Wolters Kluwer Health

Stephen Claypool

Some of today’s clinical decision support (CDS) systems are making inroads into improving care outcomes. But most still lack the features required to achieve their stated goals. This is evident in the results of a meta-analysis of 162 randomized trials, which revealed that nearly 50 percent of CDS trials failed to improve outcomes. And this trial overestimated the benefit; we tend to only study projects that we strongly believe will be successful, ignoring the projects that seem less impactful. Plus, there is a tendency to avoid publishing negative studies. So, in reality, only a small fraction of deployed CDS is effective at improving outcomes.

"CDS systems often need to be deployed in concert with robust change management that includes education and establishment of policies"

Why? One reason is alert fatigue. Most systems lack the ability to deliver highly specific alerts, which erodes clinician confidence and leads to the failure of point-of-care alerting— not to mention alert fatigue, clinician burnout and, ultimately, turnover. A well-cited article revealed that clinicians ignore electronic health record (EHR) safety notifications between 49 percent and 96 percent of the time. Alert fatigue is a major problem with alerting systems.

To improve these results, CDS solutions must be built with the capabilities necessary to address problems of alert fatigue.

Systems are more successful when interrupting alerts are pushed to clinicians and a response is required, as opposed to passive alerts. Although annoying if not deployed carefully, modal alerts are more likely to improve outcomes. Further, when alerts are overridden, systems should mandate justification and documentation for the override. This requirement reduces non-compliance—and even reduces alert fatigue.

In addition to automatic alerting, effective CDS should integrate into clinician workflow at the time of decision making. It should also include actionable, evidence-based recommendations.

Another essential factor of alert acceptance is the quality of alert display and data presentation. CDS systems must contain the functionality to determine the correct providers on the care teams to receive the alert, distinguish if the recipient is aware of the situation, and fire the alert at the appropriate time. For instance, if a patient is experiencing a severe medical problem, such as a heart attack, an alert for a less urgent medical issue may be irrelevant and cause alert fatigue.

In summary, hospitals and healthcare systems greatly benefit from CDS systems that adhere to the following 10-point checklist to ensure successful alerts and outcomes:

1. Active, interrupting alerts
2. Real-time alerts with high sensitivity and specificity— high specificity is required to reduce alert fatigue
3. Seamless integration with EHR
4. Actionable recommendations
5. Requires override justification and documentation
6. Delivered to the point of care in an easy-to-use manner
7. Directed to appropriate provider
8. Appropriately timed within clinician workflow
9. Only when relevant to the patient episode
10. Assists the provider with early detection of situations of which they may not be aware

This checklist provides a detailed outline for achieving the “Five Rights of CDS”, as outlined by the U.S. Department of Health and Human Services, Agency for Healthcare Research and Quality (AHRQ), to enable sustainable CDS alert improvements. The “Five Rights” states that healthcare providers can achieve CDS-driven outcomes improvements by providing the right information, to the right person, in the right intervention format, through the right channel, at the right time in workflow.

As healthcare organizations increasingly turn towards CDS solutions to improve care without affecting quality, it remains critical that CDS systems be based upon both the CDS checklist and the Five Rights to ensure improved, sustainable alerting, and results. By investing in superior CDS solutions, healthcare systems can effectively minimize false positives and associated complications, while improving processes that result in significant savings, early identification and treatment of illness, and improved outcomes.

Perhaps more importantly, healthcare systems need to steer away from deploying systems that do not adhere to these checklist principles. Our informatics history is riddled with failing systems, and this legacy has created a sense of alert fatigue that has produced headwinds for future CDS. To prevent deployment of poor systems, careful study should be undertaken prior to implementation. Informaticists should understand their system’s sensitivity, specificity, incidence, and positive predictive and negative predictive values. In fact, informaticists should know, off the top of their heads, these vital statistics—yet embarrassingly, few facilities have any idea how well their CDS tools will perform. They don’t know how often the CDS will fire, how often clinicians will comply with it and how big the impact will be.

No wonder we fail! Informaticists need to clearly understand the goal of the CDS—not just an intention to increase a behavior, but the overall goal of changing vital statistics of the health system. This will lead to a greater understanding of key populations, personnel, and workflows that need to be targeted for CDS. Furthermore, CDS systems often need to be deployed in concert with robust change management that includes education and establishment of policies.

Unless we embrace this all-encompassing, scientifically-minded approach, CDS systems will continue to under-deliver on the promise of improving clinical outcomes.