Hm, I think that article references this paper[1]. It looks like their "control group" was patients who were tested for COVID-19 for any reason. They then check

Hm, I think that article references this paper[1]. It looks like their "control group" was patients who were tested for COVID-19 for any reason. They then checked whether the age distributions and distributions of Charleson Comorbidity Scores were the same for people who tested positive vs those who tested negative (of positive tests, 65.3% were under 65 vs 64.5% of negative tests, 53.5% had a comorbidity score of 0 or 1 among positives vs 49.4% among negatives). However, they then checked the age distributions between mild and severe cases, found very large differences (54.5% of severe cases were over age 65 vs 20.3% of mild cases, and 74.2% of severe cases had a comorbidity score of 2 or more vs 26.4% of mild cases)[2]. And then if I'm reading correctly, they just decided that was fine and they'd calculate the hazard ratios for mild vs severe COVID without adjusting for underlying confounders anyway? Their methodology doesn't look super sketchy when it comes to COVID vs no COVID, but I am not sure what they're trying to do by comparing mild with severe COVID without even controlling for counfounding factors. [1] https://www.frontiersin.org/articles/10.3389/fmed.2021.778434/full https://www.frontiersin.org/articles/10.3389/fmed.2021.77843... [2] https://www.frontiersin.org/files/Articles/778434/fmed-08-778434-HTML-r1/image_m/fmed-08-778434-t002.jpg https://www.frontiersin.org/files/Articles/778434/fmed-08-77...