CTCA Studies Are Not Externally Valid

This is a multicenter study from Canada that looked at the diagnostic accuracy of computed tomographic coronary angiography using invasive coronary angiography as the gold standard - and they found that it's not bad.  Specifically, they found it was not bad at one of their four centers used in the study, and terrible at three of the four centers used in the study.  In a patient population with a pretest probability of CAD less than 50%, the AUC for CTCA was 0.951 at center 1, and 0.597 at centers 2, 3, and 4 combined.

So, clearly, the most important factor affecting the results of your CTCA is your institution's skill at performing and interpreting the test.  Which, if you take it one step further, means that unless your institution is a CTCA center of excellence like the ones pumping out the CTCA studies, you can't apply their results to your practice.  Specificity stays reasonable, but you lose a lot of sensitivity - and when the CTCA for low-risk rapid rule-out is predicated on the high NPV, you can't afford to lose sensitivity.

"Ontario Multidetector Computed Tomographic Coronary Angiography Study"
www.ncbi.nlm.nih.gov/pubmed/21403014