ABSTRACT
Background Viral load kinetics and the duration of viral shedding are important determinants for disease transmission. We aim i) to characterize viral load dynamics, duration of viral RNA, and viable virus shedding of SARS-CoV-2 in various body fluids and ii) to compare SARS-CoV-2 viral dynamics with SARS-CoV-1 and MERS-CoV.
Methods Medline, EMBASE, Europe PMC, preprint servers and grey literature were searched to retrieve all articles reporting viral dynamics and duration of SARS-CoV-2, SARS-CoV-1 and MERS-CoV shedding. We excluded case reports and case series with < 5 patients, or studies that did not report shedding duration from symptom onset. PROSPERO registration: CRD42020181914.
Findings Seventy-nine studies on SARS-CoV-2, 8 on SARS-CoV-1, and 11 on MERS-CoV were included. Mean SARS-CoV-2 RNA shedding duration in upper respiratory tract, lower respiratory tract, stool and serum were 17.0, 14.6, 17.2 and 16.6 days, respectively. Maximum duration of SARS-CoV-2 RNA shedding reported in URT, LRT, stool and serum were 83, 59, 35 and 60 days, respectively. Pooled mean duration of SARS-CoV-2 RNA shedding was positively associated with age (p=0.002), but not gender (p = 0.277). No study to date has cultured live virus beyond day nine of illness despite persistently high viral loads. SARS-CoV-2 viral load in the upper respiratory tract appears to peak in the first week of illness, while SARS-CoV-1 and MERS-CoV peak later.
Conclusion Although SARS-CoV-2 RNA shedding in respiratory and stool can be prolonged, duration of viable virus is relatively short-lived. Thus, detection of viral RNA cannot be used to infer infectiousness. High SARS-CoV-2 titers are detectable in the first week of illness with an early peak observed at symptom onset to day 5 of illness. This review underscores the importance of early case finding and isolation, as well as public education on the spectrum of illness. However, given potential delays in the isolation of patients, effective containment of SARS-CoV-2 may be challenging even with an early detection and isolation strategy.
Funding No funding was received.
INTRODUCTION
Viral load kinetics and the duration of viral shedding are important determinants for disease transmission. They determine the duration of infectiousness which is a critical parameter to inform effective control measures and disease modelling. While a number of studies have evaluated SARS-CoV-2 shedding, viral load dynamics and duration of viral shedding reported across studies so far have been heterogenous.1 In several case series with serial respiratory sampling, peak viral load was observed just before, or at the time of symptom onset.2-4 Viral ribonucleic acid (RNA) shedding was reported to be persistent in the upper respiratory tract and in feces, for over one month after illness onset.1 However, the duration of SARS-CoV-2 RNA detection has not been well characterized. A comprehensive understanding of viral load dynamics, length of viral shedding, and how these relate to other factors, such as age and disease severity is lacking.
The aim of this systematic review and meta-analysis was to i) characterize the viral load dynamics of SARS-CoV-2, duration of viral RNA shedding by reverse transcriptase polymerase chain reaction (RT-PCR) and viable virus shedding in various body fluids and ii) compare SARS-CoV-2 viral dynamics with that of SARS-CoV-1 and MERS-CoV.
METHODS
Search Strategy
We retrieved all articles reporting viral dynamics and/or the duration of shedding of SARS-CoV-2, SARS-CoV-1 or MERS-CoV in various specimens through systematic searches of major databases including Medline, EMBASE, Europe PMC, pre-print databases (MedRxiv, BioRxiv) and the grey literature from 1 January 2003 to 6th June 2020 using Medical Subject Headings (MeSH) terms (Supplementary Material). We also manually screened the references of included original studies to obtain additional studies. Studies prior to 2003 were excluded since the first recognized case of SARS-CoV-1 was identified in March 2003.
This systematic review was registered in PROSPERO on 29th April 2020 (CRD42020181914) and will be updated in three monthly intervals as a living systematic review.
Study Selection
Studies were eligible if they met the following inclusion criteria: (1) report on SARS-CoV-2, SARS-CoV-1 or MERS-CoV infection and (2) report viral load kinetics, duration of viral shedding or viable virus. We excluded: (1) review papers; (2) animal studies; (3) studies on environmental sampling; (4) case reports and case series with < 5 participants, due to likely reporting bias; (5) papers where the starting point of viral shedding was not clear or reported from post-discharge and (6) modelling studies with no original data.
Data Extraction
Two authors (MT and OL) screened and retrieved articles according to the eligibility criteria. Four reviewers (MT, OL, JS, MC) performed full text review and final article selection. From each study, the following variables were extracted as a minimum: name of first author, year of publication, city and country, sample size, median age, sex ratio, time from symptom onset to viral clearance detected by RT-PCR and culture in different specimens, and longest reported time to viral clearance. If these data were not reported, we also contacted the authors to request the data. If available, we extracted data on peak viral load, clinical outcome, and reported factors associated with duration of viral shedding.
Risk of bias in included studies
Two authors (OL and JS) independently assessed study quality and risk of bias using the Joanna Briggs Institute (JBI) Critical Appraisal Checklist tools,5 which comprise standardized checklists, for the different study designs included in this review. Any disagreements regarding grading of quality were resolved through discussion with a third author (MC).
Meta-Analysis
For every study included, mean duration of viral shedding and 95% confidence interval (CI) were calculated. The random-effects model (DerSimonian or Laird) was applied to estimate a pooled effect size. Forest plots illustrated the detailed representation of all studies based on the effect size and 95% CI. If not reported, means and standard deviations were derived from sample size, median, interquartile range (IQR), minimum and maximum values.6 Heterogeneity between studies were quantified by the I2 index and Cochran’s Q test. Publication bias was not assessed as usual appraisal methods are uninformative when meta-analysed studies do not include a test of significance. A weighted meta-regression using an unrestricted maximum likelihood model was performed to assess the impact of potential moderators on the pooled effect size (P-values <0.05 were considered significant). All statistical analyses were performed using Comprehensive Meta-Analysis (CMA) version 3 software (Biostat, Englewood, mNJ).
RESULTS
The systematic search identified 1486 potentially relevant articles. Three hundred and fifty articles were retrieved for full text review. After reviewing the eligibility criteria, a total of 79 studies on SARS-CoV-2, eight on SARS-CoV-1, and 11 on MERS-CoV were included (Figure 1).
From: Moher D, Liberati A, Tetzlaff J, Altman DG, The PRISMA Group (2009). Preferred Reporting Items for Systematic Reviews and Meta-Analyses: The PRISMA Statement. PLoS Med 6(6): e1000097. doi:10.1371/journal.pmed1000097
For more information, visit www.prisma-statement.org.
Summary of SARS-CoV-2 studies
Of the 79 papers included, 58 studies were conducted in China (Table 1). Six studies included outpatient or community cases, the remainder comprised hospitalized patients only. Six studies reported viral load dynamics exclusively in children.7-12 Two additional studies included children, but data on viral load dynamics were presented in aggregate with adults.13,14 One study reported findings in renal transplant patients.15
Median duration of viral shedding
In total, 61 studies reported median or maximum viral RNA shedding in at least one body fluid and six studies provided duration of shedding stratified by illness severity only. Of those, 43 (3229 individuals) reported duration of shedding in upper respiratory tract (URT), seven (260 individuals) in lower respiratory tract (LRT), 13 (586 individuals) in stool, and 2 studies (108 individuals) in serum samples were eligible for quantitative analysis. Means viral shedding durations were 17.0 days (95% CI, 15.5-18.6), 14.6 days (95% CI, 9.3-20.0), 17.2 days (95% CI, 14.4-20.1) and 16.6 days (95% CI, 3.6-29.7), respectively (Figures 2 to 5). Maximum duration of RNA shedding reported in URT, LRT, stool and serum were 83, 59, 35 and 60 days, respectively.
Note: the overall effect is plotted as a black square.
Test for heterogeneity: Q-value = 4076,08, df(Q) = 42, p < 0.001, I2 = 99%.
Note: the overall effect is plotted as a black square.
Test for heterogeneity: Q-value = 203.3, df(Q) = 6, p < 0.001, I2 = 97%.
Note: the overall effect is plotted as a black square.
Test for heterogeneity: Q-value = 77,6, df(Q) = 1, p < 0.001, I2 = 99%.
Note: the overall effect is plotted as a black square.
Test for heterogeneity: Q-value = 356.0, df(Q) = 12, p < 0.001, I2 = 96.6%.
Studies reporting duration of viral shedding in URT and stool samples were eligible for meta-regression analysis. Pooled mean viral shedding duration was positively associated with age (slope: +0.304; 95% CI, +0.115 to +0.493; p = 0.002 Fig 6), but not gender (p = 0.277, Supplementary Fig 3). When adjusted for the proportion of male subjects in a multivariable analysis, mean age was positively associated with the mean duration of viral shedding in URT specimens (p = 0.003). There was a positive but non-significant association between mean age and duration of shedding in stool (p = 0.37) (Supplementary Figure 4).
URT: upper respiratory tract.
Note: the plot was built upon 41 studies (no data on mean age from the study of Qian et al.98). A random-effects model was used.
Peak viral load
The majority of studies evaluating SARS-CoV-2 viral load in serial URT samples demonstrated peak viral loads within the first week of symptom onset. 2,4,8,16-24 The highest viral loads were reported either soon after or at the time of symptom onset2,8,16,23,24 or at day 3-5 of illness4,18,20,22 followed by a consistent decline.
Five studies that evaluated viral load dynamics in LRT samples observed a peak viral load in the second week of illness.4,18,20,23,25 In contrast, the dynamics of SARS-CoV-2 shedding in stool is more erratic, with highest viral loads reported on day 7,18 2-3 weeks,24,25 and up to 5-6 weeks after symptom onset.23 While several studies reported significantly higher viral titres in stool compared to respiratory samples,8,25 Huang et al. reported lower viral load in stool than in both LRT and URT samples early in the disease course.23
Severity and association with duration of viral shedding
In total, 20 studies evaluated duration of viral RNA shedding based on disease severity. The majority (n=13) reported longer duration of viral shedding in patients with severe illness than those with non-severe illness, 18,25-36 while five studies reported similar shedding durations according to disease severity in URT samples17,19,37-39 and one study in stool samples.40 Only one study reported shorter viral shedding in moderate to severe illness compared to mild to moderate illness.41 Six studies have performed comparative analysis based on severity of illness;18,25,27,28,38,39 the majority (n=5) demonstrated significantly longer duration of shedding among the severe illness group compared to the non-severe patients and only one study observed no difference.39 (Table 2).
Other factors associated with prolonged shedding
All but one study42 (n=10) that examined the impact of age on SARS-CoV-2 shedding identified an association between older age and prolonged viral RNA shedding.25,26,28,33,37-39,43-45 Three studies identified age as an independent risk factor for delayed viral clearance.25,26,38 Male sex was also associated with prolonged shedding,25,38,46 and the association remained significant even when patients were stratified based on illness severity.25,38 Corticosteroid treatment was associated with delayed viral clearance in four studies,33,38,47,48 and one study that recruited 120 critically ill patients, found no difference between corticosteroid and control groups.49
In a phase 2 open-label study evaluating interferon beta-1b, lopinavir–ritonavir, and ribavirin a shorter duration of viral shedding was seen with combination treatment compared to the control.50 None of the antiviral regimens (chloroquine, oseltamivir, arbidol, and lopinavir/ritonavir) independently improved viral RNA clearance.28,51 In a retrospective study of 284 patients, lopinavir/ritonavir use was associated with delayed viral clearance even after adjusting for confounders.28
Asymptomatic SARS-CoV-2 shedding
Twelve studies reported on viral load dynamics and/or duration of viral shedding among patients with asymptomatic SARS-CoV-2 infection (Table 3); two demonstrated lower viral loads among asymptomatic patients compared to symptomatic patients,8,52 while four studies found similar initial viral loads. 13,14,53,54 However, Chau et al reported significantly lower viral load in asymptomatic patients during the follow up compared to symptomatic patients.53 Faster viral clearance was observed in asymptomatic individuals in five out of six studies.13,28,53,55,56 The exception Yongchen et al., found longer shedding duration among asymptomatic cases, but the difference was not significant.36
Live virus detection
We identified 11 studies that attempted to isolate live virus. All eight studies that attempted virus isolation in respiratory samples successfully cultured viable virus within the first week of illness, 9,17,20,54,57-60 No live virus was isolated from any respiratory samples taken after day 8 of symptoms in three studies,20,57,58 or beyond day 9 in two studies17,54 despite persistently high viral RNA loads. One study demonstrated the highest probability of positive culture on day 3 of symptoms.57 Arons et al. cultured viable virus 6 days before typical symptom onset, however onset of symptom was unclear.54
The success of viral isolation correlated with viral load quantified by RT-PCR. No successful viral culture was obtained from samples with a viral load below 106 copies/ml, 20 Ct values >24,57 or >34,54,58 with culture positivity declining with increasing Ct values.58 Several other studies cultured live virus from RT-PCR positive specimens; however, they did not correlate these results with viral load titres.9,59,60
Only one study reported the duration of viable virus shedding in respiratory samples; the median time to clearance from URT and LRT samples was 3.5 and 6 days, respectively.20 Arons et al. cultured viable virus in one out of three asymptomatic cases from the respiratory tract.54
Viral culture was successful in two of three RNA-positive patients in one study, but the time points from symptom onset were not reported.61 Andersson et al. failed to culture virus from 27 RT-PCR positive serum samples.62
Summary of SARS-CoV-1 and MERS studies
Eight studies on SARS-CoV-1 were included; the majority of studies did not report mean or median duration of viral shedding thus, were not eligible for quantitative analysis. The maximum duration of viral shedding reported was 8 weeks in URT,63,64 52 days in LRT,63,65 6-7 weeks in serum,66 and 126 days in stool samples.63,65,67-69 Dialysis patients had longer viral shedding in stool compared to non-dialysis patients.68 Studies that have evaluated SARS-CoV-1 kinetics found low viral load in the initial days of illness, increasing after the first week of illness in URT samples, peaking at day 10,70 or day 12-14,67 and declining after week 3-4.64 High viral loads correlated with severity of illness64 and poor survival.64 While Chen et al. identified an association between younger age and lower viral titers, 64 Leong et al. found no difference.69 Viable SARS-CoV-1 was isolated from stool and respiratory samples up to 4 weeks, and urine specimens up to day 36.63,66 All attempts to isolate virus from RT-PCR–positive stool specimens collected >6weeks after disease onset failed.65 The isolation probability for stool samples was approximately 5 to 10 times lower compared to respiratory specimens.63
We identified 11 studies on MERS-CoV. Three studies (324 subjects) reporting MERS-CoV shedding in URT and four studies (93 subjects) in LRT were included in the quantitative analysis. The mean shedding duration was 15.3 days (95% CI, 11.6 – 19.0) and 16.6 days (95% CI, 14.8 – 18.4), respectively (Supplementary Figures 1 and 2). Only one study reported duration of viral shedding in serum with a median of 14 days and max of 38 days.71 In a small study, mortality rates were higher in patients with viraemia.72 In URT and LRT specimens, prolonged shedding was associated with illness severity73,74 and survival75 with the shortest duration observed in asymptomatic patients.73 Peak viral loads were observed between days 7 to 10 and higher viral loads was observed among patients with severe illness and fatal outcome.71,73,74,76,77 Differences in viral loads between survivors and fatal cases was more pronounced in the second week of illness (P< 0.0006).77 The proportion of successful viable culture was 6% in respiratory samples with a viral load values below 107copies/ml.78
Qualitative analysis
All but 11 studies (6 cohort studies, 2 cross-sectional studies, and 1 RCT on SARS-CoV-2 and 2 cohort studies on MERS-CoV) were case series, the majority of which recruited non-consecutive patients and therefore prone to possible selection bias. (Supplementary Table 1)
DISCUSSION
This systematic review and meta-analysis provide comprehensive data on the viral dynamics of SARS-CoV-2 including the duration of RNA shedding and viable virus isolation. Our findings suggest that while patients with SARS-CoV-2 infection may have prolonged RNA shedding, median time to live virus clearance from upper and lower respiratory tract samples were 3.5 days and 6 days, respectively. No live virus isolated from culture beyond day nine of symptoms despite persistently high viral RNA loads, thus emphasizing that the infectious period cannot be inferred from the duration of viral RNA detection. This finding is supported by several studies demonstrating a relationship between viral load and viability of virus, with no successful culture from samples below a certain viral load threshold.
SARS-CoV-2 viral load appears to peak in the URT within the first week of illness, and later in the LRT. In contrast, peaks in SARS-CoV-1 and MERS-CoV viral loads in the URT occurred at days 10-14 and 7-10 days of illness, respectively. Combined with isolation of viable virus in respiratory samples primarily within the first week of illness, patients with SARS-CoV-2 infection are likely to be most infectious in the first week of illness. Several studies report viral load peaks during the prodromal phase of illness or at the time of symptom onset,2,4,8,16-23 providing a rationale for the efficient spread of SARS-CoV-2. This is supported by the observation in contact tracing studies that the highest risk of transmission occurs during the prodromal phase or early in the disease course.79,80 No secondary cases were identified beyond 5 days after the symptom onset.81 Although modelling studies estimated potential viral load peak before symptom onset, we did not identify any study that confirms pre-symptomatic viral load peak.16
Emerging evidence suggests a correlation between virus persistence and disease severity and outcome.18,25,27-29,38 This is consistent with the viral load dynamics of influenza, MERS-CoV, and SARS-CoV-1 whereby severe disease was also associated with prolonged viral shedding.73,74,82 However, more studies are needed to understand the duration of viable virus in patients with severe illness.
Similar to SARS-CoV-1, SARS-CoV-2 can be detected in stool for prolonged periods, with high viral loads detected even after 3 weeks of illness. A clear difference between SARS-CoV-1 and MERS-CoV is the detection of viral RNA in stool. In SARS-CoV-1, RNA prevalence in stool samples was high, with almost all studies reporting shedding in stool. Although viable SARS-CoV-1 was isolated up to 4 weeks of illness, fecal-oral transmission was not considered to be a primary driver of infection. Whereas in MERS-CoV, none of the studies reported duration of viral shedding in stool and RNA detection was low.77,83 To date, only a few studies have demonstrated viable SARS-CoV-2 in stool.61,84 Thus, the role of fecal shedding in viral transmission remains unclear.
Viral loads at the start of infection appear to be comparable between asymptomatic and symptomatic patients infected with SARS-CoV-2. Nevertheless, most studies demonstrate faster viral clearance among asymptomatic individuals. This suggests similar transmission potential among both groups at the onset of infection, but a shorter period of infectiousness in asymptomatic patients. This is in keeping with viral kinetics observed with other respiratory viruses such as influenza and MERS-CoV, in which people with asymptomatic infection have a shorter duration of viral shedding than symptomatic individuals.73,85 However, there are limited data on the shedding of infectious virus in asymptomatic individuals to quantify their transmission potential to inform policy on quarantine duration in the absence of testing.
This is the first study that has comprehensively examined and compared SARS-CoV-2, SARS-CoV-1 and MERS-CoV viral dynamics and performed a meta-analysis of viral shedding duration. Our study has limitations. First, some patients in the included studies received a range of treatments, including steroids and antivirals, which may have modified the shedding dynamics. Second, most of the included studies are case series, which are particularly vulnerable to selection bias. Third, our meta-analysis identified substantial study heterogeneity, likely due to differences in study population, follow up and management approaches. Further, shedding duration is reported as median ± IQR for most studies, but meta-analysis necessitates conversion to mean ± SD.6 The validity of this conversion is based on the assumption that duration of viral shedding is normally distributed, which may not apply to some studies. Lastly, although there is likely a broad overlap, the true clinical window of infectious shedding may not entirely align with viral culture duration.
We identified a systematic review of SARS CoV-2 viral load kinetics that included studies published up until 12 May 2020.86 This review included many studies that did not meet our eligibility criteria, including 26 case reports and 13 case series involving <5 individuals; these are prone to significant selection bias, reporting atypical cases with prolonged viral shedding. Additionally, the review included studies that reported viral shedding duration from the time of hospital admission or initial PCR positivity, rather than symptom onset. Furthermore, no meta-analysis of the duration of viral shedding was performed.
This review provides detailed understanding about the available evidence to date on viral dynamics of SARS-CoV-2 and has implications for pandemic control strategies and infection control practices. Although SARS-CoV-2 RNA shedding can be prolonged in respiratory and stool samples, the duration of viable virus is short-lived, with culture success associated with viral load levels. No study has reported live SARS-CoV-2 beyond day nine to date. Most studies detected the SARS-CoV-2 viral load peak within the first week of illness. These findings highlight that isolation practices should be commenced with the start of first symptoms including mild and atypical symptoms that precede more typical COVID-19 symptoms. This systematic review underscores the importance of early case finding and isolation, as well as public education on the spectrum of illness. However, given potential delays in the isolation of patients, effective containment of SARS-CoV-2 may be challenging even with an early detection and isolation strategy.87
Data Availability
All data fully available
Authors contributions
M. Cevik: conceptualisation, methodology, investigation, data curation, writing – original draft. M. Tate: investigation, data curation, writing – original draft; O Lloyd: investigation, data curation, writing – review and editing; A. E. Maraolo: formal analysis, writing – original draft; J. Schafers: investigation, data curation, writing – review and editing; A Ho: conceptualisation, methodology, data curation, writing – original draft, supervision.
Financial support and sponsorship
No financial support received
Conflicts of interest
All authors have nothing to disclose.
Acknowledgements
We would like to thank Vicki Cormie at the University of St Andrews for assistance with the search and obtaining papers not readily accessible.
References
- 1.↵
- 2.↵
- 3.
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.
- 11.
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.
- 31.
- 32.
- 33.↵
- 34.
- 35.
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.↵
- 84.↵
- 85.↵
- 86.↵
- 87.↵
- 88.
- 89.
- 90.
- 91.
- 92.
- 93.
- 94.
- 95.
- 96.
- 97.
- 98.↵
- 99.
- 100.
- 101.
- 102.
- 103.
- 104.
- 105.
- 106.
- 107.
- 108.
- 109.
- 110.
- 111.
- 112.