Validating a Nutrition Knowledge Questionnaire (QuesCa) for Arabic-Speaking Children Aged 9-15 Years: Translation and Cultural Adaptation

Ethics

This study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving research study participants were approved by the Ethical Committee of the National Institute of Nutrition of Tunis (approval number: 15/2022), approved on 17^th June 2022. Written informed consent was obtained from all subjects.

Participants

The study aimed to assess the nutritional knowledge of Tunisian children aged 9 to 15 years old. The sample size of 600 participants was determined based on an alpha error of 5% and a fixed precision of 4%. Since there were no prior studies on this topic in Tunisian children, the prevalence (P) was set at 0.5. The chosen sample size fell within the recommended range for a questionnaire with 10 items.

Data collection took place over a 6-month period from January to June 2022. The study was conducted at two centers: the National Center for School and University Medicine in Tunis (NCSUM), which serves as a screening and health care center for students with specific health issues, and randomly selected school nurseries in Tunis.

To be eligible for participation, children had to be aged between 9 and 15 years and not have used medications affecting food intake or appetite. Those with chronic illnesses requiring specific nutritional care, such as diabetes or renal insufficiency, were excluded. Incompletely or incorrectly filled-out questionnaires were also excluded.

The participants were divided into two groups based on age: children aged 9-11 years (GR1) and pre-adolescents and adolescents aged 12-15 years (GR2) (flowchart is present in Figure 1). This division was made considering the influence of family customs on younger children and the impact of media and online information on older ones. It also took into account the intellectual maturity of the students, and the overall knowledge level of the two groups was similar to a previous study by Kruseman et al.,⁹ which was used as a reference for developing and validating the questionnaire.

Figure 1. Flowchart of the study recruitment process.

The study involved two age groups. The first group consisted of children aged 9-11 years, recruited from NCSUM and local nurseries. Incomplete responses were excluded, and parental assistance was sought when necessary. The remaining participants formed the final sample for this age group. The second group consisted of adolescents aged 12-15 years, with the same exclusion criteria applied. The flowchart illustrates the process from recruitment to the final sample selection.

A total of 524 children were invited to answer the QuesCA AR. The response rate was 90% for GR 1 and 83.6% for GR 2.

Translation, adaptation and pre-test

The development of our Arabic versions followed several steps outlined in Figure 2.

Figure 2. Flowchart detailing the process of translating and validating questionnaires from French into literary Arabic and vice versa.

The process involves various steps including expert consultation, focus groups, pre-testing and analysis.

For the first stage of the elaboration and validation process, two independent translations were made from French into literary Arabic of QuesCA 9-11 and QuesCA12-15. One by a committee of experts in the field of nutrition working at the Zouheir Kallel National Institute of Nutrition and Food Technology. It is made up of an associate professor of nutrition, 4 doctors specialized in nutrition, a dietician and an associate professor of preventive medicine. All team members were bilingual, native speakers of Arabic and had a good level of French. This committee constitutes the « focus group ». The other translation was done by a sworn translator. After being translated, the two questionnaires obtained were first revised by the focus group in order to identify translation errors and compare the two translated versions and the discrepancies between them, and to associate the Tunisian specificities with them. This step also aims to ensure the veracity and semantic equivalence of the different translated items in relation to the original version. A retrograde translation of the questionnaires from Arabic to French was carried out by a French national whose mother tongue is French and who has lived for more than twenty years in Tunisia (and is therefore familiar with Tunisian habits and culture) and who has a good command of literary Arabic. Based on the Delphi procedure, a comparison of the two questionnaires obtained with the two corresponding original versions was made by the experts of the first stage by pointing out the items that were not correctly translated by the back-translation and this was done in writing via a questionnaire to be completed by each expert individually. For each problematic item, each expert had to choose between keeping the initial version or modifying it and, if necessary, proposing a new one. Based on the results, the experts met and agreed on the most appropriate translations and modifications were made accordingly. The words “salmon”, “green beans”, “rice cake”, “non-light iced tea” are foreign to Tunisian eating habits. The fifth stage of the validation process consisted of a pre-test using the final Arabic versions. Before this stage, however, a primary and secondary school teacher was asked for her opinion on the two Arabic versions obtained in order to detect whether there were any difficulties in understanding the questionnaire items in literary Arabic and to detect whether there were any linguistic errors. Then, in a primary school in the governorate of Gabes, 30 students aged 9-11 years and 30 students aged 12-15 years who were enrolled in grades 3, 4, 5 and 6 were randomly selected.¹⁰ The questionnaires were administered to the children according to their age after obtaining their parents’ consent. Each participant completed the questionnaire individually. The time necessary to complete the questionnaire was evaluated as about 1–3 min. The aim of this step was to adapt the comprehensibility and difficulty level of the questionnaire, to improve the wording of the items and to estimate the time needed to complete the questionnaire. Finally, two final translated versions were obtained; the QuesCA 9-11 AR and the QuesCA 12-15 AR which were produced by the research team on the basis of the development and revision process and they are available on: DOI: 10.5281/zenodo.10702366.

The QuesCA AR questionnaire

The QuesCA AR questionnaire comprised 9 questions, only one of which was open question (question 3 of the QuesCA 12-15); the other eight questions can be categorized as follows: (i) those with a true/false answer (questions 1-9 of the QuesCA 9-11 and questions 1-4-5-6-9 of the QuesCA 12-15); (ii) multiple choice question with one correct answer (question 2 of the QuesCA 12-15); and (iii) multiple-choice questions with more than one correct answer question (question 7-8 of the QuesCA 12-15).

For each question the ‘I don’t know’ answer was always present.

The correct response rate was calculated. The thresholds of 25% and 75% correct answers were used to qualify the questions as “difficult” or “easy”. All calculations were done separately for each age category.

For the nine questions, all of the possible answers and the correct answer to each question, see Extended data.²²

The questionnaires were completed entirely anonymously. The procedure adopted was as follows: a value of 1 was assigned to each correct answer, while a value of 0 was assigned to each incorrect or “don’t know” answer. Then mean scores were calculated.

Interpretation of the scores was done globally by theme. Thresholds were defined to evaluate the children’s responses:

Statistical analysis

Data were analyzed using SPSS statistic 26.0.0 available on this link. Descriptive statistics were computed for all variables, including mean and standard deviation for continuous variables and frequency and percentage for categorical variables. The normality of the distribution of continuous variables was assessed using the Shapiro-Wilk test. Since the data were normally distributed, parametric tests were used.

Exploratory Factor Analysis (EFA) was conducted to explore the dimensionality of the QuesCa Arabic version. The principal component analysis (PCA) with varimax rotation method was used. Items with a factor loading of 0.4 or higher were considered significant and included in the subscales. The internal consistency reliability of the unidimensional scale and its subscales was assessed using Cronbach’s alpha coefficient. Values of 0.7 or higher were considered acceptable.

We used logistic regression modelling to determine independents socio-demographic correlates of good responses to each QuesCa.

The independent variables tested were gender and studying in a public or private school.

We used the forward stepwise logistic regression, the “Wald” type. The maximum number of iterations to arrive at the most parsimonious model was set at 100.

Population characteristics

Group 1, comprising individuals aged 9 to 11 years, accounted for a total of 142 participants, with 45% of them identifying as female. Additionally, 3.5% of participants from this group resided in rural areas, and 15% attended private schools. Regarding obesity prevalence, 8% of individuals in Group 1 were classified as obese. In Group 2, encompassing individuals aged 12 to 15 years, there were 97 participants, with 46.4% being female. Notably, a higher proportion of participants (19%) from Group 2 were from rural areas, while only 4% attended private schools. Obesity prevalence was relatively higher in Group 2, with 13% of participants being classified as obese.

Reliability analysis

The overall Cronbach’s alpha of the QuesCA 9-11 AR was 0.336 and for the QuesCA 12-15 AR it was 0.284 for all 9 items despite the many possible combinations tried. The Cronbach’s alpha of the questionnaire was recalculated after eliminating the items with low correlations. Despite these modifications, the Cronbach’s alpha value remained below the threshold (0.7). Therefore, it was decided to keep the questionnaire as it is with the same number of items.

Descriptive results

The results per question are presented in Figure 3.

Figure 3. A bar graph comparing the percentage of correct answers for each item in the QuesCA 9-11 AR and QuesCA 12-15 AR questionnaires.

The graph provides a visual representation of performance across these items.

On a maximum possible score of 9 points, scores ranged from 0 to 9/9. Only 7 children (2.2%) answered all items in the questionnaire correctly and had a score of 9/9. Furthermore, only one out of the 315 respondents had a score of 0/9. The average score was 5.77. The median score was 6. It was concluded that items 4, 5, and 9, which corresponded to the themes of protein, fruits and vegetables, and beverages, respectively, were the easiest to answer since they had a response rate of > 75%. On the other hand, the topic of starchy foods (item 3) was the most difficult question for children aged 9 to 11.

A poor score (≤ 3) was obtained by 8.6% of children and only 30.2% had a good score (≥ 7).

No respondent was able to answer all items correctly and scores ranged from 0 to 8/9. Only 2 respondents had a score of 8/9. Furthermore, only one respondent had a score of 0/9. The average score was 5.14. The median score was 5. These results show that items 1, 3, 5, 6, and 9, which corresponded respectively to breakfast, beverages, protein, fruits and vegetables, and sweets, were the easiest to answer since they had a response rate of > 75%.On the other hand, children in this age group had difficulty answering the following themes: balanced diet (item 2), starchy foods (item 4), calcium-rich foods (item 7), and fats (item 8). These items were the most difficult to answer and had a correct response rate of less than 25%.

Only 8% of children had a poor score (≤3), 79% had an average score (between 4 and 6) and only 13% had a good score (≥7).

In the multivariate analysis examining the factors influencing scores obtained in the QuesCA 12-15 AR questionnaire, several variables were considered. Regarding gender, the odds ratio (OR) was found to be 0.323, with a 95% confidence interval ranging from 0.435 to 1.080. However, the associated p-value was 0.404, suggesting that gender did not have a statistically significant influence on the scores. The odds ratio for attending a private school was 1.810, with a 95% confidence interval spanning from 0.24 to 3.77. The associated p-value was 0.023, indicating a statistically significant difference in scores between students from public and private schools.

Strengths and limitations

Our study aimed to fill the gap in the literature by examining the nutritional knowledge of Arabic children, a topic that had not been explored before. The results demonstrate the importance of our QuesCA 9-11 AR and QuesCA 12-15 AR as valuable tools for assessing nutritional knowledge in Arabic-speaking children. A notable strength of our study is that we conducted a reliability analysis by age group, a step that had not been undertaken in previous Swiss and Italian studies. This allowed us to obtain the first two Arabic versions of QuesCA 9-11 and QuesCA 12-15, making them accessible to other Arab countries.

However, it is essential to acknowledge the significant limitations we encountered in our study. The primary focus was to validate the Arabic versions of QuesCA, but we faced challenges in proving the reliability and validity of the questionnaires. The multidimensional nature of the versions, their brevity, and their target audience of children may have contributed to these limitations.

We considered content validity during the qualitative study stages through the input of the focus group, but we did not assess construct validity, as each theme in the questionnaires was represented by a single item.

In our analysis of internal consistency, we used the global Cronbach’s alpha despite the multidimensional nature of the questionnaires. This approach reflects the consistency of responses obtained for each item. As each theme in QuesCA 9-11 AR and QuesCA 12-15 AR comprises a single item, we were unable to calculate the Cronbach’s alpha by theme. In the Swiss study, a global Cronbach’s alpha of 0.66 was found for all 30 items, significantly better than the alpha values calculated for individual themes (ranging from 0.06 to 0.45). The use of the global Cronbach’s alpha allowed us to compare our results with other studies evaluating children’s nutrition knowledge using similar methods.^9,11–14

It is crucial to remember that the Cronbach’s alpha coefficient is not an absolute measure of questionnaire reliability. Its value indicates the extent to which a questionnaire is reliable for a specific population at a given moment. Thus, reliability should be estimated each time the questionnaire is administered, including during pilot tests and later validation stages.

Therefore, the two optimal Swiss scales, QuesCA 9-11 and QuesCA 12-15, served as the basis for our study. It is important to note that reliability analysis by age group was not conducted during the Swiss or Italian studies. Even in the Italian study, Tallarini et al.⁵ worked on two groups, GR1 (ages 9-11) and GR2 (ages 12-16), but they used the long version of the 30-item QuesCA to obtain their Italian version, QuesCA IT, which was used for both age groups, and no reliability analysis was conducted during their study.⁵ It is important to note that the number of items in a test, the interrelationships between its items, and its uni- or multidimensional nature all affect the value of alpha.^10,15,16 Indeed, the alpha coefficient increases with an increase in the number of items in a questionnaire.^15,16 Therefore, the shorter the questionnaire is, the lower the alpha value is. This partly explains the low values of Cronbach’s alpha obtained during our study. On the other hand, the global Cronbach’s alpha calculated during the Swiss study for all 30 items was better (0.66). It therefore seems logical to obtain similar results for questionnaires with only 9 items each. Furthermore, a weak correlation between items underestimates reliability.^10,15,16

Our study not only addressed specific nutrition-related issues, such as fruit and vegetable consumption¹⁷ and sugary and unhealthy food habits,¹⁸ but also emphasized the ability to select balanced meals from various options. Equipping young people with a clear understanding of what constitutes a healthy diet is crucial in helping them discern between safe and unhealthy nutrition messages in media.

The analysis of our obtained data revealed some important insights. The item with the lowest score in both age groups was ‘starchy foods,’ followed by ‘balanced diet,’ ‘calcium-rich foods,’ and ‘fats.’ These results suggest that more efforts are needed to improve the nutritional knowledge of students regarding the exact role of macronutrients protein and fat in a healthy diet.

Regarding the item ‘healthy diet,’ statistically significant differences were found, likely due to the presence of a complex question with a multiple-choice answer, assuming students have precise knowledge of all nutrients and how to combine them for a balanced diet.

We also observed a significant association between the level of nutritional knowledge and the type of institution among the 12-15 age group. Enrollments in private schools were indicative of a higher socio-economic status, although they followed the same curriculum. This aligns with the findings of the NUTRI-KAP study,¹⁹ which demonstrated the impact of economic inequality on nutritional knowledge, attitudes, and practices in Iranian households. Socioeconomic status has been linked to nutritional knowledge in various studies, highlighting the need to address disparities in nutrition education.²⁰

Overall, our findings suggest that nutritional education in North Africa requires improvement. Teachers should adopt multidisciplinary approaches and receive training to enhance the effectiveness of nutritional education. Additionally, health professionals and parents play crucial roles in promoting nutritional education among children. Involving parents through interfamily activities and nutrition programs offered in schools or dispensaries can contribute to fostering healthier eating habits among youth.²¹