Availability and the quality of key newborn data within routine health facility data: findings of the IMPULSE observational study in the Central African Republic, Ethiopia, Tanzania, and Uganda.

Background: With declining funding for population-based household surveys, routine health facility data offer a promising alternative for tracking newborn health and service quality. However, their utility depends on data quality. We assessed the quality of ten data elements within routine health information systems in the Central African Republic (CAR), Ethiopia, Tanzania, and Uganda, seven of which align with the Every Newborn Action Plan core newborn indicators. Methods: We conducted a cross-sectional study in 97 emergency obstetric and newborn care facilities across 4 countries between November 2022 and July 2024. We extracted three months of routine register and summary report data on ten maternal and newborn elements (two denominators, three outcome numerators, five newborn care interventions) and one tracer maternal indicator. We evaluated data quality on four dimensions (availability, completeness, accuracy, and internal consistency) and measured internal consistency using the ratio of (total births - live births)/stillbirths, with a value of 1 suggesting ideal internal consistency. Results: Denominator completeness exceeded 90% in Uganda and Tanzania, but was lower in the CAR (87%) and Ethiopia (82%). Impact numerator completeness averaged 79% for neonatal mortality and 81% for low birth weight, with Ethiopia performing worst, with scores of 45% and 32%, respectively). Completeness for newborn interventions (early breastfeeding, kangaroo mother care, bag-mask ventilation, sepsis management) remained below 90%, with the CAR lacking neonatal sepsis data and Ethiopia lacking early breastfeeding data. Accuracy was poor: concordance between register recounts and summary reports ranged from 9% to 40%. Internal consistency checks revealed mismatches in 80% of facilities, including negative ratios in Uganda and ratios >1 in the CAR. Conclusions: Significant gaps in completeness, accuracy, and internal consistency undermine the reliability of newborn and stillbirth data in routine health information systems, highlighting a need for their strengthening, the integration of standardised newborn indicators, and institutionalized quality verification processes to ensure timely, reliable, and actionable data for improving newborn care.