Establishment of Isolation and Noncongregate Hotels During COVID-19 and Symptom Evolution Among People Experiencing Homelessness—Atlanta, Georgia, 2020

Abstract

Context:

Local agencies across the United States have identified public health isolation sites for individuals with coronavirus disease 2019 (COVID-19) who are not able to isolate in residence.

Program:

We describe logistics of establishing and operating isolation and noncongregate hotels for COVID-19 mitigation and use the isolation hotel as an opportunity to understand COVID-19 symptom evolution among people experiencing homelessness (PEH).

Implementation:

Multiple agencies in Atlanta, Georgia, established an isolation hotel for PEH with COVID-19 and noncongregate hotel for PEH without COVID-19 but at risk of severe illness. PEH were referred to the isolation hotel through proactive, community-based testing and hospital-based testing. Daily symptoms were recorded prospectively. Disposition location was recorded for all clients.

Evaluation:

During April 10 to September 1, 2020, 181 isolation hotel clients (77 community referrals; 104 hospital referrals) were admitted a median 3 days after testing. Overall, 32% of community referrals and 7% of hospital referrals became symptomatic after testing positive; 83% of isolation hotel clients reported symptoms at some point; 93% completed isolation. Among 302 noncongregate hotel clients, median stay was 18 weeks; 61% were discharged to permanent housing or had a permanent housing discharge plan.

Discussion:

Overall, a high proportion of PEH completed isolation at the hotel, suggesting a high level of acceptability. Many PEH with COVID-19 diagnosed in the community developed symptoms after testing, indicating that proactive, community-based testing can facilitate early isolation. Noncongregate hotels can be a useful COVID-19 community mitigation strategy by bridging PEH at risk of severe illness to permanent housing.

The coronavirus disease 2019 (COVID-19) pandemic has raised important public health concerns and revealed stark inequities in the health system. People experiencing homelessness (PEH) may be at a higher risk for both acquiring and developing severe illness from COVID-19.^1–3 Limiting COVID-19 transmission requires social distancing, wearing masks, testing, quarantining people who are exposed, and isolating people with COVID-19.^4,5 These measures can be difficult for PEH, who have limited access to health care, are marginalized from public health outreach and messaging, and frequently shelter in congregate living facilities that can facilitate transmission of respiratory infections.^6–9

Fulton County, Georgia, which includes the city of Atlanta, experienced an initial increase in COVID-19 cases during March 2020, followed by a steep increase at the end of June, which peaked during the last week of July with a weekly case rate of 261 cases per 100 000.¹⁰ With concern for potential widespread transmission among PEH in Atlanta, homeless service agencies partnered with public health agencies and city government to establish an isolation hotel for PEH with COVID-19 and a separate hotel to temporarily shelter PEH without COVID-19 who were at an increased risk for severe COVID-19 illness (hereafter referred to as noncongregate hotel).¹¹ In this report, we describe how the 2 hotels were established in April 2020, the procedures that were put into place, and the symptom evolution in a prospective cohort of PEH with COVID-19 who were admitted to the isolation hotel.

Methods

Establishment of isolation hotel and noncongregate hotel

In Atlanta, the lead homeless Continuum of Care agency, Partners for HOME, worked with the Georgia Department of Public Health (GDPH), Fulton County Board of Health, city of Atlanta, and Atlanta’s Health Care for the Homeless clinic, Mercy Care, to launch 2 hotels to serve PEH. An isolation hotel was established for PEH who were laboratory confirmed with SARS-CoV-2, the virus that causes COVID-19. In addition, a noncongregate hotel was established to temporarily shelter PEH without COVID-19 who were at an increased risk for severe illness (ie, 65 years or older or with comorbid medical conditions). Individuals with young children were referred to a separate facility that could accommodate families and are not included in this analysis.

The Federal Emergency Management Agency provided funding for the hotels as eligible emergency protective measures in response to the COVID-19 pandemic under the Public Assistance Program.¹² Cost sharing was provided by the city of Atlanta through Coronavirus Relief Funds. GDPH, the city of Atlanta, and the Continuum of Care agency developed a Memorandum of Understanding to formalize roles and responsibilities. City leadership identified the hotels through a standard procurement process. GDPH contracted with the isolation hotel; the city contracted with the noncongregate hotel. GDPH was responsible for developing clinical guidelines for the isolation hotel, providing masks and personal protective equipment (PPE) for staff and clients, and screening and referring patients from various clinical sites. The Fulton County Board of Health provided case management staff. The Continuum of Care agency led the effort through operational planning and managing logistics and operations for both hotels, including coordinating basic services and client transportation, providing in-room meals through an external caterer, organizing hotel security, staffing resident assistants (RAs) and on-site medical staff, and training staff.

Isolation hotel operations

Proactive SARS-CoV-2 reverse-transcriptase polymerase chain reaction testing was conducted at homeless shelters and at events serving people living in unsheltered locations during April-August 2020.¹³ Testing was offered to all clients and staff affiliated with a shelter or encampment, regardless of symptoms. PEH with symptoms could also seek testing from hospitals. PEH who tested positive for SARS-CoV-2 at shelters, community events, or hospitals were referred to the isolation hotel through GDPH’s isolation referral hotline managed by the state operations center or directly to the Continuum of Care agency. GDPH sent isolation orders for referred clients to the isolation hotel operations manager. Hospital referrals included both inpatient and outpatient (eg, emergency department) discharges. Upon arrival to the isolation hotel, clients received an orientation and signed an agreement form outlining hotel rules and services, which included a symptom screening schedule, hygiene and sanitation guidelines, isolation order requirements, and communication expectations. Services included in-room meals, free access to telephone, television, and wireless Internet, and clean linens. Visitors were not allowed.

The isolation hotel was staffed 24 hours per day, in 12-hour shifts, by RAs contracted through a local homeless shelter. One RA was assigned to each floor, with a maximum of 22 clients per floor; RAs ensured client isolation and supervised outdoor breaks (eg, on a designated outdoor patio with locations marked for social distancing). RAs also contacted clients twice daily by telephone to identify any personal needs, inquire about their well-being, and assess self-reported temperature and symptoms. Medical staff (paramedics and emergency medical technicians) were on-site 20 hours per day and conducted a once-daily, in-person temperature check and symptom assessment with all clients. During in-person interactions with clients, RAs and medical staff wore full PPE, including N95 masks. On-site medical staff also evaluated and triaged clients’ medical needs. Nonemergency clinical, mental health, and prescription needs were addressed via telehealth visits in coordination with Atlanta’s Health Care for the Homeless clinic. Emergent conditions were addressed by activating the emergency medical services (EMS) system. Clients were discharged from the isolation hotel at least 14 days after date of symptom onset or test date (whichever was first) and resolution of fever (initially 72 hours without fever-reducing medication, later updated to 24 hours in accordance with Centers for Disease Control and Prevention [CDC] guidance) and symptom improvement, if symptomatic.

Noncongregate hotel operations

The Continuum of Care agency worked with street outreach teams to identify PEH at an increased risk for severe illness and refer them to the noncongregate hotel. Clients were tested for SARS-CoV-2 prior to admission or were given a SARS-CoV-2 test at the time of admission or shortly thereafter. All clients received an orientation and signed an agreement form outlining hotel rules and services. Noncongregate hotel clients could request special permission to come and go during certain hours of the day, but outside visitors were restricted. Noncongregate hotel clients and staff were required to follow CDC recommendations on social distancing, mask wearing, and hand hygiene for the general public.¹⁴ Atlanta’s Health Care for the Homeless agency offered universal, facility-wide testing events at least monthly to minimize the risk of SARS-CoV-2 introduction into the noncongregate hotel.

The procedures and 24 hours per day RA staffing model at the noncongregate hotel mirrored the isolation hotel with additional RAs due to the larger number of clients. RAs conducted twice-daily temperature and symptom screenings in person or by telephone. RAs and clients wore a mask, and RAs wore eye protection and disposable gloves during in-person symptom screenings. On-site medical staff reviewed symptom screenings, performed a physical and temperature assessment on symptomatic clients, and referred clients for testing. If any client tested positive, he or she was immediately transferred to the isolation hotel or medical facility, as needed. On-site medical staff also facilitated telehealth and in-person clinic visits, prescription refills, and addressed any other medical- or behavioral-related needs. Noncongregate hotel clients remained admitted until a permanent housing plan was identified, a significant behavioral incident occurred, or the client elected to self-discharge. Permanent housing was defined according to the Department of Housing and Urban Development as “community-based housing without a designated length of stay in which formerly homeless individuals and families live as independently as possible.”¹⁵

Data collection and analysis

Among clients admitted to the isolation hotel, we examined whether they arrived via a hospital or community referral. We documented presence, duration, and type of symptoms (individually and in combinations). Typical symptoms were defined as subjective fever, cough, and shortness of breath. We examined the numbers and duration of stay among clients admitted to the noncongregate hotel. Finally, disposition location of clients discharged from both hotels was documented. This activity was reviewed by CDC and was conducted consistent with applicable federal law and CDC policy.^*

Symptoms at the time of SARS-CoV-2 testing and test date were extracted from isolation orders. Specimen collection date was used for test date when available; otherwise, result date was used as a proxy (11% of cases). Demographic data, dates of isolation hotel admission and discharge, and symptoms recorded during admission were documented. Symptom status was categorized as symptomatic at testing, presymptomatic (developed symptoms after testing), or asymptomatic (symptom status negative at testing and for duration of stay at the hotel). Subjective fever was defined as self-reported fever or chills. We performed descriptive analysis using frequencies of demographic characteristics, symptoms, and disposition location.

Results

Isolation hotel

From April 10 through September 1, 2020, 181 people who tested positive for SARS-COV-2 were admitted to the isolation hotel. Seventy-seven clients (43%) were referred from the community, and 104 clients (57%) were referred from hospitals. Among community referrals, clients were referred primarily from shelters (n = 69) as well as health departments (n = 4), correctional facilities (n = 2), the noncongregate hotel (n = 1), and the city of Atlanta (n = 1). Table 1 displays clients’ demographic characteristics. The median time from testing to date of admission for community referrals was 3 days (interquartile range [IQR] = 2–6 days) and for hospital referrals was 3 days (IQR = 1–6). Length of stay ranged from 1 to 22 days, with a median of 12 days (IQR = 9–14 days).

TABLE 1.

Demographic Characteristics of COVID-19 Isolation Hotel Clients, Atlanta, Georgia—April 10-September 1, 2020

	Overall (N = 181)		Community Referrals (n = 77)		Hospital Referrals (n = 104)
	n	%	n	%	n	%
Age group (missing n = 0)
18–29 y	16	9	8	10	8	8
30–39 y	24	13	10	13	14	13
40–49 y	37	20	20	26	17	16
50–59 y	60	33	21	27	39	38
60–69 y	35	19	14	18	21	20
70–79 y	9	5	4	5	5	5
Gender (missing n = 0)
Male	150	83	61	79	89	86
Female	30	17	16	21	14	13
Transgender	1	1	0	0	1	1
Race and ethnicity (missing n = 0)
Black, non-Hispanic	141	78	64	83	77	74
White, non-Hispanic	31	17	11	14	20	19
Hispanic	6	3	2	3	4	4
Asian, non-Hispanic	1	1	0	0	1	1
Native Hawaiian or other Pacific Islander	1	1	0	0	1	1
Multiple	1	1	0	0	1	1
Veteran status (missing n = 2)
No	158	88	68	89	90	87
Yes	21	12	8	11	13	13
Disability (missing n = 18)
No	123	75	56	80	67	72
Yes	40	25	14	20	26	28

To adequately describe symptom evolution, subsequent analyses were restricted to the 158 clients who stayed at least 7 days. Symptom status at the time of SARS-CoV-2 testing was available for 106 of 158 (67%) clients (Table 2). Among 31 community referrals with known symptom status at testing, 12 (39%) were symptomatic and 10 (32%) were presymptomatic. Among 75 hospital referrals with known symptom status at testing, 69 (93%) were symptomatic and 5 (7%) were presymptomatic.

TABLE 2.

COVID-19 Symptom Evolution for People Experiencing Homelessness, Atlanta, Georgia—April 10-September 1, 2020

	Community Referrals		Hospital Referrals
	n	%	n	%
Total admitted for at least 7 d (n = 158)	74		84
Symptoms evaluated at testing
Known	31/74	42	75/84	89
Unknown	43/74	58	9/84	11
Among people with known symptom status at testing (n = 106)
Symptomatic	12/31	39	69/75	92
Presymptomatica	10/31	32	5/75	7
Asymptomaticb	9/31	29	1/75	1
Among people with unknown symptoms at testing (n = 52)c
Reported symptoms during hotel admission	27/43	63	8/9	89
Did not report symptoms during hotel admission	16/43	37	1/9	11
Total ever symptomatic (n = 158)	49/74	66	82/84	98

^aPresymptomatic: No symptoms at testing and symptoms reported during hotel admission.

^bAsymptomatic: No symptoms reported at testing or during hotel admission.

^cUnable to distinguish symptomatic, presymptomatic, and asymptomatic because symptoms were unknown at the time of testing.

Overall, 131 of 158 clients (83%) reported symptoms at any observed time point. A majority of community referrals (49/74; 66%) were ever symptomatic. Nearly all hospital referrals were ever symptomatic (82/84; 98%). Among clients whose symptoms were recorded for the first time while admitted, median symptom onset was on day 1 (IQR = 1–3 days; range, 1–13) (see the Appendix Figure).

Overall, 120 of 158 clients (76%) had symptoms while admitted to the isolation hotel (Table 3). Among these clients, the most common were cough (92; 77%), headache (76; 63%), and muscle aches (74; 62%). At least one typical symptom (subjective fever, cough, or shortness of breath) was reported by 107 (89%) clients, and 27 (23%) reported all 3 symptoms. A majority of clients with symptoms reported at least one gastrointestinal symptom (68; 57%), and most reported both typical symptom(s) and gastrointestinal symptom(s) (62; 52%). In total, there were 30 EMS transports (including possible repeat transports) to the emergency department for higher level of care. Although reason for transport was not systematically recorded, the isolation hotel operations manager reported that most transports were unrelated to COVID-19.

TABLE 3.

COVID-19 Symptoms Reported by People Experiencing Homelessness During Isolation Hotel Admission, Atlanta, Georgia—April 10-September 1, 2020

	n	%
Overall symptoms (N = 158)
Symptoms prior to admission only	11	7
No symptoms during admissiona	27	17
Symptoms during admissionb	120	76
Individual symptoms reported during admission (n=120)
Cough	92	77
Dry cough	76	63
Wet cough	55	46
Headache	76	63
Muscle aches	74	62
Shortness of breath	60	50
Subjective fever	54	45
Diarrhea	52	43
Sore throat	44	37
Abdominal pain	42	35
Vomiting	17	14
Symptom combinations (n = 120)
Typical symptoms (subjective fever, cough, shortness of breath)
At least one typical symptom reported	107	89
All 3 typical symptoms reported	27	23
No typical symptoms reported	13	11
GI symptoms (abdominal pain, diarrhea, vomiting)
At least one GI symptom reported	68	57
All 3 GI symptoms reported	10	8
No GI symptoms reported	52	43
Symptom combinations
Typical symptom(s) with GI symptom(s)	62	52
Typical symptom(s) without GI symptom(s)	45	38
GI symptom(s) without typical symptom(s)	6	5
No GI or typical symptom(s)	7	6

Abbreviation: GI, gastrointestinal.

^aTen had no symptoms at testing; 17 had unknown symptoms at testing.

^bFifty had symptoms during admission only, and 70 had symptoms both at testing and during admission.

Disposition locations for all 181 clients of the isolation hotel were recorded (Table 4). Emergency shelters included community homeless shelters, hotels, and motels as well as the noncongregate hotel. Twenty-eight (15%) clients were connected with inpatient substance use treatment services at the time of discharge. Seven clients (4%) self-discharged before meeting discharge criteria. Six people (3%) were transported and admitted to a hospital and did not return to the isolation hotel. Four clients (2%) refused housing placement and returned to living unsheltered.

TABLE 4.

Disposition Location for COVID-19 Isolation and Noncongregate Hotel Clients, Atlanta, Georgia—April 10-September 1, 2020

	n	%
Isolation hotel clients
Temporary housing
Emergency shelter	87	48
Noncongregate hotel	19
Other emergency sheltera	68
Transitional housing for homeless persons (including homeless youth)	25	14
Staying or living with family or friends	17	9
Medical care site
Substance use treatment or detox center	28	15
Hospital or other residential nonpsychiatric medical facilityb	6	3
Residential project or halfway house with no homeless criteria	5	3
Other
Self-discharged because of inability to comply with program	7	4
Unsheltered
Place not meant for habitationc	4	2
Unknown	2	1
Total	181
Noncongregate hotel clients
Permanent housing
Permanent housing	155	51
Still enrolled with permanent housing plan in place	30	10
Unsheltered
Place not meant for habitationc	75	25
Temporary housing
Staying or living with family or friends	17	6
Emergency sheltera	12	4
Transitional housing for homeless persons (including homeless youth)	3	1
Medical care site
Substance use treatment or detox center	3	1
Long-term care facility or nursing home	2	1
Hospital or other residential nonpsychiatric medical facilityb	2	1
Residential project or halfway house with no homeless criteria	1	0
Other
Jail, prison, juvenile detention facility	2	1
Total	302

^aIncludes noncongregate hotel, hotel or motel paid for with or without an emergency shelter voucher, or runaway and homeless youth-funded host home shelter.

^bAdmitted to hospital and did not return.

^cRefused housing placement. Examples include a vehicle, an abandoned building, bus/train/subway station/airport, or anywhere outside.

Noncongregate hotel

In total, 302 people were admitted to the noncongregate hotel during April 22 through September 1, 2020, and were followed through November 10, 2020. People were admitted for a median of 18 weeks (IQR = 8–23 weeks; range, <1–29 weeks). One noncongregate hotel client developed symptoms consistent with COVID-19, tested positive for SARS-CoV-2, and was transferred to the isolation hotel. No other clients from the noncongregate hotel developed symptoms or tested positive for SARS-CoV-2. At the time of analysis, 185 of 302 clients (61%) were discharged to permanent housing or remained admitted with a permanent housing plan in place (Table 4).

Discussion and Conclusion

The use of hotels for PEH with limited personal housing options has been an integral part of the public health response to COVID-19 in Atlanta, providing a safe space for isolation and social distancing. Profiles of the individuals admitted to the isolation hotel indicate that nearly all hospital referral clients had symptoms prior to admission. In comparison, only approximately two-fifths of community referral clients with known symptom status were symptomatic at testing; however, just over half of those who were asymptomatic at testing subsequently developed symptoms. The identification of a sizeable proportion of people prior to the onset of symptoms suggests that proactive screening and surveillance testing in communities facilitate rapid isolation, which allows for effective interruption of transmission.

Overall, 83% of isolation hotel clients were symptomatic at any time. Previous reports from Boston, Rhode Island, Seattle, and Atlanta documented symptoms among PEH with COVID-19 during universal testing events ranging from 12% to 28%.^3,13,16,17 The prevalence of reported symptoms at the time of testing depends on the time period over which symptoms are elicited, how exhaustive the list of symptoms is, indiviudal recollection and reporting, and when during the course of illness the testing is conducted. Our ability to monitor a cohort of patients prospectively might have allowed us to document a higher symptom prevalence than prior analyses. The higher prevalence was also influenced by including patients who sought diagnostic testing in response to symptoms. When testing is prompted by symptoms, as is often the case in the general population, symptom prevalence will be higher; when testing is conducted proactively, such as a universal testing event, reported symptoms are expected to be lower. Symptom screen-directed testing alone will not identify all COVID-19 cases and would likely lead to delayed case finding (given the proportion of presymptomatic cases). Reliance on this approach could perpetuate community transmission. Effective case finding will require periodic, universal testing events at shelters or other venues frequented by PEH in addition to offering testing for individuals with signs or symptoms consistent with COVID-19.^8,13

Cough, headache, and muscle aches were the most common symptoms reported, symptoms that are well documented in nonhospitalized patients.^18,19 Overall, most illness was well within the scope of on-site medical staff to manage. The prevalence of most individual symptoms, gastrointestinal symptoms, and symptom combinations (typical symptoms with and without gastrointestinal symptoms) was similar to reports from CDC case report forms describing COVID-19 cases among the US adult population during January-April 2020.¹⁹ The prevalence of typical symptoms was lower among PEH in our population than previously published for the US adult population; however, for US adult cases early during the pandemic, typical symptoms were often required for testing and were likely overrepresented compared with patients in our study.¹⁹ Half of clients with no or unknown symptoms at testing who reported symptoms during isolation hotel admission developed symptoms on the first day, indicating that the early admission period is an important period for vigilance.

Ultimately, testing is only valuable as a public health intervention insofar as it leads to isolation of cases, including in environments where individuals at risk of severe illness can be monitored and supported in the event that higher-level care is needed. In this report, a majority (93%) of clients stayed until the end of their infectious period (4% self-discharged; 3% were admitted to a hospital), suggesting that the hotel was an acceptable alternative to their current housing situation. Given the high prevalence of underlying behavioral and mental health conditions among PEH,^20–22 considerable efforts were made to meet client needs and minimize potential effects of physical and social isolation. The 24-hour availability of RAs to address behavioral health concerns and communicating clear expectations at the beginning of isolation likely contributed to the high retention rate. Such a model could be expanded as a community resource for other people who might not be experiencing homelessness but for whom isolation would otherwise be financially or logistically burdensome.

In addition to preventing disease and interrupting COVID-19 transmission, offering noncongregate housing supported a larger strategy of linking PEH to permanent housing. Rapid rehousing of noncongregate hotel clients was funded through a Department of Housing and Urban Development Emergency Solutions Grant special allocation awarded through the Coronavirus Aid, Relief, and Economic Security Act (CARES Act).^† Noncongregate hotel clients were transitioned to permanent, supportive housing using housing choice vouchers provided by the Atlanta Housing Authority and state. More long-term housing options were available to noncongregate hotel clients than isolation hotel clients due to the availability of case management services and longer lengths of stay; however, 19 individuals in the isolation hotel were discharged to the noncongregate hotel where permanent housing plans were initiated. Connecting clients with more permanent housing should be incorporated into COVID-19 temporary shelter planning; it is healthier for PEH and positions communities for more effective public health responses.

There are at least 4 limitations to this report. First, symptoms at the time of testing were not elicited for some community referral clients. It is possible that selection bias could have led to an over- or underestimation of symptom prevalence. Second, our prevalence estimate of “ever symptomatic” could be underestimated because symptoms that occurred after testing and before admission were not recorded. Third, because objective temperature measures were not recorded, we relied on subjective fever, which may have led to underdocumentation of true fever. Fourth, length of stay for individuals at the noncongregate hotel might be underestimated because 30 individuals remained admitted at the time of writing. Finally, we did not record the number of people who declined admission to the hotels or reasons for refusing housing placement from the noncongregate hotel, which limited our ability to identify opportunities for improvement.

The program described in this report was a successful public and private partnership, bringing together many participants with clear, complementary roles and areas of expertise. A coordinated and targeted cascade of testing, isolation, and protective housing for people at an increased risk of severe illness is needed for PEH, who experience limited options for self-isolation and are at an increased risk for infection. The provision of noncongregate housing to people, not only PEH, who do not have the resources to protect themselves or their communities and families can serve as a model for future public health responses. Programs such as this one, developed as part of an emergency response, can be a bridge to more stable, safe, and sustainable housing solution for PEH.

Implications for Policy & Practice.

• Proactive, community-based testing for SARS-CoV-2 among PEH can facilitate isolation by identifying individuals early in the course of illness.

• Isolation hotels for PEH with COVID-19 can be established with a high completion rate.

• Noncongregate hotels can be a useful COVID-19 community mitigation strategy by bridging PEH at risk of severe illness to permanent housing.