March 9th, 2015
How Appropriate Are Appropriate Use Criteria for Coronary Angiography?
The CardioExchange Editors interview Harindra C. Wijeysundera about his research group’s use of registry data from Ontario, Canada, in applying the 2012 appropriate use criteria (AUC) for diagnostic catheterization to patients referred for coronary angiography who had no known history of coronary artery disease (CAD). The study and its accompanying appendix are published in the Annals of Internal Medicine.
For the 48,336 patients in the final cohort, 58.2% of angiographic studies were classified as appropriate, 10.8% as inappropriate, and 31.0% as uncertain. In all, 45.5% of patients had obstructive CAD.
|Appropriateness for angiography||
% with obstructive CAD
|Patients with appropriate indications||
|Patients with inappropriate indications||
|Patients with uncertain indications||
CardioExchangeEditors: In your view, how accurately do the appropriate use criteria identify who should and who should not undergo coronary angiography?
Wijeysundera: The ultimate goal of our study was indeed to determine how well the AUC could guide patient selection. Unfortunately, our results suggest that although the AUC provides some insight, it is insufficient in isolation to select patients who either should or should not undergo diagnostic angiography. The rationale for that conclusion is that, in our study, although patients with appropriate indications were more likely than patients with inappropriate indications to have obstructive CAD, still almost 50% of the “appropriate” patients had normal coronaries. Most important, almost a third of patients with inappropriate indications had obstructive CAD, and 1 in 5 had disease severe enough for revascularization. Our study highlights the need to further refine the AUC to better guide patient selection.
Editors: Of the roughly 10% of patients whose coronary angiograms would have been deemed “inappropriate,” about 20% underwent revascularization. Did this surprise you?
Wijeysundera: This important finding did not surprise us, in that it reflects the fact that clinical decision making is often complex and not easily captured by explicit criteria such as the AUC.
Editors: How were you able to translate the chart data into AUC categories? How did you know you captured each category well?
Wijeysundera: We obtained all of our data from a clinical registry held by the Cardiac Care Network of Ontario, Canada. All patients who undergo cardiac procedures in the province of Ontario must have data entered into this registry. The data itself have been validated by selected chart audits. To place patients into the AUC categories, we used a computer algorithm as detailed in our paper. To ensure the accuracy of that categorization, we restricted our analyses to only 12 AUC indications (i.e., those for patients with suspected coronary artery disease). The full document contains 102 AUC indications covering the spectrum of acute coronary syndromes, previously documented CAD, valvular disease, and arrhythmias. By restricting ourselves to only 12 indications in a very explicitly defined population, we were able to apply a relatively simple algorithm to categorize patients. We then performed numerous checks within each of the final AUC groups to ensure that patients had been appropriately categorized.
Editors: Where can readers find the mapping? What were the numerous checks?
Wijeysundera: The mapping is part of the appendix to our paper. We did make a few assumptions that are explicitly explained in our methods section. Our checks involved looking at the characteristics of patients in each category to ensure that our algorithm placed patients as intended. We were reassured by the face validity of the categories. Ideally, we would have done chart abstractions in a subset of patients, but we did not have access to the charts.
JOIN THE DISCUSSION
How do Dr. Wijeysundera’s findings affect your perceptions of the appropriate use criteria for coronary angiography?