BMJ Open. 2026 Aug 25;16(8):e121099. doi: 10.1136/bmjopen-2026-121099.
ABSTRACT
INTRODUCTION: Perioperative monitoring requires clinicians to integrate multiple physiological data streams while managing interruptions and audible alarms. The Visual Patient Avatar (VPA) is an avatar-based monitor visualisation that integrates vital signs and sensor states into an animated virtual patient. This protocol evaluates whether VPA implementation is associated with objective changes in monitoring and alarm outcomes during routine perioperative care.
METHODS AND ANALYSIS: This investigator-initiated, dual-centre, quasi-experimental pre-post observational study uses routine perioperative monitoring and alarm data from NewYork-Presbyterian/Weill Cornell, USA and University Hospital Zurich, Switzerland. It includes approximately 58 000 anaesthesia cases (16 000 and 42 000, respectively); adult and paediatric cases will be analysed separately. Data span pre-implementation, adaptation and post implementation phases; technical deployment intervals are excluded. Each anaesthesia case is one unit of analysis. The primary outcome is the duration of vital-sign deviation per 60 min of valid measurement. Secondary outcomes are a time-based area under the curve metric and audible alarm burden (count, time and reaction time). Multivariable models will adjust for patient, procedure and care-context covariates. Analyses will be performed separately by centre.
ETHICS AND DISSEMINATION: The Weill Cornell Medicine Institutional Review Board approved the study and granted a waiver of informed consent (IRB No. 22-05024803). At University Hospital Zurich, only cases from patients with written general consent for research use of routinely collected health data were included. The Cantonal Ethics Committee Zurich issued a declaration of non-jurisdiction (BASEC Req-2021-00756; 6 July 2021), and the University Hospital Zurich Data Governance Board approved data export and anonymisation (DGR0000504; 31 January 2025). A data-sharing agreement was established on 11 April 2023. Data-collection windows are complete; extraction, linkage and quality control are ongoing, and no comparative outcome analyses have been performed. Findings will be disseminated through peer-reviewed publications and conferences.
PMID:42642085 | DOI:10.1136/bmjopen-2026-121099
