Measures Determining Dementia-Related Attitudes in Adolescents: A Scoping Review

Reducing stigma is a key benefit to intergenerational programs. However, little is known about the availability and suitability of measures that capture dementia-related attitudes in adolescents, thus limiting interpretations of the efficacy of such programs. The aim of this scoping review was to provide an overview of outcome measures used to capture dementiarelated attitudes in adolescents. Scoping review methodology was used to systematically identify relevant articles. Key search terms included dementia, attitudes, and adolescents. Fourteen studies met the inclusion criteria, of which 13 unique measures were identified. However, there are gaps in psychometric properties and a lack of underlying theoretical frameworks.


Introduction
Dementia has become an increasing public health priority in international policy.It is estimated that 55.2 million people currently live with dementia worldwide (Kafadar et al., 2021), with this number projected to increase further (GBD 2019Dementia Forecasting Collaborators, 2022).Given the projected increase, the World Health Organization (WHO) outlined a need for greater dementia awareness in the general public as part of the wider goal to achieve a dementia-friendly society (WHO, 2017).Achieving a dementiafriendly society includes tackling some of the negative attitudes toward dementia held by the general public.Two thirds of the general public have misconceptions and a widespread fear of developing dementia (Alzheimer's Disease International, 2019).Negative attitudes can have an impact on people living with dementia, their carers, and society at a psychological, economic, physical, and social level (Herrmann et al., 2018;Prince et al., 2016).For example, over 60% of people feel that it would be important to remove responsibilities from individuals living with dementia.Moreover, 20% of people stated that they would hide their dementia diagnosis from others (Alzheimer's Disease International, 2019).These can lead to individuals socially withdrawing in order to conceal their dementia diagnosis from others and delay seeking help due to the fear of having their responsibilities, such as financial freedom, taken away from them.
Improving attitudes toward dementia may help tackle the barriers that individuals face in seeking timely diagnosis and care (Prince et al., 2016), consequently impacting quality of life for people living with dementia.Drawing upon the mental health literature, education about a group, protest against the inequity faced by a group, and contact with the stigmatized group are potential ways to help tackle negative attitudes (Mukadam & Livingston, 2012).Whilst initiatives such as the "Dementia Awareness Program" and intergenerational programs (Evans & Atkinson, 2017) attempt to address inaccurate stereotypes and beliefs (Chow et al., 2018), these are currently not based on rigorous empirical evidence.There is ambiguity as to the theoretical foundations used to inform such initiatives (Hebert & Scales, 2017) and the validity of dementia-related attitudes outcomes.These may undermine our understanding about what an effective initiative looks like.
A widely adopted definition of attitudes is that an attitude is an evaluation made on a particular object (or persons) through a psychological inclination that is expressed with some degree of liking or disliking (Eagly & Chaiken, 1993).Attitudes, alongside "stereotypes," "perceptions," "beliefs," and "discrimination," are all terms associated with stigma (Celious & Oyserman, 2001;Corrigan et al., 2005).While these terms represent distinct constructs, they all align with the central components of stigma frameworks.Stigma frameworks, such as the tripartite model, argue that attitudes are made up of three main components: cognitive (belief), affective (feelings and emotions), and behavioral responses (actions) (Eagly & Chaiken, 1995;Pryor & Reeder, 2011).These components are viewed as what is best captured by attitude measures (Eagly & Chaiken, 1993).A collection of negative attitudes and beliefs (cognitive responses) lead to discrimination and avoidance behaviors (behavioral responses) toward an attitudinal target (i.e., people with dementia) (Cheng et al., 2011).
There is limited literature on dementia related attitudes in adolescents in general despite theoretical frameworks highlighting the importance of the adolescent stage in attitude formation (e.g., "impressionable years hypothesis" (Krosnick & Alwin, 1989).Schools are also a potentially an optimal setting for delivering national and widespread initiatives since adolescents spend a large amount of their time at school.Schools can also reach adolescents from marginalized groups (Green et al., 2005) and aligns with citizenship education within the school curriculum.Targeting adolescents is important since they are an over-looked group in the general public regarding dementia-related attitudes research, with many studies focussing on attitudes held by healthcare professionals, carers, and adults over the age of 18 years old (Herrmann et al., 2018).This is despite evidence highlighting that adolescents do hold negative attitudes toward dementia and may develop negative attitudes at a young age (Farina, Griffiths, et al., 2020).There is a need to tackle attitudes at a generational level given that stereotypes are more susceptible to change in early adolescence and that adolescents appear more responsive to education and related initiatives (Werner et al., 2017).Intergenerational programs are one such way we can potentially improve attitudes and should be appropriately tailored to the relevant context and different communities, including adolescents (WHO, 2017).Therefore, we should have measures of dementiarelated attitudes that are developed for this purpose, taking into consideration the unique characteristics and needs of adolescents.
Measures of dementia-related attitudes in adolescents are important to consider due to the lack of agreement on a "gold standard" measure for dementia-related attitudes in general (Harper et al., 2018).Measures not rigorously developed, may not accurately capture the underlying construct or lead to measurement error (Bound et al., 2001;Kimberlin & Winterstein, 2008).Even when a measure is rigorously developed for the general public, these cannot always be reliably used in a younger cohort (Isaac et al., 2017).Adolescents may have difficulties with the readability, complexity, and applicability of questionnaires compared to adults (i.e., 18 years and older) (Bell, 2007).As such, when intergenerational programs and other anti-stigma initiatives are developed for adolescents, it is unclear whether the measures used are accurately capturing attitudes in a standardized manner.
The main aim of this scoping review is to describe the outcome measures used when examining dementia-related attitudes in adolescents.To our knowledge, there are currently no reviews focusing on the measures used to assess dementia-related attitudes in adolescents.An overview of such measures will identify those that are validated and psychometrically sound, as well as the common pitfalls of these types of measures.This is important as it will provide a scope on further psychometric refinement before disseminating such measures widely (Farina, Hughes, et al., 2020;Parveen et al., 2020).

Methods
Given the limited literature that exists in dementia-related attitudes in adolescents and the likelihood that there is a limited number of measures in this field, a scoping review was deemed more appropriate than a systematic review.Scoping reviews are useful for covering breadth of existing knowledge.The scoping review protocol is available upon request from the corresponding author.
The Arksey and O'Malley (2005) scoping review framework was adopted which consisted of the following steps: 1) identifying a research question, 2) identifying relevant studies, 3) study selection, 4) charting the data, and 5) collating and reporting the results (synthesis).In line with the framework, a risk of bias analysis was not conducted as it falls outside the scope of this review (Arksey & O'Malley, 2005).The framework was chosen for its transparency (Munn et al., 2018) and was amalgamated with additional scoping review recommendations by Levac et al. (2010) to further strengthen the process (e.g., defining concepts and target population for clearer direction on eligibility criteria) (Levac et al., 2010).

Identifying the research question
The research question was as follows; "what measures are used to determine dementia attitudes and associated domains in adolescents?"

Identifying relevant studies
Studies were identified through systematically searching electronic databases; PubMed, Web of Science, and PsycInfo (ProQuest).For literature inclusivity, no search filters, exclusionary Boolean operators or limitation on the time period were applied to the databases.A combination of MeSH terms, synonyms, variations in spelling for search words, text words, and Boolean operators were used to formulate the search string.Key search terms were in the English language and included those associated with "dementia" (dement* OR Alzheimer*), "stigma" (stigma OR perception OR attitude* OR discrimination OR "social distance" OR prejudice), and "adolescents" (adolescent* OR teen* OR "young people" OR child* OR student* OR "college students").Final search strategies and date of searches were saved for replicability.

Study selection
Only English language papers were included in the review.All articles were stored on the reference manager Zotero and underwent de-duplication.Two reviewers independently screened and examined the titles and abstracts on the Rayyan platform (Ouzzani et al., 2016), applying the eligibility criteria.At the full text screening stage, a similar independent review was conducted, and the level of agreement between the reviewers was determined using Cohen's kappa coefficient (κ) for interrater reliability (a kappa result of 0.81-1.00indicates almost perfect agreement) (McHugh, 2012).Disagreements on study inclusion were resolved between the two reviewers by having a discussion with a third independent reviewer who would be consulted should there have been difficulty in the reviewers coming to an agreement.Further relevant studies were identified from reference lists using the snowballing method and citation searches ("cited by") (Pham et al., 2014).

Eligibility
Inclusion criteria: a) English language papers, b) adolescents (10-18 years old), c) measures for dementia related stigma, d) quantitative outcomes, and e) peer-reviewed articles.Exclusion criteria: a) population targets above the age of 18 years old, b) populations of exclusively university students, c) qualitative methods and outcomes, d) studies reporting on specialist professions views exclusively (medical or healthcare professionals), e) measures exclusively measuring self-stigma or stigma by association, f) only measures with knowledge as an outcome, and g) gray literature (gray literature refers to information not produced by commercial publishers such as conference abstracts and blogs) (Adams et al., 2016).
While the WHO's definition of adolescents includes individuals aged between 10 and 19 years (WHO, 2022), we have chosen to limit the inclusion to individuals up to 18 years old.This is to reflect standard education models.By excluding those aged 19 years old, potential heterogeneity introduced by including university students, who may have specialist training and are more likely to represent a biased cross-section of society (e.g., education level), is minimized.Studies were excluded if the average age of participants did not fall within the required age range (10-18 years old).

Data charting and extraction
Data charting was performed on a pre-designed form for full text extraction.Extracted information included the following aspects: 1) study characteristics, including descriptive data such as study design and demographics, 2) psychometric properties where reported (validity and reliability), and 3) measure characteristics, such as a Likert scale and theoretical framework.Only directly reported and available data from the eligible studies were extracted.

Data synthesis
The descriptive narrative for scoping reviews was adopted (Tricco et al., 2018).The data were categorized into themes based on the characteristics of the outcome measures such as validated measures used in the target population, validated measures used but not in the target population, and unvalidated measures.Validity and reliability outcomes indicated by p-values, α (alpha), and r (correlation coefficent were reported.The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) study flow diagram was adopted for transparency of the selection, analysis, and reporting of the literature (Page & Moher, 2017).

Results
The dates of identified publications ranged from 1979 to 2022.The full text review had a 94.6% agreement rate (κ = 0.88) between the reviewers.As a result, 12 full-text articles were included in the review, and an additional two further studies identified through snowballing (checking reference lists of eligible studies) (Felc & Felc, 2020;Werner et al., 2017) were included.A total of 14 studies were extracted and synthesized.The PRISMA flow study diagram is presented in Figure 1.

Study characteristics
Out of the 14 studies included in the review, nine studies had a quantitative study design, comprising of survey-based studies, two studies with a mixed-methods design and three studies with an intervention study design.Sample size ranged between four (Chow et al., 2018) and 5,515 (Fuh et al., 2005).The median sample size was 330.There were more females (58%) than males (42%) overall.Most of the studies did not report on ethnicity (10 studies).Where ethnicity was reported, participants were mostly homogenous (i.e., White British) (Farina, Hughes, et al., 2020, 2020;Griffiths et al., 2018).Three studies reported on nationality (Baker et al., 2018(Baker et al., , 2019;;Felc & Felc, 2020) with two studies recruiting Australians (>85%) and one study recruiting Slovenian participants.Studies were mostly conducted in England (k = 6) followed by Australia (k = 2) and Taiwan (k = 2) with one study each from the following: Canada, Israel, Macao, and Slovenia.The most common recruitment setting was schools.Specifically, three studies recruited from primary schools, while 11 studies recruited from secondary schools.Table 1 and supplementary material A provides an overview of the studies characteristics.
What measures are used to determine dementia related-attitudes in adolescents?
Thirteen unique measures were identified, which were grouped into three categories: validated in target population, validated not in target population, and not validated.See supplementary material B.

Theoretical frameworks
"Attitudes" was the most commonly measured construct with the outcome typically being "attitudes toward dementia" (KIDS, A-ADS, and brief A-ADS).Three of the measures adopted a theoretical framework.The AQ-9 (a shortened version of the AQ-27) adopted the attribution model of public discrimination (P.Corrigan et al., 2003;Werner et al., 2017), whilst the KIDS adopted the tripartite framework of attitudes (Baker et al., 2018).The brief A-ADS (Farina, Griffiths, et al., 2020) shares similarities with a public stigma framework (Rüsch et al., 2005), but it was not developed based on that framework.

Psychometric properties
An overview of the psychometric properties of the validated measures is reported in Table 3.

Validity
Construct validity was reported for the KIDS, A-ADS, and brief A-ADS measures.Exploratory factor analyses demonstrated a three-factor structure for both the KIDS (Baker et al., 2018) and the A-ADS measure (Farina, Griffiths, et al., 2020;Griffiths et al., 2018).The brief A-ADS demonstrated a single-factor structure (Farina, Griffiths, et al., 2020).Both the KIDS and A-ADS demonstrated convergent validity.The KIDS measure revealed a positive Pearson's correlation with the DAS measure (p < .01),indicating that they measure similar constructs.Similarly, the A-ADS revealed a strong, positive correlation with attitudes toward older people scale (p < .001)(Griffiths et al., 2018), further suggesting the scales measure similar constructs.There was evidence of good concurrent validity for the KIDS and brief A-ADS with both significantly correlated with other measures within the dementia literature (Baker et al., 2018;Farina, Griffiths, et al., 2020;Farina, Hughes, et al., 2020), whilst a moderate, positive correlation between the brief A-ADS and KIDS was reported (Farina, Hughes, et al., 2020).Where content validity was reported, an item pooling procedure from the existing literature and feedback from an advisory committee was observed across the KIDS, A-ADS and questionnaire of knowledge, attitude, and preventive practice of dementia care.The brief A-ADS significantly was able to distinguish between self-reported positive attitudes toward dementia (p < .001)(Farina, Griffiths, et al., 2020).No study reported on criterion validity.

Reliability
All measures were reported to have at least adequate reliability/internal consistency (>0.60), with most measures demonstrating good internal consistency (>0.70).Only one study explored test-retest reliability (Farina, Hughes, et al., 2020).The authors demonstrated that the brief A-ADS had acceptable testretest reliability (r > 0.70) whilst the KIDS demonstrated "questionable A-ADS (Griffiths et al., 2018) Brief A-ADS (Farina, Griffiths, et al., 2020) Questionnaire of knowledge, attitude and preventive practice of dementia care (Lo et al., 2020) Construct validity Good construct validity.Of 33 items, 14 items retained.EFA shows three factor solution (personhood, stigma and dementia understanding) (Baker et al., 2018).

Measure characteristics
Five studies adopted measures that were validated in populations over 18years-old.Three measures were identified from these studies: Allophilia scale (Kinney et al., 2017, used by;Farina, Hughes, et al., 2020;Farina, Griffiths, et al., 2020), the Dementia Attitudes Scale (DAS) (O'Connor & McFadden, 2010, used by;Baker et al., 2018;Griffiths et al., 2018;Liao et al., 2022), and the Young Adult Attitudes about Alzheimer's disease Measure (Lundquist & Ready, 2008, used by;Griffiths et al., 2018).The latter two measures were validated in college (O'Connor & McFadden, 2010) and university students (Lundquist & Ready, 2008).Three studies used these measures for the purpose of developing or validating measures specifically designed for individuals under the age of 18 (Baker et al., 2018;Farina, Griffiths, et al., 2020;Griffiths et al., 2018).Four of the studies needed to simplify items of the Allophilia scale, DAS, and the Young Adult Attitudes about Alzheimer's disease in order to make them more accessible to younger participants.For example, in the case of the DAS, "it is rewarding to work with people who have dementia" was reworded to "it is rewarding to play with people who have dementia" (Baker et al., 2018).In the study by Liao et al. (2022), the DAS was translated into Chinese for participants.

Validity
The reported validity of these measures are presented in Table 3, where the correlations between these measures and the validated measures in individuals under the age of 18 are presented.Content, criterion and predictive validity for all three of the measures were not reported within the adolescent samples.

Reliability
The inter-rater reliability and test-retest reliability were not reported on for any of the measures when used in adolescents.Internal consistency was not reported for the Allophilia scale within the context of these studies on adolescents.The DAS had excellent internal consistency as indicated by Cronbach's Alpha (α = 0.83) where previously validated (reported in Baker et al., 2018;Griffiths et al., 2018).This was supported by Baker et al. (2018) (ωt = .89)and Liao et al. (2022) (α = 0.85-0.87).Reliability for the young adult attitudes about Alzheimer's disease was reported as "good" (cited in Griffiths et al., 2018).

Discussion
The findings of this scoping review reveal that there is limited literature measuring dementia-related attitudes in adolescents.However, out of 13 measures identified, it is encouraging that the majority of measures were validated within the target population.The brief A-ADS and KIDS were found to be the most used validated measures and had the most comprehensive psychometric properties, although this was in a small number of studies overall.To our knowledge, this scoping review is the first in the dementiarelated attitudes literature to map out measures administered in adolescents using established scoping review methodology.In doing so, some of the limitations of the current literature base are available for researchers to consider and help with future work in capturing dementia-related attitudes in the adolescent demographic."Attitudes" was the most common construct measured amongst the studies.However, steps should be taken to reduce the ambiguity of what is being measured by defining the construct of interest.This is because the limitation of "Attitudes" is its variability in definition (Annear et al., 2015).Adopting a theoretical framework would help provide greater clarity regarding what is being measured.However, it is worth noting that there were only a limited number of measures (AQ-9 and KIDS) that adopted a theoretical framework, as mentioned by others in the past (Werner et al., 2020).It is important to note that the frameworks identified in this scoping review (attribution model of public discrimination and tripartite attitude framework) were developed with mental illness in mind (P.Corrigan et al., 2003;Eagly & Chaiken, 1993;Pryor & Reeder, 2011).
Internal consistency was reported for seven of the measures.Cronbach's alpha is widely used to demonstrate reliability, with a coefficient alpha of 0.70 and above as the general rule of thumb for good reliability (Taber, 2018).Amongst the measures validated in the target population, there was a general indication for good internal consistency.The McDonalds Omega was used for the KIDS instead of Cronbach's alpha (Baker et al., 2018) which some methodologists argue is a more optimal measure of reliability for unidimensional constructs (Goodboy & Martin, 2020;Hayes & Coutts, 2020).Test-retest reliability was not commonly reported across the measures, although this type of stability testing is not necessarily deemed the most appropriate for constructs that are expected to change over time such as attitudes (De Von et al., 2007).This may explain the "questionable" test-retest reliability score reported on the KIDS and "acceptable" reliability for the brief A-ADS (Farina, Hughes, et al., 2020).
Almost all of the measures that were validated in populations over 18year-olds required adjustments to item wording to make them more accessible to the younger participants (e.g., Baker et al., 2018).Adapting/removing items may alter the psychometric outcomes of the original measure and therefore measures with these modifications should undergo further psychometric testing to ensure they are fit for purpose within a younger demographic.Only one study reported on measure readability (Farina, Griffiths, et al., 2020).This is a notable limitation common in child measures but are important to the ease and understanding of text for children (Oakland & Lane, 2004;Patalay et al., 2018) due to cognitive effort differences between adults and children (Bell, 2007;Krosnick & Alwin, 1989).These pitfalls are important to consider since these have implications on accurately capturing attitudes in intergenerational programs.Accurately measuring attitudes in both older adults and younger people, such as adolescents, is vital in understanding the benefits of intergenerational programs.This is particularly important when comparing the effectiveness of various intergenerational programs that aim to tackle negative dementia attitudes (Farina, Griffiths, et al., 2020;Silverstein & Sherman, 2010).There are several limitations to this scoping review.First, there were relatively few studies that met this review's inclusion criteria.Whilst this highlights a need for further work in this area of the literature, we are limited as to how confidently we can apply these recommendations to other countries.Second, the psychometric properties of questionnaires were limited to those that were reported within the included studies.As such, some of conclusions may be limited by the quality of reporting.Lastly, this scoping review did not implement all the recommendations by Levac et al. (2010).Implementing the optional sixth stage (consulting) (Arksey & O'Malley, 2005) would have added methodological rigor and further sources of information to the review (Levac et al., 2010).However, there is no widely accepted consensus on how to approach consultation in scoping reviews.A review on scoping review methodology found that scoping reviews rarely report consultation exercises in meaningful detail.This may be attributed to power imbalances between researchers and stakeholder consultants, as well as ethical implications regarding whether stakeholder consulting is participatory research in itself (Buus et al., 2022).The consultation exercise was therefore considered beyond the scope of this review.

Conclusion
Whilst just over half of the studies exploring dementia-related attitudes in adolescents used validated measures, there is still a clear gap in terms of psychometric properties reported and the underlying theoretical framework.Measures that have been validated in other populations should take precedence over unvalidated measures since these types of measures have no psychometric support.To date, the brief A-ADS and KIDS have the most robust evidence of psychometric validity for measuring dementia related-attitudes in adolescents.

Table 1 .
Study characteristics of studies measuring dementia-related stigma and associated domains.

Table 2 .
Measure characteristics of validated measures in adolescents and children.

Table 3 .
Psychometric characteristics of the validated measures identified in eligible studies (<18-years-old).