The design and interpretation of pilot trials in clinical research in critical care

Background: Pilot trials are important to ensure that large randomized trials are rigorous, feasible, and economically justifiable. The objective of this review is to highlight the importance of randomized pilot trials and to describe key features of their design and interpretation using examples from critical care. Methods: We searched MEDLINE (1997-2007) and contacted experts to identify pilot randomized trials to exemplify and summarize their key methodologic features including objectives, sample size determination, outcomes, analysis, and reporting. Results: Pilot trials can have distinct and broad objectives. Investigators can predefine explicit criteria for determining their success. Surrogate outcome analyses are common in pilot trials, yet are usually underpowered to detect meaningful differences in clinically important end points and thus, should be cautiously interpreted. Pilot trials can facilitate successful conduct of large clinical trials by informing study design and streamlining protocol implementation. Recommendations: We recommend that investigators define suitable objectives, determine sample size estimates, and select outcomes that will address their specific pilot trial objectives. Clinical effects documented in pilot trials should be reported with caution to avoid undue enthusiasm or pessimism about unstable estimates. Further methodologic work is required to identity optimal pilot trial design, indexing, and reporting. (Crit Care Med 2009; 37[Suppl.]:S69-S74)