AHRQ's database on Social Determinants of Health (SDOH) was created under a project funded by the Patient Centered Outcomes Research (PCOR) Trust Fund. The purpose of this project is to create easy to use, easily linkable SDOH-focused data to use in PCOR research, inform approaches to address emerging health issues, and ultimately contribute to improved health outcomes.The database was developed to make it easier to find a range of well documented, readily linkable SDOH variables across domains without having to access multiple source files, facilitating SDOH research and analysis.Variables in the files correspond to five key SDOH domains: social context (e.g., age, race/ethnicity, veteran status), economic context (e.g., income, unemployment rate), education, physical infrastructure (e.g, housing, crime, transportation), and healthcare context (e.g., health insurance). The files can be linked to other data by geography (county, ZIP Code, and census tract). The database includes data files and codebooks by year at three levels of geography, as well as a documentation file.The data contained in the SDOH database are drawn from multiple sources and variables may have differing availability, patterns of missing, and methodological considerations across sources, geographies, and years. Users should refer to the data source documentation and codebooks, as well as the original data sources, to help identify these patterns

This archived SDOH Database (beta version) is available for reference. The most recent version of the SDOH Database replaces the beta version and is available on the main page. To ensure consistency in variable names and construction, analyses should not combine data from the beta version and the updated database.Download DataThe SDOH Data Source Documentation (PDF, 1.5 MB) file contains information for researchers about the structure and contents of the database and descriptions of each data source used to populate the database.The Variable Codebook (XLSX, 494 KB) Excel file provides descriptive statistics for each SDOH variable by year.***Microdata: YesLevel of Analysis: Local - Tract, CountyVariables Present: Separate DocumentFile Layout: .xslxCodebook: Yes Methods: YesWeights (with appropriate documentation): YesPublications: NoAggregate Data: Yes

These datasets contain data from the AHRQ Social Determinants of Health (SDOH) Database (https://www.ahrq.gov/sdoh/data-analytics/sdoh-data.html), processed to facilitate machine learning/multivariate analyses focusing on the healthcare context of counties. The datasets derive from the AHRQ 2019 and 2018 county-level SDOH files. Three sets of files are provided. The first "Raw" set has the source SDOH data with a few core pre-processing steps applied. The second, “Full” set has variables characterizing the health and healthcare context of counties (rather than outcomes), with further processing steps applied to facilitate multivariate and machine learning analytics (e.g. handling of missing data, normalizing, standardizing). The third set, labeled “Reduced”, incorporates those same data processing steps but in addition has had a further data reduction step applied in which groups of highly intercorrelated variables were removed and replaced with corresponding principal component scores, one for each group. These files would be useful for investigators interested in characterizing and comparing the broad SDOH context of US counties.

Social isolation, described as a state in which an individual lacks social connectedness and personal relationships, was declared an epidemic by the United States Surgeon General in 2023, but nationwide data for county-level isolation prevalence and its most influential social determinants of health (SDOH) is limited. This gap in data and lack of SDOH attribution prevents policy-makers from constructing targeted prevention strategies that directly impact social isolation at the county level. Our study aims to close this gap by linking all available SDOH predictor variables with known social isolation estimates, enabling us to predict isolation estimates for counties that lack them and identify dominant SDOH drivers for each county.We obtained known social isolation data from the Centers for Disease Control and Prevention Population Level Analysis and Community Estimates (PLACES) 2024 dataset and county-level SDOH data from the Agency for Healthcare Research and Quality (AHRQ) 2022 dataset. We then merged the two datasets, applied a linear regression model, and pruned predictors using statistical tools, generating a final set of 23 SDOH predictors used to predict social isolation percentages for 727 missing counties/county-equivalents. The contributions for each predictor were also calculated to find the dominant SDOH for each county and each predictor was classified as actionable or non-actionable. Of the 23, 6 were actionable, the most significant of which was broadband access, followed by public-only health insurance coverage and bachelor’s degree attainment.This study completes the CDC PLACES dataset with comprehensive nationwide county-level estimates and identifies the most dominant SDOH drivers. This can inform targeted public health interventions, aligning with PHR’s mission of moving science into policy and practical public health applications.

The first Social Drivers of Health (SDoH) dataset contains percentages of preventable hospitalizations (i.e., discharges) by Race/Ethnicity, preferred language spoken, expected payer, percent of employment, percent of home ownership, percent of park access and percent of access to basic kitchen facilities by the stated year. Preventable hospitalizations rates were created by dividing the number of patients who are 18 years and older and were admitted to a hospital for at least one of the preventable hospitalization diagnoses (see list below) by the total number of hospitalizations. List of preventable hospitalization diagnoses: diabetes with short-term complications, diabetes with long-term complications, uncontrolled diabetes without complications, diabetes with lower-extremity amputation, chronic obstructive pulmonary disease, asthma, hypertension, heart failure, angina without a cardiac procedure, dehydration, bacterial pneumonia, or urinary tract infection were counted as a preventable hospitalization. These conditions correspond with the conditions used in the Agency for Healthcare Research and Quality’s (AHRQ), Prevention Quality Indicator - Overall Composite Measure (PQI #90). The SDoH "overtime" dataset contains percentages of preventable hospitalizations (i.e., discharges) by Race/Ethnicity, preferred language spoken and expected payer overtime in the stated year range.

The first Social Drivers of Health (SDoH) dataset contains percentages of preventable hospitalizations (i.e., discharges) by Race/Ethnicity, preferred language spoken, expected payer, percent of employment, percent of home ownership, percent of park access and percent of access to basic kitchen facilities by the stated year. Preventable hospitalizations rates were created by dividing the number of patients who are 18 years and older and were admitted to a hospital for at least one of the preventable hospitalization diagnoses (see list below) by the total number of hospitalizations. List of preventable hospitalization diagnoses: diabetes with short-term complications, diabetes with long-term complications, uncontrolled diabetes without complications, diabetes with lower-extremity amputation, chronic obstructive pulmonary disease, asthma, hypertension, heart failure, angina without a cardiac procedure, dehydration, bacterial pneumonia, or urinary tract infection were counted as a preventable hospitalization. These conditions correspond with the conditions used in the Agency for Healthcare Research and Quality’s (AHRQ), Prevention Quality Indicator - Overall Composite Measure (PQI #90). The SDoH "overtime" dataset contains percentages of preventable hospitalizations (i.e., discharges) by Race/Ethnicity, preferred language spoken and expected payer overtime in the stated year range.

Social isolation, described as a state in which an individual lacks social connectedness and personal relationships, was declared an epidemic by the United States Surgeon General in 2023, but nationwide data for county-level isolation prevalence and its most influential social determinants of health (SDOH) is limited. This gap in data and lack of SDOH attribution prevents policy-makers from constructing targeted prevention strategies that directly impact social isolation at the county level. Our study aims to close this gap by linking all available SDOH predictor variables with known social isolation estimates, enabling us to predict isolation estimates for counties that lack them and identify dominant SDOH drivers for each county.We obtained known social isolation data from the Centers for Disease Control and Prevention Population Level Analysis and Community Estimates (PLACES) 2024 dataset and county-level SDOH data from the Agency for Healthcare Research and Quality (AHRQ) 2022 dataset. We then merged the two datasets, applied a linear regression model, and pruned predictors using statistical tools, generating a final set of 23 SDOH predictors used to predict social isolation percentages for 727 missing counties/county-equivalents. The contributions for each predictor were also calculated to find the dominant SDOH for each county and each predictor was classified as actionable or non-actionable. Of the 23, 6 were actionable, the most significant of which was broadband access, followed by public-only health insurance coverage and bachelor’s degree attainment.This study completes the CDC PLACES dataset with comprehensive nationwide county-level estimates and identifies the most dominant SDOH drivers. This can inform targeted public health interventions, aligning with PHR’s mission of moving science into policy and practical public health applications.