Harm From Alarm Fatigue

SPOTLIGHT CASE

Commentary by Michele M. Pelter, RN, PhD, and Barbara J. Drew, RN, PhD

Case Objectives

Case & Commentary—Part 1

A 54-year-old man with hypertension, diabetes, and end-stage renal disease on hemodialysis was admitted to the hospital with chest pain. His initial electrocardiogram (ECG) showed no evidence of significant ischemia, but cardiac biomarkers (troponin T) were slightly positive. He was admitted to the observation unit, placed on a telemetry monitor, and treated as having a non-ST segment elevation myocardial infarction (NSTEMI).

Overnight, the patient's telemetry monitor was constantly alarming with warnings of "low voltage" and "asystole." The bedside nurse initially responded to these alarms, checking on him several times and each time finding him to be well. The resident physician responsible for the patient overnight was also paged about the alarms. He came and checked the patient and the alarms and was not concerned. Both clinicians felt the alarms were misreading the telemetry tracings.

The scenario described in this case is common—skilled and well-intentioned health care providers diligently respond to repeated false alarms. Yet excessive false alarms may lead to unintended harm. This case provides an opportunity to consider the benefits and potential harms associated with the multitude of alarms in the hospital setting.

For many reasons (as in this case example), hospitalized patients are often monitored using telemetry. Unlike bedside ECG monitors in the intensive care unit where data is displayed in the patient's room, telemetry ECG systems transmit the ECG signal wirelessly to a central monitoring station where data for all of the patients is displayed. Some hospitals choose to utilize monitor watchers to identify alarms and notify nurses. Other hospitals use pager systems or enhanced sound systems on the unit to alert nurses to alarms. Since one monitor watcher is responsible for watching as many as 40 patients' data, only one ECG lead is typically displayed for each patient so that all patients' data can fit on one or two display screens.

While a standard diagnostic ECG acquires data from 12 different leads (via 10 electrodes placed on the patient's body), telemetry monitoring systems typically acquire data from fewer leads (via 3–6 electrodes placed on the patient's torso). One reason computer algorithms from telemetry monitoring systems are less diagnostic and less accurate than computer interpretations from the standard 12-lead ECG is that a limited number of leads (typically, 1–2) are used for analysis. If the telemetry algorithm uses just one ECG lead for analysis, this can more easily be misinterpreted, leading to false alarms.

Despite harnessing advanced technology, telemetry monitoring devices often misidentify heart rhythms as asystole. In this case, the providers were correct in concluding that the telemetry monitor device was misreading the patient's heart rhythm because a true asystolic event would have been clinically apparent. The most common cause of false asystole alarms is under-counting of heart rate due to failure of the device to detect low-voltage QRS complexes in the ECG leads used for monitoring. This patient's telemetry device warned of this problem with "low voltage" alarms.

After the nurse responded to these alarms by checking on the patient (multiple times) and contacting the responsible physician, the correct action would have been to search for another ECG monitoring lead with greater QRS voltage. For example, the resident and nurse could have checked the patient's full diagnostic standard 12-lead ECG to determine which of the 12 leads had the greatest QRS voltage, and then changed the telemetry monitor lead accordingly. This may have prevented the repeated alarms that were a consequence of a low-voltage QRS.

In our recent study of alarm accuracy in 461 consecutive patients treated in our 5 adult intensive care units over a 1-month period, we found that low-voltage QRS complexes were a major cause of false cardiac monitor alarms.(1) The Figure shows the standard diagnostic 12-lead ECG of the single outlier patient in our study who contributed 5,725 of the total 12,671 arrhythmia alarms (45.2%) analyzed. Similar to the case described here, under-counting of heart rate due to low-voltage QRS complexes led to repetitive false asystole alarms in our patient.

Case & Commentary—Part 2

The nurse and resident decided to silence all of the telemetry alarms (in this observation unit, there was not continuous or centralized monitoring of telemetry tracings). The patient was not checked for approximately 4 hours.

When the bedside nurse went to perform the patient's morning vital signs, he was found unresponsive and cold with no pulse. A code blue was called but the patient had been dead for some time. The cause of death was unclear, but providers felt the patient likely had a fatal arrhythmia related to his NSTEMI.

Silencing all telemetry alarms in this patient was an error that contributed to this patient's death. This patient was at risk for developing a fatal arrhythmia due to his acute myocardial infarction and co-morbid conditions (diabetes, end-stage renal failure). The arrhythmia would likely have triggered an appropriate alarm had the alarms been functioning, and the patient might have been saved. This adverse event reveals a clear hazard associated with hospital alarms. It also provides an opportunity to consider why such harms exist and what can be done to mitigate them.

What types and numbers of alarms occur with hospital monitor devices and how accurate are they?

Typically, there are three types of alarms generated with hospital monitor devices: arrhythmia alarms that detect a change in cardiac rhythm; parameter violation alarms that detect when a vital sign measurement (heart rate, respiratory rate, blood pressure, SpO2, etc.) exceeds the "too high" or "too low" alarm limit settings; and technical alarms that indicate poor signal quality (e.g., a low battery in a telemetry device, an electrode problem causing artifact, etc.).

Because monitor manufacturers never want to miss an important arrhythmia, alarms are set to "err on the safe side." As a result, the sensitivity for detecting an arrhythmia is close to 100%, but the specificity is low. That is, arrhythmia alarms are programmed to never miss true arrhythmias, but as a consequence they trigger alarms for many tracings that are not true arrhythmias, such as when a low-voltage QRS complex triggers an "asystole" alarm. In our recent analysis of monitor alarms in 77 intensive care unit beds over a 31-day period, there were 381,560 audible monitor alarms, for an average alarm burden of 187 audible alarms/bed/day.(1) Of the 12,671 arrhythmia alarms that were annotated, 88.8% were false alarms and did not signify true arrhythmias.(1)

What is alarm fatigue?

Alarm fatigue occurs when clinicians become desensitized by countless alarms, many of which are false or clinically irrelevant. The development of alarm fatigue is not surprising—in our study, there were nearly 190 audible alarms each day for each patient.(1) If only 10% of these were true alarms, then the nurse would be responding to more than 170 audible false alarms each day, more than 7 per hour. Consequently, rather than signaling that something is wrong, the cacophony becomes "background noise" that clinicians perceive as part of their normal working environment.

Patient safety concerns surrounding excessive alarm burden garnered widespread attention in 2010 after a highly publicized death at a well-known academic medical center.(2) Despite repeated low heart rate alarms before the patient's cardiac arrest, no one working that day recalled hearing the alarms. In the investigation that ensued, the Centers for Medicare & Medicaid Services (CMS) reported that alarm fatigue contributed to the patient's death.(3)

In the present case, clinicians turned off all alarms. Clinicians who find constant audible or textual messages bothersome may silence alarms at the central station without checking the patient or permanently disable them. Warnings have been issued about deaths due to silencing alarms on patient monitoring devices.(4) Moreover, several federal agencies and national organizations have disseminated alerts about alarm fatigue. In 2015, for the fourth consecutive year, ECRI listed alarm fatigue as the number one hazard of health technology.(5) In 2013, The Joint Commission issued an alarm safety alert (6); they established alarm safety as a National Patient Safety Goal in 2014, with further regulations becoming mandatory in 2016.(7)

Why is alarm fatigue dangerous?

The biggest harm that can result from alarm fatigue is that a patient develops a fatal arrhythmia or significant vital sign abnormality that is not noticed by the clinical staff because that patient's heart rhythm monitor has been plagued with false alarms. Imagine a neighbor who has a hair trigger car alarm that goes off all the time. If someone actually breaks into this car, setting off yet another alarm, would anyone be likely to call the police?

A number of different forces result in an excessive number of cardiac monitor alarms. The key contributing factors are (i) alarm settings that are not tailored for the individual patient (i.e., leaving hospital default settings in place even if they don't make sense for an individual patient); (ii) the presence of certain patient conditions such as having low ECG voltage, a pacemaker, or a bundle branch block; and (iii) deficiencies in the computer algorithms present in the devices.

What can be done to combat alarm fatigue?

Many steps can be taken to combat alarm fatigue and ensure that alarms that truly indicate a change in condition are responded to in an appropriate manner. First, devices themselves could be modified to maximize accuracy. One example would be to build in prompts for users. For instance, in patients with persistent atrial fibrillation (an irregular heart rhythm that can trigger telemetry alarms) rather than have the alarm repeatedly triggering in response to the atrial fibrillation, the monitor could generate a prompt, "do you want to continue to hear an atrial fibrillation alarm?" Another suggestion for industry is to create algorithms that analyze all of the available ECG leads, rather than only a select few leads. This could minimize the number of false alarms for asystole, pause, bradycardia, and transient myocardial ischemia. Lastly, algorithms that integrate parameters (i.e., link heart rate and blood pressure) could help determine if alarms are real or false by checking to see if there was any simultaneous physiologic impact. For instance, an algorithm-defined asystole event that was not associated with a simultaneous drop in blood pressure would be re-defined as false and would not trigger an alarm.

In addition, individual nurses and providers at the bedside can take steps to improve the usefulness of alarms. First, nurses and providers can review their hospital alarm default settings to determine whether some audible alarms that do not warrant treatment can be changed to inaudible text message alerts. Furthermore, nurses can tailor alarm settings for individual patients because hospital default settings may not make sense for the individual patient. For example, if the hospital default setting for high heart rate is set at 130, but a certain patient with atrial fibrillation has a heart rate averaging 135, then to avoid incessant alarms the alarm threshold needs to be increased while treatment is underway.

Lastly, institutions can take steps to improve the use of alarms and combat alarm fatigue. Committees charged with addressing alarm management should be formed and include all levels of the organization to ensure recommendations for practice changes can be carried out. We recently conducted a human factors analysis and determined that clinicians (nurses, physicians, and monitor watchers) found it difficult to respond to alarms or adjust alarm settings when working at the central monitoring station.(8) Importantly, most participants reported they had not had training on how to use the monitoring equipment. This highlights the need for education and training of all staff that interact with monitoring devices. Training should be provided upon employment and include periodic competency assessments.

Take-Home Points

Alarms should never be completely silenced; rather, clinical staff should problem-solve why an alarm condition is occurring and work to resolve it.
Cardiac monitor devices have a high sensitivity for detecting arrhythmias and vital sign changes, but have a low specificity; therefore, they generate a high number of false positive alarms.
Clinicians should learn how to tailor alarm thresholds to an individual patient to avoid an excessive number of alarms and alarm fatigue.
Alarm safety is a National Patient Safety Goal, highlighting the importance of developing institutional policies and practice standards to improve awareness of this problem and designing interventions to reduce the burden to clinicians, while ensuring patient safety.

Michele M. Pelter, RN, PhD
Assistant Professor
Director, ECG Monitoring Research Lab
Department of Physiological Nursing
University of California, San Francisco (UCSF)

Barbara J. Drew, RN, PhD
Emeritus Professor
Founder and Former Director, ECG Monitoring Research Lab
Department of Physiological Nursing
University of California, San Francisco (UCSF)

Faculty Disclosure: Dr. Drew has received research funding from GE Healthcare. The commentary does not include information regarding investigational or off-label use of products or devices. All conflicts of interest have been resolved in accordance with the ACCME Updated Standards for commercial support.

References

1. Drew BJ, Harris P, Z?gre-Hemsey JK, et al. Insights into the problem of alarm fatigue with physiologic monitor devices: a comprehensive observational study of consecutive intensive care unit patients. PLoS One. 2014;9:e110274. [go to PubMed]

2. Kowalczyk L. MGH death spurs review of patient monitors. Boston Globe. February 21, 2010. [Available at]

3. Kowalzyk L. 'Alarm fatigue' linked to patient's death. April 3, 2010. [Available at]

4. A siren call to action: priority issues from the medical device alarms summit. Arlington, VA: Association for the Advancement of Medical Instrumentation; 2011. [Available at]

5. ECRI Institute Announces Top 10 Health Technology Hazards for 2015. Plymouth Meeting, PA: ECRI Institute; November 25, 2014. [Available at]

6. The Joint Commission announces 2014 National Patient Safety Goal. Oakbrook Terrace, IL: The Joint Commission; 2014. [Available at]

7. Medical device alarm safety in hospitals. Sentinel Event Alert. April 8, 2013;(50):1-3. [Available at]

8. Fidler R, Bond R, Finlay D, et al. Human factors approach to evaluate the user interface of physiologic monitoring. J Electrocardiol. 2015;48:982-987. [go to PubMed]

Figure

Figure. Standard 12-lead ECG in the patient who generated more (mostly false) arrhythmia alarms than any other patient in our study (1). Low voltage QRS complexes are present in the seven leads available for monitoring (I, II, III, aVR, aVL, aVF, and V1). If the nurse or physician had recognized how much greater the QRS voltage was in leads V3 and V4, then the chest electrode could have been moved to the V3 or V4 position and the source of alarm fatigue (frequent false bradycardia type alarms) would likely have been eliminated. Reprinted with permission from (1).