Social Determinants of Health, Quality of Public Health Data, and Health Equity in the United States

Abstract

In recent years, increasing attention has been paid to the impact social determinants of health (SDOH) can have on health equity in the United States.

In this essay, we provide a framework for considering the upstream structural factors that affect the distribution of SDOH as well as the downstream consequences for individuals and groups. Improving health equity in the United States will require multiple policy streams, each requiring comprehensive data for policy development, implementation, and evaluation.

Although much progress has been made in improving these data, there remain considerable gaps and opportunities for improvement. (Am J Public Health. 2023;113(12):1301–1308. https://doi.org/10.2105/AJPH.2023.307423)

In recent years, increasing attention has been paid to the impact social determinants of health (SDOH) can have on health equity in the United States. Yet improving health equity in the United States will require comprehensive data for policy development, implementation, and evaluation. Although much progress has been made in improving these data, there remain considerable gaps and opportunities for improvement.

The recent COVID-19 pandemic exposed and exacerbated many inequities in our society, including those affecting health outcomes. American Indian or Alaska Native, Black, and Hispanic or Latino people experienced greater rates of infection, hospitalization, and death from the virus than did White people.^1,2 Although it is important to have real-time data that are sufficient to formulate and implement policies responsive to the needs of all populations during public health emergencies, it is critical to realize that these disparate effects of COVID-19 are symptomatic of longstanding health inequities in our system. To address underlying systemic drivers of such health inequities, there is an ongoing need for high-quality, timely data that extends beyond the recent pandemic.

Addressing SDOH is a component of the Biden–Harris Administration’s larger health equity agenda. As defined in Executive Order 13985, the term “equity” means the consistent and systematic fair, just, and impartial treatment of all individuals, including individuals who belong to underserved communities that have been denied such treatment, such as Black, Latino, Indigenous and Native American persons, Asian Americans, Native Hawaiians, Pacific Islanders, and other persons of color; members of religious minorities; lesbian, gay, bisexual, transgender, and queer persons; persons with disabilities; persons who live in rural areas; and persons otherwise adversely affected by persistent poverty or inequality.³ According to the Centers for Disease Control and Prevention (CDC), health equity is achieved when every person has the opportunity to “attain his or her full health potential” and no one is “disadvantaged from achieving this potential because of social position or other socially determined circumstances.”⁴

In this essay, we argue that several types of data are needed to inform and support policies to address SDOH, their upstream structural causes, their downstream consequences for individuals and groups, and their interrelationships with the medical care system. We first present a conceptual framework to describe the relationships SDOH have with health equity as a guide to data needs. We then discuss currently available data and the gaps that exist in fulfilling these needs.

CONCEPTUAL FRAMEWORK

The Office of the Assistant Secretary for Planning and Evaluation (ASPE) has developed a conceptual framework (abbreviated framework in Figure 1; detailed framework in Figure A; available as a supplement to the online version of this article at http://www.ajph.org) to guide ASPE’s work on health equity. This framework can help in considering potential policy solutions and assessing knowledge and evidence gaps. It is not, however, a fully predictive model of how key drivers affect specific disparities in outcomes. It recognizes that health equity can be assessed by examining disparities in health indicators (e.g., life expectancy, active life expectancy) and assessing the medical and social drivers of these disparities. These drivers include broad, interrelated aspects of the social, legal, and institutional environment affecting communities such as structural discrimination, physical environment, employment, income and wealth, education, and access to high-quality health care. Most importantly, it is a framework for thinking about the multiple and interrelated social and medical care drivers of health so they can be addressed more comprehensively in policy development and data needs.

FIGURE 1—

A Conceptual Framework for Addressing Social Determinants of Health (SDOH)

Note. For a full version of this framework, see Figure A, available as a supplement to the online version of this article at https://www.ajph.org.

The framework is described in detail elsewhere⁵; here, we highlight the parts most relevant to the current discussion of data needs. The first critical aspect—identified in the “Populations Negatively Affected by SDOH” bar at the top of Figure 1—is the groups for which concerns about disparities in outcomes, opportunity, and experience arise. Assessing equity requires making comparisons between groups with different levels of social advantage.⁶ In virtually every society, social advantage—and its corresponding position in social hierarchies—varies according to socioeconomic, racial, ethnic, gender, age, and geographic differences. In addition, various forms of discrimination can play a significant role in disadvantaging some groups.⁷

The rest of Figure 1 illustrates various nonmedical and medical individual and systemic drivers that combine to influence disparities in outcomes observed (example in Box D1 in Figure 1). These drivers are shown in the 3 columns of the figure: (1) nonmedical determinants, (2) access to care, and (3) experience in the medical care system. The top and middle rows, moving from left to right, provide examples of inequities that affect health. The middle row provides examples of the structural factors affecting health equity, and the top row reflects how these factors are experienced by individuals and communities. The bottom row provides examples of policies that can affect the drivers and outcomes in rows 1 and 2. Box D2 displays examples of health-related social needs (HRSNs), which are distinct from SDOH but must be addressed along with medical care to improve health outcomes.

POLICY STRATEGIES AND DATA NEEDS

A comprehensive approach to addressing SDOH to improve health equity requires strategies across all levels of government that we group into 5 categories:

• 1.Address individual-level HRSNs in communities by using incentives for medical care providers to screen for HRSNs and refer to appropriate services and community assistance to ensure that these services are available.
• 2.Address the underlying systemic determinants (e.g., structural discrimination) that affect the distribution of HRSNs, SDOH, access to care, and quality of care.
• 3.Improve the distribution of community-level SDOH using whole-of-government approaches for improving economic activities, environment, housing, food availability, transportation, and so on.
• 4.Bolster access to health care by expanding insurance coverage and the supply of services and facilities in currently underserved areas.
• 5.Advance quality of care by reducing disparities in care, increasing the provision of culturally and linguistically appropriate care and services to reduce discrimination in care, and ensuring that members of historically underserved communities have training opportunities so the future workforce will reflect the populations served.

Developing, implementing, and evaluating policies in these 5 categories will be challenging across governments at all levels and their community partners. These efforts will require improvements in public health data to identify factors driving observed disparities and how they can best be addressed via the 5 policy categories.

From a national policy perspective, data standardized across programs and data systems allow comparison and data sharing across programs, similar to current hospital discharge data and vital records. But most importantly, the data must contain elements to identify disparities between groups and to analyze the factors contributing to such disparities and the impact of policy interventions intended to address them. Arguably, current data allow researchers and policymakers to assess measures of access and quality of care (categories 4 and 5) more thoroughly than the other categories. Thus, for the remainder of the essay we focus on data for structural discrimination, SDOH, and HRSNs to support the other policy categories.

CURRENT STATUS, GAPS, AND INITIATIVES

Based on the conceptual framework described, a complete set of SDOH-related data would include elements that are not widely available in public health data at this time and are developed with communities. These include data to identify disparities in health outcomes and the social drivers (SDOH, HRSNs, and structural factors) of these disparities.

In this essay, we take a broad view in defining public health data. Specifically, we focus on resources from within and outside public health for the purpose of measuring and improving public health and health equity. Under this broader view, public health data can come from a variety of sources, including surveys, insurance claims, surveillance data, disease registries, vital records, medical records, environmental pollution, water quality, built environment, community vaccination rates, education, employment, justice system involvement, and more. Public health data can include both individual-level and aggregated data. These data sources vary in terms of availability, quality, and interoperability for populations affected, as well as their appropriateness for specific types of public health and policy uses.

Data to Identify Affected Populations

Although some of the groups of concern identified in the executive order are more often captured in public health data, others are rarely included. For example, some databases that contribute to public health data, such as public health surveillance systems, include race and ethnicity information. Other data sources that could be helpful do not, such as private insurance claims for research.⁸ Although race and ethnicity are often collected, most data sources do not include information to identify sexual orientation and gender identity or disability status. For example, Centers for Medicare & Medicaid Services (CMS) programs (Medicare, Medicaid, and the Federally Facilitated Marketplace) do not collect such data in any standardized format, although a few states do collect these data in their Medicaid programs.⁹

Even in cases in which equity data are collected, there may be limitations. For example, despite the existence of race and ethnicity fields in many data sources, there are differing definitions of race/ethnicity, varying levels of detail, challenges with missing data, and unresolved questions about the accuracy for particular groups among these data sources. Moreover, studies using race/ethnicity data often lack the methodological details needed to assess their validity and replicate results.¹⁰ Other gaps include when such data are intentionally not collected or reported to protect patient privacy, particularly when data are collected on a small number of individuals or when there may be heightened sensitivities about potential repercussions for individuals because of the nature of the data. Efforts to address these health equity–related data uses are already under way. For example, CMS is pursuing its future vision for health equity data, including, where applicable and appropriate, alignment with the Department of Health and Human Services’ United States Core Data for Interoperability overseen by the Office of the National Coordinator for Health Information Technology. Additionally, the Office of Management and Budget has proposed to update the federal government’s race and ethnicity data standards.¹¹

There are existing data sources that could start to fill these gaps as well. Claims from public and private insurers are a staple of health services research and have the potential to greatly improve the availability of individual-level data for addressing health equity issues. These data have many advantages, including large sample sizes and patient-level information, that can be used for studying costs, utilization, practice patterns, burden of illness, and access and quality issues, as well as for forming the basis for policy simulations. Eventually, these data may include information on patients’ HRSNs. There are also well-known gaps and issues with claims data that can limit the scope and usefulness of some research efforts, particularly with regard to addressing health equity. Among these gaps are the exclusion of information not needed for reimbursement, including equity information such as race and ethnicity, and the proprietary nature of many claims databases requiring analyses at higher aggregation levels and preventing the comparison of claims across payers to understand the full population.¹²

Structural Factors

For the purposes of developing data and policy, it is important to distinguish between SDOH and HRSNs. SDOH affect everyone; they are not something an individual can have or not have, and they are not inherently positive or negative.¹³ Whether they contribute to either positive or negative community- and individual-level outcomes depends on the nature of the SDOH in a community. The policy levers to address SDOH also exist above the individual level, at the community level or other geographic area. For instance, a community may have an insufficient supply of affordable housing as well as unequal distribution of such housing that may result in high levels of housing instability for some residents of the community. Policies to address the community’s supply of affordable housing require community-level interventions, such as designating that a certain portion of new housing be affordable.

Although SDOH exist at a community or other geographic level, HRSNs describe an individual’s experience. For example, individuals’ inability to access nutritious food and maintain a healthy diet may be related to their community being a “food desert,” that is, lacking in accessible supermarkets or farmers markets. To best support policy development and implementation, it is important to distinguish between SDOH and HRSNs, particularly with regard to the data that are needed.

To assist health care providers in helping to address HRSNs, there have been suggestions to use area-level deprivation indices to adjust Medicare and Medicaid payments.¹⁴ Existing measures of area deprivation have been found to be weakly correlated with HRSNs.^15,16 In practice, a particular community may lack abundant affordable housing, but a single individual in the area may have stable housing, whereas another may experience homelessness, resulting in only the second person having a HRSN. Policies to address HRSNs include having medical care providers screen patients for these needs and refer them to appropriate community-level services and assisting communities to develop platforms for such referrals and ensure the availability of these services.

As policy discussions focus on SDOH and health equity, additional attention has been paid to the structural factors in our systems that result in health disparities; that is, individuals with particular characteristics may be subject to structural inequities that produce adverse health outcomes.¹⁷ A particular focus has been placed on structural racism, which can be defined as the macrolevel systems, social forces, institutions, ideologies, and processes that interact with one another to generate and reinforce inequities among racial and ethnic groups.¹⁸ Structural racism leads to “differential access to the goods, services, and opportunities of society by race,”^19(p1212) determines societal values and power structures, and underlies persistent health disparities in the United States.²⁰ We use the term “structural discrimination” because factors that result in unequal distributions of SDOH and HRSNs and unequal access to high-quality medical care, all of which contribute to poor health outcomes, affect other groups as well (e.g. rural residents). Structural discrimination is different from individual experiences of discrimination, which happen to an individual, rather than a society. But both types of discrimination are harmful to health equity.

Each individual in their community has a unique combination of HRSN, SDOH, and structural discrimination, and their experiences and needs exist at the intersection of these. To truly address social drivers of health, data and policies need to capture not only the group differences but the intersectional impacts as well.

Although we use these concepts distinctly in this essay, in other cases they have all been referred to as SDOH or social risk factors. Here, we will use language based on the definitions we have discussed although source material may use another term.

Health-Related Social Needs Data

Much progress has been made in collecting HRSN data over the past few years. In a 2020 report to Congress, ASPE discussed the then current state of HRSN data collection.²¹ At that time, more than one third of Medicaid managed care plans had reported collecting or planning to collect data on HRSNs in the 2018 to 2019 plan year; the Centers for Medicare and Medicaid Innovation was requiring entities participating in the accountable health communities model to use the CMS-developed HRSNs screening tool (although many other tools were being used in the health care system); and health care providers were beginning to use International Classification of Disease, 10th Revision (ICD-10) Z codes to identify HRSNs in health care claims. Today, all of these efforts have been expanded, and new efforts to improve HRSNs data have begun.

One of the major obstacles to improving HRSNs data collection that ASPE identified was the lack of incentives for health care providers to collect these data, and much has been done to address this. The Centers for Medicare and Medicaid Innovation is encouraging participants in all new payment models to collect HRSNs information though a validated screening tool.⁹ Moreover, CMS has introduced hospital quality measures assessing the proportion of patients screened and the proportion that screen positive for HRSNs. Additionally, Medicare Advantage Special Needs Plans are required to screen enrollees for HRSNs as part of an annual risk assessment.⁹ CMS has also increased providers’ incentives to use ICD-10 Z codes to identify HRSNs by including documented HRSNs in clinical severity and corresponding payment rates.^22,23

Another challenge ASPE identified was the inability to share HRSNs data once collected, and much progress has been made in developing data collection and exchange standards to share HRSNs across the health and social service ecosystem. The Office of the National Coordinator for Health Information Technology has included HRSNs and health equity elements in the United States Core Data for Interoperability, version 2, potentially increasing the ease of exchanging this information across the health care system.²⁴ Over the past 4 years, the _Gravity Project has worked to “develop consensus-driven data standards to support the collection, use, and exchange of data” on HRSNs, creating value sets and use cases for a number of HRSNs that can now be more easily captured and exchanged.²⁵

Despite this progress, many of the challenges ASPE identified have not yet been addressed. Although data sharing has become easier, there are still many different screening tools being used, so information may not be comparable across data sources. Additionally, even when validated screening tools are used, they may not be used consistently, resulting in undocumented HRSNs. This is evident from the still low use of Z codes in Medicare claims.²⁶ More work will need to be done to consistently screen for HRSNs in a way that allows comparison across data sets.

Social Determinants of Health Data

In practice, SDOH data tend to be operationalized as area-level measures of deprivation. The available SDOH data include resources available in a community and community-level characteristics of the population. Although some of these SDOH data have been available for decades, new efforts are making the information more readily accessible and aggregated in a single data set. For example, the Agency for Healthcare Research and Quality has aggregated much of this information into a SDOH database,²⁷ and the CDC has developed the PLACES (population-level analysis and community estimates) data set that combines small-area SDOH and health status information.²⁸

Rather than measuring community-level resources (e.g., affordable housing stock or availability of public transport), much of the existing SDOH data come from survey responses that are aggregated to represent the social characteristics of the population in an area (e.g., proportion of homeless respondents or those with access to a private vehicle). Many CDC surveys include SDOH modules, such as the Behavioral Risk Factor Surveillance System, the Youth Risk Factor Surveillance System, and the Pregnancy Risk Assessment Monitoring System. In particular, Census’ American Community Survey provides a rich data source for small-area measures and has been used to understand specific SDOH and to create multidimension SDOH indices that summarize an area’s level of social deprivation, such as the Area Deprivation Index and the Social Vulnerability Index. However, the applicability of these indices for policy has come under scrutiny.²⁹ Additionally, not all SDOH domains are included in these indices; most include a measure of income or wealth, but population-level HRSNs are much less likely to be included in an index.⁵ Furthermore, a report from the National Commission to Transform Public Health Data Systems found that many individual data sources capture only a few of the SDOH domains, making it difficult to understand the full SDOH picture using a single data source.³⁰ Additional work on these indices will make them more useful for policymakers.

In addition to the aggregate population characteristics, SDOH can be measured by the services and infrastructure of an area, such as the availability of public transportation, healthy food, health services, affordable housing, temperature, and environmental pollutants. Although these measures are available, as evidenced by their inclusion in the Agency for Healthcare Research and Quality SDOH database, they are not typically included in area-level SDOH indices. Thus, these indices are often missing an important component: they identify the population’s needs but not the availability of services to address those needs.

Finally, some important SDOH are rarely, if ever, captured. For example, the National Commission to Transform Public Health Data Systems notes that the domains of community and civic well-being are generally missing from existing data,²⁹ limiting the ability of policies to account for these important community factors.

A challenge with SDOH data is understanding what the appropriate uses are for specific elements and how to use data to effectively target interventions. Although the availability and awareness of SDOH data are increasing, additional efforts are needed to determine which measures or indices are appropriate for different uses, so that programs and policymakers have guidance when selecting data sources. Additional information on the development of SDOH data elements, their applicability to specific populations, their validity at varying geographic levels (i.e., census block, zip code, county), and their association with outcomes of interest would make these decisions easier.

Data on Structural Factors

Across the 3 types of data, structural factors are often the hardest to measure. Regardless, measuring structural discrimination as a component of public health data needs attention, as this is an underlying factor that drives many of the disparities in SDOH, HRSNs, and health outcomes. To clearly demonstrate and understand the mechanisms that produce these differences in health outcomes and craft appropriate policies to address these disparities, we will need data that include direct measures of structural discrimination. For example, a recent study reviewed a number of efforts to develop measures of structural racism and associate them with outcomes, but the authors find that much work remains to be done to establish universally accepted measures of structural racism.³¹

CONCLUSIONS

There has been a longstanding recognition that addressing social drivers of health is critical to improving the nation’s health.³² In recent years, the policy debate has recognized the need to address social drivers of health both to improve the health status of all Americans and to address the issues that disadvantage some populations in achieving optimal health. The Biden–Harris Administration has made health equity a top priority and has begun a number of initiatives to improve access to and quality of care as well as address SDOH and HRSNs.

The objective of these efforts is to minimize disparities in key indicators of health status among groups by eliminating systematic differences in the drivers of health, such as SDOH and HRSNs. These efforts will require better data to develop, implement, and evaluate policies. As described in this essay, numerous data sources have emerged in recent years, but there are gaps and inconsistencies that may inhibit evidence-based policymaking. It is time to develop and disseminate key indicators of structural factors, SDOH, and HRSNs that are standardized, can identify key groups of interest, and are available nationally and at local levels. Developing these data would require resources as well as addressing important issues of statutory requirements, confidentiality, and proprietary concerns. Importantly, succeeding will require engaging communities to understand what data should be collected, from whom, and how. Although progress has been made in recent years, additional opportunities remain to use existing data and plan for ongoing data system improvements to support public health efforts to promote health equity.