Clinical Decision Support: The Journey Begins
Is it goodbye to radiology benefit management (RBM) companies and hello to automated decision-support systems? Not really, as the two aren’t mutually exclusive. Nonetheless, computerized decision-support tools are gaining ground in the outpatient setting. With CMS about to begin a Medicare decision-support demonstration project for high-end outpatient imaging, and with Minnesota embarking on a consortium-mandatory, statewide decision-support installation, automated protocols that guide referrers to the most appropriate high-end imaging tests for their patients are about to be put to the long-term test. Cally Vinz, RN, vice president for clinical products and strategic initiatives at the Institute for Clinical Systems Improvement (ICSI), Bloomington, Minnesota, is a true believer in clinical decision support. She calls it a win–win–win for referring physicians, health insurers, patients, and even radiology providers, who won’t have to spend time straightening out inappropriate requests for CT, MRI, and other high-tech imaging exams. The enthusiasm that Vinz expresses for clinical decision support is based on experience with a pilot project in Minnesota that has outlived itself by two years. Those results have been so positive, she says, that health-insurance providers in Minnesota are now paying to implement clinical decision support statewide and to make it mandatory for ICSI members. Minnesota is also turning to a commercial decision-support vendor to upgrade its system. If the rest of the nation wants to ride Minnesota’s coattails, Vinz says, there’s no reason that clinical decision support can’t be implemented nationally to hold down costs and guide referring physicians to appropriate imaging tests. “We hope this is how it gets done across the United States,” she says. “I could envision one set of appropriateness criteria across the country.” That would be a big step, but even states that have studied Minnesota’s program are proceeding slowly. The Washington legislature, in 2009, created an Advanced Imaging Management Workgroup to study and implement evidence-based decision making for high-end radiology, with the goal of holding down cost. The workgroup did exhaustive surveys and issued guidelines to which health plans and providers have agreed. Now, according to Jeff Thompson, MD, medical director of the state’s Health Care Authority (Lacey, Washington), both RBM prior authorization and decision-support automated tools are being used. Some health care providers, Thompson says, have been given a gold-card pass to use existing decision-support systems developed in-house. For other health plans, the use of a single RBM is allowed. For state-run health programs—Medicaid, worker’s compensation, and the Public Employees Benefits Board—a commercial physician review organization has been hired either to implement clinical decision support or to invoke prior authorization, depending on the nature of the requested imaging tests and the described clinical indications, Thompson says. For all the ink being expended on clinical decision support at the moment, many problems are still to be solved. Chief among them are what Thompson calls hard stops—the outright denials of requested imaging exams, for which RBMs are notorious. Also a caution with clinical decision support, Thompson adds, is that the embedded scoring on appropriate use for imaging exams is based primarily on ACR® utilization guidelines or those developed by other medical specialty societies. These guidelines—on which decision-support software programs rely—often lack evidence-based data and are more like recommendations, Thompson says. “We looked at Minnesota, but we don’t know if that program had a return on investment (ROI) that reduced overutilization; we haven’t seen any published studies,” Thompson says. “I think it’s a rule of thumb that we believe RBMs would have the higher ROI, but the issue is that they cause a lot of disruption of care or access, so we want to balance that in looking at utilization strategies.” It is this very disruption of access, however, along with the burden and expense of using RBMs’ prior authorization, that has turned Minnesota away from RBMs and in the direction of clinical decision support. ICSI’s Savings Minnesota’s ICSI is a collaborative funded by member health insurers and health-care providers within the state. The five largest ICSI members operate health plans and/or run hospitals and clinics that cover and care for about half the state’s population of 5 million, according to Vinz. Vinz says that one of ICSI’s five largest health-care providers found, during the initial year of ICSI’s decision-support pilot project, that physician staff spent an average of 10 minutes per advanced imaging order responding to prior-authorization restrictions put in place by RBMs. In contrast, the time needed to use computerized decision support and get authorization, Vinz says, was 10 seconds. That’s how long it takes for the ordering physician using clinical decision support to click through the steps to arrive at the appropriate imaging test, whether that turns out to be simple radiography or an MRI or CT study, Vinz says. Barry Bershow, MD, is vice president for quality and interim executive regional medical director for Fairview Health Services, Minneapolis, Minnesota, a major hospital/clinic operator that is one of the ICSI big five. Fairview Health Services ran the data that Vinz cites, and Bershow confirms the findings. “We found that staff was averaging over 300 hours per month talking to RBMs,” Bershow says. “When we threw the switch to clinical decision support, it was 10 seconds per hit. We feel we have, so far, saved staff time of 300 hours per month—times 36 months.” Bershow doesn’t have a dollar figure for those savings, he says, but multiplied across ICSI, they are considerable. He says, “The health plans are happy enough with the results that they’re paying the total costs for the decision-support vendor, so that for all the providers covered by the health plans, the health plans pay the per-click costs.” Vinz estimates that the total staff time saved amounts to the equivalent of five FTEs. The cost of running clinical decision support amounts to one-eighth of the expense of using an RBM, she says. More Savings Staff time turns out to be only a small part of the total savings that ICSI attributes to using its decision-support system. At the time the pilot began in 2007, according to a recent ICSI white paper,¹ high-end imaging costs in Minnesota had been growing at the rate of 8% per year for three years. After the decision-support system was put in place, this growth flattened dramatically. For the past three years, high-tech diagnostic imaging growth has been close to zero. The savings attributable to the use of clinical decision support during that time are estimated at $84 million, according to the white paper. Clinical decision support might result in better exam selection than prior authorization does, too. In one Minnesota experiment, where decision-support ordering was compared with prior authorization, prior authorization resulted in 79% of orders meeting appropriateness criteria. For clinical decision support, that success rate was 89%, the ICSI white paper reports. Bershow says that clinical decision support has been so successful that none of the state’s health insurers or state health programs now require prior authorization. Minnesota is on its way to becoming a decision-support state. Even for Medicare, there have been savings because of clinical decision support, Bershow adds. Despite the numbers of Medicare patients in Minnesota growing by 25% since the decision-support pilot began, there has been, overall, a slight decrease in the number of Medicare high-tech diagnostic imaging tests performed. Bershow says, “We think this clinical decision support is a highly effective system.” Brian Rank, MD, is an oncologist who also serves as medical director for HealthPartners Medical Group. The group’s parent company, HealthPartners, Bloomington, Minnesota, is both an insurer and a health-care provider. Its network includes three hospitals and 70 clinics, and it insures about 1.25 million members. The medical group, Rank says, has about 700 physicians and serves more than 425,000 outpatients. Rank says that HealthPartners was the first entity to pioneer clinical decision support in Minnesota, using a homegrown application that was later offered to other Minnesota health plans for the ICSI pilot study. He adds that the new decision-support vendor that ICSI is using will create access to far more data than could be compiled with the original homegrown system. Rank credits Minnesota employers for the push to deploy clinical decision support. Health-care costs were skyrocketing and caregivers were being dragged down by the cumbersome and entangled prior-authorization processes involving multiple companies and telephone contacts, he says. Now, as clinical decision support is being rolled out statewide, it’s those same employers who will see the financial benefits. Rank says, “Our ability to hold down costs allows them to continue to offer health care to their employees.” Medicare Patients, Too Medicare and Medicaid patients in Minnesota will now be served by clinical decision support as well. According to Vinz, the Minnesota health plans, along with health-care providers, have agreed to use the decision-support system for Medicare and Medicaid patients, even though those patients aren’t enrolled in the health plans. “We don’t want them to sort the patients,” Vinz says. “We don’t want different patients getting different care, so we have contracted with providers for all patients to have the system used for them.” There is no contract or formal approval from CMS, however. Vinz says, “CMS is not requiring us to do something different right this minute.” This is significant because CMS is in the process of rolling out its Medical Imaging Services Demonstration to test clinical decision support on Medicare patients nationally. That test might determine whether CMS uses clinical decision support or turns to RBMs to control utilization (CMS now uses neither decision support nor RBMs). Minnesota’s Medicare-patient pool of a nearly 2 million will be far larger than the patient pool in the CMS demonstration project; ICSI is hoping, Vinz says, that CMS will look at its results, as well as the demonstration project, in judging how well clinical decision support compares with RBMs’ prior authorization. “Our goal is that CMS will accept clinical decision support as an alternative to RBMs and won’t require anything different from us for a Medicare claim,” Vinz says. Leveraging One EMR Many who discuss Minnesota health care use the word collaborative. Not only is ICSI itself a collaborative, but the decision-support rollout will use a single decision-support vendor chosen by the participants in ICSI. That vendor will have an easier time interfacing its decision-support software with health-care providers’ electronic medical records (EMRs) because only one EMR product is in use in most of Minnesota. The use of a single EMR product reflects another way in which Minnesota providers have collaborated to achieve uniformity and simplicity. Vinz says that even though there’s only one EMR product, a separate decision-support interface will have to be designed for each provider network. The EMR vendor will do each installation, Vinz adds. She estimates that even for large caregivers, however, the interface won’t cost much more than $10,000—a negligible amount, given the potential for savings. For rural or small-practice physicians who don’t have an EMR system, ICSI and its decision-support vendor have created an Internet application that will allow these small providers to use clinical decision support, too. Their cost for in-office connection will be about $1,000 each, Vinz estimates. This represents the cost of a laptop computer and a wireless router to connect to clinical decision support using the ICSI website. The beauty of having either EMR or Internet access to clinical decision support is that ICSI can make the use of clinical decision support mandatory for its members—and for any other Minnesota provider choosing to join the program, which is open to clinics and physicians statewide. Mandatory use will be simple to enforce, Vinz says. If a referring physician ordering an imaging test doesn’t enter a decision-support code, then insurers won’t pay for the exam. In order to get a decision-support code, the referring physician will have to complete the decision-support process to arrive at the most appropriate imaging test. The exception is that, in some cases, the radiology provider will be the one to apply clinical decision support to an order, and the decision-support number will be entered after the initial order. Some radiology groups are offering to run clinical decision support as a service to referrers, in fact. Jim Trevis is ICSI’s director of marketing and communications. He makes the point that whether the decision-support system is used on the front end by referrers or on the back end by radiologists, the data hub will be the same. The locus will be the decision-support tool. “The appropriateness criteria are in the tool, and the user will either feed something in or pull something out of it,” Trevis says. Vinz points out that health insurers will also have access to their insured patients’ records and will be able to see whether clinical decision support has been used when their physicians ordered imaging tests. If not, the tests won’t be paid for, she says. All the interfacing and the transmission of data, Vinz adds, will be carefully protected to meet federal patient-privacy regulations. Rolling Out Slowly Because the decision-support application in the ICSI pilot project was different from the commercial decision-support system that ICSI has since chosen, the actual rollout of the permanent, statewide decision-support system is just starting in Minnesota. Different providers will implement the new decision-support system as interfaces are completed and after testing is done. Rank says that HealthPartners is upgrading its EMR and will add the new decision-support application in the second quarter of 2011. “I don’t think there will be any integration issues,” he says. He adds that one of the benefits of the decision-support system will be that referring physicians will get immediate feedback on the appropriateness of the studies that they order—something that did not happen with RBMs. “Personally, I think the RBMs are indiscriminate,” Rank says. “They make it hard for the people at the point of care to find the right study, but a physician needs that support.” As a referring oncologist, Rank reports that the decision-support system is easy to use. ”I just type in a few words, like ‘evaluate liver metastases,’ and then I choose the exam. The worst thing that used to happen when I ordered a study was that at the end of the radiology report, there would be a note that if I’d ordered an MRI instead I’d have gotten the information I needed. We’re trying, with clinical decision support, to put that information up front for physicians,” he says. Vinz says that training referring physicians to use clinical decision support will be nearly effortless. “The training is a one-time shot. For Web users, it’s an introduction, and then it becomes intuitive, like Amazon or PricePoint,” she says. For clinics, decision support is accessed through the EMR, so the training will amount to learning clicks. One major health network is integrating all 75 of its clinics at the same time, Vinz says. Radiologists Give Support Despite the benefits of clinical decision support for insurers, clinics, and referrers, radiology groups might lose some imaging volume because of the greater focus on appropriate utilization and the consequent reduction in unnecessary exams, Vinz notes. The prospect of lost volume does not concern Steve Fischer, CIO for Center for Diagnostic Imaging (CDI), a multistate radiology practice that is headquartered in Minneapolis and that operates about a dozen outpatient imaging centers in Minnesota (as part of its much larger national network). Fischer says that CDI welcomes the technology—so much so that it is providing a link between referrers and clinical decision support until integration with EMRs can be completed. “In Minnesota, we had more than 18,000 physicians who referred to CDI last year. Almost all of those referrals were for high-tech imaging,” Fischer says. “The vision is to have clinical decision support at the patient’s point of care, but since most EMRs don’t, as yet, support high-tech imaging clinical decision support, we’re bridging the gap by incorporating clinical decision support in our order-intake process.” CDI has set up a physician portal that allows referrers to issue exam orders that CDI will then run through clinical decision support. That’s what CDI is doing for the Williams IntegraCare Clinic in Sartell, Minnesota (one of many rural clinics that the radiology group serves). Mark R. Halstrom, MD, is a primary-care physician and is currently the only full-time physician on staff at Williams IntegraCare Clinic, which is largely a chiropractic clinic. Halstrom says that CDI has taken over all the preauthorization requirements for his clinic. He fills in queries on the clinic’s EMR and then sends the order sheet to CDI by fax. The radiology practice then schedules the patient for the exam. “As far as I know, I’m not getting denials from the insurer,” Halstrom says. “I assume CDI is doing its job to get its bills paid. If there’s a problem, I expect CDI to notify me of that.” Fischer says that the radiology group will be obtaining the required decision-support confirmation numbers through the group’s interface with the ICSI website. “From our perspective, since we’re dependent on the referring physicians, we make it easy and convenient for them to order,” Fischer says. “Clinical decision support has allowed us to make sure they’re ordering the right exam and that there’s not overutilization.” Fischer says that the decision-support ordering process has another big advantage: It encourages providers to accumulate patient data that radiologists can use to understand the images that they are interpreting better. In this way, patient care is upgraded, he notes. When CDI radiologists have finished their interpretations, those reports, under the new decision-support format, will be sent back to ICSI as blinded data. They are then broken down by an analytical software system and sorted for positive and negative findings and other outcomes, Fischer says. “At some point, there will be some sort of yardstick to measure the quality of radiologists’ interpretations,” Fischer says. “None of that was in the ICSI pilot. This is where no person has gone before; it’s going to take a number of years to mature.” It is through the analysis of referrers’ ordering patterns and radiologists’ findings that advocates of clinical decision support hope to stymie critics who accuse clinical decision support of not being sufficiently evidence based. “The problem is that the ACR guidelines have a lot of holes,” Bershow says. “About 30% of orders don’t have good decision support built into them.” This is exactly the criticism that planners in the state of Washington have expressed. Bershow says, however, that the new decision-support product that ICSI is using appears to have added sophistication and to have improved the decision-assistance capability of the tool. “We’ve had a number of times where the physicians have said they had been ordering the wrong test for years,” he says, “but now they’re guided to the right thing.” Vinz says that the opportunity to compile useful outcomes data is more likely to be present with clinical decision support than with RBM use. “The RBMs need to position themselves to be more transparent,” she believes. “The provider isn’t learning why they deny the exam. In addition, the RBM data are based just on utilization. With decision support, we have the opportunity not just to get utilization data, but to correlate those data with radiological exams and identify the impact on patient outcomes. High-tech imaging’s evidence is not far along, so this will help get feedback into the appropriateness criteria.” Bershow says that patients often arrive at the physician’s office demanding a CT or MRI exam. The decision-support tool, displayed on a screen in the physician’s office, has made it “a lot easier for physicians to stand up to the patient,” he says. As Vinz notes, with clinical decision support, the patient won’t be blindsided when a test is denied a day or more later, as can happen with RBMs. The authorization, through clinical decision support, is part of the ordering process. Washington’s Different View Since the state of Washington passed legislation in 2009 to require public and private health entities to take steps to control the cost of high-tech diagnostic imaging, the workgroup that the legislation created to study cost control has been listening to proponents of both clinical decision support and RBMs. As Thompson says, “The different players have been at the table throughout these discussions.” It’s not surprising, from a political standpoint, that Washington has approved the use of both an RBM and a decision-support system, in addition to creating gold-card status to let some providers continue to use in-house utilization-control mechanisms already in place. As Thompson notes, though, the larger goal has been to achieve consistency in approving high-tech imaging exams for certain uses, whether through a decision-support system or an RBM. The idea is that consistency is achieved when clinical evidence establishes clearly appropriate or inappropriate uses for high-tech diagnostic imaging. Even before Washington passed legislation, Thompson says, Medicaid had restricted PET largely to lung- and gastrointestinal-cancer exams. The state is trying to establish strict utilization guidelines for other advanced modalities, he adds, but the task is not easy. “Everybody’s got guidelines,” Thompson says. “Everybody’s got clinical decision support; everybody’s got an RBM. Some RBMs use clinical decision support and some decision-support systems are starting to look a little bit like an RBM. Everybody says that his or her system can do it better, but there’s limited information on saving money, based on what any baseline of utilization is.” So far, Thompson says, in addition to PET, the state workgroup has focused on nonspecific abdominal pain as a common inappropriate indication for CT exams. It is studying other states and other data to isolate other tests. “We’re trying to go to the source data to find out who’s done the homework,” Thompson says. The Gold Card Virginia Mason Medical Center (Seattle, Washington) is well known for a method of continuous process improvement that it uses throughout its operations. The Virginia Mason Production System (VMPS) is modeled on Japanese production techniques known to maximize efficiency. The VMPS is a fitting parent for an imaging-utilization technique like clinical decision support. It’s no surprise, then, that according to radiologist Craig Blackmore, MD, MPH, Virginia Mason Medical Center has been using an in-house decision-support system since 2005. Virginia Mason Medical Center’s hospital is licensed for 336 beds; Virginia Mason Medical Center also operates a network of outpatient clinics. It is one of the Washington health systems to be given a gold card to continue its own imaging-utilization program. Blackmore, an expert in evidence-based medicine who has a background in public health, is scientific director of the Center for Healthcare Solutions at Virginia Mason Medical Center. “The big difference between Virginia Mason Medical Center and Minnesota,” he says, “is that Virginia Mason Medical Center is targeted at specific imaging studies and indications. Where there is a lot of utilization and good evidence of overutilization, we try to fix those problem areas, instead of using more of a shotgun approach.” Where there is evidence to support inappropriate ordering, referring physicians are barred from ordering those imaging tests, Blackmore says. He uses the example of an MRI exam for lower-back pain on a patient’s first visit. “We look at the whole continuum of care for the patient, not just radiology,” he says. “We look at lower-back pain, but we also identify other benefits, like physical therapy.” A patient complaining of lower-back pain and wanting an MRI exam at the first encounter might, instead, be routed to physical therapy, Blackmore says. The patient might be disappointed that the physician couldn’t order an MRI exam, but in the end, be glad to get physical therapy. “You have to provide other services of value,” Blackmore notes. He says that Virginia Mason Medical Center has established strict, evidence-based guidelines that are built into its decision-support system to limit the use of several other exams, including MRI for headache and CT for sinusitis. There is always the chance for the ordering physician to consult with a radiologist or another specialist, Blackmore stresses. “We work with clinicians and the people who pay the bills to identify those exams that are highly utilized. We drill down for good evidence for what should be used,” Blackmore says. Decision Support Plus By sticking to strict, evidence-based standards, Virginia Mason Medical Center has been able to deny exams on good grounds and limit overutilization for specific exams more effectively than Minnesota’s generalized decision-support tool does, Blackmore contends. “Our rate of imaging is down 20% to 25% in those specific areas,” he says. Blackmore can’t put a dollar figure on those savings, but the cost of care at Virginia Mason Medical Center has been going down and not up, he reports. “Minnesota has reported slowed growth in imaging,” he says, “but we’ve gone further and shown a decrease of 20% to 25% in those areas.” Blackmore doesn’t dispute that clinical decision support is a good tool, and he praises ICSI for its decision-support rollout. He demands more evidence to support appropriateness criteria, though. “Nobody likes preauthorization; it’s effective, but it’s inefficient,” Blackmore says. “Clinical decision support is so much more efficient, and so much easier on providers and patients, that it’s definitely the way of the future.” He continues, “The standard has always been that we can always do what we want if it’s safe and somebody will pay for it, but in a world of limited resources, there is only money to pay for what works. We definitely need more evidence to show what does and doesn’t work.” The likelihood is that ICSI, Virginia Mason Medical Center, Washington, and perhaps the rest of the United States are on the same long path—all headed toward a decision-support system that selects exams based strictly on evidence supporting the best patient care. The vision that Vinz holds of the whole country using something like ICSI’s decision-support tool could be on the mark, after all—just not yet. George Wiley is a contributing writer for Radiology Business Journal