Background
The literature reports that up to 70% of patients with the top 10 diagnoses admitted through the emergency department may meet criteria for inpatient cardiac monitoring based on the Practice Standards for Electrocardiographic Monitoring in Hospital Settings: An American Heart Association Scientific Statement.1 With the dissemination of guidelines, there has been an exponential growth in the number of cardiac monitors that are connected to hospitalized patients either directly (hardwire monitoring) or wirelessly (telemetry) in all clinical settings, but particularly on general medical-surgical units.2 There is also a sense among physicians that telemetry allows for closer monitoring of patients on busy general medical-surgical floors, which contributes to the number of patients being monitored in these settings.3
Clinical Alarm Fatigue
General medical-surgical units struggle with how best to use cardiac monitor alarms to alert nursing staff to important abnormal heart rates (HRs) and rhythms. Audible alarms can produce unintended consequences that undermine quality and safety when they occur unnecessarily.4-6 If alarms are more often false than true, a culture emerges on the unit in that staff may delay response to alarms, especially when staff are engaged in other patient care activities, and important critical alarms may be missed.7-9 For patients who hear alarms and are unaware of their source, anxiety may be heightened and the healing environment may be disrupted. Patient survey questions regarding noise in and around patient rooms consistently receive the lowest ratings, with noise negatively impacting the patient's overall perception of the quality of the hospital experience.10,11
The Emergency Care Research Institute (ECRI) identifies alarm hazards and clinical alarm fatigue as the number 1 technology hazard of 2013.12 In addition to the ECRI reports, there have been several reports throughout the country of failure to rescue patients having serious arrhythmic events while being continuously monitored.9,13-16 A recent review suggests that these incidents occurred for several reasons, including (1) lapses in monitoring with patients found off monitor, (2) life-threatening critical alarms not responded to by staff, and (3) missed warning alarms that signaled clinical deterioration.17 "Clinical alarm fatigue" has become the phrase used to describe the phenomenon of medical providers becoming so desensitized to the constant noise emitted from monitors that they fail to notice alarms or react to alarms in a timely manner.16,18 The ECRI, Joint Commission, and Food and Drug Administration have reported numerous alarm-related incidents from alarms turned off, ignored, or simply missed.4-9,14,19 The Food and Drug Administration database for cardiac monitoring devices for 2011 reported 565 incidents of patient harm due to alarm fatigue, 35 of which resulted in patient death.8 Each of the above agencies believes that the actual incidence of patient harm is significantly underreported.8,9 Is the answer to clinical alarm fatigue to silence more of the audible cardiac alarms in an effort to improve staff sensitivity to critical events? Alternatively, is the answer to use additional technology, such as more viewing stations and alarm alert paging notification systems, to capture the attention of the clinician? Or is there another approach that alters how clinicians interact with monitors that can improve patient safety while minimizing the noise that has contributed to clinical alarm fatigue? The alarm changes that were made at Boston Medical Center (BMC) and are described in this article demonstrate that there is a significant opportunity to change how clinicians interact with cardiac monitoring alarms on a daily basis and that this interaction decreases clinical alarm fatigue.
We present a novel approach for recognizing and safely decreasing audible cardiac monitor alarm volume while ensuring that all important monitor alarms are heard. This information is important to all clinicians in environments with cardiac monitors and telemetry but is of particular value to nursing staff on general medical-surgical units who often rely on the manufacturer's arrhythmia, parameter limits, and system defaults for their alarm settings and who are typically not present at the patient's bedside when alarms occur.20
Description of Telemetry Monitoring Practices at Boston Medical Center
In 2008, BMC standardized its cardiac telemetry monitoring equipment for all adult inpatient clinical units. Uniform equipment provided the opportunity to standardize arrhythmia alarm levels and HR parameter limit defaults across all units. Having alarm defaults that varied from one patient care unit to another increased the potential for errors to occur when nursing staff floated between units and were unaware of unit-specific alarms. Furthermore, standardization of alarms on general medical-surgical units was felt to be particularly important because nursing staff on these units are responsible for the care of multiple patients and viewing of the patient's actual HR and rhythm is available at only 1 location on the unit at a central telemetry console.20
The monitoring system purchased in 2008 provides 4 levels of patient status arrhythmia and parameter limit alarms: message, advisory, warning, and crisis. These alarms occur with arrhythmia or parameter limit violations. In addition, there are 3 levels of system status alarms: message, advisory, and warning. These system alarms trigger when there is a mechanical fault of the monitoring system, such as when leads are disconnected or when the monitor is unable to determine the patient's rhythm. The types of alarm, whether there is a visual alert, whether it is audibly signaled, whether the alarm resets when the alarm condition is no longer met, and whether the alarm is stored in alarm history, are seen in Table 1.
With the purchase of the new cardiac telemetry monitoring equipment, BMC senior leadership convened a multidisciplinary Telemetry Task Force (TTF) to evaluate how monitors were being used in various clinical areas, identify ways to improve the management and utilization of this valuable resource, and develop consensus for a common approach to cardiac monitoring. Membership on the TTF included BMC's chief medical officer; a cardiologist; physicians from critical care, cardiology, medicine, and surgery; the director of clinical engineering; the clinical service manager nurse practitioner for cardiology; nursing directors from critical care and medical-surgical nursing; clinical instructors responsible for telemetry training; and a quality and patient safety specialist.
The TTF's initial work in 2008 focused on the standardization of default alarms for both arrhythmias and parameter limits for all adult inpatient units (Table 2). In the fall of 2011, the TTF was reconvened to explore the issue of clinical alarm fatigue after reports of sentinel events at other institutions. Discussions held with the nursing staff at that time highlighted their perception of an excessive burden of audible alarms, many of which did not necessitate a clinical response.
To reduce the number of clinically insignificant audible alarms, the TTF worked with clinicians to determine what alarms did and did not require an action by a clinician. From these discussions, the decision was made by the TTF to move R-on-T and Couplet violations from a warning alarm to a message alarm, with visual display only. This was in response to clinician input that R-on-T and Couplet violations occurred frequently, rarely impacted clinical management, created unnecessary noise, and frequently bumped out other clinically significant alarms from the patient's alarm history. In contrast, Pause violation was elevated from a warning alarm to a crisis alarm. As a warning alarm, the Pause violation was able to self-reset once the pause resolved, without a mandate that the event be viewed in real time. This reset resulted in a potential delay in detection of a clinically significant pause until later review of alarm history. Table 2 compares the 2011 default parameters for patient status arrhythmia alarms, patient status parameter limits, and system status alarms with the previous 2008 defaults.
Because only RNs can silence alarms at BMC, the TTF felt that direct observation of nursing staff response to monitor alarms could prove fruitful in further identifying opportunities for better managing alarms that contributed to alarm fatigue. Serial observations by TTF members identified that nursing staff frequently did not respond to the patient status arrhythmia and parameter limit warning alarms, which would self-reset when the patient no longer met the criteria for the alarm trigger. Nursing staff reported that these alarms were often artifact or brief violations that then self-reset and that answering these alarms pulled them away from other important patient care activities. Because these warning alarms were not viewed in real time, nursing staff missed the opportunity to adjust the monitor settings and thereby both improve alarm accuracy and reduce repetitive warning alarms.
Members of the TTF also observed that the continuous sounding of patient status arrhythmia and parameter limit HR warning affected the staff's ability to hear other important alarms, such as those associated with lead failure, loss of telemetry, arrhythmia suspension, blood pressure or oxygen saturation violations, patient call lights, and even crisis alarms. The patient status arrhythmia and patient status HR parameter limit warning alarms were felt by staff to be the most significant contributors to the excess noise level and clinical alarm fatigue on the patient care units. Extraction of alarm data from the telemetry system of a medical cardiology unit confirmed that the patient status arrhythmia and parameter limit HR warning alarms were the largest component of audible alarm noise (Table 3).
Methods
Proposal for an Alarm Management Quality Improvement Project
An Alarm Management Quality Improvement (QI) Study was proposed by the TTF to trial the elimination of the patient status arrhythmia and patient status parameter HR limit warning alarms on this medical cardiology unit. The TTF proposed raising the patient status arrhythmia alarms for accelerated ventricular rhythm, bradycardia, and tachycardia and the patient status parameter HR limit alarms for both high and low HR from a warning alarm to a crisis alarm, thus ensuring that nursing staff viewed these alarms as they occurred. Mandating nursing staff review of these violations at the time of their occurrence was believed to have the potential to significantly improve patient safety. Placing bradycardia and tachycardia at a crisis alarm was consistent with advanced cardiac life support guidelines to assess for symptomatic bradycardia or tachycardia whenever these rhythms occur.21
For the QI study, HR default limits were set to a lower HR limit of less than 45 bpm and an upper HR limit of greater than 130 bpm. These limits were selected with the recognition that nocturnal HRs often drop to the upper 40s and that many deconditioned patients may have HRs in the low 100s during daytime activity. Physicians supported these parameter changes, as the slightly widened limits were not felt to compromise patient safety; in fact, they believed raising HR limit violations to a crisis alarm improved patient safety because these violations would now be viewed in real time by the nursing staff.
The TTF also recommended adding an audible alert to the atrial fibrillation (AF) alarm raising it from a patient status arrhythmia message alarm to a patient status arrhythmia advisory alarm. Although AF is a common rhythm disturbance, its assignment as a message alarm increased the risk that episodes of paroxysmal AF would not be identified unless someone had witnessed the actual episode, an upper HR violation occurred, or the full disclosure function of the monitor was reviewed. Moving AF to a patient status arrhythmia advisory alarm would audibly signal its occurrence and record the episode in the alarm history. Embedding a process within BMC's adult cardiac monitoring order set where nursing staff could easily reassign AF to a nonaudible message alarm for patients in AF (either chronic or paroxysmal) allowed staff to appropriately control this alarm.
For the QI study, all patient status arrhythmia and patient status parameter limit alarms were assigned as either crisis or advisory alarms; warning alarms were eliminated.
* A crisis alarm would require the alarm be immediately viewed by nursing staff, and 1 of 2 actions are taken: either response to the patient for a clinically significant alarm or adjustment of the patient's alarm settings to better reflect the patient's baseline HR and rhythm to eliminate the false crisis alarm. The pilot was structured so that a change to bradycardia, tachycardia, and HR limit default settings could be made with a second registered nurse's endorsement, with a physician order later obtained for that default change, allowing nursing staff to control the alarm in real time.
* The advisory alarm, of which there was only 1, would signal the occurrence of AF requiring nursing staff to assess the patient, notify the physician, and reassign AF either transiently or permanently to a message alarm status until sinus rhythm was reestablished.
The patient status arrhythmia, patient status parameter limits, and system status alarms before and for the Alarm Management QI Study are contrasted in Table 4.
Implementing the Alarm Management Quality Improvement Project
A 6-week pilot was initially proposed for the Alarm Management QI Study. To be successful, it was imperative to have the full engagement, support, and commitment of senior leadership, physician colleagues, clinical engineering, and information technology services and, perhaps most importantly, the engagement of the nursing staff that would be implementing the pilot. After presentations of the proposed Alarm Management QI Study by TTF members, BMC's senior leadership and Quality and Patient Safety Council fully endorsed the pilot. Informational meetings and intensive education regarding the Alarm Management QI Study design and proposed alarm management changes were held for the nursing and physician staff on the pilot unit.
Adult cardiac monitoring order sets were redesigned for the pilot unit incorporating the new HR defaults. In addition, order prompts were inserted to limit the nuisance alarms from permanent or paroxysmal AF by allowing nursing staff to change the default alarm levels for these arrhythmias. The adult cardiac monitoring order that would be placed for a patient in whom paroxysmal AF is anticipated is demonstrated in Figure 1.
A prompt was also included for patients who were known to have short runs of nonsustained ventricular tachycardia, as might occur in the presence of chronic heart failure. This prompt gave physicians the option of reassigning "VT >2" from a crisis alarm to a message alarm, thus eliminating the audible alert from violations that were believed to be clinically insignificant in this patient subset. Importantly Vent Tach (VT >6 beats) remained a crisis alarm.
On the "go-live" date, clinical engineering adjusted the default values at the central monitoring station and on all bedside monitors so that all equipment defaults were synchronized. Throughout the pilot, clinical engineering staff were present on the pilot unit and interacted on a daily basis with nursing staff regarding the alarm changes. The cardiology clinical service manager nurse practitioner worked closely with physician and nursing colleagues to ensure that adult cardiac monitoring orders accurately reflected the HR limit defaults and alarms for the pilot unit and to troubleshoot alarm settings during implementation of the pilot. Admitting clinicians used only the adult cardiac monitoring telemetry alarm pilot rollout orders for new admissions to pilot unit. The nurse manager, clinical educator, and nursing director played pivotal roles in reinforcing staff accountability for responding to the alarms and elicited staff feedback regarding the alarm changes.
Findings
The discrete outcome measures that were examined to assess the effectiveness of the alarm changes were patient safety incident reports, rapid response team (RRT) activations, the incidence of code blues, volume of audible alarms, distribution of alarm types, noise decibel levels, nursing staff perception of noise levels, nursing staff satisfaction, and patient satisfaction. Two-sample t tests were used using an [alpha] level of .05 to compare the mean total weekly audible alarms as well as the mean weekly combined bradycardia, tachycardia, and HR limits alarms before the implementation of the Alarm Management QI Study to the mean combined weekly alarms after implementation.
Patient Safety Incident Reports, Rapid Response Activations, and Incidence of Code Blues
Review of incident reports placed for the pilot unit in the year preceding implementation of Alarm Management QI Study revealed 1 incident report related to cardiac monitor alarms. This incident report identified warning alarms missed at the time of their occurrence and later found in review of alarm history. Incident reports reviewed on a daily basis during the pilot and over the subsequent 6 months revealed no incident reports involving cardiac alarms on the pilot unit. Review of RRT activations and incidence of code blue on the pilot unit in the 6 months preceding and after implementation of the alarm changes demonstrated that the number of RRT activations remained constant at 11, whereas the incidence of code blues decreased by 50%, from 6 to 3.
Audible Alarms
Comparison of alarm data in the 2-week period before and after implementation of the pilot demonstrated an 89% reduction in the number of total audible alarms per week on the pilot unit (t = 8.84; P < .0001). This reflects a reduction in the average number of alarms from 87 823 to 9967 per week on the 24-bed pilot unit. (Figure 2) The largest contribution to the decrease in alarm frequency came from a 93% reduction in the combined bradycardia, tachycardia, and HR parameter limit alarms, reflecting a reduction from 62 793 to 3970 alarms per week (t = 6.34; P < .0001) (Figure 3).
Distribution of Alarm Types
Comparison of extracted alarm data before and during the pilot program demonstrated that all-cause patient status arrhythmia alarms decreased by 91%, from 52 368 to 4881 alarms per week; patient status parameter limit alarms decreased by 94%, from 31 397 to 1985 per week; and system status alarms decreased by 36%, from 3012 to 1942 alarms per week, on the pilot unit (Figure 4).
Decibel Level Values Before and After the Pilot
Decibel levels on the pilot unit were assessed before and after the pilot by 5 separate readings in each time period. Before the pilot, the decibel level on the pilot unit ranged from 54 to 90 dB. After the pilot, the range of decibel levels narrowed with a similar minimum value of 60 dB, with a maximum value of 72 dB.
Nursing Staff Perception of Noise
Nursing staff were anonymously surveyed both before and after the pilot regarding the noise level of the unit. Three choices were given for noise level: acceptable, louder than I would have liked, and unacceptable. The percentage of respondents who assessed the noise level as "acceptable" increased from 0% before the pilot to 64% at the end of the pilot.
Immediately after instituting the pilot program, the nursing staff on the pilot unit noted a significant reduction in alarm-related noise, especially when compared with other units that used standard manufacturer default alarm parameters. One nurse commented, "This unit is so much quieter than other units. As a float nurse, I want to come here." Night staff reported to the nurse manager that it was so quiet overnight that they "kept going back to the central station to reassure themselves that the central station was working."
Nursing Staff Satisfaction
During the pilot, nurses were asked to share their perceptions of how the alarm changes had effected their work environment. Their comments included the following:
* "The monitor alarms were just an irritant; they no longer seem that way."
* "Not only are the alarms decreased but even the call lights go off less often."
* "I feel so much less drained going home at the end of my shift."
* "I can spend more time on patient care instead of answering meaningless alarms."
Patient Satisfaction Scores
Even without targeted interventions to improve patient satisfaction scores on the pilot unit, there was improvement in Press Ganey percentile ranks in several metrics when comparing the 7 months before the pilot with the 7 months after the pilot: Nurse domain rank increased by 15, noise in and around the room rank increased by 12, promptness to call lights rank increased by 39, personal issues domain rank increased by 30, and overall assessment rank increased by 31 (Figure 5).
Discussion
This Alarm Management QI Study differs from others reported in the literature by its focus on general medical-surgical units; other initiatives have proposed alarm strategies for intensive care and progressive care units.22 The design of this Alarm Management QI Study resulted in a significant reduction in audible alarms as well as ambient noise on the pilot unit without an adverse effect on patient safety as validated by review of incident reports, RRT activations, and code blue incidences for the pilot unit. The success of this Alarm Management QI Study was the result of a multidisciplinary approach, using both internal alarm data as well as observations of nursing staff interactions with the cardiac monitoring system, to craft meaningful alarm changes. The nursing staff noted both an immediate decrease in the number of audible alarms and that the remaining alarms were reflective of issues that required their response. As a result, the nurses became strong advocates of the pilot project, resulting in sustained practice change and improved management of cardiac monitor alarms. This type of engagement of the nursing staff is critical in creating the culture change needed to better manage alarms and minimize alarm fatigue.7,22,23 The 93% decrease in the average weekly combined bradycardia, tachycardia, and HR limit alarms in the pilot came not simply from widening the alarm limits but, more importantly, from empowering nursing staff to alter the default settings when a crisis alarm was not indicative of a clinical event requiring intervention but, rather, an alert signaling that the alarm setting needed further modification for the specific patient.16 The patient status arrhythmia and patient parameter limit alarms in the pilot decreased as we had anticipated; however, we had not anticipated the more than one-third decrease in system status alarms. We believe this reduction in system status alarms occurred as a result of improved alarm recognition in the setting of reduced overall alarm noise, resulting in earlier correction of mechanical monitoring faults such as loss of telemetry or arrhythmia suspend due to excessive artifact, again improving patient safety. In addition, we used the pilot to reinforce daily electrode changes to ensure proper signal acquisition and reduce artifact; this likely also contributed to the reduction in system status alarms.23
In conjunction with the reduction in the total number of audible alarms, the maximal decibel values recorded on the pilot unit decreased. We believe that the overall reduction in alarms and the decrease in decibel level on the pilot unit contributed to the 64% improvement in number of staff rating the noise level on the unit as "acceptable" and played a role in why the nursing staff reported feeling less fatigued at the end of their shift as the noise level on the pilot unit fell. This is consistent with previous observation that nurses and other healthcare staff are affected by high noise levels and report increased stress and exhaustion in noisy environments.11 Given the staff's positive comments regarding the pilot and their request not to reinstate previous defaults, we continued the alarm changes on the pilot unit while evaluating the QI study pilot metrics.
Patient satisfaction scores increased on the pilot unit, supporting observations in the literature that noise reduction on patient care units may improve patients' overall perception of the quality of their hospital experience.10 We suspect that as staff spent less time responding to alarms, they spent more time in direct patient care activities and in interactions with patients and their families, thereby positively influencing patient satisfaction scores. Nursing staff specifically commented about this issue, remarking that "even the call lights go off less often because we're with our patients and not at the central monitors all day." It is possible that clinical alarm fatigue may have as much of a negative impact on patient satisfaction as it does patient safety. Research addressing this question is absent in the literature, and the opportunity exists for BMC and others to explore the impact of clinical alarms on patient satisfaction.
In June 2013, the Joint Commission announced the creation of a new national patient safety goal (NPSG) focused on clinical alarm safety. This NPSG calls on each hospital to understand its own situation and to develop a systematic, coordinated approach to alarms.24,25 This article details a methodology that other institutions can use to understand and improve management of cardiac monitor alarms without requiring additional resources or technology. Institutions should
* Establish a broad-base multidisciplinary alarm work group
* Understand their current manufacturer alarm defaults
* Extract and evaluate their alarm data
* Observe staff response to alarms, looking for the barriers to timely response
* Identify with clinician stakeholders clinically insignificant alarms
* Remove audible notification for clinically insignificant alarms
* Choose an alarm setting that requires staff response for all clinically significant alarms
* Standardize alarm defaults across patient care units wherever possible
* Empower nursing staff to eliminate false alarms, appropriately adjusting alarm in real time after validation with second registered nurse
Frontline staff can design innovative sustainable solutions for lessening clinical alarm fatigue in all institutions if presented with this challenge and given appropriate resources, as BMC's experience demonstrates. In recognition of the success of the Alarm Management QI Study, the TTF was awarded a $25 000 BMC Patient Safety Grant to extend this QI initiative to all BMC general medical-surgical telemetry units.
What's New and Important
* Review of actual alarm data, as well as observations regarding how nursing staff interact with cardiac monitor alarms, is necessary to craft meaningful quality alarm initiatives for decreasing the burden of audible alarms and clinical alarm fatigue.
* Audible alarms capable of self-resetting once alarm conditions are no longer met contribute significantly to excess audible alarms and the phenomena of clinical alarm fatigue.
* Improved management of audible cardiac monitor alarms not only improves patient safety and staff satisfaction but also may positively impact patient satisfaction scores.
Limitations
This QI study was conducted on a general cardiology medical floor. Its application on a general surgical or mixed general medical-surgical unit may yield different results. The study was performed using equipment from a single manufacturer. The ability to modify default settings with the equipment of other manufacturers was not evaluated and may impact the alarm parameter changes that can be made. Finally, the QI study required the extraction of alarm data from the monitoring system. Such data have traditionally been in the hands of the manufacturer and not easily obtained by institutions. Extracting alarm data requires a collaborative undertaking with the manufacturer, is labor intensive, and may be difficult to obtain.
Conclusion
This Alarm Management QI Study contributes to existing literature showing how audible alarm reduction can be achieved on general medical-surgical units, a setting where cardiac monitoring use has significantly grown.26 It demonstrates that by the elimination of self-resetting alarms, the volume of audible alarms contributing to clinical alarm fatigue can be significantly reduced without requiring additional resources or technology. Finally, it proposes a methodology that other institutions can use to address the new NPSG proposed by the Joint Commission for 2014 for developing a systematic coordinated approach to alarms.24,25
Acknowledgments
This Alarm Management QI Project would not have succeeded without senior leadership endorsement, nursing leadership's engagement, medical staff support, support of clinical engineering staff, information technology support, and a nursing staff so very willing to embrace change to improve patient safety. Together, they recognized that making all alarms meaningful and actionable improves patient safety and minimize clinical alarm fatigue.
Special thanks to Allison Marshall, RN, BSN, MPH, JD, CPHRM, patient safety risk specialist, for her support of telemetry patient safety initiative at BMC.
Finally, special thanks to Lana Kwong MPH, Analytics and Public Reporting, for graphics and statistical interpretation.
REFERENCES