Estimated Emergency and Observational, Quarantine Capacity Need for the US Homeless Population

Estimated Emergency and Observational/Quarantine Capacity Need for the US Homeless Population Related to COVID-19 Exposure by County; Projected Hospitalizations, Intensive Care Units and Mortality 1 Estimated Emergency and Observational/Quarantine Capacity Need for the U.S. Homeless Population Related to COVID-19 Exposure by County; Projected Hospitalizations, Intensive Care Units, and Mortality The rapid progression of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic has raised concerns about the potential impact of coronavirus disease (COVID-19) on the homeless population. According to official reports, 575,000 people in the United States meet the U.S. Department of Housing and Urban Development (HUD) definition of homelessness – living in a homeless shelter or a place not fit for human habitation.1 On any given night New York City and Los Angeles County, two areas heavily impacted by COVID-19, are estimated to have 70,000 and 58,000 homeless individuals, respectively. In recent days, both the State of California and the City of Los Angeles have made substantial funding allocations to support rapid emergency shelter for unsheltered individuals, outreach for early detection, and quarantine space. Policing reforms have also been put in place to ensure that homeless individuals are able to shelter in place should the need arise. The federal government is in the process of considering additional funding allocations aimed at protecting homeless populations from COVID-19. This report aims to establish the potential mortality and hospitalization costs of inaction along with estimating the funding needs associated with a comprehensive plan of action. Potential impacts of COVID-19 on the homeless population We estimate the potential impact of COVID-19 on the homeless population and the homeless and healthcare systems caring for them. We model our estimates on a variety of severity and fatality scenarios informed by the unique health burdens facing the homeless population, applied to the age distribution of the homeless population. Concern has been raised around the potential for widespread transmission of COVID-19 within the homeless population due to inadequate access to hygiene and sanitation and the difficulty of early detection among a population isolated from health care. Yet, given limited understanding around the exact parameters of the virus’ transmission, it is difficult to explicitly model the potential transmission, and so we simply model a range of infection rate scenarios. Less widely known—but considerably more important—is the extraordinarily high susceptibility to symptomatic infection, hospitalization, and fatality among the homeless population due not only to their advanced age, but also the accelerated physical decline and mental weathering that frequently results from repeat exposure to harsh elements. For decades, the single adult homeless population has been dominated by members of the late baby boom cohort (e.g., those born between 1955-1965, approximately) , and the age distribution of homeless populations has shifted upwards as this birth cohort phenomenon has persisted into older ages.2,3 A recent study of aging trends among homeless populations in New York City, Los Angeles County, and Boston observed that the modal age of homeless clients in all cities was between 50 and 55 years. Studies of COVID-19 severity and case fatality suggest that older populations face risk orders of magnitude higher than those facing younger populations.4,5 Concerns over chronological age are magnified by concerns over accelerated physical decline among homeless populations.6 Homeless individuals are admitted to the hospital with medical-surgical conditions 10-15 years earlier than comparable, housed individuals,7 and with age-related impairments 2 typical of housed individuals 20 years older.8 Early studies of COVID-19 risk factors point to comorbidities relating to hypertension, diabetes, cardiac disease, chronic respiratory disease, and cancer.9 In Los Angeles and other communities on the West Coast, concerns about physical susceptibility are further heightened by high rates of unshelteredness, which are thought to carry both a generalized risk of accelerated age-related decline and specific exposures to poor hygiene and respiratory distress.10 Existing studies of homeless populations have observed obstructive pulmonary disease prevalence between 20 and 30%,8,11 compared to 10% for the general adult population. 12 This study serves two distinct purposes. First, we model potential scenarios of COVID-19 severity, hospitalization and fatality among homeless populations. Second, we estimate additional homeless system capacity – through private units as quickly feasible and scalable and otherwise through current infrastructure that incorporates tenets of social distance and safety - required to manage the acute impact of the COVID-19 pandemic on the existing homeless population. We perform this exercise for all unsheltered persons and homeless individuals in emergency shelter or transitional housing, which are generally congregate in nature (no private rooms or partitions) and highly dense. We have excluded families with children, who tend to be sheltered in private rooms or apartments. This exercise outlines the additional homeless system capacity required for a sufficient response to safely care for the existing homeless population; it is not intended to prescribe specific solutions, which should incorporate local conditions and guidance from the CDC and other public health experts. Modeling Health Impacts of COVID-19 on the Homeless Population Data and Methods Because detailed national age distribution data are not available, we drew on aggregate age distributions for the adult homeless population (age 25+) from the Homeless Management Information Systems (HMIS) of New York City (NYC) for 2017 and Los Angeles County (LAC) for 2018, two municipalities with robust social service data infrastructures for which we could readily access data. (Previous research by this team has found that the age structure of adult homelessness is fairly robust across states).2,3,13 The most granular geography at which official homelessness estimates are available is the Continuum of Care (CoC), which is the geographic unit at which federal homeless assistance grants are awarded and local homeless services are coordinated. These geographies do not map uniformly onto county boundaries, and we thus interpolate county estimates of these outcomes from CoC-level data mirroring a process described by Almquist and colleagues.14 We built estimates of the distribution of hospitalization, critical care and fatality for five-year age groups by modifying the estimates included in the widely publicized Imperial College report published March 16, 2020.4 We used nonlinear regression techniques to approximate rates for five-year age groups instead of ten-year groups. To estimate potential impacts on the homeless population, we built on the intuition of existing studies that older homeless populations bear health risks comparable to those of individuals 15 to 20 years older.7,8 To be conservative, we focused primarily on a 15-year accelerated aging shift while also reporting 10-year and 20-year shift models. A future scientific report will explore a wider range of assumptions. Figure 1 compares the risk curves for homeless populations to the general population using a model that assumes infected homeless individuals would be twice as likely to be hospitalized, two to four times as likely to require critical care, and two to three times as likely to die. 3 Figure 1: Age-specific risk for homelessness-adjusted scenarios in comparison to general population Panel A: Hospitalization 40% 30% 20% 10% 25 -2 9 30 -3 4 35 -3 9 40 -4 4 45 -4 9 50 -5 4 55 -5 9 60 -6 4 65 -6 9 70 -7 4 75 -7 9 80 -8 4 85 -8 9 0% General Homeless Panel B: Critical Care 30% 25% 20% 15% 10% 5% 25 -2 9 30 -3 4 35 -3 9 40 -4 4 45 -4 9 50 -5 4 55 -5 9 60 -6 4 65 -6 9 70 -7 4 75 -7 9 80 -8 4 85 -8 9 0% General Homeless Panel C: Fatality 15% 10% 5% 25 -2 9 30 -3 4 35 -3 9 40 -4 4 45 -4 9 50 -5 4 55 -5 9 60 -6 4 65 -6 9 70 -7 4 75 -7 9 80 -8 4 85 -8 9 0% General Homeless Figure 2 shows the age distribution of the homeless clients of LAC and NYC. The total number of clients age 25+ was 44,914 in NYC and 44,054 in LAC. The modal age group in both areas was 50-54, while the 4 mean age was 45.6 in NYC and 46.7 in LAC. The proportion over age 65 was 5.8% in NYC and 7.2% in LAC, lower than the national average among the general population. The proportion over age 50, and thus potentially heavily affected by accelerated aging was 39.7% in NYC and 42.5% in LAC. Figure 2: Age distribution of adult homeless clients, New York and Los Angeles 85-89 80-84 75-79 70-74 65-69 60-64 55-59 50-54 45-49 40-44 35-39 30-34 25-29 -8,000 -6,000 -4,000 -2,000 Los Angeles 0 2,000 4,000 6,000 8,000 New York Source: Los Angeles HMIS, California Policy Lab Secure Data Enclave; New York HMIS, provided by NYC Center for Innovation through Data Intelligence Table 1 shows the projected number of hospitalizations and critical care and fatality episodes anticipated among the homeless population of the United States. These projections are imposed on the estimated 493,000 single adults experiencing homelessness on a given night (unsheltered PIT counts adjusted *1.4, see later section on unit estimation method), which creates a conservative estimate of COVID-19’s impact on the homeless population given that the total number of people experiencing homelessness over the course of a year is likely 3-4 times that. Our estimates assume that 40% of the homeless population will be infected at any given time at the peak of the crisis, and that the U.S. homeless population would have the same age distribution as NYC and LAC grounded in work from a prior study of aging homelessness. Findings • • • We estimate that 21,295 people experiencing homelessness, or 4.3% of the U.S. homeless population, could require hospitalization at the peak infection rate of 40%, with a potential range from 2.4% to 10.3% hospitalizations. Critical care needs could range from 0.6% to 4.2%, with the midpoint scenario seeing 7,145 in critical care nationally. Finally, we estimate a wide range--0.3% to 1.9%-- of potential fatality rates, with the central estimate of 0.7% implying 3,454 homeless deaths. We believe that the true likely fatality outcome would be on the higher end of this range given the challenge of actually getting 5 homeless clients to the hospital, especially when they are unsheltered, as well as the unusually high mortality risks that prevail among the homeless population. Table 1: Projected COVID-19 outcomes for U.S. homeless population at a given time assuming peak 40% infection rate Percent of total Number of cases population Range across scenarios Hospitalization 21,295 4.3% 2.4%-10.3% Critical Care 7,145 1.4% 0.6%-4.2% Fatality 3,454 0.7% 0.3%-1.9% Map 1 depicts the proportionate distribution of hospitalizations, ICU admissions, and mortality among homeless individuals across the United States as a result of the COVID-19 pandemic. It is largely reflective of the distribution of the homeless population generally, with cases concentrated in urban areas and most regions seeing very few COVID-19 cases and low mortality. Map 1 6 Estimating Additional Capacity Data and Methods We use data from the U.S. Department of Housing and Urban Development’s 2019 Annual Homelessness Assessment Report (AHAR) to create assessments of additional capacity required to manage the acute effects of the COVID-19 crisis on the homeless population. The following assumptions were used to develop bed estimates: • • • • • Adult shelters operate at high density. To reduce density sufficient to enable social distancing by allotting at least 100 square feet per bed, we assume a 50% bed reduction in existing facilities, which would require an offsetting increase in beds or private units. The 2019 AHAR unsheltered count reflects persons observed during the PIT count, which is an estimated 40% undercount due to unobserved persons, based on Glynn and colleagues15. New units are estimated to cost $25,000 per year, or $68.50 per night based on current national shelter expenditures16, although accommodations may be made in other facilities. We estimate a peak infection rate of 40% and keep that constant for our estimates and maintain that rate – which we know will vary over time -- for our analyses. Finally, for those who are infected or under observation we estimate an additional $7,500 per unit for more private accommodations, including rooms in hotels and motels. Findings Our estimates suggest the need for an additional 400,000 units to manage the COVID-19 pandemic for the current, estimated homeless population. This includes: • • • • • • Approximately 200,000 single adults were sheltered on a given night in January, 2019 (199,531; US HUD, 2019). To reduce density by 50% while maintaining current capacity would require the addition of 100,000 units. 211,293 adults and persons in families were enumerated as unsheltered in 2019. Assuming a 40% undercount, approximately 300,000 beds are needed to provide accommodations to all unsheltered persons. Of the projected 500,000 total beds needed (300,000 for unsheltered, 200,000 for sheltered), at a 40% infection rate at a given time, 200,000 of these beds should be suitable for observation of symptomatic persons or persons under quarantine. At a cost of approximately $25,000 per unit per year, we estimate the annual cost of meeting this need at $10 billion. Assuming a premium of $7,500 per unit per year for more private accommodations for persons who are infected or under observation, including placement in hotels and motels, this would require an additional $1.5 billion for those 200,000 beds. The total estimated cost to meet the additional need is approximately $11.5 billion annually. Map 2 shows the need for these additional capacity by U.S. county. This map reflects the geographical heterogeneity in unsheltered and single adult homelessness across the country, with need concentrated in a handful of areas and municipalities. Ten percent of all counties will need only one, single unit or bed to accommodate the additional need, and half will need fewer than 10 additional spaces. The counties in the top 1% of need will each require a minimum of 2,100 additional spaces, with Los Angeles County facing far and away the greatest need at 65,000 new units. 7 Map 2 Options for emergency accommodation and coordination of care Given the high transmission rate of COVID-19, concerns have been expressed about the potential risk of congregate shelters, or shelters without private sleeping and bathing facilities. Decisions about the appropriate mix of rapid housing options must be made by individual jurisdictions based on local population needs, capacity for emergency accommodations, including vacant hotels and motels, and local government resources, but it is helpful to consider the basic principles of such an effort: 1) Emergency accommodations with private sleeping and bath space should be the preferred option for all clients and would be especially beneficial for individuals with known risk factors for COVID-19 complications. 2) Individuals in urgent need of protection – including the elderly, those with severe risk factors for COVID-19 complications, and those already presenting with respiratory symptoms – must be accommodated with considerable haste 8 3) Individuals with known or suspected coronavirus infections must be accommodated in private sleeping and bathing quarters, and such spaces should be made available well in excess of the likely number of infected individuals This means that most jurisdictions will need to use all potential emergency accommodation options to protect homeless populations from disease risk. We review the specific options individually before suggesting some technologies that would assist in the broader effort. Managing Encampments and Unsheltered Homelessness The recently posted CDC guidance discourages the clearance of encampments that frequently occurs under normal conditions and can lead to temporary or permanent disruptions that could lead to more harm than benefit. It recommends basic prevention measures for encampments: 1) encampments members should be encouraged to sleep with 12x12 feet of space per individual tent 2) public restrooms with water taps, soap and sanitizer should be made available 24 hours per day, and 3) portable toilets and handwashing facilities should be made available for encampments with encampments with 10 or more people. The report does not recommend strategies for encampments where these prevention measures cannot be implemented. Nor does it make specific prevention recommendations for individuals living outside of large encampments. We note that in Los Angeles, the city with by far the largest unsheltered population, only 24% of unsheltered homeless individuals actually live in encampments of any size, with a much smaller share living in large encampments. A recent study also found that individuals with longer exposure to homelessness, and thus potential greater health risks, were actually more likely to live in a sidewalk or alley and less likely to live in an encampment.17 In other words, in many cities, the vast majority of unsheltered homeless individuals, and an even greater share of the most vulnerable, are alone, unprotected, difficult to reach, highly dependent on services that may be unsustainable through the crisis, and unlikely to receive sanitation access. Those living in encampments or even in individual unsheltered spaces who have the ability and the desire to shelter in place should be encouraged to do so, because this will reduce demand for other emergency accommodations. But concerns about the ability to detect disease, maintain sanitation and meet basic needs – particularly if outreach capacity were to diminish – means that the bar for sheltering in place would ideally be set high. Building on the CDC guidelines, the following conditions should ideally be met for any encampment sheltering in place • • • • Ability to maintain social distance among encampment residents and from others outside the encampment Ability to maintain adequate handwashing and sanitation for a long duration of delayed service Ability to assess and report suspected infections, meaning either regular access to outreach workers or to mobile phones with reliable service and electrical power Ability to maintain adequate food, water and medicine supplies through local stores or service providers Encampments or individuals failing to meet these criteria may face severe risk of infection and indirect consequences in the event of a more severe emergency. 9 Congregate shelters: Congregate shelters are the predominant form of emergency accommodation within homelessness assistance systems. Bed densities are often high, and can include bunk beds in many cases. Per the CDC guidelines18 and FEMA recommendations, 100-110 square feet of space should be allocated per bed, aligned in head-to-toe formation. For many facilities, this will require a reduction in bed capacity in a given location, and the establishment of additional offsetting bed capacity elsewhere to sustain current supply. Consideration should be given to the provision of adequate hygiene and toilet facilities, and frequent, routine cleaning. Private accommodations: The ideal scenario would involve private accommodations for all clients. Private accommodations would dramatically reduce the likely transmission of disease relative to congregate shelters. The problem lies with the supply and the speed at which options can be mobilized. In the initial stages of the pandemic, some municipalities have sought to use campers and a variety of temporary disaster shelters that could be placed in empty areas. This approach is both slow and costly, and also potentially depends on people moving to isolated areas with open space. More recently, a number of cities have begun to scale up the use of hotels. Hotels have the advantage of already existing. Under usual conditions, hotel rooms would be not available. But in the current crisis, the use of hotels for emergency accommodation offers the added benefit of generating revenue for empty hotels and the often low-wage workers who operate hotels. Cities such as San Francisco, San Diego and Philadelphia have already mobilized hotel space for isolate of confirmed or suspected cases. In San Francisco, a group of city Supervisors have mobilized a more ambitious plan to enroll hotels accounting for 8,500 rooms into a proposed program, with buy-in from both owners and worker’s unions. This approach may offer a more scalable alternative to constructing shelters. Finally, we note that a sizable share of unsheltered individuals live in vehicles, some already parked in safe parking sites and others scattered throughout cities. Additional safe parking facilities should be opened, and individuals living in vehicles should be encouraged to move to these locations for safety and security. Emergency coordination of care: Regardless of the exact mix of emergency accommodations, the COVID-19 response would be greatly enhanced by the mobilization of better information and communication technology, evidence, and disease transmission models. In particular, we recommend the rapid adoption of SMS/mobile alert systems that would allow a broad spectrum of emergency outreach activities with relatively simple technologies. Prior research shows that nearly 95% of persons experiencing homelessness own cell phones and the vast majority make use of text messaging,19 thus highlighting the feasibility of this approach in terms of reach. We also note some specific advantages of this approach, all of which would serve both to enhance the reach of existing methods while also reducing burdens on outreach worker and freeing up their time to support high-need cases: 1) An alert system could remind people of new emergency opportunities as they come onstream, and allow people to link to location-based maps of the nearest facilities. 2) Alert systems could update both the sheltered and unsheltered of any sudden changes in rules or procedure, for instance if lockdown restrictions are tightened, eased or reimplemented. 3) For those who remain unsheltered, surveys pushed out over an alert system would serve as a valuable real-time tool for early detection tool, for monitoring social distancing patterns, and for delivering hygiene, cleanups and medical care to those who request it. 4) For those moving into emergency accommodations, whether shelters or private rooms, the alert system could disseminate surveys to monitor social distancing, personal security and threat perception among those who have received shelter. 10 5) Over time, needs assessments could be adapted to assess whether unsheltered and sheltered clients have lost access to essential treatments and medicines, in order to avoid indirect consequences of the pandemic. Conclusion The COVID-19 pandemic is creating a severe and emergent health crisis for the homeless population across the United States, a crisis that our shelter and health systems are simply not adequately prepared to meet. The current virus, when scaffolded on top of the already present crisis of aged homelessness, as well as a myriad of other factors impacting this population, is likely to wreak havoc on this already highly vulnerable group. For the 500,000 single adults who experience homelessness on a given night, the current crisis is likely to cause upwards of 21,000 hospitalizations and 3,400 deaths. Given an annual shelter turnover rate of at least 3 – meaning that over the course of a year at least three times the PIT-estimate will experience homelessness and the confined and harsh conditions that come with it, the infection, critical care, and fatality rates presented here are almost certainly lower bound estimates. Compounding this, we model these rates and solutions for current (2019) levels of homelessness. While economists are only beginning to quantify the short-and long-term economic impacts of this pandemic, we are almost certain to see a recession resulting from the infection itself, resultant social distancing, and general market uncertainty. Predictions vary widely, but the current record unemployment claims will be followed by additional housing instability and homelessness that will further stretch an already taxed homelessness assistance system. There are limitations to this study that should be noted. We apply homogenous rates of infections, hospitalizations, critical cases, and mortality across the country, though they will certainly vary by region and county in ways that we were not able to model. Second, we model those rates based on a single Point-in-Time estimate of homelessness though, as noted above, households enter and exit homelessness over the course of a year. Third, this study is notably limited in its scope. It is intended to model potential infection rates and the capacity required to provide every homeless person safe accommodation. It is not meant to prescribe the specifics of those facilities, and policymakers and practitioners are advised to take their cues from research more specifically devoted to that cause, with a particular emphasis on the aforementioned CDC guidelines and the public health community more generally. There are obvious and immediate steps that we can take to mitigate this situation. By creating adequate and humane accommodations for people living unsheltered, leveraging existing private units like hotels while creating new ones as feasible, and reconfiguring existing facilities to accommodate social distancing and isolate symptomatic individuals, lives can be saved. Federal, state and local governments will need to collaborate around the funding, staffing and siting of facilities. But the urgency is clear, as is the moral imperative to act. 11