Main hypothesis. Floor and door visual exit barriers will decrease exit-seeking wandering (exit door approaches) proximal to exit doorways by persons with dementia who wander.

Introduction: Wandering behavior is manifested in ways that include lingering near exit doors and/or mimicking or "shadowing" the exiting behavior of others who depart through this door. Referenced as "exit-seeking," such behavior is a common mobility pattern that places persons with dementia who wander (PDW) at high risk of wandering away or eloping. PDW who wander away from a care setting unattended are at risk of death from exposure or accidents if lost and not quickly found, and those that do not wander away but persist in such behaviors are at risk for falls, fractures and other injuries, weight loss, fatigue and sleep disturbances. Traditional methods to manage PDW exit-seeking and wandering away include physical restraints and pharmacology, but these can be harmful. Subjective or "visual exit barriers" to deter exit-seeking offer an attractive alternative because these interventions are non-invasive, low-cost and simple to use. Nursing home tests of visual exit barriers such as plain cloth door covers and stripped floor covers have demonstrated positive results in reducing or completely exit-seeking behaviors. Confirmation of these results in a well designed, home-based trial is warranted.

Research Questions/Hypotheses: The research questions and hypotheses of this 2-year, cross-over assignment with randomized allocation design with two intervention and two no-intervention periods (at baseline and after the first intervention) are: Q1. Will installation of visual barrier interventions change exit-seeking wandering patters, controlling for cognitive impairment? H1. Compared to baseline, installation of a visual exit barrier will lower the frequency and duration of exit-seeking behaviors. Q2. Of the two visual barriers (door and floor), which will result in the greater reduction of wandering (frequency and duration of lingering and shadowing)? H2. Door covers will reduce exit seeking more than floor covers.

Five secondary questions are proposed for descriptive purposes and to generate hypotheses for future studies on wandering management interventions. These include: Q3. What are the baseline characteristics of exit-seeking (predominant loci, durations and variability of durations during waking hours)? Q4. Will the introduction of a visual barrier on one exit doorway alter exit-seeking behaviors occurring in proximity to other potential exit doors and windows within the same area? Q5. Will the introduction of visual exit barriers reduce CG (caregiver) burden? Q6. Will the introduction of visual exit barriers increase PDW agitation? and Q7. How do CG rate visual barrier effectiveness and acceptability for in-home use?

Methods: A crossover assignment with randomized order of allocation design was used. One main exit doorway was sequentially protected with the two types (door, floor) of visual barriers. Data related to exit-seeking of subjects was collected via ultrawide radio frequency identification devices on a continuous basis for four 14-day periods: Entry into a monitored zone 6-feet in front of exit doorways will be considered exit-seeking behavior; PDW locomotion in tandem with CG locomotion will be considered shadowing. Caregiver-reported PDW agitation and CG burden will be collected at baseline and immediate post-interventions. Prior to all statistical analysis, all variables will be reviewed for valid data points and normalcy for continuous data. Should erroneous data points exist, data will be verified to ensure accuracy. Should data not be normally distributed, processes such as transformations or truncations will be completed. Prior to conducting formal hypothesis testing, an order of allocation analysis will be completed to determine if carry-over effects occurred in the study. Analysis (Pocock) will be completed, which in essence examines the mean of differences by use of t-statistics. In addition, intent to treat analysis will be completed. Given we have incorporated a second baseline phase into the study, frequency and duration will be compared between the two baseline phases to ensure consistency. Descriptive statistics (frequency distributions for categorical variables, means and standard deviations for continuous variables) will be calculated for all baseline variables including the covariates.

Inclusion Criteria:

• veterans age 60 and over with an Alzheimer's-like dementia diagnosis, documented in medical record with International Codes for Diagnosis (ICD-9) codes that include all 290 diagnoses and 331.0
• Mini Mental State Examination (MMSE) 24 or less
• evidence of pre-elopement behaviors operationalized as scoring 1 standard deviation (SD) above mean on any one of the three items that comprise the Revised Algase Wandering Scale-Community Version (RAWS-CV) eloping subscale (#16 - He/she attempts to get outside; #27- He/she stands at the out door wanting to go out; #30- He/she attempts to find or go to familiar locations, even unrealistic ones)
• living in a non-institutional private family home or foster home
• independently ambulatory (with or without canes, walkers or wheelchairs to assist)
• living within a 60-mile radius of the James A. Haley Veterans Administration Medical Center (VAMC), Tampa, Florida
• living with a caregiver (CG) who is willing and able to serve as a study partner and provide informant reports, and who intends to be with the PDW for three months (allowing for being away for seven days or less during the three-month study period)
• English speaking.

Exclusion Criteria:

• living in a nursing home, group home or assisted living facility at point of entry (foster homes allowed)
• previous exposure to any visual exit barrier for wandering management
• no live-in CG
• bilaterally deaf or blind
• currently taking anti-psychotic medications
• current use of any visual exit barrier on any main exit doorway. CG will be allowed to use other wandering management interventions that are not specific to the exit doorways. For example, they may use a stop sign at the entrance to the kitchen, but may not use a stop sign on an experimentally monitored door.

Radio Frequency Identification Device measurement method cumbersome and obtrusive; small sample size (n=19). Further investigation is necessary in order to draw conclusions about visual barriers deterring wandering defined as exit door approaching.