Identifying delays impacting maternal and perinatal deaths using a facility-based death audit review system integrated with community engagement: a mixed methods study.

Background: Pakistan's persistently high maternal (178 per 100 000 livebirths) and neonatal (41 per 1000 livebirths) mortality ratios are compounded by lack of reliable data on the root causes and preventable risk factors, which hinders their effective use in improving care and progress toward safe, high-quality services. We sought to identify system delays impacting maternal and perinatal deaths using a facility-based death audit review system integrated with community engagement for implementing actionable solutions. Methods: We used a mixed methods, concurrent parallel study design at three secondary level healthcare facilities in District Matiari, Sindh, Pakistan. We reviewed 319 cases that included 19 maternal deaths (MDs), 185 neonatal deaths (NDs) and 115 cases of stillbirth (SB) using the WHO based death audit review system integrated with community engagement. We documented quantitative data for all cases using descriptive statistics and simultaneously collected qualitative narratives, which we then analysed and categorised across the four delays model. Audit committees comprising facility staff and community representatives met quarterly to review cases. Results: We observed delay patterns across death types. MDs were predominantly influenced by delays 3 (reaching facility: 74%) and 4 (receiving adequate care: 74%), reflecting system-level barriers in access and quality of care. SB cases were primarily associated with delays 1 (recognition of danger signs: 64%) and 2 (decision to seek care: 60%), highlighting household-level knowledge and decision-making gaps. NDs were primarily affected by delays 4 (receiving adequate care: 54%) and 1 (recognition of danger signs: 48%), indicating both facility capacity constraints and early recognition failures. The analysis revealed four interconnected themes explaining these delays: lack of education and awareness (delays 1 and 2), inadequate transport mechanisms (delay 3), multiple referrals (overlapping delays 3 and 4), and limited facility operational hours and delayed medical care (delay 4). Conclusions: The four delays model identified patterns of preventable factors contributing to maternal and perinatal deaths across household and health system levels. Health systems need to invest in women's access to, and the availability of, healthcare facilities both during and after pregnancy. Scaling up and implementing audit review systems with learning feedback loops is key to systematically identifying and contributing to addressing preventable delays in resource constrained settings, particularly when comprehensive national-level mortality data are lacking.