THE N-OF-1 RANDOMIZED CONTROLLED TRIAL - CLINICAL USEFULNESS OUR 3-YEAR EXPERIENCE

Objective: To review the feasibility and effectiveness of n-of-1 randomized controlled trials (n-of-1 trials) in clinical practice. Design: Individual trials were double-blind, randomized, multiple crossover trials. The impact of n-of-1 trials was determined by eliciting physicians' plans of management and confidence in those plans before and after each trial. Setting: Referral service doing n-of-1 trials at the requests of community and academic physicians. Object of Analysis: All trials were planned, started, and completed by the n-of-1 service. Measures of Outcome: The proportion of planned n-of-1 trials that were completed and the proportion that provided a definite clinical or statistical answer. A definite clinical answer was achieved if an n-of-1 trial resulted in a high level of physician's confidence in the management plan. Specific criteria were developed for classifying an n-of-1 trial as providing a definite statistical answer. Main Results: Seventy-three n-of-1 trials were planned in various clinical situations. Of 70 n-of-1 trials begun, 57 were completed. The reasons for not completing n-of-1 trials were patients' or physicians' noncompliance or patients' concurrent illness. Of 57 n-of-1 trials completed, 50 provided a definite clinical or statistical answer. In 15 trials (39% of trials in which appropriate data were available), the results prompted physicians to change their 'prior to the trial' plan of management (in 11 trials, the physicians stopped the drug therapy that they had planned to continue indefinitely). Conclusion: We interpret the results as supporting the feasibility and usefulness of n-of-1 trials in clinical practice.