Objective: Majority of neonates in developing countries are born at home and most neonatal deaths occur without receiving medical care. This retrospective analysis was undertaken to develop simple clinical criteria for use in rural community to identify neonates at risk of death.
Study design: By analyzing the observational data on two cohorts of neonates in 39 villages in different years of the Gadchiroli field trial, we selected a minimum set of clinical features. We evaluated this set for its sensitivity, specificity and predictive value to detect eventual neonatal death, the primary study outcome.
Results: The cohorts included 763 neonates with 40 deaths in 1995 to 1996, a year with minimum interventions, and 1598 neonates with 38 deaths in 1996 to 1998, the years of intensive interventions. On the day of birth, presence of any one of the three: (1) birth weight <2000 g, (2) preterm birth or (3) baby not taking feeds; or, during the rest of neonatal life, mother's report of reduced or stopped sucking by baby, were identified as the predictors of neonatal deaths. The combined set gave a sensitivity of 95%, specificity, 77.3%; predictive value, 18.8%; and the yield, 26.5% in 1995 to 1996 and, respectively, 86.8, 78, 8.8, and 23.5% in 1996 to 1998. The mean lead time gained was 3.4 to 6.6 days.
Conclusion: Presence of any one of the four predictors will identify with high sensitivity and moderate specificity nearly a quarter of the neonates in rural community as high risk, 3.4 to 6.6 days in advance, for intensive attention at home or referral.