A Staggering Number of Alarms
Over the total monitoring time of 48,173 hours, a total of 2,558,760 unique audible and inaudible (visual text message) alarms occurred in these ICUs. Of this total, 1,154,201 were arrhythmia alarms, 612,927 were vital sign parameter (too low/too high) alarms, and 791,632 were technical alarms. The default settings in this hospital system restricted audible tone alarms to those that are deemed to be clinically important. Despite this restriction, a total of 381,560 audible alarms occurred, for an average alarm burden of 187 audible alarms per bed per day. During the study period, 17 patients (3.7%) experienced a cardiac or respiratory arrest.
The most frequently occurring alarm was for premature ventricular contractions (PVCs), with a total of 854,901 alarms during the 31-day period. According to current guidelines, however, PVCs are not actionable alarms. The next most frequent category of alarms was technical alarms, most of which were inaudible text message alarms such as artifact or lead fail. Atrial fibrillation was the second most frequently occurring arrhythmia alarm, and these occurred repetitively in patients with persistent atrial fibrillation.
A chief concern of ICU nurses is not missing a "treatable" ventricular arrhythmia. According to practice guidelines, the most common ventricular arrhythmias (accelerated ventricular rhythm and nonsustained ventricular tachycardia lasting <30 seconds) do not warrant antiarrhythmic therapy in the hospital setting. Therefore, the 4361 audible accelerated ventricular alarms found during the study period were nearly all "nuisance" alarms. Through the annotation process, the investigators determined that excluding a single patient with ventricular storm, only 12 "true positive" episodes of sustained (lasting ≥30 seconds) ventricular tachycardia occurred, and all of these episodes heralded life-threatening events.
Another alarm that occurs frequently in monitored adults is for the ST segment because a change in ST amplitude lasting 1 minute or longer can signal transient myocardial ischemia. Only a small proportion (9%) of 6196 ST alarm events in this study lasted 1 minute or longer, and the remaining 91% were nonactionable alarms.
According to lead author Barbara Drew from the University of California, San Francisco, the excessive number of monitor alarms found in this study was related to one of three factors:
• Inappropriate alarm settings by the clinical staff;
• Certain patient conditions, such as pacemaker devices or other rhythm abnormalities; and
• Algorithm deficiencies in the monitors.
This study found that nurses often retain the monitor default settings rather than tailoring alarm settings appropriately for individual patients. Monitors that are more interactive and intuitive would encourage individualization and should take this a step further by actually suggesting alarm parameter changes to the user.
The investigators were able to determine whether an arrhythmia alarm was true or false in 94% of the 12,671 cases by observing the seven available ECG leads, and they rarely needed to use non-ECG waveforms (arterial pressure or pulse oximeter waveforms) to reach a diagnosis. They concluded that if the arrhythmia algorithm took advantage of all available leads, it would go a long way toward resolving the "cry wolf" phenomenon of alarm fatigue. The high false alarm rate could be significantly reduced, and without so many false alarms, clinicians would rapidly perceive that alarms were clinically meaningful and would respond accordingly.
Although poor signal quality was not found to be a frequent cause for false alarms in this study, such monitor deficiencies as a failure to detect low-amplitude QRS complexes contributed to heart rate undercounting and false asystole, pause, and bradycardia alarms.
Medscape Nurses © 2014 WebMD, LLC
Cite this: Laura A. Stokowski. Alarm Fatigue? Smarter Monitors Needed - Medscape - Nov 13, 2014.