Beyond p-values: A case for clinical relevance

Abstract

There are inherent flaws to relying on p-values, or statistical significance, when assessing intervention effectiveness. Perhaps the biggest limitation to reporting and discussing statistically significant findings is that they do not necessarily translate to improvements that are perceptible or meaningful to patients. This opinion piece urges researchers to go beyond the prevailing standard of reporting p-values by using additional approaches that approximate clinical relevance; specifically anchor-based and distribution-based methods. To this end, we discuss the relative strengths and weaknesses of each method, and how they can not only be incorporated into existing data analyses but also inform future research design.