Community Health Network Deploying Analytics Tool to Improve CDS
In 2020 the ECRI Institute ranked alert fatigue among its Top 10 Health Technology Hazards. Recognizing the danger, Indianapolis-based Community Health Network, which has five hospitals in central Indiana, has made a concerted effort over the past few years to address alert fatigue caused by an overabundance of clinical decision support rules.
“When we started on this journey five or six years ago, we were the probably the poster child for alert fatigue,” said Patrick McGill, M.D., chief analytics officer of Community Health Network. “We had over 30 million BPAs [best practice alerts] firing in our Epic environment annually.”
About three years ago the health system set out to take a methodical approach and turn off the alerts that are not being followed or that are firing in the wrong place, and tried to follow the five rights of decision support: the what (information), who (recipient), how (intervention), where (format), and when (workflow) for a proposed intervention.
“We had great success, and we turned off 12 million to 15 million of those alerts within the first six months,” McGill said. “We got buy-in from many of our providers because they saw that we were taking the situation seriously. We put governance in place around who could request an alert. Before that, it was really anybody and everybody could request an alert and pretty much it got turned on. But
The health system also tasked a clinical nurse specialist as the clinical decision support manager and expanded the definition of decision support to include not only BPAs, but also order sets and anything that could guide clinical decision-making, including predictive algorithms in the EHR. “We've had success by being more surgical about what we're turning on and turning off. To do that, you really have to have a robust analytics platform,” McGill added. Community Health built some analytics tools of its own, but to gain a more thorough view of which clinical decision support is effective and which CDS rules are not working as intended, it also is an early adopter of a tool called FDB CDS Analytics from First Databank.
FDB CDS Analytics is a cloud-based application that leverages the health system’s data warehouse, along with FDB’s drug knowledge and expertise, to offer pre-configured metrics and reports to track the effectiveness of decision support. The solution offers dashboards designed for different roles within the organization, such as chief medical information officers, pharmacy directors, analysts and other stakeholders, and the ability to view trends across the entire institution, at the departmental level and even for the individual clinician.
“FDB has historically help us with medication alerts,” McGill explained, “but now, with this suite of tools, it's also going to look at either native Epic best practice alerts that we've created or other types of alerts, and help us marry the two together. We can highlight them and say this alert’s performing well or it can help us do a deeper dive into why people are not following certain alerts,” he added. “Are they firing at the wrong time? Are they firing in the wrong place for the wrong provider? What we find a lot of times is that we have alerts created with the best intentions, but when you put them out in the wild, they either don't behave like you think they're going to or you have them firing for the wrong person who can't take action on what it's telling them to do.”
In a statement, FDB President Bob Katter said, “Clinical decision support tools are essential in modern EHR workflows, but it can be difficult for healthcare organizations to determine which CDS initiatives are providing clinicians with the optimal guidance they need at the right time and which are just contributing to alert fatigue. FDB CDS Analytics simplifies and accelerates the process of evaluating CDS across their organization with an intuitive and powerful tool that offers deeper insight into how clinicians are using CDS. This insight also informs how CDS can be adapted to more effectively support clinicians in the delivery of safe and high-quality patient care.”
