Just read a slew of industry-sponsored trials on a weightloss medication and - surprise! - even with >40% dropout, they all used LOCF In my opinion, that makes the results highly suspect
-
-
Show this threadThanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Only as part of a sensitivity analysis, to see how much the missing data could affect your results
-
That's what I thought! But these studies have done the opposite - sensitivity analyses using multiple imputation or ITT with attenuated effects, while main reported analysis is LOCF
End of conversation
New conversation -
-
-
What about for a confounder? I’ve done this before for neighborhood income. If missing in the census year closest to exposure measurement, I imputed using adjacent year when available. Is this a methods pitfall I didn’t know of? Enlighten me, pls
#EpiTwitter!Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Yes.
But do sens analyses.
Also note that the pharma/regulatory standard is that the primary analysis should be robust and conservative for the purpose at hand, not unbiased under ideal conditions of correct imputation model specification, which sees to be the epi standard. -
You are implying that LOCF is a robust and conservative approach?
- Show replies
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.