Significant challenges surround the implementation of CPOE systems in a community hospital, particularly when physicians are independent practitioners in solo and small-group practices. Our strategies focused on key physicians, representing a majority of inpatient activity, and targeted them with specific support to maximize their readiness and understanding of the upcoming change environment.  Key among these strategies was assigning each physician to an individual empowered to meet all their educational, technical and support needs.  

A broad enterprise-wide physician steering group was involved with all physician design decisions and charged with aggressively soliciting feedback and broadly communicating their decisions.  We believe our success to date is directly attributable to the early and aggressive use of these and other engagement strategies.  We were also aided by a long-standing effort to partner with our physicians on methods to mutually differentiate ourselves from other regional providers based on quality and outcomes.  Finally, an ongoing focus and priority by hospital leadership kept the project, implementation activities and shared responsibilities embedded in every activity and at the top of every meeting agenda.

MemorialCare is a five-hospital, 1,500-bed system in Southern California that has installed its second CPOE system with excellent results. The first system, TDS 7000, was installed in 1991, but direct physician entry of orders plateaued at 5 to 55 percent at the three MemorialCare hospitals that implemented the application.

Order entry was highest at the children's hospital predominantly served by housestaff and hospital-based subspecialists, with nearly 100 percent CPOE by the neonatology staff who adapted their workflows to system capabilities.

CPOE was 20 percent at the non-academic community hospital that would become the pilot hospital for a new system, and only 5 percent at the largest hospital in the system, a 462-bed adult teaching hospital with a predominantly private-practice-based, voluntary medical staff.

Epic replaced the legacy TDS system at our pilot hospital in 2006, and replaced TDS at the two other hospitals in July 2008, including the children's hospital. From day one at each of the facilities, the CPOE rate has approximated the Leapfrog Group's threshold of 75 percent. Traditional written orders have varied from 1 to 3 percent, never exceeding 5 percent. The remaining orders represent verbal and telephone orders. MemorialCare assigns orders to these various categories, and calculates CPOE percentages according to methods and definitions described elsewhere. The CPOE rates before and after activation of Epic at the three hospitals appear in the figure on page 30.

CPOE became the centerpiece of our patient safety strategy predicated on the Institute of Medicine (IOM) report in 1998. CPOE was adopted by The Leapfrog Group in 2000 as one of its first three mandated improvements, or "leaps," in patient safety. The IOM also defined CPOE as a critical element of the electronic health record in 2003. The Certification Commission for Health Information Technology has recently focused the certification criteria of inpatient clinical systems on those offering CPOE, medication administration and medication reconciliation functions because of their specific potential to improve patient safety, particularly medication safety.

Setting the agenda

MemorialCare made a conscious choice to prioritize CPOE over medication management primarily because of our experience with CPOE (and its general adoption at one of our hospitals), our need to replace an aging mainframe-based system and growing national support for this technology. The initial CPOE strategy included an integrated medical management module with embedded pharmacy verification and electronic medication administration record (e-MAR) components. MemorialCare plans to eventually extend its CPOE system to include closed-loop medication administration as a combination of barcoding, pharmacy packaging and medication distribution projects.

CPOE also served as a primary driver of value in our benefits realization model and the justification provided to governance for the EMR investment - the largest single, non-building project ever undertaken by our system. Greater benefits were expected from higher levels of direct order entry by physicians. The most obvious example is the inverse relationship between labor cost associated with transcribing orders and physician use of the system. The more direct order entry, the less transcription labor, even without counting benefits from minimizing transcription errors or the decreased number of medical errors associated with real-time decision support provided to clinicians during the ordering process.

Our challenges included MemorialCare having some 3,000 or so affiliated, but independent medical staff. MemorialCare doesn't own physician practices. Laws in California and in a handful of other states prohibit the corporate practice of medicine outside of a foundation. MemorialCare physicians are thus truly independent medical practitioners free to admit patients to any facility, not just one of MemorialCare's hospitals. In fact, many "split" their business among various hospitals, some out of necessity given specialization requirements and the contractual directives of patient insurance carriers.

The lack of strong economic alignment and employee relationships with our physicians put a premium on the value proposition of CPOE since physicians could "vote with their feet" by admitting patients to other institutions if change management and CPOE-related initiatives were not handled well. If done poorly, we risked not only poor adoption and low CPOE rates, but also the possibility that physicians would direct their patients elsewhere. And, we were acutely sensitive to the potential of adding one more frustration to the lives of physicians who, along with MemorialCare, struggle with reimbursement, collection, payer mix and other issues.

A variety of factors contributed to the high levels of clinician adoption at MemorialCare. Some of these reflect conventional wisdom with respect to CPOE. Others were unique to our situation although they undoubtedly have analogs at other institutions. The success factors highlighted here focus on strategies specific to physician engagement. However, the key dynamic was engagement of the entire organization, including our independent physicians, as a unified team behind the common goals of safer and more effective medical care supported by an electronic medical record migrating toward a paperless environment. While some factors may have had greater impact than others, we found them all to be interrelated, reinforcing and synergistic.

We pioneered a concerted effort to engage physicians in unique ways that pre-dated system selection and project funding. This proactive approach with respect to physician engagement before, during and after system activation may have been the most effective factor of all in clinician adoption.

Interestingly, this program is not peculiar to the system we selected or our organizational structure. It could be employed by others in both inpatient and outpatient settings.

Proper framework

At the outset, we must acknowledge insights gained from the experience of others - both successful installations as well as those that were troubled or failed. We studied what others had done, talked with those who had first-hand experience, and visited hospitals well in advance of considering one system or another. Much of this was done to gain insights into potential success factors, but also to understand significant risk factors and failure points. We engaged clinicians from the Gartner Group to participate in medical staff retreats to share and discuss their own research and perspectives. We also reviewed other published research with an eye to finding clues on what worked and what didn't, as well as factors that drove success, underpinning outcomes in either final result or time to value.

Two generic, but vital, factors were system architecture and the system itself. We decided early in our system-selection process that integration of certain parts of the system were essential to success. In particular, we determined that the pharmacy component of the system had to be fully integrated with the ordering component. We were not going to be satisfied with superficial or "visual" integration, but required fundamental integration at a database and operational level rather than, for example, the interface of two previously independent systems from a common vendor. The product also needed to address critical care and the emergency department. An even larger "footprint" was highly desirable long-term as we targeted other ancillary systems for subsequent replacement.

In selecting our vendor of choice, we endeavored to be as certain as we could be that the system would meet these requirements and work as advertised. We knew that any contemporary system can be considered a "work in progress." However, as we settled on a final system, we were satisfied that our core system requirements would be met or exceeded, with predictable operation and performance. The traditional work of IT system rollouts and the academic view on the subject typically centers on people, process and technology. This is usually expressed visually as a triangle to show that the system addresses all three dimensions as a condition of overall success. The certainty of technology performance meant that we could focus our attention on the people and process elements that we knew were keys to any level of success with CPOE.

Two other starting-point advantages were our prior use of CPOE and a decade-long effort to partner with our physicians in the areas of best practices. The latter was the long-standing effort to partner with our physicians on methods to mutually differentiate ourselves from other regional providers based on quality and outcomes.

This had little to do with systems per se and much more to do with ongoing dialog and debate on any number of subjects that facilitated high levels of mutual understanding and collaboration.

The pharmaceutical model

A primary goal was to achieve a "tipping point" with respect to physician use, a point at which a significant majority of orders were entered by physicians persuaded that such use was superior to paper-based methods, or at a minimum that this was the "way of the future" for a forward-looking health care organization. Broad adoption by many physicians was expected, in turn, to bring the few, inevitably more resistant, colleagues along. We thought it best, where possible, for physicians to encourage and educate other physicians as to the overall benefits of CPOE and reinforce its use by example and demonstrated proficiency.

The best, fastest way to reach this tipping point, we believed, was a strong, personal relationship with key physicians coupled with a keen understanding of the physicians and their practices. To achieve a genuine relationship with our physician partners, we sought to understand individual physician preferences, practice patterns and peer influences, as well as their use of office-based EMR systems, comfort with technology, and other related attributes that would help predict their personal adoption of CPOE as well as the positive or negative impact they might have on others.

We had seen this operate in the pharmaceutical industry where "detail persons" visit physicians regularly and develop personal relationships over a period of time. These individuals are able to gain a few minutes of valuable, but hard-to-get, physician time. Their overriding objective is to educate the physician about new and presumably more effective drugs, and convince him/her to try such drugs.

We began to think of a similar model with the goals being to educate physicians about new and presumably more effective technologies to get them to try such technologies. We planned to build upon technology already in place (remote access to results, for example), to discuss best practices developed by the medical staff and embedded in paper-based order sets, and to develop a regular, supportive dialog through periodic office visits and phone conversations.

Peer pressure

We formed a team that eventually grew to 10 individuals well in advance of system selection. The genesis of the team was two-fold. One factor was a formalization of support that grew out of facilitating remote use of TDS and general access to our systems that is today an equivalent of a "physician portal." The other was an outgrowth of activities to identify, develop and adopt best practices and evidence-based medicine apart from any implemented technology. Both were associated with activities that were already successful joint efforts between hospital staff and physicians. Over time, they naturally evolved in the direction of the EMR as that became an item of intense mutual focus.

We called this group of staff clinical training specialists (CTSs); members of the group had varied backgrounds. A few had some knowledge of or experience in IT, most had good analytical skills, and some had previously worked with physicians in some way. More important, they were all friendly, extroverted and sufficiently assertiveto make repeated attempts to contact individual physicians and develop relationships over time. They all had a strong service orientation. Working alongside systems analysts and technical staff, they learned the essential functions of existing systems, remote access and essential network methods. They had the basic computer skills to set up, support and network computers in offices and homes. CTSs participated on the team that developed and tested the EMR system, and they were sufficiently skilled in its basic operation that they could answer most questions directly or function as a gateway to an appropriate resource for more complex or technical issues.

CTS staff focused solely on physicians and, as necessary, their office staff. Their support was distinct from the established help desk function; it was multifaceted. They augmented not only the help desk staff, but also system analysts, network staff, and all parts of the project team and their associated specialized activities. And, support was in-person, at the convenience of the physician. The CTS team effectively served as a one-stop shop for most of what a physician might need. They also would hear both the good and the bad with respect to physician needs and concerns, an important source of critical feedback that was honest and direct and by no means limited to CPOE or the EMR project

Over time, the CTS team leadership developed an immensely valuable compendium of information concerning our physician community, not unlike a powerful marketing database. This accumulated information allowed selective targeting of CTS activities. For example, as Pareto would predict, we knew which 20 percent or so of our privileged physicians accounted for 80 percent of the business of MemorialCare. We knew how this group was distributed over the various specialties, their interest level in the EMR project and how it would affect their practice, and who was already using one or more of our existing systems.

As is true with other institutions, inpatient volume at our facilities is driven by a large network of referring physicians and subspecialists who are a layer removed from direct patient admission. Our database helped us identify these physicians and understand referral patterns that would be relevant to system use. For example, some referring physicians would depend more heavily on results reporting and might be infrequent users of the CPOE. They would likely need additional training, support and follow-up. As another example, we identified the 5 percent or so of our physician staff who were opinion leaders irrespective of inpatient volume. What they said really mattered, so we focused our education efforts on those staff so that everyone was "on message" with respect to the project and we knew where they stood with respect to concerns. We learned from the nearby Cedars Sinai's CPOE initiative that we didn't want a "silent majority" of initially supportive, but uninvolved, physicians who might coalesce over time into an irretrievably resistant population.

The information we gained over time is shown in what came to be known as our "stoplight report," which, as we approached activation, helped us map out the spectrum of support we could expect at go-live and where additional resources or leadership would be needed. A small portion of our Stoplight database can be seen in the figure on page 28.

We empowered CTSs and supported them as much as possible. They were our eyes and ears into the physician community, trusted partners of the physicians. The key to their success was developing supportive relationships early, not "just-in-time." The CTSs supported physicians early and often - early when helping them with technologies other than the EMR, and often as the development of proficiency with CPOE, notes, retrievals, etc., required.

Our use of the CTS team was aided by a number of other factors that contributed toward adoption of the system and direct order entry by physicians. These included participative system selection, design sessions with physicians at the helm, common design across our enterprise wherever possible, and an abundance of order sets and preference lists to name but a few. We spent an enormous amount of time cleaning up arcane dysfunctional operating practices and exposed the "magic" behind paper-based orders. Physicians were astounded at the gaps left in required specificity of their orders, for example, and the work of others behind the scenes to fill those gaps by either subsequently interrupting the clinician and delaying care, or making educated guesses based on personal knowledge of a given physician's practice patterns. The safety issues were obvious and compelling.

Mandate training

Last, we mandated training. Not use, but training. No exceptions. That was necessary because, at a minimum, results of various kinds were not going to be available on paper. Physicians needed to have retrieval proficiency. We trained to workflows and functionality. We established comfortable, "safe" training environments with physicians' peers and, often, with physician instructors and always with good food. We trained at the convenience of physicians, and organized by specialty. Afterwards, they had access to a "play" environment to experiment and practice - before and after going live.

Implementation of CPOE and an EMR in general is a marathon, not a sprint. It is a long-term, beneficial journey of process improvement, quality gains and informatics. All of it, in our view, begins with the clinician in mind who is on the front line of diagnosis and treatment, deserving and needful of useful and proven technology - if all can be done to "get over the hill" of adoption and on the road of proficiency. We are pleased with and proud of our results and our ongoing program of physician engagement which we think can be employed by others to similar ends.

Mr. Joslyn is senior vice president and CIO and Dr. Stutman is executive director of research at MemorialCare Medical Centers.

After CPOE

By Michael Gallagher, MD, MBA, MPH

El Camino Hospital (ECH) has enjoyed a lengthy history with computerized physician order entry (CPOE) systems. ECH is effectively the first hospital in the country to institute and continuously run a CPOE system. Despite this rich legacy, CPOE remains only one of the systems that make up our patient-centered decision support strategy, representing one of many knowledge silos.

CPOE remains the central system we use for managing acute patient encounters. It also provides an arc of data continuity over multiple admissions. But ultimately, even adding in certain features of an electronic medical record (EMR) at the hospital level,it still falls short of a full electronic health record (EHR).

As we explore methods to approach some form of full-on EHR for our population, we are also concerned with richer and less "atomic" data. We are moving outward from a patient-centered view, to allow ECH to explore ways of considering patients in groups, and relating those groups to the operations of the hospital itself.

Patients have distinct characteristics that allow us to roll up data in ways that will inform vital clinical decisions based on aggregations around combinations of those characteristics. For instance, finding all male patients on rate-control medications, or finding all female patients on after-load reducing medications can alert us to excellence in treatment or opportunities for intervention. Being able to find those groups in real time, and close the loop with the clinical team, has been pure speculation until recently. Relating those clinical slices with operational data has been even more of a challenge.

To make this vision a reality, ECH has initiated a comprehensive enterprise-wide data strategy with Microsoft technology at its heart. Microsoft Amalga, the unified intelligence system, will aggregate these silos of data, providing the flexibility needed to roll up patient data, better influence vital clinical decisions and enable ECH to focus on providing the highest level of patient care.

Initially, Amalga will be deployed as a stand-alone solution to provide real-time clinical data for our acute care patients. We will group, sort, filter and mine information about the patients currently admitted in order to drive excellence in health care. Using this patient data, we will track Joint Commission core measures and other indicators of hospital safety and effectiveness, in real time, to institute change or intervene clinically based on well-defined key performance indicators (KPIs).

As our use of Amalga matures, it will join with our conventional enterprise-wide data initiative named Apex. Together, Apex and Amalga will expose business-centered data for advanced reporting and analytics.

Additional technology from Microsoft rounds out the foundation for information from both the clinical and business realms. The full spectrum of functionality of the Microsoft SQL Server Reporting Services and Analysis Services is the engine. End-user experiences will be provided through Microsoft Office SharePoint Server and Microsoft Office Performance-

Point Server. Users will have a selection of targeted and context-aware Web portals, Web applications and embeddable widgets in their information toolbox.

At first blush it may appear that such a strategy, while providing improved service and lowered cost of care to the patient, would have limited clinical application. However many use cases exemplify clinical relevance. For example, rising power consumption in the ventilation system in one unit of the hospital may warn of impending decreases in air quality and risk for respiratory complication in patients with asthma or other compromised lung functions.

The enterprise-wide and unified vision for information usage has unlocked our creativity here at ECH. We are identifying KPIs and use cases in every corner of our institution to move us toward our goal of being one of the top five percent of hospitals in the country for clinical excellence.

Dr. Gallagher is director of business intelligence and outcomes for El Camino Hospital in Mountain View, Calif.