Key features of clinical trial design - power and sample size calculation

Definition

Statistical power is the probability that a study will detect a true effect if one genuinely exists - determined by sample size, expected effect size, variability in the outcome measure, and the chosen significance threshold - with power/sample size calculation performed before a study begins to ensure adequate ability to detect a clinically meaningful effect
A foundational element of clinical trial design methodology, with inadequate power being a common, important limitation to identify when critically appraising a "negative" (non-significant) trial result

An underpowered study (too small a sample size relative to the true effect size and outcome variability) risks a Type II error (failing to detect a true effect that actually exists) - this means a non-significant result from an underpowered study should not be interpreted as confidently demonstrating "no effect," since the study may simply have lacked the statistical power to detect a real, clinically meaningful effect that was genuinely present

Power calculations are performed at the study design stage, requiring pre-specification of the expected effect size (often based on prior evidence or the minimum clinically important difference), outcome variability, desired power (conventionally 80% or 90%), and significance threshold
A study's reported/achieved power (or, retrospectively, whether the confidence interval around a non-significant result excludes clinically important effect sizes) should be checked when interpreting a "negative" trial result

A non-significant trial result should prompt checking whether the study was adequately powered before concluding "no effect" - an underpowered negative trial provides much weaker evidence against a treatment effect than an adequately powered one
Sample size calculations should be based on a clinically meaningful minimum effect size, not simply the largest feasible sample given resource constraints, to ensure the study can actually answer the intended clinical question

Type II error and underpowered "negative" trials (a frequently tested, clinically important limitation), pre-specified power calculation requirements, minimum clinically important difference as the basis for sample size planning, confidence interval width as a retrospective power indicator

A non-significant ("negative") trial result does not necessarily mean a treatment has no effect - if the study was underpowered (too small a sample size for the true effect size), it may simply have failed to detect a real, clinically meaningful effect that genuinely exists (a Type II error), which is why checking whether a trial had adequate statistical power is an essential part of correctly interpreting a non-significant result rather than accepting it at face value

5 more sections, plus exam facts

Premium unlocks every note across every specialty, and the full exam fact library behind it.