Background: Child linear growth sometimes improves in both intervention and comparison groups in evaluations of nutrition interventions, possibly because of spillover intervention effects to nonintervention areas or improvements in underlying determinants of nutritional change in both areas.Objective: We aimed to understand what changes in underlying socioeconomic characteristics and behavioral factors are important in explaining improvements in child linear growth.Methods: Baseline (2010) and endline (2014) surveys from the Alive & Thrive impact evaluation were used to identify the underlying determinants of height-for-age z scores (HAZs) among children aged 24-48 mo in Bangladesh (n = 4311) and 24-59 mo in Vietnam (n = 4002). Oaxaca-Blinder regression decompositions were used to examine which underlying determinants contributed to HAZ changes over time.Results: HAZs improved significantly between 2010 and 2014 in Bangladesh (∼0.18 SDs) and Vietnam (0.25 SDs). Underlying determinants improved substantially over time and were larger in Vietnam than in Bangladesh. Multiple regression models revealed significant associations between changes in HAZs and socioeconomic status (SES), food security, maternal education, hygiene, and birth weight in both countries. Changes in HAZs were significantly associated with maternal nutrition knowledge and child dietary diversity in Bangladesh, and with prenatal visits in Vietnam. Improvements in maternal nutrition knowledge in Bangladesh accounted for 20% of the total HAZ change, followed by maternal education (13%), SES (12%), hygiene (10%), and food security (9%). HAZ improvements in Vietnam were accounted for by changes in SES (26%), prenatal visits (25%), hygiene (19%), child birth weight (10%), and maternal education (7%). The decomposition models in both countries performed well, explaining >75% of the HAZ changes.Conclusions: Decomposition is a useful and simple technique for analyzing nonintervention drivers of nutritional change in intervention and comparison areas. Improvements in underlying determinants explained rapid improvements in HAZs between 2010 and 2014 in Bangladesh and Vietnam.
Keywords: Bangladesh; HAZ; Vietnam; decomposition; linear growth; underlying determinants.