At Valley Children’s Healthcare, EMR alerts had become background noise.
Clinical staff across inpatient, outpatient, and emergency settings were closing screen pop‑ups almost reflexively. Some were redundant. Others appeared for the wrong user or at the wrong moment in care. Many told clinicians what they already knew.
The volume alone made it hard to tell what mattered.
“People were being interrupted constantly, and the alerts weren’t helping them do their jobs better,” said Jacqueline Crouch, PharmD, director of quality, patient safety, and infection prevention.
The scale of the problem was visible in the data. Clinicians were spending more than 140 hours a month interacting with alerts, and practice advisories were firing well above peer benchmarks across care settings. Over time, the interruptions eroded trust not just in individual alerts but in the system itself.
“It’s the boy who cried wolf,” said Michael Scahill, MD, pediatric hospitalist and head of informatics. “When these alerts are going off constantly, people just tune them out.”
Two years later, the picture looks much different. Valley Children’s has cut overall alert volume by more than half, reduced time spent on alerts by 55%, and nearly tripled the informational value of the alerts that remain — with no changes in ordering volume or care delivery.
“When an alert fires now,” Scahill said, “it means something.”
Making alerts matter
The project goal was simple: fewer interruptions, higher-value messages, and alerts that clinicians would actually trust.
The team began by examining the top alerts. Internal dashboards showed some advisories going off tens of thousands of times over short periods and for people who didn’t need them.
“One of the early things we noticed was alerts firing for everyone under the sun,” Crouch said. “Not the people who could actually do something with it.”
In many cases, alerts were compensating for workflows that had never been fully designed during Epic go‑live due to limiting factors during the pandemic. And many advisories were created by EHR teams who weren’t clinicians and didn’t understand clinical nuances or the final users.
“There were so many alerts telling clinicians what they already knew,” Scahill said. “There’s a smart human on the other end of these alerts. If it’s firing a lot, they probably know what’s going on.”
They also prioritized recommendations from frontline staff provided through comments in the EMR.
“The feedback was gold,” said Michael Hart, PharmD, Willow analyst and clinical pharmacist. “It explained exactly why something was wrong for a patient or in a workflow.”
The team met every week, evaluating alerts using the “five rights” of clinical decision support: right information, right person, right time, right channel, and right intervention.
When they decided to remove or adjust an alert, they submitted an IT ticket and met with the relevant stakeholders.
One early example came from the PICU. A frequently firing alert reminded nurses to turn patients every two hours to prevent pressure injuries. In practice, it interrupted nurses repeatedly, often before documentation was complete.
Based on nurse feedback, the reminder was moved to the NurseBrain, where it appeared alongside other routine tasks and allowed nurses to document turns directly. The alert itself was eventually turned off.
Similar redesigns followed across the organization:
- Sedation target alerts were removed from medication workflows and embedded into admission orders, ensuring goals were established once and consistently applied.
- Ambulatory lead screening alerts were rewritten after reviews showed they relied on procedure orders instead of lab values and missed point‑of‑care results.
- Rare disease pharmacogenomic alerts were scaled back after data showed they followed patients across settings, repeatedly interrupting clinicians who already knew the diagnosis.
As confidence grew in the redesigned system, the team made its most significant decision: disabling pediatric early warning system alerts. Clinical teams were identifying deterioration earlier than the system, and the alerts often fired during high‑stress moments, adding friction rather than clarity.
“It wasn’t a light decision,” Scahill said. “But clinicians were already acting, and the alerts were getting in the way.”
After the change, alert rates dropped sharply and have remained stable — with no evidence of delayed escalation.
Results that changed day‑to‑day work
Over time, the cumulative impact became clear:
- Total alert volume fell by more than half compared with baseline.
- Time spent interacting with alerts dropped 55%, from 143 to 65 hours per month.
- Monthly labor costs associated with alerts fell by nearly $9,000, driven largely by reduced provider time. (Reports on time spent engaging are available in Epic.)
- Information content per alert increased more than 250%, meaning fewer alerts delivered more meaningful signals.
Importantly, ordering volume remained steady, confirming the improvements were driven by redesign, not reduced clinical activity.
Nurses saw the largest absolute reduction in time spent closing alerts. Providers experienced the most consistent month‑over‑month declines.
“When an alert fires now, caregivers pay attention,” Crouch said.
Keys to success
The work was led by a small, multidisciplinary team combining clinical experience, informatics expertise, and deep familiarity with Epic build and data.
“We can move from clinical question to analysis to build change very quickly,” Scahill said. “That makes iteration possible.”
Their advice to peers:
- Start with your highest-frequency alerts. If they fire constantly, they’re unlikely to add value.
- Respond to feedback. Closing the loop builds trust and improves alert quality.
- Be relentless but patient. Cultural change takes time, but clinicians will notice when alerts finally mean something.
- Keep the focus on patients and care teams. Take care of the people taking care of patients.
“When we make life better for users, we make care better for the patients,” Hart said. “That’s what this is all about.”
This work was presented on the "Reducing Alert Fatigue to Restore Joy in Work" poster at CHA's 2026 Together for Kids Pediatric Quality Conference.
