Poverty and InequalitySexual and Reproductive HealthFamily, Maternal & Child HealthMethodology

Stunting Mediates the Association between Small-for-Gestational-Age and Postneonatal Mortality

TitleStunting Mediates the Association between Small-for-Gestational-Age and Postneonatal Mortality
Publication TypeJournal Article
Year of Publication2016
AuthorsOddo, VM, Christian, P, Katz, J, Liu, L, Kozuki, N, Black, RE, Ntozini, R, Humphrey, J
JournalJ Nutr
Date PublishedNov
ISBN Number0022-3166
Accession Number27733526
KeywordsHIV, mediation, postneonatal mortality, small for gestational age, stunting

BACKGROUND: In sub-Saharan Africa, one-third of all births are small for gestational age (SGA), and 4.4 million children are stunted; both conditions increase the risk of child mortality. SGA has also been shown to increase the risk of stunting. OBJECTIVE: We tested whether the association between SGA and postneonatal mortality is mediated by stunting. METHODS: We used longitudinal data from children aged 6 wk to 24 mo (n = 12,155) enrolled in the ZVITAMBO (Zimbabwe Vitamin A for Mothers and Babies) trial. HIV exposure was defined based on maternal HIV status at baseline. SGA was defined as birthweight <10th percentile of the INTERGROWTH-21st (International Fetal and Newborn Growth Consortium for the 21st Century) standards. We used a standard mediation approach by comparing the attenuation of the risk when the mediator was added to the model. We used Cox proportional hazards models first to regress SGA on postneonatal mortality, controlling for age. Stunting (length-for-age z score <-2) was then included in the model to test mediation. RESULTS: Approximately 20% of children were term SGA, and 23% were stunted before their last follow-up visit. In this cohort, 31% of children were exposed to HIV; the HIV-exposed group represented a pooled group of HIV-infected and HIV-exposed but uninfected children. Postneonatal mortality was significantly higher among children born SGA (HR: 1.5; 95% CI: 1.3, 1.7). This association was attenuated and not statistically significant when stunting was included in the model, suggesting a mediation effect (HR: 1.1; 95% CI: 0.91, 1.3). When stratified by HIV exposure status, we observed a significant attenuation of the risk, suggesting mediation, only among HIV-exposed children (model 1, HR: 1.3; 95% CI: 1.1, 1.6; model 2, HR: 1.1; 95% CI: 0.88, 1.3). CONCLUSIONS: This analysis aids in investigating pathways that underlie an observed SGA-mortality relation and may inform survival interventions in undernourished settings.