Previously, we published a primer on social determinants of health (SDoH), social risk, and health equity indices, providing an evaluation framework to uncover critical characteristics when considering an index.
This article offers a deeper look into 5 prominent indices, providing an overview that can be used to facilitate comparisons while offering guidance on the SDoH use cases that may be appropriate for each index.
Our objective is to empower health equity advocates and leaders to answer, “Which SDoH and social risk score indices meet our needs for 1 or more of the following 6 social risk areas?"
- Socioeconomic position: income or wealth, education, occupation
- Race, ethnicity, and cultural context: race/ethnicity, language, nativity, acculturation
- Gender: gender, gender identity, sexual orientation
- Social relationships: marital status, social support
- Residential and community context: community socioeconomic composition, built environment
- Social needs: housing instability, food insecurity, interpersonal safety
There are possibly hundreds of indices available to assess the social risks for a given geographical community. And, while social indices have been used and proven valuable for policy development, recent analyses suggest that choosing a SDoH or social risk index takes deeper vetting.
Here we provide an overview of 5 prominent indices along with considerations for aligning use cases with scores.
Recent SDoH and social risk index analysis create a foundation
Published in December 2022, RTI reviewed SDoH indices, creating an in-depth study of the 5 most commonly used. That same month, a team of researchers completed a scoping and qualitative analysis of socioeconomic deprivation indices in Health Affairs, reviewing 15 indices.
The analysis informs this overview and provide an in-depth reference when embarking on vetting available indices.
How area-level indices are used in SDoH, social risk, and population health programs
Health researchers, health equity, and population health leaders, clinicians, and policymakers want to understand how social drivers are impacting the health of the populations and communities they serve. Using an area-level socioeconomic or deprivation index can focus analysis on pinpoint-specific health or SDoH barriers. This pinpointing can lead to the identification of targeted interventions to address these barriers and mitigate gaps and inequities in healthcare access outcomes.
While socioeconomic status can be evaluated at an individual level or at an area level, broader measures are useful when individual data is not available or to further inform existing sources. Area-level measures examine the role that neighborhood or environmental context has on an individual's health. When used appropriately, area level indices may be useful in informing individual level needs or health outcomes.
There are 2 types of area-level measures: single scores or composite indices. Each have their own strengths.
1. Single area-level measures
- Have fewer missing data
- Simpler to interpret
- Useful in evaluating the impact of an aspect of socioeconomic deprivation, like education or unemployment
2. Composite indices
- Provide more comprehensive representation of neighborhood social determinants of health
- Capture multiple dimensions of socioeconomic disadvantage
- Have been shown to have a stronger relationship with health outcomes
- Can help avoid statistical problems related to collinearity among individual neighborhood-level variables
When deciding which type of index could fit the current need, it's important to ask the question as Lisa M. Lines, PhD, MPH, RTI Senior Health Services Researcher, did in a case study article about a pilot using the RTI Rarity™ Local Social Inequality (LSI) Score to determine the variance in life expectancy for residents of Ohio.
“Lacking person-level information, can ZIP-code-level information be informative? While person-level data is best, we have ample evidence of the population-level ties between neighborhood characteristics and health outcomes going back to at least the 1980s. What is new is the ability to get more granular data into our models."
RTI's SDoH index analysis led to 4 conclusions that any selection process should consider:
- Publicly available SDoH indices are not designed to be universal measures or compendia of measures that can be readily entered into statistical models. Each varies in its purpose and composition.
- These indices draw from multiple data sources, such as the American Community Survey and Behavioral Risk Factor Surveillance System. These indices also use different measures to conceptualize SDoH domains, such as economic stability, education access and quality, healthcare access and quality, neighborhood and built environment, and social and community context.
- The composition of an index has multiple analytic implications for statistical modeling, including impacts that may skew the data or introduce collinearity.
- Many of the data available today predate the pandemic and may not capture the current state of key SDoH factors.
These considerations are useful when reviewing the 5 area-level indices profiled below, particularly when evaluating SDoH, health equity, and population health use cases in the final section.
Overview of 5 area-level indices & scores
The following 5 indices are profiled below include Area Deprivation Index (ADI), Social Deprivation Index (SDI), Social Vulnerability Index (SVI), Distressed Communities Index, and the RTI Rarity LSI score.
1. Area Deprivation Index (ADI)
Level of disaggregation: Census block group
Data source: ACS 5-year estimates, current scores use 2014-2018
Number of variables/measures included: 17
Description: Created to inform healthcare delivery and policy, especially for disadvantaged areas focusing on 4 domains: income, education, employment, and housing quality. An example of its use is MassHealth. They use ADI as a census tract neighborhood stress score to help adjust payments according to a patient's social risk. Another includes the new ACO REACH model that uses ADI to adjust reimbursement rates.
How it's calculated: Originally developed using data from the 1990 census, factor analysis was conducted on 21 variables and 17 indicators were weighted using factor score coefficients. The variables were multiplied by their factor weights and summed for each geographic unit. The result is transformed into the ADI standardized index.
Special features: The ADI has been validated across a variety of health outcomes at the census block-group level. The ADI can be merged with other data sources because it has been cross-referenced to over 69 million 9-digit ZIP codes. Interactive maps and platform allow for data visualization.
2. Social Deprivation Index (SDI)
Level of disaggregation: County, Census tract, ZIP code tabulation area, primary care service area
Data source: ACS 5-year estimates
Number of variables/measures included: 7
Description: Examines relationships between levels of social disadvantage, health, and healthcare. Focuses on measures related to poverty, education, single-parent household, rented housing, overcrowding, access to a vehicle, and unemployment.
How it's calculated: A list of 14 ACS measures were converted into centiles to create a common scale. Factor analysis methods were used to investigate the relationship between a group of observed variables and an unobserved, underlying variable. To simplify the model, the SDI includes 7 measures with factor loadings greater than 0.60 with a final SDI measure based on weighted factor loading scores for each measure.
Special features: The SDI is updated annually with the most recent ACS 5-year estimates, but downloadable files should be checked for the latest data.
3. Social Vulnerability Index (SVI)
Level of disaggregation: County, Census tract
Data source: ACS 5-year estimates
Number of variables/measures included: 15
Description: Identifies areas most likely to need support related to hazardous events across 4 areas: socioeconomic status, household composition and disability, minority status and language, and housing type and transportation. An example of use includes Arizona's Medicaid agency that incorporates a modified version of the SVI in its payment model.
How it's calculated: SVI includes ranked Census tracts data for each state and the District of Columbia. These ranked tracts are compared against each another, allowing mapping and analysis of relative vulnerability in multiple or all states. Tract ranking are based on percentiles with values range from 0 to 1; higher values indicate greater vulnerability. Each tract includes a percentile rank among all tracts for 15 individual variables, 4 themes, and its overall position. Each theme includes percentile sums and percentile rankings.
Special features: Developed using SQL programming language, which can create slightly different estimates from those produced using Microsoft Excel.
4. Distressed Communities Index (DCI)
Level of disaggregation: ZIP code (with 500 residents)
Data source: ACS 5-year estimates, Census Business Patterns data sets
Number of variables/measures included: 7
Description: Provides a single, comparative measure of economic well-being across communities, including high school diploma, housing vacancy, unemployment, poverty, median income ratio, change in employment, and change in business establishments. The DCI measures the comparative economic well-being of communities to illuminate ground-level disparities, capturing 99% of the US population with 26,000 ZIP codes, representing a minimum of 500 residents.
How it's calculated: Each community score is equal to its percentile rank across all 7 measures. Each community is ranked on each measure, those rankings are then averaged, weighted equally, and used to create a preliminary score. This score is normalized into a final score, ranging from approaching 0 (most prosperous) to 100 (most distressed).
5. RTI Rarity Local Social Inequality (LSI) Score
Level of disaggregation: Census tract, ZIP code, and county levels
Data sources: More than 40 federal and private data sources
Number of variables/measures included: 200+
The RTI Rarity tool creates and delivers the Local Social Inequity (LSI) score by drawing on 40+ public and private datasets from publicly available federal, state, non-profit, and academic resources, including:
- American Community Survey (ACS) from the U.S. Census Bureau
- USDA's Food Environment Atlas
- CDC's Wide-ranging ONline Data for Epidemiologic Research (WONDER)
- Department of Housing and Urban Development (HUD)
- Child Opportunity Index (COI)
- Opportunity Atlas
- and 34+ other datasets
Description: The RTI Rarity LSI score reveals key characteristics of higher-risk populations. It aids in assessing SDoH needs, identifying target interventions to address gaps in care, and provides evidence for NCQA accreditation.
How it's calculated: The RTI Rarity tool merges artificial intelligence (AI), advanced data science methods, and geospatial analytics in a risk adjustment framework. Featuring a supervised machine learning method known as random forests (RFs) and other state-of-the-art predictive analytics methods, it provides high-resolution SDoH composite scores constructed with a health equity focus.
Special features: The RTI Rarity tool measures for bias, stress, and trauma such as residential segregation and redlining. As a more comprehensive index, it can account for the systemic racism that drives many inequities among populations.
Validation analysis: In benchmarking and validation analyses, the RTI Rarity LSI score proved to be substantially better at explaining variance in life expectancy than other commonly used SDoH composite measures. Not only can LSI scores identify which communities are at high risk of poor health outcomes, it identifies the most important social, behavioral, and contextual predictors for those outcomes within neighborhoods across the US.
Which index is most accurate when explaining life expectancy variance?
When benchmarked against the major indices, the RTI Rarity LSI score proves to be substantially better at explaining variance in life expectancy among all US neighborhoods:
- RTI Rarity Local Social Inequality (LSI) Score: 67%
- Area Deprivation Index (ADI): 43%
- Social Deprivation Index (SDI): 29%
- Social Vulnerability Index (SVI): 26%
In Dr. Lines' article, featuring a RTI Rarity LSI score pilot, she highlighted that her team's work specifically addressed potential biases in several ways.
They created a conceptual model of SDoH that guided which risk factors to include. She states, "Our conceptual model builds on the CDC's Healthy People 2020 framework for social determinants."
She and her team "explicitly called out bias, stress, and trauma by including measures of racial segregation and inequality in the model." Additionally, the RTI Rarity LSI score algorithm predicts Census tract (CT)-level life expectancy, rather than spending.
Marrying data science, proven success, and expert guidance
RTI Health Advance combines Health Equity and SDoH consulting and solutions for a comprehensive approach to address health disparities. Contact us to discuss how to put our benchmarking research and experience to work for your organization.