,
4 tweets,
1 min read
Read on Twitter
Question for folks about a research "design" that I don't like very much, but I see often enough to wonder how much I'm being a hater:
An investigator want to see whether (Intervention) reduces the incidence of (Outcome). They don't actually have a comparison arm available, just data on patients that received the intervention (for whatever reason).
So instead of testing the incidence of(Outcome) in the (Intervention) against control over the same period of time, they report the outcomes in the X months before starting the intervention against the X months after the intervention.
I don't think this design is very useful for testing whether an intervention 'reduces' the incidence of an outcome, and I think most regard this as 'weak' at best. But, if there's no data available at all, is it better than nothing? Or is this an auto-rejection for you?
