Viral Dynamics of the SARS-CoV-2 Omicron Variant in Pediatric Patients: A Prospective Cohort Study

Abstract

Background

There are limited data on the viral dynamics of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) in children. Understanding viral load changes over the course of illness and duration of viral shedding may provide insight into transmission dynamics to inform public health and infection-control decisions.

Methods

We conducted a prospective cohort study of children aged 18 years and younger with polymerase chain reaction–confirmed SARS-CoV-2 between 1 February 2022 and 14 March 2022. SARS-CoV-2 testing occurred on daily samples for 10 days; a subset of participants completed daily rapid antigen tests (RATs). Viral RNA trajectories were described in relation to symptom onset and resolution. The associations between both time since symptom onset/resolution and non-infectious viral load were evaluated using a Cox proportional hazards model.

Results

Among 101 children aged 2 to 17 years, the median time to study-defined non-infectious viral load was 5 days post–symptom onset, with 75% meeting this threshold by 7 days and 90% by 10 days. On the day of and day after symptom resolution, 43 (49%) and 52 (60%) of 87 had met the non-infectious thresholds, respectively. Of the 50 participants completing a RAT, positivity at symptom onset and on the day after symptom onset was 67% (16/24) and 75% (14/20). On the first day where the non-infectious threshold was met, 61% (n = 27/44) of participant RAT results were positive.

Conclusions

Children often met the study-defined non-infectiousness threshold on the day after symptom resolution. The RATs were often negative early in the course of illness and should not be relied on to exclude infection.

Clinical Trials Registration. clinicaltrials.org; NCT05240183.

The transmission of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is influenced by the environment, virus, and host factors and by host–virus interactions [1, 2]. One important host–virus factor affecting transmission dynamics is the viral load (VL) present in the upper respiratory tract and the duration of viral shedding. The infectious period for an individual is correlated with increased VLs and increased duration of viral shedding, thus creating the opportunity for increased transmission [3, 4].

Host factors that may impact the VL and duration of shedding include disease severity, presence of underlying medical conditions (eg, immunocompromise), and prior immunity through infection or immunization [5]. The impact of age has not been fully elucidated, with most studies assessing the VL at only 1 time point and several studies not controlling for time since symptom onset [6–9]. One study evaluated individuals with multiple respiratory samples tested for SARS-CoV-2 from 1 March 2020 to 31 January 2022 and found that children younger than 10 years had a significantly shorter time to negative polymerase chain reaction (PCR) results compared with adults [10]. However, this study had no data on symptom onset and included predominantly wild-type SARS-CoV-2 infections.

Understanding the duration of time an individual may be infectious is an important consideration when determining public health interventions, such as duration of case isolation, in order to mitigate further spread of infection and quantify the period of communicability. While VLs and duration of shedding have been evaluated early during the pandemic and described in adults with Omicron [11–14], to our knowledge there have been no studies in children.

The aim of this study was therefore to describe the viral dynamics of the SARS-CoV-2 Omicron variant in children aged 18 years and younger using daily SARS-CoV-2 testing following a positive SARS-CoV-2 test. Factors that may alter viral shedding were captured to determine their influence on viral dynamics.

METHODS

Study Design

We conducted a prospective cohort study of children aged 18 years and younger with PCR-confirmed SARS-CoV-2 infection between 1 February 2022 and 14 March 2022. Ethical approval was obtained from the Hospital for Sick Children's Research Ethics Board (REB 1000079263). The study was registered with clinicaltrials.org (NCT05240183). The protocol is available with the full text of this article online. Written informed consent of parents/caregivers or children and verbal assent of all children when parents/caregivers provided consent were obtained for participants.

Study Setting

The study was conducted at the Hospital for Sick Children, Toronto, Canada. Potential participants were identified by a positive SARS-CoV-2 PCR test result in the microbiology laboratory, which supports testing of inpatients and outpatients, and the SickKids COVID-19 Testing center (CTC), a school-based coronavirus disease 2019 (COVID-19) testing program (see the Supplementary Appendix).

Study Participants

Participants were eligible for study inclusion if they were 18 years or younger and had acute SARS-CoV-2 infection, defined by a positive molecular test and reported onset of symptoms within 72 hours or, if asymptomatic, a known household exposure where the index case was clear. Participants were ineligible if they had symptoms for more than 72 hours or if they were asymptomatic at the time of enrollment and had no known exposure (ie, the baseline for infection could not be determined).

Study Procedures

Following enrollment, participants were given 10 saliva self-collection containers (Spectrum DNA-1000; Spectrum Solutions, USA) to provide daily samples for 10 days. Participants unable to provide saliva had SARS-CoV-2 testing completed by a healthcare provider using flocked nasopharyngeal swabs (Copan Italia, Brescia, Italy). Participants who consented to the optional rapid antigen sub-study were provided with Rapid Response COVID-19 Antigen Test kits (BTNX, Canada), containing 10 tests for self- or caregiver collection to be performed daily (simultaneously to the specimen collection for PCR) according to the manufacturer’s instructions, with the exception that an oral collection was followed by nasal collection as recommended in Ontario at the time of the study [15].

All participants underwent a baseline questionnaire collecting variables of interest, including age, underlying medical conditions, vaccination status (number and timing), history of SARS-CoV-2 infection, and exposure (household vs non-household exposure). Participants tracked the time of specimen collection, rapid antigen test (RAT) result (positive or negative), if applicable, and presence of symptoms in a daily written log (see the Supplementary Appendix).

Specimen Testing

Specimens were tested at SickKids for SARS-CoV-2. Primary specimens were tested by 1 of 3 assays, including the following: (1) Allplex 2019-nCoV assay (Seegene, South Korea), (2) US Centers for Disease Control and Prevention (CDC) 2019-nCoV_N1 and 2019-nCoV_N2 (https://www.fda.gov/media/134922/download) https://www.cdc.gov/coronavirus/2019-ncov/lab/rt-pcr-panel-primer-probes.html), or (3) the Amplitude platform utilizing the TaqPath assay (ThermoFisher, USA). For each assay, a standard curve was created using a 10-fold serial dilution series of a known concentration of the Twist Synthetic SARS-CoV-2 RNA Control 1 (Twist Bioscience, USA).

After enrollment, all samples underwent SARS-CoV-2 PCR testing using the Allplex 2019-nCoV assay. An endogenous human-gene target was used as the internal control to assess specimen collection quality. SARS-CoV-2 variant status was determined utilizing the Allplex SARS-CoV-2 Variants I assay (Seegene, South Korea), which detects E484K, del69–70, N501Y. Given the SARS-CoV-2 variant prevalence at the time of specimen collection, the detection of del69–70 and N501Y was indicative of Omicron (B.1.1.529) BA.1, while N501Y detection without del69–70 indicated BA.2.

Outcome

The primary outcome was the SARS-CoV-2 VL, expressed in log₁₀ RNA copies/mL, which was determined using the standard curve of the cycle threshold (Ct) N-gene (CTN) result from the Seegene Allplex 2019-nCoV assay [16, 17]. Given that specimens were self-collected, each sample was assessed for quality using the internal control; samples with Ct values greater than 30 on the internal control were deemed of low quality and excluded from the analysis (Supplementary Figures 1 and 2).

Statistical Analysis

We used scatterplots to examine the relationship between SARS-CoV-2 VL and days since symptom onset and resolution, with a smoothed locally estimated scatterplot smoothing curve to demonstrate the trend over time.

We had originally intended to describe the VL trajectories in relation to peak VL [18]. However, the first sample obtained was the highest VL with subsequent values decreasing and no earlier data points to truly identify the peak VL for the majority of patients (Supplementary Figure 3). For this reason, we described the viral RNA trajectories in relation to symptom onset.

Kaplan-Meier curves and Cox proportional hazards models were used to assess the relationship between time to non-infectious VL and days since symptom onset. A VL of 10⁵ RNA copies/mL or greater (corresponding to a CTN of 31.85 based on the standard curve) was used to define infectiousness for the primary analysis. This VL has been demonstrated to correlate with culturable virus and was thus used as a surrogate for infectiousness [14, 16]. Time to non-infectious VL was defined as time from symptom onset (or first positive molecular test if symptoms were absent) to first VL below this threshold, with no subsequent contradicting value (within 1 log VL). Sensitivity analyses were performed using a threshold of 10⁶ RNA copies/mL or greater (CTN of 28.55). A Cox proportional hazards model was used to estimate adjusted hazard ratios for variables of interest, including age, sex, household exposure, vaccination status, and Omicron subvariant. We defined the number of vaccine doses by the self-reported number of doses that had been received at least 7 days before the positive test result [19–21]. For the antigen RAT sub-study, we calculated the percentage of tests positive on each day after symptom onset and symptom resolution. All analyses were performed with R, release 4.0.5 (R Foundation for Statistical Computing, Austria).

RESULTS

From 1 February 2022 to 14 March 2022, 139 children were assessed for eligibility, 107 provided consent and assent as appropriate, and 101 participants had data that could be included in the analysis (Figure 1). The majority of children were recruited from the outpatient testing program (n = 95, 94%), were symptomatic at the time of enrollment (n = 97, 97%), and provided saliva samples (n = 99, 99%). The remaining 4 participants were tested because of a known exposure, and 2 subsequently developed symptoms. A total of 66 participants (65%) were positive for BA.1 and 33 (33%) were positive for BA.2. The mean age of participants was 10.2 years (standard deviation, 3.8 y), 43 were female (43%), and 19 had underlying medical conditions (19%), none of which were considered to compromise the immune system, and no patients were on immunosuppressive medications (Table 1). Several participants had received 2 or more SARS-CoV-2 vaccine doses (n = 64, 63%), with a smaller proportion being unvaccinated (n = 18, 18%), having received 1 vaccine dose (n = 11, 11%), or having received 3 vaccine doses (n = 8, 8%). All those who were vaccinated received either the original, monovalent pediatric Pfizer-BioNTech COVID-19 vaccine (ages 5–11) or the original, monovalent Pfizer-BioNTech COVID-19 vaccine (ages 12 and up). The median number of days since the last vaccine dose was 56 (interquartile range [IQR], 34–170).

Clinical Symptoms

Of the 101 participants, 99 (98%) had symptoms over the course of the study. The median time to symptom resolution was 6 (IQR, 4–8) days. Twelve participants (12%) did not have symptom resolution before the end of the follow-up. There was no difference in time to symptom resolution by Omicron subvariant BA.1 compared with BA.2 (P = .3) or sex (P = .2). The most common presenting symptoms included sore throat (n = 56, 55%), fever (n = 41, 41%), fatigue/muscle aches (n = 22, 22%), and headache (n = 18, 18%). At the end of the 10-day follow-up the most common persistent symptoms were cough (n = 9, 9%) and runny nose (n = 12, 12%). The frequency of each symptom over the course of the 10-day study period is shown in Supplementary Figure 4.

After the threshold of non-infectiousness was met, the most common persistent symptoms were runny nose (n = 41, 41%), cough (n = 28, 28%), and sore throat (n = 11, 11%).

Viral Load Trajectories

A scatterplot of the VLs of each participant on each day since symptom onset (or positive test date) is shown in Figure 2A and each day since symptom resolution in Figure 2B. Individual participants’ viral trajectories are provided in Supplementary Figure 3.

Time to Non-infectious Viral Load

The median duration to non-infectious VL was 5 days post–symptom onset (Figure 3A); 75% of participants met the non-infectious threshold by 7 days and 90% by 10 days post–symptom onset. Ten participants (10%) still had an infectious VL at 10 days, only 1 of whom was symptomatic with cough. Sensitivity analyses performed using a threshold of less than 10⁶ RNA copies/mL to define non-infectious VL showed a median duration of 3 days, with 75% of participants reaching this threshold by 4 days and 90% by 8 days.

On the day of symptom resolution, 43 of 87 (49%) had met the threshold of non-infectiousness, with 52 (60%) having reached the threshold as of the first day post–symptom resolution (Figure 3B). Sensitivity analyses performed using a non-infectious threshold of less than 10⁶ RNA copies/mL showed 83% of participants reaching the non-infectious threshold on the day of symptom resolution and 91% by the day after symptom resolution.

In the multivariable model, there was no association between age, sex, vaccination status, Omicron sublineage, or household exposure and time to non-infectious VL (Table 1). In the sensitivity analysis, older participants reached the threshold of non-infectiousness later than younger participants (adjusted hazard ratio: 0.92; 95% confidence interval: .86, .99) (Supplementary Table 1).

Rapid Antigen Test Results

There were 50 children enrolled in the RAT portion of the study (Supplementary Figure 5 and Supplementary Table 2). Rapid antigen test positivity at symptom onset and on the day after symptom onset was 67% (16/24 participants who completed RATs) and 75% (14/20 participants who completed RATs), respectively (Figure 4A). The percentage of positive RATs peaked on day 3 after symptom onset at 90% (30/40). The percent positive on day 6 after symptom onset was 49% (20/41). All participants who completed RATs were negative 10 days after symptom onset. The log₁₀ RNA viral copies/mL for saliva specimens collected across time were higher in those participants with a positive as compared with those with a negative RAT (Supplementary Figure 6). There was no observed difference in RAT performance by sublineage or vaccination status (Supplementary Figures 7 and 8).

During the 10-day follow-up period, 74% (37/50 participants) of participants had complete resolution of symptoms; of these, 33% (12/36 participants that completed a RAT on the day of symptom resolution) continued to have a positive RAT on the first day where all symptoms had resolved (Figure 4B). Rapid antigen test positivity decreased daily after symptom resolution; all participants who completed RATs (14/14 participants) were negative 5 days after symptom resolution.

On the first day where the VL reached the non-infectious threshold, 61% (n = 27/44) of participant RAT results were still positive (Figure 4C). Despite PCR testing indicating a VL consistent with a lack of culturable virus, RATs remained positive for some participants (2/11) for up to 9 days post–infectious period and 4 days post–symptom resolution.

DISCUSSION

In this prospective cohort study of predominantly symptomatic children with SARS-CoV-2 infection with Omicron BA.1 and BA.2 variants, we found the median time to a study-defined non-infectious VL was 5 days after symptom onset, with 49% of participants meeting this threshold on the day of symptom resolution. There was no association between age, sex, self-reported vaccination status, Omicron sublineage, or household exposure and time to non-infectious VL.

To our knowledge, this is the first prospective study of viral dynamics of SARS-CoV-2 in children and the first to describe the period of infectiousness of children using VL as a surrogate for infectiousness. Studies of viral dynamics in adults have reported varying infectivity results. Keske et al [13] examined the duration of viral shedding of Omicron BA.1 among 53 symptomatic SARS-CoV-2–positive healthcare workers (98% vaccinated) and found high rates of viral culture positivity of 83% on day 5 post–symptom onset. Conversely, Bouton et al [11] described 92 predominantly symptomatic, vaccinated outpatients (18.5% Delta, 81.5% Omicron) at a university campus and found only 17% had positive cultures beyond 5 days from symptom onset. The range in findings may be related to challenges with viral culture. However, our finding that 50% of children met our definition of study-defined non-infectious VL at 5 days post–symptom onset is consistent with the findings of 2 adult studies [12, 14], suggesting that viral shedding between adults and children may be similar. Boucau et al [14] found that approximately 50% of newly diagnosed outpatients had replication-competent, culturable virus at day 5 (n = 56, 66% Delta, 34% BA.1, 18% unvaccinated). Luna-Muschi et al [12] evaluated viral culture of respiratory samples in 24 vaccinated healthcare workers with mild Omicron and found that viral culture was positive in 46% at day 5.

Our findings have important public health and hospital infection-prevention and -control implications. One in 4 children had not reached the threshold for study-defined non-infectiousness by 7 days post–symptom onset and 10% of children had not reached the threshold by 10 days, suggesting that some children who are infected with the Omicron variant shed potentially culturable virus beyond 5 days of symptom onset. While the replication-competent virus is only 1 component in transmission risk, these findings support the consideration for infection-prevention and -control interventions for up to 10 days post–symptom onset to reduce residual transmission risk around vulnerable or immunocompromised populations.

While most studies have focused on time since symptom onset to describe duration of infectivity, our study is unique in also assessing potential infectivity after symptom resolution. We found that 60% of participants were likely non-infectious on the day after symptom resolution and this increased to 91% when the less conservative threshold to define infectious VL was used in sensitivity analysis. These findings, together with risk tolerance, support the consideration of symptom resolution as a criterion for return to regular activities in low-risk settings (similar to guidance for other respiratory viral infections). For those individuals who do not have symptom resolution, we also noted that ongoing symptoms, such as cough, beyond a time threshold were not predictive of infectiousness. As such, time-based clearance may be considered in individuals with persistent symptoms, as all participants had VLs that met the threshold of study-defined non-infectiousness at the end of the 10-day study period.

In the RAT sub-study, we found that RATs in children lacked sensitivity early on in the course of infection, with only 67% testing positive on the day of symptom onset, increasing to 75% on the day after symptom onset. The percent positivity greatly decreased after 5 days post–symptom onset and, overall, our findings are similar to the adult literature suggesting 30–50% positive RATs between 5 and 10 days of illness [11, 22, 23]. We also found that, for several participants, the RAT remained positive after symptom resolution and when infectiousness was unlikely. Together, this suggests limited utility for early use of RATs in the detection of SARS-CoV-2 in symptomatic children and reinforces the need for a repeated test after a few days of symptoms to confirm infection, if it is being used, and questions the value of using RATs as a test to release from isolation in this population.

Study limitations include that most participants had symptomatic SARS-CoV-2 infection and therefore these findings may not be applicable to children with asymptomatic infections. In addition, many participants had received at least a single dose of a SARS-CoV-2 vaccine. Participants were also otherwise healthy children, limiting the generalizability to children with immunocompromising conditions or medications. The use of quantitative PCR methods to infer infectiousness, rather than performing viral culture to demonstrate the presence of potential transmissible virus, is a further limitation. While both PCR VL and viral culture are used to assess transmission, no agreed-upon threshold exists for either method. However, similar comparisons between VL and transmission using the same framework have been made previously [5, 24]. Furthermore, infectiousness is not binary but on a spectrum that decreases over time. Finally, we only used a single type of RAT in our sub-study, but this assay has been shown to successfully detect the Omicron variant [25].

In conclusion, we found that the median time to presumably non-infectious VL was 5 days after symptom onset, with 60–90% of children meeting the threshold of study-defined non-infectiousness on the day after symptom resolution. As some children shed the virus at levels that are likely to be associated with infectious virus beyond 5 days after symptom onset and after symptom resolution, maintaining infection-prevention interventions for up to 10 days may be considered to reduce any residual transmission risk in higher-risk scenarios.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

References

Author notes