The development and validation of the adolescent level of contact with dementia scale

As the number of people living with dementia increases, reducing stigma has become a policy priority. One way of decreasing stigma is through contact with the stigmatised group. However, the impact of this is difficult to establish due to a lack of validated measures suitable for adolescents. The aim of this study was to develop and validate a level of contact questionnaire designed to assess adolescents' contact with people living with dementia.


| BACKGROUND
Stigma towards dementia can come in many forms, from the use of negative and disempowering language 1 to the misconception that people with dementia cannot have a good quality of life. 2 Such stigma can lead to social isolation, reduced quality of life and loss of independence 3 ; whilst also being a major barrier for seeking and accessing support, diagnosis, treatment and information. 4,5 It is therefore unsurprising that reducing the stigma towards people with dementia is a key policy priority at a national (UK) and global level (Batsch and Mittelman; Department of Health; Scottish Government 3,6-8 ; particularly in light of the growing prevalence of dementia. 9 Negative attitudes and stigma towards dementia appear to exist in adolescents, 10,11 which is in line with the broader literature that negative attitudes towards mental illness form early during childhood. 12,13 It has been proposed that reducing stigma towards mental illness can be achieved through increased contact with the stigmatised group, education 14,15 and protest. Within the mental health literature there has been considerable debate as to whether education is enough to reduce stigma or whether education needs to be paired with contact with the discriminated group. 16 There have been a number of studies that have explored the impact of education and contact on mental health stigma amongst adolescents, but have produced contradictory results. 17,18 Within the field of dementia, there has been an increasing interest in the 'Human Rights based approach'. 19 This approach has tended to make use of contact based interventions to reduce discrimination against people living with dementia and increase positive opportunities for them. The approach views the stigma associated with dementia as a civil rights issue and assumed contact with those living with dementia will dispel erroneous myths and reduce the associated stigma. As such, researchers have attempted to capture the level of contact with people with mental illness, to increase understanding of its role in stigma towards mental illness. 20 In adolescents and adults more generally, there is a dearth of knowledge about what level of contact people have with dementia (and mental health). In part, this is due to a lack of validated measures to capture this level of contact, with researchers opting to use dichotomised descriptive outcomes instead. For example, 'Have you ever met a person with dementia?' 21 and 'Have you ever heard of dementia?' 22 However, whilst this is useful information, such measures do not capture the full breadth or frequency and the quality of contact that individuals may have with people with dementia due to their overly simplistic nature. As such, researchers have attempted to capture the level of contact with people with mental illness, to increase understanding of its role in stigma towards mental illness. 20 Understanding adolescents' level of contact with people with dementia will better enable us to understand how their experiences impact stigma towards dementia, and be more accurately able to identify whether we need to provide additional support to those that currently have contact with dementia. The aim of this study is therefore to develop and validate a novel level of contact questionnaire designed to assess adolescents' contact with those individual's living with dementia.

| METHOD
Ethical approvals were obtained through the Brighton and Sussex Medical School Research and Governance Ethics Committee.

| Scale development and testing
Adolescent level of contact of dementia (ALoCD) questionnaire was, in part, adapted from a previously developed measure of level of contact report of mental illness. 23 Since its initial creation, a number of The demographics of participants included in the exploratory factor analysis (n = 422)

Key points
• Negative and stigmatising views towards those living with dementia develop during adolescent years.
• One way of reducing the stigma associated with dementia is through contact with people living with dementia.
There is currently no validated tool that measures extent and quality of contact.
• The adolescent level of contact dementia scale (ALoCDs) was developed and tested with 934 adolescents aged between 13 and 18 years.
• The scale was found to be a valid and reliable tool enabling assessment of direct and indirect contact with dementia.
studies have adapted and utilised the measure in a number of samples, including adolescents. [24][25][26][27] Items from the existing level of contact questionnaire were altered to capture adolescents' level of contact with 'people living with dementia' rather than people with 'mental illness'. New items were drawn to reflect both direct contact with people with dementia, but also more passive contact with dementia (eg, through social media). The item response scale was converted from dichotomous to Likert. All items were independently reviewed and then discussed between the research team. The final items were reviewed by 10 adolescents and one secondary school teacher to ensure that the wording was appropriate, clear and meaningful.
The field version of the ALoCD questionnaire was composed of 11-items (see Table 2). The ALoCD requires participants to respond on a 5-point scale: 1 = Never, 2 = Rarely, 3 = Occasionally, 4 = A moderate amount, 5 = A great deal.
Demographic information (eg, age, gender and ethnicity) was also collected and is presented in Tables 1 and 4.

| Procedure
Nineteen schools were approached on the basis of previous interest in participating in dementia-related research, 27 of which five accepted to take part (26.3%). Head Teachers were provided an information sheet about the study, and consent was obtained. All Head Teachers acted 'in Loco parentis', however, they were also given the option to notify the parents or guardians about the research, giving them the opportunity to opt their child out of participation. A total of 947 students agreed to participate in the study.

| Data analysis
A total of 947 students participated in the study from five secondary schools. Due to the large sample size, missing data was not imputed but cases with missing data were deleted prior to analysis. To demonstrate discriminant validity between the direct and indirect scales, independent t tests were conducted between adolescents who lived with a person with dementia and those who did not. The scales would demonstrate discriminant validity if mean scores on the direct contact scale were higher for the group lived with somebody with dementia, and no significant difference in indirect contact.

| Study 1: Exploring the factor structure of the ALoCDs
To explore the factor structure of the ALoCDs, a sample of 446 adolescents was utilised. Sample descriptives are presented in Table 1. indicating good internal reliability. 33 The average inter-item correlation for factor 1 was 0.63 suggesting some items may be too closely The demographics of participants included in the confirmatory factor analysis (n = 488) related. The average inter-item correlation for factor 2 was 0.29 which is within the optimum range for good internal reliability. Item descriptives are presented in Table 2 and factor loadings are presented in Table 3.
To confirm the structure of the ALoCDs, a parallel analysis of 1000 data sets using a 95% cut-off was conducted as recommended by O'Connor (2000). Parallel analysis creates data sets with the same number of cases and variables as the actual dataset, filled with random numbers. An EFA is then performed on each data set, and any factors within the actual data set with eigenvalues that exceed those that emerge in 95% of the data sets of random numbers are defined as not having arisen due to chance variation within the data. The first five eigenvalues (and % of variance accounted for) extracted for 95% of the simulated data sets were equal to or less than 1.26 (1.14%), 1.19 (1.09%), 1.13 (1.03%), 1.08 (0.98%) and 0.99 (0.90%). In the actual data set, only the two eigenvalues of 4.70 which explained 42.68% of the variance, and 1.82 which explained 16.55% of the variance, exceeded chance values, suggesting that two factors underlie the data.
3.2 | Study 2: Testing the validity of the factor structure

| Construct validity
On a sample of 488 adolescents (sample description can be found in Table 4 and item descriptives can be found in Table 5), confirmatory factor analysis was used to test a two-factor model using maximum

| Internal reliability of new factors
Cronbach's alpha analysis produced a score of α = .89 for factor 1, and α = .62 for factor 2, indicating adequate levels of internal consistency for both factors. The average inter-item correlation for factor 1 was 0.62 and 0.28 for factor 2.

| Discriminant validity
Independent t tests were conducted to assess discriminant validity.

T A B L E 5
Item descriptives for items in ALoCDs post-confirmatory factor analysis

| CONCLUSION
While the scale has a number of practical uses, further tests of reliability and validity are warranted. Initial validation of the scale suggests that tool is adequately reliable and psychometrically sound, whilst acceptable for use with adolescents. A key strength of the scale is that it addresses an adolescent's level of contact on a continuum rather than a dichotomous variable. This enables an assessment of the extent of contact as opposed to simply establishing if there has been exposure. Although the study has led to the production of a psychometrically strong scale, it is worth noting that whilst the sample of adolescents recruited for this study was large (N = 947), the sample was relatively homogenous with regards to ethnicity and age.
The ALoCDs can be used by researchers, educators and practitioners working to improve dementia awareness. The initial validation suggests a psychometrically sound scale but further research needs to be undertaken to fully explore its practical uses and limitations.