SARS-CoV-2 Transmission Dynamics in a Sleep-Away Camp

All attendees of the camp from June 10, 2020, to July 1, 2020 were eligible for inclusion in the retrospective cohort study. We categorized persons who attended staff orientation from June 17 to June 20 as trainees if they only attended orientation and as staff members if they also worked during the camp session, which was held June 21 to June 27. Campers only attended the camp session. The camp provided attendee contact information, age, sex, type (trainee, staff, camper), and cabin. On the basis of contact information, we categorized attendees as residents of counties included in the Metro Atlanta area, counties in Georgia not part of the Metro Atlanta area, or counties out of state. From July 17 to August 25, we contacted camp attendees for a phone interview; those we did not successfully reach after 3 attempts over different times of day, including evenings, and days of the week (including weekends), were considered nonrespondents. We used a structured questionnaire to collect demographics, clinical characteristics, SARS-CoV-2 testing history, activities during camp, known exposures to COVID-19 and community exposures 14 days before and 14 days after attending camp, mask use during camp attendance, and dates of arrival and departure from camp. We also reviewed prearrival laboratory test results that were provided to the camp per Georgia executive order.¹²  We conducted a detailed interview with a senior staff member to assess mitigation measures adopted by the camp. For attendees who were Georgia residents, we obtained post-camp laboratory test results by manually matching the name and age, address, or phone number of attendees to known case patients in the Georgia Department of Public Health (DPH) State Electronic Notifiable Disease Surveillance System and collected symptom status and testing histories from state case investigations conducted from June to July. For out-of-state attendees, we contacted state health departments to obtain available information. In cases of discordant laboratory test results or symptom reports between interviews and state case investigations, a positive test result or the presence of symptoms from either source superseded a negative test result or the absence of symptoms.

We classified camp attendees as case patients with COVID-19, non–case patients, or patients having an unknown case status using the Council of State and Territorial Epidemiologists (CSTE) definitions approved on August 5, 2020.¹³  Case patients were defined as attendees who had a state- or self-reported positive viral test result or met the CSTE clinical criteria without test information. Non–case patients were defined as attendees who had a state- or self-reported negative viral test result or had not been tested and did not meet the CSTE clinical criteria. Case status was unknown for attendees whom we did not interview and who were not identified in state case investigations. We defined the date of the first positive specimen test result as the earliest specimen collection date, if available in the laboratory reports, or as the earliest specimen collection date reported during the interview. We categorized case patients as either community associated or camp associated. Case patients with symptom onset or a first positive specimen test result, whichever was earliest, 10 days before to 2 days after arrival at camp were community associated, and those with symptom onset or a first positive specimen test result 3 days after arrival to 14 days after leaving camp were camp associated.

For attendees who were Georgia residents, one commercial laboratory provided available residual specimens to the US Centers for Disease Control and Prevention (CDC) for whole genome sequencing (WGS). Twenty-two specimens with cycle threshold values <32 by real-time reverse transcription polymerase chain reaction were selected for sequencing extraction. The nucleic acid was extracted and subjected to Illumina MiSeq sequencing, following previously published protocols,¹⁴  and consensus sequences were generated with Minimap 2.17 and Samtools 1.9. We downloaded representative full-genome sequences on September 28, 2020, from GISAID and inferred phylogenetic relations using approximate maximum likelihood analyses implemented in TreeTime¹⁵  by using the Nextstrain pipeline.¹⁶ 

We tabulated demographic characteristics and exposures by case status and by attendee type. We calculated attack rates (ARs) using 2 methods: (1) the proportion of attendees with COVID-19 among all attendees and (2) the proportion of attendees with COVID-19 among attendees excluding those with an unknown case status. To estimate effective case and instantaneous reproduction numbers, we performed a probabilistic reconstruction of transmission chains based on a serial interval distribution of illness onset among case patients and time present at camp (Georgia Camp Outbreak: Probabilistic Reconstruction of Transmission Chains in the Supplemental Information). The effective case reproduction number is the average number of secondary case patients per infectious case patient under observed conditions.^17,18  The instantaneous reproduction number is the average number of secondary case patients whom each infectious case patient at time, t, would infect if the conditions remained as they were at time, t (reflecting mitigation measures in place).¹⁹ 

For attendees aged 6 to 21 years with nonmissing values for covariates of interest, we used unconditional generalized estimating equations to calculate unadjusted and adjusted risk ratios with 95% confidence intervals (CIs) for characteristics and exposures related to camp-associated case status.

We conducted statistical analyses in SAS version 9.4 (SAS Institute, Inc, Cary, NC) and R (version 4.0.2; R Foundation for Statistical Computing, Vienna, Austria).

This activity was reviewed by human subjects research advisors at the CDC and the Georgia DPH and was determined to not be human subjects research. For interviews with attendees younger than 18 years, we obtained parental or guardian permission and verbal assent from attendees.

From June 10, 2020, to July 1, 2020, 627 persons attended the camp, including 137 trainees, 127 staff, and 363 campers (Table 1). The trainee median age was 16 years (range = 14–20 years), and 61% (83 of 137) were female. The staff member median age was 17 years (range = 14–59 years), and 59% (75 of 127) were female. The camper median age was 12 years (range = 6–16 years), and 53% (193 of 363) were female. Most attendees were white (94%), non-Hispanic (96%), and Metro Atlanta area residents (77%). Attendees spent a median of 6 days (range = 2–21 days) at camp. As part of the mitigation measures implemented by the camp (Camp Description in the Supplemental Information), attendees were cohorted by cabin. During orientation, 137 trainees and 124 staff members stayed in 28 cabins, with a median occupancy of 11 (range = 1–23 occupants). During the camp session, 127 staff and 363 campers stayed in 31 cabins, with a median occupancy of 24 (range = 1–26 occupants); 98% of staff members stayed in the same cabin as during orientation.

Among 627 attendees, 598 (95%) provided negative prearrival laboratory test results to the camp and 29 (5%) attendees (8 [6%] trainees, 11 [9%] staff members, and 10 [3%] campers) did not have record of prearrival test results. A total of 476 (80%) attendees had an available specimen collection date, with a mean time from specimen collection to arrival at camp of 6 days (range = 0–13 days).

We identified 351 (56%) case patients among camp attendees, of whom 340 (97%) had a positive viral test result and the remaining 11 (3%) reported no testing but had symptoms consistent with COVID-19. Among 211 (34%) attendees categorized as non–case patients, 159 (75%) had a negative viral after-camp test result and 52 (25%) reported no testing and no symptoms consistent with COVID-19. Case status was unknown for 65 (10%) attendees who were neither interviewed nor found in state reports.

Of all 351 case patients, 288 (82%) had symptom information available; 243 (84%) reported having symptoms, and 45 (16%) reported no symptoms. Most (74%) symptomatic case patients reported developing symptoms by the last day of the camp session on June 27 (Fig 1). The most common symptoms included subjective or documented fever (56%), headache (52%), and fatigue (49%). Among case patients with available information, 6% (16 of 258) had an underlying medical condition, 5% (12 of 259) sought medical care because of COVID-19 illness, and none were hospitalized.

Among 351 case patients, 12 (3%) were categorized as community-associated case patients. Negative prearrival laboratory test results were available for 11 community-associated case patients; 1 was missing laboratory test results. Five case patients (2 asymptomatic and 3 with missing symptom information) had a positive test result on a specimen collected a median of 7 days (range = 6–8 days) before arriving at camp but retested with a negative result a median of 3 days (range = 0–5 days) after their positive test result (Supplemental Fig 4). Only negative results were supplied to the camp. Six case patients with symptoms had symptom onset from 6 days before to 2 days after arriving and had a positive test result on a specimen collected within 5 to 11 days of arriving at camp. One additional symptomatic community-associated case patient had a positive test result on a specimen collected within 2 days of arrival, but symptom onset was missing.

There were 339 camp-associated case patients; 328 (97%) had a positive viral test result, and 11 (3%) were not tested but met the CSTE clinical case definition. Among the 279 camp-associated case patients with available symptom information, 236 (85%) were symptomatic, 132 (56%) reported the symptom onset date during camp, and 104 (44%) reported the symptom onset date after leaving camp. The median number of days from camp arrival to symptom onset was 7 days (range = 3–21).

Among 338 Georgia case patients, 32 (9%) had available residual specimens. Full-genome sequencing was successful in 22 (7%) isolates; all were clustered within 0 to 2 single-nucleotide polymorphisms (SNPs) of another case isolate and were at least 6 SNPs from any other sequenced isolate available in the public database (Supplemental Fig 5). These findings indicate low viral genomic diversity, although case patients with available sequences were from 10 different cabins; 2 were community associated, and 20 were camp associated (Supplemental Table 3), with symptom onset dates between June 19 and June 30 (n = 17).

The overall AR was 56% (351 of 627) among all attendees; the AR was 62% (351 of 562) when excluding the 65 attendees with an unknown case status. Across age groups, ARs ranged from 44% (4 of 9) among attendees aged 22 to 59 years to 62% (123 of 197) among those aged 11 to 14 years (Table 1). ARs increased with increasing days spent at camp, up to 75% among attendees who spent ≥7 days at camp. Staff members had the highest AR (73%). The median cabin AR was 50% (interquartile range [IQR] = 35%–59%) during orientation and 67% (IQR = 54%–72%) during the camp session; 94% (29 of 31) of cabins had ≥1 case patient (Fig 2, Supplemental Video 1).

The mean effective case reproduction number ranged from 3.2 to 4.0 for case patients with illness onset during orientation (June 17–20) and from 0.1 to 3.5 for those with illness onset during the camp session (June 21–28) (Fig 3A). For community-associated cases, the mean effective reproduction number was 2.0, and for camp-associated cases, it ranged from 0.8 among trainees to 1.3 among staff members. The instantaneous reproduction number was highest (10.1) on June 21 (Fig 3B), indicating a high probability of transmission from infectious case patients at the beginning of the camp session when campers arrived.

We interviewed 450 (70%) attendees to ascertain exposures and activities before, during, and after camp (Supplemental Table 4). Time spent at camp varied by attendee type as follows: 99% of trainees stayed ≤4 days on-site, 96% of staff stayed ≥7 days, and 81% of campers stayed 5 to 6 days at camp (Table 2). At the beginning of orientation, there were 5 community-associated case patients in 5 separate cabins; 2 were symptomatic. At the beginning of the camp session, there were 10 case patients across 7 cabins: 5 symptomatic community-associated case patients, 3 asymptomatic community-associated case patients, and 2 symptomatic camp-associated case patients among staff members who stayed for the camp session. A total of 31 (23%) trainees, 23 (18%) staff members, and 100 (28%) campers stayed in a cabin with ≥1 case patient on the day they arrived at camp (Fig 2, Supplemental Video 1). The proportion of attendees who reported direct contact, such as hugging or kissing, or close contact, such as playing indoor sports or traveling in vehicles, with people outside their cabins was 88%. Approximately 15% of trainees and staff members reported always wearing a mask during camp, compared with 5% of campers. Although singing and cheering were not individually assessed in interviews, a senior staff member described daily vigorous singing and cheering during the camp session. Community activities that could increase the risk for a SARS-CoV-2 exposure before camp, such as eating indoors at restaurants or attending gatherings with nonhousehold members, were commonly reported (58% among staff members and 54% among campers), and 3 attendees reported a known exposure to a person who tested positive for SARS-CoV-2 before camp. Although potential community exposures after camp were less commonly reported (2%), the proportion reporting known exposures, including exposures to other attendees who became sick with COVID-19, increased after camp (12% among staff and 6% among campers).

Among the 404 attendees aged 6 to 21 years with nonmissing values for covariates of interest, staff members were 4.5 times as likely to become a camp-associated case patient compared with trainees (95% CI = 2.7–7.5), adjusting for age group, length of stay, staying in a cabin with a case patient when arriving at camp, and contact with people outside their cabins (Supplemental Table 5). Campers were 3.8 times as likely to become a camp-associated case patient compared with trainees (95% CI = 2.6–5.5), adjusting for the same covariates.

This investigation reveals rapid, widespread SARS-CoV-2 transmission in a congregate setting with children, adolescents, and young adults. Relatively few community-associated case patients were identified, but ARs were as high as 73% among staff members in this sleep-away camp. In this cohort, which included >600 persons aged 6 to 21 years, a majority of whom were tested after a well-defined period of exposure, most cases were characterized by mild or asymptomatic illness, similar to previous, smaller studies characterizing SARS-CoV-2 infection among younger populations.^8,20,21  Nearly half of symptomatic camp-associated case patients reported symptoms that started after leaving camp, suggesting that transmission from presymptomatic individuals contributed to this outbreak.²²  Assuming case patients with available sequences were representative of all case patients, WGS results support the findings that few introductions resulted in widespread transmission.

In this outbreak, estimates of the case reproduction number varied day to day and were as high as 4.0, demonstrating efficient transmission among children, adolescents, and young adults. The instantaneous case reproduction number peaked at 10.1 on June 21 (the first day of camp with an influx of susceptible individuals), indicating that the contact rate and intensity on that day, if sustained, would have resulted in 10 secondary cases per case among attendees. In the multivariable analysis, we found a higher risk of SARS-CoV-2 infection among staff and campers compared with trainees. During the camp session, when cabin occupancy increased, there were also more cases, either asymptomatic or presymptomatic, among attendees. Daily singing and cheering, which has contributed to previous outbreaks,²³  might have increased transmission within cabin cohorts. Most attendees reported having direct or close contact with others outside their cabins, and only 9% reported wearing masks at all times, which likely led to increased transmission between different cabin cohorts. These findings underscore the importance of implementing layered mitigation strategies in settings where younger populations congregate.^24,25 

This investigation is subject to at least 4 limitations. First, the interviews were performed between 2 and 9 weeks after attendees’ last day at camp, subjecting responses to recall bias. Second, misclassification of case status and community- versus camp-associated cases was possible because not all attendees were tested, and among those tested, there could have been false-positive or false-negative results; a 56% AR among all attendees is likely an underestimate. Third, the effect of mask use could not be assessed; few campers reported wearing masks, which were not required. Finally, the types of activities and intensity of contact among and within groups, mainly due to the sleeping arrangements in the camp setting, cannot be extrapolated to all settings that include children, adolescents, and young adults, although some similarities exist (eg, high school students may participate in large-group indoor school activities, college students may interact with the surrounding community, including as counselors for young children in after-school programs).

Other youth-centric settings have also used prearrival testing to reduce transmission.²⁶  In this outbreak, we found that testing within 12 days of arrival, without a mandatory 14-day quarantine, was insufficient to prevent attendees who were infected from arriving at camp and infecting others. Most attendees were residents of the Metro Atlanta area, which had a high incidence of COVID-19 in June 2020.²⁷  Many attendees reported engaging in community activities that could have increased their risk of exposure before arriving at camp. These findings underscore the challenges of preventing outbreaks in areas with substantial community transmission.

Despite mitigation measures, including prearrival testing, relatively few introductions of SARS-CoV-2 into this congregate setting resulted in a large outbreak affecting >50% of attendees. Testing should not be used as the sole mitigation measure²⁸ ; instead, it should be used as one component of a layered mitigation approach combined with adherence to prearrival quarantine, routine symptom monitoring with appropriate isolation and quarantine, cohorting, social distancing, mask wearing, enhanced disinfection, and proper hand hygiene.²⁵  Furthermore, it is important to emphasize appropriate isolation education and compliance for persons who test positive even in the absence of symptoms,²⁹  particularly among younger adults who have been reported to have lower engagement in social mitigation behaviors.^30,31  Targeted communication strategies about behavioral expectations for younger populations may be necessary to emphasize mitigation measures that should be adopted to avoid contracting and spreading COVID-19 to others in youth congregate settings.

Camp Outbreak Field Investigation Team collaborators on this article include Adebola Adebayo, MPH; Tiffiany M. Aholou, PhD, MSW; Minal M. Amin, MS, MPH; Peter Aryee, MBA; Cindy Castaneda, MPA; Trudy V. Chambers; Amy C. Fleshman, MSc; Christin Goodman, MS; Tony Holmes; Asha Ivey-Stephenson, PhD, MA; Emiko Kamitani, PhD, MPH, MS; Susan Katz, MPH; Jennifer K. Knapp, PhD, MPH; Maureen Kolasa, MPH; Maranda F. Lumsden; Erin Mayweather, MPH; Asfia Mohammed; Anne C. Moorman, MPH; Alpa Patel-Larson, MPH; Lara C. Perinet, MS; Mark Pilgard; Deirdre D. Pratt, MSc; Shanica Railey, MPH; Jaina Shah, MPH; and Dawn Tuckey, MPH.

We thank camp attendees and their household contacts. We thank members of the Georgia DPH: Luke Baertlein, Tiffany Baird, Aaron Blakney, Tom Campbell, Alicia Dunajcik, Amit Eichenbaum, Amanda Feldpausch, Pamela Logan, Amanda Mohammed, Stephanie O’Conner, Tonia Parrott, Haley Putnam, Zoe Schneider, Brandon Shih, Kat Topf, and Bill Williamson. We thank member of the CDC: Ramika Archibald, Elizabeth Dietrich, Kathy Fowler, Leah Graziano, Chad Heilig, Margaret Honein, Mark Johnson, Scott Lee, Kelsey McDavid, Robert Montierth, Krista Queen, Joe Sexton, Anupama Shankar, and Robert Slaughter. Lastly, we thank the Alabama DPH, the Arkansas Department of Health, the Colorado Department of Public Health and Environment, the Florida Department of Health, the Maryland Department of Health, the North Carolina Division of Public Health, the South Carolina Department of Health and Environmental Control, the Tennessee Department of Health, the Texas Department of State Health Services, and Ipsum Diagnostics.

AR

attack rate

CDC

US Centers for Disease Control and Prevention

CI

confidence interval

COVID-19

coronavirus disease 2019

CSTE

Council of State and Territorial Epidemiologists

DPH

Department of Public Health

IQR

interquartile range

SARS-CoV-2

severe acute respiratory syndrome coronavirus 2

SNP

single-nucleotide polymorphism

WGS

whole-genome sequencing