Laura M. Gottlieb, MD, MPH

 Health care payers are increasingly experimenting with interventions to address social risk factors. With enactment of the 2018 Bipartisan Budget Act, Medicare Advantage (MA) plans have new opportunities to offer supplemental benefits that are not “primarily health-related.” This article presents findings from interviews conducted with executives from MA plans regarding plan decision-making processes related to new social risk factor benefits.

 Semistructured qualitative interviews with MA plan leadership.

 A total of 63 plan representatives from 29 unique MA plans were interviewed about the rationale for social risk–related interventions and how data are used to inform benefits expansion decisions. This paper combines qualitative interview data from 2 separate studies with similar target groups and interview guides. Interview transcripts were qualitatively analyzed to examine underlying themes.

Three main themes emerged: (1) Plans use multiple data sources to determine how to target benefits; (2) evidence gaps hinder decision-making to expand or offer new supplemental benefits; and (3) in the absence of sufficient evidence, some plans have their own research and quality improvement processes to maximize effectiveness.

 Findings provide insights about opportunities and challenges that MA plans face in making decisions related to supplemental benefits designed to address members’ social risk factors. Barriers include collecting, generating, and analyzing data critical to informing investments. Results highlight the need to ensure interoperability of new and existing data sources, foster shared learning opportunities, and narrow evidence gaps about specific social care interventions to inform the design and implementation of effective supplemental benefits.

Despite enthusiasm around expanding Medicare Advantage plans’ supplemental benefits to social care programs, innovation or activity has been slow.

Plans use multiple data sources to determine how to target benefits, and evidence gaps persist, which hinder decision-making to offer new supplemental benefits.

In the absence of sufficient evidence, plans have their own research and quality improvement processes to maximize effectiveness.

These findings highlight the need to ensure interoperability of new and existing data sources, foster shared learning opportunities, and narrow evidence gaps about specific social care interventions to inform the design and implementation of supplemental benefits.

As evidence about the impact of social drivers of health grows, opportunities to address patients’ social and economic adversity—such as food and housing insecurity and the lack of social connections

—have increased across the US health care sector. Medicaid plans have been at the leading edge of developing payment models that incentivize or otherwise support social care, including expanding opportunities for plans and affiliated providers to provide nonmedical services and supports related to food and housing.

 New opportunities are emerging in Medicare

 to address the barriers to achieving good health that many older adults experience, such as poverty,

The 2018 Bipartisan Budget Act, which included the Creating High-Quality Results and Outcomes Necessary to Improve Chronic Care Act (referred to as the CHRONIC Care Act),

 and the 2018 CMS call letter (an annual set of guidelines and clarifications for Medicare Advantage [MA] plans) announced greater flexibility for MA plan supplemental benefits to address social drivers of health. MA plans, which currently enroll one-third of all Medicare beneficiaries,

 are required to cover traditional Medicare services, but they may also cover services not traditionally available under Medicare. MA plans now have further flexibility in offering supplemental benefits. For the first time, the federal government has authorized MA plans to offer supplemental benefits that are not “primarily health-related”

 and has given MA plans more room and discretion to design supplemental benefits that “have a reasonable expectation of improving or maintaining the health or overall function”

Criteria for benefit structure were also expanded, such that benefits should meet at least 1 of the following criteria: diagnose, prevent, or treat an illness or injury; compensate for physical impairments; act to ameliorate the functional or psychological impact of injuries or health conditions; or reduce avoidable health care utilization.

 Benefit offerings under the expanded structure can be diverse, including meal delivery, home modifications, personal support services, and cooking classes,

 and they can be targeted toward groups at highest risk of poor health, as the CHRONIC Care Act waived the requirement of uniform access to benefits.

 Plans therefore have the option to target these benefits to enrollees with specific chronic conditions.

 Importantly, insurers do not receive additional funds to offer supplemental benefits. Supplemental benefits can be paid through charging all MA enrollees a premium, through cost sharing, or through a plan’s rebate dollars.

Our previous research has found that adoption of new or expanded offerings among MA plans has been gradual and limited.

 However, research has yet to examine the underlying reasons for this slow uptake. Although MA plans have a role in addressing the social risks of their members, little is known about how plans address these risks and how they collect data to inform their approaches to addressing social risk. This paper addresses this gap in the literature by explicitly exploring MA plan decision-making, including how plans anticipate defining eligibility parameters around at-risk populations and how plans collect and apply social risk data to improve health outcomes.

This paper combines data from 2 separate qualitative studies (study 1 and study 2), both of which involved interviews with MA plan leaders. Study 1 examined how MA plans address the health of MA plan members; how they design, implement, and assess the efficacy of new programs and benefits; and plans’ receptivity to alternative payment models. Study 2 explored how MA plans were using the additional flexibility offered by CMS to direct supplemental benefits toward supports that address health-related social risk factors in their 2020 bid cycles and what shaped the plans’ decisions to offer new or expanded benefits. Although these 2 studies were funded, designed, and conducted independently, interview topics described in this report overlapped. In total, the 2 studies involved interviews with 63 leaders from 29 different MA plans. Recruitment and data collection from each of the 2 studies and the combined analysis methods are described below.

The research teams conducted semistructured interviews with health plan leaders in 2 qualitative studies. Study 1 interviewed 38 representatives from 17 plans; study 2 interviewed 25 representatives from 14 plans, yielding 31 total plans. Two plans overlapped across both studies, accounting for 16 of the 63 total participants. The 29 unique plans together cover 75% of the national MA population. The study 1 research team employed a purposive and snowball sampling strategy. Initially, participants from plans of varying size, geographic location, and quality rating were purposively recruited, and interview participants were asked for recommendations of potential participants from other plans at the completion of each interview. The study 2 project team used a random sampling approach to identify plans from a complete list of MA plans nationally, ensuring representation of plans from the following categories: geographical coverage and location, special needs plan (SNP) vs non-SNP, tax status (nonprofit vs for profit), and plan type (eg, health maintenance organizations, private fee-for-service plans). In instances in which the study team received no response to a request for an interview after multiple outreach attempts, the next plan was randomly selected from the list of MA plans. Twice when initial outreach attempts failed, plan representatives were contacted through professional network connections using a convenience sampling approach.

Interview guides in both studies included questions focused on health plans’ responses to the CHRONIC Care Act and on factors shaping plan decision-making around offering new or expanded supplemental benefits. Common interview guide topics included organizational strategy and work regarding social risk factors; eligibility for benefits that address social risk factors; current and future plans around expanding or offering new supplemental benefits that address social risk factors; new benefit decision-making processes; evaluation plans and research capacity; and barriers and facilitators to new or expanded benefit offerings. Both interview guides are included in the 

). Both study teams piloted their respective interview guides with leaders from a health plan, and both guides were subsequently refined for clarity.

Recruitment and interviews were conducted via telephone between July and November 2018 (study 1) and between April and November 2019 (study 2). The study 1 team conducted interviews immediately after the CHRONIC Care Act was passed; interviews for study 2 were conducted while organizations were planning 2020 bids in response to the CHRONIC Care Act. This approach allows for a longer range of time from which to understand MA plans’ decisions regarding supplemental benefits and provides an opportunity to understand more anticipated responses from plans (study 1) and more real-time responses from MA plans in the midst of supplemental benefits planning and implementation (study 2).

In both studies, interviews were conducted with executive leadership team members (eg, chief medical officer, chief executive officer, chief organizational officer, department heads, or institutional focal points for enrollees’ social risk factors) who could speak to overall strategy and mission related to addressing social risk factors, as well as to the plans’ bid processes. When more than 1 representative was interviewed in any given plan, group interviews were conducted. Of all plans approached in study 1, 12 plans did not respond to the request for interview and 2 refused to participate. In study 2, 5 plans did not respond to the request for interview and 1 refused to participate.

Interview participants were approached by email or phone with a request to participate in an interview focusing on MA and social risk factors. Interviews were conducted by 1 to 2 members of each research team by phone. Interviews ranged between 30 and 60 minutes; the mean interview length was 45 minutes. Participants provided verbal consent to record the interview, and the audio recordings were transcribed before analysis. The institutional review boards at the study 1 and study 2 institutions exempted the qualitative study from ethical review, as it was not considered human subjects research.

Both teams followed a similar content analysis approach for initial transcript review.

 Initial coding trees based on the study-specific interview guide topics were discussed using a consensus-based team approach to identify major concepts. Two coders individually coded each transcript, met to reconcile codes, and drafted a master codebook for use in the next stage of coding. Audit trails of ongoing decisions including revision of existing codes and emergent codes were kept for both sets of interviews to ensure analytic rigor.

For this cross-study analysis, 2 members from each research team reviewed a random sample of the partner study’s transcripts and engaged in open coding. After developing a preliminary codebook for themes emerging from sets of interviews, all 4 coders reviewed the separate codebooks for relevant overlapping codes and discussed reasoning for any discrepancies before reaching a consensus on a shared codebook. Coded excerpts from interviews were shared with study partners for review and analysis.

To ensure anonymity of the participating MA plans, organizations were identified by unique numbers assigned in the study. The label “ID” for identifier represents a unique MA plan, and ID numbers were assigned to study 1 and study 2 participants. This study is reported in accordance with the Consolidated Criteria for Reporting Qualitative Research reporting guideline.

Three main themes emerged across both studies: (1) Plans use multiple data sources to determine how to target benefits; (2) evidence gaps hinder decision-making to develop new or expand existing supplemental benefits, and (3) in the absence of sufficient evidence, participants recognize they will need to implement their own research and quality improvement processes to maximize effectiveness. These themes are described below, and supportive quotes are included in 

Plans Use Multiple Data Sources and Considerations to Determine How to Target Benefits

Plan representatives described using multiple sources of data, including patient-reported data, patient-level purchased consumer data, and population-level data, to inform decisions about eligibility and services. Some plans use both individual and population-level data sources to stratify risk and determine eligibility for social care benefits. These methods included building predictive models using different social risk factors (eg, food insecurity), behavioral risk factors (eg, depressive symptoms), and medical risk factors (eg, chronic renal failure) in efforts to proactively identify “at-risk” enrollees (ID6, study 1), as well as hot-spotting techniques combining socioeconomic, medical, and demographic data (ID17, study 1). Actively screening for social risk factors in different health care settings was another strategy used to gauge members’ needs (ID6, study 1). Informants also described using community needs assessments as important data sources to better understand plan membership in a particular coverage area (ID13, study 1). Some plan representatives mentioned exploring relationships with community-based organizations to acquire and leverage data points to more accurately identify community needs and social risk factors (as noted by ID7, study 2). Several plan representatives mentioned taking the opportunity to build data ecosystems or data banks that pull in data from multiple sources to target member needs more accurately, leveraging data sources such as claims data, provider data, or third-party vendor data (ID1; ID7, both study 2). See Table 1 for supportive quotes.

Evidence Gaps Hinder Decision-Making to Expand or Offer New Supplemental Benefits

Although plan representatives acknowledged the importance of evidence-based approaches to decision-making, they highlighted the need for better data to build the evidence base around which social needs to target and what interventions are effective in addressing social risk factors. Plan representatives recognized that some of the data generation linking social needs, interventions, and outcomes needs to happen externally and is beyond the scope of what plans can realistically do themselves. Participants articulated that the current national evidence base is not robust enough to inform plan decisions. In particular, plan representatives expressed the challenge of obtaining “cause and effect” data to inform investment decisions (ID1, study 2). Plan representatives noted compelling evidence linking food insecurity and housing instability to health outcomes (ID8, study 2), which facilitated internal decisions related to including benefits targeting these 2 social risk factors.

Plan representatives mentioned the difficulty in the health care domain of attributing which inputs generate improvements in health outcomes and the lack of evidence to support the granular data that are needed to effectively guide plan decision-making. The challenge of investing resources using rebate dollars was highlighted among several interviews. Rebate dollars are one way to finance supplemental benefits, and typically rebate dollars are invested specifically into benefits that will have the greatest impact on health outcomes and generate a return on investment (ROI) (ID14, study 1). Some representatives described having to make more pragmatic decisions in the absence of data, basing decisions to offer new or expanded benefits more on whether it would be a market opportunity for the plan, such as whether a certain benefit will be competitive in a particular market segment or have high uptake among enrollees (ID16, study 1; ID14, study 2). One plan representative expressed the need for more guidance or improved tools to calculate the potential ROI of interventions targeting social risk factors to synthesize emergent evidence (ID10, study 2). See Table 2 for supportive quotes.

In the Absence of Sufficient Evidence, Participants Recognize They Will Need to Implement Their Own Research and Quality Improvement Processes to Maximize Effectiveness

To address the challenge of limited data, MA plans are developing pilot programs to generate and test the feasibility of implementation and new benefit effectiveness as a means to shape future benefit design. Plan representatives noted that internal research and quality improvement processes are important to have in place to maximize learning from experimentation in offering benefits that address social risk factors. Plan representatives described this as the process of learning from what they are doing, tweaking processes and benefit offerings accordingly, and measuring outcomes (ID1, study 1).

Several plan representatives shared that their plan had a systematic internal procedure to review evidence and impact to steer modifications to benefit offerings (as noted by ID12, study 1); however, many plans lacked the ability or resources to build internal research capacity. Plan representatives with internal research capacity and infrastructure emphasized that data on social needs have only very recently been collected internally and that it takes time to build a credible data set on which actuaries base their decisions (ID11, study 2). Correspondingly, plan representatives noted that a longer time frame is needed to show an ROI when providing a benefit that addresses a social need compared with other outcomes such as preventing a hospital admission (ID1, study 2).

Operational and implementation data were noted as important information sources for MA plans, including uptake rates, acceptability of the benefit to members, and member engagement with the benefit over time (ID1, study 1; ID8, study 2). These data inform investment decisions or benefit modifications that are needed, as these are more granular data on whether a certain benefit works in reducing social risk factors among one target population more than another. Some plan representatives stated that once the benefit is offered, data collection processes are put in place to monitor uptake of the benefit among members and the impact of the benefit on health outcomes such as emergency department visits. In particular, one plan representative (ID7) from study 2 shared an example that internal monitoring of the implementation of a new meal benefit demonstrated low member uptake, leading the plan to only gradually expand the benefit and monitor utilization rates in 2019. See Table 3 for supportive quotes.

We leveraged 2 unique studies that involved interviews with MA plan executives to better understand plan decision-making around supplemental benefits to address social risk factors. Despite enthusiasm among plan representatives to expand benefits to social care programs, innovation or activity around the new supplemental benefits is slow, consistent with hypotheses and findings of several recent reports.

 There may be several reasons for the slow uptake, including lack of evidence, which in part stems from lack of data; not enough clarity and guidance from CMS about what is allowed; and insufficient time to develop thoughtful proposals.

Paucity of evidence around how to use existing data in determining which risk factors to address and for which populations within the member constituency is a feasible reason for plan hesitation to structure new benefits. This hesitation is reflected in the low 4.6% adoption rate of new special supplemental benefits for the chronically ill among MA plans in 2020.

 A recent analysis found that the largest areas of growth within benefit expansion were pest control, meal programs, and produce.

 Given the evidence gaps and lack of additional funding for providing new or expanded supplemental benefits, there may be hesitation among plans to invest additional resources into new benefits.

 MA plan representatives stated that limited evidence was available to guide decision-making about programs or benefits that address health-related social risk factors, including ROI in terms of both financial and health impacts. This has also been reflected in other recent studies among MA plan leadership.

 MA plans identified that both financial ROI and health impact ROI were important factors in decision-making about which services to offer or provide related to social determinants of health. Recent (2021) guidance from CMS has noted that spending on nonmedical services is generally expected to produce health returns and therefore lower health spending. Although the evidence base was considered strong for some health-related social risk factors (eg, food insecurity), there was less evidence on the operational implementation of these benefits and what the returns were financially or healthwise and over what time horizon. MA plan representatives stated that they need more granular, tailored evidence to make financial investments, with some representatives indicating that they were collecting some of this information internally.

Plan leaders also noted that gaps in the availability of data limited their ability to make informed decisions. A number of plan representatives in this study explained that they contract out services for data collection, monitoring, and evaluation efforts, particularly for evaluating pilot programs. It could also be valuable to set requirements for consistent and reliable MA plan encounter data for plans to learn to use data to drive decisions around modifications to benefit offerings to ensure better population reach or uptake and better outcomes. To advance this effort, consensus is needed on standardization of measures and metrics to drive plan decisions.

Although lack of data and financial resources may partially explain slow uptake of the new supplemental benefits, other reasons may be the relatively short period of time between the call announcement and the proposal deadline,

 a challenge reflected in several interviews in this study. CMS has an important role to play in setting requirements and guidelines to steer MA plans in this new process. The ability for MA plans to learn from each other is also crucial, and currently there are limited shared data mechanisms under the MA program. CMS could be an important player in setting requirements for shared data and in monitoring and evaluating emerging data on these new supplemental benefits. Slow uptake of new supplemental benefits may result, in part, from hesitancy on the part of plans to adjust their roles. In our prior work that used interview data from study 1,

 we found that among MA plan representatives interviewed, although there was significant interest in identifying ways in which MA plans can have greater social and health impacts, there were differences of opinion regarding the best approach. Whereas some plan representatives expressed a desire to offer supplemental benefits to meet members’ holistic needs, others preferred to engage with community-based organizations to provide support for members’ social needs. This lack of consensus across MA plans may contribute to limited benefit uptake.

There are several limitations to this work. First, although we were able to sample leaders representing 75% of the MA market, our findings may not represent the entire MA market in the United States. Second, plans may not have shared the full picture of their decision-making processes with respect to offering benefits or strategies to address members’ social risk factors, and qualitative interviews provide only a part of the picture with respect to this fast-evolving payer policy landscape. A strength of employing qualitative methods is that it is the only way to understand the contextual factors of MA decision-making. Although the interview guides were not identical, the advantages of combining data outweighed that limitation because they enabled us to develop a data set reflecting the opinions and perspectives of a larger set of MA plan leaders. Our study is among the few to provide insights into how a large sample of MA plans make decisions to address members’ social risk factors in response to the CHRONIC Care Act and covering a longer time range.

The CMS rule and CHRONIC Care Act together have provided the MA market with more flexibility to address nonmedical needs of chronically ill members. Uptake of this flexibility has been slow, overall, among plans. Emphasizing limited effectiveness evidence, our participants expressed caution about making long-term investments in this area. Efforts to strengthen data availability and interoperability will improve the quality of research and thereby support MA investments in social care programs. A coordinated research agenda would help target future research efforts in ways that can support public and private payers looking to advance social care initiatives.


 Department of Family and Community Medicine, University of California San Francisco (LSS-Z, MT, LMG), San Francisco, CA; Brown University School of Public Health (EAG, JFB, KT), Providence, RI.

 This research project (study 1) has been made possible by a grant from the Non-Profit Finance Fund.

 Dr Shields-Zeeman received personal grant funding from The Commonwealth Fund via the 2018-2019 Harkness Fellowship in Healthcare Policy and Practice.Dr Gottlieb is involved in a collaborative research project with Humana. The remaining authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

 Concept and design (LSS-Z, EAG, LMG, KST); acquisition of data (LSS-Z, EAG, MT, JFB, LMG, KST); analysis and interpretation of data (LSS-Z, EAG, MT, JFB, LMG, KST); drafting of the manuscript (LSS-Z); critical revision of the manuscript for important intellectual content (LSS-Z, EAG, MT, JFB, LMG, KST); obtaining funding (LSS-Z, LMG, KST); administrative, technical, or logistic support (LSS-Z, JFB, LMG); and supervision (LSS-Z, LMG, KST).

 Laura S. Shields-Zeeman, PhD, MS, Netherlands Institute of Mental Health and Addiction, Da Costakade 45, 3521 VS, Utrecht, the Netherlands. Email: lshields-zeeman@trimbos.nl.

1. National Academies of Sciences, Engineering, and Medicine. 

Integrating Social Care into the Delivery of Health Care: Moving Upstream to Improve the Nation’s Health

. The National Academies Press; 2019. doi:10.17226/25467

2. Gottlieb LM, Alderwick H. Integrating social and medical care: could it worsen health and increase inequity? 

3. Alderwick H, Hood-Ronick CM, Gottlieb LM. Medicaid investments to address social needs in Oregon and California. 

. 2019;38(5):774-781. doi:10.1377/hlthaff.2018.05171

4. Braveman P, Gottlieb L. The social determinants of health: it’s time to consider the causes of the causes. 

. 2014;129(suppl 2):19-31. doi:10.1177/00333549141291S206

5. Thomas KS, Durfey SNM, Gadbois EA, et al. Perspectives of Medicare Advantage plan representatives on addressing social determinants of health in response to the CHRONIC Care Act. 

. 2019;2(7):e196923. doi:10.1001/jamanetworkopen.2019.6923

6. Get the facts on economic security for seniors. National Council on Aging. March 1, 2021. Accessed March 11, 2022. https://www.ncoa.org/article/get-the-facts-on-economic-security-for-seniors

7. Cudjoe TKM, Roth DL, Szanton SL, Wolff JL, Boyd CM, Thorpe RJ. The epidemiology of social isolation: National Health and Aging Trends Study. 

. 2020;75(1):107-113. doi:10.1093/geronb/gby037

8. National Academies of Sciences, Engineering, and Medicine. 

Social Isolation and Loneliness in Older Adults: Opportunities for the Health Care System

. The National Academies Press; 2020. doi:10.17226/25663

9. Ziliak JP, Gundersen C. The state of senior hunger in America in 2018. Feeding America. May 2020. Accessed August 6, 2020. https://www.feedingamerica.org/sites/default/files/2020-05/2020-The%20State%20of%20Senior%20Hunger%20in%202018.pdf

10. DeGood K. Aging in place, stuck without options: fixing the mobility crisis threatening the Baby Boom generation. Transportation for America. 2011. Accessed September 5, 2020. https://t4america.org/docs/SeniorsMobilityCrisis.pdf

11. DuGoff EH, Buckingham W, Kind AJH, Chao S, Anderson GF. Targeting high-need beneficiaries in Medicare Advantage: opportunities to address medical and social needs. 

12. Bipartisan Budget Act of 2018, Pub L No. 115-123, 132 Stat 64.

13. Neuman P, Jacobson GA. Medicare Advantage checkup. 

. 2018;379(22):2163-2172. doi:10.1056/NEJMhpr1804089

14. Medicare Advantage’s new supplemental benefit for 2019: plan views and responses. Long-Term Quality Alliance. 2018. Accessed September 5, 2020. http://www.ltqa.org/wp-content/themes/ltqaMain/custom/
images/LTQA-Report-on-MA-Flexible-Supplemental-Benefits-FINAL-11-9-18.pdf

15. Willink A, DuGoff EH. Integrating medical and nonmedical services—the promise and pitfalls of the CHRONIC Care Act. 

. 2018;378(23):2153-2155. doi:10.1056/NEJMp1803292

16. Creating High-Quality Results and Outcomes Necessary to Improve Chronic (CHRONIC) Care Act of 2017, S 870, 115th Cong, 1st Sess (2017).

17. CMS expands Medicare Advantage coverage for social determinants of health. PricewaterhouseCoopers. 2019. Accessed September 5, 2020. https://www.pwc.com/us/en/industries/health-industries/health-research-institute/medicare-advantage-coverage-for-social-determinants.html

18. Meyers DJ, Durfey SNM, Gadbois EA, Thomas KS. Early adoption of new supplemental benefits by Medicare Advantage plans. 

. 2019;321(22):2238-2239. doi:10.1001/jama.2019.4709

19. Meyers DJ, Gadbois EA, Brazier J, Tucher E, Thomas KS. Medicare plans’ adoption of special supplemental benefits for the chronically ill for enrollees with social needs. 

. 2020;3(5):e204690. doi:10.1001/jamanetworkopen.2020.4690

20. Noel-Miller C, Sung J. Supplemental benefits under Medicare Advantage: an in-depth look at what they are today. AARP. December 2019. Accessed September 5, 2020. https://www.aarp.org/content/dam/aarp/ppi/2019/07/medicare-supplement-series-one.doi.10.26419-2Fppi.00075.001.pdf

Qualitative Data Analysis: An Expanded Sourcebook

22. Tong A, Sainsbury P, Craig J. Consolidated criteria for reporting qualitative research (COREQ): a 32-item checklist for interviews and focus groups. 

. 2007;19(6):349-357. doi:10.1093/intqhc/mzm042

23. Rinaldo F, Altman M, Cannon K, et al. Slowing Medicare spending by optimizing late-life needs. 

24. Skopec L, Ramos C, Aarons J. Are Medicare Advantage plans using new supplemental benefit flexibility to address enrollees’ health-related social needs? Urban Institute. September 19, 2019. Accessed September 5, 2020. https://www.urban.org/research/publication/are-medicare-advantage-plans-using-new-supplemental-benefit-flexibility-address-enrollees-health-related-social-needs-0

Articles

This article presents findings from interviews conducted with executives from 29 Medicare Advantage plans regarding plan decision-making processes related to new social risk factor–related benefits.

How Medicare Advantage Plans Use Data for Supplemental Benefits Decision-Making

 Computable social risk factor phenotypes derived from routinely collected structured electronic health record (EHR) or health information exchange (HIE) data may represent a feasible and robust approach to measuring social factors. This study convened an expert panel to identify and assess the quality of individual EHR and HIE structured data elements that could be used as components in future computable social risk factor phenotypes.

 A 2-round Delphi technique included 17 experts with an in-depth knowledge of available EHR and/or HIE data. The first-round identification sessions followed a nominal group approach to generate candidate data elements that may relate to socioeconomics, cultural context, social relationships, and community context. In the second-round survey, panelists rated each data element according to overall data quality and likelihood of systematic differences in quality across populations (ie, bias).

Panelists identified a total of 89 structured data elements. About half of the data elements (n = 45) were related to socioeconomic characteristics. The panelists identified a diverse set of data elements. Elements used in reimbursement-related processes were generally rated as higher quality. Panelists noted that several data elements may be subject to implicit bias or reflect biased systems of care, which may limit their utility in measuring social factors.

 Routinely collected structured data within EHR and HIE systems may reflect patient social risk factors. Identifying and assessing available data elements serves as a foundational step toward developing future computable social factor phenotypes.

Computable phenotypes are measurements of patient conditions or characteristics that can be obtained from existing data by combining a defined set of variables and logical expressions. Routinely collected structured data within electronic health records and health information exchange systems are reflective of characteristics of social and economic well-being and thus may be amenable to use in social risk factor phenotype development. An expert panel identified and assessed structured data elements to support future development of social risk factor computable phenotyping.

Computable phenotypes represent an additional method of measuring patient social factors that leverages existing data sources and workflows.

Structured data elements used in reimbursement-related processes may be of the highest quality for use in phenotyping.

Currently collected structured data elements such as 

International Classification of Diseases, Tenth Revision

 Z codes and Logical Observation Identifiers Names and Codes are potentially susceptible to bias.

Computable phenotyping will require transforming or combining data elements into novel and potentially more informative measures.

Social risk factors include patients’ nonclinical, economic, and contextual characteristics that may adversely affect health.

 As important drivers of morbidity, mortality, utilization, and health care costs, social risk factors are important for health risk assessment and both individual and population health management.

 Specifically, social risk factor information may improve risk prediction models

 and identify patients in need of social services.

 Because of the potential value of social risk factor information, federal agencies, clinical organizations, and health system experts advocate for better collection and use of patient social risk factor information.

Despite the potential value of social risk factor information to individual patient care and population health management activities, health care organizations’ current methods to measure this information are fraught with challenges. Patient-facing social risk questionnaires have not been consistently validated,

 area-level measures (eg, zip code–level demographics) can mask heterogeneity across individuals and are prone to the ecological fallacy,

 and extracting free-text clinical documentation from electronic health records (EHRs) remains difficult for many organizations.

 As a result, any one of these methods may not be sufficient for health care organizations to collect the information necessary to make inferences about patients’ and populations’ social risk factors.

Computable social risk factor phenotypes derived from routinely collected structured EHR or health information exchange (HIE) data may be an alternative approach to measuring social factors.

 Computable phenotypes are representations of patient conditions or characteristics that can be obtained from EHR data by combining a defined set of variables and logical expressions.

 Data such as demographics, insurance information, billing histories, appointment status, emergency contacts, and language preferences exist in most EHRs. Although such data may not contribute to understanding patients’ clinical status, they are reflective of characteristics of social and economic well-being and thus may be amenable to use in social risk factor phenotypes. Additionally, using structured data elements already routinely collected as part of clinical and business operation workflows may mitigate the challenges of underutilization of screening surveys and diagnosis coding, additional data collection burden, and the technical implementation hurdles of natural language processing (NLP) for textual data. Moreover, when applied to HIE data, which combine patient data across organizations over time, robust computable social risk factor phenotypes may be constructed that reduce missing data challenges

 However, biomedical informatics and health services research have devoted little attention to the potential value of developing phenotypes from existing structured data for social risk factor measurement in favor of questionnaires, area-level data linkage, and NLP.

From existing work on computable phenotypes, we know that poor data quality

 and data that are inconsistently collected across patient populations may result in inaccurate and otherwise biased phenotypes.

 Therefore, as a foundational step, we convened an expert panel to identify and assess the quality of individual EHR and HIE data elements that might be useful, accurate, and unbiased components to include in future computable social risk factor phenotype development. Our work sets the stage for the future quantitative development of computable social risk phenotypes by providing expert insight to guide selection of candidate data elements.

We used a 2-round Delphi technique to identify and preliminarily evaluate structured data elements as candidates for use in future computable phenotype development.

We recruited 17 individuals (of 18 invitations) with in-depth knowledge of EHR and/or HIE data based on publications or practice experience in 1 or more of the following 3 areas: EHR or HIE technology management in a health care organization; clinical or operational practice that involved data collection; or EHR or HIE research. The majority of respondents were affiliated with research or academic medical institutions (n = 14) and the remainder were individuals in leadership roles at health information technology organizations (n = 3). The panel represented organizations located on the East Coast and West Coast and in the South and Midwest. Five of the panel members were physicians. Expert panel members received a financial incentive of $250 for participating in the focus group and follow-up survey. Panelists were split across 2 identification sessions that each followed a common protocol.

In advance of the identification sessions, we provided each panelist with a summary of the research objectives of identifying and assessing potential structured data elements, a description of the expectations, and a shared definition of social risk factors (ie, any patient-level nonclinical economic, contextual, and psychosocial characteristics and factors). Because computable phenotypes are useful if generalizable,

 we instructed panelists to focus on structured data elements that they would expect to be commonly available in EHR or HIE data. We asked panelists to exclude unstructured data elements (eg, clinical note or other text data), data elements requiring linkage to sources outside of typical EHR or HIE systems (eg, tax records), and patient-facing social risk factor questionnaires because those are not widely adopted. These restrictions were intended to prioritize potential structured data elements that would be widely available. Panelists’ ideas were not restricted to specific age groups. We provided this information to each panel member during a short preparatory phone call in advance of the identification sessions.

The study team (consisting of authors J.R.V., H.E.K., C.M., and C.A.H.) conducted two 90-minute group identification sessions (n = 8 and n = 9) via videoconference. In each of these sessions, we followed a nominal group approach in which each panelist, in turn, was asked to suggest a data element, until all ideas were exhausted. A research team member documented the ideas generated in real time and displayed them on screen during the session. To help organize the idea generation, when suggesting data elements, panelists were asked to categorize them into 1 or more broad categories of social risk factors

: socioeconomic status (eg, employment, financial, food insecurity/hunger, housing instability); cultural context (eg, language, health literacy); social relationships (eg, social support, incarceration); and community context (eg, housing quality, transportation, safety/violence).

The 2 identification sessions were treated as independent (ie, findings from the first were not shared with the second). The research team deduplicated suggested data elements from across the 2 panels.

Panelists were asked to rate data elements identified in round 1 based on 2 characteristics (defined below): quality and likelihood of systematic differences across demographic groups. The purpose of this rating exercise was to begin evaluating the feasibility and appropriateness of real-world EHR or HIE data for future computable phenotypes. The rating survey was conducted using REDCap.

 Before administering, 4 core research team members and 2 panelists pilot tested the survey.

 We defined high-quality data elements as those that were concurrently complete, accurate, and up to date in an EHR or HIE system. These 3 dimensions are common components of data quality frameworks.

 Panel members rated each element on a 5-point scale from poor quality to excellent quality (

Differential data quality across patient demographic groups (eg, race/ethnicity, gender, age, sexual orientation) can lead to inaccurate and biased social risk factor measurement, risk prediction, and population health management activities. Systematic differences in quality could be due to a lack of patient diversity, differential work processes, structural barriers to care, or broader societal conditions.

 Panel members rated each element on a 5-point scale from extremely unlikely different data quality to extremely likely different data quality.

Finally, to better understand each panel member’s frame of reference when completing the survey, we included a single item to gauge if responses were rooted in experiences with data from hospital settings, physician/group practice settings, and/or HIE systems.

Analyses were divided by the identification and assessment phases of data collection. First, we determined counts of identified potential data elements by social risk factor category. We also determined data elements suggested in each social risk category during the identification sessions. These sessions also generated group discussion on potential risks and limitations of each data element, which we summarized. Next, we computed frequencies and percentages to describe panelists’ ratings of each data element during the assessment portion of the panel. We created 2-way scatterplots to illustrate the plurality of panelists who responded at the 2 extremes of the respective scales (eg, top 2 box approach).

 The plots help identify those factors that they generally perceived to be of higher data quality and also as less likely to have systematic differences across groups. We plotted the data elements for each social factor category separately. To facilitate visualization, we labeled reported data elements as billing and payment, diagnoses and clinical data, encounters and appointments, identifiers and contact information, language, referrals and orders, social determinants of health codes, and other. The full distribution of responses for every data element is presented in 

. To examine consistency in ratings across panel members, we also grouped average data element ratings by social factor category and stratified by panel member type (physician or nonphysician) and primary frame of reference when answering questions (EHR or HIE) (

Identification: Potential Data Elements (identification session)

Across the 2 identification sessions, panelists generated a total of 89 structured data elements (

). However, due to the cooccurring nature of social risk factors, several suggested data elements pertained to multiple categories. About half of the identified elements (n = 45 of 89) were relevant to the socioeconomic status category. Within the socioeconomic status category, most data elements were suggested as relevant to financial status followed by employment, food insecurity/hunger, and housing instability. The socioeconomic category also included several data elements that we considered to be general indicators of socioeconomic-related needs. Candidate data elements reflected billing, identifiers, orders, and utilization data.

Identification: Observations on Concerns, Considerations, and Limitations Regarding Data Elements (identification session)

During identification, panelists recognized multiple limitations related to structured data elements. These concerns included the potential for inherently biased data elements, inconsistent data collection processes, potential variation across patient populations, and the limitations of area-level measures. For example, a panelist noted that discrimination occurs in the care delivery processes and that underserved populations face barriers in accessing services, which could lead to biased data. Similarly, another panelist noted that credit scores may be very predictive of patients’ financial risks and needs, but this data element is known to be biased by race. Similarly, another noted that inconsistent data collection also limited the usefulness of some data elements. As one panelist stated about documenting homelessness, “There are some 

 codes that nobody uses.” Another panelist agreed with the limited adoption, but he noted that “…the 

 code is going to be very specific when used.” Panelists noted that computable phenotypes may need to be developed for different patient populations. For example, some data elements could be relevant for adults but would not be relevant for pediatric populations. Alternatively, a phenotype could prove useful for patients with high health care utilization only. Finally, one panelist noted the “poor overlap” between area-level measures and patients’ self-reported social risk factors.

Assessment: Perceptions of Data Quality and Likelihood of Systematic Differences in Quality

When responding to the assessment surveys, most (n = 12 of 17) of the panel members reported primarily thinking about data that come from HIE systems. Clinicians and nonclinicians did not vary substantially in their assessments of data quality and likelihood of systematic differences in data quality across populations. Panelists’ assessment of the likelihood of systematic differences in data quality across populations did vary based on whether they reported primarily thinking of HIE systems vs EHR systems. Those who reported thinking primarily about HIE systems most frequently reported that quality was likely to be different across populations for the socioeconomic, social relationship, and community context categories (eAppendix C).

Socioeconomic Status Data Element Assessment

Data elements that are both higher quality and unlikely to have systematic differences across populations are preferable. For those data elements suggested by the panelists as relevant to socioeconomic status, only identifier and contact information and billing and payment-related data elements were frequently rated as “very good” or “excellent” quality and at the same time rated as also “unlikely” or “extremely unlikely” to have differential quality across patient groups (

). These elements included date of birth, last name, address, insurance type, bills in collection, payment method, days in accounts receivable, and outstanding bills. Many more data elements were generally viewed as low quality (ie, “fair” or “poor”). Notably, panelists rated the 

 Z and Logical Observation Identifiers Names and Codes (LOINC) codes that represent various social risk factors, several data elements related to referrals to specific social services and providers, and inability to do telehealth visits as lower quality (ie, “fair” or “poor”) and simultaneously “likely” or “extremely likely” to have differential quality across patient groups.

Cultural Context Data Element Assessment

Panelists rated only address and EHR portal account presence and usage as high quality (ie, “very good” or “excellent”) (

). Panelists also considered these elements to generally be “unlikely” or “extremely unlikely” to have differential quality across patient groups. Conversely, more than half of panelists rated presence of advance directives, language of discharge instructions, primary language, and the need and use of interpreters as low quality (ie, “fair” or “poor”) and as “likely” or “extremely likely” to have differential quality across patient groups. Again, 

 Z codes and similar LOINC codes related to education and literacy were rated of low quality and likely to have different data quality across populations.

Social Relationships Data Element Assessment

Panelists rated few social relationship data elements as high quality overall (ie, “very good” or “excellent”) (

). The majority of panelists rated social relationship–relevant 

 Z and LOINC codes as low quality and “likely” or “extremely likely” to have differential quality across patient groups.

Community Context Data Element Assessment

), some data elements associated with identity and contact information, diagnoses and clinical data, and encounters and appointments, such as address, arrival by ambulance, and emergency department visits associated with trauma or injury, were viewed by more panelists as higher quality and “unlikely” or “extremely likely” to have differential data quality across populations. Other diagnoses and clinical data were also considered “unlikely” or “extremely likely” to have differential data quality across populations but were nevertheless viewed as having “fair” or “poor” data quality. Again, panelists rated community context–relevant 

 Z and LOINC codes as “likely” or “extremely likely” to have differential data quality across populations and to be of poorer data quality.

Our panel of 17 EHR and HIE data experts identified and commented on routinely collected structured data elements for potential use in the future development of computable social risk factor phenotypes. Panelists highlighted several specific concerns about overall data quality and the potential for systematic quality differences across populations that may lead to bias and other data inaccuracies. This novel and foundational work can be used to help develop future computable phenotypes for social factors.

Data quality (defined in this study as complete, accurate, and up to date) is a long-standing concern in biomedical informatics, particularly when data are used for purposes other than those for which they were originally collected.

 Panelists generally perceived data elements of the highest quality to be those used in reimbursement-related processes (eg, identifiers and contact information, billing and payment-related data, diagnoses), which is consistent with prior studies.

 In addition, panelists reported that these data elements were among those less likely to be systematically different in quality across populations. Given these advantageous qualities, reimbursement-related data elements may be viable candidates for use in computable phenotypes development.

In general, the most consistently identified quality concerns related to structured data elements that have been designed to document social risk factors: 

 Z and LOINC codes. Evidence indicates that these codes are substantially underutilized in practice.

 Not only did the expert panel results question the quality of these data, but they also indicated that these were among the most likely to have different data quality across populations. These perceptions align with a recent quantitative analysis indicating that 

 Z codes are a specific indicator of social risk, but one that is collected in a biased fashion.

 Increasing adoption, explicit reimbursement for documenting social needs, or the mapping of screening questionnaires to these standards

 may eventually increase the utility of these data elements. However, currently their application to computable phenotyping or other measurement activities appears limited.

Poor-quality data can undermine any application to care delivery. However, when data quality is systematically different across populations, the risks increase that any measurement strategy, including computed phenotypes, could perpetuate societal biases and inequitable practices in health care.

 Related applications of health data have demonstrated the risk of drawing the wrong inferences about patients. For example, a widely utilized risk stratification tool systematically recommended healthier White patients over sicker Black patients for care management programs, because it failed to account for differential levels of access.

 Similarly, disease risk models developed in homogenous majority populations do not perform well for minority groups.

 The future development of computable social risk factor phenotypes will require attention to the risks of biased and differential quality data because the processes for collecting health care data are highly variable.

 Multiple frameworks and methodologies for identifying and mitigating bias exist, which could be applied to these data.

 Additionally, like any advanced analytics interventions, computable social factor phenotypes would need to be continually evaluated and monitored for effectiveness and lack of bias.

Effectively identifying patients’ social risks is necessary for health care organizations to initiate appropriate referrals to services.

 Vendors, collaboratives, and health care organizations have successfully integrated screening questions into EHR systems and workflows to support data collection.

 Nevertheless, usage of screening questionnaires in practice is highly variable

 and, when used, they have increased staff’s data collection burden.

 As a measurement strategy reliant on existing data and one that can be potentially automated, computable social factor phenotypes could support the screening use case while avoiding the challenges of administering questionnaires. However, screening for social risk factors in health care can be a sensitive issue for patients

 and automated computable phenotypes are admittedly not as transparent a screening strategy as patient-completed questionnaires. If computable social factor phenotypes could be successfully developed, future work should include assessments of patient acceptability.

Developing computable phenotypes was beyond the scope of this Delphi panel. Nevertheless, the findings in this paper provide a candidate list of data elements that could be further evaluated for constructing computable phenotypes. As part of the identification process, panelists explained or justified their suggested data elements. Often, these explanations took the form of methods for transforming or combining data elements into novel and potentially more informative measures. As an example, panelists emphasized the potential to gain information by looking at changes in data elements over time. The most salient examples were changes in addresses to identify housing instability, in insurance status for financial status, or in emergency contact information for social relationships, which are regularly updated for billing and reimbursement reasons. Others have suggested similar uses of change in address data over time.

 Likewise, panel members noted the information to be gained by explainable missingness (ie, instances in which social circumstances would result in data not being recorded or intentionally recorded as missing, as in the case of missing zip codes, phone numbers, or addresses for homeless individuals). Other important recommendations related to combining data across patients. For example, novel data points could be created by identifying the number of individuals sharing an address to indicate financial strain or by noting reciprocal emergency contacts as indicators of social support. Panelists also surfaced the possibility that social need could be identified using evidence of discordant utilization. An example of this type included ordering nutritional supplements or specific meals without an accompanying diagnosis that suggested a clinical need. Lastly, ideas from panel members also included using the information from the components of the patient record. This included the presence, absence, length, or accessing of social worker notes or even the frequency with which the patient portal was used. Therefore, in addition to raw, untransformed data elements, future work to develop computable phenotypes should consider these data transformations or combinations—including changes over time, explainable missingness, combining data across patients, discordant utilization, and the components of the patient record.

Given the variation in the usage of social risk factors, risks, and needs in practice and the literature, it is possible that panel members had different conceptualizations of the social risk factors discussed. For identification, such variation likely had little effect, but the actual movement to phenotype construction would require clear construct definitions. Similarly, our survey did not reflect the multidimensional nature of data quality (eg, conformance, completeness, plausibility) but relied on a single question to reduce respondent burden. Additional work would be necessary to understand the different data quality ratings of each data element. For example, we cannot tell from this study if elements were rated as low quality due to perceived inaccuracy or that the data elements could not be relied upon because they were used too infrequently. Additional study would be required to compare panelists’ perceptions with actual data quality metrics. Also, the panelists were instructed to exclude nonstructured and other sources of data. Of course, these are important sources of social risk factor information, but their potential usage in computable phenotypes represents a different set of challenges than the ones explored in this panel. Still, the identified structured data elements could be combined with structured data from survey questionnaires or even unstructured data extracted from NLP, where available. Such combinations could be more informative. Although expert panel members recognized that social risk factors change over time, the issue of appropriate intervals for measuring social risk factors was not included in this Delphi panel. Lastly, our expert panel reflected individuals with knowledge about EHR and HIE data sources and processes that generated data for clinical, research, and business purposes. A different set of panel members, with different backgrounds, may have identified other data elements.

EHRs and HIE systems contain structured data elements that reflect patient social circumstances, and these data may be useful in developing computable phenotypes. Efforts to develop phenotypes should consider data quality and risks for systematic differences across populations. Future computable phenotyping research should validate strategies for incorporating concepts such as changes over time, explainable missingness, combining data across patients, discordant utilization, and the components of the patient record.

The authors thank Lindsey Sanner, MPH, for her assistance with visualizations.

 Indiana University Richard M. Fairbanks School of Public Health (JRV, HEK, WMT), Indianapolis, IN; Regenstrief Institute (JRV), Indianapolis, IN; Department of Medicine (JA-M) and Department of Family and Community Medicine (LMG), University of California, San Francisco, San Francisco, CA; Department of Health Outcomes and Biomedical Informatics, College of Medicine, University of Florida (JB, CM, EAS, CAH), Gainesville, FL; Population Health Sciences, Weill Cornell Medical College (TRC), New York, NY; Mathematica (GRC), Washington, DC; New York eHealth Collaborative (ND), New York, NY; Owl Health Works LLC (JH), Indianapolis, IN; Departments of Family and Community Medicine and Biomedical Informatics, College of Medicine, The Ohio State University (TRH), Columbus, OH; Indiana Health Information Exchange (JPK), Indianapolis, IN; Johns Hopkins School of Public Health (HK), Baltimore, MD; Department of Population Health, Dell Medical School, The University of Texas at Austin (AK), Austin, TX; Anthem, Inc (JMO), Indianapolis, IN; Department of Pediatrics, Center for Health and Community, University of California, San Francisco (MSP), San Francisco, CA; University of Rochester Medical Center (WP), Rochester, NY; Department of Pediatrics, School of Medicine, Indiana University (SW), Indianapolis, IN.

 This work was supported, in part, by the Indiana Clinical and Translational Sciences Institute Fund and in part by award No. UL1TR002529 from the National Institutes of Health, National Center for Advancing Translational Sciences, Clinical and Translational Sciences Award. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

An expert panel identified and assessed electronic health record and health information exchange structured data elements to support future development of social risk factor computable phenotyping.

Laura M. Gottlieb, MD, MPH

Articles

How Medicare Advantage Plans Use Data for Supplemental Benefits Decision-Making

Assessment of Structured Data Elements for Social Risk Factors

Clinical Interventions Addressing Nonmedical Health Determinants in Medicaid Managed Care