key: cord-0907900-ns04aqit
authors: Nash, Denis; Rane, Madhura; Chang, Mindy; Kulkarni, Sarah Gorrell; Zimba, Rebecca; You, William; Berry, Amanda; Mirzayi, Chloe; Kochhar, Shivani; Maroko, Andrew; Robertson, McKaylee M.; Westmoreland, Drew; Parcesepe, Angela; Waldron, Levi; Grov, Christian
title: Recent SARS-CoV-2 seroconversion in a national, community-based prospective cohort of U.S. adults
date: 2021-02-16
journal: medRxiv
DOI: 10.1101/2021.02.12.21251659
sha: 4405b6d109284159c15bbe9936b9335eaaa1db3d
doc_id: 907900
cord_uid: ns04aqit

BACKGROUND: Epidemiologic risk factors for incident SARS-CoV-2 infection are best characterized via prospective cohort studies, complementing case-based surveillance and cross-sectional seroprevalence studies. METHODS: We estimated the cumulative incidence of SARS-CoV-2 infection and incidence rates of seroconversion in a national prospective online cohort of 6,745 U.S. adults, enrolled March-July 2020. A subset (n=4,459) underwent serologic testing (Bio-Rad Platelia Total Ab, IgA/IgM/IgG), offered initially May-September 2020 and again November 2020-January 2021. RESULTS: A total of 303 of 4,459 individuals showed serologic evidence of past SARS-CoV-2 infection (cumulative incidence of 6.8%; 95% Confidence Interval [CI] 6.1%−7.6% [6.3%, 95% CI 5.7%−7.1% adjusting for laboratory test error]). Among 3,280 initially seronegative participants with a subsequent serologic test, we observed 145 seroconversions during 1,562 person years of follow-up (incidence rate of 9.3 per 100 person-years [95% CI 17.9–11.0]). Racial/ethnic disparities in crude incidence rates were apparent through January 2021 (rate ratio [RR(Hispanic vs Whites)]=2.1; 95% CI 1.4–3.1; RR(non-Hispanic Blacks vs Whites)=1.8; 95% CI 0.96–3.1). Incidence was higher in the southern (RR(South vs Northeast)=1.7; 95% CI 1.1–2.8) and midwestern (RR(Midwest vs Northeast)=1.6; 95% CI 0.98–2.7) regions, in rural vs urban areas (RR=1.5; 95% CI 1.0–2.2), and among essential workers (RR=1.7; 95% CI 1.1–2.5). Household crowding (RR=1.6, 95% CI 1.1–2.3), dining indoors at restaurants/bars (RR=2.0; 95% CI 1.4–2.8), visiting places of worship (RR=2.0; 95% CI 1.3–2.9), wearing masks sometimes vs always while grocery shopping (RR=2.5; 95% CI 1.3–4.4), indoor visits with people outside the household with masks (RR(always mask vs no visit)=2.6; 95% CI 1.6–4.4) and without masks (RR(sometimes mask vs no visit)=3.5; 95% CI 2.7–5.7; RR(never mask vs no visit)=5.3; 95% CI 3.1–8.9); working indoors at a place of employment with masks (RR(always mask vs no in-person)=2.0, 95% CI 1.4–2.8) and without masks (RR(sometimes mask vs no in-person)= 2.0, 95% CI 1.1–3.5; RR(never mask vs no in-person)=3.7, 95% CI 1.3–8.5); attending a salon or gym with masks (RR(always mask vs no salon/gym)=1.7 (95% CI 1.1–2.4), gathering indoors and outdoors in groups of ≥10 (RR=1.9, 95% CI 1.2–2.0); and air travel during the pandemic (RR=1.7; 95% CI 1.1–2.6) were also associated with higher incidence rates. Among 303 seropositive individuals, 27.4% had asymptomatic infection, and 32% reported a positive SARS-CoV-2 PCR test or provider diagnosis of COVID-19. In this group, there were major gaps in the coverage of public health interventions aimed at isolation (31% isolated) and contact tracing (asked about contacts [18%]; told about exposure to a confirmed case [7.6%]). CONCLUSIONS: Modifiable risk factors and low reach of public health strategies drive SARS-CoV-2 transmission across the U.S. It is critical to address inequities in incidence, reduce risk factors, and improve the reach of public health strategies in the vaccine era.

seroprevalence studies.

Methods 50 We estimated the cumulative incidence of SARS-CoV-2 infection and incidence rates of 51 seroconversion in a national prospective online cohort of 6,745 U.S. adults, enrolled March-July 52 2020. A subset (n=4,459) underwent serologic testing (Bio-Rad Platelia Total Ab, 53 IgA/IgM/IgG), offered initially May-September 2020 and again November 2020-January 2021.

A total of 303 of 4,459 individuals showed serologic evidence of past SARS-CoV-2 infection 57 (cumulative incidence of 6.8%; 95% Confidence Interval [CI] 6.1%-7.6% [6.3%, 95% CI 5.7%-58 7.1% adjusting for laboratory test error]). Among 3,280 initially seronegative participants with a 59 subsequent serologic test, we observed 145 seroconversions during 1,562 person years of follow-60 up (incidence rate of 9.3 per 100 person-years [95% CI 17.9-11.0]). Racial/ethnic disparities in 61 crude incidence rates were apparent through January 2021 (rate ratio [RR Hispanic vs Whites ]=2.1; 62 95% CI 1.4-3.1; RR non-Hispanic Blacks vs Whites =1.8; 95% CI 0.96-3.1). Incidence was higher in the 63 southern (RR South vs Northeast =1.7; 95% CI 1.1-2.8) and midwestern (RR Midwest vs Northeast =1.6; 95% 64 CI 0.98-2.7) regions, in rural vs urban areas (RR=1.5; 95% CI 1.0-2.2), and among essential INTRODUCTION 92 As of this writing, about one year after the emergence of SARS-CoV-2 in the United States 93 (U.S.), there have been more than 25 million SARS-CoV-2 cases diagnosed, 435,000 deaths 94 recorded, and 23.5 million vaccine doses administered. 1 As compared to diagnoses, the true 95 number of infections to date in the U.S. is unknown, but national estimates in the general 96 population as measured by seroprevalence have put the cumulative incidence at 10% in 97 September 2020 2 and 14% as of November 2020. 3 One recent study in a nationally 98 representative sample of U.S. adults found that 4.6% had a history of undiagnosed SARS-CoV-2 99 as of July 2020: about 5 undiagnosed infections for every diagnosed case. 4 100 A major challenge of controlling community transmission of SARS-CoV-2 is that the virus' 101 infectious period allows for onward spread prior to recognition of infection, by the substantial 102 proportion of people with asymptomatic infection, and during the pre-symptomatic period among 103 those who go on to develop symptoms. 4, 5 104 While SARS-CoV-2 is understood to be transmitted from person-to-person via droplet and 105 airborne spread, to date, the incidence of SARS-CoV-2 infection and risk factors for incident 106 infection have not been well-characterized by routine case-based surveillance of SARS-CoV-2 107 diagnoses or by cross-sectional seroprevalence studies. 1-3 As the pandemic progresses through 108 different seasons and stages, it is critical to continue to characterize COVID-19's evolving 109 epidemiology and risk factors for SARS-CoV-2 acquisition, the uptake and impact of non-110 pharmaceutical interventions (NPIs) 6 , and the reach of public health strategies aimed at 111 controlling community transmission, including testing, quarantine, isolation, contact tracing, and 112 vaccination. 113 Researchers and public health practitioners have called for prospective cohort studies to 114 describe incidence rates and how they are influenced by NPI implementation and uptake, 115 sociodemographics, knowledge and behaviors, and other potential risk factors. 7 Yet, both the 116 frequency with which many risk factors occur in community samples, as well as the SARS-CoV-117 2 risk that could be attached to them, still have not been systematically assessed and 118 epidemiologically linked to outcomes like seroconversion. Globally, few prospective 119 epidemiologic studies of SARS-CoV-2 have been published to date. One recent global 120 systematic review of observational studies of SARS-CoV-2 that employed serologic or PCR 121 testing found 18 prospective studies. 8 Most were focused on healthcare workers or other 122 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 16, 2021. ; occupational groups, individuals in congregate settings, evacuees, or cruise ships; none were 123 community-based (i.e., focused on risk factors in communities vs other higher risk 124 populations/settings). 8 A greater understanding of SARS-CoV-2 incidence and risk factors in 125 community samples can substantially complement routine case-based surveillance of new SARS- 126 CoV-2 diagnoses and cross-sectional serosurveys, serving to inform aspects of implementation 127 of the public health response and policies. 128 In late March 2020, we launched the national, prospective Communities, Households and 129 SARS-CoV-2 Epidemiology (CHASING) COVID Cohort. 9 We describe the cumulative 130 incidence of SARS-CoV-2 infection, risk factors for recent SARS-CoV-2 seroconversion, and, 131 for those with serologic evidence of prior SARS-CoV-2 infection, the reach and uptake of public 132 health strategies. 133 134 METHODS 135 We used internet-based strategies 10-12 to recruit a geographically and socio-demographically 136 diverse cohort of adults in the U.S. and U.S. territories into longitudinal follow-up with at-home 137 specimen collection. To be eligible for inclusion in the cohort, individuals had to: 1) reside in the 138 U.S. or a U.S. territory; 2) be aged 18 years or older; 3) provide a valid email address for follow-139 up; and 4) demonstrate early engagement in study activities (provided 

The characteristics of the cohort are shown in Table 1 , which includes participants from all 50 205 U.S. states, the District of Columbia, Puerto Rico, and Guam (see Fig. S1 ). Of the 6,745 206 participants enrolled in the cohort, 4,459 (66%) had at least one serologic test (Table 1) . 207 Compared with the entire cohort, those who underwent any serologic testing were older and 208 more likely to be non-Hispanic White, college educated, and less likely to be healthcare workers 209 or essential workers. 4,235 persons underwent serologic testing in Period 1, and 3,615 persons 210 tested in Period 2. A total of 3,339 persons tested at both time points, including 3,280 who were 211 seronegative on their antibody test in Period 1 and who could be followed prospectively for the 212 outcome of SARS-CoV-2 seroconversion based on their subsequent serologic test result in the 213 study. Differences between those testing in Period 1 and Period 2 were negligible (Table S1) . 214 The median time between specimen collection dates for both serologic tests was 190 days (IQR 215 152-201) ( Figure S2 ). 

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 16, 2021. ; Cumulative incidence since the beginning of the pandemic and potential risk factors 219 The crude and adjusted serology-based estimates of cumulative incidence of SARS-CoV-2 220 infection through the end of observation (January 31, 2021) were 6.8% (95% CI 6.1%-7.6%) and 221 6.3% (95% CI 5.5%-7.1%), respectively (Table S2 ). Cumulative incidence rates declined in 222 dose-response fashion with increasing age, with substantially lower incidence in those aged 60 223 years and older ( Figure 1 ). Higher cumulative incidence estimates were observed among males 224 (7.8%, 95% CI 6.6%-9.0%), Hispanics 10.2% (95% CI 7.8%-12.6%), non-Hispanic Blacks 225 (7.9%, 95% CI 5%-10.8%), those with a high school education (9.1%, 95% CI 0.9-20.9) or less 226 (8.0%, 95% CI 5.2-10.9%), and those in the Northeast (7.8%, 95% CI 6.3%-9.4%). Lower 227 cumulative incidence estimates were observed among persons aged 60 years and over (3.8%, 228 95% CI 2.5%-5.2%), those whose gender identity was non-binary (2.5%, 95% CI 0.4%-5.5%), 229 and those who were retired (3.4%, 95% CI 1.7%-4.9%).

230 231 232 Figure 1 shows cumulative incidence estimates and 95% confidence intervals (CIs) for SARS-CoV-2 by baseline demographic characteristics of 233 CHASING COVID Cohort Study participants. Estimates are adjusted for laboratory test error. Essential workers include people working in in law 234 enforcement, emergency management, delivery, transportation, and construction. High risk group is defined as participants who either reported a 235 comorbidity, were over 60 years old, or were a smoker. The size of the square box/marker is proportional to the overall sample size in each group. 236 237 nd . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 16, 2021. ;

The estimated cumulative incidence of SARS-CoV-2 infection by epidemiologic risk factors 238 reported at enrollment are shown in Figure 2 (and in Table S3 ). Higher cumulative incidence 239 estimates were observed among those who, at cohort enrollment, reported having attended mass 240 gatherings (8.6%, 95% CI 5.8%-12.4%), and lived in more crowded households (8.7%, 95% CI 241 5.7%-13.2%). Those who thought they had COVID-19 in the past, and in Period 2, those who 242 had a known SARS-CoV-2 exposure, sought a test, had a self-reported SARS-CoV-2 diagnosis 243 or COVID-19-related hospitalization also had markedly higher cumulative incidence (Table S3) . 244 People living with HIV had the highest cumulative incidence among those with comorbidities 245 (8.8%, 95% CI 5.5%-13.6%) over the entire study period, however, those with diabetes had the 246 highest cumulative incidence in Period 2 (Table S3 ). In terms of other potential risk factors, 247 those who reported recently drinking more than 6 alcoholic drinks on one occasion (9.9%, 95% 248 CI 6.6%-14.3%) and those who used public transportation (8.6%, 95% CI 5.7%-12.6%) had 249 higher cumulative incidence (Fig. 2 , Table S3 ). Lower cumulative incidence was observed 250 among those who, at enrollment, reported having asthma, immunosuppression, and depression. by . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 16, 2021. ; https://doi.org/10.1101/2021.02.12.21251659 doi: medRxiv preprint more than 6 drinks on one occasion weekly, daily, or almost daily. (NPIs: Non-pharmaceutical interventions; HH: Household). The size of the square 256 box/marker is proportional to the overall sample size in each group.

Incidence of recent SARS-CoV-2 seroconversion and potential risk factors 259 Figure 3 (and Table S4 ) shows more recent seroconversions and incidence rates among 3,280 260 initially seronegative persons who had a subsequent serologic test during November 2020-261 January 2021 (Fig. S2) , stratified by risk factors that were present prior to or between serologic 262 tests. There were 145 observed seroconversions over 1,562 person years of follow-up among the 263 3,280 participants, for an overall incidence rate of 9.3 per 100 person years (PY) (95% CI 7.9-264 11.0). The rate ratio (RR) for incident SARS-CoV-2 infection was higher for males than females 265 CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 16, 2021. ; https://doi.org/10.1101/2021.02.12.21251659 doi: medRxiv preprint Figure 3 shows risk ratio estimates and 95% confidence intervals (CIs) for recent SARS-CoV-2 seroconversion (May 2020-January 2021) in CHASING 274 COVID Cohort Study participants by demographic factors. Essential workers include people working in in law enforcement, emergency management, 275 delivery, transportation, and construction. High risk group is defined as participants who either reported a comorbidity, were over 60 years old, or were 276 a smoker. The size of the square box/marker is proportional to the overall sample size in each group. Figure 4 (and Table S5 ) shows recent seroconversions and incidence rates for epidemiologic 279 risk factors. Living in more crowded households was associated with higher incidence than 280 living in less crowded households (RR 1.61, 95% CI 1.08-2.32). Compared with referent groups 281 in Figure 4 , there was also higher incidence among those who dined indoors at restaurants/bars 282 (RR 1.95, 95% CI 1.38-2.83); those who visited a place of worship (RR 1.95, 95% CI 1.25-2.93); 283 those who wore a mask only sometimes while grocery shopping (RR 2.49, 95% CI 1.31-4.36); 284 those who visited indoors with people not in their own household while always wearing a mask 285 (RR=2.6; 95% CI 1.6-4.4), while sometimes wearing a mask (RR=3.5; 95% CI 2.7-5.7 or while 286 never wearing a mask (RR=5.3; 95% CI 3.1-8.9); those working indoors at a place of 287 employment while always wearing a mask (RR=2.0, 95% CI 1.4-2.8), while sometimes wearing 288 a mask (RR=2.0, 95% CI 1.1-3.5), or while never wearing a mask RR=3.7, 95% CI 1.3-8.5); 289 those attending a salon or gym with masks (RR=1.7, 95% CI 1.1-2.4); those who gathered 290 indoors and outdoors in groups of >10 (RR=1.9, 95% CI 1.2-2.0); and those that traveled by air 291 during the pandemic (RR 1.72, 95% CI 1.12-2.55). 292 293 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 16, 2021. ; were aware that they had a prior SARS-CoV-2 infection (Table S6) . A substantial proportion of 302 the 303 (27.4%) recalled no symptoms of COVID-like illness (i.e., were asymptomatic cases). 303 Among the 145 more recent seroconverters in Period 2, the proportion reporting no symptoms 304 was 25.5%, and another 60% reported milder symptoms that did not result in seeking care. 305 In terms of public health outcomes, 64.4% (195/303) said that they were tested for SARS- 306 CoV-2 outside the study (Table S6) , half of them (32% of total) reported having a positive 307 SARS-CoV-2 test. However, only 30.7% of all 303 seropositives said that they had isolated 308 themselves from people outside their household because of their infection, and, among those 309 who did not live alone, even fewer (13.5% overall) said they isolated themselves from others 310 within their household. In terms of contact tracing, 17.8% of all seropositives were asked about 311 G rs . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 16, 2021. ;

contacts following diagnosis and only 12.2% of all seropositives had been informed by a contact 312 tracer that they may have had contact with someone confirmed to have SARS-CoV-2. Only 5.3% 313 of those diagnosed (3.3% of all seropositives) were told by a contact tracer to stay home for a 314 period of time because they had COVID-19. (Table S7) . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 16, 2021. ; immunity, more nuanced understanding of the actual contribution of these factors over time is 343 needed. This in turn will facilitate more effective government and community public health 344 policies, actions and implementation strategies to control community spread, which is key to 345 preventing avoidable hospitalizations and deaths, and ultimately maximizing the impact of 346 vaccines. In areas where there is not an enforced lock down, our findings provide an evidence-347 base for federal, state, and local policy focus on risk factors that appear to be potential drivers of 348 late-phase pandemic spread, such as indoor dining, mass gatherings, or not wearing masks when 349 outside the home. 350 Data on more recent seroconversions occurring between May 2020 and January 2021 in our 351 cohort suggest that elevated risk among essential workers, observed early in the U.S. pandemic, 352 has persisted. Essential workers risk exposure to SARS-CoV-2 not only in their workplace(s), 353 but also in their communities and as part of their commutes to and from work if they need to use 354 public transportation. The increased burden of risk of SARS-CoV-2 infection in essential 355 workers is then shared with their household members, among whom transmission is very 356 efficient. 19 Along with this comes a higher burden of stress and anxiety among essential 357 workers. 20 The most immediate action that governments can take to help keep essential workers 358 safe is to minimize community transmission so that they are less likely to be exposed in the 359 workplace or as part of their commutes. Policies and other strategies should be considered and 360 implemented to ensure workplace safety (high quality masks, distancing, ventilation, paid sick 361 leave for those that cannot afford to miss work) and the safety of public transportation. Any 362 strategies should include prioritized access to SARS-CoV-2 vaccines for essential workers and, 363 until or in the absence of vaccination, rapid testing for essential workers and members of their 364 households. FDA-approved fully at-home rapid testing options are becoming more readily 365 available with or without a prescription, and past research has indicated an overall increase in the 366 uptake of testing with the addition of at-home testing options, 21,22 making them a potentially 367 important strategy to increase early identification of new infections and subsequent interventions 368 to prevent onward transmission. In this later phase of the pandemic, healthcare workers in our 369 cohort had borderline elevated risk. Healthcare workers in the U.S. could be expected to have 370 lower risk of SARS-CoV-2 exposure in the workplace due to focused efforts to mitigate risk via 371 improved access to personal protective equipment and engineering controls. However, healthcare 372 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 16, 2021. ; workers also have ongoing community exposure, including possibly during their commutes to 373 and from work, that could put them at higher risk of SARS-CoV-2 exposure and acquisition. 374 Our study provides epidemiologic evidence of increased risk for a number of key exposures. 375 Living in more crowded households, defined as living in a multi-unit (i.e., apartment) building 376 with 4 or more household members, was associated with a 60% higher incidence of SARS-CoV-377 2 infection compared with those living in less crowded households. Household transmission of 378 SARS-CoV-2 occurs with very high attack rates, and may also increase the risk of more severe 379 COVID-19 disease, requiring hospitalization. 23 Household crowding, more than population 380 density, was shown to be a characteristic of community transmission 'hotspots' in a recent 381 ecological analysis from the early phase of the pandemic. 24 When there are possible or confirmed 382 cases of SARS-CoV-2 among household members, when background community transmission is 383 substantial, or when the possibility of exposure outside the household is high (e.g., essential 384 workers), our study and others 25 suggest mask use in the household should be considered, 385 especially when it is otherwise not possible to limit potential exposure and spread in the 386 household by isolating or sequestering away from others. As the vaccine rollout continues, this is 387 particularly important to protect unvaccinated household members who may be at high risk for a 388 severe COVID-19 outcome. 389 We characterized the potential contribution of several other key risk factors, some of which 390 have been, or could be, the focus of state and federal policies, including dining indoors at a 391 restaurant/bar, visiting a place of worship, inconsistent mask use while grocery shopping or 392 visiting non-household members indoors (especially without masks), visiting a salon/gym, 393 gathering indoors in groups of 10 or more, and recent travel by airplane during the pandemic. 394 Working indoors, even while always masking but especially while never wearing a mask, was 395 strongly associated with higher infection rates. In addition to being potential individual policy 396 focus areas, we note that all of these risks can be mitigated by social distancing, workplace 397 safety, increased mask use, use of more effective masks (e.g., KN95), and use of more than one CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 16, 2021. ; crowding, the need to go to work to avoid income loss, and inequitable access to SARS-CoV-2 435 testing 34 , create a disparate burden of SARS-CoV-2 exposure and incidence. 35 To date, no 436 targeted strategies or policies have been deployed that aim to protect those who cannot afford 437 missing work, including, but not only, essential workers. These disparities in SARS-CoV-2 438 exposure, incidence, hospitalization and death will likely be perpetuated and likely exacerbated 439 by differential uptake of the COVID-19 vaccines, driven by potentially flawed implementation 440 strategies that prevent equitable access to vaccination. It is incumbent upon public health leaders, 441 decision makers, and policy makers to anticipate and proactively design pandemic response 442 implementation strategies, including metrics, that account for and counteract the fundamental 443 and prevailing structural forces that without fail will otherwise create, perpetuate, or exacerbate 444 inequities in safety, health and well-being. 36-39 445 Strengths of our cohort study include its prospective design, allowing direct observation of 446 seroconversions and incident COVID-19 infection among those who had no serologic evidence 447 of prior SARS-CoV-2 infection, with and without risk factors. Our cohort is also geographically 448 and sociodemographically diverse, facilitating assessment of geographic differences and 449 racial/ethnic disparities. We also measured and were able to estimate incidence rates associated 450 with a range of relevant epidemiologic risk factors. 451 Our study also has limitations worth noting. Most research studies deployed in the middle of 452 a pandemic, including ours, may produce biased estimates since they may not include complete 453 information on participants who died from COVID-19, or were too sick or otherwise too busy to 454 participate in the research activities, or enrolled but became lost to follow-up. Moreover, it is 455 possible that those who thought they had SARS-CoV-2 or were tested outside the study may 456 have been more or less likely to participate in follow-up serologic testing, which would lead our 457 seroincidence estimates to be biased in either direction. However, there was no systematic 458 difference between those who tested in Period 1 and Period 2 (Table S1 ). Separately, while we 459 have corrected our cumulative incidence estimates for laboratory test error 18 , the observed 460 cumulative incidence in our cohort may be lower than the true cumulative incidence in our 461 cohort because of waning of SARS-CoV-2 antibodies. Recent studies suggest waning of 462 antibodies to both nucleocapsid and spike proteins 2 , which combined with the timing of 463 specimen collection relative to infection for many participants in our cohort (median of 190 464 days) 9 , could mean that we have underestimated the true cumulative incidence. Next, as with any 465 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 16, 2021. ; observational cohort study, estimated associations between SARS-CoV-2 risk factors and 466 incidence are subject to confounding. The crude associations we presented could be over-or 467 underestimated, and also may vary by setting. For example, the effect of indoor dining may be 468 different in areas of lower versus higher levels of community transmission. Our study also had 469 limited statistical power due to our sample size and the relatively short duration of follow-up. 470 Finally, while our study is well-suited to characterize the risks related to SARS-CoV-2 471 acquisition, it is less well-suited to examine factors related to onward transmission from The authors wish to thank the participants of the CHASING COVID Cohort Study. We are 503 grateful to you for your contributions to the advancement of science around the SARS-CoV-2 504 pandemic. We thank Prof. Patrick Sullivan and MTL for local validation work on the serologic 505 assays for use with DBS that greatly benefited our study. We are also grateful to MTL Labs for 506 processing specimen collection kits and serologic testing of our cohort's specimens. 

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 16, 2021. ; https://doi.org/10.1101/2021.02.12.21251659 doi: medRxiv preprint 1. CDC, COVID Data Tracker (2020), (available at https://covid.cdc.gov/covid-data-tracker/). CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 16, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 needed. N. Engl. J. Med. 382, 1194 -1196 . 

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 16, 2021. ; https://doi.org/10.1101/2021.02.12.21251659 doi: medRxiv preprint

Geographic distribution of CHASING COVID Cohort participants, N=6,745 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 16, 2021. ; https://doi.org/10.1101/2021.02.12.21251659 doi: medRxiv preprint

Timing of first (red) and second (blue) dried blood spot specimen collection in the CHASING COVID Cohort Study, including follow-up interview milestones . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 16, 2021. ; is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 16, 2021.

; https://doi.org/10.1101/2021.02.12.21251659 doi: medRxiv preprint 3 value for this variable ** >60 years old, or reported co-morbidity, or current smoker. Table S1 . Baseline characteristics and serologic testing of CHASING COVID Cohort Study participants by time period of testing . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 16, 2021.

1 Table S2 . Crude and adjusted** cumulative incidence estimates of SARS-CoV-2 infection in CHASING COVID Cohort Study participants by time period of testing .

It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 16, 2021. 839.5 8.7 (6.9, 11.0) 0.87 (0.63, 1.21) *<5% of the data were missing for this variable and values for those with missing values were imputed by assigning the most common value for this variable **>60 years old, or reported co-morbidity, or current smoker. ***PY = Person Years Table S4 . Incidence rates of recent SARS-CoV-2 seroconversion (May 2020-January 2021) by sociodemographic characteristics among CHASING COVID Cohort participants . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint Table S5 .

Incidence rates of recent SARS-CoV-2 seroconversion (May 2020-January 2021) by risk factor among CHASING COVID Cohort participants . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 16, 2021. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 16, 2021. Additional serologic testing for antibodies to the SARS-CoV-2 spike protein in 392 samples* that were seropositive for antibodies to the SARS-CoV-2 nucleocapsid protein .

CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted February 16, 2021. 

the Northwell COVID-19 Research Consortium

Comorbidities, and Outcomes Among 5700 Patients Hospitalized With 631 COVID-19 in the New York City Area

Factors associated with hospital admission 634 and critical illness among 5279 people with coronavirus disease

Using Lorenz Curves to Measure Racial Inequities in 637 COVID-19 Testing

Assessment of Racial/Ethnic Disparities in 641 Hospitalization and Mortality in Patients With COVID-19 in New York City

Goal-Aligned, Epidemic Intelligence for the Public Health Response to 644 the COVID-19 Pandemic

Designing and Disseminating Metrics to

Substance use***