April 11, 2011

Decision Support: A Foundation for Success
By Alice Shepherd
For The Record
Vol. 23 No. 7 P. 10

Clinical decision-support systems can lay the groundwork for better and safer care, but the process requires plenty of heavy lifting prior to implementation.

Besides requiring computerized physician order entry (CPOE), meaningful use criteria specify that providers who want to qualify for federal financial incentives implement clinical decision support (CDS) in the form of drug-drug, drug-allergy, and drug-formulary checks as well as five CDS rules relevant to specialty or high clinical priority, including rules for diagnostic test ordering. Providers will also need to demonstrate the ability to track compliance with those rules.

As many organizations are currently struggling to meet the requirements, how can they ensure that their CDS systems will achieve the ultimate goal of improved patient safety?

So far, success has been elusive for many hospitals. A research report published in the April 2010 edition of Health Affairs (“Mixed Results in the Safety Performance of Computerized Physician Order Entry”) studied a national sample of 62 hospitals that voluntarily used a simulation tool designed to assess how well safety decision support worked when applied to medication orders in CPOE. The assessment tool mimicked what happens when a physician writes an order in an EHR using CPOE.

The simulation tool detected only 53% of the medication orders that would have resulted in fatalities and 10% to 82% of test orders that would have caused serious adverse drug events. While top-performing hospitals achieved scores of 70% to 80%, many hospitals performed poorly—the mean score was 44% of potential adverse drug events detected.

“It’s not as simple as asking, ‘Do you have decision support or not?’” says David Classen, MD, a study author and associate professor of medicine at the University of Utah in Salt Lake City and partner and chief medical officer with CSC Healthcare Group. “The question is, ‘What kind of decision support do you have, and how have you implemented it?’ Our test shows that there is tremendous variability in how effectively decision support is implemented within EHRs. It’s therefore critical to give organizations self-assessment tools to assess the effectiveness of their decision support. Our assessment tool is publicly available to any hospital through the Leapfrog Group. We also plan to release an ambulatory version of the tool soon; its application so far suggests that decision support is also not very well implemented in ambulatory EHRs.”

What can be done to improve CDS systems? “The true effectiveness of CDS systems depends on both good content and effective implementation,” says Classen. “Some improvements will ensue from the inevitable customization and cycles of improvement that accompany any new technology. Others will come from vendors adding more and more capabilities to their products, more and more content or knowledge about how to implement them properly. However, organizations are responsible for implementing decision support because it is a standard-of-care issue.”

It Starts With Solid Content
“Content” refers to the actual rules stating, for instance, that a specific drug cannot be taken with another. “There have been attempts at providing public access to rules related to existing guidelines,” says Floyd Eisenberg, MD, MPH, senior vice president of HIT at the National Quality Forum. “Much of that work has been going on from the ‘80s and ‘90s. The challenge is that implementing rules always takes local work because there is no standard rule that could be picked up and easily adopted in any system. So the task is to define a method that allows the rules to be more easily shared. Projects are currently under way, funded by the ONC [Office of the National Coordinator for Health Information Technology] and by AHRQ [the Agency for Healthcare Research and Quality], to do just that.”

Until rules become publicly available, organizations have three choices. “One option is to purchase electronic order sets from vendors,” says Judy Hanover, research manager at IDC Health Insights. “These are databases for drug-drug interaction, drug-allergy, and drug-formulary checking. Some providers choose to develop their own rules and order sets, which is usually done by a committee or an implementation team. Others purchase third-party order sets and then modify them and customize them to the care setting.”

Developing rules and keeping them up-to-date with the latest medical research is an almost insurmountable task for most organizations.

“I once heard a speaker who said that a physician who committed himself over the next 30 to 40 years to read two good articles a day in his specialty would be something like 10,000 years behind the latest evidence by the time he retired,” says Jason Hess, general manager of clinical research at KLAS Research. “Vendors that provide electronic order sets have large contingents of physicians constantly gleaning all the materials across the planet, finding the latest evidence, and using it to build tools to assist clinical decision making. Rather than building order sets in-house after getting consensus from all their specialists, organizations can get up and running faster with CDS by purchasing evidence-based order sets from a third party, such as Zynx or ProVation, and modifying them as necessary.”

ProVation’s electronic order set authoring and maintenance tool utilizes evidence-based clinical content from Wolters Kluwer’s UpToDate, an electronic medical reference system. The company also offers Medi-Span and Medi-Span Clinical, an electronic medication decision-support system that integrates with EMR/CPOE systems to provide drug-interaction, route-contraindication, and drug-allergy alerts as well as links to supporting medical evidence.

Steve Claypool, MD, Wolters Kluwer Health’s vice president of clinical development and informatics, says the system delivers a library of evidence-based electronic order sets for many medical conditions.

“The order sets provide decision support in the form of hyperlinks and narrative text from UpToDate’s reference content, which is developed and regularly updated by physicians who constantly review the latest scientific studies from around the world,” he notes. “The order sets also utilize the medication database from Medi-Span.”

A recent study by KLAS Research (“Clinical Decision Support: Striving for More Intelligent Care”) divided CDS into five key areas: evidence-based order sets, multiparameter alerting, evidence-based nursing care plans, evidence-based reference content, and drug information databases. Of the 100 hospitals surveyed (all of which had full EMRs and CPOE), 38% indicated that electronic order sets had the biggest impact on improving quality and safety, patient length of stay, and mortality rate.

“Respondents stated that this is because it standardizes the care and forces the physicians to follow the safest ordering practices,” says Hess. “Electronic order sets provide proven content based on the latest evidence, which allows clinicians to improve their craft.”

Nearly 50% of the study participants used third-party order sets, while about one-quarter opted for homegrown solutions. Nineteen percent and 10% of these survey participants, respectively, stated that evidence-based alerts and reference content were also having a positive impact on the organization.

Standardizing care with electronic order sets can also save money when the order sets are properly implemented and maintained. Hess cites an example of a hospital that used to order two units of blood every time it needed blood.

“Because in many cases only one unit was needed, a great deal of blood was wasted,” he says. “They made a change to their order sets to begin checking patients’ creatinine levels before placing the order, which would tell them how much blood was actually needed. They saved a great deal of money just by standardizing and having a rule around checking the creatinine levels.”

In addition to helping ensure medication safety, CDS order sets can also reduce unnecessary testing.

“When a physician orders diagnostic lab tests or radiology, basic CPOE decision support checks for duplicate tests from the same or other providers,” says Hanover. “In some cases, providers may still need to order duplicate tests if they’re not able to access the results. However, in an ideal care environment where all the providers that interact with a patient are connected electronically, via an HIE for example, it can reduce the number of duplicate tests, thereby saving time and money and avoiding unnecessary patient discomfort.”

“The practice patterns of clinicians can be positively impacted by effectively implementing order sets that prompt for evidence-based and best-practice orders and support care decisions by providing CDS in the form of instruction and reference content,” says Claypool. “Ideally, CPOE systems should also modify orders within an order set based upon information within the EMR. For example, a test may be unchecked in an order set if the patient recently had that test performed.”

The Real Work Begins
While content can be readily obtained from a vendor, the real challenge comes during the implementation of CDS systems. Here it’s important to realize that CDS systems are not plug-and-play.

“While vendors need to provide the infrastructure and capability to enable appropriate implementation, they cannot easily integrate CDS into the workflow,” says Eisenberg. “Localization is required because implementing an EHR and CDS system requires close attention to organizational workflow, staffing, and infrastructure. How any given rule is implemented differs from site to site.”

“It takes a significant effort to ensure the EMR supports appropriate workflows and provides user interfaces that make the use of CDS systems palatable for clinicians,” says Claypool. “An important part of that effort involves selecting appropriate alerts and reminders and implementing them properly so that they don’t cause alert fatigue.”
Providers can take several steps to prevent alert fatigue. Namely, they should prioritize.

“Organizations need to decide which alerts should be merely suggestions, which should be prominent warnings physicians have to acknowledge, and which should be hard stops that prevent physicians from ordering drugs under certain circumstances,” says Classen. “The literature shows that casual warnings are usually ignored, but hard-stop rules may prevent care from being given to a patient.”

By and large, clinicians don’t want their day impeded by a barrage of warnings. “A number of studies show that clinicians often view alerts as obtrusive to the usual clinical workflow,” says Eisenberg. “Turning on a large number of rules leading to alerts popping up all over the place leads rapidly to alert fatigue. Sometimes the information isn’t going to the right person. There also has to be a filter to deal with what is highly significant and when and how to intervene.”

Large headaches could be averted if healthcare organizations are willing to spend the time and resources to analyze data prior to implementation.

“Sophisticated technology doesn’t just prompt an alert on every possible problem but issues alerts that are situational, taking the patient’s history as well as the clinical situation into account,” says Hanover. “These alerts target specific issues of high concern for providers so they can focus on alerts that matter and not on those that don’t. The top mistake hospitals make is not spending enough time prior to implementation to drill down and discern which reports and types of alerts will be most important and have the best effect on the particular provider setting. When hospitals turn on too many alerts, it can also slow down the run-time performance of EMRs.”

“Surveillance and analysis are necessary to understand which alerts are important to an institution,” says Claypool. “Maintaining the alerts also takes organizational effort and labor. Someone has to keep up with the changes in medicine and changes in the quality of care that may necessitate a change in the alerts that are turned on or off. Surveillance is also important for real-time patient care management.”

If done correctly, there’s little doubt CDS systems can be advantageous on several levels. “While it can be a struggle to implement CDS systems and make the necessary changes to workflow, it’s almost unanimous that physicians don’t want to give them up once they’ve become accustomed to them,” says Classen.

Eisenberg sums up the goal of workflow integration by referring to HIMSS’ “Improving Medication Use and Outcomes With Clinical Decision Support: A Step-by-Step Guide.” It describes CDS’ “five rights” as getting the right information to the right stakeholder at the right point in workflow through the right channel in the right format.

For organizations that want to evaluate the effectiveness of their CDS implementation, the National Quality Forum has developed CDS Taxonomy, a classification and categorization of the information necessary for quality improvement as described in “Clinical Decision Support, Driving Quality and Performance Measurement — A Foundation for Clinical Decision Support” (2010).

Taxonomy categories include the following:

• Trigger: events or actions that initiate a CDS rule;

• Input data: additional data from the patient record or other source used as background to modify or constrain the CDS rule;

• Interventions: the possible actions taken by decision support to provide information when the conditions specified in a rule are met; and

• Action step: any action or event presented to the user of a clinical system that could lead to successful completion (or realization) of the intended mission of the rule.

“The taxonomy provides a foundation for the description of an electronic infrastructure, bridging quality measurement and health IT,” says Eisenberg. “It will enable quality measure developers, clinical system implementers, and vendors to be more effective in developing the tools, content, and procedures that are compatible with and enable comprehensive use of CDS, thereby improving delivery of evidence-based care. Because of the attention to EHR use and all the discussions around meaningful use, the industry is currently in a transition period that will likely generate some positive and negative reports about the value of CDS and EHRs in general. It’s important to maintain a strategic view that this is an evolutionary process with the goal of improving health, not just healthcare.”

— Alice Shepherd is a southern California-based business-to-business journalist specializing in healthcare topics.