. Author manuscript; available in PMC: 2023 Apr 1.

Published in final edited form as: Ann Intern Med. 2022 Sep 13;175(10):1423–1430. doi: 10.7326/M21-4204

Table 3.

Performance of Comorbidity Measures in Predicting Mortality, Indicated by c-Statistic

Variables	30-d Mortality, c-Statistic (95% CI)	In-Hospital Mortality, c-Statistic (95% CI)	1-y Mortality, c-Statistic (95% CI)
Model with demographics	0.653 (0.651–0.655)	0.610 (0.607–0.614)	0.664 (0.663–0.664)
+ S-Elixhauser summary score^*	0.705 (0.703–0.707)	0.654 (0.651–0.657)	0.743 (0.741–0.744)
+ Elixhauser comorbidity indicator (38 individual comorbidities )^*	0.711 (0.709–0.713)	0.658 (0.655–0.661)	0.748 (0.747–0.749)
+ AHRQ Elixhauser summary score^*	0.688 (0686–0.690)	0.642 (0.638–0.645)	0.719 (0.718–0.721)
+ Charlson comorbidity indicator (17 individual comorbidities)^*	0.696 (0.694–0.698)	0.641 (0.638–0.644)	0.730 (0.729–0.731)
+ Charlson summary score^*	0.690 (0.688–0.693)	0.638 (0.634–0.641)	0.726 (0.724–0.727)

AHRQ = Agency for Healthcare Research and Quality; S-Elixhauser = summary Elixhauser.

Model with S-Elixhauser score reflects internal validation in the population in which it was derived, whereas all of the other models with Charlson or Elixhauser scores reflect external validation.