In ancient Greece, wisdom was sought by the mighty and the modest alike by consulting the Oracle of Delphi, an aged seer who channeled Apollo’s thoughts while intoxicated on the fumes emanating from a python, slain by Apollo himself and cast into the depths of the earth. Sitting on a three-legged stool astride this malodorous crevasse, this ancient Oracle, always a woman of unblemished repute, could answer such burning questions of the day as (from the King of Thera), “Where shall I colonize next?” (Answer: “Libya”; second question: “Where is Libya?”). Fast-forward almost 3000 years, and today we have our own Oracle of Delphi (in the form of the modified Delphi process used by the American College of Cardiology Foundation) to answer such questions as, “Is it appropriate to perform a transthoracic echocardiogram when there is a severe deceleration injury [and] cardiac injury [is] suspected?” (Answer: “yes”). With the release of Douglas’s 2011 Appropriate Use Criteria for Echocardiography , the medical community has new consensus data summarizing when echocardiography is and is not appropriate to use in a host of clinical scenarios. The purpose of this commentary is to guide the reader through the possibly unfamiliar Appropriate Use Criteria process, describe the broad findings of the document and how to use them, indicate potential areas that the echocardiography community may find troubling, and suggest research priorities for the future to better define the benefit of echocardiography.

The method used by the American College of Cardiology (ACC) in developing the Appropriate Use Criteria (AUC) is a modification of the Delphi process designed by the RAND Corporation in the 1950s to better predict the impact of technology on the cold war. It is based on the notion that opinions of a group are superior to the opinions of a single individual and is specifically meant to force consensus among disagreeing parties or at least define those issues on which there is no agreement. Of course, we should recall Abba Eban’s observation that “consensus means that everyone agrees to say collectively what no one believes individually.” As implemented by the ACC, the AUC development is a four-step process. First, descriptions of many (202 in the case of this document) clinical scenarios in which echocardiography might be used were compiled and reviewed by external reviewers for their relevance, precision, and inclusiveness (disclaimer: I was one of the external reviewers for this document, though, to be honest, not too many of my recommendations made it into the scenarios). Then, the clinical scenarios are presented to a technical panel for initial rating. This technical panel is deliberately composed of a minority of “card-carrying” echocardiographers, with most of the panelists coming from other specialties, including a representative from a health insurance plan. These panelists voted on a 9-point scale (1 = completely inappropriate; 9 = absolutely appropriate) without any interaction with each other. Next, the panelists were brought together for a face-to-face meeting where each indication was discussed and the group received a list of their individual scores and a blinded summary of the other scores, after which another poll was taken. The expectation is that with successive votes and discussion, the scores will converge on a consensus, defined as no more than 4 of the 15 panelists falling outside a 3-point range, at which point the median score is reported. Although these median scores should be viewed as a continuum, they are grouped (in the document’s own words) into three “somewhat arbitrary” categories: an inappropriate zone (1 to 3), an appropriate zone (7 to 9), and an “uncertain” zone (4 to 6), indicating that the test may or may not be appropriate and often pointing toward areas where additional research may be needed.

What do we learn from this rigorous exercise? For the most part, the echocardiography community should find the scores entirely rational and in keeping with daily clinical practice. Of the 202 clinical scenarios, the panel found use of echocardiography appropriate in 97 of them, inappropriate for 71, with the remaining 34 deemed uncertain. For virtually any scenario in which a patient presents with new or changing signs or symptoms of heart disease, an echocardiogram is appropriate. Scenarios involving simple screening in asymptomatic individuals (e.g., healthy subjects with occasional ventricular premature contractions) are, appropriately enough, generally deemed inappropriate for echocardiography. Routine follow-up examinations are closely scrutinized. Thus, initial postoperative examination of a newly implanted prosthetic valve earns a full 9 for appropriateness, whereas follow-up echocardiography within 3 years without any evidence for dysfunction was rated inappropriate at 3, increasing back to an appropriate 7 beyond 3 years and a 9 if there is any suspicion of prosthetic dysfunction. Of course, by offering only scenarios 3 years before or after post implant (a time apparently taken from the 2006 Valve Guidelines), the results were largely preordained.

Similar observations prevail for transesophageal echocardiography (TEE). Anytime there is a clinically relevant question that is better answered with TEE than transthoracic echocardiography (TTE), that indication is appropriate. If there is low expectation of a positive finding or no suggestion that that finding will affect patient care, then the indication generally is inappropriate. It really is mostly common sense. Of course, all AUC are just general recommendations and should not mandate an echocardiogram (or preclude one) in a specific clinical situation. Even with the plethora of clinical scenarios voted on, there still are many nuances not captured in them, for which a good clinician may overrule the guidelines.

One area in which these AUC may cause agita among echocardiographers is in the realm of stress echocardiography. Even here, for the most part, the results conform to rational daily practice. Chest pain with low pretest likelihood and an interpretable electrocardiogram yields an inappropriate stress echocardiogram. Make either the pretest likelihood intermediate or high or the electrocardiogram uninterpretable and—voila!—the stress echocardiogram is entirely appropriate. Precisely congruent guidance for stress nuclear imaging is found in their guidelines. Only in a few specific scenarios are discrepancies found between stress echocardiography and nuclear cardiology, such as scenario number 172 for echo (number 58 for the nuclear document): evaluation of an asymptomatic patient 5 or more years after coronary artery bypass grafting. This earns a 6 in the echo document (uncertain) and a 7 (appropriate) in the nuclear document. Beyond the obvious fact that there is only one point separating these grades, which just happens to fall on the uncertain-appropriate boundary, one should note that neither the echocardiography nor the radionuclide imaging panel was able to come to “agreement” on these grades, defined by no more than four scores outside the 3-point range around the median. Furthermore, it is possible to dig into the details of the scores, because for all AUC documents, an anonymized tabulation of the votes is posted online, yielding some interesting observations.

First of all, there was a 1.21 spread in mean scores among the individual panelists for the 202 clinical echo scenarios, with one grader (you know who you are!) falling 0.72 below the mean ( P < 10 ⁻¹⁴ ). Second, by looking at the grades for the specific scenario of the asymptomatic patient 5 or more years after coronary artery bypass grafting, there was actually no significant difference between the two sets of scores (both mean scores rounded to 6). However, although we can take some comfort in the relatively insignificant differences between these scores, the echocardiography community should also acknowledge that there may be some gaps in the literature base that we must use to convince colleagues and payers alike that our procedures are appropriate for use. For example, the online supplements for the AUC documents contain a synopsis of the relevant literature for each of the broad clinical scenarios considered. For risk assessment post-revascularization, the echo document lists two studies, but neglects to include other relevant studies, whereas the same section in the radionuclide imaging document lists 16! This may be an area for future research or a targeted area for outcomes research funded by the American Society of Echocardiography. It should be remembered that these judgements were made in isolation without any attempt at resolution by specialists who perform and are experts at both echocardiography and single photon emission computed tomography. The ACC is now in the process of defining the relative value and appropriateness of multiple imaging modalities for common disease entities or presentations, a situation in which exercise echocardiography’s low cost and lack of radiation may prove advantageous.

So what should the echocardiography community do with these guidelines? My recommendation is that we embrace them, while still pointing out these relatively minor limitations and potential areas for refinement. Proper self-policing with adherence to AUC is our best defense against restrictions imposed by payers. Currently there is a move afoot among private insurance companies to require precertification for all echocardiograms, similar to what is already done for computed tomography and magnetic resonance imaging. In a busy laboratory, this would represent an oppressive administrative burden and potentially require additional staff at a time of decreasing revenues from echocardiography. How much better if we can show that we are proactively enforcing policies so that the majority of echocardiograms done have appropriate indications (this will never reach 100%, because there are many scenarios not considered in the AUC process and many nuances not captured by the existing scenarios). What are the data so far? Naturally, there is no experience yet with the new guidelines, but a number of studies have been published examining adherence to the earlier guidelines for TTE/TEE and stress echocardiography. These articles have examined TTE appropriateness in academic centers, community practice, and the Veterans Administration. By and large, results are encouraging, with 90% to 97% of the classifiable studies deemed appropriate (most of the 10%–30% of studies that were previously unclassifiable should now be covered by the new AUC, with its greatly expanded clinical scenarios). In the TEE laboratory, a recent study showed that fully 97% of studies were appropriate. Unfortunately, the situation is not quite as encouraging in stress echocardiography, because a recent study found up to 20% of such tests were inappropriate, generally in younger patients and ordered by noncardiologists. Note that these have all been relatively small studies involving dedicated investigators.

How can we hope to disseminate the AUC into the daily practice of echocardiography across the United States? One possibility is the development of an online point of order tool that could quickly (with just a few mouse clicks) inform the ordering clinician and performing echocardiography laboratory whether the echocardiogram being ordered adheres to the AUC. If not, there would be the possibility to override the process (there will always be exception to these rules), but this would surely reduce dramatically the number of inappropriate echocardiograms. Such an online tool was recently developed by Ward and colleagues of the University of Chicago with promising early results, allowing appropriateness to be determined in less than 1 minute. The ACC is currently developing a tool (FOCUS) to guide multimodality imaging ordering, which will incorporate this most recent echocardiography AUC guideline information. When fully implemented, this has the potential to greatly reduce inappropriate studies and hopefully to inhibit the move toward use of Radiology Benefit Managers to dictate which tests can be ordered.

Overall, the AUC process in general and the new echocardiography document in particular have been valuable to the medical community, providing guidance as to when and if an imaging test is appropriate in a given clinical scenario. Adherence to them will help ensure that health care dollars are spent in the most effective way. Although AUC development is an arduous and ongoing task, requiring reassessments as new techniques and data emerge, we can at least be thankful that our modern Delphic oracle requires fumes no stronger than freshly brewed coffee. For now, at least, rotting pythons are no longer needed for inspiration.