Persistent SARS-CoV-2 RNA Shedding without Evidence of Infectiousness: A Cohort Study of Individuals with COVID-19

Abstract Background To better understand SARS-CoV-2 shedding duration and infectivity, we estimated SARS-CoV-2 RNA shedding duration, described characteristics associated with viral RNA shedding resolution1, and determined if replication-competent viruses could be recovered ≥10 days after symptom onset among individuals with mild to moderate COVID-19. Methods We collected serial nasopharyngeal specimens at various time points from 109 individuals with rRT-PCR-confirmed COVID-19 in Utah and Wisconsin. We calculated probability of viral RNA shedding resolution using the Kaplan-Meier estimator and evaluated characteristics associated with shedding resolution using Cox proportional hazards regression. We attempted viral culture for 35 rRT-PCR-positive nasopharyngeal specimens collected ≥10 days after symptom onset. Results The likelihood of viral RNA shedding resolution at 10 days after symptom onset was approximately 3%. Time to shedding resolution was shorter among participants aged <18 years (adjusted hazards ratio [aHR]: 3.01; 95% CI: 1.6–5.6) and longer among those aged ≥50 years (aHR: 0.50; 95% CI: 0.3–0.9) compared to participants aged 18–49 years. No replication-competent viruses were recovered. Conclusions Although most patients were positive for SARS-CoV-2 for ≥10 days after symptom onset, our findings suggest that individuals with mild to moderate COVID-19 are unlikely to be infectious ≥10 days after symptom onset.

M a n u s c r i p t 5 Background SARS-CoV-2, the novel virus that causes coronavirus disease 2019 (COVID- 19), has caused a pandemic that has led to business and school closures, restrictions in travel and social gatherings, and strained healthcare systems globally. As of October 16, 2020, approximately 7.9 million cases and 216,000 deaths have been reported in the United States, with almost 367,000 new cases in the last 7 days [1]. With continued community transmission of SARS-CoV-2, it is vital to better understand the duration of viral RNA shedding and the infectious period for individuals with COVID-19 to inform decisions regarding isolation duration, infection prevention and control guidance, reopening businesses, and setting return-to-work and return-to-school policies.
SARS-CoV-2 RNA shedding (detection of viral RNA from specimens) durations of up to 63 days have been reported from China [2][3][4]. Viral RNA shedding has been shown to continue after resolution of symptoms based on samples from the nasopharynx tested by real-time reverse transcription polymerase chain reaction (rRT-PCR) [5]. Although the CDC 2019-nCoV rRT-PCR Diagnostic Panel is a qualitative assay used for diagnostic purposes only, the cycle threshold (Ct) value has been used as an indirect measure of viral load. The Ct value appears to be related to disease progression and possible infectivity [6]. Multiple studies showed higher viral loads around the time of symptom onset and significantly reduced viral loads within a week after symptom onset [7][8][9][10][11]. However, only a few studies investigated the duration of replication-competent SARS-CoV-2 shedding primarily in hospitalized patients [5,[12][13][14]. Based on available literature, shedding of replication-competent virus among patients with mild to moderate illness beyond 10 days after symptom onset is rare [5,12,13,15], but shedding of replication-competent virus can continue up to day 20 after symptom onset in patients with severe illness [14]. Further research is needed to understand A c c e p t e d M a n u s c r i p t 6 the duration of viral RNA shedding and potential infectivity among community-based individuals with mild to moderate COVID-19 to inform public health policies.
Our analysis estimates the time to resolution of SARS-CoV-2 RNA shedding in the nasopharynx, and we report on individual characteristics associated with viral RNA shedding resolution. We also evaluate whether replication-competent SARS-CoV-2 may be recovered from rRT-PCR-positive nasopharyngeal (NP) specimens collected ≥10 days after symptom onset among community-based people with mild to moderate COVID-19.

Methods
From March to May 2020, the Centers for Disease Control and Prevention (CDC) partnered with two local health departments in Wisconsin (Milwaukee City and North Shore) and three county health departments in Utah (Salt Lake, Davis, and Summit) to investigate the transmission of SARS-CoV-2 within local households. SARS-CoV-2 positive individuals (termed 'source participants') identified through diagnostic testing during clinical care and routine public health reporting were eligible for enrollment if they were not currently hospitalized, diagnosed with SARS-CoV-2 infection within 10 days prior to enrollment, and were living with ≥1 other individual. We recruited these source participants through referrals from public health nurses at the respective local health departments, and invited their household contacts to participate in the household transmission investigation as previously described [16,17]. We followed this investigation cohort for a 15-day period (day 0-day 14) to collect clinical and laboratory data, allowing us to estimate the duration of SARS-CoV-2 RNA shedding (detection of viral RNA from NP specimens by rRT-PCR) using symptom onset date, pre-investigation positive test date (if applicable), a day 0 test, and a day 14 test, and any interim tests as potential time points for onset and resolution of viral RNA shedding ( Figure 1).
A c c e p t e d M a n u s c r i p t 7 At enrollment (day 0), questionnaires were administered to all participants for collection of demographic information, underlying medical conditions, exposure history, and detailed symptom information. On day 0, NP specimens were collected from all participants. On day 1 through day 14, all participants maintained a daily symptom log and were asked to report any new or worsening symptoms to the investigation team. If new or worsening symptoms were reported among participants included in the investigation, we returned to the household to collect NP specimens from all household members. On day 14, closeout questionnaires were administered, and NP swabs were collected. To better understand potential infectivity of participants who were rRT-PCR positive beyond 9 days, we performed viral culture on 35 rRT-PCR-positive NP specimens that were collected 10-36 days after symptom onset. Presence of culturable virus was detected by limiting dilution culture in Vero CCL-81 cells as previously described [22,23]. If cytopathic effects were observed, cell monolayers were harvested, and total nucleic acid was extracted for confirmatory testing using rRT-PCR. In order to define a specimen as culture positive, the passage 1 viral isolate had to be positive by rRT-PCR and have a Ct of at least two less than the clinical specimen, indicating more viruses coming out of the cell culture than was put into the cell culture. The follow-up period for the 109 participants ranged from 1-38 days from symptom onset ( Figure 2).

Investigation Measures
Participants were classified into three categories based on their viral RNA shedding duration: persistent, not persistent, or indeterminant. Persistent viral RNA shedding was defined as a positive rRT-PCR test ≥14 days after symptom onset or first positive test. Not persistent viral RNA shedding was defined as a negative rRT-PCR test <14 days after symptom onset or positive rRT-PCR test. Participants classified as indeterminant were those who withdrew after the first day or were observed for <14 days after symptom onset at the conclusion of the investigation, and those whose shedding status could not be determined due to testing interval. Because we did not test participants daily, we could not classify some participants as having persistent viral RNA shedding although they tested negative more than 14 days after A c c e p t e d M a n u s c r i p t 9 symptom onset. This was because some days elapsed between their last positive test and first negative, and it was unclear if they stopped shedding before or after the 14-day cutoff.
We categorized symptom data reported by participants from symptom onset through the last follow-up day as constitutional (fever, chills, myalgia, or fatigue), upper respiratory (runny nose, nasal congestion, or sore throat), lower respiratory (cough, shortness of breath, wheezing, or chest pain), neurologic (headache, loss of taste, or loss of smell), and gastrointestinal (nausea/vomiting, diarrhea, or abdominal pain). This activity was reviewed by CDC and was conducted consistent with applicable federal law and CDC policy (e.g., 45 C.F.R. part 46, 21 C.F.R. part 56; 42 U.S.C. §241(d); 5 U.S.C. §552a; 44 U.S.C. §3501 et seq). Informed consent was obtained from all adult participants and legal guardians for those <18 years. Child assent was also obtained from all children ≥7 years old.

Role of the Funding source
No external funding was received for this study.

Data Analysis
We calculated frequencies and percentages for demographic factors, underlying medical conditions, and symptoms. We used the Kaplan-Meier estimator to calculate probabilities for viral RNA shedding resolution over time. The log-rank test was used to compare viral RNA shedding resolution probabilities over time among categories of demographic and underlying medical condition variables. After confirming that the data met the hazard proportionality assumption in a Cox model (p>0.2 for all covariates) [24], we conducted Cox proportional hazards regression analysis to identify participant characteristics associated with viral RNA shedding resolution. The outcome was viral RNA shedding resolution, defined as first A c c e p t e d M a n u s c r i p t 10 negative rRT-PCR test, and we included age, sex, race/ethnicity, and any underlying medical condition (yes/no) as covariates.
The number of days to resolution of viral shedding was calculated as: (1) the number of days from date of symptom onset to the date of the last positive NP specimen collection for those who withdrew, or were hospitalized after enrollment, or were still shedding at the conclusion of the investigation; or (2)  To account for clustering of participants within households and study sites, we used the robust sandwich estimator to adjust the standard errors of the hazard ratios (HR). [25] A twotailed statistical significance level of 0.05 was set a priori.
A c c e p t e d M a n u s c r i p t 11

Results
Of the 109 participants included in the analysis, 52% were male, 61% were non-Hispanic White, and 15% were non-Hispanic Black. The participants were 3-90 years of age (median: 33; IQR: 21-52 years). Forty-three percent of the participants reported at least one underlying medical condition. Participants reported neurologic (92%), upper respiratory (90%), lower respiratory (87%), constitutional (84%), and gastrointestinal (64%) symptoms. One participant who was classified as not persistent was asymptomatic through resolution (Table   1). One participant was hospitalized for three days and discharged prior to enrollment, and another was hospitalized after enrollment and was still in hospital at the conclusion of the investigation.
A total of 35 rRT-PCR-positive NP specimens (Ct values 26.3-38.4) collected ≥10 days after symptom onset from 34 participants were placed in viral culture to attempt virus isolation.
The characteristics of participants whose specimens were submitted for viral culture were similar to the total sample, except that they were slightly older (47% vs 30% were aged ≥50 years) (Table 1) (Figure 4). The median day from symptom onset (or first positive test, whichever came first) to viral RNA shedding resolution was 21 days (IQR: [18][19][20][21][22][23][24]. The viral RNA shedding resolution over time differed by age groups, with younger participants' shedding resolution occurring sooner than that of older participants (Log rank P<0.001) ( Figure 5).
Resolution of viral RNA shedding occurred sooner in participants aged <18 years (aHR: 3.01; 95% CI: 1.6-5.6) and later in those aged ≥50 years (aHR: 0.50; 95% CI: 0.3-0.9) than those aged 18-49 years, controlling for sex, race/ethnicity, and underlying medical conditions. There was no statistically significant association with other demographic or clinical factors. The unadjusted and adjusted estimates are reported in Table 2.

Discussion
Many participants in our investigation continued to shed viral RNA beyond 10 days after symptom onset. However, no replication-competent virus was isolated ≥10 days after symptom onset, suggesting it was unlikely they would be able to transmit SARS-CoV-2.
These findings support the current guidance that people who are positive for SARS-CoV-2 can be released from home isolation 10 days after symptom onset [26,27].
Our results add to our understanding about SARS-CoV-2 viral shedding among communitybased individuals with mild or moderate disease. We found prolonged viral RNA shedding among individuals with mild disease. We also found that children aged <18 years were more likely to have viral RNA shedding resolution and adults aged ≥50 years were less likely to have resolution of viral RNA shedding during the observation period compared with adults aged 18-49 years.
Our findings are important because our investigation included predominantly nonhospitalized people with mild to moderate COVID-19 and we observed viral RNA shedding periods similar to those reported for hospitalized patients [2,9,13,28,29]. Previous studies A c c e p t e d M a n u s c r i p t 13 reported viral RNA shedding duration of a median days of 20-31 in hospitalized patients [2,3,30]. In our investigation, we found that community-based participants infected with SARS-CoV-2 continued to shed virus a median of 21 days after symptom onset. Researchers in China observed that persons aged >45 years experienced delayed resolution of viral RNA shedding among 59 hospitalized patients [29]. Similarly, we found that participants <50 years ceased shedding earlier than older participants. The differences in viral RNA shedding by age may coincide with the difference in clinical syndromes experienced by younger versus older people. These findings can help with developing guidance for how to address scenarios when people with COVID-19 exhibit persistent positive rRT-PCR test results.
In this investigation, we did not recover any replication-competent virus from the 35 rRT-PCR-positive NP specimens collected 10-36 days after symptom onset. Our group has performed virus isolation for numerous other published [22,23,28] and unpublished studies and have found virus recovery of 71% during days (-4) to 0 pre-symptomatic, 47% days 0-4 post symptom onset, and 30% days 5-9 post symptom onset. These data indicate that an absence of culturable virus is not due to the laboratory method, but the time of specimen collection. Previous studies examined whether replication-competent SARS-CoV-2 could be recovered from specimens collected on different days [5,12,14,22]. One study recovered replication-competent viruses from respiratory specimens 1-9 days (median=4) after onset, [28] and in another study, replication-competent viruses were isolated from 31 of 46 rRT-PCR-positive specimens collected from 6 days before to 9 days after onset of classical symptoms [22]. However, replication-competent virus was isolated up to 12 days (median 4 days) in patients with mild to moderate disease [15] and up to 20 days (median 8 days) after symptom onset for severe and critically ill patients [14]. Further studies are still needed to better characterize shedding of replication-competent virus longitudinally from people with infection to better define the timing of infectivity.
A c c e p t e d M a n u s c r i p t 14 When considering the generalizability of our findings, the results are based on a convenience sample of people with COVID-19. The convenience sample approach is subject to potential selection bias, including selection of health-conscious participants who may be healthier than the general population. Also, the probabilities reported in this study were calculated using the Kaplan Meier estimator with no adjustment for the cluster sampling design. Another consideration is that the estimated duration of viral RNA shedding may be longer than the actual duration of shedding for some participants who had viral RNA shedding resolution which we did not detect because we did not conduct testing daily. On the other hand, by limiting viral RNA shedding resolution to first negative, we might have misclassified intermittent shedding as resolution of viral RNA shedding. Despite these limitations, this study is one of the few to describe viral RNA shedding duration in community-based individuals with mild to moderate illness and to include viral culturing of NP rRT-PCRpositive specimens collected at multiple time points.
Based on our data, community-based individuals with confirmed COVID-19 may shed viral RNA beyond 21 days after symptom onset, particularly older adults. While the majority of people diagnosed with COVID-19 will likely remain rRT-PCR positive after the recommended 10-day isolation period [26,27,31], shedding of replication-competent virus appears to be rare ≥10 after symptom onset. Our findings provide further evidence that people with mild to moderate COVID-19 are unlikely to be infectious ≥10 days after symptom as long as they do not have any risk factors for prolonged infectivity, such as immunocompromising conditions [14]. Our findings support a time-based/symptom-based strategy, versus a test-based approach for decisions about discontinuation of isolation to limit the spread of SARS-CoV-2.       Here, we plot cycle threshold values of the N1 probe against the time elapsed between initial symptom onset and NP specimen. No replication-competent virus was recovered from the 35 specimens collected days 10-36 after symptom onset, submitted for culture.      A c c e p t e d M a n u s c r i p t 33 Figure 5