ECOLOGICAL MOMENTARY ASSESSMENT FOR ASSESSING DEPRESSION IN ADVANCED DEMENTIA: A PILOT STUDY

Abstract

Introduction Up to 38% of individuals with moderate to severe dementia experience clinical depression. Although most studies demonstrate lower rates of clinical depression as dementia advances, this may be attributed to the difficulty of assessment at this stage. Though clinical interviews are thorough in assessing depression, they are often time- and resource-contingent. Instead, healthcare providers often turn to screening tools or scales. However, conventional tools for assessing depression have problems with validity in this population. Individuals with dementia who self-report may lack insight and have memory impairments. Retrospective recall bias may also impact proxy reports from caregivers, particularly in institutional settings. A comprehensive approach that incorporates both self and proxy inputs, is sensitive to changes over time and with lower risk of recall bias is crucial in improving performance of depression assessments for people with dementia. In this study, we propose sampling mood data several times daily and in real-time, in order to shorten the period of recall and to maximize representativeness. This approach, known as Ecological Momentary Assessment (EMA), provides data used to measure target symptoms with greater sensitivity than retrospective methods in a variety of populations. EMA is delivered using mobile technology and has shown to be a feasible and acceptable method of assessment. However, the use of a proxy mobile EMA for depression in dementia has yet to be developed or evaluated. The objective of this study was to design and pilot a mobile EMA tool for assessing depression in individuals with dementia. Specific aims sought to evaluate the tool's feasibility and reliability in assessing depressive symptoms. Methods A literature review was conducted to determine commonly used and validated assessments for depression in dementia. These assessments were analyzed for common items, where items less commonly used or not relevant to be collected using EMA were excluded. Wording of items were also analyzed to develop the specific structure of questions for the EMA tool. The tool was piloted in a study on the Specialized Dementia Unit (SDU) at the Toronto Rehabilitation Institute (TRI). Nine participants (22% female, mean age = 77±10.2?years, mean Mini Mental State Examination score = 5) with moderate to severe dementia were enrolled in the study. Participants were assessed for clinical outcomes of depression using the Provisional Diagnostic Criteria for Depression in Dementia (PDC-dAD), the Neuropsychiatric Inventory (NPI) Dysphoria subscale and the improved Clinical Global Impressions Scale (iCGI) at admission and discharge. Observations were completed on phones that had access to the EMA tool via RedCap survey software by research and nursing staff. Across six weeks, five researchers were involved in observing participants at different time points and in different combinations. Separately, five nurses completed observations twice per shift for every shift during their work week. Brief follow-up interviews were conducted to ascertain the nurse experience of using the tool and the Systems Usability Scale (SUS) was completed. Total SUS scores and nurse completion rates using the EMA tool were generated to demonstrate feasibility. Results Six depression assessments validated in dementia were found and demonstrated adequate performance outcomes. Items fell into either mood-related, dementia-related, vegetative, psychotic or positive mood symptom groups. The mood-related group was analyzed separately for prominent items, which included sadness, anxiety, pessimism, loss of interest and irritability. Wording of items were modified to be consistent with being collected in real-time, rather than retrospectively. These items were incorporated as core observational domains in the application to be tested. Sadness and anxiety were additionally included as self-report items as studies have shown these to be most discordant between individuals with dementia and informants. The tool was piloted on the SDU at TRI with research and nursing staff. Preliminary data analyses demonstrated that when admitted to the unit, 22% of participants met criteria for major depression as demonstrated by the PDC-dAD and 44% were experiencing at least some degree of depressive symptoms as demonstrated by the NPI Dysphoria subscale. Three nurses were recruited to observe patients. The average SUS score was 80% and average completion rate was 82% across nurses. Preliminary analyses of the EMA data demonstrated a total of 1,557 observations across five raters and three nurses, where no participants had less than 75% of their data collected. Conclusions This research provides an innovative approach in assessing depression in dementia. Future steps involve evaluation of the tool's inter-rater reliability and exploring outcomes related to validity and responsiveness to change of the EMA tool. This research was funded by: This research has been generously funded by the Walter & Maria Schroeder Institute for Brain Innovation and Recovery.