The Addition of the Charlson Comorbidity Index to the GRACE Risk Prediction Index Improves Prediction of Outcomes in Acute Coronary Syndrome

Abstract

Patients with cardiovascular disease have increased risk of poor outcomes when coexisting illnesses are present. Clinicians, administrators, and health services researchers utilize risk adjustment indices to stratify patients for various outcomes. The GRACE Risk Prediction Index (GRPI) was developed to risk stratify patients who experienced an acute coronary syndrome (ACS) event. GRPI does not account for the presence of comorbid conditions. The objective of this study was to compare the ability of the GRPI and the Charlson Comorbidity Index (CCI), used independently or combined, to predict mortality or secondary coronary events in patients admitted for ACS. Data were obtained from an academic health system's ACS registry. Outcomes included inpatient and 6-month postdischarge mortality and occurrence of secondary cardiovascular events or revascularization procedures. Logistic regression derived C statistics for CCI, GRPI, and CCI-GRPI predictive models for each outcome. Likelihood ratio tests determined the contribution of CCI when added to GRPI models. Complete data were available for 1202 patients. The GRPI model had the greatest C statistic when predicting inpatient mortality (0.73); the GRPI-CCI combined model C statistic was 0.81 when predicting death during the follow-up period; and C statistics for all 3 models were similar in predicting secondary events (0.57-0.60). The likelihood ratio analysis demonstrated that adding CCI to GRPI models was beneficial primarily for predicting secondary events. CCI is a useful addition to GRPI when predicting future cardiac-related events or mortality after an ACS event. It is an acceptable alternative to the GRPI model if data to construct GRPI are not available.