Overconsumption of fatty foods has been identified as a risk factor in major chronic diseases, including CVD, diabetes and obesity( 1 ). Moreover, fat contributes to eating pleasure through its sensory properties, and influences food choices and dietary intake( Reference Drewnowski 2 – Reference Mela 4 ). Taste ranks first among the reasons for individual food and beverage choices, ahead of price, health or convenience( Reference Glanz, Basil and Maibach 5 ); individual sensory liking thus appears to be an important determinant of dietary intake( Reference Drewnowski and Hann 6 – Reference Raynor, Polley and Wing 10 ). According to Mela( Reference Mela 11 ), liking is defined as a qualitative hedonic evaluation of a food: the degree of pleasure or displeasure experienced. In the present study, we examined overall liking for sensory sensations measured via an original questionnaire.
In order to better understand the framework in which taste sensitivity, sensory liking, dietary intake and other characteristics interact with one another, it is useful to identify individual factors associated with liking for fat. Indeed, few previous studies( Reference Duffy, Hayes and Sullivan 8 , Reference Bartoshuk, Duffy and Hayes 12 – Reference Lahteenmaki and Tuorila 15 ) explored the relationship between liking for fat and specific individual characteristics.
Available studies have shown that a strong liking for fat was associated with weaker cognitive restraint in an obese population( Reference Elfhag and Erlanson-Albertsson 14 ) and highly uncontrolled eating among dieters( Reference Lahteenmaki and Tuorila 15 ). Moreover, positive associations have been shown between liking for fat and cardiovascular risk factors such as BMI and blood pressure( Reference Duffy, Hayes and Sullivan 8 , Reference Bartoshuk, Duffy and Hayes 12 , Reference Duffy, Lanier and Hutchins 13 ). No significant association has been found between liking for fat and sex or age( Reference Drewnowski 2 , Reference Drewnowski, Almiron-Roig, Montmayeur and Le Coutre 16 ).
Major limitations of previous studies included a small sample size and the absence of demographic heterogeneity in the study population (children, females and dieters), which restricts variability in terms of liking and individual factors. Moreover, associations between liking for fat and lifestyle, socio-economic characteristics and health status have not been previously studied, even though these factors are strongly associated with taste sensitivity or dietary intake( Reference Mejean, Macouillard and Castetbon 17 , Reference Stewart, Feinle-Bisset and Golding 18 ). Finally, no study has separately considered liking for fat-and-salt and fat-and-sweet sensations, and yet we assume that specific associations might exist, such as uncontrolled eating associated with fatty-salted food consumption, and emotional eating associated with fatty-sweetened food( Reference Keskitalo, Tuorila and Spector 19 ).
The aim of the present study was to investigate the associations between liking for fat-and-salt and fat-and-sweet sensations and sociodemographic, economic, psychological, lifestyle and health characteristics in a large sample of French adults.
Subjects and methods
Study population
We used data from the NutriNet-Santé Study, a large Web-based observational cohort launched in France in May 2009 with a scheduled follow-up period of 10 years. It was implemented in a general population and targeted Internet-using adult volunteers aged 18 years or older. The study was designed to investigate the relationship between nutrition and health (incidence of IHD, cancer and all-cause mortality), as well as the determinants of dietary behaviour and nutritional status. The design, methods and rationale have been described elsewhere( Reference Hercberg, Castetbon and Czernichow 20 ). Briefly, in order to be included in the cohort, participants had to fill out an initial set of questionnaires assessing dietary intake, physical activity, anthropometry, lifestyle, socio-economic conditions and health status. As part of their follow-up, the participants complete the same set of questionnaires every year. Moreover, each month, they were invited to fill out complementary questionnaires related to the determinants of food behaviour, nutritional and health status. The present study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures were approved by the Institutional Review Board of the French Institute for Health and Medical Research (IRB Inserm no. 0000388FWA00005831) and the ‘Commission Nationale Informatique et Libertés’ (CNIL no. 908450 and 909216). Electronic informed consent was obtained from all subjects (EudraCT no. 2013-000929-31).
Data collection
Assessment of liking for fat-and-salt and fat-and-sweet sensations
Liking for fat-and-salt and fat-and-sweet sensations was assessed using the PrefQuest, an original Web-based questionnaire which was internally validated using factor analyses( Reference Deglaire, Mejean and Castetbon 21 ). This questionnaire also permitted the evaluation of liking for salt and sweet sensations. In May 2010, 65 683 participants were invited to complete this questionnaire on the ‘NutriNet-Santé’ website (https://www.etude-nutrinet-sante.fr/) available online for 6 months. The development and validation of the questionnaire has been described elsewhere( Reference Deglaire, Mejean and Castetbon 21 ). Briefly, eighty-three relevant items were divided into liking for salt (eleven items), sweet (twenty-one items), fat-and-salt (thirty-one items) and fat-and-sweet (twenty items) sensations. The questionnaire included four types of items: (1) liking for sweets, fatty-sweet and fatty-salty foods, rated on a 9-point hedonic scale; (2) preferred level of salt, sweet, fat-and-salt or fat-and-sweet seasoning, measured on a 5- or 6-point scale; (3) preferred drinks (sweet/sweetened or unsweetened) on a restaurant menu; and (4) eating behaviour regarding sweet, salty and fatty food, measured on a 5- or 9-point scale. For most items, subjects also had the option of checking a non-applicable answer, such as ‘I have never tasted (this food)’ or ‘I do not like (this food)’.
Assessment of sociodemographic, economic, psychological, lifestyle, anthropometric and health data
Sociodemographic, economic, lifestyle, anthropometric and self-reported health data were collected at baseline with the initial set of questionnaires. Psychological data were collected 14 months after inclusion using the Three-Factor Eating Questionnaire, which is a self-assessment instrument of eating behaviour( Reference Stunkard and Messick 22 ) (its completion was optional). The revised version was selected (Three-Factor Eating Questionnaire-R21) and covers three eating behaviour domains: the cognitive restraint scale (six items); the emotional eating scale (six items); the uncontrolled eating scale (nine items)( Reference Tholin, Rasmussen and Tynelius 23 ).
Statistical analysis
The present analysis focused on subjects living in metropolitan France and included in the NutriNet-Santé cohort who had completed the initial set of questionnaires, the PrefQuest and the Three-Factor Eating Questionnaire. All analyses were performed separately for men and women because most interactions between sex and individual characteristics were significant (P< 0·05).
The calculation of liking scores for the fat-and-salt and fat-and-sweet sensations has been detailed previously( Reference Deglaire, Mejean and Castetbon 21 ). Factor analyses demonstrated that the fat-and-salt and fat-and-sweet sensations were multidimensional and presented a hierarchical factorial structure. The fat-and-salt sensation was composed of two compounding factors, ‘added fat-and-salt’ and ‘fatty-salty foods’ (based on ‘meats’, ‘cheese products’ and ‘savoury snacks’); the fat-and-sweet sensation was composed of two compounding factors, ‘added fat-and-sweet’ and ‘fatty-sweet foods’ (based on ‘pastries and desserts’, ‘chocolate spread’ and ‘chocolate desserts’). Since we used different hedonic scales in the questionnaire (from 5 to 9 points), all data were linearly transformed into values ranging from 0 to 10 so as to standardise the ratings. First, for each participant, liking scores of each factor composing a sensation were calculated by summing the ratings of the items belonging to the factor and dividing by the number of items of this factor. Next, overall liking scores of a sensation were computed by averaging the liking scores of the compounding factors. If a responder had more than 25 % of non-applicable answers, i.e. ‘I have never tasted (this food)’ or ‘I do not like (this food)’ for a given sensation, the corresponding liking score was not computed and was excluded from the analysis. If a responder had 25 % of non-applicable answers or less, an imputation was performed for these missing values. Non-applicable answers were estimated within each sensation by adding the mean of all ratings of the subject to the mean of all ratings of the given item in the sample, and then subtracting by the mean of all ratings of the sensation in the sample( Reference Deglaire, Mejean and Castetbon 21 ). Quartiles of the score for fat-and-sweet and fat-and-salt sensations were computed and defined from quartile 1 ‘low liking’ to quartile 4 ‘high liking’.
We categorised sociodemographic and economic variables as follows: sex (men, women); age by quartiles (male: < 40, 40–54, 54–63, >63 years; female: < 32, 32–44, 44–56, >56 years); matrimonial status (single, couple, married, divorced/widowed); having at least one child (yes, no); living area (rural, urban < 20 000 inhabitants, urban ≥ 20 000 inhabitants, urban ≥ 100 000 inhabitants, Paris); educational level (elementary school, secondary school, college graduate, advanced degree, other); occupational category (never worked, farmers/self-employed, manual workers/employees, intermediate profession, managerial staff); income per consumption unit ( < 900, 900–1800, 1800–2700, >2700€, refused to answer). Regarding cognitive restraint and uncontrolled eating, according to tertiles, categories were created: low; medium; high. As the emotional eating variable was not normally distributed, we categorised it into three classes using the median: no emotional eating; low emotional eating; high emotional eating.
According to French recommendations( Reference Hercberg, Chat-Yung and Chauliac 24 ), male drinkers were categorised as abstainers and irregular consumers ( ≤ 4 g alcohol/d), moderate (>4 and ≤ 30 g alcohol/d) or heavy drinkers (>30 g alcohol/d), and females as abstainers and irregular consumers ( ≤ 3 g alcohol/d), moderate (>3 and ≤ 20 g alcohol/d) or heavy drinkers (>20 g alcohol/d). Smoking status was classified into three categories: never smoker; former smoker; current smoker. We also assessed current dieting (yes, no) and past dieting (yes, no). Normal weight, overweight and obesity were defined according to the WHO classification for BMI, as BMI < 25 kg/m2 (normal weight), 25 ≤ BMI < 30 kg/m2 (overweight) and BMI ≥ 30 kg/m2 (obese)( 25 ), and self-reported health problems listed were type 1 and type 2 diabetes, hypertension, hypertriacylglycerolaemia and hypercholesterolaemia. Menopause (yes, no, no answer) and pregnancy status (yes, no) were also recorded.
Comparisons between included and excluded participants and sex comparisons were performed using Student's t test and the χ2 test. First, in order to use ordinal logistic regression, we tested the hypothesis of proportionality of OR, but the latter was not valid. Therefore, associations of liking for fat-and-sweet and fat-and-salt sensations with all individual characteristics were assessed by multinomial logistic regression analysis (common reference: quartile 1 ‘low liking’) stratified by sex. Univariate logistic regressions were performed by calculating OR and 95 % CI to determine the strength of the associations between liking and each explanatory variable. According to the literature and our hypotheses, interaction effects between age and cognitive restraint( Reference Keskitalo, Tuorila and Spector 19 , Reference De Lauzon, Romon and Deschamps 26 ), age and smoking status and age and alcohol consumption( Reference Eigenbrodt, Mosley and Hutchinson 27 ) were also examined. Only explanatory variables and interaction terms associated with liking at the 0·1 significance level were retained for inclusion in the initial multivariate model. Subsequently, using stepwise backward elimination, multivariate logistic regression models were constructed. Variables were removed from the model one by one using a P value >0·05 for exclusion. Variables whose exclusion from the model caused large fluctuations in OR (>10 %), as well as variables whose exclusion gave rise to significant likelihood ratio tests (P< 0·05) were re-entered into the model. Data management and statistical analyses were performed using SAS software (version 9.3; SAS Institute, Inc.). A P value < 0·05 was considered statistically significant.
Results
Among the 65 683 participants in the NutriNet-Santé Study in May 2010, 49 066 had responded to the PrefQuest, i.e. a 75 % participation rate. Among the responders, 24 % were excluded (n 11 885) because they were non-responders to the Three-Factor Eating Questionnaire (n 10 481), had non-applicable data for a sensation (n 822) or had missing data for BMI, living area or alcohol consumption (n 582), leaving 37 181 subjects available for analysis (28 504 women and 8677 men). Compared with excluded subjects, included subjects were older (46 v. 41 years, P< 0·0001), the proportion of men was higher (23 v. 21 %, P< 0·0001), the proportion of subjects with high incomes was higher (26 v. 21 %, P< 0·0001) and the proportion of smokers was lower (15 v. 20 %, P< 0·0001) (data not shown).
For women and men, fat-and-salt-liking scores ranged from 0 to 9·48 with a mean of 3·9 (sd 1·4) and from 0 to 9·82 with a mean of 4·2 (sd 1·4), respectively, while fat-and-sweet-liking scores ranged from 0 to 9·92 with a mean of 3·8 (sd 1·8) and from 0 to 10 with a mean of 3·7 (sd 1·8), respectively. All results mentioned below were statistically significant. Men had a significantly higher liking score for the fat-and-salt sensation than women, while the opposite was observed for the fat-and-sweet sensation (Table 1). Regarding individual characteristics, women were younger and had higher psychological scores, while men had higher BMI (Table 2). The proportions of married men, those who had at least one child and those who belonged to a high-level occupational category were higher than in women (Table 2).
Quartiles of liking scores for the fat-and-salt and fat-and-sweet sensations (Mean values and standard deviations)
Individual characteristics of the sample (n 37 181, NutriNet-Santé cohort, 2009–2010) (Mean values, standard deviations or percentages)
Individual characteristics associated with liking for the fat-and-salt sensation
For greater clarity of results, only multivariate results from the ‘high liking’ level (fourth quartile) were reported here, and we defined this as the ‘high risk’ group. In men, age was negatively associated with a strong liking for the fat-and-salt sensation (Table 3). Women who had at least one child were more likely to have a strong liking for the fat-and-salt sensation (Table 4). Men and women with low income and women belonging to a low-level occupational category were more likely to have a high liking for the fat-and-salt sensation compared with those with higher income or a managerial position (Tables 3 and 4). In addition, current smokers and heavy drinkers were more likely to have a stronger liking for this sensation than never smokers, abstainers or irregular alcohol consumers.
Multivariate multinomial logistic analysis of the associations between sociodemographic, economic, psychological, lifestyle, anthropometric and health characteristics and levels of liking for the fat-and-salt sensation in men (n 8677, NutriNet-Santé cohort, 2009–2010) (Odds ratios and 95 % confidence intervals)
* Multinomial logistic regression analysis was performed using the common reference (quartile 1 ‘low liking’).
Multivariate multinomial logistic analysis of the associations between sociodemographic, economic, psychological, lifestyle, anthropometric and health characteristics and levels of liking for the fat-and-salt sensation in women (n 28 504, NutriNet-Santé cohort, 2009–2010) (Odds ratios and 95 % confidence intervals)
* Multinomial logistic regression analysis was performed using the common reference (quartile 1 ‘low liking’).
Regarding psychological characteristics, men and women with highly uncontrolled eating were more likely to strongly prefer the fat-and-salt sensation than less uncontrolled eaters (Tables 3 and 4). Moreover, men with high cognitive restraint were less likely to have a strong liking for the fat-and-salt sensation than low cognitive restrainers. In women, a significant interaction between age and cognitive restraint was found. In stratified models by age, women with high cognitive restraint were less likely to have a strong liking than those with low cognitive restraint, and this relationship was stronger in young women (data not shown).
Compared with non-dieters, women who were currently dieting were more likely to have a strong liking for the fat-and-salt sensation, while women who had already dieted at least once (former dieters) were less likely to have a strong liking. In both sexes, obese and overweight individuals were more likely to strongly prefer the fat-and-salt sensation than those with normal corpulence. Pregnant and postmenopausal women were also more likely to have a strong liking for the fat-and-salt sensation (Table 4).
Individual characteristics associated with liking for the fat-and-sweet sensation
In both sexes, age was negatively associated with a strong liking for the fat-and-sweet sensation (Tables 5 and 6). Compared with women without children, those who had at least one child were more likely to have a strong liking for the fat-and-sweet sensation. In addition, women with low income and men and women belonging to a low-level occupational category were more likely to strongly prefer the fat-and-sweet sensation than individuals with high income or a managerial position.
Multivariate multinomial logistic analysis of the associations between sociodemographic, economic, psychological, lifestyle, anthropometric and health characteristics and levels of liking for the fat-and-sweet sensation in men (n 8677, NutriNet-Santé cohort, 2009–2010) (Odds ratios and 95 % confidence intervals)
* Multinomial logistic regression analysis was performed using the common reference (quartile 1 ‘low liking’).
Multivariate multinomial logistic analysis of the associations between sociodemographic, economic, psychological, lifestyle, anthropometric and health characteristics and levels of liking for the fat-and-sweet sensation in women (n 28 504, NutriNet-Santé cohort, 2009–2010) (Odds ratios and 95 % confidence intervals)
* Multinomial logistic regression analysis was performed using the common reference (quartile 1 ‘low liking’).
Men and women with high cognitive restraint were less likely to have a strong liking for the fat-and-sweet sensation compared with low cognitive restrainers, while subjects with highly uncontrolled eating or high emotional eating were more likely to have a strong liking. In addition, women who were currently dieting were more likely to have a strong liking compared with non-dieters. In both sexes, overweight and obese participants were more likely to have a strong liking for the fat-and-sweet sensation compared with those of normal weight.
Postmenopausal women were also more likely to have a strong liking for this sensation compared with non-menopausal women (Table 6). Men who declared that they had hypertriacylglycerolaemia or hypertension were less likely to have a strong liking for the fat-and-sweet sensation than those without the diseases (Table 5).
Discussion
The present findings help to identify specific individual profiles associated with a strong liking for fat. To our knowledge, no study has investigated simultaneously the relationships between liking for fat-and-salt and fat-and-sweet sensations and various associated factors. Biological and physiological factors such as BMR, fat oxidation and insulin sensitivity, and also behavioural characteristics such as size of eating episode, frequency of eating and level of hunger have allowed to define the high-fat phenotype, according to Blundell & Cooling( Reference Blundell and Cooling 28 ). The originality of the present study is the distinction between fat-and-salt and fat-and-sweet sensations. Our findings thus allowed highlighting not only common factors associated with both sensations that define the overall fat sensation, but also specific physiological and behavioural characteristics associated with the fat-and-salt or fat-and-sweet sensation only.
For both sensations, equivalent relationships were found between liking scores and age, socio-economic status, cognitive restraint, uncontrolled eating, weight status and menopause. Specific associations were also highlighted according to the sensation: smoking and alcohol consumption were associated with high liking for the fat-and-salt sensation, while emotional eating, hypertension and hypertriacylglycerolaemia (only in men) were associated with strong liking for the fat-and-sweet sensation. Moreover, only in women, having a child and current dieting were both positively associated with the two sensations.
Liking for fat-and-salt and fat-and-sweet sensations decreased with age, as observed for intake of fatty foods in a previous study( Reference Mejean, Macouillard and Castetbon 17 ). This may be linked to the predictive role of fat liking upon fat intake, as suggested previously( Reference Raynor, Polley and Wing 10 ). One hypothesis is that liking and consumption of fatty foods may decline with age due to physiological changes such as altered taste and smell, slower gastric emptying, altered hormonal responses and decreased BMR( Reference Drewnowski and Shultz 29 ). Previous studies( Reference Drewnowski 2 , Reference Drewnowski, Almiron-Roig, Montmayeur and Le Coutre 16 ) assessing fat liking did not report this finding concerning age, probably due to the highly selected samples compared with our population. The present study also highlighted the fact that pregnancy was associated with a higher fat-and-salt-liking score. It has been previously demonstrated that during pregnancy, there is an increase in bitter sensitivity during the first trimester so as to protect against ingesting poison, and a decrease in the perception level for salt and bitter tastes in the second and third trimesters to support ingesting a varied diet( Reference Duffy, Bartoshuk and Striegel-Moore 30 ). This status could also affect fat sensitivity. Moreover, the positive associations between menopause and high liking for fat-and-salt and fat-and-sweet sensations may be due to hormonal changes. This relationship has never been studied in the literature, and few studies carried out on the relationship between the menstrual cycle and preferences or dietary intake have shown no significant associations( Reference Hormes and Rozin 31 – Reference Verma, Mahajan and Mittal 33 ).
A positive association between having at least one child and high fat liking was found in women, probably due to maternal exposure to fatty foods intended for children. Low socio-economic status was inversely associated with liking for fat-and-salt and fat-and-sweet sensations. Higher exposure to fatty foods due to the greater affordability of these products compared with healthy foods in low socio-economic populations( Reference Drewnowski and Specter 34 , Reference Drewnowski and Darmon 35 ) could influence the liking for fat.
Uncontrolled eating was positively associated with liking for both sensations, but the strength of the relationship was higher for the fat-and-salt sensation, while restriction of eating was inversely associated with liking. Our findings in a general population are in concordance with previous studies assessing these relationships in obese populations and among dieters( Reference Elfhag and Erlanson-Albertsson 14 , Reference Lahteenmaki and Tuorila 15 ). Emotional eating was positively associated with a liking for the fat-and-sweet sensation, as observed in a previous work conducted in a general population( Reference Keskitalo, Tuorila and Spector 19 ). Psychological characteristics influenced the dietary intake of energy-dense foods such as butter, cheese, pastries and sweets( Reference Lahteenmaki and Tuorila 15 , Reference De Lauzon, Romon and Deschamps 26 , Reference de Castro and Lilenfeld 36 ), and may also influence fat liking due to modified exposure to these foods. Moreover, it showed the specificity of the associations between sensory liking and individual factors according to the sensation, i.e. emotional eating and fat-and-sweet sensation. This suggests that there are distinct phenotypes whether salt or sugar is associated with fat. We could therefore identify the high fat-and-salt and high fat-and-sweet phenotypes.
Women who were currently dieting were more likely to prefer fat; thus, a recent low-fat diet may have led to frustration due to high cognitive restraint( Reference King, Herman and Polivy 37 ) and an increased liking for fat. Elsewhere, smoking and alcohol consumption were positively associated with a strong liking for the fat-and-salt sensation. Indeed, a high dose of alcohol alters the perception of salty taste and promotes the consumption of fatty-salted foods( Reference Caton, Ball and Ahern 38 ). Smoking decreases the olfactory function( Reference Frye, Schwartz and Doty 39 ) and also affects taste sensitivity( Reference Pavlos, Vasilios and Antonia 40 , Reference Pepino and Mennella 41 ). This may affect food preferences by increasing exposure to fatty-salted foods, as shown in a previous study( Reference Perrin, Krut and Bronte-Srewart 42 ).
Our findings showing positive relationships between weight status and fat liking, particularly the fat-and-salt sensation, are in agreement with previous studies( Reference Duffy, Hayes and Sullivan 8 , Reference Bartoshuk, Duffy and Hayes 12 , Reference Duffy, Lanier and Hutchins 13 ), and suggest that fat liking might be involved in the obesity epidemic through overconsumption of fatty foods.
Interpretation of the present results must take into account the characteristics of the study. Subjects were volunteers in the NutriNet-Santé Study and thus probably more concerned about healthy lifestyle and nutrition than the general population. Thus, caution is needed when interpreting and generalising the results. Moreover, compared with liking as assessed by sensory analysis, self-reported liking on a questionnaire may lead to under-reporting. However, positive correlations between the present questionnaire and sensory test measurements have been previously shown in our laboratory. Coefficients of 0·4–0·5 for salt and sweet tastes and 0·2–0·3 for the fat sensation were found (A Deglaire, C Urbano, C Mejean, et al., unpublished results). Such coefficients are in line with previous results( Reference Beauchamp, Bertino and Burke 43 – Reference Kim and Lee 45 ). The present questionnaire, shown to be repeatable, feasible and valid( Reference Deglaire, Mejean and Castetbon 21 ), can thus serve as a proxy for sensory test measurements of liking.
In addition, the large sample size induced significant associations even when differences between groups were small. However, the sample size and the diversity of collected data, such as sociodemographic, economic, psychological, lifestyle, anthropometric and health factors, enabled a highly accurate estimate and adjustment for several confounders.
In conclusion, the present study provides original information and elucidates individual characteristics associated with high liking for fat-and-salt and fat-and-sweet sensations. Indeed, relationships between common unhealthy characteristics such as low socio-economic conditions, dieting, overweight and high fat liking were highlighted, and a strong liking for the fat-and-salt sensation was specifically associated with smoking and alcohol consumption. Likewise, pregnancy, having a child, menopause, uncontrolled eating and emotional eating were positively associated with high fat liking, while increasing age and cognitive restraint were negatively associated. These individual characteristics are likely to interact with each other in the causal chain between liking and dietary intake. Our original findings should therefore enable a more precise elucidation of the causal framework of relationships between taste sensitivity, liking, dietary intake and other characteristics.
Acknowledgements
The authors thank the scientists, dietitians, technicians and assistants who helped carry out the NutriNet-Santé Study, and all dedicated and conscientious volunteers. They especially thank Gwenaël Monot, Mohand Aït-Oufella, Paul Flanzy, Yasmina Chelghoum, Véronique Gourlet, Nathalie Arnault and Laurent Bourhis. The authors thank Voluntis (a healthcare software company) and MXS (a software company specialising in dietary assessment tools) for developing the NutriNet-Santé Web-based interface according to our guidelines. The authors are grateful to Jerri Bram for English editing of the manuscript.
The present study was supported by the French National Research Agency (Agence Nationale de la Recherche) in the context of the 2008 Programme de Recherche ‘Alimentation et Industries Alimentaires’ (Epipref project, ANR-08-ALIA-006). The NutriNet-Santé Study was supported by the following institutions: Ministere de la Sante (DGS); Institut de Veille Sanitaire (InVS); Institut National de la Prevention et de l'Education pour la Sante (INPES); Fondation pour la Recherche Medicale (FRM); Institut de Recherche en Santé Publique (IRESP); Institut National de la Sante et de la Recherche Medicale (INSERM); Institut National de la Recherche Agronomique (INRA); Conservatoire National des Arts et Metiers (CNAM); Universite Paris 13. The study was supported by grants from the Region Ile-de-France.
The authors' contributions are as follows: A. L. designed the study, performed the statistical analysis, interpreted the data and drafted the manuscript; A. D., P. S. and K. C. designed the study and participated in the drafting of the manuscript; S. P. contributed to the study design and writing of the manuscript; S. H. designed and coordinated the study cohort and supervised the study; C. M. was involved in the conception and design of the study and interpretation of the data, supervised the statistical analysis and helped in the drafting of the manuscript. All authors critically reviewed the paper and approved the final version.
None of the authors has any conflict of interest to declare.
