This is much more likely to be due to people with more severe disease being more likely to be noticed in the early stages of an outbreak, while in the later stages everyone including those with mild disease are noticed.
I re-read the studies originally included in this post looking for data relating illness severity to fever. Two of the studies compared the rate of fever in ICU vs non-ICU COVID-19 patients. This one finds 13⁄13 in ICU and 27⁄28 in non-ICU have fever; this one finds 36⁄36 in ICU and 100⁄102 non-ICU had fever. This study broke down its patients into a hospitalized subset in which 6⁄6 had fever and a non-hospitals subset in which 58.3% had fever; I had mis-read this one on the first pass when writing the post, and wrote down the percent with fever in the hospitalized subset (100%).
This does seem to be consistent with fever being more common among the subset of patients which are severe enough to be hospitalized. I have edited the post to reflect this interpretation, and corrected the one study that was mis-extracted; you can see the old version of the post here. This weakens my belief that the rate of fever at onset has already declined, though I still think it is somewhat likely, and I still expect it to decline in the future.
This is much more likely to be due to people with more severe disease being more likely to be noticed in the early stages of an outbreak, while in the later stages everyone including those with mild disease are noticed.
I re-read the studies originally included in this post looking for data relating illness severity to fever. Two of the studies compared the rate of fever in ICU vs non-ICU COVID-19 patients. This one finds 13⁄13 in ICU and 27⁄28 in non-ICU have fever; this one finds 36⁄36 in ICU and 100⁄102 non-ICU had fever. This study broke down its patients into a hospitalized subset in which 6⁄6 had fever and a non-hospitals subset in which 58.3% had fever; I had mis-read this one on the first pass when writing the post, and wrote down the percent with fever in the hospitalized subset (100%).
This does seem to be consistent with fever being more common among the subset of patients which are severe enough to be hospitalized. I have edited the post to reflect this interpretation, and corrected the one study that was mis-extracted; you can see the old version of the post here. This weakens my belief that the rate of fever at onset has already declined, though I still think it is somewhat likely, and I still expect it to decline in the future.
If that’s the reason, then the same study-series should also have a declining rate of ARDS. I’ll check.