The revival anecdote that died
New BMJ paper on estimands
The BMJ has just published our ‘primer’ paper on estimands1. It was fun to write and involved several fiery but good-natured arguments about details ranging from ‘quite important’ to ‘incredibly trivial’ (mainly between Brennan, Suzie and me).
When the paper was accepted, BMJ gave us the opportunity to write an accompanying opinion piece2. I suspect because their core readership don’t automatically read articles in their Research Methods and Reporting section, so these opinion pieces are hooks intended to draw in these folks.
I proposed that we being the opinion piece with (something like):
Have you ever revived the dead? I have. In fact, I’ve done so many times.
This was the preface to an anecdote that, when writing the SAP (statistical analysis plan) for a trial, Brennan had been shocked to discover his planned method of analysis would (implicitly) impute outcome data for participants who had in fact died before the end of planned follow-up. Reviving the dead! Since ‘revival’ has Christian connotations3, I’ll call this his scales-from-eyes moment4: So that’s why people go on about the importance of estimands!
Anyway, I thought that opening would be a funny flex from statisticians in a journal with a nominally5 medical audience. ‘We’ (😉) eventually decided against it, as you can see. To make up for that, I’m making this whole post revolve around it. You, dear reader, are clearly a connoisseur who will appreciate it… ok, ok, there’s no fooling you – I confess that this post is a second hook-piece. I’m talking about the whole revival anecdote but my real purpose is shamelessly promotion of our new paper. There’s no fooling you!
Now that my cards are on the table, please read and share the paper.
B. C. Kahan, J. Hindley, M. Edwards, S. Cro, and T. P. Morris. The estimands framework: A primer on the ICH E9(R1) addendum. BMJ. 2024; vol 384. doi:10.1136/bmj-2023-076316.
B. C. Kahan, T. P. Morris, S. Cro. We must let the research question drive study methods. BMJ. 2024; vol 384. doi:10.1136/bmj.q173.
I nearly went for reanimation which has Frankenstein connotations.
Interestingly, a lot of statisticians and epidemiologists read this series and find them an invaluable source of information on methods. I read them for things I know little about. Presumably it’s successful in this regard because the pieces are supposed to be targeted at the BMJ’s core readership.