Cardiac risk prediction model for pregnant women with structural heart disease in Eastern China.

ABSTRACT

OBJECTIVE: To develop a model for predicting cardiac risk among pregnant women with structural heart disease in Eastern China. METHODS: The model was built using data from pregnant women (development cohort, n=566; validation cohort, n=314) who delivered at Shanghai Obstetrical Cardiology Intensive Care Center, Renji Hospital, Shanghai, between 2002 and 2015. Independent predictors of adverse cardiac events were determined by logistic regression. Discrimination and calibration of the model, termed the Renji score, were compared with the CARPREG score, ZAHARA score, and modified WHO risk classification. RESULTS: There were 87 (15.4%) adverse cardiac events in the development cohort. Independent predictors of adverse cardiac events included left ventricular systolic dysfunction (ejection fraction, <40%), prior cardiac event or arrhythmia, moderate-to-severe pulmonary arterial hypertension (≥50 mm Hg), mechanical valve replacement, moderate-to-severe mitral stenosis. Surgical intervention before pregnancy was protective against cardiac events. As compared with other risk assessment systems, the Renji score performed better in predicting cardiac events, with a concordance statistic of 0.844 (95% confidence interval [CI], 0.800-0.889) for the development cohort and 0.779 (95% CI, 0.684-0.873) for the validation cohort. CONCLUSION: The Renji score was applicable to predicting cardiac events among pregnant women with structural heart disease in Eastern China.