3 ms·

this is so dumb - first of all, patients weren't randomized to drug vs placebo, so almost certainly, it was the sicker patients who got the drug. If you repeate

this is so dumb - first of all, patients weren't randomized to drug vs placebo, so almost certainly, it was the sicker patients who got the drug. If you repeated the analysis with “hospital admission” as the variable of interest, you’d find… ”deaths induced by admission to hospital during first covid wave”, somehow I don’t think the authors would conclude that admitting less of these patients to hospital would be beneficial. secondly, “We did not consider studies containing only ICU patients because too many external parameters could influence the results” - this also makes no sense, if there is ANY group where the benefit might outweigh the small but non-zero risk of harm from the drug, its in the sickest patients, so why would you exclude those patients unless you’re trying to get a predetermined result? After all, all of the studies and patients had “many external parameters could influence the results” A great example of how to mislead with statistics

refulgentis 3y ago
No misleading: TFA covers this. They did the thing you complain about in the second paragraph because of your complaint in the first paragraph. If you include ICU patients then you're biasing towards hydroxy experimental prescription that had bad outcomes. From MD friend: "Compassionate use" is a term of art that means "we gave it to em because we ran out of other options and family consented to the treatment with known downsides", so excluding ICU prevents a complaint that it's obviously biased because patients that were obviously more likely to die were also more likely to get hcq. MD friend also says you can't do randomized trials with known-toxic treatments, and the implicit "they took it versus didn't" is the control vs. placebo. Not randomizing doesn't mean there isn't any control vs. placebo, and not having a literal placebo pill biases the results to being positive for hydroxy. My takeaway is they stacked the deck for HCQ to prevent self-peasantization, and yet...

refulgentis 3y ago

No misleading: TFA covers this. They did the thing you complain about in the second paragraph because of your complaint in the first paragraph. If you include ICU patients then you're biasing towards hydroxy experimental prescription that had bad outcomes. From MD friend: "Compassionate use" is a term of art that means "we gave it to em because we ran out of other options and family consented to the treatment with known downsides", so excluding ICU prevents a complaint that it's obviously biased because patients that were obviously more likely to die were also more likely to get hcq. MD friend also says you can't do randomized trials with known-toxic treatments, and the implicit "they took it versus didn't" is the control vs. placebo. Not randomizing doesn't mean there isn't any control vs. placebo, and not having a literal placebo pill biases the results to being positive for hydroxy. My takeaway is they stacked the deck for HCQ to prevent self-peasantization, and yet...