ENVIRONMENTAL TRIGGERS OF SPECIFIC SUBTYPES OF AGITATION IN PEOPLE WITH DEMENTIA: AN OBSERVATIONAL STUDY

Abstract

Introduction Among the most critical behavioral and psychological symptoms of dementia, agitation can lead to decreased quality of life of people with dementia and their caregivers. Limited research exists quantifying environmental triggers of agitation or its subtypes. In this study, we aim to quantify the relationships between specific environmental factors and agitation, as well as specific agitation subtypes, such as motor and verbal agitation. Methods Using a cross-sectional repeated measures design, 37 people with dementia, admitted to a specialized neuropsychiatric ward for patients with dementia and severe behavioral and psychological problems, were each included for one week. During this period, the Pittsburgh Agitation Scale was filled in by the nurses on the ward following an Experience Sampling methodology to assess a patient’s agitation level on a momentary basis. Continuous environmental data (light, sound and temperature) was collected from fixed sensors mounted on the ward. Generalized linear mixed models were used to quantify relationships between environmental variables and outcome variables (agitation, motor agitation, and verbal agitation). These models accounted for the hierarchical nature of our dataset as well as confounding factors, such as time of day and the room-level location of the patient. The time window for analysis was selected through comparison of beta coefficient estimates across various window lengths. Models were built up sequentially, per outcome variable, using selected features per environmental modality. Results We found that different environmental factors were most informative for different subtypes of agitation: mean light level (β: -0.61, C.I.: -1.12 –0.10, p: .02) for motor agitation, and standard deviation of sound level (β: 0.68, C.I.: 0.34-1.02, p: LESS THAN .001) for verbal agitation. Contextual factors such as time of day (β range: 0.51-0.94) and room_x0002_level location (β range: 0.85-1.08) were also significant predictors of agitation. Conclusions Integrating the key differences in predictors between verbal and motor agitation, identified here, could substantially improve subtype-specific model performance. Overall, these findings can aid in the development of predictive models for agitation and enable subsequent just-in-time interventions, improving quality of life for both patients and caregivers.