Delirium Identification in Older Patients With Alzheimer's and Other Related Dementias In the Emergency Department

Delirium Identification in Older Patients With Alzheimer's and Other Related Dementias In the Emergency Department Using Wrist Accelerometer Biosensors and Machine Learning

Delirium is highly prevalent and very bad for patients with dementia. Delirium is a dangerous medical condition that occurs in 6-38% of older Emergency Department patients and 70% of ICU patients. A person who develops delirium in the ED or hospital has a 12 times higher odds of being newly diagnosed with dementia in the next year compared to a similar patient who does not become delirious. Delirium is especially dangerous for persons living with Alzheimer Disease and Related Dementias (AD/ADRD). Persons living with ADRD have an almost 50% chance of developing delirium in the hospital.

Clinicians are bad at recognizing delirium. A recent systematic review led by the Geriatric Emergency Care Applied Research network (NIH funded) found that current delirium screening tools are at most 64% sensitive, meaning that physicians can identify some phenotypes of delirium well, but cannot easily rule out delirium in acutely ill older patients.

The investigators propose integrating wrist biosensors into the emergency management of older adults with dementia. The investigators will monitor heart rate variability, movement, and electrodermal activity (electrical activity of at the level of the skin) to determine if an array of biosensors more sensitive to delirium than current verbal screening tools.

Aim 1: (Patient and Caregiver Acceptability) Collect biosensor data on 60 older adults with dementia in the ED. The study population is any older adult with dementia in the ED who is anticipated to be in the ED or hospital for >4 hours. The investigators will place a wrist biosensor on the dominant hand (unless prohibited by IV placement, in which case the nondominant hand will be used). The investigators will ask questions about acceptability to caregiver (if available) and patient within 4-24 hour mark after biosensor placement. The acceptability questions are derived from the Theoretical Framework of Acceptability Questionnaire.

Study end data will be collected in person prior to discharge, but if unable to be completed before discharge from the hospital, the investigators will call them back at home to complete the final survey about acceptability.

Aim 2: Correlation of biosensor data with delirium phenotype. RA will perform a CAM ICU 7 upon enrollment, and at 24 and 48 hours (or upon discharge, whichever comes first). The investigators will also collect nurse scores for RASS (agitation scores documented as part of usual clinical care), times of scoring and any other delirium or cognitive assessments done as part of usual care (e.g., CAM -ICU, delirium triage screen, mini-cog).

Biosensor data will be coordinated with RASS and presence of delirium using machine learning analysis.

Other data for covariates: medication history/medication administration records, age, gender, length of stay in the ED and in the hospital if applicable. Stage of dementia or most recent cognitive assessment. Baseline skin tone using the expanded Fitzpatrick scale (skin tone scale).

Participant exclusion criteria: patients who are intubated, getting electrical cardioversion or sedation in the ED. Patients who cannot consent due to cognitive impairment and do not have a LAR or caregiver available will also be excluded.

Inclusion Criteria:

• adults patients 65+ years old with a known diagnosis of dementia or medical team highly suspects or confirms dementia this visit.
• Emergency Department visit and/or hospitalization anticipated to last >4 more hours from the time of enrollment.

Exclusion Criteria:

• Intubation/mechanical ventilation
• Patients undergoing a procedure requiring sedation
• Patients who cannot consent due to cognitive impairment and do not have a legally authorized representative or caregiver present.