Thread by @GidMK

This is a common misconception that springs up all the time, so let's talk industry bias in scientific research

Does industry funding make a study bad - a tweetorial

https://twitter.com/mabelyang/status/1103491457854423040

The basic idea is similar to the one in the tweet above. You can find it anywhere

Industry does research. But industry wants money. Therefore, their research is hopelessly biased, because they don't do research that doesn't earn them money

The thing is, this isn't just a theoretical question. We can look at industry-funded trials and compare them to non-industry funded ones

There are many ways of grading studies in terms of how "good" they are. One commonly-used (and in many ways gold standard) method is @cochranecollab risk of bias tool methods.cochrane.org/bias/resources…

Bias refers to elements that the study has failed to control for which may have influenced the outcome of the study

For example, if the study had more dropouts in the intervention group, then it may have been biased

So back to the original question

Compared to non-industry trials, are industry trials (in terms of risk of bias)

As I said, this is a question we can actually answer. Risk of bias is a somewhat subjective assessment, but even so we can compute it for different trials and see which ones are better or worse

There are plenty of studies that have done just this

Generally speaking, they find that industry funded trials are either as good as or BETTER than trials funded through other methods bmj.com/content/326/74…

So we know that industry funded trials are usually as good as or better than other research

What does that mean for their results?

To put it another way, are industry funded trials more or less likely to find positive results for the funding body?

Well, we know that they're pretty good trials. If they were all fair comparisons, they should be the same or perhaps less likely

But instead, when we look at the evidence, we find...

Exactly the opposite

Industry funded trials mostly report positive results for the funding body

How can this be????

(I'll give you a hint - the industry is definitely not always right)

A few people have answered, so let's put the cards on the table

That's right, it's publication bias!

If we look at industry funded trials, they are almost never negative (i.e. no benefit for the funding body)

It turns out that often, the negative trials just aren't published!

So individual trials can be very good - meaning that industry funding isn't necessarily an issue for individual trials - but they still make a misleading picture overall

If you're interested in this issue, you should follow @senseaboutsci who are doing great work getting studies registered and published

So ultimately, saying "the industry funded this trial" doesn't necessarily mean much in terms of its quality

What it does say is that you should be looking for more research overall because there may be unpublished work contradicting the findings