key: cord-0025976-7ecme7kv
authors: Akinbami, Lara J; Biggerstaff, Brad J; Chan, Philip A; McGibbon, Emily; Pathela, Preeti; Petersen, Lyle R
title: Reinfection with SARS-CoV-2 among previously infected healthcare personnel and first responders
date: 2021-11-15
journal: Clin Infect Dis
DOI: 10.1093/cid/ciab952
sha: 231bca912ed7d7f6c2cbc8d208b96761d525c1e9
doc_id: 25976
cord_uid: 7ecme7kv

BACKGROUND: SARS-CoV-2 virus testing among first responders and healthcare personnel who participated in a May-August 2020 serosurvey which assessed spike protein antibodies (S1 region) provided an opportunity to assess reinfection. METHODS: Serology survey data were merged with virus testing results from Rhode Island (March 1, 2020-February 17, 2021) and New York City (March 10-December 14, 2020). Participants with a positive virus test ≥14 days before their serology test were included. Reinfection was defined as a second positive SARS-CoV-2 test result ≥90 days after the first positive test. The association between serostatus and reinfection was assessed with a proportional hazards model adjusting for demographics, exposures, and virus testing frequency. RESULTS: Among 1,572 previously infected persons, 40 (2.5%) were reinfected. Reinfection differed by serostatus: 8.4% among seronegative versus 1.9% among seropositive participants (p<0.0001). Most reinfections occurred among Rhode Island nursing home and corrections (RINHC) personnel (n=30) who were most frequently tested (mean 30.3 tests versus 4.6 for other Rhode Island and 2.3 for New York City participants). The adjusted hazard ratio (aHR) for reinfection in seropositive versus seronegative persons was 0.41 (95% CI 0.20, 0.81). Exposure to a household member with COVID-19 before the serosurvey was also protective (aHR 0.34, 95% CI 0.13, 0.89). CONCLUSIONS: Reinfections were uncommon among previously infected persons over a 9-month period that preceded widespread variant circulation. Seropositivity decreased reinfection risk. Lower reinfection risk associated with exposure to a household member with COVID-19 before the serosurvey may reflect subsequently reduced household transmission among members of previously infected households.

Studies are needed to build the evidence base about the frequency of and risk factors for reinfection with severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2). Most definitively, documented reinfection has occurred among persons infected with a genetically different variant than the primary infection [1, 2] . However, few persons have stored samples to make this determination.

Probable reinfections have also been described among persons with two positive virus detection tests (antigen or viral nucleic acid amplification test [NAAT] ) separated by long intervals. In the absence of genomic sequencing, a 90-day timeframe is used to determine a reinfection for surveillance purposes [3] . Uncertainty regarding the duration of detection of viral nucleic acid due to shedding following infection previously limited this approach. However, NAAT testing data among frequently tested cohorts suggest that the duration of viral shedding is <90 days in immunocompetent persons [4] [5] [6] .

Observational studies and data from serially tested cohorts suggest that previous infection confers some degree of immunity, at least for several months. Among staff and residents of two British nursing homes, seropositive persons were 96% less likely to become infected during a second outbreak four months later [7] . In a British prospective cohort study which observed persons with and without SARS-CoV-2 antibodies for seven months, seropositivity was associated with an 84% lower risk of infection [8] . Another British cohort study found an 83% reduction in the incidence of SARS-CoV-2 infection over a five-month period among seropositive persons or those with prior infection documented by NAATs [9] . A large U.S. study that linked commercial laboratory results with medical claims data and electronic medical records found a 90% reduction in infection (measured ≥90 days after baseline antibody testing) among seropositive compared to seronegative persons [10] . A French study of healthcare workers found an 85% reduction in infection for seropositive versus seronegative persons after a 6-month observation period [11] . A study of longterm care facility (LTCF) residents and workers in England with periodic serology testing and weekly A c c e p t e d M a n u s c r i p t 4 virus testing found that seropositivity reduced risk of reinfection by 85% in residents aged ≥65 years and by 61% in staff members over a 10-month observation period [12] . Finally, a South African community study found an 84% reduction in reinfection when initial infection was defined by positive NAAT or serology test [13] .

These studies defined prior infection using positive serology status; thus, persons who were previously infected but did not develop antibodies could thereby have been misclassified. Prior studies have shown that among persons known to have been infected, a sizeable minority remain seronegative [14, 15] . Although seropositive persons appear to be protected from reinfection compared to initial infection among seronegative persons, the risk of reinfection among infected persons who do not seroconvert has not previously been assessed. To identify frequency and risk factors for SARS-CoV-2 reinfection among seropositive and seronegative persons with previous SARS-CoV-2 infection based on virus testing, we merged NAAT or antigen test results from population-based surveillance systems with study data for health care workers and first responders who participated in SARS-CoV-2 serology surveys in Rhode Island and New York City [16, 17] .

Serology surveys focused on first responder and healthcare personnel in Rhode Island during July-August 2020 and in New York City during May-July 2020 as previously described [16, 17] . In both settings, the serology survey occurred between two waves of transmission (Supplemental Figure 1 ).

Products Anti-SARS-CoV-2 IgG test directed at the S1 domain of the spike protein (https://www.orthoclinicaldiagnostics.com) [16, 17] . Only participants with an initial positive virus test result (defined below) ≥14 days before their serology test were included. Those with no positive virus test results were excluded because they were presumed never infected and therefore were not at risk for reinfection. Those with a first positive virus test result after their serology test were also excluded because the association between serology status and reinfection could not be assessed. The final sample included 373 Rhode Island participants and 1,199 New York City participants (Supplemental Figure 2) . Outcome and main exposure SARS-CoV-2 testing data included the date and result of each virus test performed. The study outcome, reinfection with SARS-CoV-2, was defined as a positive virus test (either NAAT or antigen test) collected ≥90 days after an initial positive test [3] . Participants who had two positive tests ≥90 days apart were considered to have an observed reinfection. Serology status (positive or negative) was the primary factor of interest in assessing risk of reinfection.

Serology survey participants reported their primary work agency, age, sex, race/ethnicity, underlying medical conditions and date of symptom onset. Primary workplace agency was categorized dichotomously as Rhode Island nursing home or corrections facilities (RINHC) versus all other agencies for the following reasons: nursing homes and corrections facilities were congregate settings in which personnel underwent more frequent testing than those in other workplaces; additionally, some workplace agencies had low samples sizes, which precluded robust statistical analysis; and finally, reporting frequencies for workplaces with small sample sizes could pose a disclosure risk.

Participants were asked to report an exposure to anyone known to have COVID-19, with an exposure defined as >10 minutes within 6 feet of a person with COVID-19 per the April 2020 Council of State and Territorial Epidemiologists case definition [18] prior to survey participation. Frequency of virus testing was calculated by summing reported test results and was incorporated into statistical modeling to provide adjustment for serology status to account for factors beyond those provided by adjusting for workplace agency type.

A c c e p t e d M a n u s c r i p t

Demographic and workplace characteristics were calculated overall and for subgroups. Differences were assessed using Fisher's exact test with the mid-p correction with significance level of 0.05.

Cumulative incidence of reinfection was assessed among previously infected seropositive and seronegative persons. Direct bivariate comparisons of reinfection risk by characteristic were not performed because individual differences in follow-up times could not be accounted for. Kaplan-Meier curves were generated to show the cumulative probability of a subsequent positive virus test by antibody status. To allow for varying individual follow-up times and right-censored observations, a multivariable Cox proportional hazards regression (CPHR) was used to assess associations of covariates with reinfection risk. Exploratory analyses were used to assess the proportional hazards assumption and model selection was based on likelihood ratio tests. Estimates of hazard ratios and 95% confidence intervals (CI) were calculated. Length of follow-up was defined as the interval from the first positive virus test to either a subsequent positive virus test that occurred ≥90 days later, or to the last negative virus test that occurred ≥90 days later for those with no subsequent positive virus test, or to the end of the virus test reporting period for those with neither of these two occurrences.

The study included 1,572 persons, 373 from Rhode Island and 1,199 from New York City who tested positive for the presence of SARS-CoV-2 RNA ≥14 days prior to collection of a serum sample. Preliminary analyses indicated the proportional hazards assumption was tenuous by workplace agency (RINHC versus other), so we conservatively included this variable in CPHR analyses as a stratification variable. We evaluated testing frequency by including it in preliminary CPHR models as a penalized spline (degree 3) term, separately for each dichotomous workplace group to evaluate the functional shape of the resulting curves, as numerical instability precluded using spline fits interacted with the dichotomous workplace agency variable. Fitted spline curves appeared wellapproximated by linear and quadratic functions of virus testing frequency for RINHC and other workplace groups, respectively. We therefore included terms for serology status and linear and quadratic terms for testing frequency, each interacted with workplace agency, and then evaluated other covariates for model inclusion ( Table 3 ). The final model selected using likelihood ratio tests contained serology status, exposure to COVID-19 positive household members, and main effects and interaction terms for testing frequency and workplace agency dichotomy. From this model fit, the resulting estimated adjusted hazard ratio for serostatus was 0.41 (95% CI 0.20, 0.81). Exposure to a A c c e p t e d M a n u s c r i p t 9 household member with COVID-19 prior to the serology survey was also inversely associated with reinfection (aHR 0.34, 95% CI 0.13, 0.89).

This is among the first studies to evaluate the risk of reinfection by antibody status among previously infected individuals. Reinfection with SARS-CoV-2 was uncommon (2.5%) in this large cohort of first responders and healthcare workers in Rhode Island and New York City, and occurred less frequently in seropositive versus seronegative persons. The 9-to 11-month observation period included the summer months of 2020 when community transmission was low, as well as the fall and early winter months when the rates of community transmission were similar to or greater than the early pandemic period. This observation period also preceded widespread full vaccination with mRNA vaccines, especially for New York City given the shorter observation period in this site. These results agree with prior studies that showed reinfection was uncommon and provided additional evidence that even in those with NAAT-confirmed re-infections, seropositivity is associated with lower risk of reinfection [19] .

Previous studies that assessed cumulative risk of infection among seropositive versus seronegative persons observed between 83% and 96% reduction in reinfection risk [7, 9, 12, 19] , higher than the 59% reduction observed in this study. This lower risk reduction could be due to several factors. First, the impact of serology status on reinfection risk was assessed among persons with a positive virus test rather than a positive serology test, and thus included seropositive and seronegative persons with known prior infection. Most prior studies compared infection rates among seropositive versus seronegative persons at baseline with the assumption that seronegative persons had not been previously infected. These studies all found antibodies were associated with reduced risk of A c c e p t e d M a n u s c r i p t 10 reinfection among seropositive persons [7-9, 11, 12, 20] but in comparison to initial infection rates among presumably never infected persons. If initial infection is determined by virus testing rather than serology testing, a lower protective effect of seropositivity is plausible. That is, compared to a seronegative group that includes persons never infected, a seronegative group with documented positive virus tests may have some immunity, including cellular immunity [21] . Second, reinfection was most frequently observed in the cohort that underwent frequent testing (RINHC personnel).

Asymptomatic reinfections may have been more likely to be detected with mandated testing at frequent intervals as opposed to testing indicated by symptoms or suspected/known exposures.

Lower protection of seropositivity could be observed if serum antibodies are less effective in preventing asymptomatic infection. Third, this study had a long observation period (9-11 months) compared to some previous studies that ranged from 4 months to just under 8 months [7] [8] [9] 19] . A study of LTCF staff and residents with a similar follow-up period (up to 10 months) found a similar level of protection of antibodies among LTCF staff (61%) to our study [12] .

Reinfection was less likely among persons with reported exposure to a household member with COVID-19 prior to participation in the serology survey but was not associated with other potential risk factors in adjusted models, including the presence of chronic health conditions. Studies of healthcare personnel have found that community transmission was a strong risk factor for initial infection [17, [22] [23] [24] [25] [26] [27] . Persons initially infected in their households may have a lower risk of subsequent household infection if household members also developed immunity. Household exposure may also be more easily identified compared to exposure to other persons with COVID-19.

The main limitation of this study was lack of clinical data associated with subsequent positive virus tests. Thus, we could not assess clinical severity of reinfections. One study that obtained paired genomic sequences to determine reinfection observed that among those with antibodies, reinfections were less severe than primary infections [19] . A prospective study of LTCF staff and A c c e p t e d M a n u s c r i p t 11 residents identified 14 reinfections during a 10-month period, of which 11 were symptomatic but none required hospitalization [12] . Given that most reinfections occurred in RINHC personnel and may have been detected in the workplace after symptom-based screening, it is likely that many of the reinfections detected were not clinically severe or even symptomatic. Our study population included only working adults. Thus, risk factors for severe symptomatic SARS-CoV-2 reinfection among unvaccinated individuals observed in a nationwide study in Mexico[28]-severe initial infection, older age, and comorbid conditions-were less likely to be present in this study cohort than in the general population. Another limitation is lack of data of vaccination status. COVID-19

vaccinations began in both sites in mid-December. While full immunity would not have been achieved until up to 5 weeks later among the first to be vaccinated (2 weeks after the second dose), vaccine-derived immunity may have impacted observed results in Rhode Island, which had a longer observation period. Next, a higher percentage of previously infected Rhode Island staff were seronegative, which could raise the possibility of false positive NAATs. However, viral tests generally have high specificity (above 98%) [29] . Additionally, previous studies have shown up to 16% of previously infected persons are seronegative, even in study populations with few immunocompromised persons [14, 15] . In other words, failure to develop detectable antibodies after SARS-CoV-2 infection may not be a rare phenomenon. The differences in testing frequency between sites could also have resulted in fewer primary infections being identified in setting where testing availability was limited, especially early in the pandemic. Because our study population was a healthy, younger working population with few immunocompromising conditions, results may not be generalizable. The study period preceded widespread transmission of variants of concern or interest, for which observed protection afforded by antibodies acquired after infection may differ from that of earlier circulating viral strains. It is unknown to what extent asymptomatic reinfections pose transmission risk, but previous studies found the mean cycle threshold value was lower among persons with symptomatic reinfection versus all reinfection [9] and among primary infections versus reinfections [12] .

A c c e p t e d M a n u s c r i p t 12 Strengths of our study included a large study cohort with information on demographics, workplace, and exposures during the first wave of the pandemic. Additionally, participants had known first positive NAAT and serology dates. This permitted inclusion criteria to specify a 2-week period between NAAT and serology testing. Current CDC guidance notes seroconversion may take up to 3 weeks [30]. However, all 8 participants with <21-day interval between the initial positive NAAT and serology testing were seropositive. Combining a cross sectional serology survey with longitudinal SARS-CoV-2 test results to examine reinfection extended the utility of the original serology survey and is a novel approach to assessing protection of seropositivity among previously infected persons.

This study of a large cohort of previously infected persons (based on an initial positive NAAT test)

found that protection from reinfection was associated with seropositivity. A c c e p t e d M a n u s c r i p t 23 A c c e p t e d M a n u s c r i p t 26 Figure 1 

Symptomatic SARS-CoV-2 reinfection of a health care worker in a Belgian nosocomial outbreak despite primary neutralizing antibody response

Temporal profiles of viral load in posterior oropharyngeal saliva samples and serum antibody responses during infection by SARS-CoV-2: an observational cohort study

21-ID-01: Update to the standardized surveillance case definition and national notification for 2019 novel coronavirus disease (COVID-19)

Duration of viral shedding of discharged patients with severe COVID-19

Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study

Duration of SARS-CoV-2 shedding: A population-based, Canadian study

Antibodies to SARS-CoV-2 protect against reinfection during outbreaks in care homes

Antibody status and incidence of SARS-CoV-2 infection in health care workers

Association of SARS-CoV-2 seropositive antibody test with risk of future infection

Protection of healthcare workers against SARS-CoV-2 reinfection

Incidence of SARS-CoV-2 infection according to baseline antibody status in staff and residents of 100 long-term care facilities (VIVALDI): a prospective cohort study

SARS-CoV-2 incidence, transmission and reinfection in a rural and an urban setting: results of the PHIRST-C cohort study

Lack of antibodies to SARS-CoV-2 in a large cohort of previously infected persons

Humoral immune response to SARS-CoV-2 in Iceland

Severe Acute Respiratory Syndrome Coronavirus 2 seropositivity among healthcare personnel in hospitals and nursing homes

Prevalence of SARS-CoV-2 antibodies in first responders and public safety personnel

COVID-19) 2020 interim case definition

SARS-CoV-2 antibody-positivity protects against reinfection for at least seven months with 95% efficacy

Real-world data suggest antibody positivity to SARS-CoV-2 is associated with a decreased risk of future infection

Intrafamilial Exposure to SARS-CoV-2 Associated with Cellular Immune Response without Seroconversion

SARS-CoV-2 seroprevalence among healthcare, first response, and public safety personnel

COVID-19 community transmission among healthcare workers at a tertiary care cardiac center

Epidemiology, clinical characteristics, household transmission, and lethality of severe acute respiratory syndrome coronavirus-2 infection among healthcare workers in Ontario

Coronavirus disease 2019 (COVID-2019) infection among health care workers and implications for prevention measures in a tertiary hospital in Wuhan, China

First experience of COVID-19 screening of health-care workers in England

Hospital-wide SARS-CoV-2 antibody screening in 3056 staff in a tertiary center in Belgium