How does Cochrane normally handle wide confidence intervals?
Editor and chief Soares disgracefully caved to pressure
Our new pre-print is out, let me walk you through it.
When I read the review by Jefferson and colleagues on physical interventions to prevent the spread of respiratory viruses, and I saw that pooled randomized trials failed to show masks slowed the spread. My conclusion was simple: there is no evidence masks work.
That's the exact same conclusion that Tom Jefferson reached, and he said as much in an interview.
Why did Jefferson and I agree? Because the standard in medicine is if you run many randomized studies and you get a null result, your practice doesn't work. That's how it's always been.
The Cochrane editor-in-chief, and many internet pundits, did not agree. They said the evidence was compatible with benefit. Why? Because the confidence interval is wide.
The question then becomes: do they say that every other time this has happened? Are they consistent?
We sought to investigate.
We pulled the last 20 Cochrane reviews that all have similar results, and I'm going to show you a plot of all of them here.
As you can clearly see: The Jefferson review (*) is one of many. It falls in line with the rest of them.
So what sort of language do the rest of the reviews use? Here are some quotes:
And you can read more in the pre-print.
What's the point?
What am I to think if the only time you say the wide confidence interval means it's inconclusive is for the one thing that you have a religious devotion to? You literally never say it elsewhere. It speaks so poorly of you to be inconsistent.
The editor-in-chief of Cochrane disgraced herself by caving into pressure from mask zealots. She is not a consistent actor.
Ultimately, masking was as simple empirical question that could have been settled with even more randomized trials, which would tighten up that confidence interval… probably around null, nobody wanted to run those studies.
Now, they invent a new way to talk about negative results. Pathetic.