Thread by @lymanstoneky

This study of long COVID in Norway has a nice design: PCR tests, long follow-up, large sample, and a plausible mechanism for identifying cases: a huge pre-existing panel survey using frequent mobile-device questionnaires. medrxiv.org/content/10.110…

My only real critique is that while they do have about 80,000 people in the sample, it's Norway, so they only managed to find about 800 COVID cases. lol. problems of low attack rate!

But, among those 800, they found that no single symptom of "long COVID" occurred among more than 22% of infected people. The most common symptoms among COVID+ people were loss of taste/smell and fatigue. Loss of taste was NOT correlated with other cognitive symptoms.

Overall, about 57% of COVID+ reported at least one new symptom in the post-acute period, whereas only about 21% of non-infected people reported such new symptoms. That said, most people reported just 1 or 2 of 22 different "long COVID" symptoms.

Also, worth noting this wasn't blind: people obviously know if they tested positive or not, so you can't rule out changes in self-perception.

Regardless, the interesting bit is on severity.

How common were "long COVID" symptoms among people with *mild* COVID cases? That's what we really want to know. And it turns out, "Moderate-to-severe" COVID cases had about 2-6 times as many "long COVID" symptoms. They lump moderate and severe due to sample size issues.

So take so take a common long-COVID symptom like chronic cough. About 2% of uninfected people developed such a cough. Among 3% of mild cases did so. But about 9% of moderate-to-severe cases did. So case severity drove long-COVID symptoms.

Or consider headaches. About 7% of the uninfected population reported developing headaches, vs. 9% of mild cases, vs. *19%* of severe cases.

Some conditions saw bigger gaps between uninfected and mild: brain fog went 4%, 9%, 18%. Fatigue went 4%, 10%, 27%.

So mild cases are associated with some increase in self-reported symptoms, with the biggest increases in harder-to-identify symptoms. Moreover, symptoms are not highly correlated with each other: people who got one symptom were not extremely likely to get others.

The authors go so far as to suggest this lack of symptom correlation calls into question the existence of "long COVID" as a single syndrome: rather, it seems to suggest that as case severity rises, some individual symptoms for some bad cases become recurrent.

But it's not a coherent post-viral syndrome. It's just that some people experience recurrences of specific symptoms for several months afterwards. But they're not likely to develop more symptoms or see increasing severity.

This is a nice study that to me shows pretty clearly that people with a positive test experience worse subjective health across several plausible symptoms for several months. Some-version-of Long COVID is real.

That said, it seems like this is mostly driven by severe cases, and even among severe cases it's a small minority who get any given symptom, and the vast majority of "long COVID" symptoms are mild discomforts which are common in the general population.

Which is to say, it doesn't seem like thre's any evidence of e.g. long COVID excess mortality or even disability. So this nudges my priors towards thinking long COVID is real, but also nudges them towards thinking it's not a big enough issue to factor into public health decisions

0.5% of adults getting a headache is not a high price to pay for keeping kids in school.