Time to Get Ahead of the Curve: Clinical Decision Support Mandate Delayed, Not Dead

Clinical decision support (CDS) tools have the potential to curb inappropriate imaging, reduce healthcare costs and improve the quality of care, according to data from the small number of early adopters in the United States. These clinicians are a step or two ahead of the Centers for Medicare & Medicaid Services (CMS), which last fall delayed implementation of the provision in the Protecting Access to Medicare Act (PAMA) that requires physicians to use CDS tools, and document their use, whenever they order advanced imaging tests. Cardiovascular Business spoke with physicians who are ahead of the CDS curve and others who are getting ready for its arrival.

The goal of the legislation requiring CDS was to help clinicians who order CT, MR, SPECT and PET imaging—often primary care providers and general cardiologists—to select the best test for each patient, depending on his or her clinical history, symptoms and demographics. According to the law, the CDS mandate was to take effect Jan. 1, 2017. Responding to a request from a coalition of 17 associations, CMS acknowledged that more time is needed to work out what will be required of practices and allow time to plan for implementation. At press time, CMS hadn’t set a new deadline for CDS use; however, a handful of hospitals have moved forward.

Why clinical decision support & why now?

“Studies show when clinicians don’t have [guidelines] information in front of them they don’t follow radiology guidelines about half of the time,” says Keith Dreyer, DO, PhD, vice chairman of radiology computing and information sciences at Massachusetts General Hospital (MGH) in Boston. That can increase costs, translate into poorer quality patient care and decrease patient safety by exposing patients to unnecessary radiation. In contrast, when guidelines are integrated into workflow at the point of care, physician adherence to guideline standards increases to 90 percent—“without the use of carrots or sticks,” Dreyer says. 

MGH started using decision support in 2004, initially for CT, MRI, nuclear medicine and PET. Computerized order entry systems and CDS tools weren’t yet available off-the-shelf, so MGH built its own.

CDS should be tied to electronic order entry and the electronic health record (EHR) to maximize efficiency and minimize workflow disruption. It’s also one of the reasons CMS had not pushed CDS until now: because workflow integration requires EHR integration and EHRs hadn’t reached the critical level of penetration. Now, EHR adoption has passed the 90 percent mark, spurring policymakers to push CDS.

Like EHRs, decision support requires well-devised operational and workflow planning. “The most arduous part of using clinical decision support is entering the reason for the exam,” Dreyer says. It has to be a structured reason. Every CDS system, whether from a vendor or homegrown, ties to structured content or guidelines, which are mapped to the reason for the exam and the exam itself. If the reason for the exam is entered as free text, then the CDS tool will show the physician all of the possible guidelines that relate to that reason—a figure that could stretch into the hundreds.

“If you give me the [structured] reason for the exam, I’ll show you the one targeted guideline that is appropriate for you to reference,” Dreyer says. Using a structured reason enables a straightforward connection between the indication and the guideline.

Another essential prerequisite for effective CDS implementation, according to Dreyer, is to integrate decision support into the workflow, providing ordering physicians with the information they need at the point of care without slowing them down. When a physician orders an imaging exam at MGH, he or she enters an exam and the reason for the exam. The system responds similar to a traffic light with green, yellow or red, indicating whether the exam is indicated, marginal or of low utility, respectively.

After the hospital implemented a hard-stop-on-red policy—it stopped performing exams likely to be negative unless authorized by a radiologist—low-yield CT, MRI and nuclear imaging exams dropped from 5.43 percent to 1.92 percent, and the probability of canceling those exams increased by 3.5-fold (Radiology 2010;255[3]:842-9). MGH also saw its per member/per month high-cost imaging exam volume drop 12 percent with CDS. 

Clinical decision support also has led to better collaboration between radiologists and ordering physicians at MGH, says Dreyer. When clinicians are off guidelines, the system pulls in specialists to consult, answer questions about why an exam is flagged as red (inappropriate) and in some cases suggest an alternate exam or indication. The result is that radiologists and other imaging specialists are better integrated into care pathways and can “fix problems.”

The University of Pittsburgh Medical Center (UPMC) launched a pilot CDS project at one hospital late in 2015. The homegrown system was instituted by the organization’s health plan and is integrated into the EHR. When a referring physician orders an imaging exam, the CDS tool generates several questions. Based on the physician’s responses, a software algorithm indicates whether the test is appropriate. While the system does not stop a physician from ordering an exam, it does provide feedback about the appropriateness of the order.

Prem Soman, MD, PhD, director of nuclear cardiology at the UPMC Heart and Vascular Institute, estimates approximately 90 percent of the health system’s nuclear studies are already appropriate without CDS, so they are “starting off in a very good place." 

[[{"fid":"22901","view_mode":"media_original","type":"media","attributes":{"height":771,"width":276,"style":"width: 276px; height: 771px; float: left; margin: 5px;","alt":" - preparation-checklist","class":"media-element file-media-original"}}]]

Beware unintended consequences

Clinical decision support may have the potential to help physicians adhere to guidelines and reduce costs, but some experts believe the legislative mandate puts the cart before the horse, or in this context, leads with the last step—clinical decisions—without critical prerequisites.

“Providers need to understand the value and role of imaging first and then appropriate use criteria,” says David Wolinsky, MD, immediate past president of the American Society of Nuclear Cardiology (ASNC), the group that led the coalition advocating to delay the decision support requirement.

Pushing CDS without those prerequisites also makes for more challenging implementation, according to Wolinsky. “We don’t want to just decrease the cost of imaging,” says Wolinsky, who is section head of Nuclear Cardiology at Cleveland Clinic Florida in Weston. “We want to achieve quality. That’s much more complicated than whether or not a test is appropriate. It’s not black and white.”

Some providers worry that CDS will prescribe a cookbook approach to medicine that is not feasible in practice. “A certain percentage of exams are going to be classified as possibly appropriate and inappropriate because criteria don’t exist for all clinical scenarios,” says Soman. “A certain number of non-appropriate exams are inevitable because that’s the nature of medicine. Not all patients fit into [CDS] categories.” 

Wolinsky cautions that inflexible requirements for CDS could have unintended consequences, including a negative impact on other quality metrics and even costs. He describes a possible scenario where a patient visits his primary care provider, who enters the patient’s demographics and symptoms into his health system’s CDS only to have the test he wants flagged asw possibly inappropriate. Phone tag between the ordering physician and the nuclear cardiologist or radiologist ensues. The patient is eventually referred to a cardiologist, but it’s another visit, another claim and another co-pay for the patient. In this scenario, the patient still gets the test, but patient satisfaction drops and the total costs are higher than they would have been if the primary care provider had ordered the test initially.

“The patient care issues may overwhelm the potential benefit of finding 5 percent of [inappropriate] outliers,” says Wolinsky.

Best practices

Despite CMS’s decision to delay the decision support start date and unknowns about when the requirement details will be announced or exactly what they will say, insiders agree that CDS is coming.  Practices would be wise to get prepared because the window for purchasing, installing and learning how to use CDS tools is likely to be shorter than most clinicians would prefer.

The first and most important key to success is to select a solution integrated into the EHR, advises Dreyer. Buying directly from the health system’s EHR vendor may be the most streamlined route for many hospitals. 

The law requires that a free option be available, but the attractive price point may mask a variety of drawbacks, including workflow inefficiencies. Physicians using a free portal will have to manually enter patient demographics, exam type and indications and then obtain a pre-authorization number that must be entered back into the order entry system. The process would slow workflow and set the stage for errors.

Even the most expensive CDS programs won’t run themselves and they shouldn’t be viewed as simple add-ons, Dreyer says. Instead, think of CDS as an essential part of a program that will help providers effectively utilize imaging services. This means the decision support tools shouldn’t be tailored to one exam type. Rather, the tools need to encompass all of the indications and modalities.

“Otherwise, physicians won’t know when to use the system and may wind up going around it by picking an indication that isn’t managed,” Dreyer says. “[When CDS is comprehensive,] you truly have a gatekeeper in front of the modality. That’s why I’ve encouraged CMS to be very prescriptive in terms of being comprehensive for all of the advanced diagnostic imaging modalities that they are going to require a mandate for.” 

Around the web

Several key trends were evident at the Radiological Society of North America 2024 meeting, including new CT and MR technology and evolving adoption of artificial intelligence.

Ron Blankstein, MD, professor of radiology, Harvard Medical School, explains the use of artificial intelligence to detect heart disease in non-cardiac CT exams.

Eleven medical societies have signed on to a consensus statement aimed at standardizing imaging for suspected cardiovascular infections.