• Building Effective Clinical Decision Support

    By Lola Butcherfeature-art-clinical-decision-support

    Embracing the power of EHR technology, leading providers are using clinical decision support to
    standardize care, improve compliance with evidence-based protocols, reduce length of stay, and save lives.


    As electronic health record (EHR) technology spreads quickly across the healthcare industry, providers are developing a wide range of clinical decision support tools-such as electronic alerts, order sets, protocols, and smart documentation forms-to improve and standardize clinical care.

    For instance, at Southeast Texas Medical Associates, clinical decision support helps physicians at the multidisciplinary clinic comply with more than 250 quality measures, with physician performance on each measure reported on the clinic's website. "Once you open the books to public reporting, the only place you have to hide is in excellence," says James L. Holly, MD, the practice's CEO.

    In addition to the purchase and implementation of EHR technology, a successful clinical decision support program requires continuous monitoring and updating, which means an ongoing commitment of staff and financial resources. But, within a matter of years, physicians and health systems that do not use clinical decision support will find themselves unable to offer the quality of care that will become routine because of the technology.

    The investment is money well spent, says Matthew Eisenberg, MD, medical vice president-clinical informatics at MultiCare Health System, which is considered one of the more sophisticated clinical decision support users in the country. "Quality prints money," he says. "We have seen our length of stay decrease and we have fewer adverse events. That translates into good, solid financial success for our organization."

    The Right Support at the Right Time

    Eisenberg defines clinical decision support this way: Providing clinicians and patients with clinical knowledge and patient-related information that is intelligently filtered and presented at appropriate times to enhance patient care and improve clinical outcomes (see the exhibit below).


    Clinical decision support is possible because of EHR technology, but EHR systems do not automatically include clinical decision support functions. "There are two ways that you can use an EHR: You can use it as a big, fancy typewriter and file cabinet, or you can actually leverage the information to improve the quality of care, improve patient safety, and improve efficiencies," says Timothy C. Birdsall, ND, FABNO, chief medical information officer, Cancer Treatment Centers of America.

    The following case studies highlight the various ways that progressive organizations are using clinical decision support to improve quality and reduce costs.

    Building templates to support preventive care. Recognized as a patient-centered medical home, Southeast Texas Medical Associates' 24 physicians and 265 other staff have access to an EHR system in each of the practice's five clinics as well as four hospitals, 28 nursing homes, and anywhere else they have an Internet connection.

    Clinical decision support-screening tools, reminders, templates, and order sets-guide virtually every patient interaction. For instance, before an outpatient visit, a nurse uses a pre-visit screening tool to identify preventive care needs. In addition, Southeast Texas' disease-specific clinics for diabetes, congestive heart failure, and other conditions are all supported by template-driven disease management tools. These help physicians and nurses provide consistent, evidence-based monitoring, treatment, and coaching for patients with these medical conditions.

    Age-specific benchmarks built into the EHR system alert clinicians if, for example, a child is overweight and this information helps all members of the care team give consistent information as they coach patients and their parents on creating healthy habits. Meanwhile, EHR templates and order sets help ensure that women get appropriate gynecological exams and other needed tests.

    "We transformed our vision to focus less on how many X-rays and lab tests are done and to focus more on ensuring that patient care meets measurable standards of excellence and on taking actions to reduce the cost of care," says CEO James L. Holly, MD.

    The multidisciplinary practice, in Beaumont, Texas, is supported by a six-member IT department. It has spent $8 million on EHR technology since 1999, says Holly. Of that, at least $3 million went to building and deploying clinical decision support. The pay-off: The practice has achieved significant results, including the following:

    • Eliminated ethnic disparities for diabetes and hypertension care
    • Decreased preventable readmissions by 22 percent in the past three years
    • Lowered the average blood glucose levels of patients with diabetes significantly below the national average

    Using "hard stops" to improve outpatient care. None of the 93 cardiologists practicing at Cardiology Consultants of Philadelphia would fail to prescribe a beta-blocker to a heart attack patient, if appropriate-because their EHR system would not allow it.

    The system is embedded with quality checking algorithms for nine indicators of good cardiac care. As a physician documents in the EHR during a patient visit, he or she is flagged to address the indicators relevant to that patient's condition. The program will not advance to the next screen until the physician has either indicated compliance with the quality measure or explained the reason for not doing so.


    "These are absolute hard stops," said Scott E. Hessen, MD, the practice's chief medical information officer. "You cannot bypass them."

    Cardiology Consultants introduced these clinical decision support functions in 2007 shortly after implementing its EHR system. Hessen and his colleagues recognized that merely documenting the care being provided did not support compliance with evidence-based guidelines, such as the use of anti-platelet therapy in coronary artery disease or warfarin for patients with atrial fibrillation.

    "To get a report card after you have seen a year's worth of patients-and find out that you got a 'C'-doesn't really help you improve," he says.

    Hessen, who programmed the clinical decision support features for the practice, said it was easy to get his colleagues' support because he proposed starting with six quality measures backed by so much evidence that no one disagreed about their worth. "All of our doctors agreed because they knew absolutely that they would score at least 97 percent or better on all the measures," he says. "And then, after this went live, they were shocked to find out that they didn't score 97 percent on all these measures."

    The lower-than-expected scores did not reflect poor care but rather a lack of systematic documentation of why an evidence-based protocol was not appropriate for a given patient. "More often than not, there's a very good reason, but insurers don't know what that reason is. All they can see is, 'You've only got 80 percent of your patients on a beta blocker so you must be doing a marginal job,'" Hessen says. "Now we have documentation to counter with, 'We have every patient on a beta blocker who needs one, or here's the reason why some patients are not on them.'"

    In addition to finding favor with insurers, Cardiology Consultants receives annual bonus checks from the federal government for its participation in the Physician Quality Reporting System. Plus, it earned a meaningful use incentive check in the first year of that program- thanks, in part, to clinical decision support functions.

    Over time, three additional quality measures have been added for a total of nine "hard stops." Obviously, they don't all apply to each diagnosis, and the EHR displays only those that are relevant for a specific patient's situation. For example, if a patient is a non-smoker, the smoking-cessation advice hard-stop screen does not appear.

    "People shouldn't be afraid of a hard stop or two because their value after the fact is incredible," Hessen says.

    Standardizing care in a stand-alone hospital. As a participant in the Institute for Healthcare Improvement's (IHI's) Impacting Cost + Quality initiative, King's Daughters Medical Center set a goal of reducing its annual operating budget by 1 percent, or $6 million. The biggest share of that savings is coming from clinical effectiveness bundles, or evidence-based EHR order sets and care plans for cardiac surgery, total joint replacements, heart failure, and other common procedures and diagnoses.

    IHI defines a bundle this way: A small set of evidence-based interventions for a defined patient segment/population and care setting that, when implemented together, will result in significantly better outcomes than when implemented individually. IHI's central line bundle includes five interventions:

    • Hand hygiene
    • Maximal barrier precautions
    • Chlorhexidine skin antisepsis
    • Optimal catheter site selection, with avoidance of using the femoral vein for central venous access in adult patients
    • Daily review of line necessity, with prompt removal of unnecessary lines

    Angela Graham, RN, director of quality outcomes, King's Daughters, said the most important part of creating a clinical effectiveness bundle is fixing the broken processes that are responsible for waste. "You have to walk through each step that the patient has to go through and ask, 'Does it lend value? Is the process the best that we can make it?'" she said. "The goal is to get as close to perfect as you can."

    The bundles concept is a very good approach to reducing unwarranted variation in clinical practices, says Katharine Luther, RN, vice president of IHI's hospital portfolio planning and administration. "We know that variation in practice is one of the single biggest drivers of cost and a contributor to poor quality. These bundles can also be a time-saver for clinicians, organizing their work and allowing them to spend time on adapting and individualizing care for patients."

    In the first 18 months of its initiative, King's Daughters, a 465-bed hospital in Ashland, Ky., saved nearly $2.1 million by implementing clinical effectiveness bundles for 13 common conditions. In addition, the medical center improved patient satisfaction and started to win buy-in from some recalcitrant physicians who were not on board at the program's inception.

    "Our physicians, just by nature, are extremely competitive," says Graham. "So when we tell Dr. B that Dr. A is using the bundle and that his results are better, they instinctively want to start using it."

    To create these bundles, Graham assembles a multidisciplinary team, including physicians, case managers, pharmacists, nurses, physical therapists, and other relevant staff to discuss how the current processes compare to evidence-based best practices. Physicians are recruited to help create the order sets that will drive the new care process. Each physician's performance on an array of measures specific to the clinical effectiveness bundle-for example, percentage of medications prescribed that are not on the order set, length-of-stay, and cost per case-is monitored on a dashboard.

    Using clinical decision support to save lives. In a bid to reduce death from sepsis, MultiCare, a four-hospital system based in Tacoma, Wash., has developed a standardized approach to sepsis management. Unlike typical order sets for a given diagnosis, these orders are issued in a step-wise fashion that reflects changes in a patient's health status. Each time the patient's clinical parameters are updated in the EHR, an algorithm determines the next step in the care plan.

    "You place some orders, see how your patient is progressing and, based on that status, you go in and place additional orders," Eisenberg says. "We've seen a substantial drop in our sepsis mortality, and this is really helping to be a differentiator in terms of our ability to deliver quality care."


    Similarly, the health system developed an order set to standardize its approach to the intravenous use of heparin, a high-risk anticoagulant that is often associated with medication errors and complications. After implementing the order set, MultiCare saw medication errors decrease and associated costs decline.

    "This is an important collaboration with our pharmacists because there's an opportunity to co-manage these patients through the pharmacy, with nurses monitoring to make sure patients are safely treated with intravenous heparin," Eisenberg says.

    MultiCare is one of the more advanced clinical decision support users in the country, with an inventory that includes:

    • Alerts, warnings, and reminders to improve drug safety
    • Clinical checklists, protocols, and pathways
    • Features that improve diagnostic accuracy and test appropriateness
    • Order menus, order sets, and order calculators
    • Intelligent clinical data processing and monitoring

    Clinical decision support, along with other EHR features, has helped MultiCare decrease its average length of stay from 4.32 days in 2006 to 3.85 days currently, says Vince Schmitz, the system's CFO.

    Getting Started

    Based on their experience, the four providers featured in this article offer the following tips for launching clinical decision support.

    Don't look for off-the-shelf products. Because clinical decision support is based on evidence-based guidelines, it may be tempting to think that one health system's clinical decision support can be used by another. However, national best practice guidelines are built on varying levels of evidence. So each health system must set its own standards. "In some instances, there's pretty clear evidence, and it's pretty easy," Eisenberg says. "But if there's a gap in the evidence, you still have to get people in a room to agree: 'What's our best practice?'"

    Another barrier: Clinical decision support functions have to be programmed to work in each health system's EHR. "This is not yet even close to plug-and-play," he says.

    Plan ahead to avoid user fatigue. One of the most common clinical decision support functions is alerts and warnings related to drug/drug interactions. The information for that functionality comes from proprietary pharmacy databases. Rules within the EHR system grade the severity of the drug/drug interaction and identify the level of evidence that supports a concern. "If you don't filter out nuisance drug/drug interactions, you have a tremendous problem with alert fatigue," Eisenberg says. The drug databases are updated every month, so reviewing and filtering drug/drug interactions requires ongoing work.

    Hessen at Cardiology Consultants of Philadelphia says "click fatigue" is also a problem. "What the doctors loathe the most is the fact that you have to sit there and endlessly click, click, click," he says. "Our lesson is you have to minimize the clicks whenever possible. You have to automate as many things as can be automated."

    Make life easy for clinicians-and patients. Whenever possible, clinical decision support tools should be context-sensitive to the patient. When a patient visits a MultiCare primary care office, the EHR screen will alert clinicians to that patient's individual health maintenance requirements-screenings, immunizations, and medication prevention, such as aspirin therapy-based on the recommendations from the U.S. Preventive Services Task Force.

    The list can be provided to the patient via MultiCare's patient portal, the patient's personal health record, or a printed after-visit summary. "This is a great example of where the information is personalized for that patient," Eisenberg says. "If a patient has had a mammogram, that should be off the list. If she has not and it is due, that should be indicated-specifically for her."

    Providing patient decision support. Southeast Texas' most innovative use of clinical decision support may be the functions that help patients make decisions about their health habits. Shortly after the medical home introduced its EHR system in 1999, Holly recognized that its power would only be realized if the practice moved to electronic patient management. That meant designing disease management and population health tools, including follow-up documents for patients that summarize the patient's health goals and the steps needed to achieve them.

    Holly's favorite example: the use of Framingham Cardiovascular Stroke Risk Assessment tool to help patients understand their risks of a heart attack or stroke within the next decade-and how they can reduce those risks. At each patient visit, the EHR system runs 12 Framingham calculations, including one that compares a patient's "relative heart age" based on his or her health factors with the patient's "real heart age." The system then automatically creates a report that shows "what if" scenarios that help a patient understand how lifestyle changes can decrease his or her risk.

    For example, if a 45-year-old man with several risk factors has a "relative heart age" of 80, his 10-year risk is 30 percent. The report shows that, if he improves his blood pressure and other modifiable risk factors by 20 percent, his risk drops to 18.4 percent.

    Another innovation: Southeast Texas' LESS (Lose Weight, Exercise and Stop Smoking and/or Avoid Second-Hand Smoke) initiative. Using a template-driven process, nurses use the EHR to calculate six metrics to assess each patient's weight-related health risks at every visit (unless the patient was seen in the previous two months). The system automatically creates a customized exercise "prescription" and, if necessary, an electronic "tickler file" that alerts care coordinators to follow up with a patient about his or her progress in quitting smoking.


    Results: While the nation's obesity crisis has worsened in the past decade, the average body mass index for Southeast Texas patients has remained stable during that time, and the percentage of patients who are overweight increased by less than 1 percent.

    Boil down the information. Southeast Texas Medical Associates boiled down the National Kidney Foundation's 460-page summary of renal disease into a few clinical decision support tools. And it used the American Medical Association's 220-page adult weight management notebook to create two support options: a full version that might take 30 minutes to complete and a short version that takes only one minute.

    "The mistake would be to assume that, unless the entire evidence-based information is used every time, it does not have value," Holly says.

    He encourages those who are not yet using clinical decision support to extend that thinking to the task ahead. "Many people are kind of dismayed by the hugeness of rolling out clinical decision support," he says. "Just accept the fact that maybe you can't do it all at first, but you can get started."

    Addressing the Challenges

    Introducing and maintaining clinical decision support requires careful planning and adequate management resources, says pioneers.

    Managing tool development. Whether it's a template, an order set, or an alert, any poorly designed clinical decision support function is worse than none at all. A multidisciplinary team of clinicians, operational leaders, and IT specialists must collaborate. The process includes many steps: analyzing workflow, building the feature, testing it, communicating about it, and more. "You need to follow mature project management processes, and you can't do this on the cheap," Eisenberg says.

    As much as possible, the development process should consider how new clinical support-driven processes will affect the cost of care. For example, the first year after Southeast Texas Medical Associates implemented a screening tool to identify patients who needed immunizations, it spent an extra $1 million in buying vaccines. "Because many of those are given every 10 years, that initial investment paid off in the next nine years as the costs dropped dramatically," Holly says.

    Working with physicians. The physicians who helped create the clinical effectiveness bundles used at King's Daughters are enthusiastic about their benefits, but other physicians have proved difficult to deal with. "They felt like it was cookie-cutter medicine-that's the terminology that they used quite often," Graham says.

    Getting competing orthopedic practices to agree on an order set and care plan proved particularly difficult. Development of the bundles was driven by hospital management, but reaching a consensus ultimately had to be delegated to physicians. "We said, 'We expect you to create this because it is yours, and you are going to be the ones using it. How can we make it easy and better?" she says.

    On the other hand, King's Daughters underestimated the engagement of some physician groups who were eager to see how the clinical effectiveness bundles affected their care. "They immediately said, 'Where are my results? I want to see how I'm doing,'" she says. "And we didn't have the results ready right off the bat. So that was a big lesson learned for us."

    Ongoing Maintenance

    MultiCare has nearly 500 inpatient and emergency department order sets, plus additional sets for outpatient care, and all of these must be updated when new guidelines and best practices emerge. "Having a governance structure and a process for doing that is important," Eisenberg says. "The work doesn't stop after implementation. And it takes a lot of time and effort."

    He monitors the use of order sets to identify any that are not being used. That may indicate that work is needed to inform clinicians of the availability or value of the order sets-or it may indicate that they need to be removed. For example, MultiCare is retiring some neonatology order sets that are rarely used because physicians have saved the orders on a "personal preference list" in the EHR system. "It's kind of like a tree-you have to keep taking care of it and some limbs need to be trimmed back so that the order set catalog is alive, thriving, and used," he says.

    Access related sidebar: Preparing for Mobile Clinical Decision Support

    Lola Butcher is a freelance writer and editor based in Missouri.

    Interviewed for this article (in order of appearance): 

    James L. Holly is CEO, Southeast Texas Medical Associates, Beaumont, Texas (Jholly@setma.com).

    Matthew A. Eisenberg, MD, is medical vice president for clinical informatics, MultiCare Health System, Tacoma, Wash. (Matthew.Eisenberg@multicare.org).

    Timothy C. Birdsall, ND, FABNO, is chief medical information officer, Cancer Treatment Centers of America, Schaumburg, Ill. (tim.birdsall@ctca-hope.com).

    Scott E. Hessen, MD, is chief medical information officer, Cardiology Consultants of Philadelphia, Philadelphia. (ScottH@ccpdocs.com).

    Angela Graham, RN, is director of quality outcomes, King's Daughters Medical Center, Ashland, Ky. (Angie.Graham@kdmc.net).

    Katharine Luther, RN, vice president of IHI's hospital portfolio planning and administration, Cambridge, Mass. (kluther@ihi.org).