Daily Rapid Antigen Testing in a University Setting to Inform COVID-19 Isolation Duration Policy

Introduction

In December 2021, the Centers for Disease Control and Prevention (CDC) reduced the recommended COVID-19 isolation period for the general population from 10 to 5 days following the symptom onset or a positive viral test.¹ To end isolation, persons must have resolving symptoms and wear a mask for an additional 5 days; however, a negative exit test was not required. The rationale for the shortened isolation is based on practical and scientific considerations, namely weighing the societal and economic burdens against the diminishing risk of onward transmission as a positive person proceeds through their infection. The CDC revised its guidelines as Omicron rapidly grew to dominance in the United States, increasing from 1% to > 50% of reported sequences over a 2-week period in December 2021.² Early analysis suggests potentially different viral dynamics for Omicron versus Delta, with lower peak viral RNA and shorter clearance periods for Omicron versus Delta, but similar proliferation times and clearance rates.³ As the current recommendations based on estimates for earlier SARS-CoV-2 variants, more data are needed to understand their appropriateness for Omicron.

Additionally, the updated guidance acknowledges the possibility of onward transmission following a 5-day isolation, citing an earlier United Kingdom modeling study estimating that 31% of persons remain infectious on day 5.⁴ Emerging literature on exit testing from an Omicron-dominant period further indicates that high proportions of persons remain potentially infectious beyond day 5. Studies of managed isolation programs through schools or employers found positivity via rapid antigen test (RAT) on days 5-9 ranging from 31-58%, although daily testing among all persons was not conducted.^5–7 A near-daily PCR testing study reported a day 5 positivity of ranging from 39-52%.³ However, key gaps in our understanding persist. Firstly, some studies did not test all persons daily from day 5 and are right-censored (i.e. some persons did not test negative). In addition, non-mandatory testing raised the possibility of selection bias, with sicker persons potentially opting not to exit test. Lastly, most studies did not adjust for the approximate time of infection.

While PCR tests are a preferred initial diagnostic option due to their high sensitivity, RATs are more suitable for exit testing when the goal is to determine when a person is likely no longer infectious. High PCR sensitivity may result in positive tests beyond the infectious period, leading to unnecessarily long isolations.^8,9 RAT positivity is generally associated with culturable virus, which itself is often a proxy for infectiousness.^9–11 To investigate concerns that RATs may have inferior performance for Omicron versus Delta infections, a study compared same-day positivity between the variants, finding similar sensitivity of RAT and PCR tests.¹² Lastly, RATs have the advantage of relative affordability, fast turnaround time, and at-home self-administration compared to PCR, making them the only viable exit test option for much of the population.^13,14

To address the evidence gaps regarding optimal isolation durations and how SARS-CoV-2-positive persons can best utilize exit RATs when resources are limited, we evaluated daily RAT data from 324 persons in a university-managed isolation program who initially tested positive for SARS-CoV-2 between January 1 and February 11, 2022. We posed two questions: 1) what percentage of SARS-CoV-2-positive persons remained positive via RAT on day 5 of isolation and each subsequent day until testing negative and 2) what factors were associated with isolation duration?

Methods

The university required undergraduate students to screen upon campus arrival and subsequently twice-weekly on designated days. SARS-CoV-2 positive students isolated and participated in mandatory daily rapid antigen self-testing beginning on day 5 following diagnosis (the CDC’s recommended end of isolation¹) until they tested negative. Diagnosis (day 0) was defined as the earliest positive or inconclusive test date. All inconclusive persons subsequently tested positive. Trained staff observed the exit testing process and confirmed the result. Upon testing negative, students ended isolation but continued mandatory masking until day 10. All testing was conducted using the Quidel QuickVue At-Home OTC COVID-19 Test, a lateral flow immunoassay that qualitatively detects the SARS-CoV-2 nucleocapsid protein antigen.¹⁵ The test received a FDA Emergency Use Authorization for prescribed home use with self-collected anterior nares swab specimens and has a sensitivity of 84.8% (95% CI 71.8-92.4) and a specificity of 99.1% (95% CI 95.2-99.8).

We used RStudio v1.4.1106 for our analyses.¹⁶ We calculated the percent still positive as the number of positive persons each day divided by the total number of positive persons. To assess prognostic factors associated with the time to event (i.e. testing negative), we coded an accelerated failure time (AFT) lognormal regression model using the survival R package v3.2-13.^17,18 We selected the AFT model due to its suitability for interval-censored data.¹⁹ Since students enter the study on day 0 but are not rapid tested until day 5, any persons testing negative on day 5 are interval-censored, with their true negative time falling between day 1 and day 5. We compared model fits using various distributions and selected the fit resulting in the lowest Akaike Information Criterion value. We exponentiated the regression coefficients to calculate the event-time-ratio (ETR), which are associated with prolonged isolation duration when > 1 and decreased duration when < 1. We checked the assumption that the ratio of survival times (i.e. the event-time-ratio) is constant for all fixed probabilities of S(t), the survival function, by a visual inspection of QQ plots generated for each covariate level comparison using the R package AFTtools v0.2.1.²⁰

Results

Our study population comprised primarily 18-22 year old students living in university dormitory housing (N=323, Table 1). Among them, 63% self-reported symptoms prior to or at diagnosis. We did not track symptoms beyond diagnosis, although 18/205 symptomatic persons had a symptom onset date 1 day post-diagnosis, potentially reflecting when they received their results and discussed symptoms. We found that 7% had a confirmed SARS-CoV-2 infection > 90 days before their recent diagnosis. The university did not screen asymptomatic persons with an infection ≤ 90 days due to the false positive risk. We categorized vaccinations into 1-4 doses. Generally, a non-mRNA vaccine primary series counted as 1 dose, a mRNA vaccine primary series as 2 doses, and a booster as an additional dose. Two individuals reported receiving 2 boosters, resulting in 4 doses. Only doses administered ≤ 14 days prior to diagnosis were counted toward the total.²¹ The breakdown of doses was as follows: 3% of persons had 1 dose, 27% 2 doses, 68% 3 doses, 1% 4 doses, and 2% of persons had missing data. eTable 1 presents detailed vaccination history. Isolation duration is dependent on where a person is in their infection course when they are diagnosed. To address this, we used the time since the last negative test as an approximation of the time since infection, with 56% of persons testing negative ≤ 4 days prior to diagnosis, 15% 5-9 days prior, and 29% ≥ 10 days prior. 1 person had missing data. The ≤ 4 days group represents students compliant with university twice-weekly screening policy, the 5-9 day group a mix of non-compliant routine screeners and arrival screeners, and the ≥ 10 day group arrival screeners.

To calculate the percent still positive on day 5+, we dropped 1 person with an unknown last negative test time and 7 persons who initially tested inconclusive but used the subsequent positive test date as the isolation start, resulting in a final dataset of 315 persons. Among twice-weekly screeners, 47% of all diagnosed (N=177) remained positive on day 5, 22% on day 6, 8% on day 7, and 1-2% on days 8-13 (Figure 1A). Among students last testing negative 5-9 days before diagnosis, 28% of all diagnosed (N=47) remained positive on day 5, 17% on day 6, 6% on day 7, and 2-4% on days 8-9 (Figure 1B). Students last testing negative ≥10 days prior (N=91) experienced similar daily positivity (Figure 1C). A density plot of these data shows isolation duration distributions for all groups centered on day 5, with relatively larger peaks for the ≤ 4 day category on days 6+ (Figure 1D).

• Download figure
• Open in new tab

Figure 1: Rapid antigen testing results by isolation day and isolation duration by days since the last negative test category.

Percent still positive of the original study population on the left axis and the number of positive persons on each isolation day on the right axis. 1 person was removed due to missing last negative test data and 10 persons were removed due to testing inconclusive initially but counted the first positive test as day 0. Last negative test A) ≤ 4 days (N=177), B) 5-9 days (N=47), and C) ≥ 10 days prior to the earliest test (inconclusive or positive) (N=91). D) Isolation duration (time from earliest test to test negative) by time since last negative test category. The vertical line indicates day 5, the first day of rapid antigen testing.

To evaluate possible prognostic variables for isolation duration, we conducted a survival analysis using an AFT lognormal regression model. We subset the dataset used in Figure 1 to exclude those with 1 (N=8), 4 (N=2), or an unknown (N=6) number of vaccine doses due to small category sizes, a missing PCR cycle threshold (CT) value at diagnosis due to an external test or RAT (N=27), a missing symptom status (N=2), and those who received an international vaccine (N=8), resulting in a final sample of N=263 persons. We included time since the last negative test category as a covariate to account for possible confounding as persons in different infection stages would necessarily experience differing isolation durations. We also included symptom status, PCR CT value, and prior infection > 90 days as covariates. We created a new variable combining the number of vaccine doses (2 or 3) and the time since the last dose (< 5 months or ≥ 5 months).²² All “3 vaccine dose” persons except 1 received their last dose < 5 months. Finally, we included the primary series vaccine brand grouped into mRNA vaccines (Pfizer and Moderna) and J&J. Table 2 presents the regression results and eFigure 1 displays the isolation duration distribution for each covariate category excluding the time since the last negative test, which is displayed in Figure 1. We found that having a last negative test ≥ 10 days prior significantly decreased isolation duration by 15% (ETR 0.85 (95% CI 0.75-0.96)) compared to having a last negative test ≤ 4 days prior. Being symptomatic significantly increased isolation duration by 13% (ETR 1.13 (95% CI 1.02-1.25)). Having a prior infection > 90 days significantly decreased isolation duration by 50% (ETR 0.50 (95% CI 0.33-0.76)). Receipt of 3 vaccine doses significantly increased isolation duration by 20% (ETR 1.20 (95% CI 1.04-1.39)) compared to the 2 doses ≥ 5 months group. The other covariates were not significant.

Discussion

We analyzed data from a mandatory, daily RAT program among university students in isolation to assess the percent still positive on day 5+ and determine possible prognostic factors for isolation duration. In addition, we approximately accounted for time since infection by stratifying our analysis by the time since the last negative test. We found a day 5 positivity of 47% in the twice-weekly screening group and 26-28% in the other less frequently screened groups (Figures 1A-C). For all groups, positivity approximately halved with each additional day. These results align with the expectation that more frequent screeners are diagnosed earlier in their infection, thus experiencing a longer isolation. Our findings are similar to results reported in other analyses, although most did not conduct daily mandatory testing. Studies of managed isolation programs reported RAT positivity ranging from 31-58% on days 5-9 of isolation and PCR positivity ranging from 39-52% on day 5, 25-33% on day 6, and 13-22% on day 7.^3,5–7

Our results indicate that a 7-day isolation period is required to achieve at least a 90% probability of being negative without exit testing (Figure 1A-C). 5-day and 6-day isolation periods are associated, respectively, with a 53-74% and 78-85% probability of testing negative (range across last negative test categories). In cases where only a single RAT per person is available, the ideal exit test timing should be informed by the risk of onward transmission. For a person living or working in a higher-risk environment due to density, vulnerable populations, or both (e.g., university dormitories or long-term care facilities), onward transmission is potentially riskier and costlier and a day 7 exit test may be most appropriate. Persons in lower-risk environments could consider using a single RAT on day 5 and, if positive, remain in isolation until day 7. Persons with multiple RATs could begin daily exit testing on day 5. While we defined the isolation start as the initial test, CDC guidelines define it as the initial test or the symptom onset. 63% of our study population reported symptoms before or at diagnosis (Table 1). Symptomatic persons reported symptom onset a mean of 0.88 days before their initial test (IQR 0-1.25 days) in the last negative test ≤ 4 days group, 0.93 (IQR 0-1.25) in the 5-9 days group, and 3.04 days (IQR 0-4 days) in ≥ 10 days group. Persons using symptom onset as their isolation start accordingly may have longer isolation durations than those we measured in our study.

In our survival analysis, having a negative test ≥ 10 days, symptom status, prior infection > 90 days, and receipt of 3 vaccine doses were significantly associated with isolation duration (Table 2). Other covariates were not significant. For the last negative test covariate, we observed an association with shorter duration time for the ≥ 10 days and 5-9 days groups compared to the ≤ 4 days group, although only the ETR for the ≥ 10 days group was significant. The relationship between less frequent screening and shorter isolation duration is intuitive as these persons are more likely to be diagnosed later in their infection. Reporting symptoms significantly increased isolation duration. Symptomatic persons may be diagnosed earlier in their infection, even in the case of routine screening, resulting in a longer isolation period. Experiencing a prior infection > 90 days significantly decreased isolation duration. In a highly vaccinated population, having a previous infection may confer greater immunity relative to those without ²³, reducing the isolation duration. Receipt of 3 vaccine doses was significantly associated with a longer isolation duration compared to receipt of 2 doses ≥ 5 months prior to diagnosis, an unexpected direction of effect. This finding was consistent under various formulations of the model during our exploratory phase. Another study found that vaccine boosted persons were twice as likely to test positive on an initial RAT on days 5-10 compared to unboosted persons, although not all persons tested daily.⁵ The authors suggested that boosted persons may develop symptoms earlier, leading to speedier detection and, resultantly, longer isolation durations. Accounting for the time since the last negative test in our model would likely reduce some of the bias toward earlier detection of symptomatic individuals, however this explanation remains possible. In addition, more persons in the 2 dose groups may have been infected with Delta compared to those in the 3 dose group. We observed a higher proportion of persons belonging to the 2 dose groups earlier in our study, when Delta still circulated at low levels (eFigure 2). If the incubation period or infection duration differ between Delta and Omicron infections, this could contribute to our findings. However, Omicron reached 97% frequency among sequenced samples in New Haven County by January 1, 2022 (with the remaining 3% composed of Delta).²⁴ We observed a substantially larger sample size for the 3 dose group (N=188) compared to the 2 dose ≥ 5 (N=44) and < 5 (N=31) month groups. The larger sample may have captured more isolation duration outliers. Finally, our analysis assesses the relationship between these factors and duration of isolation, not infection. There may be other unaccounted factors associated with both the 3 dose group and isolation duration.

Conclusion

A high percentage (26-47%) of university students during an Omicron-dominant period remained positive after the currently recommended isolation period. Thus, the currently recommended 5-day isolation period may be too short, especially for persons using symptom onset as their isolation start or those diagnosed early in their infections. However, it is important to note that these factors are associated with duration of isolation, not infection, and thus unmeasured factors associated with diagnosis timing could affect our findings. Future research assessing daily RAT positivity beginning at the isolation start in the general population would further refine our understanding of optimal isolation periods, as would analyses characterizing what, if any, onward transmission has resulted from the current 5-day isolation period. Factors such as social distancing and masking may reduce transmission risk posed by RAT-positive persons. These considerations illustrate the complexity of recommending isolation durations for the general population, but our study adds to growing evidence that the current 5-day isolation period may be 1-2 days too short to sufficiently reduce the risk of onward transmission, especially in dense settings or ones with vulnerable populations.

Author Contributions

Conceptualization, R.E., C.Chen, N.D.G., M.S.W

Collected and provided isolation data: C.Chen, M.S.W., Yale COVID-19 Resulting and Isolation Team

Methods development, R.E., C.Chaguza

Data analysis and interpretation, R.E., C.Chaguza, N.D.G.

Supervision, N.D.G., M.S.W

Writing - Original Draft, R.E., N.D.G;

Writing - Review & Editing, all authors.

Yale COVID-19 Resulting and Isolation Team Authors

Adam Cairns, Alyssa Cooksey, Arielle Pensiero, Asnath Mosha, Christina Pivirotto, Colleen Traub, Gabriel Velazquez, Heejun Lee, Jasper Larioza, Jelissa Neal, Jenna Bourgeois, Kathleen Donnelly, Karen Otterson, Kiley Carbone, Kristin Smith, Lauren Gillingham, Lauren Greenberg, Linda DiGangi Ehrenfels, Maen Adileh, Marie Toupou, Nathan Lubich, Nathan Yuen, Nick Davies, Nikkiyah Brown, Noor Khalid, Olivia Dumont, Onyi Umeugo, Oscar Monteagudo, Presley Hill, Quinn O’Leary, Rashea Banks, Richard Rousseau, Sam Duplantis, Sara Khalid, Saskia César, Sean Doran, Shannan Charney, Shinelle Wilkins, Sofia Martinez, Spencer Volpe, Victor Martinez Garcia, Vittoria Cappuccia

Declarations of Interests

NDG is a paid consultant for Tempus Labs and the National Basketball Association and has received speaking fees from Goldman Sachs. The remaining authors declare no competing interests.

Ethics

The Institutional Review Board from the Yale University Human Research Protection Program determined that the use of information, including information about biospecimens, is recorded by the investigator in such a manner that the identity of the human subjects cannot readily be ascertained directly or through identifiers linked to the subject and thus is exempt from IRB review of human subjects research (IRB Protocol ID: 2000032111).

Data and code availability

All original code and data have been deposited at Github and are publicly available (https://github.com/rebecca-earnest/2022_paper_isolation-rapid-antigen). Any additional information required to reanalyze the data reported in this paper is available from the corresponding author upon request.

Supplementary Information

• Download figure
• Open in new tab

eFigure 1: Relationship between model covariates and isolation duration, related to Table 2.

N=263 persons included in the survival model analysis (Table 2). Isolation duration measured as the number of days from testing positive or inconclusive to testing negative.The vertical line indicates day 5, the first day of rapid antigen testing. We show isolation duration by A) symptom status, (B) prior infection > 90 days, C) primary vaccine brand, (D) the number of and time since the last vaccine dose, and (E) CT value at diagnosis. For (E), note the inverted y-axis as lower CT values correspond to higher viral loads. We present the mean and 95% confidence interval.

• Download figure
• Open in new tab

eFigure 2: Number of and time since the last vaccine dose by diagnosis date, related to Table 2.

N=263 persons included in the survival model analysis (Table 2). The diagnosis date is the date of the first positive or inconclusive test. (A) Number of persons in each number and timing of vaccine dose category over time by diagnosis date. (B) Smoothed proportion of total daily diagnosed persons in (A) belong to each number and timing of vaccine dose category over time.