Population variation in biomonitoring data for persistent organic pollutants (POPs): An examination of multiple population-based datasets for application to Australian pooled biomonitoring data
Introduction
National biomonitoring efforts designed to characterize the levels and distribution of environmental chemical pollutants in populations have typically relied upon a sampling and analysis strategy including hundreds or thousands of individual biological samples. Such efforts are underway in countries including the United States (the National Health and Nutrition Examination Survey, or NHANES), Canada (the Canadian Health Measures Survey, or CHMS), Germany (the German Environmental Survey, GerES), and others (Porta et al., 2008, Schoeters et al., 2011). Biomonitoring programs are resource intensive, often requiring an extensive infrastructure and expertise dedicated to population sampling and significant financial resources to analyze the collected samples (Porta et al., 2009, Porta et al., 2012). Such detailed sampling strategies allow cross-sectional analyses for potential associations with both environmental sources of contamination and with health outcomes. However, if the goal of the effort is to characterize population exposure levels and document spatial and temporal trends, other approaches may be considered.
In Australia and elsewhere, a series of studies relying on multiple pooled serum samples, a less resource-intensive approach, has been employed (Caudill, 2010, Karrman et al., 2006, Toms et al., 2009a, Toms et al., 2009b). Reliance on analysis of multiple pooled samples allows characterization of central tendencies of concentrations in the age, gender, and geographical groups sampled. This approach has produced information allowing evaluation of patterns with age, gender, region of Australia, and, over time, information about temporal trends in the central tendency of measured chemical concentrations. Analytes examined to date include dioxin-like compounds, polychlorinated biphenyls (PCBs), organochlorine pesticides (OCPs), polybrominated diphenylethers (PBDEs), and perfluorinated compounds (PFCs).
While the pooling approach allows estimation of central tendencies (specifically, arithmetic means, under the condition of equally weighted samples contributing to each pool) of concentrations in the sampled subgroups, it does not provide direct information about population variation in biomarker concentrations. Characterization of population variation – and in particular, estimation of typical upper bound concentrations – may provide benchmarks for assessing whether individuals have biomarker concentrations beyond those typically observed in the population. For example, the German Human Biomonitoring Commission estimates reference values at the 95th percentile (RV95), for use in identifying individuals with unusual exposure levels (Angerer et al., 2011). Characterization of population upper bounds for toxic compounds also allows comparison with health-based screening values, when available, to assess whether general population levels approach or exceed such guidance values.
Statistical approaches have been published to estimate population variation based on variation in concentrations measured in multiple pools from a given population or subpopulation (Caudill, 2010, Caudill, 2011, Caudill, 2012, Caudill et al., 2007). These approaches rely upon parametric assumptions regarding the shape of the underlying population distribution. As an alternative, this analysis examines empirical patterns in survey data based on the hypothesis that biological variability in processes relative to the accumulation and elimination of such compounds may be similar across populations, leading to similar degrees of variation in biomarker concentrations in different populations, even when the absolute levels of exposure differ. Specifically, variations in biomarker concentrations of persistent compounds within a population and age group that are exposed to a given matrix of environmental concentrations usually are controlled by a few factors. These include interindividual variation in metabolism and elimination rates, interindividual variation in long-term exposure rates (for example, high fat vs. low fat diet, for lipophilic compounds), history of breast feeding, and, in the case of lipophilic compounds, interindividual variation in lipid content of blood (Bernert et al., 2007, Phillips et al., 1989, Porta et al., 2009). The net effect of these factors on the degree of variation of biomarker concentrations may be similar across different populations, even if absolute exposure levels are not.
This project examined available biomonitoring datasets from the US National Health and Nutrition Examination Survey (NHANES), the Canadian Health Measures Survey (CHMS), the German Environmental Survey (GerES), the Catalan Health Interview Survey (CHIS), and the Flemish Environmental Health Survey II (FLEHS II) to inform interpretation of the Australian pooled biomonitoring data. The objectives of the present study were to:
- •
Compare population subgroup arithmetic mean concentrations of the subject analytes in Australian pools to those observed in datasets from the US, Canada, Germany, and Catalonia; and
- •
Assess variation within population subgroups across available datasets for similarity and differences for key persistent organic pollutants (POPs).
We use the results of this analysis to estimate population 95th percentiles (with confidence limits) for the Australian age groups based on the pooled sample means and discuss the estimated levels in the context of health risk screening values, where available.
Section snippets
Target analytes
The focus of this analysis is on persistent chemicals that are found widely in the general populations in countries around the world. Specifically, we included analytes that are likely to be detected at high rates (so that methods for imputation of non-detected concentrations would not be influential in the calculation of mean or upper percentile statistics) and that have been studied in multiple population-representative studies. For this effort we selected three indicator polychlorinated
Results
Each of the datasets included in this analysis covered somewhat different age ranges and numbers of sampled participants (Table 1). Because the purpose of the analysis is to inform the interpretation of the Australian pooled biomonitoring surveys, age groups were defined using the same cutpoints used by the Australian sampling program. Data representing the youngest age group (0 to 4 years) was not available for these analytes from any of the available datasets, although arguably the cord blood
Discussion
The results from this analysis suggest that, for many of the persistent chemicals examined here, the degree of variation between typical upper bound concentrations as estimated by the 95th percentile in the population and the arithmetic mean in the population is reasonably similar between populations, even when the absolute levels of the analyte vary substantially across these populations. This observation is consistent with the hypothesis that population variation in biomarker concentrations
Competing financial interests
The authors declare no competing financial interests.
Acknowledgments
The analyses presented here were funded under contract to the Australian Department of the Environment. The authors thank Suzelle Giroux from Statistics Canada; Tomàs López and Yolanda Rovira from the Hospital del Mar Institute of Medical Research; and Margarete Seiwert and all other colleagues of the German Environmental Survey team for provision of data and scientific and technical advice. We are also grateful to the colleagues from the Flemish Centre of Expertise on Environment and Health,
References (38)
- et al.
Human biomonitoring assessment values: approaches and data requirements
Int J Hyg Environ Health
(2011) - et al.
Biomonitoring equivalents for hexachlorobenzene
Regul Toxicol Pharmacol
(2010) - et al.
Calculation of serum “total lipid” concentrations for the adjustment of persistent organohalogen toxicant measurements in human samples
Chemosphere
(2007) - et al.
Geometric mean estimation from pooled samples
Chemosphere
(2007) - et al.
Guidelines for the derivation of biomonitoring equivalents: report from the Biomonitoring Equivalents Expert Workshop
Regul Toxicol Pharmacol
(2008) - et al.
Biomonitoring equivalents for DDT/DDE
Regul Toxicol Pharmacol
(2011) - et al.
Environmental surveys, specimen bank and health related environmental monitoring in Germany
Int J Hyg Environ Health
(2012) Human contamination by environmental chemical pollutants: can we assess it more properly?
Prev Med
(2012)- et al.
Monitoring concentrations of persistent organic pollutants in the general population: The international experience
Environ Int
(2008) - et al.
Correcting serum concentrations of organochlorine compounds by lipids: alternatives to the organochlorine/total lipids ratio
Environ Int
(2009)
Distribution of blood concentrations of persistent organic pollutants in a representative sample of the population of Catalonia
Environ Int
Number of persistent organic pollutants detected at high concentrations in a general population
Environ Int
Internal exposure to pollutants measured in blood and urine of Flemish adolescents in function of area of residence
Chemosphere
Twenty years of the German Environmental Survey (GerES): human biomonitoring—temporal and spatial (West Germany/East Germany) differences in population exposure
Int J Hyg Environ Health
Overview of the study design, participation and field work of the German Environmental Survey on Children 2003–2006 (GerES IV)
Int J Hyg Environ Health
Renal function, cytogenetic measurements, and sexual development in adolescents in relation to environmental pollutants: a feasibility study of biomarkers
Lancet
Opinion of the French Food Safety Agency on interpreting the health impact of PCB concentration levels in the French population
Interpreting variability in population biomonitoring data: role of elimination kinetics
J Expo Sci Environ Epidemiol
Evaluation of biomonitoring data from the CDC National Exposure Report in a risk assessment context: perspectives across chemicals
Environ Health Perspect
Cited by (22)
Developing human biomonitoring as a 21st century toolbox within the European exposure science strategy 2020–2030
2022, Environment InternationalOrganohalogens: A persisting burden in Slovenia?
2021, Environmental ResearchPer- and polyfluoroalkyl substances (PFAS) in Australia: Current levels and estimated population reference values for selected compounds
2019, International Journal of Hygiene and Environmental HealthCitation Excerpt :The P95:AM ratios and 95% CIs calculated for each age group for each of the six frequently-detected compounds based on the NHANES 2013–14 and 2015-16 datasets are presented in Fig. 2. NHANES 1999–2000 cycle data for PFOA and PFOS are also presented based on a previous assessment (Aylward et al., 2014). The 95% CIs on the ratios are somewhat skewed, reflecting the influence of upper tail of the distribution of measured concentrations in the NHANES datasets.
Integration of biomonitoring data into risk assessment
2018, Current Opinion in ToxicologyCitation Excerpt :While cause and effect cannot necessarily be established, data from the NHANES program support the efficacy of exposure reduction efforts for second-hand smoke exposure (Fig. 2) and inorganic arsenic in drinking water [43]. Persistent organic chemicals listed under the Stockholm convention and subject to various bans and restrictions on use also show clear declining trends in the general population, including perfluorinated compounds [44] and persistent organochlorine pesticides [39,45,46]. Finally, changes in blood lead concentrations observed following the reduction and elimination of the use of lead in gasoline demonstrated clearly the impact of this regulatory action on childhood lead exposure [1].