Reducing Practice Variation
"Language is the archives of history." — Ralph Waldo Emerson
The words we choose influence the way we think about the things to which the words refer. We tend to think about things that are named systematically in systematic ways. Standardized medical terms and definitions correlated with consistent clinical processes and practices reinforce a systematic approach to the delivery of patient care. But, though standardization represents a core purpose for building a clinical information system, the word itself, used in the context of physician practice, connotes an impersonal, rote, "cookbook" approach to delivering patient care. For this reason, when discussing the role of physicians and other clinicians in the design, implementation, and adoption of clinical information systems, the "s" word can connote too much structure. From the physician's perspective, the goal of a CIS is to reduce variations from leading clinical practices.
The Evidence Is In
There is a growing body of clinical evidence supporting medical interventions and their value. These interventions may be surgical, procedural, cognitive, pharmaceutical or other types of patient management. The evidence shows that there are specific, leading-practice interventions that deliver the best patient outcome for treating certain medical conditions. If we introduce or institutionalize this information through appropriately designed clinical processes, we should be able to reduce variation in the management of those conditions for which there is a reliable body of knowledge. Embedding these leading practices within a technology framework further reinforces their use and helps to sustain the practices over time.
Variations in clinical practice can occur for many reasons. Among them is the fact that there are many kinds of evidence that point with varying degrees of confidence at potential treatment options. Randomized clinical trials may clearly establish a leading practice for a particular medical condition/intervention. Less helpful in defining leading practices are anecdotal reports or studies that establish only a small statistical advantage to a particular treatment modality. Differences in medical training and in a physician's practical experience with a particular medical condition also contribute to variations. The fact is, it's often hard to decide which clinical intervention will produce the most favorable outcome. A computerized physician ordering system must accommodate variations due to uncertain medical knowledge. At the same time, interventions offering a clinical benefit or cost savings are sometimes too far below a busy physician's radar screen. A computerized ordering system can remind the physician to consider them.
Critical Success Factors for Evidence-Based Medicine
Physicians need to have confidence that the source of the medical evidence supporting the recommended clinical practices is current and reliable. The medical evidence should be available in a timely manner so that physicians can begin to access it soon as possible after it is published. A well-designed computerized physician order entry (CPOE) system links medical evidence to the relevant medical conditions at the point of care as the physician is developing and implementing the plan of care. Beyond improving the accessibility of medical knowledge, a CPOE system should make it easy to develop tools that will make knowledge easier to access and easier to use. A computerized physician order entry system, enabled by the latest information technology, makes it easier for physicians to help translate the current explosion of medical knowledge into better, more cost-effective patient outcomes.
The Tools for Reducing Variations: Order Sets
Within the CPOE system, there are a number of different types of order sets:
- First, there are standing order sets that help the physician facilitate "common" activities such as admitting, transferring, or discharging the patient.
- A second type of order set addresses the common conditions that patients experience while in the hospital, such as pain, insomnia, and nausea.
- A third type of order set is used to address specific diagnoses or procedures, including the management of pneumonia or the preparation of patients for hip replacement.
Standing order sets make it easy to access accurate and state-of-the-art orders for those problems where there is a good body of evidence or there is good consensus on the best way to render care. A CPOE system makes it easy for the physician to do the right thing. Consistently. Order sets facilitate entering orders without recreating each time a patient is admitted to the hospital. Some order sets organize the orders associated with a particular procedure, such as admitting the patient, or a particular medical condition, so that physicians don't have to laboriously choose individual orders. A list of patient-specific or disease-specific orders pop up on the screen for the physician to choose or modify.
For example, physicians may need to admit a patient into the hospital for congestive heart failure. Physicians can access evidence-based and locally approved and adopted order sets for managing a patient with this condition. Physicians can choose to follow the endorsed treatment options, which the order sets present in a consistent, systematic way. If physicians feel that the standing order set does not quite meet the needs of their particular patients (for example, if the patient has a comorbidity), they are always able to tailor the patient's treatment to meet the need.
Each health care organization will strike its own balance between steering physicians toward leading practices and giving them unlimited freedom to customize their own individualized order sets. Highly individualized order sets risk automating practice variations into perpetuity. Our experience has shown that the preferred approach is to build order sets from evidence-based leading practices, where there is agreement about the management of common diagnoses, but also make it easy to tailor an order set to a particular patient, based on their own health profile, personal history, and to some degree, physician preference.
Categories of Order Sets
- Common order sets deal with non-disease specific procedures such as admitting
a patient, admitting a pediatric patient, admitting a preoperative patient,
writing postoperative orders, transferring the patient from the ICU to the
floor, and discharging the patient from the hospital.
- Disease-specific order sets, sometimes called departmental order sets, are
driven by the patient's specific diagnosis. These order sets specify recommended
interventions, cost-effective medications, and leading practices for specific
diseases and patient categories.
- Condition-specific order sets may address multiple diagnoses and deal with issues such as pain management and managing conditions such as fever, anxiety, sleep disorders, nausea, DVT prophylaxis, dehydration, and others.
In addition to supporting leading practices, order sets support resource management by suggesting cost-effective treatments, and medications.
An important principle in designing order sets is to break them down into modular components that are consistent with how physicians do their work, what their workflow is like, and how the patient progresses from one phase of care to another. For example, if a patient is admitted to the hospital for acute myocardial infarction, the tendency might be to design an order set that addresses everything the physician needs to admit a patient to the hospital with acute MI. The more helpful alternative is to access an order set when the patient initially arrives in the emergency room to address the first phase of evaluation in establishing a diagnosis and initial treatment. A subsequent order set helps prepare the patient for the intervention, such as admission to the ICU or setting up a cardiac catheterization or balloon angioplasty. If the MI is mild, a different order set addresses getting the patient quickly evaluated and into a rehabilitation program.
The access to current medical knowledge, careful design and sequencing of order
sets to make it easier to manage a patient's treatment, and the benefits of
reducing practice variations are powerful reasons why we should strive to achieve
high physician adoption of computerized physician order entry. Physicians need
to collaborate with other clinicians in designing the order sets and sequences
of orders that they will eventually use to improve patient care. Physician involvement
in developing locally approved order sets from evidence-based leading practices
helps give them confidence in the validity and usability of new clinical tools.
Physicians are much more likely to adopt a system they trust and have helped
design.
| Characteristics of Effective Order Sets |
|
Support quality care and resource management Support evidence-based leading practices
|
Case Study: Enhancing Physician Adoption
One of the biggest challenges of implementing a CPOE system is to persuade physicians to use it. The many benefits of institutionalizing leading practices and integrating patient data come to nothing if physicians choose not to place their treatment orders using the CIS. ProHealth, a relatively new health system created by the merger of two previously independent hospitals, realized that physician adoption of computerized order entry was key to getting the desired benefits from their CIS implementation. Decision makers at the new health system knew that getting physicians to adapt to a CIS wouldn't be easy. Physicians at the two hospitals were already accustomed to reviewing patient data online. Nevertheless, an earlier initiative to promote physician interest in computerized order entry by giving them their own laptop computers failed to generate physician buy-in.
The issue of computerized physician order entry became a higher priority when ProHealth recognized the competitive advantage of computerizing physician order entry and standardizing leading clinical practices into a system-wide clinical information system. Decision makers quickly realized that the key to physician adoption is physician involvement. An existing physician computer committee was supported by a carefully selected mix of physicians that included proponents of computerization, influential opinion leaders with no strong feelings either for or against computerized order entry, and computer skeptics. This physician design committee rapidly took up collaborative positions in other implementation teams and was instrumental in selecting the system vendor as well as contributing to the design of clinical documentation; pharmacy and other CIS design teams.
"It's also given us an opportunity to do some standardization of care. We've gotten people to agree on things that I never thought we'd ever get them to agree to." — Jane Neumann, M.D., Coordinator of ProHealth CIS Implementation Project
Physicians became CIS advocates and discussion leaders throughout three Rapid Design workshops, each a three- day, highly-structured, and intensive process design session. The nursing staff also participated energetically in these workshops. The Rapid Design sessions became the springboard for frequent group meetings, usually conducted in the evenings to accommodate clinicians' schedules.
"Probably the biggest thing we've done to expand physician involvement is to work on order sets, or groupings of order by specialty, and we're meeting anywhere from three to six groups a month, usually, inviting specialty physicians to come in the evening to sit down and write orders." — Jane Neumann, M.D.
The enthusiastic involvement of physicians and clinicians across the hospital system has surprised even the participants themselves. The physician groups are well on the way to writing the 400 to 450 order sets that must be in place when the CIS goes live in the spring of 2004.
One reason why group sessions are so well-attended and productive is that considerable preparation time goes into recruiting the right team members, preparing an agenda, and even putting together prototype screens for the team to discuss.
"They work with a straw model, usually, and that has helped a great deal, and whether it's a 'yea' or a 'nay' on each particular order, it helps to have something there to start with." — Jane Neumann, M.D.
Much credit for the extent of physician involvement in ProHealth's CIS implementation goes to the project's coordinator, Jane Neumann, a practicing pulmonologist. She devotes nearly 20 hours a week to the project. Over the project's duration of several years, her time, and the time spent by her physician and clinician colleagues represent a substantial investment by ProHealth. Though ProHealth's CIS implementation is a work in progress, the health system's willingness to devote the right people and the right resources to a mission-critical project seems to be generating the physician buy-in that will buy the organization a competitive edge.
Rules and Alerts
During design of a CPOE system, hospital physicians and clinicians collaborate to build in rules and alerts, which in addition to order sets, help steer clinical interventions toward leading practices. Based on patient-specific information, and clinical design committee decisions, the system can prompt, alert or remind physicians and other clinicians when an action could adversely affect the patient, such as a drug-drug interaction or when an action is unnecessary, such as when the patient is already on an equivalent medication, or if a particular test has already been ordered. Other examples of useful rules and alerts include when an important piece of clinical information becomes available that should be acted upon and when a specific action or order becomes appropriate. Rules and alerts can also make the physician or clinician aware that they may be doing something that is not a leading practice.
Rules and alerts help prevent physicians from making careless errors. Clinical alerts can also help prevent "deliberate errors," when a physician placing a treatment order is unaware of the current leading practice for a medical condition for which there is a body of clinical knowledge. There are two general categories of rules and alerts:
- Basic medication and order checking warns of potential drug interactions
and checks drug allergies, proper dosage, and duplicate therapy.
- Advanced clinical rules compare patient information and treatment orders to alert physicians to potential risks and patterns and can often suggest corrective actions. For example, the system would alert the physician to adjust medication dosages or take other corrective action in cases of abnormal electrolyte levels due to medication use or abnormal renal function. The system can also support the physician by assisting with complicated drug dose calculations in children and patients with abnormal kidney function. If a patient is on warfarin, the system automatically triggers an INR test. The system can look for patterns of symptoms and signs and alert the physician to a potential diagnosis. This is helpful in looking for very specific combinations of symptoms that are easily overlooked in isolation, such as rare communicable diseases. Rules and alerts can also improve the quality of care by ensuring the key preventive measures are not forgotten, such as immunizations in children and the elderly.
Rules and alerts also help to manage the clinical pathway, putting patients on a treatment strategy based on their individual outcomes. This can help nurses and other clinicians to deliver the right treatments at the right times.
Help Is Appreciated
In general, physicians are increasingly comfortable with the notion that reducing practice variations is a virtue. An automated order entry system that reinforces leading clinical practices and checks for potential adverse events while providing the flexibility to tailor treatment as desired is a welcome ally.
