Quality Assurance and Harmonization for Targeted Biomonitoring Measurements of Environmental Organic Chemicals Across the Children’s Health Exposure Analysis Resource Laboratory Network (2024)

2.1. Participation in external proficiency testing (PT) programs

PT providers distribute common samples to participating laboratories at regular intervals, (typically 2–3 times a year). Laboratories measure the target analyte(s) and report the results back to the PT provider. The PT provider performs a statistical analysis of results reported from all participating laboratories, assigns target values, scores the responses and then submits a performance report back to each laboratory. The PT provider also sets acceptance criteria or quality specifications for the analytical data reported by participating laboratories. How these acceptance criteria are established varies between EQA/PT programs. Some EQA schemes may use a z-scoring model while others may adopt fixed criteria. For other schemes, an acceptance criterion could be that a laboratory’s result must be within mean ±3 SD of data from all participating laboratories or based on the values determined in “reference laboratories.” Participation in a PT program enables evaluation of the performance of the laboratory, competency for a particular biomarker analysis and accuracy of the analysis.

The CHEAR laboratories participated in several commercial EQA/PT programs for environmental chemical analysis in urine and/or serum matrices. The targeted organic chemical analysis laboratories were encouraged to participate in the External Quality Assessment Scheme for Organic Substances in Urine (OSEQAS) and the Arctic Monitoring and Assessment Program (AMAP) for persistent organic pollutants in serum offered by the Centre de Toxicologie du Québec (CTQ; https://www.inspq.qc.ca/en/ctq/about). Participation in the German External Quality Assessment Scheme (G-EQUAS) was required, which included the analyses of biological materials (serum or urine) for various organic analytes (http://www.g-equas.de/). A few CHEAR laboratories also participated in other EQA programs such as the Centers for Disease Control and Prevention’s (CDC) Biomonitoring Quality Assurance for State Programs (BQASP) and Ensuring the Quality of Urinary Iodine Procedures (EQUIP) as well as European Human Biomonitoring Initiative (HBM4EU) – Inter-laboratory Comparison Investigations and External Quality Assurance Scheme (ICI/EQUAS).

In this article, we describe results from the CHEAR laboratories’ participation in the G-EQUAS, which comprises two rounds per year (spring and fall). The G-EQUAS comprises determination of toxicological parameters including environmental toxicants in blood, plasma/serum and urine at occupational and environmental exposure concentration ranges. This program has offered EQA exercises since 1992, with a successive increase of parameter numbers; over 50 organic chemical parameters (in urine and serum) of public health relevance were included in “Round 63”, offered in the spring of 2019. In 2010, the number of international laboratories enrolled in this program ranged from 3 to 37, depending on the parameter analyzed (Göen et al., 2012). Two samples with different concentrations of biomarkers were sent to the participating laboratories. The target values as well as the tolerance ranges (i.e., acceptable criteria) were assigned based on results from laboratories invited to serve as reference laboratories. A few CHEAR laboratories also participated as reference laboratories for select classes of environmental chemicals. Successful participation was indicated when the laboratory’s results were within the tolerance ranges for both samples, which were calculated as 3 × SD of the results of the reference laboratories (Göen et al., 2012). Since 2016, five CHEAR laboratories participated in G-EQUAS with each of them analyzed varying numbers (>10) of biomarkers.

2.2. CHEAR Quality Control Pool

Two urine pools (designated as ‘Pool A’ and ‘Pool B’) were prepared and distributed to the CHEAR laboratories for analysis, along with every batch of CHEAR project/study samples for organic contaminants. In 2017, urine was collected into borosilicate containers from anonymous volunteers at Emory University, Atlanta, Georgia. To create Pool A, 10 L of urine was filtered through a paper filter to remove particulates. To create Pool B, 4.5 L of filtered urine was mixed with 500 mL of a combined smokers’ urine pool provided by the University of Minnesota. The smokers’ urine pool was derived from nine 24-h urine collections obtained in 2015. Because Pool A urine was not expected to contain tobacco smoke biomarkers (such as cotinine), smokers’ urine was added to enable quantifiable levels of these biomarkers in Pool B urine. The two pools were assumed to have different levels of organic contaminants. Both pools (Pool A and Pool B) were then divided into 1.5 mL aliquots using a liquid handling robot, labeled, and stored at −20° C until shipment to the CHEAR laboratories. Laboratories received these CHEAR urine QC pools on dry ice and the pools were immediately stored in −20°C or −80°C freezers. The laboratories were not informed of the identity of the two different pools and they were inserted into analytical runs as if they were unknown samples. The laboratories were instructed to analyze three separate aliquots of each urine QC pool (A and B) for every 100 CHEAR study samples. At the time of analysis of CHEAR study samples, the QC pool samples were randomized and analyzed in the same manner as the samples. Each vial of QC pool sample was intended for a single use to avoid stability issues from multiple freeze-thaw cycles. The QC pool samples were analyzed for a wide range of environmental chemicals including phthalate metabolites, environmental phenols, monohydroxy polycyclic aromatic hydrocarbons, dialkylphosphate pesticide metabolites, and tobacco metabolites. Concentrations of target analytes in QC pool samples were reported in the same manner as the CHEAR study samples. Insertion of QC pool samples at two different concentrations in every batch of study sample analysis by all participating laboratories enabled assessment of batch-to-batch variabilities (intra-lab variabilities) and inter-laboratory variabilities in the analysis. QC pool data for various chemical groups obtained from each of the participating laboratories were combined. The mean and standard deviation (SD) were calculated for each analyte group. Outliers were removed if the values were greater than ±3 SD of the group mean. With the large number of QC pool samples analyzed by multiple CHEAR laboratories over a period of time, the coefficient of variation (CV) was calculated to describe precision of the data as:

Where SD is the standard deviation and X is the mean.

2.3. NIST Standard Reference Materials^® (SRMs)

Among several internal QC measures taken by the CHEAR laboratories, inclusion of NIST SRMs in every batch of study samples was recommended. The CHEAR laboratories were instructed to include NIST Standard Reference Material^® (SRM^®) 3672 (organic contaminants in smokers’ urine-frozen) and SRM 3673 (organic contaminants in non-smokers’ urine-frozen) with every batch of urine samples analyzed for target analytes for which the SRMs have certified or reference values. Similarly, the laboratories were instructed to include SRM 1958 (organic contaminants in fortified human serum-freeze-dried) into every batch of 100 serum analyses for target chemicals for which this reference material has certified or reference values. The SRMs 3672 and 3673 were intended for use in evaluating the accuracy and precision of methods for PAHs (hydroxylated PAHs), phthalates, environmental phenols, VOC metabolites, tobacco metabolites and creatinine in urine. Similarly, SRM 1958 was intended for use in evaluating accuracy and precision of the analytical methods for determination of selected PCB congeners, chlorinated pesticides, polybrominated diphenyl ether (PBDE) congeners, PFAS and lipids in human serum. Reference values are available for selected polychlorinated dibenzo-p-dioxins (PCDDs), polychlorinated dibenzofurans (PCDFs), and PFAS. Further details of these SRMs including instructions for storage, handling and analysis have been described elsewhere (https://www.nist.gov/srm); these SRMs (when purchased) are accompanied with certificates of analysis with detailed use instructions. A NIST-certified value is a value for which there is highest confidence in its accuracy, with all known or suspected sources of uncertainty having been taken into account. Certified values are metrologically traceable to the SI derived unit for mass fraction. Reference values are noncertified values that are estimates of the true values but for which all sources of uncertainty may not be fully characterized. The SRM found values, generated in each of the CHEAR laboratories with every batch of study samples, were submitted for the calculation of recoveries (a measure of accuracy) of target analytes and precision (from replicate analyses of SRMs across multiple batches). The mean recovery of each analyte was calculated as:

where X is the mean. The SRM data have also been used in the calculation of expanded uncertainty in analysis, as described in Galusha et al. (2021).

2.4. Blinded duplicate samples

The CHEAR program encouraged individual investigators who submitted study samples to the laboratories to insert duplicate samples, the identities of which were blinded to the laboratories. The number of blinded duplicate samples inserted into a batch varied depending on the study, based on the availability of additional volumes of samples and study size. Once the data were generated and submitted to the CHEAR Data Center, statistical analyses were performed for reporting the results of blinded duplicate samples to individual investigators. As a measure of precision, the relative percent difference (RPD) was calculated for each valid duplicate pair, which is a duplicate pair with both sample concentrations above the limit of detection (LOD) for that analyte. RPD is the percent of the difference relative to the mean of the duplicate value and was calculated as:

An RPD of zero indicates that results for the two duplicate samples were identical. RPDs greater than zero indicate a difference between the two sample concentrations. A median RPD of valid duplicate pairs exceeding 30% for a given analyte was flagged for further investigation, since this can be indicative of poor precision (Udesky et al., 2019). Any deviations from these QC guidelines were expected to be rigorously checked and appropriate corrective actions undertaken including the reanalysis of samples where warranted. Further, an intra-class correlation (ICC) was calculated for these duplicate pairs, which is a composite measure of reliability and is used to assess both the degree of correlation and agreement between measurements. The ICC was estimated from a one-way random effects model (Koo and Li, 2016). An ICC of 1 indicates that results of the two samples in a duplicate pair, for each pair, were equal. Although there is no standard value for acceptable reliability using the ICC, Koo and Li (2016) stated that ICC values less than 0.5 indicate poor reliability, with values between 0.5 and 0.75 indicating moderate reliability, values between 0.75 and 0.90 indicating good reliability, and values greater than 0.90 indicating excellent reliability.

3.1. Participation in external proficiency testing programs

The CHEAR laboratories participated in several EQA schemes; however, the primary participation was in G-EQUAS and CTQ-EQAS, which are the subject of this report. There are some overlapping organic analytes offered between the two EQA schemes. Between 2016 and 2019, five CHEAR laboratories routinely participated in G-EQUAS (Table 1). The number of biomarkers analyzed (i.e., capability) varied among laboratories, from 2–20 in 2016 to 10–37 in 2019. The percentage of samples that were scored within the tolerance range varied from 60–90% in 2016 to 91–97% in 2019. Across the 4-year period, CHEAR laboratory capabilities grew, i.e., the number of biomarkers analyzed increased in all five participating laboratories. Overall laboratory performance also improved (i.e., improved harmonization) for each laboratory. It is also worth noting that the number of individual analytes prescribed under each chemical class varied (Table 2). For example, only seven phthalate metabolites were included in G-EQUAS whereas several CHEAR laboratories had the capability to analyze up to 20 different phthalate metabolites in urine (Table S1). The CDC’s BQASP included up to 11 phthalate metabolites in their inter-laboratory study, but participation in that program is limited to US-based State Public Health Laboratories funded under the CDC’s National Biomonitoring Program. For PFAS, only two target analytes (PFOS and PFOA) were included in G-EQUAS, whereas up to nine PFAS were offered in AMAP (i.e., PFDA, PFHpS, PFHpA, PFHxA, PFHxS, PFNA, PFOA, PFOS [total, linear and branched] and PFUdA) in 2019. The results from AMAP are also available for the participating CHEAR laboratories (data not shown). Participation in more than one EQA scheme was advantageous to document a laboratory’s capabilities and performance. The CHEAR laboratories were encouraged to participate in multiple EQA/PT programs, if and when available, to demonstrate capabilities for a comprehensive suite of analytes within a chemical class. For those analytes that were not covered under the existing EQA/PT programs, several laboratories validated the analyses through internal method validation protocols (e.g., fortified samples, use of labeled standards) and/or inter-laboratory comparison studies (splitting of same samples between laboratories).

One of the deficiencies identified during the course of participation in EQA/PT programs was that not all programs cover every target organic analyte measured in the CHEAR laboratories. There exists a need for inclusion of a wide range of organic chemicals in the PT programs, especially for those analytes that are routinely determined in human biomonitoring studies. As many laboratories throughout the world are actively engaged in the analysis of human specimens for trace organic chemicals, there is a critical need to assess laboratory performance and improve the harmonization of analytic measurements. However, establishment of a comprehensive EQA/PT program requires well-developed infrastructure and resources. For instance, preparation of EQA/PT samples involves selecting and handling of a large amount of suitable materials. The material has to be thoroughly hom*ogenized, dispensed into appropriate bottles, labelled and the final products tested for hom*ogeneity and stability over time, then the materials are dispatched to the laboratories. The results returned by participating laboratories should be statistically analyzed, target values assigned and performance reported back to the laboratories. Thus, the establishment and organization of an EQA/PT program is a complex task that can only be successfully carried out with the assistance of well-equipped and trained laboratories; recommended protocols for such inter-laboratory programs have been established to meet internationally accepted standards (ISO, 2017).

Another observation that could be made from participation in the EQA/PT programs was the lack of harmonization of a target chemical’s identity, i.e., chemical identification information was sometimes so fragmentary that it could mislead an analyst. Abbreviations such as 5-OH-MEHP, 5-oxo-MEHP and 5-carboxy-MEPP are used in some cases to represent mono-(2-ethyl-5-hydroxyhexyl) phthalate (mEHHP), mono-(2-ethyl-5-oxyhexyl) phthalate (mEOHP) and mono-(2-ethyl-5-carboxypentyl) phthalate (mECPP), respectively. Lack of full chemical names or chemical abstracts service (CAS) numbers have occasionally led to misidentification of 5-oxo-MEHP for mEHHP and 5-OH-MEHP for mEOHP. Similarly, the abbreviation MEP is used frequently for both monoethyl phthalate and methyl paraben. Unless full chemical names with CAS numbers are provided, these ad hoc abbreviations can be misleading. Furthermore, attention should be paid to structural isomers of target analytes and in some cases EQA/PT programs request analysis for select isomers (e.g., n-PFOS versus total PFOS). In certain cases, it is not clearly specified if analytes are to be measured in ‘free’ form or ‘total’ form (e.g., cotinine). To overcome the challenges associated with multiple chemical abbreviations, the CHEAR program developed a list of analyte codes for each of the target chemicals measured in the laboratories along with their CAS numbers (Table S1).

3.2. CHEAR QC Pools

Two CHEAR urine QC pools (A and B) were prepared and distributed to each of the participating laboratories. In total, concentrations of up to 30 different environmental phenols and 27 different phthalates were measured, among several other target chemical classes (Tables 3 and ​and4).4). As stated earlier, not all laboratories measured all 30 phenols and 27 phthalates (see Tables S3 and S4 for lab specific results). The reported concentrations were collected for 240 QC samples (120 for Pool A and 120 for Pool B) across seven different CHEAR studies performed in four different laboratories from December 2017 to May 2018. In total, over 4,700 individual concentration values above the limits of detection (LODs) were reported for environmental phenols and phthalate metabolites for the QC pools included within the seven CHEAR studies. The laboratory-specific LOD values were similar to or lower than the CDC’s published NHANES detection limits (Tables 3 and ​and4).4). Nevertheless, LOD values varied among the CHEAR laboratories. In some cases, LOD values varied between projects/studies performed, even within a single laboratory, which resulted in multiple LOD values being reported from the same laboratory. Harmonization of LOD values across the laboratories can be a daunting task, if not impossible, as laboratories use different models of analytical instrumentation and the laboratories differ in approaches to calculate LOD values. Presence of background levels of contamination and sample volume available for analysis can affect LOD values. These parameters likely contributed to the differences in LOD values reported between projects/studies analyzed in a single laboratory. The differences in LOD values among laboratories may pose challenges in combinability of datasets across studies, especially for those analytes that are close to or below the LOD values. Nevertheless, it should be noted that most of the analytes were found above the LOD values in the two urine QC pool samples (Tables 3 and ​and4).4). The measured concentrations of environmental phenols and phthalate metabolites in urine QC pools were similar to the values typically found for the US general population (https://www.cdc.gov/exposurereport/pdf/FourthReport_UpdatedTables_Volume1_Jan2019-508.pdf). However, we did not find marked differences in the concentrations of phthalate metabolites and environmental phenols between urine QC Pools A and B, even though our original intent was to create pools with two different concentrations (e.g., high and low). The urine Pool B was specifically blended with smoker’s urine at a 10% level (v/v), to attain measurable concentrations of tobacco smoke biomarkers and VOCs in that pool. With extensive analyses performed on these QC pool samples among the five CHEAR laboratories, it is possible to thoroughly characterize the materials for use in inter-laboratory studies as reference materials. A subset of data from the CHEAR QC pools have been used in formulating statistical approaches for pooling study data from different batches, studies and laboratories (Mazzella et al., 2021).

The QC pool data were combined and grouped by the pool (i.e., A or B), lab ID and analyte to compare results across various participating laboratories (Tables S3 and S4). Within-laboratory variability was calculated as %CV for representative analytes (for phthalates: monobenzyl phthalate [mBzP], mEOHP and monoisobutyl phthalate [mIBP]; for environmental phenols: bisphenol A [BPA], methyl paraben [MEPB] and triclosan [TCS]) within each chemical class. The three representative analytes from each chemical class were chosen based on their environmental significance, and to illustrate a range of low, medium and high concentrations typically found in human urine. The values for within-laboratory CV are presented in Figures 1 (phthalates: Pool A and B) and 2 (environmental phenols: Pool A and B). It is worth noting that the number of QC pools analyzed and projects completed by each of the three laboratories examined here varied. The CV can be influenced by the sample size and the concentration of analytes. In general, the %CV statistic increases non-linearly as analyte concentrations approach the LOD. The CV values for example target analytes varied between 6 and 23%, in general, across the three laboratories. Nevertheless, CV values for the two QC pools were similar among the three laboratories, which indicates that this level of variance is typical among biomonitoring laboratories that perform trace organics analysis. The U.S. CDC reported CV values in the range of 7–24% associated with the analysis of NHANES urine QC pools for environmental phenols and phthalate metabolites (https://www.cdc.gov/nchs/data/nhanes/nhanes_09_10/EPH_F_met_phenols_parabens.pdf). We also calculated an overall (total) CV, by combining all QC pool data from the four laboratories for seven studies conducted over a 6-month period. The overall CV was between 10 and 33% for environmental phenols, and between 16 and 22% for phthalate metabolites (Figures 1 and ​and2).2). A CV value of <30% was considered ‘acceptable’ for analytical methods assessed by the ISO (Taniyasu et al., 2013). Within the European biomonitoring program, inter-laboratory studies on the measurement of five core urinary phthalate metabolites among seven qualified laboratories showed CV values of between 19 and 45% (Schindler et al., 2014). The CV values for urinary BPA measurements in inter-laboratory studies performed among ‘qualified’ European biomonitoring laboratories were in the range of 14–21% (Schindler et al., 2014). Whereas the measured CV values within and between the CHEAR laboratories are more precise than those reported in various other laboratories performing trace organic biomonitoring, the sources of variability can be purity of analytical standards, sample preparation steps, and calibration of instruments. Relatively higher imprecision encountered for phthalate monoesters can be due to external contamination, lack of chromatographic separation of the structural isomers (e.g., MnBP/MiBP and MEHP/MnOP) and insufficient purity of some batches of commercially available monoester standards (Langlois et al., 2012). Nevertheless, it should be noted that the measured analytical variances are lower than the biological variances (measured as CV%) encountered in biomonitoring studies (Aylward et al., 2012).

Although urine QC pool samples were not characterized for stability and hom*ogeneity at the time of distribution to laboratories, they proved valuable in efforts to harmonize the data generated from multiple laboratories for the assessment of between-lab variances (see Mazzella et al., 2021). The two urine QC pools did not vary in concentrations of target analytes significantly, due to the fact that these pools were not characterized adequately prior to use. Besides characterizing analytical uncertainty from the analysis of QC pools, stability of the analytes and hom*ogeneity of the material should be evaluated.

3.3. NIST Standard Reference Materials

The NIST SRMs 3672 and 3673 were periodically analyzed by three participating CHEAR laboratories for up to 26 environmental phenols and 14 phthalate metabolites. Data were collected across five studies completed by these CHEAR laboratories from February 2018 to January 2019. In total, over 1,800 concentration values greater than the LOD were reported for these SRMs. The data were grouped by NIST SRM, lab ID and analyte in order to compare results from each lab, as well as results from all laboratories combined to the NIST certified value. The SRMs have been certified by NIST (based on the highest confidence) for only a selected number of analytes within a chemical class. For instance, SRMs 3672 and 3673 were certified only for 10 hydroxylated PAHs. Reference values, which are assigned by NIST based on more limited definition of confidence, are available for 11 phthalate metabolites and 8 environmental phenols. The overall recoveries of environmental phenols and phthalate metabolites analyzed in NIST SRMs across three laboratories and five studies ranged 97±5.2% and 95±11%, respectively. We selected three representative analytes from each chemical class and the results are shown in Table 5. Although the number of SRMs analyzed by each of the three laboratories varied (as did the number of projects completed by each laboratory), the degree of accuracy (defined as a % recovery) ranged from 79 to 103% for SRM 3672 and from 79 to 112% for SRM 3673 (Table 5). The analytical imprecision (measured as CV%) calculated from repeated analysis of SRMs for environmental phenols and phthalate metabolites by the three CHEAR laboratories over several months was <10% for over 90% of the dataset (data not shown). The results of SRM analysis among the CHEAR laboratories demonstrate acceptable levels of accuracy and precision for the specific target analytes measured.

3.4. Blinded duplicates

The CHEAR project/study investigators were given the opportunity to insert blinded duplicate samples as an additional measure of QC in the analysis of study samples. The duplicate samples were blinded to the CHEAR laboratories. Following analysis, results were submitted to the CHEAR Data Center, and RPD and ICC were calculated for each valid duplicate pair (Table 6). The RPD values calculated for phthalate metabolites and environmental phenols were generally <25%, but values as high as 63% were found in a few cases. In general, analyte concentrations that were at or close to the LOD showed higher RPD values given the larger relative analytical uncertainty at these low levels. Similarly, the ICC values were mostly greater than 0.90 (Table 6), which indicated excellent reliability of the data (Koo and Li, 2016), except for a few compounds that had values as low as 0.68 (MMP). Although an ICC value between 0.5 and 0.75 indicates moderate reliability, there is a background level of contamination in the analysis of MMP, which contributed to increased unreliability in the measurements. Fifteen duplicate samples from a single CHEAR study were randomly assigned as sample 1 or sample 2 and a plot of concentration in sample 1 on the x-axis and sample 2 on the y-axis was used to inspect agreement between measurements (Figure S1). Excellent correlations (R²>0.90) were found between concentrations measured in blinded duplicate samples (Figure S1). Furthermore, the difference between the concentrations of the two samples was plotted against the average concentration of the two samples (Bland-Altman plot). Majority of measurements fell within the limits of agreement (dashed red lines), which was calculated to be ± 1.96 standard deviations from the mean difference (Figure 3). Only duplicate pairs of elevated concentrations, higher than those typically found in human urine, fell outside the limits of agreement. The mean difference (solid blue line) was close to 0 for all analytes except for MEPB, which suggests that there was a slight bias between measurements for this compound, found at extremely high concentrations. Omitting these two points resulted in a mean difference of 1.86 which was significantly near zero. Overall, the blinded duplicate sample measurements in different analytical cycles resulted in concentrations within the range of analytical uncertainty and yielded measurements deemed fit for purpose.

1

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Supplementary data

List of target organic analytes measured in the CHEAR program with analyte codes, list of internal quality assurance and control protocols followed in each CHEAR lab hub, and lab-specific QC pool results for phthalates and environmental phenols.

• Aylward LL, Kirman CR, Adgate JL, McKenzie LM and Hays SM (2012). Interpreting variability in population biomonitoring: role of elimination kinetics. J. Exp. Sci. Env. Epi, 22, 398–408. [PubMed] [Google Scholar]
• Balshaw DM, Collman GW, Gray KA and Thompson CL (2017). The Children’s Health Exposure Analysis Resource: enabling research into the environmental influences on children’s health outcomes. Curr. Opin. Pediatr, 29, 385–389. [PMC free article] [PubMed] [Google Scholar]
• CLSI (2012). Clinical and Laboratory Standards Institute (CLSI) Harmonized Terminology Database. Available from: http://www.clsi.org/source/custom/termsall.cfm?Section=Harmonized_Terminology_Database. (AccessedJune 29, 2012)
• European Commission (2006). Quality Control Procedures for Pesticide Residues Analysis, Document No. SANCO/10232/2006
• Galusha AL, Merrill L, Palmer CD, Amarasiriwardena C and Parsons PJ (2021). Measurement harmonization and traceability for trace element analyses across the Children’s Health Exposure Analysis Resource laboratory network. Environ. Res, 10.1016/j.envres.2020.110302 [PMC free article] [PubMed] [CrossRef] [Google Scholar]
• Goen T, Schaller K-H and Drexler H (2012). External quality assessment of human biomonitoring in the range of environmental exposure levels. Int. J. Hyg. Env. Health, 215, 229–232. [PubMed] [Google Scholar]
• Greenberg N (2014). Update on current concepts and meanings in laboratory medicine — standardization, traceability and harmonization. Clin. Chim. Acta, 432, 49–54. doi: 10.1016/j.cca.2013.12.04 [PubMed] [CrossRef] [Google Scholar]
• ISO (2005). ISO/IEC 17025: 2005 General requirements for the competence of testing and calibration laboratories. International Standard Organization, Geneva, Switzerland. [Google Scholar]
• ISO (2017). Guide 35:2017. Reference materials-guidance for characterization and assessment of hom*ogeneity and stability
• King B (1995). Quality in the Analytical Laboratory. In 6th International Symposium on Quality Assurance and TQM for Analytical Laboratories, ed. Parkany MThe Royal Society of Chemistry; pp. 8–18. [Google Scholar]
• Koo TK, & Li MY (2016). A guideline of selecting and reporting intra-class correlation coefficients for reliability research. J. Chiropra. Med, 15, 155–163. [PMC free article] [PubMed] [Google Scholar]
• Langlois É, Leblanc A, Simard Y, Thellen C, 2012. Accuracy investigation of phthalate metabolite standards. J. Anal. Toxicol, 36 (4), 270–279. 10.1093/jat/bks016. [PubMed] [CrossRef] [Google Scholar]
• Mazzella MJ, Barr DB, Kannan K, Amarasiriwardena C, Andra SS and Gennings C (2021). Evaluating Interstudy Variability in Phthalate and Trace Element Analyses within the Children’s Health Exposure Analysis Resource (CHEAR) Using Multivariate Control Charts. J. Exp. Sci. Env Epi(in press). [PMC free article] [PubMed] [Google Scholar]
• OECD (1998). Series on Principles of Good Laboratory Practice and Compliance Monitoring No. 1–11, OECD Environmental Directorate, Paris. http://www.oecd.org/document/63/0,2340,en_2649_34365_2346175_1_1_1_1,00.html (accessed onNovember 20, 2020). [Google Scholar]
• Schindler BK, Esteban M, Koch HM, Castano A, Koslitz S, Canas A, Casteleyn L, Kolossa-Gehring M, Schwedler G, Schoeters G, Hond ED, Sepai O, Exley K, Bloemen L, Horvat M, Knudsen LE, Joas A, Joas R, Biot P, Aerts D, Lopez A, Huetos O, Katsonouri A, Maurer-Chronakis K, Kasparova L, Vrbik K, Rudnai P, Naray M, Guignard C, Fischer ME, Ligocka D, Janasik B, Reis MF, Namorado S, Pop C, Dumitrascu I, Halzlova K, Fabianova E, Mazej D, Tratnik JS, Berglund M, Jonsson B, Lehmann A, Crettaz P, Frederiksen H, Nielsen F, McGrath H, Nesbitt I, De Cremer K, Vanermen G, Koppen G, Wilhelm M, Becker K and Angerer J (2014). The European COPHES/DEMOCOPHES project: towards transnational comparability and reliability of human biomonitoring results. Int. J. Hyg. Environ. Health, 217, 653–661. [PubMed] [Google Scholar]
• Taniyasu S, Kannan K, Wu Q, Kwok K, Yeung LWY, Lam PKS, Chittim B, Kida T, Takasuga T, Tsuchiya Y and Yamash*ta N (2013). Interlaboratory trials for analysis of perfluorocotanesulfonate and perfluorocotanoate in water Samples: performance and recommendations. Anal. Chim. Acta, 770, 111–120. [PubMed] [Google Scholar]
• Udesky JO, Dodson RE, Perovich LJ, and Rudel RA (2019). Wrangling environmental exposure data: guidance for getting the best information from your laboratory measurements. Environmental health : a global access science source, 18(1), 99. 10.1186/s12940-019-0537-8. [PMC free article] [PubMed] [CrossRef] [Google Scholar]
• Vesper HW and Thienpont LM (2009). Traceability in laboratory medicine. Clin. Chem, 55, 1067–1075. [PubMed] [Google Scholar]
• Wright RO, Teitelbaum S, Thompson C, Balshaw D and CHEAR Network. (2018). The child health exposure analysis resource as a vehicle to measure environment in the environmental influences on child health outcomes program. Curr. Opin. Pediatr, 30, 285–291. [PMC free article] [PubMed] [Google Scholar]