I recently attended the AMIA Clinical Informatics Conference (CIC) in Minneapolis, where the ongoing focus was the utilization of AI and tools like ambient listening to enhance the clinician experience. Clinician burnout is a multifaceted problem and it’s clear that a component of this issue is tied to the demands of documentation. However, the EHR more broadly plays a contributing factor. For instance, a study highlighted in JAMIA noted that physicians at one health system needed to make up to 61 clicks in the EHR just to order Tylenol. This is a glaring example of how cumbersome these systems can be.
Clinician burnout intersects with various factors like reimbursement requirements and payor documentation. Originally, EHRs were envisioned as a tool to simplify clinicians' duties and tasks. However, they have instead blossomed new issues related to note bloat and “pajama time,” where clinicians spend hours after work completing documentation. This indicates a divergence between the purpose of EHR systems and their real-world implementation. It’s become so painful that health systems have come together in groups like the Arch Collaborative to share their struggles and successes.
Moreover, while clinical decision support (CDS) systems within EHRs should ideally guide clinicians efficiently and reduce cognitive load, they are often bluntly implemented. After their initial implementation, these systems are rarely optimized post-launch, leading to significant disruptions in daily clinical workflows that often go unreported.
One key approach to reducing EHR burden is refining alert systems within the EHR. Interruptive alerts are notorious for breaking the concentration of healthcare providers, and cutting down on these can lead to real time savings. For example, a paper from JAMIA attempted to quantify “alert dwell time,” which measures the time spent dealing with alerts. The authors describe the median drug-drug alert dwell time is 8 seconds. In other words, a reduction in 1 million interruptive alert firings on an annual basis can give back over 2,000 hours to the organization’s workforce.
However, looking at total alert firings can miss out on significant pain points within the organization. At Phrase Health, we run an analysis that assesses the impact of these alerts through what we call the “Phrase Burden Index” (PBI). This tool identifies specific alerts that create significant disruptions at the individual provider-level. For instance, during a recent implementation, we discovered an alert that fired interpretively 70 times per day (only about 2,000 times per day). You might think “no big deal”, so why is the PBI off-the-charts. Well, we found it was unintentionally built in a way where it only fired to a single ambulatory provider. Of note, the alert had been live for several years and was overridden every. single. time. When the EHR vendor’s new feedback functionality was turned on, the physician end-user gave consistently negative feedback for about 2 weeks, but then the user gave up again. This is a striking example of how the physician learned to live with a malfunctioning workflow for years, while the organization worked through an alert review process that focused on total firings across the enterprise.
To effectively reduce the EHR burden on clinicians, a multi-faceted approach is necessary. This involves not only technological improvements but also a fundamental shift in how we think about the implementation of these very complex systems into healthcare. We must:
At Phrase Health, we’re actively working on solutions that make EHRs a tool for empowerment rather than a source of frustration. By fostering a culture of continuous iteration and adaptation, informatics and quality groups can significantly alleviate the burden EHRs place on our healthcare providers, ultimately leading to better patient care and clinician well-being.