Construction of a New Predictive Model and Scoring System for Hypertensive Disorder Complicating Pregnancy

Abstract

Objective: To construct a new predictive model and scoring system for hypertensive disorder complicating pregnancy by using a combination of simple clinical variables that can easily be obtained at antenatal visit, and evaluate its clinical diagnostic value. Study design: Included in this study were 2,466 nulliparous pregnant women at the gestational age of 16~20 weeks who came to our hospital for the first antenatal visit between 2011 and 2013. They were classified as a derivation cohort (n=1,421) and a validation cohort (n=1,045). A predictive model was developed from a logistic regression model, scoring system was developed by using the regression coefficients obtained from the new model and then internally and externally validated. Results: The predictive model comprised 8 variables: body mass index (BMI), mean arterial pressure (MAP), drinking history, gestational diabetes, occupational physical activity, family history of hypertension, platelet count (PLT) and uric acid (UA), with a scoring system ranging from 0 to 13. The risk of HDCP in participants with low-risk (≤ 4 scores) or high-risk (>4 scores) in the validation cohort was 4.5% and 24.4% respectively (P<0.001), indicating that the predictive model had good discrimination (area under the receiver operating characteristic curve = 0.783), 95% confidence interval (0.746, 0.820) and calibration (P=0.745). The sensitivity and specificity were 76.6% and 67.7% respectively. Conclusion: We have developed and validated a new predictive model comprising 8 variables: BMI, MAP, drinking history, gestational diabetes, occupational physical activity, family history of hypertension, PLT and UA. Due to cost/ effectiveness and cost/utility, the new model can be applied to clinical screening of HDCP.