Noninferiority trials have serious inherent limitations , the main one being that (unlike in a superiority trial), given a finding of noninferiority, it is impossible to distinguish between true equivalence and poor execution or faulty pretrial a priori assumptions regarding expectations of benefit and event rates. In this particular trial, a 15% absolute difference in the primary outcome was arguably too large to use as a noninferiority threshold. That threshold implies acceptance of 1 additional COPD exacerbation for each 7 patients treated. If a superiority trial showed a week's extra steroids prevented 1 COPD exacerbation for 10 or even 20 patients treated (, a 5-10% noninferiority margin), I think many doctors and patients would endorse longer-course prednisone. This trial (as designed) would fail to detect that.
Register for alerts
If you have registered for alerts, you should use your registered email address as your username