Background: Quality improvement is a recommended strategy to improve implementation levels for evidence-based essential interventions, but experience of and evidence for its effects in low-resource settings are limited. We hypothesised that a systemic and collaborative quality improvement approach covering district, facility and community levels, supported by report cards generated through continuous household and health facility surveys, could improve the implementation levels and have a measurable population-level impact on coverage and quality of essential services.

Methods: Collaborative quality improvement teams tested self-identified strategies (change ideas) to support the implementation of essential maternal and newborn interventions recommended by the World Health Organization. In Tanzania and Uganda, we used a plausibility design to compare the changes over time in one intervention district with those in a comparison district in each country. Evaluation included indicators of process, coverage and implementation practice analysed with a difference-of-differences and a time-series approach, using data from independent continuous household and health facility surveys from 2011 to 2014. Primary outcomes for both countries were birth in health facilities, breastfeeding within 1 h after birth, oxytocin administration after birth and knowledge of danger signs for mothers and babies. Interpretation of the results considered contextual factors.

Results: The intervention was associated with improvements on one of four primary outcomes. We observed a 26-percentage-point increase (95% CI 25–28%) in the proportion of live births where mothers received uterotonics within 1 min after birth in the intervention compared to the comparison district in Tanzania and an 8-percentage-point increase (95% CI 6–9%) in Uganda. The other primary indicators showed no evidence of improvement. In Tanzania, we saw positive changes for two other outcomes reflecting locally identified improvement topics. The intervention was associated with an increase in preparation of clean birth kits for home deliveries (31 percentage points, 95% CI 2–60%) and an increase in health facility supervision by district staff (14 percentage points, 95% CI 0–28%).

Conclusions: The systemic quality improvement approach was associated with improvements of only one of four primary outcomes, as well as two Tanzania-specific secondary outcomes. Reasons for the lack of effects included limited implementation strength as well a relatively short follow-up period in combination with a 1-year recall period for population-based estimates and a limited power of the study to detect changes smaller than 10 percentage points.