CPOE primer.

An interesting event took place in March 2003. The U.S. House of Representatives' Ways and Means Committee passed the Patient Safety Improvement Act by a wide margin. (1)

This remarkable piece of proposed legislation specified that within 18 months the Secretary of Health and Human Services was to develop voluntary national standards for interoperability of health care information systems, specify medical terminology and evaluate technologies such as computerized physician order entry (CPOE) and medication bar coding.

The ultimate goal was to show a measurable reduction in the medical error rate. Hard on the heels of this announcement, the Food and Drug Administration announced two proposed rules to reduce medical errors.

1. The first was the requirement to bar code all medications so that in combination with computerized order entry systems they would enable a much more robust safety check.

2. The second was enhanced rules for reporting medical errors.

This legislative action is the result of a nationwide debate kicked off by a seminal report from the Institute of Medicine in 1999, "To Err is Human," (2) where between 50,000 to 100,00 deaths were ascribed to medical errors. A substantial number of these were adverse drug events.

In order to clarify the role of technologies such as computerized physician order entry, it is important to define the above two terms clearly. A medication error is an error in the process of ordering, dispensing or administering a medication, regardless of whether an injury occurred or whether the potential for injury was present. (3)

An adverse drug event is an injury resulting from the use of a drug, which may or may not result from an error. (4) It is important to realize that the vast majority of medication errors do not result in any harm to the patient.

The literature is replete with data that identify the incidence of adverse drug events in both inpatient and outpatient settings.

In general, adverse drug events constitute about 19 percent of all adverse events, the largest category being after surgical events. Overall, it is estimated that adverse drug events occur in 6.5 percent to more than 20 percent of hospitalized patients. On an ambulatory basis, adverse events occur at a rate of 50.1 per 1,000 person years, of which 13.8 per 1000 are preventable. (5,6,7,8,9)

Awareness of this problem extends four decades back when data showed 30 percent of inpatients suffered at least one adverse drug event and 3 percent of hospital stays were as a result of adverse drug events. (10) The prevailing opinion at that time was that adverse drug events were an unavoidable part of modern therapy.

Costs related to adverse drug events have also been studied. The average additional cost for an adverse drug event per hospital stay was found to be approximately $2,500, (11) while the average claim related to liability for an adverse drug related event was $376,000. (12)

Adverse drug events can occur as a result of an error made at literally any point in the medication management process. However, the literature shows that medication ordering is responsible for the majority of these events. (13,14)

Technology available today can significantly reduce the incidence of these events. Pharmacy systems with alerts, electronic medical records with medication administration record modules, bar coding, medication cabinets, computerized order entry are all examples that have been shown to impact medication errors.

The potential of computerized physician order entry to make a significant impact has been shown clearly in studies that documented a 55 percent reduction in serious medication errors. (15) In addition to its obvious role in medication order reduction, CPOE has the potential to reduce variation in care by providing embedded guidelines and standardized order sets. (16)

Workflow is also positively impacted as these order entry systems reduce duplicative documentation, allow earlier processing of orders and impact delivery of medications. (17)

Definition and terminology

Consider for a moment that clinicians perform two common tasks as part of daily patient care:

1. Documentation of evaluation

2. Initiation of care through orders for therapy and procedures

CPOE is a software application that enables the second task. However, there are several levels to this application.

The most basic application (order communication) merely substitutes the multi-copy paper documents for an electronic template. While the template is available for all providers to review and far more legible and audit-friendly than paper, decision support is the feature that truly defines CPOE.

From simple alerts to allergies and interactions to complex clinical algorithms that drive disease specific protocols, the goals of this application remain the same. Reduce adverse events, reduce variation and improve efficiency by improving productivity.

Due to the publicity surrounding patient safety in recent times, the adverse event reduction function has commonly been mistaken to mean the entire capability of this application. Indeed, most of the purchases of CPOE are being driven by the patient safety function it is perceived to improve.

At its most advanced level, the order entry systems are tied to data warehouses that enable data mining thereby supporting long-term outcomes management and population based care. Inpatient CPOE is the focus of most of the legislative and safety activity, but the lessons are equally applicable to the ambulatory arena and quality advocates such as the Leapfrog Group, (18) an industry coalition that seeks to modify practice through financial incentives and pushes the outpatient adoption of CPOE.

A recent article in the Journal of the American Medical Association clearly documented the incidence of adverse drug events in an ambulatory setting (8). A review by an industry magazine in March 2003 shows that CPOE is being rolled out into ambulatory areas of the hospital as well as physician's offices.

CPOE myths

There are two common myths about CPOE:

1. It is a stand-alone application.

By the very fact that order entry spans the entire continuum of care and requires decision support of various levels means that certain applications need to be in place to enable this application. This includes a clinical data repository, medical vocabulary or terminology, clinical documentation, clinical data support, pharmacy and a methodology to enable the wireless use of this function.

With medical error reduction as a prime concern, a tight integration of the pharmacy and CPOE functions along with interaction checking database support is a fundamental requirement. This lack of understanding has been the cause of some failures of implementation as clinicians quickly realize that the purported benefits of the electronic system were simply not there.

2. The second myth is that CPOE applications can be judged by the functionality they offer.

CPOE is an enterprise-wide undertaking in the true sense of the word. Since it is required to take into account the workflows present in the vast number of care settings, flexibility of design and some configuration is a requirement.

Integration with nurse documentation systems, especially at the medication administration record level, is also a fundamental prerequisite. In addition, being part of a service-oriented architecture is critical, as few other applications will need to be as robust, stable and scalable as CPOE. (19)

A recent article related the confusion raised by computer generated medication administration records where dosage strength was shown before actual patient dosages, which led to nurses mistaking the first line as the actual order. (20) This requires a simultaneous configuration of CPOE and the medication administration record.

Leapfrog and CPOE

The Leapfrog group is a consortium of Fortune 500 companies--private and public health care purchasers--that launched an effort to reimburse providers for high quality care in November 2000.

Leapfrog initially identified three main standards:

1. CPOE

2. ICU physician staffing

3. Evidence-based hospital referral to rate providers

By implementing a pilot program, the Leapfrog group developed an evaluation tool for CPOE applications that rated their capability in meeting several core order categories:

1. Therapeutic duplication

2. Single and cumulative dose limits

3. Allergies and cross allergies

4. Contraindicated route of administration

5. Drug-drug and drug-food interactions

6. Contraindication dose limits based on patient diagnosis

7. Contraindication dose limits based on patient age and weight

8. Contraindication dose limits based on laboratory studies

9. Corollary (requiring an associated or secondary order)

10. Cost of care

11. Nuisance

The last three categories recognize that CPOE was designed to do more than order medications and interaction between different data sets was necessary to accomplish the goal of comprehensive facility-wide order entry. (20)

Recognizing that the majority of U.S. health care is provided in ambulatory settings, the structure of CPOE in the ambulatory environment is somewhat different. The Center for Information Technology Leadership defined ambulatory CPOE as a software application that supports the ordering of medications, lab tests, radiology studies, nursing interventions and referrals.

As with the inpatient CPOE, a key component is clinical decision support that is different in structure and goals. The basic functions of this version of CPOE are:

* Medication order entry

* Medication decision support

* Diagnostic order entry

* Diagnostic decision support

Adoption and use

Coming on the heels of the recently concluded Health Information Management Society (HIMSS) meeting in February 2003, a CIO leadership survey conducted by the Superior Consulting Group revealed that 64 percent of these executives said CPOE would be among the most important applications in the next two years.

This closely relates to the Deloitte and Touche Survey of Fall 2002 where 61 percent of CIOs said they would be implementing CPOE in the next two years and the more recent ACPE Health Care Technology Survey where nearly 78 percent of respondents reported that they were either planning, bidding, testing or already using CPOE.

These data clearly show that the issue of patient safety is now prime among providers and the public.

In stark contrast, First Consulting Group estimated that about 5 percent to 7 percent of acute care hospitals in the United States had installed CPOE as of 2002. Data of usage of these systems from the Dorenfest complete database of 2002 is even more fascinating. It states that there is limited use of CPOE in the industry today.

Less than 25 of the 1,426 health care systems in the database had "most" physicians entering "most" of the data into the CPOE system. Only 8.7 percent of the systems had over half of physicians entering "some" data into the system. And 21 percent of the systems had "some" physicians entering "some" data into the system.

This clearly shows a disconnect between the installation momentum and actual use. The reasons for this are beginning to be reported.

A recent study from the Oregon Health Sciences University reported that there were four identifiable themes linked to use of these systems.

1. Organizational issues such as culture, egos, power and politics

2. Clinical and professional issues linked to workflow practices

3. Technical and implementation issues including usability, time, training and support

4. Issues related to organization of information and flexibility of the systems

The prime conclusion of this study was that active engagement of clinicians was a critical factor in the success of implementation. (21) An industry newsletter, Inside Healthcare Computing, in its February 24, 2003 issue examined the very public failure of the Cedars-Sinai CPOE implementation. What was revealed was that far from being a technical issue, the prime drivers were lack of involvement of critical physician leaders prior to going live as well as a perception of lack of ease of use related to insufficient education.

Finally, a lack of workflow analysis prior to going live meant there was a poor understanding of how the system affected daily tasks and interaction between departments. Meanwhile, data continue to accumulate showing that the system reduces errors, improves turnaround times and productivity. (22-28)

It is clear that there are issues with the design of the system and acceptance that need to be further examined.

As of the first quarter of 2003, there were over a dozen offerings from vendors with multiple beta and live installs available for prospective users to review.

While most of these applications serve the inpatient market, there is still a lack of true enterprise-wide order entry capability as defined by the ability to address the comprehensive ordering needs of a clinician across care settings.

In order to really evaluate the order entry function, a basic understanding of the medical management process is necessary, as CPOE is a process of medical management and not merely the entry of orders to facilitate care. In effect, the CPOE tool is evolving to be an advanced form of clinical decision support. It is most effective when linked in real time to many clinical data sources.

Most of the functionality and alerting capability is linked to the existence of a comprehensive data repository that integrates information from multiple applications that support CPOE. The core areas that must be in place as an integrated data repository include the nursing system incorporating the medication administration record, the pharmacy system with checking databases, a laboratory system with multiple alerts based on institution-specific protocols and a dietary system that enables drug to food alerts.

This core system enables any of the non-medication related orders to be processed in a similarly intelligent manner.

For example, an order for a dye study in radiology will be automatically checked against a patient's renal function because the lab system is now enabled by algorithms to monitor orders as well as results.

To date, much of CPOE has focused on digitizing the paperbased processes surrounding order writing and fulfillment with an added layer of alert-focused decision support. For CPOE to reach its full promise it will extend beyond its origins and start addressing the full spectrum of computer enhanced medical management. This will require making CPOE and decision support integrated ubiquitously into the entire workflow of medical management--a process much larger than medication management. This would extend CPOE beyond the point of ordering directly into every aspect of patient management.

Decision support can be used to improve patient issue identification, therapeutic and diagnostic decision determinations and patient care delivery. For patient issue identification, decision support should identify potential problem patients and bring them to the appropriate person's attention.

This may be as simple as notification of a given abnormal lab value but, more interestingly, may extend to analysis of numerous parameters to identify downward patient trends early enough that the trajectory can be reversed prior to the onset of severe morbidity.

Much of the work in CPOE so far has been in the area of therapeutic and diagnostic decisions, but much of the focus has been on post order placement alerting to prevent errors.

In the next generation, decision support should be used to gather the right information to the point of ordering that alerts are much less frequent. Many of the alerts presented today are the result of the ordering physician not having enough information present at the moment of decision. This lack of information finally presents itself as an alert, which may interrupt workflow and potentially become irritating to the user.

If, while developing an order, the computer system provided information to guide the physician's decision making, then many of the alerts could be prevented.

Finally, point-of-care decision support should focus on the issue that a patient my change dramatically between the point an order is issued and the point care is delivered. As decision support advances it should assess if the order to be carried out is still appropriate in the current patient context.

Additionally, point-of-care decision support should be used to verify the correct execution of orders and to monitor the effects on the patient. So, as applications progress from CPOE to full spectrum computer aided medical management, patient information and care team actions will be integrated and assessed by decision support to facilitate the execution of the best decisions for a patient at any point in their care.

Costs and implementation

There is enough literature to show that the principal cost-saving opportunity from the installation of CPOE is the reduction of preventable adverse drug events. However, with evidence from early adopters of this application beginning to be shared nationally, it is clear that there are also several other categories of cost savings that are related to decreased utilization.

The Advisory Board Company, in a monograph published in late 2001, described a method of quantifying adverse drug events based on prevalence published in the literature. It described an average hospital of 300 to 399 beds as experiencing a rate of 379.7 adverse drug events for an inpatient census of about 76,000 or five per 1,000 days.

If the assumption that an advanced CPOE system could reduce this by 62 percent is accepted, then that number dropped to 235.4. Given data that suggested an average cost of $4,685 per adverse drug event, savings of approximately $1.1 million could be realized on avoidance alone.

Further data published by the company show that savings of up to $1.8 million per year could be realized by reductions in ancillary test usage (mostly laboratory) and reductions in length of stay for a large public hospital. The initial cost of installation of CPOE at the average 400 bed general hospital is estimated at somewhere between $4 million and $7 million (Frost and Sullivan; Advisory Board Company). This includes software as well as implementation and training.

Ongoing support costs are generally in the industry accepted average of 18 percent to 21 percent of the software cost. This is clearly an expensive proposition and when it is realized that the clinical data repository and pharmacy systems are needed as fundamental infrastructure, the figure could easily double.

Such a project requires extremely detailed planning with the realization that the ultimate success depends on the use of the system by physicians. A thorough evaluation of all support systems and level of integration is a valuable first step in implementation.

Determining what the priorities for physicians are in terms of actual functions is a next step realizing that ease of use and consistency of the system are major factors in success. The adoption of clinical use cases that show typical facility physician workflow can be a tremendous tool in evaluating CPOE applications for ease of use and clinical intuitive feel.

Other critical issues in implementation of CPOE include the use of order sets that are pre-approved by physicians, a regularly updated decision support rules engine, a very deliberate roll out supported by physician leaders and extensive support in the early days of roll out.

Having onsite vendor support or immediate response when going live may be a make or break issue as a delay or shut down can prove disastrous.

In the end, several issues are paramount in the adoption of CPOE.

* CPOE should be recognized as a decision support tool.

* CPOE can work effectively only after a supporting infrastructure of the clinical data repository, laboratory, pharmacy and dietary systems is in place.

* Adverse drug event reduction is a big opportunity for CPOE, although decreased utilization, guideline embedding and variance analysis are others.

Physician concern about increases in time ordering and ease of use are the two biggest barriers to success. A good versus bad implementation may be differentiated by early involvement of influential physicians. Technology is not the primary factor in failure; it is the human factors.

A good CPOE application is the foundation of an enterprise-wide digital medical record and not an "isolated software, flavor-of-the-day" install.

IN THIS ARTICLE ...

If you don't already have it, get ready for it. Computerized physician order entry (CPOE) is most likely coming your way. Discover some of the ifs, ands and buts of identifying and implementing a CPOE systems

Check out ACPE's E-Health Strategies course for more on what you should be doing to meet the challenges of E-Health and the new consumer.

www.acpe.org/interact

References

1. March 13, 2003: U.S. House of Representatives; The Patient Safety Improvement Act. (HR 877).

2. To Err is Human: Building a safer health system. Institute of Medicine, John Lindo, Janet M. Corrigan, Mella Donaldson. Eds. National Academy Press, 1999.

3. Bates DW, Teich JM, Lee J, and others. "The Impact of Computerized Physician Order Entry on Medication Error Prevention. JAMA 1999, 6(4):313-321.

4. Cullen DJ, Bates DW, Small SD and others. "The Incident Reporting System Does Not Detect Adverse Drug Events." Joint Commission Journal on Quality Improvement 1995, 21(10):541-548.

5. Kaushal R, Bates DW, Landigran C and others. "Medication Errors and Adverse Drug Events in Pediatric Inpatients." JAMA 2001 285(16):2114-20.

6. Bates DW, Cullen DJ, Laird N, and others. "Incidence of Adverse Drug Events and Potential Adverse Drug Events: Implications for Prevention." JAMA 1995; 274:29-34.

7. Gurwitz JH, Field TS, Harrold LR and others. "Incidence and Preventability of Adverse Drug Events Among Older Persons in the Ambulatory Setting." JAMA. 2003, 289:1107-1116.

8. Gurwitz JH, Field TS, Avorn J and others. "Incidence and Preventability of Adverse Drug Events in Nursing Homes." Am J Med. 2000,109: 87-94.

9. Krozer E, Scolnik D, and others. "Variables Associated with Medication Errors in Pediatric Emergency Medicine." Pediatrics. 2002, 110, 4:737-741.

10. Jick H, Miettinen OS, Shapiro S, Lewis GP, Siskind V, Slone D. "Comprehensive Drug Surveillance." JAMA. 1970, 213:1455-1460.

11. Bates DW, Spell N, Cullen DJ, and others. "The Costs of Adverse Events in Hospitalized Patients." JAMA. 1997, 277:307-311.

12. Rothschild JM, and others. "Analysis of Medication Related Malpractice Claims: Causes, Preventability and Costs." Archives of Internal Medicine 2002, 162:2414-2420.

13. Bates DW, and others. "Incidence of Adverse Drug Events and Potential Adverse Drug Events." JAMA 1995, 274:29-34.

14. Kilbridge P and Classen D. "A Process Model of Inpatient Medication Management and Information Technology Interventions to Improve Patient Safety." VHA Research Series 2001, Volume 1, VHA Inc.

15. Bates DW, Leape LL, Cullen DJ, and others. "Effect of Computerized Physician Order Entry and a Team Intervention on Prevention of Serious Medication Errors." JAMA 1998, 280:1311-1316.

16. Chin HL and Wallace P. "Embedding Guidelines into Direct Physician Order Entry." Proceedings of the 1999 AMIA Annual Symposium, p.221-225.

17. Bates DW, Kuperman GJ, Rittenberg E, and others. "A Randomized Trial of a Computer Based Intervention to Reduce Utilization of Redundant Laboratory Tests." American Journal of Medicine. 1999, 106:144-150.

18. "CPOE: Order from Chaos," Health Data Management. March 10, 2003. 1-11

19. Classen D. "Medication Safety; Moving from Illusion to Reality." JAMA, March 5, 2003, 289,9: 1154-1156

20. Grissinger M. "Computer-Generated MAR format: A source of Errors?" P & T. Feb 2003,28;2:74.

21. Ash JS, Gorman PN, Lavelle M and others. "A Cross Site Qualitative Study of Physician Order Entry." J Am Med Inform Assoc 2003, Mar-Apr;10(2):188-200.

22. Taylor R, Manzo J, Sinnett M. "Quantfying Value for Physician Order Entry Systems: A Balance of Cost and Quality." Healthc Financ Manage. 2002, S56(7):44-8.

23. Shu K, Boyle D, Horsky J and others. "Comparision of Time Spent Writing Orders on Paper with Computerized Physician Order Entry." Medinfo. 2001, 10(Pt 2):1207-11.

24. Papshev D, Peterson AM. "Electronic Prescribing in Ambulatory Practice: Promises, Pitfalls, and Potential solutions." Am J Manag Care. 2001, 7(7):725-36.

25. Overhage JM, Perkins S, Tierney WM, Mcdonald CJ. "Controlled Trial of Direct Physician Order Entry: Effects on Physician Time Allocation in Ambulatory Primary Care Internal Medicine Practices." J Am Med Inform Assoc. 2001, Jul-Aug; 8(4):361-71.

26. Mekhijan HS, Kumar RR, Kuehn L and others. "Immediate Benefits Realized Following Implementation of Physican Order Entry at an Academic Medical Center." J Am Med Inform Assoc. 2002, Sep-oct;9(5):529-39.

27. Chin HL, Wallace P. "Embedding Guidelines into Direct Physician Order Entry: Simple Methods, Powerful Results." Proc AMIA Symp 1999;221-5.

28. Weir C, Johnsen V, Roscoe D, Cribbs A. "The Impact of Physician Order Entry on Nursing Roles." Proc AMIA Annu Fall Symp 1996, 714-7.

Narendra Kini, MD, is director of the GE Healthcare Leadership Institute at GE Medical Systems. He can be reached by e-mail at [e-mail]

[ILLUSTRATION OMITTED]

Brandon Savage, MD, is general manager for enterprise systems at GE Medical Systems Information Technology. He can be reached by e-mail at [e-mail]

[ILLUSTRATION OMITTED]

By Narendra Kini, MD, MHA and Brandon Savage, MD