A web-based survey for expert review of monitor alarms

Bedside monitors trigger alarms if the state of a patient needs attention. However, many studies have shown that there is a high rate of false or clinically irrelevant alarms. Despite technological and methodological advances false alarms persist in today's monitoring systems. There is still a lack of annotated, standardized and publicly available alarm databases that would foster promising results of studies that use data mining methods, knowledge-based systems or alike for alarm rate reduction. We present a web-based survey conducted at the German Heart Center Munich in which physicians had to rate given monitoring events. Our results unveiled significant differences in the ratings within the group of participants (deviations of 21.16%) but also in comparison to an established alarm set of a second group of reviewers (31.28%). Manual alarm ratings depend on subjective assessments and are therefore error-prone. Our web-interface gives access to reviewers on a large scale. This allows substantial analysis and can help to validate existing alarm databases as well as to establish new ones.