AUTHOR=Twabi Halima S. , Manda Samuel O. M. , Small Dylan S. 

TITLE=Evaluating the Effect of Appropriate Complementary Feeding Practices on Child Growth in Malawi Using Cross-Sectional Data: An Application of Propensity Score Matching

JOURNAL=Frontiers in Nutrition

VOLUME=Volume 8 - 2021

YEAR=2021

URL=https://www.frontiersin.org/journals/nutrition/articles/10.3389/fnut.2021.714232

DOI=10.3389/fnut.2021.714232

ISSN=2296-861X

ABSTRACT=Introduction: Estimating the causal association between appropriate complementary feeding and child growth using a randomized control trial is practically and ethically close to impossible. Using observational data, this study shows how potential confounder variables can be statistically balanced using propensity score matching to assess the causal association between appropriate complementary feeding and child growth, namely length/height-for-age, weight-for-length/height, and weight-for-age z-scores (HAZ, WHZ and WAZ, respectively) in Malawi. 
Methods: Data on 4722 children aged 6 to 23 months from the 2015-16 Malawi Demographic and Health Survey (MDHS) were analyzed. Propensity scores were obtained by modeling the probability of a child being on appropriate complementary feeding as a function of potential confounders. The derived propensity scores were used to match children who received appropriate complementary feeding with those who did not receive appropriate complementary feeding. 
Results: The prevalence of stunting (HAZ < -2) was 31.9% (95% CI:29.3%,34.6%), wasting (WHZ<-2) 3.5%(95% CI:2.6%,4.7%) and underweight (WAZ < -2) 9.9%(95% CI:8.4%,11.8%). Of the 4722 children, 7.7% (95% CI: 6.9%, 8.5%) were appropriately complementary fed. Appropriate complementary feeding was beneficial in significantly reducing levels of stunting (OR (95% CI); 0.7 (0.4 - 0.95)). On the other hand, it lowered level of wasting and underweight but it was not statistically significant (OR (95% CI); 0.4 (0.1 - 1.7) and 0.6 (0.2 - 1.7), respectively). 
Conclusion: This study has illustrated the application of propensity score matching to balance potential confounder variables when assessing causal effect in child health research based on observational data. We have found that appropriate complementary feeding is associated with reduced levels of adverse child growth. In the absence of randomized control trials, we recommend the use of statistical control confounder bias when assessing the causal effect of interventions on child health using observational data.