Social and behavioral determinants of health (SBDH) greatly affect health outcomes and include factors outside of the traditional health care setting, such as the environment in which individuals live, work, and play and other social and behavioral factors. 1–3 Negative SBDH, such as low income, limited neighborhood resources, and limited job opportunities, place populations at risk of health disparities. 1–5 Achieving health equity requires addressing SBDH disparities, but there are conflicting views on how best to identify and address SBDH needs within a clinical setting. Through a review of various frameworks and interviews and focus groups with key stakeholders, we sought to begin to understand the barriers and facilitators to accessing SBDH information and provide guidance on how to address the challenges of SBDH collection. Academic medical centers (AMCs) are increasingly called upon to be leaders in promoting health equity. 2,3 Collecting and using SBDH could play an important role in addressing health disparities. The role of AMCs in identifying and addressing SBDH that may place patients at risk remains in question. Several recent reports have highlighted the potential of the electronic health record (EHR) to house SBDH data, which would allow clinical care teams to integrate SBDH information into their care plans and promote better health of the patient and adherence to treatment plans. 2,3,6–11 There are multiple efforts from national organizations such as the Centers for Medicare and Medicaid Services (CMS) and the Office of the National Coordinator for Health Information Technology and many private EHR organizations to standardize SBDH information captured in EHRs. 8,11–14
It is not yet clear if the benefits of capturing SBDH information outweigh the burden of collecting each variable in a system that lacks infrastructure to address SBDH effectively. Clear strategies to collect and address SBDH are needed, but numerous challenges must be overcome before meaningful integration of SBDH data collection into practice is a reality. Traditionally, EHR systems have been used to capture clinical information such as medical history, procedures, labs, and other information to enable diagnosis and treatment. During their inception, EHR systems collected limited SBDH information, such as smoking status, alcohol use, and race, in a structured and standard manner. 2,3,8,10 SBDH data collection is a growing area of interest among EHR vendors, and many vendors have increased structured SBDH data fields in their products. Significant limitations persist, however. A qualitative study with vendor participants reported a need from health systems to articulate clearer priorities and standards for SBDH information and for measure consensus to assist in development. 11 Despite some increases in availability of structured EHR forms for SBDH data, the free-text portion of an EHR, such as encounter notes, remains a common location for SBDH information, preventing automated data retrieval. 9,10,15,16
The definitions of SBDH and how they affect outcomes are constantly evolving. AMCs provide a unique environment in which the growing potential of SBDH information can be harnessed, allowing the collection and use of SBDH information to move from recommended to standard practice. A variety of SBDH variables are recommended for capture in an EHR by various organizations, including the National Academy of Medicine (formerly the Institute of Medicine) and the University of California, San Francisco, Social Interventions Research and Evaluation Network. 1,17 CMS programs such as meaningful use and Quality Reporting Document Architecture are beginning to require hospitals to report on specific SBDH for their populations. 12,13 Unfortunately, collecting SBDH information in the EHR is evolving at a slow pace.
SBDH frameworks can provide valuable insight to inform EHR data fields and to maximize the utility of SBDH data to health care. AMCs need to decide what variables are most important and how information should be collected (i.e., in a structured form or in free text). Currently, universally used standards are lacking for SBDH information. 2,3,6 AMCs have the option to collect SBDH at an individual level (e.g., age, employment status, income) or geographic or neighborhood level (e.g., public transportation access, walkability index, presence of food deserts). However, it is not yet clear which level of SBDH variables will be most useful for patient or population health. 1–5,8–10 Community-level information can point to potential issues a patient may face or be beneficial in understanding neighborhood barriers affecting health, such as living in a food desert. Individual-level information may be a higher priority for immediate patient care. Both provide specific, actionable information when the right resources are available.
Although there are exceptions, AMCs have not routinely prioritized a systematic, reliable approach to SBDH data collection and are inconsistently prepared to take action on data collected, limiting rigorous assessments of the impact of addressing SBDH in health care. As such, this limitation provides mixed messages to policymakers and payers regarding investment and reduces incentives for EHR developers to increase usability of SBDH data collection, storage, and response tools in their products. The current state impedes progress in addressing SBDH in clinical care and leaves unrealized the potential to improve patient health and address health disparities. To move forward, there is a clear need to develop and test systematic approaches to integrating SBDH data collection and resultant actions based on data into the clinical care practices of AMCs. It is our perspective that next steps require a practical, theory-informed approach that leverages key learnings from implementation science regarding the importance of stakeholder engagement, clinical workflow integration, contextual facilitators and constraints, and dissemination-ready solutions. The purpose of our work is to generate guiding recommendations for systematic implementation of SBDH data collection in the EHR through (1) reviewing SBDH conceptual and theoretical frameworks and (2) eliciting stakeholder perspectives to understand the current barriers to and facilitators of using SBDH information in the EHR and priorities for systematic implementation of data collection.
Method
Frameworks to guide SBDH variable selection
Conceptual frameworks were reviewed to understand the categories of SBDH and relationships between SBDH factors and various health outcomes and inform the interview and focus group guides. We examined 2 of the 3 suggested frameworks by the National Academy of Medicine in its original report on SBDH data collection and a distinct, newer framework, the Maryland Population Health Framework by Hatef et al (2018), which modified National Academy of Medicine frameworks to fit specific state needs. 1,2,10,18,19 Each framework explores how SBDH information is characterized, categorized, and affects health outcomes. 1,18–24Table 1 presents an overview of the 3 frameworks we reviewed.
;)
Table 1: Frameworks Describing Social and Behavioral Determinants of Health and Potential Applications to Patient Care, Population Health, Research, and Health Outcomes, From a Study of Social and Behavioral Variables in Electronic Health Records, Johns Hopkins Health System, March–May 2018
Gathering stakeholder perspectives on SBDH
Given that current frameworks do not provide clear guidance for prioritization of SBDH variables for collection, we conducted interviews and focus groups to understand how clinicians and researchers are accessing SBDH information from EHRs (e.g., Epic software) and to inform SBDH collection implementation. We used snowball sampling to identify participants, starting with a list of individuals with known interest in SBDH. We originally identified 15 researchers and clinicians across multiple hospitals and departments from a single large academic medical institution, Johns Hopkins Health System, who were interested in participating. Based on schedule and participant availability, 4 one-on-one interviews and 2 focus groups were conducted between March and May 2018. The focus groups contained 5 and 8 participants from the same department or division, bringing our total to 17 participants.
Semistructured interviews and focus groups lasted approximately 45 to 60 minutes and were conducted by the same facilitator (E.C.L.). The interview guide (Supplemental Digital Appendix 1 at https://links.lww.com/ACADMED/B83), used for both focus groups and individual interviews, was specifically developed to help the study team understand (1) facilitators of and barriers to collecting and accessing social and behavioral variables from the EHR for research and clinical care, (2) how SBDH information is used in practice, and (3) how SBDH data should be collected and what factors would be most important to collect. The SBDH frameworks informed development of the initial interview guide. The guide was then adapted and modified for clarity after the first focus group. Participants were promised anonymity in any final report, allowing for open communication about their experiences—both positive and negative—with using or considering use of the EHR as a source for SBDH data. Each interview and focus group was audio-recorded and transcribed by a professional transcriptionist.
Transcripts were coded using MAXQDA, version 12 (VERBI Software, Berlin, Germany), a qualitative data management tool, to identify overarching themes in the collection and uses of SBDH information in an EHR.
We (E.C.L.) developed a codebook deductively, using the 3 categories of questions from the interview guide. Through our review of the transcripts, we inductively developed subcodes within each category. Study team members (J.M.K., H.K., L.R.D.) then reviewed each category with the coder (E.C.L.) to understand common themes as they relate to types of variables needed and how information should be captured. The study team and coder then reached consensus through discussion regarding the analytic conclusions based on the categories and codes in the data. Direct quotes from participants are provided in Table 2 to illustrate the themes. This study was reviewed and approved as not human subjects research by the Johns Hopkins Bloomberg School of Public Health institutional review board (IRB 00008765).
Table 2: Selected Quotes From Key Stakeholders About the Facilitators of and Barriers to Collection of Data on Social and Behavioral Determinants of Health in Electronic Health Records, Johns Hopkins Health System, March–May 2018a
Results
Frameworks to guide SBDH variable selection
While some frameworks assume a traditional causal model that shows a direct link of SBDH to health outcomes, other frameworks explore a more detailed pathway that links socioeconomic status to health. 1,10 In Table 1, we describe 3 SBDH frameworks in detail, including how SBDH information is broken down into different domains, how each can be used to improve health, and how each can answer SBDH-related research questions.
Although all frameworks agree that SBDH affect health outcomes, the impact may vary on setting and how factors are used and integrated, and each has distinct SBDH variables and pathways (Table 1). The general themes among the frameworks were as follows: (1) SBDH can be collected at an individual or neighborhood level; (2) SBDH information can be categorized as social, behavioral, or economic; and (3) the specific impact that SBDH have on health outcomes is largely unknown. Currently, no consensus exists on what framework best shows the relationship between SBDH and health outcomes. Lack of model consensus complicates the development of specific recommendations on what SBDH should be captured and incorporated into an EHR.
Gathering stakeholder perspectives on SBDH
All study participants were from one large academic medical institution but from different hospitals and different departments and divisions, including pediatrics, adolescent medicine, pulmonology, general internal medicine, and gastroenterology. All focus group participants and 1 interview participant had primary responsibilities as clinicians in ambulatory and inpatient care. Two participants saw patients as general internists but were primarily researchers, and 1 participant conducted research exclusively. Two overarching themes were repeated during the interviews: (1) the domains of SBDH information that the researchers and clinicians were interested in varied greatly and (2) information should be captured through patient-reported methods (Table 2).
There was no consensus on which SBDH should be captured. Both researchers and clinicians acknowledged the need to collect SBDH and expressed frustration with the lack of data collection standards. Clinicians were often interested in collecting SBDH information (e.g., physical activity level) if it was relevant to the patients’ age, diagnoses, or treatment plan. Researchers also desired a variety of SBDH but reported they did not generally use the information currently available in the EHR because of concerns about the validity and quality of such data. Researchers reported directly requesting SBDH data from participants to ensure a standardized data collection practice.
Clinicians expressed the need for SBDH to draw a complete picture of their patients. They believed social and behavioral information could be useful to help identify barriers to care or provide an understanding of adherence. Clinicians also felt that SBDH could be useful for care management and health assessment. Some posited that SBDH variables, collected over a longer time frame, could help clinicians identify why patients are nonadherent to their treatment plan.
Without being prompted, participants offered several approaches on how to use SBDH that are similar to the uses described in the frameworks in Table 1. For research, interview participants described using data found in the AMC’s EHR to identify patients who would qualify for specific interventions or studies. Contact information was then used to inform patients and gain their consent. Patients would then fill out study-specific documents that captured SBDH information. This information was neither stored nor transferred to the EHR, nor was it available to be used by anyone outside of the study team. Some participants reported that SBDH variables have been used to conduct trend analysis and report on clinical performance, but they said that an abundance of missing data makes these reports preliminary and unreliable.
Most participants reported that the SBDH information collected in the EHR was not consistent and there was no standard way to check the accuracy or capture all needed SBDH. Standardized data appropriate for research or quality reporting increase the burden of data collection at the point of care, which lowers feasibility unless consensus can be reached on a limited number of high-priority SBDH variables. Nevertheless, various stakeholders had differing priorities for SBDH information. In fact, variables needed often varied by health care provider, specialty, and patient needs (see Table 2 for direct quotes from participants). For example, pediatricians expressed their desire to know what schools their patients attend or to link patient records to siblings, while internists expressed the desire for more information about diet and physical activity.
Challenges exist in how SBDH information should be captured. Most participants believed adding more information to be collected in a structured manner (e.g., in drop-down menus) would increase the clinical burden. Clinicians presented a strong preference for patient-reported data but were unclear how to standardize or ensure that data are collected consistently. Suggestions included creating a structure that ensures self-reporting of SBDH through the patient portal. Clinicians believed this method would help decrease the burden on health care providers and yield consistent data quality because they would know who provided the information and how it was collected (Table 2).
As barriers to SBDH data collection, participants reported the following: no direct reimbursement, time pressures, and competing demands. Despite these barriers, clinicians reported collecting data. They captured information that was important to them, their interests, and their patient population generally as free text. They did not collect information considering about how it could also be used for research or population health measurement.
Discussion
Understanding and addressing patients’ SBDH are new frontiers in health care. Clinicians, researchers, and policy experts acknowledge that SBDH play a large part in health outcomes, yet it is unclear how to fully integrate the information at the point of care and support patients in mitigating negative outcomes. Improving clinical care actions in response to negative SBDH requires a foundation of reliable, accessible SBDH data in the EHR. Stakeholders who participated in this study revealed 2 key challenges facing AMCs in constructing an SBDH data collection foundation: (1) lack of consensus on which variables to collect and (2) how to implement and sustain reliable data collection. These challenges are interrelated, and addressing both should occur simultaneously. Addressing negative SBDH that underlie the persistent and pervasive health disparities in the United States requires urgent action.
As evidenced by the wide range of opinions of the stakeholders, reaching consensus on which SBDH to prioritize for EHR data collection is a formidable task. Key factors to consider include availability of a valid measure for SBDH, the ease of use of the measure for collecting information, the ability to intervene on the negative SBDH, and the impact of addressing SBDH on a person’s health. Physicians and care teams may not be able to identify which SBDH may be most impactful to address. Adding to the complexity is that how SBDH can be used may depend on the framework applied and the context of care. In our review, each SBDH framework was developed with a certain use in mind (e.g., policy, clinical care, research focus), but a framework that integrates both operational and research needs is lacking. Health care providers practicing at AMCs have varied responsibilities that influence their opinions about the importance of the role of SBDH in care and prioritization of SBDH information for collection. It is critical for AMCs to understand the divergence of opinions but also to prioritize several domains of SBDH data collection and move forward with systematic collection implementation initiatives.
Systematic implementation of SBDH data collection requires investment of time and resources. Stakeholders who participated in this study underscored the importance of consideration of multiple modes of SBDH data collection. Clinicians and researchers believe that collecting SBDH data directly from patients in an automated fashion, such as via patient portals, would reduce the workflow burden, as compared with point-of-care data collection, and provide relevant, patient-centered information for care and treatment. Point-of-care data collection for some SBDH and for some patient populations would still play a role. The complexity of collection renders SBDH data collection akin to other complex health care interventions, which implementation science methods are well poised to address. For example, normalization process theory is primarily concerned with embedding and sustaining an intervention, in this case data collection, into routine workflows and provides a mechanism to address the varied contexts in which the work must be done. 25,26 It has been used previously in implementing and understanding other complex health system interventions and can integrate with established quality improvement infrastructure. 26–28 We acknowledge there are other valid theories and methods that could support systematic SBDH data collection. The critical element, however, is the investment in implementation and an evaluation plan. Current varied processes, generally limited resources, and time investment in systematic data collection are a clear point of frustration for stakeholders and underpin the persistent health care challenges in incorporating and addressing SBDH in clinical practice, despite the increased emphasis on SBDH. 2–10,13
The dynamic interplay between prioritization of SBDH for data collection and systematic implementation highlights a potential role for extracting SBDH data from an EHR’s unstructured data (or free text), given the common practice of incorporating this information in clinical notes. 9,15 Natural language processing (NLP) methods could be used to explore free-text documentation to extract specific SBDH. With NLP, the clinical workflows would not need to be modified to collect variables in a structured manner and could allow for use of SBDH information in clinical care before other data collection systems are established. 9,16 It should be noted, however, that NLP methods face challenges with context and language used to express SBDH. Despite these challenges, advancing the NLP methods may be less challenging than changing the routine clinical workflow to capture SBDH within structured EHR flow sheets in the short term. NLP, however, would not allow for standardized data collection, which could introduce bias into measurement of SBDH and thus is not favored as a long-term or stand-alone solution to the problem of reliable SBDH data collection. In the present time of transition, we believe the ideal use of NLP methods is to increase evidence about the utility of EHR-derived SBDH information on improving patient health and thus increase the willingness of health care systems to invest in standardized data collection that fits clinical workflows.
Some limitations exist in this study. Participants were from one academic medical institution and expressed similar themes and recommendations, although they represented different hospitals, roles, and settings. We did not include all stakeholder groups, such as clinical support staff or patients and their families. Further studies are needed to investigate perspectives of other EHR stakeholders to develop best practices and to assist health systems in integrating SBDH into their EHRs and making them usable by multiple stakeholders.
Conclusions
Our findings highlight the lack of consensus in SBDH frameworks and the current system limitations preventing systematic and reliable collection of SBDH data. AMCs are at a fork in the road: either choose to not focus on collecting SBDH for their patient populations or implement and iteratively improve new clinical workflows to collect standardized SBDH information. Based on our findings and the supporting literature, it is recommended that health systems begin by identifying the key SBDH variables prioritized by their clinicians and researchers that can be captured in structured fields of EHRs and experiment with NLP methods to extract other variables from unstructured notes.
Choosing key SBDH variables can be a challenge on its own. We have used the SBDH frameworks to help outline which types of variables AMCs may focus on in the short term. Recognizing varied and competing priorities, it is difficult to make clear recommendations on specific variables. AMCs will certainly not be able to satisfy all stakeholders with their initial selection. Within existing frameworks and the health system’s priorities, the chosen SBDH variables should affect care or be amenable to change themselves, represent a range of domains, and be captured in a standard and structured manner. Data collection efforts should be an iterative process with the EHR vendor and fit into the current workflow to minimize the increased burden on health care providers. Once data collection has improved, evidence can be built to understand the impact on outcomes and provide the basis for developing methods to improve data collection on other variables in the future.
Starting with a few SBDH variables and developing a clear road map for integration and collection, AMCs can build a data-driven approach to collecting and addressing SBDH using EHRs and perhaps have a more meaningful impact on health equity.
References
1. Committee on the Recommended Social and Behavioral Domains and Measures for Electronic Health Records; Board on Population Health and Public Health Practice; Institute of Medicine.. Capturing Social and Behavioral Domains and Measures in Electronic Health Records: Phase 2. 2014. Washington, DC: National Academies Press;
2. Gottlieb L, Tobey R, Cantor J, Hessler D, Adler NE. Integrating social and medical data to improve population health: Opportunities and barriers. Health Aff (Millwood). 2016;35:2116–2123.
3. Cantor MN, Thorpe L. Integrating data on social determinants of health into electronic health records. Health Aff (Millwood). 2018;37:585–590.
4. Hughes LS, Phillips RL Jr, DeVoe JE, Bazemore AW. Community vital signs: Taking the pulse of the community while caring for patients. J Am Board Fam Med. 2016;29:419–422.
5. Bazemore AW, Cottrell EK, Gold R, et al. “Community vital signs”: Incorporating geocoded social determinants into electronic records to promote patient and population health. J Am Med Inform Assoc. 2016;23:407–412.
6. Kasthurirathne SN, Vest JR, Menachemi N, Halverson PK, Grannis SJ. Assessing the capacity of social determinants of health data to augment predictive models identifying patients in need of wraparound social services. J Am Med Inform Assoc. 2018;25:47–53.
7. Palacio A, Suarez M, Tamariz L, Seo D. A road map to integrate social determinants of health into electronic health records. Popul Health Manag. 2017;20:424–426.
8. Gold R, Cottrell E, Bunce A, et al. Developing electronic health record (EHR) strategies related to health center patients’ social determinants of health. J Am Board Fam Med. 2017;30:428–447.
9. Kharrazi H, Anzaldi LJ, Hernandez L, et al. The value of unstructured electronic health record data in geriatric syndrome case identification. J Am Geriatr Soc. 2018;66:1499–1507.
10. Hatef E, Rouhizadeh M, Tia I, et al. Assessing the availability of data on social and behavioral determinants in structured and unstructured electronic health records: A retrospective analysis of a multilevel health care system. JMIR Med Inform. 2019;7:e13802.
11. Freij M, Dullabh P, Lewis S, Smith SR, Hovey L, Dhopeshwarkar R. Incorporating social determinants of health in electronic health records: Qualitative study of current practices among top vendors. JMIR Med Inform. 2019;7:e13849.
12. Centers for Medicare and Medicaid Services.. Medicare & Medicaid EHR Incentive Program: Meaningful use stage 1 requirements overview.
https://www.cms.gov/Regulations-and-Guidance/Legislation/EHRIncentivePrograms/downloads/mu_stage1_reqoverview.pdf. Published 2010. Accessed February 19, 2021.
13. Centers for Medicare and Medicaid Services. Electronic Clinical Quality Improvement Resource Center. Now available: Updated 2018 CMS QRDA III Implementation Guide for Eligible Clinicians and Eligible Professionals.
https://ecqi.healthit.gov/now-available-updated-2018-cms-qrda-iii-implementation-guide-eligible-clinicians-and-eligible-professionals. Published March 12, 2018. Accessed February 19, 2021.
14. EHRIntelligence.. Researchers add social determinants of health to Epic EHR.
https://ehrintelligence.com/news/researchers-add-social-determinants-of-health-to-epic-ehr. Published July 19, 2017. Accessed February 19, 2021.
15. Chen ES, Manaktala S, Sarkar IN, Melton GB. A multi-site content analysis of social history information in clinical notes. AMIA Annu Symp Proc. 2011;2011:227–236.
16. Cifuentes M, Davis M, Fernald D, Gunn R, Dickinson P, Cohen DJ. Electronic health record challenges, workarounds, and solutions observed in practices integrating behavioral health and primary care. J Am Board Fam Med. 2015;28suppl 1S63–S72.
17. University of California, San Francisco. Social Interventions Research and Evaluation Network.
https://sirenetwork.ucsf.edu. Accessed February 19, 2021.
18. Predmore Z, Hatef E, Weiner JP. Integrating social and behavioral determinants of health into population health analytics: A conceptual framework and suggested road map. Popul Health Manag. 2019;22:488–494.
19. Hatef E, Lasser EC, Kharrazi HHK, Perman C, Montgomery R, Weiner JP. A population health measurement framework: Evidence-based metrics for assessing community-level population health in the global budget context. Popul Health Manag. 2018;21:261–270.
20. Ansari Z, Carson NJ, Ackland MJ, Vaughan L, Serraglio A. A public health model of the social determinants of health. Soz Praventivmed. 2003;48:242–251.
21. Feinstein JS. The relationship between socioeconomic status and health: A review of the literature. Milbank Q. 1993;71:279–322.
22. Wen M, Hawkley LC, Cacioppo JT. Objective and perceived neighborhood environment, individual SES and psychosocial factors, and self-rated health: An analysis of older adults in Cook County, Illinois. Soc Sci Med. 2006;63:2575–2590.
23. DeVoe JE, Bazemore AW, Cottrell EK, et al. Perspectives in primary care: A conceptual framework and path for integrating social determinants of health into primary care practice. Ann Fam Med. 2016;14:104–108.
24. Rosen AK, Reid R, Broemeling AM, Rakovski CC. Applying a risk-adjustment framework to primary care: Can we improve on existing measures? Ann Fam Med. 2003;1:44–51.
25. Normalization Process Theory.. What is NPT?
http://normalizationprocess.org. Accessed February 19, 2021.
26. May CR, Cummings A, Girling M, et al. Using normalization process theory in feasibility studies and process evaluations of complex healthcare interventions: A systematic review. Implement Sci. 2018;13:80.
27. Ziegler E, Valaitis R, Yost J, Carter N, Risdon C. “Primary care is primary care”: Use of normalization process theory to explore the implementation of primary care services for transgender individuals in Ontario. PLoS One. 2019;14:e0215873.
28. Gillespie BM, Harbeck E, Lavin J, Gardiner T, Withers TK, Marshall AP. Using normalisation process theory to evaluate the implementation of a complex intervention to embed the surgical safety checklist. BMC Health Serv Res. 2018;18:170.
