Evaluating the applicability of replication success metrics in animal-to-human translation: A simulation study.

Most replication success metrics, except meta-analysis and Bayes factor, become liberal with increasing heterogeneity, limiting their reliability in animal-to-human translation assessment.

• Why it matters: Accurately predicting whether animal study results will replicate in humans is crucial for advancing biomedical research and avoiding costly failures, yet current metrics' applicability remains uncertain.
• What they did: A simulation study evaluated nine metrics across 648 scenarios, varying effect sizes, heterogeneity, sample sizes, and study pooling, to determine their effectiveness in assessing translation success.
• The result: No single metric proved universally optimal; combining multiple metrics, especially those like skeptical p-values and weighted Edgington's method, enhances reliability in evaluating translation outcomes.