Multiple pesticide analysis in hair samples of pregnant French women: Results from the ELFE national birth cohort

Multiple pesticides in mothers' hair samples and children's measurements at birth: Results from the French national birth cohort (ELFE)

“…Elastic net regression modeling presents the advantage of handling multicollinearity between pesticides, while limiting the risk of false positives due to multiple testing (Chadeau-Hyam et al, 2013;Agier et al, 2016). Correlations between the 64 compounds studied, weak in most of the situations, have previously been described (Béranger et al, 2018). Among the pesticides associated with measurements at birth in our study, we observed polychoric/tetrachoric correlation coefficients above 0.6 only for mecoprop and 2,4D; fipronil and fipronil sulfone; and isoproturon and chlortoluron.…”

“…We chose to measure pesticides in hair to estimate exposure during the entire pregnancy, assuming that hair would aggregate exposure over a long time period. This would present the advantage of limiting intraindividual variability, as previously suggested (Béranger et al, 2018), and of capturing sporadic exposure to chemicals with short half-lives. Nonetheless, using this biological matrix did not enable us to distinguish the pattern of exposure (low and chronic, or sporadic and intense), or to assess the point during pregnancy when the exposure occurred, or to determine whether the period of contamination corresponds to a relevant window of exposure for fetal growth.…”

“…Elastic net regression modeling results in a balanced approach that makes it possible to select the strongest predictors, while enabling collinear predictors to be included in the model. In their comparison of different multipollutant approaches with a simulated dataset, Agier et al (2016) suggested that an elastic net regression modeling approach performs well in our situation (i.e., environmental exposure to chemicals modestly correlated between them -see Béranger et al, 2018).…”

“…Overall 111 pesticides and 29 pesticide metabolites were screened in hair samples. These have been fully described elsewhere (Béranger et al, 2018).…”

“…The analytical methods used to measure multiple pesticide residues in hair samples have previously been published in detail (Duca et al, 2014b;Hardy et al, 2015;Salquebre et al, 2012;Béranger et al, 2018). Hair samples were successively washed with sodium dodecyl sulfate (SDS) and then methanol, to remove chemicals that might have been deposited on the hair surface (Duca et al, 2014a).…”