What they said was that in 33% of cases, even with all the test...

What they said was that in 33% of cases, even with all the test results, xrays etc, two doctors could not agree on the affliction, so they had to get a third doctor to get a majority.  Assuming the third doctor went 50-50, This means that potentially RAP could have been 15% more accurate (as 33% @50% correct/ 50% incorrect = 15% ) BUT lets not go there.

What this shows is that in 33% of cases, RAP got 7x% and a single  Doctor got 5x% .  (based on the 2nd doctor not agreeing, if you went and saw one doctor/GP (normal visit) then that single doctor was incorrect in 50% of 33% of the cases even when they had all the tests (which they normally don't have initially) .  With RAP, based on cough alone, RAP got it right (verified right) 7x% of the time at it's worst.  

The patient would be better off having an initial screen with RAP and a follow up for a negative with a doctor just to be sure.