Predicting the Risk of Stillbirth for Small Infants Using Maternal and Pregnancy Characteristics

The risk of stillbirth increases as birthweight falls below the 25th centile. To mitigate this risk, many infants are delivered at late preterm or early term gestations, which increases the risk of neonatal and longer-term complications.

To develop a model to predict the risk of stillbirth for small infants after 34⁺⁰ weeks using information available antenatally.

This was a retrospective cohort study of non-anomalous singleton infants ≥ 34⁺⁰ weeks of gestation with birthweight (BW) < 25th centile, born between 2000 and 2021 in Queensland, Australia. The study outcome was antepartum stillbirth. We used survival analysis to model the number of gestational weeks each pregnancy is at risk of antepartum stillbirth. Competing risks regression models were then built to account for informative censoring due to planned birth. For ease of clinical translatability, we developed a clinical scoring rule to present the probabilities of stillbirth at each gestational week. The dataset was split into testing and discovery cohorts for assessment of internal validation, model discrimination, and concordance.

There were 259,378 infants with BW < 25th centile, including 646 stillbirths (0.25%). The strongest predictors were BW centile < 1 (sub-hazard ratio [SHR] 5.84, 95% confidence interval [CI] 4.60, 7.42), BW centiles 1-2 (SHR 3.10, 95% CI 2.46, 3.91), < 5 antenatal visits (SHR 2.91, 95% CI 2.39, 3.55), BW centiles 3-4 (SHR 2.23, 95% CI 1.72, 2.89) and BMI ≥ 35 kg/m² (SHR 2.24, 95% CI 1.52, 3.29). Other predictors in the model were hypertension, anaemia, and asthma. Across all point scores, the cumulative incidence of stillbirth increased with gestation. Harrell's C-statistics were consistent across the discovery (0.71) and testing cohorts (0.73).

Our prediction model for small infants may be useful to risk-stratify women according to their stillbirth risk and support decisions around the timing of birth.