Developing Appropriateness Criteria for Imaging
Decision support might be the technology that brings imaging out of external control and back into self-regulation, since its effectiveness appears to equal or exceed that of other utilization-management strategies (including burdensome preauthorization). Every decision-support program, however, must be able to tell the ordering physician how useful an imaging exam is likely to be. That simple indication of probable appropriateness—often a numbered ranking—must be based, in turn, on a prior determination that is anything but simple: Valid appropriateness criteria take into consideration not only expert opinion, but up-to-date research findings and the broadest possible range of relevant clinical experience. The provision of decision support itself, whether through computerized provider order entry (CPOE) or nonelectronic methods, is relatively straightforward, in comparison with the vast pile of information that must be considered and condensed to create appropriateness criteria. For each patient condition that the referring physician will encounter, the developers of appropriateness criteria must determine not only the degree to which specific imaging exams might be helpful, but also whether imaging is indicated at all. Who makes these decisions today, and how might that change tomorrow? These were among the questions asked on November 29, 2010, by the four presenters of a special-interest session, “The ACR Appropriateness Criteria: Are You Trying to Tell Me What to Do?” at the annual meeting of the RSNA in Chicago, Illinois. Two of the session’s sections deal with appropriateness criteria and what radiology can expect from them (and contribute to them). Michael Bettman, MD, vice chair of the ACR® Commission on Quality and Safety, describes how the ACR’s guidelines are constructed by and for radiologists; a prediction that nonradiologists will become more involved in imaging-guideline development is expressed by Earl Steinberg, MD. Steinberg is senior vice president for integrated health strategy and evaluation at WellPoint (Indianapolis, Indiana); president and CEO of WellPoint’s Resolution Health subsidiary; and chair of the US Institute of Medicine’s Committee on Standards for Developing Trustworthy Clinical Practice Guidelines. ACR Methods Bettman explains that while the ACR’s appropriateness criteria are meant to guide clinical decisions, they are also intended for use in medical education. For a given clinical condition, the ACR guidelines should tell the physician who is making an imaging decision whether an imaging exam is typically indicated for that condition. If so, the guidelines then specify whether each exam that could be ordered, under the existing circumstances, falls into one of four categories: usually appropriate, may be appropriate, usually not appropriate, or rarely necessary. Bettman notes that these terms were chosen with care, since the number of variables involved in patient care makes it unrealistic to categorize a study as always appropriate or never appropriate. In response to growing concern about the cumulative radiation dose attributable to medical imaging, the ACR guidelines now include relative radiation levels (for both adults and children) for each applicable exam. If there are inadequate data on the value of an exam, or if there is disagreement concerning the meaning of those data, that exam will appear in the guidelines as unrated. The perfect appropriateness criteria, Bettman says, are evidence based and chosen by experts whose methods are both transparent and consistent. They are updated regularly and are available online, as well as within CPOE systems. They also have broad acceptance in the field. Unfortunately, he adds, all of these are difficult goals to meet. Evidence alone is never sufficient, so expert opinion is always needed. For this reason, the ACR’s method for guideline development involves selecting a topic author to serve as an expert. The topic author then gathers and reviews the relevant medical literature, condensing it into an evidence table that lists the most important published findings (with notes on the quality and reliability of the studies involved). The topic author also designs clinical scenarios representing the possible and probable imaging orders for the condition being reviewed. The chair of the guidelines panel reviews this material, which is later voted on by the panel’s members before it becomes the official ACR guideline. Multispecialty Criteria Another approach to the creation of appropriateness guidelines for imaging welcomes the involvement of nonradiologists—in the interest of balance, according to Steinberg. In the future, he says, this model might become more common, since many sets of guidelines created within various medical specialties clearly seem biased in favor of the procedures performed by members of those specialties. This is one of the reasons that the Medicare Improvements for Patients and Providers Act of 2008 mandated the creation of the guideline-quality committee on which Steinberg serves. The committee has noted that several problems are common in guidelines. Many are created without sufficient evidence (especially where multiple morbidities are involved). Reviews of evidence have been of varying quality, and conflicts of interest have been apparent. The creation of guidelines has been exuberant, but confidence in the results has been limited, and adoption of guidelines has often been disappointingly narrow. The Institute of Medicine’s committee recommends taking several steps to increase confidence in guidelines. First, they should be objective. Second, they should be developed by balanced (multispecialty) panels. Third, they should be transparent, containing complete information about the methods used and the strength of the evidence on which they are based. Fourth, they should be subject to external review. Fifth, the relevant literature should be monitored continuously. Sixth, organizations developing guidelines should attempt to increase consensus among specialties instead of taking a single-specialty approach. Steinberg finds the ACR’s practice of having a single topic author manage the creation of guidelines for a particular condition to be inadvisable. Instead, he recommends broadening the responsibility to include several people who will perform the literature review; few individuals, he says, could simultaneously have the necessary clinical expertise and understanding of research methods to conduct the review alone. In cases where the imaging guidelines created by the ACR are in conflict with those published by other specialties, Steinberg feels that it should be the ACR’s responsibility to lead the effort to determine why there is a discrepancy. Above all, Steinberg says, it is vital to ensure that appropriateness criteria are generated within health care, not outside it. If physicians are unable to cut imaging costs with precision, using a scalpel, then they can expect someone else to do it—with an ax. High Global Stakes The existence of appropriateness criteria is not, in itself, enough to ensure their acceptance or application on any useful scale. The National Guideline Clearinghouse, a searchable database maintained by the Agency for Healthcare Research and Quality, now contains more than 2,700 sets of practice guidelines for health care, Steinberg says. Nonetheless, many physicians maintain a hearty dislike for guidelines and make no use of them. If all of the effort that goes into creating appropriateness criteria is not to be wasted, then CPOE and other decision-support technologies must make those criteria part of everyday practice. As the session presenters note, inappropriate utilization of imaging should not be considered a problem limited to the United States, even though incentives for self-referral and fear of malpractice suits could be stronger driving forces here. Worldwide, even though many regions lack such powerful reasons to order inappropriate studies, 30% to 50% of imaging exams could be inappropriate. Of course, this excess adds to the cost of health care, but it also creates other burdens (including those of additional exposure to radiation and of unnecessary follow-up care). Therefore, the development of sound appropriateness criteria for imaging—coupled with the adoption of decision-support systems that will make those criteria part of referring physicians’ ordering methods—will do more than reduce costs. It will help imaging decisions stay in the hands of those who order and provide (and undergo) exams, and it will make imaging a more solid foundation for better patient care, with more accurate diagnoses accompanied by reduced risks. Kris Kyes is technical editor of