Revisiting sample size: are big trials the answer?

Abstract

The superiority of the evidence generated in randomized controlled trials over observational data is not only conditional to randomization. Randomized controlled trials require proper design and implementation to provide a reliable effect estimate. Adequate random sequence generation, allocation implementation, analyses based on the intention-to-treat principle, and sufficient power are crucial to the quality of a randomized controlled trial. Power, or the probability of the trial to detect a difference when a real difference between treatments exists, strongly depends on sample size. The quality of orthopaedic randomized controlled trials is frequently threatened by a limited sample size. This paper reviews basic concepts and pitfalls in sample-size estimation and focuses on the importance of large trials in the generation of valid evidence.