Statistical tests answer a narrow, mathematical question - is there likely a real difference between group means? But science aims to produce clinically useful knowledge to benefit individual patients. This nuclear medicine study serves as a sobering reminder that statistical significance does not reliably predict clinical usefulness.
While group averages may differ in a statistically compelling fashion, typical cutoffs misclassify a full third of healthy individuals. So a statistically significant finding applied blindly in clinical practice, without further consideration of its clinical implications, is prone to harm individuals.
The mathematical abstractions of statistics have undeniable value. But statistical significance is easily misinterpreted as clinical importance when groups are viewed as interchangeable with individuals. As this study shows, what seems persuasive in aggregate often breaks down at the individual level. Statistical findings should be interpreted judiciously, with due consideration of clinical context and implications.
Citation: Heston TF, Wahl RL. How often are statistically significant results clinically relevant? Not often. J Nucl Med. 2009;50(Suppl 2):1370. https://zenodo.org/doi/10.5281/zenodo.10198933