A head-to-head comparison of the validity and predictive ability for health outcomes of diagnosis versus medication-based comorbidity indices

Moritz Platen; Maresa Buchholz; Anika Rädke; Fabian Kleinke; Annelie Scharf; Michelle Pfaff; Audrey Iskandar; Neeltje van den Berg; Wolfgang Hoffmann; Bernhard Michalowsky

doi:10.1007/s40520-025-03073-w

A head-to-head comparison of the validity and predictive ability for health outcomes of diagnosis versus medication-based comorbidity indices

Aging Clin Exp Res. 2025 May 27;37(1):172. doi: 10.1007/s40520-025-03073-w.

Affiliations

¹ German Center for Neurodegenerative Diseases (DZNE), site Rostock/Greifswald, Ellernholzstrasse 1-2, D-17489, Greifswald, Germany. moritz.platen@dzne.de.
² German Center for Neurodegenerative Diseases (DZNE), site Rostock/Greifswald, Ellernholzstrasse 1-2, D-17489, Greifswald, Germany.
³ Institute for Community Medicine, Section Epidemiology of Health Care and Community Health, University Medicine Greifswald (UMG), Ellernholzstrasse 1-2, D-17489, Greifswald, Germany.

Abstract

Background: Various comorbidity indices have been validated for individual health outcomes. However, systematic concurrent comparisons of multiple outcome measures in a single study remain relatively underrepresented but needed for practical decision support.

Aims: To compare the performance of the Charlson Comorbidity Index (CCI) and Rx-Risk Comorbidity Index (Rx-Risk).

Methods: Baseline and six-month follow-up data from n = 221 patients recruited in n = 70 practices were used. CCI and Rx-Risk scores were calculated using documented diagnoses and prescribed medications. Outcomes assessed were health-related quality of life (HRQoL, EQ-5D-5 L), functional impairment (B-ADL), cognitive decline (MMSE), and healthcare utilization (physician visits, hospitalizations). Indices performance was evaluated regarding agreement (Cohens Kappa (k)), known-groups validity (ANOVA, t-test), convergent validity (correlation coefficient (r_s)) and predictive ability (R², Akaike information criterion (AIC)).

Results: Patients were, on average, 80 years old, mostly female (55%), with 12 diagnoses and seven medications. Agreement between both indices was poor for all conditions except diabetes (k = 0.645) and chronic airway diseases (k = 0.486). Rx-Risk differed more in known groups, especially for HRQoL and hospitalizations, and showed stronger correlations with the EQ-5D index (r_s, -0.215 vs. -0.134) and risk of hospitalization (r_s, 0.145 vs. 0.128) than CCI. Rx-Risk, again, performed better in predicting the change of EQ-5D index (R², 30 vs. 28%) and all EQ-5D dimensions, functional (R², 55 vs. 52%) and cognitive decline (R², 47 vs. 46%) and physician consultations (AIC, 649.2 vs. 651.0), except for hospitalization (AIC, 149.2 vs. 147.1).

Conclusions: Rx-Risk demonstrated slightly superior validity and predictive ability for HRQoL and healthcare utilization, making it a promising option for studies focused on these outcomes. However, limitations regarding functional and cognitive impairment suggest alternative instruments are needed.

Keywords: Comorbidity indices; Multimorbidity; Patient-reported outcomes; Predictive models; Risk adjustment.

Publication types

Comparative Study

MeSH terms

Aged
Aged, 80 and over
Comorbidity*
Female
Hospitalization
Humans
Male
Quality of Life
Reproducibility of Results