Community Connector: AHRQ's Visualization of Community-Level Social Determinants of Health Challenge (SDoH)
- Through developing the tool, we learned there is vast heterogeneity across community needs, outcomes, and data availability. This tool can help users understand community differences and explore similarities.
- In late January 2020, the Agency for Healthcare Research and Quality (AHRQ) announced this tool as the grand prize winner of the AHRQ’s Visualization Resources of Community-Level Social Determinants of Health Challenge.
- This tool provides further resources and opportunities for communities to use SDoH data when identifying intervention opportunities.
Clinical care accounts for a small part of the factors that affect population health. The rest is determined by social determinants of health (SDoH)—social, behavioral, and environmental factors that interact dynamically to keep people healthy. Many tools, such as the Opportunity Atlas, the City Health Dashboard, and the Robert Wood Johnson Foundation rankings, describe the social needs of populations but do not integrate a variety of federal, state, and local data sources and summarize only one community at a time, limiting the potential for peer-to-peer, comprehensive learning. Thus, we sought to build a common definition of local SDoH and a way to identify communities with similar needs and demographics, particularly those that have had success in addressing social needs and improving health and well-being.
Our Community Connector tool is designed to summarize a community’s social needs in one picture, or fingerprint; identify communities with similar fingerprints; and compare communities across key indicators of health care utilization and cost. This fingerprint is based on outcome-agnostic county-level scores for six domains of SDoH identified by the Kaiser Family Foundation. We used sparse principal component analysis to determine which variables would be used in the defined SDoH scores and assigned additional variables that were not selected by this approach based on prior knowledge and strong association with the health outcomes. Each of the domain scores are a weighted average of the selected, normalized variables. The tool also provides a comparison of a county’s fingerprint to other counties with similar demographic and nonmodifiable SDoH characteristics, where the similarity is determined using a Lasso regression model.
We collected open-source federal, state, and local data for the state of Colorado and targeted health outcomes related to obesity, diabetes, and kidney disease for the prototype. We used data sources such as the Colorado Department of Public Health and Environment and the Centers for Disease Control and Prevention’s Diabetes Atlas. With additional time and funding, the tool can expand and scale nationally and present analyses on a larger set of health outcomes and utilization data.