International Journal of Hygiene and Environmental Health 234 (2021) 113711 Available online 10 March 20211438-4639/© 2021 The Author(s). Published by Elsevier GmbH. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Interlaboratory comparison investigations (ICI) and external quality assurance schemes (EQUAS) for cadmium in urine and blood: Results from the HBM4EU project Stefanie Nübler a, Marta Esteban Lopez b, Argelia Casta~no b, Hans Mol c, Moritz Schafer a, Karin Haji-Abbas-Zarrabi a, Daniel Bury d, Holger M. Koch d, Vincent Vaccher e, Jean-Philippe Antignac e, Darina Dvorakova f, Jana Hajslova f, Cathrine Thomsen g, Katrin Vorkamp h, Thomas Goen a,* a Institute and Outpatient Clinic of Occupational, Social and Environmental Medicine, Friedrich-Alexander Universitat Erlangen-Nürnberg, Henkestraße 9-11, 91054, Erlangen, Germany b National Center for Environmental Health, Instituto de Salud Carlos III, Ctra. Majadahonda a Pozuelo km2,2, 28220, Madrid, Spain c Wageningen Food Safety Research, part of Wageningen University and Research, Akkermaalsbos 2, 6708, WB, Wageningen, the Netherlands d Institute for Prevention and Occupational Medicine of the German Social Accident Insurance, Institute of the Ruhr-University Bochum (IPA), Bürkle-de-la-Camp-Platz 1, 44789, Bochum, Germany e Oniris, INRAE, UMR 1329 Laboratoire d’Etude des Residus et Contaminants dans les Aliments (LABERCA), F-44307 Nantes, France f University of Chemistry and Technology Prague, Department of Food Analysis and Nutrition, Technicka 5, 16000, Prague, Czech Republic g Environmental Exposure and Epidemiology, Norwegian Institute of Public Health, Oslo, Norway h Aarhus University, Department of Environmental Science, Frederiksborgvej 399, 4000, Roskilde, Denmark A R T I C L E I N F O Keywords: Cadmium Human biomonitoring (HBM) HBM4EU Interlaboratory comparison investigation (ICI) External quality assurance scheme (EQUAS) Inductively-coupled plasma mass spectrometry (ICP-MS) Atomic absorption spectrometry (AAS) A B S T R A C T Human biomonitoring (HBM) of cadmium is essential to assess and prevent toxic exposure. Generally, low cadmium levels in urine and blood of the general population place particularly high demands on quality assurance and control measures (QA/QC) for cadmium determination. One of the aims of the HBM4EU project is to harmonize and advance HBM in Europe. Cadmium is one of the chemicals selected as a priority substance for HBM implementation in the 30 European countries under HBM4EU. For this purpose, analytical comparability and accuracy of the analytical laboratories of participating countries was investigated in a QA/QC programme comprising interlaboratory comparison investigations (ICI) and external quality assurance schemes (EQUAS). This paper presents the evaluation process and discusses the results of four ICI/EQUAS rounds for the deter- mination of cadmium in urine and blood. The majority of the 43 participating laboratories achieved satisfactory results, although low limits of quantification were required to quantify Cd concentrations at general population exposure levels. The relative standard deviation of the participants’ results obtained from all ICI and EQUAS runs ranged from 8 to 36% for cadmium in urine and 8–28% for cadmium in blood. Applying inductively-coupled plasma mass spectrometry (ICP-MS), using an internal standard, and eliminating molybdenum oxide in- terferences was favourable for the accurate determination of cadmium in urine and blood. Furthermore, the analysis of cadmium in urine was found to have a critical point at approximately 0.05 μg/l, below which vari- ability increased and laboratory proficiency decreased. This QA/QC programme succeeded in establishing a network of laboratories with high analytical comparability and accuracy for the analysis of cadmium across 20 European countries. Abbreviations: Interlaboratory Comparison Investigation, (ICI); External Quality Assurance Scheme, (EQUAS); control material, (CM); consensus value, (C); assigned value, (A); relative standard deviation, (RSD); robust RSD for the CM of each round, (study RSDR); RSD of the mean values from the expert laboratories, (RSDmean-of-means); geometric mean, (GM). * Corresponding author. E-mail address: thomas.goeen@fau.de (T. Goen). Contents lists available at ScienceDirect International Journal of Hygiene and Environmental Health journal homepage: www.elsevier.com/locate/ijheh https://doi.org/10.1016/j.ijheh.2021.113711 Received 13 November 2020; Received in revised form 23 January 2021; Accepted 4 February 2021 International Journal of Hygiene and Environmental Health 234 (2021) 113711 2 1. Introduction Cadmium is a heavy metal which has been used in many industrial products and processes, such as in the production of pigments and more recently in the manufacture of cadmium telluride solar panels (Hetherington et al., 2008; IARC, 2012; Nordberg et al., 2015). Indus- trial emissions, massive use of fertilizers, leaching processes and inad- equate recycling strategies, but also geological sources have led to a ubiquitous presence of cadmium in the environment (Thornton, 1992; Is¸ikli et al., 2006; Akram et al., 2019; Hou et al., 2019). Dietary intake and smoking are the main determinants of cadmium exposure of the general population in industrialised countries (Mezynska and Brzoska, 2018; EFSA, 2009), whereas in developing countries other additional relevant exposure routes exist, for example from electronic waste recy- cling (Motawei and Gouda, 2016; Kim et al., 2019; Adam et al., 2021). Cadmium can be stored and accumulated in various organs, especially in the liver and kidneys (Jarup and Akesson, 2009). Biochemical, morphological and functional disorders of renal function are the pri- mary toxic effects of chronic cadmium exposure (Jarup, 2003; Satarug et al., 2010). In addition, cadmium and cadmium compounds are clas- sified as carcinogenic to humans (IARC, 2012). Due to the low but prevalent exposure of the general population and the toxic effects following chronic low dose exposure, the assessment and prevention of cadmium exposure is a major public health issue (Jarup and Akesson, 2009; Satarug et al., 2003). For human biomonitoring (HBM) of cadmium exposure, primarily the determination of cadmium in urine, but also of cadmium in blood is applied (Vacchi-Suzzi et al., 2016; Klotz et al., 2013; Fransson et al., 2014; Aoki et al., 2017; Garner et al., 2017). The available HBM data have revealed generally low cadmium levels in both matrices in the general population of industrialised countries (Ruiz et al., 2010; Ber- glund et al., 2015; Bonberg et al., 2017; Nisse et al., 2017; Sar- avanabhavan et al., 2017), requiring limits of quantification (LOQ) of 0.1 μg/l or below. This LOQ was met in almost all recent HBM studies by using inductively-coupled plasma mass spectrometry (ICP-MS). How- ever, this technique involves the challenge of controlling the impact of interfering element clusters, particularly of molybdenum oxide, which is generated in the ICP from background molybdenum content (Jarrett et al., 2008; Akerstrom et al., 2013; Schindler et al., 2014; Ca~nas et al., 2013). Thus, a high level of quality assurance is crucial for the deter- mination of cadmium in biological matrices of the general population. HBM4EU is a European project which represents a joint effort of 30 countries and European Commission authorities, co-funded under Ho- rizon 2020 (https://www.hbm4eu.eu). The main aim of this initiative is the harmonization and advancement of HBM in Europe. HBM4EU tar- gets the exposure of EU citizens to a variety of chemicals and their possible health effects to support policy making (Ganzleben et al., 2017). Cadmium was included in the first priority substance list of HBM4EU and in the first joint HBM studies of the project (Louro et al., 2019; Vorkamp et al., 2021). One of the objectives within the HBM4EU project is the establish- ment of a network of analytical laboratories across Europe (Esteban Lopez et al., 2021) for the HBM of environmental pollutants, generating high-quality and comparable HBM data for the prioritized substances. Thus, HBM4EU has implemented a quality assurance/quality control (QA/QC) scheme to verify analytical comparability between candidate laboratories for the analysis of samples within HBM4EU. An essential component of this QA/QC scheme is the design and implementation of interlaboratory comparison investigations (ICI) and external quality assurance schemes (EQUAS). The ICIs included candidate laboratories from the HBM4EU consortium and investigated the results with a view to comparability between these laboratories, whereas the EQUAS involved additional external expert laboratories that have experience with HBM population studies in other regions of the world and that applied comprehensively validated analytical procedures (Esteban Lopez et al., 2021). This paper presents the ICI/EQUAS programme for cadmium devel- oped in HBM4EU, including the evaluation process, difficulties encountered and the results obtained. 2. Materials and methods 2.1. Design of the HBM4EU ICI/EQUAS programme for cadmium in urine and blood In the QA/QC programme, four rounds of tailor-made ICI and EQUAS exercises were conducted from February 2018 to November 2019 to assess the proficiency of laboratories for cadmium in urine (Cd (U)) and cadmium in blood (Cd (B)). The organisational processes and conditions of ICI and EQUAS exercises for all substance groups in the HBM4EU project are described in detail in Esteban Lopez et al. (2021). Successful participation in the ICI/EQUAS for Cd (U) and Cd (B) was mandatory for laboratories to analyse the respective HBM4EU project samples. Candidate laboratories were requested to apply the same pro- cedure in the ICI/EQUAS as they would use for analysis of samples in the frame of the HBM4EU project. A total of four ICI/EQUAS rounds for Cd (U) and Cd (B) were organized. The first round was conducted as an ICI and the following three rounds were conducted as EQUAS to be consistent with the overall HBM4EU QA/QC programme (Esteban Lopez et al., 2021). Each round comprised the analysis of control materials (CMs) with two concentrations of Cd (U) and Cd (B), respectively. The results were reported to the laboratories before the next round. After the 1st round, a web conference was held for the participants to solve possible difficulties and improve future results. A web conference was also offered after each of the following rounds, but was not deemed necessary by the participants as no major difficulties had been encountered. 2.2. Invitation of participants The process for selecting the candidate laboratories that participated in the Cd ICI/EQUAS has been described elsewhere (Esteban Lopez et al., 2021; short description in the Supplemental Material). In brief, two calls to identify candidate laboratories to perform Cd analysis in HBM4EU were carried out, resulting in a list of 38 candidate laboratories from 22 countries after the first call. This number increased to 58 laboratories from 25 countries after the second call. All laboratories that had previously registered as candidate labora- tories for the analysis of cadmium in HBM4EU samples were invited to the ICI/EQUAS programme for Cd (U) and Cd (B). Candidate labora- tories could participate in the ICI/EQUAS for either Cd (U) or Cd (B) or both. The 38 candidate laboratories established after the first call were invited to participate in the 1st round of the programme. The partici- pants were asked to report LOQs of the analysis and the details of the applied methods in addition to the measured concentrations. The candidate laboratories were also informed that LOQs of 0.05 μg/l for Cd (U) and 0.15 μg/l for Cd (B) were advisable for successful participation. The setting of these LOQ target values was aligned according to the existing HBM data on cadmium in population studies (Casta~no et al., 2012; Jarup and Åkesson, 2009). After the 1st round, the revised candidate list was used to invite 58 laboratories to participate in the 2nd, 3rd and 4th round for Cd (U) and/ or Cd (B). 2.3. Selection of expert laboratories For the rounds organized as EQUAS (2nd, 3rd and 4th), six expert laboratories for Cd (U) and Cd (B) were selected by the HBM4EU Quality Assurance Unit (QAU). Experts were laboratories with experience in the analysis of Cd (U) and Cd (B), having documented their expertise in peer-reviewed publications. In addition, the following selection criteria were considered, although none of them was mandatory: number of S. Nübler et al. International Journal of Hygiene and Environmental Health 234 (2021) 113711 3 years of experience in the analysis of Cd (U) and Cd (B), application of highly sensitive and selective analytical techniques with sufficiently low limit of detection (LOD) and LOQ, application of isotope-labelled stan- dards for quantification, availability of in-house validation reports, data on on-going intra-laboratory performance (e.g. control charts), ISO 17025 accreditation for the biomarker of interest and success rate in ICI/ EQUAS or comparative results in HBM studies. For Cd (U) and Cd (B), two expert laboratories were from outside Europe. Four expert labora- tories were from Europe and three of them also participated as candi- dates in the ICI/EQUAS. 2.4. Preparation and testing of control materials The control materials were freshly prepared and tested for homo- geneity before each round of the ICI/EQUAS programme. The native control material consisted of human urine (Cd (U)native) or bovine blood (Fiebig-Nahrstofftechnik, Idstein-Niederauroff, Germany) in EDTA so- lution (Cd (B)native), both with the addition of sodium azide. For the animal materials, health conditions were certified. Stock solutions (Cadmium ICP standard, Cd(NO3)2 in HNO3 2–3%, 1000 mg/l, Merck) were diluted to two different concentrations to obtain the spiking so- lutions. The addition of these spiking solutions to Cd (U)native and Cd (B)native yielded the target concentrations for the CM (Cd (U)low, Cd (U)high, Cd (B)low, Cd (B)high) (Suppl. Table. 1). Five millilitres of the CMs for Cd (U) were filled into tubes with caps (82  13 mm, polypropylene, Sarstedt). Three millilitres of the CMs for Cd (B) were filled into tubes with caps (57  15.3 mm, polypropylene, Sarstedt). Previous investigations did not show any Cd contamination of the tubes used in the programme. CMs were prepared for each round, stored at  18 C until shipment and then shipped under ambient conditions (Suppl. Fig. 1). Details of the analytical method for the determination of homoge- neity and stability of Cd (U) and Cd (B) are given in Table 1. A more detailed method description can be found in the Supplementary mate- rial. For the determination of homogeneity, ten randomly selected tubes of each CM of each round were taken from storage, thawed, re- homogenised by vortex shaking and simultaneously analysed in duplicate. Homogeneity testing: Homogeneity was evaluated according to ISO 13528:2015 (Fearn and Thompson, 2001; Thompson, 2000) using ICP-MS (see Supplementary material). Stability testing: The stability of the CM was tested in accordance with ISO 13528 (Statistical methods for use in proficiency testing by inter- laboratory comparison, 2015) and the International Harmonised Pro- tocol for the Proficiency Testing of Analytical Laboratories (Thompson et al., 2006). For stability assessment, samples were stored under con- ditions representative of storage at the participants’ laboratories ( 18 C) and at 80 C (considered the maximum stability). Stability was determined by simultaneous ICP-MS analysis of six randomly selected samples from each concentration and after storage at both 18 C and 80 C for a time interval covering the time between shipment and the deadline for result submission. Stability was assessed by comparing the means of the six samples at 18 C and 80 C using the T-test. 2.5. Distribution of control materials The control materials were dispatched to the participants under ambient conditions (Suppl. Fig. 1). Each participant received samples for Cd (U)low, Cd (U)high and/or Cd (B)low, Cd (B)high, according to their registration. In the 1st ICI/EQUAS round, three samples of Cdlow and three samples of Cdhigh were sent to the participants. Additionally, the participants received three samples of Cdnative, with the purpose of determining potential background levels. As no background was found, no further Cdnative samples were sent to the participants in subsequent rounds. From the 3rd round onwards, the participants received only one sample of each concentration (Cdlow, Cdhigh) in order to mimic best the real analysis situation of the HBM4EU project samples. At the time of shipment, a letter with instructions on sample handling, a sample receipt form, a result submission form and a method information form were sent to the participants. Participants were asked to perform a single analysis of each sample using the same procedure as used for analysis of samples in the frame of HMB4EU and to submit their results within four weeks after sample shipment. In the 2nd, 3rd and 4th round, the selected expert laboratories received three samples of each CM (Cdlow and Cdhigh) and were asked to provide a single or duplicate analysis so that they should submit at least six results per material (Cdlow and Cdhigh, both for Cd(U) and Cd(B)). 2.6. Assessment of laboratory performance In brief, for an ICI, a minimum of seven quantitative results from participating candidate laboratories was required for regular evaluation. For Cd (U)low, Cd (U)high, Cd (B)low and Cd (B)high, the following values were calculated using robust statistics so that outliers were not excluded, but only had a minor impact on the performance parameters (Thompson et al., 2006; Analytical Methods Committee, 1989a, b; ISO 13528:2015): robust mean of the participants’ results taken as consensus value (C), uncertainty of the consensus value (uICI), robust ICI standard deviation of the consensus value (σICI) and the Z-scores for each participant. The uncertainty of the consensus value was calculated as follows: uICI ˆ 1:25 σICI  n p (1) with: n ˆ number of results used for calculation of the consensus value with n  7. In the EQUAS, the evaluation of the candidate results was based on the data generated by a minimum of three and a maximum of six designated expert laboratories. The mean-of-means of the individual expert laboratories was used as the assigned value. The uncertainty (uEQUAS) was defined as the relative standard deviation of the expert means (RSDmean-of-means) divided by the square root of the number of expert laboratories: uEQUAS ˆ RSDmean of means  n p (2) The mean-of-means was considered suitable for use as assigned value (A) in EQUAS studies if uEQUAS did not exceed a value of 17.5% derived from the following equation: uEQUAS  0:7 σT (3) Table 1 Analytical method parameters for determination of homogeneity and stability of Cd (U), Cd (B) and Mo background levels. Quantitated ion and matrix Instrument Reagent gas Sample volume Dilution Internal standard Calibration LOD Cd/Mo (μg/l) LOQ Cd/Mo (μg/l) 114Cd or 98Mo in urine ICP-triple quadrupole-MS Argon 0.4 ml 1:10 Rhodium external, matrix-based, multi-level 0.023/0.040 0.050/0.127 114Cd or 98Mo in blood ICP-triple quadrupole-MS Argon 0.2 ml 1:20 Rhodium external, matrix-based, multi-level 0.040/0.050 0.050/0.270 S. Nübler et al. International Journal of Hygiene and Environmental Health 234 (2021) 113711 4 with σT ˆ a pre-set relative target standard deviation for proficiency of 25% This target relative standard deviation (σT) reflected the maximum variability that was considered acceptable for the candidate results and was also used for the Z-score calculation in the ICI/EQUAS. The value of σT (25%) was set based on expert opinion, taking into account what was technically feasible and realistic in current routine practice. As measure of proficiency, Z-scores were calculated using the consensus value derived from the participants’ results (ICI) or the mean of the expert laboratories (EQUAS) as the assigned value, and the σT of 25% (Equation (4)). In the first round, conducted as an ICI, the value of uICI was negligible for Cd (U)low, Cd (U)high, Cd (B)low and Cd (B)high so that the Z-scores (Z) of the results submitted by the participants (x) were calculated accord- ing to the equation: Z ˆ x C σT*C (4) In the 2nd, 3rd and 4th round, conducted as EQUAS, the Z-scores of the participants’ results were calculated according to: Z ˆ x A σT*A (5) In ICI/EQUAS, Z-scores were classified in three categories:  jZj  2 ⇒ satisfactory  2 < jZj < 3 ⇒ questionable  jZj  3 ⇒ unsatisfactory The results of the participating laboratories were evaluated on an individual biomarker/matrix/concentration basis. If no numerical value for a CM was reported by a participant, the specified LOQ that was reported by the participant was used for the Z- score calculation according to equation (5). These LOQ-Z-scores (LOQ- Z) were not included in the final evaluation, but were only used to assess the laboratory’s performance with respect to its reported LOQ. Statistical analyses were conducted using IBM Corp. Released 2019. IBM SPSS Statistics for Windows, Version 26.0. Armonk, NY: IBM Corp. 3. Results and discussion 3.1. Spiking concentrations, homogeneity and stability testing Spiking concentrations in the CM (Cd (U)low, Cd (U)high, Cd (B)low, Cd (B)high), as shown in Suppl. Table 1, were selected in accordance with the expected exposure levels of the general population, which was the target population in HBM4EU aligned studies. DEMOCOPHES, a previ- ous pan-European HBM project in mother-child pairs, used urinary CM with ranges of 0.2–1.0 μg/l for Cd on the basis of data from the German Environmental Survey 1998 and the German Environmental Survey for Children (Schindler et al., 2014). A more recent study that analysed Cd (U) in the general population of northern France reported a geometric mean (GM) of 0.39 μg/l and 0.37 μg/l Cd (U) for women and men, respectively (Nisse et al., 2017). For Cd (B), low levels starting from 0.02 μg/l, high levels up to 4.4 μg/l and mean values ranging from 0.3 μg/l to 1.53 μg/l have been documented for smoking and non-smoking men and women in the general population of several European countries (Mezynska and Brzoska, 2018). Nisse et al. (2017) reported a GM of 0.375 μg/l for Cd (B). The spiking concentrations chosen for the ICI/E- QUAS of 0.000 (native urine unspiked) to 0.350 μg/l Cd (U) and 0.120–0.720 μg/l Cd (B) were therefore considered adequate to test the accuracy of the participating laboratories for quantitative de- terminations within the environmental exposure range. Homogeneity and stability testing of the CMs was conducted sepa- rately for each of the four ICI/EQUAS rounds by the same laboratory (IPASUM). The results of the homogeneity testing for Cd (U) and Cd (B) in the four ICI/EQUAS rounds are shown in Table 2. No outliers were detected in any ICI/EQUAS round for Cd (U) and Cd (B), the homoge- neity was adequate and the method was considered suitable. The Mo background levels determined by single analysis for each CM were below 10 μg/l and below 5 μg/l in urine and blood, respectively (Table 2). The results of the stability testing for Cd (U) and Cd (B) in the four ICI/EQUAS rounds are shown in Suppl. Table 2. No statistically signif- icant instability was detected in any ICI/EQUAS round, minor deviations were caused by the day-to-day imprecision of the applied method. 3.2. Participation and range of reported LOQs In the first round of the ICI/EQUAS programme, 21 and 19 labora- tories (55% and 50%) out of 38 candidate laboratories participated (Table 3) for urine and blood, respectively. 58 laboratories were invited to the following ICI/EQUAS rounds. The range of LOQs reported by the participants in the four ICI/ EQUAS rounds is shown in Suppl. Table 6. The recommended LOQs were 0.05 μg/l for Cd (U) and 0.15 μg/l for Cd (B). Two candidates did not provide their LOQs. In the 1st round, 14 candidates met the LOQ re- quirements for Cd (U) and Cd (B), representing 67% and 74% of all reported LOQs, respectively. In the following ICI/EQUAS rounds, the proportion of candidates meeting the required LOQs increased for Cd (U) and remained fairly constant for Cd (B). The expert laboratories also reported their LOQs and met the respective requirements except for one laboratory in the 2nd round for Cd (B) (Suppl. Table 6). 3.3. Establishment of assigned values derived from expert laboratories (EQUAS) Each expert laboratory analysed either three samples of each CM (Cdlow and Cdhigh) in duplicate (round 2) or six samples of each CM in single or duplicate analysis (round 3 and 4). In the 2nd and 3rd round, all six selected expert laboratories submitted results, while in the fourth round one expert (Exp4) was missing. The details of the expert analyses for Cd (U) and Cd (B) are shown in Suppl. Fig. 2 and Suppl. Fig. 3, the corresponding assigned values and uncertainties can be found in Table 4 and Table 5. The RSD of the assigned values derived from the expert laboratories (RSDexpert labs) decreased from the 2nd to the 4th round for all CMs except for Cd (B)high. Overall, the RSDexpert labs for the high CM was lower than for the low CM in each round. The precision of the mean values (2*SD) varied considerably among the expert laboratories from round to round and for the different CMs. 3.4. Method characteristics Details of the methods used by candidates and experts to analyse Cd (U) and/or Cd (B) are shown in Suppl. Table 3 and Suppl. Table 4. All experts and most candidates applied ICP-MS to analyse Cd in urine and blood. The use of atomic absorption spectrometry (AAS) was very Table 2 Results of the homogeneity testing for Cd (n ˆ 10) and Mo (n ˆ 1) background levels in urine (Ulow and Uhigh) and blood materials (Blow and Bhigh) of the four ICI/EQUAS rounds (mean  SD in μg/l). Round 1 Round 2 Round 3 Round 4 Cd (U)low 0.198  0.010 0.063  0.007 0.055  0.007 0.076  0.005 Mo (U)low 9.47 8.61 9.59 7.38 Cd (U)high 0.451  0.013 0.213  0.022 0.156  0.009 0.156  0.008 Mo (U)high 9.51 8.50 9.61 7.31 Cd (B)low 0.238  0.016 0.105  0.013 0.168  0.017 0.174  0.023 Mo (B)low 4.51 1.19 1.29 3.56 Cd (B)high 0.508  0.017 0.749  0.083 0.300  0.025 0.407  0.051 Mo (B)high 4.56 1.18 1.35 3.58 S. Nübler et al. International Journal of Hygiene and Environmental Health 234 (2021) 113711 5 limited among the participants of the ICI/EQUAS and ranged between 5% and 11% over all rounds. For Cd (U), an internal standard and the respective normalisation were used by 71% of all participants in the first ICI/EQUAS round. In the following rounds, the percentage of laboratories using the response normalised to an internal standard increased to 89% in the 4th round. The percentage of candidates using an internal standard without nor- malising the response declined from 10% in the first round to 3%–5% in the following rounds. In the first round, 19% of the participants did not use an internal standard, while this percentage decreased to 5% in the last round. Measures to suppress molybdenum oxide interferences were applied by around half of the participants. More details on the sup- pression measures are given in Suppl. Table 8. For the analysis of Cd (B), around 80% of the participants used an internal standard with normalised response in the first three ICI/EQUAS rounds (Suppl. Table 4). In the last round, the proportion of laboratories that normalised their response to an internal standard was highest, reaching 94%. Molybdenum interferences were eliminated by a variable percentage of candidates across the four ICI/EQUAS rounds. While in the 1st round only 27% of the participants reported the elimination of molybdenum oxides by effective reaction/collision cell application, half of all laboratories applied such a procedure in the last round for Cd (B). Among the six experts, the methods to determine Cd in urine and blood were quite uniform as all of them used ICP-MS and normalised their results to an internal standard (Suppl. Table 3; Suppl. Table 4). The percentage of expert laboratories that eliminated molybdenum oxide interferences was 60% for Cd (U) and 40% for Cd (B) in the 2nd and in the 3rd round. In the last round, one expert could no longer participate so that 50% of the experts applied elimination of molybdenum oxides. 3.5. Assessment of laboratory performance When comparing the overall performance of all participating labo- ratories in the different rounds, various aspects of the test design, such as organisation as ICI or EQUAS, target concentrations of the CMs and applied methods, should be taken into account. One interesting aspect of the EQUAS exercises is the comparison of the assigned values for Cd calculated from the results of five to six selected expert laboratories with the consensus values achieved by the participating candidates in the three EQUAS rounds. The switch from ICI to EQUAS in the 2nd round was not due to problems with the ICI eval- uation, but aimed to harmonize the exercises for all substance groups in the HBM4EU QA/QC programme targeting information on the accuracy (Esteban Lopez et al., 2021). For all CMs of Cd (U) and Cd (B), the dif- ference between the assigned and the subsequently calculated consensus values was within the range of the standard deviation of the assigned value (except for Cd(B) low in 4th round) and of the consensus value (Table 4). This indicates that the different evaluation schemes of ICI and EQUAS generated comparable Z-score results for the participating laboratories. For the appraisal of the participant results, Z-score values were determined based on a pre-set relative target standard deviation for proficiency (σT) of 25%, which was extracted from previous experience for these parameters (Esteban Lopez et al., 2021). The Z-scores of the participants who submitted quantitative results in the ICI/EQUAS rounds are shown in Suppl. Fig. 4. The performance of the participants who did not provide quantitative data but indicated ‘