Title: Extending synthetic control methods to evaluate the effectiveness of global maternal health programs
Authors: Isabel Fulcher - Harvard Medical School (United States) [presenting]
Abstract: Community health worker (CHW) programs are commonly used to improve access to prenatal, postnatal, and neonatal care in low- and middle-income countries by connecting families to facility-based care. In CHW-led programs operating at scale, rich individual-level data is often collected among persons receiving the intervention as a product of program implementation; however, data is often not collected at baseline or among a comparison group post-intervention. Recent advances in causal inference provide methods for emulating a trial when randomization is not possible or, in the case of synthetic control methods, when there is not an obvious comparison group. We present an extension of synthetic control methods to account for scenarios when data is not collected among the intervention group at baseline. We apply these methods to a large maternal health program in Zanzibar, Tanzania, utilizing data from the program and publicly available Demographic and Health Surveys data.