Asymptomatic and presymptomatic transmission of SARS-CoV-2: A systematic review

Abstract

Background and purpose: Many of the statutes comprising the shelter-in-place and phased-reopening orders are centered around minimizing asymptomatic and presymptomatic transmission. Assumptions about the presence and relative importance of asymptomatic and presymptomatic transmission are based on case reports, the failing of quarantine measures aimed at sequestering ill patients, viral dynamic studies suggesting SARS-CoV-2 production peaks before symptoms appear, and modeling evidence that calculates serial interval between successive generations of infection. In aggregate, these data offer compelling evidence of asymptomatic and presymptomatic transmission, but individually these studies have notable shortcomings that undermine their conclusions. The purpose of this review is to discuss the literature of asymptomatic and presymptomatic transmission, highlight limitations of recent studies, and propose experiments that, if conducted, would provide a more definitive analysis of the relative role of asymptomatic and presymptomatic transmission in the ongoing SARS-CoV-2 pandemic.

Methods: We conducted a systematic review of literature on PubMed using search filters that relate to asymptomatic and presymptomatic transmission as well as serial interval and viral dynamics. We focused on studies that provided primary clinical data.

Results: 34 studies were eligible for inclusion in this systematic review: 11 case reports pertaining to asymptomatic transmission, 9 viral kinetic studies, 13 serial interval studies, and 1 study with viral kinetics and serial interval.

Conclusion: Different approaches to determining the presence and prevalence of asymptomatic and presymptomatic SARS-CoV-2 transmission have notable shortcomings, which were highlighted in this review and limit our ability to draw definitive conclusions. Conducting high quality studies with the aim of understanding the relative role of asymptomatic and presymptomatic transmission is instrumental to developing the most informed policies on reopening our cities, states, and countries.

Keywords: RT-PCR; SARS-CoV-2; asymptomatic transmission; natural history of infection; presymptomatic transmission; serial interval; systematic review; viral kinetics; viral load.

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(7): e1000097. doi:10.1371/journal.pmed1000097. A complete list of studies retrieved from the search can be found in the appendix.

Figure 2.

Hypothetical distributions of SARS-CoV-2 viral load. Different assumptions about the shape of the distributions will impact when and if presymptomatic transmission will occur. A line indicating the threshold of transmissibility is shown in purple, which is currently believed to be 10⁶ copies per mL. The intersection of the purple line with the various curves would show when an individual becomes contagious. In these hypothetical distributions, a normal and Weibull distribution suggest significant presymptomatic transmission, while a gamma and lognormal distribution seem to suggest limited presymptomatic transmission. These conclusions can change with different transmission thresholds and distribution parameters. A vertical dashed line in grey shows when an individual might seek medical consultation, which Zhang and colleagues report as being 2.5 days after symptom onset in China during the COVID-19 pandemic.^[53] Although this number decreased from 3.0 to 1.6 days as the pandemic progressed. Assuming patients don’t seek medical care for 2.5 days, the light-yellow shaded region refers to the area where data is lacking. While many studies concluded viral load peaks when observation begins, for almost all of the studies, a significant portion of time elapsed between when symptoms first appeared and observation began.

Figure 3.

Green dotted line shows the reported mean incubation period of 5.2 days. Green shaded area shows 95% CI of incubation period as reported by Li et al. We preferentially reported the mean serial interval (red circle). If mean was not reported, median was used (red triangle). However, it should be noted that in skewed distributions such as gamma and lognormal, median is often less than mean. In the case of Wu et al. the mean was noted as 6.3, but no error terms were reported, therefore median was used in the figure. Error bars default to show 95% CI on serial interval on statistic, however if 95% CI was not reported, 1^st and 3^rd quartiles were used (denoted by *) or +/− 1 standard deviation (denoted by †). Error bars that extended below zero were not shown but are reported in supplemental Table 2. The two studies from Du et al. may use overlapping data, and if so, these serial intervals cannot be considered independently.