Development and Validation of Multivariable Risk Prediction Models for Adverse Maternal and Neonatal Outcomes in Severe Preeclampsia: Findings from a Cross-Sectional Study in a Low-Resource Setting at Mpilo Central Hospital, Bulawayo, Zimbabwe

Abstract

Background: Hypertensive disorders of pregnancy are major causes of global maternal, fetal and neonatal morbidity and mortality. This research aimed to develop and validate multivariable prediction models for composite adverse maternal outcomes and neonatal outcomes in severe preeclampsia for use in low-resource settings. Methods: This was a retrospective cross-sectional study. Predictor variables with a p<0·2 from the univariable logistic regression were entered into a multivariable stepwise backward elimination logistic regression model to predict composite adverse maternal and neonatal outcomes. Models performance were assessed using calibration slopes and the area under receiver operating characteristic curve (AU ROC). Internal validation was performed using bootstrap-based methods. External validation was done with the Preeclampsia Integrated Estimate of RiSk (miniPIERS) dataset for the maternal model. Findings: 549 women had severe preeclampsia during the study period from whom 567 neonates were born. The model for composite adverse maternal outcomes included maternal age, gestational age on admission, epigastric pain, vaginal bleeding with pain, haemoglobin and platelets; with good discriminatory ability, the AU ROC was 0·796 (95% CI 0·758-0·833). The model for composite adverse neonatal outcomes included gestational age, maternal platelets, maternal alanine transaminase and birth weight; with good discriminatory ability, the AU ROC was 0·902 (95% CI 0·876-0·927). There was minimal overfitting in both models. Interpretation: The final models accurately predicted composite adverse maternal and neonatal outcomes in the study population. Such parsimonious models have the potential to prevent poor outcomes in low-resource settings, especially in rural areas where early recognition and referral could be crucially life-saving. Funding Statement: None - AEPH is supported by the National institute for Health Research (NIHR) Global Health Research Group for the Prevention of Stillbirth in Sub-Saharan Africa. Declaration of Interests: The authors declare that they have no conflicts of interest. Ethics Approval Statement: The Ethics Committee at Mpilo Central Hospital gave a waiver for all retrospective and non-intervention studies to go ahead in the institution in 2016 as long as the data remained anonymous. The institutional Ethics Committee approved the study.