Clinical Trials: Design Flaws Associated with Use of a Placebo

BACKGROUND: The placebo arm of a trial, although it controls for psychological placebo effects and natural history of a disease and facilitates effective blinding, is susceptible to bias just like the true treatment arm is. The placebo arm of a trial may even be particularly vulnerable to specific types of bias. The present article focuses on design flaws associated with the use of a placebo. RESULTS: (1) Placebo-controlled trials using mortality rather than morbidity as the end point may have both ethical and statistical problems. (2) Bias may affect placebo data to a larger extent than active treatment data. (3) Trials may provide an atmosphere of exaggerated placebo effects and, consequently, inadequate power to determine differences between placebo and active treatment groups. (4) A biased placebo period due to carryover effect is a common problem of controlled trials with a crossover or self-controlled design. (5) Adjustment for asymmetries between placebo and treatment group in a parallel-group trial is inappropriate if such asymmetries are not completely independent on the main variables of the study. (6) Samples may not be representative when a large number of patients inappreciative of a possible treatment with placebo refuse randomization. CONCLUSIONS: It is emphasized that routinely accounting for design flaws associated with use of placebo may further improve the powerful method of controlled clinical trials.