Effect of a conditional cash transfer program on length-for-age and weight-for-age in Brazilian infants at 24 months using doubly-robust, targeted estimation


OBJECTIVE: Conditional cash transfer programs are popular internationally and represent a large investment in child health. Evidence of their impact on child nutrition status remains weak and inconsistent, particularly for Bolsa Familia, the Brazilian conditional cash transfer program and one of the worlds largest. Our objective was to estimate the effect of the Brazilian conditional cash transfer program, Bolsa Familia (BF), on child nutritional status as measured by length-for-age z-score (LAZ) and weight-for-age z-score (WAZ) at 24 months.
METHODS: We analyzed the 1703 children eligible for BF from the 2004 Pelotas Birth Cohort. Children were divided into three exposure groups by total amount of money their household received from BF in 24 months: no BF, low BF (???R$1000) and high BF (>R$1000). Using a doubly robust semiparametric estimation method we estimated the effect of receiving low and high levels of BF on LAZ and WAZ at 24 months. RESULTS: After adjustment for measured confounders, the expected difference in LAZ between children that received low or high levels of BF compared to no BF was -0.14 [95% confidence interval (CI): -0.27, -0.02] and -0.20 (95% CI: -0.33, -0.08) respectively. For WAZ the estimated differences were -0.04 (95% CI: -0.17, 0.08) for low levels versus no BF and -0.18 (95% CI: -0.30, -0.05) for high levels versus no BF. The expected difference in population LAZ had all eligible households received it and population LAZ under no BF was -0.15 (95% CI: -0.26, -0.04). Sensitivity analyses suggested only a strong confounder could explain away these results. CONCLUSIONS: Among participants of the 2004 Pelotas Birth Cohort, BF was associated with a reduction in LAZ and WAZ in 24 month old children

Social Science & Medicine
Jeremy A. Labrecque
Assistant professor, Epidemiology and causal inference

My research is on how we know what we know.