Related Experiment Videos

Practical considerations on the use of the Charlson comorbidity index with administrative data bases

W D'Hoore1, A Bouckaert, C Tilquin

  • 1Université de Montréal, Départment d'Administration de la Santé, Equipe de Recherche Opérationnelle en Santé, Montréal, Canada.

Insights

The Charlson comorbidity index, adapted for ICD-9 codes, effectively predicts inpatient death in ischemic heart disease patients. This validated index aids risk adjustment in administrative databases.

Area of Science:

  • Health Services Research
  • Epidemiology
  • Biostatistics

Background:

  • Assessing the burden of comorbid disease is crucial for patient outcomes.
  • Administrative databases offer valuable data for health services research.
  • Existing comorbidity measures may require adaptation for specific data sources.

Purpose of the Study:

  • To adapt the Charlson comorbidity index for use with International Classification of Disease (ICD-9) codes.
  • To evaluate the index's ability to predict inpatient death in patients with ischemic heart disease.
  • To assess the index's utility for risk adjustment in administrative data.

Main Methods:

  • Adapted the Charlson index using ICD-9 codes from the MED-ECHO database.
  • Applied multiple logistic regression to analyze predictors of inpatient death.
  • Examined various transformations of the comorbidity score for predictive accuracy.
  • Validated the index on independent MED-ECHO data from 1990-1991.

Main Results:

  • The adapted comorbidity index was strongly and consistently associated with inpatient death.
  • Transforming the index into four dummy variables yielded the best predictive accuracy (AUC = 0.87).
  • Validation analysis confirmed the index's statistical properties in a separate cohort.

Conclusions:

  • The adapted Charlson index is an efficient tool for risk adjustment using administrative data.
  • The index demonstrates strong predictive value for inpatient mortality in ischemic heart disease.
  • Further testing on other conditions is recommended to broaden its applicability.

Related Concept Videos