Administrative data outperformed single-day chart review for comorbidity measure

Area of Science:

• Healthcare research
• Epidemiology
• Health services research

Background:

• The Charlson comorbidity index is a widely used tool for measuring disease burden.
• Accurate comorbidity assessment is crucial for predicting patient outcomes and evaluating healthcare quality.
• Traditional methods like chart review can be resource-intensive.

Purpose of the Study:

• To compare the predictive accuracy of the Charlson comorbidity index derived from administrative data versus single-day chart review.
• To evaluate the performance of each method in predicting inpatient mortality and nosocomial infections.

Main Methods:

• Cross-sectional study involving 890 adult patients in six Swiss hospitals.
• Charlson comorbidity index calculated from both administrative data (ICD-10 codes) and a single-day chart review.
• Logistic regression used to compare prediction of hospital mortality and nosocomial infection.

Main Results:

• Administrative data showed higher prevalence for 11 out of 17 comorbidities compared to chart review.
• Charlson index from administrative data demonstrated superior predictive performance for hospital mortality (C=0.863 vs. 0.795) and nosocomial infection (C=0.645 vs. 0.614).
• Agreement between the two methods varied, with Kappa values ranging from -0.001 to 0.56.

Conclusions:

• The Charlson comorbidity index derived from administrative data is superior to that from single-day chart review for predicting patient outcomes.
• Administrative data offers a more robust and potentially efficient method for comorbidity assessment in healthcare evaluations.
• Recommendations favor using administrative data for evaluating nosocomial infections and patient comorbidities.