Deaths induced by use of hydroxychloroquine during first Covid wave: an estimate | Hacker News Reader

Deaths induced by use of hydroxychloroquine during first Covid wave: an estimate

sciencedirect.com

14 pointsj4yav7 comments

I wonder how they determined that the patients died from HCQ and not from covid..

The paper describes the methodology: for broad analyses like this you’re looking at whether people in the test group died at a higher rate than their demographics would otherwise suggest rather than digging into an individual patient’s details.

What you will never see /s

“Deaths induced by compassionate use of vaccines during the first COVID-19 wave: an estimate”

“We estimated the worldwide in-hospital mortality attributable to vaccines use by combining the mortality rate, vaccine exposure, number of hospitalised patients, and the increased relative risk of death with vaccines.”

Only 'cause you don't phrase negative numbers like that, the research itself has absolutely been done:

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9848037/

Seems to be modelled from a single estimation model. not hard data, just estimates on estimates using estimates with no true correlation

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

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...

Reply on news.ycombinator.com