Clinical inference: critically weighing the evidence from trials and registries to make clinical decisions

The results of even excellent randomized trials are applicable primarily to patients who would have met all of the inclusion criteria and none of the exclusion criteria, specified in the study protocol (generalizability or external validity). In addition, three important sources of potential systematic error, or bias, are as follows: selection bias, confounding, and information bias. Selection bias is relevant to nearly all randomized clinical trials in terms of external validity, or generalizability, of even internally valid results. Selection bias, confounding, and information bias can all limit the internal validity of prospective, registry data. Carefully reading the methods sections of both randomized trial and registry reports, with these potential limitations in mind, is the first step in critically using clinical research to help make more informed clinical decisions.