Overview of clinical research design.

PURPOSE Basic concepts and terminology of clinical research design are presented for new clinical investigators. SUMMARY Clinical research, research involving human subjects, can be described as either observational or experimental. The findings of all clinical research can be threatened by issues of bias and confounding. Biases are systematic errors in how study subjects are selected or measured, which result in false inferences. Confounding is a distortion in findings that is attributable to mixing variable effects. Uncontrolled observation research is generally more prone to bias and confounding than experimental research. Observational research includes designs such as the cohort study, case-control study, and cross-sectional study, while experimental research typically involves a randomized controlled trial (RCT). The cohort study, which includes the RCT, defines subject allocation on the basis of exposure interest (e.g., drug, disease-management program) and follows the patients to assess the outcomes. The case-control study uses the primary outcome of interest (e.g., adverse event) to define subject allocation, and different exposures are assessed in a retrospective manner. Cross-sectional research evaluates both exposure and outcome concurrently. Each of these design methods possesses different strengths and weaknesses in answering research questions, as well as underlying many study subtypes. CONCLUSION While experimental research is the strongest method for establishing causality, it can be difficult to accomplish under many scenarios. Observational clinical research offers many design alternatives that may be appropriate if planned and executed carefully.