Blindness and Thrombolysis

Abstract

A major concern in clinical research is that of bias. Protocol design, data collection, analysis of results and presentation of data are all areas of potential bias. Trials shown to have been biased tend to lose their validity. Consequently, it is crucial to recognize when and where bias may occur and to guard against it. A significant research technique for minimizing bias is blinding, that is, keeping some or all participants in the protocol from knowing the treatment assignment, active agent or placebo. Bias may be either negative or, more likely, positive. There is often an assumption by the patient and investigator that the intervention will be helpful, and reporting of side effects and outcome measures may therefore not be accurate. The impact of bias on a study is particularly critical when the study is small and the end-points subjective [1, 2]. It is the purpose of this review to identify sources of bias and to suggest mechanisms of prevention. Potential sources include: the research subject, the physician-investigator, the research staff, the laboratory, the study medication, and the sponsor. The sponsor is assumed to be a pharmaceutical company.