Study design
The ENVIRonmental influences
ON early AGEing (ENVIR
ONAGE) birth cohort was established in 2011 at Hasselt University and recruits pairs of mothers and neonates (singleton births only) at birth at the East-Limburg Hospital (Genk, Belgium). The catchment area of the cohort is located in the north-east of Belgium, in the province of Limburg, Flanders [
9]. A nation-wide survey of Belgian pregnant women showed that nearly 60% was iodine deficient [
10], which corroborated with findings in the ENVIR
ONAGE cohort [
8]. The Ethics Committee of Hasselt University and the East-Limburg Hospital approved the study protocol that was carried out following the Declaration of Helsinki. Written informed consent was obtained from all participating mothers.
The inclusion criteria were mothers who provided informed consent and were able to fill out questionnaires in Dutch, while planned caesareans were excluded. Participants were recruited equally during all seasons of the year, and midwives recorded the main reasons for non-participation (such as failure to ask about participation, communication barriers, or complications during labour). Full details of the study design were published previously [
9].
Medical records of the hospital were used to retrieve information on newborns’ sex, gestational age, date of delivery, maternal age, maternal pre-pregnancy BMI, maternal weight gain during the pregnancy, and gestational hypertension. Gestational age was determined based on the mother’s last menstrual period in combination with ultrasound data. Mothers completed questionnaires, which provided us with detailed information about socio-demographic and lifestyle factors such as parity, newborns’ ethnicity, maternal education, household smoking habits, vitamin use, and consumption of alcoholic beverages, fish, and fruit and vegetables.
In the present study, 500 bio-banked placentas were randomly selected from 799 eligible mother-neonate pairs who were recruited between March 1st, 2013, and April 1st, 2017. 462 participants remained for statistical analysis after the exclusion of failed iodine extractions (n = 2), children born before 37 weeks of pregnancy (n = 3), mothers with previous thyroid complications (i.e. hypo- [n = 12] or hyperthyroidism [n = 4]), pre-eclamptic pregnancies (n = 8), or missing data on fruit and vegetable consumption (n = 9).
Blood collection and thyroid hormone measurements
Plastic BD Vacutainer® Lithium Heparin Tubes (BD, Franklin Lakes, New Jersey, USA) were used to collect 8 mL of cord blood immediately after delivery and 8 mL of maternal blood at one day after delivery via venepuncture. Plasma from these samples was obtained after centrifugation at 3,200 rpm for 15 min and was frozen at − 80 °C. At the clinical laboratory of the East-Limburg Hospital, the plasma levels of the thyroid hormones FT4 (pmol/L), FT3 (pmol/L), and TSH (mIU/L) were measured with an electro-chemiluminescence immunoassay using the Modular E170 automatic analyser (Roche, Basel, Switzerland).
Statistical methods
Database management and statistical analysis were performed with the SAS software, version 9.4 (SAS Institute, Cary, NC, USA). Mean ± standard deviation (SD) is given for continuous variables and the proportion for categorical variables. The normality of the data distributions was tested with the Shapiro–Wilk statistic and quantile–quantile plots.
First, we assessed the distributions of continuous variables (ANOVA) and the proportions of categorical variables (χ2-statistics) across tertiles of the placental iodine concentrations.
Second, we identified determinants of placental iodine concentrations by forward stepwise linear regression procedures in which we set the p-value at 0.15 for the independent variables to enter and to stay in the model. The following 20 variables were considered: maternal age, pre-pregnancy BMI, gestational weight gain, maternal education (coded as low [no diploma or primary school], middle [high school], or high [college or university degree]), gestational hypertension (coded as yes or no), maternal smoking habits (coded as never-smoker, cessation before pregnancy, or current smoker), maternal exposure to indoor second-hand smoke (SHS; coded as non-smokers not exposed to SHS, non-smoker exposed to SHS, or current smoker), consumption of alcoholic beverages during pregnancy (coded as none or maximum one glass per day), maternal fish consumption (coded as did not consume during pregnancy, consumed fish less than once per week, or at least once per week), maternal fruit and vegetable consumption (coded as less than once per day, daily, or more than once per day), multi-vitamin use (coded as yes or no), anti-inflammatory medication use (coded as yes or no), blood pressure medication use (coded as yes or no), antibiotics use (coded as yes or no), gestational age, neonate’s sex, parity (coded as one, two, or at least three children), ethnicity [classified on the basis of the native country of the neonates’ grandparents as either European (at least two grandparents were European) or non-European (at least three grandparents were of non-European origin)], date of delivery as a proxy for the time trend, and season of delivery [coded as winter (December 21st to March 20th), spring (March 21st to June 20th), summer (21st to September 20nd), or autumn (September 21st to December 20th)]. Effect estimates are presented for the final model with the determinants selected by the stepwise selection procedure.
Third, to evaluate the biological significance of our placental iodine findings, we measured the plasma FT
4, FT
3, and TSH hormone levels in cord blood and maternal blood. After exclusion of outliers (n = 6), and missing blood samples for either child or mother (n = 78), a total of 378 participants remained for the thyroid hormone analyses. We assessed the associations between these hormones and placental iodine with a linear regression model, adjusted for common variables such as the neonate’s sex, birth weight, gestational age, parity, maternal pre-pregnancy BMI, maternal age at delivery, maternal education, and maternal smoking status (Model A) as tested previously [
11], or for the determinants selected by the stepwise selection procedure (Model B).
In sensitivity analyses, we re-ran the stepwise linear regression model with the same twenty variables, except for pre-pregnancy BMI and gestational weight gain, which were in the sensitivity analysis recoded into a new variable based on the United States Institute of Medicine recommendations on the maternal gestational weight gain [
12]. Recoding comprised: ‘low weight gain’ (n = 103), ‘normal weight gain’ (n = 146), and ‘excess weight gain’ (n = 213). For underweight women, the ‘normal weight gain’ is between 12.5 and 18 kg; for normal, overweight, and obese women, the ranges are 11.5–16 kg, 7–11.5 kg, and 5–9 kg, respectively. Women who gained weight below these ranges were categorized as ‘low weight gain’, while those who gained more weight were categorized as ‘excess weight gain’ [
12]. In further sensitivity analyses, we excluded women who consumed any alcoholic beverages during their pregnancy (n = 62) or those who smoked during the pregnancy (n = 42).