Introducing Data Driven Alarm Management to the Intensive Care Unit of a German University Hospital

Whenever the vital signs of an intensive care unit’s patient deviate from values that are considered safe, staff members receive automatically triggered alarms from the bedside monitoring systems. If there are too many alarms and if the vast majority of these turn out to be false or non-actionable, staff can become alarm fatigued, that is desensitised to alarms. Alarm management interventions can help to reduce the overall number of alarms. These interventions are most effective, if they are adapted to a unit’s unique situation. With this thesis, I provide the intensive care unit of a German university hospital with the foundation for a repeatable, data driven alarm management. To this end, I analyse 90 days of alarm log- data in order to create an overview of the unit’s current alarm situation and to understand the way staff members are being confronted with alarms. Additionally, I survey staff members with a questionnaire that aims to assess their alarm fatigue and their approach to the unit’s alarm situation. The results of both, the data analysis and the questionnaire, call for alarm management interventions that effectively reduce the number of alarms in order to ensure patient safety and ICU staff’s work satisfaction. I provide suggestions for how one should proceed in implementing alarm management, as well as concrete starting points that can be used to design interventions that effectively and efficiently reduce the amount of alarms.