Thursday, December 27, 2012

Day 193 - non inferiority

The highlight of today was helping Emily buy a pair of her own Dansko's. Otherwise, I helped Ellie survive her cold and had coffee with Patrick. Tomorrow I go back to work again unless the snow keeps me in.

Since today was not a very exciting day, I'll finish telling you about the Dec 26 JAMA. There were a bunch of articles that I found uninteresting and two that I thought were really interesting. There was one about how to use a noninferiority trial and a clinical crossroads on varicose veins. I had not really thought about noninferiority trials in any organized way so this was helpful to me. The basic idea is that the new treatment is compared to the old treatment in terms of efficacy and in order to be "non-inferior," it needs to be within a pre-specified range of efficacy. The efficacy range is determined by deciding how much of an improvement the new treatment is over the old in terms of convenience or acceptability to patients. When I learned about non-inferiority trials as a resident, our local statistician told us that a trial that was designed as a non-inferiority trial could not show superiority, that non-inferiority was the best it could do. This article states otherwise, but does talk about how a failed superiority trial can be recast as a successful non-inferiority trial. It is against the "rules," but not all journals go back and check the design of the trial as submitted before it was started and not all trials even are entered in the clinical trial registry although they are supposed to be. I think what this means is that non-inferiority trials from non-first tier journals are suspect.

They also talk about the importance of the "range of efficacy." For instance, in one trial, a new beta blocker was found to be non-inferior to an old ace inhibitor for cardiovascular death or hospitalization. The range of efficacy was 5%. The burden of taking a beta blocker might actually be higher than that of an ace inhibitor and cardiovascular death is a big-ticket item so a 5% non-inferiority margin is not really appropriate. This echoes what I remember being taught in residency about reading old fashioned superiority trials, that before you read the results section of the abstract even, you have to decide what amount of difference between the two arms would be enough for you to care about. Each new fact takes up valuable real estate in your brain and time spent reading and evaluating study A cannot be spent reading study B or doing homework with kids or cooking cheese puffs so it's not a "no-brainer" decision. Some studies report differences that are statistically significant, but might not be enough of a difference to make it worth your while to read them. These studies ask the same of us. One has to decide what would constitute clinical equivalence. If the burden is not great, a 5% decrease in efficacy is too high a price. If the burden of the old treatment is huge or the clinical outcome in question is trivial enough, a larger decrease in efficacy might be acceptable.

The other article I found interesting was a clinical review of varicose veins. I don't remember being taught a single word on varicose veins and they are pretty darned common so it was nice to have a primer.

Want to see a recent picture of me? Sure you do.

By the way, here is the Christmas present I was working on a month ago or so. It was for Terry and he liked it. Ellie made the ceramic mug in front of it. Pretty nifty, huh?

For me, for tomorrow, I will wish for a safe commute. For you, I will wish safety in the storm, regardless of your interaction with it.


  1. You look fabulous!!! The picture you have woven for Terry is amazing.. Going to Vermont for the weekend hope to get together next week. Hope Elie is getting better anecdotal Emily is enjoying her time at home. Be well and stay safe!

  2. Thank you! Have fun on the farm.