Individual and food environment determinants of Mediterranean diet adherence among Lebanese adults: a cross-sectional study

Background: Many studies revealed positive health impacts of the Mediterranean diet (MD) especially on metabolic disorders. Conversely, information about determinants of adherence to the MD are limited. This study aims to examine the individual and environmental determinants of MD adherence among adults in Lebanon. Methods: A cross-sectional study was carried out during July 2021 among a convenient sample of 326 Lebanese adults. Data from participants were collected via an online survey developed by study researchers, that is composed of 3 well-structured questionnaires: a) background questionnaire; b) Perceived Nutrition Environment Measures Survey in the Mediterranean Context (NEMS-P-MED), and c) 14-item Mediterranean Diet Adherence Screener (14-MEDAS). Results: Mean MD score in the total sample was 7.59±2.22, reecting moderate-to-fair MD adherence. Older age and having a regular routine of physical activity emerged as signicant independent determinants of better MD adherence. Equally, availability of whole wheat pasta, rice or our and sh at home and perceived importance of availability of easy to cook foods at food stores showed signicant positive association, while healthy food availability, and in- store characteristics had no association with MD adherence score. Conclusion: Future public health interventions aiming at promoting MD adherence among young adults Lebanon shall take account of availability of healthy food at home alongside individual factors such as meal preparation skills and regular physical activity.


Background
The Mediterranean diet (MD) is the traditional dietary pattern commonly embraced by the populations bordering the Mediterranean Sea [1]. The MD is characterized by high consumption of carbohydrates (plant foods, cereals and legumes) with low glycemic index, olive oil, a moderate intake of sh, poultry and red wine, and a low intake of red and processed meats. Adherence to the MD has been associated with numerous health bene ts including reduced risk of cognitive decline [2], reduced incidence of obesity [3], and prevention of cardiovascular diseases [4,5], and diabetes mellitus (DM) [5,6]. Lebanon, among other Mediterranean countries in the Middle East and North Africa (MENA) region, has been experiencing a gradual shift from the traditional Lebanese Mediterranean dietary pattern to the Western dietary pattern, characterized by increased intake of animal-based foods accompanied by a higher prevalence of dietrelated diseases [7][8][9].
Several studies reported low-to-moderate adherence to the traditional MD among Lebanese adults. Only 13% among a nationally representative sample of 2,610 Lebanese adults had higher adherence level to the MD as per the Lebanese Mediterranean Diet (LMD) score (LMD) [10]. Similar results were reported in another cross-sectional study [11]. While previous research studied the relationship between individual factors and MD adherence level [12], the role of the food environment in determining the level of MD adherence has not been examined among Lebanese adults. We presume that individuals' adherence to the MD is in uenced by both individual and food environment factors. Examining the independent associations between individual and food environment factors and MD adherence level would help identify the most important determinants of MD adherence and design effective public health intervention campaigns towards better lifestyle habits. This study aims to assess the independent associations of several individual, home and in-store food environment factors with MD adherence level among Lebanese adults prior to the COVID-19 pandemic and economic crises that have hit Lebanon since February 2020.

Methods
Sampling A cross-sectional study was carried out during July 2021 among a convenient sample of 326 subjects of recruited from urban Lebanese governorates. The study was approved by the Institutional Review Board of Notre Dame University-Lebanon and performed in accordance with the Declaration of Helsinki.
Potential study subjects were invited to participate via an e-yer. Inclusion criteria included: being a healthy Lebanese adult aged 18-65 years and responsible for most or all household food and grocery shopping. Exclusion criteria included: being pregnant or/and lactating and having chronic diseases (diabetes, kidney disease, cancer) that require dietary modi cations. Those interested and eligible to participate were invited to sign an electronic consent form and complete an online self-administered survey.

Data Collection
Data from participants were collected via an online survey developed by study researchers composed of 3 questionnaires: a) background questionnaire; b) Perceived Nutrition Environment Measures Survey in the Mediterranean Context (NEMS-P-MED), and c) 14-item Mediterranean Diet Adherence Screener (14-MEDAS). All questionnaires were translated from English to Arabic by a professional translator, pre-tested on a pilot sample of 10 participants, selected in the same way as participants in the actual study, and amended for clarity and consistency before use in the actual study.

Background information: Socio-Demographic, Lifestyle and Anthropometric Factors
Data on socio-demographic factors included age, gender, having children, area of residence (within Lebanon), marital status, educational status, specialty in a health-related major, and employment status.
Data on lifestyle factors included eating habits such as frequency of meals/day, frequency of main meals/week, frequency of breakfast intake/week, frequency of eating out/week, smoking status, and physical activity status. Anthropometrics included self-reported weight (kg) and height (cm) measurements for determination of body mass index (BMI), a main indicator of obesity. BMI (kg/m 2 ) was calculated by dividing the weight (kg) by the height squared (m 2 ) and classi ed according to the Center for Disease Control and Prevention (CDC) where participants with a BMI ≥ 30 kg/m 2 were classi ed as obese and those with a BMI less than 30 kg/m 2 were considered to be non-obese [13]. Moreover, perceived overall sleep quality and health status were measured.

Food Environment Assessment
Perception of food environment in two different settings, home and stores, was assessed using an adapted version of the NEMS-P-MED questionnaire composed of 5 main sections: (1) Home Food Environment, (2) Perceived Food Environment in Stores, (3) Perceived Food Environment in Restaurants, (4) Your Food Habits and Thoughts About Food, and (5) General/Background Information Questions [14].
To adapt the NEMS-P-MED to the Lebanese context, the questionnaire was translated from Spanish to Arabic by a professional translator and some sections were modi ed. Speci cally, we added 2 questions to Section 1 "How often did you share mealtimes with your household members?" and "How often did your parents/siblings encourage you to have healthy food choices when you were tempted to eat junk foods?". We also revised responses to 3 questions related to type of food stores, transportation to food stores and in-store characteristics, in Section 2, and added 1 question "Why did you read nutrition labels?" to this section. Two questions were removed from Section 4: one was unrelated to any of our study objectives (on factors affecting eating out at restaurants and ordering take-out) and the other was a question in the 14-MEDAS tool "How often do you eat fruits and vegetables?". We eliminated Sections 3 and 5 in the original NEMS-P-MED because they probe the consumer's experience in restaurants (unrelated to any of our study objectives) and participants' sociodemographic and lifestyle factors (covered in the background questionnaire), respectively.
The nal adapted questionnaire (NEMS-P-MED) was, therefore, composed of 3 sections (Home Food Environment, Perceived Food Environment in Stores, and Your Food Habits and Thoughts About Food) that include a total of 13 questions. The questions had different types of responses: dichotomous (yes/no), ordinal with a Likert-type scale ranging between 3 and 5 options (degree of agreement, importance or frequency). The complete adapted NEMS-P-MED questionnaire is available in Additional le 1.

MD Adherence Score
To determine participants' MD adherence level, we used the 14-MEDAS tool, a 14-item questionnaire developed in the context of the PREDIMED study [15], which has been validated in several Mediterranean (Greece, Cyprus, Italy, Spain, Portugal) and non-Mediterranean (Germany, USA, UK, Korea) populations [16][17][18][19]. The 14-MEDAS assesses the consumption frequency of 12 main components of the MD and 2 MD-related food habits [15]. Each of the 14 items is scored 1 (adherence to a MD component) or 0 (nonadherence to a MD component). The MD adherence scores were summed and the resulting score ranged from 0 to 14 with higher scores indicating greater MD adherence. MD adherence was characterized according to the following criteria: "low adherence", ≤ 5; "moderate to fair adherence", 6-9; "good or very good adherence" ≥10 [20].

Statistical Analyses
Analysis of data was carried out using the Statistical Package for the Social Science (SPSS) software version 22 for Windows. Descriptive statistical analyses were performed to determine means and standard deviations (SD) for continuous variables, and frequencies and percentages for categorical ones. Chi square test/Fisher's Exact test was used to explore relationships between categorical variables. Group differences on continuous variables were tested using one-way ANOVA when there were more than two groups to be compared. Multiple linear regression analyses were used to assess the individual and environmental determinants (home and food store) of MD adherence after controlling for the effects of confounders. In one regression analysis, the effects of individual characteristics on MD adherence was evaluated in Model 1. These effects were then re-evaluated after the introduction of home and store food environment variables into Model 2 and Model 3, respectively. In another regression analysis, the effects of home food environment variables on MD adherence were evaluated in Model 1. These effects were then re-evaluated after the introduction of individual characteristics in Model 2. In the third regression analysis, the effects of store food environment variables on MD adherence were evaluated in Model 1. These effects were then re-evaluated after the introduction of individual characteristics (Model 2). A pvalue < 0.05 was considered to be indicative of statistical signi cance. Preliminary analyses were conducted to ensure no violation of the assumptions of normality, linearity, multicollinearity and homoscedasticity.

Socio-Demographic, Anthropometric and Lifestyle Sample Characteristics
The socio-demographic, anthropometrics and lifestyle characteristics of study participants are presented in additional le 3. Overall, the sample consisted of 326 individuals (~ 79% females) with a mean age of 37.14 ± 11.84 years. Most of the study participants lived in Mount Lebanon (~ 79%), were married (~ 54%), had children (~ 88%), were holders of a university degree (~ 80%) and majored in non-health-related majors (~ 70%). In general, the study participants had normal body weight (~ 58%) and reported 3 or more meals per day (~ 86%), around 6-7 days per week (54%). They take breakfast almost daily (~ 54%), perceive their sleep quality to be good/very good (~ 81%), perceive their health status to be good/excellent (~ 84%) and eat out at a restaurant 0-2 days per week (~ 83%). The majority of the study participants reported to read/understand food labels (~ 56%, 61%, respectively) in order to help them make healthy food choices (58.6%). In the total sample, the mean MD adherence score was 7.59 ± 2.22.
The majority of the study participants (~ 83%) had a good to fair adherence level (MD score ≥ 6) and only about 17% had a low adherence level (MD score ≤ 5).

Adherence to the Mediterranean Diet
Additional le 2 shows the percentage of study participants who met the recommended consumption frequency of the 12 main components of the MD and reported to have two food habits related to the MD.
The majority of the study participants (> 50%) met the recommended consumption frequency of nine out of the twelve main components of the MD (seasonings with tomato, onion, garlic and olive oil (81.6%); legumes (74.8%); commercial baked goods (70.6%); vegetables (66%); red/processed meat (55.5%); fruits (54.6%); olive oil (54.3%); sugar-sweetened beverages (51.8%) and butter, margarine or full-fat cream (50.6%). In addition, most of the participants endorsed the two food habits related to the MD; speci cally: about 60% reported using olive oil as the main culinary fat and consuming chicken or turkey rather than beef/pork/sausage. Less than half of the participants; however, met the recommended consumption of unsalted nuts (48.2%), sh or seafood (23%) and wine (8.6%).

Description of the Food Environment
Home Food Environment A description of the home food environment is presented in additional le 4. Most of the study participants reported availability of certain healthy foods at home. These foods included fruits (~ 97%), vegetables (~ 99%), legumes (~ 95%) whole-grain or brown bread (~ 70%), fresh/frozen sh (~ 56%) and diet soft drinks (~ 53%). However, less than half of the individuals included in the study reported availability of other healthy foods at home -speci cally, low-fat or non-fat milk (~ 47%), whole wheat pasta, rice or our/low-fat or non-fat dairy (~ 41%). At least two-thirds of the study participants reported frequent easy access to fruits and vegetables. On the other hand, a high percentage (≥ 2/3) of the study participants also reported availability of certain unhealthy foods at home such as re ned pasta rice/ our (~ 97%), meat products (92%), sweets and pastries (~ 85%), cold cuts and charcuterie (~ 79%), white bread (~ 77%), chips and snacks (~ 73%), full-fat dairy (~ 74%) and full-fat milk (~ 65%). In addition, half of the participants reported availability of regular soft drinks at home. It is worth mentioning; however, that about 50% of the study participants reported infrequent easy access (sometimes/rarely, never) to sweets and pastries (sweets and pastries on the kitchen counter). As for family-level factors that affect dietary habits, 54% reported having meals with the family all the time and ~ 63% reported having consistently parental and sibling support to make healthy dietary choices.

Food Environment In Stores
A description of the food environment in stores is presented in additional le 5. The majority of the study participants reported going to supermarkets for food shopping (81.6%) while an equal percentage of the remaining participants reported going to minimarkets (35.3%) or small grocery stores (34.4%). Reasons for selecting a certain food shopping place include, in decreasing order: quality (92%), price (~ 88%) and variety (~ 84%) of food available at a food store, proximity of the food store to the participant's house (~ 78%)/other places the participant goes to (~ 76%), and same shopping store as that of friends/family (~ 33%). The majority of the study participants (~ 68%) indicated that the prices of fresh fruits and vegetables at the food store where they buy their food were not expensive as compared to those in other food stores. Furthermore, at least 70% of the participants agreed that healthy, unhealthy and a variety of food products are easy to buy at the food store where they buy their food (healthy (68-76%): fruits, vegetables, lean meats, low-fat products; unhealthy (71-80%): sweets, snacks and sodas or sugary drinks; a variety of food products: 85.9%). The majority of the study participants reported that they use their own car (~ 87%) to visit food stores. Other methods of transportation include walking (~ 18%), use delivery services (17.2%), or ride with a neighbor (5.5%).
As for food placement and promotions in stores where participants buy most of the food, only about 22% agreed that unhealthy foods were placed at one of the ends of the aisles, whereas 32% agreed that they bought food items placed at eye-level on shelves. While about 51% agreed that food items placed next to cash registers were usually unhealthy, only about 18% agreed that they bought food items placed next to the cash registers. Moreover, about 43% participants reported that they see signs that encourage the customer to buy healthy food compared to only 34% for unhealthy food. Approximately 62% of the study participants were aware of nutrition facts labels' presence on most pre-packed food products. The proportion of participants indicating each of the suggested characteristics of food as important when considering purchase of a particular food was as follows: food taste: ~98%; healthy food: ~94%; price: 91%; convenience/easy to cook: 88%, and weight control: ~82%. Furthermore, most of the participants reported use of a shopping list to buy their groceries (sometimes: ~43%; always: ~33%).

Individual Determinants
Associations of socio-demographic, anthropometric, and lifestyle characteristics with MD adherence level are presented in Table 1. In the total sample, older age, having children, having a health-related major, and having a regular routine of physical exercise were associated with higher MD adherence level. Speci cally, individuals with good MD adherence levels were older (Good: 40.55 ± 11.88 vs. Low: 34.52 ± 11.33, p = 0.016). In addition, a signi cantly higher percentage of individuals who reported having children and majored in health-related elds had good MD adherence levels compared to their counterparts (Children: Yes: 24% vs. No: 4.8%, p = 0.023; health-related elds: Yes: 29.5% vs. No: 14.8%, p = 0.020). Lastly, a signi cantly higher percentage of the study participants who reported following a regular routine of physical exercise was found to have good/moderate MD adherence levels compared to those who reported not to follow a regular routine of physical exercise (Yes: ~93% vs. No: ~78%, p = 0.001).    Table 4. After controlling for gender, having children, educational level, having a degree in health-related disciplines, smoking, and physical exercise, a 1-year increase in age was found to be associated with an increase of 0.042 in the MD score. In addition, lack of a regular routine of physical exercise was associated with a decrease of 0.908 points in the MD score (Model 1). After controlling for the effects of home food environment variables (availability of whole wheat pasta, rice or our, whole grain or brown bread, fresh/frozen sh, low-fat or non-fat dairy, sweets and pastries and regular soft drinks) in Model 2, and the effects of store food environment variables (e.g., motivation to select place of food shopping (variety, quality and price) and importance of taste, nutrition, price and convenience when shopping for food) in Model 3, age and regular routine of physical exercise remained to be positively associated with the MD score. Speci cally, a 1-year increase in age and lack of a regular routine of physical exercise were found to be associated with an increase of 0.028 and a decrease of 0.775 points in the MD score, respectively (Model 2). Similarly, in Model 3, a 1- year increase in age and lack of a regular routine of physical exercise were found to be associated with an increase of 0.042 and a decrease of 0.990 points in the MD score, respectively. Model 2: adjusted for home food environment variables (e.g., availability of whole wheat pasta, rice or our, whole grain or brown bread, fresh/frozen sh, low-fat or non-fat dairy, sweets and pastries and regular soft drinks).
Model 3: adjusted for food environment in stores variables (e.g., motivation to select place of food shopping (variety, quality and price) and importance of taste, nutrition, price and convenience when shopping for food).

CI = con dence interval Home Food Environment and MD Adherence as Assessed by Multivariable Linear Regression
Association of home food environment with MD score in the study population, as assessed by multivariable linear regression is presented in Table 5. Before adjustment for the effects of individual characteristics, there were signi cant associations between availability of whole wheat pasta, rice or our; whole grain or brown bread, sh at home and a borderline signi cant association between availability of regular soft drinks at home and MD score. Speci cally, unavailability of whole wheat pasta, rice or our; wholegrain or brown bread, and sh at home were found to be associated with a decrease of 0.747, 0.541 and 1.137 points in the MD score, respectively (Model 1). However, unavailability of regular soft drinks was found to be associated with an increase of 0.475 points in the MD score. After controlling for the effects of individual characteristics, availability of whole wheat pasta, rice or our and sh at home remained to be signi cantly associated with MD score whereas the associations between availability of whole grain or brown bread and regular soft drinks at home and MD score vanished.
Speci cally, unavailability of whole wheat pasta, rice or our and sh at home were found to be associated with a decrease of 0.714 and 1.084 points in the MD score, respectively (Model 2). Association of food store environment with MD score in the study population, as assessed by multivariable linear regression is presented in Table 6. Before adjustment for the effects of individual characteristics, there was a signi cant association between perception of availability of easy to cook foods at food stores as important and MD adherence score. Speci cally, perception of availability of easy to cook foods as important was found to be associated with a decrease of 0.783 points in the MD score (Model 1). After controlling for the effects of the individual characteristics, perception of availability of easy to cook foods as important remained to be signi cantly associated with MD score and was found to be associated with a decrease of 0.765 points in the MD score (Model 2).

Individual Determinants of MD adherence
Older age, having children, having a health-related major, and following a regular routine of physical exercise were identi ed as characteristics associated with increased adherence to the MD while gender, living area, marital status, educational level, eating habits and BMI were not associated with adherence levels. According to our analyses, old age and regular routine of physical activity were the only determinants which maintained statistically signi cant association with adherence levels, after controlling for multiple individual and food environment variables. The observed association between age and MD adherence is similar to previous ndings from studies conducted in Lebanon and other Mediterranean countries [11,12,21,22]. One explanation could be that older adults tend to maintain traditional dietary habits whereas younger adults are more exposed to new and "fashionable" food products (high-calorie and nutrient-poor ultra-processed foods). In this study, regular physical activity was strongly associated with higher MD adherence levels, a nding similar to those reported among Spanish [23,24] and Lebanese adults [12].
In view of the inconsistent ndings from the literature and the disproportionate gender distribution in the study sample (~ 80% females), the lack of a statistically signi cant association between gender and adherence to the MD was expected. No statistically signi cant association was found between educational level and adherence to the MD, contrary to the ndings from the literature where higher educational levels were associated with higher MD adherence [12,21,25]. However, the majority of the study participants (80%) had university level education which may account for the lack of a statistically signi cant association between educational level and MD adherence. No statistically signi cant association was found in this study between BMI and MD adherence: no association between BMI and MD adherence [21,26], or negative association [27].

Home Environment Determinants of MD adherence
Studies examining the association of home food environment with MD adherence particularly among adults living in Mediterranean countries are very few. Our nding that the availability of healthy foods (e.g. whole wheat pasta, rice or our, fresh/frozen sh, low-fat or non-fat dairy) and unavailability of unhealthy foods (e.g. pastries, chips, regular soft drinks) at home were signi cantly associated with higher MD adherence is consistent with previous ndings which revealed positive associations between fruit and vegetables at home and healthy eating/fruit and vegetable intake in adults, proxies for the MD [28,29]. Moreover, our ndings were in line with previous studies that reported negative associations between salty snacks and sweets availability at home and healthy eating/fruit and vegetable intake in adults [30]. The availability of whole wheat pasta, rice or our and sh at home remained signi cantly positively associated with MD score after controlling for the effects of individual characteristics.
Moreover, an inverse signi cant association was found between accessibility to unhealthy foods (e.g., sweets and pastries on kitchen counter) and MD adherence level, similar to previous ndings on accessibility to and increased consumption of fruits and vegetables at home [28,30].
In addition, social-environmental in uences within the home such as parent's and siblings' support to make healthy food choices and frequent family meals were not found to be signi cantly associated with MD adherence level. The only existing evidence is found in studies on children and adolescents where signi cant positive associations were found, probably due to participants' young age, implying that these participants' dietary choices were still highly in uenced by familial behaviors and communication patterns [31].

Food Store Environment Determinants of MD adherence
Our study found a signi cant association between those who perceived no importance of food stores' proximity to their houses and higher MD adherence level, contrary to previous ndings that reported insigni cant associations of food stores' proximity with fruit/vegetable consumption [32,33], possibly due to either Lebanese neighborhoods being easier to walk around to food stores without public transportation, or the Lebanese's preference to purchase high-quality, nutritious foods even if outside the neighborhood. In addition, our study found that participants who perceived food quality and variety as important in selecting place of food shopping had lower MD adherence level; these associations, however, became insigni cant after controlling for confounders and are consistent with two studies which found no association between perceived store quality and food availability to a healthy eating index [29] after controlling for sociodemographic factors [34]. One study, which showed an inverse association between availability/variety of in-store fruits and vegetables and fruit and vegetable consumption was only partly consistent with our ndings [28] whereas two studies, which showed positive associations between perceived availability of healthy foods [35] and quality of fruits and vegetables [28] with daily fruit and vegetable consumption disagreed with our ndings.
The only food store environment variable which emerged as an independent determinant of MD adherence was perceived importance of convenient meals during food shopping. Participants who perceive convenient or easy to cook meals when shopping in food stores as important are less likely to follow the MD.

Strengths and Limitations
To the best of our knowledge, this study is the rst to examine the relationship between food environment at home and in food stores and adherence to MD among Lebanese adults. In our study, we collected data on many covariates reported to have associations with MD adherence level, exploring the associations between individual and food environment factors and MD adherence, before and after controlling for confounders. The ndings of this study ought to be considered considering several limitations. First, the cross-sectional design of the study does not allow the identi cation of causal associations between home and store food environment and MD adherence. Second, the target sample was a convenient sample making over-sampling of a particular subgroup of the population cannot be ruled out. Third, the use of self-lled frequency questionnaires may have resulted in an overestimation of foods considered healthy and underestimation of foods considered unhealthy. The interpretation of frequencies and importance may have been different among participants, when responding to questions about the food environment. Fourth, individuals were asked to re ect on the conditions prior to February 2020, preceding the COVID-19 pandemic and the economic crisis. This raises possibility of memory bias among participants. Fifth, we did not carry out a study of the validity and reliability of the adapted NEMS-P-MED questionnaire prior to its use. Finally, the 14-MEDAS adherence score represents a valid and easy tool for a rapid screening rather than exact assessment of the adherence to the Mediterranean dietary pattern.

Conclusion
Overall, our study showed moderate-to-fair adherence (64%) to the Mediterranean Diet among Lebanese adults. Younger individuals and those who do not engage in regular physical exercise were less adherent to the MD, highlighting the need for health promotion efforts directed at younger age groups, across genders, with fairly low levels of physical activity to emphasize the advantages of the MD and encourage adherence. We identi ed the home environment as the most in uential environmental setting associated with MD adherence. Participants with increased access to healthy foods at home, speci cally sh and whole wheat pasta, rice, or our, had higher MD adherence levels. As for the food environment in stores, perceived importance of easy to cook foods when shopping for food in stores was found to be the strongest predictor of the MD score. The over-representation of young, highly educated females in our