Urate levels as a predictor of cardiac deaths: causal relation or mere association?

Over the last 50 years, through a series of animal experimental studies, physiological studies, observational studies in various human populations, and randomized clinical trials, a number of risk factors have been shown to be related to atherosclerotic vascular disease. Some of these risk factors are undoubtedly causal and include elevated cholesterol, blood pressure and glucose and tobacco use. Other risk factors have been proposed such as some markers of thrombosis, markers of renal dysfunction such as creatinine or microalbuminuria, inflammatory markers, markers of poverty and social disparity, psychological factors and markers of oxidant activity. It is not clear if these associations are real or spurious; and if real, whether they are causal or mere associations. These distinctions are only possible through a range of different types of studies, that include large epidemiological studies, complemented by randomized interventional trials with measurements of both clinical outcomes and atherosclerosis. In this issue Wong et al. report a relationship between serum urate concentrations measured 1 year after strokes in 354 individuals and subsequent cardiac mortality. Adjustment or stratification for a range of confounders that may increase both urate levels and risk of cardiac deaths such as diuretic use, history of diabetes or hypertension, or creatinine levels, did not alter the findings. How does one interpret this study? First how robust are the findings in Wong’s study? Although consistency across different ways of analysing the data should add to our confidence, the fact remains that the findings are based on only 19 cardiac deaths (with 10 occurring among those not on diuretics). This low number of events, the lack of impact on other vascular deaths (presumably because the authors do not report the analysis, despite having the data) and lack of information on related non-fatal events, adds considerable uncertainty to their claims. With very few events, statistical adjustments are unreliable and hence should be interpreted with considerable caution.