INTRODUCTION:
Approximately 80% of 2.3 million global neonatal deaths occur in sub-Saharan Africa and South Asia1. However, there is a paucity of studies that account for nuanced diagnostic challenges in neonates. The aim of this study was to determine the accuracy of antemortem clinician diagnoses compared to postmortem causes of neonatal deaths in the Child Health and Mortality Prevention Surveillance (CHAMPS) network.

METHODS:
This was a cross-sectional analysis of a prospective surveillance study utilising neonatal mortality data from 2016 to 2024 in seven countries in the CHAMPS network. We included cases that died from the leading five causes of death in the causal chain. Diagnostic accuracy was assessed via Youden’s index and the area under the curve (AUC)2, and reliability was assessed using Kappa’s analysis3. Diagnostic accuracy and reliability values ≥0.80 were considered to have good clinical utility2,3. All analyses were done at 5% significance level using Stata 18.

RESULTS:
Of the 4296 total deceased cases, 1831 (42.6%) were included in the analyses. The median age at the time of death was 2 days (IQR 1-5 days). Perinatal asphyxia/hypoxia was the leading cause of death in the causal chain (43.3%). The diagnostic accuracy of clinician antemortem diagnoses was lowest for lower respiratory tract infection (LRTI - Youden’s index 0.04, 95% CI 0.01, 0.09) and highest for perinatal asphyxia/hypoxia (Youden’s index 0.45, 95% CI 0.39, 0.51). The AUC for antemortem clinical diagnoses compared with postmortem determined cause of death ranged from 0.58 for LRTI to 0.74 for perinatal asphyxia/hypoxia. The Kappa values ranged from 0.06 (95% CI 0.02, 0.11) for LRTI to 0.46 (95% CI 0.42, 0.50) for perinatal asphyxia/hypoxia.

CONCLUSIONS:
The antemortem clinical diagnostic performance and reliability were suboptimal. Diagnostic capacity building of clinicians and enhanced allocation of diagnostic resources are imperative to inform clinical care and reduce neonatal mortality.