Current concepts of risk in psychiatric disorders

Purpose of review A biostatistical 'autopsy' was done on a sample of psychiatric risk studies published in 2002, in order to evaluate the approaches currently used to seek risk factors. The comments are directed not towards criticism or blame, but towards improved understanding of current risk research results, and to the development of better ways of dealing with risk research problems in the future. Summary The following recommendations are made: (1) there should be greater emphasis on longitudinal studies in which there is a clear time-line; (2) care should be taken to avoid 'muddling' populations in which the risks and risk factors might differ; (3) there should be greater emphasis on seeking to understand how risk factors work together (proxies, overlapping, moderators and mediators); (4) statistical hypothesis testing should be de-emphasized; (5) there should be greater emphasis on developing potency measures interpretable for clinical and policy decision-making; and (6) there should be an avoidance of overgeneralization of results.