asymptomatic and pre-sympyomatic spread of Covid is statistically 0%. According to the study it isn’t happening.
I'll point out, first, that the claim that presymptomatic transmission doesn't occur is totally false. The authors of the UF study you cite themselves concede that it does occur, and they could hardly say otherwise, given the substantial evidence (which i discussed upthread) that shows that infectiousness peaks before the emergence of symptoms in people who develop the latter.
With regard to asymptomatics--people who never develop any symptoms--I begin by pointing out that the UF study was a meta-analysis, and only 4 studies were used as a source of data for asymptomatics. This is a problem with such studies. The sample sizes are generally very small.
In fact, I looked at those four studies. One reported only a single asymptomatic, who had just two close contacts. Two of the other studies (one of which was the famous Korean call center super spreading event) reported just four asymptomatics each; in one of those studies, one of the asymptomatics did transmit the virus to a household contact. The authors even conclude that this confirms that asymptomatic transmission is possible.
The fourth study included a total of 31 asymptomatics or presymptomatics, and for this relatively larger sample size (about 80% of all the subjects the meta-analysis based its conclusion on), there were numerous transmission events. In fact, what the authors of this fourth study describe as the crude attack rate (transmission events/number of close contacts) was not different from that of symptomatics. However, the authors note that the household attack rates (which is what the UF study is focussed on) were much lower for asymptomatics. Outside the household, symptomatic transmission, relative to asymptomatic transmission, goes down, because symptoms predispose people to wearing masks, social distancing, and other behavioral measures that would reduce spread. So this study found that asymptomatic spread was in fact very important outside of households. It was less important within households, but they still found that it was significant:
Nonetheless, an AR [attack rate] of 4.4% and 6.1% in asymptomatic and presymptomatic cases, respectively, is not negligible.
I repeat: this is the conclusion of one of four studies that are the basis for the UF meta-analysis, and this study contains about 80% of all subjects. So how did the UF authors conclude there is virtually no asymptomatic transmission? They have performed what seems like magic. They determine the attack rate of symptomatics in each individual study--infections caused/number of contacts--and through a technique called restricted maximum likelihood, come up with a weighted average of 18%. If you just use the raw data--add up all the infections, and divide by the total of contacts--you get about 12%. So the weighted technique bumps up the rate by about 50%. This is understandable. Some studies, for various reasons, are thought to count more than others. The technique is performed on about half a dozen other symptomatic data sets, and each time it bumps up the raw rate, but not too much, never more than 50%.
But if you perform the same analysis on the asymptomatics, something different happens. The raw rate is 4.5%, but after the statistical technique, it's bumped down, and way, way down: to 0.7%. Why the difference? Why does it bump the symptomatic rate up, and the asymptomatic rate way down?
This is above my pay grade, but it seems that the technique sees that two of the four asymptomatic studies had no infections, and somehow incorporates this heavily into the analysis. I really don't understand this, but JAMA is a medical journal, not a mathematical one. Most of its readers are doctors. If I don't understand a statistical tool, you can be sure most of those doctors don't, either. And for something that has such a critical effect--turning a robust rate into virtually nothing--an explanation is owed. In any case, their conclusion remains starkly at odds with the one study that is the source of most of their data. Obviously, those authors saw no reason to alter their attack rate so dramatically.
The UF authors cite two other studies as evidence that asymptomatic transmission is rare (not included in the meta-analysis, I assume because they couldn't determine when infection took place). One of these studies did report a very low attack rate; there was just one incident of asymptomatic transmission. But the other study reported an attack rate of asymptomatics that was about the same as that of symptomatics. The authors of the UF meta-analysis also cite evidence that presymptomatic transmission occurs; as I said earlier, they could hardly claim otherwise. I'll just add that many other studies have reported transmission by asymnptomatics.
Moving on from meta-analysis, of the single studies finding no transmission by asymptomatics, this is probably the best, at least in terms of sample size (it was not included in the UF meta-anaysis; again, I assume because infection times could not be identified):
Large-scale population screening can provide insights to levels of ongoing SARS-CoV-2 transmission. Here, the authors report a citywide screening of ~10,000,000 residents of Wuhan and show that SARS-CoV-2 infection prevalence was very low five to eight weeks after the end of lockdown.
www.nature.com
This was carried out in Wuhan last summer, where most of the more than 10 million residents were tested, and there were only 300 asymptomatic cases. None of them had active virus. So asymptomatic is simply defined as someone who tests positive by PCR, but has no symptoms. But someone who tests positive by PCR might have non-functional virus or viral fragments for some time after initially being infected, and might have been infectious during an earlier period. Studies in China have actually shown that viral fragments can be found in some people as long as five weeks after symptoms subside. People like this might not be common, but 300/10 million is definitely not common. The researchers may have just identified the relatively rare individuals who retain viral sequence some time after the infection has been resolved. I note that more of them were in older people, who you would expect would have more trouble eradicating the virus, even if it was no longer infectious. What you really want to know is whether asymptomatics are infectious within a week or so after having been infected. The studies included in the UF meta-analysis did address this issue, but as I noted earlier, these studies included very few asymptomatics, and the one that had the largest sample size found there was significant transmission, and emphasized it.
There were also no symptomatic cases in the entire nearly 10 million pool of tested individuals, so there isn't even a control for the conclusion that asymptomatics don't spread the virus. This study didn't find any spread, period.
We believe these data, together with results of several other studies, suggest that SARS-CoV-2 infection provides strong immunity to reinfection that lasts for at least several months.
Yes, that is consistent with emerging findings. Some people, we know, have been infected a second time, but they seem to be fairly rare.
Facebook
Twitter
Instagram