Abstract
Introduction/Objectives:
Primary care organizations are increasingly collecting data on patients’ social risks, bringing forth an unprecedented opportunity to present combined health and social data that clinical and social care providers could leverage to improve patient care and outcomes. Little is known, however, about how these data could be used and what combinations of specific data elements are most helpful. We explored how primary care staff who provide clinical or social care services view potential benefits of and use cases for combined patient-level clinical and social data.
Methods:
We conducted qualitative interviews or focus groups with 39 social and clinical care providers representing 6 healthcare organizations in San Diego County, California. Interviews were transcribed and analyzed using a deductive thematic analysis approach.
Results:
Overall, both clinical and social care providers noted the value of access to both types of data. Participants highlighted 3 benefits from integrating social and clinical data. The data could: (1) offer providers a more holistic view of patients’ circumstances; (2) strengthen their ability to tailor care to patients’ medical and social conditions concurrently; and (3) enhance coordination across care team members. Interviewees cited specific examples of ways social and clinical data could be paired to improve care.
Conclusions:
Social and clinical care providers alike envisioned multiple uses and benefits of accessing combined individual-level clinical and social data, highlighting the potential for practice and policy innovations to facilitate access and uptake of combined data. Future research should focus on ways to increase accessibility of cross-sector data and evaluate the impact of care informed by combined data on patient social and health outcomes.
Keywords
Introduction
Given mounting evidence on the influence of social risks such as food insecurity, housing instability, and financial strain on health outcomes and avoidable healthcare utilization, primary care organizations are increasingly collecting data on patients’ social risks,1 -4 defined as individual-level adverse social determinants of health. 5 Most commonly, these data are collected through a social risk screening protocol, which may be administered through face-to-face, paper, or electronic questionnaires, and the results of which are commonly stored on electronic health record (EHR) software. 4 Social risk data are primarily used to deliver social care interventions such as referral to community resources (eg, social service organization), or direct provision of social services (eg, on-site food pantry).6 -13 Social risk screening and social care activities are typically performed by navigators, community health workers, medical assistants, care coordinators, or social workers.4,14 -16 The literature to date indicates that in general, clinically focused team members like physicians or advanced practice clinicians have not been routinely viewing or acting upon social risk data. 17
While the biopsychosocial model emphasized the interconnectedness between biological and social factors for an individual’s health over 45 years ago, 18 the application of theory into practice and our understanding of successfully integrating this information into clinical care in electronic data continues. 19 The rapidly increasing availability of social risk data in healthcare settings brings forth an unprecedented opportunity to design ways to combine health and social data in EHRs and other software platforms so that clinical and social care providers alike could simultaneously leverage both sets of data to improve patient care and outcomes. For example, clinical providers could adjust clinical care plans in light of patients’ social circumstances.20 -22 Scholars have also written about the power of social risk data for better understanding and empathizing with patients and have linked this to better patient–provider communication.20,23 To date, however, this has almost exclusively been examined among clinical providers. There is a dearth of research on healthcare-based social care providers’ use of combined health and social data. One reason for this may be that while clinicians have access to both social risk and clinical data, social care providers often do not have access to clinical data, due to privacy and data use concerns. There may be untapped opportunities to consider how social care providers may use social and clinical data in tandem to deliver social care, such as determining eligibility or appropriateness of social services, but to our knowledge, this has not yet been explored in the literature.
We therefore sought to bolster the emerging literature on clinical provider use of combined clinical and social data and shed new light on how staff in social care roles could also leverage these data. The objectives of the current study were to examine how social care providers and clinical providers might use combined clinical and social data and how the uses might differ based on role.
Methods
Setting
This qualitative study was conducted in San Diego County, California, a region with a population of 3.29 million. 24 Our study population was drawn from users and stakeholders of the 211 San Diego Community Information Exchange (CIE), a network of community-based social service and healthcare organizations in San Diego County. The CIE network uses a cloud-based software platform that contains unique longitudinal client/patient records, a resource directory, social needs assessments, and electronic referral management system to support community care coordination.25,26 Created in 2011 with the initial goal of connecting housing providers and other community-based organizations, the CIE has evolved to bring together approximately 130 agencies and over 300 000 clients to date. 26 As part of CIE expansion efforts, clinical information from 3 Federally Qualified Health Centers (FQHCs) began to be integrated into the CIE in 2021, and there is a long-term goal for social information to flow into the EHR, further creating bidirectional opportunities for primary care based social and clinical care providers alike to access social and clinical information. Our team leveraged this unique opportunity both to directly inform the design of clinical data integration into the CIE as well as to advance our understanding more generally of the opportunities for utilizing joint social and clinical patient information to improve patient care and outcomes.
Data Collection
This study is part of a larger formative evaluation for which we conducted semi-structured key informant interviews and focus groups in April to September 2020. For the current study we recruited a purposive sample of 39 social and clinical care providers representing 6 healthcare organizations in San Diego County, CA. Purposive selection took place on 2 levels: first, we endeavored to maximize representation from the primary care organizations within the CIE network but focused recruitment efforts especially on 2 FQHCs that had recently agreed to share limited patient clinical data on the CIE platform. Secondly, within organizations, we sought to recruit participants with diverse social or clinical care roles. Two study team members who were staff at 211 San Diego conducted initial outreach with champions at healthcare organizations with active participation in the CIE network to determine organizational interest in participation. Organizational leaders, in turn, identified staff engaged in CIE to participate in either a focus group or individual interview, based on feasibility and number of potential participants in a specific role. For example, we scheduled a focus group for a large group of patient navigators working in a community health center as they held the same role, whereas leaders and physicians in the same health center each had distinct areas of responsibility.
While interviews and focus groups were initially planned as in-person activities, the arrival of the COVID-19 pandemic led the study team to conduct all activities on Zoom, with participants given the choice of participating with or without video. Interviews were 30 to 60 min long; focus groups were 60 min long. Interviews were conducted by 1 of the 2 co-Principal Investigators following a semi-structured interview guide. After providing informed verbal consent, participants were asked about their interest in and use of respective types of social and clinical data, perceived utility of combined social and clinical data, and examples of combinations of social and clinical data they would find beneficial for care. Individuals were not offered incentives to participate in interviews and focus groups; however, the 2 FQHCs with data-sharing agreements signed a memorandum of understanding regarding the study and were provided with an organizational incentive to make their staff available for study activities during work hours.
Data Analysis
Interviews were audio recorded and transcribed by Rev.com. Transcriptions of interviews and focus groups were anonymized before being uploaded to the qualitative analysis software program Dedoose. We conducted a deductive thematic analysis. Deductive approaches to qualitative data define a priori the categories of concepts that are going to be identified in the data, in contrast to inductive approaches that identify concepts as they emerge from close reading of the data. Our study took place within a larger data dashboard design project that sought answers to very specific questions regarding the use and potential benefits for combined social and clinical data that could inform future dashboards (electronic displays) of these data; a deductive approach was therefore most appropriate. A research assistant and a study team member (CF) piloted a codebook; once definitions stabilized on a first batch of transcripts, the research assistant completed the rest of the coding. To analyze interview data for the purposes of the present study, we selected the codes that corresponded to our research questions, developed preliminary themes from the coded data, and then 2 of us (DHJ and YC) ran a secondary check of the transcripts to identify any additional data that mapped to the themes.
This study was approved by the Institutional Review Board of the University of California, San Francisco (#19-28947).
Results
We conducted 5 focus groups and 18 individual interviews with a total of 39 providers (8 clinical providers and 31 social care providers) working in 6 organizations that provided or supported primary care services in San Diego County, including 3 FQHCs, 1 independent physician association (IPA) primary care practice, 1 health access organization (where volunteer physicians provide no-cost care), and 1 payor that provides case management services. Interviewees had either primarily clinical (eg, physician and registered nurse) or social care (eg, social worker, care coordinator) roles, in both frontline and leadership positions. See Table 1. In addition, interviewees’ frontline social care roles varied greatly, ranging from low-touch (ie, relatively lower intensity of intervention) navigation to some high-touch interventions with behavioral components.
Study Participants.
Benefits of Combined Clinical and Social Data
Overall, both clinical and social care providers believed that access to both social and clinical data could improve care. They reported 3 main benefits to having access to both social and clinical data.
Provide a more holistic view of individual
Both clinical and social care providers underscored that having access to both social and clinical data would help them form a more complete picture of their patients and gain a better understanding of their life outside the clinic, a perspective that was shared by this social care provider, who routinely asked their patients about both clinical and social factors:
We need to make sure we’re collectively always treating the whole person and asking about everything else [besides what they present with]. (Social care provider)
This holistic understanding of patients gave clinical providers more empathy for how patients’ socioeconomic challenges could affect their ability to follow treatment guidelines, such as this physician encountering a food insecure patient with diabetes:
So if I’m treating a patient with diabetes and they don’t have good access to food, and they come in with a sugar of 300 and I’m like, “What did you eat today?” And they’re like, “Well, I had a can of Coke this morning.” I’m just like, “Okay, you’re diabetic. You can’t drink a can of Coke for breakfast.” And they’re like, “Well, there was no food in the house. And I gave myself insulin last night and that was the only thing I had.” So like how can you be angry? You know what I mean? Like you can’t be upset at them for doing that if they don’t have access to food. (Clinical provider)
Social care providers also saw clinical information as helpful for informing how to approach a patient interaction, and some with access to the EHR (eg, social workers) mentioned reviewing the medical information before meeting the patient. One social care provider noted how being aware of the patient’s chronic conditions and medications could help them understand the patient’s self-management burdens and shed light on when to increase coaching and encouragement:
Keeping them motivated [and being] empathetic about [it when] they get tired of taking their meds, whatever amount of meds they take, they get tired of poking themselves. Keeping them motivated because they do tend to fall into this exhaustion, and that spirals to many other things. (Social care provider)
Tailoring care
Given the wide variety of services social care providers can deliver (eg, navigation to social services, behavioral services, or even primary care; benefit enrollment assistance; follow-up on obstacles to connecting to resources), nearly all the social care providers we spoke with saw the utility of the clinical data for tailoring the care they provided. Social care providers saw patient clinical data (eg, behavioral health diagnoses, disabilities, and other health conditions) as immediately useful to guide initial care, as the combination of medical data with social needs data could signal more urgent needs that should be addressed first. This was especially true for higher-touch social care providers who worked on programs designed for patients with both medical and social needs, for example, the California Health Homes program; their intake process included needs assessments for physical health, behavioral health, social risks, long-term services, and social support.
Social care providers used clinical data to tailor referrals for social services in 2 ways: by finding programs for which they may be eligible because the programs target patients with a specific diagnosis; and by making referrals based on the known constraints of a patient’s medical issues, the way this social care provider tailored a referral to a food resource based on their patient’s mobility and disease status:
If [the social worker has] seen someone who recently had surgery, and has difficulty walking, is on medication, [the care plan] wouldn’t recommend signing them up for our food distribution program where we have tons of people. It would let them know, “Hey, we can have this delivered to your home.” That would be a different referral versus someone else who would do automatic food distribution program, first. Based on their medical needs, or their age, or their specific situation, [e.g. they] have active cancer, and we don’t want them out there. (Social care provider)
A social care provider at a health plan also expressed that tailoring the care plan to medical needs could increase the plan’s potential to succeed.
All of those different things (social risk factors), those are just a handful of examples that medically I don’t know if they always think of, and so then the care plan are the things that they, the goals that they have for the member are unrealistic, it’s never going to be met. (Social care provider)
In contrast, most clinical providers did not proactively seek social data at the outset, and only perceived social data as useful when patients were not responding to medications or other clinical interventions, as this clinical provider noted regarding a housing insecure patient with asthma:
An example that I have is a patient whose asthma wouldn’t get better with all the treatment that we provide. When we reviewed the social determinants of health we came to realize that the patient was living in the car. The patient didn’t have a permanent residence, so once we were able to help them place that patient into a permanent residence, the asthma got much better. (Clinical provider)
In these cases where clinicians were trying to better understand why patients had little clinical improvement, they reported using social risk information to modify clinical care plans. For instance, they might prescribe an alternative to a refrigeration-required medication for individuals experiencing housing instability or schedule additional follow-up visits for patients with social risks to stay engaged (such visits could include clinical and social care).
Strengthen coordination with other care team members
Lastly, social care providers and a few clinical providers expressed how knowing about the clinical and social needs, respectively, of their patients facilitated communication and coordination with members of the care team to ensure patients were getting connected to the care or resources they needed, as in the quote in the previous section on finding housing for an asthmatic patient. This communication could take different forms; while some social care providers referred to engaging in informal communication with clinical providers to alert them to social needs (or even clinical issues that were brought up by the patient), 1 social care provider described wanting a centralized place where all providers could use the combined information to coordinate medical and social care:
If everything can be in one place, that way we can all be on the same page and really go back to it and look at, I don’t know, let’s say like a blurb or a checklist, that kind of information I think that would be helpful, just to get a better sense. (Social care provider)
A clinical provider gave a similar example of why it would be useful to have access to data about referrals and referral outcomes:
It would help me then in my next visit with the patients. If the last visit it was like, “Well, I’m not eating healthy because I don’t have food,” I can go back and say like, “Okay, well you met with a social worker and you got resources for SNAP or for these food banks,” or like, “How did that go?” And sometimes they’ll tell me like, “Oh yeah, now I’m getting food delivery, but it’s all like old bread and carbs. So like the stuff you’re telling me to eat isn’t the stuff I’m getting from the food bank.” And then we need to rethink like, “Okay, well what might be better food resources for you?” Yeah. (Clinical provider)
However, most clinical providers saw their role around care coordination as limited to making referrals or warm handoffs to social care providers on their team; they did not report revisiting patients’ social information, closing the loop with social care providers, or participating in other coordination activities as part of their role.
Examples of How Paired Social and Clinical Data Were Anticipated to Influence Care Decisions
In addition to the general themes, potential applications and perceived utility of access to clinical and social information noted above, several interviewees were able to pinpoint types of paired social and clinical data that would inform decision-making. In Table 2, we present examples of combinations of social and clinical data elements that were cited by both clinical and social care providers alongside their respective use cases. These uses cases illustrate specific applications of the themes explained above.
Examples of Combinations of Social and Clinical Data Elements and Their Potential Use Cases, as Reported by Interviewees.
Concerns About Integrated Data
Participants highlighted 2 categories of concerns about accessing both medical and social information. The first was where the data would be stored and later accessed. This concern was voiced by clinical providers, who were reluctant to access data through a different platform than the EHR, such as community information exchanges or community resource referral platforms. A clinical provider whose EHR already contained some social data noted that the social data were typically buried in free text notes and thought that to be useful the information needed to be easily visible in the patient record.
Two clinical provider interviewees additionally raised medical data privacy and HIPAA compliance concerns about clinic-based social care providers having access to specific clinical information and/or EHR systems. One clinical provider suggested a potential solution could be to use a second layer of log-in providing access to a limited view of the clinical portions: “Let’s say you have somebody that is not clinical, that doesn’t need to see the medical part, if you had a special login for that person, so that person could have access to the clinical part but kind of like the clinical part to be a little bit more protected than the social part only. I think that will solve a lot of our HIPAA problems.” No informants voiced data privacy concerns about clinical care providers having access to social information.
Discussion
In our study, primary care-based social and clinical care providers described multiple ways in which combining social and clinical data could positively impact patient care. Social care providers described using combined data to gain a holistic perspective on the patient that informed their interactions, tailor their social care interventions and social service connections, and inform conversation with clinical providers, but noted limited access to clinical data. Clinical care providers described awareness of the availability of social data and mainly reported using combined data to empathize with patients’ nonmedical challenges and to some extent, tailor clinical care and better coordinate with social care providers. Both clinical and social care providers named examples of specific combinations of social and clinical data, such as homelessness status and medications, that were or would be particularly beneficial for improving care.
This is the first paper of which we are aware to explore how primary care-based social care providers anticipate using combined data. The social care providers we interviewed were already collecting some patient self-reported clinical data on their own and were enthusiastic about the benefits of access to other data in the future to improve social care provision, care coordination, and patient relationships. For example, better quality referrals that are more tailored to patients’ unique social and medical situation could help patients resolve their needs more effectively, which could both bolster patients’ trust in the social care provider and bring potential quality of life or health benefits. In fact, social care providers in our study seemed to have more interest, receptivity, and thoughtfulness in using combined social and medical data than clinical providers.
There were, however, barriers for some social care roles to accessing EHR data as well as concerns from healthcare providers about granting social care providers access to such data. One potential solution to this tension is to facilitate interoperability between EHRs and the platforms used by social care providers (eg, community resource referral platforms) through such means as leveraging data standards (eg, ICD z codes), building integrations, or implementing single sign-on with native EHRs or other electronic platforms. Interoperability efforts could not only facilitate combined social and clinical data views for both social and clinical providers within an organization’s data privacy constraints, but also, for those who must work in multiple platforms, easing dual documentation and sign in fatigue could increase use.
Our findings among clinical providers suggest that increased availability of social data may not be sufficient to spur clinical providers to more routinely leverage combined social and clinical data in patient care. This is consistent with other literature that has documented the limited use of social risk information (25%-35% of clinical encounters27,28) and that social data are considered most important when patients were not doing well.20,29,30 Overlooking opportunities for clinical providers to apply the social risk data to clinical decision-making may lead to error: 1 study found error-free care in only 22% of patients with social risks and other “contextual complications.” 22 This raises the question of how to design both clinical training supports as well as electronic systems so that clinical providers are both able and encouraged to leverage the increased patient social data coming down the pike. As other work has consistently noted, 31 clinical providers in our study strongly preferred to access social data in the EHR. However, the mere availability of social data in the EHR is insufficient, as some clinical providers in our study did have data in EHR that they were aware of and not consulting, which is consistent with other studies. 17 The technical implication is that clinicians need data presented in the right way at the right time, for example, via pop-up alerts. 21 Our study suggests that there are specific combinations of social and clinical health data perceived as especially useful by both clinical and social care providers in primary care, which is a potential indicator of sweet spots for data visualization innovations.
Nevertheless, technical innovations will likely be of limited utility in the current context, where physicians and other clinical providers are neither trained nor paid for a social risk-first approach. 32 Despite greater than ever interest in value-based care, it still only accounts for 6.7% of medical revenue in primary care, 33 so the incentives are currently not aligned. Yet even in the face of these barriers, it is encouraging to note that primary care providers in 1 study reported that looking up social risk information in the EHR was not time-consuming 27 and in another study of hospital data, provider engagement with (ie, documenting and/or reviewing) social determinants of health data was associated with lower probability of hospital readmission. 17
The perceived value and potential use cases for combined data expressed by both social and clinical providers are also encouraging in the context of recent federal and state initiatives that impose regulatory requirements on assessing social risks and/or reimbursing social care activities (eg, California’s CalAIM Medicaid reforms, social risk screening quality measures).34 -37 These incentives will increase the social care activities performed in primary care settings; by extension, due to their documentation requirements, they will make much more social data available. Translating the greater availability of social data into greater use will require increasing investment into documenting and disseminating best practices on accessing and leveraging cross-sector data.
Limitations
Findings from this work should be considered in light of 4 key limitations.
First, interviews were conducted during the first wave of COVID-19 pandemic, a time during which there were massive shifts in the delivery of both clinical and social care, so participants may have responded differently than prior to the pandemic or in later, more routinized phases of the pandemic.
Second, our sample focused on CIE partner organizations who, by dint of their participation in the CIE, were invested in social care delivery; therefore, while not generalizable to all social and clinical care providers, it’s unlikely that our findings underestimate, and may even oversample clinical providers with a willingness to consider social information. This limitation has implications on the utility of these findings as a baseline against which to compare changes in perspectives following interventions increasing access to combined clinical health and social data.
Third, fewer clinical care providers were successfully recruited than social care providers, which potentially means that we captured less variability or nuance in their perspectives than in those of social care providers. However, we had to take a pragmatic approach that was sensitive to burden on clinics during the pandemic.
Fourth, not every participant was asked about their concerns, so this theme may be underdeveloped.
Conclusion
There are multiple feasible ways for primary care-based providers of clinical or social care to use combined social and clinical data to improve the care they provide and their relationships with patients in the short term, while awaiting policy incentives and other structural changes that enable deeper engagement with combined data from multiple sources. For example, even without access to EHR-derived clinical data, social care providers could include more questions about patients’ clinical concerns in their intake processes so that they can better tailor services. Clinical providers with social data available to them also have opportunities to incorporate some use cases for combined data into even brief appointments.
Future research should explore if those working in social service organizations also perceive benefit in having access to clinical health data and for which purposes, and investigate care recipient perspectives around the ethical implications for data sharing of patient clinical and social information alongside perceived impacts (both negative and positive). These additional perspectives can aid in informing the content and design of data sharing and visualization innovations to increase the value of cross-sector care coordination systems like the San Diego Community Information Exchange.
Footnotes
Acknowledgements
We are grateful to our interviewees for generously lending their time and insights to this study. We also wish to acknowledge Alicia DiGiammarino for her assistance in coding.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This project was funded under grant number R18HS027394 from the Agency for Healthcare Research and Quality (AHRQ), U.S. Department of Health and Human Services (HHS). The authors are solely responsible for this document’s contents, findings, and conclusions, which do not necessarily represent the views of AHRQ. Readers should not interpret any statement in this article as an official position of AHRQ or of HHS. None of the authors has any affiliation or financial involvement that conflicts with the material presented in this article.
