Objective: There is no consensus on which comorbidity index is optimal for rheumatic health outcomes research. We compared a new Rheumatic Disease Comorbidity Index (RDCI) with the Charlson-Deyo Index (CDI), Functional Comorbidity Index (FCI), Elixhauser Total Score (ETS), Elixhauser Point System (EPS), and a simple comorbidity count (COUNT) using a US cohort of rheumatoid arthritis (RA) patients.
Methods: Using administrative diagnostic codes and patient self-reporting, we tested predictive values of the RDCI, CDI, FCI, ETS, EPS, and COUNT for 2 outcomes: all-cause mortality and physical functioning. Indices were compared using 3 models: bare (consisting of age, sex, and race), administrative (bare plus visit frequency, body mass index, and treatments), and clinic (administrative plus erythrocyte sedimentation rate, nodules, rheumatoid factor positivity, and patient activity scale).
Results: The ETS and RDCI best predicted death, with FCI performing the worst. The FCI best predicted function, with ETS and RDCI performing nearly as well. CDI predicted function poorly. The order of indices remained relatively unchanged in the different models, though the magnitude of improvement in Akaike's information criterion decreased in the administrative and clinic models.
Conclusion: The RDCI and ETS are excellent indices as a means of accounting for comorbid illness when the RA-related outcomes of death and physical functioning are studied using administrative data. The RDCI is a versatile index and appears to perform well with self-report data as well as administrative data. Further studies are warranted to compare these indices using other outcomes in diverse study populations.
© 2015, American College of Rheumatology.