Comorbidity scores for administrative data benefited from adaptation to local coding and diagnostic practices

Area of Science:

• Health Services Research
• Medical Informatics
• Epidemiology

Background:

• Comorbidity indices are crucial for risk adjustment in healthcare using administrative data.
• The Charlson and Elixhauser indices are widely adopted for predicting patient outcomes.
• Evaluating and refining these indices is essential for accurate risk prediction models.

Purpose of the Study:

• To compare the predictive performance of the original Charlson index, a modified Charlson index, and a modified Elixhauser index.
• To assess these indices in predicting in-hospital mortality using administrative data from English public hospitals.
• To evaluate the impact of code modifications and England-specific recalibration on index performance.

Main Methods:

• Logistic regression analysis was employed.
• Two years of administrative hospital data from all acute nonspecialist public hospitals in England were utilized.
• Comparisons included the original Charlson index, Charlson index with England-calibrated weights, modified Charlson codes, and a modified Elixhauser index.

Main Results:

• The modified Elixhauser index demonstrated the best predictive performance (c=0.80).
• Modified Charlson codes significantly improved discrimination (c=0.76) and model fit compared to the original version.
• England-calibrated weights for the original Charlson index offered modest improvements in model fit but similar discrimination (c=0.73).

Conclusions:

• The modified Elixhauser index is superior for predicting in-hospital mortality in the studied English population.
• Modifications to Charlson codes and weights enhance its utility for English administrative data.
• While effective, the modified Elixhauser index's weight consistency across diverse patient groups warrants further investigation.