Comment by macksd | Hacker News Reader

Comment by macksd | Hacker News Reader

Parent Full thread

I've heard quite a few stories like this, and anecdata though it may be, I find it very intriguing. How WOULD we know? There seems to me to be a pretty consistent timeline: confirmed cases and rampant spread in other countries, the first confirmed cases in the US and rampant spread among vulnerable populations (nursing homes, etc.), and subsequently the overwhelming of ICUs. Each step seemed to take a few weeks, and progress consistently. There's a lot of suspicion from some that things were misreported, and some of that is the typical lies-with-statistics, etc. but I know enough people who were working in nursing homes and in ICUs that I personally trust the general timeline to be reality.

So what's up with these early possible cases? I wonder how plausible it is that it there was some early spread of the same disease, but vulnerable populations were just luckily avoiding exposure. It really seemed to take off in the US in those nursing homes. IIRC that's what happened in Seattle, and that's what happened in my county. THAT'S what wiped out most of our body count, and that's what initially overwhelmed ICUs. And that's what raised a lot of awareness, and probably a few self-diagnosed false positives too. Could genuine COVID cases have been going around before then and just not getting the attention, or causing as much damage until that point?

No. We have plenty of ways of detecting a large scale Covid epidemic after the fact, even if it ended up somehow avoiding all people from the risk groups. Analysis of historical blood samples for antibodies. Early population-wide antibody tests. Phylogenetic analysis. Wastewater analysis.

Think of just how contagious Covid would have been before we knew about it and started taking precautions. You could only introduce Covid into a country a few times before it'd inevitably start a self-sustaining transmission chain. We have no credible proof of that happening in the west until 2020.

Isn't that what the study is suggesting though? They are looking back (like you point out) and finding it earlier than we all thought?

That is the kind of study they did, yes. And the results are consistent with commonly accepted timeline, not with the "I had Covid in October 2019 but they covered it up" crackpots.

The prevalence they found is extremely low, less than 0.05%, despite their samples going all the way back to Mid-March. Since they did not test on any samples that are definitely pre-Covid (e.g. early 2019, late 2018), we can't calibrate their false positive rate. But if we assume the specificity of the tests was 99% which seems on the high end for antibody tests, and that the false positives are uncorrelated, we're already in the region of false positives feasibly explaining literally all of their samples.

The data is just totally inconsistent with any kind of widespread transmission of Covid in the US in 2019.

How many of your detection mechanisms were actually carried out?

I think the most useful measure that was actually carried out were the handful of places that take samples for flu surveilance that were able to retest the samples for covid. Of course, in many places, you can't actually convince a doctor to take a sample for flu-like illness, so there's no data.

All of them have been done in practice, of course.

>How WOULD we know?

Antibody tests after the fact would reveal whether or not these were Covid cases. You'd have to be tested before the antibodies disappeared, though.

Reply on news.ycombinator.com