The conservativeness of standard C statistics in the prediction of clinical events

Le résumé fourni par la source

The C statistic, also known as the concordance index (C-index), is widely used in clinical research to assess the discriminative ability of risk prediction models. Its appeal lies in its intuitive interpretation and broad applicability, particularly in fields such as cardiovascular medicine and oncology, where accurate risk stratification is essential. However, despite its popularity, the C statistic has notable limitations that can undermine its utility in both research and clinical practice. Chief among these is its inherent conservativeness: the C statistic is often insensitive to meaningful improvements in model performance when new biomarkers or risk factors are added to an already robust model. This insensitivity stems from its rank-based nature, which focuses solely on the correct ordering of risk predictions rather than the magnitude of improvement. As a result, significant advances in risk estimation may be overlooked, potentially discouraging the adoption of clinically valuable innovations. Furthermore, the C statistic does not account for calibration-the agreement between predicted and observed outcomes-or the clinical consequences of misclassification. Alternative metrics, such as the Mean Absolute Difference (MAD), Brier score and Net Reclassification Improvement (NRI), offer complementary perspectives by capturing aspects of predictive accuracy and clinical relevance that the C statistic may miss. A comprehensive evaluation of risk models should therefore integrate these alternative measures to ensure that predictive tools are both statistically robust and clinically meaningful, ultimately advancing patient care and the practice of precision medicine.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.