Variability of urinary concentrations of non-persistent chemicals in pregnant women and school-aged children

Background Exposome studies are challenged by exposure misclassification for non-persistent chemicals, whose temporal variability contributes to bias in dose-response functions. Objectives We evaluated the variability of urinary concentrations of 24 non-persistent chemicals: 10 phthalate metabolites, 7 phenols, 6 organophosphate (OP) pesticide metabolites, and cotinine, between weeks from different pregnancy trimesters in pregnant women, and between days and between seasons in children. Methods 154 pregnant women and 152 children from six European countries were enrolled in 2014–2015. Pregnant women provided three urine samples over a day (morning, midday, and night), for one week in the 2nd and 3rd pregnancy trimesters. Children provided two urines a day (morning and night), over two one-week periods, six months apart. We pooled all samples for a given subject that were collected within a week. In children, we also made four daily pools (combining morning and night voids) during the last four days of the first follow-up week. Pools were analyzed for all 24 metabolites of interest. We calculated intraclass-correlation coefficients (ICC) and estimated the number of pools needed to obtain an ICC above 0.80. Results All phthalate metabolites and phenols were detected in >90% of pools whereas certain OP pesticide metabolites and cotinine were detected in <43% of pools. We observed fair (ICC = 0.40–0.59) to good (0.60–0.74) between-day reliability of the pools of two samples in children for all chemicals. Reliability was poor (<0.40) to fair between trimesters in pregnant women and between seasons in children. For most chemicals, three daily pools of two urines each (for weekly exposure windows) and four weekly pools of 15–20 urines each would be necessary to obtain an ICC above 0.80. Conclusions This quantification of the variability of biomarker measurements of many non-persistent chemicals during several time windows shows that for many of these compounds a few dozen samples are required to accurately assess exposure over periods encompassing several trimesters or months.