Thread by @GidMK

A study recently made massive, international news for reportedly showing that normal glasses may protect against COVID-19

Let's do a brief peer-review on twitter, because this is wild 1/n

2/n The study is here. Altmetric of >2,000, hundreds of news articles about it already

jamanetwork.com/journals/jamao…

3/n The basic idea of the study is simple - we know that COVID-19 can be spread through droplets. Sometimes these droplets might go into eyes. Wearing glasses might prevent this, so do people who wear regular corrective glasses get COVID-19 less than people who don't?

4/n Now, usually what you'd do here is identify one group of people with glasses, one without, and follow them up through time to see if the infection rates were different

This is what's known as a cohort study

5/n Another way of doing this research would be to look at people who have caught COVID-19, and another group who haven't, and compare them in terms of the proportion who wear glasses

This is called a case-control study, and is what these researchers appear to have done

6/n Essentially, the cases (COVID-19 patients) wore glasses less than controls (non-COVID-19 patients), which might indicate that glasses prevented COVID-19 infection

So far so good, right?

7/n Well, there are a few immediate and obvious snags

Snag 1: small sample size. We have <300 cases and no information on controls. That's weird, and problematic (usually these studies are much bigger)

8/n Snag 2: The outcome measure isn't wearing glasses. The study actually appears to compare short-sightedness percentages between the cases and controls, which is an obviously imperfect measure

9/n (It's imperfect because not everyone who wears glasses is myopic and not everyone who is myopic wears glasses >8hrs per day)

10/n But then we hit the biggest snag of all

The control population

You see, the authors didn't actually collect a control sample. The rate of myopia here is from ANOTHER STUDY ENTIRELY

11/n Not only that, but the study was conducted IN 1987 and the extrapolation is just weirdly moved forward from then

This makes NO SENSE AT ALL

12/n I honestly don't know what to say here. They're comparing the rate of myopia in a group of hospitalized patients with COVID-19 to the research findings of physical fitness of students from 1987

You can't draw any meaning from that!

13/n These two groups are so wildly different that it's entirely impossible to say whether the glasses had the slightest impact on anything at all, it's just very strange speculation

14/n Maybe glasses DO prevent COVID-19 transmission, but based on this research we have absolutely no idea if that's true

15/n The ABSOLUTE BEST you can say from this research is that it appears that fewer patients diagnosed with COVID-19 from a single hospital in Hubei had myopia than some high-school students in '87

It's not even about glasses!