Why do neonates die in rural Gadchiroli, India? (Part II): estimating population attributable risks and contribution of multiple morbidities for identifying a strategy to prevent deaths.

OBJECTIVES: The understanding about why neonates die in rural areas in developing countries is limited. In the first year (1995 to 1996) of the field trial of home-based neonatal care in rural Gadchiroli, India, we prospectively observed a cohort of neonates in 39 villages. In Part I of this article, we presented the primary causes of death. The data were further analyzed: To estimate the population attributable risk (PAR) of death for the main causes of neonatal mortality. To evaluate the effect of a multiplicity of morbidities and to identify which morbidity combinations cause neonatal deaths. To develop a hypothesis about how best to reduce neonatal mortality.

STUDY DESIGN: We analyzed the observational data by logistic regression to estimate the PAR of death for six major morbidities. The effect of the number of morbidities per neonate on case fatality (CF) was estimated. Then we identified the main combinations of morbidities as the component causes leading to death. We estimated the excess deaths attributable to sepsis.

RESULTS: This cohort included 763 neonates among whom 40 neonatal deaths occurred. Six major morbidities were associated with the following proportion of deaths: preterm, 62.5%; sepsis, 60%; intrauterine growth restriction (IUGR), 27.5%; asphyxia, 25%; hypothermia, 22.5%, and feeding problems, 15%. The estimated PARs were: preterm, 0.74; IUGR, 0.55; sepsis, 0.55; asphyxia, 0.35; hypothermia, 0.08, and feeding problems, 0.04. The CF associated with the number of morbidities per neonate was: with no morbidity, 0.3%; one morbidity, 2.1%; two morbidities, 15.3%; three or more morbidities, 41.4% (p<0.001). In all, 82.5% of all deaths occurred in neonates with two or more morbidities. The proportion of total deaths associated with only preterm was 7.5%, and with only IUGR was 2.5%; however, with the main morbidity combinations it was preterm+sepsis, 35%; IUGR+sepsis, 22.5%; preterm+asphyxia, 20%; preterm+hypothermia, 15%; and preterm+feeding problem, 12.5%. The % CF with low birth weight (LBW) <2500 g alone was 5.2% and with infection alone was 1.9%, but with LBW+infection it was 31.9%. The estimated excess deaths caused by sepsis over and above LBW was 44% of the total deaths.

CONCLUSIONS: Preterm and IUGR are ubiquitous components, but usually not sufficient to cause death. Most deaths occur due to a combination of preterm or IUGR with other comorbidities. If preterm birth or IUGR cannot be prevented, the strategy should be to ensure neonatal survival by addressing comorbidities, that is, infections, asphyxia, hypothermia, and feeding problems in that order of priority. We hypothesize that the prevention and/or management of neonatal infections will reduce neonatal mortality by 40 to 50%.