Objective:

This study aimed to use video monitoring to identify patient behaviors and environmental characteristics associated with delirium to develop efficient markers of delirium.

Background:

Delirium is an acute confusional state common in older hospitalized patients, associated with increased morbidity and mortality. However, delirium is under-recognized. Current clinical tools to assess delirium are subjective, require in-person evaluation, and can be time-consuming. There is a need for methods to improve detection and prediction of delirium that could be applied efficiently and broadly.

Design/Methods:

We conducted an observational cohort study of hospitalized patients expected to stay at least one night, who were evaluated up to twice daily using video followed by bedside evaluations. Video evaluations were performed using structured observation of patient behaviors and environmental features via the hospital’s real-time video monitoring system, visualized outside of patients’ rooms. Bedside evaluations for delirium were performed using the 3-Minute Diagnostic Interview for Confusion Assessment Method (3D-CAM). Pearson’s chi-squared test and generalized linear mixed-effects models were used to identify features differentiating delirious and non-delirious patients.

Results:

A total of 318 video evaluations (46 unique patients) were analyzed. Among these video evaluations, 199 corresponded with a positive delirium assessment (median video evaluation duration 2.05 minutes), and 119 with no delirium (median video evaluation duration 1.60 minutes). Analysis revealed video features significantly associated with delirium, including reduced spontaneous and purposeful movements, eyes either closed or fixed in a stare, mouth being open, and less interaction with the environment (ex. other people, phone, TV).

Conclusions: