Session S61.4

Alarming Data from the Intensive Cardiac Care Unit

JA Lipton*, MJB van Ettinger, RJ Barendse, TB van Dam,
NHJJ van der Putten, SP Nelwan

Erasmus University
Rotterdam, Netherlands

Patients admitted to the Intensive Cardiac Care Unit (ICCU) are closely monitored by different devices. When an abnormality is detected, an alarm is generated. However, most alarms do not signify a life-threatening event. The distribution of alarms by category and over time is not well known, but may provide information on how to improve the accuracy.
During a four month period all alarms from the 8-bed ICCU at the Erasmus MC were collected in a database by receiving the pager messages generated by the patient monitoring network. The alarms were subsequently categorized by type, hour and day of week.
Over a four month period 34827 alarms were collected during 547 patient admissions with a median (IQR) duration of 6.1(3-23) hours. The most frequent alarm categories were related to mechanical ventilation (42.2%), blood pressure (32.3%), electrocardiogram (9.8%) and heart rate (8.1%). 2750 (7.9%) of the alarms were not limit violation alarms, but technical advisory messages. Main causes of these advisory messages were: ECG artifacts (64.9%), disconnected devices (29.3%) and SpO2 artifacts (1.6%).
The mean alarm frequency was 2.2 per patient and 13.4 for the ICCU per hour. The time between alarms varied greatly with a median (IQR) of 1.3 (0.5-4.1) minutes, mainly due to the occurrence of alarm “bursts”: in 50% of the alarms the interval was less than 90 seconds. There was overall decreased alarm frequency during nighttime (0.00 until 6.00) when compared to day: 9.7 vs. 14.6 per hour (P<0.001). A peak in alarm frequency was seen at the time of drawing blood samples from the arterial line and awakening of the patients. Another peak occurred between 8.00 and 10.00 when patients are washed. The alarm frequency during these peaks was 20.3 per hour vs. 12.4 during other hours (P<0.001). There was no difference in frequency of alarms by day of week.
Overall alarm frequency on the ICCU was relatively low. However, the distribution over time varied greatly due to the occurrence of alarm “bursts”. Combining the alarms within these “bursts” to display only the most important information could be an interesting target for clinical decision support.

(Abstract Control Number: 241)