Validation of ageism scale for dental students in India: (Ageism Scale for Dental Students- India) – a cross sectional study.

Study design

This was a cross-sectional analytical study carried out in Manipal College of Dental Sciences, Mangalore. The mode of conducting was through online platform conducted concurrently from March 2021 to June 2021.

Study population

Residents and final-year undergraduate dentistry students were the eligible participants for this study. The sample size for the present study was based on at least five individuals responding for each item while subjecting to factor analysis.¹² By implementing this method, the sample size would be 205. Following the universal sampling approach, the ageism scale for dental students was distributed to 250 final-year undergraduate dental students and residents after approval from the institution's head had been obtained in accordance with the inclusion criteria. The use of the universal sampling method eliminated the possibility of bias in the selection of participants. Finally, 213 students responded to the ageism scale. The method of self-reporting allowed for the identification of demographic information such as gender about the participants (Table 1).

Ethical considerations

Permission of Head of the Institution was taken. Institutional Ethics Committee clearance was obtained before commencement of the study (Protocol no. 21002, 9^th April 2021). Informed consent was electronically collected along with the responses to the ageism scale questionnaire,¹³ and participation was completely voluntary. Confidentiality of the participants was maintained.

Study instrument

Dental students' ageism scale: The initial version of the survey had a total of 27 items to measure ageism attitudes among dental students. A team of professors with experience in teaching geriatric dentistry from the United States and Europe created the original version.³ A 27-item ageism scale specifically tailored for assessing ageism among dental students, which considers the specificities related to geriatric dentistry. The original version was tailored to developed countries health set –up. This is measured on six-point Likert scale (“Strongly disagree” = 1; “Disagree” = 2; “Slightly disagree” = 3; “Slightly agree” = 4; “Agree” = 5; “Strongly agree” = 6). The total score value is directly proportion to the understanding of the individual about ageism.

Content validity by the subject experts

We asked six subject specialists with at least ten years of teaching experience and knowledge in working with senior citizens to take part in the study. None of these specialists were a part of the research team of this study. Through the online platform, the content was validated. The form's filling instructions was explained in detail through the information sheet.¹⁴ The specialists were asked to review each item critically and were asked to provide score for each item from 1 = “The item is not relevant to the measured domain”; 2 = “The item is somewhat relevant to the measured domain”; 3 = “The item is quite relevant to the measured domain”; 4 = “The item is highly relevant to the measured domain” and written comment to improve the relevance of items. Once they had fully provided the scores on all the items, the experts were requested to submit their comments to the researcher. We all agreed that “Old age home/Senior citizen house/Senior dwelling” should be used in place of “nursing home” for Question 8 and Question 24. Since none of the other items on the agism scale made reference to where the elderly resided, none of the phrases in those 25 remaining questions needed to be modified. This was then administered to the 213 final-year undergraduate dentistry students and residents who were open to taking part in an online platform. They received detailed directions¹⁴ on how to complete the electronic form.

Data handling and statistical analysis

Statistical Package for Social Sciences (SPSS) version 26.0 (SPSS, Inc., Chicago, IL, USA) was used to calculate the results based on the data obtained.¹⁵ Values obtained from the participants were computed to obtain descriptive and inferential statistics. A test of discriminant validity was conducted to look for differences based on gender, life history with senior people, year of study, and living in an urban or rural region. Mean and standard deviation was calculated for “age” and frequency with percentage was calculated for other demographic data. The Independent t test was used wherever applicable for the quantitative data obtained and participant responses which had missing data were excluded from the analysis. To measure item-scale correlation and inter item correlation Pearson correlation coefficient was used.

The reliability of the scale was evaluated using item analysis, internal consistency, and split-half test.

The goal of principal component analysis is to increase the sum of the squared loadings for each factor that is separately extracted. The loadings obtained from any other method of factoring would not explain as much variance as the principal component factor so, Principal component method of factor extraction used to carry out factor analysis for construct validity, structural validity, and internal consistency. In order to achieve what is referred to as “simple structure” in data, the varimax rotation approach is used. This method maximises the variance of the loadings inside each factor. Large loadings on a small number of variables describe the factors produced by the varimax rotations’ solution and increases the accuracy of construct validity, structural validity, and internal consistency while using the principal component method.¹⁶ The Kaiser-Meyer-Olkin (KMO) test and Bartlett’s sphericity test was used to check the adequacy of the factorial analysis. A KMO greater than 0.60 and Bartlett’s significance p < 0.05 were considered acceptable.

Eigen values, also known as latent root in Principal Component method is the sum of squared values of factor loadings relating to a factor/component.¹⁶ All components with Eigen values > 1.0 were initially extracted, and the number to keep was chosen using a combination of Scree Plot analysis and the overall variance that each component contributed. Items that (a) loaded poorly on any component or (b) cross-loaded across several components were removed as the Principal Component Analysis (PCA) was run iteratively. PCA was continued until the strongest and most parsimonious final set of items was reached. Cronbach's Alpha was used to assess the internal consistency of the final components, with a minimum acceptable value of 0.60 and an ideal value of 0.80. To identify items that were either possibly redundant due to multicollinearity (r|0.80|) or not meaningfully associated to any other item (max r|0.30|), correlation matrix of all 27 items was evaluated. P < 0.05 was considered to be the level of significance. The content validity of the scale was analysed using Content Validity Index (CVI).¹⁷

Calculating CVI¹⁷

There are two types of CVI: scale-based CVI and item-based CVI (S-CVI). There are two ways to calculate S-CVI: the proportion of items on the scale that receive a relevance value of 3 or 4 from all experts (S-CVI/UA), and the average of the I-CVI scores for all items on the scale (S-CVI/Ave).⁶ The following table summarises the formula and definition of the CVI indices.

The definition and formula of I-CVI, S-CVI/Ave and S-CVI/UA.

The above-mentioned calculation led to the conclusion that I-CVI, S-CVI/Ave, and S-CVI/UA meet satisfactory levels, indicating that the scale of questionnaire’s content validity has been attained to a satisfactory level. When writing the report, this study used the STROBE checklist.

PCA results

First, all items were coded so that a higher value indicated more Ageism, so ten items were reverse-coded where 1 = “Strongly Agree” and 6 = “Strongly Disagree”. We indicate in our results whenever one of these items are mentioned that it was reverse coded for clarity.

There were no items that posed a risk of biasing the PCA results due to a multicollinear or insignificant relationship with other items. The factorability of the data was confirmed with KMO = 0.61 and Bartlett’s Test of Sphericity resulting in p < 0.001. The final PCA model resulted in 15 items and six components that together accounted for 70.37% of overall variance (Table 2). The six components had reliability ranging from marginal 0.51 (Component 6) to a high of 0.81 (Component 3). While not ideal, the small number of items on each scale, which directly impacts Cronbach’s Alpha, may be the cause of these coefficients.¹⁸

Statistically significant findings are highlighted in the tables for easy identification. Gender differences by component shows a statistically significant difference in scores on Component 4 where males (M = 6.93, SD = 2.08) scored 0.91 points lower than females (M = 7.84, SD = 1.94), p = 0.02 and Component 6 where Males (M = 5.93, SD = 1.66) scored 1.43 points higher on average, (95% CI = 0.84 – 2.03) than females (M = 4.5, SD = 1.5), p < 0.001 (Table 3). Overall, there was one missing data when testing the gender differences and also 3 missing data for component 1, 2 and 5.

Results of independent t- test for history of living with elderly based on the components had no statistically significant results (Table 4). There was one missing data in this comparison and 3 missing data for component 1, 2 and 5. While comparing differences by component for year of study statistically there was significant difference in component one which was barriers/concerns in dental treatment of elderly. The residents (M = 11.56, SD = 3.89) and final year students (M = 10.16, SD = 3.26) had a mean difference of 1.4 (95% CI = 0.41 – 2.38) indicating residents had more barriers/concerns in treating the elderly (p = 0.06) and there were 3 missing data for component 1, 2 and 5.

The students residing in urban areas (M = 7.86, SD = 1.86) felt that it was more time consuming to treat the elderly than when compared to those residing in not urban areas (M = 7.07, SD = 2.32) where the mean difference was 0.79 (p= 0.04) (95% CI = -1.57 - -0.02). Components 3, 4, and 6 each had one missing data, while 1, 2, and 5 each had three. All of the missing data entries were excluded for component-based analysis. Regardless of statistical significance, it is very important to note that the actual differences between groups are quite small, so results must be interpreted with caution.

Component 1: barriers/concerns in dental treatment of elderly

There were 4 items from the agism scale under component 1 which are 16, 17, 8 and 19 in descending order of factor loading. Item 16 was also seen in the dual institution validation study in USA version of agism scale under the component “Preconceived notions about dental treatment”,¹⁰ under “Values/ethics about older people” in the Greek version¹⁹ and under the same category as this study in the German version,²⁰ in component 2 (“education around older patients care”) of the French version²² and factor 1 (“Negative view of older adults’ life and dental treatment”) of the Serbian version.²³ The Q17 was “whether elderly patients make it long enough to make investment worthwhile” and was seen in the US version under “Preconceived notions about dental treatment”,¹⁰ under the component “negative view of older patients” in Brazil version,²¹ under “Values/ethics about older people” in the Greek version,¹⁹ component 2 (“education around older patients care”) of the French version²² and factor 1 (“Negative view of older adults’ life and dental treatment”) of the Serbian version.²³

‘If the elderly are better off in old age homes?’ was the eighth question in the agism scale and it was seen under the components “negative view of older patients” in Brazil version,²¹ under “Values/ethics about older people” in the Greek version¹⁹ and “Patient compliance” factor in German version.²⁰ The last item in this component in this study was Q19 which was seen in complexity of providing care for older adults” in Brazilian version,²¹ under “Barriers to dental care” factor in Greek version,¹⁹ “Practitioner’s perspective” in German version²⁰ and factor 1 (“Negative view of older adults’ life and dental treatment”) of the Serbian version.²³

Component 2: patient compliance

This component had 3 items; 5, 6 and 9 where all three were reverse-coded so that higher score indicated more agism. The Q5 which asked if they pay more attention to their elderly patients was also seen in the Brazilian version of agism scale under the component “Positive view of older adults”,²¹ under “Dentist–older patient interaction” in the Greek version,¹⁹ component “Perceptions of the older and their value in society” in the Romanian version⁹ and “Dental care in younger and older patients” in the Serbian version.²³ The question 6 was seen in component “Positive view of older adults” in Brazilian version,²¹ under “Dentist–older patient interaction” in the Greek version,¹⁹ under “Opinions about elderly” in the German version,¹⁷ “Perceptions of the older and their value in society” in the Romanian version⁹ and “Dental care in younger and older patients” in the Serbian version.²³ The item 9 which was regarding elderly patients being more appreciative of the care provided was seen in the component Dentist–older patient interaction” in the Greek version¹⁹ and “Patient compliance” factor in German version.²¹

Component 3: opinions of elderly

The questions 13 and 14, were under this component. They were also seen in the component “Patient compliance” in USA version¹⁰ and Greek version,¹⁹ under “negative view of older patients” in the Brazil version,²¹ “general negative view” in the French version²² and “ethical values about older people and patient compliance” in the Serbian version.²³

Component 4: time consuming

The Q2 which was regarding the opinion that taking medical history from elderly patients was time consuming and Q3 being that taking medical history was complex were present in this component of our study, in the Brazilian version of agism scale under “complexity of providing care for older adults”²¹ and “difficulties in medical history taking” in the Serbian version.²³

Component 5: treatment success rate

The items 12 and 21 were grouped in the ‘Treatment success rate’ component. In the German version Q12 was seen under the component “Opinions about elderly” and Q21 was seen in “Barriers/concerns on dental treatment in elderly”²⁰ and under “Exposure to geriatric dental training and experiences” in the Romanian version.⁹

Component 6: practitioner perspective

The questions 24 and 25 in this component were reversed so that higher score indicated more agism. Q24 was not seen in any of the versions of the agism scale and Q25 was seen Under “Barriers to dental care” in the Greek version,¹⁹ in “education around older patients care” of the French version²² and under “barriers to dental treatment” in the Serbian version.²³

As seen above, questions 13, 14, 16 and 17 are included in the dual institution validation of agism scale carried out in USA¹⁰ but under different components than seen in our study. In the Greek version questions 5, 6, 8, 13, 14, 16, 17, 19 and 25 were included in the scale where Q19 is under the same factor “Barriers/concerns”.¹⁹ The questions that were included in the German version were 6, 8, 9, 12, 16, 19 and 21 among which Q9 and Q16 were under same components of “Patient Compliance” and “Barriers/Concerns”.²⁰ In the Brazilian versions the similar questions included were 2, 3, 5, 6, 8, 13, 14, 17 and 19 but none of the components were similar.²¹ Only three questions i.e., 5, 6 and 21 were similar in the Romanian version of agism scale.⁹ The following five questions were present in the French version: Q16, 17, 13, 14, and 25.²² Highest similarity was seen with the Serbian version as the questions 2,3,5,6,8,13,14,16,17 and 25 were included in their version of agism scale and Q8, 16 and 17 were under the component “Negative view of older adults’ life and dental treatment”.²³ In this study it was included in “barriers/concerns in dental treatment of elderly” indicating that though named differently they point towards the negative aspect of treating an elderly patient. Although the results of this study are comparable to those of other studies on the ageism scale for a diverse nation like ours, generalizability is constrained.

Limitations

Since this study is done on a single institution the generalizability of its result is limited. In-spite of being statistically significant the differences are small and interpretation must be done cautiously. Similar studies of validation should be carried out in different geographical and cultural areas which can help in constructing a single scale for our country. The paper could offer suggestions for future research, such as exploring the relationship between ageism and the quality of dental care provided to older adults to reduce ageism among dental students.