Methods for estimating maternal, newborn, and child health and nutrition effective coverage cascades from household and health facility surveys

BackgroundEffective coverage cascades have been proposed to understand to what extent populations are able to benefit from interventions to address their health needs. Theoretical effective coverage cascades have been developed for reproductive, maternal, newborn, child, and adolescent health and nutrition (RMNCAH&N), but there is no consensus regarding the methods to estimate effective coverage cascades. We operationalized the proposed effective coverage cascades for selected RMNCAH&N services; this paper presents the overall methods, challenges, and lessons learned. MethodsWe used data from Demographic and Health Surveys, Multiple Indicator Cluster Surveys, Service Provision Assessments, and the Service Availability and Readiness Assessment to estimate effective coverage cascades in seven low- and middle- income countries for the following service areas: antenatal care, care for small and/or sick newborns, postnatal care, sick child care, and maternal and child nutrition. We developed operational definitions for each of the seven steps of the effective coverage cascade and developed readiness, and, where data allowed, process quality indices for each service area. Readiness- and process quality-adjusted coverage were estimated using ecological linking by stratum. We propose approaches for dealing with multiple observations per facility; multiple care-seeking episodes; and empty strata, as well as a jackknife approach to estimate the standard errors for readiness- and process quality-adjusted coverage. ResultsWe were able to estimate effective coverage cascades through intervention coverage (step 4) for postnatal care and through process quality-adjusted coverage (step 5) for antenatal care, sick child care, and maternal and child nutrition. For small and/or sick newborn care, we did not have an appropriate denominator or measure of service contact coverage and had to modify the cascade significantly. Data gaps were the largest barrier to the estimation of effective coverage cascades for RMNCAH&N. Other challenges included accounting for community- and home-based interventions, determining whether the cascade should be nested, and interpreting the cascade. ConclusionsTo make effective coverage cascades feasible for routine use, clear guidance is needed on cascade methods and definitions, accounting for the full spectrum of RMNCAH&N interventions, and developing our understanding of how coverage cascades can be used by stakeholders to improve health systems and programs.