Whenever possible, clinical decision making should be evidence-based. In reality, the evidence may be incomplete and the practitioner must use personal clinical judgement. This should be based on recommendations or guidelines provided by authoritative groups.

An example of clinical practice reliant on guidelines with an incomplete evidence base is the use of low-dose aspirin for primary prevention of cardiovascular disease (CVD) events in people with diabetes mellitus. Diabetes Australia, the Royal Australian College of General Practitioners, and the National Health and Medical Research Council (NHMRC), as well as numerous associations in the United States and United Kingdom, have all recommended low-dose aspirin for men and women with diabetes (Box).

Until very recently, there were no randomised controlled trials of aspirin use for people with diabetes to provide the evidence for these guidelines. Rather, the recommendations are based on the rationale that low-dose aspirin should benefit people with diabetes because aspirin has proven benefit for secondary prevention of CVD events, and people with diabetes have a risk of CVD events equivalent to the risk found in secondary prevention populations. However, people with diabetes represent a heterogeneous population, and an individual patient’s risk of CVD events varies according to factors such as age at diagnosis and duration of diabetes.8 Therefore, the guidelines for aspirin use in people with diabetes, based on extrapolation from secondary to primary prevention, may be flawed.

Two substudies of the Bettering the Evaluation And Care of Health (BEACH) program, one undertaken in 20069 and the other in 2007,10 generated data about patients with diagnosed type 2 diabetes and allowed us to determine the compliance of Australian general practitioners with the relevant guidelines. Using the BEACH substudy data, we analysed aspirin usage by patients with and without diagnosed CVD comorbidity. The results indicated that 39% of people with type 2 diabetes were taking aspirin for primary prevention.

In late 2008, two randomised controlled trials of low-dose aspirin in people with diabetes were reported.11,12 They showed no significant effect of aspirin on the primary endpoint (CVD events). These studies attracted a good deal of attention, not only in the medical press but the lay press as well (eg, Norman Swan’s The health report on Radio National13). They also left medical practitioners with a conundrum — whether or not to prescribe low-dose aspirin to their patients with diabetes but no overt CVD. In this editorial, we highlight the deficiencies in these trials; suggest that the evidence for the use of aspirin in this context is still lacking; and inform clinicians of ongoing trials that should provide adequate power to address this issue reliably.

The two 2008 trials of low-dose aspirin therapy for people with diabetes were substantially underpowered to address the question of aspirin effectiveness for reducing CVD events.14,15 The Prevention of Progression of Arterial Disease and Diabetes (POPADAD) study of low-dose aspirin (100 mg daily)11 followed 1276 participants with diabetes and asymptomatic peripheral arterial disease but no CVD events for a median of 6.7 years. The annual CVD event rate was less than 3% in those assigned placebo (lower than the expected rate of 8% per annum). There were 116 primary CVD events in subjects receiving aspirin, compared with 117 in those not receiving aspirin (hazard ratio, 0.98). In the open-label Japanese Primary Prevention of Atherosclerosis with Aspirin for Diabetes (JPAD) study,12 2539 patients with type 2 diabetes and no history of CVD were treated either with aspirin (81 mg or 100 mg daily) or with no aspirin. The median follow-up period was 4.37 years. In both groups, actual event rates were about three times lower than anticipated rates. A 20% reduction in the primary endpoint of composite CVD events for the group assigned aspirin was not statistically significant because of wide confidence intervals. Nevertheless, the secondary endpoint of CVD mortality was significantly reduced, and, in a sub-analysis of subjects aged 65 years and over at baseline, aspirin significantly reduced CVD events. For both the POPADAD and JPAD trials, there was no significant difference in the occurrence of adverse events among subjects receiving or not receiving aspirin. As a result of the lower than expected event rates (possibly related to effects of more optimal background therapies) and the relatively low numbers of people included in the trials, results of POPADAD and JPAD have not ruled out important benefits or risks of aspirin. Larger trials are needed.

A major study (A Study of Cardiovascular Events in Diabetes [ASCEND])16 is underway in Britain to determine whether low-dose aspirin has a benefit for primary prevention of CVD in people with diabetes. About half of its estimated sample size of 10 000 men and women has been recruited. In Australia, our own trial of aspirin therapy for primary prevention in 19 000 people aged 70 years and over, the Aspirin in Reducing Events in the Elderly (ASPREE) study17 (clinical trial registration number ISRCTN83772183), will also be including participants with diabetes on the basis that equipoise prevails between the benefits and risks of aspirin therapy. In the ASPREE study, a GP co-investigator will help decide whether the patient is a suitable candidate for the placebo-controlled trial.

A recent editorial in the Journal of the American Medical Association concluded that

We concur that an enhanced evidence base can inform a more rational approach to therapy.