Well-designed and managed community health worker (CHW) programs can lead to large improvements in health provision and consequent reductions in child morbidity and mortality. Yet, evidence of whether and how these effects can be sustained when programs are operated at scale is largely missing. We use a large cluster-randomized controlled trial to assess the impact of two scaled-up programs (Living Goods and BRAC CHW programs), versions of which previously, in a proof-of-concept trial, were effective in reducing all-cause under-five mortality in Uganda. The scaled-up program trial used elements of an adaptive design where performance outcomes were collected and reported to the program implementers mid-way through the trial period, allowing for evidence-based modifications of the intervention design and implementation guidelines and extension of the trial period while leaving the initial treatment assignment unchanged. Results from the performance review after 18 months showed small improvements in outcomes in both sub-trials. The design and implementation guidelines of the Living Goods CHW program were modified in response, while the BRAC CHW program underwent modest changes. The trial period was extended for two years. In the revised design period, and for the sample of children fully exposed to the program, the scaled-up Living Goods CHW program reduced U5MR by 29%. We find no significant difference in child mortality between the intervention and control clusters in the BRAC CHW program.

Written with P. Awor (Makerere University), A. Guariso (University of Bicocca), and J. Svensson (IIES, Stockholm University)