Background. Ethical, logistical, and funding approaches preclude conducting randomized control trials (RCTs) in some humanitarian crises. A lack of RCTs and other intervention research has contributed to a limited evidence-base for mental health and psychosocial support (MHPS) programs after disasters, war, and disease outbreaks. Propensity score methods (PSMs) are an alternative analysis technique with potential application for evaluating MHPS programs in humanitarian emergencies.
Methods. PSMs were used to evaluate impacts of education reintegration packages (ERPs) and other (vocational or economic) reintegration packages (ORPs) v. no reintegration programs on mental health of child soldiers. Propensity scores were used to determine weighting of child soldiers in each of the three treatment arms. Multiple linear regression was used to estimate adjusted changes in symptom score severity on culturally validated measures of depression, posttraumatic stress disorder (PTSD), and functional impairment from baseline to 1-year follow-up.
Results. Among 258 Nepali child soldiers participating in reintegration programs, 54.7% completed ERP and 22.9% completed ORP. There was a non-significant reduction in depression by 0.59 (95% CI −1.97 to 0.70) for ERP and by 0.60 (95% CI −2.16 to 0.96) for ORP compared with no treatment. There were non-significant increases in PTSD (1.15, 95% CI −1.55 to 3.86) and functional impairment (0.91, 95% CI −0.31 to 2.14) associated with ERP and similar findings for ORP (PTSD: 0.66, 95% CI −2.24 to 3.57; functional impairment (1.05, 95% CI −0.71 to 2.80).
Conclusion. In a humanitarian crisis in which a non-randomized intervention assignment protocol was employed, the statistical technique of PSMs addressed differences in covariate distribution between child soldiers who received different integration packages. Our analysis did not demonstrate significant changes in psychosocial outcomes for ERPs and ORPs. We suggest the use of PSMs in evaluating non-randomized interventions in humanitarian crises when nonrandomized conditions are not utilized.