Re: 'Association of small artery elasticity with incident cardiovascular disease in older adults: The multi-ethnic study of atherosclerosis'

摘要

We read with interest the study by Duprez et al. and were especially intrigued by the conditional net reclassifi-cation index (NRI) statistic promoted in the paper. While the idea of presenting an NRI for each baseline risk stratum has obvious utility, we were concerned that the authors did not correct for the expected value in each stratum. While the expected value for the overall NRI is 0, using the same assumptions of diagonal symmetry we have previously demonstrated that the expected NRIs for individual baseline strata are nonnull . These are easily calculable and can be incorporated to generate a bias-corrected conditional NRI .