Background: While telemetry was designed to improve patient safety, the volume and low positive predictive value of telemetry alarms can endanger patients by desensitizing healthcare providers to those alerts. To address this safety concern, the Joint Commission requires hospitals to establish specific policies and procedures to balance alarm fatigue and patient safety in their alarm systems by 2016. However, the ideal approach to safely reduce the alarm burden is unknown. We sought to characterize the distribution and etiology of telemetry alarms on the adult general wards to identify targets for intervention to decrease alarm burden and associated fatigue.
Methods: We conducted a prospective direct observation study of telemetry alarms on the wards in an academic medical center. A convenience sample of 32 hours of observation by a trained observer on a mix of general medicine, cardiology, surgical, and cardiac surgery units was included. The observer was positioned at the nurse station telemetry screens and received real-time notification by pager or phone alerts for all alarms during the observation period and followed the alarm to the patient room to determine the cause, caregiver response, and if it reflected a true physiologic instability, as defined by a combination of artifact-free telemetry readings consistent with the alarm and patient and nurse report.
Results: A total of 390 telemetry alarms were observed. The majority (n=326, 84%) were technical in nature, comprised primarily of “leads off” alarms. Only 4% represented true physiologic instability, with cardiac alarms the most likely to be real (8 of 28, 29%), followed by respiratory alarms (6 of 36, 17%), p<0.01. Clinical staff responded to less than half (n=156, 40%) of alerts overall, with respiratory alarms being the most likely to garner a response (55% vs 39% for cardiac and 38% for technical alarms, p<0.01). Caregivers responded more often to real alarms (n=12 of 14, 86%) than false alarms (n=144 of 376, 38%; p<0.01). False alarms were most often caused by patient movement (n=147 of 368, 40%), followed by caregiver actions, (n=121 of 368, 33%), and equipment problems (n=100 of 368, 27%); 8 alarms were excluded (cause undetermined). Of the alarms observed, 50 (13%) were from patients who never received orders for continuous monitoring.
Conclusions: Our study confirmed a high burden of false alarms and associated alarm fatigue, with 14% of true alarms not receiving a response. We identified the “leads off” alarm as the primary contributor and a potential target for intervention to decrease the non-actionable alarm burden.