masterofmonkeys

Angy Old Man
10+ Year Member
7+ Year Member
Sep 9, 2008
1,582
257
Status
Attending Physician
I've been grumbling about this for a while and it seems to me that we may place too much store in intention-to-treat analysis.

I think it makes sense at a population level because in real life, people are going to drop out of therapies and they're NOT going to take the medicine. So at the level of writing clinical practice guidelines and estimating population-level health effects it's absolutely a valid and valuable method of assessing the treatment effect of therapy.

However, I've been put out by several studies for therapies that ONLY report using ITT principles and don't give enough info to infer treatment effect in perfectly compliant patients.

To me this is problematic, especially in the case of things like physical therapy, group therapy, or psychotherapy in which there is an additive effect of going to more sessions and/or completing a course.

When a patient comes to me and asks about an intervention, or when I suggest one and he asks how effective it's going to be, I don't want to say "well, based on the fact that there's a 33% chance you're going to stop taking this drug or a 50% chance that you're not going to complete your mindfulness-based stress reduction class, you'll see an improvement of 30%". I'd like to be able to say "People who actually take their meds every day or complete the course see on average a 50% reduction in symptoms".

Thoughts?
 

WellWornLad

10+ Year Member
Feb 5, 2008
1,088
32
Status
Resident [Any Field]
It'd be nice to always have both intent-to-treat and efficacy results, but efficacy is much harder to gauge correctly. It requires a lot of complacency checking, which is not always practical or accurate for large outpatient studies. A lot of efficacy stats get cut after review because the investigators weren't rigorous enough about establishing criteria and checking compliance thoroughly.
 
About the Ads