This is a thread on the basics of reviewing epidemiological trials. I'll be looking at a few common pitfalls and red flags
-
-
When you see people saying "this research is garbage" without elucidation of HOW and WHY the issue they've raised influences the results, they are talking nonsense
Show this thread -
So, what is bias? There are many and varied types - I could spend another entire thread on just a handful - but the easiest way to think of it is anything that could've influenced the outcome
Show this thread -
So, an example. This is the recent Virta Health trial that has attracted so much attention. If you aren't familiar with the research, you can find it here: https://link.springer.com/article/10.1007%2Fs13300-018-0373-9 …pic.twitter.com/X5SUDmYmdw
Show this thread -
It's a great study to look at, because there are a truly wonderful number of potential sources of bias in the methods Note: I say POTENTIAL sources of bias for a reason. Generally, you can't tell definitively whether things have biased a study without access to the raw data
Show this thread -
Our first example of potential bias: recruitment bias. It is possible - even likely - that these recruitment methods might have influenced the people enrolled in the studypic.twitter.com/O9CJwjCkfY
Show this thread -
-
Another potential source of bias: they allowed patients to self-select into treatment groupspic.twitter.com/vPI2Iah9Tw
Show this thread -
-
The answer to both polls is, of course, all of these things. Most potential sources of bias have many causes and indeed many impacts
Show this thread -
Both of these sources of bias would likely make the treatment group appear better - if they did have an impact - and so reduce the strength of the study's results
Show this thread -
HOWEVER it's important to remember that this is not true for all sources of bias What you need to look for is whether the bias applies equally to all groups, or unequally to one or more groups in the study
Show this thread -
What do I mean by this? Well, here's an example from the recent study on carbs and health in the Lancet: reporting biaspic.twitter.com/ef2Q4Kw7Hg
Show this thread -
We know that people under-report their food intake to medical professionals This is not surprising. No one likes to admit to their doctor that they aren't following the diet they're supposed topic.twitter.com/7MMFIzIlGe
Show this thread -
But, AND THIS IS IMPORTANT, this reporting bias impacts every single person in the study in a similar way Thus, it shouldn't impact the results, except to potentially make them seem LESS meaningful than they actually are
Show this thread -
If someone is bashing a piece of research, take a look at their arguments and think: "Is this really going to influence the results?" "Will it impact all groups equally?" "WHAT impact will it have?" These are all vital questions to answer
Show this thread -
They are also important for you yourself if you want to know more about the research and its usefulness
Show this thread -
This has been my brief thread on bias. If anyone's interested, let me know and I'll do another on different sources of bias or expand on this one
Show this thread
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.