Adolescence is a transitional stage during which many changes take place at physiological and behavioural levels, representing an important life stage for the development of healthy nutrition behaviour(Reference McNaughton, Ball and Mishra1). Many different factors influence food habits in a complex interactive way(Reference Aranceta, Pérez-Rodrigo and Ribas2). Socio-economic and lifestyle factors (parental occupational status, maternal level of education, cultural and/or religious habits, the role of family and patterns of beauty) have a strong influence on eating habits in adolescents(Reference Neumark-Sztainer, Hannas and Story3).
Epidemiological evidence suggests that dietary patterns in the Mediterranean countries are changing rapidly, with an increased consumption of animal products and saturated fat and a decline in intake of basic foodstuffs based on vegetables(4). Recent nutritional surveys carried out in Spain also confirmed a progressive departure from the traditional Mediterranean diet towards a Western dietary pattern, mainly in young generations(Reference Aranceta5–Reference Martínez, Llull and Bibiloni11). Because there is evidence that nutritional behaviours track from adolescence into adulthood, the promotion of healthy nutrition during adolescence has the potential to confer significant long-term health benefits(Reference McNaughton, Ball and Mishra1).
Despite the worldwide promotion of the Mediterranean dietary pattern, a progressive shift to a non-Mediterranean pattern could be also developing among Balearic Islands’ adolescents. Therefore, the aim of the present study was to assess prevailing food patterns among adolescents living in the Balearic Islands, as well as socio-economic and lifestyle determinants.
Methods
Study design
The study is a population-based cross-sectional nutritional survey carried out (2007–2008) in the Balearic Islands, a Mediterranean region.
Selection of participants, recruitment and approval
The target population consisted of all inhabitants living in the Balearic Islands aged 12–17 years. The sample population was derived from residents aged 12–17 years registered in the scholar census of the Balearic Islands. The theoretical sample size was set at 1500 individuals in order to provide a specific relative precision of 5 % (type I error = 0·05; type II error = 0·10), taking into account an anticipated 70 % participation rate. The sampling technique included stratification according to municipality size, age and sex of inhabitants, and randomisation into subgroups, with Balearic Islands municipalities being the primary sampling units and individuals within the schools of these municipalities comprising the final sampling units. The interviews were performed at the schools. The final sample size was 1231 individuals (82 % participation). The main reason for non-participation was the adolescent declined to be interviewed, or the parents did not authorise the interview.
Ethics
The study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving human subjects/patients were approved by the Balearic Islands Ethics Committee. Written informed consent was obtained from all adolescents and their parents or legal guardians.
Dietary assessment
Dietary questionnaires included two non-consecutive 24 h diet recall periods, one in the warm season (May–September) and one in the cold season (November–March) to account for the influence of seasonal variations, and a semi-quantitative FFQ that has previously been validated(Reference Martin-Moreno, Boyle and Gorgojo12) and applied in other studies and surveys on the Spanish population(Reference Tur, Romaguera and Pons9, Reference Tur, Romaguera and Pons10, Reference Serra-Majem, Morales and Domingo13, Reference Bondia-Pons, Serra-Majem and Castellote14). The FFQ, which asked the participant to recall average use over the past year, consisted of 145 items (118 of the original validated FFQ plus the most characteristic Balearic Islands foods in order to make it easy for the interviewee to answer) arranged by food type and meal pattern. Frequency of food consumption was based on times that food items were consumed (per day, week or month). Consumption < 1/month was considered no consumption. The period of consumption of seasonal items was also considered. Edible fractions of foods were recorded in the database. The FFQ foods items were collapsed to twenty-nine food groups (Table 2) that may have practical importance in the daily diet and closely approximated food groups previously reported(Reference Lazarou, Panagiotakos and Matalas15, Reference Lazarou, Panagiotakos and Kouta16). To account for day-to-day intake variability, the questionnaires were administered homogeneously from Monday to Sunday. Well-trained dietitians administered the recalls and verified and quantified the food records.
To estimate volumes and portion sizes, the household measures found in the participants’ own homes were used. Conversion of food into nutrients was made using a computer program (ALIMENTA®; NUCOX, Palma, Spain) based on Spanish(Reference Mataix, Mañas and Llopis17, Reference Ortega, López and Requejo18) and European(Reference Feinberg, Favier and Ireland-Ripert19) food composition tables, and complemented with food composition data available for Majorcan food items(Reference Ripoll20). Identification of misreporters was conducted on the basis of the ratio of energy intake (EI) to BMR. EI:BMR < 0·92 (boys) and < 0·85 (girls) was considered to represent under-reporting(Reference Livingstone and Black21), while EI:BMR ≥ 2·4 was considered to represent over-reporting(Reference Johansson, Solvoll and Bjørneboe22, Reference Mendez, Wynter and Wilks23). Under-reporters (20 %) and over-reporters (2 %) were excluded from the analysis.
Socio-economic and lifestyle determinants
A questionnaire incorporating the following questions was used: age group; parental educational level (grouped according to years and type of education: low, < 6 years; medium, 6–12 years; high, > 12 years); and parental socio-economic level (based on the occupation of parents and classified as low, medium and high, according to the Spanish Society of Epidemiology)(Reference Alvarez, Alonso and Domingo24). The number of daily meals and snacks was calculated from the total eating occasions that participants declared among the following: breakfast; mid-morning snack; lunch; mid-afternoon snack; dinner; before going to sleep; others. Three groups of eating frequency were considered: ≤ 3, 4 and ≥ 5 times/d.
Physical activity patterns
Physical activity was evaluated according to the guidelines for data processing and analysis of the International Physical Activity Questionnaire (IPAQ)(25) in the short form and its specific modification for adolescents (IPAQ-A)(Reference Hagströmer, Bergman and De Bourdeaudhuij26). The specific types of activity assessed were walking, moderate-intensity activities (i.e. physical activity at school) and vigorous-intensity activities (i.e. sports practice). In accordance with the AVENA (Food and Assessment of Nutritional Status of Adolescents) study(Reference Vicente-Rodríguez, Rey-López and Martín-Matillas27), the questionnaire also included information on television (TV) viewing, computer use and video games in h/d, and usual sleep duration to the nearest 10 min. Physical inactivity was established with a cut-off level of 300 min/week of moderate/vigorous physical activity, in accordance with current guidelines of physical activity for adolescents(Reference Ferreira, Molena and Marques28, Reference Strong, Malina and Blimkie29).
Assessment of other covariables
Anthropometry and adiposity
Height was determined using a mobile anthropometer (model KaWe 44444; Kirchner & Wilhelm GmbH Co. KG, Asperg, Germany) to the nearest millimetre, with the subject's head in the Frankfurt plane. Body weight was determined to the nearest 100 g using a digital scale (model sc9210; Tefal, Rumilly, France). The subjects were weighed in bare feet and light underwear, which was accounted for by subtracting 300 g from the measured weight. Triceps and subscapular skinfold thickness (ST) were measured using a Holtain skinfold calliper (Tanner/Whitehouse, Crosswell, Crymych, UK) and the mean of three measurements (right arm) was used. Height and weight measures were used to calculate BMI (kg/m2). Body fat percentage (%BF) was measured from triceps and subscapular ST according to Slaughter et al.(Reference Slaughter, Lohman and Boileau30). This equation has been proposed as the most accurate for estimation of %BF from ST in this particular population of adolescents(Reference Rodríguez, Moreno and Blay31). %BF and height were used to calculate fat mass index (FMI; kg/m2).
Defining overweight and obesity
In children and adolescents, BMI for age has been established as the main measurement to define overweight and obesity(Reference Krebs, Himes and Jacobson32). However, there are some limitations associated with its use as an indicator of fatness. For example, individuals with increased muscle mass may also have increased BMI(Reference Daniels33). On the other hand, individuals with decreased lean body mass and increased adiposity may also be misclassified by assessment with BMI(Reference Daniels33). Alvero-Cruz et al.(Reference Alvero-Cruz, Alvarez Carnero and Fernández-García34) showed that the FMI had higher accuracy for overweight screening than BMI. The FMI is a useful measure to evaluate body composition parameters by effectively eliminating differences in body fat associated with height(Reference Baumgartner, Koehler and Gallagher35). Nevertheless, it is difficult to exclude BMI from the normal-weight and overweight definition. For this reason, a combination of BMI and FMI was used to define overweight and obesity in the present study.
First, subjects were classified using the age- and sex-specific cut-offs developed and proposed for international comparisons by Cole et al.(Reference Cole, Bellizzi and Flegal36), recommended for use also by the International Obesity Taskforce. Then, subjects were classified as normal-fat and overfat according to their FMI using the sex-specific cut-offs proposed by Alvero-Cruz et al.(Reference Alvero-Cruz, Alvarez Carnero and Fernández-García34) for adolescents: 4·58 kg/m2 in boys and 7·76 kg/m2 in girls. Thus, adolescents were classified into two groups as follows: (i) not at risk (BMI for age and sex < 25 kg/m2 or BMI ≥ 25 kg/m2 but FMI < 4·58 kg/m2 in boys and FMI < 7·76 kg/m2 in girls); (ii) overweight/obesity (BMI for age and sex < 25 kg/m2 or BMI ≥ 25 kg/m2 but FMI ≥ 4·58 kg/m2 in boys and FMI ≥ 7·76 kg/m2 in girls). The variable was labelled as ‘body composition’.
Body image
Self-perceived body image was measured using the Stunkard scale(Reference Stunkard, Sorensen and Schulsinger37), which consists of silhouette drawings ranging from 1 to 9 with monotonic increments in overweight percentage where 1 is the leanest and 9 the heaviest. Separate figures for boys and girls were used. Participants were asked to identify of the nine body figures: ‘Which silhouette looks most like yourself?’ and ‘Which silhouette would you like to look like?’ The difference between perceived body image and desired body image was used to determine the level of dissatisfaction with current body image. Values other than zero represent dissatisfaction with perceived body image. A positive value was indicative of the participant's desire to be thinner than his/her perceived current size, while a negative value reflected the participant's desire to be thicker than his/her current perceived size(Reference Bulik, Wade and Heath38, Reference Baptiste-Roberts, Gary and Bone39).
Statistical analyses
Analyses were performed with the SPSS statistical software package version 19·0 (SPSS Inc., Chicago, IL, USA). Factor analyses by the principal components method and varimax rotation were run on Z-scored transformed food consumption variables in order to identify salient food patterns in the group. To interpret the data, only food groups with factor loading > 0·250 were retained for each factor. For each adolescent, the factor score for each pattern was calculated. These scores were categorised (quintiles), as is often done in dietary epidemiological studies that relate food patterns to health outcomes when there is not a priori knowledge of the function that best fits the data(Reference Lioret, Touvier and Lafay40). Significant differences in prevalence were calculated by means of the χ2 test. Logistic regression models with the calculations of corresponding adjusted odds ratio and 95 % confidence interval were used to examine differences between adolescents’ characteristics and dietary patterns (the fifth quintile (Q5) v. the first quintile (Q1) in each dietary pattern). Univariate analysis was first carried out for all of the socio-economic and lifestyle variables that could be associated with dietary patterns. Any factor was considered a candidate for the multivariate model. Multiple logistic regression analyses were used to simultaneously examine the effect of different socio-economic and lifestyle variables on the dietary patterns. Level of significance for acceptance was P < 0·05.
Results
Participants’ characteristics according to gender are shown in Table 1. Factor analysis retained two major dietary pattern factors which explained 24 % of the total variance: the ‘Western’ pattern (factor 1, explaining 13·4 % of total variance) and the ‘Mediterranean’ pattern (factor 2, explaining 10·6 % of total variance; Table 2). Food group categories identified in the ‘Western’ pattern were yoghurt and cheese, dairy desserts, red meat, poultry, sausages, eggs, bread, cereals, pasta, rice dishes, pizza, fruit juices, canned fruits, nuts, soft drinks, high-fat foods, other oils and fats, sweets and chocolates. Those identified in the ‘Mediterranean’ pattern included yoghurt and cheese, red meat, poultry, fish and seafood, eggs, legumes, pasta, fresh fruit, fruit juices, vegetables, potatoes and tubercles and olive oil. ‘Western’ and ‘Mediterranean’ labels were chosen according to similarities to Western and Mediterranean dietary patterns described elsewhere(Reference Hu, Rimm and Stampfer41, Reference Trichopoulou, Costacou and Christina42). Despite that nuts are a common Mediterranean food(Reference Willett, Sacks and Trichopoulou43), in the present study a higher consumption was associated with the ‘Western’ pattern. However, this association may be associated with the higher consumption of nuts found in the Balearic Islands population than in the overall Spanish population(Reference Tur, Romaguera and Pons10).
Table 1 Characteristics of the study population: representative sample of adolescents (aged 12–17 years) living in the Balearic Islands, Spain, 2007–2008
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary-alt:20160626132954-13611-mediumThumb-S1368980011002199_tab1.jpg?pub-status=live)
Table 2 Food patterns identified by factor analyses using the principal components method and varimax rotation: representative sample of adolescents (aged 12–17 years) living in the Balearic Islands, Spain, 2007–2008
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary-alt:20160626132943-14115-mediumThumb-S1368980011002199_tab2.jpg?pub-status=live)
Food group factor loadings: only food groups with factor loading > 0·250 were retained for each factor. Only adolescents who did not misreport their energy intake were considered for this analysis.
The ‘Western’ dietary pattern was found more often among boys (OR = 4·47, 95 % CI 2·81, 7·11, P < 0·001), while the ‘Mediterranean’ dietary pattern was mainly followed by girls (OR = 1·43, 95 % CI 0·93, 2·20, P ≥ 0·05). Associations between dietary patterns and socio-economic and lifestyle determinants stratified by gender were assessed (Table 3 for boys and Table 4 for girls). The ‘Western’ dietary pattern was associated with spending ≥ 4 h/d on media screen time in both sexes and with being in the youngest age group (12–13 years) in girls, but was less prevalent among those adolescents who wished to be thinner and those girls who desired to remain the same weight. The ‘Mediterranean’ dietary pattern was mainly followed by boys who spent < 2 h/d on media screen time and girls with low parental socio-economic status. Results also revealed that adolescents who spent ≥ 4 h/d on media screen time and were in the highest quintile for the ‘Western’ dietary pattern showed a higher frequency of consumption of milk (OR = 7·86, 95 % CI 1·15, 53·48; P < 0·05), soft drinks (OR = 6·15, 95 % CI 1·23, 30·79; P<0·05) and nuts (OR = 4·01, 95 % CI 1·02, 15·80; P < 0·05) than their counterparts who spent < 4 h/d.
Table 3 Socio-economic and lifestyle determinants of ‘Western’ and ‘Mediterranean’ dietary patterns among adolescent boys, Balearic Islands, Spain, 2007–2008
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary-alt:20160626132950-79416-mediumThumb-S1368980011002199_tab3.jpg?pub-status=live)
Q1, first quintile; Q5, fifth quintile; Ref., reference category.
†Univariate analysis (logistic regression analysis considering the effect of one explanatory variable).
‡Multivariate analysis (multiple logistic regression analysis considering the simultaneous effect of all explanatory variables). Only adolescents who did not misreport their energy intake were considered for this analysis.
Significant differences between Q5 and Q1 in each dietary pattern (χ 2 test): *P < 0·05, **P < 0·01, ***P < 0·001.
Table 4 Socio-economic and lifestyle determinants of ‘Western’ and ‘Mediterranean’ dietary patterns among adolescent girls, Balearic Islands, Spain, 2007–2008
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary-alt:20160626132947-82760-mediumThumb-S1368980011002199_tab4.jpg?pub-status=live)
Q1, first quintile; Q5, fifth quintile; Ref., reference category.
†Univariate analysis (logistic regression analysis considering the effect of one explanatory variable).
‡Multivariate analysis (multiple logistic regression analysis considering the simultaneous effect of all explanatory variables). Only adolescents who did not misreport their energy intake were considered for this analysis.
Significant differences between Q5 and Q1 in each dietary pattern (χ2 test): *P < 0·05, **P < 0·01, ***P < 0·001.
In boys, the univariate analysis also showed that the ‘Western’ dietary pattern was associated with being in the youngest age group (12–13 years), low parental socio-economic status, spending ≥ 4 h/d on media screen time, not being at risk of overweight or obesity and the wish to be thinner; whereas being in the middle age group (14–15 years) and spending < 2 h/d on media screen time were associated with the ‘Mediterranean’ pattern. In girls, low parental educational level, low socio-economic status and spending ≥ 2 h/d on media screen time were associated with the ‘Western’ dietary pattern; whereas medium or high parental educational level, high parental socio-economic status and spending < 2 h/d on media screen time were associated with the ‘Mediterranean’ pattern. Results also revealed a negative association between age and a desire to maintain the same body shape or a lower one and ‘Western’ dietary pattern in girls; whereas it was positively related to the ‘Mediterranean’ dietary pattern (despite the lack of statistical significance). To consume < 5 daily meals was associated with a low probability to follow a ‘Western’ dietary pattern among girls, but this variable lost statistical significance after being adjusted for all explanatory variables.
Discussion
The Mediterranean dietary pattern has been widely reported to be a model of healthy eating for its contribution to favourable health status and best quality of life. However, a progressive shift to a non-Mediterranean pattern, even in countries bordering the Mediterranean Sea, has been observed(Reference Sofi, Cesari and Abbate44). Previous studies carried out in Spain confirmed that young generations(Reference Aranceta5–Reference Martínez, Llull and Bibiloni11) are further away from the Mediterranean dietary pattern. Lately, we found that the average adherence to the Mediterranean dietary pattern in our adolescent population was 58 %(Reference Martínez, Llull and Bibiloni11). The present study has identified two major dietary patterns among adolescents (‘Western’ and ‘Mediterranean’), supporting evidence for a nutrition transition from a traditional healthy diet towards a Western diet among Mediterranean youth. The literature has reported that girls paid more attention to foods than boys, met nutritional recommendations, and tried to prevent or reverse the obese state and improve health status(Reference Sweeting45). Accordingly, we found that girls were more likely to follow the ‘Mediterranean’ but not the ‘Western’ dietary pattern than boys.
Socio-economic and lifestyle characteristics are important determinants of the health status in a community. Parental educational level and socio-economic status have a marked effect on children's and adolescents’ lifestyles and dietary habits(Reference Fernández46). In previous studies, we found that maternal educational level was associated with diet quality(Reference Tur, Romaguera and Pons9) and adherence to the Mediterranean dietary pattern(Reference Martínez, Llull and Bibiloni11). Moreover, low socio-economic status and maternal educational level have also been related to high consumption of sweets, high-fat bakery products, sugary and salty snacks(Reference Aranceta5, Reference Lowry, Kann and Collins47–Reference Zive, Frank-Spohrer and Sallis49). It has also been pointed out that food cost is another factor that may influence people's dietary choices, and to follow a ‘Mediterranean’ dietary patterns is usually more expensive than following a ‘Western’ one(Reference Lopez, Martinez-Gonzalez and Sanchez-Villegas50). Our study shows that low parental socio-economic status is strongly associated with the ‘Western’ pattern, whereas high parental socio-economic status is associated with the ‘Mediterranean’ pattern.
Children and adolescents usually spend much leisure time on sedentary behaviours, such as TV viewing, computer use and video games, collectively known as screen time(Reference Ferreira, Molena and Marques28). An association between watching TV for > 2 h/d and major consumption of high-fat snacks and high-sugar drinks has been demonstrated(Reference Aranceta5, Reference Rey-López, Vicente-Rodríguez and Répásy51). TV is clearly related to exposure to the advertising of unhealthy foods(Reference Sweeting45). However, whereas media screen time was a good proxy for either dietary pattern in the present study, no significant differences were observed when TV and computer use were assessed separately. Thus, adolescents who spent ≥ 4 h/d on media screen were more likely to follow a ‘Western’ dietary pattern. However, those who spent < 2 h/d showed the highest probability for a ‘Mediterranean’ dietary pattern.
Physical activity level has been associated with food choice, and cereals, fruits and vegetables often appear in the diet of active adults and children(Reference Bellisle52). Children who follow a healthy diet are those who might also maintain high levels of physical activity(Reference Lazarou, Panagiotakos and Matalas53). Despite no significant association between physical activity and the dietary patterns being obtained in the present study, in a previous study we found that sedentary and low-active adolescents showed the lowest adherence to the Mediterranean dietary pattern(Reference Martínez, Llull and Bibiloni11).
On the other hand, no statistically significant association between dietary patterns (‘Western’ and ‘Mediterranean’) and sleep time was found in the present study. Taheri(Reference Taheri54) suggested that sleep loss leads to more opportunities for food intake. The evidence linking sleep and weight has been stronger for younger children and adults(Reference Capuccio, Taggart and Kandala55, Reference Patel and Hu56). In a previous work we found that short sleep (< 7 h/d) was associated with obesity compared with longer sleep in boys, whereas no association was found in girls(Reference Bibiloni, Martinez and Llull57). Recently Lytle et al.(Reference Lytle, Pasch and Farbakhsh58) have found more evidence for younger adolescents as compared with older adolescents. Despite that further studies are needed, adequate sleep during this critical time period is important(Reference Lytle, Pasch and Farbakhsh58).
Despite finding no significant association between number of daily meals and snacks and dietary patterns, the present results indicated that ≥ 5 eating occasions per day may promote high daily consumption of energy-dense foods and drinks. However, meal pattern and omission of meals, especially skipped breakfast, have also been suggested as markers of an inappropriate dietary intake among adolescents(Reference Sjöberg, Hallberg and Höglund59), and in a previous study eating frequency was identified as a risk factor for obesity in both boys and girls(Reference Bibiloni, Martinez and Llull57). Therefore, a promotion of at least 5 daily meals and snacks should be considered in nutrition education programmes for adolescents aimed at reducing risk of disease. However, education programmes could also focus on strategies to promote healthy food choices following the Mediterranean diet.
Our results also revealed that adolescents without risk to be overweight or obese were more likely to follow the ‘Western’ dietary pattern than the ‘Mediterranean’ dietary pattern (despite that statistical significance was found only in boys). The Mediterranean diet is an example of a healthy diet(Reference Trichopoulou, Costacou and Christina42). However, controversial results in adherence to the Mediterranean diet and obesity in children and adolescents have been found in the literature. Some studies observed an inverse association between Mediterranean dietary pattern and BMI(Reference Lazarou, Panagiotakos and Matalas53, Reference Kontogianni, Vidra and Farmaki60, Reference Tsartsali, Thompson and Jago61), but others found no correlation(Reference Klesges, Klesges and Brown62). On the other hand, a study among 2513 Spanish children and adolescents aged 10–24 years which assessed adiposity by waist circumference and weight:hip ratio found an inverse association of both indicators with adherence to the Mediterranean diet(Reference Schröder, Mendez and Ribas-Barba63). We have also previously observed that both obese boys and girls avoided sweets and salty snacks consumption to counteract the obesity(Reference Bibiloni, Martinez and Llull57).
Body image has been found to be a powerful determinant of adolescent nutritional habits and food choices(Reference Middleman and Duran64); and in the present study a desire for a thinner body shape was associated with low risk for the ‘Western’ dietary pattern. Therefore, understanding how satisfaction with current body shape affects food preferences and the overall adolescent diet is a key issue for the development of strategies aimed at influencing dietary behaviour.
Limitations of the study
Dietary and physical activity data were based on self-reports(Reference Lazarou, Panagiotakos and Kouta16). The literature reports that food under-reporting is usually associated with gender and weight status(Reference Baxter, Smith and Litaker65, Reference Ventura, Loken and Mitchel66). Self-report of physical activity also can lead to over-report the physical activity due to social desirability bias, and therefore the number of inactive individuals may be greater than that reported(Reference Tammelin, Näyhä and Laitinen67, Reference Motl, McAuley and DiStefano68), especially among children and adolescents, and also among the obese(Reference Tammelin, Näyhä and Laitinen67). However, in many cases, self-reporting is the only feasible method of assessing physical activity(Reference Ekelund, Neovius and Linné69) and dietary intake in epidemiological studies. Although epidemiologists made every effort to get as accurate data as possible, there is a possibility that misreporting occurred(Reference Lazarou, Panagiotakos and Kouta16).
The FFQ did not differentiate between wholegrain and white bread. However, Ribas-Barba et al. (Reference Ribas-Barba, Serra-Majem and Salvador70) found a slight percentage of daily wholegrain consumers (14·2 %) among the Catalonia population aged 10–75 years, with a high percentage of never consumers of wholegrain bread (71·3 %). Therefore, we would expect that bread, which was one of the main foods related to the ‘Western’ dietary pattern in the present study, was consumed mostly as white bread.
The statistical methods usually applied in nutritional epidemiology to define dietary patterns based on collected dietary information contain a posteriori approaches, such as cluster and factor analyses, and a priori dietary index approaches(Reference Lioret, Touvier and Lafay40). In the present study an exploratory statistical analysis based on factor analyses was chosen. Factor analysis is a useful technique to summarise food patterns and relate them to different socio-economic and lifestyle factors(Reference Aranceta5) and body self-perception. However, it must be acknowledged that the method is data-specific, thus the patterns and their associations extracted in one study population may not be reproduced in other populations(Reference Aranceta5, Reference Schulze, Hoffman and Kroke71). This kind of analysis can facilitate the development of interventions aimed at modifying eating patterns, rather than specific components of the diet(Reference Aranceta5).
Body fat was calculated using Slaughter et al.'s equations (Reference Slaughter, Lohman and Boileau30) which have been suggested previously by Rodríguez et al.(Reference Rodríguez, Moreno and Blay31). However, the present study did not take into account pubertal development despite that chronological age may vary dramatically during this phase. Therefore, as in a previous study(Reference Alió-Sanz, Iglesias-Conde and Pernía72) in which adolescents have been classified according to their pubertal stage, boys were divided into two groups: pubertal (12–14 years old) and post-pubertal (15–17 years old).
Conclusions
The present study shows the existence of two major dietary patterns among Balearic Islands’ adolescents: ‘Western’ and ‘Mediterranean’, but girls are more ‘Mediterranean’ than boys. This evidence supports that the food pattern of Balearic Islands’ adolescents is in a transitional state characterised by the loss of the traditional Mediterranean dietary pattern towards a Western dietary pattern. Low parental socio-economic status, much leisure time on sedentary behaviours such as media screen time and body image are factors associated with the ‘Western’ dietary pattern.
Adolescents constitute priority targets for action and should be more aware about the Mediterranean diet and its health benefits. Programmes to promote the traditional Mediterranean dietary pattern among not only adolescents but also their families, combined with an active lifestyle, would likely result in a more favourable future health profile.
Acknowledgements
The study was supported by the Spanish Ministry of Health and Consumption Affairs (Programme of Promotion of Biomedical Research and Health Sciences, Projects 05/1276 and 08/1259; and Red Predimed-RETIC RD06/0045/1004) and the Spanish Ministry of Education and Science (FPU Programme, PhD fellowship to M.M.B.). The authors state that there are no conflicts of interest. A.P. and J.A.T. conceived, designed and devised the study; M.M.B., E.M., R.L. and J.A.T. collected and supervised the samples; M.M.B. and J.A.T. analysed the data and wrote the manuscript; A.P. and J.A.T. supervised the study; A.P. and J.A.T. obtained funding.