Two units at Palo Alto, Calif.-based Lucile Packard Children’s Hospital Stanford cut bedside monitor alarms from 60,000 to 26,000 without an increase in emergencies, The Wall Street Journal reported Sept. 28.
The result points to a shift in how hospitals are tackling alarm fatigue. Rather than simply silencing monitors, health systems are making alarms wait, sending quieter risk signals and hardwiring what clinicians must do when a signal fires.
The stakes of reducing alarm noise run in two directions. For patients, constant beeping makes sleep nearly impossible: Before Stanford’s changes, about 40% of families said monitor alarms disrupted their child’s sleep. The sheer number of alarms encourages nurses to tune them out, including those that signal deterioration.
Hospitals have long been louder than recommended. A Johns Hopkins study in 2005 found daily average sound levels of 50 to 60 decibels, well above the World Health Organization’s proposed 35 to 45 decibels during the day and 20 to 35 at night, Becker’s reported in 2023.
Making alarms wait
Stanford’s pediatric ICU and acute cardiac care units were producing large numbers of “nonactionable” alarms, set off by brief, self-correcting or clinically unimportant changes. Nurses still had to address every one.
“As we were adding new technology for more things we can do at the bedside, the alarm burden was only increasing,” Felice Su, MD, medical director of the pediatric ICU, told the Journal.
Stanford reprogrammed monitors with time delays and conditional triggers. A brief dip in a patient’s blood-oxygen reading now gets time to correct itself, while a sustained or more serious drop sets off an alarm sooner.
Over a one-month study period, alarms per patient per day fell 75% in the pediatric ICU and 82% in the cardiac unit. Nurses reported they could respond to alarms more appropriately. The effect on sleep was less clear: Family-reported disruptions fell, but the change was not statistically significant.
San Francisco-based UCSF Health found similar math in adult care. Continuous pulse oximeters generated an average of 58 low-oxygen alarms per patient per day. Of the alarms set to fire when oxygen saturation fell below 92%, 70% never led to a drop below 88%, and half resolved within 30 seconds.
In a pilot, moving the threshold to 88% delayed detection by 11 seconds, and those 11 seconds cut alarms by 70%, Wendy Horton, PharmD, senior vice president and president of university hospitals at UCSF Health, wrote in a January column for Becker’s.
“Sometimes, the most powerful question isn’t, ‘What new thing should we build or buy?’ but, ‘What should we stop doing or do differently?'” she wrote.
Both approaches follow long-standing nursing guidance. In 2018, the American Association of Critical-Care Nurses recommended hospitals set unit-specific default alarm parameters and monitor only patients with a clinical indication.
Quieter signals
The same logic has been applied to the EHR. In 2020, Nashville, Tenn.-based Vanderbilt University Medical Center’s Clickbusters program cut about 49,000 of the 488,000 best practice advisories that fired each week, roughly 10% of alerts.
The goal was not fewer alerts for their own sake. “We believe in alerts. There should be a lot of good alerts in the system that are accepted a lot of the time and that people find to be useful,” said Adam Wright, PhD, director of the Vanderbilt Clinical Informatics Center, who led the effort.
A newer class of tools skips the alarm altogether. Concern, a machine-learning early warning system, scans nursing documentation in the EHR for signs that nurses are worried about a patient, such as checking vital signs more often than usual.
Nursing documentation is well suited to that kind of monitoring.
“Nursing documentation occurs throughout the entire shift. It’s every time we interact with a patient,” Stephanie Clements, BSN, RN, senior vice president and chief nurse executive at Chesterfield, Mo.-based Mercy, previously told Becker’s.
Concern recalculates each patient’s risk hourly, but it does not send an alert. Instead, a green, yellow or red icon appears beside the patient’s name in the EHR, and clinicians can click it to see what is driving the score.
“Unlike many early warning systems that generate interruptive alerts, we designed Concern as a noninterruptive alert,” Patricia Dykes, PhD, RN, director of patient safety and innovation at Atlanta-based Emory Healthcare, told the Journal.
In a randomized trial of more than 60,000 hospital stays, patients on units using Concern had a significantly lower risk of dying in the hospital and shorter stays than patients receiving usual care.
A response nobody can skip
A quiet signal only works if someone acts on it. At Somerville, Mass.-based Mass General Brigham, an early warning system in the Epic EHR flags patients who appear to be deteriorating. When it fires, paging the attending physician is mandatory, and the attending must come to the bedside.
“You wouldn’t get on a plane where one human who may have had a bad night does the safety check and flies the plane,” Rachel Sisodia, MD, chief quality officer at Mass General Brigham, told Becker’s.
What remains unsettled
Some researchers are redesigning how alarms sound. At Vanderbilt, anesthesiologist and musician Joseph Schlesinger II, MD, and McMaster University music-cognition researcher Michael Schutz, PhD, built blood pressure alarms around the “lub-dub” of a heartbeat, with shifts in pitch and harmony that signal whether pressure is high or low and how severe the problem is.
In a 2023 study, participants recognized an urgent high-blood-pressure alarm about 10% faster when it used musical cues, according to the Journal. A later version was identified correctly 90% of the time, compared with 80% for an earlier design.
“It doesn’t need to be the New York Philharmonic in the operating room, but as a system for communicating information effectively, music has a lot to teach us,” Dr. Schutz told the Journal.
Several questions remain open. Stanford’s results cover one month in pediatric units. Patients on Concern units were also more likely to be moved to intensive care, which researchers said could reflect deterioration being caught earlier.
