Determining the communicable period of SARS-CoV-2: A rapid review of the literature, March to September 2020

Introduction Standard testing for infection with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is based on RT-PCR tests, but detection of viral genetic material alone does not indicate ongoing infectious potential. The ability to isolate whole virus represents a better proxy for infectivity. Aim The objective of this study was to gain an understanding of the current literature and compare the reported periods of positive SARS-CoV-2 detection from studies that conducted RT-PCR testing in addition to experiments isolating whole virus. Methods Using a rapid review approach, studies reporting empirical data on the duration of positive RT-PCR results and/or successful viral isolation following SARS-CoV-2 infection in humans were identified through searches of peer-reviewed and pre-print health sciences literature. Articles were screened for relevance, then data were extracted, analysed, and synthesised. Results Of the 160 studies included for qualitative analysis, 84% (n = 135) investigated duration of positive RT-PCR tests only, 5% (n = 8) investigated duration of successful viral isolations, while 11% (n = 17) included measurements on both. There was significant heterogeneity in reported data. There was a prolonged time to viral clearance when deduced from RT-PCR tests compared with viral isolations (median: 26 vs 9 days). Discussion Findings from this review support a minimum 10-day period of isolation but certain cases where virus was isolated after 10 days were identified. Given the extended time to viral clearance from RT-PCR tests, future research should ensure standard reporting of RT-PCR protocols and results to help inform testing policies aimed at clearance from isolation.


Introduction
Understanding how long individuals may continue to transmit virus after infection with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is important to inform testing policies and isolation procedures required to prevent nosocomial and community spread. Numerous studies have been conducted to date to address this question; however, the literature has been predominantly represented by studies using methods based on reverse transcription-PCR (RT-PCR) tests. As RT-PCR tests detect the presence of viral genetic material and thus do not differentiate between live (or viable) and non-infective virus [1], inferences from these results as to potential infectious periods are limited.
A more accurate proxy of infectious potential is based on the ability of whole virus to be successfully isolated and cultured in laboratory settings, with clinical confirmation of transmission where possible. While fewer studies have been conducted that have successfully isolated and cultured live virus, results from this literature have contributed to a number of reviews [2][3][4][5][6][7]. The findings from these reviews have drawn consistent conclusions that align with previously recommended isolation strategies [8]: overall, infectious potential appears to be greatly reduced by day 10 following symptom onset of coronavirus disease   [2,7,9].
However, a growing concern relates to individuals who continue to test positive by RT-PCR over extended periods of time [10][11][12][13][14][15][16], including those who re-test positive after an initial negative test result [10]. As RT-PCR tests of respiratory samples have been and will remain the de facto method of confirming initial and ongoing infection with SARS-CoV-2, it is important to be able to interpret positive test results that are obtained throughout the disease course, including during convalescence.
Against a backdrop of dramatically rising cases internationally, limited testing resources and public health capacity for ongoing case management, and ongoing restrictions on mobility, there is a need for as much evidence as possible to help understand the likely implication of ongoing positive tests or re-tests after cessation of the recommended period of isolation, on potential disease spread.
The purpose of this review was to conduct a rapid review of existing literature in order to directly compare the duration of potential infectivity of SARS-CoV-2 from studies that obtained measurements of the duration of infectious potential using both RT-PCR and viral isolation methods. An additional, and related, objective was to understand and be able to provide an overview of the literature at the time the review was conducted (until end of September 2020), especially given the evolving nature of the pandemic.

Methods
We conducted a rapid review of the literature using the methods outlined in the National Collaborating Centre for Methods and Tools Guidebook [17]. We used the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines to report our rapid review [18]. We did not publish or pre-register a protocol for our review given time pressures for this information. From the available data, duration (maximums, medians or means values) was retrieved or calculated. Ethics approval was not required.

Search strategy
Databases of peer-reviewed and pre-print health sciences literature (Ovid MEDLINE, Embase, Google Scholar, medRxiv and arXiv) and the grey literature for reports or guidelines on discontinuation of isolation for SARS-CoV-2 from international and national public health organisations (World Health Organization, European Centre for Disease Prevention and Control, United States Centre for Disease Control websites) were searched using two search strategies: (i) terms for 'SARS-CoV-2' and 'viral clearance/shedding,' on 23 May 2020 and (ii) terms for 'SARS-CoV-2' and 'viral isolation/culture,' on 1 July 2020. The latter search was done with specific terms because there was a paucity of studies with data from viral isolation/cultures identified in the previous iteration of the search; moreover, in order to include more recently published studies, we examined other reviews, including a living evidence review, until 29 September 2020 and added relevant references to our results [1,2,[5][6][7]9]. All databases were searched from inception and searches were limited to English. Detailed information on search strategies undertaken in each database can be found in the Supplement (Appendix A). Additional studies were identified by reviewing the references of select highimpact articles, reports from reputable sources, and existing reviews. Any studies identified through other sources or contacts were manually added.

Inclusion criteria
We included studies presenting primary empirical data on duration of possible infectivity of SARS-CoV-2 in human populations using respiratory samples and reported in English. Articles that did not report data on duration of potential infectiousness in text, figures or tables were excluded. Studies reporting solely on pre-symptomatic or on convalescent period durations, incubation periods, serial intervals or on results based on statistical modelling were excluded. Studies, including reviews, that used data from other investigations were excluded; where relevant studies were identified, their references were reviewed to add relevant primary studies. When duplicate study reports were identified (i.e. a pre-print and a peer-reviewed journal article), the most recent version was included. Inclusion criteria are outlined in more detail in the Supplement (Appendix B).

Screening process
Titles and abstracts were screened for relevance independently by two reviewers. Full-text review was conducted independently by two reviewers for articles where relevance was not readily determined from title and abstracts, and any conflicts were resolved through consensus by the two reviewers.

Data extraction process
A draft data extraction form was developed and trialled across multiple reviewers to develop the final version. Extracted data fields included study characteristics (first author, publication status, study type, sample size), study population characteristics (age, hospitalisation, disease severity), method of determining infectious period (viral shedding, viral isolation), type of respiratory specimen(s) collected, reported durations (minimum, mean, median, maximum), whether cases without symptoms (asymptomatic or pre-symptomatic) were reported, whether the study focused solely on duration of communicability during the convalescent phase, how measurement of duration start and end was defined, and study quality. All extracted data were reviewed by a second reviewer.

Definitions
Sample size was defined as the total number of participants for whom data on communicability period was assessed. Disease severity associated with SARS-CoV-2 infection was classified according to the following definitions: mild referred to study populations reporting no symptoms or non-serious symptoms that did not require healthcare intervention, moderate severity included participants who required acute care and/or intervention, and severe disease referred to cases that required admission to the intensive care unit, critical intervention and/or resulted in death. Studies that included participants with mild, moderate and severe cases were classified as 'mixed'; otherwise, if they included cases that were mild/moderate or moderate/severe, they were categorised as the higher level of severity. Hospitalisation status was determined as described in the study. Notably, in some jurisdictions, admission to hospital appeared to be part of routine isolation policies and so this description alone was not taken as an indicator of disease severity. Studies with children were those that included participants aged 19 years or younger. Respiratory samples included those taken from the upper (naso/oro-pharyngeal, nasal, throat, or saliva swabs) or lower respiratory tract (from sputum or bronchial lavage specimens). For the purposes of this review, the start of measurement of duration is referred to as 'symptom onset' and measurement end as 'viral clearance'. The end point for measuring duration of viral isolation was the last reported day on which virus could be isolated and cultured (captured under 'other').

Assessment of study quality
An adaptation of the Mixed Methods Appraisal Tool was used to assess study quality [19]. Questions were concerned with the following: (i) the study had clear research questions or objectives, (ii) the collected data allowed the study to address the stated research The PRISMA flowchart details the number of studies identified through our search and then included for analysis after screening for eligibility [18]. Table 1 Characteristics of included studies, overall and broken down by method of assessment of duration of communicability of SARS-CoV-2, March-September 2020 (n = 160) question, (iii) the research question was aimed at understanding duration of communicability, (iv) there was a complete follow-up period defined to measure duration of communicability and (v) there was clarity about when measurement of communicability period started, about sample types collected and frequency of sample collection, about how long patients were followed (until viral clearance, study end, hospital discharge, death) and about the method of assessment of communicability. Owing to the emerging nature of this topic, we did not exclude studies from our synthesis or analysis based on study quality.

Analysis
All data processing and analysis was conducted using the statistical programming language R (version 4.0.0) [20]. For studies where more than one value was reported for duration (i.e. when data for multiple sample types were reported or results were presented in a stratified manner), the values corresponding to the higher duration were included for analysis. As analyses were generally aimed at identifying maximum reported durations, these values were pulled from each study and summarised. Raw data are available in Supplementary Table S1.

Results of literature search
We retrieved, 2,174 records from database searches and 91 additional studies from reference chaining and other sources (2,265 total), of which 1,481 remained after removing duplicate records. Of these, 1,234 were excluded in screening and 87 in full-text review as they did not report data on duration of potential infectivity, used secondary data from the literature, reported only on the convalescent period, assessed non-respiratory samples only, the full text was not accessible, or they were duplicate reports of the same study. 160 studies were included in the final synthesis. The PRISMA flowchart in Figure 1 illustrates the study selection process.

Study populations
Information on study characteristics is presented in Table 1 Table 1). The results of the study quality assessment are included in Table 1.

Study quality
At the time of writing, 26 (16%) of the studies included in this review were pre-prints and thus had not undergone peer-review. Many of the remaining studies published in peer-review journals were letters to the editor or other short communications that do not fully report their methods. There was variation in study quality in both preprints and studies published in peer-reviewed journals; roughly half of the included studies (n = 86; 54%) had two or more study quality concerns ( Table 1).

Comparison of maximum duration of positive RT-PCR tests and successful viral isolation reported from studies that measured both
All studies that measured the longest time to viral clearance in both RT-PCR tests and viral isolation experiments reported positive RT-PCR test results after virus was no longer able to be isolated and cultured. The median duration after symptom onset that virus was successfully isolated was 9 days (IQR: 2.25; range: 2-21), while the corresponding median value for longest duration until viral clearance by RT-PCR was 26 days (IQR: 16.8; range: 8-63) (Figure 2). Three studies reported successful viral isolation beyond 10 days (Figure 2).

Maximum reported durations of positive RT-PCR tests and successful viral isolation across all studies
We conducted a final analysis on all included studies that had measurements of maximum duration to viral clearance (n = 142). In 20 studies that successfully achieved viral isolations, viable virus could be isolated from a case across a range of 2-32 days after symptom onset; the median and mean durations of these values across studies were 10.5 days (IQR: 10) and 12.8 days (SD: 7.3), respectively. From the 134 studies with data on maximum duration of positive RT-PCR test results, the shortest reported time until viral clearance was 5 days from symptom onset, while the maximum was 95 days, with respective median and mean values of 25 days (IQR: 19) and 28.8 days (SD: 15.8) (Table 3).

Table 2b
Overview of studies that included measurements of duration to SARS-CoV-2 clearance from both RT-PCR and viral isolation experiments, March-September 2020 (n = 12) Young

Discussion
This review contributes to the evidence on the period of potential infectivity after infection with SARS-CoV-2, showing that positive detection of viral nucleic acid by RT-PCR can far exceed the duration after which viral transmission may occur. Results from this review suggest that positive RT-PCR results after day 10 following symptom onset of COVID-19 are unlikely to indicate infectiousness. The conclusions of this review are also in line with recently updated guidelines that support relying on symptoms rather than RT-PCR test results for ending isolation precautions in non-healthcare settings [45].
Overall, results from this review concur with previous studies that also support a general 10-day isolation period. However, in reviewing the data from all studies reporting viral isolation, we identified important exceptions. The longest period that virus was isolated after symptom onset was 32 days from a hospitalised patient with severe disease [46]. A duration of 21 days was identified from a report of a single individual with mild symptoms but with significant underlying conditions including a recent heart transplant [21]. A separate case report of an individual without any significant comorbidities and also with mild disease symptoms reported viable virus until 18 days [11]. One case series of 129 hospitalised patients with moderate to severe disease, and mixed degrees of comorbidities, reported a maximum duration of viable virus of 20 days, with a median time to viral clearance in this sample of eight days [38].
Although these exceptions may not occur frequently, they highlight the need to be cautious when COVID-19 cases are being released from isolation into high-risk settings. Previous studies have suggested that there may be differences in duration of viral viability by disease severity with cases of more severe disease potentially having longer infectious viral shedding [7,39]. Despite such trends, the select cases highlighted herein suggest symptom presentation and disease severity do not always follow a pattern with respect to duration of viral viability. Of the studies assessing viral isolation, some also isolated and cultured virus from patients who were asymptomatic or during their convalescent disease period [21][22][23][24].
Given the prolonged detection of viral nucleic acid by RT-PCR testing, there have been suggestions to incorporate additional testing metrics when results are being used to inform release from isolation [9,40], such as Ct values. This would be an important and relevant qualification to current protocol especially when considering that positive tests from RT-PCR may persist for several weeks [12][13][14][15][16]41,47]. While the available data on RT-PCR Ct values and viral viability were inconsistent, a number of studies have reported an upper Ct limit for successful viral isolation [9,22,[25][26][27][28]40,41,46] which may help inform future threshold values.
This review highlights significant heterogeneity in the content and quality of the underlying literature which limits the ability to draw robust inferences to inform isolation protocols from the available data. There was a general paucity of high-quality evidence. We identified few studies that investigated viral isolation and many of the included studies had small sample sizes, making it difficult to draw robust findings from the single subject studies. The included studies suffer from selection bias and lack of generalisability, as case reports and case series often focus on highly specific clinical populations that do not represent the majority of COVID-19 cases. Given that data were often drawn from applied clinical settings, several of the included studies did not follow participants until viral clearance or collect samples at consistent intervals throughout the communicable period. A lack of reporting standards also resulted in incomplete information on valuable metrics such as viral load and Ct values. Finally, although this was not a focus of this review, studies that are able to demonstrate clinical confirmation of transmission are needed to better understand infectivity [47].
This study builds off of previous research by directly comparing results from all studies that measured both the maximum duration of positive RT-PCR test results and longest time to successful viral isolation, thereby adding robustness to interpreting overall findings from the literature. We have made the raw data extracted from all studies in this review available in Supplementary  Table S1 for the benefit of other research or public health groups. This research identifies improvements for future research and reporting that should enable more robust syntheses of newly emerging evidence to better inform infection control policies for SARS-CoV-2.