Reviewing the variability in urinary concentrations of non-persistent organic chemicals: evaluation across classes, sampling strategies and dilution corrections

Various biomonitoring studies have been carried out to investigate the exposure of populations by measuring non-persistent organic chemicals in urine. To accurately assess the exposure, study designs should be carefully developed to maximise reproducibility and achieve good characterization of the temporal variability. To test these parameters, the intraclass correlation coefficients (ICCs) are calculated from repeated measurements and range from poor (<0.4) to excellent (≥0.75). Several studies have reported ICCs based on diverse study designs, but an overview, including recommendations for future studies, was lacking. Therefore, this review aimed to collect studies describing ICCs of non-persistent organic chemicals, discuss variations due to study design and formulate recommendations for future studies. More than 60 studies were selected, considering various chemical classes: bisphenols, pyrethroids, parabens, phthalates, alternative plasticizers and phosphate flame retardants. The variation in ICCs for an individual chemical was high (e.g. ICC of propyl paraben = 0.28-0.91), showing the large impact of the study design and of the specific exposure sources. The highest ICCs were reported for parabens (median = 0.52), while lowest ICCs were for 3-phenoxybenzoic acid (median = 0.08) and bisphenol A (median = 0.20). Overall, chemicals that had an exposure source with high variation, such as the diet, showed lower ICCs than those with more stable exposure sources, such as indoor materials. Urine correction by specific gravity had an overall positive effect on reducing the variability of ICCs. However, this effect was mostly seen in the adult population, while specific compounds showed less variation with creatinine correction. Single samples might not accurately capture the exposure to most non-persistent organic chemicals, especially when small populations are sampled. Future studies that examine compounds with low ICCs should take adequate measures to improve accuracy, such as correcting dilution with specific gravity or collecting multiple samples for one participant.