Transmission Dynamics of Large Coronavirus Disease Outbreak in Homeless Shelter, Chicago, Illinois, USA, 2020

Yi-Shin Chang; Stockton Mayer; Elizabeth S. Davis; Evelyn Figueroa; Paul Leo; Patricia W. Finn; David L. Perkins

Emerging Infectious Diseases. 2022;28(1):76-84.

Abstract and Introduction

Abstract

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has the potential for rapid transmission in congregate settings. We describe the multidisciplinary response to an outbreak of coronavirus disease (COVID-19) in a large homeless shelter in Chicago, Illinois, USA. The response to the outbreak included 4 rounds of mass PCR testing of all staff and residents and subsequent isolation of persons who tested positive for SARS-CoV-2. We further describe the dynamics of the shelter outbreak by fitting a modified susceptible-exposed-infectious-recovered compartmental model incorporating the widespread SARS-CoV-2 testing and isolation measures implemented in this shelter. Our model demonstrates that rapid transmission of COVID-19 in the shelter occurred before the outbreak was detected; rates of transmission declined after widespread testing and isolation measures were put in place. Overall, we demonstrate the feasibility of mass PCR testing and isolation in congregate settings and suggest the necessity of prompt response to suspected COVID-19 outbreaks in homeless shelters.

Introduction

The coronavirus disease (COVID-19) pandemic, caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has disproportionately affected persons living in congregate settings, including homeless shelters.^[1,2] People experiencing homelessness are at increased risk for SARS-CoV-2 infection because of shared living spaces and difficulty maintaining physical distance and are at increased risk for severe COVID-19 because of the high prevalence of underlying medical conditions.^[3,4]

Previous studies of COVID-19 in homeless shelters have reported testing results from 1 or 2 cross-sectional time points of an outbreak,^[1,2] but data are limited regarding the dynamics of SARS-CoV-2 transmission in homeless shelters. Community transmission was documented in Chicago, Illinois, USA, in early March,^[5] and a statewide stay-at-home order was implemented on March 14, 2020. During March–May 2020, many homeless shelters in Chicago experienced COVID-19 outbreaks.^[4] We describe an outbreak of COVID-19 in Chicago's largest homeless shelter, including the results of repeated rounds of SARS-CoV-2 reverse transcription PCR (RT-PCR) testing. On the basis of these data, we developed a compartmental mathematical model to characterize the extent and temporal dynamics of SARS-CoV-2 infection within this shelter.

1 2 3 4

Next Section

Abstract and Introduction
Methods
Results
Discussion
References

Table 1. Model parameters for fitting in study of transmission dynamics of coronavirus disease outbreak in homeless shelter, Chicago, Illinois, USA, 2020*
Fitted variables	Range of values fitted	Description of variable	Referenced ranges for fitting	Directly dependent model parameters	Dependent model phases	Dependent model compartments
β₀	0–445	Initial β	NA	β, R₀	1, 2, 3, 4	S, E
β_f_pct_β₀	0–1	Final β as percentage of β₀	NA	β	1, 2, 3, 4	S, E
k	0.01–2	Rate of transformation of β	NA	β	1, 2, 3, 4	S, E
t_Trans	1–50	Day where β reaches halfway between β₀ and β_f	NA	β	1, 2, 3, 4	S, E
Incubation period	2.8–4.0	Time between E and I compartments	(8)	σ	1, 2, 3, 4	E, Ia, Is
Asymptomatic percentage	0.18–0.87	Asymptomatic percentage	(9,10)	σ_s, σ_a, R₀	1, 2, 3, 4	E, Ia, Is
Infectious period for symptomatic persons, d	3–8	Infectious duration for symptomatic persons	(11,12)	ϒ_sp, R₀	1, 2, 3, 4	Is, R+_s
Infectious period for asymptomatic persons, d	3–8	Infectious duration for asymptomatic persons	(11,12)	ϒ_ap, R₀	1, 2, 3, 4	Is, R+_a
Period of RT-PCR–positivity for symptomatic persons, d	16–35	Duration of RT-PCR–positivity of symptomatic persons	(13,14)	ϒ_sn	1, 2, 3, 4	R+_s, R–_s
Period of RT-PCR–positivity for asymptomatic persons, d	3–35	Duration of RT-PCR–positivity for asymptomatic persons	(13,14)	ϒ_an	1, 2, 3, 4	R+_a, R–_a
α	0.01–1	Rate of detection of symptomatic infectious persons through screening	NA	α	2, 3, 4	Is, Isol_soft, Hotel
λ₀_pct_β	0–1	Rate of transmission between persons in Isol_soft and S compartment, as a percentage of β	NA	λ₀	2, 3	S, E
λ_isol_pct_β	0–0.5	Rate of transmission between persons in isolation units and S compartment, as a percentage of β	NA	λ_isol	4	S, E
Isolation duration, d	14	Rate of return from isolation units to R compartment = 1/[14 d]†	NA	ρ	2, 3, 4	Isol, R
RT-PCR sensitivity	0.72–0.90	RT-PCR sensitivity	N.S. Padhye, unpub. data‡	–	1 and 2,§ 4	-
ω₀	0.05–1.0	Rate of hospital admission of Infectious symptomatic persons before screening	NA	ω₀	1	Is, Hosp
ω	0.05–1.0	Rate of hospital admission of Isol_soft symptomatic persons during phase 2	NA	ω	2	Isol_soft, Hosp

*Details of optimization and calculation can be found in the Appendix (https://wwwnc.cdc.gov/EID/article/28/1/21-0780-App1.pdf). E, exposed; Ia, infectious asymptomatic; Is, infectious symptomatic; NA, not applicable; R+, recovered PCT-positive; R–, recovered PCR-negative; RT-PCR, reverse transcription PCR; R0, basic reproduction number; S, susceptible.
†Value not fitted.
‡https://www.medrxiv.org/content/10.1101/2020.04.24.20078949v2.
§Fitting based on hospital-based test results.

Table 2. Self-reported characteristics of Pacific Garden Mission staff and residents, Chicago, Illinois, USA, 2020
Characteristic	No. (%)
All	429 (100)
Role
Resident	362 (83)
Staff member	67 (17)
Age group, y
20–29	22 (5)
30–39	40 (10)
40–49	81 (20)
50–59	144 (35)
60–69	109 (26)
≥70	18 (4)
Sex
M	301 (70)
F	131 (30)
Other	1 (0)
Race and ethnicity
Non-Hispanic Black	266 (62)
Non-Hispanic White	92 (21)
Hispanic	48 (13)
Non-Hispanic Other	22 (5)
Smoking status
Current smoker	133 (33)
Former smoker	112 (28)
Nonsmoker	156 (39)
Medical history
Cardiovascular disease	85 (22)
Chronic lung disease	53 (13)
Diabetes mellitus	54 (14)
Neurologic disease	20 (5)
Chronic kidney disease	13 (3)
Immunocompromised	10 (3)
Chronic liver disease	7 (2)

Table 3. Fitted model parameter values in study of transmission dynamics of coronavirus disease outbreak in homeless shelter, Chicago, Illinois, USA, 2020*
Parameter	Fitted value	Description of parameter
β₀	0.60	Initial β
β_f_pct_β₀	0.11	Final β as percentage of β₀
k	2.0	Rate of transformation of β
t_Trans	23	Day where β reaches halfway between β₀ and β_f
σ_s	0.098	Rate of transition from E to Is compartment = 1/(incubation period) × (% symptomatic)
σ_a	0.26	Rate of transition from E to Ia compartment = 1/(incubation period) × (% asymptomatic)
Asymptomatic percentage	0.73	Asymptomatic percentage
ϒ_sp	0.15	Rate of transition from Is to Rs+ compartment = 1/(infectious period for symptomatic persons)
ϒ_ap	0.13	Rate of transition from Ia to Ra+ compartment = 1/(infectious period for asymptomatic persons)
ϒ_sn	0.046	Rate of transition from Rs+ to R- compartment = 1/[(duration of RT-PCR–positivity) – (infectious period)] for symptomatic persons
ϒ_an	0.12	Rate of transition from Ra+ to R- compartment = 1/[(duration of RT-PCR–positivity) – (infectious period]) for asymptomatic persons
α	0.32	Rate of detection of I symptomatic persons through screening
λ₀_pct_β	1	Rate of transmission between persons in Isol_soft and Susceptible compartment, as a percentage of β
λ_isol_pct_β	1	Rate of transmission between persons in isolation units and Susceptible compartment, as a percentage of β
ρ	1/14 d*	Rate of return from isolation units to Recovered compartment = 1/[14 d]†
PCR sensitivity	0.90	RT-PCR sensitivity
ω₀	0.75	Rate of hospital admission of Infectious symptomatic persons before screening
ω	0.39	Rate of hospital admission of Isol_soft symptomatic persons during phase 2

*Details of optimization and calculation can be found in the Appendix (https://wwwnc.cdc.gov/EID/article/28/1/21-0780-App1.pdf). E, exposed; I, infectious; RT-PCR, reverse transcription PCR.
†Value not fitted.