In Backchannel on Medium, you’ll find a harrowing, fascinating story: How Medical Tech Gave a Patient a Massive Overdose.
He noticed something that stopped him cold. Six hours earlier, Levitt had given the patient not one Septra pill—a tried-and-true antibiotic used principally for urinary and skin infections — but 38½ of them.
Levitt recalls that moment as the worst of her life. “Wait, look at this Septra dose,” the resident said to her. “This is a huge dose. Oh my God, did you give this dose?”
“Oh my God,” she said. “I did.”
The whole story is definitely worth reading. Much of the narrative turns on how ignoring an alert nearly turned deadly for one patient. Nurses are exposed to thousands of alerts and alarms each day. Sue Sendelbach, RN, PhD, CCNS and Marjorie Funk, RN, PhD define alarm fatigue as “sensory overload when clinicians are exposed to an excessive number of alarms, which can result in desensitization to alarms and missed alarms,” in AACN Advanced Critical Care.
Less an alarm, more a noise in the background
In 2014, clinicians were exposed each day to tens of thousands of alarms. A sample study of an intensive care unit at the Johns Hopkins Hospital in Baltimore showed that hospital staff are exposed to an average of 350 alarms per bed, per day.
The typical ICU bedside is littered with devices that produce, on average, more than 40 different alarm and information signals.
Devices with alarms include:
- infusion pumps
- feeding pumps
- pulse oximeters
- intraaortic balloon pumps
- sequential compression devices
The vast majority of clinical alarms, 72% to 99% according to some research, are false. A 2008 questionnaire of US hospitals found that only 1% of all alarms resulted in a therapeutic intervention. Only .6% of alarms were necessary in ED patients with angina and low coronary risk.
Clinical alarms exist to alert clinicians to deviations from a predetermined normal status. But when alarms themselves become normal, they stop enhancing safety and start putting patients at risk.
“The biggest culprit” in the boy’s hospital poisoning “may well have been the hospital’s incessant electronic alerts. Some automated warnings misled the medical staff; others were lost in the cacophony of alarms going off throughout the hospital.”
You need alarms to be sensitive enough not to miss any critical events, but discerning enough to only alert when attention is absolutely needed. Turning off unnecessary alarms is a waste of nurses’ time. Worse, when alarms cry wolf too often, nurses respond by turning down the volume of audible alarm signals, adjusting the alarm settings outside limits that are safe and appropriate for the patient, ignoring alarm signals, or even deactivating alarms.
Another study of US hospitals showed nurses taking up to 40 minutes to respond to alarms. Still another showed caregivers attending to only 10% of all alarms. A third study demonstrated that caregivers could not correctly identify half of all relevant alarms.
The deadly consequences of alarm fatigue
A Boston Globe investigation identified at least 216 deaths nationwide linked to alarms which monitor heart function, breathing, and other vital signs between January 2005 and June 2010. Many of the alarms for the patients who died were ignored in a cacophony of beeps.
In one such case, an alarm signaled that the patient’s telemetry battery was dying went off for about 75 minutes. When the battery died, the patient went into cardiac arrest. But no loud alarm sounded. The patient was found unresponsive and could not be resuscitated.
The 2015 edition of the ECRI Institute’s top 10 patient safety concerns for healthcare organizations puts alarm hazards at number one. It held the top spot in 2013 and 2014 as well. A 2013 Joint Commission (TJC) Sentinel Event Alert cited 98 reports of alarm-related events over a 3.5-year period, with 80 of those events resulting in deaths and 12 in permanent loss of functions.
And last year 19 out of 20 hospitals surveyed ranked alarm fatigue as a top patient safety concern, according to a national survey presented at the annual meeting of the Society for Technology in Anesthesia.
Alarms also bother patients. Interrupted sleep actually delayed convalescence for ICU patients and reduced their patient satisfaction.
A national survey showed that effective alarm management required hospitals devote the necessary resources to develop effective alarm management schemes. Alarms have to be a supplement to assessment, they can’t replace human judgment.
Here are some processes that can help limit false alarms.
Proper skin prep
One researcher reduced the average number of alarms per bed per day by 46% by properly preparing patient skin before placing ECG electrodes. Dead skin cells impede signal quality and degrade the quality of measurement. In two separate studies, lightly sanding with fine sandpaper decreased skin resistance and minimized artifacts that could mimic a tachycardia and set off an alarm. It also helps to change the electrodes every day.
Delay the alarm
Irrelevant alarms can be reduced by up to 80% by increasing the alarm delay to 19 seconds. Most alarms don’t need to be immediate and can be delayed without any risk to patient safety, with the exception of life-threatening alarms, such as asystole or atrial fibrillation. While these must sound immediately, the average heart rate alarm could be delayed by up to 30 seconds.
This requires that systems allow clinicians to lengthen the maximum alarm delays in the standard specifications.
Widen the defaults
When you combine nurse education with revising defaults on monitor alarms, including parameter limits and levels, you can see a 43% reduction in critical monitor alarms.
According to one study, increasing alarm delays in SpO2 monitors from five to 15 seconds decreased alarms by 70%. Decreasing alarm thresholds from 90% to 88% decreased alarms by 45%.
Abbott Northwestern Hospital in Minneapolis provides an example. Administrators at the hospital took on alarm fatigue by changing the default settings for pulse-rate alarms. They sound when beats per minute are too high or low. But the threshold to go off was very conservative. By widening the acceptable range, and getting rid of duplicate alarms for cardiac issues, they reduced pulse rate alarms by 76% in six months. And they did this without missing any patient emergencies. In fact, patient safety improved as nurses responded more quickly to fewer, more serious, alarms.
The initiative has been successful enough that Abbott Northwestern has expanded it to its neuro ICU and is planning on reviewing the frequency of alarms for IV lines.
In a similar vein, researchers at Cincinnati Children’s Hospital Medical Center changed their cardiac monitor procedure for the hospital’s pediatric bone marrow transplant unit. They set alarm thresholds according to age, instead of using one default across the board. They started changing electrodes and personalizing assessments of cardiac monitor parameters daily, while appropriately discontinuing monitors.
The project yielded a decrease of 140 fewer median daily cardiac alarms, from 180 to 40, while caregiver compliance increased from 38% to 95%.
Boston Medical Center (BMC) reduced its weekly audible cardiac alarm rate by 89% by adjusting monitor alarms for bradycardia, tachycardia, and heart rate limits.
Prioritize and limit alarms
Get input from staff and published best practices and guidelines to determine which signals are most urgent. Many systems convey essentially the same information with different alarms. For example, you don’t need both a tachycardia and a high heart rate alarm, but many hospitals will have both.
Hospitals should also turn off alarms for conditions that doctors no longer treat, such as premature ventricular contractions (PVCs). Twenty years ago the Cardiac Arrhythmia Suppression Trial showed excess mortality related to the use of antiarrhythmics for asymptomatic ventricular arrhythmias. However, PVC alarms remain one of the primary causes of nonactionable alarms.
In many ways hospitals are behind another high-stakes, high-signal volume industry: the airlines. The author of the Medium story interviewed Captain Chesley “Sully” Sullenberger, the famed “Miracle on the Hudson” pilot to see what medicine might learn from the aviation industry.
They prioritize alarms carefully, with a hierarchy of signals which target different senses, including touch, sight, and hearing. For some signals, a visual alarm can work better than another beeping sound.
Statistical signal extraction
Technology that utilizes robust statistical signal extraction can also help limit false alarms. Outliers are a huge source of false alarms. You want to ignore outliers, or extremely short deflections in measurements, caused by movement artifacts or other technical reasons such as flushing an arterial line and not by physiologic changes. Robust regression methods, with sliding windows, extracts signal from noise. The system produces a cleaned dataset and compares the signal with alarm thresholds for the alarming decision.
“Among the most critical considerations are the capabilities and configuration of technology,” said Michael Wong, executive director of the Physician-Patient Alliance for Health & Safety, who presented the findings at the annual meeting of the Society for Technology in Anesthesia.
Changing and personalizing defaults around thresholds and what gets monitored is essential, but it requires flexible, user-friendly technology. The monitors have to be configurable, and most are not. Manufacturers establish alarm settings most often right now, according to the AACN. And for configurable devices, nurses need to know how to set them.
Connecting medical devices helps reduce instances where the same situation triggers multiple alarms from different devices. And so-called “smart” alarms consider other parameters before sounding. For example, they can take blood pressure into account before alarming for asystole.
Between 1987 and 1993, a group of Stanford computer scientists and physicians developed a cardiac surgical ICU system which correctly detected and diagnosed approximately 30% of the typical ICU complications.
They used root cause analysis to intelligently interpret alarms. The goal is to figure out automatically why an alarm is sounding. For example, a machine can diagnose a patient with arterial blood pressure above a set threshold with hypertension. This could replace alarm chains with one alarm that signals the underlying reason for this problem.
The Stanford system isn’t commercially available, but RCA is now widely deployed in health care.
UCSF Medical Center, where Pablo Garcia nearly died, formed a committee in the aftermath to review all of their alerts. “This is painstaking work, the digital equivalent of weeding the lawn.” After two years of combing through them one after the other they have removed less than a third.
New processes including proper skin prep, along with delayed alarms, wider, more individualized parameters, prioritization, robust statistical signal extraction, root cause analysis, and flexible, intuitive electronic medical records software can help limit alarms, and alarm fatigue, making everyone safer and more comfortable.
Do you experience alarm fatigue? How do you cope? Let us know in the comments!
Header by Abby Kahler