Lifestyle and dietary habits in the Mexican population have changed dramatically in the past 20 years, reflected in an increased prevalence of overweight and obesity in both urban and rural adult populations – from 33·4 % in 1988 to 59·6 % in 2000 (Mexican National Health and Nutrition Survey(Reference Rivera-Dommarco, Shamah-Levy and Villalpando-Hernández1)) and to 71·9 % in 2006(Reference Olaiz, Rivera and Shamah2). This represents a 41 % and 160 % increase in prevalence of overweight and obesity, respectively, in just a decade(Reference Barquera, Campos-Nonato and Hernández-Barrera3). Among adults, obesity is more prevalent among women than among men. In 2006, 36·9 % of women were obese (BMI ≥30·0 kg/m2), whereas only 23·5 % of men belonged to the obese group(Reference Barquera, Campos-Nonato and Hernández-Barrera3). A total of 35·1 % of Mexican adult women are estimated to be at risk for excessive carbohydrate intake and 12·6 % are deemed to be at risk for excessive fat intake(Reference Barquera, Hernández-Barrera and Campos-Nonato4). In women over 60 years of age, overweight and obesity rates are even higher – approximately 40 % are overweight and 35·5 % are obese – and a high prevalence of hypertension and diabetes is observed(Reference Shamah-Levy, Cuevas-Nasu and Mundo-Rosas5). Moreover, it appears that Mexicans have genetic susceptibility to insulin resistance and to altered carbohydrate and lipid metabolism(Reference Mitchell, Blangero and Comuzzie6). It is therefore important to determine dietary factors that are most strongly associated with obesity in this population to provide public awareness and prevention strategies.
Assessment of dietary patterns is an approach that analyses intakes of specific foods in the context of the whole diet(Reference Murtaugh, Herrick and Sweeney7) and may be of particular interest to public health, providing a basis to make recommendations for eating practices that prevent disease. Several dietary patterns have been previously defined and associated with chronic diseases(Reference Fung, Hu and Holmes8–Reference Slattery, Boucher and Caan13); dietary patterns are culturally defined, however, and differ between ethnically diverse populations. There exist very few data on dietary patterns associated with obesity in the Mexican population. Recently, Carrera et al.(Reference Carrera, Gao and Tucker14) evaluated data from the National Health and Nutrition Examination Survey (NHANES) on a small subsample of 659 Mexican-American adults without identifying a specific dietary pattern associated with obesity in this population.
Body shape silhouette is a promising tool to examine body size and image. Studies conducted in different countries have shown that the simple use of silhouette could adequately rank individuals according to body size(Reference Muñoz-Cachón, Salces and Arroyo15–Reference Tehard, van Liere and Com Nougué18). In a study conducted in Mexico, a very good correlation (0·77) was observed between BMI and classification of body shape silhouette among adult women(Reference Kaufer-Horwitz, Martínez and Goti-Rodríguez19). This approach has special value in large populations for which anthropometric measurements cannot be obtained.
As part of a large follow-up study of Mexican women to investigate the role of diet and lifestyle in the risk for chronic diseases, the association of self-perceived body shape silhouette and BMI with dietary patterns was evaluated. The dietary pattern most strongly associated with obesity and the usefulness of self-perceived body silhouette as proxy for body size in epidemiological studies were also assessed.
Methods
Study population
The source population for the present study consisted of a cohort of female teachers from the public education system who were aged ≥35 years and who were active members of the Federal and State level Ministry of Education's Economic Incentives Program called ‘Magisterial Career’ (MC). Using databases provided by the MC state programmes, we obtained the full name and place of work for active participants in the MC programme, identifying a total of 27 456 female teachers who met the age inclusion criterion (≥35 years old) in the state of Veracruz, a representative state of the Mexican population. These women work in 9699 different schools distributed across the entire geographical area, including urban and rural areas.
All 27 456 women were invited to participate in a cohort study to evaluate lifestyle and chronic disease incidence. In addition to being part of the MC programme and being ≥35 years of age, inclusion criteria included their consent to participate in the cohort study with planned long-term follow-up. The rationale for inviting MC programme participants was that MC members are highly motivated and trained individuals who are capable of responding to complex questionnaires in the setting of a developing country. Each potential participant received an invitation letter, a consent form, the baseline questionnaire, a promotional brochure and a fibreglass measuring tape in a personalized sealed envelope. Letters were sent through the state-level MC internal delivery network between 2006 and 2008. The questionnaire was self-administered, and women were asked to return the completed questionnaire to the National Institute of Public Heath in a return envelope provided, ensuring confidentiality. The protocol for the present study was approved by the Ethics Committee of the National Institute of Public Health, the Education Ministry, and the Health Ministry. A total of 20 258 questionnaires were returned, corresponding to a participation rate of 73·7 %. Respondents and non-respondents had a similar age distribution and were evenly distributed within the localities (with the exception of the city of Veracruz where respondents represented 15 % of the total sample and non-respondents represented 21 %). Among respondents, 18 875 questionnaires (93·2 %) had complete data on current silhouettes and anthropometric measures. The general characteristics of women did not differ between those who did and those who did not provide complete data on silhouette and anthropometric measures.
Baseline questionnaire
The questionnaire included general information on demographics, socio-economic status (SES; electrical appliance and car ownership, number of bedrooms and number of persons in the household), reproductive history and use of oral contraceptives, menopausal hormone therapy, clinical history, anthropometry, lifestyle (including an FFQ), physical activity (PA), smoking habits and early-life risk factors.
Anthropometry and silhouettes
Women were provided with a plastic measuring tape and a short set of instructions to measure their height, waist and hip circumference and weight. Waist-to-hip ratio (WHR) was determined from these measurements. BMI was calculated as weight in kilograms divided by the square of height in metres (kg/m2).
In addition, women reported their weight at 18 years and the maximum weight attained at any age during their lifetime. Women were also asked to select a body silhouette from nine options (from very thin to very fat) at six different ages: 2 years after menarche, between 18 and 20 years of age, before their first pregnancy, between 25 and 35 years of age and at their current age. Responses to these pictograms have been validated in different settings and have proven to be reliable(Reference Tehard, van Liere and Com Nougue20).
To validate the results reported on the questionnaire, 1000 women participating in the study were randomly selected and measured, weighed according to recommendations by Lohman et al.(Reference Lohman, Roche and Martorell21) and classified according to their body silhouette by a health professional during a clinical examination conducted in 2008. The reliability (concordance) of women reporting their silhouette when compared with that observed by a health professional was 0·70; the correlation between BMI based on weight and height reported on the questionnaire and that measured during the validation study was 0·90.
FFQ
Dietary intake was assessed using a semi-quantitative FFQ that included 116 food items with a standard portion size using the Mexican version of that developed by Willett(Reference Willett22). This questionnaire has been validated and has been shown to perform well among women residing in Mexico City(Reference Hernández-Avila, Romieu and Parra23, Reference Romieu, Parra and Hernández24). We added twenty-three food items to evaluate new food consumption patterns. The inclusion of these foods was based on the food reported in 24 h reports conducted as part of the Mexican National Health Survey(Reference Rivera-Dommarco, Shamah-Levy and Villalpando-Hernández1). For each food item, a commonly used unit or portion size was indicated (specified serving size: slice, glass or natural unit such as one apple) and participants were asked how often, on average over the previous year, they had consumed the specified amount of each food, choosing among ten frequencies of consumption: ≥6 times/d, 4–5 times/d, 2–3 times/d, 1 time/d, 5–6 times/week, 2–4 times/week, 1 time/week, 2–3 times/month, ≤1 time/month or never. Participants with an unreasonably high intake (>14 644 kJ (>3500 kcal)/d) or an unreasonably low intake (<2510 kJ (<600 kcal)/d) and those with more than seventy items left blank were excluded from the analysis (n 1752 or 9·3 %).
Physical activity
Women were asked to report a representative week in terms of the number of hours of mild, moderate and vigorous activity during work time, work at home or during recreational time on weekdays and weekends, as well as the number of hours of sleep. They were also asked to report the number of hours per week of watching television (TV). Specific definitions and examples were provided to the women on these different types of activities. Hours were summed to calculate the number of hours per week of mild, moderate and vigorous activity and to calculate MET-h/week (MET = metabolic equivalents), using the following coefficient to multiply the number of hours in each specific type of activity: mild (2·2), moderate (4·7) and vigorous (6·0).
Dietary pattern
Forty-eight food groups were defined (see Appendix) but only forty-six groups were included in the analysis because alcohol and atole (a Mexican drink) had an unsuitable distribution and a low consumption in our population. Food items were classified individually when their composition differed substantially from that of other foods or when they represented particular dietary habits, as mentioned by Hu(Reference Hu25).
Statistical analysis
Mean and sd were calculated for continuous variables, and frequencies were calculated for categorical variables. When the variable was not normally distributed, we used a transformation (in general, log transformation) to normalize the distribution. Categorical variables were compared using the χ 2 test, and continuous variables were calculated using the difference of mean tests. The SES index was calculated using factor analysis including the following variables: electrical appliance, car ownership, number of bedrooms and number of persons in the household. A single factor explained most of the variance. The SES factor was categorized into tertiles to define three levels of SES: low, medium and high.
Dietary pattern was defined using factor analysis as described earlier(Reference Hu25, Reference Varraso, Fung and Barr26). Factor analysis is a type of cluster analysis(Reference Dallas27) that determines those features that are most important when classifying a group of items and that generates ‘factor scores’ representing values of the underlying constructs for use in other analyses(Reference Dallas27). In our context, this analysis allowed the grouping of individuals on the basis of food group intakes. Factors were rotated by orthogonal transformation (Varimax) to achieve simpler structures with greater interpretation. The number of factors retained was determined using a diagram for eigenvalues, the scree plot and the percentage of variance explained. Foods that loaded ≥0·30 were considered to contribute to the factor, although the value for meaningful loading is arbitrary and we included two foods in the second factor with a load of 0·28 (milk beverages) and 0·24 (milk). The factor score for each pattern was constructed summing observed intakes of the component food items weighted by factor loading. Three major dietary patterns were identified: (i) fruit and vegetables; (ii) meat and dairy; and (iii) carbohydrates, sweet drinks and refined foods. To reduce measurement errors and represent long-term dietary patterns, the cumulative average of pattern scores was calculated and then divided into tertiles.
Current silhouette was categorized into three groups by re-grouping the nine categories provided on the questionnaire: lean women silhouette, 1–3; medium, 4–6; and large, 7–9. For change in silhouette from 18–20 years to current age we also defined three groups: (i) women who did not change silhouette or who decreased silhouette; (ii) women who increased one or two silhouettes; and (iii) women who increased three or more silhouettes. For BMI we used cut-off points as follows: <25·0 kg/m2, ≥25·0–<30·0 kg/m2 (overweight) and ≥30·0 kg/m2 (obese). For change in BMI from 18–20 years to current age, we categorized the difference into tertiles. Association of dietary pattern with current silhouette and change in silhouette and with current BMI and change in BMI was determined using logistic regression models, adjusting for potential confounders including age, socio-economic level, total energy intake, hours of watching TV and PA. Other potential confounders (marital status, smoking and parity) were not significant – they did not alter the results by >1 % – and were therefore not included in the final models. All analyses were conducted using the STATA statistical software package version 9·2 (StataCorp., College Station, TX, USA).
Results
Population characteristics
Table 1 presents general and sociodemographic characteristics of the 18 875 women included in the analysis.
Table 1 Sociodemographic characteristics of women from Veracruz; EsMaestras Cohort Study, Mexico
MET, metabolic equivalents.
*The total number for each variable shows some discrepancies because of different numbers of missing values.
†Data are presented as mean and sd.
‡On the basis of the number of hours of mild, moderate and vigorous activities during work time, work at home or during recreational time on weekdays and weekend days in a representative week, as well as the number of hours of sleep.
§Data are presented as median and 25th and 75th percentiles.
The mean age was 44 years in a range from 35 to 77 years. Most of the women lived with a partner. Approximately 12 % spoke a native language. Most of the women were non-smokers or former smokers and only 8·5 % reported being current smokers. The majority of women reported having one or two children with an age at first pregnancy between 16 and 39 years. Among teachers with children, 82·3 % reported breast-feeding. Women reported watching TV on average 5 h/week. PA was low; only 13·6 % reported some amount of vigorous PA (such as running or playing tennis). The mean moderate activity (such as playing volleyball, light bicycling, swimming, walking, dancing and home activities) was 6·7 (sd 3·2) h/week. In all, 74 % of women were premenopausal and 26 % reported natural menopause; 70 % of women reported living with a partner.
Table 2 presents anthropometric measurements. The median BMI was 26·8 kg/m2 (25th percentile = 24·2, 75th percentile = 30·0); 67 % of women were either overweight (43 %) or obese (24 %). In addition, women who reported living with a partner were heavier than women who reported being single (69 % v. 31 % with BMI ≥ 30·0 kg/m2; P = 0·000).
Table 2 Baseline anthropometric measurements of women from Veracruz; EsMaestras Cohort Study, Mexico
*Median and 5th and 95th percentiles are presented for women within this category of BMI.
Silhouette and change in silhouette
Overall, 34·3 % of women classified themselves as currently having a large body silhouette (7–9) and only 10·9 % as having a thin body silhouette (1–3). This contrasts with reports regarding body silhouette close to menarche, at which age 78·8 % of women reported a thin body silhouette and only 1·6 % reported a large body silhouette. A clear shift over time from a thinner to a larger silhouette was observed (Table 3, Fig. 1).
Table 3 Self-perception of body image at baseline and at different ages for women from Veracruz; EsMaestras Cohort Study, Mexico
*Some missing values.
†Only women with parity.
Fig. 1 (colour online) Frequency distribution of women from Veracruz by perceived silhouette at different ages; EsMaestras Cohort Study, Mexico
The majority of women increased silhouette from menarche to their current age, gaining up to three silhouettes over time. The change in silhouette from 18–20 years to current age was less, with the highest proportion of change corresponding to an increase of two silhouettes (25·7 %).
Dietary pattern and body silhouette
Foods most frequently consumed were tortilla and sweet breads, fruity and leafy vegetables (including tomato), rice and pasta, beans, fruit, milk, cereal, cream and butter. The fruit and vegetable pattern was composed mainly of fruit and vegetables, nuts and cereals. The meat and dairy pattern was composed mainly of meat, processed meat, fish, cream and butter, cheese and milk. The carbohydrates, sweet drinks and refined foods pattern was composed of tortilla and sweet bread, biscuits and pastry, rice and pasta, beans, jam and soft drinks. Table 4 displays factors for particular foods consumed in each of the three intake patterns.
Table 4 Factor-loading matrix for the major factors (dietary patterns) among women (n 19 705) from Veracruz; EsMaestras Cohort Study, Mexico
Factor 1, fruit and vegetables pattern; Factor 2, meat and dairy pattern; Factor 3, carbohydrates, sweet drinks and refined foods pattern.
Women with the highest (3rd tertile) intakes of carbohydrates, sweet drinks and refined foods were more likely to have a large body silhouette (7–9) when compared with women with the lowest intake (1st tertile; 1–3; ORT1−3 = 1·86; 95 % CI 1·56, 2·22, test for trend P = 0·001). In addition, women with the highest intakes of carbohydrates, sweet drinks and refined foods were more likely to have increased by ≥3 silhouettes from 18–20 years of age to current age (ORT1−3 = 1.56; 95 % CI 1·31, 1·85, test for trend P = 0·10).
In contrast, women with the highest fruit and vegetable intakes were less likely to have a large silhouette (ORT1−3 = 0·68; 95 % CI 0·57, 0·82) compared with women with the lowest intake, and to increase silhouette by ≥3 from 18–20 years of age to current age (ORT1−3 = 0·76; 95 % CI 0·64, 0·90). Significant trends were observed for both outcomes.
For the meat and dairy pattern, no significant relation was observed with current silhouette or change in silhouette (Table 5).
Table 5 Association of current silhouette and change in silhouette with dietary patterns in women from Veracruz; the EsMaestras Cohort Study, Mexico
All models were adjusted for age, socio-economic level, total energy intake, hours spent watching television and physical activiy in MET-h/week (MET = metabolic equivalents).
*Data are presented as multivariate OR and 95 % CI.
†Test for trend.
Dietary pattern and BMI
Women with the highest (3rd tertile) intakes of carbohydrates, sweet drinks and refined foods were also more likely to have a large BMI (≥30·0 kg/m2v. <25·0 kg/m2) when compared with women with the lowest intake (1st tertile), with a significant increasing trend with increasing tertiles of intake (ORT1−3 = 1·47; 95 % CI 1·28, 1·69, test for trend P < 0·00). In addition, women with the highest intakes of carbohydrates, sweet drinks and refined foods were more likely to have a larger increase in BMI, with a significant increasing trend with increasing intake (ORT1−3 = 1·27; 95 % CI 1·11, 1·44, test for trend P = 001).
As observed for silhouette, women with the highest fruit and vegetable intakes were less likely to have a large BMI and to increase BMI between 18 years and current age, with a significant trend with increasing tertiles of intake (ORT1−3 = 0·77; 95 % CI 0·67, 0·88 and ORT1−3 = 0·79; 95 % CI 0·69, 0·90, respectively). No clear effect was observed for the meat and dairy pattern (Table 6).
Table 6 Association of current BMI and change in BMI with dietary pattern in women from Veracruz; the EsMaestras Cohort Study, Mexico
All models were adjusted for age, socio-economic level, total energy intake, hours spent watching television and physical activity in MET-h/week (MET = metabolic equivalents).
*Data are presented as multivariate OR and 95 % CI.
†Test for trend.
‡Tertiles of BMI: −4·9–4·8 kg/m2; >4·8–8·3 kg/m2 and >8·3–30·5 kg/m2.
Discussion
A dietary pattern characterized by high intakes of carbohydrates, sweet drinks and refined foods was significantly associated with current large body size and obesity. Significant trends were observed with increasing tertiles of intake. In contrast, a dietary pattern characterized by high intakes of fruit, vegetables, grains and nuts appears to protect against large body size and obesity. These patterns also appeared to be related to a change in silhouette over time.
The prevalence of obesity and overweight is increasing in the Mexican population(Reference Shamah-Levy, Villalpando-Hernández and Rivera-Dommarco28) and this trend is made apparent by comparing national nutritional surveys conducted over the past 20 years. Overweight and obesity affect approximately 70 % of the Mexican population aged between 30 and 60 years for both sexes, with a greater percentage of obesity (approximately 32 %) among women (BMI ≥30·0 kg/m2). In our study, women from the state of Veracruz were included for strategic reasons: Veracruz had been the first state to be included in our teachers’ cohort. In all, 61 % of the teachers were either overweight (39 %) or obese (22 %). The National Nutrition and Health Survey 2006(Reference Olaiz, Rivera and Shamah2) is a cross-sectional survey conducted on a representative sample of the Mexican population and included objective anthropometric measurements among other evaluations (n 33 624 participants, among whom 9848 women were ≥20 years old). In that study, the median BMI was 26·8 kg/m2, similar to our results(Reference Barquera, Hernández-Barrera and Campos-Nonato4, Reference Shamah-Levy, Villalpando-Hernández and Rivera-Dommarco28).
High intakes of carbohydrates, sweet drinks and refined foods were strongly related to larger silhouettes and BMI. Although we were able to measure only current diet, our results also suggest that this dietary pattern was related to an increase in silhouette and BMI over time. This pattern was highly weighted by fast-absorbed carbohydrates (simple carbohydrates), in particular sugar-sweetened soft drinks, which are major contributors to energy intake and have been associated with high body fat and increased weight gain in other populations(Reference Carrera, Gao and Tucker14, Reference Novotny, Daida and Acharya29). The expert committee in charge of developing the beverage consumption recommendations for the Mexican population has observed that beverages contribute to one-fifth of all energy consumed by Mexicans and are strongly related to obesity. Among the biggest contributors are beverages high in sugar and low in nutritional value (soft drinks and other beverages with significant amounts of added sugar, such as juices, flavoured water, coffee and tea)(Reference Rivera, Barquera and González-Cossío30).
In our population, an intake pattern rich in fruits, vegetables and cereals was associated with a thin figure, low BMI and the smallest change in silhouette and gain in BMI over time. Other studies have observed that a ‘prudent pattern’ with high intakes of fruit and vegetables is negatively associated with higher BMI(Reference Togo, Osler and Sørensen31). In prospective settings, Newby et al. reported an association of a diet high in reduced-fat dairy products, whole grains and fruit and low in refined grains, processed and red meat, fast food and soda with smallest gain in BMI and waist circumference(Reference Newby, Muller and Hallfrisch11, Reference Newby, Muller and Hallfrisch32).
Two recent studies have evaluated dietary patterns in Mexican Americans and identified differences among them. In the NHANES 2001–2002, which included 659 Mexican-American adults, four dietary patterns were identified: poultry and alcohol, milk and baked products, traditional Mexican food and meat. None of these patterns was specifically related to obesity. The traditional Mexican pattern had a high percentage of energy from tortillas, tacos, flavoured and sweetened drinks and legumes. Surprisingly, no ‘healthy’ pattern group was identified in this population. The small sample size and assessment of dietary intake based on 24 h recall may explain the lack of significant findings. That analysis with NHANES data as well as ours observed low levels of leisure-time PA. In another study conducted among women living in southwestern USA, including 871 Hispanic women and 1599 non-Hispanic women, five dietary patterns were determined: Western, native Hispanic, prudent, Mediterranean and dieter. The native Hispanic pattern was heavily loaded with Mexican cheese, soups, meat, legumes, tomato sauce and sugar-sweetened drinks. Although not significant, the highest tertile of a native Mexican diet was related to a 64 % increased risk of being obese (95 % CI 0·83, 3·18) and the highest tertile of a prudent diet was related to a decreased risk of being obese (OR = 0·54; 95 % CI 0·29, 1·03). Although these results are concordant with ours, the small sample size of Mexican women might have limited the ability of the study to detect significant associations(Reference Murtaugh, Herrick and Sweeney7). A recently published study of data from the National Health Survey in Mexico has also observed in a sample of men and women that overweight and obesity are significantly related to a dietary pattern rich in refined foods and sweets(Reference Flores, Macias and Rivera33).
Several issues need to be discussed. A limitation for interpretation of our results is the cross-sectional nature of our data. Current body silhouette and BMI were reported at the same time as dietary intake; hence, we cannot conclude causality. Heavy women might have gone through changes in dietary pattern to lose weight and may therefore report a diet different from their usual one. However, had this occurred, we would have found a positive association between obesity and the fruit and vegetable pattern, which is not the case. In addition, the consistency of the results using body silhouette, which women appear to underestimate, and BMI strengthens the validity of our results. Our results are consistent with the fact that intake of a diet high in carbohydrates, sweet drinks and refined foods is associated with obesity, whereas a diet rich in fruit and vegetables is protective against obesity. Another limitation is that because dietary intake was reported at baseline, misclassification might have occurred, in particular when considering diet over a long period of time, as in our analysis of change in silhouette. Although previous studies have shown a reasonable tracking of diet over time(Reference Willett34), we do not have data to confirm this in our population. However, our results suggest that dietary pattern may explain part of the change of body silhouette over time, given that random misclassification in dietary pattern would tend to underestimate the association.
The overall consistency of our observations with both body shape silhouette and BMI suggests that in our population women who follow the carbohydrates, sweet drinks and refined foods pattern have a risk factor for large body size and for an increase in body size over time. Body silhouette appears to classify individuals by body size well and hence could be used in epidemiological studies to study obesity and its determinants.
The present study included a large sample of Mexican women, using a validated FFQ to evaluate common diet, anthropometric measurements and self-reported body silhouette. Self-reported body silhouette had a good agreement with observed body silhouette in other studies(Reference Muñoz-Cachón, Salces and Arroyo15–Reference Tehard, van Liere and Com Nougué18). In addition, the validation study showed good agreement between reported and measured BMI and perceived and observed body silhouettes. Must et al.(Reference Must, Willett and Dietz35) have shown good concordance between recalled and observed body silhouette. Although some misclassifications cannot be avoided, they would most likely be random because women were not influenced by their dietary pattern when reporting body silhouette at different ages. Among Mexican women, knowledge of dietary factors related to obesity is very limited. The association that we observed would therefore be underestimated.
Conclusion
The significant associations observed between a dietary pattern high in carbohydrates, sweet drinks and refined foods and larger body figures and BMI plus an increase in body figure and BMI over time, as well as the protective effect of a dietary pattern high in fruit and vegetables, strongly support the fact that a diet with a high glycaemic load (corresponding to the carbohydrate content of one serving multiplied by the glycaemic index value of that food) is a risk factor for obesity and weight gain in Mexican women. Such foods, in particular sugar-sweetened drinks, pastry and jam, should be restricted. This is particularly important because the intake of carbohydrates is high for the Mexican population (a mean daily carbohydrate intake of 357 g/d), accounting for 64 % of total energy intake(Reference Rivera-Dommarco, Shamah-Levy and Villalpando-Hernández1), whereas the intake of fruit and vegetables is low (97 % of Mexican women consume <400 g/d). In addition, Mexicans appear to have a genetic susceptibility to insulin resistance and altered carbohydrate and lipid metabolism(Reference Mitchell, Blangero and Comuzzie6), with a potential for severe health consequences. As part of a weight-control programme, a decreased intake of quickly absorbed carbohydrates, sweet drinks and refined foods coupled with an increased intake of fruit and vegetables should be promoted along with PA. Several actions could be further developed: improving knowledge on food-based dietary guidelines for our population; providing healthy food choices within canteens and promoting PA such as walking to work and gym activity during lunch break; emphasizing on the responsibility of role modelling of teachers towards students; and supporting the regulation of refined foods, high sugar-sweetened foods and sweet drinks.
Acknowledgements
The present study was supported by the American Institute for Cancer Research (Grant no. 05B-047) and by Mexican Sciences and Technology Council (CONACYT; Grant no. 2005-02-14429). The authors have no conflict of interest to declare. I.R. developed the protocol, obtained funding for the project and directed the data analysis and the writing of the manuscript; M.C.E.-N. participated in the protocol and in data collection; L.M.S.-Z. participated in data analysis; R.L. participated in the protocol, in coordination of data collection and in writing of the manuscript; G.T.-M. participated in the protocol and in interpretation of data; E.M.Y. participated in data collection and in standardization of the questionnaires; M.L. participated in the protocol and in interpretation of data; J.A.R.-D and E.L.-P. participated in the interpretation of data and in writing of the manuscript. All authors approved the final report. We are grateful to all participant teachers, who provided invaluable information for the study, for their time and commitment. We would like to thank the Federal Coordinación Nacional de Carrera Magisterial (Coordination Office of the Magisterial Career Program), with special thanks to Victor Sastré, Director of Regulation, Dolores Cruz and María del Carmen Placencia, coordinators from Veracruz and Jalisco States, as well as all their staff for their continued support. We thank also all educational and health authorities who have actively collaborated in the project.
Appendix
