ICU Background Early Awareness for Critical deterioratiON

Single-center, stratified, cluster-randomized (the ICU units will be randomized) crossover analysis with outcome-assessor blinding. The analysis will be conducted in the adult intensive care units (ICUs) of Inselspital, Bern, comprising two distinct ICU units: the Blue and the Yellow ICU Unit. Each unit contains two sub-units, which will serve as the clustering units for randomization. The study follows a two-phase design. In the initial phase, sub-units within each ICU unit will be randomly allocated in a 1:1 ratio to either the additional BEACON scoring group or control group. Patients admitted to sub-units allocated to the BEACON group will receive standard care plus BEACON scores, while those admitted to control group will receive standard care only. After predefined cluster periods, allocations will be swapped between the BEACON and control groups. A wash-out period of two weeks will be observed between swaps to minimize carryover effects. During the wash-out period, no new study patients will be enrolled. Once the swap is complete, only newly admitted patients will be enrolled under the sub-unit's new allocation. Patients still occupying sub-units from the preceding period will be censored from the study at the time of the swap and excluded from outcome analysis beyond that point. The same approach will be applied for patients who will be transferred from one unit to a different unit during the ICU admission for logistical/organizational reasons. Importantly, individual patients will only be exposed to one analysis condition-either the BEACON or control-based on the bed allocation at the time of their admission. Only the sub-units themselves undergo crossover, ensuring temporal balance while maintaining patient-level exposure to a single condition.