Powering Neurology Drug Development Trials to Reduce False Negative Results

Sensitive and reliable measurement of therapeutic outcomes in neurology drug trials remains challenging. Clinical rating scales and their total scores are often considered the gold standard and used as primary efficacy endpoints in studies because of their acceptance by the clinical community and regulators. However, rating scales have limitations when used in drug development, and we hypothesize that their lower statistical power is one possible reason leading to false negative results. This paper proposes a framework to address four systemic components of statistical power-size of the treatment effect, variability, sample size, and statistical threshold-to minimize the odds of observing false negatives, and thereby maximize the odds of trial success. Drawing from the literature and biopharmaceutical anecdotal evidence, we identify key factors that affect each of these power components and assess their influence. We also highlight challenges with rating scale endpoints associated with total scores and large sample sizes. We identify potential causes of rating scale endpoint failure and provide an indication-agnostic framework for understanding the relationship between clinical rating scale performance and study power. The proposed methods can be applied to both study design and post hoc analysis.