Using the Appropriate Fishing Net for Computed Tomographic Coronary Angiography in Daily Clinical Practice

We read with interest the Invited Commentary by Diamond and Kaul.¹ The authors correctly noted the variability in accuracy of computed tomographic coronary angiography (CTCA) across the different centers.² The Invited Commentary attributes substantial distortions in the operating characteristics to diagnostic verification bias. Although we agree that this bias is true for other studies where invasive coronary angiography (ICA) referrals are influenced by CTCA results, this was not an issue in our study. In the Ontario Multidetector Computed Tomographic Coronary Angiography Study (OMCAS), all screened and enrolled patients underwent ICA irrespective of CTCA results.² The results of CTCA were blinded to patients and treating physicians, and therefore we purposely avoided the possibility of verification bias. There is no need to correct for verification bias. Rather, we were intrigued that the authors proposed an optimal patient population, in particular patients with a pretest probability between 0.10 and 0.86. In fact, our patient population was purposely enrolled with this range of probability of coronary artery disease and used the previously published assessment of probability of coronary artery disease by Diamond and Forrester³ to make this determination. This suggests that CTA may be applicable in a large population of symptomatic patients. Our principal approach was to define explicitly the “net” so that careful assessment of our admission criteria would elucidate for readers the “fish” that we were about to catch.