INTRODUCTION

As health systems shift toward addressing social determinants of health as part of routine care,1,2,3,4 they face implementation challenges to screening and referring patients to services,5, 6 including determining appropriate screenings, administration logistics, and systematic integration of data into the electronic health record (EHR).2, 7 The Veterans Health Administration (VHA) has a long history of addressing social determinants of health, including housing instability. Housing instability has been linked to a wide array of poor health outcomes and increased risk of mortality.8,9,10,11 In 2009, responding to concerns over high rates of housing instability among Veterans (approximately 1 in 3 homeless people were Veterans at that time), VHA launched an initiative to end homelessness among Veterans.12 Key features of this initiative included the creation of “medical homes” for chronically homeless individuals (i.e., Homeless Patient Aligned Care Teams [H-PACTs]) and increased funding toward housing assistance and placement activities through the Supportive Services for Veteran Families (SSVF) program and, in collaboration with the US Department of Housing and Urban Development (HUD), the HUD-VA Supportive Housing (HUD-VASH) program. To identify and connect eligible Veterans to these services, VHA implemented a universal screening program across all outpatient clinics in 2012 using the EHR-embedded Homelessness Screening Clinical Reminder (HSCR).13

The HSCR, administered at least annually during outpatient visits to Veterans not already receiving VA housing support services, prompts providers to ask a patient two questions:

  1. 1.

    In the past 2 months, have you been living in stable housing that you own, rent, or stay in as part of a household? (A negative response indicates housing instability.)

  2. 2.

    Are you worried or concerned that in the next 2 months you may not have stable housing that you own, rent, or stay in as part of a household? (A positive response indicates risk.)

Veterans who screen positive for housing instability may then be referred to services, though this process varies by site.

Little is known about how providers tasked with screening for and addressing housing instability and other social determinants of health view the process and their involvement in it. The few studies examining providers’ views toward addressing housing instability have been conducted primarily in emergency department settings and highlight physicians’ feelings of frustration and futility in meeting the needs of these patients.14 Additionally, providers have reported difficulty identifying who is homeless; many do not ask about housing status routinely and express surprise that a patient who did not have the “typical appearance” of a homeless individual was indeed homeless.15

Implementation of the HSCR presents an opportunity to ask providers about: (1) their experiences asking about housing, both before and after implementation; (2) their role in addressing housing status in clinical settings; and, (3) their views on how patient housing status impacts clinical decision-making. These questions were explored through qualitative interviews with VA clinical providers (i.e., physicians and non-physician practitioners) as part of a larger study examining HSCR implementation.

METHODS

The study was conducted at the Corporal Michael J. Crescenz VA Medical Center (CMCVAMC) in Philadelphia, PA, or an associated community-based outpatient clinic (CBOC), and was approved by the CMCVAMC Institutional Review Board.

Sample and Interview Procedures

All providers (physicians, advanced practice nurses, and physician assistants) who had administered the HSCR and documented five or more positive screens between 2013 and 2015 were eligible to participate. Research staff contacted eligible providers by email and followed up by telephone to invite them to participate; interviews were conducted at a time and location most convenient to providers. Providers were not reimbursed for study participation. Recruitment was purposive, with an effort to include providers across different outpatient settings including primary care, women’s health, geriatrics, and outpatient mental health.

Interviews, lasting approximately 1 hour, were conducted by an experienced qualitative researcher in a private setting between March and September 2016. The semi-structured interview guide included questions about provider interpretations of HSCR questions and experiences with administering the HSCR and addressing affirmative responses. The guide also elicited details about how providers assessed patients’ housing status (if at all) prior to implementation of universal screening, and how practices changed following HSCR implementation.

Data Analysis

Digital recordings of interviews were transcribed verbatim and de-identified. The research team met regularly to review and discuss interview transcripts as they became available to determine when thematic saturation had been reached, after which no new interviews were scheduled.16

Transcripts were coded in Atlas.ti qualitative data analysis software using a modified grounded theory approach.17 The authors developed a codebook through open coding of several transcripts and met regularly to discuss the coding process, resolve coding discrepancies, and identify emerging patterns in the data. The coded transcripts were then further organized and analyzed using a template analysis approach.18 Domains relevant to the interview guide and study objectives (e.g., experiences administering the HSCR; how housing status impacts clinical decision-making) were organized into a template; members of the research team then used the template to summarize and compare responses across transcripts, identifying common categories of responses, outlier responses, and exemplar quotes.

RESULTS

In our sample, 104 providers were eligible for participation, and 22 providers were interviewed, including 20 physicians and two nurses (one nurse practitioner and one advanced practice psychiatric nurse). Twelve providers declined to participate. Providers practiced in a variety of outpatient clinics, including primary care, women’s health, H-PACT, geriatrics, and psychiatry. In response to interview questions, providers reflected on their experiences with asking patients about housing both before and after HSCR implementation. They also discussed how they viewed their role in addressing patient housing status, and how patient housing status impacted their clinical decision-making.

Few Providers Routinely Assessed Housing Prior to HSCR Implementation

Prior to the HSCR, most providers did not routinely assess housing status as part of the patients’ social history. Instead, inquiries about housing were prompted by other contextual disclosures, and were often limited by providers’ assumptions about who might be at risk. As one primary care provider observed:

Typically I [didn’t] specifically ask about housing a lot of times. Someone might say, “I couldn’t afford my medication, so I wanted to get it here at the VA.” “I lost my job. I’m now living with my sister,” or something like that. So usually it’s related to something that has to do with financing their medical care, and then housing will come up as a piece of that. (participant M)

In other examples, a patient experienced a problem associated with housing issues (e.g., bedbugs, food insecurity, or uncontrolled symptoms such as high A1c’s), which then cued providers to ask about housing. Providers also described awareness of, and attunement to, potential indicators of housing instability. These included changes in a patient’s appearance, reports of relational or financial instability, or a history of mental illness or substance use disorder. However, they also reported that prior to the HSCR, their preconceptions about what a person who was homeless “looked like” resulted in missed opportunities. One PCP recounted his surprise:

He was just not who I would have typically thought about as homeless... I found out that he was homeless through trying to contact him about an abnormal result. (participant N)

Providers cited lack of training, during medical school and residency, as one reason for not asking routinely about housing prior to the HSCR. Additionally, prior to working at the VA, most providers typically had not worked in settings that sought to systematically address housing instability. Providers would occasionally ask about household composition (“Who do you live with?”) as part of the social history, but rarely about the stability of the housing context. A women’s health clinic provider described her experience prior to HSCR implementation:

We’re not really trained and we don’t have [asking about housing] incorporated. Even though we’re aware of many social determinants of health, homelessness hasn’t really been part of [the] paradigm we’ve used. (participant P)

This differed somewhat for providers who worked in specialty clinics such as psychiatry or geriatrics. While individuals in these specialties did not routinely screen for housing instability prior to the HSCR, they discovered patients’ housing instability through standard intake questions about home environment and self-management. As one mental health provider practicing in a geriatrics clinic described:

Sometimes [I would] use specific examples…Who pays the utility bills? Who pays the rent? Or, is there a mortgage? Or, tell me what your expenses are. How do you manage that? We may ask very explicitly about things related to maintaining stable housing. (participant A)

HSCR Prompted Consideration of Patient Housing Status and Broadened Conceptions of Risk

Providers noted the HSCR broadened their understanding of housing instability beyond literal homelessness and alerted them to less obvious or visible cases of housing instability: patients who were “well kempt, groomed, showing up, taking their meds” (participant L) and whose housing issues were not apparent. For example, one provider spoke about how the HSCR changed his practice of asking about housing, resulting in identification of Veterans experiencing homelessness who might otherwise have been missed:

I think it [the HSCR] changed [my practice]. [The HSCR] is making [housing] the 8th vital sign or the 10th vital sign, whatever we are up to now. It is something that you should always be looking for in patients. I think [the HSCR] has served that role very well. (participant E)

In general, providers reported the presence of the HSCR prompted incorporation of housing status into routine assessment. One primary care provider noted:

I find the reminder actually very helpful, even if I do not necessarily sit there and click through the reminder asking the prompts. It has prompted me to incorporate it into what I normally do. I know I am going to have to click on it, so it has trained me to ask about it in my own way. (participant B)

Asking About Housing Is Important, but May Best Be Done by Someone Other than the Provider

Discussing their role in addressing housing instability, many providers distinguished between medical and social aspects of patients’ presenting problems, and felt their training best positioned them to manage the former. This focus on the purely medical aspects of care was all the more important in the context of brief clinical visits. As one primary care provider noted:

I still feel my primary obligation is helping them manage their illness…Once the disease, the illness, the symptoms, the signs, the side effects, and the medications are taken care of, then the next big thing is housing. (participant B)

Similarly, most providers expressed the view that others on the care team—social workers in particular—had the knowledge and expertise and were better positioned to address patients’ housing needs:

I’m not the one providing them with the services.…it’s not my job to get them housing…My job is to connect them to the people who can get them housing…I’m trying to decide what the right medication to give is, so I don’t see myself as a social worker. I don’t know the services. (participant N)

These views also shaped perspectives on who should be responsible for HSCR administration. While providers agreed the health system has an important role in addressing housing concerns, opinions about providers’ role in HSCR administration were mixed. One primary care provider questioned the need for screening to be done by a primary provider:

A nurse can do it, an LPN can do it, and a clerk can probably do it. It does not really require any medical training to do it…but my time probably could be spent doing other things that I am more trained to do. There is nothing special about my medical training that allows me to help them with their homelessness necessarily. I just refer them to our resources here for that. (participant R)

Even those who felt they should have a role in the screening process believed their role should end after administration and referral, given their expertise was primarily medical, their lack of familiarity with available resources, and their need to prioritize patient care needs during time-limited visits. As one primary care provider described, attending fully to a positive screen could disrupt other goals for a visit:

If somebody answers “yes” to that question, it’s like suicidality. If you’re really going to take this seriously as opposed to just checking off a box...that could hijack the whole visit. “Oh, so you’d be out on the street. Hmm. Let’s talk about that.” There goes everything else. (participant H)

Primary care providers also cited the burden of completing multiple screening reminders as a deterrent to addressing social determinants of health during clinical visits. The same provider continued:

If you went through the 20 clinical reminders and really were open to hearing and bringing up all these points—suicide, depression, smoking, homeless—I mean, any one of those could tip a person over. So, it’s unrealistic. And they should be really asked by different health professionals, not the physician. When you have time, face time with your doc, these are the things the doc really can make a clinical decision on. Not all this. It’s in the wrong hands. (participant H)

However, a few providers defined their role more expansively and saw consideration of social factors as critical to providing good medical care. This came up more frequently among specialty providers, including those in the women’s health clinic, H-PACT, and geriatrics. This view was typified by a women’s health provider:

My job is to not just take care of them medically, but also to figure out what social determinants will affect them being able to take care of their health, and housing is one of them. (participant U)

One reason noted for this difference was that social work support was more readily available in specialty clinics than in routine primary care. This made it easier for providers to incorporate screening into their existing role. As one provider in the women’s health clinic observed:

When you have someone, a social worker, who can work in your clinic with you, you realize how valuable it is to have that role and how necessary it is. (participant G)

Knowledge of Patients’ Housing Instability Shapes Clinical Decision-making

While providers’ views on who was best suited to screen for housing instability varied, participants clearly felt that knowing a patient’s housing status influenced care decisions. When asked how housing instability impacts patients’ health, most providers offered vivid descriptions, using terms such as “crisis” and comparing it to other medical emergencies encountered in a primary care setting, like elevated blood pressure or extreme hyperglycemia. As one provider said:

It is sort of “antennae go up” when you hear somebody does not have stable housing. I think of that as one of the most fundamental...If you do not have shelter, you cannot do much else...It is like if somebody comes in with low blood pressure...it is a crisis. (participant A)

Providers described how housing problems impacted patient health directly as well as indirectly, by limiting patients’ ability to engage in health-related behaviors required to manage their conditions. As a result, knowing about patients’ housing instability led some providers to take a more active role in supporting self-management, such as changing or simplifying medication regimens. A primary care provider related a typical example:

You have to think about medications that need to be kept cool. It’s like, “Do they have…?” Insulin is supposed to be in the fridge. It would be much preferable if the medication was once a day, minimizing the total number of medications. (participant N)

Some providers reported ordering additional tests for conditions such as HIV, hepatitis C, or tuberculosis, for which they believed patients experiencing homelessness and housing instability were at higher risk. They were also more likely to elevate patients to higher levels of care, fearing that patients may not be able to effectively manage their condition at home. Conversely, some providers said they were less likely to consider preventative screening for patients who were homeless:

That woman that I knew was living out of her car…Her mind about cancer screening and stuff like that is not anywhere…Your threshold to talk about things that are ten years in the future, I do not even bring it up. (participant U)

Providers described strategies to help patients experiencing housing instability manage their health conditions, including setting up systems for tracking and following up with patients in the EHR, scheduling more frequent follow-up appointments with those patients, and arranging primary care and specialty appointments on the same day to limit barriers to care:

I keep them on my radar a little closely, especially if they have uncontrolled chronic diseases, which they often do. I spend a little bit more time going over our follow-up plan. And sometimes I will try to say, “if you are having issues with your medicines because you are homeless, then maybe we need to just simplify it if we can.” I will try to meet their needs in terms of treating their disease while they are homeless until hopefully they get a home. So it does change my practice. (participant R)

While providers were more likely to be more aggressive with some interventions, they also reported being less likely to prescribe certain medications such as opioids or benzodiazepines due to concerns about abuse or diversion:

Some of the medicines that we give have a significant street value. So you are concerned that if a patient does not have a place to stay where he can safely store the medicine and he is sleeping on the street somewhere, you worry it is going to be diverted. Or you are worried that he is going to become desperate and sell it. (participant K)

When participants were asked about their experience caring for patients who were homeless, they highlighted both a sense of reward in caring for a vulnerable population and a sense of frustration and, at times, futility. Not knowing how best to manage or care for a patient who may not return for further evaluation added a level of uncertainty that was challenging for providers.

It is hard. It creates a sense of vulnerability in the people who take care of these patients; you are pretty sure that if you set up a CT scan for three months in the future because there was a suspicious thing that doesn’t need anything but follow-up, that that could not happen 50% of the time. So I think some people would just not do the initial study. I think that is the wrong way to go. If you are going to do it for anyone, you may as well do it for this person, too…It is harder work, I think. (participant E)

At times, such unpredictability conflicted directly with providers’ desire to provide equitable treatment to all patients in their care and confronted them with the reality that this was not always possible.

It gets me upset if I have to choose a different target of healthcare or provide a different level of healthcare because the person isn’t housed. That’s frustrating. You try to take into account the person’s actual living situation and provide the best care you can considering the situation. But I hate to lower my standards. (participant N)

DISCUSSION

Previous literature highlights challenges providers face in asking about housing status and addressing housing instability.14, 15 Many of the providers studied here said they previously felt ill-equipped to discuss housing instability with their patients due to lack of training or biases about what a homeless patient “looks like.” These findings support recommendations by Behforouz19 and others to teach students to obtain comprehensive social histories that include housing, particularly given the impact housing status has on health outcomes.20

Despite a lack of training and prior experience with asking patients about housing status, VA providers reported routinely incorporating the assessment of a patients’ housing stability during the social history following HSCR implementation. Overall, providers indicated that they believed in the value of knowing about a patient’s housing situation. These positive views may in part be related to the availability of resources within VA to address housing instability among Veteran patients. These resources may mitigate against barriers to screening reported in previous studies, such as insufficient time and information to help patients who disclose vulnerabilities during screening.21 However, providers’ views were also clearly influenced by understanding how a patients’ housing status impacts the clinical care plan. Thus, arguments against screening in the absence of immediately available resources22 may be reconsidered in light of the potential benefit to clinical care of understanding of the patient’s social context, even in the absence of readily available remedies. In fact, it may be more important to understand the social context and potential impact on patient health and care access when such remedies are not so readily available, as the social condition is then more likely to persist.

Although providers in our sample viewed information regarding patients’ housing instability as directly applicable to their clinical care, there was disagreement regarding who was best equipped to conduct the screening. In addition, while providers shared many commonalities in terms of their practices and views, our analysis revealed some differences between the experiences and perspectives of providers in specialty clinics (i.e., women’s health, H-PACT, geriatrics, and psychiatry) compared with those working in a general primary care setting. One reason for this difference may have been the difficulty primary care providers faced in adding one more screening reminder to an already busy visit. The US Preventive Services Task Force (USPSTF) recommends approximately 30 screenings to be conducted during primary care visits, to which adding additional screening questions is perceived as burdensome.23 This difference is also likely attributed to the resource distribution in terms of time available for an appointment in some of the specialty clinics, as well as greater availability of social work staff support. Thus, there is likely a role to have screenings for housing stability and other social determinants of health be performed by an interdisciplinary team, with findings relayed to the provider.

This study was conducted at a single VAMC; as a result, it is a small sample of providers in a single region. Additionally, this study only included providers who had experience with multiple positive screens, and as a result may not reflect the perspectives of providers who had not administered the reminder or received positive screens. Additionally, participants’ responses may have been influenced by social desirability, although this potential bias was minimized by using well-established techniques, including establishing rapport, ensuring confidentiality, asking open-ended questions, and encouraging participants to provide examples.

Despite its limitations, our study provides novel insight into how providers ask about housing status and how information about patients’ housing instability may shape plans of care. This study raises questions for future exploration, particularly regarding the implementation of different strategies for screening for housing instability in primary care—including who should conduct the screen and the best ways to respond to positive screens—as well as how to address housing status in health systems where additional resources may not be available. Our findings suggest that even in settings without housing-related resources, asking about housing status and risk of instability are important to clinical care, and may be integrated into the visit by a team member other than the clinical provider. As above, training clinicians to integrate this into their social history and providing additional clinical resources and support staff in the form of social workers are possible mechanisms to make the assessment of housing instability more routine. Although focused within the VA, with integrated service provision, findings on the impact of an embedded screening for identifying and understanding patient housing status may be applicable to other health systems looking to expand attention to social determinants of health such as housing instability.