Understanding Viral Shedding of SARS-CoV-2: Review of Current Literature.

OBJECTIVE
Transmission of SARS-CoV-2 has significant implications for hospital infection prevention and control, discharge management, and public health. We reviewed available literature to reach an evidenced-based consensus on the expected duration of viral shedding.


DESIGN
We queried four scholarly repositories/search engines for studies reporting SARS-CoV-2 viral shedding dynamics by PCR and/or culture available through September 8, 2020. We calculated the pooled median duration of viral RNA shedding from respiratory and fecal sources.


RESULTS
Seventy-seven studies on SARS-CoV-2 were included. All studies reported PCR-based testing and 12 also included viral culture data. The overall pooled median duration of RNA shedding from respiratory sources was 18.4 days (95% CI: 15.5 days - 21.3 days; I2=98.87%, p<0.01) among 28 studies. When stratified by disease severity, the pooled median duration of viral RNA shedding from respiratory sources was 19.8 days (95% CI: 16.2 days - 23.5 days; I2=96.42%, p<0.01) among severely ill patients and 17.2 days (95% CI: 14.0 days - 20.5 days; I2=95.64%, p<0.01) in mild/moderate illness. Viral RNA was detected up to 92 days after symptom onset. Viable virus was isolated by culture from -6 days to 20 days relative to symptom onset.


CONCLUSIONS
SARS-COV-2 RNA shedding can be prolonged, yet high heterogeneity exists. Detection of viral RNA may not correlate with infectivity since available viral culture data suggests shorter durations of shedding of viable virus. Additional data is needed to determine the duration of shedding of viable virus and the implications for risk of transmission.


Duration of Viral RNA Shedding
Seventy-seven reports included data on viral RNA shedding by PCR. 4-80 Box 1 summarizes the key points of viral shedding duration. The duration of viral RNA shedding ranged from a minimum of 1 day 4,7,21,33,46 to a maximum of 83 days. 48 Intermittent PCR positivity did occur through day 92 from symptom onset in one case report; that patient had previously tested negative at day 72 followed by repeat positive PCR. 57 In a study of 56 serially tested hospitalized patients with mild and moderate disease, 66.1% of NP/OP swabs were still positive at 3 weeks.
Positivity rates then declined weekly and all PCR tests were negative by week 6. 15 Based on the 28 studies that provided sufficient data (see Appendix Table), the pooled median duration of RNA shedding from respiratory samples was 18.4 days (95%CI: 15.5 days -21.3 days). High heterogeneity was observed among these studies (I 2 =98.87%; p<0.01).
We reviewed shedding data for patients with mild to moderate illness. Based on parametric regression modeling, Sun et al. concluded that detection of viral RNA in throat swabs beyond 50 days post symptom onset in patients with mild illness would be a low probability event occurring beyond the 95 th percentile. 66 Despite this calculation, there are case reports of patients with viral RNA shedding  45 days from symptom onset. 48,58,[67][68][69]78,80 Among all studies we reviewed, the longest duration of PCR positivity in a patient with mild illness was 92 days after symptom onset from a NP swab. 57 The pooled median duration of viral RNA shedding from respiratory sources among patients with mild to moderate illness, based upon 10 studies that reported sufficient data (see Appendix Table), was 17.2 (95% CI: 14.0 days -20.5 days).
There were multiple reports of patients with intermittently positive PCR results from respiratory specimens. 17,21,25,27,28,51,[56][57][58]79,81 Although not consistently defined, cessation of shedding was most often described as 2 consecutive negative PCR results  24-48 hours apart. 21,23,25,38,51,56,58,81 Tests were frequently done in anticipation of discharge from the hospital. 57,81 One report estimated that 26-49% of patients were re-positive after a negative test, but in other studies re-positivity varied between 3-35%. 17,21,25,27,28,51,56,81 Wang, et al. described a case report of a patient that was discharged 75 days after illness onset following 3 consecutive negative tests. The patient then tested positive on days 82 and 92 followed by negative PCR on days 101 and 105. 57 Another case report described a woman with mild COVID- Wölfel et al. observed the pharyngeal rate of detection was highest in the first 5 days of symptom onset and then declined. 10 NP swabs may have a higher rate of detection than OP swabs, but they were only compared in two of the studies included in this review. 63,71 Negative upper tract specimens may not correlate with lower tract specimens, though the significance of these findings is not well understood. In a post-mortem analysis of a patient whose NP sample tested PCR negative, lung tissue was PCR positive and histology revealed coronavirus particles in bronchiolar epithelial cells. 59 Some studies included data for presymptomatic or asymptomatic patients and observed that PCR positivity can occur as early as 5 days prior to symptom onset. 9,10,60,61 Multiple case series reported that the viral load of asymptomatic patients are as high as those with symptoms. 9,10,62 In one case series, the asymptomatic individual in a family cluster had similar viral RNA loads in nasal and throat swabs to those of symptomatic family members. 63 The majority of the subjects in this case series converted to a negative PCR by day 18. 63 Five studies included saliva samples. 18,31,48,55,62 In a series of 13 patients with mild disease, viral RNA load was highest in saliva in the first week of illness; 3 of the patients still had detectable viral load in their saliva at day 20 of illness. 48 In another series, PCR turned negative in the saliva of 13 mildly ill patients before nasal swab PCR -an average of 13.33 +/-5.27 days and 15.67 +/-6.68 days, respectively. 55 In the same study, the average duration of positive PCR in sputum was shorter in non-ICU patients than ICU patients, who were positive for an average of 16.5 +/-6.19 days. 55

Predictors of extended duration of viral RNA shedding in respiratory samples
The most frequently identified predictor of prolonged viral RNA shedding was disease severity.
Patients with severe disease have been observed to shed RNA for longer and have higher viral RNA loads at symptom onset followed by a gradual decline in viral RNA 3 weeks after symptom onset. 29,32,50,53,64,65 Based on 10 studies (see Appendix Table), the pooled median duration of viral RNA shedding from respiratory samples in patients with severe illness was 19.8 days (95% CI: 16.2 days -23.5 days). Again, significant high heterogeneity exists (I 2 =96.42%; p<0.01). In one cohort of patients, the median duration of positive NP PCRs was 22.25 days +/-3.62 (SD) days in patients admitted to the ICU, compared to 15.67 days +/-6.68 (SD) days in non-ICU patients. 55 Sun et al. also observed prolonged duration of RNA shedding from NP swabs in those with severe illness compared to those with mild disease, with median durations of 33.5 days and 22.7 days, respectively. 66 Predictors of severe disease and duration of shedding  15 days in hospitalized patients included older age, hypertension, coronary artery disease, and diabetes mellitus. 17,27,50,52,53,62 Gender was not consistently identified as a risk factor for severe disease or prolonged shedding but comparisons were limited by small sample sizes. 47,49,52,54,62 Downloaded from https://www.cambridge.org/core. 22 Oct 2020 at 06:22:37, subject to the Cambridge Core terms of use.

Viral RNA shedding in non-respiratory samples
A subset of studies presented PCR data from both respiratory and fecal samples. [10][11][12][13][14]16,20,24,25,33,36,37,45,47,50,51,55,65,66,[70][71][72][73][74][75]78 Rectal/stool PCR pooled median duration of positivity based on 5 studies was 22.1 days (95% CI: 14.4 days -29.8 days; I 2 =95.86%, p<0.01). Stool PCR positivity has been observed to lag behind both PCR positivity of pharyngeal specimens and symptom improvement and even may become positive after the OP PCR has become negative. 16 RNA replication in the stool was observed 2 weeks after symptom onset. 10,20,50,51,73 In one study, the number of PCR-positive stool samples increased between the first and third weeks of illness, with a median time to detection in the stool of 19-22 days. 50,70 Based on the limited data available thus far, illness severity does not seem to impact stool RNA detection, as similar durations of RNA shedding in the stool have been observed in mild and severe illness. 16 Park et al. detected SARS-CoV-2 RNA in stool 50-55 days after initial diagnosis of asymptomatic or mild SARS-CoV-2 illness; in this study, people with higher viral loads were more likely to have viral RNA in the stool. 72 However, stool shedding was not consistently observed, and some studies showed that virus was detectable in only 35-59% of patients screened. 50,75 Data for serum and blood is limited but evolving. Among studies reporting serum/blood testing, viral RNA was detected in 30-87.5% of patients with COVID-19, though one smaller study did not detect viral RNA in any of the 14 patients tested. 47,50,55,75,82 The ability to detect RNA in blood and serum may be reflective of disease severity. 55,82 Virus was detected by PCR for longer in blood samples of ICU patients (14.63 +/-5.88 (SD) days) compared to non-ICU patients (10.17 +/-6.13 (SD) days). 55 Downloaded from https://www.cambridge.org/core. 22 Oct 2020 at 06:22:37, subject to the Cambridge Core terms of use.

Correlation between viral culture and PCR
Twelve studies also included both PCR and viral culture information. 4-15 Sequential viral cultures were not performed in all studies, which is a key limitation. Growth of SARS-CoV-2 on viral respiratory culture was reported ranging from 6 days before symptom onset through day 20 after symptom onset. 4,5,9,10 A position statement published in Singapore reported that viable cultured virus was not isolated past day 11. 15 Culture data suggest that duration shedding of viable virus may vary according to illness severity. In a study of patients with moderate to severe illness, Van Kampen et al. found the median duration of shedding viable virus was 8 days (IQR: 5-11 days, range: 0-20 days) with the probability of detecting virus  5% after 15.2 days. 4 In contrast, four studies of mildly ill patients did not find viable virus past day 8 or 9 of illness, but viral culture was not consistently reattempted. 5,[9][10][11] Liu, et al. described a patient with mild disease whose sputum viral culture was positive on day 18, but continued to have viral RNA detection until day 63, 45 days longer than detection of viable virus. 12 The correlation of SAR-CoV-2 viral loads and PCR cycle thresholds (Ct)  compared to 17.2 (95% CI: 14.0 days -20.5 days) for mild illness. Though these medians should be interpreted with caution given the high heterogeneity of the studies and overlapping confidence intervals, viral culture data appears to support this conclusion. In reviewed studies, viable virus from respiratory cultures was not recovered past day 9 of illness for mildly ill patients but was cultured from severely ill patients through day 20. 4,5,9,10 Interpreting positive PCR samples beyond 2-3 weeks of illness is complex. Potential explanations for these intermittently negative PCR tests include a viral load below the detection limit of the assay, specimen source, quality of specimen collection, timing of specimen collection or reinfection. 83,84 Although viral culture positivity may also not correlate perfectly with transmissibility, the correlation between culture data and Ct thresholds may help predict infectiousness. Further data is needed to understand the correlation between transmission risk, culture positivity and Ct thresholds. The studies that examined viral culture were limited by small size, inclusion of patients with mostly mild illness, and lack of serial cultures on all patients. Isolation of viable virus in respiratory samples correlates with the timing of peak viral loads which occur within 1-2 weeks of illness onset. Only one study reported culturing viable virus from a respiratory sample beyond the second week of illness. Based on this information, it seems more likely that a positive PCR past 2-3 weeks of illness represents shedding of nonviable virus. Although the pooled median viral RNA shedding duration from patients with mild/moderate and severe disease do not differ greatly, reports of positive viral cultures through day 20 in severely ill patients support the potential for a prolonged infectious period for sicker patients. In addition, viable virus has been recovered from stool cultures, but further studies are needed to determine the implications for person-to-person spread.
Our review supports the US Centers for Disease Control and Prevention (CDC) interim guidance, which recommends maintaining transmission-based precautions for 10 days after symptom onset in asymptomatic or mildly ill patients and for 20 days in severely ill patients. 85 The decision to extend the duration of transmission-based precautions is complicated given the potentially profound impact on patients and their families, hospital systems, and public health.
Prolonged home isolation may lead to longer periods of unemployment, social separation, and feelings of isolation. In the hospital, the supply of personal protective equipment, staff allocation, availability of patient beds, and the health system budget are impacted by the duration of isolation for patients with COVID-19. That said, aggressive infection control measures are required in the setting of an outbreak to control the virus and to avoid overwhelming healthcare systems.
In calculating the pooled median duration of shedding, we identified a significantly high degree of heterogeneity between studies. In a standard meta-analysis, we would not report a pooled measure of association when heterogeneity was high. However, the pooled median is not intended to inform our knowledge of causality or effect size, but rather to best inform the policy decisions that currently must be made on the very limited data available at this time in the SARS-CoV-2 pandemic. Factors contributing heterogeneity may include the variable timing of sample collection for PCR or viral culture, Ct threshold, sample types, SARS-CoV-2 genotype, Downloaded from https://www.cambridge.org/core. 22 Oct 2020 at 06:22:37, subject to the Cambridge Core terms of use. and host factors such as pharmacotherapy, comorbidities, and disease severity. We noted broad variability in the definitions of disease severity applied. While initially no formal definitions existed, the National Commission of China developed a classification scheme for mild, moderate, and severe illness that include specific clinical variables. 70 The National Institutes of Health and World Health Organization have since also developed similar severity scales. 85,86 Going forward, these definitions will facilitate the conduct of generalizable studies of viral dynamics.
This comprehensive review details the evidence available to date pertaining to SARS-CoV-2 viral dynamics. Although PCR positivity can be prolonged, culture data suggests that virus viability is typically shorter in duration. Continued reporting of viral shedding data via PCR and viral culture with improved standardization in methods and definitions, in coordination with transmission data, will facilitate evidence-based decision making for the infection control and public health measures necessary to control the pandemic.