Epistemic status: Not quite settled science, but preprints seem to agree.
Strong evidence points to presymptomatic sources as a major source of COVID-19 infections, possibly the majority. The exact proportion is environment-dependent; awareness and public health measures reduce symptomatic transmission more than they reduce presymptomatic transmission.
The main reasons for thinking presymptomatic transmission is significant are direct measurements of the serial interval and incubation period, and the outside view of what level of public health measures have and haven't succeeded at containment.
Before delving into papers, a quick aside. If COVID-19 were only transmissible when people were coughing or feverish, containing it would be pretty easy; just tell people to stay home if they have those symptoms. Some people might try to go out anyways, so you might also set up checkpoints where people have their temperature taken and have someone listen to whether they're coughing, but that would pretty much be sufficient. Empirically, however, COVID-19 is successfully spreading in countries which have taken these measures and other more extreme measures, which is what we would expect given presymptomatic transmission, but not what we would expect without it.
(Note: You might think this means that symptomatic people aren’t contagious, but actually it just means that people who show symptoms are doing a good job of isolating themselves. People with COVID-19 symptoms are definitely contagious and need to isolate themselves and notify people they might have spread it to.)
(Note: Presymptomatic transmission is a separate issue from asymptomatic carriers. Presymptomatic transmission is when someone is contagious when they aren't symptomatic yet. An asymptomatic carrier is someone who is contagious but who never develops symptoms. Asymptomatic carriers seem to be rare, though not completely nonexistent.)
Serial Interval and Incubation Period
The serial interval is the average length of time between transmissions in a transmission chain; that is, given pairs of people A and B where A was infected on a known date and then infected B on a known date, the serial interval is the average amount of time between those dates. The incubation period is the amount of time between when someone is infected, and when they display symptoms.
If the serial interval is shorter than the incubation period, this implies that a large fraction of transmission must be presymptomatic. So, with that in mind, I went looking for studies which measure COVID-19's incubation period and serial interval. These are in two tables below.
One of these studies, Tapiwa Ganyani et al, estimated the proportion of transmission which was pre-symptomatic: 48% (95% CI 32-67%) for Singapore and 62% (95%CI 50-76%) for Tianjin. No other studies estimated this quantitatively, but most stated that their results provided qualitative evidence that presymptomatic transmission is occurring.
Tapiwa Ganyani et al and Lauren C Tindale et al appear to have used overlapping public data sources. The sample size column for serial interval studies is unusually painful, as sample-size columns go, because many of the studies needed to account for uncertainty in who infected who; as such, sample sizes are reported varyingly in units of (in order from most to least reliable per sample) pairs, chains, clusters, and cases.
The study with the longest estimated serial interval, Qun Li et al, looks at a small number of clusters and guesses which cases infected which other cases. While it estimates a mean serial interval of 7.5, Its data is also compatible with an interpretation in which the mean serial interval is shorter and some of the transmissions are indirect. This change in interpretation would bring it in line with other studies in this set, which estimate shorter intervals.
One of these studies, Zhanwe Du et al, estimated the serial interval using when people became symptomatic (rather than when they were exposed), and found that in 13% of cases, the infectee showed symptoms before the infector did. This would imply that either in those cases the infector transmitted presymptomatically, the infector had a relatively long incubation period, and the infectee had a relatively short incubation period; or that this data set had major issues identifying who affected who. The distribution of SIs fits a nice Gaussian, which is some evidence that it's the former.
Anecdotal Reports and Case Studies
To understand what presymptomatic transmission of COVID-19 would look like, I went looking for anecdotes and case studies of known COVID-19 transmission events. You can't use these to infer much about rates, but they're helpful for internalizing what presymptomatic transmission would look like.
"I believe I caught it when attending a small house party at which no one was coughing, sneezing or otherwise displaying any symptoms of illness. It appears that 40% of the attendees of this party ended up sick."
(via Google Translate) "On January 24, Li and his grandfather, grandma, and father went to aunt's house for dinner, a total of 9 people. On January 28, Li developed fever. … all 9 people participating in the dinner were confirmed as confirmed cases."
The main practical implication is that contact tracing is really important.
Contact tracing is where, when you find someone with COVID-19, you identify everyone they might have spread it to and warn them that they've been exposed. People who've been exposed are expected to quarantine themselves for 14 days, which is long enough that if they are in fact infected, there's only a 1% chance they are infected but not yet symptomatic . Back in January, this served two purposes: it ensured that if they had a cough, they wouldn't brush it off as something minor and keep going to work, and it ensured that if they didn't have a cough, they wouldn't transmit it while presymptomatic. The first issue is now less of a concern; everyone knows that if someone has a cough, they aren't supposed to go to work, even if it's definitely rhinovirus. The second issue is exactly as much a concern as it was before.
Looks like I was about a day ahead of the curve. I wonder whether Elizabeth Cohen reads LessWrong? In any case, I'm glad the media is now making this more well known. That article is mostly based on the same papers as my post, but adds some other evidence, such as Sandra Ciesek's measurements of viral shedding,
The earliest swabs were taken on day 1 of symptoms,
with symptoms often being very mild or prodromal.
patients except one, throat swab RNA concentrations seemed to be already on the decline at
the time of first presentation.
BTW, my speculation of how presymptomatic transmission happens is that, in addition to droplets caused by normal speaking, during that time the virus also causes very mild itching/tingling in the throat, which causes very mild coughs (throat clearings) that don't even get noticed as symptoms. (This is based on a family member having a cold recently, and me noticing such very mild symptoms in myself that I probably wouldn't have noticed if I hadn't been looking for them.)
ETA: The above linked paper is also interesting for these reasons:
these data indicate active replication of SARS-CoV-2 in the throat during the first 5 days after
When aligned to viral load courses, it seems there is no abrupt virus elimination at the time of
seroconversion. Rather, seroconversion early in week 2 coincides with a slow but steady
decline of sputum viral load. Whether certain properties such as glycosylation pattern at
critical sites of the glycoprotein play a role in the attenuation of neutralizing antibody
response needs further clarification. In any case, vaccine approaches targeting mainly the
induction of antibody responses should aim to induce particularly strong antibody responses
in order to be effective.
COVID-19 is successfully spreading in countries which have taken these measures ["tell people to stay home if they have those symptoms"] and other more extreme measures
How true is this? I haven't delved in that closely, but my impression is that a big part of what's been successful in containing the spread in places like Hong Kong and mainland China has involved identifying & isolating people as soon as they show symptoms.
When they identify a person with symptoms, they certainly isolate that person, but they also do contact tracing. Standard practice is to tell everyone who's interacted with an infected person to self-quarantine for 14 days, even if they aren't symptomatic. If that happens soon enough and is thorough enough, it can in principle prevent all presymptomatic transmission.
Yeah, I agree that contact tracing & testing/quarantining contacts is good, and that presymptomatic transmission is possible.
It looked to me like you were claiming that the hypothesis "stopping all symptomatic transmission is sufficient to prevent the number of COVID-19 cases from curving upwards" has been tested by some countries' measures and found to be false, and I am questioning that apparent assertion.
It's funny how counterintuitive this post was at the time, and yet how obvious it feels now. Thanks again for writing it, if I hadn't learned this I would've been much worse off in my qualitative thinking.
Promoted to curated: I think this is the kind of literature review and analysis that is really useful at this point, in particular given that lots and lots of papers are skipping peer-review, and so it's often hard to judge the quality of things.
Asked on Facebook crosspost: If it's transmitted by cough, how do asymptomatic people transmit it? Is it dependent on other diseases or allergies to provoke coughing?
None of the studies cited address the question of how presymptomatic transmission occurs, but my guess is that it's the same: respiratory droplets. People also emit droplets when they talk and breathe. There aren't as many and they don't travel as far as when they cough, but they're there.
Abstract We report temporal patterns of viral shedding in 94 patients with laboratory-confirmed COVID-19 and modeled COVID-19 infectiousness profiles from a separate sample of 77 infector–infectee transmission pairs.
We observed the highest viral load in throat swabs at the time of symptom onset, and inferred that infectiousness peaked on or before symptom onset.
We estimated that 44% (95% confidence interval, 25–69%) of secondary cases were infected during the index cases’ presymptomatic stage, in settings with substantial household clustering, active case finding and quarantine outside the home.
Disease control measures should be adjusted to account for probable substantial presymptomatic transmission.
I notice that the estimates of serial interval (almost?) all come from places that had pretty aggressive & successful containment measures in place, such as identifying & isolating potential carriers (including people who show symptoms, traced contacts, and high-risk travelers). That would tend to shorten the serial interval, since people who are identified early in their infection lose the opportunity to transmit during the later portion of their illness.
Are there estimates of what R was for these populations? If it's a lot less than the 2-3 that other studies have found that would be some evidence that a lot of later-stage transmissions were prevented.
I was looking at this paper (for other reasons) and saw that it estimated a mean serial interval of 6.3 days in Shenzhen while there was aggressive testing, contact tracing, and isolating. They report that the mean serial interval was 3.6 days among patients who were infected by someone who was isolated within 2 days of symptom onset, and 8.1 days among patients who were infected by someone who wasn't isolated until 3+ days after symptom onset, for an overall average serial interval of 6.3 in their population. They found R=0.4 - an average of 0.4 known transmissions from each infected person.
Insofar as the virus mostly spreads through presymptomatic transmission in some countries, that's almost certainly because the people with symptoms are all isolated. Symptomatic people definitely spread the disease.
It'd be interesting if the R of these populations were <1 at the time the studies were done, though. If so, presymptomatic transmission might be insufficient to sustain exponential growth, as long as all symptomatic transmission is prevented.
Tapiwa Ganyani et al, the one with quantitative estimates of presymptomatic transmission, used data from Singapore up to Feb 26th and Tianjin up to Feb 27th. Both are cities which seem to have achieved containment over the relevant time period. I'm going to focus on Singapore, because information about what Singapore has been doing is easier to come by.
This article has a graph of Singapore's case counts over the relevant time period, and appears to show it as being contained (R<1) during the relevant time period. This paper estimates the latency between people becoming symptomatic and being isolated in Singapore, and finds that it's about 3 days for local cases as of Feb 26th, but longer on earlier dates. Their endpoint is "hospitalization or quarantine", but reading Singapore's FAQ, it sounds like they have a tiered system with two lesser levels of isolation that the paper doesn't mention: leave of absence (advice to isolate with no legal force) and stay-home notice (which has legal force against leaving the house, but only a non-binding advisory against having visitors). The lesser isolation tiers are for traced contacts, and would be effective at preventing asymptomatic transmission as well.
My impression from all this is that Singapore's measures would have driven symptomatic transmission down more than asymptomatic transmission, but substantially driven down both.
Asymptomatic carriers seem to be rare, though not completely nonexistent.
I think one should expect asymptomatic people to be less infectious than pre-symptomatic people, so it may not be important from a "changing current responses" perspective to understand the prevalence of asymptomatic-ness. That being said, here is a paper that analyzes the Diamond Princess data and claims the asymptomatic rate to be >15%.
I haven't understood their methodology well enough to have a strong opinion about its validity. I think they're using time series data along with some priors about the shape of the distribution of the random variable "number of days between testing positive and showing symptoms, given that the patient will at some point show symptoms" to conclude that a decent fraction of the Diamond Princess people won't show symptoms.
The claim is made in paragraph 5 of the discussion section.
A question I always have about these studies is at what level symptoms are defined and self-reported. E.g. presumably "you have an itchy throat or a mild headache in the morning/mildly increased fever over your baseline" is pre-symptomatic. Self-isolating with mild symptoms is probably hard to measure but can be at least socially enforced.
If the thought here is true does that suggest the actual transmission mechanism is perhaps not yet understood as well?
The claim has been it is not an airborne transmission (which perhaps still makes sense on a technical level) but if infection is occurring without things like people coughing or sneezing (dispelling a larger amount of internal matter than simple breathing) how is it spread?
Is the view that it must be more about things like an infected person touches their mouth in some manner, then touches something else or shakes someone's hand...
What are the confidence internal for the estimates for the serial intervals? That doesn't seem like something that would have been well measured during the outbreaks so far but perhaps I'm not crediting the medical community enough here.
I found this article which makes that claim, and cites the WHO joint mission report. However, I can't find the claim (or the number "25%" used in any capacity) in the cited document at all. I suspect that the article mixed up percent of cases which are asymptomatic with the percent of transmissions which are from asymptomatic sources.