Development of Dietary Knowledge and Adherence Questionnaires for Lebanese Adolescents and Their Parents

The availability of practical tools to assess dietary knowledge and adherence is essential to evaluate the effectiveness of dietary interventions. The aims of this paper were to develop reliable dietary knowledge and adherence questionnaires, suitable for Lebanese adolescents and their parents, and to estimate the feasibility of conducting studies involving such participants in the school-based setting. Eight Lebanese high schools participated in this study (involving 220 adolescents aged 15–18 years). Self-administered dietary knowledge and adherence questionnaires (the Dietary Knowledge Questionnaire (DKQ) and the Dietary Adherence Questionnaire (DAQ), respectively) were completed by the high school students and their parents. A 24 h recall was additionally administered for the adolescents by a dietitian and a trained interviewer at school, in order to validate the adolescents’ answers in the DAQ. The cognitive interview method was used to qualitatively evaluate the questionnaires. The resulting Cronbach’s alpha ranged from 0.61 to 0.78 for the adolescent questionnaires and from 0.46 to 0.89 for the parental ones. In addition, 23 items (out of 25) of the adolescent DAQ matched with the administered 24 h recall. A significant negative correlation was found between the knowledge score (DKQ) and the unhealthy items of the adolescent DAQ. There was a significant positive correlation between the DKQ of the parents and the knowledge score of their children. This is the first study of dietary questionnaires involving Lebanese high school students from different regions, while also including their parents.


Introduction
Pediatric obesity is the most common childhood nutritional disorder in the world [1]. This global epidemic is the cause of many physical complications, such as diabetes mellitus, metabolic syndrome, dyslipidemia, cardiovascular diseases, obstructive sleep apnea and hypertension, as well as psychosocial complications such as low self-esteem, higher rates of anxiety disorders, decreased health-related quality of life (QOL) and decreased educational and financial attainment [2,3]. These health consequences constitute a serious economic burden. The incremental lifetime direct costs from the perspective of a 10-year-old obese child relative to a normal weight child reaches an average of $16,000 [4]. Obesity is preventable, however [5]. According to the World Health Organization (WHO) [5], over 340 million children and adolescents aged 5-19 years were overweight or obese in self-administered questionnaires present no interviewer-related bias, are time-saving, and permit more careful responding (since respondents are not hurried) [23]. Questionnaires can be a feasible, reliable and valid method to evaluate nutrition knowledge and the correlation between dietary awareness and eating behavior [24].
There is a paucity of validated dietary knowledge and adherence questionnaires for Lebanese adolescents. Previous dietary knowledge questionnaires were often designed specifically for a particular study, and lacked psychometric validation [16]. In addition, the few available studies involving Lebanese youth did not describe the details of the process for developing such questionnaires. In addition, despite the difference between urban and rural regions as described above, previous studies often did not include participants living in rural regions [25]. This highlights the need to develop reliable and valid, user-friendly and inexpensive tools to assess levels of dietary knowledge and adherence among Lebanese adolescents living in urban and rural areas.
The current study assessed the psychometric properties of new instruments to measure the dietary knowledge and adherence to the nutritional guidelines of Lebanese adolescents. The aims of this paper are to describe the step-by-step process of developing these questionnaires for Lebanese adolescents living in urban and rural areas and their parents, to analyze their feasibility and internal reliability in the target population, and to examine the correlation between parental and adolescent scores.

Materials and Methods
We developed two questionnaires: The Dietary Knowledge Questionnaire (DKQ), to assess the level of dietary knowledge of Lebanese adolescents and their parents, and the Dietary Adherence Questionnaire (DAQ), to assess the level of adherence to nutritional guidelines [12]. Both questionnaires were developed following the same steps (see Figure 1). Int. J. Environ. Res. Public Health 2020, 17, x 3 of 32 a feasible, reliable and valid method to evaluate nutrition knowledge and the correlation between dietary awareness and eating behavior [24]. There is a paucity of validated dietary knowledge and adherence questionnaires for Lebanese adolescents. Previous dietary knowledge questionnaires were often designed specifically for a particular study, and lacked psychometric validation [16]. In addition, the few available studies involving Lebanese youth did not describe the details of the process for developing such questionnaires. In addition, despite the difference between urban and rural regions as described above, previous studies often did not include participants living in rural regions [25]. This highlights the need to develop reliable and valid, user-friendly and inexpensive tools to assess levels of dietary knowledge and adherence among Lebanese adolescents living in urban and rural areas.
The current study assessed the psychometric properties of new instruments to measure the dietary knowledge and adherence to the nutritional guidelines of Lebanese adolescents. The aims of this paper are to describe the step-by-step process of developing these questionnaires for Lebanese adolescents living in urban and rural areas and their parents, to analyze their feasibility and internal reliability in the target population, and to examine the correlation between parental and adolescent scores.

Materials and Methods
We developed two questionnaires: The Dietary Knowledge Questionnaire (DKQ), to assess the level of dietary knowledge of Lebanese adolescents and their parents, and the Dietary Adherence Questionnaire (DAQ), to assess the level of adherence to nutritional guidelines [12]. Both questionnaires were developed following the same steps (see Figure 1).

Phase 1-Literature Review
First, a thorough literature review of dietary knowledge, general dietary recommendations for adolescents and dietary behaviors linked to overweight and obesity, was done to identify the main parts of the questionnaires. Second, the format of both questionnaires was determined (e.g., type of questions, method of administration, duration, language, etc.). Both questionnaires are self-

Phase 1-Literature Review
First, a thorough literature review of dietary knowledge, general dietary recommendations for adolescents and dietary behaviors linked to overweight and obesity, was done to identify the main parts of the questionnaires. Second, the format of both questionnaires was determined (e.g., type of questions, method of administration, duration, language, etc.). Both questionnaires are self-administered structured ones with closed questions. All questions were designed to suit both target groups (adolescents and adults) and participants with different socioeconomic backgrounds.
Particular attention was paid to the first few questions, as they influence the respondent's attitude and cooperation [26]. The first page of both questionnaires included statements summarizing the purpose of the study and confirming the consent to participate once the questionnaire is completed and submitted.
Both questionnaires were administered during classes and in Arabic. They were designed to be convenient for Lebanese adolescents and their parents. It should be possible for 20 to 35 students to fill out the questionnaires during a class session lasting 40-50 min. This explains the reason behind choosing a self-administered format, closed questions and a relatively small number of answer options. To make sure that the selected items suited the Lebanese population and culture, some were taken from existing questionnaires (see below), while others were added or adapted to the Lebanese eating culture.
The third step involves translation and back translation. Both questionnaires were originally developed in English by a dietitian with practical experience in nutrition in Lebanon, then translated into Arabic, and back translated to English. The initial translation and back-translation were done by two different individuals. To avoid bias, the back-translator was not aware of the concepts the questionnaires intend to measure [27].

Dietary Knowledge Questionnaire
The Dietary Knowledge Questionnaire (DKQ) aims to assess the level of dietary knowledge of Lebanese adolescents (see Supplementary Materials S1 and S2), and their parents (see Supplementary Materials S3 and S4). It was based on the Nutrition Knowledge Questionnaire (NKQ) developed by Parmenter and Wardle [17], which meets the criterion for construct validity, and has a total Cronbach's alpha of 0.97, reflecting a high internal reliability. It was adapted to suit the Lebanese culture: for instance, Lebanese food items, such as Akawi cheese and labneh, replaced foods less common in Lebanon, like edam cheese. The selected items were meant to be representative of a broad range of dietary knowledge [17]. The questions ask about very basic and general nutrition information to assess whether the participants have a minimum level of knowledge, allowing them to make decisions regarding their food intake.
The DKQ consists of five parts and a total of 56 items (including questions and sub-questions) and takes 15-20 min to fill out. Some questions are divided into additional sub-questions, making 40 sub-questions in total. Each question has three to five answer options, and only one correct answer per question/sub-question. A "do not know" option was added to all questions.
The five parts are: (1) Knowledge of dietary recommendations, to determine if participants are familiar with the current ones; (2) knowledge of nutrient sources, to determine whether participants know if the indicated food items are good or bad sources of the nutrient in question; (3) knowledge of common misconceptions in nutrition, to determine whether participants are aware of some of those related to nutrition; (4) using knowledge of nutrition to make dietary choices, to determine whether participants are able to choose the healthiest option when given different food choices; (5) knowledge of associations between nutrition and diseases, to determine if the participants are aware of the correlation between eating habits and health. The original NKQ questionnaire included the same parts [17], but the "understanding of terms" part was renamed "knowledge of common misconceptions in nutrition", although the questions remained the same. As more than 92.7% of young adults relies on online resources to obtain nutrition information and only 4.7% refers to healthcare professionals, it is important to verify if the received information is correct [28].
Answers were coded as follows: zero for the wrong answer or if the participant did not know the answer, and one point for the correct answer.
The weighting of the various items is indicated in Appendix A. The total knowledge score was calculated by summing up the points on all items. Thus, the maximum possible score is 56 and the minimum is zero. The same DKQ was administered to both the parents and their children.

Dietary Adherence Questionnaire
The Dietary Adherence Questionnaire (DAQ) (see Supplementary Materials S5 and S6) was based on the Schools Physical Activity and Nutrition Survey (SPANS) years 8 and 10 student questionnaire [28].
The questions and answer options were modified to lower the burden for participants by making them shorter, to avoid response fatigue and inaccurate results [29]. It is a self-administered questionnaire with close-ended questions, and takes about 20 min to complete. It consists of 30 questions and is divided into four parts: (1) food choices, related to the consumption of specific food groups like lean meats, sweets, etc.; (2) eating habits, related to meal pattern consumption such as the number of meals, snacks and breakfast consumption; (3) physical activity (PA) and screen-viewing time; (4) smoking. The number of options per question ranges from two to six.
Even though the DAQ included one question regarding smoking and four related to engaging in PA and screen time, it was still called the Dietary Adherence Questionnaire because those questions were added to examine the correlation of diet with smoking and PA to identify potentially clustered behaviors. Based on a cross-sectional study of Brazilian adolescents, a significant positive correlation was found between smoking and a high intake of sweets, low fruit intake and high intake of soft drinks [30]. The questions regarding PA and screen-viewing time were added because studies have shown that active individuals tend to have healthier diets [31]. The questions and scoring system were identical for both the parents and their children, except for the question about PA in school, which was omitted as irrelevant from the parents' DAQ (see Supplementary Materials S7 and S8).
This questionnaire was designed to assess each question separately by checking, for instance, whether the participant eats high fat meat or lean meat, or eats breakfast or not. However, to facilitate the evaluation of the overall dietary behavior, instead of looking at each question separately, the total adherence index was created. Therefore, it presents a useful, user-friendly and rapid tool for health promotors to detect unhealthy eating behavior, potentially leading to overweight or obesity and correcting it. It also allows us to compare the answers before and after an intervention.
Scoring was zero for "No, I did not eat any of the foods listed above yesterday", one point for "Yes, I ate it once", two points for "Yes, I ate it twice", and three points for "Yes, I ate it three times or more". In the question related to breakfast consumption, the option "yes" was equal to two points, while "no" was equal to zero, as breakfast cannot be consumed more than once a day, as well as to emphasize the importance of this meal. In questions 26 and 27, related to PA practice, "yes" was equal to one point and "no" to zero. The weekly duration of PA practiced was coded as follows: zero for "0 h", one point for "1 h", two points for "2-4 h" and three points for "more than 5 h". As for smoking, "yes" was equal to two points, and "no/stopped smoking" was equal to one point. The question related to screen-viewing time (duration spent watching television, using phones or tablets, or playing video games) was removed to improve reliability (increased Cronbach's alpha).
Next, to simplify the use of this questionnaire in interventional studies, and to be able to assess the overall adherence to dietary guidelines instead of looking at each question separately, a total adherence index was developed by dividing the score of healthy items by the score of unhealthy items: Total adherence index = Healthy items score/Unhealthy items score If the score is above 1, then the healthy food choices and habits outweigh the unhealthy ones, and if the score is below 1, there are more unhealthy choices compared to healthy choices (see Table 1 and Appendix B). All individuals started with one point on the unhealthy score to avoid zero as a denominator. The maximum possible index score is 37 (reflecting a total adherence to dietary guidelines) and the minimum score is zero. The items related to the number of snacks and juices consumed were removed from the calculation of the index score, as it is hard to distinguish whether the snack itself is healthy or unhealthy. For instance, an apple is not the same as soft drinks. The same goes for natural juices: to avoid controversy about whether to consider the item healthy or unhealthy, it was not included in the index. Thus, a maximum of 37 points for the healthy items and a maximum of 38 points for unhealthy items is obtained. It is important to note that the DAQ does not investigate the reasons behind the participants' actions and attitudes.

Phase 2-Cognitive Interview
The cognitive interview method [32] was selected to test the clarity and interpretation of the Arabic version of the questionnaires. According to Miller [33], large sample sizes are not necessary at this stage because problems in questionnaires are supposed to be revealed immediately. Therefore, 12-25 cases are sufficient to pre-test questionnaires, although sample sizes are often influenced by the survey schedule, leading to a typical sample of 5-15 participants [33].
A group of 12 students was asked to fill out both questionnaires (DKQ and DAQ) and to add a (+) or a (−) next to the questions they liked or did not like for different reasons. For instance, they could add a (−) next to the questions they did not understand or if something was not clear, and a (+) sign next to the comprehensible questions. After filling out the questionnaires, each student handed them back and discussed their comments with the interviewer (see Table 2). Adjustments were made to the final draft of the questionnaires according to the results obtained during the cognitive interviews, before testing both questionnaires on a larger sample size. Table 2. List of comments reported by the adolescents and the corresponding adjustments to the questionnaires, based on the cognitive interview results.

Feedback Categories
Comments Adjustments

Language Problems
DAQ: A participant asked to clarify the part "fish (not fried)".
The question was modified to include the following words "grilled/boiled/not fried".
DKQ: Many adolescents did not know what fibers and saturated fatty acids are.
No modifications made because this is related to the level of knowledge, and is an essential part of the DKQ.

Logical Problems
DKQ: One of the participants asked if smoking includes water pipes.
The words "cigarettes and waterpipes" were added to make it clear that both ways of smoking are included.
DAQ/DKQ: One of the respondents asked if the word "vegetables" in question 10 (DKQ) and question 13 (DAQ) included pickles too, or just fresh vegetables like cucumbers.
In the DKQ, participants are asked if vegetables are high or low in sodium, and the word "fresh" was added to make it clearer. In the DAQ, it was left as is because both pickles and fresh vegetables are included. The purpose of this question is not related to sodium consumption, but rather to fiber intake.

Inclusion Criteria and Sampling Method
Lebanese public and private high schools located in Beirut, Baalbeck and Rayak were included, thus covering both urban and rural regions. The urban region refers to Beirut, the capital and largest city of Lebanon, and the rural region refers to Baalbeck and Rayak, located in the Bekaa region, which is considered Lebanon's most important farming region (see Figure 2).

Inclusion Criteria and Sampling Method
Lebanese public and private high schools located in Beirut, Baalbeck and Rayak were included, thus covering both urban and rural regions. The urban region refers to Beirut, the capital and largest city of Lebanon, and the rural region refers to Baalbeck and Rayak, located in the Bekaa region, which is considered Lebanon's most important farming region (see Figure 2). Even though Baalbeck is a small city, it is considered part of the rural area along with Rayak, as schools in Baalbeck include many children coming from the villages nearby. In addition, no international fast-food chains, shopping malls or buses (connecting different locations in the same town) are located in Baalbeck. There are only some local restaurants and coffee shops.
As few studies involved Lebanese adolescents, assessing the feasibility of conducting research in Lebanese school-settings is one of the first key learnings of the current study. This includes experiences with school and participant recruitment, questionnaire administration in classes, data collection scheduling during class time, and facing challenges and opportunities in the implementation of the study protocol.
Several inclusion criteria for the participating adolescents were used: (1) being Lebanese and enrolled in Lebanese public or private high schools located in Beirut, Baalbeck and Rayak; (2) aged Even though Baalbeck is a small city, it is considered part of the rural area along with Rayak, as schools in Baalbeck include many children coming from the villages nearby. In addition, no international fast-food chains, shopping malls or buses (connecting different locations in the same town) are located in Baalbeck. There are only some local restaurants and coffee shops.
As few studies involved Lebanese adolescents, assessing the feasibility of conducting research in Lebanese school-settings is one of the first key learnings of the current study. This includes experiences with school and participant recruitment, questionnaire administration in classes, data collection scheduling during class time, and facing challenges and opportunities in the implementation of the study protocol.
Several inclusion criteria for the participating adolescents were used: (1) being Lebanese and enrolled in Lebanese public or private high schools located in Beirut, Baalbeck and Rayak; (2) aged 15-18 years; (3) being fully capable (cognitive, psychiatric and physical ability) of communicating (as reported by parents or by school administration); (4) not having any chronic or genetic disease (such as chronic kidney diseases or other diseases, as reported by parents or by school administration). Lebanese high schools are either public or private. Public high schools are managed by the Lebanese Ministry of Education and Higher Education, while private schools are managed by individuals or organizations. Due to the higher tuition fees in private schools, Lebanese families with higher incomes tend to enroll their children in private schools, while children from lower socioeconomic groups more often attend public schools. In this study, both types of schools were taken in consideration to include adolescents from different socioeconomic backgrounds.
The sampling method was nonselective, meaning that all participants meeting the inclusion criteria and present at the moment of data collection were selected. All subjects gave their informed consent for inclusion before they participated in the study. The questionnaire stated clearly that once it was filled in and handed in, this meant that the adolescent accepted and assented to participate in the study. Consent was also orally confirmed by the PI before administering the questionnaire. A similar assent statement was also included in the parents' version of the questionnaires. The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Lebanese Ministry of Education and Higher Education (3/15465; date: 24 October 2016).
The total number of adolescents enrolled in high schools located in Beirut (urban area) is 12,983 [34]. The number of public schools located in Beirut is 18 (obtained from the Lebanese Ministry of Education and Higher Education). However, the total number of private high schools in this region is not reported. The number of adolescents enrolled in high schools located in Bekaa (Baalbeck, Rayak and other towns and villages) is 15,843 [34]. The number of public high schools located in Baalbeck and Rayak is four, the number of private schools is not reported. For the current study, a total of seven schools were recruited: four public (one from Beirut, two from Baalbeck and one from Rayak) and three private (one from each city/town) schools.
According to Hulley et al. [35] and taking into consideration that this observational study involves correlations as the principal form of analysis, the total sample size is calculated as follows: Total sample size = N= [(Zα + Zβ)/C]2 + 3; taking into consideration: Asaad et al. [36], the obtained correlation between fruits and vegetable servings, high sugar foods and high fiber foods obtained from the perceived dietary adherence questionnaire and the 24 h recalls were 0.30, 0.40 and 0.46, respectively. Other correlations were not considered, as they were irrelevant in our case, e.g., low glycemic index foods) This results in a minimally required sample of N = 85 participants. The actual obtained sample size in the current study was 220.

Questionnaire Administration and Data Collection
Data collection took place between January and May 2017. A total of seven high schools were included. School administrations assigned one session per class during the normal schedule (40-50 min). The interviewers visited each class, explained the purpose of the study in detail, gave the necessary instructions to fill out the DKQ and DAQ questionnaires, and distributed and collected them. In addition, a 24 h dietary recall was administered to each student by the interviewers during the same session. The 24 h recalls were unstructured, although a stepwise protocol was used. A blank sheet was filled by the interviewer based on what the participants answered to the question concerning what they ate in the previous 24 h. In the next step, the interviewers asked about the missing information, such as the portion sizes, brand names of food items and cooking methods (grilled, boiled, fried, etc.). At the end, the interviewers went over the answers one more time, to make sure that the adolescents did not forget anything. The recall was used to compare the DAQ filled in by the students with the DAQ filled in by an expert based on the 24 h recall, so as to check the validity of the DAQ answers related to food and meal consumption. In addition, the parents' DKQ and DAQ were handed to the children, for their parents to complete at home. After one to two weeks, the completed parental questionnaires were collected at the schools.

Statistical Analyses
Data were entered and analyzed using the Statistical Package for Social Sciences version 25.0 (SPSS Inc., Chicago, IL, USA). p-values of <0.05 were considered statistically significant. Descriptive statistics were used to analyze the participant characteristics. Cronbach's alpha test was used to assess the internal consistency and reliability of the questionnaires. Coefficients from 0.5 to 0.75 suggest moderate reliability, and values >0.75 reflect good reliability [37].
Independent t-tests were employed to determine significant differences between subgroup means (boys vs. girls, urban vs. rural and public vs. private). Pearson's correlations were used to explore potential relations between knowledge and the adherence scores of adolescents and parents. One-Way analysis of variance (ANOVA) Post Hoc was used to determine any differences between the three study locations.

Cognitive Interview
A total of 12 adolescents was asked to fill out both questionnaires (DKQ and DAQ) and comment on them in a cognitive interview. These opinions were divided into three categories: (1) language problems (e.g., not knowing the meaning of a word); (2) logical problems (interpreting questions the wrong way); and (3) general feedback (e.g., positive feedback; not shown). The comments and subsequent adjustments to the questionnaires are listed in Table 2.

Demographic Characteristics
A total of 220 adolescents, recruited from seven high schools, submitted both questionnaires. As for the parents, 108 (49%) returned the DKQ and 89 (40.5%) filled out the DAQ. Their demographic characteristics are shown in Table 3.

Feasibility
All adolescents agreed to participate, and they were positively engaged. They were able to fill out both questionnaires during the assigned time. The presence of a teacher in class was experienced as favorable, as the adolescents were more disciplined and worked faster.

Internal Reliability
The internal reliability reflects how well the individual items of the scores fit together and assess the same construct [22]. Values for Cronbach's alpha ranged from 0.46 to 0.89 (see Table 4).

Dietary Knowledge and Adherence Scores
Adolescents scored an average of 30.7 (SD ± 6.9) on the DKQ, ranging from 7 to 50, and 34.54% answered more than half of the questions incorrectly. The mean total adherence index score was 2.1 (SD ± 1.5), ranging from 0.2 to 12, and 10% of adolescents obtained a score below 1. In terms of demographic characteristics such as age, gender and type of school, there was no significant difference, either for the knowledge score or for the total adherence score between groups (see Table 5). There was a significant difference for the knowledge score between urban and rural regions (p < 0.01). When comparing the three locations separately, there was a significant difference for the total knowledge and adherence scores between them (see Appendix D). Post-hoc analyses revealed that the difference was significant between Beirut and Baalbeck (p < 0.05). Parents had a mean of 32.4 (SD ± 10.0) on the DKQ, ranging from 2 to 49, and there were no significant differences between the groups based on demographics. Detailed responses of both questionnaires are shown in Appendices C and E.

Correlations between Scores
The DAQ of the parents was not included in the correlation analyses, as its Cronbach's alpha was low (<0.50). There was no statistically significant correlation between the knowledge score of adolescents and the corresponding total adherence index (Table 6). However, a significant negative correlation (r = 0.158, p = 0.02) was found between the DKQ and the unhealthy items score of the DAQ. There was a significant positive correlation (r = 0.359, p < 0.001) between the knowledge score of the parents and the knowledge score of their children. The parental knowledge score further correlated positively with the total adherence score of the children (r = 0.242, p = 0.02). A significant positive correlation was found between the unhealthy items and the healthy items of DAQ (r = 0.33, p < 0.001).
When comparing the DAQ items completed by the adolescents and those by the dietitian based on the 24 h recall, we found a significant, positive correlation between all the items related to food intake and meal consumption (questions 1-25) in the DAQ and their correspondents based on the 24 h recall, except for two items (question 8-skimmed milk and question 18-diet soft drinks) (see Appendix F).

Discussion
Overweight and obesity are multifactorial and largely preventable conditions [38], affecting more than 40% of Lebanese adolescents [6]. However, prior to planning any health promotion program targeting this particular age category, it is necessary to have reliable tools to assess the current levels of dietary knowledge and adherence to determine the effectiveness of such interventions later on. To our knowledge, studies assessing the relationship between dietary knowledge and adherence are scarce [25], particularly in the Middle East. Correspondingly, there is a paucity of reliable Arabic questionnaires measuring dietary knowledge and adherence.
Few studies have described the detailed process of developing such questionnaires, and none of them investigated the potential correlation between the dietary knowledge and adherence of the parents and of their children.
Many dietary questionnaires do not include physical activity questions, even though it is an essential part of energy expenditure [39]. Nor are all methods of dietary assessment (e.g., 24 h recall, diet records, etc.) feasible for administration in school settings due to their high cost and respondent burden [39]. The aim of the present paper was to describe the step-by-step process of developing dietary knowledge and adherence questionnaires suitable for Lebanese adolescents and their parents living in urban and rural areas, to examine their feasibility and internal validity in the target population, and to analyze the correlation between parental and adolescent scores.

Feasibility
As few previous studies involved Lebanese adolescents, it was important to assess the feasibility of conducting research in Lebanese school-settings (i.e., recruiting schools and participants, administering questionnaires to large groups of individuals at once, investigating potential challenges and opportunities). This step is essential, not only to guide future assessment studies, but also to plan future nutrition intervention in Lebanese schools. All adolescents agreed to participate in the study. By contrast, their parents showed low participation rates (49.0% for DKQ and 40.5% for DAQ). This suggests that involving the parents in future nutrition studies might be quite challenging, in line with previous findings [40]. There might be several reasons behind the low participation rate of the parents, including a lack of motivation to participate in research studies or events organized in schools. This was confirmed by school principals of the schools participating in the current study, reporting that many parents do not attend parents' meetings or lectures organized by the schools. This was in line with another study conducted in Lebanese schools [40]. Some parents might not have received the questionnaires, as some adolescents reported forgetting to hand the questionnaires to their parents, even though many teachers reminded them. In addition, parents might have a lack of interest in activities that do not affect the educational attainments of the children. Besides recruitment of parents, another challenge was recruiting private schools to participate in the study. There was no difficulty in convincing public schools, since we obtained the approval of the Lebanese Ministry of Education and Higher Education, allowing direct access to public schools. The private schools had to be approached individually for participation. Similar challenges were also reported in another study involving Lebanese children aged 9-11 years [40].
As for the feasibility of administering the questionnaires, all students were able to fill in both DKQ and DAQ during the assigned time, and the interviewer was also able to collect a 24 h recall in the same session. For classes with more than 20 students, two interviewers are required to cover all participants. Also, in some cases, it was beneficial to work with the students in the presence of a teacher or teacher assistant, as they were more disciplined and worked faster then.

Internal Reliability
The internal reliability of the dietary knowledge questionnaires we developed for both adolescents and their parents was more than acceptable [37]. The parental version had a somewhat higher Cronbach's alpha value, which can be explained by the fact that the current DKQ was based on a knowledge questionnaire that was originally developed for adults [17]. The internal validity of the current questionnaire is higher compared to previously developed knowledge questionnaires for younger Lebanese children (aged 9-11 years; alpha: 0.66) [25], and for older adolescents (aged 17-18 years; alpha ranged from 0.5 to 0.75) [41].
As for the dietary adherence questionnaire, the Cronbach's alpha was moderate for both components of the total adherence score for adolescents (healthy and unhealthy items), as well as for the unhealthy items for parents. A lower Cronbach's alpha was found for the healthy items of the adherence questionnaire of the parents, which can potentially be attributed to the small number of participants who filled in the DAQ (N = 89). Therefore, we suggest retesting it on a larger sample size in future studies. It is difficult to compare the current DAQ to similar questionnaires in other studies, because the scoring system and the target sample are often different, and Cronbach's alpha values were not always calculated [22].
Nonetheless, a nutrition questionnaire developed for Australian children aged 10-12 years and the New Zealand Diet Quality Index for Adolescents aged 14-18 years had both a lower Cronbach's alpha (0.5) for healthy behavior score and for measuring diet quality, respectively [22,42].

Dietary Knowledge and Adherence Scores
Adolescents scored a mean of 30.7 ± 6.95 for their dietary knowledge score, while their parents scored 32.4 ± 10.03, both groups scoring positive on more than half of the items on average. Even though only around 30% of adolescents answered incorrectly on more than half of the DKQ, this does not exclude the fact that some adolescents might have scored correctly by chance. The findings underline the necessity of studying the dietary adherence of adolescents within the contextual influence of dietary knowledge and of developing interventions to improve their dietary knowledge level. Another study involving adolescents aged 17-19 years showed that 27.4% had inadequate nutrition knowledge [14], which is similar to the current findings. A higher weighting was used for the nutrient sources, compared to the other parts, partly because we believe that it will be easier to understand the difference between healthy and unhealthy foods after explaining the nutrients present in different foods; the same weighting was used in the original knowledge questionnaire [17].
It is important to note that the absence of a significant difference between the dietary knowledge level of Lebanese adolescents enrolled in public vs. private schools suggests that using the same educational material and strategy when planning nutrition education is appropriate.
As for the total adherence index score, adolescents scored a mean of 2.1 ± 1.48, indicating that the healthy items outweighed the unhealthy ones. However, although only 10% had an adherence score of less than 1, this does not mean that the remaining 90% had adopted healthy lifestyles. The DAQ focuses on the frequency of consumption of a certain item and not on the portion size, as it uses a non-quantified food frequency approach [39]. Its main purpose was to compare the effectiveness of dietary interventions and to track eating patterns over time. It included short questions about the previous day, and the number of questions was limited to address the specific behaviors related to weight gain and to ease the respondent burden [39].

Correlations Between Scores
In addition to the reliability analyses, a validity testing of the adolescent DAQ was done by comparing results from the food choice and eating habits sections and 24 h recall. This showed significant correlations for the vast majority of the items (23 out of 25), indicating that the DAQ is a valid tool in reflecting the food choices and eating habits of Lebanese adolescents [21]. These findings differ from those in other studies. Hoelscher, Day, Kelder and Ward [39] reported that the percentage agreement between the School-Based Nutrition Monitoring questionnaire and the 24 h recall ranged from 38% to 89%, and Asaad, Sadegian, Lau, Xu, Soria-Contreras, Bell and Chan [36] reported that six out of nine items (67%) of the Perceived Dietary Adherence Questionnaire designed for diabetic patients were significantly correlated with the 24 h recall. However, the current study showed no significant correlation between the DAQ and the dietary recall for the items regarding skimmed milk and soft drinks. Perhaps the adolescents missed the word "skimmed" (for milk) and "diet" (for soft drinks), since both questions related to similar food items in the previous question (i.e., regular milk; regular soft drinks) with only a one word difference, unlike all the other questions.
The correlation between the total knowledge and total adherence index scores of adolescents was not significant. This was similar to the finding in a previous study of adolescent athletes, which also showed no correlation between nutritional knowledge and dietary intake [15]. However, the knowledge score of adolescents was strongly correlated with the unhealthy items of the total adherence score rather than with the healthy items. This suggests that when adolescents know more about nutrition, they seem to eat less unhealthily. In other words, dietary knowledge may act as an inhibitor of unhealthy behavior, but to a lesser extent as a stimulant for healthy food choices. Future studies should consider interpreting the total adherence index versus examining the healthy and unhealthy components of the DAQ separately.
Very few studies have previously investigated the correlation between dietary adherence and dietary knowledge, making it difficult to compare these results. Our findings are somewhat similar to another study reporting that a higher nutrition knowledge score was associated with a significantly higher consumption of some healthy food items and a significantly lower consumption of some unhealthy food items [43]. In addition, Kresic et al. [44] reported an association between the nutrition knowledge of university students and their dietary intake, but this relationship was significantly influenced by other factors, such as gender and university status.
The positive correlation between both adherence score components (healthy and unhealthy items) can be regarded as surprising. It might be explained by the fact that when an individual is eating excessive amounts of foods, he/she is eating unhealthy and healthy food items at the same time, especially as the traditional Lebanese diet includes both healthy (e.g., labneh, salads, etc.) and unhealthy food (e.g., Arabic sweets, high fat cheese, etc.). More studies are needed to examine these findings further.
From the present study, it seems that their parents' knowledge affects adolescents' food choices and practices more than their own knowledge does. Similarly, parental knowledge regarding the daily recommended servings of many food groups, has previously been found to be positively associated with adequate average child intakes of dairy products, fruit, vegetables, grains and cereals and meat [20]. This might not be surprising, because parents play a pivotal role in the development of their child's energy intake habits (food preference, availability of energy-dense food, etc.) [45]. There was also a positive correlation between the knowledge score of the parents and that of their children.

Strengths and Limitations
The current study has various strengths. A systematic mixed methods approach, combining both qualitative and quantitative analyses, was used to develop and test both questionnaires. They were designed by a dietitian with practical experience in nutrition in Lebanon. Adolescents from three different areas in Lebanon with differing socioeconomic status were included in the sample (both urban and rural areas). The study also had an appropriate sample size (N = 220). According to Oldewage-Theron et al. [29], a minimum sample size of 94 adolescents is required to study nutrition knowledge and dietary intake patterns.
As for the limitations, while most of the DAQ questions were compared against the 24 h recall, this was not the case for the DKQ and the rest of the DAQ questions related to physical activity, screen-viewing time and smoking. Further validation of the DKQ and the rest of the DAQ is desirable. In addition, more girls than boys participated. This can be attributed to the higher number of girls enrolled in Lebanese schools [30]. Some Lebanese public high schools are for girls only, but there are no schools for boys only. Another limitation is the unequal number of adolescents in the urban and rural regions. Rural regions have fewer schools, but the number of students per class is higher compared to urban high schools. The unequal number between public and private school students is due to the higher number of students per school in the recruited public schools compared to private high schools.

Implications for Research
The current pilot study is the first one including Lebanese adolescents from both rural and urban areas, and their parents. The development of a measure of dietary knowledge and adherence designed specifically for use among Lebanese adolescents is likely to be a useful addition to existing methods of dietary assessment. Both questionnaires are inexpensive, time-saving, and present a low burden. The questionnaires can be used in epidemiological studies to assess the general knowledge level, gaps in certain areas in nutrition, food choices and eating patterns. As nutrition knowledge is a broad topic, specifying the areas to focus on can guide future nutrition interventions and education strategies in nutrition, and elucidate the hidden problems.

Implications for Practice
The DKQ and DAQ can be used by public health promoters and researchers when planning and assessing nutrition interventions for adolescents, and by policy makers to develop effective strategies to promote healthy eating and prevent obesity. The DAQ is helpful to track progress by looking at the total adherence index reflecting the overall dietary adherence, and then looking closer at separate items separately to identify specific targets (e.g., low fruit and vegetable intake, frequent soft drink consumption, meal skipping). In addition, both questionnaires can be used by nutrition professionals and dietitians in face-to-face counseling to evaluate levels of dietary knowledge and adherence of adolescents, and therefore specify the areas and topics to focus on when educating individuals and planning diets.

Conclusions
The Dietary Knowledge (DKQ) and Dietary adherence questionnaires (DAQ) are brief and relatively inexpensive assessment questionnaires of dietary knowledge and adherence to nutritional recommendations that can be used in Lebanon. Future testing should investigate whether these questionnaires can be used in other Arabic countries. This study paved the way for subsequent nutrition education campaigns by assessing the feasibility of conducting such interventions and providing them with useful tools to evaluate their success.

Acknowledgments:
We are grateful to Cecilia Said who helped with the back translation, data collection, and data entry and to all the staff and participants of the high schools involved in this study.

Conflicts of Interest:
The authors declare no conflict of interest.   Table A4. Total adherence score components for the parents.

Items Questions Maximum Points
Healthy Items 1-Yesterday, did you eat meat (like chopped meat in stews), chicken breast (grilled/boiled/not fried), or fish (grilled/boiled/not fried)? 4-Yesterday, did you eat any of these foods? Labneh, shanklish, kareesha 5-Yesterday, did you eat any of these cheeses: Mozzarella, feta, akawi, baladiye, khadra? 8-Yesterday, did you drink skimmed milk (reduced or 0% fat) or skimmed laban? 10-Yesterday, did you eat whole bread, oat bread, brown bread, tortillas? 12-Yesterday, did you eat beans like lentils, white beans, fava beans (do not count green beans)? 13-Yesterday, did you eat any vegetables (do not count potatoes)? 14-Yesterday, did you eat fresh fruits? Do not count fruit juice and dried fruits. Remarks: Some items were not included in the total adherence score for the following reasons: • To improve reliability, by removing the question related to refined bread and commercial juices, our Cronbach's alpha value increases. • Some items are acceptable to eat by children, but are then less recommended to consume as adults, and vice versa. This explains the reason behind considering whole milk/laban as part of the unhealthy items for adults.

•
To avoid controversy, diet soft drinks were removed from both scores for adults, as it depends on the overall health of each individual separately.    Table A5. Cont.

Number of Answers Percent
What do you think will help prevent heart diseases?      Are you aware that many chronic diseases (such as diabetes, heart diseases and certain types of cancer) are related to a low intake of fruits and vegetables?
False/Do not know 54 51.9 Correct answers 50 48.1 Are you aware that smoking causes many chronic diseases (such as lung cancer and heart diseases)?
False/Do not know 13 12.5 Correct answers 91 87.5 Table A10. Answers to the dietary adherence questionnaire per question by parents.

Number of Answers Percent
Yesterday, did you eat meat (like chopped meat in stews), chicken breast (not fried), or fish (not fried)? *  Table A10. Cont.

Number of Answers Percent
Yesterday, did you have a snack? A snack is a food or drink (except for water) that you eat or drink between meals.