BackgroundEvidence-based practices that reduce childbirth-related morbidity and mortality are core processes to quality of care. In the BetterBirth trial, a matched-pair, cluster-randomised controlled trial of a coaching-based implementation of the WHO Safe Childbirth Checklist (SCC) in Uttar Pradesh, India, we observed a significant increase in adherence to practices, but no reduction in perinatal mortality.MethodsWithin the BetterBirth trial, we observed birth attendants in a subset of study sites providing care to labouring women to assess the adherence to individual and groups of practices. We observed care from admission to the facility until 1 hour post partum. We followed observed women/newborns for 7-day perinatal health outcomes. Using this observational data, we conducted a post-hoc, exploratory analysis to understand the relationship of birth attendants’ practice adherence to perinatal mortality.FindingsAcross 30 primary health facilities, we observed 3274 deliveries and obtained 7-day health outcomes. Adherence to individual practices, containing supply preparation and direct provider care, varied widely (0·51 to 99·78%). We recorded 166 perinatal deaths (50·71 per 1000 births), including 56 (17·1 per 1000) stillbirths. Each additional practice performed was significantly associated with reduced odds of perinatal (OR: 0·82, 95% CI: 0·72, 0·93) and early neonatal mortality (OR: 0·78, 95% CI: 0·71, 0·85). Each additional practice as part of direct provider care was associated strongly with reduced odds of perinatal (OR: 0·73, 95% CI: 0·62, 0·86) and early neonatal mortality (OR: 0·67, 95% CI: 0·56, 0·80). No individual practice or single supply preparation was associated with perinatal mortality.InterpretationAdherence to practices on the WHO SCC is associated with reduced mortality, indicating that adherence is a valid indicator of higher quality of care. However, the causal relationships between practices and outcomes are complex.FundingBill & Melinda Gates Foundation.Trial registration detailsClinicalTrials.gov:NCT02148952; Universal Trial Number: U1111-1131-5647.