The role of asymptomatic SARS-CoV-2 infections: rapid living systematic review and meta-analysis

Background: There is substantial disagreement about the level of asymptomatic severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection in a population. The disagreement results, in part, from the interpretation of studies that report a proportion of asymptomatic people with SARS-CoV-2 detected at a single point. Review questions: 1. Amongst people who become infected with SARS-CoV-2, what proportion does not experience symptoms at all during their infection? 2. Amongst people with SARS-CoV-2 infection who are asymptomatic when diagnosed, what proportion will develop symptoms later? 3. What proportion of SARS-CoV-2 transmission is accounted for by people who are either asymptomatic throughout infection, or pre-symptomatic? Methods: Rapid living systematic review (protocol https://osf.io/9ewys/). We searched Pubmed, Embase, bioRxiv and medRxiv using a living evidence database of SARS-CoV-2 literature on 25.03.2020. We included studies of people with SARS-CoV-2 diagnosed by reverse transcriptase PCR (RT-PCR) that documented follow-up and symptom status at the beginning and end of follow-up and modelling studies. Study selection, data extraction and bias assessment were done by one reviewer and verified by a second, with disagreement resolved by discussion or a third reviewer. We used a common-effect model to synthesise proportions from comparable studies. Results: We screened 89 studies and included 11. We estimated an upper bound for the proportion of asymptomatic SARS-CoV-2 infections of 29% (95% confidence interval 23 to 37%) in eight studies. Selection bias and likely publication bias affected the family case investigation studies. One statistical modelling study estimated the true proportion of asymptomatic infections at 18% (95% credibility interval 16 to 20%). Estimates of the proportions of pre-symptomatic individual in four studies were too heterogeneous to combine. In modelling studies, 40-60% of all SARS-CoV-2 infections are the result of transmission from pre-symptomatic individuals, with a smaller contribution from asymptomatic individuals. Conclusions: An intermediate contribution of pre-symptomatic and asymptomatic infections to overall SARS-CoV-2 transmission means that combination prevention, with enhanced hand and respiratory hygiene, testing tracing and isolation strategies and social distancing, will continue to be needed. The findings of this systematic review of publications early in the pandemic suggests that most SARS-CoV-2 infections are not asymptomatic throughout the course of infection.


Background
There is substantial disagreement about the level of asymptomatic severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection in a population. The differences are extreme.
Authors of a World Health Organization report stated that, "The proportion of truly asymptomatic infections is unclear but appears to be relatively rare and does not appear to be a major driver of transmission." 1 In contrast, reports of new infections found on a single day have led to statements that "the large majority of coronavirus infections do not result in symptoms." 2 The disagreement results, in part, from the interpretation of studies that report a proportion of asymptomatic people with SARS-CoV-2 detected at a single point. 2 3 These studies include both people who will remain asymptomatic throughout and those, known as pre-symptomatic, 4 who will develop symptoms of coronavirus disease 2019 (COVID- 19) if followed until at least the end of the incubation period of 14 days. 5 The full spectrum and distribution of COVID-19, from completely asymptomatic, to mild and non-specific symptoms, viral pneumonia, respiratory distress syndrome and death are not yet known. 6 Without follow up, however, the proportions of asymptomatic and pre-symptomatic infections cannot be determined.
Accurate estimates of the proportions of true asymptomatic and pre-symptomatic infections are needed urgently because their contribution to overall SARS-CoV-2 transmission at the population level will determine the appropriate balance of control measures. 6 If the predominant route of transmission is from people who have symptoms, then strategies should focus on testing, followed by isolation of infected individuals and quarantine of their contacts. If, however, most transmission is from people without symptoms, social distancing measures that reduce contact with people who might be infectious, should be prioritised. 6 The objectives of this study were to address three questions: 1. Amongst people who become infected with SARS-CoV-2, what proportion does not experience symptoms at all during their infection? 2. Amongst people with SARS-CoV-2 infection who are asymptomatic when diagnosed, what proportion will develop symptoms later? 3. What . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted April 29, 2020. . https://doi.org/10.1101/2020.04.25.20079103 doi: medRxiv preprint proportion of SARS-CoV-2 transmission is accounted for by people who are either asymptomatic throughout infection, or pre-symptomatic?

Methods
We conducted a rapid systematic review to provide relevant evidence during this public health emergency of international concern. 7 The protocol for this review is published in the Open Science Framework (https://osf.io/9ewys/). We report our findings according to the preferred reporting items for systematic reviews and meta-analyses. 8 We conducted the search on March 25, 2020.

Search strategy
We searched the living evidence database at the University of Bern Institute of Social and Preventive Medicine (ISPM) (https://ispmbern.github.io/covid-19/living-review/collectingdata.html), which includes daily updates of searches of four electronic databases: Medline-Pubmed, Embase, bioRxiv and medRxiv, for articles with medical subject headings and keywords for SARS-CoV-2 infection and COVID-19. We reviewed results that contained "asymp*" in the title or abstract (https://osf.io/9ewys/). We also examined articles suggested by experts and the reference lists of retrieved mathematical modelling studies and systematic reviews.

Eligibility criteria
We included studies of people with SARS-CoV-2 diagnosed by reverse transcriptase PCR (RT-PCR) that documented follow-up and symptom status at the beginning and end of follow-up. We included contact tracing investigations, case series, cohort studies, case control studies and statistical and mathematical modelling studies. We did not apply any language restrictions to the search. We excluded the following study types: case reports of a single patient, surveillance data, case series where patients were not enrolled consecutively, reports in which the primary data were found in another included publication, and mathematical modelling studies that did not specifically estimate the quantities specified in the review questions.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 29, 2020

Study selection and data extraction
One reviewer selected studies based on our predefined inclusion and exclusion criteria. A second reviewer verified all included and excluded studies. We reported the identification, exclusion and inclusion of studies in a flowchart ( Figure S1).
One reviewer extracted data using a pre-piloted extraction form in an electronic data capture system (REDCap, Vanderbilt University, USA). A second reviewer verified the extracted data using the query system in REDCap. In case of disagreements, a third reviewer was consulted. When disagreements were not resolved by discussion, we contacted study authors for clarification. The extracted variables included, but were not limited to, study design, country and/or region, study setting, population, age, review outcomes and length of follow-up.
The primary outcomes for each review question were: 1. Proportion with asymptomatic SARS-CoV-2 infection who did not experience symptoms at all during follow-up; 2. Proportion with SARS-CoV-2 infections who did not have symptoms at the time of testing but developed symptoms during follow-up.
3. Estimated proportion (with including uncertainty interval) of SARS-CoV-2 transmission accounted for by people who are asymptomatic or pre-symptomatic.

Risk of bias in included studies
Two authors independently assessed the risk of bias. A third reviewer resolved disagreements when consensus was not achieved. For review questions 1 and 2, we adapted the Joanna Briggs Institute Critical Appraisal Checklist for Case Series. 9 The adapted tool included items about inclusion criteria, measurement of asymptomatic status, follow-up of course of disease, and availability of numerator and denominator. We added items about the representativeness of source and target populations . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 29, 2020. . https://doi.org/10.1101/2020.04.25.20079103 doi: medRxiv preprint from a tool for the assessment of risk of bias in prevalence studies. 10 For review question 3, we used a tool for assessing the credibility of mathematical modelling studies. 11

Synthesis of the evidence
We used the metaprop function from the meta package (version 4.11-0) 12 in R (version 3.5.1) to display the study findings for questions 1 and 2 as forest plots. The 95% confidence intervals (CI) are estimated using the Clopper-Pearson method. 13 To synthesise proportions from comparable studies, in terms of design and population, we used both random and common-effect models. When the between-studies variance was close to zero we present only the results from the common-effect analysis.

Results
We screened 89 studies ( Figure S1). We included six studies that reported family contact tracing investigations in China, [14][15][16][17][18][19][20] two studies of populations evacuated from Wuhan and the Diamond Princess cruise ship, 21 22 one statistical modelling study based on passengers on the Diamond Princess, 23 and two mathematical modelling studies. 24 25 Proportion of asymptomatic SARS-CoV-2 infections The six family contact investigations included 39 people, with one or more individuals who remained asymptomatic throughout follow-up (Table 1). [14][15][16][17][18][19][20] One study contributed three clusters. The study included adolescents and young adults admitted to hospital in Chonqing with COVID-19. 16 The authors reported nine case investigations it total, of which four were all symptomatic and two included a person who was asymptomatic on admission, but who developed symptoms later. Nine asymptomatic cases were identified in total, with follow-up from the last possible day of exposure ranging from 17 to 33 days (Table 1), which exceeds the accepted SARS-CoV-2 incubation period of 14 days. 5 The ages were reported for six of the asymptomatic cases; two were children, aged 13 months and 10 years. For three of the asymptomatic cases, transmission to other individuals was . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted April 29, 2020. . https://doi.org/10.1101/2020.04.25.20079103 doi: medRxiv preprint reported. The common-effect summary of the proportion of asymptomatic SARS-CoV-2 from contact investigations was 23% (95% CI 12 to 39%) ( Figure 1). is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted April 29, 2020. Two studies reported on study populations evacuated from a setting where SARS-CoV-2 transmission was confirmed (Table 1, Figure 1). Nishiura et al. reported on 565 Japanese nationals evacuated from Wuhan. 22 Of these, 13 tested positive by RT-PCR for SARS-CoV-2 and four were asymptomatic. The authors reported that after 30 days of follow up, all four remained asymptomatic. Tabata et al. reported on a sample of passengers evacuated from the Diamond Princess cruise ship who tested positive by RT-PCR for SARS-CoV-2 and were followed up at a hospital in Japan. 21 Of 107 SARS-CoV-2-infected people in the hospital, 104 were followed for a median of 10 days (interquartile range 7 to 10 days). Thirty-three remained asymptomatic by the end of the follow-up period (February 26, 2020). The common-effect summary proportion of infected individuals remaining asymptomatic was 32% (95% CI 24 to 41%). The estimates from the two study types were statistically compatible. The common-effect summary of all studies was 29% (95% CI 23 to 37%) ( Figure 1). is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted April 29, 2020.  23 Of these, 328 were reported to be asymptomatic when tested. The authors estimated the timing of infection, using published data, and adjusted for the proportion of people who would develop symptoms (right censoring). In a Bayesian framework, they estimated that, if all were followed up until the end of the incubation period, the true proportion of asymptomatic infections would be 17.9% (95% credibility interval, Cr I 15.5 to 20.2%).

Proportion of pre-symptomatic SARS-CoV-2 infections
We included four studies (Table 2, Figure 2). 15 17 21 26 In three studies in China, people who had SARS-CoV-2 detected by RT-PCR were followed in hospital. In all three studies, authors reported that the asymptomatic cases were detected during contact investigations of patients who had presented with symptoms of COVID-19. In two studies, most patients developed symptoms, 15 26 including 43 of 55 patients in Shenzhen, which was the only study in which patients were followed up until clearance of SARS-CoV-2, measured by repeated negative RT-PCR test results. 26 In the third hospitalbased study, in Nanjing, five of 24 asymptomatic SARS-CoV-2 cases developed symptoms within one to three days of admission. 17 Another thirteen showed viral clearance, with two or more negative RT-PCR test results. The remainder still had positive RT-PCR test results, but had not developed symptoms. The fourth study reported the outcome of SARS-CoV-2 infection in passengers of the Diamond Princess cruise ship who were hospitalised in Japan. 21 Of 43 who were asymptomatic at the time of diagnosis, 10 developed symptoms during median follow-up of 10 days (IQR 7-10 days). The findings of the four studies were too disparate to pool in a meta-analysis (Figure 2).
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted April 29, 2020.

Contribution of asymptomatic and pre-symptomatic infection to SARS-CoV-2 to transmission
We included two transmission dynamic mathematical modelling studies that explicitly addressed this question. 24  Tianjin, China and from Singapore. The applied statistical models in a Bayesian framework to examine generation and serial intervals for linked cases, with an assumption of the incubation period from a published study. A generation period shorter than the incubation period of the infector indicates pre-symptomatic transmission. They found that the proportion of presymptomatic transmission was 48% (95% CI 32 to 67%) for Singapore and 62% (95% CI 50 to 76%) for Tianjin, China. 24 Ferretti et al. developed a compartmental mathematical model, informed by data on linked COVID-19 cases in Hubei province, China. They separated the transmission parameter into symptomatic, asymptomatic, pre-symptomatic and environmental components. In their baseline scenario, they assumed a fraction of 46% asymptomatic SARS-CoV-2 infections (citing data from the Diamond Princess) and reduced infectiousness from asymptomatic cases. They found, in their base case scenario, that pre-symptomatic patients account for 47% (95% credibility interval 11 to 58%) of the total transmission, and asymptomatic transmission 6% (0 to 57%) of the total. They provide a shiny app [ref:link], where different assumptions can be examined. 25

Summary of main findings
This rapid systematic review found that an upper bound for of 29% (95% CI 23 to 37%) for the proportion of asymptomatic SARS-CoV-2 infection. A statistical modelling study estimated that 17.9% (95% CrI 15.5 to 20.2%) remain asymptomatic. In empirical studies, the estimated proportions of people who are pre-symptomatic but who go on to develop symptoms were too heterogeneous to combine. In modelling studies, 40-60% of all SARS-CoV-2 infections are the result of transmission from pre-symptomatic individuals, with a smaller contribution from asymptomatic individuals.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Strengths and weaknesses
A strength of this review is that we clearly distinguish between SARS-CoV-2 infections that remain asymptomatic throughout their course from those that become symptomatic, and we separate the proportions of people with infection from their contribution to overall transmission in a population.
This rapid systematic review follows a published protocol and uses methods to minimise bias whilst increasing the speed of the review process. 7 As a living systematic review, we will update it regularly using a living evidence database that includes preprints. One main limitation of the review is that the database does not include all sources. The four databases cover the majority of publications and we do not believe that we have missed studies that would change our conclusions.

Comparison with other reviews
We identified one online summary of asymptomatic SARS-CoV-2 infection that listed 21 articles found through searches of five electronic databases. 27 There was no published protocol. The authors did not distinguish between asymptomatic and pre-symptomatic infection and present crosssectional studies alongside longitudinal studies and mathematical modelling studies. The review gave a wide range (5 to 80%) of infections that might be asymptomatic. The advantage of our review is that it is a systematic review and, although we included fewer studies, we only report findings from empirical studies that followed participants for at least 14 days from the last exposure and we quantify the proportion of asymptomatic infection in a meta-analysis.

Interpretation
The findings from this systematic review do not support the claim that a large majority of SARS-CoV-2 infections is asymptomatic. 2 Our estimate, 29% (95% CI 23 to 37%) is an upper bound of SARS-CoV-2 infections that remain asymptomatic throughout. We believe that our analysis is likely to have overestimate the true proportion because all the included contact investigations included at least one asymptomatic individual. At this stage of a pandemic of a new infectious disease, there is likely publication bias, with rapid publication of case reports of newsworthy findings, such as person-to person transmission by asymptomatic individuals. The denominator of all contact investigations is not known, but the study by Liao et al. also included clusters with no asymptomatic case. 16 The large . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted April 29, 2020. . https://doi.org/10.1101/2020.04.25.20079103 doi: medRxiv preprint number of case reports of novel manifestations of disease was also a feature of the Zika epidemic.
Analysis of published studies showed that estimates of the duration of infection decreased over time as less biased studies were published. 28 We anticipate that, as larger, more representative studies are conducted and published, the proportion estimated to be asymptomatic will decrease. The statistical modelling study by Mizumoto, which accounted for right censoring, provides an upper uncertainty interval of 20% asymptomatic infections, 23 which might be closer to the true proportion if that assumed incubation period distribution is accurate. The level and duration of transmissibility from individuals with asymptomatic infections is not known, but in this review, three asymptomatic cases were reported to be the source of subsequent SARS-CoV-2 infections within families. The proportion of people who are pre-symptomatic could not be determined with certainty, presumably because the distribution of dates at which SARS-CoV-2 was detected differed between studies.
Nevertheless, transmission from pre-symptomatic individuals has been found to occur one to three days before symptom onset. 29 Only mathematical modelling studies can determine the overall contribution to SARS-CoV-2 transmission of individuals with asymptomatic and pre-symptomatic infection at the population level because of the non-linear dynamics of infection transmission. In the studies included in this review, three estimates from two studies found that 40-60% of all SARS-CoV-2 infections are the result of transmission whilst pre-symptomatic, with a smaller contribution from asymptomatic infections. 24 25 The consistency of the findings is reassuring because the studies used different methods. Of note, Ganyani et al. used data about the serial interval to derive their estimate and did not use assumptions about the proportion of asymptomatic infections. 24 Our finding differs from a modelling study that has been interpreted as showing that up to 90% of SARS-CoV-2 transmission is asymptomatic. 30 Li and colleagues found that 86% of infections were defined as undocumented, 30 but it appears that this proportion is a composite of asymptomatic infections, infections that were symptomatic and diagnosed but unreported, and infections that were mild and undiagnosed.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Implications and unanswered questions
This systematic review gives important information that can inform future research and public health decision-making. An intermediate contribution of pre-symptomatic and asymptomatic infections to overall SARS-CoV-2 transmission 24 25 means that combination prevention, with enhanced hand and respiratory hygiene, testing tracing and isolation strategies and social distancing, will continue to be needed. Social distancing measures will need to be sustained at some level because droplet transmission from close contact with people with asymptomatic and presymptomatic infection is occurring. Easing of restrictions will, however, only be possible with wide access to testing, contact tracing and rapid isolation of infected individuals. Quarantine of close contacts is also essential to prevent onward transmission during asymptomatic or pre-symptomatic periods of those that have become infected. Digital, proximity tracking will need to supplement . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted April 29, 2020 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted April 29, 2020.  Included in question 2 n=4 Figure S1. Flow chart of identified, excluded and included records, as of 25 March 2020 Included in metaanalysis for question 1 n=8 Included in metaanalysis for question 2 n=4 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted April 29, 2020.  Figure S2. Risk of bias in studies of asymptomatic and pre-symptomatic infection . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 29, 2020.

METHODS
Protocol and registration 5 Indicate if a review protocol exists, if and where it can be accessed (e.g., Web address), and, if available, provide registration information including registration number.

5
Eligibility criteria 6 Specify study characteristics (e.g., PICOS, length of follow-up) and report characteristics (e.g., years considered, language, publication status) used as criteria for eligibility, giving rationale.

5
Information sources 7 Describe all information sources (e.g., databases with dates of coverage, contact with study authors to identify additional studies) in the search and date last searched.

5
Search 8 Present full electronic search strategy for at least one database, including any limits used, such that it could be repeated.

5
Study selection 9 State the process for selecting studies (i.e., screening, eligibility, included in systematic review, and, if applicable, included in the meta-analysis).

6
Data collection process 10 Describe method of data extraction from reports (e.g., piloted forms, independently, in duplicate) and any processes for obtaining and confirming data from investigators.

6
Data items 11 List and define all variables for which data were sought (e.g., PICOS, funding sources) and any assumptions and simplifications made.

6
Risk of bias in individual studies 12 Describe methods used for assessing risk of bias of individual studies (including specification of whether this was done at the study or outcome level), and how this information is to be used in any data synthesis.

6
Summary measures 13 State the principal summary measures (e.g., risk ratio, difference in means).