Abstract
Primary care practices are under pressure to address patients' social determinants of health (SDOH). However, the extent to which these practices have this ability remains unknown. The objective of this study was to examine the association between physician, practice, and community characteristics and the ability of family medicine practices to address patients' SDOH. This cross-sectional study used data from the American Board of Family Medicine Continuing Certification Questionnaire from 2017 to 2019, with a 100% response rate. Respondents rated their practice's ability to address SDOH, which was dichotomized as high or low. Sequential multivariate logistic regression determined the association of the reported ability to address SDOH with physician, practice, and community characteristics. Among 19,300 respondents, 55.6% reported a high ability to address patients' SDOH. Across models controlling for different groups of variables, characteristics persistently positively associated with ability to address SDOH included employment at a federally qualified health center (Odds Ratios [OR] = 2.111–3.012), federally funded clinic (OR = 1.999–2.897), managed care organization (OR = 2.038–2.303), and working collaboratively with a social worker (OR = 2.000–2.523) or care coordinator (OR = 1.482–1.681). Characteristics persistently negatively associated with the ability to address SDOH were practicing at an independently owned (OR = 0.726–0.812) or small practice (OR = 0.512–0.863). While results varied across models, these findings are important for developing evidence-based policies and recommendations for resource sharing and allocation in clinics and communities. Ensuring availability and access to allied health professionals and community resources may be key components in Family Medicine clinics addressing SDOH.
Background
The health care system is often the main focus to improve health outcomes. However, it is now known that medical care alone is insufficient for improving patient health. Recent studies have shown that up to 80% of health outcomes are due to health-related behaviors, socioeconomic factors, and environmental factors, 1 otherwise known as social determinants of health (SDOH). Furthermore, evidence from social needs' interventions has shown promise for improving health outcomes. 2 –5 This knowledge has led to a “call to action” to develop care delivery strategies that address social needs in health care practice. 6 –8
Primary care has emerged as one of the most essential settings to address SDOH, arguably due to the preventive efforts, regular patient contact, and better continuity of care compared to other specialty settings. In fact, both the American Academy of Family Physicians and the Society for General Internal Medicine have issued statements on the importance of physicians addressing SDOH with patients. 9,10 Second, several toolkits have been developed and validated to expand SDOH interventions in primary care settings. 11 –13 Finally, the movement of public and private payers toward value-based care models encourages physicians and practices to target SDOH by incentivizing them to address social needs in clinical practice. 14,15
While the health system continues to develop methods to integrate SDOH interventions into clinical practice, little is known about the tendency of primary care practices to address SDOH or whether they are associated with patient needs. For example, a small survey of family physicians found the majority participated in at least one clinical action to address SDOH, and higher levels of clinical engagement were associated with physician experience, practice ownership, type of clinic, and the presence of other professionals, including social workers and community health workers. 16
This study intended to build on this past work by using a survey from a large, nationally representative cohort of family physicians with a 100% response rate. This study also includes additional data on physician characteristics, practice site, location and size, and support staff. The objective of this study was to examine the association of physician, practice, and community characteristics with the ability of family physician practices to address patients' SDOH needs.
Conceptual model
The conceptual framework for this study was based on the Chronic Care Model (CCM), 17 a widely used approach in health services research; the CCM explains how patient outcomes are improved by promoting productive interactions between informed, activated patients and prepared, proactive health care teams. These interactions depend on elements within the health care system, including the larger community. For this study, specific physician, practice, and community characteristics were evaluated as health care system elements to determine their relationship in predicting the ability of a primary care practice to address SDOH.
Methods
Datasets and sample
The American Board of Family Medicine (ABFM) Continuing Certification Questionnaire 2017–2019 was used to obtain data on practice characteristics. This questionnaire was completed by family physicians to continue their ABFM certification as a required component of the certification process. 18 Each respondent was given a core set of questions and 1 of 5 modules that produced representative subsamples. For the 2017–2019 examination years, 26,923 nationally representative physicians completed the questionnaire with a 100% response rate. After excluding those who did not provide direct patient care or continuity care (n = 7090), those with incomplete demographic data (n = 310), and those living outside the United States or with addresses unable to be geocoded (n = 223), there were 19,300 participants included in the final sample. Physician characteristics—age, gender, degree type, and location of training—were gathered from the ABFM administrative data sets.
To obtain community characteristics, the Area Health Resources File (AHRF), the Food Environmental Index (FEI), and the Social Deprivation Index (SDI) were used. Maintained by the Bureau of Health Workforce, the AHRF aggregates health workforce, health infrastructure, and census data at the county level. 19 FEI is available on the County Health Rankings website 20 and includes index factors that contribute to a healthy food environment, from 0 being the worst to 10 being the best.
The 2021 County Health Rankings used data from 2015 and 2018 for the FEI measure used in this study. Created and managed by the Robert Graham Center, SDI is a composite measure of area-level deprivation based on 7 demographic characteristics used to quantify the socioeconomic variation in health outcomes. 21 County settings were captured using the Office of Management and Budget (OMB) statistical area definitions for metropolitan, micropolitan, and noncore (rural) status. 22 The county was used for community-level variables as the catchment area of primary care practices can vary substantially by setting and local health care markets.
Dependent variable
The primary outcome of this study was physicians' self-reported ability of their clinic to address patients' social needs in their clinic. The ABFM questionnaire asks physicians to report whether their “clinic has the resources and tools to address patients' social needs, such as dedicated staff and linkages to community programs.” Responses to this question were given on a 10-point scale, with 1 indicating “strongly disagree” and 10 indicating “strongly agree.” Consistent with past work, these results were dichotomized into a response of 1–6 as low ability to address patients' social needs and 7–10 as high ability. 23,24 In addition, the dependent variable was modeled as binary, as opposed to continuous, as the objective was more to understand characteristics associated with high ability to address SDOH rather than an increase in average ability to address SDOH, which may still be in the “low” range.
Independent variables
Variables were selected to tap different domains of the CCM, specifically physician, practice, and community characteristics. Physician characteristics included age, gender, degree type, location of training, race, and ethnicity. Practice characteristics included the type of practice (eg, health-system owned, independently owned, Federally Qualified Health Center [FQHC], rural clinic), practice size, practice specialty mix, and other health professionals at the practice site, specifically licensed social workers (LSW) and care coordinators. The ABFM questionnaire contains an item for physicians to report the percentage of their patient population that is part of a vulnerable group, defined as uninsured, Medicaid users, homeless, low income, non-English speaking, or racial/ethnic minority. This variable was included in the model as it may indicate the care of patients with higher social needs. Community characteristics included county-level percent of the population living in poverty, SDI, FEI, and rurality.
In addition, data from one module question were used to determine the practice's exposure to value-based payment models. The practice address was collected and geocoded by county to obtain community characteristics. These variables included county-level social environments and settings captured using metropolitan, micropolitan, and noncore (rural) status, poverty rate (total and child), FEI, and SDI.
Analysis
The sample was restricted to physicians who primarily provided outpatient continuity care. Descriptive statistics were then generated for all variables. Next, Chi-square and t-tests were used to compare demographics and practice characteristics by OMB rural categories and the ability to address SDOH. Also assessed was whether a practice's ability to address social needs was associated with practice organization and the percentage of the vulnerable population served.
Collinearity of county-level variables was assessed, and the child poverty variable was removed due to its strong correlation with the overall poverty rate. Finally, a series of multivariate logistic regression analyses was performed. For Model 1, the adjusted associations were determined between a clinic's capacity to address SDOH needs and practice-level characteristics. A practice working collaboratively with an LSW or care coordinator was controlled for in Model 2. Model 3 controlled for community characteristics, which indicated a higher population social need. Finally, Model 4 was run, utilizing all variables, on the subset of physicians who answered the module question on value-based payment exposure. This last model was used to determine the associations between this practice element and the ability to address SDOH.
A sensitivity analysis was conducted in which the dependent variable was defined as continuous rather than dichotomous, using linear regression to check for the robustness of findings to variable specification. SAS v9.0 (Cary, NC) was used for all analyses. This study was approved by the Institutional Review Board of the American Academy of Family Physicians.
Results
Of the 19,300 physicians in the analytical sample, 55.6% reported that their clinic had a high ability to address SDOH. The majority of participants were White (72.7%), non-Hispanic (93.2%), male (56.0%), and 40 years of age and older (90.6%) (Table 1). Most participants worked in a metropolitan area (84.7%) and were employed in either a health system-owned (34.8%) or independently owned (32.7%) practice. In bivariate analyses, physician characteristics associated with a high ability to address SDOH included identifying as male, having a DO degree, and being an international medical graduate (IMG).
Physician, Practice, and Community Characteristics by Ability to Address Social Determinants of Health
P-value <0.01, ** P-value <0.001, P-value <0.0001.
IMG, international medical graduate.
The adjusted associations of physician characteristics across all models are shown in Table 2. Across all models, females had lower odds of reporting that their clinic had a high ability to address SDOH needs than their male counterparts. Compared to MDs, DOs had higher odds of reporting a high ability to address social needs; however, significance was lost when adjusting for having an LSW or care coordinator on staff. IMGs had higher odds of reporting high ability to address SDOH compared to US or Canadian graduates in all models.
Adjusted Associations of Physician, Practice, and Community Characteristics with Ability to Address Social Determinants of Health
P-value <0.01, ** P-value <0.001, *** P-value <0.0001.
IMG, international medical graduate.
The adjusted associations of practice characteristics across all models are shown in Table 2. Using a comparison group of physicians employed at health system-owned clinics, significant variation was observed in practices' ability to address SDOH based on the practice site. Respondents working at independently owned clinics had lower odds of reporting the ability to address SDOH (Odds Ratios [OR] ranging from 0.726 to 0.812). However, this significance was lost when adjusting for exposure to value-based payment models.
Those working for managed care practices and FQHCs had higher odds of reporting a high ability to address SDOH across all models. Physicians working in an academic medical center or nonfederal government clinic had higher odds of addressing SDOH only in Model 1 when adjusting for demographic and practice-level characteristics. For all other models, no significant difference was observed between these groups. For physicians working in rural health clinics, workplace clinics, or other clinics, no significant difference was observed in the reported practice ability to address SDOH for their patients.
Other practice characteristics that showed a significant difference in the ability to address SDOH included practice size, the percentage of vulnerable population served, and having an LSW or care coordinator on staff. A smaller practice size is associated with significantly lower odds of addressing SDOH compared to practices that employ more than 20 physicians. There was no association observed for practices with 2–20 physicians when adjusted for exposure to value-based payments.
Compared to physicians who saw a small percentage of vulnerable patients (less than 10%) in their practice, physicians with a vulnerable patient population of greater than 50% had significantly higher odds of addressing SDOH when adjusted for demographic and practice-level characteristics. At the same time, all other models showed no significant difference between the 2 groups. Finally, physicians who reported having a social worker or care coordinator on staff had significantly higher odds of addressing SDOH across all models than physicians who did not have such staff (OR ranged from 2.000 to 2.523 and 1.482 to 1.681, respectively).
When evaluating county characteristics, no association was found between practices' ability to address SDOH and the percentage of the county population living in poverty, county SDI, or county FEI (Table 2). No difference was found in the ability to address social determinants in micropolitan and nonmetropolitan counties compared to metropolitan counties.
No association was found between practices' ability to address social needs and exposure to value-based payment models.
A sensitivity test was performed with the ability to address SDOH as a continuous outcome for Model 3. The results were similar in direction and magnitude to those of the logistic regression (Supplementary Table S1).
Discussion
This large national study on Family Medicine practices' abilities to address SDOH demonstrated that, while a slight majority of physicians reported that their clinic had a high ability to address patients' social needs, this ability varies greatly.
Variation was seen by practice size, with smaller practices (≤20 physicians) having less ability to address SDOH than those employed in larger practices. This may be explained by larger practices benefiting from economies of scale, having more financial resources and capacity for support staff compared to smaller practices. Supporting this is the finding that practices with social workers and care coordinators had higher ability to address SDOH. However, these professionals were present in a minority of practices with high ability to address SDOH, indicating that other professionals may serve this function in other practices.
These results provide insights into care delivery and payment reform to facilitate addressing SDOH in clinical practice. While Kovach et al. 16 suggest that clinical settings should adopt the characteristics of practices that are successfully engaged in addressing SDOH, results from this study indicate that policies addressing the availability of and access to community resources may be a more effective solution. Policies promoting resource sharing among small practices may provide more equity in the ability of practices to address SDOH. 25,26 For example, small practices could share health care personnel such as social workers, care coordinators, or health navigators through a community-based network to connect patients to needed social services.
There is a wide variation in the ability to address SDOH among practice types. In particular, respondents in Health Management Organizations/managed care practices, FQHCs, and Federal sites reported their clinic had higher ability to address SDOH compared to other practices. These results were robust across models that accounted for community characteristics and practice features, indicating these practice types may have best practices other organizations could adopt to address SDOH.
Interestingly, none of the community characteristics studied had a significant association with the practice's ability to address SDOH. This may indicate that, while the burden of addressing patients' social needs is equally distributed among areas, practices respond equally well to local needs. Alternatively, there may be a misalignment of practice catchment area with county boundaries, which would lead to a null finding. The study included a variable on the percentage of vulnerable patients the practice serves, which should be a better marker of social needs of patients; however, it was only in one model. Future work could use practice catchment area to further elucidate how community context impacts the ability to address SDOH.
The dependent variable is self-reported and is likely interpreted as an “absolute” ability to address SDOH rather than relative value. Physicians in practices with high ability to address SDOH also reported a higher proportion of vulnerable patients. In adjusted analysis, the proportion of vulnerable patients lost significance once practice features were accounted for, which indicates the ability to address SDOH can occur even in practices with a low need.
This study is subject to limitations. First, this was a cross-sectional study, and causality cannot be confirmed. Second, the primary outcome does not assess how practices are addressing SDOH. Future studies should assess specific methods to address social needs. Third, as responses to this question were self-reported, there was potential for response biases. Finally, although this study controlled for many physician, practice, and community characteristics, there may have been unmeasured confounders.
Conclusion
In conclusion, this study indicates that the ability of a practice to address SDOH is mostly associated with practice site, practice size, and working with a social worker or care coordinator. These findings are important for developing evidence-based policies and recommendations for resource sharing and allocation in clinics and communities. Future studies should examine interventions that focus on clinical collaborations with allied health professionals and community and social service organizations.
Footnotes
Authors' Contributions
Ms. Sand: Writing-original draft preparation and visualization; Dr. Peterson: Writing-review and editing, supervision, and methodology; and Mr. Morgan: Validation and formal analysis.
Author Disclosure Statement
No competing financial interests exist.
Funding Information
Dr. Peterson and Mr. Morgan were supported by the Federal Office of Rural Health Policy (FORHP), Health Resources and Services Administration (HRSA), and US Department of Health and Human Services (HHS) under cooperative agreement no. U1CRH30041. The information, conclusions, and opinions expressed in this article are those of the authors, and no endorsement by FORHP, HRSA, HHS, or the University of Kentucky is intended or should be inferred. Ms. Sand was supported by the American Board of Family Medicine Foundation.
Supplementary Material
Supplementary Table S1
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
