The problems with some epidemiological studies

Epidemiological studies provide a valuable tool for the investigation of the causes of disease. However, such studies alone are rarely able to prove cause. No study is perfect and the evaluation of results must take account of the design and execution of the study together with the analytic methods used. Before imputing cause it is important to consider the findings against the criteria set out by Bradford Hill. Some of the problems with studies are discussed including the specification of the data collected, accuracy of information, end point definition, study size and the way results are presented. The issues are illustrated by reference to published papers. It is concluded that unless studies are evaluated and interpreted with care they may result in more harm than good.