(733) Identifying Patients at Risk for Unplanned Extubation

Introduction: Unplanned Extubation (UE) is a serious adverse event that can lead to instability, cardiac arrest, and emergent re-intubation. It is also associated with prolonged mechanical ventilation and length of stay. We aim to develop a UE risk assessment score (UERAS) tool that reflects risk factors to alert clinicians. We hypothesize our new UERAS scores are reliable and feasible.

Methods: At a 75-bed quaternary non-cardiac ICU in the USA, we developed UERAS after modifying a published UE risk assessment tool by Vats (PCCM 2017) and adding contributing factors from our local UE review. Once a list of candidate variables is developed, each item is evaluated for feasibility for reliable data capture. We piloted UERAS to evaluate the reliability of total score and utilization of each item. We provided a standardized training to ICU Respiratory Therapists (RTs). Two independent RTs evaluated an ICU patient simultaneously using UERAS. Interrater reliability (IRR) with Pearson’s correlation coefficient.

Results: The developed UERAS includes 14 items: anatomical risk, significant movement, agitation, State Sedative Scale (SBS) > goal, more than 3 as needed sedatives in 6 hours, clinical concern for delirium, clinician’s intuition, multiple procedures/transports, proning, UE history, excessive oral secretions, emesis & gagging, planned extubation ≤ 24 hours, and frequent ETT taping. The score ranges from 0-18.

During March to June 2023, 50 paired observations were performed by 16 RTs with IRR 0.997 (p < 0.001). The score difference between two raters was 0 in 48 pairs, 1 in 2 pairs.

Subsequently, 562 assessments (73 patients, median age 13 months [IQR 6-168 months], 65% male) were completed. 12% met a priori defined high risk criteria (UERAS≥6). Item level analysis revealed anatomical risk (67%), agitation (27%), more than 3 as needed sedatives bolus (26%), clinician’s intuition (25%), significant movement (24%). There was only one UE event in this period; the UERAS=1 at 6 hours before the UE event.

Conclusions: Our UERAS tool is highly reliable and feasible to use at bedside, however, prognostic validity may be difficult to prove given low UE occurrence. Next step is to automate items from electronic health records and to generate an alert system to bedside clinicians.