Abstract

As regards study designs, randomised controlled trials are judged as the gold standard for quantitatively evaluating treatment effect sizes with less bias than observational trials. In some cases, the RCTs can be considered unethical, not feasible and impractical to conduct. In such cases, when RCTs are not appropriate for evaluating interventions, observational studies, which generate valuable health data and are readily available, have been used. A major disadvantage of observational studies is that they cannot be used for investigating cause–effect relationships due to confounding factors. Propensity score approaches are one of the strategies that have been developed to control for confounder bias in observational studies and allow for the estimation of causal association. This chapter provides a description and theoretical fundamentals of two propensity-score-based approaches, namely the propensity score matching and propensity score weighting for facilitating the assessment of causal exposure effects using observational data. The two methods are illustrated with an evaluation of the effects of: (a) exclusive breastfeeding or (b) appropriate complementary feeding on nutritional outcomes of infants or children using survey data from Malawi and Zambia.

Original language	en
Pages (from-to)	41-86
Publication status	Published - 2022