BCIS estimate the MACCE rate for each patient based on parameters which are reported in a study from data obtained some 10 years ago called NWQIP. This model predicts, apparently, with 70% accuracy according to the patient's age, sex, emergent PCI, urgency of treatment, cardiogenic shock and whether a bypass graft or the left main stem was intervened on. From this BCIS estimate the 95% confidence intervals (CI) of the predicted MACCE and provided the cardiologists actual MACCE rate is lower than the upper CI then everything is apparently fine.
So statistically if there were 1000 cardiologists you would expect that by chance alone then 25 of them would be above the upper 95% confidence interval and 25 would lie below the lower confidence interval.
So what is the chance that not one of the 557 cardiologists in the BCIS report would be outside the top 95% confidence interval? Well its a very small number (actually p=0.000000751). So currently this is an exam which a cardiologist has a 1 in 100 million chance of failing.
Now as BCIS say you can't directly compared different operators but what you can do is look at people who have a similar practice. If you pool data from cardiologists who do more than 75 angioplasty procedures per year (the BCIS recommended minimum) and who do more than 10 primary PCI procedures (treating patients with acute heart attacks who are the sickest patients) then the average MACCE rate is 2.2%. This is much lower than 8.2% MACCE rate which is predicted by the BCIS model.
Although we might like to regard interventional cardiologists are superhuman creatures what this data really tells us is that the model BCIS has used to predict risk vastly over predicts the actual MACCE making this truly an exam which cannot be failed. To be fair BCIS do acknowledge the limitations of the data but there is an urgent need for more contemporary risk prediction models going forward.