Bias in Clinical Research.

Abstract

Clinical epidemiological research entails assessing the burden and etiology of disease, the diagnosis and prognosis of disease, the efficacy of preventive measures or treatments, the analysis of the risks and benefits of diagnostic and therapeutic maneuvers, and the evaluation of health care services. In all areas, the main focus is to describe the relationship between exposure and outcome and to determine one of the following: prevalence, incidence, cause, prognosis, or effect of treatment. The accuracy of these conclusions is determined by the validity of the study. Validity is determined by addressing potential biases and possible confounders that may be responsible for the observed association. Therefore, it is important to understand the types of bias that exist and also to be able to assess their impact on the magnitude and direction of the observed effect. The following chapter reviews the epidemiological concepts of selection bias, information bias, intervention bias, and confounding and discusses ways in which these sources of bias can be minimized.