Early Pregnancy Systolic Blood Pressure Patterns Predict Early- and Later-Onset Preeclampsia and Gestational Hypertension Among Ostensibly Low-to-Moderate Risk Groups

Abstract

Background Clinical risk factors, a single blood pressure (BP) measurement, current biomarkers, and biophysical parameters can effectively identify risk of early-onset preeclampsia but have limited ability to predict later-onset preeclampsia and gestational hypertension. Clinical BP patterns hold promise to improve early risk stratification for hypertensive disorders of pregnancy. Methods and Results After excluding preexisting hypertension, heart, kidney, or liver disease, or prior preeclampsia, the retrospective cohort (n=249 892) all had systolic BP <140 mm Hg and diastolic BP <90 mm Hg or a single BP elevation ≤20 weeks' gestation, prenatal care at <14 weeks' gestation, and a still or live birth delivery at Kaiser Permanente Northern California hospitals (2009-2019). The sample was randomly split into development (N=174 925; 70%) and validation (n=74 967; 30%) data sets. Predictive performance of multinomial logistic regression models for early-onset (<34 weeks) preeclampsia, later-onset (≥34 weeks) preeclampsia, and gestational hypertension was evaluated in the validation data set. There were 1008 (0.4%), 10 766 (4.3%), and 11 514 (4.6%) patients with early-onset preeclampsia, later-onset preeclampsia, and gestation hypertension, respectively. Models with 6 systolic BP trajectory groups (0-20 weeks' gestation) plus standard clinical risk factors performed substantially better than risk factors alone to predict early- and later-onset preeclampsia and gestational hypertension, with C-statistics (95% CIs) of 0.747 (0.720-0.775), 0.730 (0.722-0.739), and 0.768 (0.761-0.776) versus 0.688 (0.659-0.717), 0.695 (0.686-0.704) and 0.692 (0.683-0.701), respectively, with excellent calibration (Hosmer-Lemeshow P=0.99, 0.99, and 0.74, respectively). Conclusions Early pregnancy BP patterns up to 20 weeks' gestation plus clinical, social, and behavioral factors more accurately discriminate hypertensive disorders of pregnancy risk among low-to-moderate risk pregnancies. Early pregnancy BP trajectories improve risk stratification to reveal higher-risk individuals hidden within ostensibly low-to-moderate risk groups and lower-risk individuals considered at higher risk by US Preventive Services Task Force criteria.

Keywords: blood pressure; hypertensive disorders; longitudinal trajectory analysis; prediction; preeclampsia; pregnancy; risk stratification.

Figure 1. Six early pregnancy systolic blood pressure trajectory (BPT) groups from 0 to 20 weeks' gestation (from initial to last average blood pressure measurement in mm Hg for each BPT group).

Six early pregnancy BPT groups: ultra‐low declining (light blue), low declining (green), moderate‐fast decline (red), low increasing (yellow), moderate stable (black), and elevated stable (purple). Adapted from Gunderson et al with permission. Copyright ©2022 Wolters Kluwer Health, Inc.

Figure 2. Number and percentages of early‐onset preeclampsia (PE), later‐onset PE, and gestational hypertension (GH) among systolic blood pressure trajectory (BPT) groups for the development and validation data sets.

Color gradient definitions: darkest shade=early‐onset PE, medium shade dots=later‐onset PE, and lightest shade vertical dashes=GH. HDP indicates hypertensive disorders of pregnancy.

Figure 3. Early pregnancy systolic blood pressure trajectory (BPT) groups stratified by the number of low‐to‐moderate risk factors based on US Preventive Services Task Force: model average predicted probabilities (percentages) and observed incidence rate of preeclampsia and gestational hypertension (percentages and 95% CIs) for the internal validation sample.

Figure 4. Early pregnancy blood pressure trajectory (BPT) groups stratified by racial and ethnic groups with or without moderate risk factors: model average predicted probabilities (percentages) and observed incidence rate of preeclampsia and gestational hypertension (percentages and 95% CIs) for the internal validation sample.