Thread by @GidMK

So, one of the biggest studies to date on ivermectin for COVID-19 has issues significant enough that, if not fraud, are so serious that it invalidates the study without further explanation

I promised a thread

Here we go 1/n

https://twitter.com/GidMK/status/1415764372362649601

2/n Firstly, none of this would've been possible without @JackMLawrence, who uncovered this in the first place and investigated it himself well before any of us gronks got on board

Read his piece here: grftr.news/why-was-a-majo…

3/n The basic story is pretty astonishing. I've previously written about this study, Elgazzar et al, and why there are some indications that it's low-quality and potentially very unreliable

https://twitter.com/GidMK/status/1407140636281638912?s=20

4/n But @JackMLawrence noticed something weirder when he read the paper

It appears that pretty much the entire introduction is plagiarized

That's...not great

Left is Elgazzar et al, right another study published 3 months before

5/n But that's just the start of this wild story

You see, the authors had shared their data online

They locked it, but our tenacious hero @JackMLawrence figured out the password - 1234

6/n To cut a very long story short, the file uploaded by the authors as the data for the study is filled with errors, and certainly not data collected from real patients in a real study

@sTeamTraen with the in-depth analysis here:
steamtraen.blogspot.com/2021/07/Some-p…

7/n One tiny example of the issues with the data - most of the patients in one of the control groups are clones of each other, with very minor changes

This is a very common feature of scientific fraud

8/n Moreover, the study is meant to have started on the 8th of June according to the authors

Of the people who died, fully 1/3 had been hospitalized and died BEFORE this date

That's...bad

9/n You can actually download the file yourself to check, although the website is a morass of malware so it's a bit dicey

10/n So is this the data file used in this study?

Well...we don't know. Maybe

11/n It's certainly what the authors uploaded, and in many ways matches their results, but it is also inconsistent in some other ways that make the whole thing extraordinarily confusing

12/n Problem is, even if this ISN'T the data that the authors actually collected (assuming they collected any data at all), there are still issues in the study

For example, they used the wrong statistical tests several times

13/n Their statistical methodology is...frankly really bizarre

They had SPSS but looked up p-values by hand in paper tables? Wild

14/n And let's not forget, pretty much the entire introduction WAS PLAGIARIZED

Not good. Not good at all

14.5/n Is the study fraud?

We may never know. It is really, really hard to prove anything in cases like these, and unless the authors go public we might never know what actually happened beyond the issues that we've seen so far

15/n What does this mean more broadly?

Well, here's where the story gets both more interesting and somewhat darker

This is currently THE BIGGEST RCT of ivermectin. It shows a mortality benefit of 90%

That's huge

16/n Being such a big study, it has been included in multiple meta-analyses. With such a vast benefit, it has a very large influence on the meta-analytic results

17/n Indeed, removing ~just this single piece of research~ from recent meta-analyses either mostly or entirely overturns the positive results that they found for ivermectin 👀👀👀

18/n Worse still, the paper has been viewed over 130,000 times, and the comments are filled with medical doctors praising the authors and recommending ivermectin based on their results

19/n In other words, this single study has drive ivermectin as a treatment to thousands, perhaps millions, of people

And it is either so flawed as to be totally unreliable, or potentially outright fraud

20/n But the story doesn't stop there. Remember, several meta-analyses have included this study in their results. Why would they do that?

Well, they thought it was at LOW risk of bias (i.e. high-quality)

21/n The pretty obvious issue we now have is that this study is clearly filled with problems, yet it's been included in meta-analyses that considered it to be a piece of top-quality research

22/n At the absolute minimum, studies including this paper should revise their main analysis to exclude it until we have a reasonable explanation for all these issues

23/n But we're left with the depressing realization that somehow EVERYONE MISSED THIS

I reviewed the study. While I noted concerns with the way it was reported, I never even checked to see if it was fraud

24/n How can it be that a study this problematic was used as evidence to treat 1,000s of people since November 2020, and no one noticed? No one cared?

25/n Anyway, the TL:DR is that the largest study to date of ivermectin for COVID-19, which found a HUGE benefit for the drug, has just been retracted amid very serious concerns about plagiarism and fraud

This will echo in the scientific community for years to come

26/n While it is still certainly possible that ivermectin works for COVID-19, this has made a huge dent in that possibility. I await the large studies that are currently being conducted, because we really have no good evidence to rely on