by Mark Hagland
When executives at MemorialSloan-KetteringCancerCenter decided in the late 1990s to expand their clinical operations beyond the organization's core 432-bed inpatient hospital facility on Manhattan's Upper East Side, they realized that automation would be required. Creating six satellite clinics across the New York metropolitan area (including Westchester County to the north, New Jersey to the west, and Long Island to the east), the renowned cancer center's leaders wanted to enhance the availability of their services to patients across the metro area.
Faced with the potential for a medical records and clinical decision support nightmare, Memorial Sloan-Kettering's leaders decided to forge ahead with a comprehensive automation plan, implementing a system-wide electronic medical record (EMR), including a picture archiving and communications system (PACS).
"We decided we wanted to become broader; but we realized quickly that we couldn't have an efficient operation while moving paper records and diagnostic images around," explains Patricia Skarulis, Memorial Sloan-Kettering's vice president and CIO. "So we imaged our paper records, installed our core EMR and PACS system, and got rid of all our films, and did this while we opened the new facility in Midtown."
Skarulis describes the synchronicity of the expansion and the need to move forward into automation as "fortuitous." Now, not only are all of the organization's 600 staff physicians and its nurse practitioners entering orders through its CPOE system, the organization's clinicians have been working collaboratively with its EMR vendor (Atlanta-based Eclipsys) to customize and optimize oncology order sets, confirms David Artz, M.D., Memorial Sloan-Kettering's medical director of information systems.
The work has been complex and time-consuming, but the results have been very positive, Artz says. To take just one example, he cites the situation around blood clots, something to which many patients are post-surgically disposed. Depending on the type of surgery, most patients (but definitely not all) should receive low-molecular-weight heparin, an injectable drug. At Memorial Sloan-Kettering, for patients with a predisposition to blood clots in their legs, the order for low-molecular-weight heparin is now forced, meaning that any physician who wishes not to order heparin under designated conditions must consciously and specifically choose not to do so. "Preventing blood clots in cancer patients' legs is one of the hot-button topics in cancer care," Artz explains. And this particular innovation helps to avert a common oversight in that area.
Such innovations speak to the core of what information technology can offer physicians in complex clinical areas as cancer care, Skarulis adds. "If you didn't have that particular order set," she says, "you'd have to be trying to educate everyone, and checking on whether they did it. This way, it forces them to act, and we also can do analysis," to determine what is happening with individual patients and with the organization's patients collectively, and can act on data-facilitated findings.
Meanwhile, in terms of clinician efficiency, Memorial Sloan-Kettering has self-developed an application built into the Eclipsys EMR, one that gives a broad view of any patient's illness at a glance, with fields pre-populated in the system that show medication history and current meds, patient demographic data, and current physician notes. The application, Skarulis and Artz note, is especially helpful for residents, as they do handoffs of daily patient oversight. In fact, its implementation dovetails with the publicly expressed desire on the part of the Oakbrook Terrace, Ill.-based Joint Commission on Accreditation of Healthcare Organizations (JCAHO) that hospital organizations do a better job of patient supervision handoffs, a widely observed point of medical error.
Optimizing order entry
The experience at the 744-bed NorthwesternMemorialHospital in Chicago is illustrative of one of the biggest challenges — and opportunities — in automation of clinical processes. That's because off-the-shelf software programs inevitably must be custom-tailored and optimized for the use of clinicians in particular organizations, says David Liebovitz, M.D., the organization's medical director for clinical information systems. In fact, Liebovitz reports that, contrary to the typical approach taken to developing order sets for CPOE use, at Northwestern, "We did not build many order sets based on medical diagnoses, but rather, we built them based on surgical procedures. The reason for that is that the average patient in an academic medical center has so many complex situations —multiple myeloma, COPD (chronic obstructive pulmonary disease), while happening to have community-acquired pneumonia, for example, coming in. So building order sets based on medical decisions seemed inappropriate," he says. As a result, Liebovitz and his colleagues spent months building academic medical center-appropriate order sets for their physicians.
In addition, Northwestern clinicians and IT professionals completely reworked the medication order alerts in their vendor's CPOE system. As Liebovitz explains it, "The standard vendor alerts built into applications are too jumpy. It's in the literature that 90 percent of the alerts are ignored or turned off. So we turned them off entirely," painstakingly creating their own alerts in order to adjust the proportion of times when a physician entering a medication order would receive an alert "ding."
For example, he says, rather than having a physician be alerted when ordering aspirin and Coumadin together — a situation whose risks he says virtually all physicians would be aware of — the system instead focuses on alerting for more unusual combinations of drugs, whose adverse interactions are less well known: the combination of the antifungal drug Itraconazole and the cholesterol drug Simvastatin (Zocor), for example. It is only when such customizations are created that patient care quality and clinician workflow can be significantly improved, Liebovitz says.
Improving clinical workflow has also been a major focus at Novi, Mich.-based Trinity Health. Narendra Kini, M.D., the 24-hospital system's chief medical information officer, reports that executives at Trinity spent about a year analyzing and reworking clinical workflow, before completing EMR implementation at several of the hospitals that had already begun.
Among many changes made has been the initiative around acute myocardial infarction (AMI), or heart attack, care. Based on a recently modified recommendation from the Washington, D.C.-based American College of Cardiology (ACC) in that clinical area, Trinity clinicians wanted to decrease the so-called "time to needle" — the amount of time from the moment a patient arrives in the emergency department with chest pain until the moment at which a needle is injected to start a stent in the cardiac catheterization lab.
Two years ago, the ACC's recommendation was 120 minutes for "time to needle," but last year, Kini notes, the specialty society changed that to 90 minutes. While there is variation among different Trinity Health System hospitals, clinical workflow improvements, facilitated by the system's EMR, has been improving "time to needle" at the system's facilities.
Essentially, Kini explains, "The faster you can get the initial EKG into the cardiologist's hands, the faster you can get the patient into the cath lab. And to improve that process, you need to digitize the EKGs so the results can go from the ambulances directly into the cath labs of the hospital facilities." In addition, at least one test is needed to confirm EKG results, "and we're beginning to try to address how to bring that test to bear as soon as possible," he says.
Reengineering the process from the paper-based world — in which an EKG would arrive as a paper strip in the ED, and then be faxed to a cardiologist's home or office — to an automated process in which an EKG would be automatically forwarded to the clinical data repository and messaged out to the doctor's PDA — is one of the tasks Trinity has been taking on.
In some cases, a leadership facility has emerged from within a health system, one that has broken new ground on EMR -based performance improvement, and whose innovations are being used to spur similar work at sister facilities. That's what's been happening at St. Vincent's Health System in Birmingham, Ala., confirms Senior Vice President and CIO Tim Stettheimer.
The core EMR system at St. Vincent's has evolved considerably since the go-live of CPOE two years ago, he reports. And because St. Vincent's has been strategically selected as the showcase hospital for the Ascension Health System — the 70-hospital Catholic system based in St. Louis that is the nation's largest Catholic system — the innovations taking place at St. Vincent's are rippling out across the Ascension galaxy, he confirms.
One of the most exciting areas of progress, Stettheimer says, has been in the development of true closed-loop medication management, which has been reducing medication errors across St. Vincent's. Getting to true closed-loop is quite a challenge, Stettheimer notes, as it requires full CPOE integrated with a sophisticated pharmacy system and an electronic medication administration record system (eMAR), something only a tiny percentage of hospitals have yet to fully implement, he notes. As the various Ascension hospitals move forward on their EMR path toward such advanced innovations, St. Vincent's work will help guide its sister hospitals forward, he says.
The lesson in all these case studies is clear, says Trinity's Kini. "Going live itself is a pretty rigorous process, because it involves a cultural change as much as a technological change. But going live just means setting up a basic framework," he adds. "After you go live, you have to change the workflow and then embed the content. You never get systems live out of the box with all the evidence-based guidelines, appropriate alerts, and so on. These are all things you need to build in, and then tweak in order to meet the needs of your clinicians. And you have to optimize it all. It is both the adoption process and the outcomes-based use of these systems that takes a substantial amount of time."
Mark Hagland is a contributing writer based in Chicago.
