Abstract / Summary
Every day, 1990 children die from an injury, with the majority in low- and middle-income countries. In Tanzania, reaching definitive care, or the appropriate level of care for a child's injury, often requires referral to a higher level facility. This hierarchical system introduces challenges that may lead to delayed care and, for severely injured children, cause mortality. The aim was to identify patient, healthcare provider, and health system-level barriers to timely referral for pediatric injury patients at the first site of care in order to identify targets for an intervention to reduce time to definitive care. In this mixed-methods study, a pediatric injury registry was established at three health facilities spanning three health system levels in Northern Tanzania. Thirty family caregivers of pediatric injury patients participated in interviews, while healthcare providers participated in four focus group discussions. The Consolidated Framework for Implementation Research framed interview guides and thematic analysis of the qualitative data. Of 782 patients enrolled in the pediatric injury registry from Aug 2024-Aug 2025, 269 were referred. Most visited 1-4 facilities before reaching definitive care (range 1-6). The median time from injury to definitive care was 15.1 hours (IQR 1.8-33). Qualitative data collection with family caregivers and healthcare providers revealed barriers to explain the delays to care in the outer setting (such as non-standardized referral pathways), the inner setting (such as delays in the decision to refer), and individual characteristics (such as inadequate training in pediatric injury care). According to the World Health Organization standards, every child who needs referral should be sent within 2 hours of arrival. To improve the referral delays found in this study, a locally co-designed, multicomponent health systems intervention for healthcare providers at the first site of care must be developed that includes pediatric injury care training and standardized referral decision support.