Background: Stillbirth remains a major public-health concern in South Africa. Evidence from routine hospital records that jointly examines maternal, fetal, placental, umbilical-cord and care-related characteristics is limited.
Methods: We conducted a retrospective comparative study using routinely collected maternity records from selected Gauteng hospitals between June 2023 and 2024. The analytical dataset comprised 18,672 birth records: 512 stillbirths and 18,160 live births. The outcome was coded as live birth (1) and stillbirth (0). We used descriptive statistics, group-difference tests and progressively adjusted non-linear logistic-regression models. Free-text documented causes among stillbirths were re-coded into hypertensive/placental disorders, fetal growth restriction or hypoxia, infection, congenital abnormality, maternal medical/metabolic disorder, obstetric/labour complication, unexplained, and excluded/missing categories.
Results: Stillbirths were associated with lower fetal weight and earlier gestation than live births. In adjusted models, hypertension, obesity, diabetes, and selected fetal, placental and umbilical-cord characteristics were associated with lower odds of live birth. Fetal weight, gestational age and maternal age showed non-linear associations with the odds of live birth. Hypertensive and placental disorders were commonly documented among stillbirths and frequently co-occurred with fetal growth restriction, hypoxia, infection and maternal medical conditions. Infection was prominent in descriptive cause classifications but was not independently associated with the outcome in the final adjusted model.
Conclusion: In these selected Gauteng hospitals, stillbirth was associated with interacting maternal, fetal, placental, infectious and care-related characteristics. The findings support strengthened antenatal booking, fetal-growth surveillance, management of hypertension, diabetes and infection, and timely obstetric response to suspected placental or cord complications. Owing to the retrospective design and use of routine clinical records, the findings represent associations and should not be interpreted as causal effects.