A revised concept for deriving reference values for internal exposures to chemical substances and its application to population-representative biomonitoring data in German children and adolescents 2014-2017 (GerES V)
Michael Hoopmann,Aline Murawski,Michael Schümann,Thomas Göen,Petra Apel,Nina Vogel,Marike Kolossa-Gehring,Claudia Röhl,Human Biomonitoring Commission of the German Environment Agency
DOI: https://doi.org/10.1016/j.ijheh.2023.114236
Abstract:HBM reference values, in contrast to toxicologically derived values, are statistically derived values that provide information on the exposure of the population. The exceedance frequency (if applicable for individual population groups) is often a first assessment standard for the local exposure situation for municipalities. More than 25 years have passed since the German Human Biomonitoring Commission (HBMC) formulated the first recommendations for the derivation of population-based reference values (HBM reference values, RV95) for substance concentrations based on HBM studies. A fundamental revision is timely, for several reasons. There have been considerable advances in relevant statistical methods, which meant that previously time-consuming and inaccessible procedures and calculations are now widely available. Furthermore, not all steps for the derivation of HBM reference values were clearly elaborated in the first recommendations. With this revision we intended to achieve a rigorous standardization of the entire process of deriving HBM reference values, also to realise a higher degree of transparency. In accordance with established international practice, it is recommended to use the 95th percentile of the reference distribution as the HBM reference value. To this end, the empirical 95th percentile of a suitable sample should be rounded, ensuring that the rounded value is within the two-sided 95% confidence interval of the percentile. All estimates should be based on distribution-free methods, and the confidence interval should be estimated using a bootstrap approach, if possible, according to the BCa ("bias-corrected and accelerated bootstrap"). A minimum sample size of 80 observations is considered necessary. The entire procedure ensures that the derived HBM reference value is robust against at least two extreme values and can also be used for underlying mixed distributions. If it is known in advance that certain subgroups (different age groups, smokers, etc.) show differing internal exposures, it is recommended that group-specific HBM reference values should be derived. Especially when the sample sizes for individual subgroups are too small, individual datasets with potential outliers can be excluded in advance to homogenize the reference value population. In the second part, new HBM reference values based on data of the German Environmental Survey for Children and Adolescents (GerES V, 2014-2017) were derived in accordance with the revised recommendations. The GerES V is the most recent population-representative monitoring of human exposure to pollutants in Germany on children and adolescents aged 3-17 years (N = 2294). RV95 for GerES V are reported for four subgroups (males/females and 3-11/12-17 years) for 108 different substances including phthalates and alternative plasticisers, metals, organochlorine pesticides, polychlorinated biphenyls (PCB), per- and polyfluoroalkyl substances (PFAS), parabens, aprotic solvents, chlorophenols, polycyclic aromatic hydrocarbons (PAH) and UV filter, in total 135 biomarkers. Algorithms implemented in R were used for the statistics and the determination of the HBM reference values. To facilitate a quality control of the study data, the corresponding R source code is given, together with graphical representations of results. The HBM reference values listed in this article replace earlier RV95 values derived by the HBMC for children and adolescents from data of precedent GerES studies (e.g. published in Apel et al., 2017).