Design and setting
This cross-sectional, explorative study reports on research undertaken within the “Life outside home for people with dementia” (OUTDEM) setting as part of a larger project in leadership of Karolinska Institutet in Sweden. Standardized questionnaires were used in interviews with participants with and without dementia living in the French-speaking region of Switzerland, in both rural and urban regions.
Participants and recruitment
All participants (n=70) were community-dwelling older adults (65+). Recruitment started in December 2015 and ended in May 2017. Participants with dementia (n=35) were recruited through memory clinics, day hospitals, and the Swiss Alzheimer's association. Diagnosis of dementia was established by physicians at memory clinics. Participants in the comparison group (n=35) were recruited through senior associations and advertisements in grocery stores. The comparison group without known dementia (n=35) was aimed to match, but not paired with the dementia group regarding age, gender, living areas and settings, and education level; thus, recruitment strategies for the comparison group targeted specific regions, age groups, or living areas, for example, to bring the distribution of the comparison group closer to the dementia group on those variables.
An approximate required sample size was calculated based on the difference of the total number of places visited between the 26 older adults and the five persons living with dementia who took part in the study presenting ACT-OUT development (40). The same approximate sample size of groups of 35 were used in similar studies (13,41). No formal power calculation was conducted due to the exploratory design. Findings of this study might be used with former ones to generate power calculations for future research using ACT-OUT.
Data collection procedures
The interviews were conducted by two registered occupational therapists of which one was the first author. Both interviewers had prior knowledge of using ACT-OUT and had harmonized the way they conducted the interviews (42). The interview comprised of three standardized questionnaires, performed in this order: (i) the Participation in ACTivities and Places OUTside Home Questionnaire (ACT-OUT) (40); (ii) the Montreal Cognitive Assessment (MoCA) (43); (iii) socio-demographic questions. Written and verbal informed consent was obtained from each participant prior to data collection. The process consent method proposed by Dewing (2002) (44) was used in this study, because it is person-centred; and it enables researchers to include consent communicated through behaviour and non-verbal means by the person with dementia. Thus, following Dewing (2007) (45), an ongoing consent monitoring was implemented throughout the whole data collection to ensure no stress or burden from participating in the project occurred (46). To mitigate against fatigue and potential burden, interviews occurred in the participant’s home and were adapted to each participant e.g. inviting a significant other for emotional support or spreading the sessions, of no more than two hours. An ethical authorization (protocol 452/15) was obtained from the "Commission cantonale d'éthique de la recherche sur l'être humain (CER-VD)" in Lausanne, Switzerland.
Data analysis: questionnaires and variables
ACT-OUT has three parts. Part I includes a list of 25 pre-determined types of places, grouped into four domains: A, consumer, administrative, and self-care places (n=7); B, places for medical care (n=5); C, social, cultural, and spiritual places (n=6); and D, places for recreational and physical activities (n=7). Part I asks questions whether respondents visit these places in the past, present, and future. Part II poses detailed questions about factors potentially influencing participation in places retained and abandoned, like activities performed, transportation means, accompanying persons, risk perception, and familiarity. Part III consists of general questions about perceived out-of-home participation, life satisfaction, and attitudes towards risk-taking and stress factors.
The Montreal Cognitive Assessment (MoCA) (43) was used as a comprehensive screening tool for assessing and describing the level of cognitive functioning for both groups. It is made of sections focusing on diverse cognitive functions (memory, time and space orientation, visual perception, for example). The total score reflects the cognitive level of participants.
In this study, data from ACT-OUT (Part I and III), MoCA total score and socio-demographic questions were used.
Dependant variable: perceived participation
The dependent variable is the perceived out-of-home participation question asked in Part III of ACT-OUT (How do you perceive your participation in all situations outside home to be?) to which participants responded using a Likert scale (4 = I participate as I wish; 3 = I participate almost as I wish; 2 = I rather do not participate as I wish; 1 = I do not participate as I wish). Of the four levels only one participant with dementia responded do not participate as I wish, thus allowing the aggregation of levels 1 and 2.
Measures: independent variables
The 15 independent variables were derived from the ACT-OUT questionnaire (Part I: number of places visited (n=5) and Part III: risks (n=4)) and socio-demographic questions (n=6). The five variables from part I are first the total number of places visited (n=25), which is the sum of the number of places in all four domains, the following four are the number of places visited in each of the four domains A/consumer, administrative, and self-care places (n=7), B/places for medical care (n=5), C/social, cultural, and spiritual places (n=6), and D/places for recreational and physical activities (n=7). The categorization into domains allows for the type of places to be highlighted.
The independent variables from the socio-demographic questionnaire are: (i) age (years); (ii) gender (male or female); (iii) living situation (alone or with someone); (iv) education level, adapted into three levels from The International Standard Classification of Education (ISCED 2011) (47) (primary/secondary school, apprenticeship and higher education degree); (v) rural/urban, adapted to the context as village, small town and city; (vi) use of car, (access to a car as self-driving, someone else drives, and no use of car).
In ACT-OUT Part III, four risks influencing out-of-home participation , identified in the literature about people with dementia (37,39) are questioned with a four level Likert-type rating scale about the concern that respondents perceive (very concerned; concerned; unconcerned; very unconcerned). The four types of perceived risks were (i) falling; (ii) getting lost; (iii) feeling stressed; (iv) feeling embarrassed. As no participants responded very concerned on any of the risks, these variables were transformed into a three-level Likert-type rating scale.
First, we computed descriptive statistics for all variables, and we systematically tested the difference between the dementia and comparison groups using the Fisher exact test for categorical variables, and the t-test for continuous variables. To explore the associations between total number of places and the perception of out-of-home participation, Spearman correlations coefficients were evaluated for the dementia group and the comparison group separately, as well as for the full group of 70 participants. The strength of association was classified using Cohen’s guidelines for social sciences: .1–.3 (small); .3–.5 (medium); and .5–1.0 (large) (48). For testing the differences between the total number of places by the level of perceived out-of-home participation, we used the Mann-Whitney U test for the comparison group (two levels of perceived participation) and the Kruskal-Wallis test for the dementia group (three levels of perceived participation). Notice that we had to use non-parametric tests because of the non-normality of data and of the small sample sizes.
Second, to explore the associations between each of our independent variables and perceived out-of-home participation, bivariate regressions were run. As the group variable was significant (p<.001) in the full sample analyses, it was decided to run the regressions for the dementia and comparison groups separately. We used logistic regressions for the comparison group, since the dependent variables had two categories, and multinomial regressions for the dementia group, since the dependent variable had three categories. At this point, we selected all independent variables significant at the bivariate level and we entered them together into multivariate regressions.
For all regressions, the highest level of perceived participation was used as the reference category. The type I error was set to .05 for all analyses. Results are given as relative risk ratios (RRR) with 95% confidence intervals for the multinomial regressions, and as odds ratio (OR) with 95% confidence intervals for the logistic regressions. Analyses were undertaken in the Statistical Package for Social Sciences (SPSS) computer software, version 25.