Effective Medication-Related Decision Support in CPOE

SUBSCRIBE
to the
PSQH e-Newsletter

September / October 2008

Effective Use of Medication-Related Decision Support in CPOE

By Jane B. Metzger, BA; Emily Welebob, RN, MS;
Fran Turisco, BES, MBA; and David C. Classen, MD, MS

The authors published Part I of this article, "The Leapfrog Group's CPOE Standard and Evaluation Tool," in the July/ August issue of Patient Safety & Quality Healthcare, which is available at http://psqh.com/
julaug08/cpoe.html. Preparation for the evaluation is basically doing a good job with medication- related decision support. Part II provides advice about doing that based on what the authors have observed in many visits to hospitals in the course of developing the evaluation tool.

If the rollout of computerized physician order entry (CPOE) follows on the heels of other advanced clinical system applications with significant decision support, such as electronic medication administration (eMAR) or clinical documentation, a core team of nurses, physicians, and pharmacists has already learned about the responsibilities and work of applying decision support on a large scale. However, in many hospitals, CPOE is the first advanced clinical application, and clinical decision support (CDS) applied so broadly is new and uncharted territory. Taking full advantage of this new capability to improve medication safety requires mastering each of the areas discussed below.

Build the structure for accomplishing the work of applying clinical decision support.
The logic of clinical decision support tools in CPOE that can guide and critique ordering is eventually applied to every targeted type of physician order. Mindful of what is at stake, hospitals proceed cautiously and deliberately. By now, most hospitals have invested in new group and individual roles and responsibilities to address quality and patient safety overall, and that structure is charged with integrating CDS into the strategy and tactics for improving medication safety (Provonost et al. 2008). Accomplishing this requires clear accountability and new processes to cover some areas such as the work of setting up and testing CDS, as well as ongoing monitoring and updating.

The major tasks of managing CDS tools are listed in Table 1 (pg. 18), along with recommendations and observations from many hospitals visited during development and testing of the evaluation tool and in other CPOE-related research (Metzger & Fortin, 2003; First Consulting Group, 2006). Although staff from the Information Systems Department (IS) — usually one or more physicians or nurse analysts who have mastered the CDS tools — play an important role, physician leaders (and to some extent pharmacists) with responsibility for patient safety own and lead this process.

Click here to view a larger, PDF version of this table in a separate window.

Hospitals have standing committees such as Patient Safety or Medication Safety and often now a patient safety officer or medical director for patient safety for whom CDS in CPOE represents a new safety net to be applied during the ordering process. Strategy and direction for applying CDS comes from these individuals and groups. The group that ends up doing the day-to-day work of applying the CDS toolset to the identified targets includes physicians, clinical analysts who manage advanced clinical applications, and representatives of pharmacy, nursing, and other departments, often under the leadership of the chief medical information officer (CMIO) or patient safety officer.

Fold the Leapfrog order categories into the quality program.
The additional safety net that decision support in CPOE offers should be managed in coordination with that overall medication safety effort. The order categories in the Leapfrog CPOE evaluation provide a useful framework for applying CDS tools in CPOE (Table 2, pg. 19) because they were selected based on available evidence concerning the frequency and severity of preventable medication errors (Kilbridge et al., 2001).

Click here to view a larger, PDF version of this table in a separate window.

These categories should be considered along with local knowledge of high-risk situations as priorities and plans for working on CDS are developed.

Become expert in the CDS toolset in the CPOE application.
CPOE includes many features that work in different ways to improve ordering and physician decision-making about orders (Metzger & Turisco, 2001; Kuperman & Bobb, 2007). Vendor solutions differ significantly in the scope and design of the toolset and how easy it is to implement and manage. Hospitals purchasing CPOE should consider the design, presentation options, and management of decision support tools as one factor in selecting the right solution for their hospital. In hospitals further down the road with advanced clinical systems, getting to know the details about every function of the toolset is an important step in learning how best to apply them. Especially important is understanding the limitations of the current toolset. For what types of problem orders are the tools too rudimentary or lacking altogether? Can rules-based alerting be sufficiently fine-tuned to reduce nuisance alerting to an acceptable level? (Kuperman & Bobb, 2007). Vendors continue to enhance the toolsets, and requests from customer hospitals will likely speed progress.

Vendors take different approaches to integrating CDS into CPOE. Some tools reside in the application itself. For example, they may be set in the templates in the order master file or tables controlling the type and level of order checking against a third-party medication database such as First Data Bank and Multum. However, other tools and the logic or tables that control activation and use may be bundled with the clinical data repository (CDR) or in a separate application (sometimes called the knowledge or rules engine).

Every hospital needs experts who understand the details for the CPOE solution being used. Often the CMIO and a pharmacy or nurse analyst in IS play this role; some hospitals are starting to add a pharmacist informaticist.

Becoming an expert requires not just inventorying all of the specific tools, but also gaining an understanding of what situations they target, how they work, and how to manage them. Table 3 (pg. 20�21) organizes the likely tools and describes how each category of tools applied to CPOE contributes to the goal of improving safety and quality of care. The last category of tools — rules-based surveillance — is strictly speaking not part of CPOE, though often thought of, and implemented in conjunction with CPOE. In fact, rules-based surveillance can be put to use long before CPOE (Classen & Metzger, 2003), and has been shown to speed response to new patient information by notifying physicians of the need to reevaluate one or more orders (typically involving medication orders).

Click here to view a larger, PDF version of this table in a separate window.

Integrate CDS rollout into CPOE plan.
"When to start applying CDS" and "How quickly to implement it" are two questions with which every hospital wrestles. On the one hand, everyone wants to proceed cautiously and not further complicate the task of getting physicians to enter their orders electronically. On the other hand, getting to "live" CPOE is a multi-year journey and delaying CDS further delays the objective — to improve ordering and patient safety.

Some of the most basic, but powerful, CDS tools are features of the order templates for each orderable item, which make up the order master file and determine the content, arrangement, and order-specific rules in the displays physicians use to enter their orders. Basic functions such as these are used to configure the system for the hospital and users, and common sense dictates employing them from the start.

Groups of pre-defined orders — in the form of order sets for specific clinical situations (i.e., admission, diagnosis) and commonly used orders (or departmental or personal "favorites") — are also fairly basic CDS tools, which most hospitals stress at initial "go-live." Aside from ensuring that orders are complete and appropriate, pre-defined orders and order groups also contribute to physician acceptance because they speed order entry significantly.

Groups and individuals charged with system configuration and set-up invest a great deal of time and effort to make CPOE quick ("speed is everything") and easy to learn and use (Bates et al., 2003). In one large Veterans Administration hospital where significant resources were devoted to set-up, the effort yielded 667 order dialogs, 5,982 preconfigured (quick) orders, and 513 order sets organized in 703 order menus (e.g., for admission for a particular diagnosis). Project leaders reported that the investment paid off in both physician adoption and immediate curtailing of many common errors in ordering (Payne et al., 2003).

Beyond order set-up to encourage appropriate dosing and discourage inappropriate routes of administration and order sets, many hospitals move more slowly with other decision support to guide and critique medication ordering. Drug-allergy, drug-drug interaction, and sometimes therapeutic overlap checking are typically the first types of order screening employed. Unfortunately, as demonstrated in a recent study, these types of medication errors only account for a small percentage of the preventable medication adverse events (4%, 2%, and 1%, respectively; Adams et al., 2008). The tools for addressing medication errors that account for more adverse events, such as renal status (19%), drug-lab (27%), and drug-age (9%) checking are more difficult to implement given the CDS tools in some vendor solutions (Adams et al., 2008). Teams in some hospitals address these with software customizations. Another look at the potential impact of CPOE concluded that "advanced clinical decision support features" are needed to address 50% of clinically significant prescribing errors (Bobb et al., 2004). Clearly more work remains for both hospitals and vendors to advance CDS tools and their use.

Be sure to deliver value.
Applying CDS effectively requires first that the specific tools used guide practice in the desired direction. As shown in Table 4 (pp. 22�23), the toolset provides a number of options that can be used singly or in combination to address each order category in the Leapfrog CPOE evaluation (patient safety experts generally recommend using more than one). It also requires that the guidance be delivered in a manner acceptable to physicians.

Click here to view a larger, PDF version of this table in a separate window.

In keeping with the "speed is everything" caution of many early CPOE adopters (Bates et al., 2003), one dimension to consider when applying decision support tools is to fit within physician workflow so that interruptions and time are minimized. Decision support can be delivered in different forms. The least effective (and acceptable) form is a message that pops up and requires a physician response before getting back to order writing. Hence these must be used sparingly, if at all. A much better fit is information displayed alongside or choices pre-screened for appropriateness.

Clinical decision support also can be delivered at different times in the electronic ordering process, as shown in Table 5. The guiding principle with respect to timing of decision support is real-time at the right time.

Click here to view a larger, PDF version of Tables 5 and 6 in a separate window.

Also important to physician acceptance is the relevance of CDS-generated advice and information to the current clinical situation (the physician perception of clinical value). Generally, the more the advice or choices are customized to this patient right now, the better (Bates et al., 2003). Moving to more customized decision support as displayed in Table 6 requires both more complex logic (and usually tools) and more comprehensive and up-to-date information about the patient. As hospital teams work on applying CDS, a second guiding principle is to provide tailored or customized advice most of the time. This leads to speedier ordering for physicians, as well as increased physician acceptance.

Click here to view a larger, PDF version of Tables 5 and 6 in a separate window.

"Nuisance messages" is the term usually applied to CDS-delivered content the physician considers irrelevant to the situation at hand. (Note that combining information that is not useful with a delivery form often viewed as disruptive is especially annoying to physicians!) Because of the importance of minimizing nuisance messages, the team of expert advisors that assisted with development of the Leapfrog CPOE evaluation included detection of nuisance messages as part of the test (Kilbridge et al., 2001).

Constant fine-tuning of decision support is needed to reach and sustain high value of the assistance delivered (a very high "hit rate"). Critical input to that effort comes from system reports on the frequency of CDS firing an alert or delivering a reminder and how often physicians take the advice given (by canceling or changing the order in some way). Hospital teams also actively solicit feedback at staff meetings, during rounds, or sometimes via a feedback button that makes it easy to submit comments or complaints while ordering.

Summing Up
The Leapfrog CPOE standard expects that the CDS tools in CPOE are aggressively applied to avoid potential adverse events that can be detected at ordering. It requires that physicians are entering most orders directly and benefiting from the safety net provided by CDS. Hospitals must work to gain the value available from the CDS toolset. Making effective use of the potential will require a concerted effort.

Jane Metzger has done research and consulting on bringing computer support to the point of care for the past 36 years. She is a principal in Emerging Practices, CSC's applied research arm. She has completed several studies on the successful adoption of CPOE in community hospitals and on using the clinical decision support tools in CPOE to improve patient safety and quality. She has also developed organizational readiness assessments and other tools that aid in planning for implementation of clinical IT that meets quality improvement objectives, including the development of the Leapfrog CPOE evaluation tool. Metzger may be contacted at jmetzger2@csc.com.

Emily Welebob has worked in leadership roles in clinical nursing, the clinical systems marketplace, and consulting related to clinical systems/EHRs for over 15 years, with a focus on patient safety.�For the past 6 years, she was responsible for day-to-day management of the Leapfrog evaluation project in all phases. Welebob has published, lectured, and researched in the areas of clinical informatics and patient safety, including numerous efforts concerning CPOE. She is currently with the Indiana Health Information Exchange. Welebob may be contacted at ewelebob@ihie.com.

Fran Turisco is a principal in CSC Emerging Practices. Her more than 25-year career in health information technology includes programming, application implementation and management in delivery organizations, consulting, and research. Turisco's recent research areas include physician order entry/clinical decision support, mobile computing to support point of care use, planning for emerging technologies, and IT support to clinical research. As a member of the Leapfrog project core team, she was responsible for the development and testing of the Web application for the CPOE evaluation tool and a survey of the CPOE vendor marketplace. Turisco may be contacted at fturisco@csc.com.

David Classen is a partner and chief medical officer at CSC — Healthcare. He is also an associate professor of medicine at the University of Utah and a consultant in infectious diseases at The University of Utah School of Medicine in Salt Lake City, Utah. Classen consults and has authored numerous publications on the use of IT-facilitated process change and decision support and epidemiologic techniques to improve medication safety. Classen serves as a member of advisory and working groups of the Institute of Medicine, Joint Commission, and National Quality Forum. He may be contacted at dclassen@csc.com.

References

Adams, M., Bates, D. W., Coffman, G., & Everett, W. (2008). Saving lives, saving money. The imperative for computerized physician order entry in Massachusetts hospitals. Massachusetts Technology Collaborative and New England Healthcare Institute.

Bates, D., Kuperman, G., Wang, G. S., et al. (2003). Ten commandments for effective clinical decision support: Making the practice of evidence-based medicine a reality. Journal of the American Medical Informatics Association, 10(6).

Bobb, A., Gleason,K., Hurch, M., et al. (2004). The epidemiology or prescribing errors. The potential impact of computerized prescriber order entry. Archives of Internal Medicine, 164, 785-792.

Classen, D. C., Cain, C., & Vaida, A. (2007, January 12). A national web conference on the First Consulting Group inpatient and ambulatory CPOE assessment tool. Available at http://healthit.ahrq.gov/portal/
server.pt?open=514&objID=5553&mode=2&holderDisplay
URL=http://prodportallb.ahrq.gov:7087/publishedcontent/publish/
communities/a_e/events/events/events/a_national_web_
conference_on_the_first_consulting_group_inpatient
_and_ambulatory_cpoe_evaluation_tool.html

Classen, D., & Metzger, J. (2003). Improving medication safety: The measurement conundrum and where to start. International Journal for Quality in Health Care, 15, Supplement 1, International Society for Quality in Health Care and Oxford University Press.

First Consulting Group. (2006, September). Saving lives, saving money. Computerized physician order entry lessons learned in community hospitals. Prepared for Massachusetts Technology Collaborative and New England Healthcare Institute.

Kilbridge, P., Welebob, E., & Classen, D. (2001, December). Overview of the Leapfrog Group Evaluation Tool for Computerized Physician Order Entry. Retreived from http://www.leapfroggroup.org/media/file/
Leapfrog-CPOE_Evaluation2.pdf.

Kuperman, G., Bobb, A., Payne, T. H., et al. (2007). Medication-related clinical decision support in computerized provider order entry systems. A review. Journal of the American Medical Informatics Association, 14(1), 29-40.

Metzger, J., & Fortin, J. (2003, June). Computerized physician order entry in community hospitals: Lessons from the field. California HealthCare Foundation and First Consulting Group.

Metzger, J., & Turisco, F. (2001, December). Computerized physician order entry: A look at the vendor marketplace and getting started, The Leapfrog Group and First Consulting Group.

National Quality Forum. (2002). Serious reportable events. Washington, DC: Author. Retrieved March 26, 2008, from http://www.qualityforum.org/pdf/reports/sre.pdf.

Payne, T., Hoey, P., Nichol, P. J., & Lovis, C. (2003). Preparation and use of preconstructed orders, order sets, and order menus in a computerized provider order entry system. Journal of the American Medical Informatics Association, 10(4), 322-329.

Provonost, P.J., Rosenstein, B.J., Paine, L. et al. (2008). Paying the piper: Investing in the infrastructure for patient safety. Joint Commission Journal on Quality and Patient Safety, 34(6, 342-348.