Executive Summary

Boston Medical Center (Boston, MA), one of the finalists for the 2013 Health Devices Achievement Award, reduced audible telemetry alarms by 89% during a six-week program.

A task force had observed that alarms for life-threatening conditions frequently had to compete for staff attention with large numbers of clinically insignificant alarms. The team designed a study to test the effect that alarm-feature changes would have on clinicians' response to alarms and on the environment (e.g., noise levels) within a general cardiology medical care area. The most notable finding of the team's initial investigations was the number of warning-level alarms—comparatively low-level alarms that didn't necessarily require an immediate response—and these became the target of the study.

A key remedy the hospital applied was a policy that all audible alarms would become actionable—that is, any time an alarm sounded, immediate action was expected. Consequently, some of the previous "warning" alarms would be converted from an audible tone to a visual message status, while others would be elevated to "crisis" level to prompt an immediate staff response. This latter change required that staff immediately respond to the patient's need if the alarm was clinically significant or, to eliminate future clinically insignificant crisis alarms, take steps such as adjusting the alarm settings to better reflect the baseline heart rate and rhythm for that patient.

In addition, to reduce artifact, patient electrodes were changed every 24 hours, and the default alarm limits for some parameters were slightly widened.

During the six weeks of the study, the number of audible alarms dropped from nearly 90,000 per week to just under 10,000—a decrease of 89%—with no negative impact on patient safety. Significantly, this approach to process control was instituted with no additional technology or capital investment.

 

(Image courtesy of Boston Medical Center.)

Who Should Read This

Table of Contents

Telemetry care areas struggle with how best to use cardiac monitor alarms to alert staff to important abnormal heart rates and rhythms. No clinician wants to miss a significant cardiac event. However, if monitoring is implemented in a way that yields more false (or clinically insignificant) alarms than true ones, a culture can emerge in which alarms aren’t always addressed in a timely manner. For instance, clinicians may be reluctant to interrupt other patient care activities to address an alarm that likely will prove to be insignificant.

The danger, of course, is that a truly significant alarm will go unheeded. At Boston Medical Center (Boston, MA), a multidisciplinary telemetry task force is in place to investigate ways to minimize such risks by improving the overall management of telemetry alarms. A recent six-week trial spearheaded by the task force led to its recommending significant changes in how the facility managed cardiac telemetry patients and how various technologies were used in a general cardiology medical care area.

The idea for the study emerged from the task force’s years of work studying alarm metrics and clinician workflow. The task force observed that alarms for life-threatening conditions frequently had to compete for staff attention with large numbers of clinically insignificant alarms. Not only did these insignificant alarms potentially impede staff from recognizing and responding to more critical alarms, they also contributed to noise in and around patient rooms. Patient surveys had shown that noise negatively affected the patient’s overall perception of the quality of the hospital experience.

The team designed a study to test the effect that alarm-feature changes would have on clinicians’ response to alarms and on the environment (e.g., noise levels) within a general cardiology medical care area. The task force obtained alarm data, extracted by the clinical engineering department, to help better understand the alarms in that care area, and used that data to develop a list of changes to default settings that would safely decrease the number of total alarms while still ensuring that alarms would activate for events that required immediate attention. These changes would be implemented in the care area being studied for a six-week trial period.

Although getting the data in a usable form for this analysis was a “laborious process,” James Piepenbrink, Boston Medical Center’s director of clinical engineering, remarked that having the data really helped the task force get a sense of what was going on. What “jumped off the page,” he noted, was the number of warning-level alarms. These are comparatively low-level alarms that often sounded for nonserious changes in heart rate, for example, and didn’t necessarily require an immediate response. Thus, it was determined that the study would target those alarms. A key principle the hospital applied was that all audible alarms became actionable—that is, any time an alarm sounded, immediate action was expected. Consequently, some of the previous “warning” alarms would be converted from an audible tone to a visual message status, while others would be elevated to “crisis” level to prompt an immediate staff response.

Previously, bradycardia, tachycardia, and high/low heart rate alarms had been set to “warning” status and configured to initiate an audible tone with a self-reset capability. This meant that an alarm would sound as long as the alarm condition existed, but then would reset on its own when the alarm condition was no longer met. Artifact was a common cause of such alarms, and staff would often allow the alarm to resolve on its own. The result, however, was that the alarms would activate over and over again, since the underlying cause of the alarm condition was not addressed. For this study, patient electrodes were changed every 24 hours to help reduce instances of artifact, and the alarms for these conditions were set to "crisis" instead of "warning" status. This latter change required that staff immediately view and act on the alarm each time it sounded. The required actions would be to respond to the patient’s need if the alarm was clinically significant or, to eliminate future clinically insignificant crisis alarms, take steps such as adjusting the alarm settings to better reflect the baseline heart rate and rhythm for that patient. A change to the alarm settings required a second nurse’s endorsement, and the physician order would later be obtained for the change.

The study also involved slightly widening the default alarm limits for some parameters. For example, for this study a low heart rate of <45 bpm (lowered from 50 bpm) was used for bradycardia, and a high heart rate of >130 bpm (raised from 120 bpm) was used for tachycardia. These limits were selected recognizing that many patients would have known transient changes in heart rate during certain times of day. The new alarm limit values were supported by physicians, who judged that the slightly widened limits would not compromise patient safety.

During the six weeks of the study, the number of audible alarms dropped from nearly 90,000 per week to just under 10,000—a decrease of 89%—with no negative impact on patient safety. This change caused many nurses to remark on the quietness of the unit and to note that they could now spend more time caring for patients. Improvements were also seen in patient satisfaction scores in several metrics.

Significantly, this approach to process control was instituted with no additional technology or capital investment. The success of the pilot study led to the implementation of this approach in the facility’s nine other telemetry care areas.

Best Practices

With the Joint Commission establishing alarm management as a new National Patient Safety Goal, many healthcare facilities will need to conduct their own studies to determine which clinical alarms pose a significant risk to patient safety and what strategies can be used to reduce the risks of harm. The initiative described here illustrates many points that such facilities will want to consider. Following are a few of the specific observations shared by the team at Boston Medical Center:

  • Institutions can benefit from looking at their actual alarm data over time, as well as how their nursing staff interacts with alarms on a day-to-day basis.

Much can be learned from visiting a unit and observing what’s happening: How noisy is it? How are nurses responding to various types of alarms? That can help you target where to focus your attention. Relevant data can then be collected and analyzed to help you craft meaningful solutions, such as identifying appropriate changes to alarm levels and parameter limits (e.g., to reduce warning alarms).

Having the data can also ease the process of initiating change. Telling frontline caregivers that they need to do things differently can be a hard conversation to have. But reviewing the data with all stakeholders and including them in the process produces more of a collegial discussion than a confrontational event. As Mr. Piepenbrink observed: “Data is binary. It is not emotional.” Further, he stressed that the process at Boston Medical Center “was very inclusive,” specifying that “we were clear about what we were doing and why, and we shared the results with users.”

  • Alarms with self-reset capabilities may lead to an excess number of audible alarms. Nurses informed members of the task force that they were reluctant to be drawn away from other important patient care activities to address alarms that often resulted from artifact or were clinically insignificant. Rather, nurses expected that the alarm would self-reset if the condition resolved itself or, if the alarm persisted, that a nurse who was not occupied with other patient care activities would respond. However, the task force observed that noise from these alarms continuously sounding in the background could affect staff’s ability to hear other, important alarms.
  • Alarm management can improve not only patient safety, but also patient and staff satisfaction. Although the interventions in the study were not targeted to address patient satisfaction scores, higher scores were nevertheless observed when comparing data from before and after the pilot. Nurses remarked that the reduced alarm load allowed them to spend more time with patients, a factor that may have contributed to the improved patient satisfaction scores. In addition, nurses noted that they "felt less drained at the end of their shift," suggesting that the changes could also positively influence overall staff satisfaction.
  • Another key element of success, according to Mr. Piepenbrink, was that the task force placed a lot of stock in training and supporting users. For example, a team of super-users was created so that, when questions arose, a nurse could go to a peer on the floor who had a really good understanding of the system and what the goals of the pilot study were. This provided a comfort level for the staff.

ECRI Institute has published guidance for improving the management of clinical alarms in its series of articles published in the August, September, and December issues of Health Devices.

Congratulations and thanks to James Piepenbrink for submitting Boston Medical Center’s application for the Health Devices Achievement Award.

Glossary

Bibliography

References

Topics and Metadata

Topics

Alarm Management; Quality Assurance/Risk Management

Caresetting

Hospital Inpatient

Clinical Specialty

Cardiovascular Medicine

Roles

Biomedical/Clinical Engineer; Nurse

Information Type

Guidance

Phase of Diffusion

 

Technology Class

 

Clinical Category

 

UMDNS

SourceBase Supplier

Product Catalog

MeSH

ICD9/ICD10

FDA SPN

SNOMED

HCPCS

Disease/Condition

 

Publication History