<< Previous Section | < Previous Page | Next Page > | Next Section >> Appropriate Risk to Benefit Ratio (page 2 of 3) Power can be defined as the adequacy of the number of research participants (treatment and controls) to confidently achieve or rule out statistically significant results for its principal end point. Estimation of power should always allow for dropouts and recruitment difficulties. Problems with recruitment and retention of participants to completion of the study impair power, sometimes making an investigation hopelessly biased or useless. A particular problem is the pursuit of subset analyses under conditions where the main result is negative. The subsets may not have enough power for a sound conclusion. Normal Controls are research participants who do not have the condition under study. In physiological and behavioral interventions they undergo the same protocol as the participants with the condition under study in order to compare the two responses. Subjecting them to any significant risk may be inappropriate. However, Phase 1 clinical trials may be carried out in small numbers of normal control subjects who should be sure to understand the significant risks of the intervention. Controls are research participants who receive an inactive treatment. In most trials they are selected by computer lottery from the group of eligible candidates with the condition under study. Historical controls are subjects from prior studies or observational investigations whose data are compared with those of the current participants. Historical controls were used for years in clinical research and are still sometimes employed because they do not require additional data collection and risk. They often produce biases because the research population rarely duplicates the historical population. Blinding refers to a process whereby the participant does not know whether he/she is receiving an active agent or a similar appearing inactive substance or mock procedure. Blinding is also used in research to refer to investigators who analyze components of a study such as X-rays or EKGs without knowing the identity and treatment of the participant. "The X-rays were read blind." Double blinding is a process whereby neither the investigator nor the participant knows which agent the participant is receiving. Usually the research pharmacy holds the master list in case there are complications. Over the course of the last 30 years it became apparent that blinding both participants and research teams reduced biases in the results of studies where subjective elements were important. One result that is almost invariably subjective is the adverse event profile. In the absence of blinding very serious biases have occurred. Sometimes the effects of the agent in question are so obvious that true blinding is impossible. For example, if a weight loss drug were immediately effective, then the results would be obvious to everyone. Under those circumstances special attention has to be given to unbiased evaluation of adverse events, and conflicts of interest (see below) must be avoided. << Previous Section | < Previous Page | Next Page > | Next Section >> Chapter 3 Quick Links Ethics and Study Design Introductory Ethics Design Appropriate Risk to Benefit Ratio Selection of Subject Populations Cases Bibliography Chapter 3 Download (PDF)