7.7 Rates of: Creatinine clearance (mL/min) 20 90–150 Urine sodium excretion (mmol/day) 28 40–100 Urine potassium excretion (mmol/day) 29 50–140 Urine calcium excretion (mmol/day) 1.56 2.0–7.5 2 Ankle x-ray and full bone scan

Andy K H Lim · David H Hooke · Peter G Kerr

Why all patients with a pituitary mass need to be seen by an endocrinologist For the average general practitioner, pituitary disorders are relatively uncommon. However, post-mortem and magnetic resonance imaging (MRI) studies show that about 10% of the population may harbour a pituitary mass, although most are small, non-functioning microadenomas.1 Pituitary adenomas are the commonest intracranial neoplasm, making up about 10%–15% of such lesions.2 In most large series, prolactinomas are the most prevalent subtype, at between 40% and 50%; non-functioning adenomas account for about 30%, with other functioning adenomas (secreting growth hormone, causing acromegaly; secreting adrenocorticotropic hormone, causing Cushing’s disease; or secreting glycoprotein hormones, such as intact follicle-stimulating hormone, luteinising hormone and thyroid-stimulating hormone) making up the remainder.3 We would like to put forward the view that the endocrinologist should be the key practitioner in a multidisciplinary approach to pituitary masses. Within the Australian health system, most patients in whom a pituitary mass is discovered are referred directly to a neurosurgeon. Many non-endocrinologists making such referrals do not realise the critical importance of the neurosurgeon’s specific pituitary surgical experience in patient outcome. Initial assessment of a patient with pituitary disease includes differential diagnosis, making decisions regarding necessity for surgical intervention, and medical management, including correction of any underlying hypopituitarism or hormone excess. On occasion, an endocrinologist may detect subtle variations in a patient’s presentation that can significantly alter management. Therefore, we encourage practitioners to refer patients initially to an established multidisciplinary pituitary team using a shared management approach — a team in which an endocrinologist works in conjunction with a pituitary neurosurgeon. Other relevant health professionals may include a neuroradiologist, ophthalmologist and radiation oncologist. Such teamwork can ensure optimal benefits for patients at all stages of management. With respect to functioning adenomas, most prolactinomas do not require surgery, even when a significant visual field defect is present. Mild increases in serum prolactin levels can be seen with some non-functioning pituitary lesions because of the interruption of dopamine delivery down the pituitary stalk. A level elevated more than 10-fold has traditionally been diagnostic of prolactinoma, for which the initial management should, in most circumstances, be with dopamine agonists.4 More recent evidence indicates that, for about 99% of histologically confirmed non-functioning adenomas, the serum prolactin level can be up to 2000 mIU/L; the authors of that article recommended a trial of dopamine agonist in patients with a macroadenoma and serum prolactin levels of more than 2000 mIU/L.5 There are tumours that secrete enough adrenocorticotropic hormone or growth hormone to produce only subtle clinical features of Cushing’s syndrome or acromegaly, respectively. Because the perioperative management of a patient with Cushing’s disease and acromegaly differs from the management of a patient with a non-functioning mass, it is important to detect these cases of subclinical pituitary hypersecretion before surgery, so the correct functional diagnosis is made from the outset.6 This also gives the endocrinologist the opportunity to use medical therapies such as somatostatin receptor analogues in the pre- or postoperative setting where appropriate. Patients with hypopituitarism benefit from hormone replacement.7 Although giant, vision-threatening tumours do occur, most cases do not generally require emergency surgical intervention, so there is time, in consultation with the neurosurgeon, for adequate assessment and stabilisation of the patient’s hormone state before surgery. Patients with clear hypopituitarism should begin taking maintenance doses of glucocorticoid and thyroxine before surgery. Sex steroid replacement can be delayed until after postoperative assessment. Successful resection of a macroadenoma may result in reversal of the hypopituitarism.8 Appropriately timed postoperative pituitary function testing and correct interpretation of the results is required for optimal outcome. Not all pituitary or suprasellar masses are pituitary adenomas. Sometimes characteristics on MRI can distinguish between the other various causes.9 The presence of diabetes insipidus is a strong clue that the mass is not a pituitary adenoma.7 Inflammatory (eg, Langerhans’ cell histiocytosis and lymphocytic hypophysitis) and other neoplastic lesions (eg, craniopharyngioma and germinoma) are commonly associated with diabetes insipidus, whereas it is very rare for an uncomplicated pituitary adenoma to present in this way. Preoperative recognition of non-adenomatous lesions may alter management. For example, lymphocytic hypophysitis often responds to high doses of glucocorticoids, even when a large mass with visual field compromise exists;10 Langerhans’ cell histiocytosis responds well to low-dose radiotherapy;11 and germinomas, which may be associated with increased serum or cerebrospinal fluid β-human chorionic gonadotropin levels, are best treated with a combination of chemotherapy and radiotherapy.12 With a collaborative, multidisciplinary approach to patient care, the endocrinologist can evaluate the likely differential diagnosis, in conjunction with the pituitary neurosurgeon and neuroradiologist, relevant investigations can then be performed, and a collective decision can made on the most appropriate management for each particular patient. Trans-sphenoidal pituitary surgery is a highly specialised procedure, and it is clear from published data that the experience and skill of the neurosurgeon is of paramount importance in determining outcome, particularly for secretory lesions. A review from the United States showed that higher surgical volume is associated with significantly reduced mortality, fewer complications and a reduced length of hospital stay.13 The best results were achieved by units where more than 25 patients were operated on per year.13 It has been estimated in the United Kingdom that only one pituitary neurosurgeon is needed for each five million of population.14 In a country such as Australia, it would seem pragmatic to have a small number of specialised neurosurgeons who do virtually all the pituitary surgery. However, such a scenario seems a long way off and would be fraught with political, personal and inter-hospital rivalries. Nevertheless, informal management networks can be forged by altering referral patterns. In Birmingham, UK, the cure rate for acromegaly almost doubled when the endocrinologists sent all their patients to one neurosurgeon, compared with when eight neurosurgeons in the city were performing pituitary surgery.15 This is another key reason why we believe referrals for patients with pituitary masses would be best directed to an established multidisciplinary team involving, as two key members, an endocrinologist with primary responsibility for pre- and postoperative assessment, and an experienced pituitary neurosurgeon. Pituitary masses are the commonest intracranial neoplasm. The initial point of referral should be an endocrinologist within an established multidisciplinary team. We are aware that a number of teaching hospitals in the major Australian cities have developed considerable expertise in the multidisciplinary management of pituitary lesions. In our opinion, this approach provides a high standard of care and contributes to the ongoing training experience of endocrinology and neurosurgery registrars. We believe patients who wish to be cared for in the private sector would also benefit from being managed jointly by an endocrinologist and a specialised pituitary neurosurgeon, if not by a full multidisciplinary team.

Warrick J Inder MB ChB, MD, FRACP · Frank P Alford MB BS, MD, FRACP

To the Editor: A recent editorial in the Journal suggested that blood could be sent overseas for genetic testing for maturity onset diabetes of the young (MODY).1 We are pleased to be able to point out that genetic testing, including clinical and laboratory support with full gene sequencing for both MODY1 and MODY3 and for neonatal diabetes (mutations in SUR1 and Kir6.2), is available in Australia. Testing for MODY2 and for a number of other disorders of the pituitary–adrenal and pituitary–gonadal axis in children is also available. We are happy to receive specimens and referrals from clinicians who would prefer to use an Australian clinical laboratory accredited by the National Association of Testing Authorities. More information is available via: http://www.mater.org.au/Home/Services/Pathology.aspx.

Mark F Harris · Ivan N McGown · David M Cowley

Potentially serious metabolic sequelae make diagnosis and intervention imperatives Polycystic ovary syndrome (PCOS) is the most common endocrine abnormality of women of reproductive age. The diagnosis is based on the presence of two of three criteria — ovulatory disturbance, hyperandrogenism, and polycystic ovaries on ultrasound. However, in most women, insulin resistance is central to the pathogenesis of the syndrome, with hyperinsulinaemia driving both androgen production and androgen bioavailability as the key diagnostic feature.1-3 In PCOS, insulin resistance not only contributes to symptoms, but also has serious sequelae including infertility, impaired glucose tolerance, a fourfold to sevenfold increase in diabetes and a potentially increased risk of cardiovascular disease.3 PCOS affects an estimated 400 000 Australian women; this is 5%–10% of the reproductive age group. Obesity exacerbates insulin resistance and glucose intolerance in PCOS. As obesity in the community increases, the prevalence of the PCOS phenotype and its associated glucose intolerance — including diabetes — are expected to rise significantly. In 2006, the estimated economic burden of PCOS in the United States was $6 billion, equating to $400 million in health care costs in Australia (with menstrual dysfunction consuming 31%, infertility 12% and PCOS-associated diabetes 40% of total costs), representing a major health and economic burden.4 An economic evaluation of PCOS recently advocated screening, diagnosis and intervention, justifiable by ameliorating or preventing serious sequelae.4 However, greater understanding of appropriate screening, long-term risks and effective interventions is urgently needed. In this issue of the Journal, Dabadghao and colleagues report a retrospective study of a large cohort of Australian women with PCOS presenting to an infertility service, and tackle the important issues of abnormal glucose tolerance and metabolic syndrome in PCOS (→ Glucose tolerance abnormalities in Australian women with polycystic ovary syndrome).5 In the study by Dabadghao et al, most women with PCOS were obese, with a mean body mass index (BMI) in their cohort of 35 kg/m2. Impaired glucose tolerance was noted in 15.6% and diabetes in 4%.5 The key predictors of abnormalities in glucose tolerance were age, BMI, metabolic syndrome and a family history of diabetes.5 Previous reports on the prevalence of metabolic complications of PCOS have shown inconsistent results, related to the diversity of populations studied (age, BMI, ethnicity) and the different diagnostic criteria applied for PCOS and for impaired glucose tolerance and diabetes. The study by Dabadghao et al used the current Rotterdam criteria for diagnosing PCOS and the World Health Organization criteria for impaired glucose tolerance and diabetes. While the ethnicity of the population is not described in the Dabadghao et al study, and the population is selected (all women had attended a fertility service), their findings in this large cohort highlight the high prevalence of metabolic complications in women with PCOS in Australia. PCOS is a heterogeneous condition, and there are undoubtedly varied genetic and environmental influences on its development and expression, with challenging clinical and research questions still to be answered. However, given the aetiological and exacerbating roles of obesity and insulin resistance in most women with PCOS, it is imperative that clinicians are aware of the metabolic implications of this syndrome. Screening for metabolic complications, as recommended by Dabadghao et al, needs to include a 75 g 2-hour oral glucose tolerance test and lipid profile determination at diagnosis, and regularly over time.5,6 Frequency of screening should be based on the key predictors for development of diabetes (as noted by Dabadghao et al) — age, BMI and family history of diabetes.5 It should be noted that measuring insulin levels does not have a clinical role in screening or in guiding management and remains a research tool. Prevention and treatment strategies should also be more aggressively pursued in these higher risk subgroups. Lifestyle modifications are first-line interventions in the treatment of insulin-resistance states (obesity, PCOS, prediabetes and diabetes). In PCOS, improvements in insulin resistance, ovulation, androgen levels and fertility have been shown with as little as a 4%–5% drop in bodyweight achieved with caloric restriction (independent of dietary composition), with or without exercise programs.7,8 In populations not affected by PCOS, lifestyle changes as well as insulin sensitisers (including metformin) significantly delay the onset of diabetes in those with impaired glucose tolerance;9 similar delay is likely with such changes in patients with PCOS. Lifestyle therapy should be realistic (initial goal of about 5% weight loss), feasible, achieved through sustainable lifestyle change rather than short-term caloric restriction, and supported by a multidisciplinary approach.7,8 In combination with lifestyle change, insulin sensitisers are likely to have a role in those at highest risk, especially where impaired fasting glucose or impaired glucose tolerance is already established.10-12 It is now recognised that PCOS is not simply a reproductive condition characterised by the appearance of the ovary on ultrasound, but represents a complex interplay between insulin metabolism and androgen production. The high prevalence of abnormalities of glucose metabolism and of the metabolic syndrome in women with PCOS mandates a proactive approach to screening and prevention. A recent survey of Australian clinicians treating women with PCOS found that screening for metabolic complications of PCOS has not been as widely practised as is advocated and supported by the literature in general,13 including the study by Dabadghao et al. A change to more proactive screening and intervention is likely to reduce the burden of disease associated with PCOS.

Helena J Teede FRACP, PhD · Bronwyn G A Stuckey BA, FRACP

Objectives: To determine the prevalence of glucose tolerance abnormalities and to identify associated risk factors in women with polycystic ovary syndrome (PCOS) attending a reproductive endocrinology clinic.Design: Retrospective chart review.Participants and setting: 372 women with confirmed PCOS attending a reproductive endocrinology clinic at Adelaide University’s Research Centre for Reproductive Health.Main outcome measures: Prevalence of glucose tolerance abnormalities and association of such abnormalities with potential risk factors.Results: 4.0% (15 women) had diabetes mellitus, 15.6% (58) had impaired glucose tolerance and 80.4% (299) had normal glucose tolerance. There was a significant trend towards increasing prevalence of diabetes with increasing age (odds ratio [OR], 0.60; P = 0.0085). The prevalence of abnormal glucose tolerance (diabetes and impaired glucose tolerance together) was significantly higher with higher waist circumference (OR, 2.9; P = 0.05), higher body mass index (OR, 8.02; P = 0.0253), a family history of diabetes (OR, 1.56; P = 0.0192) and the presence of metabolic syndrome (OR, 5.62; P < 0.001).Conclusion: The prevalence of diabetes and impaired glucose tolerance is high in women with PCOS, especially in older women and those with abdominal obesity and a family history of diabetes.

Preeti Dabadghao MD · Bronwen J Roberts RN · Jim Wang PhD · Michael J Davies BA(Hons), MPH, PhD · Robert J Norman MD, FRACOG, CREI

To the Editor: Awareness about atypical and malignant modes of presentation of a clinical condition can avoid catastrophic outcomes, assist in correct diagnosis in the appropriate clinical setting and, as typified by the following case, offer complete cure. A 39-year-old woman presented with a 5-year history of intermittent, recurrent brief syncopal episodes. During an episode at presentation, telemetry showed torsade de pointes with ventricular fibrillation (Box), and external defibrillation was required to restore sinus rhythm. Amiodarone infusion was initiated at a local hospital before the patient was referred to our institution for further investigation and management. On presentation, her heart rate was 50 beats/min and her blood pressure was 170/95 mmHg. No other abnormalities were detected on examination. A resting electrocardiogram (ECG) showed prominent U waves, with a long QT interval (QTc of 540 ms). As the patient had mild hypokalaemia (serum potassium level, 3.1 mmol/L), mild hypocalcaemia (serum calcium level, 2.10 mmol/L) and a prolonged QT interval, the amiodarone infusion was discontinued, and supplementation with potassium and calcium was initiated. In view of the hypertension and hypokalaemia, primary aldosteronism was suspected. Serum cortisol, 24-hour urinary cortisol and 24-hour urinary catecholamine levels were normal. The plasma aldosterone/renin ratio was markedly elevated (1920/1.2 = 1595; normal, < 99). Failure of aldosterone suppression after acute saline loading was also noted. Computed tomography of the abdomen showed a right adrenal ovoid mass (1.9 × 1.2 cm). Adrenal vein sampling confirmed right lateralisation (right to left ratio, 40 : 1; aldosterone level in the right vein was 224 000 pmol/L while that in the left vein was 5570 pmol/L). Despite initial potassium supplementation, the hypokalaemia persisted and only improved after initiating diuretic therapy with amiloride. Two weeks later, laparoscopic right adrenalectomy was performed, and adrenocortical adenoma was confirmed histologically. After surgery, plasma aldosterone and renin levels normalised to 106 pmol/L and 8.9 mU/L, respectively. Nine months later, the patient was normotensive (without treatment) and had a normal ECG with no further recurrence of arrhythmias. This was a case of primary aldosteronism presenting as aborted sudden cardiac death and malignant syncope secondary to hypokalaemia-induced torsade de pointes. Most patients with primary aldosteronism are either asymptomatic or have symptoms related to hypertension or hypokalaemia (eg, polyuria, cramps, paraesthesia or muscle weakness); the diagnosis is often missed because of the non-specific clinical features. Primary aldosteronism presenting with cardiovascular collapse caused by hypokalaemic torsade de pointes and recurrent ventricular fibrillation is extremely rare, as is presentation as sudden cardiac death secondary to ventricular fibrillation.1 A prolonged QT interval has been reported in cases of primary aldosteronism,2 with values normalising after adrenalectomy.3 Torsade de pointes noted on telemetry in a 39-year-old woman

Aditya Kapoor · Timothy A Wells · Daniel Wong · John P O’Shea

To the Editor: Recent studies provide evidence that having a low serum vitamin D level is a risk factor for autoimmune disease, including type 1 diabetes mellitus (T1DM).1,2 Available data come from northern hemisphere countries where sunlight exposure levels and the genetic background of the population are different from those in Australia. We compared vitamin D levels in stored serum from Brisbane children and adolescents with T1DM who attended the Mater Children’s Hospital clinic with local historical control data from a previous study.3 Levels of 25-hydroxyvitamin D (25-OHD; the major circulating form of vitamin D) were lower in those with T1DM than in the control group, with no difference in levels of 1,25-dihydroxyvitamin D (1,25-[OH]2D; the biologically active form). Children and adolescents with T1DM were more than three times as likely to have vitamin D deficiency4 as those in the control group. There was a trend towards seasonal variation in 25-OHD levels, with mean levels (95% CI) being 53.8 nmol/L (47.0–60.6 nmol/L) in summer, 61.4 nmol/L (54.9–67.9 nmol/L) in autumn, 56.4 nmol/L (51.7–61.0 nmol/L) in winter and 64.7 nmol/L (58.8–70.6 nmol/L) in spring (P = 0.06), but no difference in seasonal variation between T1DM and control groups (P = 0.73). There was no difference in the ages or proportions of males and females in the two groups (Box). There were no differences in vitamin D levels between the sexes in either T1DM or control groups, nor any correlation with duration of diabetes. These observations support previous reports. One found low 25-OHD levels in 459 Swedish patients aged between 15 and 34 years who were newly diagnosed with T1DM compared with age-matched and place-matched controls.1 Another found low 25-OHD levels in 88 newly diagnosed children and adolescents.2 Understanding the nature of low vitamin D levels in people with diabetes is important because it potentially clarifies the mechanisms of autoimmune β-cell destruction, and may lead to interventions for preventing or delaying insulin dependence by using vitamin D or its analogues. Vitamin D probably acts by modifying the autoimmune response, as 1,25-(OH)2D modulates dendritic cell function to promote tolerogenic T cells. It may be relevant that we have recently found low blood dendritic cell counts in children and adolescents with T1DM.5 Vitamin D levels in our Queensland sample of children and adolescents were lower overall than those found in the subjects of the Swedish study, (mean 25-OHD levels [± SEM] were 96.7 ± 2.7 nmol/L for the control group and 82.5 ± 1.3 nmol/L for those with T1DM); this is unexpected given Brisbane’s latitude (29°S) compared with that of Sweden (about 55–65°N). These differences might be explained by differences in dietary intake, sun avoidance behaviours promoted in Queensland, or differences in the assays used, as the Swedish group used the Nichols chemiluminescence assay (Nichols Institute, San Juan Capistrano, Calif, USA) and we used the DiaSorin radioimmunoassay (DiaSorin Inc, Stillwater, Minn, USA). The observation in the Swedish study that the deficit in 25-OHD level did not resolve over time after diagnosis concurs with our finding of low levels in children and adolescents several years after diagnosis. While our pilot data cannot support causal inference, and is limited by being retrospective and our lack of information about history of sunlight exposure, dietary vitamin D intake, cultural factors such as sun avoidance or veiling, skin tone, and not having contemporaneous controls, it strongly supports the case for prospective clinical studies of vitamin D in T1DM. Comparison of clinical characteristics and vitamin D levels in healthy children and adolescents and those with type 1 diabetes mellitus Variable Control group Type 1 diabetes mellitus group P No. of children and adolescents 94 47 Age (range) 13.2 years (12.5–13.8 years) 13.6 years (12.6–14.6 years) 0.47* No. of males/females 44/50 21/26 0.81† Mean duration of diabetes (95% CI) — 4.7 years (3.9–5.5 years) Sample collection period July 2000 – December 2001 June 2001 – July 2006 Mean 25-OHD level (95%CI)‡ 64.6 nmol/L (61.3–67.9 nmol/L) 54.7 nmol/L (50.3–58.9 nmol/L) 0.0005* Mean 1,25-(OH)2D level (95% CI)‡ 126.7 pmol/L (115.8–137.6 pmol/L)§ 127.6 pmol/L (114.8–140.4 pmol/L) 0.92* Proportion 25-OHD-deficient (≤ 50 nmol/L) 18% (17/94) 43% (20/47) 0.002† (OR,¶ 3.4; 95% CI, 1.5–7.3) Proportion with 1,25-(OH)2D level below reference range (40–150 pmol/L) 0 (0/84) 4% (2/47) 0.13** (OR,¶ 9.3; 95% CI, 0.4–197.6) * t test. † χ2 test. ‡ DiaSorin radioimmunoassay double antibody assay (DiaSorin Inc, Stillwater, Minn, USA), performed by Queensland Health Pathology Services. § 84 controls; insufficient serum for analysis in 10. ¶ Odds ratio for deficiency in type 1 diabetes mellitus. ** Fisher’s exact test. 25-OHD = 25-hydroxyvitamin D. 1,25-(OH)2D = 1,25-dihydroxyvitamin D.

Ristan M Greer · Meredith A Rogers · Francis G Bowling · Helen M Buntain · Mark Harris · Gary M Leong · Andrew M Cotterill

To the Editor: I read with interest the excellent review article on the metabolic syndrome by Chew et al in the 16 October 2006 issue of the Journal.1 In their article the authors claim there is a lack of data about the relationship between hyperinsulinaemia and changes in free testosterone levels. As part of the Kuopio Ischaemic Heart Disease (KIHD) Risk Factor Study, an ongoing prospective epidemiological study of 2682 middle-aged Finnish men investigating risk factors for chronic disease, our research group has shown an association between the presence of metabolic syndrome at baseline and a change in sex hormone levels at follow-up after 11 years.2 In our study, men who met the World Health Organization criteria for metabolic syndrome both at baseline and at 11-year follow-up were at 2.6-fold increased risk of developing hypogonadism (serum total testosterone concentration < 11 nmol/L) during the study period compared with men who did not have metabolic syndrome. There was also a non-significant trend for men with metabolic syndrome to develop hypogonadism as defined by calculated free testosterone levels of < 225 pmol/L at 11-year follow-up.2 In the same cohort, we also reported a reverse association — that is, hypogonadism predicting metabolic syndrome.3,4 However, as the question posed by Chew et al was whether hyperinsulinaemia affects free testosterone levels, I examined the KIHD data further for evidence of such an association. I found that subjects grouped in ascending baseline fasting serum insulin quartiles had baseline mean free testosterone levels of 316 pmol/L (SD, 72 pmol/L), 312 pmol/L (SD, 77 pmol/L), 299 pmol/L (SD, 74 pmol/L) and 271 pmol/L (SD, 79 pmol/L), respectively (P < 0.001 for trend). At 11-year follow-up, mean free testosterone levels for subjects in each quartile were 248 pmol/L (SD, 64 pmol/L), 242 pmol/L (SD, 68 pmol/L), 229 pmol/L (SD, 70 pmol/L) and 216 pmol/L (SD, 67 pmol/L), respectively (P < 0.001 for trend). The proportional drop in free testosterone levels over 11 years was approximately the same in each quartile, ranging from 20% to 23%. On the basis of these data, it seems that hyperinsulinaemia is associated not only with a fall in serum total testosterone levels but also with a fall in free testosterone levels in a general population.

Tomi-Pekka Tuomainen

In reply: We thank Tuomainen for his interest in our review article, and for sharing with us his data showing an inverse association between fasting serum insulin levels and calculated serum free testosterone levels. We were cautious in our statement about the relationship between hyperinsulinaemia and free testosterone levels, as there are conflicting data in the literature regarding this,1,2 and few studies that directly measure free or bioavailable testosterone. Moreover, there is ongoing controversy about the calculation of free testosterone levels using total testosterone and sex hormone-binding globulin concentrations, with the validity and assumptions of some of these widely used estimation equations being called into question.3,4 We also echo the concerns of Allan et al5 about the potential pitfalls of diagnosing hypogonadism based on testosterone levels only. As the presence of low total (and even calculated free) testosterone in obese men may not necessarily reflect deficient androgen action, the diagnosis of androgen deficiency should only be made in the context of supportive clinical features. Furthermore, in abdominally obese men with the metabolic syndrome, levels of sex hormone-binding globulin and both total and calculated free testosterone can increase following weight loss,6 thereby obviating the inappropriate use of testosterone supplementation in such patients.

Gerard T Chew · Seng Khee Gan · Gerald F Watts

Objective: To assess the impact of iodine fortification of bread on the iodine status of pregnant women, and to determine if studies of iodine levels in school-age children were indicative of women’s gestational iodine status.Design: Urinary iodine surveys of pregnant Tasmanian women before and after bread was fortified with iodine in October 2001.Participants and setting: 285 women attending the Royal Hobart Hospital (RHH) antenatal clinic from 1 October 2000 to 30 September 2001 and 517 women attending the RHH antenatal clinic or primary health care centres in 2003–2006.Main outcome measures: Median urinary iodine concentration (UIC) for comparison against the World Health Organization recommendation of of 150–249 μg/L for pregnant women.Results: Before supplementation, the median UIC of the 285 women attending the RHH antenatal clinic was 76 μg/L. After supplementation, median UICs were 81 μg/L for 288 women attending primary health care centres and 86 μg/L for 229 women attending the RHH antenatal clinic. Differences in mean UIC were not significant for either the antenatal clinic group (P = 0.237) or the primary health care group (P = 0.809) compared with the pre-supplementation group.Conclusions: Iodine deficiency in pregnancy persists despite being corrected in Tasmanian children. Successful iodine supplementation must target reproductive-age and pregnant women and be substantiated by ongoing monitoring during pregnancy and lactation. A robust national program for correcting iodine deficiency is urgently needed. Mandatory universal salt iodisation has international endorsement, and should be considered the preferred strategy for eliminating iodine deficiency in Australia.

John R Burgess BMedSc, MD, FRACP · Judy A Seal MPH, AdvAPD · Georgina M Stilwell MB BS · Peter J Reynolds MB BS, FRACOG · E Roscoe Taylor GradDipEpid, MRNZCGP, FAFPHM · Venkat Parameswaran PhD

Reducing the burden of diabetes will require action well beyond the health service sphere Type 2 diabetes represents a serious public health problem for Indigenous Australians, occurring at a much higher prevalence than in the non-Indigenous population, and with a much earlier age of onset of the disease and its micro- and macrovascular complications.1,2 It is likely that diabetes is an important contributor to the considerably higher circulatory disease mortality rate among Indigenous Australians at young ages (9–10 times higher in Indigenous men aged 25–44 years, and 12–13 times higher in Indigenous women aged 35–54 years).1 Thus, diabetes imposes significant financial and human costs on Australian society, which are disproportionately borne by Indigenous individuals, families and communities. Of three articles about diabetes in Indigenous Australians in this issue of the Journal, two provide evidence that the problem is escalating. Craig et al3 analysed data from the Australasian Paediatric Endocrine Group NSW Diabetes Register and found that type 2 diabetes accounts for 11% of new diabetes cases among 10–18-year-olds, and that the incidence in Indigenous children was about six times higher than that in non-Indigenous children (→ Type 2 diabetes in Indigenous and non-Indigenous children and adolescents in New South Wales). McDermott et al4 found that, for Torres Strait Islanders, there were significant increases in body mass index (BMI) — the major risk factor — between 1999 and 2005, and a very high 5-year incidence of diabetes (→ Diabetes in the Torres Strait Islands of Australia: better clinical systems but significant increase in weight and other risk conditions among adults, 1999-2005). Is it possible to prevent type 2 diabetes? International studies indicate that, in people with impaired glucose tolerance, an intensive focus on diet and physical activity can substantially reduce progression to diabetes, and to an equal or greater extent than pharmacological interventions.5 Although BMI and age are the two strongest predictors of diabetes for Indigenous Australians, leanness is protective.6,7 As in all Australians, preventing diabetes goes hand in hand with preventing excessive weight gain, but trends in overweight and obesity are unambiguously upwards. Preventing excessive weight gain in Indigenous communities, which are profoundly disadvantaged relative to mainstream Australia, is complicated by the strong link between poverty and obesity. People living in poverty tend to maximise calories per dollar spent on food,8 and energy-dense foods rich in fats, refined starches and sugars represent the lowest cost options. Healthy diets based on lean meats, whole grains, and fresh vegetables and fruits are much more costly. Poverty in Indigenous communities is related to high unemployment and welfare dependency; living conditions are overcrowded, and community infrastructure is poor, with limited access to good quality foods.1 Many of these factors are compounded by remote living, although successful prevention of obesity in some outstation communities has been associated with greater physical activity, consumption of bush foods, and ownership of and access to traditional homelands.9 Are there any opportunities for practical intervention? Low birthweight, which is linked to an increased risk of central obesity and type 2 diabetes in adult life, is more common in Indigenous and other socially disadvantaged communities, and is linked to maternal smoking, overcrowded living conditions and mothers’ perceived stress.10 Furthermore, diabetes in pregnancy increases the risk of early onset obesity and diabetes in the offspring. This can be attenuated by improved control of gestational diabetes, and by the mother breastfeeding for at least 2–3 months. A systematic approach to improving nutritional status of infants should be a priority — including the option (controversial in some circles) of providing subsidised food. We have observed that a community decision to provide a healthy breakfast and lunch 5 days a week for primary school children was a major step towards the children achieving their recommended daily intakes for a number of key nutrients (unpublished data). Improved maternal and child health could be an important and cost-effective contributor to diabetes prevention programs at the population level. Preventing and managing the complications of diabetes, such as cardiovascular risk factors, also involve lifestyle modification. Through changes in food supply, increased opportunities for physical activity, and health promotion, Indigenous communities were able to achieve amelioration of dyslipidaemia, improved insulin action (even in the absence of weight loss), and increased in red cell folate and reduced homocysteine levels.11 A large international trial has shown that fish and fish-oil supplements reduce coronary heart disease mortality.12 There seems little argument that improving the quality use of medicines (including through greater access) is one of the most cost-effective approaches to reducing the additional and preventable burden of chronic illness among Indigenous people.13,14 Angiotensin-converting enzyme (ACE) inhibitors have been shown to reduce mortality in an Aboriginal community with a high prevalence of end-stage kidney failure.15 Internationally, numerous trials have reported the effectiveness of statin therapy in reducing vascular mortality. Metformin improves glycaemic control in diabetes, without weight gain. Yet, the gaps between the evidence and actual practice, in both Indigenous communities and the broader community, remain unacceptably large, and are limiting gains for those at risk of diabetes and for those who already have the disease and related conditions. The article by McDermott et al4 illustrates the value of systematic primary health care approaches to diabetes control, including electronic health information systems, screening, management protocols, recall systems, improved specialist access, quality improvement activities, and staff support and training (→ Diabetes in the Torres Strait Islands of Australia: better clinical systems but significant increase in weight and other risk conditions among adults, 1999-2005). Such systems are integral to improving the quality and outcomes of clinical care. All three articles on diabetes in Indigenous people in this issue3,4,16 also address aspects of screening. Taken together, they support a critical role of coordinated health system approaches to diabetes identification and control. Simple point-of-care procedures, as developed by Marley et al16, could form the basis of cost-effective screening for diabetes (and other vascular risk factors) in high-risk populations, and may be able to accurately identify those who could benefit from more immediate pharmacological and non-pharmacological therapies (→ Point-of-care testing of capillary glucose in the exclusion and diagnosis of diabetes in remote Australia). The prevention and management of diabetes are critical to the future health of Indigenous as well as non-Indigenous Australians. But, there is no simple solution. The effectiveness of clinical and public health interventions is limited in Indigenous people, by the added burden of systematic historical and contemporary discrimination. Getting it right will require better clinical treatment and action well beyond the health service sphere. This is one of contemporary Australia’s greatest challenges.

Kerin O'Dea AO, BSc, PhD · Kevin G Rowley PhD · Alex Brown BMed, MPH, FCSANZ