Song of Silence: Making Alarms’ Din Manageable

Bleeps, chirps and a cacophony of warning alarms assault nurses and physicians in today’s hospital units. But it doesn’t have to be that way. Several hospitals have taken steps to reduce this auditory overload in an effort to help caregivers discern between important notifications and mere noise.

“There are so many different devices that produce alarms,” according to Marjorie Funk, RN, PhD, a professor at the Yale University School of Nursing. “Alarms go off not only when a patient is in trouble, but also if a device’s batteries are running low or something else is wrong that needs attending to.”  

The increasing use of alarms has contributed to the problem. The Oklahoma Heart Hospital, Christiana Care Health System and Johns Hopkins Hospital are showing that effectively managing alarms means cutting some out. 

Teaming up with IT

“One of the problems with a past monitoring system was that nurses would get false alarms for every little crimped cable or moving finger,” says Steve Miller, the CIO of Oklahoma Heart Hospital in Oklahoma City. “They couldn’t tell the difference between a life-threatening alarm and a nuisance alarm.”

When clinicians and the biomedical engineering team at Oklahoma Heart came to the IT department with the issue, Miller shopped around for an appropriate tool to improve alarm management. He settled on an unobtrusive information exchange client that could read data from patient monitors and transmit that information wirelessly to nurses’ mobile devices.

Oklahoma Heart customized alarm parameters so that only the most crucial would reach nurses. The information exchange client is able to read every data field in a patient monitor, interpret the data according to the customized alarm parameters and transmit notifications of serious events along with patient information and room to nurses. Using the system, the hospital has reduced false alarms by setting time delays for certain events. For instance, an alarm caused by an SpO2 sensor will not reach a nurse until it has sounded for 30 seconds, which eliminates alarms caused by crimped cables or moving patients.

Now, nurses know that they need to respond when they receive an alarm. “Before we implemented this, nurses might not have been too responsive to alarms because they knew that maybe 80 percent of them were nuisances,” Miller says. The initiative has strong support from the chief nursing officer, who has encouraged her staff to respond to alarms transmitted through the new monitoring system.

There are also multiple redundancies installed in the system to ensure alarms receive attention. If a nurse does not respond to a notification by pressing the acknowledge button on her device, it is sent to another nurse after a set amount of time. Additionally, test messages are sent to the mobile devices twice a day to ensure they are receiving notifications.

Technology for technology’s sake is worthless; technology to improve clinical workflow is the goal. While Miller isn’t sure if the customized alarm parameters or the mobile monitoring system are improving patient outcomes, he believes they improve care by letting nurses know what is going on with and giving them more time with their patients. “We think clinical outcomes are better because happy nurses provide better care and that’s what we want: happy nurses and happy patients.”

Unique solutions for unique needs

The Christiana Care Health System in Wilmington, Del., embarked on an alarm redesign in the late 1990s. Christiana standardized alarms across all units, turning off many unnecessary alarms and setting the same defaults for all patients. Standardizing alarms at Christiana made sense because it allowed nurses to move between units and still know how the alarms should be set, according to Maureen Seckel, RN, MSN, a clinical nurse specialist.

Turning off unnecessary devices reduced the number of alarms at Christiana, according to Seckel, but a more recent initiative has tweaked the hospital’s alarm management strategy even further. While Oklahoma Heart went with an intuitive mobile monitoring system to avoid the space and financial requirements of central monitoring, nurses at Christiana monitor clinical alarms from a post across the street from the main facility.

A central monitoring room now receives alarms from all units at Christiana except for those originating in intensive care. Alarm silence has been achieved. “There are no alarm sounds on nursing units anymore,” says Anita K. Witzke, MSN, manager of flex monitoring. “All the noise is on our end.”

Personnel receiving alarms in the central monitoring room have all of a patient’s data available to them and can determine whether or not a patient requires attention. All problems requiring attention are communicated to nursing units via telephone, but potentially lethal problems are communicated over a reserved network. The “lethal” phone located on each unit demands an immediate response when it rings, which has not been a problem. Because it rings so infrequently, sounds so different and indicates a crisis, the “lethal” phone always gets attention.  “When that phone rings, there is none of that alarm amnesia we saw in the past,” Witzke says.

Christiana investigated mobile monitoring as an alternative to a central monitoring room, but decided that the latter route better suited the hospital’s patient and provider needs. “We didn’t see that mobile monitoring did anything to prevent oversaturating bedside providers with information,” Seckel says. “We wanted to provide an alarm filter and none of the computer systems do that quite as well as a human being.”

The clinical alarm management strategy at Christiana may differ significantly from Oklahoma Heart, but that doesn’t necessarily mean that either healthcare organization has implemented ineffective strategies; it more likely indicates that the circumstances affecting individual hospitals will inspire them to take different alarm management routes. “There are no simple solutions and potentially successful solutions vary from one environment to another,” Funk notes. “Even within the same hospital, different units have different issues with alarms and, therefore, different solutions would be appropriate.”

Making a determination with data

At Johns Hopkins Hospital in Baltimore, Maria Cvach, RN, MS, and her co-workers scoured reams of unorganized data captured by clinical information systems to determine that a vast majority of alarms sounded at the 560-bed facility were categorized as technical or advisory rather than warning or crisis.  

In the surgical intensive care unit alone, an average of 317 alarms were sounding per patient per day, of which 245 were advisory alarms. Cvach, an assistant director of nursing, estimated that 80 to 90 percent of these were false and, therefore, liable to distract nurses from more serious matters requiring attention. Designed for sensitivity rather than patient specificity, these types of alarms were sounding too frequently and often alerting nurses to minor equipment errors and clinically insignificant events.

Unimportant alarms had the potential to divert nurses’ attention from serious incidents to relatively minor ones and hampered their ability to make snap judgments about the need to respond to an alarm, according to Cvach. “That’s why it was so important for us to get those alarms down.”

Having learned the exact number and nature of alarms sounding at the hospital, Cvach went with numbers in hand to the clinical leaders on each unit to discuss how the default settings on patient monitors could be adjusted to reduce clinically insignificant and false alarms. In the surgical intensive care unit this meant adjusting the alarm parameters of SpO2 sensors and arterial lines, which caused approximately 70 percent of total alarms on the floor, and shutting off alarms indicating events that didn’t require treatment.

After the intervention, the average number of alarms sounding per patient per day in the unit decreased to 212. Without data to back it up, Cvach is uncertain whether or not the alarm system redesign has had a positive effect on response time or patient outcomes. Nonetheless, she is optimistic. “I’ve observed that it does seem to be calmer and it seems to be working,” she says. “Noise competes for people’s attention, but if you can give them reliable noise, they will respond.” 

Moving alarm management forward

Despite ongoing efforts and recent success, Johns Hopkins’ clinical team can’t address all the issues with clinical alarms on their own. When Cvach looked at the alarms occurring in the hospital’s cardiac surgery intensive care unit, she noted that 86 percent were system warnings coming from SpO2 sensors. That number barely budged after adjusting alarm settings, indicating deficiencies in the equipment, which Johns Hopkins has since brought up with its vendors. Making alarm management easier will take effort from many of healthcare’s stakeholders, not just providers.

Additionally, hospitals shouldn’t use the technology solely because it’s there. “Don’t even put certain patients on monitors to begin with,” Funk says. “That’s almost a bigger problem. We have the capability to put everybody on a monitor, but for so many patients there’s no reason for it.”


Keeping EHR alert fatigue at bay

Devices in hospitals that warn nurses and physicians that a patient is in need of care are not the only technologies that contribute to alert overload. EHRs also can produce alert fatigue if they swamp physicians with prompts following tasks such as prescribing a medication that interferes with another drug. Rather than mute the alerts, one research team found a way to bypass frazzled physicians treating patients on the anticoagulant warfarin by sending alerts to a patient-management team instead.

Kate E. Kaplan, MD, MPH, of Harvard Vanguard Medical Associates in Newton, Mass., and colleagues used their multispecialty medical group’s EHR system to test whether a stealth alert system would improve anticoagulation monitoring in patients whose physicians prescribed a medication that potentially interfered with warfarin (J Gen Intern Med, online July). Rather than alert the prescribing physician, the EHR automatically sent an alert to a centralized Anticoagulation Management Service (AMS), which consisted of a team of nurses, pharmacists and physicians with anticoagulation expertise responsible for anticoagulation care.

The test included a 12-month pre-intervention period, a two-month implementation and a six-month post-intervention period. The primary outcome measure was the proportion of patients who completed international normalized ratio (INR) monitoring within five days of having a new medication prescribed. The researchers reported that 34 percent of patients completed INR monitoring within five days in the pre-intervention group, compared with 39 percent in the post-intervention group.

The benefit was modest, they acknowledged, but the stealth alert approach succeeded in reducing physicians’ alert fatigue and fostered sharing of responsibilities between the primary care and disease management groups. It helped the AMS provide timely follow-up if clinically indicated. While their medical group uses Epic, they said the system could be replicated in Epic or any EHR that meets Certification Commission for Health IT standards.

The study did not measure outcomes or the time in therapeutic range, but they pointed out that increased INR monitoring would likely reduce the risk of over-anticoagulation or under-anticoagulation.