Related Experiment Videos
Evaluation of two competing methods for calculating Charlson's comorbidity index when analyzing short-term mortality
M A Cleves1, N Sanchez, M Draheim
1Department of Epidemiology and Biostatistics. Rammelkamp Center for Education and Research, Case West Reserve University, Cleveland, Ohio 44109, USA.
Journal of Clinical Epidemiology
|August 1, 1997
Summary
The Deyo-Charlson and Romano-Charlson comorbidity indices showed poor predictive power for short-term mortality after hospitalization. Empirical weights slightly improved performance, but overall applicability remains limited for these common comorbidity adjustment tools.
Area of Science:
- Medical Informatics
- Health Services Research
- Epidemiology
Background:
- Comorbidity indices like Deyo-Charlson and Romano-Charlson are vital for adjusting administrative data in comparative studies.
- Accurate adjustment is crucial for reliable comparative effectiveness research and health outcomes analysis.
Purpose of the Study:
- To compare the performance and predictive power of Deyo-Charlson and Romano-Charlson comorbidity indices.
- To evaluate their effectiveness in predicting short-term mortality following hospitalization.
Main Methods:
- Logistic regression models were used to assess the explanatory power of the indices.
- Medicare claims data from six medical categories were analyzed for 30, 90, and 180-day mortality.
- Likelihood ratio chi-square (G2) and ROC curve analysis (C statistic) were employed to evaluate index contribution and predictive power.
Main Results:
- Both Deyo-Charlson and Romano-Charlson indices demonstrated poor predictive power for short-term mortality (C statistics: 0.60-0.78).
- The indices offered equal improvement to the base model.
- Empirically derived weights showed a slight increase in predictive power compared to standard weights.
Conclusions:
- The Deyo-Charlson and Romano-Charlson comorbidity indices have limited applicability for predicting short-term mortality.
- Further research may be needed to develop more accurate comorbidity adjustment methods for administrative data analysis.