Inappropriate stress echo referrals persist despite AUC, education

Twitter icon
Facebook icon
LinkedIn icon
e-mail icon
Google icon
 - Heart Rate

Almost one-third of stress echocardiograms are ordered for inappropriate indications and this proportion persisted over time despite interventions directed toward referring physicians, according to a recent analysis. It also found only a fair correlation between radiology benefits managers (RBMs) and appropriate use criteria (AUC) for stress echocardiography.

These findings were published online Feb. 20 in the Journal of the American College of Cardiology: Cardiovascular Imaging, along with an editorial that suggested that objective standards can augment, but not replace, clinical judgment.

Growth in diagnostic imaging for coronary artery disease (CAD) in recent years has led to concerns about overutilization. In response. the American College of Cardiology (ACC) developed AUC for several imaging procedures, including stress echocardiograms (SE) in 2008, which it updated in 2011. The ACC also recently revised its AUC terminology.

Howard Willens, MD, of the University of Miami Miller School of Medicine in Florida, and colleagues conducted a retrospective survey of 209 consecutive SEs performed in August and September in 2008, and 209 consecutive SEs performed between July and September in 2011, analyzing them for appropriateness in accordance with both the 2008 and 2011 AUC. The third cohort comprised 111 SEs ordered after an educational intervention aimed at cardiologists on appropriate indications for SE. The researchers also evaluated whether the studies would be covered under two large RBMs.

Using EHRs, the researchers extracted demographics, cardiac history, clinical and laboratory assessment of heart disease including CAD and prior procedures, and used these data to estimate a pretest probability of CAD. They rated the SEs according to the 2008 and 2011 AUCs, denoting any test for which indications could not be assigned as “unclassified.” The others were categorized as appropriate, inappropriate or uncertain.

The percentage rated inappropriate was virtually the same in both periods: 29.7 percent in the 2008 SEs according to the 2008 AUC and 30.6 percent in the 2011 SEs according to the 2011 AUC. Of the 529 studies analyzed, 43.5 percent were requested for indications the AUCs deemed appropriate, 22.9 percent were requested for uncertain indications and 30.8 percent were inappropriate according to the 2011 AUC. The percentage of inappropriate tests was similar among the three cohorts, but the SEs were significantly more likely to be inappropriate when the patients were female or under 65 years of age.

The updated AUC made a substantial difference in the classification of the SEs; 90 percent of the studies that were unclassified according to the 2008 AUC were able to be assigned a category under the 2011 AUC. Overall, 25 percent of the SEs changed categories under the 2011 AUC. Only 2 percent of studies were unclassified using the 2011 AUC, but the percentage of studies categorized as uncertain increased from 11 percent under the 2008 AUC to 21.1 percent under the 2011 criteria.

More than 40 percent of the SEs deemed inappropriate were ordered for the AUC 2011 indication “Evaluation of ischemic equivalent, low pre-test probability of CAD, interpretable EKG and able to exercise.” The next most common inappropriate indication was “Evaluation of asymptomatic patients or stable symptoms, intermediate to low global CAD risk, normal stress imaging study more than two years ago,” which was used in 7.4 percent of the inappropriate studies.

The researchers found no changes in the appropriateness ratings of SEs ordered by cardiologists before and after an educational intervention to explain the 2011 AUCs. Although the numbers of SEs ordered for appropriate indications increased after the educational initiative, there was no decrease in the number of inappropriate studies ordered. “Overall, there was no significant difference in appropriateness ratings before and after the initiative,” the authors wrote.

Comparing the appropriateness of studies under the AUCs with the preauthorization criteria published by two RBMs, the researchers found 84 percent agreement between the 2008 AUC and RBM 1, and 65 percent agreement between the 2008 AUC and RBM 2. Using the 2011 AUC, there was 83 percent agreement with RBM 1 and 61 percent agreement with RBM 2. “The large majority of SEs with a discordant pre-authorization determination were rated appropriate by both versions of the AUC as opposed to uncertain,” the authors pointed out. This finding “suggests a need for greater collaboration between the medical societies and health plan RBMs,” they concluded.

In an accompanying editorial, James K. Min, MD, of Cedars-Sinai Medical Center in Los Angeles, pointed out that the ACC was currently developing AUC for all testing modalities capable of diagnosing CAD that will “homogenize appropriate use categories across different test types in an easy-to-apply manner, and Willens’ data suggest that the scenarios addressed in this multimodality application will encompass nearly all commonly used clinical applications.”

Min also noted that the finding of only fair correspondence between published RBMs and AUC “speaks to the issue of physician-preferred, rather than policy-based, testing...Algorithmic flow charts are useful for general guidance, but can never be a substitute for the art of medicine.”    

The authors and Min cited the small cohort sizes, the fact that the study was conducted at a single center and the subjectivity of AUC, RBM algorithms and EHRs as limitations of the study.