Thursday, February 22, 2007

TORCH trial

The TORCH trial is the lead article on today's NEJM. No statistically significant change in mortality with Advair but reduced annual rate of exacerbations and improved health status and spirometric values. There was also an increase in pneumonias in the fluticasone-treated groups (FP alone or Advair).
How do you think this data will affect your prescribing patterns for ICS in COPD?

4 comments - CLICK HERE to read & add your own!:

Anonymous said...

I like the transparency of the study. Anyway I have a different opinion from the editorial. I believe there is a survival benefit with SFC based on this study. They just did not enroll enough patients or too many dropouts in the placebo group may have skewed the study results.

My take is, in a moderate to severe COPD, the choice of inhaler is
1. Advair 500/50
2. Salmeterol
And never use fluticasone alone.

What I would do in the real world is start with Advair and add on tiotropium if necessary based on an early release on Annals.

The true winner in the Wonderland of Alice is GSK. Dodos went extinct a long time ago anyway.

Anonymous said...

The either/or significance testing based on the idea that a p value of 0.052 is insignificant, but 0.049 is significant seems out of context. Based on this study, the combination treatment group was likely to have a modest mortality reduction compared to the placebo group. The borderline p value raises the likelihood that there was no difference between groups, but to make a crude analogy, if you were a placing bets in Vegas on death or no death in COPD patients, where would you put your money?

Furthermore, aside from the pneumonia finding, there is not a major downside to the treatment. If we were talking about something more risky, like transplant surgery for COPD, then this p value would more understandably be unacceptable.

Anonymous said...

Whoever wrote the comment above, you should send it to Jeff Drazen. So he could use a better editorialist next time.

Jeff H said...

I agree that a set point p of less than 0.05 as "significant" is by definition arbitrary. So we accept a 4.9% chance of "no difference" as more statistically meaningful than a 5.2% chance that there is really no difference between treatments. Realistically, these numbers are no different. Nevertheless, I'd congratulate the authors for the design and execution of a well done trial. And while it did not meet their a priori definition of "significance" for the primary endpoint of mortality, we are all given some direction here.

And while the absolute differences in mortality and in the secondary endpoints are small, given the prevelence of disease and the number neaded to treat, there is potential for a large impact. If we saw a similar decrease in mortality for treatment of hypertension or breast cancer, it would be all over the news.

I think another positive point can be raised. The mortality in the control group was much lower than the estimates used for the study design. Although this led to the study being under-powered, it also suggests that our overall care of patients with moderate and severe COPD (as the patients in this study were) may have improved with the current standard of care.

At the bottom line, this trial will give amunition to support whichever treatment approach one chooses. Those biased against using the combination inhaled steroid and LABA will use the lack of "significance" to support their arguement, while those who use the combination therapy will cite the "near-significance" and the impact on secondary endpoints to support continued use.

So, unfortunately, I think that this well done, important, and long awaited trial will have very little impact on daily practice.