Preliminary results of the RESPECT study are in, and from now on, every stroke patient with a PFO must get it closed.
Well, maybe not. Here’s what we know so far (data here–please take a look for yourself):
In the primary intention-to-treat analysis, the risk of stroke over an average of 2.1 years of follow-up was 1.8% in the closure arm vs. 3.3% in the medical arm. Relative risk was 0.53; 95% CI 0.23-1.2; p=0.16. That is, the study at first blush appears to have failed.
But wait! This primary analysis was deemed invalid due to an imbalance in the number of dropouts in the medical arm (people going for off-protocol PFO closure). Per pre-specified rules, they re-analyzed the data using a survival analysis (i.e., how long do patients survive free of outcome events? Also known as Kaplan-Meier curves). Using an intention-to-treat analysis, there was a “trend” toward better outcome in the closure arm, but the confidence intervals still crossed null. Using a per-protocol analysis, the effect looked more robust. Using an as-treated analysis, the effect looked even more robust.
So, if you focus just on the people who actually stuck with their treatment assignment–didn’t drop out, didn’t cross over and get their PFO closed, etc.–there does seem to be an effect. The relative risk reduction is actually quite large, although the absolute risk reductions are small.
There were some interesting exploratory analyses as well. First, it appears that having a larger shunt or an atrial septal aneurysm was associated with greater benefit from PFO closure. Second, recurrent strokes in the medical arm tended to be larger than those in the closure arm.
Expect a whole bunch of hoopla about this, but I recommend not getting too worked up until we have a published manuscript to review and clinical trialists smarter than I have the chance to review the statistical methods.