Related Experiment Videos

Searching for an improved clinical comorbidity index for use with ICD-9-CM administrative data

W A Ghali1, R E Hall, A K Rosen

  • 1Health Care Research Unit, Boston University Medical Center, MA 02118, USA.

Insights

Developing a new comorbidity index using study-specific weights improved mortality prediction in cardiac surgery patients. This enhanced risk adjustment offers better insights than standard Charlson comorbidity index adaptations.

Area of Science:

  • Health Services Research
  • Medical Informatics
  • Cardiovascular Surgery

Background:

  • Accurate comorbidity risk adjustment is crucial for evaluating patient outcomes in healthcare.
  • The Charlson comorbidity index is a widely used tool, but its adaptations may vary in effectiveness.
  • Coronary artery bypass surgery (CABS) patient data from Massachusetts were used to assess comorbidity measures.

Purpose of the Study:

  • To compare the performance of two International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) adaptations of the Charlson comorbidity index.
  • To develop and validate a novel comorbidity index using study-specific weights for improved mortality prediction.
  • To assess the impact of different comorbidity scoring methods on predicting mortality in CABS patients.

Main Methods:

  • Two established ICD-9-CM adaptations of the Charlson comorbidity index (Deyo and Dartmouth-Manitoba) were applied to CABS patient data.
  • A new comorbidity index was created by assigning unique weights to the original Charlson comorbidity variables based on study data.
  • Statistical analysis, including kappa statistics and model performance metrics (c-statistic), was used for comparison.

Main Results:

  • The two ICD-9-CM adaptations of the Charlson index demonstrated high agreement (90% identical scores) and consistency in identifying comorbidities (kappa 0.72-1.0).
  • The novel, study-specific weighted index identified a 10% high-risk patient group with 15% mortality, outperforming the standard Charlson index (5% group, 8% mortality; p=0.01).
  • The model incorporating the new index showed superior predictive performance (c=0.74) compared to the model using the original Charlson index (c=0.70).

Conclusions:

  • The choice of ICD-9-CM adaptation for the Charlson comorbidity index had minimal impact on scoring in this CABS cohort.
  • Utilizing study-specific weights with Charlson comorbidity variables significantly enhances the predictive power for mortality.
  • This customized approach offers a more accurate risk adjustment strategy for patient populations undergoing coronary artery bypass surgery.

Related Concept Videos