Performance of a population-based cardiac risk stratification tool in Asian patients with chest pain.

OBJECTIVES: Most contemporary cardiac risk stratification tools have been derived and validated in mixed-race populations. Their validity in single-race populations has not been tested. The authors sought to compare the performance of a risk stratification tool between a mixed-race U.S. patient population and an Asian patient population. METHODS: This study is an analysis of data from the Internet Tracking Registry for Acute Coronary Syndromes (i(*)trACS) registry of patients with chest pain presenting to the emergency departments of eight U.S. centers and one site in Singapore. The Acute Cardiac Ischemia Time-Insensitive Predictive Instrument (ACI-TIPI) was computed for included patients, and its performance in predicting acute coronary syndrome (ACS) was compared between patients from the United States and Singapore. RESULTS: Of the 11,991 included patients, 1,120 experienced ACS. Although the ACI-TIPI demonstrated similar accuracy among groups (area under the curve, 0.729 [U.S.] vs. 0.719 [Singapore]; p = 0.5611), sensitivity and specificity were different when equal ACI-TIPI thresholds were considered. Recreating the logistic regression models used to create the ACI-TIPI showed similar results between the derived parameters and the parameters estimated for the U.S. group. In contrast, age older than 50 years (log-odds ratio [LOR], 0.107; 95% confidence interval [CI] = 0.518 to 0.713), male gender (LOR, 0.487; 95% CI = 0.149 to 1.122), and chest pain as a primary complaint (LOR, 0.237; 95% CI = 0.139 to 0.613) had little predictive power in patients from Singapore. CONCLUSIONS: Differences exist in presentation and factors associated with ACS among patients from the United States and Singapore that may affect the performance of risk stratification tools. These findings suggest that cardiac clinical decision rules need international validation.

Duke Scholars

Author Christopher Lindsell Biostatistics & Bioinformatics, Division of Biostatistics