Data integration under The Demographic and Health Survey Program
On March 22, 2022, Data for Impact (D4I) hosted the second webinar in a series focusing on integration in global health monitoring, evaluation, and learning (MEL).
This webinar featured two staff members, Sara Riese and Rose Donohue, from The Demographic and Health Surveys (DHS) Program. They discussed two recent DHS studies that integrated data; specifically, household survey and health facility assessment data and geospatial covariate and household survey data.
Integration is of increasing importance to global health MEL as more resources are dedicated to improving the integration and use of data from routine health and other sector information systems for evaluation. An increased number of projects tackling a range of health issues at once and projects spanning multiple sectors—health and education, health and environment, etc.—also means a greater need to evaluate and learn from these integrated models.
Integration of household survey and health facility assessment data to calculate effective coverage estimates (Sara Riese)
Effective coverage is a measure of health system performance that combines multiple aspects of health care services delivery into a single measure. Most often this includes measures of need for services, utilization of services, and quality of services. However, these component measures frequently come from different data sources—household surveys provide estimates of need for and utilization of services, while health facility surveys provide estimates of quality of services. This presentation described a recent study which explored approaches to combining these different data sources to calculate effective coverage estimates for antenatal care and sick childcare.
Integration of geospatial covariate and DHS survey data to model DHS indicators at the second subnational administrative level (Rose Donohue)
The DHS are designed to provide reliable estimates of survey indicators at the national level, as well as the first subnational administrative level (Admin 1), often referred to as regions. In recent years, there has been an expressed need to estimate survey indicators at the second subnational administrative level (Admin 2), often referred to as districts, as decision making and program implementation often occur at this level. Geospatial modeling techniques can be used to produce estimates at lower administrative levels, and do not require a costly increase in survey sample size. Rather, these techniques integrate geospatial covariate data—ancillary geospatial data on topics including climate, population, infrastructure, and environment—with the DHS survey data to produced modeled estimates. Here, we presented a geospatial modeling approach we have been using to produce estimates at the Admin 2 level. Results were presented from a recent case study in Zambia and describe how policymakers and program managers can use these results in DHS Program partner countries.
Dr. Sara Riese is a Senior Demographic and Health Researcher at The DHS Program, focusing on quality of care, maternal and child health services, family planning, and inequality. She holds master’s degrees in international affairs and public health from Columbia University and a PhD in Population, Family, and Reproductive Health from Johns Hopkins University.
Dr. Rose Donohue is a Senior Geospatial Data Scientist at The DHS Program. She holds a PhD in Biological Sciences and an MS in Global Health from the University of Notre Dame.