Guidelines for reading case-control studies.

Abstract

To develop guidelines for evaluation of case-control reports a survey of 37 experts in the performance and evaluation of case-control studies was conducted. A majority of the respondents listed 20 items as essential; 80-90% considered identification of case and control sources, exclusion criteria, and response rate as essential; 75-80% considered information on methods of data collection, "blinding" of interviewers, investigation of bias, and methods of dealing with confounding variables essential; 70% considered a description of the analytic methods and 57% the presentation of confidence limits essential. Twenty items judged essential by more than half the survey participants were used as guidelines to appraise 48 case-control studies published in 1984. In 88% of these studies information was lacking on at least one of the items. The proposed guidelines serve as a framework for readers to effectively assess the validity of a case-control report.