EXAS statistics vs statistics used in traditional colonoscopy or non-invasive colonoscopy:
When comparing traditional colonoscopy vs non-invasive colonoscopy I believe that they are comparing number of missed individual polyps etc.
But while EXAS statistics sound the same I suspect they are, de facto, about the whole patient. For example, imagine the following:
a) 70% of patients with any polyps have more than 1.
b) Virtual colonoscopy catches 60% of all polyps - so in the 30% of patients with just one polyp they catch it 60% of the time.
c) But if EXAS is making their calc on a per-patient basis then they might well be catching almost none of the patients with single polyps less than 1 cm and yet still claim they have a sensitivity of 60% - because they catch 90% of the patients with multiple polyps. Obviously this is a worse outcome for the patients (since the only goal of virtual colonoscopy or Cologuard is to trigger a real colonoscopy and excision.)
Note that, in fact, I expect some variant of #c to be true - that Cologuard is much more likely to catch a patient with 4 polyps less than 1 cm than it is to catch a patient with just 1 polyp. Thus I would expect that for similar sounding statistics the cologuard is not as good as virtual colonoscopy. The only question is how much so.