Get Complete Project Material File(s) Now! »

Data collection


Two TDS were conducted in France, the first one between 2000 and 2004 and the second one (TDS2), used in the present thesis, in 2006. The TDS2 (Sirot et al., 2009) provided the concentration level of 440 substances in 212 core foods (Arnich et al., 2012; Bemrah et al., 2012; Nougadere et al., 2012; Riviere et al., 2014; Sirot et al., 2013; Sirot et al., 2012a; Sirot et al., 2012b; Veyrand et al., 2013). These core foods, defined from the classification of the 1,280 food items included in the second French national food consumption survey (INCA 2) (Dubuisson et al., 2010; Lioret et al., 2010), covered about 90% of the whole diet of the French population. In order to be representative of French dietary habits, each sample was composed by 15 sub-samples of the same core food, covering different varieties, purchase locations, preparation methods and cooking methods. In total, 19,785 different food products were purchased, in eight French regions over different seasons from 2007 to 2009, to make up the 1,319 composite samples of core foods to be analysed for additives, environmental chemicals, pesticide residues, trace elements and minerals, mycotoxins, phytoestrogens and acrylamide.


Detection limits

Concentrations below the analytical limits of detection (LOD) or quantification (LOQ) could not be detected or quantified, respectively. These data are left-censored and a solution is to replace censored data by a fixed value (GEMS/Food-EURO, 2013) according to different scenarios. In order to focus on substances with a detected value, an LB (Lower Bound) scenario derived from the one recommended in GEMS/Food-EURO (2013) was used in this thesis. This consists in replacing non-detected values by 0 and detected values unable to be quantified, by the LOD value. Under the LB scenario, the exposure of the whole population was equal to zero for 51% of the food chemicals (223 chemicals: 212 pesticides, four perfluorinated compounds and seven mycotoxins).

Chemicals available

In this thesis, we have focused on chemicals or neoformed components and have not included ten minerals (Ca, Cu, Fe, K, Mg, Mn, Mo, Na, Se, Zn). Then, 207 out of 440 initial substances were included in the present study (Appendix 1) EDEN mother-child cohort

Study presentation

The overall objective of the EDEN study (study on the pre- and early postnatal determinants of child health and development) was to examine the relations and potential interactions between maternal exposures and health status during pregnancy, foetal development, health status of the new-born and the child’s health and development (Heude et al. 2015). The exclusion criteria were twin pregnancy, known medical history of diabetes, plan to move outside of the region in the three following years and French illiteracy. Among women who fulfilled these inclusion criteria, 55% agreed to participate (Deschamps et al. 2009). This mother-child cohort had recruited 2002 pregnant women from February 2003 to January 2006, before their 24th week of gestation in two French university hospitals in the cities of Poitiers and Nancy (Figure 3).
From the 2002 recruited women, 95 had left the study upon delivery due to personal reasons. Among the 1907 remaining women, birth weight was not available for 8 of the children and 117 did not have a clinical examination at birth, mainly due to transfer to another service or delivery in another maternity department. Figure 4 illustrates the attrition rate from birth to 5–6 years for the 1899 children with birth weight available: 66% of them were followed up to 5–6 years. Mothers of children lost to follow-up were younger at delivery (28.3 ± 5.2 vs 29.2 ± 4.8 years, P<10-3) and had less frequently attained a high school diploma (49% vs 74%, P<10-3) than included mothers; however, no difference was observed for maternal pre-pregnancy BMI, child’s birth weight or preterm birth rate (Heude et al., 2015).
Mothers had three clinical examinations, one between 24 and 28 weeks of amenorrhea, one at delivery and one 5–6 years after delivery (Figure 5). The child was examined clinically four times: at birth, at 1 year and 3 years and at 5–6 years. Concomitantly and in-between clinical visits, mothers and fathers answered questionnaires, administered by midwives or self-administered about their offspring, themselves and their household. Biological samples were collected from the mother during pregnancy and at birth, and from the child (or cord) at birth and at 5–6 years. Blood samples were also collected from fathers after the child’s birth (Figure 5).
The study has received the approval from the ethics committee of Kremlin-Bicêtre and from the National Commission for Data Protection and Liberties (CNIL).

Data collection

At the 24-28thweek examination, maternal height was measured by research midwives, and pre-pregnancy weight (g), education level (highest degree attained) was reported by the participants during the interview. Maternal smoking during first and second trimester was collected through interview at inclusion. Mothers reported their current smoking status (used to characterize the 2nd trimester status), as well as smoking habits at the beginning of pregnancy (used for the 1st trimester status), both including daily cigarette consumption. After delivery, similar information was collected by the research midwives concerning smoking at the end of the 3rd trimester of pregnancy (used for the 3rd trimester status). Mothers were weighed after the delivery using electronic Terraillon SL 351 scales (Hanson Ltd, UK) to the nearest 0.1 kg. Maternal parity (primiparous / multiparous) and the gestational age (weeks), weight (g), length (cm), head circumference (cm) at birth were collected from medical records.
The three variables on smoking status were used to create the average number of cigarettes smoked during all the pregnancy, used as a categorical variable (none, 1 to 9, ≥10 per day). Maternal pre-pregnancy body mass index (BMI) was calculated as weight (in kg) for height (in m) squared. The education level was classified as high school diploma attained or not. Maternal specific weight gain during pregnancy was calculated by subtracting reported weight before pregnancy from measured weight after delivery.

Characteristics of women and new-borns

The mean age of the women included was 29 years (range: 18–44) and 30% of the women were pregnant for the first time. The mean BMI was 23 kg/m² (standard deviation (SD): 4.5) and 53% of the women in the cohort attained high school diploma. The mean birth weight of their child was 3279 g (SD: 512). 52.6% of the newborns of the cohort were males and 94.4% were born at 37 weeks or more. Compared with the 2003 French National Perinatal Survey (Enquête National Perinatale, ENP) (Blondel et al., 2012), a national sample of births, women included in EDEN and still followed up at delivery had a higher level of education (Table 2). Rates of preterm births or admissions of the new-born to a neonatal or intensive care unit were, however, similar in both studies (Heude et al., 2015).
Table 2: Characteristics of mothers and their children followed up until delivery in the EDEN study and comparison with the 2003 National Perinatal survey (table from (Heude et al., 2015)).


Food frequency questionnaire

A self-administered food frequency questionnaire (FFQ) was available at two time points in the EDEN study. One was completed at the inclusion and is related to the diet in the year prior to pregnancy and the other one was completed few days following delivery at the maternity and is related to the diet during the last three months of pregnancy. This FFQ was slightly modified from a FFQ developed and validated in another French study (Deschamps et al. 2009). Consumption frequencies were recorded for 137 different food items, with a 7-item scale from “never” to “more than once a day”. We first generated a frequency from the midpoint of the category (i.e. two servings/month for the category of 1 – 3 servings/month). However, some changes were made after a validation step for the questionnaire by comparing it to repeated 24-hour recalls: a minimum assignment was performed for groups of 7 or more items (e.g. fruits, vegetables) or high energy density foods (e.g. “Boiled potatoes”, “Fried potatoes”). For these categories, if the frequency reported was between 1 to 3 times per month, we assigned 1 time per month rather than 2 times. Usual portion sizes were estimated using photos for different food types on a 3-level scale, derived from the Supplémentation en Vitamines et Minéraux Antioxydants (SU.VI.MAX) portion book (SU.VI.MAX., 2002) for the 12 food groups (e.g. meat, French fries, pasta, raw vegetables, cooked vegetables, cakes, cheese, beverages) corresponding to FFQ photographs. The portion size of the remaining food groups was estimated from the middle portion of the corresponding item in the SU.VI.MAX portion book. To calculate individual intake for each food item, the portion consumed (in g) was multiplied by the frequency declared (per day). The 137 food items were then grouped into 37 food groups. Women for which more than 3 items of the FFQ were missing were excluded and, when at the most 3 items were missing, missing values were imputed by the sample median. Individual total energy intakes were calculated for all subjects by multiplying the intake (in g per day) by the energy content of each food item, derived from the SU.VI.MAX nutrient composition database (SU.VI.MAX., 2006). Subjects with likely misreport of their food intake, i.e. with an estimated total energy intake under 1000 kcal/day (68 participants) or over 5000 kcal/day (30 participants), were excluded in some analysis (cf Chapter 4).

Study presentation

The purpose of the ELFE (Etude Longitudinale Française depuis l’Enfance) study was to build a nationally representative cohort of 20,000 children to be followed from birth to adulthood using a multidisciplinary approach in order to characterize the relationship between environmental exposures and the socio-economic context on health and behaviours more thoroughly (Vandentorren et al., 2009). It is a multidisciplinary, nationally representative, birth cohort, which included 18,329 children born in a random sample of 349 maternity units in France in 2011. From April 2011, all babies born in the included maternity wards on given recruitment days could be included with parental consent. Inclusion criteria were as follows: children born after 33 weeks of gestation, of mothers aged 18 years or older, not planning to move outside Metropolitan France in the following 3 years. Foreign families could also participate in the study, if mothers were able to read French, Arabic, Turkish or English. The recruitment phase took place during 4 waves of 4 to 8 days distributed along the year to take into account seasonal variability (Pirus et al., 2010). Among the 349 selected maternity units, 320 agreed to participate (Figure 6).
Participating mothers and children were recruited in maternity wards (51% participation rate). Infants born out of the days of inclusion (n = 71) and those whose parents withdrew consent within first year (n=55) were excluded from the analyses. Infants for whom it was not possible to verify the eligibility criteria due to missing data were also excluded (n = 350). Thus, 17,853 infants were actually eligible. Data were collected in standardized interviews conducted by trained interviewers and through self-completed questionnaires. Mothers were interviewed at the maternity ward after delivery and biological samples were taken during the delivery. Clinical and medical data related to pregnancy and delivery were also accessible from medical records. Two months after delivery, telephone interviews took place with the mothers and fathers. The family was contacted again by phone around the child’s first birthday, and again on his or her second birthday. At 3 years old, an interview was held in the child’s home. Medical information was recorded and non-invasive samples were taken. In this thesis, we used the data collected at birth and at the 2-month interview (Figure 7).
Families signed a consent form presenting the general aim of the study and all data were analysed anonymous. The ELFE study received approval of bodies overseeing ethical data collection in France (Comité Consultatif sur le Traitement des Informations pour la Recherche en Santé: CCTIRS, Commission National Informatique et Libertés: CNIL).

Table of contents :

Food chemicals during pregnancy
Environmental contaminants and child’s growth
Assessment of prenatal exposure to environmental chemicals
Growth parameters
Acrylamide and diet quality
Acrylamide and growth in the EDEN study
Acrylamide and postnatal growth in the MoBa study
Single contaminant approach
Mixture of contaminant approach
Beneficial nutrients vs environmental contaminants from maternal diet
Strengths and limitation
Potential mechanisms
Public health implications


Related Posts