JAMA Health Forum – Health Policy, Health Care Reform, Health Affairs | JAMA Health Forum | JAMA Network
[Skip to Navigation]
Sign In
Table 1.  Definitions and Data Sources for Social Risk Factors
Definitions and Data Sources for Social Risk Factors
Table 2.  Identification of Hospitals Caring for a High Proportion of Patients With Social Risk Factors Across CMS Hospital Outcome and Payment Measures
Identification of Hospitals Caring for a High Proportion of Patients With Social Risk Factors Across CMS Hospital Outcome and Payment Measures
Table 3.  Combinations of Risk Factors That Account for Most Hospitals Caring for a High Proportion of Patients With Social Risk Factors
Combinations of Risk Factors That Account for Most Hospitals Caring for a High Proportion of Patients With Social Risk Factors
1.
HR34—21st Century Cures Act. 114th Congress (2015-2016). December 13, 2016. Accessed May 31, 2021. https://www.congress.gov/bill/114th-congress/house-bill/34/
2.
Joynt  KE, De Lew  N, Sheingold  SH, Conway  PH, Goodrich  K, Epstein  AM.  Should Medicare value-based purchasing take social risk into account?   N Engl J Med. 2017;376(6):510-513. doi:10.1056/NEJMp1616278 PubMedGoogle ScholarCrossref
3.
Buntin  MB, Ayanian  JZ.  Social risk factors and equity in Medicare payment.   N Engl J Med. 2017;376(6):507-510. doi:10.1056/NEJMp1700081 PubMedGoogle ScholarCrossref
4.
US Department of Health and Human Services; Office of the Assistant Secretary for Planning and Evalution. Report to Congress: social risk factors and performance under Medicare’s value-based purchasing programs. December 21, 2016. Accessed May 31, 2021. https://aspe.hhs.gov/pdf-report/report-congress-social-risk-factors-and-performance-under-medicares-value-based-purchasing-programs
5.
National Quality Forum. Risk adjustment for socioeconomic status or other sociodemographic factors: technical report. August 15, 2014. Accessed July 9, 2015. https://www.qualityforum.org/Publications/2014/08/Risk_Adjustment_for_Socioeconomic_Status_or_Other_Sociodemographic_Factors.aspx
6.
Institute of Medicine; National Academies of Sciences and Engineering.  Accounting for Social Risk Factors in Medicare Payment: Identifying Social Risk Factors. National Academies Press; 2016.
7.
Boyd  RW, Lindo  EG, Weeks  LD, McLemore  MR. On racism: a new standard for publishing on racial health inequities. Health Affairs Blog. July 2, 2020. Accessed May 31, 2021. https://www.healthaffairs.org/do/10.1377/hblog20200630.939347/full/
8.
US Department of Health & Human Services. Creation of new race-ethnicity codes and SES indicators for Medicare beneficiaries: chapter 3: creating and validating an index of socioeconomic status. Agency for Healthcare Research and Quality publication 08-0029-EF. January 2008. Accessed June 30, 2015. https://archive.ahrq.gov/research/findings/final-reports/medicareindicators/medicareindicators3.html
9.
American Hospital Association. AHA Annual Survey database fiscal year 2013. Accessed June 26, 2015. https://www.ahadata.com/aha-annual-survey-database
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    Views 1,492
    Original Investigation
    July 2, 2021

    Identification of Hospitals That Care for a High Proportion of Patients With Social Risk Factors

    Author Affiliations
    • 1Center for Outcomes Research and Evaluation, Yale–New Haven Health System, New Haven, Connecticut
    • 2Boston University School of Public Health, Boston, Massachusetts
    • 3Department of Emergency Medicine, Yale University School of Medicine, New Haven, Connecticut
    • 4Frank H. Netter School of Medicine, Quinnipiac University, Hamden, Connecticut
    • 5Section of Rheumatology, Department of Medicine, Yale University School of Medicine, New Haven, Connecticut
    • 6Department of Medicine, Division of Rheumatology, Veterans Affairs Connecticut Health System, New Haven, Connecticut
    • 7Division of General Internal Medicine, Department of Medicine, Yale University School of Medicine, New Haven, Connecticut
    • 8Yale University School of Public Health, New Haven, Connecticut
    JAMA Health Forum. 2021;2(7):e211323. doi:10.1001/jamahealthforum.2021.1323
    Key Points

    Question  Are hospitals that care for a high proportion of patients with social risk factors consistently identified using varied definitions of these factors?

    Findings  In this cross-sectional study, among 4465 US hospitals qualified for Centers for Medicare & Medicaid Services hospital performance measures, one-third were identified as caring for a high proportion of patients with social risk factors across 7 definitions of social risk; fewer than 1% met all 7 definitions. Most hospitals serving patients with social risk factors could be identified using 3 or 4 definitions.

    Meaning  Inconsistencies in identifying hospitals caring for high proportions of patients with social risk factors suggest value in developing a common definition of social risk.

    Abstract

    Importance  Hospitals can face significant clinical and financial challenges in caring for patients with social risk factors. Currently the Hospital Readmission Reduction Program stratifies hospitals by proportion of patients eligible for both Medicare and Medicaid when calculating payment penalties to account for the patient population. However, additional social risk factors should be considered.

    Objective  To evaluate 7 different definitions of social risk and understand the degree to which differing definitions identify the same hospitals caring for a high proportion of patients with social risk factors.

    Design, Setting, and Participants  Across 18 publicly reported Centers for Medicare & Medicaid Services (CMS) hospital performance measures, highly disadvantaged hospitals were identified by the the proportion of patients with social risk factors using the following 7 commonly used definitions of social risk: living below the US poverty line, educational attainment of less than high school, unemployment, living in a crowded household, African American race (as a proxy for the social risk factor of exposure to racism), Medicaid coverage, and Agency for Healthcare Research and Quality index of socioeconomic status score. In this cross-sectional study, social risk factors were evaluated by measure because hospitals may serve a disadvantaged patient population for one measure but not another. Data were collected from April 1, 2014, to June 30, 2017, and analyzed from July 25, 2019, to April 25, 2021.

    Main Outcomes and Measures  The proportion of hospitals identified as caring for patients with social risk factors using 7 definitions of social risk, across 18 publicly reported CMS hospital performance measures.

    Results  Among 4465 hospitals, a mean of 31.0% (range, 28.9%-32.3%) were identified at least once when using the 7 definitions of social risk as caring for a high proportion of patients with social risk factors. Among all hospitals meeting at least 1 definition of social risk, a mean of 0.7% (range, 0%-1.0%) were identified as highly disadvantaged by all 7 definitions. Among hospitals meeting at least 1 definition of social risk, a mean of 2.7% (range, 1.3%-5.1%) were identified by 6 definitions; 6.5% (range, 5.9%-7.1%), by 5 definitions; 10.4% (range, 9.5%-12.1%), by 4 definitions; 13.2% (range, 10.1%-14.4%), by 3 definitions; 21.4% (range, 20.1%-22.4%), by 2 definitions; and 45.2% (range, 42.6%-47.1%), by only 1 definition. This pattern was consistent across all 18 performance measures.

    Conclusions and Relevance  In this cross-sectional study, there were inconsistencies in the identification of hospitals caring for disadvantaged populations using different definitions of social risk factors. Without consensus on how to define disadvantaged hospitals, policies to support such hospitals may be applied inconsistently.

    Introduction

    The 2016 21st Century Cures Act states that hospitals assessed by Centers for Medicare & Medicaid Services (CMS) outcome and payment measures are assigned to groups to allow for separate comparison of hospitals by their overall proportion of patients eligible for both Medicare and Medicaid (dual eligibility).1 This peer-group stratification of hospitals is used to assign a payment penalty threshold that is different for each group of hospitals. Using dual eligibility to group hospitals may address concerns that pay-for-performance programs do not account for social risk. Consideration of the social risk profile of a provider’s patients to classify providers in pay-for-performance programs will account for factors such as patient social support and functional status that may affect patient outcomes, hospital performance, and the resulting financial penalties.2 By financially penalizing hospitals without accounting for these social risk factors, there is a chance that providers caring for patients with social risk who are penalized will have fewer needed resources for improving quality, potentially accelerating disparities.3 However, a challenge in incorporating social risk into quality and payment measures is that no consensus among experts exists as to which social risk factors are best for identifying and stratifying providers.

    Hospital-level indices for social risk are a valuable way to assess the population that a hospital serves. Such approaches are currently used for a variety of purposes. They are used by researchers to identify safety net hospitals to understand quality and safety, financial resources, and disparities. As noted above, they are also used to identify hospitals to assess performance and target resources.2 The Office of the Assistant Secretary for Planning and Evaluation supported dual eligibility as “the most powerful predictor of poor outcomes.”4(p8) Although dual eligibility may include a high proportion of racial minority populations and low-income patients, it is not intended to be used as a composite measure for all social risk factors. For example, dual eligibility may not indicate someone’s educational attainment or whether they are living in a crowded household. Furthermore, dual eligibility is affected by state-to-state variations in Medicaid eligibility and enrollment, thus reducing ability to compare hospital service of disadvantaged populations between states. Experts have flagged additional social risk factors. The National Quality Forum5 prioritized social risk factors, including income, educational attainment, race, primary language, and patient living environment, for potential use in measuring risk adjustment or stratification. The National Academy of Medicine identified socioeconomic position; race, ethnicity, and cultural context; sex; social relationships; and residential and community context as social risk factors that affect health outcomes.6 The objective of this study was to evaluate 7 different definitions of social risk to understand the degree to which differing definitions identify the same hospitals.

    Methods

    This cross-sectional study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. This work received Yale University Institutional Review Board approval under a protocol for research on CMS data. The Human Investigation Committee at Yale University approved an exemption for this study to use CMS claims and enrollment data and waived the requirement for informed consent because the research involved no more than minimal risk and could not be practicably performed without the waiver.

    This cross-sectional study included US hospitals that publicly reported 1 or more of the 18 CMS 2018 hospital readmission, mortality, complication, and payment measures, which include data from April 1, 2014, to June 30, 2017. For each measure, highly disadvantaged hospitals were identified as those caring for a high proportion of patients with social risk factors. Highly disadvantaged hospitals were compared using 7 different definitions of social risk. A social risk factor was deemed useful to test based on (1) established face and/or empirical validity as a measure of social risk (eg, shown to be associated with greater risk of care disparities and/or worse patient outcomes) and (2) availability in CMS administrative data or publicly available data to ensure universal assessment of patient social risk. These definitions included living below the US poverty line, educational attainment below high school, unemployment, living in a crowded household, African American race (included as a proxy for the social risk factor of exposure to racism),7 Medicaid coverage, and Agency for Healthcare Research and Quality (AHRQ) index of socioeconomic status (SES) score.8 The first 6 social risk factors were chosen based on their relevance in other research and availability in national data sources.5,6 The AHRQ index of SES score was chosen because it is a weighted composite score of the first 4 social risk factors. For all the social risk factors, except for race, a patient is considered to have a social risk if they are in the 75th percentile (25th for AHRQ index of SES) on a continuous scale. For example, to identify patients with a low AHRQ index of SES, the 25th percentile of the national distribution of AHRQ index of SES scores was used. Patients with scores below the 25th percentile were considered to have low AHRQ index of SES.

    To calculate a hospital’s proportion of patients who live below the US poverty line, possess educational attainment below high school, are unemployed, and live in crowded households, data from the American Community Survey 2013-2017 5-year estimate, as well as Medicare Part A Inpatient Claims (2016), were used. The proportion of African American patients treated at a hospital was calculated using data from Medicare Part A Inpatient Claims, which captures patient self-reported race. The proportion of Medicaid patients for each hospital was calculated using data from the American Hospital Association Survey from 2016.9 The AHRQ index of SES score used data from both the American Community Survey and Medicare Part A claims. The American Community Survey data used zip code–level data for calculating risk factors for each patient, whereas Medicare Part A Inpatient Claims uses patient-level data to identify social risk factors for the hospital population (Table 1).

    Hospitals were evaluated as serving a high proportion of patients with social risk factors by each measure because the patient population and number of hospitals included in the calculation differed by quality and payment measure. Therefore, a hospital may serve a high proportion of patients for a given social risk factor for one measure but not another. To identify a hospital as highly disadvantaged for a given social risk factor, the proportion of patients with each of the social risk factors was calculated for each hospital. For each measure, hospitals with fewer than 25 admissions were excluded because there is not enough information to calculate hospital performance, and therefore these hospitals are not included in payment programs (for that measure). Hospitals were ranked from the lowest proportion of patients with the social risk factor of interest to the highest proportion within each measure. Hospitals that ranked in the top decile for each measure were considered to have a high proportion of patients for that social risk factor. This approach was used to account for varying numbers of hospitals reporting each measure. Then the number of definitions that identified a given hospital as having a high proportion of patients with social risk factors (0-7 social risk factor definitions) was counted. Hospitals were categorized as highly disadvantaged if their patient population was identified in the top decile of 1 or more social risk factors. Of those highly disadvantaged hospitals, the percentage of hospitals identified by only 1 through all 7 social risk factors for each outcome and payment measure was calculated and summarized as a mean across measures. This process was repeated for each of the 7 social risk factors to understand how hospitals are identified with various social risk factors across multiple measures.

    An additional analysis was completed to understand whether any combination of risk factors could account for most hospitals that were identified as caring for a high proportion of patients with social risk factors by at least 1 factor. For each measure, the combination of 3 and the combination of 4 risk factors that identified the most hospitals serving patients with social risk factors were calculated. This study consisted solely of descriptive statistics (ie, frequencies and percentages) and therefore did not include statistical tests as part of the analysis. Data were collected from April 1, 2014, to June 30, 2017, and analyzed from July 25, 2019, to April 25, 2021.

    Results

    In the US, 4465 hospitals were evaluated using one of the CMS mortality, readmission, complication, or payment measures. Almost one-third (mean, 31.0% [range, 28.9%-32.3%] across measures) of all hospitals were identified as highly disadvantaged (Table 2). Of these hospitals, almost one-half (mean, 45.2% [range, 42.6%-47.1%] across measures) were identified by only 1 risk factor. As the number of social risk factors used to identify highly disadvantaged hospitals increased, fewer hospitals were identified. When the mean was taken across measures, among the hospitals meeting at least 1 definition of social risk, a mean of 0.7% (range, 0%-1.0%) of hospitals were identified by all 7 definitions; 2.7% (range, 1.3%-5.1%), by 6 definitions; 6.5% (range, 5.9%-7.1%), by 5 definitions; 10.4% (range, 9.5%-12.1%), by 4 definitions; 13.2% (range, 10.1%-14.4%), by 3 definitions; and 21.4% (range, 20.1%-22.4%), by 2 definitions.

    When evaluating combinations of risk factors that could account for most hospitals that were identified as caring for a high proportion of patients with social risk factors (Table 3), a mean of 74.9% (range, 73.0%-77.2%) of hospitals across measures could be identified by a combination of 3 risk factors, and a mean of 87.7% (range, 85.7%-89.6%) of hospitals could be identified by a combination of 4 risk factors. The combination of 3 risk factors that was most common for identifying the greatest proportion of hospitals identified by at least 1 social risk factor consisted of African American race, low educational attainment, and Medicaid coverage. The combination of 4 risk factors that was most common for identifying the greatest proportion of hospitals identified by at least 1 social risk factor was African American race, AHRQ index of SES score, living in a crowded household, and Medicaid coverage. Living below the poverty line was the only other factor that was included as part of a 3– or 4–risk factor combination.

    Discussion

    In this cross-sectional study, approximately one-third of hospitals were identified as highly disadvantaged hospitals by at least 1 definition of social risk. Almost one-half of hospitals were only identified by 1 social risk definition, and less than 1% were identified by all 7 definitions. This supports our hypothesis that each definition identifies a different set of hospitals. These findings are critical to the discussion about how to address social risk in the CMS pay-for-performance programs. Recent legislation advises CMS to stratify hospitals for payment penalties by their proportion of patients with dual eligibility. However, our results suggest that a single definition of social risk may miss hospitals with significant populations of at-risk patients. This may limit the hospitals that are helped by this change in policy.

    The findings of this study show that there is no single best definition of social risk that can be used to identify hospitals serving disadvantaged populations, but rather each definition of social risk identifies a different group of hospitals. These social risk factors can represent important different vulnerabilities and may be valuable to use in different contexts. Therefore, the best definition of social risk is highly dependent on the application of the social risk index, be it for research or payment policy. Using a combination of 3 or 4 risk factors may allow a more robust and parsimonious identification of hospitals serving high proportions of patients with social risk factors.

    Limitations

    A limitation of this study is that the top decile of hospitals for any definition of social risk was used to identify hospitals with a high proportion of patients for that social risk factor. If this criterion were relaxed, more hospitals would be identified as serving high proportions of patients with social risk factors, and therefore slightly more hospitals would be identified by multiple social risk factors. However, using deciles instead of quintiles or quartiles ensured consistent identification of hospitals truly caring for a disproportionate share of patients with social risk. In addition, there are other social risk factors that could be considered to identify hospitals that serve a high proportion of patients with social risk factors; however, this study used only social risk factors that were available as part of the existing data that met our criteria.

    Conclusions

    The findings of this cross-sectional study reveal variability in identifying hospitals caring for a disproportionate share of patients with social risk factors, supporting the need for continued research in this area. The use of multiple definitions of social risk may provide better insight into health care disparities than any individual definition of social risk.

    Back to top
    Article Information

    Accepted for Publication: April 28, 2021.

    Published: July 2, 2021. doi:10.1001/jamahealthforum.2021.1323

    Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2021 Matty R et al. JAMA Health Forum.

    Corresponding Author: Susannah M. Bernheim, MD, MHS, Division of General Internal Medicine, Department of Medicine, Yale University School of Medicine, 333 Cedar St, New Haven, CT 06510 (susannah.bernheim@yale.edu).

    Author Contributions: Dr Bernheim and Ms Barthel had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

    Concept and design: Matty, Heckmann, George, Suter, Bernheim.

    Acquisition, analysis, or interpretation of data: Matty, George, Barthel, Suter, Ross, Bernheim.

    Drafting of the manuscript: Matty, George, Barthel.

    Critical revision of the manuscript for important intellectual content: Heckmann, Barthel, Suter, Ross, Bernheim.

    Statistical analysis: Barthel.

    Obtained funding: Bernheim.

    Administrative, technical, or material support: George.

    Supervision: Suter, Bernheim.

    Conflict of Interest Disclosures: Ms Matty reported receiving salary support from the Centers for Medicare & Medicaid Services (CMS) to develop, implement, and maintain hospital performance outcome measures, including those related to this report, that are publicly reported. Dr Heckmann reported receiving salary support from the CMS to develop, implement, and maintain hospital performance outcome measures, including those related to this report, that are publicly reported, in addition to receiving research support from the FDA as part of a Yale-Mayo Clinic Center for Excellence in Regulatory Science and Innovation program through Yale as part of a Centers for Disease Control and Prevention project designed to strengthen prescription drug overdose prevention efforts, from Connecticut Department of Public Health as part of a public health project designed to assess the impact of Good Samaritan Laws, and from the Community Health Network of Connecticut for her work as a medical consultant. Ms George reported receiving salary support from the CMS to develop, implement, and maintain hospital performance outcome measures, including those related to this report, that are publicly reported. Ms Barthel reported receiving salary support from the CMS to develop, implement, and maintain hospital performance outcome measures, including those related to this report, that are publicly reported. Dr Suter reported receiving salary support from the CMS to develop, implement, and maintain hospital performance outcome measures, including those related to this report, that are publicly reported. Dr Ross reported receiving salary support from the CMS to develop, implement, and maintain hospital performance outcome measures, including those related to this report, that are publicly reported; receiving research support through Yale University from Medtronic, Inc, and the US Food and Drug Administration (FDA) to develop methods for postmarket surveillance of medical devices and from the Blue Cross Blue Shield Association to better understand medical technology evaluation, through Yale University from Johnson & Johnson to develop methods of clinical trial data sharing, and from the FDA to establish the Yale–Mayo Clinic Center for Excellence in Regulatory Science and Innovation program; and receiving grants from the Medical Device Innovation Consortium as part of the National Evaluation System for Health Technology, the Agency for Healthcare Research and Quality, the National Heart, Lung and Blood Institute of the National Institutes of Health, and from the Laura and John Arnold Foundation to establish the Good Pharma Scorecard at Bioethics International and the Collaboration for Research Integrity and Transparency at Yale. Dr Bernheim reported receiving salary support from the CMS to develop, implement, and maintain hospital performance outcome measures, including those related to this report, that are publicly reported, and Humana, Inc, to advise on quality strategy.

    Funding/Support: This work was supported by grant HHSM-500-2013-13018I from the CMS.

    Role of the Funder/Sponsor: The funding organization had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript. The funding organization approved the decision to submit the manuscript for publication.

    References
    1.
    HR34—21st Century Cures Act. 114th Congress (2015-2016). December 13, 2016. Accessed May 31, 2021. https://www.congress.gov/bill/114th-congress/house-bill/34/
    2.
    Joynt  KE, De Lew  N, Sheingold  SH, Conway  PH, Goodrich  K, Epstein  AM.  Should Medicare value-based purchasing take social risk into account?   N Engl J Med. 2017;376(6):510-513. doi:10.1056/NEJMp1616278 PubMedGoogle ScholarCrossref
    3.
    Buntin  MB, Ayanian  JZ.  Social risk factors and equity in Medicare payment.   N Engl J Med. 2017;376(6):507-510. doi:10.1056/NEJMp1700081 PubMedGoogle ScholarCrossref
    4.
    US Department of Health and Human Services; Office of the Assistant Secretary for Planning and Evalution. Report to Congress: social risk factors and performance under Medicare’s value-based purchasing programs. December 21, 2016. Accessed May 31, 2021. https://aspe.hhs.gov/pdf-report/report-congress-social-risk-factors-and-performance-under-medicares-value-based-purchasing-programs
    5.
    National Quality Forum. Risk adjustment for socioeconomic status or other sociodemographic factors: technical report. August 15, 2014. Accessed July 9, 2015. https://www.qualityforum.org/Publications/2014/08/Risk_Adjustment_for_Socioeconomic_Status_or_Other_Sociodemographic_Factors.aspx
    6.
    Institute of Medicine; National Academies of Sciences and Engineering.  Accounting for Social Risk Factors in Medicare Payment: Identifying Social Risk Factors. National Academies Press; 2016.
    7.
    Boyd  RW, Lindo  EG, Weeks  LD, McLemore  MR. On racism: a new standard for publishing on racial health inequities. Health Affairs Blog. July 2, 2020. Accessed May 31, 2021. https://www.healthaffairs.org/do/10.1377/hblog20200630.939347/full/
    8.
    US Department of Health & Human Services. Creation of new race-ethnicity codes and SES indicators for Medicare beneficiaries: chapter 3: creating and validating an index of socioeconomic status. Agency for Healthcare Research and Quality publication 08-0029-EF. January 2008. Accessed June 30, 2015. https://archive.ahrq.gov/research/findings/final-reports/medicareindicators/medicareindicators3.html
    9.
    American Hospital Association. AHA Annual Survey database fiscal year 2013. Accessed June 26, 2015. https://www.ahadata.com/aha-annual-survey-database
    ×