Current Controversies Does clinical epidemiology have a role in clinical practice?

摘要

In this issue of the Journal, Jelinek challenges the utility of clinical epidemiology, given his experience in clinical cardiology practice.1 In a research-study population of subjects selected for coronary angiography, he found no predictive relationship between the clinical features of anginal symptoms and the severity of angiographically determined coronary artery disease (CAD). However, on entering clinical practice, his experience was that classical symptoms of angina were highly predictive of the presence of CAD. Given this apparent discrepancy, he argues the case for the deconstruction of clinical epidemiology.Influences on the predictive ability of any particular diagnostic test are an important consideration to the discussion. The practice setting of the consultant physician is clearly an important influence, as evidenced by the relative likelihood of CAD in a patient with epigastric pain referred to a cardiologist, as opposed to a gastro-enterologist. The post-test likelihood of coronary disease can be expressed as a function of pre-test likelihood, sensitivity and specificity according to Bayes' theorem of conditional probability. While the use of a Bayesian approach may aid, it does not replace the physician's judgement.