gms | German Medical Science

Introduction: False alarms are a frequent occurrence in critical care units. The consequential phenomenon – alarm fatigue – threatens staff wellbeing and patient safety [1], [2] as it leads to clinicians ignoring or deactivating alarms, resulting in missed events or delayed response times [3]. Computational approaches are increasingly applied to address the problem; however, the developed approaches lack clinical implementation to date [4]. For real-world translation, engineers should involve clinical domain experts in finding solutions to this socio-technical problem. This study aims to investigate the current state of clinician-engineer collaboration on publications related to computational methods of false alarm reduction.

Methods: A literature search was conducted on 6 databases (PubMed, CINAHL, EMBASE, Web of Science, IEEE Xplore, and Cochrane) to identify publications on computational approaches for false alarm reduction published between 2010 and 2022.The search strategy included the topic domain (i.e., “alarm”) and setting (i.e., “critical care”). The inclusion criteria included studies on computational approaches to reduce false alarms in critical care. Exclusion criteria were publications on non-clinical alarms or alarm fatigue, information system alert fatigue, and short conference abstracts. We extracted the author's professional background (nursing, medicine, engineering, or other) and their role (any authorship, first author, last author). If one author fit two categories, an individual decision on the more dominant category was made. Data was extracted from the publication itself or secondary sources (i.e., university website). The individual proportions of author backgrounds were averaged to calculate the overall mean proportion of author backgrounds.

Results: Overall, 50 publications were included. The mean number of authors of all publications was n=4.60 (SD 2.08). Most authors in the publications were from the engineering field, with a small percentage from medical, nursing, or other backgrounds (Table 1 [Tab. 1]). Only a few articles had first or last authors from clinical professions.

Discussion: Our findings suggest a lack of equitable collaboration between engineers and clinicians in developing solutions for false alarm reduction. The overwhelming majority of publications have engineers as first and last authors, with little clinical input. This finding resembles a trend shown in another review of clinical artificial intelligence literature, showing data experts generally succeed the number of domain experts as authors [5]. This indicates a potential mismatch between the development of solutions and the need to draw on necessary clinical insight for addressing a problem holistically, e.g., labeling alarms as clinically actionable. Engineers may have the technical expertise to develop computational approaches, but clinical expertise is essential in ensuring that solutions are clinically viable and meet the needs of clinicians and patients. Otherwise, solutions may be ineffective or introduce new safety risks. Due to the eligibility criteria, passive involvement of clinicians (i.e., as study subject) was not possible nor regarded as legitimate involvement.

Conclusion: There is a compelling need for intensified collaboration between engineers and clinicians. To address the challenges of alarm fatigue effectively, one must involve experts with real-world clinical expertise for developing clinically relevant and applicable solutions. Ultimately, this will improve patient outcomes and staff wellbeing in critical care units.

The authors declare that they have no competing interests.

The authors declare that an ethics committee vote is not required.