Patient and Care Team Perspectives on Social Determinants of Health Screening (2024)

Key Points

Question Are patient and clinician factors associated with early implementation of social determinants of health (SDOH) screening in primary care, and what strategies can improve these efforts?

Findings In this qualitative study of 78 928 primary care visits from the inception of primary care–based SDOH screening, visits with a physician assistant, belonging to a racial minority group, and having noncommercial/nonprivate health insurance were associated with greater screening likelihood. Stakeholders suggest that patient-clinician rapport, practice champions, streamlined questions, and referral follow-up ability may improve screening implementation.

Meaning Results of this study suggest that primary care SDOH screening is feasible but limited by barriers that can be overcome with consideration of stakeholder feedback.

Importance Health systems in the US are increasingly screening for social determinants of health (SDOH). However, guidance incorporating stakeholder feedback is limited.

Objective To examine patient and care team experiences in early implementation of SDOH screening in primary care.

Design, Setting, and Participants This qualitative study included cross-sectional analysis of SDOH screenings during primary care visits from February 22 to May 10, 2022, primary care team member interviews from July 6, 2022, to March 8, 2023, and patient stakeholder engagement on June 30, 2022. The setting was a large southeastern US health care system. Eligible patients were aged 18 years or older with completed visits in primary care.

Exposure Screening for SDOH in primary care.

Main outcomes and Measures Multivariable logistic regression evaluated patient (eg, age, race and ethnicity) and care team characteristics (eg, practice type), and screening completeness. Interviews contextualized the quantitative analysis.

Results There were 78 928 visits in practices conducting any SDOH screening. The population with visits had a mean (SD) age of 57.6 (18.1) years; 48 086 (60.9%) were female, 12 569 (15.9%) Black, 60 578 (76.8%) White, and 3088 (3.9%) Hispanic. A total of 54 611 visits (69.2%) were with a doctor of medicine and 13 035 (16.5%) with a nurse practitioner. Most had no SDOH questions answered (75 298 [95.4%]) followed by all questions (2976 [3.77%]). Logistic regression analysis found that clinician type, patient race, and primary payer were associated with screening likelihood: for clinician type, nurse practitioner (odds ratio [OR], 0.13; 95% CI, 0.03-0.62; P = .01) and physician assistant (OR, 3.11; 95% CI, 1.19-8.10; P = .02); for patient race, Asian (OR, 1.69; 95% CI, 1.25-2.28; P = .001); Black (OR, 1.49; 95% CI, 1.10-2.01; P = .009); or 2 or more races (OR, 1.48; 95% CI, 1.12-1.94; P = .006); and for primary payer, Medicaid (OR, 0.62; 95% CI, 0.48-0.80; P < .001); managed care (OR, 1.17; 95% CI, 1.07-1.29; P = .001); uninsured or with Access Health (OR, 0.26; 95% CI, 0.10-0.67; P = .005), and Tricare (OR, 0.71; 95% CI, 0.55-0.92; P = .01). Interview themes included barriers (patient hesitancy, time and resources for screening and referrals, and number of questions/content overlap) and facilitators (communication, practice champions, and support for patient needs).

Conclusions and Relevance This qualitative study presents potential guidance regarding factors that could improve SDOH screening within busy clinical workflows.

Health systems in the US recognize the importance of social determinants of health (SDOH) in patient outcomes and care. The SDOH are economic and social conditions affecting health outcomes,¹ health care use,² and health inequities.³ Health systems are increasingly engaging in SDOH screening.⁴ Although such screening can potentially improve health outcomes and reduce health care use,^5,6 there is limited peer-reviewed evidence incorporating patient and clinician or care team characteristics and perspectives when describing early screening initiatives.

Given the personal nature and limited evidence guiding SDOH screening adoption,^7-9 it is critical to understand stakeholder perspectives. Prior research indicates that health care professionals recognize the importance of addressing patient SDOH needs and strive to adopt patient-centered approaches¹⁰ but face ethical and time-related challenges.^8,11,12 Existing work reports greater SDOH screening uptake in primary care vs specialist visits and lower completion among patients requiring interpreters and patients with racial and ethnic minority status.⁷ Studies on patient and caregiver perspectives have documented SDOH screening acceptability and preferences.¹³ The role of practice and care team characteristics in screening uptake has not been assessed within a multistakeholder analysis.

To address this research gap, we conducted a qualitative study of a large southeastern US health care system's experiences during the early stages of SDOH screening in primary care. Quantitative analysis examined practice, care team, and patient characteristics and SDOH screening uptake. Qualitative analysis engaged team member feedback. Patient experts informed interview protocols and finding interpretation. Our goal was to identify barriers and facilitators to SDOH screening within primary care to inform future screening.

This qualitative study was classified as exempt by the Prisma Health institutional review board in accordance with 45 CFR §46. In February 2022, Prisma Health, South Carolina’s largest nonprofit health system with approximately 1.5 million unique patients annually, began screening adults for SDOH needs in primary care practices with the goal of annual screening. Practices had implementation flexibility and determined how and when to screen during the clinical workflow. Patients were screened using a 16-question electronic health record (EHR)–embedded survey (eTable 1 in Supplement 1). Questions were chosen using validated questionnaires and clinical input on system priorities and resource availability. Answers triggered automated input of community-based service information curated to patient SDOH needs and location into patient after-visit summaries using an EHR-compatible platform connecting patients to community-based organizations (NowPow; Unite Us). Practices provided the after-visit summaries to patients at visit end. Reporting follows the 21-item Standards for Reporting Qualitative Research (SRQR) reporting guideline.

The study population included patients aged 18 years or older with a visit in a family or internal medicine practice in the northwestern region of South Carolina from February 22 to May 10, 2022. Visits classified as future, cancelled, no show, or left without being seen were excluded. The last screen on a day was the patient final value, and the same patient could have multiple visits over the study period. In 2021, the northwestern region (4 counties) had 813 069 inhabitants, with 14.2% in poverty (11.4% nationally) and 13.9% uninsured (10.2% nationally). The population is 75.8% White, 14.6% Black, 6.5% Hispanic, 0.4% American Indian or Alaska Native, 1.6% Asian, and 0.1% Native Hawaiian or Other Pacific Islander.¹⁴

The primary outcome was SDOH screening completion status. Visits with a response to at least 1 question were deemed partial screening while complete screening included responses to all questions. Our primary outcome compared visits with complete or partial screening (any screening) with no screening. Secondary outcomes compared visits with complete vs partial or no screening and visits with complete screening vs partial screening.

Potential explanatory variables included practice type (family or internal medicine), clinician qualification (medical doctor, doctor of osteopathic medicine, nurse practitioner, and physician assistant), patient demographic characteristics (age, sex, race and ethnicity [treated as classified in the electronic medical records as separate fields], preferred language, primary payer), and SDOH risk (calculated as the ratio of screener questions with positive responses to the total number of questions answered by patients). Race and ethnicity came from the EHR and thus were primarily patient self-reported. Race is reported as Asian, Black, White, 2 or more races, other race, patient refused, or unknown. Other race comprises American Indian or Alaska Native, Native Hawaiian or Other Pacific Islander, and other as reported in the EHR. Ethnicity is reported as in the EHR. We included SDOH risk to test whether patients with a need might be more likely to be screened (ie, care team members suspect a need or patients are more likely to answer questions).

Binary logistic regression was used to determine the odds of screening completion. Standard errors were clustered by practice to account for practice-specific differences. A 95% CI not including 1 indicated statistical significance. We tested for multicollinearity using variance inflation factors and omitted variable bias using the Ramsey Regression Equation Specification Error Test (RESET). Analysis was conducted using Stata/MP, version 11 (StataCorp LLC).

Six practices were categorized as higher-adopting facilities as they performed SDOH screening during at least 4.0% of visits over the study period. Two of these practices were excluded because of involvement in other SDOH-related studies. Lower-adopting practices performed at least 10 screenings but in less than 2.0% of visits. Four practices met this criterion, but 1 practice was excluded because of involvement in SDOH pilot efforts. Higher- and lower-adopting was defined by quantitative analysis. We excluded practices performing no or minimal screening because we wanted to learn from those practices with some screening familiarity and those screening at both higher and lower levels. These 7 practices were approached for interviews of primary care team members (ie, physicians, administrative staff, nursing staff, and allied health professionals). Six practices participated in a total of 9 interviews (at least 1 interviewee from each of these 6 practices). Interview findings contextualized the quantitative analysis.

Two trained medical students (E.K. and M.J.) conducted and recorded 9 semistructured interviews online between July 6, 2022, and March 8, 2023. The students had not met the interviewees or worked in these clinics prior to the interviews. Interview questions focused on potential barriers and facilitators to screening (eMethods 1 in Supplement 1). Oral consent was obtained prior to interviews. Interviews were transcribed verbatim by a speech-to-text service (rev.com). Interview recordings were accessible only to interviewers and the team member uploading for transcription. Interviewers asked questions aimed to not yield identifying information. Additionally, transcripts were kept either on secure file-sharing systems or on password-protected computers. Using a web application (Dedoose), transcripts were coded by 2 research team members (D.G. and M.M.) and analyzed using an inductive grounded theory approach, in which important concepts and themes are derived from close reading of the text, and similar concepts are grouped into conceptual categories (codes). No further interviews were necessary as theme saturation was achieved.

To ensure the research was relevant and ethical for patients and the broader community, we included a meeting with patient experts from the University of South Carolina Patient Engagement Studio (PES) in our research strategy.^15-17 The PES is built on guidance from the Patient-Centered Outcomes Research Institute and provides structured opportunities for research teams to engage with community-recruited patient experts. Patient expert refers to individuals or caregivers with substantial health system interaction due to their health conditions who are trained in communication, research methods, and team building.

The research team met with patient experts on June 30, 2022, prior to interviews with primary care practices. In accordance with standard PES processes,¹⁸ patient experts were provided the health system SDOH screening tool as presession reading material. Discussion topics at that meeting included screening and referral processes (eMethods 2 in Supplement 1). Patient expert feedback was incorporated into the research process through practice interview topics and by incorporating what we heard from patient experts when discussing study results.

Over the study period, there were 147 096 practice visits, with 3630 (2.5%) involving complete (2976 [3.8%]) or partial (654 [0.8%]) SDOH screening. In the restricted sample, 22 of 58 practices (37.9%) performed any screening during the study period (Table 1). Of the 78 928 visits (mean [SD] age of 57.6 [18.1] years; 48 086 [60.9%] were female, 12 569 [15.9%] Black, 60 578 [76.8%] White and 3088 [3.9%] Hispanic) in the restricted sample, 41 574 (52.7%) were in family medicine and 37 354 (47.3%) in internal medicine practices. Most visits were with medical doctors (54 611[69.2%]), followed by nurse practitioners (13 035 [16.5%]), doctors of osteopathic medicine (5877 [7.4%]), and physician assistants (2958 [3.8%]). On average, patients had a mean (SD) of 0.08 (0.13) (95% CI, 0.08-0.09) positive responses per SDOH question answered.

The SDOH screener responses in order of question appearance are given in eTable 1 in Supplement 1. Earlier questions were more likely to be asked and answered. Overall, patient response refusal was low (≤3.3%). Descriptive statistics for the unrestricted sample (visits to all practices) are given in eTable 2 in Supplement 1.

Table 2 displays regression results examining factors associated with any SDOH screening (complete or partial screening vs no screening) in the restricted (model 1) and unrestricted (model 2) practice samples. In model 1 (restricted), compared with visits with a medical doctor, visits with a physician assistant had 3.11 (95% CI, 1.19-8.10; P = .02) greater odds of any screening done, while visits with nurse practitioners had significantly lower odds (odds ratio [OR], 0.13; 95% 0.03-0.62; P = .01) of any screening done. Visits with patients identifying as Asian (OR, 1.69; 95% CI, 1.25-2.28; P = .001), Black (OR, 1.49; 95% CI, 1.10-2.01; P = .009), or 2 or more races (OR, 1.48; 95% CI, 1.12-1.94; P = .006) were more likely to have any screening compared with visits with patients identifying as White. With regard to primary payer, visits where patients had managed care had 1.17 (95% CI, 1.07-1.29; P = .001) greater odds of any screening compared to visits where patients had private or commercial payers. Visits where patients had Medicaid (OR, 0.62; 95% CI, 0.48-0.80; P < .001), were uninsured or had Access Health (OR, 0.26; 95% CI, 0.10-0.67; P = .005) or had Tricare (OR, 0.71; 95% CI, 0.55-0.92; P = .01) had lower odds of any screening. Practice type, patient age, sex, language, and ethnicity had no significant associations with screening likelihood. Results were consistent in model 2 (unrestricted) except for visits with physician assistants and uninsured patients, where the finding was not significant.

We also compared visits completing the entire screening questionnaire vs partial or no screening (Table 3) for the restricted practice sample. In model 3, compared with visits with a medical doctor, visits with a physician assistant had 3.78 times (95% CI; 1.43-10.0; P = .007) greater odds of screening completion while visits with a nurse practitioner had lower screening completion odds (OR, 0.15; 95% CI, 0.03-0.75; P = .02). Visits where patients identified as Black had greater odds of screening completion (OR, 1.33; 95% CI, 1.01-1.74; P = .04) than visits where patients identified as White. Visits where patients had managed care had 1.15 (95% CI, 1.05-1.26; P = .002) times greater screening completion odds than visits where patients had private or commercial payers. However, screenings were less likely to be complete if patients had Medicaid (OR, 0.53; 95% CI, 0.40-0.72; P < .001), Tricare (OR, 0.76; 95% CI, 0.58-0.98; P = .04), or were uninsured or had Access Health (OR, 0.14; 95% CI, 0.05-0.40; P < .001). Results were consistent in model 4 comparing the odds of complete vs partial screening.

Model 5 extended model 4 to include patient SDOH risk from screening responses. Patient SDOH risk was not associated with screening completion (OR, 1.03; 95% CI, 0.56-1.88; P = .93). Results in model 5 are consistent with model 4.

All models had variance inflation factors of less than 10 indicating absence of multicollinearity. Models 4 and 5 had omitted variable bias.

We identified 7 themes regarding barriers and facilitators from health care team member interviews for implementing SDOH screening (Table 4). Care team members reported patient reluctance in responding to screener questions. Hesitancy was attributed to perceptions about questions being intrusive or offensive. Interviewees reported patients reacting unfavorably to sensitive questions (eg, violence/abuse, financial strain). Time to administer the screener, interpret results, and address identified needs posed challenges with existing workloads.

Clinicians expressed concerns about potential patient response burden and overlap with routine care questions (eg, stress and Patient Health Questionnaire 2). Clinicians suggested streamlining the screener by combining multiple related questions and then tailoring subsequent questions based on patient initial responses.

Some clinicians felt inadequately trained in navigating the screening tool and expressed uncertainty about effective use of screening results. Many practices lacked social workers or resource navigators to connect patients with resources and follow up on referrals. Clinicians felt their attention diverted from the primary goal of medical care provision.

Care team members reported that screening facilitated patient care by uncovering socioeconomic issues not identified in routine care. Practices that informed patients about the screening purpose, assured them it would not affect care, and obtained verbal consent prior to screener administration perceived more successful uptake.

Some practices identified practice champions as being responsible for screening implementation and supporting patient needs. Some practices had a referral coordinator or social worker who connected patients to community-based resources and provided follow-up support. Clinicians reported they would benefit from training on how to best use screening.

Table 5 presents feedback from patient experts. Patient experts preferred that screening be done at annual appointments to allow for discussion time and in the examination room to ensure privacy. Patient experts emphasized rapport building between patients and care teams and providing information about the screening purpose. They expressed the importance of empathetic clinicians performing screening. Recommendations for rephrasing questions included expanding the partner violence or abuse questions (eTable 2 in Supplement 1) to include safety concerns related to family members, neighborhoods, and caretakers. Patient experts expressed concern about timely referral follow-up.

This qualitative study assessed factors associated with SDOH screening completion in primary care and explored patient and care team member perspectives on screening. We found that clinician type, patient race, and primary payer were linked to any screening but that practice type, patient age, sex, language, ethnicity and SDOH risk were not.

Completion rates differed in this study (3.8%) from previous research (58.7%)⁷ also examining systemwide SDOH screening implementation. This may be related to study duration, timing (intra–COVID-19 pandemic vs pre–COVID-19 pandemic), or implementation (recommendation for all primary care patients vs preassigned screening).⁷ Based on qualitative interviews, our study completion rates may be affected by the desire to receive more resources to support patient referrals.

Our findings suggest that primary care visits with nonphysician clinicians, such as physician assistants, may be favorable for SDOH screening. However, this result did not hold for nurse practitioners and deserves further research, as previous studies demonstrated nonphysician clinician confidence in addressing SDOH needs and greater community-based resource awareness.¹⁹ Clinician type could be serving as a proxy for visit type as our data set did not include visit reason. Consistent with previous studies,²⁰ our interview-based findings suggest that clinicians faced an additional time burden from incorporating SDOH screening, which they perceived to affect care provision.

We found patients with managed care to be more likely to be screened, while those with Medicaid and those who were uninsured or had Access Health and Tricare were less likely. Medicare and Medicare Advantage had no effect relative to private or commercial payer status. Patients with Medicaid and uninsured or had Access Health may benefit most from screening; therefore this finding is critical for further implementation. Of note, these patients may have been screened via other programs at the health system thus, lack of screening in primary care is not necessarily reflective of screening otherwise.

A lack of association between screening and other patient characteristics (age, gender, language, ethnicity, SDOH risk) suggests that perhaps these characteristics are not associated with SDOH needs in the perceptions of those performing screening. These results differed from previous research that found members of racial and ethnic minority groups less likely to be screened,⁷ thereby providing support for universal implementation across primary care practices as a potential mitigation against screening disparities.⁷

In our quantitative analysis, questions appearing later in the screener were less likely to be completed. Interviews further explained this finding as questionnaire length and repetitive questions led to a greater perceived patient response burden by health care clinicians. Although there is no consensus on screener length, existing tools range from 6 to 23 questions.²¹ Generally, short-form surveys are more acceptable to patients.²² Notably, patients did not express the same concerns as clinicians about survey length or repetitiveness.

Interviews and patient expert feedback found that patient–care team communication is crucial for screener uptake. Sensitive questions about patient needs may lead to incomplete or untruthful responses if patients have privacy concerns,^10,23 feel embarrassed, or fear stigmatization.²⁴ Patient experts and health care team members emphasized rapport building and communicating the screening purpose to mitigate patient concerns and build trust. Future investigation should include assessment of standard phrasing to introduce the screener rationale and consideration of the best location and visit type for screening. Last, patient experts and care team members expressed concerns about referral follow-up, perceiving that care would benefit from an enhanced ability to follow up on referral outcomes.

Our study has a few limitations to be considered. First, findings are restricted to primary care practices within 1 health system in 1 region, limiting generalizability. However, this study is comprehensive by including all primary care practices in 1 region covered by a large health system that statewide serves approximately 25% of residents.¹⁴ Second, we used a convenience sample of practice staff for our qualitative assessment. This restricted our examination of how qualitative themes differed based on practice characteristics. However, practice choice for interviews was based on screening implementation to intentionally capture those screening at higher and lower adoption rates. Third, our data set included whether a survey was taken on MyChart (Epic). No surveys were done on MyChart. Accordingly, we were unable to test screening modality association with screening completion. We also had no information on screening completion via telemedicine vs office visits and did not include this topic in our interview guide. In addition, we do not know at what rate patients refused to verbally consent to screener administration if a practice asked for such consent.

Although health systems face different challenges in implementing SDOH screening, identifying and addressing common barriers are critical for improved patient activation and care collaboration. Future research should focus on robust assessment of strategies to improve screening uptake.

Accepted for Publication: October 19, 2023.

Published: November 28, 2023. doi:10.1001/jamanetworkopen.2023.45444

Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2023 Rudisill AC et al. JAMA Network Open.

Corresponding Author: A. Caroline Rudisill, PhD, MSc, Department of Health Promotion, Education, and Behavior, Arnold School of Public Health, University of South Carolina, 300 E McBee Ave, Ste 401, Greenville, SC 29601 (caroline.rudisill@sc.edu).

Author Contributions: Dr Rudisill and Ms Gupta had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Rudisill, Eicken, Macauda, Self, Thomas, Hartley.

Acquisition, analysis, or interpretation of data: Rudisill, Eicken, Gupta, Macauda, Self, Kennedy, Kao, Jeanty.

Drafting of the manuscript: Gupta, Kao, Hartley.

Critical review of the manuscript for important intellectual content: Rudisill, Eicken, Macauda, Self, Kennedy, Thomas, Jeanty.

Statistical analysis: Rudisill, Gupta, Self.

Obtained funding: Rudisill, Eicken.

Administrative, technical, or material support: Rudisill, Kennedy, Thomas, Kao, Jeanty.

Supervision: Rudisill, Eicken, Macauda.

Conflict of Interest Disclosures: Dr Rudisill reported grants from the Prisma Health Transformative Seed Grant Program during the conduct of the study and The Duke Endowment, Centers for Disease Control and Prevention, Viiv Healthcare, University of Michigan/National Institute on Aging/National Institutes of Health, South Carolina(SC)/NIA/NIH, SC Research Foundation (SCRF)/BlueCross/BlueShield Foundation of SC and National Heart, Lung, and Blood Institute/NIH. Dr Eicken reported grants from Prisma Health Transformative Seed Grant Program during the conduct of the study; grants from the Duke Endowment and grants from the Prisma Health Transformative Seed Grant Program outside the submitted work; Dr Eicken sits on the board of the Piedmont Health Foundation. Ms Gupta reported grants from Prisma Health during the conduct of the study; and support from the Duke Endowment. Dr Self reported grants from Prisma Health during the conduct of the study; personal fees from Companion Animal Parasite Council and personal fees from Merck outside the submitted work. Dr Kennedy reported grants from Prisma Health The Patient Engagement Studio received a portion of the grant to provide feedback during the conduct of the study; and has received 2 Eugene Washington Engagement Awards for capacity building with patients from the Patient-Centered Outcomes Research Institute in 2020 and in 2021. Ms Kao reported grants from Prisma Health Seed Grant during the conduct of the study. Ms Jeanty reported grants from Prisma Health Seed Grant Program during the conduct of the study. Mr Hartley reported grants from Prisma Health Seed Grant Program during the conduct of the study. No other disclosures were reported.

Funding/Support: This research was funded by the Prisma Health Research Seed Grant program.

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

Data Sharing Statement: See Supplement 2.