CTA appropriateness criteria: Improved, but gaps remain

The 2010 update to the appropriateness criteria (AC) for cardiac CT made them more clinically useful than the original 2006 criteria, but high interobserver variability remains, according to a study published online Jan. 30 in the Journal of the American College of Radiology.

By comparing the results of using the 2010 AC for cardiac CT angiography (CTA) to a previous study using the 2006 AC, Ethany L. Cullen, MD, and colleagues at the Mayo Clinic in Rochester, Minn., found the number of patients who could not be classified was cut by two-thirds.

“Although this was a significant decrease compared with the 2006 criteria, the fact that the reviewers continue to disagree on the specific indications for nearly half the cases calls into question how reliably the 2010 AC can be used in practice or by third-party payers,” wrote Cullen and colleagues.

The AC was created in 2006 by the American College of Cardiology Foundation and the American College of Radiology, among other societies, to offer guidance to clinicians for using CTA, which was a rapidly changing technology and often performed inappropriately, explained the authors.

A previous study of the 2006 AC conducted at the Mayo Clinic found some major problems, however, as 46 percent of patients could not be classified and observers disagreed on over 60 percent of cases.

The 2010 AC expanded the number of indications from 39 to 60, with many of the indications receiving separate appropriateness ratings based on patient history, noted Cullen et al. This was thought to provide further guidance on the appropriate use of CTA.

To evaluate these updates, Cullen and colleagues applied the 2010 AC, using two observers, to the same 251 patients used in the previous study of the 2006 AC. Patients for whom no indications could be found were considered not classifiable.

Results showed that the number of unclassifiable patients fell from 115 to 39 by switching to the updated criteria. Using the 2010 criteria, 16 percent of patients could not be classified. “Given that the AC were not intended to cover every clinical scenario, the 16 percent rate of patients who were not classifiable seems reasonable and suggests that the 2010 AC are usable.”

Observer disagreement was also reduced, but more modestly. Disagreements on specific indications occurred in 61 percent of patients with the 2006 AC, and with the 2010 criteria the observers disagreed for 47 percent of patients.

The authors pointed to four distinct sources of disagreement:

1. Difficulty in consistently identifying prior test results in the medical record;
2. Once agreed that patients had prior tests, there was variability in abstracting the required data;
3. Complex patient histories could fit multiple indications; and
4. The AC organizes indications into tables, and one of the tables used to classify patients—Table 7: “Evaluation of Cardiac Structure and Function”—was not included in the criteria’s flow chart. This flow chart is used to indicate in which order the tables should be considered, excluding patients who could be assigned to Table 7, which created uncertainty.

“Although the rate of patients not classifiable has decreased to an acceptable level, the interobserver variability remains concerning,” wrote the authors.