Prevalence, risk factors and early neonatal outcomes of preterm birth at four selected referral hospitals in Kigali, Rwanda: a cross-sectional analysis of delivery register data

Preterm birth delivery before 37 completed weeks of gestation is a leading cause of neonatal mortality and long-term morbidity worldwide. Rwandan evidence comes mainly from community cohorts and single district hospitals, and the burden, determinants and immediate neonatal consequences of preterm birth in the referral hospitals that manage the country’s most complicated pregnancies have not been characterised. We determined the prevalence, risk factors and early neonatal outcomes of preterm birth at four selected referral hospitals in Kigali, and distinguished spontaneous from provider-initiated preterm birth. We conducted a cross-sectional analysis of routine delivery-register data from four selected referral hospitals in Kigali (Kibagabaga, La Croix du Sud, Masaka and Muhima), covering all deliveries recorded between November 2023 and March 2024. Preterm birth was defined as gestational age < 37 completed weeks, recorded categorically in the register (< 34, 35–37, 38–40 and > 40 weeks). We hypothesised that, in a referral population, acute obstetric complications rather than the nutritional and infectious exposures reported from Rwandan community cohorts would be the dominant correlates of preterm birth. Preterm births were classified as spontaneous (spontaneous labour onset) or provider-initiated (induction or pre-labour caesarean section), and each phenotype was modelled separately against term deliveries. Factors independently associated with preterm birth were identified using multivariable logistic regression with hospital fixed effects. Early neonatal outcomes were compared between preterm and term neonates and reported with 95% confidence intervals (CI). Of 5,520 deliveries, 623 (11.3%, 95% CI 10.5–12.1) were preterm: 121 (2.2%) early preterm (< 34 weeks) and 502 (9.1%) late preterm (35–37 weeks). Preterm births were predominantly spontaneous (488; 78.3%) rather than provider-initiated (135; 21.7%), and the two phenotypes had distinct correlate profiles. Spontaneous preterm birth was associated with antenatal hospitalisation (adjusted odds ratio [aOR] 12.85, 95% CI 7.03–23.50), antepartum haemorrhage (aOR 2.86, 1.22–6.67) and lower body-mass index (aOR 0.93 per kg/m², 0.90–0.96), whereas provider-initiated preterm birth was associated with multiple pregnancy (aOR 6.90, 2.11–22.55), pre-eclampsia (aOR 5.38, 2.37–12.24), antepartum haemorrhage (aOR 3.57, 1.00–12.75) and older maternal age (aOR 1.05 per year, 1.03–1.08). Preterm neonates had higher rates of NICU admission (23.8%, 95% CI 20.6–27.3 versus 3.3%, 2.8–3.8), resuscitation (10.6%, 8.4–13.3 versus 3.3%, 2.9–3.9), low birth weight (41.1%, 37.0–45.3 versus 3.5%, 3.0–4.1) and in-hospital death (7.7%, 5.9–10.1 versus 0.9%, 0.7–1.2; all p < 0.001). In-hospital mortality reached 28.1% (20.9–36.7) among the 121 neonates born before 34 weeks. Preterm birth affected more than one in ten deliveries at these four referral hospitals and carried a heavy early-neonatal mortality burden. Separating the two phenotypes shows that pre-eclampsia and multiple pregnancy operate through provider-initiated delivery, while antenatal hospitalisation marks the severity of pregnancies that end in spontaneous preterm labour; the pooled estimate obscures this distinction. Because the design is cross-sectional and the setting is referral-based, these are risk markers for antenatal stratification rather than demonstrated causes, and the prevalence should not be read as a population estimate for Kigali or for Rwanda.