Widespread Implementation of Appropriate Use Criteria for Cardiac Imaging-Which Are "Appropriate"?

Abstract

For more than a decade, organizations such as the American College of Radiology (ACR) and theAmericanCollege of CardiologyFoundation(ACCF)havepublishedcriteriadelineatingthe appropriateuseofcardiac imaging.Thesespecialtysocietyguidancedocumentsaredesigned toprovideclinicianswith recommendations regarding the use of imaging and are focused on reducing unnecessary and inappropriate testing. In general,boththeACRandACCFusesimilarmethods in theconstructionof appropriateusecriteria (AUC), byperformingcarefulevidencereviewsandprovidingastandardizedratingofmultiple clinical indications by the use of the UCLA/RAND methodologywithamodifiedDelphiapproach.1However, substantial methodological differences exist between each organization’s approach to AUC, including the ACCF’s greater relianceof riskstratificationbasedonclinical factors,whichresults in a far greater specificity of clinical indicationswith theACCF version.Furthermore, thehierarchicalnatureof theACCFAUC readily lend these criteria to be adopted into clinical decision support tools.1 In this issue of JAMA Cardiology, Winchester et al2 compare2setsofappropriateusecriteria,onederivedfromtheACR and the other led by the ACCF. Substantial discordance in appropriatenesscategorizationwasnoted(22.3%),andmorecomplete “coverage” of all clinical situations was demonstrated with the ACCF AUC. A significant limitation of this article relates to theuseofanolderversionof theACCFAUC.This singlemodality set ofAUCwaspublished7years ago3 andhas largely been superseded by the 2013 multimodality AUC,4 which should serve as the current standard for the evaluation of appropriateuse.Whilemany indicationswereratedsimilarlywith regard to radionuclide imaging, several important differences in appropriateness classificationwere noted, including the appropriateness of radionuclide imaging after CABG, the value inhigh-risk asymptomatic individuals, and its use in the preoperative assessment of patients before intermediate-risk surgery.5 A concern raised by Winchester et al2 is the finding that there were far more abnormal imaging studies or individuals with ischemiawithan inappropriatedesignation fromtheACR (35.4%and 17.5%, respectively) thanwith theACCFAUC (7.3% and2.6%, respectively).Although it iswelldescribed that studies designated as inappropriate may still have abnormal test results, as reported in6 trials,5 the fact that the frequencymay be higher using the ACR criteria is worrisome. Another important consideration related to theuseofAUC ishowoften testing for inappropriate indicationsprovidesusefulprognostic information, thuspotentially supporting the recommendation not to perform a test for an inappropriate indication.Dataarenowbecomingavailable, includingstudiessuch as thosebyDoukkyetal,6whichdemonstrate that the truepredictive power of radionuclide imaging regarding the identification of patients at risk for a subsequent cardiac event lies in thepopulationwhoundergoappropriate testing, notwhen radionuclide imaging is applied for inappropriate indications.To myknowledge,nosuchdatahavebeenprovidedusing theACR criteria, and based on the discrepancies between the 2 sets of AUCandthesubstantiallyhigher incidenceofabnormal singlephotonemissioncomputed tomographystudiesperformedfor ACR inappropriate indications, it seems reasonable to conclude that theremay be performance issues related to the use of the ACR criteria for this purpose. It is critical to recognize that the mere creation of these AUC, regardless of the source, is almost certainly insufficient to promote change in utilization. It must first be shown that the levels of appropriateuse can successfully be trackedusing a specific set of AUC, so as to demonstrate current performance and permit initiatives to be developed to improve imaging utilization. Notably, there are multiple publications demonstrating the feasibility of tracking AUC using the ACCF criteria but very limited literature regarding theuseof theACR AUC. Publications over the last 10 years, which includemore than 20000 patients, have revealed that appropriate utilizationof cardiac imagingmaybe successfully assessedusing the ACCF AUC,5 with very few gaps in appropriateness determination through the entire spectrum of patient presentations. Furthermore, tools must be available to place this valuable information directly in the hands of the clinician and ideally provideguidanceat thetimeof testordering.Thishasnowbeen shown in several cohorts using the ACCF AUC.7,8 Despite efforts to promote utilization of AUC and encourage practitioners to be mindful of utilization patterns, neither set of AUC has been fully incorporated into clinical practices, and are incompletely used by radiology benefits managers, who often develop their own, non–providerbased criteria. Theopportunities providedbyAUC implementation appear great and offer situations for tracking provider practice to support continuous quality improvement and to provide education for active practitioners and trainees alike. Recent federal legislation will undoubtedly alter the current landscape and may impact whether published, adjudicated AUC are fully implemented by all payers. The Protecting Access to Medicare Act of 2014 (which became law on April 1, 2014)mandates the use of AUC for advanced cardiac imaging, including nuclear cardiology, cardiac magnetic resonance, and cardiac computed tomography, for all Medicare beneficiaries.9 The current program by the Centers for Medicare &Medicaid Services (CMS) requires the selection of a provider-led entity’s AUC and the subseRelated article page 207 Appropriate Use Criteria for Cardiac Imaging Brief Report Research