Abstract / Summary
Abstract Background The postnatal period accounts for up to one-third of maternal deaths and half of newborn deaths globally. Early intervention can prevent progression along the continuum of illness severity. The aim of this study was to develop prognostic models to predict post-discharge outcomes in women and newborns following hospital delivery in Uganda. Methods This prospective observational cohort study enrolled women presenting for delivery at two regional referral hospitals. The maternal outcome was readmission due to obstetric infection within the six-week postnatal period; newborn outcomes were six-week mortality and readmission resulting from acute illness likely due to an infectious cause. Parsimonious models were derived for each outcome using the top eight variables based on variable importance of available predictors and internally validated using 10-fold cross-validation. Results Among 6,996 women discharged postpartum with six-week follow-up data: none died; 224 (3.2%) were readmitted, with 126 (1.8%) due to infection. The maternal model’s mean (range) cross-validated area under the receiver operating characteristic curve (AUROC) was 0.784 (0.738 to 0.824) and the area under the precision-recall curve (PRAUC) was 0.099 (0.053 to 0.263); at an 80% sensitivity threshold, it had mean specificity 0.697 and Brier score 0.017. Among 6,044 newborns discharged alive: 35 (0.6%) died; 228 (3.8%) were readmitted. The newborn mortality model’s AUROC was 0.731 (0.600 to 0.854) and PRAUC was 0.029 (0.010 to 0.071); at 80% sensitivity, it had mean specificity 0.539 and Brier score 0.006. The newborn readmission model’s mean (95% confidence interval) bootstrapped AUROC was 0.712 (0.672 to 0.747) and PRAUC was 0.080 (0.065 to 0.109); at an 80% sensitivity threshold, it had mean specificity 0.532 and Brier score 0.036. Conclusions Maternal and newborn morbidity remains high in Uganda. Routine parameters used in clinical care at discharge can identify mothers and newborns at risk of serious infection or death and inform risk-based postnatal care. Trial registration This observational study was retrospectively registered at clinicaltrials.gov (NCT05730387, 15-Feb-2023).