Predicting In-Hospital Mortality in Patients With Acute Coronary Syndrome in China

Currently available risk scores (RSs) were derived from populations with very few participants from China. We aimed to develop an RS based on data from patients with acute coronary syndrome in China and to compare its performance with the commonly promoted Global Registry of Acute Coronary Events (GRACE) RS. Clinical Pathways for Acute Coronary Syndromes—Phase 2 was a trial of a quality improvement intervention in China. Patients recruited from 75 hospitals from October 2007 to August 2010 were divided into training and validation sets based on immediate or delayed implementation. A Clinical Pathways for Acute Coronary Syndromes (CPACS) RS for in-hospital mortality was developed separately by gender, using the training set (6,790 patients). Discrimination and calibration of the CPACS RS and GRACE RS were compared on the validation set (3,801 patients). Although discrimination of the GRACE RS was acceptable, this was improved with the CPACS RS (c-statistic 0.82 vs 0.87, p = 0.012 for men; c-statistic 0.78 vs 0.85, p = 0.006 for women). The absolute bias was significantly lower with CPACS RS for both genders (7.6% vs 97.5% in men and 21.5% vs 77.2% in women), compared with the GRACE RS, which systematically overestimated risk. The CPACS RS underestimated risk in women, but only in those already above threshold levels currently used to define a clinical high-risk population. In conclusion, the GRACE RS substantially overestimates the risk of in-hospital death in patients presenting to the hospital with a suspected acute coronary syndrome in China. We have developed and independently validated a new RS utilizing data from Chinese patients.