Comparative performance of comorbidity indices in discriminating health-related behaviors and outcomes

Huang Tz Ou, Bhramar Mukherjee, Steven R. Erickson, John D. Piette, Richard P. Bagozzi, Rajesh Balkrishnan

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

BACKGROUND AND OBJECTIVE: Although the predictive ability of the Charlson Index, Elixhauser Index (EI), Chronic Disease Score (CDS), and Health-related Quality of Life Comorbidity Index (HRQL-CI) for health care outcomes has been assessed individually, little research has compared the discriminative performance of these indices directly in a single study. The current study compared these indices in discriminating among type 2 diabetes patients varying in demographics and health care outcomes characteristics. STUDY DESIGN: There were 9832 Medicaid patients with type 2 diabetes from 8 states evaluated. Endpoints included demographics (age, race), health care behaviors (physician's diabetes care standard adherence, patient's medication adherence), and health care utilization and expenditures. Discriminative power of comorbidity indices was determined by c-statistics from logistic regression, the shape of receiver operator characteristic curve, and area under the curve. RESULTS: The CDS demonstrated the best ability in discriminating between age subgroups (c = 0.61) and patients who were or were not adherent to their medication (c = 0.56). The CDS and HRQL-CI mental index performed similarly in discriminating based on diabetes care standard adherence (c = 0.60). The EI had the best discrimination for health care utilization and costs, while HRQL-CI physical index performed similarly to EI in predicting hospitalization admission (c = 0.62), and the HRQL-CI mental index performed similarly to the EI in predicting outpatient visits (c = 0.74). CONCLUSIONS: The CDS was found to be the best metric for differentiating among patients varying in demographics, physician's diabetes care standard adherence, and patient's medication adherence, while the EI should be the first choice to identify patients at risk of high medical resource use.

Original languageEnglish
Pages (from-to)e91-e104
JournalHealth Outcomes Research in Medicine
Volume2
Issue number2
DOIs
Publication statusPublished - 2011 May

All Science Journal Classification (ASJC) codes

  • Health Policy

Fingerprint Dive into the research topics of 'Comparative performance of comorbidity indices in discriminating health-related behaviors and outcomes'. Together they form a unique fingerprint.

  • Cite this