Soleil Research | Evidence Review
Published August 26, 2026 in Nature Medicine
Plain-Language Summary
A large hospital-based trial across Benin, Malawi, Tanzania, and Uganda found that a coordinated maternity-care improvement program was associated with lower odds of a composite outcome combining fresh stillbirth and neonatal death within 24 hours. The fresh-stillbirth outcome alone did not significantly improve. Because ALERT combined co-design, clinical training, quality-improvement work, data feedback, and leadership mentoring, the trial supports the package as a whole but cannot tell us which component produced the benefit.
Why This Study Matters
Many deaths around childbirth occur even when effective clinical practices are already known. The practical challenge is whether hospitals can recognize deterioration, monitor labor reliably, escalate care promptly, and sustain better teamwork under real-world resource constraints. ALERT tested a systems-level implementation strategy rather than a new drug or device.
What the Study Tested
ALERT was a stepped-wedge cluster-randomized trial conducted in 16 high-caseload hospitals, with four hospitals each in Benin, Malawi, Tanzania, and Uganda. Hospitals crossed from comparison to intervention periods according to a randomized sequence over five six-month periods. The intention-to-treat analysis included 134,630 women and 139,300 neonates from July 2021 through December 2023.
The intervention combined four major elements: co-design with women, families, and maternity providers; a three-day competency-based training program; quality improvement using Plan-Do-Study-Act cycles supported by an electronic perinatal registry; and mentoring of maternity-unit leaders.
The co-primary outcomes were fresh stillbirth and early perinatal mortality, defined as the composite of fresh stillbirth plus neonatal death within 24 hours.
What the Study Found
- Early perinatal mortality: the intervention was associated with 22% lower adjusted odds of the composite outcome (adjusted OR 0.78, 95% CI 0.65–0.94; P = .008).
- Fresh stillbirth: the co-primary outcome did not show a statistically significant reduction (adjusted OR 0.88, 95% CI 0.68–1.13; P = .32).
- Hypoxic-ischemic events: a secondary outcome defined using a five-minute Apgar score below 7 was lower during intervention periods (adjusted OR 0.81, 95% CI 0.69–0.95; P = .008).
- Cesarean delivery: the odds increased modestly (adjusted OR 1.14, 95% CI 1.01–1.27; P = .027).
- Experience of care: the investigators did not detect significant improvements in responsiveness or respectful care.
The mortality result should be described precisely: an odds ratio of 0.78 means lower odds of the composite outcome under the study conditions; it is not equivalent to saying that every hospital would experience 22% fewer deaths. The secondary-outcome findings also deserve caution because the trial evaluated multiple outcomes without a formal multiplicity adjustment.
Methodological Quality
Overall assessment: strong pragmatic cluster-randomized evidence for the ALERT package on the 24-hour composite outcome, with important limits on precision, attribution, and generalizability.
Strengths
- Randomized stepped-wedge allocation at the hospital level, appropriate for an intervention delivered to maternity teams rather than individual patients.
- Four-country implementation across 16 hospitals, improving relevance across varied sub-Saharan African hospital settings.
- A clinically important primary composite outcome rather than reliance only on process measures.
- Intention-to-treat analysis with very low missingness for the primary outcome.
- Per-protocol and additional sensitivity analyses that were broadly consistent with the main mortality estimate.
- Prospective trial registration, a published protocol, and independent data and safety monitoring.
Limitations and Cautions
- Only 16 randomized clusters: the very large number of births does not create 134,630 independent randomized units. Statistical precision is ultimately constrained by the number of hospitals.
- Time and stepped-wedge effects: although the analysis adjusted for study period, secular changes in staffing, referral patterns, supplies, or clinical practice can still complicate causal interpretation.
- Heterogeneity: hospital-specific effects varied, with substantial between-site heterogeneity. A pooled estimate should not be interpreted as a uniform effect in every hospital.
- Composite versus component outcome: the composite early perinatal mortality outcome improved, while fresh stillbirth alone did not. These results are not interchangeable.
- Multiplicity: two co-primary outcomes and numerous secondary and process outcomes were evaluated without a formal multiple-testing correction, so secondary findings are best treated as supportive rather than definitive.
- Short neonatal follow-up: neonatal mortality was captured only through the first 24 hours, so the trial does not establish an effect on later neonatal mortality.
- Outcome ascertainment: the authors acknowledge that underreporting of perinatal events cannot be completely excluded in routine hospital systems, although sensitivity analyses were used to test robustness.
- Process-data missingness: some operational measures were incompletely documented, limiting conclusions about exactly how the intervention produced its effect.
- Package-level inference: the study cannot isolate whether co-design, training, quality improvement, registry feedback, leadership mentoring, or their interaction was responsible for the observed outcome.
- Cesarean delivery: the higher cesarean rate may reflect more timely recognition and escalation, as the authors propose, but an increase in surgery should not automatically be classified as beneficial without considering appropriateness, safety, and local capacity.
Funding and Conflicts of Interest
The ALERT project was supported by a European Commission Horizon 2020 implementation-science grant (grant 847824) to Claudia Hanson. Hanson also disclosed support from the Swedish Research Council (2019-01906_VR). The authors state that the funders had no role in study design, data collection or analysis, publication decisions, or manuscript preparation. Open-access funding was provided by Karolinska Institutet. The authors declared no competing interests.
Soleil Insight
In high-burden maternity settings, implementation is itself an intervention. Evidence-based practices can only improve outcomes when clinical systems reliably recognize deterioration and act. ALERT suggests that a coordinated, context-specific package can improve an important early perinatal outcome, but the result should not be reduced to a single training effect or assumed to transfer unchanged to every hospital. The lack of improvement in respectful-care measures is equally instructive: technical quality, mortality reduction, and patient experience are related but distinct dimensions of care.
Original Publication
Hanson C, Akuze J, Annerstedt KS, et al. A multi-faceted hospital-based intervention for intrapartum care in sub-Saharan Africa: a stepped-wedge cluster-randomized trial. Nature Medicine. Published August 26, 2026. DOI: 10.1038/s41591-026-04599-w.
Published protocol: Akuze J, Annerstedt KS, Benova L, et al. Action leveraging evidence to reduce perinatal mortality and morbidity (ALERT): study protocol for a stepped-wedge cluster-randomised trial in Benin, Malawi, Tanzania and Uganda. BMC Health Services Research. 2021;21:1324. DOI: 10.1186/s12913-021-07155-z. Trial registration: PACTR202006793783148.
Featured Image Attribution
Featured image: Nurse with neonate, photographed by Hodge120 at Nkhoma Mission Hospital, Malawi, via Wikimedia Commons, licensed under CC BY-SA 4.0. The image illustrates neonatal care in one ALERT study country and does not depict an ALERT trial participant or study procedure.
Advancing Science. Empowering Humanity.
Soleil Research™ provides evidence-based scientific analysis for educational purposes and does not replace clinical judgment, local protocols, or individualized medical care.