Developing an adapted Charlson comorbidity index for ischemic stroke outcome studies

Area of Science:

• Neurology
• Epidemiology
• Health Services Research

Background:

• The Charlson Comorbidity Index (CCI) is widely used for adjusting patient casemix in research.
• Its relevance in specific patient cohorts, such as ischemic stroke (IS), requires reevaluation.

Purpose of the Study:

• To reevaluate the Charlson Comorbidity Index (CCI) in an ischemic stroke (IS) cohort.
• To determine the relevance of the original 17 comorbidities and their assigned weights for IS patients.

Main Methods:

• An IS cohort of 6988 patients was identified from the Ontario Stroke Registry (OSR).
• Hospital discharge data (ICD-10-CA) was used to identify Charlson comorbidities.
• A multivariable Cox model predicted one-year mortality, retaining significant comorbidities (P < 0.05) with hazard ratios ≥1.2 to generate revised weights for the ISCCI.

Main Results:

• Ten of the original 17 comorbidities were retained in the ISCCI model, with 7 having reassigned weights.
• The ISCCI demonstrated a small but significant increase in the c-statistic for 30-day mortality (0.746 vs. 0.732, p=0.009) compared to the CCI.
• No significant improvement in c-statistic for in-hospital or one-year mortality, nor in the continuous Net Reclassification Index (cNRI), was observed.

Conclusions:

• The ISCCI model exhibited similar overall performance to the original CCI model in predicting mortality.
• The primary advantage of the ISCCI is its reduced complexity, incorporating seven fewer comorbidities.
• This simplification enhances its practical implementation, especially where coded data availability is limited.