Abstract 144: Development of a Risk Score for Predicting Hypertension in Women with a History of Gestational Diabetes Mellitus

Abstract

Introduction: Gestational diabetes mellitus (GDM) is associated with increased maternal cardiovascular disease (CVD) risk manifest as hypertension subsequent to pregnancy. The impacts of dysglycemia and race/ethnicity on hypertension risk have been examined; however, the study of putative risk factors has been limited. Hypothesis: We hypothesize that GDM and gestational dysglycemia are sufficiently predictive of hypertension subsequent to pregnancy and that a risk score based on degree of dysglycemia can be used to identify at-risk patients. Methods: We examined a cohort of 23,223 women who presented for prenatal care to the Massachusetts General Hospital Obstetrical Department between September 1998 and January 2007. We selected women, ages 18 to 40 years, free of CVD or pre-GDM, with complete glycemic and demographic data who delivered a singleton, live birth. Each woman’s initial pregnancy was selected in order to not violate the independence assumption. The population was stratified by incident hypertension subsequent to pregnancy. Logistic regression models (with and without stepwise selection) were used to identify significant maternal risk factors based on their predictive performance. Finally, the risk prediction score was developed from β-coefficients in multivariable Weibull models. Results: The study sample comprised of 11,161 women, 547 (4.9%) of whom developed hypertension during 4.2 years (median) follow-up. Significant predictors of hypertension included age, body mass index, systolic and diastolic blood pressures, parity, dysglycemia, and race/ethnicity. These factors were used to develop a risk score for subsequent hypertension risk. The final risk score achieved a c-statistic of 0.76, indicating moderate discriminative ability (Figure 1). Conclusions: We developed a risk score that may be used to identify patients with an elevated risk of hypertension subsequent to a pregnancy complicated by dysglycemia. Further external validation is warranted to determine clinical effectiveness.