When disconcerted by their illness and disappointed by traditional medicine, cancer patients can move towards complementary and alternative medicines, including the use of dietary supplements (DS). However, the impact of DS on cancer prognosis, risk of recurrence and risk of second cancer remains unclear( Reference Giovannucci and Chan 1 ). Although the beneficial effects of DS taken during or after cancer treatment are not excluded( Reference Block, Koch and Mead 2 – Reference Bougnoux, Hajjaji and Ferrasson 4 ), several studies have reported adverse effects of some DS on cancer prognoses( Reference Giovannucci and Chan 1 , Reference Latino-Martel and Bachmann 5 – Reference Greenlee, Kwan and Kushi 8 ). Presently, it is recommended that phyto-oestrogen DS should be avoided with hormone-dependent cancer patients( 6 , Reference Tsai, Lin and Simon Pickard 9 ). Similarly, β-carotene DS should not be used with cancer patients or with healthy people who are smokers( Reference Druesne-Pecollo, Latino-Martel and Norat 10 – Reference Meyer, Bairati and Fortin 12 ). In addition, studies have shown that some vitamin/mineral or herbal DS may interact with anti-cancer or other active treatments( Reference Tsai, Lin and Simon Pickard 9 , Reference Tsai, Lin and Lu 13 – Reference Sparreboom, Cox and Acharya 15 ). This may be particularly problematic if attending physicians have no knowledge of the self-medication practices in their patients. According to a recent review of US studies, 56–68 % of physicians are not aware of DS use among their cancer patients( Reference Velicer and Ulrich 16 ).
In this context, it is essential to assess DS use and its correlates in cancer patients and survivors. Such observational studies have been published in the United States( Reference Greenlee, Kwan and Kushi 8 , Reference Velicer and Ulrich 16 – Reference Greenlee, Kwan and Ergas 21 ). A recent review showed that DS use was widespread in American cancer patients, with a prevalence of vitamin and mineral DS use that varied between 64 and 81 %( Reference Velicer and Ulrich 16 ). Between 14 and 32 % of the survivors began taking their supplements after diagnosis( Reference Velicer and Ulrich 16 ). Fewer studies provided information on DS use in cancer patients in European countries( Reference Fox, Butler and Coughlan 22 – Reference Molassiotis, Fernandez-Ortega and Pud 28 ), but those that did showed a prevalence of DS use that varied from 9 to 48 % across countries.
Several sociodemographic and lifestyle characteristics, such as female sex, higher education level and higher income, have been reported to be associated with the use of DS. However, no epidemiological study has been published for France on this topic. Perceptions and attitudes towards food and nutrition may vary significantly across countries. Thus, specific research is needed, especially in the context of the increasing availability of DS from traditional distribution channels but also from Internet sources.
The objectives of the present study were (1) to assess DS use and its sociodemographic, lifestyle and dietary correlates in a large sample of French cancer survivors involved in the NutriNet-Santé cohort; (2) to evaluate the role and involvement of physicians in DS use by their cancer patients; (3) to assess the extent of potentially harmful practices, such as use of β-carotene DS in smokers, the use of phyto-oestrogen DS in hormone-dependent cancer patients and contraindicated DS–medication simultaneous use.
Methods
Subjects
The NutriNet-Santé study is a population-based prospective observational cohort that is followed via the Internet (http://www.etude-nutrinet-sante.fr). Its main objective is to assess the relationships between nutrition and chronic disease risk as well as the determinants of eating behaviour( Reference Hercberg, Castetbon and Czernichow 29 ). It was launched in France in May 2009. Using mass-media campaigns, adults aged 18 years or older who had access to the Internet were invited to participate. All subjects signed an electronic informed consent. The NutriNet-Santé study was approved by the Institutional Review Board of the French Institute for Health and Medical Research (no. 0000388FWA00005831) and the Commission Nationale de l'Informatique et des Libertés (no. 908450 and no. 909216).
Data collection
Participants completed several self-administered web-based questionnaires at baseline and then regularly during the follow-up. All baseline questionnaires were tested and compared against traditional assessment methods (paper questionnaires or interviews by a dietitian)( Reference Touvier, Kesse-Guyot and Mejean 30 – Reference Touvier, Mejean and Kesse-Guyot 32 ).
Case ascertainment
Participants self-declared any cancer diagnoses during the study follow-up through regular questionnaires and a permanent web-interface. Medical data were then gathered from participants, physicians, and/or hospitals and reviewed by an independent physician expert committee. Pathological reports were used to validate cancer cases and to extract cancer clinical data. Cases were classified using the International Chronic Diseases Classification, 10th Revision, Clinical Modification (ICD-10)( 33 ).
Dietary supplement and medication use
The questionnaire that assessed DS and medication use, which was specifically designed for cancer survivors, was sent in January 2014 to all first-incident primary cancer cases (who were diagnosed after their enrolment in the cohort). In the present study, we considered as DS both regular DS and medicinal supplements that are mainly composed of vitamins and minerals, which are treated as pharmaceutical products in France. In contrast, clinical oral nutritional supplements, such as Renutryl® or Clinutren®, were not considered as DS in the survey. Indeed, these specific products, which are rich in energy and/or protein, are administered to undernourished patients to stimulate weight gain and avoid malnutrition.
Participants were asked about their DS use at the time that they completed of the questionnaire (‘current use’). In addition, cancer patients were asked to report their DS use since the date of their cancer diagnosis. For each DS, they reported the commercial name, brand, form, number of days of use per year, and number of units per day of use. The questionnaire included an assessment of the circumstances and motivations for DS use (close-ended questions) as well as questions about DS use changes after cancer diagnosis. For each DS, participants were asked if they had told at least one of their attending physicians about their DS use and, when appropriate, the reasons why they did not. Subjects also reported all anti-cancer or other medications (chosen among an exhaustive list based on the VIDAL, a French drug reference database( 34 )) that they were taking at the time of the DS questionnaire. A DS composition database was created and implemented based on information found on official brands' websites or direct contact with manufacturers.
Sociodemographic, lifestyle and anthropometric data
Self-administered questionnaires were annually administered during follow-up to collect and update data on sociodemographic, lifestyle and behavioural characteristics, including sex, age, geographical region, marital status, number of children, educational level, smoking status, anthropometry and leisure-time physical activity (which was estimated with the validated international physical activity questionnaire( Reference Craig, Marshall and Sjöström 35 )). Data collected after cancer diagnosis and closest to the DS questionnaire administration were used in the present study.
Dietary data
Each year, participants were asked to complete three non-consecutive self-administered web-based 24 h dietary records, the days for which are randomly assigned during a 2-week period (2 d during the week and 1 d during the weekend). All foods and beverages consumed at breakfast, lunch, dinner and at all other occasions were recorded. For foods with potentially high nutrient variability, participants were also asked to provide the brand name. Participants were asked to estimate the portion size for each reported food and beverage item using a validated photograph booklet( Reference Hercberg, Deheeger and Preziosi 36 ). These photographs represented more than 250 foods (corresponding to about 1000 generic food items) in three different portion sizes. Along with the two intermediate and two extreme quantities, participants had seven choices of amounts. Daily dietary intakes of energy and nutrients were then calculated using the NutriNet-Santé food composition table( Reference Hercberg 37 ), which includes more than 2500 different foods. Dietary records from the year closest to the DS questionnaire administration (i.e. 2013 or 2014) were used in dietary data analyses.
Statistical analyses
Proportions of DS users (i.e. at least one DS taken after cancer diagnosis) and type of and motivation for DS use were calculated overall and also by sex and main cancer sites (breast and prostate) in a cross-sectional design. For each DS reported in the study, circumstances of DS purchase, duration of use and communication with a physician about this use were described.
DS users and non-users were compared by age and sex-adjusted unconditional logistic regression analyses regarding their sociodemographic, anthropometric and lifestyle characteristics (sex, age, geographical region, marital status, number of children, educational level, smoking status, BMI and leisure-time physical activity). OR and their 95 % CI were calculated.
Participants who provided at least two dietary records after cancer diagnosis (328 subjects excluded) and those who were not classified as under-reporters of energy intake according to the Black( Reference Black 38 ) criteria (eighty-five subjects excluded) were included in dietary data analyses (n 668). Most participants provided three 24 h dietary records, but 7 % provided two records. The mean daily intake of dietary macro- and micronutrients from food only and from food and supplements were compared by unconditional logistic regression between DS users and non-users after adjustment for age, sex, number of 24 h dietary records and energy intake. The aim of the present analysis was to quantify the extent to which total nutrient intake increased when total intake from food and supplements was considered in comparison to intake from food only and which nutrients were more specifically impacted. For each nutrient, the proportion of total daily intake brought by DS was also estimated both in overall DS users and in DS users of the specific nutrient. The number of subjects who practised potentially harmful DS use was assessed. Such practices included: (1) use of β-carotene DS by smokers( Reference Latino-Martel and Bachmann 5 , Reference Druesne-Pecollo, Latino-Martel and Norat 10 , Reference Touvier, Kesse and Clavel-Chapelon 11 ); (2) the use of DS that should be avoided in certain cancer patients, such as phyto-oestrogen use by hormone-dependent cancer patients( 6 , Reference Tsai, Lin and Simon Pickard 9 ); and (3) the simultaneous use of DS (current use at the time of the DS and medication questionnaire) and drugs for which harmful interactions of moderate-to-major severity have been described in the literature( Reference Tsai, Lin and Simon Pickard 9 , Reference Tsai, Lin and Lu 13 – Reference Sparreboom, Cox and Acharya 15 ).
P< 0·05 was considered statistically significant. All tests were two-sided. Analyses were carried out with SAS software release 9.3 (SAS Institute, Inc.).
Results
A total of 1490 subjects developed an incident primary cancer between the beginning of the study and January 2014. Among them, 1081 answered the DS questionnaire (participation rate = 73 %). The average age of cancer survivors was 60·2 (sd 10·9) years. Women constituted 68 % of the sample. The main cancer types were breast (43 %), followed by prostate (19 %) and melanoma skin cancer (15 %). The mean time between cancer diagnosis and administration of the DS questionnaire was 22·3 (sd 14·6) months.
Overall and specific dietary supplement use
DS use by sex and among breast and prostate cancer survivors is presented in Table 1. 62 % of women (n 455) and 29 % of men (n 101) reported using at least one DS after cancer diagnosis (n 556 in all): 442 were current users (at the time of the DS questionnaire), and 114 had taken DS since their cancer diagnosis but were not currently using DS. In current users, the number of DS taken simultaneously was 1, 2 or ≥ 3 in 52, 24 and 24 % of the subjects, respectively. The mean number of DS taken was 1·9 (sd 1·4). Vitamin D, vitamin B6, Mg and vitamin C were the most frequently used DS nutrients. After cancer diagnosis, 14 % of subjects started DS use, 9 % kept using DS but modified the types of DS used and 4 % stopped using DS (data not tabulated).
Dietary supplement (DS) use in cancer survivors (n 1081, NutriNet-Santé cohort study, France, 2014) (Number of subjects and percentages)
* At least one DS taken after cancer diagnosis.
† In decreasing order of frequency. Nutrients and other substances were consumed alone or in combination in a single DS. Percentages were calculated among overall DS users.
‡ K, Cu, Li, Mn, Cr and others.
Motivations for dietary supplement use, circumstances of purchase and involvement of attending physicians
The main stated reasons for DS use included: to reduce the adverse effects of cancer treatments (28·2 %), to decrease the risk of cancer recurrence (13·8 %), to strengthen the immune system (54 %), and to overcome tiredness (39 %) (Table 2, several answers possible). Similar motivations were reported by men and women and for breast and prostate cancer patients.
Reasons for dietary supplement (DS) use in cancer survivors (n 522, NutriNet-Santé study, 2014) (Number of subjects and percentages)
* At least one DS taken after cancer diagnosis. Data regarding reasons for DS use were missing for thirty-four DS users. Several answers were possible.
Among the 1140 DS reported by the participants, only about half of them (54 %) were taken following medical prescription or advice (Table 3). Medical prescription or advice was higher for vitamin D DS (81 %), medium for many vitamins and minerals (e.g. 43 % for vitamin C and Mg and 47 % for vitamin B6) and lower for herbal/natural DS (36 % for acerola/guarana and 24 % for phyto-oestrogen DS) (data not tabulated). Among the 1140 DS reported, 56 % were used for more than 1 year. For 35 % of reported DS, participants did not inform any of their attending physicians (general practitioner, oncologist or other) about their use. The main reason cited for this was that physicians did not ask their patients about their DS use (86 %).
Circumstances of purchase, duration of use, and communication with the attending physician for the 1140 dietary supplements (DS) reported by cancer survivors (NutriNet-Santé cohort study, France, 2014) (Number of subjects and percentages)
* Several answers were possible.
† General practitioner, oncologist or other attending physician.
‡ Among the 398 DS for which participants answered ‘no’ to the previous question. Several answers were possible.
Sociodemographic, lifestyle and anthropometric correlates of dietary supplement use
As compared to non-users (Table 4), DS users were more likely to be women (P< 0·0001) those with higher educational levels (P= 0·0001), never smokers (P= 0·03) and those with a normal weight (P= 0·0005).
Sociodemographic, lifestyle and behavioural correlates of dietary supplement (DS) use in cancer survivors (n 1081, NutriNet-Santé cohort study, France, 2014) (Number of subjects, percentages, odds ratios and 95 % confidence intervals)
* At least one DS taken after cancer diagnosis.
† Overseas departments/territories.
‡ As measures by the international physical activity questionnaire. Because of missing values, the proportions of subjects were calculated on 883 participants (459 DS users and 424 non-users).
Dietary correlates of dietary supplement use and its contribution to nutrient intake
When considering food intake only (Table 5), DS users already had significantly higher intakes of fibre, vitamins B6, B9 and C, Fe, Mg, K and Zn. When considering total intake (food+DS), DS users additionally had higher intakes of thiamin, riboflavin, niacin, pantothenic acid, vitamins D and E and Ca compared to non-users.
Comparison of daily dietary energy and nutrient intakes from food only and from food and dietary supplements (DS), between DS users (n 270) and non-users (n 398) among cancer survivors (NutriNet-Santé cohort study, France, 2014)* (Mean values, standard errors, number of subjects and percentages)
* In subjects with at least one 24 h dietary record after cancer diagnosis, normo-energy reporters. DS users were defined as subjects who used at least two DS after cancer diagnosis.
† Logistic regression analysis (performed to derive the P-value for the comparison between DS users and non-users) were adjusted for sex, age, number of 24 h dietary records and energy intake.
‡ Very small amounts of alcohol can be found in some DS or medicinal products as excipients (not as main active substance).
The contribution of DS to total nutrient intake was very high for some nutrients: in DS users of the specific nutrient, DS represented more than half of the total daily intakes for vitamins D (89 %), B6 (68 %), E (68 %) and B12 (50 %) and more than one-third of the total daily intake for fibre, thiamin, pantothenic acid, folate, retinol, vitamin C and Ca.
Potentially harmful dietary supplement use practices
Among the 442 DS current users (i.e. participants who were using DS at the time of the DS and medication questionnaire), eighty-one (18 %) reported DS practices that were contraindicated because of potential for adverse effects documented in the literature. Such practices are listed in Table 6 (one patient may cumulate several of these practices): thirty participants with smoking history (six current and twenty-four former smokers) used β-carotene DS, forty-four participants used DS that were contraindicated for specific cancer sites and thirty-nine participants simultaneously used DS and drugs for which a drug–nutrient interaction of moderate-to-major severity would be predicted.
Dietary supplement (DS) use that may convey adverse effects in cancer survivors (NutriNet-Santé cohort study, France, 2014) (Number of subjects)
Discussion
DS use was widespread in the present large sample of French cancer patients or survivors (29 % of men and 62 % of women). These proportions were lower than those observed in various American studies (50 to 81 % DS users)( Reference Velicer and Ulrich 16 , Reference Ferrucci, McCorkle and Smith 18 , Reference Miller, Vasey and Short 39 ) but were closer to figures observed in other European countries( Reference Molassiotis, Panteli and Patiraki 23 , Reference Westerlund, Steineck and Bälter 24 , Reference Molassiotis, Fernandez-Ortega and Pud 28 , Reference Velentzis, Keshtgar and Woodside 40 ). As suggested earlier( Reference Patterson, Neuhouser and Hedderson 41 , Reference Wiygul, Evans and Peterson 42 ), cancer diagnosis may have influenced DS use. Indeed, the present findings showed that 28 % of participants started, stopped or modified their DS use after cancer diagnosis. Further research into patients' beliefs regarding DS and cancer is needed to better understand these changes.
As observed in the NutriNet-Santé cohort for the general population( Reference Pouchieu, Andreeva and Péneau 43 ), vitamins B6 and C and Mg DS were among the most frequently consumed DS. In addition, in the present population of cancer survivors, vitamin D emerged as the most frequent supplementation (used by 45 % of DS users). This high prevalence of vitamin D DS use, especially in women, was consistent with prior investigations conducted in the United States( Reference Zirpoli, Brennan and Hong 20 , Reference Patterson, Neuhouser and Hedderson 41 , Reference Osian, Leal and Allmer 44 ). Indeed, several anti-cancer treatments, such as aromatase inhibitors, which are used to treat postmenopausal breast cancers and androgen-deprivation therapy, which is used to treat prostate cancer( Reference Limburg, Maxwell and Mautner 45 ), may induce bone loss and thus justify vitamin D supplementation.
Cancer itself and its consequences appeared as a strong motivation for DS use. Another frequently cited motivation was to ‘strengthen immune system’, which is in line with previous studies( Reference Velicer and Ulrich 16 , Reference Ferrucci, McCorkle and Smith 18 ) and with the fact that the immune system is often altered by chemotherapy and radiotherapy treatments. To ‘overcome tiredness’ was also among the main motivations for DS use in the NutriNet-Santé general cohort( Reference Pouchieu, Andreeva and Péneau 43 ), and this was in line with the frequent use of vitamins B6 and C and Mg DS, which are recognised by the European Food Safety Authorities for their anti-fatigue properties( 46 – 48 ).
About 45 % of DS use corresponded to self-medication, and 55 % were taken following medical prescription or advice. This proportion varied across DS types. Whereas some types of self-medicated DS should be avoided by cancer survivors, others may truly be useful if consumed under medical supervision. For instance, this is the case for vitamin D, as discussed earlier. We consistently observed that the proportion of medical prescription or advice was higher for this nutrient (81 %).
A substantial proportion (35 %) of DS use was not discussed with a physician. Although this proportion was lower than that found in US studies (56–68 %)( Reference Velicer and Ulrich 16 ), it remains a concern because of the potential interactions between DS and adjuvant therapy or other medications( Reference Tsai, Lin and Simon Pickard 9 , Reference Tsai, Lin and Lu 13 , Reference Cassileth and Lucarelli 14 ). The main reason why patients did not inform their physician was simply that the latter did not ask about DS use by their patient, which is in line with the literature( Reference Hall, Bissonette and Boyd 49 ). This suggests that physicians may not be fully aware of the medical importance of knowing whether their cancer patients are using DS. It has also been hypothesised that the perception of physician neutrality or indifference, or possibly even a fear of physician opposition, are reasons why patients may be hesitant to communicate with their physicians about DS use( Reference Velicer and Ulrich 16 ). Another possibility is limited physician time during medical consultations.
Previous studies of DS use in cancer survivors in other countries also found that female sex( Reference Ferrucci, McCorkle and Smith 18 , Reference Miller, Vasey and Short 39 , Reference Osian, Leal and Allmer 44 ), higher education( Reference Velicer and Ulrich 16 , Reference Ferrucci, McCorkle and Smith 18 , Reference Miller, Vasey and Short 39 , Reference Walshe, James and Macdonald-Wicks 50 ), non-smoking( Reference Bright-Gbebry, Makambi and Rohan 17 ) and lower BMI( Reference Bright-Gbebry, Makambi and Rohan 17 , Reference Miller, Vasey and Short 39 ) were associated with DS use. These factors have also been associated with DS use in the general French population( Reference Pouchieu, Andreeva and Péneau 43 , Reference Touvier, Kesse and Volatier 51 ). Although some studies have identified older age( Reference Miller, Vasey and Short 39 , Reference Osian, Leal and Allmer 44 ) as a correlate of DS use, it was not a significant predictor in the present study's population. Likewise, a review on vitamin and mineral supplement use among US cancer survivors( Reference Velicer and Ulrich 16 ) pointed out that DS use may differ by the type of cancer, but we did not find significant association in the present population between main cancer types after age and sex adjustment (data not shown).
As it has been observed in the general population( Reference Pouchieu, Andreeva and Péneau 43 ), DS use was associated with a healthier diet that already includes higher intakes of fibre and several vitamins and minerals from food only. The high contribution of DS to total nutrient intake highlights the importance of taking this source of intake into consideration both in clinical practice and in nutritional epidemiological studies. Few studies have provided detailed information about total micronutrient intake from food and supplements in cancer survivors( Reference Saquib, Rock and Natarajan 19 , Reference Inoue-Choi, Greenlee and Oppeneer 52 ). Compared to DS users in the American Women's Healthy Eating and Living Study( Reference Saquib, Rock and Natarajan 19 ), DS users in the present study had a lower total daily intake for almost all micronutrients, except for vitamin D, folate, Ca and Mg, which were similar in intake amount. These differences may notably be explained by the variability of DS formulations between the United States( 53 ) and Europe( Reference Lentjes, Bhaniani and Mulligan 54 ).
Several interactions between herbal or vitamin/mineral DS and drugs that may lead to moderate-to-major adverse effects have been well documented( Reference Tsai, Lin and Simon Pickard 9 , Reference Tsai, Lin and Lu 13 – Reference Sparreboom, Cox and Acharya 15 ). As recommended by official authorities, some DS should also be avoided by certain cancer patients, such as phyto-oestrogens in oestrogen receptor-positive cancers( 6 , Reference Tsai, Lin and Simon Pickard 9 , Reference Cassileth and Lucarelli 14 ). Another potentially harmful practice is the use of β-carotene DS, which is recognised for its association with increased lung and gastric cancers risk( Reference Latino-Martel and Bachmann 5 , Reference Druesne-Pecollo, Latino-Martel and Norat 10 , 55 ) in current smokers but also maybe in former smokers( Reference Touvier, Kesse and Clavel-Chapelon 11 ). A relatively high proportion of DS users in the present study (18 %) were concerned by such potentially harmful practices. Potentially problematic drug–nutrient interactions reported in the present study mainly concerned anticoagulants/antiplatelet agents and not specifically anti-cancer treatments, which may be explained by the fact that the DS questionnaire was administered 22 months after diagnosis on average.
The strengths of the present study are that it included a large number of cancer cases, detailed data about DS use (including precise doses and composition) and its dietary and lifestyle correlates, information about drugs taken simultaneously, and it was original in a French/European context.
The main limitation was that the present study's population may not be representative of all French male and female cancer survivors, given that the NutriNet-Santé study involved volunteers who accepted to participate in a cohort on nutrition and health. Compared to national estimates( 56 , 57 ), the NutriNet-Santé study included more women than men and more individuals who belong to higher socio-professional categories. In addition, some eligible cancer survivors did not answer the DS questionnaire. Compared to respondents, non-respondents were more likely to be women, to have more children and to have cancers other than those mainly represented in the cohort (P< 0·0001, data not shown). However, response rates to the DS questionnaire was high (73 %) compared to similar epidemiological studies( Reference Ferrucci, McCorkle and Smith 18 , Reference Greenlee, Kwan and Ergas 21 , Reference Wiygul, Evans and Peterson 42 , Reference Walshe, James and Macdonald-Wicks 50 , Reference Greenlee, Gammon and Abrahamson 58 ). In addition, because the DS questionnaire was administered 22 months after diagnosis on average, a recall bias may have occurred for DS that were not currently being taken by the participants. Finally, although the sample size was reasonably large overall, only the main cancer types (i.e. breast and prostate) could be analysed, and other types with lower incidence could not be investigated separately.
In conclusion, the present study provides new and detailed information about DS use and its correlates in a large sample of French cancer survivors. The findings suggest that DS use was widespread in the present population and substantially contributed to total nutrient intake, with 14 % of subjects initiating DS use after diagnosis. DS users had an overall healthier profile (in terms of diet, smoking status and BMI) than non-users. The present study highlighted the absence of communication about DS use between patients and healthcare professionals (for 35 % of reported DS) and the importance of self-medication (45 %). Worryingly, a substantial proportion of potentially harmful practices were detected (in 18 % of DS users), including the simultaneous use of vitamin E and anticoagulant/antiplatelet agents, the use of β-carotene by smokers and the use of phyto-oestrogens by hormone-dependent cancer patients. Physicians should be encouraged to more routinely discuss DS use with their cancer patients and to warn them about potential adverse effects. Finally, given the widespread use of DS in cancer patients and survivors, further etiological research is needed to better elucidate its impact on cancer prognosis, the risk of recurrence and the risk of second cancer.
Acknowledgements
The authors thank Younes Essedic, Paul Flanzy, Mohand Ait Oufella, Yasmina Chelghoum and Than Duong Van (computer scientists), Florence Charpentier (dietitian), Elise Martin (communication assistant), Nathalie Arnault, Véronique Gourlet, Fabien Szabo, Laurent Bourhis and Stephen Besseau (statisticians) and Rachida Mehroug (logistics assistant) for their technical contributions to the NutriNet-Santé study.
The present work was supported by the 2013 Research Prize from the French Nutrition Society and the French National Institute of Cancer (no. DEPREV14-027). Camille Pouchieu was funded by a grant from the French Ministry of Research and Higher Education (no. 11/ED G/UREN/2011). Philippine Fassier was funded by a grant from the Cancéropôle Ile-de-France (public funding from the Paris region). The NutriNet-Santé study was supported by the following institutions: Ministère de la Sante; Institut de Veille Sanitaire; Institut National de la Prévention et de l'Education pour la Santé; Fondation pour la Recherche Médicale; Institut de Recherche en Santé Publique; Institut National de la Santé et de la Recherche Médicale; Institut National de la Recherche Agronomique; Conservatoire National des Arts et Métiers; and Université Paris 13.
The author's responsibilities were as follows: C. P. and M. T. designed the research; S. H. and M. T. conducted the research; C. P. performed statistical analysis; C. P. and M. T. wrote the paper; P. F., N. D.-P., L. Z., P. B., M. T., I. B., S. H., P. G., P. C. and P. L.-M. contributed to the data interpretation and revised each draft for important intellectual content; and M. T. had primary responsibility for final content. All authors have read and approved the final manuscript.
None of the authors has any conflict of interest to declare.
