Reducing Ineffective and False Alarms in the ICU
Abstract & Commentary
By David J. Pierson, MD, Editor, Professor, Pulmonary and Critical Care Medicine, Harborview Medical Center, University of Washington, Seattle, is Editor for Critical Care Alert.
Synopsis: In this study of alarms in a medical ICU, in each patient room there were approximately 6 alarms per hour, only 23% of which were judged as effective (action taken within 5 min). The authors calculated that 67% of ineffective and ignored alarms could be eliminated if the detected change was required to persist for 19 sec before the alarm sounded.
Source: Görges M, et al. Improving alarm performance in the medical intensive care unit using delays and clinical context. Anesth Analg 2009;108:1546-1552.
Investigators at the University of Utah hospital carried out a comprehensive observational study of alarms that sounded in the medical ICU. On each of 24 days, an observer spent approximately 9 hours in the room of a patient who was intubated and receiving ventilator support, the room being randomly selected each day. Using a computer, the observer logged the time and duration of every alarm, the characteristics of the alarm (device, threshold setting, variable triggering the alarm, etc.), which providers entered and left the room (e.g., nurse, respiratory therapist, physician, etc.) and when they did so, and the action taken (e.g., alarm silenced, drug administered or modified, patient assessed, ventilator settings changed, etc.). Alarms were classified as either effective (alarm-related action was performed by an appropriate health care provider within 5 min), ineffective (no provider entered room in response to alarm), or ignored (provider was present in room but no alarm-related action taken within 5 min or alarm silenced at nursing station with no action taken).
Two hundred hours of data were collected on 22 patients over 24 days, with observations made for an average of 9.16 hours/day (range, 6.25-10.5 hours). During the observation period, 1214 alarms occurred (6.07 per hour). Of these, 23% were effective; 5% were patient-related, such as hypotension, agitation, need for suctioning, etc.; and 18% were technically related, such as restarting infusion pumps, changing sensor positions, etc. Thirty-six percent of all alarms were ineffective, and 41% were ignored. Alarms sounded for an average of 3.28 min per hour in the room being observed. Median alarm duration was 17 sec, with a range of 1 sec to 17.25 min; 45% of alarms lasted < 15 sec, 74% sounded for < 30 sec, and 89% for < 60 sec. Thirty-four percent of all alarms canceled themselves without any health care team member being present in the room. Based on the observed durations of the recorded alarms, a 14-sec delay would have reduced the number of ineffective and ignored alarms by 51%, and a 19-sec delay would have reduced it by 67%. The authors calculated that introducing a 19-sec alarm delay and automatically detecting suctioning, repositioning, oral care, and washing could reduce the number of ineffective and ignored alarms from 934 to 274.
Commentary
The ICU is a noisy place, with a substantial proportion of the noise coming from alarms. The majority of those alarms are either ignored or undetected by the clinicians they are intended to alert. As pointed out in an editorial1 accompanying the article by Görges et al, if their findings from individual beds were extrapolated to an entire 10-bed ICU, some alarm would be sounding about 50% of the time — a figure that is consistent with the results of other studies.
The authors' assertion that introducing a 19-sec delay before a given alarm begins to sound would greatly reduce the number of ineffective and ignored alarms is appealing. But would such a delay be safe for the patient? It might be, if alarms could be made to sound immediately with certain crucial events — asystole or apnea, for example — while deferring an audible signal until the detected phenomenon had persisted for a specified interval. In fact, repetitive alarming has been shown to desensitize clinicians, who become less likely to respond the more often an alarm sounds: If the number of false and irrelevant alarms could be greatly reduced it might permit caregivers to become re-sensitized to them and more alert to their signal that something important may be happening — which is their original purpose. Development of "smarter" alarms, capable not only of detecting clinically relevant events, but also of filtering out changes accompanying routine care or patient movement in bed, would be a major advance. For now, though, we are stuck with trying to recognize and respond to alarms that indicate important events amid the welter of false and nuisance alarms in the ICU.
Reference
- Imhoff M, Fried R. The crying wolf: still crying? Anesth Analg 2009;108:1382-1383.
You have reached your article limit for the month. Subscribe now to access this article plus other member-only content.
- Award-winning Medical Content
- Latest Advances & Development in Medicine
- Unbiased Content