Clinical Decision Support: Workflow Integration Is Vital for Optimizing Care

Healthcare organizations of all sizes are enlisting clinical decision support (CDS) systems to assist providers with decision making. However, to ensure proper utilization and improve the practice of evidence-based medicine, workflow integration at the point of care needs to improve.

In a review of computer-based systems, 66 percent of CDS systems significantly improved clinical practice, while 34 percent did not, according to 1998 study in the Journal of the American Medical Association. Although the technology has improved since that article was published, the message is clear: For clinicians to get the full benefit of CDS, these systems must better engage those who use them.

"Most clinicians are so busy that they can't accommodate any extraneous engagement. Pull technologies that require action on the part of the provider to bring in decision support [are] not conducive to these time constraints," says David F. Lobach, MD, PhD, chief of the clinical informatics division at Duke University Medical Center in Durham, N.C. "Therefore, push technologies that seamlessly integrate into workflow are clearly preferred and more effective."

As a result, "highly directive messages" are necessary for CDS to properly fit into the clinical workflow, says David W. Bates, MD, chief of the division of general medicine and primary care at Brigham and Women's Hospital in Boston.

For example, when a physician prescribes a medication for a patient with kidney complications, the CDS can take into account the patient's kidney function, and recommend the proper dose. "Ideally, the system does not interrupt the provider, but simply steers him or her in the proper direction," Bates says. To avoid interruption, some systems suggest a dose appropriate for that specific patient, and will only present a warning if the provider selects too high a dose.

Even small CDS alterations can make major differences. For example, Bates and colleagues at Brigham determined that ondansetron, which is used to prevent nausea and vomiting caused by cancer therapies, was as effective if given at a lower dose three times daily rather than the previous routine of four times daily. Simply changing the default dose and frequency on the ordering screen had an effect: In the four weeks before changing the default frequency, 89.7 percent of the orders for ondansetron were for four times daily and 5.9 percent were for three times daily. In the four weeks after the change, 13.7 percent of ondansetron orders were for four times daily and 75.3 percent were for three times daily. The change resulted in a cost savings of approximately $250,000 in the first year, according to a study in the November/December 2003 issue of Journal of the American Medical Informatics Association.   

Integration at order entry

CDS should be integrated into the order entry process, as many providers have moved to computerized provider order entry. "Order entry is the point at which decisions are made, and therefore, the point at which a provider should be informed of drug interactions or testing suggestions," Lobach says.

Ten Commandments for Effective Clinical Decision Support: Making the Practice of Evidence-based Medicine a Reality

Speed is everything Anticipate needs and deliver in real time Fit into the user’s workflow Little things can make a big difference Recognize that physicians will strongly resist stopping Changing direction is easier than stopping Simple interventions work best Ask for additional information only when you really need it Monitor impact, get feedback & respond Manage & maintain your knowledge-based systems Source: Bates et al, J Am Med Inform Assoc 2003;10(6):523–530

While CDS systems should provide recommendations, the systems need to draw on patient-specific data, says Ben-Tzion Karsh, PhD, associate professor, at the industrial and systems engineering department at the University of Wisconsin in Madison. "Therefore, pertinent patient-specific data need to be documented during the visit, or preferably prior to the visit to provide proper support of clinical decisions," he says.

"Particularly, in busy or ambulatory settings, providers cannot always allot time for proper documentation, and patient-specific information is entered after the visit," Karsh says. "Therefore, the CDS makes recommendations after the patient has departed. For improved utilization, patients should start inputting data via web-based portals, PHRs [personal health records] or electronic waiting-room kiosks that could inform clinical decisions."

Technological functionality and better workflow integration are still works in progress, but providers must appreciate the benefits of CDS. "Physicians will recognize the benefits of utilizing CDS when the system suggests a drug-interaction or makes a recommendation that they did not previously consider," Bates says. "A gradual CDS introduction into workflow is key."

CDS success begins with design

As for the systems themselves, many efficacious CDS systems are tied to a particular application, and they tend to be homegrown, Lobach explains. "However, for this field to grow, the decision-making process shouldn't be tied to any single application, and each provider shouldn't have to re-invent the wheel," he adds.

Many systems haven't evolved in the past decade, especially with regard to their interruptive nature, notes Bates, who, along with his colleagues, created 10 commandments for making the practice of evidence-based medicine a reality via CDS (see chart).

Lobach and Bates suggest that CDS vendors need to be supremely aware of the user's workflow and minute-by-minute clinical activities and knowledge needs.  

K. Kawamoto, along with Lobach and others, conducted a systematic review of 70 studies to identify features of CDS that are critical for improving practice. They found four features were independent predictors of improved clinical practice:

• Automatic provision of decision support as part of clinician workflow;
• Provision of recommendations rather than just assessments;
• Provision of decision support at the time and location of decision making; and
• Computer-based decision support.

"Of the 32 systems possessing all four, 94 percent significantly improved clinical practice," the authors wrote in the British Medical Journal in April 2005.

"It is critical for CDS [system] manufacturers to understand the physician workflow, and then evaluate where in the physician workflow to position CDS-driven interventions where they will be the least obtrusive, and where they can most easily integrate with existing systems," Lobach says.

Throughout the care continuum

"Particularly in the inpatient setting, there should be actions that follow each clinical decision to ensure optimal care," Lobach says. CDS recommendations will vary based on patient care settings. For instance, in inpatient settings, certain intravenous antibiotics may require reminders to monitor kidney function after the antibiotic is given.

"In busy facilities, CDS needs to complement the clinicians' strengths," Lobach suggests. "Traditionally, clinicians have relied on deductive reasoning using disparate pieces of data for diagnosis and decision making. However, individual clinicians cannot always be expected to also remember the list of required tests and contraindications. [This] provides an opportunity to rely on the computer's strength for retrieving comprehensive lists of information."

Alert fatigue is a major concern in this busy setting, and clinicians will begin to ignore any system that consistently interrupts their workflow with irrelevant recommendations or warnings. To establish a threshold for strongly "action-oriented" suggestions, clinicians should respond positively more than 60 percent of the time, according to the 2003 JAMIA study. The authors recommended that providers monitor impact early when supplying decision support by assessing how often suggestions are followed, and then make appropriate midcourse corrections.

"We need to start thinking that all data in EHRs can be used as a means to support clinical decisions," Larsh says. "Therefore, instead of only thinking of CDS in terms of pop-ups with discreet information, the data with the appropriate context could be more organically integrated into the order entry or the other data displays."

"Once the required knowledge base within an environment is fully understood, the CDS should be fine-tuned to meet the desires of the providers," Bates says. "Unfortunately, this type of intensive monitoring and tweaking is rare. Mostly, larger institutions have the resources required to undertake such adjustments, but every organization should have at least one medically trained individual who is dedicated to the evaluation and implementation of CDS."

"Providing decision support to patients and various caregivers could be very beneficial, because the systems could be used across the continuum of healthcare delivery," Lobach concludes.