Browse the corpus

Interpreting trial results-time for confidence and magnitude and not P values please. The problems with the P value as the single metric to summarize the results of a study are being recognized. It captures a single domain-random error-but it is relatively uninformative about more critical domains for deciding whether the results should be applied to clinical care and policy. Alternatives include the components of the outcomes reported (relevance, magnitude, frailty, and net benefit) and confidence (risk of bias and directness).