Health Care for People Experiencing Homelessness—What Outcomes Matter?

The cohort study by Stewart et al¹ describes the Safe, Healthy, Empowered (SHE) Clinic, which is colocated with a Seattle, Washington, drop-in community center and serves women experiencing homelessness 4 hours per week on a walk-in basis. Stewart et al¹ compared nonemergent (as judged by 2 investigators) emergency department (ED) visits for women who accessed the clinic vs those who had not. While comparison between these groups is limited by selection bias and potential unmeasured confounders, Stewart et al¹ report that the 41 women who accessed the clinic 1 or more times had a reduction from 37 nonemergent ED visits in the 6 months before they used the clinic to 22 visits in the next 6 months. A similar reduction was not observed for 35 women who had not used the clinic.

More than 567 000 people were homeless on a single night in 2019 in the US, a number that was both an underestimate at the time and widely expected to increase in the wake of the COVID-19 pandemic. Since the onset of modern-day homelessness, health care practitioners have struggled to provide quality care to people who lack one of the most basic necessities for health, housing. The first demonstration programs of specialty clinics for people experiencing homelessness appeared in the mid-1980s, funded by the Robert Wood Johnson Foundation and the Pew Charitable Trust.² The wrap-around services model was subsequently formalized as the Health Care for the Homeless (HCH) Program in the McKinney-Vento Homeless Assistance Act of 1987.

Since that time, specialized health care services for people experiencing homelessness have proliferated. In 2019, 300 health centers received federal funding as HCH sites. Countless other clinics not formally funded under the HCH program—staffed by an assortment of paid employees or volunteers and offering a range of operating hours and services—help to fill various cracks in the health care landscape for people experiencing homelessness.

Ultimately, we hope to someday no longer need a specialized system of health care for people experiencing homelessness because the US has committed to making homelessness something that is rare and, when it occurs, brief. Absent that, the value of health clinics serving people experiencing homelessness should be measured by much more than their ability to reduce ED visits. As described by Stewart et al,¹ the SHE Clinic emulates many best practices not only for patients experiencing homelessness but for all patients, including meeting people where they are (in this case, both figuratively and literally) and providing on-demand, walk-in availability. Given these strengths, and with the knowledge that multiple studies with more robust designs have found ED visit reduction outcomes in similar populations elusive, we are interested in learning more about other outcomes, such as improvements in housing status, increases in physical and mental well-being, continuity of care, medication management, number of women using preexposure prophylaxis for HIV, and reductions in risk for overdose. These outcomes would not necessarily be reflected in reduced ED use, which is a poor proxy for health, given its multifactorial drivers.

Prior to the COVID-19 pandemic, ED visits had continued to outpace what would be expected based on population growth. While we are not sure why, of all the positive outcomes that could result from receiving care in a clinic like the SHE Clinic, Stewart et al¹ reported only on ED use, they are certainly not alone in this focus. We suspect the focus on reducing ED use relates to real or perceived pressure from health systems, insurers, and policy makers. Simultaneously, this focus is intertwined with a narrative asserting that increased insurance coverage and access to primary care leads to improved health and, ultimately, reductions in ED use. This simple and intuitive story is the one that sticks, despite the fact that there is not compelling evidence for the existence of such a clear-cut path.

People experiencing homelessness in particular have been a primary focus for efforts to reduce ED use that accelerated after publication of Malcolm Gladwell’s widely read 2006 New Yorker article, “Million-Dollar Murray,”³ about “a bear of a man” from Reno who was brought to the ED for alcohol intoxication several times weekly. Gladwell’s article³ acknowledged Murray was on the far right-hand tail of the distribution curve, one of the top sliver of individuals who account for an outsized proportion of health care costs. While overall, people experiencing homelessness have been found to have disproportionate use of the ED compared with people who are adequately housed—not surprisingly given their high burden of health conditions, among other factors—most are not truly frequent ED users.⁴ Nonetheless, the narrative around homelessness and ED use has taken hold. Some policy makers have even expressed disappointment that permanent supportive housing, a best practice intervention for ending chronic homelessness, does not pay for itself solely through reductions in ED and other health care use,⁵ despite the fact that housing is a human right and should be valued on those grounds alone.⁶

As researchers who have studied and published on frequent ED use, including among people experiencing homelessness, we acknowledge that we may have ourselves lent fuel to this fire. We would like to pull on the brakes by offering a few suggestions to researchers and policy makers aiming to examine ED use as an outcome in research, programs, or policy. First, research has shown that it is nearly impossible to accurately distinguish so-called nonemergent ED use from other ED use. Making such distinctions is particularly fraught when using discharge diagnoses, which, by definition, are assigned by clinicians with the benefit of hindsight only after a full work-up has been completed.⁷ Certainly such determinations should not be made based on investigator gestalt, even if it may seem like common sense. Second, it is important to consider the baseline rate of ED use among the group of interest. Expecting large reductions in ED use from an intervention for a group with relatively low rates of ED use to begin with is unrealistic, not to mention of questionable importance. Third, especially with many EDs having a renewed focus on addressing patients’ social needs, ED visits among people experiencing homelessness and other marginalized populations could present opportunities to provide low-barrier access to needed health and social service resources and referrals and should not be uniformly viewed as a negative.

We should think critically about why ED use is consistently selected as an outcome, as opposed to other metrics that may be more important, especially given the relatively small contribution ED use makes to overall health care spending. Additionally, it is now clear that ED crowding, another common rationale given for studying low-acuity ED use, is not driven by such visits but results instead from hospital-driven factors such as boarding of admitted patients in EDs owing to lack of inpatient beds.⁸

We have often observed studies examining the effects of interventions on ED use, but have rarely seen reports of programmatic changes when such interventions fail to demonstrate the desired effects, which raises questions regarding how critical reduced ED use really is to a program’s goals. Humanistic care addressing the health and social needs of people experiencing homelessness is essential, regardless of whether it results in a few less visits to an ED.

Published: March 25, 2021. doi:10.1001/jamanetworkopen.2021.3837

Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2021 Doran KM et al. JAMA Network Open.

Corresponding Author: Kelly M. Doran, MD, MHS, Department of Emergency Medicine, NYU Grossman School of Medicine, 227 E 30th St, New York, NY 10016 (kelly.doran@nyulangone.org).

Conflict of Interest Disclosures: Dr Raven reported receiving grants from Arnold Ventures, and San Francisco Whole Person Care, and serving on the advisory board for the UCSF Benioff Homelessness and Housing Initiative outside the submitted work. No other disclosures were reported.

Disclaimer: The views in this article reflect those of the authors alone and not of their employers.