Abstract
Introduction: Social determinants of health (SDOH) significantly impact health outcomes, yet their integration into clinical decision making is inconsistent. We examined how family physicians document SDOH in electronic health records (EHRs) and identified factors influencing this practice.
Methods: We performed a cross-sectional analysis of 2,089 family physicians completing the 2022 American Board of Family Medicine Continuous Certification Questionnaire. The outcome was physicians’ self-reported SDOH documentation by checking a box within the EHR, writing it in a note, or entering it as a diagnosis. Physician, practice, and community characteristics associated with SDOH documentation were assessed, using logistic regression.
Results: We found that 61% of family physicians documented SDOH in notes, with fewer using checkboxes (46%) or diagnosis codes (35%). Across models, factors persistently positively associated with documenting SDOH included participating in value-based programs, having more resources for social needs, collaborating with neighborhood organizations, and working in a more disadvantaged area (higher Social Deprivation Index [SDI] score). For example, family physicians who worked in areas with the third quartile of SDI (OR = 1.366, 95% CI = 1.037 - 1.799) and the fourth quartile of SDI (OR = 1.364, 95% CI = 1.032 - 1.804) were more likely to enter SDOH as a diagnosis, compared with those in the least disadvantaged areas.
Discussion: Socioeconomic aspects of the communities and a practice-level capacity to address SDOH were the biggest predictors of documenting SDOH, rather than the physicians’ own characteristics. These findings affirm the necessity of financial incentives and well-resourced care teams to successfully achieve integrated SDOH in primary care practice.
- Clinical Decision-Making
- Cross-Sectional Studies
- Documentation
- Electronic Health Records
- Family Physicians
- Health Equity
- Logistic Regression
- Outcomes Assessment
- Primary Health Care
- Social Deprivation
- Social Determinants of Health
- Socioeconomic Factors
- Surveys and Questionnaires
Introduction
A growing body of evidence has consistently shown that social determinants of health (SDOH) at both the community and individual level - such as housing, food insecurity, lack of transportation, or economic instability - significantly influence health outcomes more than medical care alone. According to the World Health Organization, SDOH account for 30 to 55% of health outcomes and are also the primary driver of health inequities.1 Unmet social needs have been associated with an increased incidence and higher complication rates of health conditions such as cardiovascular disease, musculoskeletal complaints, and diabetes.2–9 In addition, patients who report more social risk factors have experienced a higher likelihood of unplanned hospital readmissions and emergency department (ED) visits.10,11 Concurrently, the emphasis on redesigning care delivery and payment models to improve the value in health care instead of the volume of services is encouraging health care providers and systems to invest in the integration of social care into health care, including screening patients for unmet social needs and linking them with services in their local community to stay well and out of the hospital. Therefore, collecting and understanding SDOH in clinical settings is a critical first step toward improving care for high-needs, high-cost populations and reducing longstanding inequities in health.
Unlike well-defined community-level social risk measures from Census data such as the Social Deprivation Index (SDI) and the Area Deprivation Index (ADI), there is a lack of standardization around how individual-level health-related social needs are documented in clinical records.12 One potential source of this information is ICD-10-CM Z-codes in electronic health records (EHRs). Risk factors may be documented elsewhere by clinicians, for example using free-text clinical notes in EHRs or structured data fields (eg, drop-down menus or checkboxes).13 However, previous studies have found that most clinicians rarely capture social needs with Z-codes or structured data inputs.14–16 In particular, the utilization of Z-codes is very low, with 1.59% of the Medicare fee-for-service (FFS) population having claims with Z-codes and 1.07% of Medicare Advantage enrollees having claims with Z-codes, according to a 2019 study.17,18 Documentation of social needs of Medicaid enrollees is also low: 1.56% with Z-code claims for managed care enrollees compared with 0.79% for FFS enrollees in 2018.19
Addressing unmet social needs in the delivery of health care is critical to achieve more equitable health outcomes, yet their integration into clinical decision making is inconsistent.20,21 In this study, we conducted a cross-sectional analysis of a nationally representative cohort of family physicians to explore how family physicians document SDOH in EHRs and identify factors influencing this practice. Family physicians are especially important for social needs screening since they provide health care in disadvantaged communities more often than other medical specialties.22
Methods
Data and Study Population
This study used the 2022 American Board of Family Medicine (ABFM) Continuing Certification Questionnaire (CCQ). The 2022 CCQ was completed by 5,458 family physicians in the US who sought to continue their ABFM certification. Because completing the CCQ is mandatory when registering to recertify, it has a 100% response rate. The survey collected detailed information about physician demographic and practice characteristics. Among the 5,458 respondents, 2,706 family physicians were randomized to questions that included documentation of SDOH. We analyzed 2,089 family physicians after excluding those who did not provide direct patient care or continuity care (n = 617).
We used 5 supplemental data sources to create measures not available in the CCQ. To construct the location of residency training programs such as a community-based residency training setting, the list of rural residency programs from the RTT Collaborative (https://rttcollaborative.net/rttc-participating-programs), the Health Resources and Services Administration (HRSA) Teaching Health Center Graduate Medical Education (THC GME) program dashboards (https://data.hrsa.gov/tools/find-grants), and the American Medical Association (AMA) Historic Residency File were used. The Robert Graham Center (RGC) SDI was used to construct zip code tabulation area (ZCTA)-level deprivation to quantify the socioeconomic variation in health outcomes.23 The 2013 Rural-Urban Continuum Codes were used to identify county-level rurality, which we also included in our model.
Documenting SDOH in EHRs
The primary outcome of interest was physicians’ self-reported frequency of documented screening for social needs by checking a box/button within the EHR, writing it in a note within the EHR, or entering it as a diagnosis within the EHR (Appendix 1). Documenting SDOH in EHRs was defined as a binary measure - yes (often/sometimes) or no (rarely/never).
Physician Characteristics
Physician characteristics included age, gender, race (White, American Indian or Alaska Native, Asian, Black or African American, Native Hawaiian or Other Pacific Islander, and Other), ethnicity (Hispanic or Latino and Non-Hispanic), and whether physicians completed community-based residency programs. To examine potential benefits of community-based residency training, a program with rural training track (RTT) where residents spend more than 50% of their training in a rural place by at least 2 federal definitions, or HRSA THC GME grant was classified as community-based training. Both programs not only focus on training in the community, but also on training in the most medically vulnerable communities in the US. We identified family physicians who completed a community-based residency training program, using their residency start/end year and training program information available in the ABFM CCQ and AMA Historic Residency File.
Practice Characteristics
Practice characteristics included the type of practice (academic, government, hospital/health maintenance organization [HMO], independent, and other), size (solo, 2 to 5, 6 to 20, and >20 providers), participation in value-based payment (VBP) initiatives, SDOH resources or tools, other health professionals at the practice (specifically social workers and care coordinators), and collaboration with local government, neighborhood organizations, and transit (Appendix 1).
Community Characteristics
The ZCTA-level SDI quartiles and county-level rurality (Codes 4 to 9 are considered to be rural) were included as community factors to examine whether or not family physicians who practice in areas where people have higher social needs would be more likely to address SDOH than their counterparts. SDI is a composite measure developed and maintained by the RGC. It is based on factor analysis of 7 demographic characteristics collected in the American Community Survey: the percentage of population living below 100% federal poverty level, the percentage of population aged 25 years or over with less than 12 years of education, the percentage of single-parent households with children aged less than 18 years, the percentage of households living in rental units, the percentage of households living in the overcrowded housing units, the percentage of households without a car, and the percentage of unemployed adults aged 16 to 64 years old.23
Analysis
Using a separate multivariate logistic regression model for each of the 3 documentation methods, we assessed the associations between documenting SDOH and the physician, practice, and community characteristics described above. The number of observations in each model varied due to missing data for key variables, including “do not know” responses to 3 outcome variables. Models included robust standard errors, and the 95% confidence interval (CI) was used to estimate the precision of the odds ratio (OR). All analyses were performed using Stata version 17 (StataCorp LLC, College Station, TX). This study was ruled exempt by the American Academy of Family Physicians Institutional Review Board.
Results
As shown in Table 1, the study population was predominantly White (65%) and non-Hispanic (90%). Only 4% completed any community-based residency training. Most worked in either hospital or HMO settings (43%). 72% worked in practices of 20 or fewer providers. Two-thirds of family physicians reported their organizations participated in one or more VBP initiatives, such as a patient-centered medical home, accountable organization, or pay-for-performance arrangement. The self-reported mean score of their clinic’s ability to address social needs was 6.6 (SD = 2.6) on a 10-point scale, where higher scores indicate greater ability to meet patient social needs. Almost one-third of respondents worked collaboratively with a licensed social worker or care coordinator at their principal practice site. Respondents worked in a practice with a variety of agency collaboration: local government (44%), neighborhood organizations (33%), or transit (25%). About 13% worked in rural areas.
Characteristics of Family Physicians from 2022 American Board of Family Medicine Continuous Certification Questionnaire (n = 2,089)
Of 2,089 family physicians, 61% reported documenting SDOH in free-text clinical notes in EHRs; 46% used structured electronic forms within the EHR and 35% used diagnosis codes (Figure 1). The most common form of documenting SDOH in EHRs was through free-text within the clinical note.
Self-reported frequency and method of documenting social determinants of health in electronic health records by family physicians from 2022 American Board of Family Medicine Continuous Certification Questionnaire (n = 2,089).
Factors consistently associated with increased documentation of SDOH across models included: employment at a practice participating in VBP, having the resources to address social needs, collaborating with neighborhood organizations, and practicing in a more disadvantaged area (Figure 2). For example, family physicians who worked in practices participating in VBP had higher odds of documenting SDOH using electronic forms (OR = 1.605, 95% CI = 1.295 - 1.990, p=<0.001) or notes (OR = 1.383, 95% CI = 1.115 - 1.716, P < 0.01) compared with those who did not participate in VBP. Having dedicated staff and linkages to community programs was positively associated with documenting SDOH using structured data fields (OR = 1.106, 95% CI = 1.063 - 1.152, p=<0.001), notes (OR = 1.081, 95% CI = 1.038 - 1.126, P < .001), or diagnosis codes (OR = 1.082, 95% CI = 1.038 - 1.128, P < .001). Family physicians who worked in a practice collaborating with neighborhood organizations had higher odds of documenting SDOH using structured data fields (OR = 1.526, 95% CI = 1.208 - 1.927, p=<0.001), notes (OR = 1.441, 95% CI = 1.125 - 1.844, P < .01), or diagnosis codes (OR = 1.662, 95% CI = 1.313 - 2.102, P < .001) compared with those did not work with neighborhood organizations. Family physicians who worked in areas with the third quartile of SDI (OR = 1.366, 95% CI = 1.037 - 1.799, P < .05) and the fourth quartile of SDI (OR = 1.364, 95% CI = 1.032 - 1.804, P < .05) were more likely to enter SDOH as a diagnosis, compared with those in ZCTAs with the first quartile of SDI (least disadvantaged).
Factors associated with documenting social determinants of health in electronic health records by family physicians from 2022 American Board of Family Medicine Continuous Certification Questionnaire.
In addition, compared with their male counterparts, female family physicians were more likely to document SDOH in their notes (OR = 1.342 95% CI = 1.095 - 1.644, P < .01). The likelihood of documenting SDOH using electronic forms was higher for family physicians in government clinics (OR = 1.527, 95% CI = 1.050 – 2.220, P < .05) compared with those in independently owned clinics. Family physicians who worked collaboratively with a care coordinator at their principal practice site had higher odds of documenting SDOH using electronic forms (OR = 1.359, 95% CI = 1.080 - 1.710, p=<0.01) than those did not. There were no statistically significant associations between documenting SDOH and training in a community-based residency training program or rural location. Detailed regression results are available in Appendix 2.
Discussion
This study examined a nationally representative cohort of 2,089 family physicians to explore how they document SDOH in EHRs and identify factors associated with the documentation of SDOH. Socioeconomic aspects of the communities and practice-level capacity to address SDOH were the biggest predictors of documenting SDOH, rather than the physicians’ own characteristics. These findings affirm the need for financial incentives and well-resourced primary care teams to successfully achieve integrated SDOH in primary care practice.
First and foremost, this study suggests clear associations between increased documentation of SDOH and practices receiving VBP, possessing more resources for social needs, and collaborating with neighborhood organizations. The family physicians in these practices were significantly more likely to document SDOH across various methods, highlighting the importance of supportive environments and partnerships in addressing SDOH, regardless of the modality, in clinical settings. The integration of social health interventions requires resources for social risk screening and navigation to social services.24–26 Stable and sufficient funding like VBP is critical to invest in a workforce and workflow changes that lead to effective screening.
This study further reinforces the understanding that robust community partnerships, aimed at improving service coordination and enhancing referral processes, significantly increase the likelihood of screening for and documenting SDOH. There are many efforts underway to foster community collaboration, and a recent set of studies demonstrates that having trusted social service or community-based organizations are key to improving individual and community health. For example, the Center for Medicare and Medicaid Innovation (CMMI) established the Accountable Health Communities (AHC) model, a 5-year initiative (2017 to 2022) to test whether connecting patients to community resources and addressing SDOH can improve health outcomes and reduce costs for Medicare and Medicaid beneficiaries. According to the most recent evaluation report on the AHC model, beneficiaries who were eligible for social needs navigation services used community services and experienced the ED visits less often than beneficiaries in a control group.27
Second, increased social need prevalence in an area also made screening and recording more likely. Specifically those working in communities with a higher SDI or those working in a federally qualified health center or rural health clinic were more likely to document SDOH in their EHR. Supportive payment policies addressing SDOH screening and intervention should prioritize these practices. Despite the widespread acceptance of the role of SDOH in determining health,28 most major insurances do not directly reimburse for social risk screening and navigation to social services.
Several Medicare programs are currently using or considering the community-level social risk adjusted payments. The Centers for Medicare and Medicaid Services (CMS) has recently launched the Accountable Care Organization Realizing Equity, Access, and Community Health (ACO REACH) Model. The ACO REACH initiative incorporates social risk adjustment, which factors in community-level deprivation indices, to better support ACOs to deliver accountable care for beneficiaries in underserved communities. By linking payment incentives to the management of social risk factors, CMS aims to create a sustainable framework for addressing SDOH at both the community and individual levels.29 The Medicare Shared Savings Program also offers an upfront payment of $250,000 and 2 years of quarterly payments, to promote equity by holistically addressing beneficiary needs, including social needs, via Advance Investment Payments.30 The Medicare Advantage plans are considering the feasibility and utility of incorporating ADI into the health equity index incentivizing practices to perform well for socially at-risk beneficiaries.31 Despite this movement by CMS and others to incorporate community-level social risk into reimbursements, recent studies raise concern that community-level social risk may be an inaccurate proxy for individual social needs.32,33 To directly address patient-specific social risk, Medicare has introduced coding (G0136) for direct payments for SDOH risk assessments to recognize when clinicians spend time and resources assessing SDOH that may be impacting their ability to treat the patient.34
CMS has also implemented Section 1115 waivers which authorize certain states to finance additional social services under state Medicaid programs.19 For example, since 2012, Oregon has fostered partnerships between its Medicaid ACOs (called coordinated care organizations (CCOs)) and local community-based organizations to address both individual social needs and community SDOH. In 2018, the Oregon Legislature passed the Supporting Health for All through Reinvestment Initiative, which requires CCOs to spend a predetermined portion of their revenue on SDOH such as housing supports, economic stability, neighborhood and built environment, education, and social and community health. Concurrently CCOs might be rewarded through a new performance-based payment model for the efficient provision of health-related services.35 In 2020, Massachusetts Medicaid (MassHealth) launched a 3-year pilot of the Flexible Services Program, which requires MassHealth’s ACOs to screen their members for health-related social needs and to provide navigation support to connect members with services related to housing, food insecurity, transportation, and utility support.36 While CMS encourages state Medicaid ACO programs to address social needs systematically through universal screening and referral, and partnerships with community-based organizations that specialize in social resource provision, many of these initiatives are pilot projects rather than standard component of care. Long-term financial resources, workforce supports, and stronger community partnerships are required to increase practice-level capacity to identify and address SDOH.
Third, as opposed to patient and practice factors, not many physician level factors contributed to a higher odds of documenting SDOH. Female physicians were more likely to screen for SDOH, perhaps related to prior findings that they spend more time with patients and generally are rated more highly for empathy.37,38 Interestingly, location of training, specifically training in an RTT or THC was not associated with the future likelihood of capturing SDOH, perhaps because the impact of this training was trumped by patient and practice factors, both of which can be associated with training in these locations.
Our last major finding was that the most common method of documenting SDOH in EHRs was through free-text clinical notes. It is important to emphasize that capturing SDOH in clinical notes, rather than as structured data elements, limits the ease to which these data can be used in efforts to address SDOH on a population level or to risk stratify patient panels within practices for payment purposes. It also highlights low awareness and the need for policy changes to incentivize the systematic documentation and addressing of SDOH in health care settings to improve health outcomes and equity. Alternatively, it signals a need for natural language processing, as opposed to traditional methods of gleaning data from EHRs, to accurately capture and report on SDOH needs within a population.
Limitations
This study has several limitations. First, the cross-sectional design limits the ability to establish causality between the documentation of SDOH and the associated factors. Our findings reflect associations but cannot determine whether the identified factors directly influence SDOH documentation. Longitudinal studies are needed to explore causal relationships and the impact of SDOH documentation on clinical outcomes over time. Second, the reliance on self-reported data from family physicians introduces the possibility of both recall bias and social desirability bias. Physicians may overestimate their frequency of documenting SDOH, particularly in settings where SDOH screening is highly encouraged or incentivized. This may explain the higher rates of documentation observed in this study compared with prior research using observational data. Future studies could combine survey data with EHR or claims data to validate self-reported practices and provide a more objective measure of documentation rates. Third, approximately 10% of family physicians took CCQ yearly during the 10-year recertification process. Although the study population is a representative sample with nearly complete responses, there may still be nonrandomness to the cohort. Fourth, SDOH screening can be completed by various members of a clinic team. the CCQ specifically captures self-reported physician-documented SDOH, without accounting for SDOH documentation or interventions completed by other members of the care team (eg, medical assistants, social workers, or care coordinators). Lastly, we were unable to assess any SDOH-related interventions that providers or practices were engaged in. Future research should explore how documentation of SDOH influences the level of family physician engagement in clinical and population-based activities and referrals that patients receive using EHR and claims data. Moreover, how meeting the social needs can improve health outcomes should be also evaluated.
Conclusion
Family physicians who participated in VBP models or worked in practices with more SDOH-related staffing resources or community relationships were more likely to screen for and record SDOH. The study highlights the important role of policies supporting VBP, team-based practice and community engagement in meeting social needs for patients in primary care and promoting health equity. Caring for communities with higher social needs also increased screening and recording, emphasizing that supportive policy should prioritize primary care practices in these locations. These policies could be direct payments for screening, population social risk-adjusted payments, as done in many CMMI payment demonstrations underway, or practice payment combined with social service resources, as is done in several states under their 1115 waiver. With larger physician survey samples linked to these different payment policies, it may be possible to evaluate which of these models are more likely to support screening and recording of SDOH.
Appendix
Appendices.
Appendix 1.
Survey Questions from the 2022 American Board of Family Medicine Continuing Certification Questionnaire
Appendix 2.
Full Logistic Regression Results
Notes
This article was externally peer reviewed.
This article was externally peer reviewed.
Funding: None.
Conflict of interest: None.
To see this article online, please go to: http://jabfm.org/content/38/2/290.full.
- Received for publication July 25, 2024.
- Revision received October 21, 2024.
- Accepted for publication November 4, 2024.
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.