To capture a more holistic picture of eating behaviour by investigating the impact of the social living conditions and weight status of parents and daughters on food consumption frequency, the context of meals and daily portion sizes.

Cross-sectional Berlin School Children's Cohort study.

A total of sixty-nine schools in Berlin (3 400 000 inhabitants, eastern Germany) participated in the present study.

A total of 1519 girls aged 11–14 years were selected. Bi- and multivariate analyses were performed to examine the impact of age, migration background, socio-economic status (SES), parental education, family situation and the weight status of parents and daughters on three different eating behaviour scores according to nutritional recommendations.

For the three dependent eating behaviour variables, different patterns of influencing factors emerged. Multivariate regression (model 1) revealed that low and middle SES, two-parent migration background and older age were significant risk factors. Meal context was also significantly influenced by living with a single parent. Similar results were obtained for the daily portion size scores and maternal overweight status was the most influential. Model 2 succeeded in showing that, within the composite variable of family SES, mothers’ level of education was the dominant component.

SES as a whole, and especially the component of mothers’ level of education and two-parent migration background, was the strongest risk factor for an unfavourable eating pattern among adolescent girls. The results clearly indicated preventive potential. Using three different measures of eating behaviour simultaneously provided an in-depth understanding of general patterns and potential risk factors.

To identify child and parental factors associated with screen time in 3-year-old children.

Observational study.

Participants were recruited from a large primary-care paediatric group practice in Toronto, Canada.

Healthy 3-year-old children were included. A questionnaire was completed by their parents on screen time. Descriptive statistics and linear regression models were used to assess associations between child screen time and selected factors. Multivariable models included factors from the univariate analysis with P < 0·1. Estimated effects and 95 % CI are reported.

A total of 157 children were enrolled (91 % recruitment). The mean screen time per weekday was 104 min (similar for weekend day). In all, 10 % of children had a television (TV) in their bedroom; 59 % consumed at least one meal while watching TV; and 81 % of parents had household rules about screen time. Controlling for maternal education and age, eating lunch and dinner in front of the screen and mother being employed were associated with an increase in child weekday screen time of 96 (95 % CI 30, 192), 42 (95 % CI 12, 90) and 36 (95 % CI 6, 72) min/d, respectively. Eating lunch in front of the screen and an increase of 1 h of parental screen time were associated with an increase of 78 (95 % CI 36, 132) and 12 (95 % CI 6, 18) min/d in child weekend screen time. Family rules decreased child weekend screen time by 30 (95 % CI 6, 54) min/d.

Interventions that include these important parental factors should be evaluated for their effectiveness in reducing screen time.

The purpose of the present study was to investigate the nutrition information needs of the Pacific Islander and African migrant communities of Logan.

The present study was structured as a needs assessment and used qualitative research methods. An integrative review was used to gather and analyse data relating to comparative and normative needs, while semi-structured interviews were undertaken for the felt and expressed needs.

Logan City, Queensland, Australia.

African and Pacific Islander migrants and health and social service providers within the Logan region.

The study identified the need for more accessible means of information delivery such as visual and face-to-face methods or hands-on demonstrations. The study found that information should be delivered in a staged approach on topics including food safety, Australian foods, healthy eating and diet–disease relationships, according to a migrant’s length of residence in Australia.

The present study contributes to our understanding of the nutrition information needs of African and Pacific Islander migrant communities of Logan. These findings will enable the development of more appropriate nutrition information and health services for these Logan communities and other similar communities across Australia.

To compare food consumption during television (TV) viewing among adolescents who watched >2 h/d v. ≤2 h/d; and to examine the association between sociodemographic variables (age, gender and socio-economic status (SES)) and the consumption of energy-dense foods and drinks during TV viewing.

The data are part of the Healthy Lifestyle in Europe by Nutrition in Adolescence (HELENA) cross-sectional survey. Data on time watching TV, types of foods and drinks consuming during TV viewing and parental SES (parental education, parental occupation and family affluence) were measured by questionnaires completed by adolescents. Binary logistic regression tested the association between energy-dense foods and drinks and (i) sociodemographic variables and (ii) TV time.

Ghent (Belgium), Heraklion (Greece), Pecs (Hungary) and Zaragoza (Spain).

Girls (n 699) and boys (n 637) aged 12·5–17·5 years.

Boys reported more frequent consumption of beer and soft drinks whereas girls selected more fruit juice, water, herbal infusions and sweets (all P ≤ 0·05). Watching TV for >2 h/d was associated with the consumption of energy-dense foods and drinks. Girls whose mothers achieved the lowest education level had an adjusted OR of 3·22 (95 % CI 1·81, 5·72) for the consumption of energy-dense drinks during TV viewing v. those whose mothers had the highest educational level.

Excessive TV watching may favour concurrent consumption of energy-dense snacks and beverages. Adolescents from low-SES families are more likely to consume unhealthy drinks while watching TV.

Little is known of the beverage intake patterns of Canadian children or of characteristics within these patterns. The objective was to determine beverage intake patterns among Canadian children and compare intakes of fourteen types of beverages, along with intakes of vitamin C and Ca, and sociodemographic factors across clusters.

Dietary information was collected using one 24 h recall. Sociodemographic data were collected by interview. Cluster analysis was used to determine beverage intake patterns. Pearson's χ2 and 95 % CI were used to test differences across clusters.

Data from the Canadian Community Health Survey Cycle 2·2.

Children aged 2–18 years with plausible energy intake and complete sociodemographic data (n 10 038) were grouped into the following categories: 2–5-year-old boys and girls, 6–11-year-old girls, 6–11-year-old boys, 12–18-year-old girls and 12–18-year-old boys.

Five beverage clusters emerged for children aged 2–5 years, six clusters for children aged 6–11 years (both sexes) and four clusters for those aged 12–18 years (both sexes). Sweetened beverage clusters appeared in all age–sex groups. Intakes of sweetened beverages ranged from 553 to 1059 g/d and contributed between 2 % and 18 % of total energy intake. Girls 6–11 years of age in the ‘soft drink’ cluster had lower Ca intake compared with other clusters in that age–sex group. Age and ethnicity differed across clusters for most age–sex groups. Differences for household food security status and income were found; however, no pattern emerged.

Patterns in beverage intake among Canadian children include beverages that are predominantly sugar sweetened. Public health nutrition professionals can use knowledge about beverage patterns among children, as well as the characteristics of these groups, in the development of nutritional programmes and policies.

The present study was undertaken to assess the status of vitamins A and E (VA and VE, respectively) and their main determinants in Tunisian children.

Cross-sectional population-based study.

Kasserine Governorate in the centre west of Tunisia.

A total of 7407 children attending the first grade of elementary school were included. VA and VE were assessed by HPLC.

The prevalence of moderate VA deficiency (VAD; <0·70 μmol/l) was 2·3 % and VE deficiency (VED; <6·97 μmol/l) was 5·4 %. Low status in VA (0·70–1·05 μmol/l) and VE (6·97–11·61 μmol/l) was observed in 17 % and 20·2 % of children, respectively. No child exhibited severe VA or VE deficiency (<0·35 and <2·32 μmol/l, respectively). The main predictors of VAD were advanced age (OR = 1·65; 95 % CI 1·13, 2·41; P = 0·05) and sickness within the past 2 weeks (OR = 1·51; 95 % CI 1·09, 2·09; P = 0·01). Predictors of VED were living in the peri-urban region (OR = 1·60; 95 % CI 1·28, 2·01; P < 0·001) and sickness within the past 2 weeks (OR = 0·75; 95 % CI 0·60, 0·94; P = 0·01).

Moderate VAD and VED were uncommon in Tunisian children. However, low status in VA and/or VE remains frequent. A reinforcement of the national strategies for children's nutrition and health is needed, particularly in disadvantaged regions. Supplementation of VA and VE is not necessary in Tunisia, but food fortification may be beneficial.

The Dutch Obesity Intervention in Teenagers (DOiT) is a school-based randomised controlled trial that was effective in decreasing the consumption of sugar-sweetened beverages among adolescents. The present study examined, using mediation analysis, whether this decrease in consumption of sugar-sweetened beverages could be explained by an increase in the consumption of water or diet drinks.

Participants completed a questionnaire about their beverage consumption at baseline and at 8 months (immediately post-intervention), 12- and 20-month follow-ups. A series of multi-level linear regression analyses were performed to examine water and diet drink consumption as potential mediators of the intervention effect on the consumption of sugar-sweetened beverages.

Eighteen Dutch secondary schools.

A total of 747 adolescents (mean age: 12·7 years).

In addition to the DoiT intervention effect of a reduction in the consumption of sugar-sweetened beverages at 8 months (−284 ml/d; 95 % CI −420, −148) and 12 months (−260 ml/d; 95 % CI −360, −160), there was also a significant reduction in diet drinks at 8 months (−52 ml/d; 95 % CI −89, −16). There was no significant difference in water consumption at any follow-up. The decrease in sugar-sweetened beverage consumption could not be explained by an increase in water or diet drink consumption at any time point.

Interventions aimed at reducing sugar-sweetened beverage consumption may be effective without changing consumption of other beverages. Reducing sugar-sweetened beverages was, however, a main message of the DOiT intervention. It is possible that a concomitant promotion of water may have resulted in a greater increase in water intake and replacement of sugar-sweetened beverages with water.

To identify and characterize dietary patterns in Lebanon and assess their association with sociodemographic factors, BMI and waist circumference (WC).

A cross-sectional population-based survey. In a face-to-face interview, participants completed a brief sociodemographic and semiquantitative FFQ. In addition, anthropometric measurements were obtained following standard techniques. Dietary patterns were identified by factor analysis. Multivariate linear regression was used to assess determinants of the various patterns and their association with BMI and WC.

National Nutrition and Non-Communicable Disease Risk Factor Survey (2009), Lebanon.

A nationally representative sample of 2048 Lebanese adults aged 20–55 years.

Four dietary patterns were identified: ‘Western’, ‘Traditional Lebanese’, ‘Prudent’ and ‘Fish and alcohol’. Factor scores of the identified patterns increased with age, except for the Western pattern in which a negative association was noted. Women had higher scores for the prudent pattern. Adults with higher levels of education had significantly higher scores for the prudent pattern. The frequency of breakfast consumption was significantly associated with scores of both traditional Lebanese and prudent patterns. Multivariate-adjusted analysis revealed a positive association between scores of the Western pattern and the BMI and WC of study participants.

The findings show the presence of four distinct dietary patterns in the Lebanese population, which were associated with age, sex, education and meal pattern. Only the Western pattern was associated with higher BMI.

To compare the prevalence rates of obesity based on BMI/anthropometric measures, using WHO standard and ethnicity-specific criteria, the National Cholesterol Education Program–Adult Treatment Panel III (NCEP-ATPIII) and the International Diabetes Federation (IDF) definitions, among a migrant South Asian Indian population.

Cross-sectional study conducted in October 2007.

A total of 213 participants of South Asian descent over the age of 18 years. Measures included a questionnaire with basic demographic information and self-reported histories of diabetes, coronary artery disease and/or hypercholesterolaemia. Height, weight, waist and hip circumference and blood pressure measurements were obtained.

Houston and surrounding suburbs.

WHO-modified (WHO-mod) BMI and IDF waist circumference (WC) criteria independently identified higher numbers of overweight/obese participants; however, when the WHO-mod BMI or IDF WC criteria were applied, nearly 75 % of participants were categorized as overweight/obese – a proven risk factor for the future development of metabolic syndrome.

Obesity is likely under-diagnosed using the standard WHO and NCEP-ATPIII guidelines. Stressing the use of modified criteria more universally to classify obesity among South Asian Indians may be optimal to identify obesity and help appropriately risk stratify for intervention to prevent chronic diseases.

To examine the prevalence and correlates of fruit and vegetable (FV) consumption in Cypriot adolescents.

A cross-sectional study.

The Republic of Cyprus.

A total of 1966 adolescents with a mean age of 14·7 (sd 2·2) years from nine elementary (n 448), six middle (n 657), five high (n 475) and five technical/vocational schools (n 386) in Cyprus. Participants completed a questionnaire assessing FV consumption using a two-item screening measure and a number of social, attitudinal and behavioural correlates of FV consumption.

Overall, 19·3 % of adolescents reported consuming five or more portions of FV daily, with elementary and middle school students more likely to meet recommendations (23·8 % and 24·4 %, respectively) compared with high and technical/vocational school students (14·0 % and 12·5 %, respectively). Consuming five or more portions of FV was associated with preference for FV (OR = 2·2), family eating patterns (OR = 1·5), friends’ FV consumption (OR = 1·2) and school support for FV consumption (OR = 0·8). Consuming at least one portion of fruit daily was significantly associated with preference for FV (OR = 2·0) and family eating patterns (OR = 1·7). Consuming at least one portion of vegetables daily was associated with preference for FV (OR = 4·2) and eating while watching television (OR = 0·8).

Targeting individual and family-based components may enhance the effectiveness of intervention programmes to promote FV consumption.

To determine whether video-based instructions improve the accuracy of self-measures of waist and hip circumference compared with written instructions.

Population-based, cross-sectional study. Self-measurements of waist circumference (WC) and hip circumference (HC) of fifty-seven participants randomly allocated to receive either written instruction or video instruction were compared with those of a trained technician.

Aberdeen, Scotland, and Brussels, Belgium, between February and April 2010.

Adults aged 18–62 years with a high level of English language and no prior training in anthropometry.

WC was significantly overestimated by the written method (1·75 cm bias; P = 0·007) but not the video method (0·95 cm bias; P = 0·239). HC was significantly underestimated in both written (−0·35 cm bias; P = 0·009) and video methods (−0·75 cm bias; P = 0·046). Reliability was not significantly affected by age, sex, BMI or WC. Bland–Altman plots demonstrated wide limits of agreement for WC (−6·83, 6·08 cm for written method; −10·14, 6·72 cm for video method) and HC (−12·85, 1·60 cm for written method; −10·82, 2·50 cm for video method).

Video technology can support more accurate self-measurements of anthropometric data in epidemiological studies. Further research is warranted using larger and more heterogeneous samples in order that results can be generalised.

The Groningen Overweight and Lifestyle (GOAL) study primarily aims at preventing weight gain by nurse practitioners (NP) guided by a standardized computerized software program. Since favourable changes in physical activity (PA) and diet may improve health independently of weight (loss), insight into effects on lifestyle habits is essential. We examined the 1-year effects of lifestyle counselling by NP on PA and diet, compared with usual care from the general practitioner (GP-UC).

A randomized controlled trial.

Eleven general practice locations in the Netherlands.

A total of 341 GOAL participants with overweight or obesity and either hypertension or dyslipidaemia, or both, who completed an FFQ and Short Questionnaire to Assess Health-Enhancing Physical Activity (SQUASH) at baseline and after 1 year.

After 1 year, the NP group spent 33 min/week more on walking compared with the GP-UC group who spent −5 min/week on walking (P = 0·05). No significant differences were found between the NP and GP-UC groups on the percentage of persons complying with the PA guidelines. In both groups, nutrient intake changed in a favourable direction and participants complied more often with dietary guidelines, but without overall difference between the NP and GP-UC groups.

With the exception of an increase in walking (based on self-reported data) in the NP group, no intervention effects on PA and diet occurred. Positive changes in nutrient intake were seen in both groups.

The Mediterranean diet has been related to a lower risk of some chronic diseases, including cancer. We aim to gain insight into the effects of the main source of fat of this diet on breast cancer, the most common type of malignancy in women.

Data from sixteen experimental series analysing the effects of dietary lipids on mammary carcinogenesis in an animal model, in the context of the international literature on the Mediterranean diet, olive oil and breast cancer risk.

Experimental and human data on the effects of olive oil and Mediterranean diet on breast cancer.

An animal model of induced breast cancer and other human and experimental studies in the literature.

Diets rich in extra virgin olive oil (EVOO) exert a negative modulatory effect on experimental breast cancer to a weak promoting effect, much lower than that obtained with a high-corn oil diet. EVOO confers to the mammary adenocarcinomas a clinical behaviour and morphological features compatible with low tumour aggressiveness. This differential effect, in relation to other dietary lipids, may be related to a lower effect on body weight and sexual maturation. In addition, EVOO induced different molecular changes in tumours, such as in the composition of cell membranes, activity of signalling proteins and gene expression. All these modifications could induce lower proliferation, higher apoptosis and lower DNA damage. These results, together with the favourable effect of olive oil reported in the literature when it is consumed in moderate quantities, suggest a beneficial influence of EVOO on breast cancer risk.

Consumption of EVOO in moderate quantities and throughout the lifetime appears to be a healthy choice and may favourably influence breast cancer risk.

The objective was to examine the tracking of obesity from childhood to young adulthood in Slovenia.

A subsample included 4833 children from the Slovenian national sample (n 21 777) who were included in the SLOFIT monitoring system from 1997 to 2008, with complete data at 7 years and 18 years.

A 12-year prospective cohort study. Height, weight and triceps skinfold thickness (TSF) were measured at 7, 11, 14 and 18 years. The BMI cut-off points of the International Obesity Taskforce were used to identify excess weight and obesity.

The survey was conducted in Slovenian primary and secondary schools from 1988 to 2008 but only the data from the measurements in 1997, 2001, 2004 and 2008 are presented.

Height, weight and BMI at 18 years were well predicted from childhood and became more predictable with age, while TSF was not. Obese and overweight children had the greatest risk of becoming obese or overweight young adults. The history of their weight shows that 40·0 % of males and 48·6 % of females who were obese at 18 years had already been obese at 7 years.

Overweight or obese Slovenian children are more likely to become overweight or obese adolescents and young adults than reported in other similar European and American studies, which indicates the need for early prevention and treatment of excess weight and obesity.

To examine the adequacy and inadequacy of dietary patterns in pregnant women for which information is absolutely lacking.

Diet was assessed by a validated, self-administered diet history questionnaire (DHQ). Dietary patterns were extracted from the intake of thirty-three food groups (g/4184 kJ (1000 kcal)), which were summarized from 147 foods assessed with the DHQ, by cluster analysis. Nutritional inadequacy for selected twenty nutrients in each dietary pattern was examined using the reference values given in the Dietary Reference Intakes (DRI) for Japanese as the temporal gold standard.

Japan.

Nine hundred and ninety-seven pregnant Japanese women aged 18–43 years.

The three dietary patterns identified were labelled as ‘meat and eggs’ (n 423), ‘wheat products’ (n 371) and ‘rice, fish and vegetables’ (n 203). The ‘rice, fish and vegetables’ pattern characterized by high intake of rice, vegetables, potatoes, pulses, fruits, seaweed, fish and miso soup showed significantly the lowest prevalence of inadequate intake for fifteen nutrients and significantly the highest prevalence of inadequate sodium intake. In contrast, the ‘wheat products’ pattern characterized by high intake of bread, noodles, confectioneries and soft drinks showed the highest prevalence of inadequate intake for fourteen nutrients. The median number of nutrients not meeting the DRI as a marker of overall nutritional inadequacy was eight in the ‘rice, fish and vegetables’ pattern. It was significantly lower at ten in the ‘meat and eggs’ and eleven in the ‘wheat products’ patterns (P < 0·001).

In pregnant Japanese women, the dietary pattern high in rice, fish, vegetables, fruit and some others showed a better profile of nutritional adequacy except for sodium.

To examine socio-economic differences in weight-control behaviours (WCB) and barriers to weight control.

A cross-sectional study.

Data were obtained by means of a postal questionnaire.

A total of 1013 men and women aged 45–60 years residing in Brisbane, Australia (69·8 % response rate).

Binary and multinomial logistic regression analyses were performed, adjusted for age, gender and BMI. Socio-economically disadvantaged groups were less likely to engage in weight control (OR for lowest income quartile = 0·60, 95 % CI 0·39, 0·94); among those who engaged in weight control, the disadvantaged group had a likelihood of 0·52 (95 % CI 0·30, 0·90) of adopting exercise strategies, including moderate (OR = 0·56, 95 % CI 0·33, 0·96) and vigorous (OR = 0·47, 95 % CI 0·25, 0·89) physical activities, compared with their more-advantaged counterparts. However, lower socio-economic groups were more likely to decrease their sitting time to control their weight compared with their advantaged counterparts (OR for secondary school or lower education = 1·78, 95 % CI 1·11, 2·84). They were also more likely to believe that losing weight was expensive, not of high priority, required a lot of cooking skills and involved eating differently from others in the household.

Marked socio-economic inequalities existed with regard to engaging in WCB, the type of weight-control strategies used and the perceived barriers to weight control; these differences are consistent with socio-economic gradients in weight status. These factors may need to be included in health promotion strategies that address socio-economic inequalities in weight status, as well as inequalities in weight-related health outcomes.

To investigate the reliability and the validity of the long format, Chinese version of the International Physical Activity Questionnaire (IPAQ-LC).

Cross-sectional study, examining the reliability and validity of the IPAQ-LC compared with a physical activity log (PA-log) and objective accelerometry.

Self-reported physical activity (PA) in Hong Kong adults.

A total of eighty-three Chinese adults (forty-seven males, thirty-six females) were asked to wear an ActiTrainer accelerometer (MTI-ActiGraph, Fort Walton Beach, FL, USA) for >10 h over 7 d, to complete a PA-log at the end of each day and to complete the IPAQ-LC on day 8. On a sub-sample of twenty-eight adults the IPAQ-LC was also administered on day 11 to assess its reliability.

The IPAQ-LC had good test–retest reliability for grouped activities, with intra-class correlation coefficients ranging from 0·74 to 0·97 for vigorous, moderate, walking and total PA, with between-test effect sizes that were small (<0·49). The Spearman correlation coefficients were statistically significant for vigorous PA (r = 0·28), moderate + walking PA (r = 0·27), as well as overall PA (r = 0·35), when compared with the accelerometry-based criterion measures, but none of the IPAQ activity categories correlated significantly with the PA-log. In absolute units, only the IPAQ light and overall PA did not differ significantly from the accelerometry measures, yet overall PA was able to faithfully discriminate between quartiles of PA (P = 0·019) when compared to accelerometry.

The IPAQ-LC demonstrated adequate reliability and showed sufficient evidence of validity in assessing overall levels of habitual PA to be used on Hong Kong adults.

The purpose of the present study was to analyse the evolution of the Spanish population’s global and regional adherence to the Mediterranean diet (MD) since 1987.

A cross-sectional study was conducted analysing food consumption from the Spanish Household Consumption Surveys administered by the Spanish Ministry of Agriculture, Fisheries and Food.

The Mediterranean Adequacy Index (MAI) was applied to measure the adherence to the MD. The significance of the MAI trend was evaluated using a regression model.

The sample comprises a household consumption data set, which consisted of 2500 households in 1987 and increased to 6200 in 2005.

There was a significant decrease in the MAI (P = 0·001) during the past two decades in Spanish households. From 1987 to 1997, MD adherence significantly decreased, but since 1998 it appears to have stabilised. In recent years, the daily household availability of several of the food components of the MD has increased. The MAI was observed to be significantly higher during the entire period in some Mediterranean regions, such as Andalusia, and significantly lower in more inland regions, such as Castile-Leon.

During the past two decades, Spain has experienced a significant downward trend in adherence to the predominant dietary pattern, likely as a result of numerous socio-economic changes. However, a stabilisation and a recent slight recovery have been observed during the past decade.

To study the association between the availability of healthy foods and BMI by neighbourhood race and socio-economic status (SES).

Trained staff collected demographic information, height, weight and 24 h dietary recalls between 2004 and 2008. Healthy food availability was determined in thirty-four census tracts of varying racial and SES composition using the Nutrition Environment Measures Survey–Stores in 2007. Multilevel linear regression was used to estimate associations between healthy food availability and BMI.

Baltimore City, Maryland, USA.

Adults aged 30–64 years (n 2616) who participated in the Healthy Aging in Neighborhoods of Diversity across the Life Span study.

Among individuals living in predominantly white neighbourhoods, high availability of healthy foods was associated with significantly higher BMI compared with individuals living in neighbourhoods with low availability of healthy food after adjustment for demographic variables (β = 3·22, P = 0·001). Associations were attenuated but remained significant after controlling for dietary quality (β = 2·81, P = 0·012).

Contrary to expectations, there was a positive association between the availability of healthy food and higher BMI among individuals living in predominantly white neighbourhoods. This result could be due to individuals in neighbourhoods with low healthy food availability travelling outside their neighbourhood to obtain healthy food.

To identify food sources of nutrients in adolescents’ diets and to identify differences in food sources according to individual characteristics.

A cross-sectional evaluation was carried out in the 2003/2004 school year. Self-administered questionnaires were used and a physical examination was performed. Diet was evaluated using an FFQ.

Public and private schools in Porto, Portugal.

Adolescents aged 13 years (n 1522) enrolled at school.

The main sources of energy were starchy foods (26·5 %), dairy (12·5 %) and meat (12·0 %). The major contributors to carbohydrate intake were starchy foods (38·2 %) and fruit (13·8 %) and to protein intake were meat (28·0 %), dairy products (20·3 %), starchy foods (15·3 %) and seafood (13·6 %). The main sources of total fat were meat (22·0 %), starchy foods (13·4 %) and dairy products (12·7 %). Sweets and pastries presented important contributions to energy (11·1 %), carbohydrate (12·4 %), total fat (13·3 %) and saturated fat (16·6 %) intakes. Parental education was inversely associated with the contribution of sweets and pastries to energy, carbohydrate and fat intakes and it was positively associated with the seafood contribution to protein intake.

The major sources of carbohydrates were starchy foods, which also accounted for a quarter of energy intake. Dairy products plus meat accounted for another quarter of energy. Meat was a major source of protein and fats. Sweets and pastries contributed more than 10 % to energy, carbohydrates, total and saturated fat. Parental education was the strongest determinant of food sources and was positively associated with a healthier contribution of food groups.