Study population
The NutriNet-Santé study (http://info.etude-nutrinet-sante.fr) is an ongoing web-based prospective observational cohort study launched in France in May 2009 with a scheduled follow-up of 10 years. It aims to investigate the relationship between nutrition and chronic disease risk, as well as the determinants of dietary behavior and nutritional status. The study was implemented in the French general population (internet-using adult volunteers, aged ≥18 years). The rationale, design and methodology of the study have been fully described elsewhere [38]. In brief, to be included into the study, participants have to complete a baseline set of self-administered web-based questionnaires assessing dietary intake, physical activity, anthropometric characteristics, lifestyle, socioeconomic conditions and health status. As part of the follow-up, participants are asked to complete the same set of questionnaires each year. Moreover, each month, participants are invited by e-mail to fill in optional questionnaires related to dietary intake, determinants of eating behaviors, nutritional and health status. This study is conducted in accordance with 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 n°0000388FWA00005831) and the Commission Nationale de l’Informatique et des Libertés (CNIL n°908450 and n°909216). All participants provided informed consent with an electronic signature. This study is registered in EudraCT (n°2013-000929-31).
Data collectionMeal planning questionnaire
Meal planning practices were assessed via an optional questionnaire launched in the NutriNet-Santé cohort study in April 2014.
First, grocery shopping and cooking practices were evaluated. In particular, participants were asked to indicate whether they were involved in grocery shopping (every day, several times a week, once a week, less than once a week) and cooking (every day twice a day, every day once a day, several times a week but not every day, once a week, less than once a week, never) in their household. Then, participants were asked the following question “Generally, when do you choose the foods you are going to eat for meal?” (just before meal, during the day, the day before, few days before, one week before, never). Participants responding “never” were exempted to complete the rest of the questionnaire.
Participants were also asked whether having to think about what they have to cook is a constraint for them. The responses were rated on a 5-point Likert scale ranging from one (strongly disagree) to five (strongly agree).
Participants were then asked whether they planned meals, even in an irregular manner (yes I do, yes I did but not anymore, no I never planned meals). The definition of “planning meals” given to the participants was “to plan ahead the foods that will be eaten for the next few days”. Participants who reported planning meal currently were considered as “meal planners” whereas others were categorized as “non-meal planners”.
Finally, the questionnaire included questions about meal planning frequency (several times a week, once a week, once every two weeks, two to three times a month, not regularly), duration (a few days, one week, two weeks or more), period of the week (weekdays, weekend, weekdays and weekend) and sources of inspiration (personal recipe repertoire, Internet or apps, ingredients available during grocery shopping).
Socio-demographic and economic characteristics
At baseline and annually thereafter, participants in the NutriNet-Santé study are asked to provide socio-demographic data, including sex, age (18–30, 30–50, 50–65, >65 years), educational level (up to secondary, some college or university degree), monthly income (<1,200 €, 1,200–1,800 €, 1,800–2,700 € and >2,700 € per consumption unit), presence of children in the household (yes, no), history of dieting to lose weight during the past year (yes, no) and physical activity (low, moderate, high). Monthly household income is calculated per “consumption unit” (CU), where one CU is attributed for the first adult in the household, 0.5 CU for other persons aged 14 or older, and 0.3 CU for children under 14, following national statistics methodology and guidelines [39].
Physical activity was assessed using a short form of the French version of the International Physical Activity Questionnaire (IPAQ). The weekly energy expenditure expressed in metabolic equivalent task minutes per week was estimated, and three scores of physical activity were constituted [i.e., low (<30 min/day), moderate (30–59 min/day), and high (≥60 min/day)] according to the French guidelines for physical activity [40].
For the present study, we used the closest available data with respect to the assessment of meal planning practices.
Dietary measurements
At inclusion and once a year thereafter, participants are invited to complete three non-consecutive 24-h dietary records, randomly assigned over a 2-week period (two weekdays and one weekend day). For the present analysis, we selected participants who completed at least three 24-h dietary records since their inclusion in the cohort study (i.e. completed between May 2009 and December 2014). Participants reported all foods and beverages consumed at each eating occasion. They estimated the amounts eaten using validated photographs of portion sizes [41], using household measures or by indicating the exact quantity (grams) or volume (milliliters). Daily mean food intakes were calculated, weighted for the type of day of the week. Energy, nutrient and food group intakes were estimated using the NutriNet-Santé composition table including more than 2000 foods [42]. Dietary underreporting was identified on the basis of the method proposed by Black [43]. We hypothesize that meal planning encourages food preparation and therefore considered food groups that can be used in food preparation (e.g. eggs). In addition, we considered food groups that have nutritional interest (e.g. fruits). Thus, the following food groups were included in the study: fruits, vegetables, fish (including seafood and processed seafood), meat (including cooked ham, offal), eggs, milk, cheese, added fats (including oil, butter, margarine, vinaigrette), sugary products (e.g. cake, biscuits, sugars, honey, jam, chocolate) and starchy foods (including potato, legumes, pasta, rice, other cereals) with a specific focus on legumes and whole grain starchy foods (including whole grain pasta, rice, other cereals).
Adherence to nutritional guidelines was assessed using the PNNS Guideline Score (PNNS-GS). The 15-point PNNS-GS is a validated a priori score reflecting the adherence to the official French nutritional guidelines which has been extensively described elsewhere [44]. Details on computation of this score are in Additional file 1. Briefly, it includes 13 components: eight refer to food-serving recommendations (fruit and vegetables; starchy foods; whole grain products; dairy products; meat, eggs and fish; fish and seafood; vegetable fat; water vs. soda), four refer to moderation in consumption (added fat; salt; sweets; alcohol) and one component pertains to physical activity [44, 45]. Points are deducted for overconsumption of salt (>12 g/day), added sugars (>17.5% of energy intake), or when energy intake exceeds the needed energy level by more than 5%. Each component cut-off was that of the threshold defined by the PNNS public health objectives when available [45] otherwise they were established according to the French Recommended Dietary Allowances [46]. For the present analysis, we consider the mPNNS-GS, a modified version of the PNNS-GS, which takes into account only the dietary components, therefore excluding the physical activity component. Thus, the maximum score was 13.5.
Food variety score
Food variety has been defined as the number of different food items reported to be eaten over a given reference period [47]. Considering that seasonality is likely to influence food variety and that a period of 10 to 15 days has been recommended to accurately assess food variety [48], the food variety was evaluated using a Food Frequency Questionnaire (FFQ).
Sixteen months after baseline, participants were invited to complete a self-administrated 240-items FFQ to assess their usual dietary intake over the past year [49]. Participants were asked to report their consumption frequency on the basis of how many times they ate the standard portion size proposed (typical household measurements such as spoon or standard unit such as a yogurt). The frequency of consumption referred to usual consumption over the past year on an increasing scale including yearly, monthly, weekly or daily units, as suitable, and participants were asked to provide only one answer.
The food variety score corresponded to the number of FFQ items reported to be consumed at least once during the last year [47]. The maximum score was therefore 240. Fruit and vegetable variety scores were also computed based on the number of different fruits and vegetables reported by the participants.
Anthropometric data
Height and weight were assessed by using an anthropometric questionnaire, which was self-administered online, at baseline and each year thereafter [50, 51]. For each participant, the closest available data to the meal planning questionnaire were used for the analysis.
Data were not collected for pregnant women. BMI (in kg/m2) was calculated as the ratio of weight to squared height. Participants were classified as underweight or normal weight (BMI < 25), overweight (25 ≤ BMI < 30) and obesity (BMI ≥ 30) according to WHO references values [52].
Statistical analysis
The analysis focused on participants who had completed the meal planning questionnaire, had declared being involved in meal preparation in their household, and who had completed at least three 24-h dietary records since they were included in the study, as well as the FFQ.
Chi-square tests and Student’s t tests were used to compare characteristics of included vs. excluded participants, as well as meal planners vs. non-meal planners. Meal planners’ practices were also described. Continuous variables are presented as means ± SDs and categorical variables as percentages.
ANCOVAs were performed to investigate the relationship between meal planning and energy, macronutrients and food groups. However, for some particular food groups which did not exhibit normal distribution (i.e. eggs, milk, legumes, and whole grain starchy foods), mainly due to a high proportion of non-consumers, a binary variable (consumer/non-consumer) was created and a logistic regression analysis was performed. Logistic regression models were also used to assess the associations between meal planning and quartiles of mPNNS-GS, as well as quartiles of food variety scores (overall, fruit and vegetable) and BMI categories. Due to significant interactions and differences on the associations with meal planning, analyses on BMI were performed separately by sex.
Meal planning has been described as a cooking skill [53]. Thus, characteristics that have been shown to influence cooking practices, dietary intakes or weight status were considered as confounders in the present analyses. Models were therefore all adjusted for sex [1, 54, 55], age [56], educational level, monthly income [6], presence of children in the household [6], history of dieting to lose weight during the past year [57], physical activity [58], and cooking frequency. Models evaluating the associations with mPNNS-GS, macronutrient and food groups intakes were further adjusted for daily energy intake and number of 24-h dietary records completed by participants. The energy model was only adjusted on the number of 24-h records while the food variety models were adjusted on daily energy intake. Missing covariate data were imputed using multiple imputation method.
Sensitivity analyses were conducted on a subsample of individuals having responded to at least one of the dietary assessments (i.e. FFQ, dietary records). In addition, analyses were conducted using another definition of food variety score (number of FFQ items reported to be consumed more than once a week) [59].
All tests of statistical significance were two-sided and the type I error was set at 5%. Statistical analyses were performed using SAS software (version 9.3, SAS Institute Inc, Cary, NC, USA).
Dining and Cooking