ENVIRONMENTAL MONITORING AND BIOMONITORING OF HUMAN ARSENIC EXPOSURE

A thesis submitted to The University of Manchester for the degree of

Doctor of Philosophy

in the Faculty of Engineering and Physical Sciences

2016

DANIEL RICHARD SIMON MIDDLETON BSc (Hons)

SCHOOL OF EARTH, ATMOSPHERIC AND ENVIRONMENTAL SCIENCES

1

LIST OF CONTENTS

LIST OF TABLES ...... 6 LIST OF FIGURES ...... 8 LIST OF ABBREVIATIONS ...... 11 ABSTRACT ...... 14 DECLARATION ...... 15 COPYRIGHT STATEMENT ...... 16 DEDICATION ...... 17 ACKNOWLEDGMENTS ...... 17 FUNDING ...... 18 RATIONALE FOR SUBMISSION OF ALTERNATIVE FORMAT ...... 19 RESEARCH CONTEXT AND AUTHORSHIP CONTRIBUTIONS ...... 19 CHAPTER 1: INTRODUCTION ...... 23 1.1. Background and rationale ...... 23 1.1.1. Environmental geochemistry of arsenic ...... 23 1.1.2. Arsenic exposure: pathways and guidance values ...... 24 1.1.3. Arsenic speciation and biotransformation ...... 25 1.1.4. Toxicity and health implications of chronic exposure to arsenic ...... 27 1.1.5. Global arsenic exposure ...... 28 1.1.6. Study area: south west ...... 30 1.1.7. Biomonitoring of arsenic exposure ...... 30 1.2. Work previously conducted ...... 32 1.2.1. Arsenic biomonitoring: exposure assessment and interpretation...... 32 1.2.2. Arsenic biomonitoring: technical considerations ...... 36 1.2.3. Arsenic in : environmental monitoring...... 38 1.2.4. Arsenic in south west England: human exposure biomonitoring ...... 45 1.2.5. Arsenic in south west England: public health implications ...... 47 1.3. Aims and objectives ...... 50 CHAPTER 2: METHODOLOGY ...... 51 2.1. Chapter overview ...... 51 2.2. Ethical approval ...... 51 2.3. Volunteer recruitment...... 52 2.3.1. Sampling design ...... 52 2.3.2. Recruitment logistics...... 54

2

2.3.3. Response rates ...... 55 2.3.4. Invitation letter, FAQ leaflet and GP letter ...... 55 2.4. Field work and sample collection ...... 65 2.4.1. Sample collection guide, consent and withdrawal forms ...... 65 2.4.2. Fieldwork ...... 73 2.4.3. Urine, toenail and hair collection...... 75 2.4.4. Drinking water collection ...... 75 2.4.5. Residential soil collection ...... 76 2.4.6. Indoor dust collection ...... 76 2.4.7. Rice collection...... 76 2.4.8. Exposure assessment questionnaire...... 77 2.5. Chemical analysis ...... 87 2.5.1. Storage considerations ...... 87 2.5.2. Analytical measurements and techniques ...... 88 2.5.3. Quality assurance and quality control ...... 90 2.6. Data management and statistical analysis ...... 90 2.6.1. Software ...... 90 2.6.2. Safe handling ...... 90 2.6.3. Data preparation: reduction and censoring ...... 91 2.6.4. Exploratory data analysis ...... 91 2.7. Reporting of results to volunteers ...... 91 CHAPTER 3 ...... 92 CHAPTER 4 ...... 93 CHAPTER 5 ...... 94 CHAPTER 6 ...... 95 CHAPTER 7: SUMMARY AND CONCLUSIONS ...... 96 7.1. Key findings and further work ...... 96 7.1.1. Arsenic exposure from private water supplies ...... 96 7.1.2. The performance of urinary hydration adjustments ...... 97 7.1.3. Prolonged arsenic exposure from private water supplies ...... 99 7.1.4. Toenails and hair as viable biomarkers of arsenic exposure ...... 99 7.1.5. Potential arsenic exposure from residential soil and dust ...... 100 7.1.6. Ongoing analysis ...... 101 7.2. Conclusions ...... 102 REFERENCES ...... 103

3

APPENDIX A: ETHICAL APPROVAL LETTER ...... 132 APPENDIX B: RESULTS LETTERS ...... 135 APPENDIX C: LIST OF CONFERENCE PRESENTATIONS AND OUTREACH ACTIVITIES ...... 152 APPENDIX D: EXTENDED ABSTRACT ...... 155

4

Final word count: 75,490

5

LIST OF TABLES Page numbering of individual publication chapters is not in sequence with the main thesis. For tables in these chapters, both thesis page numbers that mark the start of the chapters and the chapter-specific page numbers are provided.

Chapter 1 Thesis p. (publication p.) Table 1 35 Published reference values for total arsenic and toxicologically relevant arsenic species in urine from nationally representative population-based surveys and smaller studies. Chapter 2 Table 2 53 An example of the data layout using the adopted coding system. Table 3 73 Study group characteristics and details of environmental and biological samples collected during the field work campaign. Table 4 87 Storage considerations and measures employed for certain environmental and biological samples. Table 5 89 A summary of the analytical measurements, techniques and instrumentation for all environmental and biological samples collected. Chapter 3 Table 1 92 (4) Descriptive statistics for drinking water and urinary arsenic concentrations. Table 2 92 (5) Correlation analysis of exposure and outcome variables for all volunteers. Table 3 92 (6) Correlation analysis of exposure and outcome variables for single volunteers per household. Table S1 92 (12) Households categorised by ONS rural-urban classification. Chapter 4 Table 1 93 (6) Suggested criteria for assessing performance of urinary biomonitoring adjustment methods. Table 2 93 (13) Demographic characteristics and unadjusted analyte geometric means and ranges for training and testing datasets. Table 3 93 (15) Araki’s b values derived for lead, cadmium, arsenic and iodine in the present study (NHANES 2009-2012 training dataset) compared with previously reported literature values.

6

Table 4 93 (18) Geometric means and ranges of urinary analytes following adjustment by various methods (testing dataset). Table 5 93 (19) Pearson correlations for performance Criterion A across the range of adjustment methods investigated for NHANES 2009-2012 (testing dataset). Table 6 93 (20) Pearson correlations for performance Criterion B across the range of adjustment methods investigated for NHANES 2009-2012 (testing dataset). Table 7 93 (23) Araki’s b values derived for arsenic, iodine, lead and cadmium on specific demographic sub-groups of the present study group (training dataset). Chapter 5 Table 1 94 (567) Household and study group characteristics. Table 2 94 (568) Drinking water arsenic arithmetic mean differences, initial and follow-up arsenic concentration geometric means and results from Pearson correlations between initial and follow-up arsenic concentrations (ln transformed variables) for different private water supply subsets. Table 3 94 (569) Summary statistics for total arsenic in toenail and hair samples for different demographic and behavioural characteristic subsets of the study group. Table 4 94 (571) Pearson correlations for drinking water arsenic and toenail and hair arsenic for different drinking water arsenic concentration ranges. Table 5 94 (572) Predictors of toenail and hair arsenic concentrations on the basis of multiple linear regression models. Chapter 6 Table 1 95 (14) Summary statistics for total arsenic, bioaccessible arsenic and arsenic bioaccessible fraction for various sample groups and normal background concentration domains. Table 2 95 (16) Summary statistics of total arsenic in composite indoor dust samples. Table 3 95 (18) Exceedances of the C4SL generic assessment criteria for total arsenic in residential soils and the normal background concentrations of total arsenic in English soils. Table 4 95 (26) Regression model results for total arsenic concentration in residential soil (mineralised domain) as a function of proximity to arsenic-specific former mining sites.

7

LIST OF FIGURES Page numbering of individual publication chapters is not in sequence with the main thesis. For figures in these chapters, both thesis page numbers that mark the start of the chapters and the chapter-specific page numbers are provided.

Chapter 1 Thesis p. (publication p.) Figure 1 29 Global scenario of population exposure to arsenic above the WHO guidance value of 10 µg L-1 in drinking water. Figure 2 39 Interpolated maps of soil total arsenic concentrations (A) and stream sediment total arsenic concentrations (B) measured in samples collected across England and Wales for the Geochemical Baseline Survey of the Environment. Figure 3 41 Photograph of an arsenic calciner at the former Devon Great Consols mine. Figure 4 41 Photograph of a spoil heap of arsenic-bearing mine tailings at the former Devon Great Consols mine. Figure 5 44 The number of single domestic private water supplies plotted by local authority district in England and Wales. Chapter 2 Figure 6 54 Map showing the distribution of the total number of households in the sampling frame and those recruited to the study. Figure 7 78 Print screen of the exposure assessment questionnaire in Microsoft Access format as it appeared to researchers and volunteers during appointments. Chapter 3 Figure 1 92 (3) Spatial distribution (map) of sampled households. Figure 2 92 (3) Study group age and gender distribution. Figure 3 92 (5) Box and whisker plots of private water supply drinking water and urinary arsenic. Figure 4 92 (5) Unadjusted and adjusted urinary arsenic versus private water supply drinking water arsenic. Figure 5 92 (6) Comparison of urinary arsenic hydration adjustment methods. Figure 6 92 (6) Log-log plot of urinary arsenic versus private water supply drinking water arsenic divided into drinking water exposure levels.

8

Figure 7 92 (9) Arsenic speciation standard chromatogram. Figure S1 92 (12) Rural-urban classification map of the study area and households. Chapter 4 Figure 1 93 (14) Unadjusted urinary lead, cadmium, arsenic, iodine and creatinine plotted against urinary flow rate (NHANES 2009-2012 training data). Multiple spot cadmium measurements (Meharg et al. 2014) from a single volunteer are shown for comparison. Figure 2 93 (16) Sensitivity of Pearson correlations to Araki’s b value for NHANES 2009-2012 training data for lead, cadmium, arsenic and iodine for Criterion A and Criterion B. Figure 3 93 (21) Scatterplots of unadjusted urinary lead concentrations and those adjusted by various methods versus urinary flow rate (Criterion A) or blood lead concentrations (Criterion B).

Figure 4 93 (24) Optimum Criterion A and Criterion B Araki’s b values derived for different age groups for the range of analytes investigated. Chapter 5 Figure 1 94 (566) Map of the study area, shown in the context of the UK (excluding Northern Ireland), and the spatial distribution of sampled households. Figure 2 94 (568) Follow-up drinking water arsenic concentrations plotted against initial counterparts. Figure 3 94 (569) Bar plot of geometric mean arsenic concentrations in toenail samples, initial and final rinse fractions for volunteers with and without observed/reported nail polish. Figure 4 94 (570) Initial rinse fraction arsenic concentrations and final rinse fraction arsenic concentrations plotted against toenail digest arsenic concentrations. Figure 5 94 (570) Significantly positive Pearson correlations between toenail and hair biomarker arsenic concentrations and those measured in drinking water. Figure S1 94 (575) Diagram of washing procedure administered to toenail and hair samples. Figure S2 94 (576) Determination of the minimum mass required for the analysis of toenail and hair samples.

9

Chapter 6 Figure 1 95 (4) Mapped domains used for the derivation of normal background concentrations of arsenic in English soils. Figure 2 95 (20) Histogram of soil arsenic bioaccessible fractions and the 25th and 75th percentiles. Figure 3 95 (21) Total arsenic concentrations for individual residential soils in relation to the C4SL generic assessment criterion and the two derived site-specific assessment criteria (A). Modelled average daily exposures using the three relative bioavailability scenarios (100, 23 and 13 %) in relation to the lowest level of toxicological concern (LLTC) of 0.3 µg kg-1 day-1 (B). Figure 4 95 (24) The spatial distribution of arsenic-specific former mining sites, located from BRITPITS and Dines (1956), in relation to sampled households. Figure S1 95 (38) The spatial distribution of arsenic-specific former mining sites, located from BRITPITS and Dines (1956), in relation to sampled households. Detailed map with individual named sites and arsenic outputs.

10

LIST OF ABBREVIATIONS AAS – atomic absorption spectroscopy AB – arsenobetaine AC – arsenocholine ACGIH - American Conference of Government Industrial Hygienists ADE – average daily exposure ADP – adenosine diphosphate AIC – Akaike’s information criterion AM – arithmetic mean ANOVA – analysis of variance As – arsenic AsIII – arsenous acid (Chapter 4 only)* AsIII –arsenite AsIMM – arsenic: Inorganic and Methylated Metabolites (AsIII + AsV + MA + DMA) ASTM – American Society for Testing Materials AsV – arsenate AsV – arsenic acid (Chapter 4 only)* ATP – adenosine triphosphate ATSDR – Agency for Toxic Substances and Disease Registry (US) Au – gold BAF – bioaccessible fraction BARGE – Bioaccessibility Research Group of Europe BE – biomonitoring equivalent BEI – biological exposure indices BEPOD – biomonitoring equivalent point of departure BGS – British Geological Survey BW – bodyweight C4SL – Category 4 Screening Level CCA – chromated copper arsenate Cd – cadmium CDC –Centers for Disease Control and Prevention (US) CHMS – Canadian Health Measures Survey CI – confidence interval CKD – chronic kidney disease CL:AIRE – Contaminated Land: Applications in Real Environments CLEA – contaminated land exposure assessment (model) COPHES – Consortium to Perform Human Biomonitoring on a European Scale Cre – creatinine CRM – certified reference material Cu – copper DEFRA – Department for Environment, Food and Rural Affairs (UK) DGC – Devon Great Consols DIW – deionised water DMA – dimethylarsinate DMA – dimethylarsonic acid (Chapter 4 only)* DNA – deoxyribonucleic acid DWI – Drinking Water Inspectorate (UK)

* For consistency with CDC/NHANES nomenclature.

11

EA – Environment Agency (England) eGFR – estimated glomerular filtration rate ENNS – Etude Nationale Nutrition Santé (French Nutrition and Health Survey) EPHT – environmental public health tracking ER – excretion rate ERBW – bodyweight adjusted excretion rate FAO – Food and Agricultural Organisation (UN) Fe – iron FMV – first morning void FSA – Food Standards Agency (UK) GAC – generic assessment criteria (soil) G-BASE – Geochemical Baseline Survey of the Environment Ge – germanium GerES – German Environmental Survey GLS – generalised least squares GM – geometric mean GP – general practitioner He – helium HPA – Health Protection Agency HPLC – high performance liquid chromatography HR – hazard ratio I – iodine IARC – International Agency for Research on Cancer (WHO) ICP-AES – inductively coupled plasma atomic emission spectrometry ICP-DRC-MS – inductively coupled plasma dynamic reaction cell mass spectrometry ICP-MS – inductively coupled plasma mass spectrometry IQR – interquartile range JEFCA – Joint FAO/WHO Expert Committee on Food Additives LLTC – lowest level of toxicological concern ln – natural logarithm LOD – limit of detection MA - methylarsonate MARS – microwave assisted reaction system MEC – mobile examination center MMA – monomethylarsonic acid (Chapter 4 only)* NBC – normal background concentration (soil) NCS – National Analysis Centre for Iron and Steel (China) NHANES – (US) National Health and Nutrition Examination Survey NHS – National Health Service (UK) Ni – nickel NIES – National Institute for Environmental Studies (Japan) NIST – National Institute of Standards and Technology (US) NOAEL – no observed adverse effect level NRES – National Research Ethics Service (UK) ONS – Office for National Statistics (UK) P25 – 25th percentile P75 – 75th percentile

* For consistency with CDC/NHANES nomenclature.

12

Pb – lead PCV – prescribed concentration or value (drinking water in England and Wales) PHE – Public Health England PII – personally identifiable information PIXE – particle induced X-ray emission PM10 – particulate matter <10 µm in diameter POD – point of departure PWS – private water supply QC/QA – quality control/quality assurance Ra – radon RBA – relative bioavailable fraction rp – Pearson’s correlation coefficient rs – Spearman’s correlation coefficient RSD – relative standard deviation SG – specific gravity Sn – tin SSAC – site specific assessment criteria (soil) SSAC13 – SSAC derived using 13 % RBA (see SSAC and RBA) SSAC23 – SSAC derived using 23 % RBA (see SSAC and RBA) Te – tellurium U-AsIMM – urinary AsIMM (see AsIMM) UBM – unified bioaccessibility method UBM – unified bioaccessibility method UCL – upper confidence limit UFR – urinary flow rate UFRA – urinary flow rate adjustment optimised to Criterion A UFRB – urinary flow rate adjustment optimised to Criterion B UREC – University of Manchester Research Ethics Committee UV – ultraviolet radiation WHO – World Health Organisation XRF/XRFS – X-ray fluorescence spectrometry

13

The University of Manchester Daniel Richard Simon Middleton BSc (Hons) Doctor of Philosophy ENVIRONMENTAL MONITORING AND BIOMONITORING OF HUMAN ARSENIC EXPOSURE 2016 ABSTRACT This study investigated human exposure to inorganic arsenic (As), a risk factor for cancer and non-cancerous health effects, in Cornwall, UK - a region of elevated environmental As resulting from naturally occurring mineralisation and historical mining. Recent exposures to As from private water supplies (PWS) were detected by measuring As in drinking water samples (n=127) and urine samples (n=207). Exceedances of the WHO 10 As µg L-1 guidance value were measured in drinking waters from 5 % of households. The Spearman correlation calculated for drinking water versus unadjusted total urinary As concentrations was 0.36. Urinary As speciation was used to distinguish between environmental inorganic As exposure and non-toxic dietary sources. Seafood derived urinary arsenobetaine exclusion and osmolality hydration adjustment yielded an improved correlation of 0.62 between drinking water and urinary As concentrations. Urinary hydration adjustment methods were improved and comparatively assessed using data from the US National Health and Nutrition Examination Survey (NHANES). Correlations of urinary concentrations of As, iodine (I), lead (Pb) and cadmium (Cd) against urinary flow rate (UFR) (low correlations desired) and urinary Pb and Cd against respective blood concentrations (high correlations desired) were used as independent performance criteria. Osmolality adjustment and a modified UFR-based adjustment method using empirically derived coefficients (slopes of analyte concentrations as a function of UFR) generally performed better than creatinine, excretion rate and bodyweight-adjusted excretion rate methods. The findings demonstrated the analyte specific nature of adjustment methods, their misuse in the literature and suggested a pathway to a more robust adjustment framework. Prolonged exposure to As from PWS was identified by the stability of 127 drinking water As concentrations measured up to 31 months apart. Drinking water As concentrations were correlated with those measured in toenails (Pearson’s r: 0.53; n=200) and hair (Pearson’s r: 0.38; n=104). The successful elimination of external contamination of toenail samples was indicated by low As concentrations in final- stage rinse solutions (geometric mean contribution: 0.4 %). A positive association between seafood consumption and toenail As and a negative association between home-grown vegetable consumption and hair As was observed when As in drinking water was <1 As µg/L. Elevated As concentrations measured in residential soil (12-992 mg kg-1; n=127) and household dust (3-1079 mg kg-1; n=99), particularly on mineralised geological domains and in the vicinity of former As mining sites, were indicative of additional As exposure routes. Bioaccessibility-adjusted assessment criteria of 190 (13 % bioaccessibility) and 129 (23 % bioaccessibility) As mg kg-1 were derived and 10 and 17 % of residential soils were in exceedance, respectively. The relative importance of different exposure routes in the study region, namely whether As intake from soil and dust is evident in the study population, will form the basis of further work. This will be addressed using multivariate analyses of drinking water, soil and dust in conjunction with urine, toenail and hair As concentrations.

14

DECLARATION

No portion of the work referred to in the thesis has been submitted in support of an application for another degree or qualification of this or any other university or other institute of learning.

15

COPYRIGHT STATEMENT (i) The author of this thesis (including any appendices and/or schedules to this thesis) owns certain copyright or related rights in it (the “Copyright”) and s/he has given The University of Manchester certain rights to use such Copyright, including for administrative purposes. (ii) Copies of this thesis, either in full or in extracts and whether in hard or electronic copy, may be made only in accordance with the Copyright, Designs and Patents Act 1988 (as amended) and regulations issued under it or, where appropriate, in accordance with licensing agreements which the University has from time to time. This page must form part of any such copies made. (iii) The ownership of certain Copyright, patents, designs, trademarks and other intellectual property (the “Intellectual Property”) and any reproductions of copyright works in the thesis, for example graphs and tables (“Reproductions”), which may be described in this thesis, may not be owned by the author and may be owned by third parties. Such Intellectual Property and Reproductions cannot and must not be made available for use without the prior written permission of the owner(s) of the relevant Intellectual Property and/or Reproductions. (iv) Further information on the conditions under which disclosure, publication and commercialisation of this thesis, the Copyright and any Intellectual Property University IP Policy (see http://documents.manchester.ac.uk/display.aspx?DocID=24420), in any relevant Thesis restriction declarations deposited in the University Library, The University Library’s regulations (see http://www.library.manchester.ac.uk/about/regulations/) and in The University’s policy on Presentation of Theses.

16

DEDICATION

This thesis is dedicated to the 215 residents of Cornwall, England who volunteered for the study. Without their generosity, time, willingness, patience and hospitality, this work would not have been possible.

ACKNOWLEDGMENTS First and foremost, I thank my supervisors Prof David Polya and Dr Michael Watts for the tremendous amount of support and faith in my ability that they have shown throughout my PhD. Michael, thank you for all of the encouragement, advice and infectious enthusiasm that has helped to drive this project forward. Thank you for making me feel welcome in a fantastic team at BGS and putting a huge amount of trust and responsibility into me that has undoubtedly shaped my career and given me my independence as a researcher. Dave, thank you for the opportunities, support and good humour that you have provided, not just during my PhD, but as a key figure in my development over the last seven years. Thank you for listening to my ideas, good and bad, and giving me the freedom and encouragement to pursue them. From Public Health England, I thank Dr Tony Fletcher, Rebecca Close, Amy Rimell, Dr Giovanni Leonardi, Dr Mike Studden, Dr Helen Crabbe and Dr Karen Exley for what has been a fantastic opportunity to work with all of you on this project. I thank my good friends and colleagues at BGS. Elliott Hamilton, thanks for making the last three years as enjoyable as they have been and going above and beyond by giving up many weekends to train me and help me generate data of a high standard. Dr Louise Ander, thank you for sharing with me a wealth of experience and training and a memorable field work trip to Cornwall. Dr Andy Marriott and Andrew Dunne, thanks for your commitment and hard work in the field and good spirits in the office. Dr Darren Beriro, Dr Mark Cave, Andrea Mills, Dr Simon Chenery, Dr Murray Lark, Dr Chris Milne, Paul Turner, Cynthia Turner, Amanda Gardner, Dr Charles Gowing, Stephen Lowe, Oliver Humphrey, Josh Coe, Dr Helen Taylor, Bob Lister, Tom Barlow and Keith Adlam (the list could go on), I thank all of you for help and advice and making my time at BGS a memorable one. From the University of Manchester, thanks go to Catherine Davies for training and support when I first started my PhD and to Paul Lythgoe and Alistair Bewsher (for an excellent on screen performance in the biomonitoring video).

17

Last but not least I thank my family. Magdalena Rodycz, for (eventually!) understanding my need to work strange hours and providing me with support and encouragement throughout. To my parents, Kathy and Richard Middleton, thank you for always believing in me and providing me with the patience and backing that has been necessary for me to achieve what I have. I would also like to extend a special thanks to my brother, Nick, for understanding the financial constraints of being a student and having to fill in for me at the bar on numerous occasions over the last seven years or more.

FUNDING This research was funded by the Natural Environment Research Council (NERC) in the form of a University of Manchester/British Geological Survey University Funding Initiative (BUFI) studentship (NERC Contract No. GA/125/017; BUFI Ref: S204.2). The funding contract was signed by the University of Manchester, British Geological Survey and the Health Protection Agency (hence Public Health England). In-kind contributions were provided by all three parties.

18

RATIONALE FOR SUBMISSION OF ALTERNATIVE FORMAT Due to the nature of this research project, a number of outputs were pre-empted for publication by peer-review. The decision was agreed upon, at the start of the PhD programme, to submit a thesis by alternative format. This thesis is built primarily around four manuscripts for publication that, while being standalone, form components of a broader, continuous research effort.

RESEARCH CONTEXT AND AUTHORSHIP CONTRIBUTIONS The work undertaken during this PhD programme forms part of a wider Environmental Public Health Tracking (EPHT) initiative: to investigate human exposure to chemicals and chemical parameters in private water supplies (PWS) in England. Between 2011 and 2013, the British Geological Survey (BGS) was contracted by the Health Protection Agency (HPA) (now part of Public Health England - PHE) to monitor the point-of-source and point-of-use water quality of 515 households with a PWS in Cornwall, south west England. Cornwall’s significance as a high-As area and its diverse geology led to its selection as a study location (Ander et al. 2016).

As a follow-up to the PWS survey, the same cohort of households was invited to participate in a biomonitoring study – a collaboration between the University of Manchester, BGS and PHE in the form of this co-funded PhD studentship. The work presented in this thesis consists of material produced for the studentship. The majority of the material is first-authored by the PhD student and co-authored by members of the study team listed below. Specific contributions to peer-reviewed outputs were also made by other scientists.

Study team (* PhD supervisors):

Daniel Middleton (DRSM) – PhD student

Professor David Polya* (DAP) – University of Manchester project PI Dr Michael Watts* (MJW) – BGS project PI Dr Tony Fletcher (TF) – PHE project PI Dr Giovanni Leonardi* (GSL) – PHE Rebecca Close (RC) – PHE Mike Studden (MS) – PHE Amy Rimell (AR) - PHE Elliott Hamilton (EH) – BGS Dr Louise Ander (ELA) – BGS

19

Dr Helen Crabbe (HC) – PHE Dr Karen Exley (KE) - PHE

The contributions made to individual study elements and the chapters in which they appear are outlined below:

Ethical approval application (Chapter 2):

The ethical approval application discussed in Chapter 2 (2.2) was initially drafted by DRSM. Amendments were made primarily by DAP with contributions from RC, MW, MS, TF, GSL, HC and KE. The document was updated throughout by DRSM. The letter of ethical approval is in Appendix A.

Stakeholder material (Chapter 2):

The collection of material compiled for stakeholder communications was co- authored by the study team. DRSM took the lead role on compiling the sample collection guide (2.4.1) and contributed to the content of all other documents presented which includes the invitation letter (2.3.4), FAQ leaflet (2.3.4), GP letter (2.3.4), consent form (2.4.1), withdrawal form (2.4.1) and results letters for drinking water, biomonitoring and soil samples (Appendix B). The final versions of all these documents were as agreed by all University of Manchester, British Geological Survey and Health Protection Agency/Public Health England through their representatives and DRSM. Public Health England took primary responsibility for communication of public health advice to volunteers.

Scientific papers (Chapters 3-6):

The four scientific papers that form the central chapters of the thesis were all first authored by DRSM. All co-authors contributed to the content of the manuscripts and all manuscripts underwent review by the study team. Detailed contributions for each publication are provided below:

Chapter 3

Middleton, DRS, Watts, MJ, Hamilton, EM, Ander, EL, Close, RM, Exley, KS, Crabbe, H, Leonardi, GS, Fletcher, T and Polya, DA (2016). Urinary arsenic profiles reveal substantial exposures to inorganic arsenic from private drinking water supplies in Cornwall, UK, Scientific Reports. DOI: SREP25656

20

Conceived by MJW, GSL, RC and DAP; and largely managed and executed by DRSM under the supervision of MJW, GSL, TF and DAP. DRSM and EH coordinated field sampling activities and produced the data. LA assisted in the set-up of the project database architecture. RC and HC coordinated public health communication activities. All authors contributed to the design of the study and to writing and/or review of the paper.

Chapter 4

Middleton, DRS, Watts, MJ, Lark, RM, Milne, CJ and Polya, DA (2016). Assessing urinary flow rate, creatinine, osmolality and other hydration adjustment methods for urinary biomonitoring using NHANES arsenic, iodine, lead and cadmium data. Environmental Health. In submission.

Conceived by DRSM, MJW and DAP. DRSM acquired the data and performed statistical analyses. RML coordinated the data management statistical methodologies employed and CJM constructed the R programming scripts to conduct analyses. All authors contributed to and reviewed the manuscript.

Chapter 5

Middleton, DRS, Watts, MJ, Hamilton, EM, Fletcher, T, Leonardi, GS, Close, RM, Exley, KS, Crabbe, H and Polya, DA. (2016). Prolonged exposure to arsenic in UK private water supplies: toenail, hair and drinking water concentrations. Environmental Science: Processes & Impacts. DOI: 10.1039/C6EM00072J

Conceived by MJW, GSL, RC and DAP; and largely managed and executed by DRSM under the supervision of MJW, GSL, TF and DAP. DRSM and EH coordinated field sampling activities and produced the data. RC and HC coordinated public health communication activities with toxicological advice from KE. All authors contributed to the writing and/or review of the manuscript.

Chapter 6

Middleton, DRS, Watts, MJ, Beriro, DJ, Hamilton, EM, Leonardi, GS, Fletcher, T, Close, RM and Polya, DA. Arsenic in residential soil and household dust: human exposure assessment criteria and the influence of historical mining. Manuscript in preparation.

Conceived by MJW, GSL, RC and DAP; and largely managed and executed by DRSM under the supervision of MJW, GSL, TF and DAP. DRSM and EH coordinated field sampling activities and produced the data. DRSM and DJB

21

conducted statistical analyses and prepared the manuscript. DJB performed assessment criteria derivations and average daily exposure calculations. All authors contributed to the writing and/or review of the manuscript.

Additional related outputs (not included in this thesis):

Polya, DA and Middleton, DRS (2016) Arsenic in drinking water: sources & human exposure routes. In Bhattacharya, P, Polya, DA and Jovanovic. D (Eds.) Best Practice Guide for the Control of Arsenic in Drinking Water, IWA Publishing, Chapter 1, ISBN13: 9781843393856. Crabbe, H, Close, R, Rimell, A, Leonardi, G, Watts, MJ, Ander, EL, Hamilton, EM, Middleton, DRS, Smedley, PL, Gregory, M, Robjohns, S, Sepai, O, Studden, M, Polya, DA, and Fletcher, T (2016) Estimating the population exposed to arsenic from groundwater-sourced private drinking water supplies in Cornwall, UK. In Bhattacharya, P. Polya, DA and Jovanovic. D (Eds.) Best Practice Guide for the Control of Arsenic in Drinking Water, IWA Publishing, Chapter A4, ISBN13: 9781843393856. Zia MH, Watts MJ, Niaz A, Middleton DRS and Kim A. Health risk assessment of potentially harmful elements (PHEs) and dietary minerals (DMs) from soils and vegetables irrigated with untreated wastewater, Pakistan. Environmental Geochemistry and Health. Accepted for publication. A list of oral and poster presentations and public engagement activities is presented in Appendix C.

22

CHAPTER 1: INTRODUCTION

1.1. Background and rationale 1.1.1. Environmental geochemistry of arsenic Arsenic is environmentally ubiquitous and present in a range of media including rocks, soil, sediments, water and biota. Arsenic is estimated to rank 47th in crustal abundance (estimates vary and are reported as high as 20th (Mandal and Suzuki 2002) and also as low as 52nd (Hanh et al. 2010). The highest concentrations of As are found among mudstones, shales, slates and some coals (Vaughan 2006). There are upwards of 300 arsenic-bearing minerals known to occur in nature, of which ~60 % are arsenates, ~20 % are sulphides and sulphosalts with the remaining ~10 % being shared among arsenates, arsenides, native elements and metal alloys (Drahota and Filippi 2009). The most common As-bearing mineral, arsenopyrite (FeAsS), is an example of a primary As mineral which forms a group with other ores containing iron (Fe), sulphide (SII-), copper (Cu), nickel (Ni) and lead (Pb) plus others (Mandal and Suzuki 2002), hence the association of As concentrations with mineral processing and mine tailings.

The uptake of As by biota and subsequent cycling of methylation and demethylation gives rise to a variety of inorganic and organic forms in the environment. Sources of environmental arsenic are numerous but can be broadly categorized into two groups of the As biogeochemical cycle: (i) natural processes and (ii) anthropogenic activities (in some cases one exacerbating the other). These inputs are expanded below and are adapted from information provided in the literature (Chilvers and Peterson 1987; Hanh et al. 2010; Smedley and Kinniburgh 2002; WHO 2001). Anthropogenic inputs reflect the various historic and current applications of As in industry and agriculture.

(i) Natural processes:

. Mineralisation – pedogenesis (soil formation) of naturally high-As parent material and dissolution of As-bearing aquifer host material in mineralised areas, producing soils and groundwaters with elevated concentrations, respectively.

23

. Low temperature volatilisation: methylation and volatilization of labile As in soils by microbial action, responsible for the largest atmospheric flux of As. . Volcanic activity: volcanic ash emissions are accountable for the second largest atmospheric flux of As. Acid crater lakes, the lakes that form in volcanic calderas once eruptions have ceased, are also an important source of environmental As and concentrations of as high 279 µg L-1 have been reported in waters from these lakes (Sriwana et al. 1998). . Wind erosion: atmospheric dispersion of As from wind-blown/eroded soils. . Forest fires: arsenic released from burning vegetation during natural forest fires.

(ii) Anthropogenic activities:

. Mining and mineral processing: waste rock and mine tailings from workings of As-bearing ores or other associated metalliferous ores (e.g. tin (Sn) and Cu). . Metal smelting . Fossil fuel combustion . Waste incineration . Application of biocides: arsenic containing herbicides and pesticides. . Wood preservation: chromated copper arsenate (CCA) used to treat timber. . Agricultural feed additives

1.1.2. Arsenic exposure: pathways and guidance values The different sources of As in the environment give rise to several possible pathways of human exposure. Exposure can result from ingestion (e.g. drinking water and food), inhalation (e.g. soil and dust) and dermal absorption (e.g. soil and occupational chemicals) (Environment Agency 2009). The most significant pathway globally, discussed later, is the ingestion of contaminated groundwater used for human drinking water supplies. Important factors that determine the level of exposure are the intake rate of the As source and the bioaccessibility of As in a given matrix. Arsenic bioaccessibility in different media varies from 100 % in drinking water by definition, 36-99 % in rice (Signes‐Pastor et al. 2012), to 10-20 % in soil (Cave et al. 2002).

24

As of the most recent review of scientific evidence from epidemiological studies, conducted in 2010 by the Joint Food and Agricultural Organisation (FAO) of the United Nations and the World Health Organisation (WHO) Expert Committee on Food Additives (JECFA), the lower benchmark dose of a 0.5 % increase in the incidence of lung cancer was determined to be 3 µg/kg bodyweight (bw) per day (WHO 2011). The current WHO provisional drinking water guidance value (WHO 2011) is 10 µg L-1 on the basis of practical achievability i.e. treatment system performance and analytical detection capabilities. The WHO guidance value has been adopted by many countries and forms the basis for the European Union (EU) directive (Directive 98/83/EC) (Council of the European Union 1998) implemented by the UK in the form of the 10 µg L-1 prescribed concentration or value (PCV) (Water Supply (Water Quality) Regulations 2000). The English Category 4 Screening Levels (C4SLs) for As in residential soils (CL:AIRE 2014) were derived to represent the concentration of a As that, under generic exposure parameters, would result in an intake equal to 0.3 µg kg-1 bw day-1. This is the equivalent daily intake resulting from a 2 L consumption of water containing 10 As µg/L - to align with the UK drinking water PCV of 10 As µg L-1. The C4SLs for As in English residential soils are 37 and 40 mg kg-1 for properties with and without home-grown produce, respectively. No UK guidance currently exists for As in foodstuffs. However, EU limits were recently established for As in rice and rice products (EC No 2015/1006) (European Commission 2015) as follows: non-parboiled milled rice: 0.2 mg kg-1; parboiled rice and husked rice: 0.25 mg kg-1; rice waffles, rice wafers, rice crackers and rice cakes: 0.3 mg kg-1; rice destined for the production of food for infants and young children: 0.1 mg kg-1.

1.1.3. Arsenic speciation and biotransformation A plethora of different organic and inorganic As species occur in the environment, those which are considered relevant to human exposure and toxicology are discussed here. The differing mobilities, toxicological and biochemical properties of As species make it essential to distinguish between them in clinical and environmental analysis. Total As determination is, therefore, often insufficient for specific applications (Gong et al. 2002). A Number of excellent reviews of the analytical techniques applied to the determination and speciation of As compounds are available, two fine examples being Francesconi and Kuehnelt (2004) and Tyson (2013). The

25

nomenclature for As speciation abbreviations is variable within the literature. For uniformity within this thesis, the suggestions of Francesconi and Kuehnelt’s 2004 review have been adopted. For the duration of this study the following abbreviations will be used:

. AsIII will refer to arsenite and AsV to arsenate. . MA or MAV will refer to pentavalent methylarsonate instead of MMA or MMAA in reference to the corresponding acid; monomethylarsonic acid. . DMA or DMAV will refer to pentavalent dimethylarsinate. . MAIII and DMAIII will refer to the reduced methylated species; methylarsonite and dimethylarsinite. . AB and AC will refer to arsenobetaine and arsenocholine, respectively.

The abbreviation AsIMM has been coined to refer to inorganic As and methylated metabolites (the sum of AsIII, AsV, MA and DMA) – a routine urinary As biomarker that excludes the contribution of AB and other toxicologically irrelevant arsenicals.

The array of As species present in the environment have differing degrees of toxicity. It is generally accepted that following entry to the body, As undergoes biotransformation via means of a series of oxidation/reduction and methylation reactions, starting with the rapid reduction of inorganic AsV to AsIII, with the end products being the methylated organic metabolites MA and DMA. Both MA and DMA are more readily excreted in the urine, with methylation primarily occurring in the liver (Vahter 2002). The intermediate stages of this pathway are questioned and there is uncertainty regarding the order in which the trivalent and pentavalent organic metabolites are formed. It was proposed that MAV and DMAV are formed prior to the reductive methylation of their trivalent counterparts, but an alternative pathway has been suggested in which the trivalent metabolites MAIII and DMAIII are formed as intermediates (Tseng 2009). While methylation has previously been considered a detoxification process, the formation of trivalent intermediates implies a transitory increase in toxicity, as both MAIII and DMAIII are highly reactive and their presence in tissues may be significant to the carcinogenicity of As (Cohen et al. 2006). Inter- individual variation in the methylation capacity of As species is a factor regulating the toxicity of As and the ratio of different urinary As species has been employed to assess the capacity of different steps in the methylation process (Tseng 2009).

26

1.1.4. Toxicity and health implications of chronic exposure to arsenic Chronic exposure to As is a risk factor of cancers of the lung, skin, bladder, kidney, liver and prostate and is categorised by the International Agency for Research on Cancer (IARC) as a Group 1 carcinogen - ‘carcinogenic to humans’ (IARC 2012). Arsenic is also responsible for non-carcinogenic effects such as cardiovascular disease, reproductive complications, neurological effects, diabetes mellitus and dermal manifestations including hyperpigmentation and hyperkeratosis (Kapaj et al. 2006; NRC 1999).

The mode of action linking As exposure to the abovementioned health end-points is a subject of ongoing biochemical research. As mentioned, the speciation of As plays an important role in toxicity, namely the species valence or oxidation state, with trivalent arsenicals being more toxicologically potent than pentavalent ones (Hughes et al. 2011). It is generally accepted that inorganic As (i.e. AsIII and AsV), those most commonly present in water and soil, are more toxic than organic species (Hughes 2002). The toxicology of As is determined by its chemistry, notably (i) its affinity for sulfhydryl groups which cause it to inhibit numerous enzymes and proteins and (ii) its and ability to emulate phosphate causing the disruption of cellular processes V (e.g. the substitution of inorganic phosphate (PO4) for As (AsO4) in glycolytic and cellular respiration pathways, forming adenosine diphosphate (ADP)-arsenate instead of adenosine triphosphate (ATP)) (Hughes 2002; Jomova et al. 2011). Other modes of action of As toxicity include the formation of reactive oxygen species, genotoxicity, altered deoxyribonucleic acid (DNA) repair, altered DNA methylation and cell proliferation (Jomova et al. 2011).

The majority of epidemiological data linking As exposure to adverse health effects comes from populations exposed to high concentrations in drinking water, notably an isolated population in and As-endemic region of Taiwan (Chen et al. 1985). Local residents were exposed to As in their drinking water for a period of about 60 years and were found to have a higher prevalence of skin cancer and blackfoot disease with increasing artesian well water concentrations (Tseng 1977). The health effects of chronic As exposure to concentrations <150 µg L-1 are less well documented. However, increased risk of skin cancer has been observed in epidemiological studies conducted, for example, in the US (Beane Freeman et al. 2004; García-Esquinas et al. 2013; Gilbert-Diamond et al. 2013), Finland (although limited to smokers)

27

(Michaud et al. 2004) and Hungary, Romania and Slovakia (Leonardi et al. 2012). Similar associations have been observed for bladder cancer in the US (Karagas et al. 2004), but increased risk was limited to smokers, indicating that As may act as a co- carcinogen with other substances.

1.1.5. Global arsenic exposure It is estimated that 105 countries worldwide are affected by human exposure to As from both natural and anthropogenic sources (Murcott 2012). Populations with the highest levels of exposure are those utilising As-bearing groundwater as their primary drinking water source. Figure 1 shows the estimated (Polya and Middleton 2016) global distribution of As-contaminated drinking-water. The extent of the problem is such that no continent is without occurrence of concentrations exceeding 10 µg As/L. The most notably affected location is Bangladesh (Chakraborti et al. 2010), where it is estimated that 45 million people are exposed to As concentrations >10 As µg L-1 in drinking water. The crisis in Bangladesh arose in response to the drilling of tube wells in the 1970s. This initiative was to combat high rates of bacterial infections and diarrheal disease resulting from the consumption of stagnant surface waters. The deeper, cleaner groundwaters contained high natural concentrations of inorganic As. Berkeley epidemiologist Alan Smith famously described the disaster as ‘the largest mass poisoning of a population in history’, worse than the Chernobyl nuclear accident and the Bhopal Gas Tragedy (Smith et al. 2000). It is evident in Figure 1 that the problem is not restricted to Bangladesh. Four additional As-endemic regions have yielded the strongest evidence of the link between As exposure and cancer. These are West Bengal (Chakraborti et al. 2009) north-west and south-east Taiwan (Chen et al. 1985), northern Chile (Marshall et al. 2007); and Argentina (Hopenhayn-Rich et al. 1996; O’Reilly et al. 2010). A number of developed countries have more localised, less severe, exposure occurrences. In particular, rural communities that rely on private water supplies (PWS). Studies have identified lower (e.g. <150 As µg L-1) exposures in North American and European communities in municipal supplies and PWS where treated water supply has not been implemented or is not accessible. In the USA, an estimated 15 % of people rely on PWS (Steinmaus et al. 2005), and as many as 40 % in New Hampshire, Vermont and Maine (Peters et al. 1999). Notable examples in Europe include Serbia (Jovanovic et al. 2011), Hungary, Romania and

28

Switzerland (5,000) Hungary (1,250,000) France (200,000) Serbia (200,000) Croatia (200,000) USA (30,000,000) Greece (5,000) China (15,000,000) Afghanistan (500,000) Nepal (2,500,000) Bangladesh (45,000,000) Mexico (2,000,000) Pakistan (6,000,000) Myanmar (6,250,000) Guatemala (7,500) India (30,000,000) Vietnam (8,000,000) El Salvador (7,500) Honduras (7,500) Thailand (37,500) Nicaragua (7,500) Cameroon (4,000) Cambodia (250,000) Colombia (7,500) Ecuador (7,500) Peru (625,000) Estimated exposed population* Bolivia (200,000)

<100,000 or not known 100,000-1,000,000 Argentina (4,000,000) 1,000,000-10,000,000 Chile (500,000) >10,000,000

Figure 1: Global scenario of population exposure to As above the WHO guidance value of 10 µg L-1 in drinking water. Estimates are those presented in Polya and Middleton (2016) (references therein).

29

Slovakia (Lindberg et al. 2006) and England (Ander et al. 2016). The latter example forms the focus of this thesis and is discussed in depth in Section 1.2.3. For an extensive review of the worldwide distribution of As contamination, the reader is referred to Murcott (2012). In addition to the 68 countries/territories affected by geogenic related As contamination of groundwaters, a further 173 are affected by anthropogenic related As contamination (including 74 by historical or ongoing mining operations) and 35 by natural and anthropogenically manipulated volcanic and geothermal activity (Murcott 2012).

1.1.6. Study area: south west England Population exposure to As is much less problematic in the UK in comparison to places such as Bangladesh. Similar to much of Europe, cases are localised and relate to specific geological domains and/or anthropogenic activities such as legacies of mining and mineral processing. One long-recognised region of concern in the UK is south west England: specifically the counties of Cornwall and Devon. South west England is an area of elevated environmental As and other elements in soils and stream sediments (Webb et al. 1978). High As concentrations in the region are associated with the highly mineralised areas around granitic intrusions and a legacy of extensive mineral exploitation (Webb et al. 1978). It was estimated (Abrahams and Thornton 1987) that 722 km2 (8 % of the area of south west England) is contaminated with As to some extent. The public health implications of the extent of As contamination in this particular region has been debated and, although its extent of As contamination is well documented, the impact on local public health is yet to be quantified.

1.1.7. Biomonitoring of arsenic exposure The traditional approach to risk assessments for environmental chemicals is a four- stage process, typically consisting of: (i) hazard identification (such as the measurement of As in drinking-water), (ii) exposure assessment (estimations of intake based on values acquired from questionnaires, intake diaries or similar tools), (iii) dose-response assessment and (iv) risk characterization (LaKind et al. 2008). An alternative or additional approach to risk assessments is the use of human biomarkers, a practice known as biomonitoring.

30

Biomarkers can be defined as indicators that signal events in biological systems in response to exposure to an external compound. Biomarkers may be of exposure, effect and susceptibility (Henderson et al. 1987). Biomarkers of effect signal induced changes or impairments in the functional capacity or state of a biological system (e.g. the COMET assay (Button et al. 2010) or measuring micronuclei frequency (Banerjee et al. 2013) in response to DNA damage), biomarkers of susceptibility indicate the particular sensitivity of a biological system to a xenobiotic agent and biomarkers of exposure are the identification of xenobiotics within a system. Biomonitoring of exposure involves the collection of biological material such as bodily fluids or tissues (e.g. blood, urine, nails, breast milk and saliva) and analysing them for the presence of chemicals and their metabolites (Metcalf and Orloff 2004).

31

1.2. Work previously conducted 1.2.1. Arsenic biomonitoring: exposure assessment and interpretation Following absorption into the bloodstream, the majority of As is cleared within the first hour (NRC 1999). This rapid elimination and the comparably low concentrations of As in blood than in other biomarkers (e.g. urine, nails and hair) have led to the limited use of blood sampling in environmental As exposure studies (Orloff et al. 2009). Blood As biomonitoring may still be suitable for recent, high concentration exposure incidents and acute poisonings. Furthermore, blood collection is invasive and the handling of blood poses additional risk to researchers from blood-borne pathogens.

Inorganic As is primarily metabolised in the liver and excreted, along with its methylated metabolites, in the urine (ATSDR 2007; Vahter 2002). This makes the analysis of human urine samples the most common approach to biomonitoring for exposure to environmental As (Orloff et al. 2009). In a study examining the urinary elimination of As species in humans following a single 500 µg g-1 dose of either inorganic sodium arsenate, MA or DMA, the half-time of As in urine was demonstrated to be 4, 11 and 28 hours for MA, DMA and inorganic As, respectively (Buchet et al. 1981). Excreted DMA passed through unchanged, MA was partially (13 %) converted to DMA and 75 % of inorganic As was excreted as methylated metabolites consisting of ~1/3 MA and ~2/3 DMA (Buchet et al. 1981). As mentioned, methylation capacity is understood to show significant inter-individual variation (Tseng 2009), but these findings, and that 46, 78 and 75 % of inorganic sodium arsenite, MA and DMA were cleared via urine respectively after 4 days following dosage (Buchet et al. 1981), highlighted urinary As biomonitoring as a useful tool for assessing recent exposures.

While urinary As profiles reflect recent (3-4 days) exposure, the analysis of nails and hair can reflect a longer window of exposure. The rich blood supply to the nail matrix and hair root and the affinity of As to sulfhydryl groups in the keratin matrix of hair and nails means that it is readily bound to, and thus excreted, via these routes (Raab and Feldmann 2005). The relevant window of exposure reflected by these biomarkers depends on their growth rate, with hair growing faster than nails (Slotnick and Nriagu 2006). The length of the hair sample taken dictates the reflected duration of exposure with a 1 inch section typically reflecting ~2.5 months (Orloff et

32

al. 2009). This is based on an average growth rate of 1 cm per month (Harkey 1993). Fingernails grow by ~0.1 mm per day and toenails between 0.03 and 0.05 mm per day (Fleckman 1997), taking about 6 and 12 to 18 months to grow out respectively (Slotnick and Nriagu 2006). These longer windows of exposure and the less invasive nature of their collection make nails and hair attractive biomarkers of As exposure, however, a lack of standardised protocols, sample mass requirements and susceptibility to external contamination are recognised limitations (Marchiset-Ferlay et al. 2012).

Many previously published investigations have used non-invasive biomonitoring to assess human exposure to As. Correlations have been reported between As concentrations in drinking water and those measured in urine (Calderon et al. 1999), toenails (Karagas et al. 2000) and hair (Gault et al. 2008). Correlations, albeit weaker ones, have been reported between As concentrations in residential soils and those in urine (Hinwood et al. 2004) and toenails and hair (Hinwood et al. 2003).

The interpretation of human biomonitoring data requires the availability of comparison values that correspond to observed deleterious health effects and identify potentially concerning levels in tissues with a view to setting protective standards (Angerer et al. 2011). Such values may represent the upper background concentration of the general population (reverence values) or be health-based thresholds derived from epidemiological analyses. Values are more commonly reported for As concentrations (total and species-specific) in urine than in toenails and hair, due to the wider use of urinary As biomonitoring and the technical challenges associated with toenail and hair biomonitoring.

Reference values are defined by Ewers et al. (1999) as values “intended to indicate the upper margin of the current background exposure of the general population to a given environmental toxin at a given time”. Such datasets are useful for recognizing when biomonitoring results of a group of subjects are higher than the background concentration to be expected in the general unexposed population (Ewers et al. 1999). Values should be based on a sample that is large enough to be representative of the wider population by capturing the effects of confounding factors such as age, gender, smoking status and other nutritional and behavioural traits. The upper 95 % confidence interval (CI) of the 90th or 95th percentile is often adopted as the reference

33

value (Schulz et al. 2011) or the 90th or 95th percentiles themselves. Examples of population-scale biomonitoring programmes that can be used to derive reference values are the US National Health and Nutrition Examination Survey (NHANES), Etude Nationale Nutrition Santé (French Nutrition and Health Survey - ENNS), the German Environmental Survey (GerES) and the Canadian Health Measures Survey (CHMS). Reference values are not representative of health-based endpoints derived from epidemiological and toxicological data but aim to represent the normal background concentrations of a presumably unexposed population.

Examples of existing As reference values for different matrices, derived from some population-scale biomonitoring initiatives from around the world, are presented in Table 1 with values that have been derived from a smaller UK sample with greater relevance to this study. There are currently no reference values available for As concentrations in toenails and the few that are reported for hair are difficult to interpret. This is due to confounding factors on elemental concentrations in hair such as age, gender and cosmetic usage and the inability of older instrumental techniques, e.g. inductively coupled plasma atomic emission spectrometry (ICP-AES), to detect the low As concentrations typical of hair samples collected from unexposed individuals (Miekeley et al. 1998). There is good agreement between the urinary As reference ranges presented in Table 1, providing that speciation analysis is performed to quantify AsIMM (i.e. the toxicologically relevant species- the sum of AsIII, AsV, MA and DMA) or that seafood consumption can be controlled prior to sampling. Hydration adjustment also effects the reference values determined, e.g. creatinine adjustment, discussed later.

Although reference values can be useful for comparison purposes and identifying elevated exposures, they do not identify biomarker concentrations that correspond to an increased risk of adverse health effects. The US Centers for Disease Control and Prevention (CDC) define clinical cases of inorganic As poisonings as >50 µg L-1 of total As in urine (CDC 2013). This is not appropriate for assessing results from non- specific chronic exposure cohorts because speciation is not considered. Examples of health-based values are biomonitoring equivalents (BE) - defined as concentrations of chemicals or their respective metabolites in biological media that are consistent with existing exposure guidance values (Hays and Aylward 2009). Biomonitoring

34

Table 1 Published reference values for total As and toxicologically relevant As species (AsIMM) in urine from nationally representative population-based surveys and smaller studies. Some studies present creatinine-adjusted and unadjusted values.

Exposure metric Biological matrix Country Population/study group n Reference value (reference) Total As Urine Germany GerES III (1998) adults (aged 18-69) 3,924 95th percentile: 13.1 µg L-1 with no prior 48 hr fish consumption. (Wilhelm et al. 2004a) Total As Urine Germany GerES IV (2003-2006) children 1,487 95th percentile: 12.4 µg L-1 (aged 3-14) with no prior 48 hr fish consumption. (Schulz et al. 2009) Total As Urine USA NHANES (2003-2004) All ages (≥6) 2,557 95th percentile: 65.4 µg L-1 (Caldwell et al. 2008) Total As (creatinine-adjusted) Urine USA NHANES (2003-2004) All ages (≥6) 2,557 95th percentile: 50.2 µg g-1 (Caldwell et al. 2008) creatinine Total As (creatinine-adjusted) Urine France ENNS (2006-2007) adults (aged 18- 1,515 95th percentile: 61.3 µg g-1 74). (Saoudi et al. 2012) creatinine AsIMM Urine USA NHANES (2003-2004) All ages (≥6) 2,557 95th percentile: 18.9 µg L-1 (Caldwell et al. 2008) AsIMM Urine Canada CHMS (2009-2010) All ages (6-79) 2,022 95th percentile: 20 µg L-1 (St-Amand et al. 2014) AsIMM Urine UK Non-occupationally exposed UK 95 a95th percentile: 15.8 µg L-1 volunteers (Leese et al. 2014) AsIMM (creatinine-adjusted) Urine France ENNS (2006-2007) adults (aged 18- 1,515 95th percentile: 8.9 µg g-1 74). (Saoudi et al. 2012) creatinine Total As Hair Italy Urban schoolboys (aged 3-15) 412 95th percentile: 0.24 mg kg-1 (Senofonte et al. 2000) Total As Hair Poland Urban students (aged 21-22) 117 90th percentile: 1.03 mg kg-1 (Chojnacka et al. 2010) Total As Hair Brazil Rio de Janeiro residents (aged >20) 1,091 GM+1 SD: <0.15 mg kg-1 (Miekeley et al. 1998) a Estimated by adding the 95th percentiles of individual species.

35

equivalents can be derived from human (if available) or animal pharmacokinetic data to convert external doses into corresponding internal doses (Angerer et al. 2011).

Biomonitoring equivalent points of departure (BEPOD) are calculated as biomarker concentrations that correspond with risk assessment POD values, e.g. the no observed adverse effect level (NOAEL) or the lowest observed adverse effect level

(LOAEL) (Meek et al. 2011). Biomonitoring equivalents and BEPODs have been derived (Hays et al. 2010) for urinary AsIMM in relation to cancer and non-cancerous health effects. For example, the AsIMM BE corresponding to a 1 in 10,000 cancer risk specific dose of 0.027 µg kg-1 day-1 was calculated as 0.84 µg g-1 creatinine (Hays et al. 2010). Biomonitoring equivalents are not to be used as diagnostic tools to clinically interpret biomonitoring results at either the individual or population level, but instead are to be used to aid in the communication of biomonitoring results and categorize results in the order of high, medium and low priority for follow-up assessment (LaKind et al. 2008). High priority results are those that exceed the

BEPOD, medium priority results are below the BEPOD but exceed the BE and low priority results fall below the BE (LaKind et al. 2008).

Other health-based biomarker As concentrations available for comparisons are the American Conference of Government Industrial Hygienists (ACGIH) Biological Exposure Indices (BEI). These values were developed for occupational exposure settings and indicate concentrations below which there should be no adverse health effects (Morgan 1997). The BEI for AsIMM is 35 µg L-1. Data from epidemiological studies, although scarce, also offer values for comparison. For example, a prospective cohort study (García-Esquinas et al. 2013) associated lung, prostate and pancreatic cancer hazard ratios of 2.0 (a doubling of risk) with approximate urinary AsIMM concentrations of 22 µg g-1 creatinine.

1.2.2. Arsenic biomonitoring: technical considerations Many of the reference and comparison values discussed previously were derived by various analytical and data adjustment methodologies which may alter their relevance to given studies. Advances in analytical chemistry and the increased sensitivity in instrumental techniques have resulted in significant progressions in the field of human biomonitoring. Measurements on biological matrices are now made at increasingly lower concentrations, such that expertise in data interpretation is lagging behind (Clewell et al. 2008). Although much attention is paid to instrumental

36

analyte determination, opportunities for error arise during the sample collection, storage and data analysis stages. Challenges exist with the synthesis of robust biomonitoring results and their subsequent interpretation for urine, toenails and hair protocols. Considerations are analyte-specific in nature and those relevant to As biomonitoring are discussed here.

Studies have investigated the effect of different collection treatments protocols and storage conditions on the species stability of urinary As metabolites. Feldmann et al. (1999) compared the effects of a variety of conditions including temperature (25, 4 and -20 °C), chemical additives (hydrochloric acid, benzoic acid, sodium azide, benzyltrimethylammonium chloride, and cetylpyridinium chloride) and storage duration (2, 4 and 8 months) on the As species stability in human urine. Storage at both 4 and -20 °C without the use of additives was reported to maintain stability of all five of the species determine (AsIII, AsV, MA, DMA and AB) for up to 2 months. The study also reported that additives offered no particular benefits to sample storage and, if anything, their addition only increased contamination opportunities. Furthermore, strong acidification was shown to induce substantial changes to species composition (Feldmann et al. 1999). Another study (Chen et al. 2002) investigated the effects of sample pre-treatment, specifically filtration and centrifugation, on the loss of both soluble and insoluble forms of As in human urine. The study found that no significant losses (< 5%) occurred when samples were filtered at 0.45 µm, however, in some cases, centrifugation resulted in up to 50 % loss of soluble As, resulting in the potential underestimation of exposure (Chen et al. 2002).

In addition to sample treatment and storage considerations, important interpretational considerations apply to urinary biomonitoring. Urinary analyte concentrations vary due to factors beyond exposure (Aylward et al. 2014). One major consideration is dilution variation among spot urine samples collected from volunteers at different hydration states in the absence of 24 hr sampling – often not feasible for logistical reasons. First morning void (FMV) and spot samples are used as an alternative (Nermell et al. 2008). Providing that the dilution factor is corrected for before reporting results, both FMV and spot urine samples have been proposed as adequate methods of urinary As exposure assessment (Rivera-Núñez et al. 2010). There remains a high degree of inconsistency in the scientific literature at to the most

37

appropriate method of hydration adjustment. A detailed review of the alternative urinary adjustment methods is presented in Chapter 4 of this thesis.

The technical considerations relevant to toenail and hair biomonitoring are predominantly related to their susceptibility to external contamination. This is widely cited and compromises the ability to distinguish endogenously bound As resulting from exposure to that from sample contamination (Orloff et al. 2009; Slotnick and Nriagu 2006). A number of studies have investigated the effectiveness of different washing procedures for toenail and hair samples (Button et al. 2009a; Kempson and Skinner 2012; Morton et al. 2002; Salmela et al. 1981). Morton et al. 2002 reported that only 40 % of As was removed from hair samples spiked with simulated sweat using a 1 % sodium lauryl sulphate solution due to the irreversible binding of As to the hair matrix. Button et al. 2009a used an alternating cycle of acetone and deionised water and sonication to remove exogenous contamination from toenail samples. The effectiveness of the washing procedure was assessed by retaining rinse solutions for analysis by ICP-MS. The authors reported that small amounts of endogenously bound As were potentially leached from samples during the rinsing process but also provided evidence of the effective removal of exogenous contamination using this method (Button et al. 2009a). Collection methodologies can also be adapted in an attempt to minimise the contamination of toenail and hair samples. For example, nail polish has been estimated to contribute over 600 % to As concentrations if present (Favaro et al. 2005). Requesting volunteers to refrain from using polish prior to providing nail samples, and collecting information of the usage of such cosmetic products, may help to address this. The location that hair samples are collected from the head has been suggested as a factor controlling their susceptibility to contamination, with the nape of the neck suggested as the most appropriate location to sample (Hindmarsh 2002).

1.2.3. Arsenic in south west England: environmental monitoring Figure 2 shows the distribution of As concentrations in soils (Figure 2A) and stream sediments (Figure 2B) in England and Wales produced by interpolating data collected from the Geochemical Baseline Survey of the Environment (G-BASE) (Johnson and Breward 2004). Soils and sediments in south west England would be expected to contain elevated concentrations of As in the absence of a historical mining and smelting legacy. Pedogenesis (soil formation) of naturally high-As

38

parent material results in higher element concentrations in soils (Hanh et al. 2010) and it has been demonstrated that these concentrations are significantly determined by parent material (Rawlins et al. 2003). This was evident in a recent study, where soils taken from an un-worked mineralised area in Cornwall in south west England contained As concentrations ranging from 123 to 205 mg kg-1 (Klinck et al. 2005). Nevertheless, mining and smelting have played a significant role in the widespread contamination of As in the region (Mitchell and Barr 1995).

A B Soil total As Stream sediment (mg/kg) total As (mg/kg) 95.7 - 2,161 195.1 - 12,000 37.1 - 95.7 67.1 - 195 27.1 - 37.1 40.1 - 67 19.1 - 27.1 21.1 - 40 14.2 - 19.1 14 - 21 10.7 - 14.2 10.1 - 13.9 9 - 10.7 8.1 - 10 8 - 9 6.7 - 8 6.8 - 8 5.1 - 6.6 0.1 - 6.8 0 - 5

Figure 2 Interpolated maps of soil total As concentrations (A) and stream sediment total As concentrations (B) measured in samples collected across England and Wales for the Geochemical Baseline Survey of the Environment (G-BASE) (Johnson and Breward 2004). Contains British Geological Survey materials © NERC 2016.

Arsenic contamination from mining activity arises from two sources: (i) waste rock from mineral exploration and (ii) mineral processing tailings of which the As concentration varies depending on the initial grade of the ore, the efficiency of the processing and the cut-off point at which the ore was deemed to be economically worthy of processing (Mitchell and Barr 1995). Arsenopyrite ore, or ‘mispickel’, was processed by sublimation in calciners (Figure 3). It was first crushed and then roasted, with the vapour passing through a labyrinth of chambers or ‘flues’ before condensing in the form of crude As or ‘arsenic soot’ which was scraped off and roasted again. The final refined product of arsenious oxide was sold by the barrel

39

(Richardson 1992). The local landscape in south west England bears many ruins of calciners as well as spoil heaps (Figure 4), the tailings of which have yielded concentrations >204,000 As mg kg-1 (20 %) (Klinck et al. 2005). Heaps are devoid of vegetation barring only the hardiest of plant species, making them more susceptible to wind erosion. Former mining sites have been used for recreational purposes. One report (Hamilton 2000) described a 100 m plume of dust that was generated when the former Devon Great Consols (DGC) mine was used as a circuit for a car rally. Such former mining sites are potential sources of exposure to As and other elements.

Numerous studies have reported elevated As concentrations in both environmental and biological matrices collected in south west England. The early geochemical reconnaissance work of the Applied Geochemical Research Group at Imperial College in London, led by Professor John Webb, laid the foundations for further, more detailed, surveys. Studies such as that presented by Nichol et al. (1971) in connection with the compilation of the Wolfson Geochemical Atlas (Webb et al. 1978) not only highlight the area as one of high elemental concentrations (up to 1,500 As mg kg-1), but demonstrate the applicability of stream sediment sampling as a proxy for catchment soil concentrations. This had important implications for the fields of mineral exploration, agriculture and environmental science. More detailed, regional, geochemical reconnaissance work was undertaken in the Tamar Valley and Dartmoor areas of Devon (Colbourn et al. 1975). This linked high soil As concentrations to geochemical anomalies on existing stream sediment maps. The highest soil As concentrations (max: 2,500 As mg kg-1) were found in soils collected at the Tamar Valley mining area (Colbourn et al. 1975). A study focussing specifically on As was conducted on the Tamar, Tavy and Lyhner river systems (Aston et al. 1975) and found large variations in As concentrations in stream sediments (3 to >5,000 As mg kg-1) that were reflected by corresponding stream water concentrations. Waters containing >250 As µg L-1 were associated with sediments ranging from 500 to 5,000 As mg kg-1 in Gunnislake – a former mining area. Waters containing between 10 and 50 As µg L-1 were associated with sediment concentrations ranging from 200-1,500 As mg kg-1 in the River Lyhner. Lower water concentrations in the Tamar (10 As µg L-1) were found where sediments contained

40

Figure 3: Photograph of an arsenic calciner at the former Devon Great Consols (DGC) mine.

Figure 4: Photograph of a spoil heap of arsenic-bearing mine tailings at the former Devon Great Consols (DGC) mine.

41

between 3 and 100 As mg kg-1 (Aston et al. 1975). These early findings had implications regarding the quality of surface water supplies in the region. Other studies have reported high As concentrations in local surface waters e.g. mean concentrations of up to 42 As µg L-1 (AsIII + AsV) in surveys of the and Restronguet Creek in Cornwall (Klumpp and Peterson 1979).The accumulation of As by fruit and vegetable produce and the implications of human exposure in south west England has been investigated. Locally grown strawberries and lettuces were found to contain increased concentrations of As when grown in high-As soils (Thoresby and Thornton 1979). Concentrations were still relatively low, with a maximum concentration of 1.4 As mg kg-1 in lettuce and 0.7 As mg kg-1 in strawberries. Xu and Thornton (1985) sampled home-grown vegetable plot soils and six commonly grown salad crops: lettuce, carrots, beetroots, onions, beans and peas for total As determination. A wide range of total As soil concentrations were observed (144 - 892 As mg kg-1) and were significantly correlated with As concentrations in beetroot (r=0.85, p<0.01), lettuce (r=0.70, p<0.01), onion (r=0.54, p=<0.01) and peas (r=0.46, p=<0.05), but not concentrations in carrots and beans. Of the six vegetables tested, none exceeded 1 As mg kg-1 fresh weight. These findings suggested that these plants acted as a ‘geochemical barrier’ to As and that other exposure pathways, e.g. the direct ingestion of soil particles adhered to produce, may be more relevant in home-grown settings. A more recent study, conducted by the University of Aberdeen and the Food Standards Agency (FSA) (Norton et al. 2013), undertook a market basket and field sampling survey of locally grown produce in south west England. Arsenic concentrations in 33 products were compared to those in samples collected from a region without elevated environmental As - Aberdeenshire, Scotland. Produce grown in south west England contained higher total As than that from Aberdeenshire. Vegetables categorised as ‘open leaf’ (e.g. lettuce, kale and chard) exhibited the highest concentrations. A lower correlation was reported between soil As concentration and produce As concentration for open leaf vegetables compared to most other types. This indicated that deposition from wind- blown particulates may be a more significant contribution (Norton et al. 2013) to the above ground portion of produce. The authors used the potato, a UK dietary staple, to estimate average daily intake for a 60 kg individual of 0.02 As µg kg-1 day-1. In this example, As intake from potatoes in south west England equated to 0.2 % of that from rice (Norton et al. 2013).

42

Another potential As exposure media that has been investigated in south west England is household dust. Arsenic in household vacuum cleaner dust was investigated by (Culbard and Johnson 1984), who found elevated concentrations in the former mining area of and Hale (range: 1-330 mg kg-1) compared to samples collected from a control area in Lincolnshire (range 4-50 mg kg-1). In another study (Rieuwerts et al. 2006), mean concentrations in dust collected from homes near a former mining site ranged from 43 to 486 As mg kg-1 compared to 1.7 to 29 As mg kg-1 in samples collected from a non-mining control area. The high As concentrations in disused mine tailings has the potential to be atmospherically dispersed and accumulate in local properties (Hamilton 2000; Mitchell and Barr 1995). Ambient air particulate As concentrations were monitored at nine sites in Cornwall between 2005 and 2006 in relation to proximity and surrounding density of former mining sites (Barnes et al. 2006). No statistically significant relationship was observed.

A little investigated potential source of As exposure in south west England is drinking water from PWS. In England, PWS are those not provided by water/utilities companies having been statutorily appointed (Water Industry Act 1991) and, if serving a single domestic dwelling, are not subject to such rigorous potability testing regimes. Supplies vary in their sophistication and source nature, ranging from simple standing wells to boreholes and spring capture systems. Treatment systems also vary by property.

In 2014, local authority records, compiled by the drinking water inspectorate (DWI), had the details of 37, 717 PWS in England. An estimated 1 % of the English population live or work on a property with a PWS (DWI 2015). Private water supplies tend to serve those living in rural areas outside of the catchment areas of utilities companies or those opting not to utilise municipal supplies. Figure 5 shows the number of single domestic dwellings served by PWS in England and Wales by local authority district.

43

Figure 5: The number of single domestic PWS plotted by local authority district in England and Wales. Cornwall is highlighted as an area of relatively high usage. Data were obtained from the Drinking Water Inspectorate (DWI 2015). Contains Ordnance Survey data © Crown Copyright and database rights 2016.

South west England is an area of relatively high PWS usage, with reports of 20,000- 30,000 PWS in Devon and Cornwall (Farago et al. 1997). As of 2015, in Cornwall, the focal area of this study, 2,462 single domestic supplies were registered with the DWI and an estimated 5 % of the population were thought to be using a PWS.

Despite the awareness of As contamination in Cornwall, concentration data from PWS in the region was, until recently, relatively scarce. Local council records from three supplies in Cornwall were reported (Sage 1994) at 11, 60 and 80 µg L-1, all exceeding the current 10 µg L-1 UK PCV (Water Supply (Water Quality) Regulations 2000) and WHO guidance value (WHO 2011). Conversely, the baseline report series (BGS commissioned by the Environment Agency) on groundwater chemistry found median, mean and maximum As concentrations of 0.2, 0.8 and 5.8 µg L-1 out of 76 samples collected from granite aquifers in south west England (Smedley 2004). In 2010, the Private Water Supplies Regulations (2009) were

44

implemented. This prompted an initiative to investigate the chemical quality of PWS in England and the implications of PWS consumption on public health. The relatively high frequency of PWS in Cornwall and the elevated environmental concentrations of As and other elements led to its selection as a study site. Between 2011 and 2013, in the first substantial study of its kind in the UK, two field campaigns were carried out by BGS, commissioned by the Health Protection Agency (HPA - now part of PHE). Samples were analysed for 55 trace and major elemental components in addition to the on-site measurement of unstable parameters such as pH and temperature. Twenty-seven out of 497 (5 %) point-of-use drinking water samples exceeded 10 As µg L-1 (Ander et al. 2016), comparable to the 5 % of As test failures across England and Wales in 2013 (DWI 2015). The 25th, 50th and 75th percentile total As concentrations measured by Ander et al. (2016) were 0.2, 0.4 and 1.4 µg L-1, respectively. However, with a maximum concentration of 435 µg L-1, some exceedances were severe.

1.2.4. Arsenic in south west England: human exposure biomonitoring

Fewer studies in south west England have investigated direct human exposure to environmental As via methods such as exposure biomonitoring. The first study to do so (Johnson and Farmer 1989) compared the urinary As profile of 37 volunteers from the Camborne and former mining district in Cornwall to a group of 40 volunteers from Glasgow (control area). Following creatinine adjustment, higher AsIMM concentrations were observed in the Cornwall group for both adults (geometric mean: 6.1 versus 4.4 µg L-1) and children (geometric mean: 10.8 versus 8.6 µg L-1) compared to those from Glasgow. The AsIMM concentrations of two pre- school age children in the Cornwall group were 49 and 21 µg g creatinine-1. More samples collected from Cornwall volunteers had detectable concentrations of AsIII (17 versus 7 samples) and MA (23 versus 6 samples) than those from Glasgow. The main limitation of the study was a lack of standardised protocols at the time it was conducted. Lower As speciation detection limits and the unavailability of certified reference materials reduce the quantitative value of results. Nevertheless, the study made key findings in the difference in exposure between the two groups and the notably high values for the two pre-school children. Children have been identified as a group more vulnerable to exposure to As and other elements for behavioural reasons, particularly in places such as south west England with high concentrations

45

(i.e. >100 mg kg-1) in soil and dust (Elghali 1994; Rieuwerts et al. 2006; Wilson et al. 2006).

Kavanagh et al. (1998) compared the urinary As profiles of two groups living in areas affected by past mining activities (DGC in Devon and Gunnislake in Cornwall) to a group living in an area with no history with mining or As processing (Cargreen in Cornwall). Urinary concentrations measured in the two exposed locations were both significantly different (p=0.01) to those in Cargreen. The sum of AsIII, AsV , MA and DMA (AsIMM) ranged from 2.7 to 59 µg g-1 creatinine in the mining areas and from 2.5 to 5.3 µg g-1 creatinine in the control area. The advantage of this study (Kavanagh et al. 1998) over that conducted by Johnson and Farmer (1989) was that high performance liquid chromatography coupled to inductively coupled plasma mass spectrometry (HPLC-ICP-MS) was used to perform As speciation. This is a more sensitive technique compared to atomic absorption spectroscopy (AAS) - the method used by Johnson and Farmer (1989). However, was limited by small sample sizes (n=7, 17 and 7 for Cargreen, Gunnislake and DGC respectively).

The hair As concentrations of children from Cornwall was investigated in an unpublished study (Elghali 1994) in relation to residential soil, household dust and wipes taken from the children’s hands. Results were compared between four ‘exposed’ areas: Leedstown; Camborne; Carnkie and and a control area: . It was concluded that children were more exposed in the four high As areas compared to the control area. Hand-to-mouth activity (pica) and inhalation were postulated to be the predominant pathways of exposure. One child was deemed to be ‘chronically exposed’ based on a hair As concentration of 1.4 As µg g-1.

Hair As concentrations of Cornwall residents were investigated in a separate study and compared to a group of residents from Oxfordshire and Wiltshire (Peach and Lane 1998). In this study, particle induced X-Ray emission (PIXE) was used to quantify As concentrations in the samples of two groups consisting of 36 age/sex matched volunteers. Those from Cornwall were from the former mining areas of Camborne, Redruth, and Gunnislake. The mean hair As concentration of the samples collected from these areas was 2.5 µg g-1 compared to 0.7 µg g-1 in the control group. A possible limitation of the study was that no effort was made to control the location on the scalp from which the hair was removed. It is usually

46

recommended that samples be collected from the nape of the neck; an area of fresh growth and not as prone to external contamination (Hindmarsh 2002).

Button et al. (2009a) used toenails as a biomarker of As exposure in a group of volunteers (n=8) living in the vicinity of the DGC mine. Comparisons were made with a control group (n=9) from the Nottinghamshire area. Geometric mean As concentrations in milled and microwave digested toenail samples were 5.4 µg g-1 (range: 0.9-26) in the exposed group compared to 0.12 µg g-1 (range: 0.07-0.1) in the control group. Toenail total As concentrations were positively correlated (r=0.60, p<0.001) with soil As concentrations. Additionally, HPLC-ICP-MS was used to determine the As speciation of toenail samples with average findings of 85, 13 and 8.5 % of AsIII, AsV and DMA, respectively. In two samples from the DGC group, MA was present. As a pilot study, small sample numbers were analysed. This was the first study to use toenails as a biomarker in the south west England region and the first to quantify the degree of exogenous contamination of toenail samples.

1.2.5. Arsenic in south west England: public health implications

Despite the large body of work that has been published on the elevated environmental concentrations in south west England, it remains inconclusive as to whether there is an increased health burden resulting from human exposure. Clough (1980) suggested a link between local elevated As concentrations and elevated rates of skin cancer, noting that rates of malignant melanoma for both males and females were higher in Devon and Cornwall than adjacent counties with identical sunshine hours. The author speculated that the distribution of As in the local environment may be a factor and, in addition, proposed that occupational groups such as farmers may be more susceptible due to the combined effects of ultraviolet radiation (UV) exposure and contact with As contaminated soils (Clough 1980).

Another study (Allen-Price 1960) reported that ‘local “folklore” has it that certain parts of west Devon are rife with cancer’. The study presented the high inter-parish variability of cancer rates and, while accepting the large demographic differences, noted several trends. High incidences over a 20 year period reportedly occurred in parishes residing on highly mineralised Devonian strata compared to adjacent geologies. Furthermore, after plotting cancer deaths and deaths from all causes on a map of a cluster of hamlets, a difference was observed in the ratio of cancer to other

47

deaths between boundaries separated by water supply source (Allen-Price 1960). A divide was mapped with higher incidences appearing in a hamlet supplied by underground wells and springs, leading the author to summarise that ‘there may be a cancer-provoking ingredient in these water-supplies’. The study was later criticised by (Shaper et al. 1979), who reasoned that the high population variation between parishes and lack of demographic information regarding age and gender and the nature of the cancers rendered the results misleading and difficult to interpret. Shaper et al. (1979) concluded that, while mapping the spatial distribution of diseases identifies trends and hypotheses, there remains the need to conduct population based epidemiological studies to test their significance (Shaper et al. 1979).

An ecological pilot study (Leonardi et al. 1995) used As concentrations from the Wolfson Geochemical Atlas (Webb et al. 1978) as a measure of As exposure, defining exposed populations as those living in areas where stream sediment concentrations exceeded 100 As mg kg-1 and calculated bladder cancer incidences for the populations living in those areas and compared to those living in unexposed areas (<100 As mg kg-1). The study failed to identify a difference in rate ratios (sex, gender and Carstairs deprivation score adjusted) between the two groups. In a more recent geographical study (Wheeler et al. 2013), age and sex standardised regional registration data of non-melanoma skin cancer for local authority areas in England were used in an attempt to find a link between different environmental risk factors. Exposure to UV, environmental As and household radon (Ra) were investigated. Data were obtained from Wolfson Geochemical Atlas (As data), Met Office (mean daily sunshine duration) and Health Protection Agency (Ra data). The authors found no association between environmental As and skin cancer. Several limitations are reported in this study. Firstly, regional variation exists in the adequacy of non- melanoma skin cancer registry data as it is currently not mandatory to register cases (Wheeler et al. 2013). Secondly, a limitation shared with the study conducted by Leonardi et al. (1995), is the use of the Wolfson Geochemical Atlas to estimate As exposure. While the applications regional geochemical reconnaissance surveys are abundant, their utility as a proxy for human exposure is insufficient. The approach fails to consider the possible exposure routes that may result in an increased health burden.

48

In their review, Mitchell and Barr (1995) conclude that previous studies lack sufficient statistical power and fail to identify what is likely to be a small number of additional deaths resulting from chronic As exposure in south west England. They suggest that one area warranting further investigation the exposure of vulnerable population sub-groups. One such proposed group is the PWS using communities in rural areas, for whom limited data is available on exposure to As via the ingestion of drinking water (Mitchell and Barr 1995).

49

1.3. Aims and objectives The primary aim of this study was to assess human exposure to environmental As from PWS in Cornwall, south west England, using up-to-date environmental monitoring and non-invasive biomonitoring methodologies. Exposure from additional sources (i.e. residential soil and dust) were also investigated.

To meet this aim, the following objectives were undertaken:

(i) Assess recent and ongoing exposures to As from PWS by collecting and determining total As concentrations in drinking water and urine samples, and As speciation in urine samples. (ii) Conduct a comparative assessment of urinary analyte hydration adjustment techniques to inform best practice for urinary biomonitoring studies using data from a wider population based survey. (iii) Assess prolonged exposure to As from PWS by monitoring the repeatability of total As concentrations in drinking water samples collected up to 31 months apart and total As concentrations in toenail and hair samples. (iv) Assess human exposure to As via additional pathways by collecting and determining total As concentrations in other environmental media (residential soil and dust).

Dietary exposures were considered but were largely beyond the scope of this PhD project.

50

CHAPTER 2: METHODOLOGY

2.1. Chapter overview This chapter describes the methodology employed to conduct the follow-up biomonitoring study of the 2011-2013 PWS survey. The data collected from this research effort relates Chapters 3, 5 and 6 and comprised the majority data source for this thesis. The work presented in Chapter 4 uses data from the US National Health and Nutrition Examination Survey. The specific data acquisition methodologies relating to Chapter 4 are described therein. Specific methodologies for individual outputs (Chapters 3, 5 and 6) of the biomonitoring study described here are also described within these individual chapters. This chapter describes, in greater detail, the wider methodological considerations and logistical operations used to conduct the project, including: acquisition of ethical approval; sampling design; recruitment strategy; field work and sample collection; sample storage and preparation considerations; summarised analytical methodologies; data management protocols; statistical approach; dissemination of results to volunteers and, where relevant, documentation to support these aspects. For methodologies that are described in depth in publication chapters (e.g. analytical chemistry techniques, sample-specific quality control measures and statistical applications), general overviews are provided in this chapter to avoid repetition. Justifications for method selection are also discussed, with references to the relevant scientific literature. While not all of the samples described in this chapter have been analysed yet (e.g. rice), their collection is still reported for the purpose of highlighting ongoing research of the wider project.

2.2. Ethical approval This research involved the participation of human subjects and, therefore, ethical approval was sought, and granted, by the University of Manchester Research Ethics Committee (UREC) and the National Health Service (NHS) National Research Ethics Service (NRES), prior to commencing recruitment. The full application submitted to UREC, used as a template by PHE for their application to NRES, is in Appendix A with the letter of approval. Below is a brief summary of the foreseen ethical issues:

. Protecting the confidentiality of medical data and personally identifiable information (PII).

51

. Working with human tissues. . Implications of the research findings and communication of appropriate advice. . Requesting sensitive information from volunteers such as lifestyle and behaviour choices e.g. tobacco and alcohol usage.

To address these issues, relevant guidance was sought from the Medical Research Council on ‘Human Tissue and Biological Samples for use in research’, the Data Protection Act (1998) and the Caldicott Principles. The following measures were employed:

. Written informed consent was obtained from all volunteers (see Section 2.4.1). . Only volunteers who were able to provide informed consent and >18 years old were included. . Data was anonymised ASAP following collection using unique volunteer codes (see Section 2.3.1). . Data was stored in an encrypted database (see Section 2.7). . Only data relevant to aims of the research was collected. . Collection and analysis of biological samples was conducted in accordance with the Human Tissues Act (2004), where relevant. . Volunteers were made aware of what analyses their samples underwent and any additional testing would require the re-acquisition of consent. . Results were reported back to volunteers and relevant advice was provided by PHE depending on the findings made. . Local authorities and GPs were informed of the study and kept up-to-date on progress. . A training day was attended by researchers participating in field work and sample collection with members of PHE staff.

2.3. Volunteer recruitment 2.3.1. Sampling design The target population consisted of adults residing in Cornwall in south west England at a property supplied by a PWS. The sampling frame consisted of households from the target population who had previously had their PWS tested as per their participation in the precursor survey during either the 2011 (East Cornwall) or 2013

52

(West Cornwall) sampling campaign. These households formed the sampling units, at which multiple human volunteers – observational units – could reside. The following coding system was used to distinguish between sampling and observational units:

1001 V1

The six-digit Volunteer Code shown above consisted of a four-digit House Code (1001) and a Volunteer Suffix (V1). The House Codes were unique to sampling units and no two households shared the same House Code. To distinguish between observational units, no two volunteers from the same household shared the same Volunteer Suffix. An example of the data layout for this coding system is shown below in Table 2.

Table 2 An example of the data layout using the adopted coding system. Volunteer Code House Code Name Address 1001 V1 1001 Mrs A. Name Cornwall Farm 1001 V2 1001 Mr A. Name Cornwall Farm 1002 V1 1002 Mrs B. Name Coastal View 1003 V1 1003 Mr C. Name Sunny Cottage

For the purpose of environmental observations, such as drinking water samples. A similar approach was employed:

1001 W1

The above example is a water sample code consisting of a House Code and a Sample Suffix. This functioned in the same way as a Volunteer Code to link samples from the same sampling units but also distinguish between observations.

The sampling strategy was intentionally biased for the purpose of this study. The As measurements made in the precursor PWS survey were used to prioritise household recruitment. An approximate maximum sample size of 150 households was budgeted for and, therefore, households with higher As measurements in their PWS were targeted first. This was to ensure that a range of drinking water As concentrations were captured to investigate the relationship between exposure (environmental measurements) and outcome (biomarker measurements) variables.

53

2.3.2. Recruitment logistics A sampling frame of 476 households was established and plotted using ESRI ArcMap version 9.0 (Figure 6).

")!( !( !(

!(

!( !(") ")!( ¯ ")!( !( !( ")!( !( !( !( ")!( !(!( ")!( !( !(") ")!( !( ")!( !( !( ")!( ")!( !( !( !( !( !( !( !( !(!( !(!( ")!( !( !( !( !( !( ")!( !( !( ")!( !( !( !(!( !( !( !( ")!( ")!( !( !( !( !(!( !( Former local authority districts !( !( !( !(!( ") !( ")!( !( ")!( ")!(!( !( !( !( !( ")!( !( Caradon !( ")!( ")!( !(!( !( !( !( !( !(!( !( !( !( !( !( ")!(")!( !(!( !( ")!( !(")!( !( ")!( ")!( !( !( !( !( !( !( !( Carrick !( !( !( !( !(!(")!( ")!( !(!( !( ")!( ")!( !( !( !( !( !( !( !( !( !( Kerrier !( !( !( !( !(!(!(!(!(!( !(!( !( ")!( !( !( ")!( !( !(") !( ")!( !(!(!( !( !( !( ")!( ")!( !(!(!( !( !( North Cornwall !( ")!( ")!( !( ")!( ")!( !(")!(")!( !( !( ")!(")!(!( !( !(!(!( !( ")!( Penwith ")!( !( !( !( !(!(!( !( !( !( !( !( ")!( !( Restormel !( !( !( !( !( !( !( !(!(!( !(!( !( !( !( !( ")!( ")!( !( !(!( ")!( !( ) Recruited households !(") !(!( !( !( ")!(!( !(!( !( ") !( ")!( ")!( !( !( !(!( !( Unrecruited households !( ")!( !( ")!( ")!( !( !( ")!( ")!( ")!( !( !( !(!( ")!( !( ")!( !( ")!(")!( ")!( !(")!( !( ")!( ")!(!(!(!( !(!( !( !( ")!( !( !( !( !( ")!(!( !(!( !(!(!( ")!( !(!( !( !( ")!(!(")!(!( !(!( !( !( !( !(")!(!( !( !( ")!( !(!(")!( !(!( Initial survey campaigns ")!(!( !(!(")!( ")!( ")!(!( !( ")!( !(")!(!( ")!(!( ")!( ")!(!( !(!( !( ")!( !( !( East Cornwall - 2011 ")!(!(!( !( ")!( !(!( ")!( !( ")!( ")!( !( !( ")!( ")!(!( ")!( West Cornwall - 2013 ")!(!( ")!( ")!( ")!( ")!(!( !(!( !( ")!(")!(")!(!( ")!( ")!( !( ")!( !( ")!(!( !( !( ")!( ")!( !( ")!(!( ")!( !( !(")!( ")!( !( !(!( !(!( !(!( !( !( !( !( 0 5 10 20 30 40 Kilometers

Figure 6 Map showing the distribution of the total number of households in the sampling frame and those recruited to the study. The former local authority districts to which they were assigned for logistical purposes is shown. The areas covered by the initial PWS surveys (Ander et al. 2016) are shown in the bottom right inset map. The map presented in this figure was compiled using ESRI ArcMap version 10.1. Contains Ordnance Survey data © Crown Copyright and database rights 2016.

Twelve days were allocated for fieldwork. This would be conducted by two separate teams under the management of DRSM, with a third team available for three days. As shown in Figure 6, the study area was divided into six parts of roughly equal area. For convenience, the former local authority districts (Caradon, Carrick, Kerrier, North Cornwall, Penwith and Restormel) were chosen. To save time, fuel consumption, and driving hours, prospective households were grouped into one of the six areas, which were allocated days on the calendar. Some households were grouped into adjacent areas if they were closer to a particular cluster (e.g. at the Restormel/North Cornwall boundary in Figure 6). This strategy resulted in a trade-

54

off between the number of household appointments that could be completed per day and the number of appointment options available to volunteers. Furthermore, allocating days to given areas ensured a roughly equal coverage of the study area. This approach was very successful and only a small number of households were unavailable on allocated days. The involvement of a third team allowed for a reserve day where a smaller number of appointments could be booked at further distances apart on days when people were most likely to be at home.

All 476 households were sent an invitation letter and asked to register their interest by telephone or email. Telephone calls were then made to households the following week by DRSM and AR, firstly to those who had already registered their interest. The recruitment process was aided by separate spreadsheets for each recruitment area, sorted in descending order by previously measured drinking water As concentrations. A customised encrypted event booking software (Orbitals booking system) was hosted on a secure server to maintain control of recruitment and allocate appointments.

2.3.3. Response rates Of the 476 households identified in the sampling frame, 209 (44 %) were contacted by telephone before appointment slots were filled. Of these 209 households, 130 (62 %) were recruited to the study and 48 (23 %) declined to participate. A further 31 (15 %) households were unable to participate for other reasons such as volunteers not being in the area at the time of the sampling campaign or not answering the telephone. This resulted in 225 volunteers from 130 households being recruited to the study and mailed a sample collection guide. Of these households, one dropped out of the study between recruitment and the start of the fieldwork campaign and 10 (4 %) volunteers, including additional volunteers from participating households, dropped out. The final sample size was 215 volunteers from 129 households.

2.3.4. Invitation letter, FAQ leaflet and GP letter The following pages display the invitation letter and FAQ leaflet mailed to participants and the briefing letter sent to local GPs. An information video was also filmed at the University of Manchester, available at: http://bgs.ac.uk/sciencefacilities/laboratories/geochemistry/igf/Biomonitoring/arsenic SW.html

55

[Name] [Address] [Date]

“A study of the association between arsenic consumption from private drinking water supplies and measured biological levels in the population of Cornwall, UK” Dear [Name],

Your drinking water supply was tested on the [Date] as part of a study to determine exposure to environmental elements of people living in South West England. Every one of us is exposed to environmental elements to some extent or other. South West England is an area in which there has been extensive mining of metal rich rocks – a consequence of which is that several metals (including arsenic) may be found in higher than average levels in rocks, dust, soils, crops grown in those soils and private water supplies. We are studying how significant exposure to these metals from private water supplies may be. The first step in this study was to determine the concentration of various elements in private water supplies, including arsenic. The second step is to determine how much arsenic is consumed through drinking from these water supplies. As you have a private water supply, we would like to invite you to participate in this study to help gather information on your exposure to arsenic. This second step involves an exercise in which we will ask those who consent to provide where possible (i) an additional tap/drinking water sample, a garden soil sample, a dust wipe sample collected from your windowsills and a small sample of rice from your kitchen, (ii) a sample of urine, toenails and hair to be tested for the presence of arsenic, and (iii) fill out a short questionnaire. This project is a collaboration between Public Health England (PHE), The British Geological Survey (BGS) and University of Manchester (UoM). It is being conducted as part of the study for a PhD by a student from UoM. We are also interested in your response to the results of the water testing and therefore questions regarding what water treatments you have installed are included within the questionnaire. The urine, toenail and hair samples will help us to determine if arsenic is actually being taken up by people living here and to what extent. The questionnaire will help us to determine what other exposure sources there might be – for example from food or dust. The outcome of the study, we hope, will be of value not only to those who have responsibility for minimising long-term public health risks from the environment, but also to you and others who take part, who will be offered personalised advice at the end of the study.

56

We are required to obtain your informed consent to involve you in this study, which we will do by collecting a signed consent form from you at the time of your visit. This study has been given ethical approval by the University of Manchester Research Ethics Committee (ref 13068) and the NHS National Research Ethics Service (NRES) (ref 13/EE/0234). If you are happy to be involved in this study, then appointments will be arranged and you will be visited by two members of the sampling team. A sample collection kit will be provided nearer the time of your appointment and will include information on how to provide your samples on the day. More information regarding your potential involvement in the study is attached in the Frequently Ask Questions leaflet and the Further Information sheet overleaf. If you would like to be involved in the study, please email us at [email protected] in the first instance to register your interest. You can also call us on 01235 825 042. We will then contact you to provide further details of the study. There are a limited number of volunteers we can accept for the study, so please register your interest as soon as possible.

Kind Regards

[Signature of D. Middleton]

Daniel Middleton PhD Researcher on behalf of the research team:

Professor David Polya (University of Manchester Principal Investigator) Dr Michael Watts (British Geological Survey Principal Investigator) Dr Tony Fletcher (Public Health England Principal Investigator)

57

Further information about the study: By agreeing to volunteer in the study you will benefit from gaining unique information about your exposure to environmental arsenic and receive individual confidential results and advice. All samples will be collected and analysed free of charge and will include:

 60 mL of first morning urine  Toenail clippings, ideally from all 10 toes  A snippet of hair from the back of the neck (about a pencil width)

 30 mL of water from the tap you use for drinking  2-4 dust wipe samples from your windowsills  20 g (approx) of rice if you have any in the house  A few small trowels of soil from the land surrounding the property (in most cases this will be the front or back garden)

In addition to the samples above we will ask you to complete an exposure questionnaire which we will bring along on one of our password protected tablets/laptops. We expect this to take about half an hour for a single volunteer to complete. You do not have to provide all of the samples above but for you participation we require a minimum of a tap water and 1 biological (urine, toenail, hair) sample plus the exposure questionnaire. However, results will be more meaningful having collected the full range of samples. For you to be eligible to take part it is required that you use your private water supply for drinking and/or cooking, are aged 18 or over and have no medical conditions that prevent you from participating or having a reduced capacity to provide fully informed consent. Providing that they meet the criteria, additional members of the same household can participate. When booking your appointment please state the number of additional participants so that we are able to reserve sufficient time for your visit. We have prepared a short video to give more information about the study and what we are asking people to contribute. For more information about the study please view the following link: [Website URL]

58

Preparation for our visit If you agree to participate in this study, then in order to maximise the usefulness of the study, we ask that you: (i) Avoid eating any seafood (e.g. fish, mussels, shrimps, crabsticks) for 4 days prior to providing urine, hair and nail samples.

(ii) Allow your toenails and to grow for at least 4-6 weeks prior to our visit and ideally avoid the use of nail varnish during this time.

(iii) Wipe down your windowsills and any other convenient locations that we could take a dust wipe sample from (e.g. your mantelpiece or a shelf). Please make a note of the date on which you do this to allow us to account for time differences between different samples. Leave them untouched before the time of our visit and be careful not to disturb the dust. If possible we would like to collect 2-4 dust samples from your home.

59

Why do you need my written consent? Can I get advice about my results? Your participation in the project is entirely voluntary. By signing the consent Yes. If your results are high, you will be advised to see your GP. On doing so, form, you would be confirming your willingness to take part and whether or please take your results sheet along with you. You will also be able to contact not you would like to receive your results. In particular, you would be agreeing the Devon, Cornwall & Somerset Public Health England Centre on 0844 225 Measuring biological levels of Arsenic in the population to: 3557 for further advice or information. Also, please keep in mind that this study and the results represent only a of Cornwall  Answer some questions snapshot of your exposure to these environmental chemicals. Elevated values  Give small samples of urine, toenails and hair – optionally one, can occur due to a temporary exposure to a chemical and may decrease Questions you might be asking… two or all of these naturally if exposure is reduced. Why do you want to take my biological samples?  Agree to have the samples and information from the interview We are all exposed to chemicals – both natural and those released by human

used, in confidence, by approved researchers. We will also ask activity – in the environment. Many of these exposures are essential for our you to agree to have the samples stored for up to 10 years If you are happy to be involved in the study: health and well-being, others have little or no demonstrable impact, whilst high exposures to some chemicals can lead to varying degrees of adverse health Even if you do consent to participate, you would be free to withdraw later at Please email: [email protected] effects. We can estimate chemical exposures through questionnaires and any time if you wished to do so (see below). analysing chemicals in food, water, dusts and soils, but there are considerable Or call us on: 01235 825 042 uncertainties in these indirect measurement calculations. Are there any costs involved? Biomonitoring directly measures chemicals in the body – from these There is no cost whatsoever involved in volunteering – other than that we ask We will then contact you to provide further details of the measurements on parts of the body that are normally shed or discarded (hair, you to give us about 1 hour of your time, and donation of samples so that we toenails, urine) we can get a more direct measure of exposure. If we have both can undertake the project. study. questionnaire results and biomonitoring results, then we can compare the two, to give us more confidence in our conclusions. Do I get paid? No. But we would very much appreciate your time in participating in the What do I do if I am concerned/want to complain about any aspect of the What environmental element exposures will you be testing for? survey. Arsenic is our focus, due to the large amounts present in the local area as a project? In the first instance, make the researchers administering the questionnaire result of natural processes and human activities (e.g. mining) and the long term What if I change my mind? Can I withdraw from participating in the study? health risks associated with the element. Yes – you can withdraw from the study at any time. aware of any concerns you have. They will address any questions or concerns that you have on the spot or, in the event that they cannot answer questions What are the benefits of taking part? Do I need to give a reason to withdraw from the study? to your satisfaction they will, with your permission, pass on your query to the project managers. You will receive results on your personal exposure to arsenic. In the vast No. You can withdraw for any reason & you do not even need to tell us what majority of volunteers we do not expect to see any values of particular concern, your reason is (although we would like to know, if you are happy to tell us). We hope that we would never do anything that would give you occasion to complain, but if there is anything that you would wish to complain about, but if we do, you would also be sent a further advice sheet on what the results mean, what possible action you could take to avoid or minimise exposure, and Will I be able to see my own results? please call Dr Michael Watts: 0115 936 3042 or email: [email protected] who you can talk to for more advice and information. Yes. Once we have completed our analysis and irrespective of what the results Additionally, you would be helping others in the community, by enabling us to are (i.e. be they “low”, “average” or “high”) you will be sent a results sheet improve our measurements and understanding of arsenic exposure in South containing the different values found in your samples. If your results are high If you would like further information or a large West England. you will receive a specific advice sheet alongside the results and will be advised to contact your GP for follow up testing. PHE will send out a briefing note to all print copy of this leaflet please call: Is the sample collection safe? GP’s within the study area to make them aware that the study is taking place. Yes - as safe as going to the hairdressers or the barbers or getting a manicure! At this stage, we anticipate that very few participants, if any, of those 01235 825 042 Scissors for cutting hair will be sterilised between household visits. If you participating in this study would have results that would require any urgent prefer, we can clip and collect your toenails. If you are due a visit to the remedial action. High values, though, might indicate that some action would hairdressers before our arrival, you could ask them to save the hair from the be recommended to reduce certain exposures in the medium to long term, in back of your neck for you to store until we arrive. If not, we would prefer one particular treatment of drinking water to remove excessive arsenic. Results will of the team to take your hair sample with your permission, to ensure that the be sent to you 4-5 months following samples being taken. samples are collected in the same way. 60

outdoor activities. Taking a soil sample will allow us to understand the soil that combined with those for all the other volunteers will enable us to better Why do you want a urine sample? individual residents come into contact with. This sample will help us support estimate health risks arising from exposure to arsenic. We are also interested We will analyse urine for arsenic and its chemical forms. These data will tell us some of the information we collect from our questionnaire regarding home in your attitude to such exposures. whether arsenic has actually entered the body and will help us to estimate grown vegetables and garden usage. health risks arising from this exposure to the whole study population. I don’t want my personal information to be passed on to unauthorised Why do you want to collect a dust sample? people or to be used in such a way that I can be readily identified. What I don’t want to give a urine sample whilst there are other people in the house. Previous studies have shown concentrations of arsenic in house dust to be systems do you have in place to protect my personal data? We understand this. We are asking those participating in the study to give their higher than those collected elsewhere in the country. We have a quick and Protection of personal data is of particular importance to us and to the urine sample into a pre-provided plastic pot, on the morning of the simple method that will allow us to collect a large number of dust samples in authorities who regulate our activities. All data and samples collected shall be appointment before any of our sampling team arrives. Indeed, for purposes of the region by using specially manufactured wet wipes. This will allow us to take anonymised and handled in accordance with the UK Data Protection Act this study we require a sample collected first thing in the morning and a number of samples from different locations in a single household (2 to 4) and (1998), the Human Tissues Act (2004) and the Standard Operating Procedure refrigerated until we arrive. compare results to those found in soils as well as information about the “Protection of Data & Samples from Participants” for the research project surrounding landscape and industrial history. We are asking people to clean “Biomonitoring of Arsenic in Private Water Supplies, South West England” as Why do you want nail and hair samples? their windowsills as soon as they have booked an appointment over the phone approved by our relevant institutional ethics committees. All the researchers Many environmental elements tend to accumulate in human tissues, such as and record the date. By the time we arrive, enough dust should have settled handling data and samples are required to sign statements that they will hair or nails, so analysis of hair and nails will tell us about exposure to arsenic. for us to be able to take a sample. comply with these procedures and there are severe penalties that may be The factors controlling how arsenic concentrates into hair and toenails are very imposed for any individual or manager who fails to comply with these complex and vary not only from individual-to-individual but also for a given Do I have to provide all three samples (hair, nails and urine) to participate in procedures. individual – so taking both hair and toenail samples will better our the study? understanding of potential exposures. Concentrations of arsenic in urine tend It would really help us if you provided all of the samples. However, you do not Will my personal data be passed on to other agencies? to reflect very recent exposures (hours, days), whilst those in hair and nails need to provide all three to participate in the study, if you wish you can just Absolutely not. This would constitute a breach of the terms of our Ethical reflect longer timescale exposures (weeks, months). provide one or two, for example just hair and nails. Approval. All researchers involved in the handling of data and samples will be required to sign a form to indicate their compliance with the data protection Why do you want me grow my toenails for at least 4-6 weeks? I am very busy and don’t have time to wait around for a sampler to turn up. rules governing the project. So that we have enough material to analyse! Typically we need around 500 Can you provide me with appointment timeslot that is at my convenience? milligrams (mg) of toenail to allow us to get a good analysis. Yes. If you agree to participate, then our staff will call you to arrange a time for What about my DNA? a visit at your convenience. Typically, we will arrange appointments within 2 We will not be undertaking any analysis of human DNA, indeed as part of our Why do you want me to avoid the use of nail varnish prior to your visit? hour timeslots between 8 am and 6 pm, but we will have some time slots after sampling procedures. Any DNA present in urine samples will be destroyed after The chemical content of these products would contaminate toenail samples 5pm. We will telephone you about 10-15 minutes before we anticipate arriving. sampling by adding a cell-breaking chemical to the samples. externally. We are only interested in arsenic that gets into the body, so A visit will typically last about one hour. analysing toenails which are covered in varnish would confuse these results. What will happen to my samples after they have been collected? Why do you want me to not eat seafood for 4 days prior to your visit? Firstly, your name will not be placed on any samples or questionnaires. Rather Why do you want to collect a rice sample? Seafood is well known as providing exposure to some compounds of arsenic we will use a unique code that is sufficient for the purposes of the study – your Rice is grown in waterlogged conditions and can accumulate arsenic through and, although non-toxic forms, this can mask exposure from other sources – so name and the code will only be linked for the purposes of communicating with root uptake of contaminated water during its cultivation. This means that the it will make our data much easier to interpret if we can avoid seafood as a you, for example, the results. rice grains that we eat can contain varying concentrations of the element which possible exposure route. Seafood includes white fish (e.g. cod, haddock), Your samples will be preserved and transported to the British Geological could be significant depending on the amount and frequency of rice in our salmon, mackerel, shellfish (e.g. prawns, mussels, shrimps, cockles), tuna, oily Survey/University of Manchester laboratories where they will be prepared for diets. Taking samples from households will allow us to assess first-hand the fish (e.g. sardines, anchovies) and other fish products such as crab sticks, fish analysis. Urine will be treated to destroy cells and the DNA contained within types of rice people are using to cook and the arsenic content of this rice. fingers etc. them. We do not plan to analyse DNA and are not permitted to do so for this study, and this step will ensure that your DNA is not analysable from the Why do you want to collect a soil sample? What sort of information will I be asked on the exposure questionnaire? samples. Samples will then be analysed for concentrations of arsenic as well as Soils in the South West and much of the UK have been shown to be above The questionnaire will ask for basic information that will help us identify its chemical forms. Unless you provide consent for us to retain your samples average. If people are using the soil in their gardens for growing vegetables or potential exposure routes to arsenic– these exposure routes include through for future analysis we will keep the samples for no longer than the duration of gardening there is a risk of this soil being unintentionally consumed, possibly eating food, through cooking, through drinking and through breathing in dust. the study, after which they will be properly destroyed and disposed of. by not properly washing vegetables, or transfer from hands to food following Basic information about age, gender, smoking habits and activities when 61

To all GPs in Cornwall, via the Director of Public Health, or CCG “A cross-sectional study of the association between arsenic consumption from private drinking water supplies and measured biological levels in the population of Cornwall, UK”

Public Health England, the British Geological Survey and the University of Manchester have undertaken a cross sectional study of arsenic exposure from private drinking water supplies and have measured biological levels in the population of Cornwall, UK. The aim of the study is: To determine the association between arsenic consumption from private drinking water supplies and biological levels of arsenic in Cornwall, UK study population.

We have agreed the reviewed ethical approval to provide individual results to participants of biological and soil values. Following receipt of their individual results it is possible that participants may contact you to discuss their results. This letter provides information about the study and what is involved for participants.

Yours sincerely

[Signature of T. Fletcher]

Dr Tony Fletcher, Principal Investigator, Public Health England

The research team for the project includes Daniel Middleton (PhD Researcher, University of Manchester), Professor David Polya (University of Manchester, Principal Investigator), Dr Michael Watts (British Geological Survey, Principal Investigator) and Dr Tony Fletcher (Public Health England).

62

Summary of the Project

Every one of us is exposed to environmental elements to some extent or other. South-west England is an area in which there has been extensive mining of metal rich rocks – a consequence of which means that several metals (including arsenic) may be found in higher than average levels in rocks, dust, soils, crops grown in those soils and groundwater. We are studying the significance of exposure to these metals via groundwater in private water supplies. It is estimated that private drinking water supplies supply around about half a million people in England, mostly in rural and remote parts of the countryside. The first step in this study was to determine the concentration of various elements in private water supplies, the second step is to determine how much of these elements are consumed through drinking from these water supplies. This second step involved a biomonitoring exercise in which we asked those who gave consent to (i) provide a sample of urine, hair and toenails and to be tested for the presence of arsenic, and (ii) fill out a short exposure questionnaire. In November 2013, biomonitoring, soil and drinking water samples were provided by 214 people from 129 Cornish households with private water supplies and were tested as part of a study to determine exposure to environmental elements of people living in south-west England. The householder’s urine, hair and nail samples will help us to determine if arsenic is actually being taken up by people living in Cornwall and to what extent. The questionnaire will help us to determine what other exposure routes there might be – for example from soil, food or dust. The outcome of the study, we hope, will be of value not only to those who have responsibility for minimising long-term public health risks from the environment. By better understanding the relationship of exposure (from soil, water or other routes) future advice on avoiding exposures can be improved. Also we are offering personalised advice to those who take part and have results exceeding corresponding guidance values, advice to help them avoid potentially hazardous exposure. The study has obtained ethics approval from the NHS National Research Ethics Service (ref 13/EE/0234) and the University of Manchester Ethics Committee (ref 13068). We obtained householder’s informed written consent to be involved in this study. For those individuals who have given consent to participate in this study, appointments were arranged and they were visited by sampling team members. Feeding back biological results to the participants Following the analysis of the samples, Public Health England will shortly send individuals the results of their biological samples by personalised letter; the letter will explain the results in the form of a table showing how their results compare to the results found in the study population and the health based guideline value of 35 µg/l of arsenic in urine set by the American Conference of Governmental Industrial Hygienists (ACGIH). If it is found that residents have high levels of arsenic in their water sample they will be provided with advice on exposure reduction as well as being contacted personally to explain the results. The presence of arsenic in hair or nail samples, are mainly a reflection of past exposures and there are no guidance values available. These were collected for the purpose of looking at their associations with uptake across the whole population, not at an individual level)

63

We will make it clear that the results are only based on one sample. Levels can be affected by a range of factors – for example, eating seafood a few days before a urine sample is collected may give rise to higher than expected concentrations of arsenic in urine, but mainly in a non-toxic form. Therefore, the results must be interpreted with caution. GP action and advice Study participants with health concerns about the levels of arsenic in their biological samples are advised to contact their GP. The arsenic measurements of urine samples provided by eight residents have been found to be higher than the 35 µg/l guidance value set by ACGIH. One individual’s urinary arsenic was greater than 100 µg/L; and they were contacted directly and advised accordingly. GPs can contact the National Poisons Information Service (NPIS) centre (contact details below) for further advice on the clinical effects of this chemical. Please note that finding a measurable amount of arsenic in individual’s urine, hair or nails does not mean that the level of arsenic will cause an adverse health effect.

Contact details If you require further advice please contact any of the following:  For more information on the study please contact the Epidemiology Department, Centre for Radiation, Chemical and Environmental Hazards (CRCE), Public Health England, Chilton, Oxfordshire. OX11 0RQ. Email: [email protected], Telephone 01235 825042

 National Poisons Information Service (NPIS) for clinical management information on the effects of chemicals : Tel 0844 892 0111, Website: www.toxbase.org

Attached is the ‘Questions you might be asking...’ leaflet which was included in the initial letter inviting individuals to participate in the study

64

2.4. Field work and sample collection 2.4.1. Sample collection guide, consent and withdrawal forms Following the arrangement of convenient appointments with recruited volunteers, sample collection guides, containing information on how to self-collect relevant samples and prepare for appointments, were sent via mail. Consent forms required completion prior to participating in the research and the opportunity was available to withdraw from the study at any time. The sample collection guide, consent form and withdrawal form are presented in the following pages.

65

Consent to take part in: “Study of the association between arsenic consumption from private drinking water supplies and measured biological levels in the population of Cornwall” Participation in this study is entirely voluntary. If you wish to take part, then please sign the form, if the following statements are true.

I have read and understood the Information Sheet provided on the study entitled “Measuring biological levels of Arsenic in the population of Cornwall: How to provide your samples”. I have been given a full explanation by the investigator(s) of the nature, purpose, location and likely duration of the study, and of what I will be expected to do. I have been assured that there will be no discomfort or possible ill-effects on my health or well-being arising from participating in the study. I have been given the opportunity to ask questions on all aspects of the study and have understood the advice and information given. I consent to the collection of the following (please tick as appropriate) as outlined in the accompanying Information Sheets: Personal data Toenails Hair Urine Rice

Water Soil Dust

And their use for the research project “Study of the association between arsenic consumption from private drinking water supplies and measured biological levels in the population of Cornwall”, and agree that anonymised data collected may be shared with other researchers involved in this study. I understand that all personal data relating to me will be held and processed in the strictest confidence and in accordance with the UK Data Protection Act (1998).

I consent to my samples being retained for future research or follow up testing: Yes No

I understand that I am free to withdraw from any part of the study at any time without needing to justify my decision and without prejudice.

I understand that should I lose the capacity to consent DURING the study I will be withdrawn but data that has already been collected will still be used.

66

I confirm that I have read and understood the above and freely consent to participating in this study “Study of the association between arsenic consumption from private drinking water supplies and measured biological levels in the population of Cornwall”. I have been given adequate time to consider my participation. Name of Participant (BLOCK CAPITALS) ______

Signed ______Date ______

Statement of Person Obtaining Informed Consent I hereby certify that all data and samples collected by me shall be handled by me in accordance with the UK Data Protection Act (1998), the Human Tissues Act (2004) and the Standard Operating Procedure “Protection of Data & Samples from Participants” for the research project “A study of the association between arsenic consumption from private drinking water supplies and measured biological levels in the population of Cornwall, UK” as approved by the relevant institutional Ethics Committees, by the University of Manchester Research Ethics Committee (ref 13068) and the NHS National Research Ethics Service (NRES) (ref 13/EE/0234).

Name of Researcher (BLOCK CAPITALS) ______

Signed ______Date ______

67

Measuring biological levels of Arsenic in the population of Cornwall: How to provide your samples

Dear [Name] Thank you for agreeing to participate in our study. We really appreciate your time and commitment. This guide is designed to help you prepare for your visit from the sampling team and make the collection of your samples as convenient for you as we can. Instructions are given for each type of sample that we’ve requested.

This pack should contain (per participant): 1 pre-labelled 60 mL urine bottle 1 unlabelled bag for urine bottles 1 pre-labelled toenail bag 1 pre-labelled hair bag If applicable: additional sample collection sheets (in houses with more than one volunteer).

Please read the following instructions very carefully to prevent any complications.

1. Appointment Information

Your visit is scheduled for [Number of volunteers] people on the [Appointment date].

2. How the labelling system works

Each participant has been given a 6 digit volunteer code.

[Name], your code is: [Volunteer code]

Please be sure to put your samples in the container marked with your own code.

Important:

 In houses with more than one volunteer, unless we already have your name, additional sheets will be addressed to:

‘Surname’ additional volunteer (No. 1, 2 etc. if applicable)

 In this case, please be sure that each person puts their name on their individual sheet to ensure that everyone has a unique code and samples are not mixed up.

Name: ______

68

Toenail bags are labelled with your unique 6 digit code followed by ‘Toenails’ Urine bottles are labelled with your unique 6 digit code followed by ‘Urine’ Hair bags are labelled with your unique 6 digit code followed by ‘Hair’

3. Toenails (we are NOT collecting fingernails in this study)

Please note: If you suffer from any conditions (e.g. diabetes), which mean that you require the care of a chiropodist to cut your toenails and are therefore unable to provide them, you will not be required to participate in this aspect of the study.

If you would prefer one of the samplers to clip and collect your toenails on the day of your appointment this is perfectly fine. However, you may be more comfortable collecting your own nails. Either way we ask that you please read the guidelines below.

If you have already collected your toenails, transfer them into the labelled sample bag provided.

 Please remove any nail varnish prior to providing the samples and preferably clip after washing to reduce contamination.  Ideally we would like you to provide a toenail sample from all 10 toes (see photo). Please use a pair of nail clippers/scissors and remove the clipping as you would normally do so.  Try where possible to keep clippings in one piece as they are easier for us to handle during cleaning prior to analysis.  Tip: clip nails over a plastic bag or a newspaper to prevent any from being lost.  Place all of the toenail clippings in your labelled toenail sample bag and seal the top of the bag. Store the nails in a safe place that is cool and dry and unlikely to get damaged.

69

4. Urine

Please refrain from eating any type of seafood (e.g. white fish, oily fish, tuna, mussels, cockles, shrimps, crab sticks) 4 days prior to giving a urine sample. This is due to seafood containing harmless organic arsenic compounds that could interfere with your results. For your visit on the [Appointment date], this would mean not eating seafood from the [Appointment date-4] onwards. We ask that you provide us with a sample from the first urination of the day.  Please wash your hands prior to providing this sample.  Please allow the urine to flow for a few seconds before passing into the bottle to collect urine mid-stream.  Fill the bottle to just below the neck.  Please fasten the screw-top lid ensuring that it is properly secured.  Please wipe off any urine from the outside of the bottle.  Next, place the bottle into one of the clear plastic bags provided and fasten at the top.  Please store your urine sample in the fridge until your visit and we will collect it just before we leave the house. 5. Hair

A small amount of hair (see photos) will be collected by a member of our sampling team at the time of your appointment. If your hair is shorter than usual, then with your permission we would take a wider sample, unless your hair is so short or absent that this is not possible. We prefer to take the sample ourselves, as we need a sample from the back of your neck just

70

above your hairline (nape). Please don’t attempt to take this sample by yourself. All you need to do is keep the labelled provided bag safe until our visit. Reminder of what to expect on the day of our visit You will also be asked to partake in the following activities:  Hand over your collected samples to our sampling team.  If you prefer, have a nail sample collected.  Have a hair sample collected.  Complete a short exposure questionnaire with a member of the sampling team.  Have your height and weight recorded. Or, if you are not comfortable with this you will be able to provide an estimate on the questionnaire.  If you have any in the house, you will also be asked if you can provide a small uncooked rice sample (roughly 20 g).  If available we will also ask to take a small amount of topsoil from your garden (roughly 1-15 cm below surface).  A water sample from your kitchen tap.  A dust wipe sample from the locations you have allocated. If you have any additional queries or something is missing from your sample pack, please don’t hesitate to contact us on: 01235 825 042 or email [email protected]. We will write to you with the results as soon as we are able to after they have been analysed and checked - we anticipate that this may take up to 4-5 months.

We would like to thank you again for your participation and look forward to your appointment.

Kind Regards

Daniel Middleton

PhD Researcher

71

Withdrawal from: “Study of the association between arsenic consumption from private drinking water supplies and measured biological levels in the population of Cornwall” This form is to confirm your withdrawal from the “Study of the association between arsenic consumption from private drinking water supplies and measured biological levels in the population of Cornwall” project. It can be completed by you, or if you are not able to do so for some reason (such as illness), by someone able to act on your behalf. I wish to withdraw from the “Study of the association between arsenic consumption from private drinking water supplies and measured biological levels in the population of Cornwall” project.

Please tick ONE

NO FURTHER CONTACT The “Study of the association between arsenic consumption from private drinking water supplies and measured biological levels in the population of Cornwall” project team would no longer contact you directly, but would still have your permission to retain and use information and samples provided previously. NO FURTHER USE In addition to no longer contacting you or obtaining further information about you, any information and samples collected previously would no longer be available to researchers. The “Study of the association between arsenic consumption from private drinking water supplies and measured biological levels in the population of Cornwall” project team would destroy your samples (although it may not be possible to trace all distributed sample remnants) and would only hold your information for archival audit purposes. Your signed consent and withdrawal would be kept as a record of your wishes. Such a withdrawal would prevent information about you from contributing to further analyses, but it would not be possible to remove your data from analyses that had already been done

Name of Participant (BLOCK CAPITALS) ______

Signed ______Date______

Member of project team (BLOCK CAPITALS) ______

Signed ______Date______

72

2.4.2. Fieldwork Field work was conducted from the 16th - 27th of November 2013 (12 days). An average of 5 household appointments were attended per day, between the hours of 08:00 and 19:00. Two sampling teams operated for the majority of the campaign and consisted of two researchers per team. Team pairings were rotated throughout the campaign. Details of the samples collected in shown in Table 3 and individual collection methods employed are discussed in the following sections. It is noted that, in individual publication chapters, sample sizes may differ to those in Table 3 because subsets were taken to investigate specific research questions.

Table 3 Study group characteristics and details of environmental and biological samples collected during the field work campaign. Study group Households n Total households 129 Single participant 44 (34 %) 2 participants 80 (62 %) 3 participants 4 (3 %) 4 participants 1 (<1 %) Volunteers Total volunteers 215 Male 110 (51 %) Female 105 (49 %) <18 0 18-65 118 (55 %) >65 97 (45 %) Mean age 62 Max age 90 Min age 18 Environmental monitoring PWS Drinking water Total PWS water samples (non QC) 137 Total PWS drinking water samples 134 Total PWS source/storage samples 3 Total supplies/households sampled 128/129 (>99 %) Alternative/additional PWS drinking 6 water samples in same household Field dups 7 Field blanks 6 Soil Total soil samples (non QC) 153 Garden soils 81 (53 %) Vegetable patch soils 71 (46 %) Unclassified 1 (<1 %)

73

Field dups 7 Total households sampled 127/129 (98%) Dust wipes Total wipes collected 409 Sample wipes 351 Field blanks 28 (7 %) Field dups 30 (7 %) Total households sampled 128/129 (>99 %) Mean sample wipe per sampled 3 household Min sample wipes per sampled 1 household Max sample wipes per sampled 8 household Composite indoor dust (vacuum cleaner bags) Total bags collected 109 Total households sampled 109/129 (84 %) Rice Total samples collected 158 Total households samples 116/129 (90 %) Mean rice samples per sampled 1.4 household Min rice samples per sampled household 1 Max rice samples per sampled household 4 Biomonitoring Urine samples Total urine samples collected 212 First Morning Void (FMV) collection 181 (85 %) Spot collection 31 (15 %) Field duplicates 9 Filter blanks 3 Toenail samples Total toenail samples collected 209 Hair Arsenic Total hair samples collected 213

74

2.4.3. Urine, toenail and hair collection Volunteers were asked to refrain from eating seafood for 4 days prior their appointment. Although samples would undergo As speciation, a portion of the DMA present in urine samples can result from the metabolism of arsenolipids and arsenosugars following seafood intake (Navas-Acien et al. 2011) and is therefore not reflective of exposure to environmental inorganic As. Volunteers were asked to self- collecting mid-stream, FMV urine samples (60 mL approx.) in pre-provided HDPE bottled (Nalgene, USA) on the morning of their visit and keep them refrigerated prior to appointments. Before leaving the household, samplers collected urine samples and placed them into cool boxes where they were stored during transit. On reception at the field laboratory, urine samples were registered and filtered through 0.45 µm Acrodisc® syringe filters (PALL Life Sciences, USA) and frozen at -30 °C prior to analysis. The decision was made not to centrifuge urine samples, as this has been reported to remove up to 50 % of insoluble As from samples (Chen et al. 2002).

Volunteers were asked to refrain from using, or remove existing, nail polish from their toes as this has the potential to influence analytical measurements (Favaro et al. 2005). Toenails from all ten toes were self-collected and stored pre-labelled polyethylene bags.

Hair samples were collected by members of the sampling team following the Consortium to Perform Human Biomonitoring on a European Scale (COPHES) protocol (Esteban et al. 2014). Hair >3 cm in length was collected from the nape with stainless steel scissors. Hair <3 cm in length was collected in smaller amounts from several locations on the back of the head. Samples were taken as close to the scalp as safely possible and sections of hair >5 cm were discarded. The hair closest to the scalp was retained for analysis. Scissors were cleaned with ethanol between volunteers.

2.4.4. Drinking water collection Point-of-use drinking water samples were collected from the most frequently used drinking tap (e.g. kitchen tap). Taps were ran for 3 minutes to purge pipes of any standing water. Pre-rinsed (with the water being sampled) LDPE containers (Nalgene, USA) were used to collect samples, which were stored in a cool box

75

during transit. Upon reception at the field laboratory, samples were acidified with 1

% v/v HNO3, and, on return to BGS, with an additional 0.5 % v/v of HCl.

2.4.5. Residential soil collection Home-grown vegetable patches, where present, were prioritised as soil sampling locations. Where a vegetable patch was not present, other uncovered patches of soil were selected (e.g. flowerbeds). To allow the future opportunity to investigate geological controls on soil As concentrations, householders were asked if their soil was modified (e.g. imported or composted). If so, an additional soil sample was collected from an area on the property that was representative of local soil. Top soil (15 cm) was collected by taking 5 sub-samples from the four corners and centre of a 5×5 m square using a Dutch screw auger, as per the Geochemical Baseline Survey of the Environment GBASE (Johnson and Breward 2004) sampling protocol. Samples were stored in Kraft strengthened paper bags.

2.4.6. Indoor dust collection Surface wipe sampling was used to obtain indoor dust As loadings (µg m-2). Although originally developed for Pb monitoring, The Ghost Wipe™ brand was selected as these wipes have been demonstrated as suitable for a range of other elements including As (McDonald et al. 2011). Wipes were collected from windowsills and shelves following the American Society for Testing Materials (ASTM) E 1728 protocol (ASTM 2002) using a 10×10 cm plastic template (Environmental Express, Charleston, USA). Wipes were stored in polypropylene DigiTUBE™ vessels prior to dissolution (SCP Science, Québec, Canada). Volunteers were asked to clean their surfaces prior to appointments and record the date to allow dust loading rates to be calculated.

Composite indoor dust samples were collected by emptying the contents of the household vacuum cleaner into a large sample bag to allow a mass/mass As concentration to be obtained.

2.4.7. Rice collection Rice samples (50 g approx.) were collected in polypropylene centrifuge tubes. Samples were taken of the varieties of rice most frequently consumed, from the packet currently being consumed.

76

2.4.8. Exposure assessment questionnaire An exposure assessment questionnaire was designed to collect basic information such as gender and age as well as behavioural factors that could result in exposure to environmental As, such as consumption of home-grown produce and visits to former mining sites. Whilst some fields were designed to be used as statistical variables in exposure modelling, additional information was included for sample checking purposes. A tape measure and digital weighing scales were used to measure volunteer height and bodyweight. If volunteers were not comfortable having these measurements taken, they were asked to estimate. Volunteers were informed that they were free to skip any questions that they were not comfortable with prior to the start of the interview.

The questionnaire was field tested on members of staff at BGS and the content was reviewed several times by the study team. The final version of the questionnaire was approved by UREC and NRES ethics committees. To administer the questionnaire, a form was designed in Microsoft Access that linked directly to a database. This allowed answers to be input directly into the database on a laptop/tablet device during interviews, saving time and minimising the risk of transcription errors. The full duration of the interview was typically between 15 and 20 minutes. An example of how the questionnaire appeared to interviewers in the Microsoft Access form layout is shown in Figure 7, followed by the full list of fields.

77

Figure 7 Print screen of the exposure assessment questionnaire in Microsoft Access format as it appeared to researchers and volunteers during appointments.

SECTION A: EXPOSURE AND PERSONAL INFORMATION 1. Source reference number: ______2. Volunteer code: ______3. Date of interview: ______4. Name of interviewer: ______5. Name of volunteer: ______6. Address: ______7. Postcode: ______8a. Number of participants at this address: ______8b. Number of residents (including participants and children) at this address: ______8c. Number of children under 16 at this address: ______9. Date of birth: 10. Please tick as appropriate: □ Male □ Female

78

11. How many years have you lived (or regularly visited) the present address? ______12. Typically how many months of the year do you live at the present address? ______13a. At you previous address, was the water supply from a private source? □ Yes □ No 13b. If the answer above is “Yes”, then please give the approximate location of your previous address (postcode or district): ______14. How many years have you been using a private water supply? At any address: ______15. What is your occupation? ______16a. How many hours a week do you spend at work? ______16b. If your work involves outdoor work (e.g. farming, construction, horticulture) please describe the outdoor tasks you do during a typical week (recently) and location. ______16c. How many hours a week do you spend working outside? 17. Please confirm whether you use your water supply for: □ Drinking □ Cooking □ Other e.g. washing, gardening etc. □ All Note to interviewers & participants: For the food and activity related questions (Q18-32), the answers should reflect food consumption and activities over the past 3 months. 18a. On WEEKDAYS, how much of the following do you drink each day? (For each item, please type in a number for each measure (glass, pint etc.) consumed per day).

Small tumbler (¼ pint) Glass (½ pint) Pint Litre½ Litre Mug Can (330mL) Bottle (500mL) Other amount (please specify) Water (or mixed with squash/juice) from private supply Water (NOT from a private supply e.g. bottled) Tea/coffee/

79

similar Soups (using water e.g. not tinned) both instant and homemade Other instant snacks containing water e.g. Noodles/pasta snacks Fruit juice Fizzy drinks Beer Wine Cider Other alcoholic beverages

18b. On WEEKENDS, how much of the following do you drink each day? (For each item, please type in a number for each measure (glass, pint etc.) consumed per day).

Small tumbler (¼ pint) Glass (½ pint) Pint Litre½ Litre Mug Can (330mL) Bottle (500mL) Other amount (please specify) Water (or mixed with squash/juice) from private supply Water (NOT from a private supply e.g. bottled) Tea/coffee/ similar Soups (using water e.g. not tinned) both instant and homemade Other instant snacks containing water e.g.

80

Noodles/pasta snacks Fruit juice Fizzy drinks Beer Wine Cider Other alcoholic beverages

19. Do you ever use a filter jug or fridge filter for any of the above? □ Yes □ No Details e.g. patterns of use, product name: ______20. Do you grow and eat your own fruit/vegetables? □ Yes □ No 21. Do you grow and eat locally grown (if known) fruit/vegetables? □ Yes □ No 22. How many times a week do you eat the following home-grown or locally grown produce? Leaf vegetables Root vegetables Salad vegetables Fruit

23. Do you wash the above? (Please tick)

24. Do you peel the above? (Please tick)

25. If you grow your own produce, do you use the following? □ Fertilizers □ Herbicides/pesticides 26. How many times do you typically eat rice each week (including takeaways)? ______27. If applicable, regarding Q26, what is the typical portion size? ______28. Do you wash your rice prior to cooking? □ Yes □ No

29. What kind of rice do you typically eat? Please tick all boxes below that are appropriate.

81

□ Long-grain rice □ Short-grain rice □ Basmati rice □ Fragrant rice □ Brown rice □ Other kinds of rice e.g. microwave (please specify) □ Don’t know

30. Please indicate by ticking the appropriate boxes below how often you typically eat other rice-based products.

Don’t Never Rarely Weekly Daily know Rice cereals Hot rice cereals Rice cakes and crackers Rice pasta Rice flour Rice drinks Rice milk Rice syrup Rice vinegar Rice pudding Other (please specify)

31. How many times do you typically eat seafood each week? ______

32. Please indicate if you do any of the following activities, and if so the amount of time you typically spend on them and where.

Yes/No Days/Year Hours/Day Location – if applicable do you do this activity on or dear a mining site/spoil heap? Gardening Walking Mountain Biking Hiking

82

Climbing Running Motocross Quad Biking Other outdoor activities (please specify)

32a. Have your activities (Q32) and, other than seasonal home-grown vegetables, diet (Q18, Q26-Q31) been noticeably different over the past 72 hours compared to the last 3 months. □ Yes □ No

32b. If the answer to Q 32A is “Yes”, then please record what was noticeably different over the last 72 hours. ______33. Have you ever smoked (e.g. cigarettes, pipe, cigars)? □ Yes □ No

34. If answer to Q33 is “Yes”, then for how many years of your life have you smoked? ______

35. If answer to Q33 is “Yes”, then do you still smoke?

□ Yes □ No

36. If you still smoke, roughly how many cigarettes/cigars/times do you smoke per day? ______

37. If you have smoked, but no longer do, how long is it since you stopped smoking? ______

SECTION B: ECONOMICS QUESTIONNAIRE Please note: this section was designed by and was for the use of Dr Jonathan Gibson. A former University of Manchester PhD Student. 38. Have you installed additional treatment since water sampling by the BGS?

□ Yes □ No Cost: ______Date: ______

39. Treatment unit(s)/product name:

□ UV □ Particulate □ R.O □ pH □ Fe rem. □ Mg rem. □ Other (please specify):

83

40. Since you have received the results of your water testing have you switched to drinking bottled water?

□ Yes □ Yes, prior to new treatment □ No

41. What is the maximum you would be prepared to pay for a water filter which would provide safe drinking water? (Please circle)

[Less than £250] [£251-£500] [£501-£750] [£751-£1,000] [£1,001-£1,500] [More than £1,501]

42. Since your participation in the previous study have you changed the source of your drinking water to a piped mains connection? □ Yes □ No

43. If you have answered “Yes” to Q42, roughly how much did it cost to change drinking water source? Please also provide cost detail if you have looked into getting a piped connection but decided against it.

£ ______

44. How do you rate the quality of your drinking water, based on appearance, smell and taste of the water as well as how safe you feel it is to drink, compared to the average water quality from a mains water supply?

□ Much better □ Better □ Similar □ Worse □ Much worse

45. If you have answered “Yes” to Q38 or 42, was there one or more elements which were identified from the testing as being present in your drinking water which particularly prompted your decision to purchase a filter or consider purchasing a filter? (Please circle)

[Aluminium] [Antimony] [Arsenic] [Copper] [Fluoride] [Iron] [Lead] [Manganese] [Nickel] [Nitrate]

46. Are you considering moving house in the next year?

□ Yes □ No

47. If you have decided against changing water source or installing new treatments, what best describes your reasons?

□ No problems found by testing □ Too expensive □ Do not trust the results of tests □ Intend to change in the future

48. What is your highest level of education? (Please circle)

[Secondary School (16)] [6th Form/college (18)] [Undergraduate University]

84

[Postgraduate University] [Professional Qualifications]

49. How safe did you/do you believe your untreated well water to be for human consumption? (On a scale of 1-5 with 1 being ‘Extremely Dangerous’ and 5 being ‘Perfectly Safe’) a) Before testing? ______b) After receiving test results? ______

50. Which band of total net household income (in £) best describes your household? (Please circle)

[Under 15,000] [15,001-25,000] [25,001-50,000] [50,001-75,000] [75,001+] [Prefer not to say]

NOTES/ADDITIONAL DETAILS:

______

SECTION C: SAMPLE/FIELD INFORMATION

Sample number: ______Date of appointment: ______Time: ______Name of Researcher Collecting Sample: ______

Height: ______recorded/estimated (circle) Weight: ______recorded/estimated (circle)

How many days since you last ate seafood? (E.g. White fish/shellfish/crab sticks/tuna/prawns) ______

How many days since you last drank an alcohol-based drink? (E.g. Beer/wine/cider/spirits) ______

Time of urine sample: ______First urination of the day? □ Yes □ No

How long since you last cut your toenails? ______

Nail varnish used prior to cutting? □ Yes □ No

Are you regularly or have you recently been exposed to cigarette smoke? □ Yes, Details: ______□ No

When was the last time you used hair products? (E.g. shampoo/gel etc.)

85

Details ______□ Not used Local garden soil? □ Yes □ No Biological samples collected: □ Hair □ Toenails □ Urine

Environmental samples collected: □ Drinking water Sample point: ______□ Garden topsoil Sample point: ______□ Rice Variety/brand: ______□ Dust wipe Additional info (age of house (if known), recent renovations, pets): ______NOTES/ADDITIONAL DETAILS:

86

2.5. Chemical analysis Detailed analytical methodologies are described in the individual publication chapters to which they apply. This section summarises the different analytical and instrumental techniques employed across the full range of samples. This study made use, where possible, of previously developed analytical methodologies and, as such, little method development was required. In particular, the adaptation of methods developed by previous BGS PhD students (Button 2009; O’Reilly 2010) were made. The required quality assurance (QA) and quality control (QC) procedures to ensure data validity are also described.

2.5.1. Storage considerations Some samples required specific storage and preparation considerations. For these samples, the specific methods employed and the problems that they aimed to address are summarised in Table 4.

Table 4 Storage considerations and measured employed for certain environmental and biological samples. Sample Storage/preparation Measures employed considerations Drinking water Prevent As from Acidification with 0. 1 % precipitating out of v/v HNO3 and 0.5 % v/v solution. of HCl. Drinking water Maintain sample stability Refrigerated at 4-6 °C. and prevent evaporation. Urine Prevent bacterial growth; Frozen at -30° C. prevent conversion of As species – AsIII reduction to AsV. Urine Prevent ICP-MS Passed through 0.45 µm nebuliser blockage. syringe filter. Toenails and hair Remove external Manually scraped and contamination from washed with alternative samples. steps of deionised water and acetone.

87

2.5.2. Analytical measurements and techniques The instrumental techniques employed for total As determination were inductively coupled plasma mass spectrometry (ICP-MS) (drinking water, urine, toenails, hair, dust) and wavelength dispersive X-ray fluorescence (XRF) (soil). Arsenic speciation (urine) was performed using high performance liquid chromatography coupled to ICP-MS (HPLC-ICP-MS). Bioaccessibility (vegetable patch soils) was determined using the Unified Bioaccessibility Method (UBM) (Denys et al. 2012). Creatinine, SG and osmolality (urine) were measured using the Jaffe method (Jaffe 1886), refractometry and crypscopic osmometry, respectively. The measurements made on individual sample types, the techniques and instrumentation used are summarised in greater detail in Table 5. All analyses were conducted at the Inorganic Geochemistry Laboratories at the BGS with the exception of creatinine/osmolality and total As in soil which were conducted at the School of Veterinary Medicine and Science (University of Nottingham) and PANalytical, respectively. Analysis of total As and As speciation is ongoing at the Manchester Analytical Geochemistry Unit (University of Manchester).

88

Table 5 A summary of the analytical measurements, techniques and instrumentation for all environmental and biological samples collected.

Sample n Measurement Preparation Analytical technique Instrumentation Environmental samples Pre-acidified 134 Total As Analysed neat ICP-MS Agilent 7500cx drinking water Residential soil 160 Total As Oven dried (40 °C); disaggregated; sieved (<2 Wavelength-dispersive XRF Axios Advanced (all) mm); milled in agate ball mill; 10 g pelletised with 2.5 g binder wax. Vegetable patch 71 Bioaccessible From 2 mm fraction: sieved (<250 µm); UBM ICP-MS Agilent 7500cx soil As extraction (0.6 g); ×100 dilution (1 % HNO3, 0.5 % HCl). Dust wipes 409 Total As Graphite hot block digestion; ×4 dilution (1 % ICP-MS Agilent 7500cx loading HNO3, 0.5 % HCl) Vacuum cleaner 101 Total As Sieved (<250 µm); graphite hot block digestion ICP-MS Agilent 7500cx dust (0.25 g); ×40 dilution (1 % HNO3, 0.5 % HCl). Biological samples Pre-filtered urine 212 Total As Thawed at room temperature; ×10 dilution (1 % ICP-MS Agilent 7500cx HNO3, 0.5 % HCl). Pre-filtered urine 212 As speciation Thawed at room temperature; ×10 dilution HPLC-ICP-MS Dionex GP50 gradient pump; (deionised water). Hamilton PRP-X100 anion exchange column and guard column; Agilent 7500cx Pre-filtered urine 206 Creatinine Thawed at room temperature; ×10 dilution Colorimetric (Jaffe method) Randox liquid assay kit; Randox (deionised water). liquid assay RX Imola chemistry analyser Pre-filtered urine 212 Specific Measured prior to freezing during fieldwork. Refractometry Atago PAL-10-S digital Gravity refractometer Pre-filtered urine 211 Osmolality Thawed at room temperature. Cryoscopic osmomotery Gonotect Osmomat 030 cryoscopic osmometer Pre-cleaned (206; 185) Total As Microwave assisted digestion (0.1 g if available); ICP-MS MARS Xpress; Agilent 7500cx toenails and hair ×4 dilution (1 % HNO3, 0.5 % HCl).

89

2.5.3. Quality assurance and quality control The BGS Inorganic Geochemistry Laboratories are accredited by the Accreditation Service (UKAS Testing Laboratory 1816). Whilst not all methods at the BGS are accredited to ISO 17025, all analyses were conducted under the accredited management system. Analytical accuracy and precision was monitored throughout the duration of laboratory analysis on all sample matrices. Measurement accuracy was assessed by calculating recoveries of certified reference materials (CRM) and independent QC standards. Precision was assessed using duplicate collection/analysis for a portion (typically 5 %) of each sample type. These were implemented at several stages of analysis: field duplicates were collected for drinking water, soil and dust wipe samples; duplicate filtered urine samples were obtained from samples provided; duplicate digests were performed of toenail, hair, dust wipe and vacuum cleaner dust samples; duplicate UBM extractions were performed and analytical duplicates (both intra and inter-run) were included in all ICP-MS analyses. Background contamination was monitored using reagent blanks for dissolutions in addition to analytical run blanks.

2.6. Data management and statistical analysis 2.6.1. Software Data management was performed using a combination of Microsoft Excel and Microsoft Access. Statistical analysis was performed using the R programming environment (R Core Team 2013) with R Studio as a user interface. Spatial mapping and analysis was performed using ArcGIS (ESRI, USA) – specifically ArcMap versions 9 to 10.1.

2.6.2. Safe handling Due to the sensitive nature of the data in the custody of the study team, protocols were employed to ensure that data were stored and transferred securely. Where possible, PII such as names and addressed were kept separate from analytical data except for the purpose of reporting data to volunteers. All PII was stored in encrypted databases and spreadsheets. All data were stored on secure servers on encrypted devices. For the purposes of merging multiple datasets together, e.g. analytical results and data from the exposure assessment questionnaire, Microsoft Access was used at all times. The use of the coding system described in Section

90

2.3.1 allowed variables to be joined to correct households and volunteers without the risk of ‘copy and paste’ type errors.

2.6.3. Data preparation: reduction and censoring Data required some degree of preparation prior to particular applications. Preparation, such as the formatting of spreadsheet headers for ease of interpretation and readability in the R programming environment was performed using Microsoft Excel. Data reduction, such as the derivation of categorical variables from continuous variables, was performed in Excel. Data subsets were generated suing Microsoft Access queries or subset code in R to minimise the risk of errors. Due to the instrumental origin of the data, some observations were below the analytical LOD. Negative values occurred when samples with lower determined concentrations than blanks had been blank corrected. Alternatively, in the case of As speciation, zero values occurred in the absence of quantifiable chromatogram peaks. To enable statistical analysis of these data, a censorship consisting of replacement with half of the LOD was applied.

2.6.4. Exploratory data analysis Full descriptions of the statistical methods used in this study are given in the publications chapters in which they were used. As a rule, all data underwent exploratory analysis in R prior to the application of specific statistical tests. This was to check for violations of assumptions of various tests and inform appropriate test selection. Distributions were assessed using boxplots, histograms and the Shapiro- Wilk test. Due to the geochemical nature of much of the data, distributions were often positively skewed, requiring the use of non-parametric statistics or suitable transformation (e.g. natural log) of the data.

2.7. Reporting of results to volunteers Individual results were reported to volunteers as they became available and the relevant documentation had been prepared, reviewed and agreed by all parties. Advice was given to volunteers by PHE. Results letters for As concentrations in drinking water, biomonitoring samples (urine, toenails and hair) and residential soil samples were mailed to participants and are presented in Appendix B.

91

CHAPTER 3

Publication:

Urinary arsenic profiles reveal exposures to inorganic arsenic from private drinking water supplies in Cornwall, UK

Journal:

Scientific Reports

Status as of thesis submission:

Published by Nature Publishing Group (Open Access)

Citation: Middleton, DRS, Watts, MJ, Hamilton, EM, Ander, EL, Close, RM, Exley, KS, Crabbe, H, Leonardi, GS, Fletcher, T and Polya, DA (2016). Urinary arsenic profiles reveal substantial exposures to inorganic arsenic from private drinking water supplies in Cornwall, UK, Scientific Reports. DOI: SREP25656

Web link: http://www.nature.com/articles/srep25656

Supplementary information:

Follows publication (p. 12)

92

www.nature.com/scientificreports

OPEN Urinary arsenic profiles reveal exposures to inorganic arsenic from private drinking water supplies in received: 22 May 2015 accepted: 14 April 2016 Cornwall, UK Published: 09 May 2016 D. R. S. Middleton1,2,3, M. J. Watts2, E. M. Hamilton2, E. L. Ander2, R. M. Close3, K. S. Exley3, H. Crabbe3, G. S. Leonardi3, T. Fletcher3 & D. A. Polya1

Private water supplies (PWS) in Cornwall, South West England exceeded the current WHO guidance value and UK prescribed concentration or value (PCV) for arsenic of 10 μg/L in 5% of properties surveyed (n = 497). In this follow-up study, the first of its kind in the UK, volunteers (n = 207) from 127 households who used their PWS for drinking, provided urine and drinking water samples for total As determination by inductively coupled plasma mass spectrometry (ICP-MS) and urinary As speciation by high performance liquid chromatography ICP-MS (HPLC-ICP-MS). Arsenic concentrations exceeding 10 μg/L were found in the PWS of 10% of the volunteers. Unadjusted total urinary As concentrations were poorly correlated (Spearman’s ρ = 0.36 (P < 0.001)) with PWS As largely due to the use of spot urine samples and the dominance of arsenobetaine (AB) from seafood sources. However, the osmolality adjusted sum, U-AsIMM, of urinary inorganic As species, arsenite (AsIII) and arsenate (AsV), and their metabolites, methylarsonate (MA) and dimethylarsinate (DMA), was found to strongly correlate (Spearman’s ρ: 0.62 (P < 0.001)) with PWS As, indicating private water supplies as the dominant source of inorganic As exposure in the study population of PWS users.

Chronic exposure to arsenic (As) in drinking water is a well-documented cause of numerous cancerous and non-cancerous health defects1, including cancers of the lung, bladder and skin. While most cases of chronic As exposure in drinking water have been reported in Bangladesh and West Bengal2,3, countries on all continents are affected4. Recent studies have identified lower (e.g.< 150 As μ g/L) exposures in European and North American populations in both municipal and private supplies in rural locations where centralized treated water supply has not been implemented. Examples of this scenario include Serbia5, Hungary6, Romania7, Slovakia7 and the USA8, where for the latter it is estimated that approximately 15% of the population rely on private groundwater supplies (PWS)9, and as much as 40% of people in New Hampshire, Vermont and Maine10. Many communities in rural parts of the UK also use PWS and a reported 567,261 people in the UK live or work in properties served by PWS11. One area warranting further investigation is Cornwall in South West England, where PWS usage is estimated to range from 20,000–30,000 wells12 although only 2,462 single domestic PWS are currently registered on the Drinking Water Inspectorate (DWI) database11. Cornwall’s diverse geology and extensive history of mineral exploitation make it a region of elevated environmental inorganic As13, with an estimated 722 km2 of As contam- inated land14. Elevated concentrations have previously been reported in soils15,16, stream waters17/sediments18 and household dusts19,20. In 2010, the Private Water Supplies Regulations (2009)21 came into force and prompted an initiative to investigate the possible public health implications of PWS consumption. The abovementioned findings and the high frequency of PWS in Cornwall relative to most of the UK led to its selection for a study22 investigating the trace metal content of UK PWS. This study found As in PWS drinking water exceeding the 10 As μ g/L UK prescribed concentration or value23 (PCV) in 27 out of 497 (5%) households22. This suggests that

1School of Earth, Atmospheric and Environmental Sciences & Williamson Research Centre for Molecular Environmental Science, University of Manchester, Oxford Rd, Manchester, M13 9PL, UK. 2Inorganic Geochemistry, Centre for Environmental Geochemistry, British Geological Survey, Nicker Hill, Keyworth, Nottinghamshire, NG12 5GG, UK. 3Centre for Radiation, Chemicals and Environmental Hazards (CRCE), Public Health England, Chilton, Didcot, Oxfordshire, OX11 0RQ, UK. Correspondence and requests for materials should be addressed to M.J.W. (email: [email protected])

Scientific Reports | 6:25656 | DOI: 10.1038/srep25656 1 www.nature.com/scientificreports/

a considerable number of people in the region may be subject to elevated levels of As in their drinking water, an exposure route not comprehensively investigated in Cornwall, nor indeed the UK as a whole, to date. The identification of elevated concentrations of As in drinking water alone can help provide an indication of the population at risk. However, the use of exposure biomonitoring, the analysis of biological material for the presence of chemicals and their metabolites, allows for a more direct quantification of internal exposure24 underpinning environmental chemical attributable health risks. A common approach to As biomonitoring is the analysis of urine samples for inorganic arsenite (AsIII), arsenate (AsV) and methylated metabolites methylar- sonate (MA) and dimethylarsinate (DMA) that are excreted in the urine following metabolism in the liver25. It is accepted that post intake, inorganic AsV is reduced to AsIII followed by methylation to MA and DMA26. The process was formerly considered to be a detoxification pathway, but findings of genotoxic intermediate trivalent forms of MA and DMA suggest otherwise27. The exact mechanisms of As biomethylation are subject to ongoing investigation28. For the purpose of exposure assessment, the methylation of inorganic to organic As species jus- tifies the quantification of MA and DMA, whilst acknowledging that direct intake of both of these species from dietary sources has been reported29. The majority of As is excreted within 4 days of dosage30, making urinary As a useful measure of recent exposure, and has been used, for example, to demonstrate rice as a significant dietary exposure pathway in rice consumers in the UK31, USA32 and West Bengal33. Several studies have used urinary As to model the risk of health end-points and toxicological responses resulting from exposure to inorganic As. These include type 2 diabetes34, a mortality follow-up of a population with baseline urine measurements which found a significant association with lung cancer35 and increased genotoxicity measured by micronuclei frequency in urothelial cells36. Urinary As biomonitoring, albeit on a small number of volunteers and in relation to soil and dust exposure, has been carried out in Cornwall on two previous occasions37,38 and elevated concentrations were observed relative to control areas with low environmental As. A number of considerations need to be taken into account when urinary As is used as a biomarker of exposure. Firstly, total urinary As results can be influenced by high concentrations of arsenobetaine (AB), an organo-arsenical found in seafood, widely thought to be non-toxic39 and readily excreted unaltered following dietary intake. This makes it necessary to perform speciation analysis on urine samples to quantify the individual As species and exclude the contribution from AB which does not reflect exposure to more hazardous environ- mental inorganic As. Secondly, the variation in hydration status among volunteers means that both first morning void (FMV) and spot urine samples differ markedly in their dilution, both giving imperfect estimates of 24 hr excretion40. Therefore, in order to be used as a robust indicator of exposure, urinary As concentrations require adjustment for dilution to eliminate variation from fluid balance. Creatinine and specific gravity (SG) adjust- ment are widely used, but both methods are susceptible to interferences. Variation in urinary creatinine has been demonstrated between demographic groups41 and in response to variations in muscle mass42 and malnutrition43, while possibly a more relevant deterrent of applying this adjustment factor is its observed relationship with As methylation efficiency44. Alternatively, because SG is routinely measured by refractometry, the presence of uri- nary solutes such as protein (proteinuria), glucose (glucosuria) and ketones (ketonuria) alters the refractive index of the liquid irrespective of its dilution, thus giving inaccurate dilution estimates45. One alternative adjustment factor, seldom used in biomonitoring studies, is urinary osmolality. Previously overlooked due to the lack of wide- spread availability and relative cost of the instrumentation required46, osmolality is regarded as the ‘gold standard’ and definitive measure of urinary concentration in the clinical and veterinary sciences community47. In the case of cryoscopic osmometry, freezing point depression is measured. Freezing point is a colligative property reflective of solute content, expressed here by osmolality (osmoles of solute per mass unit of solvent) and is not vulnerable to the same interferences as SG measurement by refractometry. Given the absence of 24 hr or timed excretion data in the present study, osmolality adjustment was preferred over the two alternative options. This study aimed to: (1) assess human exposure to inorganic environmental As in a population of PWS users in Cornwall using non-invasive urinary As exposure biomonitoring, (2) assess to what extent the biomarker response can be attributed to PWS drinking water as an exposure route and (3) observe the effect of osmolality adjustment to better define the relationship between urinary As and PWS drinking water As. Results Study group demographics. The extent of the study area and spatial distribution of households is shown in Fig. 1. A total of 215 volunteers from 129 households participated in the study. Of these volunteers, 207 from 127 households consisting of 108 males (52%) and 99 females (48%), reported using their water for drinking and provided both a drinking water and urine sample. Henceforth, unless otherwise stated, this sub-group will be the focus of the present article. The mean volunteer age was 62 years old (range: 18–90). The age and gender distribu- tion is shown in Fig. 2. The study group was classified as a 99% rural population (see supplementary information).

PWS drinking water and urine samples. Summary statistics for total As in drinking water samples and total and speciated As in urine samples (unadjusted and osmolality adjusted) are displayed in Table 1 and plot- ted in Fig. 3. Geometric means (GM) were calculated in addition to arithmetic means (AM) as the data were positively skewed. Of the 127 households, 126 (99%) had detectable (> 0.02 μ g/L) As in their drinking water, 62 (49%) had ≥1 μ g/L and 15 (12%) exceeded the current WHO guidance value48 and UK PCV23 of 10 μ g/L. This corresponds to 21 of the 207 (10%) volunteers being exposed to drinking water As concentrations above 10 μ g/L. The maximum PWS drinking water arsenic concentration was 233 μ g/L. All volunteers had detectable (> 0.2 μ g/L) concentrations of unadjusted urinary total As; with a maximum observed concentration of 426 μ g/L. Speciation data yielded a 98% mean recovery of total As and precision, expressed as relative standard deviation (RSD), was 9%. Despite requesting volunteers to refrain from eating sea- food for the 4 days prior to sample collection, a large contribution of total As was from organic AB. Arsenobetaine was detected (LOD 1.3 μ g/L) in 152 (73%) samples whilst the mean contribution of AB to total urinary As was

Scientific Reports | 6:25656 | DOI: 10.1038/srep25656 2 www.nature.com/scientificreports/

Figure 1. Spatial distribution of sampled households. Compiled using ArcMap 10.1.

Figure 2. Study group age and gender distribution. While population risk assessment is not the focus of this aspect of the study, it is noted for future reference that the present sample is not wholly representative of the underlying population of rural Cornwall. *Office for National Statistics (ONS) Rural-urban classification 2011 (RUC11) was used to determine the underlying population (see supplementary information). Adapted from data from the Office for National Statistics licensed under the Open Government Licence v.3.0.

49%; (range: 0–98%). Findings of inorganic AsIII and AsV were lower, with 56 (27%) and 10 (5%) of samples having detectable concentrations (> 0.8 μ g/L; > 1.5 μ g/L) respectively. The sum of AsIII and AsV ranged from < LOD (0.8 μ g/L and 1.5 μ g/L respectively) to 19.2 μ g/L. All samples had detectable concentrations of DMA and 107 (52%) had detectable concentrations of MA. Dimethylarsinate was the dominant arsenic species with the exception of AB. The sum of inorganic As (AsIII and AsV) and its organic methylated metabolites (MA and DMA), referred to here as U-AsIMM, ranged from 0.9 to 124 μ g/L with an arithmetic mean (AM) of 9.0 μ g/L and a GM of 5.8 μ g/L. Urinary osmolality ranged from 181–1161 mOsm/kg, reflecting a large variation in urinary dilution amongst volunteers. Post osmolality adjustment, AM urinary total As moderately decreased from 36.8 to 36.1 μ g/L and the GM slightly increased from 15.8 to 17.1 (range: 2.2–404 μ g/L). The osmolality adjusted U-AsIMM AM and GM also decreased to 8.6 and increased to 6.3, respectively. Additionally, 30 (14%) urine samples were collected as spot samples at the time of visit as opposed to first morning voids (FMV). To address this, a Welch’s independent two-group t-test was used to assess the difference between the two collection methods. For unadjusted and osmolality adjusted U-AsIMM and urinary osmolality no significant difference was observed (P = 0.20, P = 0.30 and p = 0.43, respectively).

Scientific Reports | 6:25656 | DOI: 10.1038/srep25656 3 www.nature.com/scientificreports/

Drinking Urinary Urinary Urinary Urinary water total As total As U-AsIMM Urinary AsIII Urinary Urinary DMA (μg/L) (μg/L) (μg/L) AB (μg/L) (μg/L) AsV (μg/L) MA (μg/L) (μg/L) n 127 207 207 207 207 207 207 207 Unadjusted Arithmetic mean 7.0 37 9.0 26 0.7 0.5 1.3 6.5 Geometric mean 1.0 16 5.8 6.2 0.6 0.8 0.7 4.3 Median 0.9 15 5.3 6.9 0.3 0.3 0.7 4.0 Min < LOD 1.6 0.9 < LOD < LOD < LOD < LOD 0.8 Max 233 426 124 363 13 12 25 79 Osmolality Adjusted Arithmetic mean 36 8.6 26 0.6 0.5 1.2 6.3 Geometric mean 17 6.3 6.5 0.6 0.8 0.7 4.7 Median 15 5.7 7.3 0.4 0.3 0.8 4.4 Min 2.2 1.7 < LOD < LOD < LOD < LOD 1.4 Max 404 131 360 14 9.1 27 84

Table 1. Descriptive statistics for drinking water and urinary arsenic concentrations. Drinking water As results are shown on the basis of collected samples/sampled supplies (n = 127). Between 1 and 4 volunteers were associated with any one drinking water sample. Urinary As results are shown both with and without osmolality adjustment.

Correlation analysis. Scatterplots showing urinary As vs PWS drinking water total As, both before and after AB and dilution adjustment, are shown in Fig. 4. Figure 4a shows that total As in drinking water was not a good predictor of urinary total As, with a large variation in urinary total As even for volunteers with low PWS drinking water As concentrations. However, when corrected for AB (Fig. 4b) a more positive correlation was observed. Correcting for urinary dilution using osmolality measurements further improved the correlation between urinary As (U-AsIMM) and PWS drinking water As (Fig. 4c). To test the strength of these correlations, Spearman’s rank correlation coefficient was used as both variables were non-normally distributed (Shapiro-Wilk test: P < 0.001 for drinking water total As, and both unadjusted and osmolality adjusted urinary total and U-AsIMM) and the results from this analysis are shown in Table 2. Following adjustment for AB, a stronger correlation was observed between drinking water and urine samples (Spearman’s ρ = 0.36 (P < 0.001) and 0.58 (P < 0.001) pre and post AB exclusion respectively). This correlation strengthened slightly (Spearman’s ρ = 0.62, P < 0.001) following osmo- lality adjustment. The correlation between creatinine adjusted U-AsIMM (μ g/g Cre) and drinking water As is also shown in comparison to unadjusted and osmolality adjusted results (Fig. 5) and is weaker (Spearman’s ρ = 0.53, P < 0.001) than both. In addition, correlations were calculated on subsets of different drinking water As con- centrations and were found to weaken with decreasing concentration. For drinking water As versus osmolality corrected U-AsIMM, Spearman’s ρ was 0.81 (P < 0.001) when drinking water As was >10 μ g/L compared to 0.21 (P = 0.031) when < 1μ g/L. This is shown in Fig. 6. Finally, 74 households consisted of >1 volunteer, all of whom were included in correlation analyses. Volunteers (observations) sharing a household (sampling unit) were therefore not independent. Correlations between U-AsIMM concentrations of volunteers from the same household (n = 74) were calculated as ρ = 0.59 (P < 0.001) and ρ = 0.66 (P < 0.001) for unadjusted and osmolality adjusted concentrations, respectively. This had the potential to influence the strength of correlations and, therefore, correlations were re-calculated by ran- domly selecting one volunteer per household for inclusion. These results are presented in Table 3 and, although some correlations (particularly those calculated for lower drinking water As concentration groups) were numer- ically different, the overall pattern remained the same. Furthermore, the correlations re-calculated on osmolality adjusted U-AsIMM concentrations agreed strongly across drinking water concentrations groups with those origi- nally calculated with the inclusion of all volunteers. Discussion The present study shows that exposure to inorganic As in drinking water, although not widespread, is occurring within the Cornwall study population with 10% of the present study group exposed to > 10 μ g As/L in drinking water. Although not a true representation of the actual proportion of population exposure, this study builds on the findings22 of its precursor survey by confirming human exposure from PWS that exceeded the PCV, with high As concentrations in drinking water reflected by dilution and AB adjusted U-AsIMM. The maximum U-AsIMM concentration measured in the present study (124 μ g/L) was comparable with val- ues found in West Bengal49, one of the world’s worst affected regions, some of the highest recorded elsewhere in Europe50, and was higher than any found previously in Cornwall37,51. In 1998 Kavanagh and co-workers51 reported a range of 2.7–58.9 μ g As/g creatinine (U-AsIMM) in urine collected from residents (8 boys aged 3–8; 9 adults aged 30–43) of Gunnislake, Cornwall, although the drinking water supply status of the volunteers was not reported. This demonstrates that the larger sample population in this study revealed further exposure incidences in the region and a previously uninvestigated exposure route, both in Cornwall and the UK to date. Correlation analysis of exposure and response variables showed that the strength of the correlation between drinking water and U-AsIMM reduced with decreasing levels of exposure to total As in drinking water. Variation among U-AsIMM results in volunteers with < 1 μ g/L in drinking water was evident, with some urinary U-AsIMM

Scientific Reports | 6:25656 | DOI: 10.1038/srep25656 4 www.nature.com/scientificreports/

Figure 3. Box and whisker plots of private water supply (PWS) drinking water and urinary As. (a) Total As in drinking water samples plotted with its analytical limit of detection (LOD) (lower dashed line) and the UK As PCV (upper solid line). (b) Individual urinary As species plotted with their respective LODs (dashed lines). (c) Urinary total As and urinary sum of species excluding AB (U-AsIMM). Boxes range from 1st to 3rd quartiles with a median line, lower and upper whiskers are the lowest and highest datum within 1.5 inter quartile range (IQR) of the lowest and upper quartile respectively and circles are outliers. For plotting purposes, speciation data were censored by replacing < LOD values with ½ LOD.

Figure 4. Unadjusted and adjusted urinary As – PWS drinking water As. (a) Unadjusted urinary total As. (b) Unadjusted U-AsIMM (adjusted for AB). (c) Osmolality adjusted U-AsIMM. Linear regression lines are for reference only. A poor relationship between drinking water total As and unadjusted urinary total As is evident (a) due to seafood intake and the large contribution of AB on urinary total As results. This is illustrated by the red dashed line showing high urinary total As results at low drinking water As exposure.

Spearman’s ρ (P value) Drinking Water Drinking Water As Drinking Water Full range As < 1 μg/L (n = 109) 1–10 μg/L (n = 77) As > 10 μg/L (n = 21) (n = 207) Drinking Water As vs Urinary Total As 0.19 (P = 0.048) 0.36 (P = 0.001) 0.55 (P = 0.009) 0.36 (P < 0.001) Drinking Water As vs U-AsIMM 0.18 (P = 0.060) 0.38 (P < 0.001) 0.69 (P < 0.001) 0.58 (P < 0.001) Drinking Water As vs Osmolality 0.21 (P = 0.031) 0.49 (P < 0.001) 0.81 (P < 0.001) 0.62 (P < 0.001) Adjusted U-AsIMM

Table 2. Correlation analysis of exposure and outcome variables for all volunteers. A strong correlation (bold font) is only observed for U-AsIMM (osmolality adjusted) for drinking water As > 10 μ g/L. All household volunteers were included in analyses.

results still higher than 10 μ g/L. As mentioned, Cornwall is an area of high environmental As and these obser- vations suggest that in low drinking water As concentration scenarios, confounding exposure variables such as direct soil ingestion from home grown produce consumption, dust ingestion/inhalation or contact with high As bearing mine wastes could be more prominent. The importance of these exposure routes will be the focus of fur- ther research incorporating the analyses of garden/vegetable patch soils and household dust.

Scientific Reports | 6:25656 | DOI: 10.1038/srep25656 5 www.nature.com/scientificreports/

Figure 5. Comparison of alternative U-AsIMM adjustment methods. The comparison between unadjusted results (a), creatinine adjusted results (b) and osmolality adjusted results (c). The Spearman correlation is stronger in osmolality adjusted results than both alternatives.

Figure 6. Log-log plot of U-AsIMM vs PWS drinking water As divided into drinking water exposure levels. Variables from Fig. 3c plotted on log scale axes to show contrasting exposure-response relationships of participants exposed to different concentrations of As in drinking water. Spearman’s correlation coefficients (ρ ) are displayed for the different drinking water As ranges.

Spearman’s ρ (P value) Drinking Water As Drinking Water As Drinking Water As Full range <1 μg/L (n = 65) 1–10 μg/L (n = 47) >10 μg/L (n = 15) (n = 127) Drinking Water As 0.40 0.27 (P = 0.03) 0.42 (P = 0.004) 0.48 (P = 0.07) vs Urinary Total As (P < 0.001) Drinking Water As 0.60 0.23 (P = 0.06) 0.49 (P < 0.001) 0.69 (P = 0.005) vs U-AsIMM (P < 0.001) Drinking Water 0.62 As vs Osmolality 0.19 (P = 0.13) 0.54 (P < 0.001) 0.82 (P < 0.001) (P < 0.001) Adjusted U-AsIMM

Table 3. Correlation analysis of exposure and outcome variables for single volunteers per household. A strong correlation (bold font) is only observed for U-AsIMM (osmolality adjusted) for drinking water As >10 μ g/L. One volunteer per household was chosen at random for inclusion in analyses.

Additionally, with the exception of AB, DMA was the dominant species measured in urine samples. This is not unexpected, as DMA is the major endpoint of As metabolism in mammals, typically accounting for 60–80% of stable urinary As species excluding AB52. This outcome requires further consideration given the low drinking

Scientific Reports | 6:25656 | DOI: 10.1038/srep25656 6 www.nature.com/scientificreports/

water concentrations of the majority of individuals. This is in agreement with the study of Leese et al.29 who reported high concentrations of AB in urine samples from an unexposed population29, in which DMA was also the dominant species after AB. Given the unexposed status of their study population, Leese et al.29 conclude that dietary sources are responsible for the presence of DMA as well as AB. In addition, they advise that organic meth- ylated species in urine samples do not necessarily indicate exposure to inorganic As. In the case of individuals not exposed to As in their drinking water, future efforts should be made to model the proportion of DMA likely to derive from direct dietary intake versus that excreted as a product of the metabolism of inorganic species. No robust reference value for U-AsIMM applicable to a UK population currently exists and existing values applicable elsewhere are discussed. Commonly cited is the Agency for Toxic Disease Registry (ATSDR) 100 μ g/L total urinary As53. This was not selected for comparison in the present study due to the large contribution of AB to urinary total As and unless seafood consumption can be categorically ruled out then this value is not recom- mended. Of 207 urine samples, 12 (6%) exceeded 21.5 μ g/L of unadjusted U-AsIMM, the approximate creatinine adjusted concentration found in a recent study35 to correspond to a lung cancer hazard ratio (HR) of 2.0 which is equivalent to double the risk of developing the disease. This value is more appropriate for comparison as it is not affected by AB, however it is noted that because it refers to creatinine adjusted urinary As results from a sample of almost 4000 American Indians, it is not directly applicable to the group studied here. An arguably more appropri- ate value is the occupational biological effect index (BEI) provided by the American Conference of Government Industrial Hygienists54 (ACGIH) (35 μ g/L of unadjusted U-AsIMM), of which 8 (4%) samples exceeded. Whilst acknowledging that this was derived for use with occupational exposure, the BEI was chosen as the comparison value provided to volunteers on feeding back their individual urinary As results (unadjusted U-AsIMM). In order to assess the magnitude of exposure it is important to consider how the sample in the present study relates to the underlying population of PWS users in Cornwall and elsewhere in the UK. As demonstrated in Fig. 2a, the sample of volunteers obtained in the present study was biased and is unlikely to reflect the true proportion of exposure in the underlying population. Furthermore, high-As bearing PWS were over-sampled to ensure that a range of exposure scenarios were captured to model the biomarker response. Therefore, the proportion of drinking water As PCV exceedances in the present study is higher than that observed in the wider population of PWS users (12% in the present biomonitoring study versus 5% in the 2011–2013 PWS survey). The relationship between the current sample and the underlying population is a matter for further investigation. Urinary As biomonitoring is useful in assessing recent exposure25, and therefore results offer a snapshot of a relatively narrow exposure window, especially given that FMV/spot samples were taken as opposed to 24 hr col- lections, making it impossible to assess day-to-day variation in individual excretion patterns. Chronic exposure to As cannot be fully assessed by exposure incidence alone, an assessment of longevity is also needed. The analysis of alternative biomarkers such as hair and toenails is ongoing and may provide evidence of longer term exposure, as will analysis of the temporal stability of As in drinking water samples. In conclusion, it has been demonstrated that, following the necessary adjustments of urinary As concentra- tions for AB intake and urinary fluid balance, a strong positive correlation was observed between As concentra- tions in PWS drinking water and urinary As excretion-indicative of ongoing human exposure to inorganic As in PWS drinking water in Cornwall. Given the comparisons to existing guidance values for other populations, the results of the present study are a cause for concern, albeit for a minority of cases. Efforts should be made to raise wider public awareness of the potential hazards associated with PWS usage and, where analytes exceed the PCV, recommendations for treatment should be made given that it has been demonstrated55 that installation of appro- priate treatment systems is effective in reducing exposure to As and other elements. This work has raised points for further investigation which should include: whether chronic/long-term exposure is evident; the importance of additional exposure routes; further refinement of As biomonitoring techniques to account for dietary sources of organic As species in addition to AB; identification of specific population groups at risk. Such groups may be dictated geographically or as a result of individual susceptibility or behavioural risk factors. Particular ‘hotspots’ of high exposure require identification using spatial/geostatistical methods and ongoing questionnaire analysis. Finally, the health implications of PWS usage in the UK warrant more investigation by detailed analysis of supply distribution, consumption patterns, geochemical risk modelling in conjunction with health surveillance datasets. Methods Ethical approval and consent. In accordance with approved guidelines, written informed consent was obtained from all volunteers and only those who were able to provide such were included in the study. In addition, all methods were followed in accordance with approved guidelines. Ethical approval for the study was provided by the University of Manchester Research Ethics Committee (Ref 13068) and the NHS Health Research Authority National Research Ethics Committee (NRES) (Ref 13/EE/0234).

Sampling strategy and recruitment methods. The sampling frame consisted of volunteers previously involved in the 2011–2013 PWS survey carried out by the BGS on behalf of the former Health Protection Agency (HPA), now part of Public Health England (PHE). Households with a PWS at which volunteers resided formed the sampling units. Observational units consisted of those individual volunteers who met the following inclusion criteria: ≥ 18 years of age; did not suffer from a health condition that could prevent them from participating in the study; had not been identified from the previous phase as unwilling/unable to participate further; provided informed consent. Prospective volunteers were contacted via an information/invitation letter prior to receiving a telephone call. All of those with > 1 As μ g/L being found in their drinking water in the previous survey were contacted to include as many as possible in the study. Numbers were then made up with households in the < 1 As μ g/L category. This approach was designed to maximise the range of observed exposures in the study group.

Scientific Reports | 6:25656 | DOI: 10.1038/srep25656 7 www.nature.com/scientificreports/

Sample collection and pre-treatment. Household visits were made to volunteers by sampling teams. Urine and point of use drinking water samples were collected and an exposure assessment questionnaire admin- istered to volunteers using Microsoft Access 2007 on a laptop/tablet device to ascertain whether volunteers were using their PWS for drinking. Drinking water samples were collected by running the tap most frequently used for drinking for a minimum of 3 minutes to purge any standing water from the pipes before collecting the water in pre-rinsed (with the water being sampled) LDPE containers (Nalgene, USA). Samples were stored in a cool box during transit. Samples were acidified with 1% v/v HNO3 on return to the field laboratory, and then with an additional 0.5% v/v of HCl on return to the Inorganic Geochemistry Facility at the British Geological Survey. For urine collection, volunteers were asked to refrain from eating seafood for a minimum of 4 days prior to providing a sample. HDPE containers (60 mL) (Nalgene, USA) were mailed in advance to volunteers who were asked to provide a FMV, mid-stream urine sample on the day of their visit and store it in the refrigerator until collection by the sampling team. Where instructions were not followed (n = 30), a spot urine sample was collected at the time of the visit where possible. Samples were stored in a cool box during transit and, on return to the field laboratory, filtered through 0.45 μ m Acrodisc syringe filters (PALL Life Sciences, USA) into 30 mL HDPE con- tainers (Nalgene, USA) and then frozen at −30® °C until analysis.

Reagents and standards. The aqueous solutions used throughout the study were prepared using 18.2 MΩ deionised water (Millipore, UK). Nitric (HNO3) and hydrochloric (HCl) acids were Romil-SpA™ super purity grade (Romil, UK). Ammonium nitrate (NH4NO3) solutions were prepared from a solid stock of BioXtra ≥ 99.5% purity (Sigma-Aldrich, USA) and pH adjusted using Aristar grade 25% ammonia (NH3) solution (BDH, UK). Arsenic calibration standards were prepared from an in-house® multi-element stock in which the As contribu- tion was from a 1000 mg/L PrimAg grade mono-elemental stock solution (Romil, UK). Arsenic QC standards (5 μ g/L) were prepared from a multi-element® stock solution of various concentrations with As at 20 mg/L (Ultra Scientific, USA). A Tellurium (Te) ICP-MS internal standard was prepared from a PlasmaCAL 10,000 mg/L stock solution (SCP Science, Canada). The following standards were used for the calibration of individual As species as III v follows: As : 1000 As mg/L stock solution of arsenic trioxide (As2O3) (Inorganic Ventures, USA); As : 1000 As mg/L stock solution of arsenic (V) oxide hydrate (As2O5·xH2O) (Inorganic Ventures, USA); MA: 50 As mg/L in-house stock solution of monomethylarsonic acid ((CH3AsO(OH)2) prepared from solid (Sigma-Aldrich, USA); DMA: 50 As mg/L in-house stock solution of dimethylarsinic acid ((CH3)2AsO(OH)) prepared from solid (Greyhound + − Chromatography, UK); AB: 1031 As mg/L BCR-626 standard solution of arsenobetaine ((CH3)3As CH2COO ) (LGC, UK).

Total arsenic determination by ICP-MS. Urine samples were thawed at room temperature and refrig- erated at 4 °C prior to analysis. Due to the high matrix of urine, samples (1 mL) were diluted x10 with 1% v/v HNO3 and 0.5% v/v HCl to reduce the effects of high concentrations of sodium (Na) on signal stability. Acidified PWS drinking water samples were refrigerated at 4 °C prior to analysis and analysed neat. Total As concentra- tions in both water and urine samples were determined using inductively coupled plasma mass spectrometry (ICP-MS). An Agilent 7500 Series ICP-MS instrument (Agilent Technologies, USA) was used under the operat- ing conditions described by Watts et al.56. The instrument was fitted with a MicroMist low-flow nebulizer (Glass Expansion, Australia) and sample introduction was accelerated using an ASXpress rapid sample introduction system (Teledyne CETAC Technologies, USA). A three-point calibration was used with As concentrations at 1, 10 and 100 μ g/L. Arsenic was detected in helium (He) collision cell mode to reduce potential mass 75 pol- yatomic interferences such as argon chloride (40Ar35Cl+). A Te internal standard was introduced simultane- ously via a T-piece and the Te signal response used to fit urinary As data. The limits of detection (LOD) were calculated as 3σ of analytical run blanks and were 0.02 and 0.2 As μ g/L for drinking water and urine samples respectively.

Arsenic speciation by HPLC-ICP-MS. Urine samples (150 μ L) were diluted × 10 with deionised water and As speciation was measured using high performance liquid chromatography coupled to ICP-MS (HPLC-ICP-MS) using the method described by Button et al.57. In summary, a GP50 gradient pump and an AS auto-sampler (Dionex, USA) were coupled to the ICP-MS instrument with PEEK tubing. Chromatography was performed with a PRP-X100 anion exchange column and a PRP-X100 guard column (Hamilton, USA) using gradient elution with the mobile phase (pH 8.65, 1 mL/min) alternating between 4 and 60 mM NH4NO3. A 3-point calibration was used with 1, 10 and 50 As μ g/L solutions of AsIII and a mixed solution of 1, 10 and 50 As μ g/L AsV, MA, DMA and AB. Figure 7 shows a standard chromatogram obtained for calibration solutions. The LODs for this method (3σ of blank values) are reported by Watts et al.58: 0.8; 1.5; 0.7; 0.3; 1.3 As μ g/L for AsIII, AsV, MA, DMA and AB respectively. It is noted that this method cannot distinguish the trivalent and pentavalent forms of both MA and DMA which vary in genotoxicity59.

Urinary dilution measurement and adjustment factor. Urinary osmolality was measured using an Osmomat 030 cryoscopic osmometer (Gonotec, Germany). The osmolality of the urine samples was determined by comparative measurement of their freezing point with that of pure water. The normalisation procedure applied was adapted from that used for creatinine and SG in a recent study on the normalization of urinary drug concen- trations60, and based on the Levine-Fahy equation61 as follows:

Scientific Reports | 6:25656 | DOI: 10.1038/srep25656 8 www.nature.com/scientificreports/

Figure 7. Arsenic speciation standard chromatogram. Chromatograms obtained for standard calibration solutions at 1, 10 and 50 μ g/ L. Calibration of arsenate (AsV), methylarsonate (MA), dimethylarsinate (DMA) and arsenobetaine (AB) was performed with mixed solutions of arsenic (V) oxide hydrate (As2O5·xH2O), monomethylarsonic acid ((CH3AsO(OH)2), dimethylarsinic acid ((CH3)2AsO(OH)) and arsenobetaine III ((CH3)3As + CH2COO-) respectively. Calibration of arsenite (As ) has been plotted simultaneously and was achieved with separate solutions of arsenic trioxide (As2O3).

urinaryAsu=×rinary As osmolality /osmolality, osmolality normalized specimen ()reference specimen where: osmolalityreferenceis thestudy population mean urinaryosmolality(563 mOsm/kg) (1)

Quality control (QC). Field duplicates for drinking water (5% of samples) and urine (4% of samples) were collected with the following mean percentage differences (where analytes > LOD): drinking water total As: 7% (n = 6), urinary total As: 10% (n = 9), AB: 6% (n = 6) and MA: 13% (n = 2). Inter-run duplicates were analysed for urinary total As and AB to assess method reproducibility (total As mean percentage difference: 3% (n = 6), AB: 8% (n = 12). To assess signal stability and the possibility of drift resulting from high urinary matrices, intra-run duplicates were analysed for urinary total As and speciation (species > LOD) (total As mean percentage difference: 8% (n = 5), AB: 2% (n = 5), MA: 2% (n = 1), DMA: 5% (n = 6). Certified reference materials (CRM) were analysed with drinking water and urine samples: NIST SRM 1643e Trace Elements in Water (National Institute of Standards and Technology, USA) (certified value: 58.98 ± 0.70 As μ g/L, recovery: 100% (n = 4), precision: 3%) and NIES No.18 Human Urine (National Institute for Environmental Studies, Japan) (total As certified value: 137 ± 11 As μ g/L, recovery: 99% (n = 14), precision: 5%, AB certified value: 69 ± 12 As μ g/L, recovery: 92% (n = 18), precision: 5%, DMA certified value: 36 ± 9 As μ g/L, recovery: 115% (n = 18), precision: 12%). Independent matrix matched QC standards (total As: 5 μ g/L) were also analysed with urine samples (recovery: 94% (n = 16), precision: 8%). Background contamination was monitored using run blanks for urine and drinking water analyses, reagent (acid) blanks for drinking water analysis and filter blanks for urine analysis. Duplicate measurements were made on 12% (n = 25) of urine samples for osmolality with a mean percentage difference of 1%.

Statistical analysis. Statistical analysis (including the production of exploratory plots) was performed using R version 3.0.0 (base package)62. Welch’s independent two sample t-test was used to assess the difference between results of spot and FMV urine collections. A Shapiro-Wilk test was used to determine the normality of exposure and outcome variables before and after applying log transformation. Correlation tests were performed using Spearman’s rank correlation coefficient accompanied by a significance test to exclude the possibility of the observed correlations resulting from random sampling. Descriptive statistics, with the exception of the geometric mean, were obtained using the ‘psych’ package63. In the case of speciation data, where manual peak integration resulted in samples with zero or negative values for particular species (AsIII and AsV), left censoring was required to enable data for log transformation and the calculation of geometric means. Values < LOD were therefore replaced with that of half the appropriate LOD.

Mapping. All maps displayed as figures in this manuscript were compiled using ESRI ArcGIS Desktop version 10.1. (ArcMap) Environmental Systems Research Institute. Redlands, CA.

Dissemination of results to households. A letter containing individual result data was fed back to households. Where a PCV exceedance was highlighted, specific advice was provided to participants on any poten- tial health risks and suggested corrective actions were given. All participants were provided with appropriate contact details for any follow-up enquiries. The letter and guidance were developed by PHE along with BGS and the Local Authority. The letter was sent from the Local Authority, as the regulator for PWS in England. References 1. IARC. Arsenic and arsenic compounds. IARC Monogr. Eval. Carcinogen. Risk Chem. Hum. 7, 100–106 (1987). 2. Chakraborti, D. et al. Status of groundwater arsenic contamination in Bangladesh: A 14-year study report. Water Res. 44, 5789–5802 (2010).

Scientific Reports | 6:25656 | DOI: 10.1038/srep25656 9 www.nature.com/scientificreports/

3. Chakraborti, D. et al. Status of groundwater arsenic contamination in the state of West Bengal, India: A 20‐year study report. Mol. Nutr. Food Res. 53, 542–551 (2009). 4. NRC. Arsenic in drinking water. National Research Council. Subcommittee on Arsenic in Drinking Water. National Academies Press (1999). 5. Jovanovic, D. et al. Arsenic occurrence in drinking water supply systems in ten municipalities in Vojvodina Region, Serbia. Environ. Res. 111, 315–318 (2011). 6. Sugár, É., Tatár, E., Záray, G. & Mihucz, V. G. Field separation–based speciation analysis of inorganic arsenic in public well water in Hungary. Microchem. J. 107, 131–135 (2012). 7. Leonardi, G. et al. Inorganic Arsenic and Basal Cell Carcinoma in Areas of Hungary, Romania, and Slovakia: A Case–Control Study. Environ. Health Perspect. 120, 721 (2012). 8. Ayotte, J. D., Montgomery, D. L., Flanagan, S. M. & Robinson, K. W. Arsenic in groundwater in eastern New England: occurrence, controls, and human health implications. Environ. Sci. Technol. 37, 2075–2083 (2003). 9. Steinmaus, C. et al. Dietary intake and arsenic methylation in a US population. Environ. Health Perspect. 113, 1153 (2005). 10. Peters, S. C., Blum, J. D., Klaue, B. & Karagas, M. R. Arsenic occurrence in New Hampshire drinking water. Environ. Sci. Technol. 33, 1328–1333 (1999). 11. DWI. Drinking Water 2014, Private water supplies in England-A report by the Chief Inspector of Drinking Water, July 2015 (2015). 12. Farago, M., Thornton, I., Kavanagh, P., Elliott, P. & Leonardi, G. In Arsenic 210–226 (Springer, 1997). 13. Mitchell, P. & Barr, D. The nature and significance of public exposure to arsenic: a review of its relevance to South West England. Environ. Geochem. Health 17, 57–82 (1995). 14. Abrahams, P. & Thornton, I. Distribution and extent of land contaminated by arsenic and associated metals in mining regions of southwest England. Trans. Inst. Min. Metall. Sect. B. Applied Earth Science 96, PB1–B8 (1987). 15. Colbourn, P., Alloway, B. J. & Thornton, I. Arsenic and heavy metals in soils associated with regional geochemical anomalies in South-West England. Sci. Total Environ. 4, 359–363 (1975). 16. Button, M., Jenkin, G. R., Harrington, C. F. & Watts, M. J. Arsenic biotransformation in earthworms from contaminated soils. J. Environ. Monit. 11, 1484–1491 (2009). 17. Rawlins, B., O’Donnell, K. & Ingham, M. Geochemical survey of the Tamar catchment (south-west England). British Geological Survey Report. (CR/03/027, Keyworth, Nottingham, 2003). 18. Aston, S., Thornton, I., Webb, J., Milford, B. & Purves, J. Arsenic in stream sediments and waters of south west England.Sci. Total Environ. 4, 347–358 (1975). 19. Rieuwerts, J. S., Searle, P. & Buck, R. Bioaccessible arsenic in the home environment in southwest England. Sci. Total Environ. 371, 89–98 (2006). 20. Culbard, E. & Johnson, L. Elevated arsenic concentrations in house dusts located in a mineralised area of south-west England: implication for human health. Trace Substances in Environmental Health. XVIII. University of Missouri, Columbia, 311-319 (1984). 21. Private Water Supplies Regulations (2009). Applying in England and coming into force on 1st of January 2010. Available at: http:// www.legislation.gov.uk/uksi/2009/3101/introduction/made. (Accessed 12th April 2015) 22. Ander, E. et al. Variability in the chemistry of private drinking water supplies and the impact of domestic treatment systems on water quality. Environ. Geochem. Health, 1–20, doi: 10.1007/s10653-016-9798-0 (2016). 23. Water Supply (Water Quality) Regulations 2000 (Amendment) Regulations (2007). Coming into force on 22nd December 2007. Available at: http://www.legislation.gov.uk/uksi/2007/2734/introduction/made. (Accessed 12th April 2015) 24. Metcalf, S. W. & Orloff, K. G. Biomarkers of Exposure in Community Settings. J. Toxicol. Environ. Health, Part A 67, 715–726 (2004). 25. Orloff, K., Mistry, K. & Metcalf, S. Biomonitoring for environmental exposures to arsenic. J. Toxicol. Environ. Health, Part B 12, 509–524 (2009). 26. Vahter, M. Mechanisms of arsenic biotransformation. Toxicology 181–182, 211–217 (2002). 27. Wang, Q. Q., Thomas, D. J. & Naranmandura, H. Importance of Being Thiomethylated: Formation, Fate, and Effects of Methylated Thioarsenicals.Chem. Res. Toxicol. 28, 281–289 (2015). 28. Cullen, W. R. Chemical mechanism of arsenic biomethylation. Chem. Res. Toxicol. 27, 457–461 (2014). 29. Leese, E., Morton, J., Tan, E., Gardiner, P. H. & Carolan, V. A. μ LC–ICP-MS Determinations of Unexposed UK Urinary Arsenic Speciation Reference Values. J. Anal. Toxicol. 38, 24–30 (2014). 30. Buchet, J. P., Lauwerys, R. & Roels, H. Comparison of the urinary excretion of arsenic metabolites after a single oral dose of sodium arsenite, monomethylarsonate, or dimethylarsinate in man. Int. Arch. Occup. Environ. Health. 48, 71–79 (1981). 31. Brima, E. I. et al. Understanding arsenic metabolism through a comparative study of arsenic levels in the urine, hair and fingernails of healthy volunteers from three unexposed ethnic groups in the United Kingdom. Toxicol. Appl. Pharmacol. 216, 122–130 (2006). 32. Gilbert-Diamond, D. et al. Rice consumption contributes to arsenic exposure in US women. Proc. Natl. Acad. Sci. USA 108, 20656–20660 (2011). 33. Cascio, C. et al. The impact of a rice based diet on urinary arsenic. J. Environ. Monit. 13, 257–265 (2011). 34. Kim, Yangho & Byung-Kook Lee. “Association between urinary arsenic and diabetes mellitus in the Korean general population according to KNHANES 2008.” Sci. Total Environ. 409.19, 4054-4062 (2011). 35. García-Esquinas, E. et al. Arsenic exposure and cancer mortality in a US-based prospective cohort: the Strong Heart Study. Cancer Epidemiol. Biomarkers Prev. 22, 1944–1953 (2013). 36. Banerjee, M. et al. High arsenic in rice is associated with elevated genotoxic effects in humans. Sci. Rep. 3, 2195 (2013). 37. Johnson, L. R. & Farmer, J. G. Urinary arsenic concentrations and speciation in Cornwall residents. Environ. Geochem. Health 11, 39–44 (1989). 38. Kavanagh, P. et al. Urinary arsenic concentration in a high arsenic area of south west England. Occup. Environ. Med. 54, 840 (1997). 39. Kaise, T., Watanabe, S. & Itoh, K. The acute toxicity of arsenobetaine. Chemosphere 14, 1327–1332 (1985). 40. Aylward, L. L. et al. Sources of variability in biomarker concentrations. J. Toxicol. Environ. Health Part B. 17, 45–61 (2014). 41. Barr, D. B. et al. Urinary creatinine concentrations in the US population: implications for urinary biologic monitoring measurements. Environ. Health Perspect. 113, 192 (2005). 42. Baxmann, A. C. et al. Influence of muscle mass and physical activity on serum and urinary creatinine and serum cystatin C. Clin J Am Soc Nephrol. 3, 348–354 (2008). 43. Nermell, B. et al. Urinary arsenic concentration adjustment factors and malnutrition. Environ. Res. 106, 212–218 (2008). 44. Basu, A. et al. Creatinine, diet, micronutrients, and arsenic methylation in West Bengal, India. Environ. Health Perspect. 119, 1308 (2011). 45. Imran, S., Eva, G., Christopher, S., Flynn, E. & Henner, D. Is specific gravity a good estimate of urine osmolality? J. Clin. Lab. Anal. 24, 426–430 (2010). 46. Dossin, O., Germain, C. & Braun, J. P. Comparison of the Techniques of Evaluation of Urine Dilution/Concentration in the Dog. J Vet Med A. 50, 322–325 (2003). 47. Leech, S. & Penney, M. Correlation of specific gravity and osmolality of urine in neonates and adults. Arch Dis Child. 62, 671–673 (1987). 48. WHO. Arsenic in drinking-water-Background document for development of WHO Guidelines for Drinking-water Quality, World Helath Organisation, WHO/SDE/WSH/03.04/75/Rev/1. (2011).

Scientific Reports | 6:25656 | DOI: 10.1038/srep25656 10 www.nature.com/scientificreports/

49. Samanta, G. et al. Flow injection hydride generation atomic absorption spectrometry for determination of arsenic in water and biological samples from arsenic-affected districts of West Bengal, India, and Bangladesh. Microchem. J. 62, 174–191 (1999). 50. Lindberg, A.-L. et al. Arsenic exposure in Hungary, Romania and Slovakia. J. Environ. Monit. 8, 203–208 (2006). 51. Kavanagh, P. et al. Urinary arsenic species in Devon and Cornwall residents, UK. A pilot study†. Analyst 123, 27–29 (1998). 52. Chowdhury, U. K. et al. Pattern of excretion of arsenic compounds [arsenite, arsenate, MMA (V), DMA (V)] in urine of children compared to adults from an arsenic exposed area in Bangladesh. J. Environ. Sci. Health. Part A, 38, 87–113 (2003). 53. ATSDR. Agency for Toxic Substances and Disease Registry (ATSDR). Toxicological profile for arsenic. (2007) Available at: http:// www.atsdr.cdc.gov/toxprofiles/tp.asp? id = 22&tid = 3. (Accessed: 30th August 2014). 54. ACGIH. American Conference of Government Industrial Hygienists (ACGIH). Documentation of biological exposure indices. 7th edition. Cincinnati (OH): ACGIH Worldwide; 2001 (2001). 55. Spayd, S. E., Robson, M. G. & Buckley, B. T. Whole-house arsenic water treatment provided more effective arsenic exposure reduction than point-of-use water treatment at New Jersey homes with arsenic in well water. Sci. Total Environ. 505, 1361–1369 (2015). 56. Watts, M. et al. Arsenic speciation in polychaetes (Annelida) and sediments from the intertidal mudflat of Sundarban mangrove wetland, India. Environ. Geochem. Health 35, 13–25 (2013). 57. Button, M., Jenkin, G. R., Harrington, C. F. & Watts, M. J. Human toenails as a biomarker of exposure to elevated environmental arsenic. J. Environ. Monit. 11, 610–617 (2009). 58. Watts, M. J., Button, M., Brewer, T. S., Jenkin, G. R. & Harrington, C. F. Quantitative arsenic speciation in two species of earthworms from a former mine site. J. Environ. Monit. 10, 753–759 (2008). 59. Kligerman, A. D. et al. Methylated trivalent arsenicals as candidate ultimate genotoxic forms of arsenic: induction of chromosomal mutations but not gene mutations. Environ. Mol. Mutagen. 42, 192–205 (2003). 60. Cone, E. J. et al. Normalization of urinary drug concentrations with specific gravity and creatinine. J. Anal. Toxicol. 33, 1–7 (2009). 61. Levine, L. & Fahy, J. P. Evaluation of urinary lead concentrations. I. The significance of the specific gravity. J Ind Hyg Toxicol. 27, 217–223 (1945). 62. R Core Team. R: A Language and Environment for Statistical Computing, R Foundation for Statistical Computing, Vienna, Austria. (2013) Available at: www.R-project.org. (Accessed: 17th April 2015) 63. Revelle, W. psych: Procedures for Psychological, Psychometric, and Personality Research, Northwestern University, Evanston, Illinois. (2014) Available at: http://CRAN.R-project.org/package = psych. (Accessed:17th April 2015). Acknowledgements The authors gratefully acknowledge the contributions of Andrew Dunne and Dr Andrew Marriott during the field work and sample collection process and the extensive efforts of Amy Rimell and Dr Mike Studden in their invaluable contribution to logistical and management operations of the project. Thanks are given to Dr David Gardner and Dr Nigel Kendall of the Nottingham School of Veterinary Medicine and Science for the use of laboratory facilities and equipment to measure urinary osmolality. Funding for this research was provided by the Natural Environment Research Council (NERC) via a University of Manchester/British Geological Survey University Funding Initiative (BUFI) PhD studentship (Contract No. GA/125/017, BUFI Ref: S204.2). The study team is grateful for the participation of the 215 volunteers who took part in the wider study of which more information can be found at www.bgs.ac.uk/research/highlights/2013/arsenicSW.html. Author Contributions The study was conceived by M.J.W., G.S.L., R.M.C. and D.A.P.; and largely managed and executed by D.R.S.M. under the supervision of M.J.W., G.S.L., T.F. and D.A.P. D.R.S.M. and E.M.H. coordinated field sampling activities and produced the data. E.L.A. assisted in the set-up of the project database architecture. R.M.C. and H.C. coordinated public health communication activities. All authors contributed to the design of the study and to writing and/or review of the paper. Additional Information Supplementary information accompanies this paper at http://www.nature.com/srep Competing financial interests: The authors declare no competing financial interests. How to cite this article: Middleton, D. R. S. et al. Urinary arsenic profiles reveal substantial exposures to inorganic arsenic from private drinking water supplies in Cornwall, UK. Sci. Rep. 6, 25656; doi: 10.1038/ srep25656 (2016). This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Scientific Reports | 6:25656 | DOI: 10.1038/srep25656 11 Urinary arsenic profiles reveal exposures to inorganic arsenic from private drinking water supplies in Cornwall, UK

Middleton, D.R.S.1,2,3, Watts, M.J.2, Hamilton, E.M.2, Ander, E.L.2, Close, R.M.3, Exley, K.S.3, Crabbe, H.3, Leonardi, G.S.3, Fletcher, T.3 and Polya, D.A.1

Supplementary Information

Figure S1 |ONS Rural-urban classification 2011 (RUC11) of study area and households. © Crown copyright and database rights 2015 Ordnance Survey 100019153. Compiled using ArcMap 10.1.

RUC11 Number of households (%) Urban city and town 3 (2%) Rural town and fringe 11 (9%) Rural village and dispersed 96 (76%) Rural village and dispersed in a sparse setting 17 (13%)

Table S1 | Households categorised by ONS RUC11. As 76% and 13% of households were classified as ‘Rural village and dispersed’ and ‘Rural village and dispersed in a sparse setting’ respectively, age-group population statistics for these two categories were used to compare against the study group population.

Source data from the Office for National Statistics licensed under the Open Government Licence v.3.0.

12

CHAPTER 4

Publication:

Assessing urinary flow rate, creatinine, osmolality and other hydration adjustment methods for urinary biomonitoring using NHANES arsenic, iodine, lead and cadmium data

Journal:

Environmental Health

Status as of thesis submission:

Resubmitted following reviewers’ comments

93

Assessing urinary flow rate, creatinine, osmolality and other hydration adjustment methods for urinary biomonitoring using NHANES arsenic, iodine, lead and cadmium data

Daniel R. S. Middleton1, 2, Michael J. Watts2, R. Murray Lark2, Chris J. Milne2 and David A.

Polya1

1School of Earth, Atmospheric and Environmental Sciences & William Research Centre for

Molecular Environmental Science, University of Manchester, Oxford Rd, Manchester, UK,

M13 9PL

2Inorganic Geochemistry, Centre for Environmental Geochemistry, British Geological Survey,

Nicker Hill, Keyworth, Nottinghamshire, UK, NG12 5GG

Address correspondence to D. A. Polya, School of Earth, Atmospheric and Environmental

Sciences & William Research Centre for Molecular Environmental Science, University of

Manchester, Oxford Rd, Manchester, UK, M13 9PL. Telephone: (+44) 161 275 3818. E-mail: [email protected].

Abstract

Background

There are numerous methods for adjusting measured concentrations of urinary biomarkers for hydration variation. Few studies use objective criteria to quantify the relative performance of these methods. Our aim was to compare the performance of existing methods for adjusting urinary biomarkers for hydration variation.

1

Methods

Creatinine, osmolality, excretion rate (ER), bodyweight adjusted ER (ERBW) and empirical analyte-specific urinary flow rate (UFR) adjustment methods on spot urinary concentrations of lead (Pb), cadmium (Cd), non-arsenobetaine arsenic (AsIMM) and iodine (I) from the US

National Health and Nutrition Examination Survey (NHANES) (2009-2010 and 2011-2012) were evaluated. The data were divided into a training dataset (n=1723) from which empirical adjustment coefficients were derived and a testing dataset (n=428) on which quantification of the performance of the adjustment methods was done by calculating, primarily, the correlation of the adjusted parameter with UFR, with lower correlations indicating better performance and, secondarily, the correlation of the adjusted parameters with blood analyte concentrations (Pb and Cd), with higher correlations indicating better performance.

Results

Overall performance across analytes was better for Osmolality and UFR based methods.

Excretion rate and ERBW consistently performed worse, often no better than unadjusted concentrations.

Conclusions

Osmolality adjustment of urinary biomonitoring data provides for more robust adjustment than either creatinine based or ER or ERBW methods, the latter two of which tend to overcompensate for UFR. Modified UFR methods perform significantly better than all but osmolality in removing hydration variation, but depend on the accuracy of UFR calculations.

Hydration adjustment performance is analyte specific and further research is needed to establish a robust and consistent framework.

2

Keywords: biomonitoring, hydration adjustment, creatinine, osmolality, urinary flow rate,

NHANES

Background

Urinary biomonitoring is the preferred method of exposure and nutritional assessment for many chemical elements and metabolites given its non-invasiveness, logistical appeal and ease of measurement with modern analytical techniques [1]. This is true for potentially harmful elements like arsenic (As) [2], essential nutrients like iodine (I) [3] and drug and organic compounds [4, 5], making the meaningful interpretation of urinary data a requirement with implications for public health, occupational health and forensic applications. The US National

Health and Nutrition Examination Survey (NHANES) has proved an invaluable, growing resource of chemical biomonitoring data [6], but the value of such data depend on their correct interpretation [7], challenges with which are currently limiting the full potential of urinary chemical biomarkers [6].

Urinary analyte concentrations are susceptible to variation from factors extending beyond exposure and are categorised [1] as follows: (i) time of sampling relative to exposure; (ii) inter- individual toxico-kinetic factors and (iii) physiological characteristics of the biomonitoring matrix. While the first two factors should not be ignored, the third, specifically the variation in dilution among spot urine samples, is addressed here.

While collection of 24 hr urine samples is preferred, it is not feasible for large biomonitoring studies due to resource limitations, cumbersome sample nature and volunteer compliance issues [8]. First morning void (FMV) or spot collections are common substitutes, but are limited in that they reflect the hydration status of the individual at the time of collection thus may differ markedly in dilution as a result of differences in urinary flow rate (UFR). Spot/FMV samples are nevertheless widely deemed acceptable provided that the effect of sample dilution

3 is quantified and appropriately adjusted [9]. Several methods for adjusting spot/FMV data are currently employed but there is no consensus on which is the most appropriate.

The most common technique employed is creatinine adjustment, whereby urinary analyte concentrations are ratioed to creatinine concentrations. This method implicitly assumes that urinary creatinine is excreted at a constant rate and varies only as a function of UFR. However, these assumptions are of questionable validity, since creatinine concentrations have been shown to depend upon all of demographic group [10], protein intake [11], muscle mass [12] and malnutrition [13].

Alternative methods such as specific gravity (SG) or osmolality adjustment are commonly reported. Close agreement has been demonstrated between these methods [14] but osmolality, as measured by osmometry, has been described as the definitive measure of urinary concentration [15] despite being previously considered prohibitively expensive [16].

Osmometry is not susceptible to the same interferences as SG, conventionally measured by refractometry, which may be confounded in subjects with, for example, proteinuria and glucosuria. Urinary osmolality was a post-2008 inclusion in NHANES and, while similar factors affecting creatinine excretion were found to be responsible for variation in urinary osmolality, less influence was observed on osmolality than on creatinine in the US population

[8].

Creatinine, SG and osmolality are all surrogate estimators of UFR with various degrees of effectiveness [17, 18]. Direct UFR measurements are now included in post-2008 NHANES cycles, with UFR determined as follows:

UFR = V/t, [1] where t is the time elapsed between two urine voids and V is the volume of the second void.

Excretion rates (ER), typically expressed in ng/hr, of the analyte can be calculated:

4

ER = (Cvol×V)/t, [2]

where Cvol is the urinary analyte concentration, typically in nanograms per millilitre, V is volume, typically in millilitres and t is time, typically in hours. Additionally, bodyweight

(BW) adjusted ER (ERBW), typically expressed in ng/kg-hr, can be calculated as follows:

ERBW = (Cvol × V)/(t × BW). [3]

Adjusting urinary biomonitoring results according to Equations 2 and 3 was recently proposed to directly account for hydration status and address demographic variations in UFR more effectively than creatinine and osmolality adjustments [19]. While UFR, providing accurate measurement of time and volume, is more reflective of hydration status than surrogate measures, such as creatinine and osmolality, its application in equations 2 and 3 directly incorporates hydration bias into results. Given that ERs can still apply to restricted time periods, they are dependent on UFR, i.e. hydration, at that time. Strong positive Spearman’s correlation coefficients (rs) have been reported between ERs of urinary analytes and UFR [20,

21], indicating that the adjustment in Equation 2 is not theoretically robust.

Specifically, application of Equation 2 implicitly assumes that analyte concentrations vary inversely proportionally with UFR. However, this was disputed by Araki et al. (1990), who observed analyte-specific, log-linear relationships between analyte concentrations and UFR of the form:

log Cvol = a-b log UFR, [4] where a and b (referred to here as Araki’s b value) are analyte-dependent, empirically determined regression coefficients.

Araki et al. (1990) therefore proposed a modified UFR adjustment whereby analyte concentrations were adjusted to a standard UFR of 1 mL/min:

5

b CUFR-adj = Cvol × UFR , [5] where Araki’s b values were derived for a number of analytes using multiple voids from single individuals subjected to water loading and water restrictive conditions. This UFR adjustment was found to be more effective than ER, creatinine and SG adjustment in removing UFR- dependent variation from adjusted urinary analyte concentrations [20, 21]. For datasets, such as NHANES, that do not contain extensive analyte concentration data for multiple voids from single individuals, a previously reported iterative method [22] may be used to calculate population-level Araki’s b values by optimising appropriate performance criteria.

Comparing the performance of urinary hydration adjustment methods requires assessment criteria appropriate to the needs of a given study. Suggested criteria are summarised in Table

1.

Table 1 Suggested criteria for assessing performance of urinary biomonitoring adjustment methods. Criteria A and B are used in this study.

Criterion Description Performance metric A Correlation between adjusted spot analyte Weaker correlations concentrations and UFR. indicating good performance. B Correlation between adjusted spot analyte Stronger correlations concentration and an independent measure of indicating good internal dose, e.g. analyte concentration in performance. blood. C Correlation of spot analyte concentrations with Closer agreement/lower analyte excretion over 24 hr or composite 24 hr variation in spot samples concentrations. indicating good performance. D Correlation of spot analyte concentration with Stronger correlations an independent measure of/proxy for external indicating good exposure e.g. drinking water analyte performance. concentration.

For this work, the primary assessment criterion was the extent of removal of systematic dependence on UFR of adjusted urinary analyte concentrations (Criterion A, Table 1) as has been used previously [20, 22, 23]. Additionally, since blood biomonitoring is the preferred

6 measure of exposure for several chemicals, notably lead (Pb), and is not as susceptible to the same level of variation as urinary concentrations [24], the correlation between adjusted urinary concentrations and blood concentrations was also used as a secondary assessment criterion

(Criterion B, Table 1) and has been previously explored [25] for Pb. Similarly, in the case of cadmium (Cd), blood is a biomarker of both recent and cumulative Cd exposure, and urinary concentrations reflect cumulative exposures and Cd levels in the kidney [26]. Agreement between urinary and blood Cd concentrations have been reported [27], making it reasonable to hypothesize that the effective removal of hydration variation from spot urine samples may strengthen this relationship. Other possible assessment criteria include agreement with 24 hr excretion rates or composite concentrations, used previously [14], but the lack of 24 hr data in the NHANES survey does not permit this. Lastly, an independent external measure of exposure, e.g. drinking water analyte concentrations [28] could be used as an assessment criterion, but was not used in this study because of a lack of appropriate environmental data in

NHANES.

Methods

Aims

This paper aims to compare the performance of urinary biomonitoring hydration adjustment techniques using NHANES (2009-2010, 2011-2012) spot urinary concentrations of selected chemical analytes and, in particular, test whether or not analyte-specific UFR adjusted concentrations provide a more robust adjustment than creatinine, osmolality, ER or ERBW adjustments. Arsenic and iodine were selected as chemicals on which to make these tests as respectively toxic and essential elements for which urinary biomonitoring is widely used.

Additionally, Pb and Cd were selected for study, due to the availability of paired urine-blood samples in the NHANES database and the applicability of criterion B to these elements. This

7 provides the opportunity to compare the adjustment performance characteristics of two independent assessment criteria.

Data acquisition

Data from the NHANES 2009-2010 and 2011-2012 surveys were acquired from the NHANES website [29]. Volunteer consent information and dataset access can be found online: http://www.cdc.gov/nchs/nhanes.htm. Data on demographic variables; body measurements; standard biochemistry profile; diabetes; kidney conditions; plasma fasting glucose; urinary flow rates; urinary creatinine; urinary osmolality; urinary metals; total and speciation urinary

As; urinary I and blood metals were downloaded in SAS (.xpt) format. Data were converted to

MS Excel (.xlsx) format using the R programming environment SASxport and xlsx packages

[30, 31], before being matched by sequence number (SQN) in MS Access. Volunteers with data present on gender, age, bodyweight, urinary creatinine, urinary osmolality, UFR, As speciation and blood metals, of either non-Hispanic white, non-Hispanic black of Mexican

American ethnicity were initially included.

Volunteers with evidence of health conditions that could affect the performance of urinary adjustment calculations were excluded using previously published criteria [10, 19]: urinary albumin-creatinine ratio >30 mg/g creatinine was treated as albuminuria and diabetics were identified by self-reported physician diagnosis or plasma glucose ≥126 mg/dL (≥8 hr fasting) or ≥200 mg/dL (<8 hr fasting). Chronic kidney disease (CKD) was identified by self-reported physician diagnosis or an estimated glomerular filtration rate (eGFR) <60 mL/min/1.73 m2 using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation [32]. For volunteers aged <18, eGFR was estimated using the Bedside Schwartz equation [33]. Finally, volunteers without detectable concentrations of urinary Pb, Cd, Total As and I were excluded to limit the effects of censoring on analyses.

8

Additional data from a single volunteer consisting of multiple spot Cd concentrations and UFR measurements were reproduced [34] for observational purposes only.

Analytical measurements

Detailed analytical methodologies for the analytes investigated in this paper, plus other analytical components of NHANES, are reported online: http://www.cdc.gov/nchs/nhanes/nhanes2011-2012/lab_methods_11_12.htm. Urine samples were provided by volunteers at the NHANES mobile examination center (MEC). Volunteers were asked to fully void the bladder and report the last time of previously doing so. Volumes of the samples provided at the MEC were measured and used, with the previous void times reported by volunteers, to calculate UFR as per Equation 1. For volunteers with initial urinary volumes below requirement, subsequent voids were collected and composite UFRs were calculated using the total volumes and times covered by all voids. This ensured that laboratory measurements made on pooled samples consisting of multiple voids corresponded to the correct UFRs. Urinary osmolality was measured using freezing-point depression (cryoscopic) osmometry performed with an Osmette II, Model 5005 Automatic Osmometer (Precision

Systems Inc.). Urinary creatinine was determined using an enzymatic (creatininase) reaction and a Roche/Hitachi Modular P Chemistry Analyzer. Urinary total I, Pb and Cd and whole blood Pb and Cd were determined using inductively coupled plasma dynamic reaction cell mass spectrometry (ICP-DRC-MS) (PerkinElmer ELAN® 6100 DRCPlus or ELAN® DRC II).

Urinary As speciation was performed using high performance liquid chromatography (HPLC) coupled to ICP-DRC-MS. Combined urinary inorganic As and methylated metabolites (AsIMM) was calculated as the sum of arsenous acid (AsIII), arsenic acid (AsV), monomethylarsonic acid

(MMA) and dimethylarsonic acid (DMA) species, as this is the routine biomarker of As exposure and does not incorporate non-toxic arsenobetaine. Osmolality and UFR were both determined at the MEC shortly after urine collection. The remaining measurements were made

9 after urine and blood samples had been frozen at -20 °C and shipped to relevant laboratories where they remained frozen until analysis to prevent evaporation and the inter-conversion of

As species.

Urinary analyte adjustment calculations

Data were read into the R programming environment [35] and partitioned into two subsets by simple random sampling using the caret package [36]:

(1) Training dataset: 80 %, reserved for Araki’s b value derivation.

(2) Testing dataset: 20 %, for applying and assessing the performance of adjustment calculations.

The partition of 80:20 % was deemed suitable [37] and was selected to (i) ensure sufficient training data were available; (ii) retain a testing dataset of a size comparable to that of a biomonitoring study in which these kinds of adjustments may be employed and (iii) to preserve the distribution of demographic and analytical variables between both datasets.

Urinary analyte excretion rates (ER, ng/hr) and bodyweight adjusted excretion rates (ERBW, ng/kg-hr) were calculated using Equations 2 and 3, respectively [19]. Conventional creatinine- adjusted analyte concentrations were expressed in µg/g creatinine as follows:

Ccr-adj= Cvol/Ccr, [6] where Ccr is the specimen creatinine concentration in grams per litre. Osmolality adjustment was performed using an equation based on the Levine-Fahy specific gravity adjustment [38] as follows:

Cosm-adj = Cvol × (Osmref/Osmmeas), [7] where, for consistency with recent publications [19], Osmref is the median osmolality

10

(mOsm/kg) of training data volunteers (734 mOsm/kg) and Osmmeas is that measured in the individual specimen. Araki’s b values were extracted using an adaption of a previously published approach [22] which involved a simple numeric method. Pearson correlation coefficients (rp) for criteria A and criteria B were calculated from the training data for values of Araki’s b from 0 to 1.5 at intervals of 0.01. Araki’s b values that yielded optimum correlations for criteria A (minimizing absolute value of rp) and separately for criteria B

(maximizing rp) were determined. Araki’s b values were also derived on demographic subsets of specific age groups and specific genders/ethnicities to identify patterns between groups.

Optimum Araki’s b values were used to perform Araki’s modified UFR adjustment (Equation

5), henceforth referred to as UFRA and UFRB when adjusted using optimum Araki’s b values for criteria A and B, respectively. An R script has been provided in supporting information to allow other groups to derive Araki’s b values and perform hydration adjustments.

Statistical analyses

Due to the specific application of NHANES data in assessing adjustment methods rather than making inferences of biomonitoring measurements in the US population, sample weights were not incorporated into analyses. Statistical tests (and graphical presentations) were performed using R version 3.0.0 (base package) [35]. Urinary and blood analyte data were positively skewed and, therefore, geometric means (GM) were calculated as opposed to arithmetic means.

For the same reason, Pearson correlations of urinary analyte concentrations against UFR and blood analyte concentrations were calculated on natural log (ln) transformed data with significance tests (p-values) and 95 % confidence intervals (CI) using the ‘cor.test’ function.

Pearson’s, as opposed to Spearman’s, correlation was selected to prevent the loss of information that occurs when data are reduced to ranks in the process of calculating Spearman’s correlation. It was necessary to test the significance of the difference between correlations of, for example, urinary Pb with blood Pb adjusted by different methods. These correlations are

11 not independent (because of the common variable, blood Pb), so the Williams’s test [39] was performed using the r.test function in the psych package [40]. Point density contour lines were added to plots using two-dimensional kernel density estimation in the MASS package [41].

Results

Exploratory analyses – training data

Inclusion of data for volunteers with the appropriate demographic, examination and laboratory variables yielded records for 3539 individuals. This was reduced to 2668 following the exclusion of volunteers with evidence of albuminuria, diabetes or CKD. A reduction to 2151 records was made after excluding those with urinary analyte concentrations below analytical detection limits. These 2151 records were partitioned independently and at random into a training dataset of 1723 records and a testing dataset of 428 records reserved for independent adjustment comparisons. Study group characteristics, GMs and ranges of creatinine osmolality,

UFR, unadjusted urinary AsIMM, I, Pb Cd and blood Pb and Cd of training and testing datasets are shown in Table 2, demonstrating the preservation of characteristic and analyte distributions following the partitioning of the data.

12

Table 2 Demographic characteristics and unadjusted analyte geometric means (GM) and ranges for training and testing datasets. Training data Testing data Demographic group, n (%) All 1,723 428 Male 887 (51) 243 (57) Female 836 (49) 185 (43) Non-Hispanic white 841 (49) 216 (51)

Non-Hispanic black 489 (28) 116 (27) Mexican American 393 (23) 96 (22) 6-11 148 (8) 39 (9) 12-19 268 (16) 63 (15) 20-39 544 (32) 125 (29) 40-59 461 (27) 120 (28) >60 302 (17) 81 (19) Analytical measurement, GM (range) Creatinine, g/L 1.1 (0.1-8) 1.1 (0.09-5.6) UFR, mL/min 0.7 (0.03-34.5) 0.7 (0.06-5.5) Osmolality, mOsm/kg 635 (91-1,394) 630 (84-1,350) Urinary AsIMM, µg/L 6.3 (2.9-386) 6.4 (2.9-105) Urinary I, µg/L 149 (8-15,651) 161 (16.9-9,322) Urinary Pb, µg/L 0.5 (0.08-49.6) 0.5 (0.08-14.3) Urinary Cd, µg/L 0.2 (0.04-6.2) 0.2 (0.04-4.8) Blood Pb, µg/dL 1.1 (0.2-33.7) 1.1 (0.2-22) Blood Cd, µg/L 0.3 (0.1-8.7) 0.3 (0.1-4)

Training dataset log transformed urinary analytes, including creatinine, showed (Figure 1) significant (p<0.001) negative, log-linear relationships with UFR. This confirmed previous findings [20]. The weak (Pb, Cd, AsIMM and I in Figures 1A-D) to moderate (creatinine, Figure

1E) r2 values indicated that the majority of variation in analyte concentrations were not explained solely by UFR – this was most pronounced for Cd and least pronounced for creatinine. Large variations in urinary analyte concentration relative to variations in UFR would be expected to result in criterion A Araki’s b values >1, however the calculated Araki’s b values for all urinary analytes were substantially <1, indicating other controls on urinary analyte concentrations. Notably, for As, the relationship between urinary AsIMM and UFR was particularly impacted by the range of concentrations of DMA (Figure 1C). It is noteworthy that

13

Figure 1 Unadjusted urinary Pb (A), Cd (B), AsIMM (C), I (D) and creatinine (E) plotted against UFR (NHANES 2009-2012 (CDC, 2015) training data). Multiple spot Cd measurements (Meharg et al., 2014) from a single volunteer are shown for comparison (F). Linear regression lines (blue) are displayed with regression slopes and r2 values. *** denotes significance to p<0.001. Point density contours were plotted using two-dimensional kernel density estimation. In (C), the transition from green to red depicts increasing concentration of urinary dimethylarsonic acid (DMA).

14

the Araki’s b value calculated from data for a single individual [34] for Cd (Figure 1F) (0.87) is substantially different from that calculated data from multiple individuals (0.32).

Derivation of Araki’s b values – training data

Araki’s b values derived in the present study from pooled NHANES data (single voids from multiple individuals) by optimising criterion A or criterion B are presented in Table 3 along with published values (where available) from mean data derived previously [20, 21] (multiple voids from single individuals). The criterion A optimised Araki’s b values for Pb (0.38), Cd

(0.32) and creatinine (0.52) are all somewhat lower than values reported previously [20, 21].

The criterion B optimised values for Pb (0.56) and Cd (0.62) are closer to those published previously [20, 21]. We note that the Araki’s b values derived to optimise Criterion A agree with the b values that describe the slopes of the linear relationships shown in Figure 1.

Table 3 Araki’s b values derived for Pb, Cd, AsIMM and I in the present study (NHANES 2009- 2012 (CDC, 2015) training dataset) compared with previously reported literature values.

Araki’s b value

Criterion A Criterion B (UFR) (Blood) Araki et al. Araki et al. Optimised Optimised (1986)a,c (1990)b,c Analyte (present study) (present study) Pb 0.38 0.56 0.50 0.45 (0.45-0.91) (0.39-0.54) Cd 0.32 0.62 - 0.58 (0.42-0.66) AsIMM 0.27 - - - I 0.45 - - - Creatinine 0.52 - 0.87 0.68 (0.67-1.01) (0.58-0.75) a Mean value of 10 subjects derived by linear regression b Mean value of 4 time quadrants derived by linear regression c Range of significant values in parentheses

15

The sensitivity of Pearson correlations to model Araki’s b values for criteria A (Pb, Cd, AsIMM,

I) and B (Pb, Cd) are illustrated in Figure 2. In all cases, the criterion A and criterion B optimised b values are all lower than the b value (b=1) implicit in conventional ER approaches and are all better adjustments based on using Pearson correlation as the metric.

Figure 2 Sensitivity of Pearson correlations to Araki’s b value for NHANES 2009-2012 (CDC, 2015) training data for Pb (A), Cd (B), AsIMM (C) and I (D) for criterion A (urinary analyte versus UFR, blue lines) and criterion B (urinary analyte versus blood analyte, red lines) with 95 % confidence intervals (grey lines). Optimum criterion A (blue diamonds) and criterion B (red diamonds) Araki’s b values are displayed and, in the case of Pb and Cd, the difference between these values is highlighted by double-headed arrows. Single-headed arrows illustrate the improvement in criterion A (decreasing correlation) and criterion B (increasing correlation) correlations relative to the equivalent Araki’s b value implicit of ER.

16

Comparison of adjustment methods – testing data

The different adjustment methods were performed on analyte concentrations and the resulting

GM concentrations and ranges are presented in Table 4. The performance of adjustment methods for urinary analyte concentrations from the testing dataset were assessed against criterion A (Pb, Cd, AsIMM and I) (Table 5) and criterion B (Pb and Cd) (Table 6). The relative performance of these methods, using Pb as an example, is also illustrated in Figure 3 and summarized below:

Criterion A

Pb: UFRA, Osmolality > Creatinine, UFRB > Unadjusted > ERBW > ER

Cd: UFRA, Osmolality > Creatinine, UFRB > Unadjusted > ERBW > ER

AsIMM: UFRA > Osmolality > Creatinine > Unadjusted > ERBW > ER

I: UFRA > Osmolality > Creatinine > ERBW > Unadjusted > ER

Criterion B

Pb: Osmolality ≥ Creatinine ≥ UFRB ≥ UFRA > ER ≥ Unadjusted ≥ ERBW

Cd: Osmolality ≥ Creatinine ≥ UFRB ≥ UFRA ≥ ER ≥ Unadjusted ≥ ERBW

17

Table 4 Geometric means (GM) and ranges of urinary analytes following adjustment by various methods (testing dataset). Urinary Unadjusted, Creatinine- Osmolality- ER, ng/hr ERBW, ng/hr- UFRA, µg/L, UFRB, µg/L, analyte, GM µg/L adjusted, µg/g adjusted, µg/L, kg UFR 1 mL/min UFR 1 mL/min

(range) creatinine 734 mOsm/kg

AsIMM 6.4 5.7 7.5 279 3.9 5.9 -

(2.9-105) (1.3-55.6) (2-82.4) (18.5-3,921) (0.2-64.8) (2.2-92.4)

I 161 143 188 7,024 97.5 140 -

(16.9-9,322) (23.3-4,615) (28.6-8,661) (558-352,368) (9.6-4,199) (24.2-7,572)

Pb 0.5 0.5 0.6 22.4 0.3 0.5 0.4

(0.08-14.3) (0.06-16.1) (0.06-17.5) (0.8-1,042) (0.008-13.9) (0.04-15.4) (0.03-15.9)

Cd 0.2 0.2 0.25 9.2 0.1 0.2 0.2

(0.04-4.8) (0.03-2.6) (0.03-4.9) (0.5-97.5) (0.006-1.5) (0.02-3) (0.02-2.2)

18

Table 5 Pearson correlations for performance Criterion A across the range of adjustment methods investigated for NHANES 2009-2012 (CDC, 2015) (testing dataset). Correlations were calculated on natural log transformed data. ***, **, and * denote significance to p<0.001, <0.01 and <0.05, respectively. Bold font denotes the best performing adjustment method for each analyte. Correlations share a letter when not significantly different from one another.

Adjustment method rp (95 % CI) Pb Cd AsIMM I Unadjusted -0.33*** -0.25*** -0.37*** -0.39*** (-0.41, -0.24) (-0.34, -0.26) (-0.45, -0.28) (-0.47, -0.31) Creatinine 0.18*** b 0.18*** b 0.32*** 0.09 (0.09, 0.27) (0.09, 0.27) (0.24-0.41) (-0.001, 0.19) Osmolality -0.001 a 0.02 a 0.10* -0.09 (-0.10, 0.09) (-0.07, 0.12) (0.003-0.19) (-0.18, 0.004) ER 0.52*** 0.47*** 0.70*** 0.45*** (0.45, 0.59) (0.39, 0.54) (0.65, 0.74) (0.37, 052) ERBW 0.43*** 0.42*** 0.60*** 0.35*** (0.35, 0.50) (0.34, 0.50) (0.53, 0.66) (0.26, 0.43) UFRA 0.01 a -0.01 a -0.03 -0.01 (-0.08, 0.11) (-0.10, 0.09) (-0.12, 0.07) (-0.10, 0.09) UFRB 0.18*** b 0.22*** b (0.09, 0.27) (0.13, 0.31)

19

Table 6 Pearson correlations for performance Criterion B across the range of adjustment methods investigated for NHANES 2009-2012 (CDC, 2015) (testing dataset). Correlations were calculated on natural log transformed data. All correlations are significance to p<0.001. Bold font denotes the best performing adjustment method for each analyte. Correlations share a letter when not significantly different from one another.

Adjustment method rp (95 % CI) Pb Cd Unadjusted 0.67 de 0.58 d (0.61, 0.72) (0.51, 0.64) Creatinine 0.79 ab 0.65 ab (0.75, 0.82) (0.59, 0.70) Osmolality 0.81 a 0.66 a (0.78, 0.84) (0.60, 0.71) ER 0.69 d 0.59 dc (0.63,0.73) (0.52, 0.64) ERBW 0.63 e 0.57 d (0.57, 0.68) (0.50, 0.63) UFRA 0.74 c 0.62 bc (0.70, 0.78) (0.56, 0.67) UFRB 0.75 bc 0.62 b (0.70, 0.79) (0.56, 0.68)

20

Figure 3 Scatterplots of unadjusted (A, B); creatinine adjusted (C, D); osmolality adjusted (E, F); ER adjusted (G, H); ERBW adjusted (I, J); UFR adjusted with Araki’s b optimised to criterion A (K, L) and criterion B (M, N) urinary Pb concentrations vs UFR (criterion A) (A, C, E, G, I, K, M) or blood Pb concentrations (criterion B) (B, D, F, H, J, L, N). Data: NHANES 2009-2012 (CDC, 2015) testing dataset. ***, **, and * denote significance to p<0.001, 0.01 and 0.05, respectively.

21

Irrespective of whether criterion A or criterion B is used to assess adjustment method performance, it is evident that UFRA, UFRB, creatinine and osmolality adjustment methods all provided for a statistically significant improvement relative to unadjusted analyte concentrations. Of these, osmolality adjustment was determined to be the optimal adjustment method except for AsIMM and I for which UFRA showed a marginally better performance. The criterion A-based performance of osmolality and UFRA adjustments were equally good for Pb and Cd. Indeed, osmolality adjustment resulted in a weak (rp=0.10) significant (p<0.05) correlation only in the case of osmolality adjusted AsIMM versus UFR. Creatinine adjustment, in contrast, yielded significant positive correlations with UFR for Pb (rp=0.18), Cd (rp=0.18)

IMM and As (rp=0.32). Iodine was an exception, with no significant correlation of creatinine adjusted I concentrations against UFR.

Excretion rate adjustment methods (ER, ERBW) performed worse than any of UFRA, UFRB, creatinine or osmolality adjustments according to both criteria. Furthermore, although ER and

ERBW adjustments removed observed negative correlations of unadjusted analyte concentrations with UFR, they mostly resulted in positive correlations of an equal or greater magnitude. Excretion rate and ERBW adjustments thus performed no better than implementing no adjustment at all, with the sole exception of I, for which ERBW adjustment performed marginally better (rp=0.35 cf. -0.39).

An exercise was undertaken to derive Araki’s b values for specific demographic groups of the data. The values derived for specific genders and ethnicities and specific age groups are presented in Table 7. Optimum Araki’s b values were plotted against age group for Criterion

A (Figure 4A) and Criterion B (Figure 4B). Large differences in optimum b values were observed between different genders and ethnicities across the range of analytes but no obvious patterns were observed.

22

Table 7 Araki’s b values derived for AsIMM, I, Pb and Cd on specific demographic sub-groups of the present study group (training dataset). In the case of Criterion A, ***, **, and * denote the significance of the relevant UFR-analyte regression slopes to p<0.001, 0.01 and 0.05, respectively.

Demographic group n Optimum Araki’s b value Criterion A Criterion A Criterion A Criterion A Criterion B Criterion B AsIMM I Pb Cd Pb Cd Non-Hispanic white male 416 0.28*** 0.50*** 0.42*** 0.31*** 0.68 0.96 Non-Hispanic white female 425 0.33*** 0.58*** 0.50*** 0.46*** 0.54 0.63 Non-Hispanic black male 265 0.27*** 0.51*** 0.46*** 0.29*** 0.56 0.53 Non-Hispanic black female 224 0.24*** 0.32*** 0.24*** 0.34*** 0.44 0.53 Mexican American male 206 0.11* 0.30*** 0.10 0.02 0.26 0.18 Mexican American female 187 0.34*** 0.40*** 0.46*** 0.37*** 0.58 0.56 6-11 148 0.14* 0.28*** 0.38*** 0.08 0.10 1.38 12-19 268 0.22*** 0.34*** 0.23*** 0.21*** 0.38 0.24 20-39 544 0.27*** 0.41*** 0.39*** 0.46*** 0.58 0.82 40-59 461 0.36*** 0.48*** 0.46*** 0.58*** 0.50 0.61 >60 302 0.33*** 0.55*** 0.60*** 0.66*** 0.70 0.79

23

Figure 4 Optimum Criterion A (A) and Criterion B (B) Araki’s b values derived for different age groups for the range of analytes investigated.

For example, the optimum Araki’s b values for non-Hispanic white males were all lower than non-Hispanic white females for Criterion A but higher for Criterion B. For non-Hispanic black males, Araki’s b values were generally higher for both Criteria than for females. There was a general increase in optimum Araki’s b values with increasing age group across the range of analytes for both Criteria. In the case of Criterion A, for both gender/ethnicity and age groups, we attributed the difference in b values to the difference in group sizes. This is evident from the results presented in Table 7, where we also show the significance of the relationship between UFR and analyte concentrations. These slopes are synonymous with the optimum

Criterion A values presented in Table 7 and, in smaller groups, some of the slopes are not

24 significant. Nevertheless, Criterion B is independent of the relationship between UFR and analyte and, therefore, we chose to pursue the investigation of age-specific Araki’s b values for

Criterion B. When UFRB adjustment was performed using b values specific to the volunteers’ age groups, no improvement in Criterion B correlations were observed relative to adjustment with a single group-wide b value (Pb: 0.73 versus 0.75; Cd: 0.60 versus 0.62).

Discussion

Osmolality and UFRA adjustment methods provided the best or near best performance of the adjustment methods tested using criterion A. Osmolality and creatinine adjusted concentrations yielded the strongest correlations using criterion B. In its nature, UFRA is tailored to optimise criterion A performance, however, osmolality and creatinine methods tended to perform better than Araki’s UFR-based adjustment methods with respect to criterion B. This may reflect greater uncertainties in volunteer reported times of initial voids than in uncertainties in objectively measured UFR surrogates such as osmolality or creatinine. The issue of reliance on the accuracy of volunteer reported void times when calculating UFR in NHANES has been raised previously [19]. We note this as a limitation of the present study, in that no efforts were made to refine UFR data or quantify their accuracy.

Alternatively, differences in Araki’s b values for a given analyte, between individuals of different ages, genders and ethnicities may partly account for deficiencies in UFR based adjustments. Attempts were made to derive Araki’s b values on demographic subsets of the training dataset. Differences in optimum values were observed between groups but this was possibly an artefact of different group sizes. When age-specific b values were implemented to the adjustment of urinary Pb and Cd, no significant improvements were observed in Criterion

B correlations. Future efforts should be made to derive Araki’s b values for multiple individuals, collecting multiple voids at various states of hydration, to more closely represent

25 the relationship between analyte concentrations and UFR, as illustrated in Figure 1F. As proposed previously [23], these derivations should be made on specific demographic groups to investigate whether the relationship between analyte concentrations and UFR vary in a characteristic manner, something which was not achievable given the constraints of the data utilised in the present study.

The finding that Araki’s b values calculated here by the UFRA method are generally lower than those calculated previously [20, 21], might give rise to questioning of the validity of the

UFRA based values. We therefore identify a potential flaw in the validity of criterion A when using the present dataset. As discussed, it was not possible to derive Araki’s b values in the conventional manner using the NHANES dataset. This requires multiple voids from single volunteers at different hydration states, circumstances under which the ‘true’ relationship between UFR and analyte concentration is observed due to the relatively constant internal dose of a given analyte, for a single individual, over the timescale investigated. NHANES data consist of single voids from multiple individuals with greater inter-individual ranges of internal doses which have the potential to alter the observed slope (b value). Under circumstances where the standard deviation of the distribution of internal dose is considerably bigger than that of the distribution of UFR for the studied population, the calculated Araki’s b value may be positively biased. This in-turn determines the optimum value for criterion A - the b value that describes the slope between UFR and analyte – evident in the agreement between the b values that describe the slopes presented in Figure 1 and those derived using the numeric method. This is further illustrated by the difference between criteria A and B optimum values for Pb and Cd

(Figure 2A and B), the criterion B optimums could be considered more robust for the NHANES dataset used here as they are independent of UFR. We note, however, that this paper does not purport to suggest specific Araki’s b values for application elsewhere, but is successful in reiterating proof of concept of their necessary implementation to remove hydration variation

26 from spot analyte concentrations and, in doing so (Figure 2A and B), better reflect internal dose.

Our findings reiterate the analyte-specific nature of hydration adjustment, exemplified by the difference in criteria A correlations between creatinine adjusted AsIMM and I. Creatinine adjusted I concentrations yielded no significant correlation with UFR, whereas creatinine adjusted AsIMM concentrations were more strongly correlated with UFR than I, Pb and Cd.

Inspection of Figure 1C indicates that differing relationships with UFR of urinary AsIMM for high-DMA and low-DMA samples – this suggests that the biochemistry of different species of the same element (and different elements) influence Araki’s b values.

As noted previously [19], for studies outside the NHANES framework or similar population- scale biomonitoring designs, the collection of UFRs may provide additional value providing that accurate recordings of time and volume are obtained. It is recognised that this may not be logistically feasible for all studies and surrogates such as creatinine and osmolality are attractive alternatives. Modifications of these surrogate based adjustments have also been explored [22, 23, 42] using methodologies based on the work of Araki et al. (1986). These approaches, such as modified SG [23] or modified creatinine [25] adjustment, were not addressed in the present study and further exploration of such alternatives may prove valuable for studies that are restricted in their ability to directly measure UFR, particularly in low-budget circumstances or developing countries. Similarly, measurements of additional urinary constituents that are indicative of medical conditions, such as glucose, protein, ketones and bilirubin, could have provided additional data exclusion criteria had they been available to us.

A comparison of the performance of different adjustment methods between volunteers with and without the presence of such analytes, and the medical conditions that were available as exclusion criteria (e.g. CKD and diabetes) will make for an important matter of further research.

27

The implications of the findings made in this investigation, and the questions that remain unanswered, have implications for environmental and epidemiological studies using urinary biomonitoring to assess human exposures and investigate the dose-response relationships between environmental chemicals and health end-points. The differences in biomarker levels yielded by different adjustment methods is evident (Table 4). This impacts the interpretation of results when making comparisons to existing guidance or reference values. Furthermore, it impacts the derivation of reference values themselves. For example, a large and much needed body of work has been undertaken to derive biomonitoring equivalents to be used in comparison with various urinary biomarkers [43]. Some derivations have utilised creatinine adjustment, which may have limited their applicability to studies using alternative adjustment methods, or other studies using creatinine adjustment with different demographic structures.

This problem also extends to studies that explore relationships between urinary analyte concentrations and health outcomes- a widely used application of NHANES data [44-46]. A robust, standardised framework of urinary hydration adjustment is warranted to ensure the validity and sensitivity of such analyses.

Conclusions

We have demonstrated for the urinary analytes studied (Pb; Cd; AsIMM and I), and the adjustment methods considered (osmolality; creatinine; excretion rate (ER); body weight adjusted excretion rate (ERBW); urinary flow rate adjustment with Araki’s b optimised to minimise correlation with urinary flow rate (UFRA) and urinary flow rate adjustment with

Araki’s b optimised to maximise correlation with blood analyte concentrations (UFRB):

(i) Osmolality consistently performs as the best or near best adjustment method against

two performance criteria: minimum correlation of adjusted urinary analyte

28

concentration with UFR; maximum correlation of adjusted urinary analyte

concentration with blood analyte concentration.

(ii) The method of Araki et al. (1986, 1990) for objectively determining Araki’s b

values to adjust urinary analyte concentrations also performs well, but is limited by

the requirement for accurately determined UFR data and age/gender/ethnicity

specific b values.

(iii) Creatinine adjustment methods may be suitable for some analytes (e.g. I) that have

similar b values to creatinine, but can otherwise result in significant biases.

(iv) ER and ERBW based adjustment methods are shown here to overcompensate for

UFR and invariable performed worse than osmolality, creatinine, UFRA, UFRB

and often worse than unadjusted concentrations.

Thus, we demonstrate that conventional application of UFR is limiting the full potential of this metric. The under-performance of both ER and ERBW in relation to two independent performance criteria support previous findings [20, 21] that using UFR to calculate excretion rates in the conventional manner propagates inaccurate results. The inclusion of Araki’s b values into adjustment calculations was demonstrated to significantly improve the performance of UFR adjustment for both criteria relative to ER and ERBW adjustments.

The derivation of specific Araki’s b values requires substantial further work by collecting multiple voids from the same individuals. By compiling a range of Araki’s b values for a range of analytes and demographic characteristics, their determining factors can be assessed, enabling the development of more sophisticated framework for urinary biomarker adjustment. Finally, additional constraints on the interpretation of urinary biomarker concentrations need addressing, such as time of sampling relative to exposure [1], which hydration adjustment cannot overcome.

29

List of abbreviations

AsIII: arsenous acid

AsIMM: Inorganic arsenic and Methylated Metabolites

AsV: arsenic acid

BW: Bodyweight

CI: Confidence Interval

CKD: Chronic Kidney Disease

DMA: dimethylarsonic acid eGFR: estimated Glomerular Filtration Rate

ER: Excretion Rate

ERBW: Excretion Rate adjusted for Body Weight

FMV: First Morning Void

GM: Geometric Mean

HPLC: High Performance Liquid Chromatography

ICP-DRC-MS: Inductively Coupled Plasma Dynamic Reaction Cell Mass Spectrometry

MEC: Mobile Examination Center

MMA: monomethylarsonic acid

NHANES: National Health and Nutrition Examination Survey

rp: Pearson’s correlation coefficient

30 rs: Spearman’s correlation coefficient

SG: Specific Gravity

UFR: Urinary Flow Rate

UFRA: Urinary Flow Rate adjustment optimised to criterion A

UFRB: Urinary Flow Rate adjustment optimised to criterion B

Ethics approval and consent to participate

Data reported in this manuscript from human subjects were obtained from NHANES 2009-

2010 and 2011-2012, which reports obtaining suitable written informed consent. Additional data were provided for comparison purposes from a previously published study (Meharg et al.

2014, Environ Pollut, 194: 181-187) in which written informed consent was obtained from all volunteers.

Availability of data and material

The data forming this investigation is from NHANES 2009-2010 and 2011-2012. Data are available for download from the NHANES website: http://www.cdc.gov/nchs/nhanes.htm.

Competing Financial Interests

All authors declare no conflicts of interest.

Funding

Funding was provided by the Natural Environment Research Council (NERC) via a University of Manchester/BUFI (Centre for Environmental Geochemistry) studentship (Contract No.

GA/125/017, BUFI Ref: S204.2).

31

Author contributions

DRSM, MJW and DAP conceived the investigation. DRSM acquired the data and performed statistical analyses. RML coordinated the data management statistical methodologies employed and CJM constructed the R programming scripts to conduct analyses. All authors contributed to and reviewed the manuscript.

Acknowledgements

We gratefully acknowledge Professor Andrew Meharg for the provision of comparison data to compliment our investigation and Dr Simon Chenery and Dr Mark Cave for scientific review and advice. We thank Olivier S. Humphrey and Ahmed A. N. Al Bualy for testing the R script provided in supporting information.

Author’s information

DRSM is completing a PhD on human biomonitoring of arsenic exposure with emphasis on the interpretation of urinary biomonitoring data. DAP is a Professor of Environmental

Geochemistry at the School of Earth, Atmospheric and Environmental Sciences and

Williamson Research Centre for Molecular Environmental Science at the University of

Manchester. DAP’s current research interests include human biomonitoring of exposure and subsequent genetic damage from consumption of arsenic in food and drinking water. MJW is the Head of Inorganic Geochemistry at the Centre for Environmental Geochemistry, British

Geological Survey and Honorary Associate Professor at the University of Nottingham, School of Biosciences, conducting research on assessing human exposures and deficiencies to harmful elements and essential minerals, respectively. RML is an Environmental Statistician at the

British Geological Survey whose interests include sampling design and environmental

32 monitoring. CJM is a geochemist with a background in groundwater chemistry and statistical programming.

33

References

1. Aylward LL, Hays SM, Smolders R, Koch HM, Cocker J, Jones K, et al. 2014. Sources of variability in biomarker concentrations. J Toxicol Environ Health B Crit Rev 17:45-61.

2. Middleton DRS, Watts MJ, Hamilton EM, Ander EL, Close RM, Exley KS, Crabbe H,

Leonardi GS, Fletcher T and Polya D.A. 2016. Urinary arsenic profiles reveal substantial exposures to inorganic arsenic from private drinking water supplies in Cornwall, UK, Sci Rep,

DOI: SREP25656.

3. Watts M, Joy E, Young S, Broadley M, Chilimba A, Gibson R, et al. 2015. Iodine source apportionment in the Malawian diet. Sci Rep 5: 15251. DOI: 10.1038/srep15251.

4. Cone EJ, Caplan YH, Moser F, Robert T, Shelby MK, Black DL. 2009. Normalization of urinary drug concentrations with specific gravity and creatinine. J Anal Toxicol 33:1-7.

5. Moeller KE, Lee KC, Kissack JC. Urine drug screening: Practical guide for clinicians. In:

Proceedings of the Mayo Clinic Proceedings, 2008, Elsevier. Vol. 83: 66-76.

6. Sobus JR, DeWoskin RS, Tan Y-M, Pleil JD, Phillips MB, George BJ, et al. 2015. Uses of

NHANES biomarker data for chemical risk assessment: Trends, challenges, and opportunities.

Environ Health Perspect 123:919-927.

7. Barrett JR. 2015. Urinary biomarkers as exposure surrogates: Controlling for possible bias.

Environ Health Perspect 123:A97.

8. Yeh HC, Lin YS, Kuo CC, Weidemann D, Weaver V, Fadrowski J, et al. 2015. Urine osmolality in the US population: Implications for environmental biomonitoring. Environ Res

136:482-490.

34

9. Rivera-Núñez Z, Meliker JR, Linder AM, Nriagu JO. 2010. Reliability of spot urine samples in assessing arsenic exposure. Int J Hyg Environ Health 213:259-264.

10. Barr DB, Wilder LC, Caudill SP, Gonzalez AJ, Needham LL, Pirkle JL. 2005. Urinary creatinine concentrations in the US population: Implications for urinary biologic monitoring measurements. Environ Health Perspect 113:192.

11. Mayersohn M, Conrad KA, Achari R. 1983. The influence of a cooked meat meal on creatinine plasma concentration and creatinine clearance. Br J Clin Pharmacol 15:227-230.

12. Baxmann AC, Ahmed MS, Marques NC, Menon VB, Pereira AB, Kirsztajn GM, et al.

2008. Influence of muscle mass and physical activity on serum and urinary creatinine and serum cystatin C. Clin J Am Soc Nephrol 3:348-354.

13. Nermell B, Lindberg AL, Rahman M, Berglund M, Åke Persson L, El Arifeen S, et al.

2008. Urinary arsenic concentration adjustment factors and malnutrition. Environ Res 106:212-

218.

14. Barber T, Wallis G. 1986. Correction of urinary mercury concentration by specific gravity, osmolality, and creatinine. J Occup Environ Med 28:354-359.

15. Leech S, Penney M. 1987. Correlation of specific gravity and osmolality of urine in neonates and adults. Arch Dis Child 62:671-673.

16. Dossin O, Germain C, Braun JP. 2003. Comparison of the techniques of evaluation of urine dilution/concentration in the dog. J Vet Med A Physiol Pathol Clin Med 50:322-325.

17. Imran S, Eva G, Christopher S, Flynn E, Henner D. 2010. Is specific gravity a good estimate of urine osmolality? J Clin Lab Anal 24:426-430.

35

18. Sauvé JF, Lévesque M, Huard M, Drolet D, Lavoué J, Tardif R, et al. 2015. Creatinine and specific gravity normalization in biological monitoring of occupational exposures. J Occup

Environ Hyg 12:123-129.

19. Hays SM, Aylward LL, Blount BC. 2015. Variation in urinary flow rates according to demographic characteristics and body mass index in NHANES: Potential confounding of associations between health outcomes and urinary biomarker concentrations. Environ Health

Perspect 123:293-300.

20. Araki S, Murata K, Aono H, Yanagihara S, Niinuma Y, Yamamoto R, et al. 1986.

Comparison of the effects of urinary flow on adjusted and non‐adjusted excretion of heavy metals and organic substances in ‘healthy’ men. J Appl Toxicol 6:245-251.

21. Araki S, Sata F, Murata K. 1990. Adjustment for urinary flow rate: An improved approach to biological monitoring. Int Arch Occup Environ Health 62:471-477.

22. Sorahan T, Pang D, Esmen N, Sadhra S. 2008. Urinary concentrations of toxic substances:

An assessment of alternative approaches to adjusting for specific gravity. J Occup Environ Hyg

5:721-723.

23. Vij HS, Howell S. 1998. Improving the specific gravity adjustment method for assessing urinary concentrations of toxic substances. Am Ind Hyg Assoc J 59:375-380.

24. CDC (Centers for Disease Control and Prevention). 2013. National Biomonitoring Program

Lead Biomonitoring Summary Webpage. Available: http://www.cdc.gov/biomonitoring/Lead_BiomonitoringSummary.html [accessed 05

September 2015].

26. CDC (Centers for Disease Control and Prevention). 2013. National Biomonitoring Program

Cadmium Biomonitoring Summary Webpage. Available:

36 http://www.cdc.gov/biomonitoring/Cadmium_BiomonitoringSummary.html [accessed 05

September 2015].

27. Akerstrom M, Barregard L, Lundh T, Sallsten G. 2013. The relationship between cadmium in kidney and cadmium in urine and blood in an environmentally exposed population. Toxicol

Appl Pharmacol 268:286-293.

28. Middleton D, Watts M, Hamilton E, Fletcher T, Leonardi G, Close R, et al. 2016.

Prolonged exposure to arsenic in UK private water supplies: Toenail, hair and drinking water concentrations. Env Sci Process Impact. DOI: 10.1039/C6EM00072J.

29. CDC (Centers for Disease Control and Prevention). 2015. National Health and Nutrition

Examination Survey Homepage. Available: http://www.cdc.gov/nchs/nhanes.htm [accessed 01

October 2015].

30. Warnes GR. 2014. Sasxport: Read and write sas xport files. http://CRANR- projectorg/package=SASxport.

31. Dragulescu AA. 2014. Xlsx: Read, write, format excel 2007 and excel 97/2000/xp/2003 files. http://CRANR-projectorg/package=xlsx.

32. Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF, Feldman HI, et al. 2009. A new equation to estimate glomerular filtration rate. Ann Intern Med 150:604-612.

33. Schwartz GJ, Muñoz A, Schneider MF, Mak RH, Kaskel F, Warady BA, et al. 2009. New equations to estimate GFR in children with CKD. J Am Soc Nephrol 20:629-637.

34. Meharg A, Williams P, Deacon C, Norton G, Hossain M, Louhing D, et al. 2014. Urinary excretion of arsenic following rice consumption. Environ Pollut 194:181-187.

37

35. R Core Team. 2013. R: A language and environment for statistical computing, r foundation for statistical computing, Vienna, Austria, www.R-project.Org.

36. Kuhn M. 2015. Caret: Classification and regression training. http://cran.R- project.Org/package=caret.

37. Dobbin KK, Simon RM. 2011. Optimally splitting cases for training and testing high dimensional classifiers. BMC Med Genomics 4:1.

38. Levine L, Fahy JP. 1945. Evaluation of urinary lead concentrations. 1. The significance of the specific gravity. J Ind Hyg Toxicol 27:217-223.

39. Wilcox RR, Tian T. 2008. Comparing dependent correlations. J Gen Psychol 135:105-112.

40. Revelle W. 2014. Psych: Procedures for psychological, psychometric, and personality research, Northwestern University, Evanston, Illinois, http://cran.R- project.Org/package=psych.

41. Venables WNR, Ripley BD. 2002. Modern applied statistics with S. Fourth edition.

Springer, New York. ISBN 0-387-95457-0.

42. Sata F, Araki S. 1996. Adjustment of creatinine-adjusted value to urine flow rate in lead workers. Archives of Environmental Health: An International

43. Hays S, Becker R, Leung H, Aylward L, Pyatt D. 2007. Biomonitoring equivalents: A screening approach for interpreting biomonitoring results from a public health risk perspective. Regulatory Toxicology and Pharmacology 47:96-109.

44. Navas-Acien A, Silbergeld EK, Pastor-Barriuso R, Guallar E. 2008. Arsenic exposure and prevalence of type 2 diabetes in US adults. Jama 300:814-822.

38

45. Tellez-Plaza M, Navas-Acien A, Crainiceanu CM, Guallar E. 2008. Cadmium exposure and hypertension in the 1999-2004 national health and nutrition examination survey

(NHANES). Environ Health Perspect: 51-56.

46. Haddow JE, McClain MR, Palomaki GE, Hollowell JG. 2007. Urine iodine measurements, creatinine adjustment, and thyroid deficiency in an adult United States population. The Journal of Clinical Endocrinology & Metabolism 92:1019-1022.

39

CHAPTER 5

Publication:

Prolonged exposure to arsenic in UK private water supplies: toenail, hair and drinking water concentrations

Journal:

Environmental Science: Processes and Impacts

Status as of thesis submission:

Published by the Royal Society of Chemistry (Open Access)

Citation: Middleton, DRS, Watts, MJ, Hamilton, EM, Fletcher, T, Leonardi, GS, Close, RM, Exley, KS, Crabbe, H and Polya, DA. (2016). Prolonged exposure to arsenic in UK private water supplies: toenail, hair and drinking water concentrations. Environmental Science: Processes & Impacts. DOI: 10.1039/C6EM00072J

Web link: http://pubs.rsc.org/en/content/articlelanding/2016/em/c6em00072j#!divAbstract

Supplementary information:

Follows publication (pp. 575-576)

94

Environmental Science Processes & Impacts

View Article Online PAPER View Journal | View Issue

Prolonged exposure to arsenic in UK private water supplies: toenail, hair and drinking water Cite this: Environ. Sci.: Processes Impacts,2016,18,562 concentrations†

D. R. S. Middleton,abc M. J. Watts,*b E. M. Hamilton,b T. Fletcher,c G. S. Leonardi,c R. M. Close,c K. S. Exley,c H. Crabbec and D. A. Polyaa

Chronic exposure to arsenic (As) in drinking water is an established cause of cancer and other adverse health effects. Arsenic concentrations >10 mgL1 were previously measured in 5% of private water supplies (PWS) in Cornwall, UK. The present study investigated prolongued exposure to As by measuring biomarkers in hair and toenail samples from 212 volunteers and repeated measurements of As in drinking water from 127

households served by PWS. Strong positive Pearson correlations (rp ¼ 0.95) indicated stability of water As concentrations over the time period investigated (up to 31 months). Drinking water As concentrations

were positively correlated with toenail (rp ¼ 0.53) and hair (rp ¼ 0.38) As concentrations – indicative of

Creative Commons Attribution 3.0 Unported Licence. prolonged exposure. Analysis of washing procedure solutions provided strong evidence of the effective removal of exogenous As from toenail samples. Significantly higher As concentrations were measured in hair samples from males and smokers and As concentrations in toenails were negatively associated with age. A positive association between seafood consumption and toenail As and a negative association Received 8th February 2016 between home-grown vegetable consumption and hair As was observed for volunteers exposed to <1 As Accepted 18th April 2016 mgL1 in drinking water. These findings have important implications regarding the interpretation of DOI: 10.1039/c6em00072j toenail and hair biomarkers. Substantial variation in biomarker As concentrations remained unaccounted rsc.li/process-impacts for, with soil and dust exposure as possible explanations. This article is licensed under a

Environmental impact Arsenic is an established carcinogen, chronic exposure to which has been linked to several cancers (lung, bladder, skin) as well as non-cancerous (cardiovascular

Open Access Article. Published on 19 April 2016. Downloaded 18/05/2016 16:41:47. disease, diabetes mellitus) health effects. This work consists of a human biomonitoring study (a collaboration between the University of Manchester, British Geological Survey and Public Health England) of 212 volunteers from 127 households with private water supplies from across Cornwall, UK. It is the largest scale exposure biomonitoring study conducted for As and drinking water in the UK to-date and investigates an exposure source for As that, until recently, had not been investigated in depth in the region. The sampling protocol consists of an initial and follow-up water collection spanning a period of either 8 or 31 months which, together with long-term biomarkers such as toenails and hair, allows for the assessment of prolonged arsenic exposure. Furthermore, the methods employed in this paper allow for an assessment of the efficacy of toenail washing procedures given the recognition of the susceptibility of this biomonitoring matrix to external contamination. The demonstration of effective contamination removal from samples in this study will be of great benet to the wider eld.

1. Introduction health concern. Five major As endemic regions of the world provide the strongest evidence of this association: north-west Chronic exposure to arsenic (As) in contaminated drinking and south-east Taiwan;2 northern Chile;3 Argentina;4,5 Bangla- water is an established cause of lung, skin, bladder and kidney desh6 and West Bengal.7 Although the aforementioned areas are cancer1 as well as other adverse health effects, posing a global more severely affected, As contaminated municipal and private water supplies (PWS) have been reported in countries across all 8 aSchool of Earth, Atmospheric and Environmental Sciences & William Research Centre inhabited continents. Notable European examples include 9 9 9 10 for Molecular Environmental Science, University of Manchester, Oxford Rd, Hungary, Romania, Slovakia and Serbia. Manchester, M13 9PL, UK A survey11 of PWS in Cornwall, south-west England, reported bInorganic Geochemistry, Centre for Environmental Geochemistry, British Geological concentrations exceeding the 10 As mgL1 UK prescribed Survey, Nicker Hill, Keyworth, Nottinghamshire, NG12 5GG, UK. E-mail: mwatts@ concentration or value12 (PCV) and WHO guidance value13 in 5% bgs.ac.uk of drinking water samples collected (n ¼ 497). In a follow-up cCentre for Radiation, Chemicals and Environmental Hazards (CRCE), Public Health 14 England, Chilton, Didcot, Oxfordshire, OX11 0RQ, UK biomonitoring study, a subset of the same cohort, drinking † Electronic supplementary information (ESI) available. See DOI: water As concentrations were positively correlated with urinary 10.1039/c6em00072j As concentrations aer the exclusion of arsenobetaine (AB) and

562 | Environ. Sci.: Processes Impacts,2016,18,562–574 This journal is © The Royal Society of Chemistry 2016 View Article Online Paper Environmental Science: Processes & Impacts

adjustment for hydration (osmolality adjustment). These Average growth rates for ngernails are 0.1 mm per day urinary As concentrations reected exposure in the preceding whereas toenails are estimated to grow by 0.03–0.5 mm per day, 2–4 days.15 Information on the longevity and temporal variation meaning that ngernails and toenails reect exposure windows of exposure in this study group was still outstanding. Two dating back approximately 6 and 12–18 months, respectively.34 methods that can assess exposure over extended timescales are Hair reects a period of just a few months, with reported scalp repeat monitoring of drinking water As concentrations and hair growth rates ranging from 0.2 to 1.12 mm per day.35 Growth monitoring of biological matrices, such as toenails and hair, rates for both matrices have been demonstrated to vary with that reect a longer exposure window than urine. Both demographic factors e.g. age and gender,29,34–36 with obvious approaches were employed in the present study. implications for interpreting exposure assessments conducted There are currently 2460 registered single domestic dwell- on diverse populations. ings served by PWS in Cornwall,16 with the true number likely to The susceptibility of nails and hair to external contamina- be much greater. No published data on the temporal variation tion is well documented, with a range of washing procedures of As concentrations in UK PWS were previously available, but having been implemented.29,37 The degree of sample contami- studies elsewhere reported mixed ndings. In Nevada, USA, nation likely depends on personal hygiene, hobbies, other although concentration changes (mean ¼3AsmgL 1) were behavioural variables and the relative ubiquity of the chemical measured in some supplies,17 with greater changes associated element of interest. Fingernails are reportedly more prone to with higher As concentrations, no clear temporal trends were contamination than toenails38 but this does not likely apply to observed between wet and dry seasons. In a related study,18 communities who are oen barefoot or wear open toed foot- ¼ strong Spearman correlations (rs)(rs 0.95) were reported wear. Contamination of hair and nails from cosmetic products between As concentrations in the same wells over a period of such as shampoos, hair colourings and nail polish is another 11–20 years, with both studies concluding that, for the region, important consideration. A study39 of the trace element limited measurements are sufficient for predicting exposures composition of nail polish estimated that the As contribution

Creative Commons Attribution 3.0 Unported Licence. over such timescales. Similarly, in Michigan, USA, strong from polish, if present, can range from 16 to 633%. 19 Pearson correlations (rp)(rp ¼ 0.88) were reported between As Whilst studies now routinely report the washing of nail and measurements taken an average of 14 months apart. Concen- hair samples prior to analysis, few have quantied the degree of trations were affected by point-of-use (POU) treatment systems, exogenous As versus As in toenails, or conrmed the removal of highlighting the necessity of collecting treatment usage data. exogenous As from samples. One investigation40 of exposure to Conversely, a study conducted in Washington, USA20 reported As in soils, also conducted in Cornwall, retained toenail changes as high as 19-fold in As concentrations measured in the washing solutions for As determination. Both the nal rinse same supply 12 months apart, suggesting that temporal stability fractions and a pooled solution of all preceding fractions were of As concentrations varied by region due to geological and retained to quantify exogenous As contamination and conrm

This article is licensed under a geochemical variables, if not inconsistencies in sampling its removal from samples. The As content of nal rinse fractions methodologies.19 accounted for 0.2 to 1.6% of the total As measured in toenails.40 The use of toenail and hair biomonitoring for As exposure This provided strong evidence of the efficacy of the washing

Open Access Article. Published on 19 April 2016. Downloaded 18/05/2016 16:41:47. offers the assessment of a longer exposure window than that procedure but, with a sample of 17 volunteers, the performance reected by urine sampling. The affinity of As for sulydryl of this method remained to be validated on a greater scale. groups in the keratin of nails and hair, the isolation of these The present study aimed to assess exposure to inorganic As via matrices from other metabolic processes following their drinking water consumption in a population served by PWS in formation and the time taken for them to ‘grow out,’ makes Cornwall, UK, using hair and toenail biomarkers in addition to them attractive for measuring biomarkers of past As exposure.21 initial and follow-up drinking water samples collected up to 31 Nails and hair have the added value of a non-invasive collection months apart. Specic objectives were to (i) compare repeat PWS protocol and few sample transport/storage requirements. Posi- drinking water As concentrations measured either 8 or 31 months tive correlations between drinking water and biomarker As apart; (ii) investigate the effects of As concentration, duration concentrations have been reported in numerous studies for between measurements, sourcetypeandtreatmentusageon both toenails22–25 and hair.21,26 Increased risk of various cancers, changes in drinking water As concentrations; (iii) measure the including cutaneous melanoma27 and small and squamous-cell total As concentrations in toenail and hair samples collected from carcinoma of the lung,28 have also been positively associated volunteers and assess their relationship with drinking water As with toenail As concentrations. concentrations adjusted for other covariables (demographic, Despite the advantages of toenail and hair biomonitoring, behavioural and dietary) and (iv) quantify the potential for external caveats apply when using these matrices to assess exposure. sample contamination to affectAsconcentrationsintoenailand Factors unrelated to exposure have been reported to inuence hair samples, including the use of nail polish and hair dye. As concentrations in hair and nails: namely, the inter-individual variability of growth rates of the biomonitoring matrices, 2. Experimental demographic and behavioural factors such as age, gender and smoking,23 their susceptibility to external contamination29,30 Ethical approval and volunteer communication and the consumption of dietary items such as fruit juices,31 Ethical approval was granted by the University of Manchester beer,32 wine32 and dark-meat sh.33 Research Ethics Committee (Ref 13068) and the NHS Health

This journal is © The Royal Society of Chemistry 2016 Environ. Sci.: Processes Impacts,2016,18,562–574 | 563 View Article Online Environmental Science: Processes & Impacts Paper

Research Authority National Research Ethics Service (NRES) day); home-grown vegetable consumption (all year, seasonally, (Ref 13/EE/0234). All volunteers provided written informed only in pots or never); rice (servings/week); seafood (servings/ consent prior to participating. Individual data feedback to week); most oen consumed seafood type (if reported): white participants was provided through a letter containing specic sh (e.g. cod, plaice, haddock etc.), shellsh (e.g. mussels, guidance developed by PHE along with BGS and Cornwall prawns, cockles etc.) and dark-meat sh (e.g. salmon, tuna, County Council. Participants were given advice on any potential mackerel, sardines etc.); beer (L per day); wine (L per day); cider health risks and suggested corrective actions if they had one or (L per day) and fruit juice (L per day). more exceedances of the water quality standards. All partici- pants were provided with appropriate contact details for any follow-up enquiries. Chemical analyses Reagents and standards. All aqueous solutions were prepared using 18.2 MU deionised water (DIW) (Millipore, UK). Recruitment and sample collection Nitric acid (HNO3), hydrochloric acid (HCl) and 30% hydrogen ™ Environmental monitoring. The sampling frame consisted peroxide (H2O2) were Romil-SpA super purity grade (Romil, of 476 households using a PWS that had provided drinking UK). The acetone used for sample cleaning was HPLC grade water samples during a previous survey11 – henceforth referred (Fisher Scientic, UK). Arsenic calibration standards were made to as initial sampling (drinking water only). The initial survey using an in-house multi-element stock in which the As contri- was conducted in two parts, with households in east and west bution was from a 1000 mg L 1 PrimAg® grade mono-elemental Cornwall surveyed in March–April 2011 and March–April 2013, solution (Romil, UK). Independent 25 mgL 1 As QC standards respectively. Information letters were sent to households that were prepared from a multi-element stock solution of various participated in initial sampling and, aer being contacted by concentrations with As at 20 mg L 1 (Ultra Scientic, USA). A telephone, 127 households were recruited to provide a follow-up germanium (Ge) ICP-MS internal standard was prepared from 1

Creative Commons Attribution 3.0 Unported Licence. drinking water sample. Follow-up sampling took place in a Fluka Analytical 1000 mg L stock solution (Sigma-Aldrich, November 2013. This resulted in 127 drinking water samples USA). collected either 31 (n ¼ 51) or 8 (n ¼ 76) months apart Sample pre-treatment and dissolution. Toenail samples were depending on whether households were in east (2011 initial cleaned and digested by adapting a previously reported collection) or west (2013 initial collection) Cornwall, respec- protocol.40 Visible exogenous debris was removed using a PTFE tively. Point-of-use drinking water samples were collected using policeman/stirring rod (Chemware, USA) in a HEPA ltered a previously reported protocol11 clean room. Samples with visible nail polish residue (regardless Biomonitoring. Biomonitoring was conducted on one occa- of whether reported in the questionnaire) were further cleaned sion only – at the time of the follow-up drinking water collection with acetone and cotton wool. Samples were transferred into

This article is licensed under a in November 2013. Sample collection packs were mailed to clean 25 mL Duran® borosilicate beakers (Schott, Germany), volunteers prior to household visits. Volunteers were asked to placed in an ultrasonic bath (Fisher Scientic, UK), sonicated at allow a minimum of 4 weeks for toenail growth (to ensure 37 MHz at room temperature for 5 minutes (15 minutes for

Open Access Article. Published on 19 April 2016. Downloaded 18/05/2016 16:41:47. sufficient mass for analysis) before self-collecting from all 10 those with visible varnish) in 3 mL of acetone, rinsed with 2 mL toes and storing in polyethylene bags. Hair samples were of DIW and then 2 mL of acetone, sonicated for 10 minutes in 3 collected by researchers during visits using an amended version mL of DIW and twice rinsed with 3 mL of DIW. All rinse aliquots of the COPHES project protocol.41 Hair >3 cm in length was prior to the nal, which remained separate, were pooled in PFA removed from the nape by twisting into a pencil-width strand vials (Savillex, USA) and evaporated to dryness overnight on before tagging with masking tape. The tape was labelled with an a graphite hot block before reconstitution in 5 mL of 1% v/v

arrow pointing towards the root. Strands were removed with HNO3 + 0.5% v/v HCl. Both initial and nal rinse fractions were ethanol-rinsed stainless steel scissors as close to the scalp as analysed by ICP-MS for total As. The nal fraction was analysed possible. Hair <3 cm in length was collected in smaller amounts separately to assess the effectiveness of the washing procedure from several locations on the back of the head. The portion of and conrm the elimination of exogenous contamination. A hair >5 cm was discarded with the portion closest to the scalp schematic of the abovementioned procedure can be viewed in being retained for analysis. ESI (Fig. S1†). Additional variables. An exposure/food frequency question- Toenail samples were dried to constant weight (12 h approx.) naire was administered to volunteers using Microso Access on in a clean laminar ow hood (Envair, UK) and stored in a laptop/tablet device. For drinking water related analysis, data microcentrifuge tubes in a silica gel desiccator before being on PWS source type, treatment usage, system storage and weighed (0.1 g or as much as available) into PTFA MARS Xpress borehole depth were collected at the time of initial water vessels (CEM Corporation, UK). Four millilitres of concentrated

sampling. For biomonitoring analysis, demographic and HNO3 +1mLofH2O2 were added and samples were le to rest behavioural variables – age, gender, current smoking status, for 30 minutes until effervescence subsided. Vessels were cap- nail polish and hair product usage – were collected and, addi- ped and digested in a microwave assisted reaction system tionally, information on the consumption of select dietary items (MARS Xpress, CEM Corporation, UK) on the following heating that have been reported31–33 to contain As in relatively high program: ramped to 100 C and held for 5 minutes; ramped to concentrations. These were: PWS water consumption (L per 200 C and held for 30 minutes (100% power: 1200 W). Vessels

564 | Environ. Sci.: Processes Impacts,2016,18,562–574 This journal is © The Royal Society of Chemistry 2016 View Article Online Paper Environmental Science: Processes & Impacts

were le to cool overnight before their contents were trans- with signicance tests (p-values) and 95% condence intervals ferred into PFA vials with DIW and reduced to a gel at 80 Con (C.I) were used to assess the strength in relationship between:

a graphite hot block. One millilitre of 10% v/v HNO3 was added initial versus follow-up drinking water As; well depth versus As to the vessels, which were then heated for 20 minutes at 50 C concentration difference; rinse versus digest As concentrations followed by the addition of 4 mL of DIW. Digests were stored in and drinking water versus toenail/hair As concentrations. Wel- polystyrene ICP-MS tubes. ch's independent unequal variance tests were used to test for Hair samples underwent the same cleaning and digestion differences in toenail, hair and rinse solution As concentrations procedure as toenail samples. Whatman Grade B-2 weighting between different subsets to account for unequal sample sizes. papers (GE Healthcare Life Sciences, UK) and a Milty Zerostat 3 One-way analysis of variance (ANOVA) was used to test for anti-static gun were used to aid the transfer of hair samples differences in toenail and hair As concentrations between between vessels. different age groups. Multiple linear regression models were Total As determination by ICP-MS. Analysis was performed constructed to assess signicant predictors of toenail and hair using an Agilent 7500cx series ICP-MS (Agilent Technologies, As in addition to drinking water As. Exploratory analyses USA) tted with a MicroMist low-ow nebulizer (Glass Expan- revealed positively skewed distributions for drinking water, sion, Australia) and an ASXpress rapid sample introduction toenail and hair As concentrations and As concentrations in system (Teledyne CETAC Technologies, USA) using previously rinse solutions. To address this, natural log(ln) transformations reported42 operating conditions. Drinking water samples were were applied to these variables prior to Pearson correlations, analysed using a previously reported method.11 Rinse solutions Welch's tests, ANOVA and multiple regression modelling. For were diluted 2 and analysed by a method used previously11 for the same reason, geometric means (GM) were calculated water samples. Those with visible suspended particulate were instead of arithmetic means. Le censoring was applied to hair passed through a 0.45 mm Acrodisc® syringe lter (PALL Life As concentrations (n ¼ 8) below the analytical limit of detection Sciences, USA). Toenail and hair digests were diluted 4 with (LOD) by replacing values with half of the LOD.

Creative Commons Attribution 3.0 Unported Licence. 1% v/v HNO3 + 0.5% v/v HCl. Helium (He) collision cell mode was used to remove potential polyatomic interferences with the same mass/charge ratio as As (m/z 75). Signal dri was corrected 3. Results and discussion using a Ge internal standard introduced via a T-piece. Analytical Study group limits of detection (LOD) were calculated as 3 the standard The spatial extent of the study is presented in Fig. 1 and char- deviation of run blanks for drinking water analysis and 3 the acteristics of households and volunteers are shown in Table 1. standard deviation of reagent blanks for toenail and hair anal- Two hundred and twelve volunteers from 129 households re- ysis. The LODs for As in drinking water and toenails/hair were ported using their PWS for human consumption and provided m 1 m 1 0.02 gL and 10 gkg , respectively. either a toenail sample, hair sample or both. This made the This article is licensed under a ffi Quality control. Toenail and hair samples of su cient mass present study the largest investigation of long-term exposure to were chosen for duplicate analysis. Samples were milled to As in drinking water in the UK to-date. Repeated water samples  a ne powder using a 6850 Freezer Mill (SPEX Sample Prep, were available for comparison from 127 households, the

Open Access Article. Published on 19 April 2016. Downloaded 18/05/2016 16:41:47. – USA) a cryogenic impact grinder cooled with liquid nitrogen. majority of which were supplied by a borehole. The age distri- One pair of duplicates was digested per batch, in addition to 3 bution of the study group was not representative of the corre-  reagent blanks. Method accuracy was assessed using Certi ed sponding local rural population, with 63% of volunteers aged Reference Materials (CRMs). Two samples (0.1 g) of NCS DC over 60. It is noted here that population-based exposure esti- 73347 Hair (China National analysis Centre for iron and steel, mates were not the focus of the present paper. Nail polish usage Beijing, China) were digested per batch of hair and toenail was reported by 17 of the 206 (8%) volunteers who provided samples. Two additional samples (0.1 g) of in-house human toenails, whereas polish was observed on the toenail samples of toenail reference material (BAPS2014 Human Toenail) were almost double that number (30, 15%). This underlines the digested per batch of toenail samples. BAPS2014 was produced importance of checking nails for visible polish prior to analysis by pooling the toenail clippings, saved over a period of 2 years, and not relying on questionnaire data alone. of 2 male volunteers (aged 23 and 38) not knowingly exposed to substantial environmental or occupational As. A homogeneous powder was prepared using the cleaning and milling procedure Repeated drinking water measurements already described prior to mixing end-over-end for several At the initial sampling phase, 125 out of 127 households (98%) hours. The accuracy of drinking water and toenail washing had detectable (>0.02 mgL1) concentrations of total As solution measurements was assessed using NIST SRM 1643e measured in their drinking water. At follow-up sampling, Trace Elements in Water (National Institute of Standards and 126 out of 127 (99%) households had As concentrations >0.02 Technology, USA). mgL 1. Only fourteen of the 127 households (11%) exceeded the 10 As mgL1 UK PCV and WHO guidance value at initial sampling with a maximum As concentration of 231 mgL 1. Two Statistical analyses households had borderline results (>9 As mgL 1), one of which Statistical tests and plot production were performed using R exceeded the PCV (17 As mgL 1) at follow-up sampling. One version 3.0.0 (base package).43 Pearson correlation coefficients further household, below PCV at initial sampling, exceeded at

This journal is © The Royal Society of Chemistry 2016 Environ. Sci.: Processes Impacts,2016,18,562–574 | 565 View Article Online Environmental Science: Processes & Impacts Paper Creative Commons Attribution 3.0 Unported Licence.

Fig. 1 Map of the Cornwall study area, shown in the context of the UK (excluding Northern Ireland), and the spatial distribution of sampled households. Total As concentrations measured in drinking water samples collected during the initial survey are plotted for reference. Note: no assessment of the spatial controls on As distribution was made in this study. Compiled using ESRI ArcMap 10.1. This article is licensed under a

follow-up (from 6 to 17 As mgL 1). Only one exceedance drop- The strongest correlation observed was for the subset of 1 Open Access Article. Published on 19 April 2016. Downloaded 18/05/2016 16:41:47. ped below PCV at follow-up sampling, from 14 to <1 As mgL . households with both iron (Fe) and manganese (Mn) removal

Households who had high As concentrations in their PWS were systems and pH buffering systems (rp ¼ 0.998) in addition to advised to install appropriate remediation. Changes were not alowermeandifference to supplies with neither treatment attributed to installation of treatment systems. Of three system. This is not unexpected given that supplies with treat- households that reported installation of any kind of treatment ment systems installed are not subject to underlying system between initial and follow-up measurements, none were geochemical variations. Although no household in this study among those exceeding the PCV and the impacts on As group reported using As-specic treatment systems, Fe/Mn concentrations were minimal. Of the 14 households above PCV removal units have been reported to reduce As concentra- at initial sampling, 11 reported not installing any additional tions.11 Of the 62 households where borehole depth informa- treatment and data were missing for the remaining three. This tion was available, no signicant correlation was observed has important implications regarding risk awareness and the between depth and the difference in As concentration between advice given to households above PCV. initial and follow-up sampling. This is consistent with previous Overall, As concentrations in PWS were stable over both 8 studies.17 Source type inuence was only assessed between well and 31 month periods. Mean differences in As concentrations, and borehole sources due to a limited number of other source initial and follow-up GM As concentrations and Pearson types. There was no apparent difference in As concentration correlation coefficients between initial and follow-up As changes between well or borehole source types or system concentrations are shown in Table 2. Follow-up As concentra- storage. An observation was made regarding the correct cate- tions are plotted against their initial counterparts in Fig. 2. In gorisation of source type. One household in the present study agreement with previous studies,17,18 strong Pearson correla- reported using a borehole at initial sampling but on receiving tions were observed between initial and follow-up samples initial results (80.5 As mgL 1) it was discovered to be a disused ¼ ¼ ‘ ’ collected both 8 (rp 0.95) and 31 (rp 0.95) months apart. A mine adit (categorised as other in Table 1). This highlights the greater mean difference was observed for PWS with >10 As mg importance of homeowners seeking the correct character- L 1 due to the higher concentrations reported in this group. isation of their PWS when acquiring a new property.

566 | Environ. Sci.: Processes Impacts,2016,18,562–574 This journal is © The Royal Society of Chemistry 2016 View Article Online Paper Environmental Science: Processes & Impacts

Table 1 Household and study group characteristics (Fig. S2a†), in the context of the certied value and upper/lower limits. So too were mean recovery and relative standard devia- Households 129 tion (RSD) (Fig. S2b†). On the basis of these results, 0.02 g was Initial and follow-up water sample, n (%)a 127 (98.4) chosen as the minimum mass requirement, being the lowest Initial sample year, n (%) mass at which As concentrations were found to be consistently 2011 51 (40.2) within upper/lower certied limits of the CRM. This value is not 2013 76 (59.8) universal and may not apply to other studies but was selected to try and maximise the usage of a compromised sample set. Source type, n (%) Borehole 111 (87.4) Depending on the amount of As in samples, requirements may Well 11 (8.7) be lower or higher. The RSD calculated for triplicates at lower Spring capture 2 (1.6) masses may also reect reduced homogeneity of the CRM. Other 3 (2.4) Following the exclusion of samples below the minimum Borehole depth reported, n (%) 62 (48.8) mass, As data were available for the toenails and hair of 200 and Mean borehole depth (m) 48 104 volunteers, respectively. All toenail and 96 (92%) hair 1 Treatment system, n (%) samples were above the 10 mgkg LOD. Arsenic measured in Fe/Mn removal 18 (14.2) CRM NCS DC 73347 was 273 10 As mgkg 1 (n ¼ 40), within the ff pH bu ering 60 (47.2) certied range of 280 50 As mgkg 1, yielding a mean recovery of 98% with 5% precision. The mean As measured in BAPS 2014 Storage (e.g. water tank) in system, n (%) m 1 ¼ Yes 62 (48.8) Human Toenail was 93 5As gkg (n 20). The accuracy of No 65 (51.2) BAPS 2014 measurements could not be assessed, but good Volunteers 212 precision (5% RSD) was maintained. The mean difference between duplicate digests was 1.1% (7 pairs) and 3.4% (6 pairs) Gender, n (%)

Creative Commons Attribution 3.0 Unported Licence. for toenail and hair, respectively. Male 109 (51.4) Female 103 (48.6) Summary statistics for toenail and hair As concentrations are Mean age, years (range) 62 (18–90) shown in Table 3 for different demographic and behavioural subsets. The GM toenail As concentration of all 200 volunteers Age group, n (%) was 151 As mgkg 1 and ranged from 27 to 3354 As mgkg 1.This 18–29 6 (2.8) falls within previously published ranges, with a higher GM and 30–39 3 (1.4) 23 40–49 28 (13.2) maximum concentration than a study conducted in New 1 1 50–59 42 (19.8) Hampshire, USA (GM: 90 As mgkg ;range:10–810 As mgkg ), 60–69 75 (35.4) with comparable levels of drinking water exposure (<0.02–66 As This article is licensed under a – 70 79 44 (20.8) mgL 1). A previous study,40 conducted in south west England, 80–90 14 (6.6) reported a range of 858 to 25 981 As mgkg 1 for individuals Smoking status, n (%) exposed to high As in soil, with no exposure to As in drinking

Open Access Article. Published on 19 April 2016. Downloaded 18/05/2016 16:41:47. Currently smoking 13 (6.1) water. Although conducted in the same geographic region as the Not currently smoking 191 (90.1) present study, Button et al. (2009)40 investigated individuals Not reported 8 (3.8) living in the direct vicinity of a former As mine, possibly Provided toenails, n (%) 206 (97.2) explaining the much higher reported concentrations than the Provided hair, n (%) 186 (87.7) 26 Provided both, n (%) 180 (84.9) present study. Hinwood et al. (2003) investigated the toenail As concentrations of volunteers in different exposure categories in Cosmetic usage, n (%) rural Australia: high soil (>30 As mg kg 1); high water (>10 As mg Polish usage reported (if toenails provided) 17 (8.3) L 1) and low exposure (<10 As mgL 1 in drinking water and <30 Polish observed on toenails 30 (14.6) As mg kg 1 in soil). Overall, much higher toenail As concentra- Dye usage reported (if hair provided) 31 (16.7) tions were reported by Hinwood et al. (2003), across all cate- a Subsequent characteristics and percentages in households section gories, than those in the present study. For example, the refer to this subset. minimum toenail As concentration in the low exposure group was 1350 mgkg 1, of which only eight volunteers exceeded in the present study. Quantication/removal of exogenous As from toenail samples was cited as a limitation by Hinwood et al. (2003) Toenail and hair total As and, therefore, few meaningful conclusions can be drawn from Due to difficulties with sample collection and handling, many this comparison. Slotnick et al. (2007)44 reported a lower drinking hair samples were of low mass at the point of digestion. This water As GM to the present study (0.59 versus 0.88 As mgL 1)and prompted the determination of a minimum mass requirement a lower toenail As GM (70 versus 151 As mgkg 1). Maximum for toenail and hair samples by digesting triplicate samples of drinking water and toenail as concentrations were also higher in NCS DC 73347 CRM in decreasing mass increments (0.1, 0.08, the present study than those reported by Slotnick et al. (2007): 0.07, 0.06, 0.05, 0.04, 0.03, 0.02, 0.01, 0.005, 0.002 and 0.001 g). 233 versus 99 As mgL1 and 3353 versus 1260 As mgkg1, Measured As concentrations were plotted against mass of CRM respectively. Other comparable studies include Rivera-Nu´nez˜

This journal is © The Royal Society of Chemistry 2016 Environ. Sci.: Processes Impacts,2016,18,562–574 | 567 View Article Online Environmental Science: Processes & Impacts Paper

Table 2 Drinking water As arithmetic mean differences, initial and follow-up As concentration geometric means (GM) and results from Pearson correlations between initial and follow-up As concentrations (ln transformed variables) for different PWS subsets

Mean difference Initial total Follow-up total Pearson correlation m 1 m 1 m 1 Subsets n (As gL ) As GM (As gL ) As GM (As gL ) (rp)

All households 127 0.7 1.0 1.0 0.95

Initial sample year 2011 51 1.1 0.8 0.9 0.95 2013 76 0.5 1.2 1.2 0.95

Initial total As concentration <1 mgL 1 67 0.1 0.2 0.3 0.87 1–10 mgL 1 46 0.1 3.2 2.7 0.68 >10 mgL 1 14 6.6 36.5 27.9 0.79

Source type Borehole 111 0.8 1.2 1.1 0.95 Well 11 0.4 0.3 0.4 0.97

Treatment system Fe/Mn removal only 12 0.2 1.7 1.6 0.95 pH buffering only 54 0.2 0.8 0.8 0.94 Both of above 6 0.3 0.5 0.5 1 (0.998) Neither of above 55 1.8 1.2 1.3 0.94

Creative Commons Attribution 3.0 Unported Licence. Storage (e.g. water tank) in system Yes 62 1.7 1.1 1.1 0.94 No 65 0.3 0.9 1.0 0.95

drinking water and toenail As concentrations, was low in the present study compared to those reported in severely affected areas. Nevertheless, 10 volunteers in the present study exhibited This article is licensed under a toenail As concentrations above the GM (1010 As mgkg 1) re- ported by Kile et al. (2005)46 across three villages in Bangladesh – the world's worst affected region – with drinking water As 1 1 Open Access Article. Published on 19 April 2016. Downloaded 18/05/2016 16:41:47. concentrations between 1 and 752 As mgL (GM: 6.2 As mgL ). The GM hair concentration measured in the present study was 82 As mgkg 1 (range: 500 gkg . Of these 40 individuals, 10 were exposed to p 1 measurements taken 31 (2011 initial collection) and 8 (2013 initial >10 mgL of As in their drinking water. While it is not possible collection) months apart. to conclude that these volunteers are either chronically or acutely exposed, where elevations correspond with drinking water As concentrations above PCV, attention is warranted. et al. (2011)45 and Yu et al. (2014)24, with drinking water As GMs of Welch's tests (Table 3) detected no signicant differences in 0.74 and 0.28 mgL 1 and toenail As GMs of 90 and 57 mgkg 1, toenail As between any subsets. Signicantly lower hair As respectively. Widespread As exposure, on the basis of both concentrations were detected for females (p < 0.001) and

568 | Environ. Sci.: Processes Impacts,2016,18,562–574 This journal is © The Royal Society of Chemistry 2016 View Article Online Paper Environmental Science: Processes & Impacts

Table 3 Summary statistics for total As in toenail and hair samples for different demographic and behavioural characteristic subsets of the study group. Statistically significant As concentrations between subsets are in bold type with p-values calculated by Welch's independent t-test on natural log transformed data in adjacent columns. Age group differences were assessed using one-way analysis of variance (ANOVA)

n Toenail total As p-Value, Welch test Hair total As p-Value for Welch's test (toenails, hair) (mgkg 1), GM (range) (ANOVA for age groups) (mgkg 1), GM (range) (ANOVA for age groups)

All 200, 104 151 (26.9–3354) — 82.6 (

Gender Male 102, 45 155 (26.9–1896) 0.63 150 (28.8–2908) <0.001 Female 98, 59 146 (39.1–3354) 52.5 (

Age group 18–39 6, 3 214 (8.1–1497) 0.28 89.9 (56.8–128) 0.76 (ANOVA) 40–49 27, 17 204 (57.9–3354) 121 (10.9–2396) 50–59 41, 20 154 (43–2578) 79.2 (

Smoking status Currently smoking 11,7 209 (100–2578) 0.25 324 (28.8–2908) 0.04 Not currently smoking 181,93 146 (26.9–1982) 74.6 (

Nail polish usage Reported/observed 34 131 (44.6–1497) 0.34 ——

Creative Commons Attribution 3.0 Unported Licence. Not reported/observed 166 155 (26.9–3354) —

Hair dye usage Reported 20 —— 41.4 (10.8–756) 0.003 Not reported 84 — 97.4 (

volunteers who reported using hair dye (p ¼ 0.003). Signicantly ¼

This article is licensed under a higher hair As concentrations were detected for smokers (p 0.04). These ndings were compared with a previous study49 investigating demographic and behavioural controls on the composition of hair: Chojnacka et al. (2006) reported 150% Open Access Article. Published on 19 April 2016. Downloaded 18/05/2016 16:41:47. more As in the hair of smokers, 210% more As in the hair of males and articially coloured hair was reported to contain 200% more As than naturally coloured hair.49

Exogenous As quantication Analysis of rinse solutions from the toenail washing procedure provided a useful insight into exogenous As contamination. The bar plot in Fig. 3 shows the hypothetical contribution of exog- enous As to that measured in toenails if they had not been washed. Rinse concentrations were normalised to the mass of toenail washed to allow comparison with digest concentrations. For toenails without polish, the GM As measured in initial rinse fractions was 9% of that measured in digested toenails, whereas the GM nal rinse fraction As concentration only accounted for 0.4%. Firstly, this conrmed the necessity of washing toenails, with a maximum percentage contribution of 716% in the case of  one volunteer. Secondly, the low contribution from nal rinse Fig. 3 Geometric mean (GM) As concentrations in toenail samples, fractions indicated the effective removal of exogenous As initial and final rinse fractions for volunteers with and without (maximum contribution: 5%). Furthermore, in agreement with observed/reported nail polish. Initial and final rinse fraction As previous ndings,40,50 the washing procedure appeared to have concentrations as a percentage of the As measured in toenail digests are printed on plots. begun to leach endogenous As from toenails by the nal rinsing stage. This is indicated in Fig. 4, where no signicant

This journal is © The Royal Society of Chemistry 2016 Environ. Sci.: Processes Impacts,2016,18,562–574 | 569 View Article Online Environmental Science: Processes & Impacts Paper

correlation was observed (rp ¼0.05; p ¼ 0.43) between initial without nail polish. This nding does not dismiss the effects of rinse As concentrations and toenail digest As concentrations polish on sample concentrations, as substantial contributions  ¼ 39 (Fig. 4a). Conversely, a signi cant positive correlation (rp 0.71; have been demonstrated elsewhere. Several factors may have p < 0.001) was observed between nal rinse As concentrations limited ndings on this occasion: misreporting of polish usage/ and toenail digest As concentrations (Fig. 4b). The relatively failure to identify polish on samples; ineffective polish removal small hypothetical contributions (5% maximum) of nal rinse during washing; low sample size of volunteers with polish and As concentrations to those in toenail digests suggests that a lack of digestion procedure for rinse solutions/the inability to a small degree of leaching is of no great concern in the present solubilise As present from polish. Contribution from polish has study. It is noted that future efforts could be made to determine also been demonstrated39 as brand dependent and further work an optimum degree of washing for toenail samples and maxi- is needed to quantify/mitigate the effects of polish usage on mise the removal of exogenous As whilst minimising endoge- biomonitoring studies using human nails as part of a wider nous As leaching. It is likely that the optimum number of rinses review of the effects of surface contamination. would depend on the level of contamination on the nail surface – adifficult metric to quantify. Welch's independent t-tests detected no signicant differ- Drinking water and biomarker relationships ¼ ences in digest As concentrations (p 0.34), initial rinse As Due to the difference in duration between initial and follow-up ¼  ¼ concentrations (p 0.85), nal rinse As concentrations (p drinking water samples, follow-up water samples (all of which ¼ 0.74) or percentage contributions from either initial (p 0.52) were collected during the same sampling campaign as the hair  ¼ or nal (p 0.35) rinse fractions between samples with and and toenail collections) were used as explanatory variables of Creative Commons Attribution 3.0 Unported Licence. This article is licensed under a Open Access Article. Published on 19 April 2016. Downloaded 18/05/2016 16:41:47.

Fig. 4 Initial rinse fraction As concentrations (a) and final rinse fraction As concentrations (b) plotted against toenail digest As concentrations. No

significant relationship (rp) was observed for initial rinse fractions, but a strong significant correlation was evident for final rinse fractions. This suggests (i) effective exogenous As contamination removal and (ii) subsequent leaching of As from toenails.

Fig. 5 Significantly positive Pearson correlations (rp) between toenail (a) and hair (b) biomarker As concentrations and those measured in drinking water.

570 | Environ. Sci.: Processes Impacts,2016,18,562–574 This journal is © The Royal Society of Chemistry 2016 View Article Online Paper Environmental Science: Processes & Impacts

Table 4 Pearson correlations (rp) for drinking water As and toenail and hair As for different drinking water As concentration ranges. Moderate/ strong correlations (bold type) were only observed where drinking water As exceeded 10 mgL1

Pearson's rp (p-value, [95% C.I])

Drinking water As <1 mgL 1 Drinking water As 1–10 mgL 1 Drinking water As >10 mgL 1 Full range

Toenail 0.15 (p ¼ 0.13, 0.12 (p ¼ 0.32, 0.86 (p < 0.001, 0.53 (p < 0.001, total As [0.04, 0.33]) (n ¼ 107) [0.12, 0.34]) (n ¼ 73) [0.66, 0.94]) (n ¼ 19) [0.43, 0.63]) (n ¼ 199) Hair 0.11 (p ¼ 0.45, 0.15 (p ¼ 0.34, 0.62 (p ¼ 0.02, 0.38 (p < 0.001, total As [0.18, 0.38]) (n ¼ 48) [0.16, 0.43]) (n ¼ 43) [0.10, 0.87]) (n ¼ 13) [0.20, 0.53]) (n ¼ 104)

biomarker As concentrations for consistency. In agreement volunteers with drinking water containing <1 As mgL 1 (low) with previous ndings,21,23,24,26 signicant positive correlations and $1AsmgL 1 (high). This was to maintain consistency with 45 were observed between drinking water and toenail (Fig. 5a, rp previous studies that reported a greater predominance of ¼ 0.53; p < 0.001; 95% C.I: 0.43, 0.63) and drinking water and additional, notably dietary, sources of As intake when drinking 1 hair (Fig. 5b, rp ¼ 0.38; p < 0.001; 95% C.I: 0.20, 0.53) As water concentrations were <1 mgL . This stratication resulted concentrations. This conrmed previous ndings14 of human in four initial models for toenail (Model 1a, 1b) and hair (Model exposure to As from PWS but over a longer timescale. When 2a, 2b) As concentrations as a function of demographic and grouped by drinking water As concentration (Table 4), strong behavioural variables only. signicant correlations were only observed where drinking Coefficients for each model are shown in Table 5. There were water As was >10 mgL 1 for both toenails and hair. Fig. 5a no signicant demographic/behavioural predictors of toenail As

Creative Commons Attribution 3.0 Unported Licence. showsthat,forvolunteersexposedtodrinkingwaterwith in the low drinking water As group (Model 1a) but both <10 As mgL 1, a considerable number toenail samples con- increasing drinking water As and age resulted in a signicant tained notable As concentrations. Given the encouraging increase in toenail As when As in drinking water was >1 mgL 1. results from the assessment of the washing procedure, sample The effect of age on toenail As concentration has been reported contamination was unlikely to account for these results. by previous studies23 but in the opposite direction to the effect Cornwall is a region of elevated environmental As51 and, as found in the present study. The mechanism of this relationship notedpreviouslybyButtonet al. (2009), alternative exposure has not been elucidated. For example, Kile et al. (2005) note that routes, such as the ingestion of As-bearing soil and dust, are toenail growth decreases with age. This may result in a higher possible explanations for elevated toenail As where drinking concentration of As relative to a lower mass of nail. The high

This article is licensed under a waterAsislow.40 The investigation of additional exposure proportion of volunteers in older age groups in the present routes in the present study population will form the basis of study may have limited the detection of a positive relationship further research. on this occasion.

Open Access Article. Published on 19 April 2016. Downloaded 18/05/2016 16:41:47. Fig. 5b depicts similar results for hair to those observed for Male gender had a signicant positive effect on hair As in the toenails, albeit with a weaker correlation. Due to problems low drinking water group. Drinking water As, age, gender encountered with sample handling and the difficulty deter- (male), dye usage and smoking were all signicantly positively mining the mass of hair washed, assessing the performance of associated hair As in the high drinking water group. Findings of washing was not possible for hair samples. Sample contami- the model for hair As in the high drinking water group com- nation cannot be ruled out as a possible explanation for this plimented those of Welch's tests, namely the signicantly lower weaker correlation. Based on the results from Welch's t-tests, As concentrations in hair collected from females and those who cigarette smoking might have accounted for elevated As in the reported using dye. The association with dye usage strength- hair of some individuals. Tobacco smoke has been demon- ened with the omission of the gender term. Furthermore, with strated52 to cause elevated As in hair samples from non-occu- all but one volunteer reporting dye usage being female and 29% pationally exposed smokers and passive smokers. This pattern of hair providing volunteers being females that did not report was not evident for toenail As concentrations, suggesting dye usage, the apparent effect of dye implied by Welch's test was external contamination of hair from tobacco smoke among an indirect effect of gender. This would be consistent with smokers as a possible explanation. Although statistically previous ndings49,53 already discussed regarding lower As in signicant, caution is advised when interpreting these results the hair of females. Wolfsperger et al. (1994) attributed the due to the small number of smokers in the present study group. higher As in male hair samples to smoking and a higher intake of seafood and wine than females.53 To test the inuence of food and drink items known to Demographic, behavioural and dietary covariables contain As, dietary terms were added to the abovementioned Multiple linear regression was used to determine signicant models. None of the dietary variables tested had a signicant predictors of toenail and hair As concentrations in addition to effect on either toenail or hair As concentrations in the high drinking water As. These included demographic, behavioural drinking water group. In the low drinking water group, more and dietary covariables. Data were stratied into two groups: servings of seafood per week resulted in a signicant increase in

This journal is © The Royal Society of Chemistry 2016 Environ. Sci.: Processes Impacts,2016,18,562–574 | 571 View Article Online Environmental Science: Processes & Impacts Paper

Table 5 Predictors of toenail and hair As concentrations on the basis of multiple linear regression models. Significant coefficients are labelled with (***), (**), (*), and (.) denoting significance to <0.001, <0.01, <0.05 and <0.1, respectively

Model Terms b coefficient (signicance)

1a. ln(toenail As), drinking water <1 As mgL 1 Intercept 5.309 (***) ln(drinking water As) 0.072 Age (continuous) 0.01 (.) Gender (male) 0.137 Nail polish usage (true) 0.268 Adjusted R2 ¼ 0.07 Smoking status (smoker) 0.31 1b. ln(toenail As), drinking water $1AsmgL 1 Intercept 5.916 (***) ln(drinking water As) 0.469 (***) Age (continuous) 0.018 (**) Gender (male) 0.101 Nail polish usage (true) 0.157 Adjusted R2 ¼ 0.29 Smoking status (smoker) 0.005 2a. ln(hair As), drinking water <1 As mgL 1 Intercept 2.646 (**) ln(drinking water As) 0.08 Age (continuous) 0.017 Gender (male) 0.826 (**) Dye usage (true) 0.159 Adjusted R2 ¼ 0.24 Smoking status (smoker) 0.77 2b. ln(hair As), drinking water $1AsmgL 1 Intercept 5.349 (***) ln(drinking water As) 0.433 (***) Age (continuous) 0.025 (*) Gender (male) 0.810 (**)

Creative Commons Attribution 3.0 Unported Licence. Dye usage (true) 0.76 (.) Adjusted R2 ¼ 0.42 Smoking status (smoker) 2.08 (**) 3. ln(toenail As), drinking water < 1 As mgL 1 Intercept 4.662 (***) ln(drinking water As) 0.089 (.) Adjusted R2 ¼ 0.04 Seafood (continuous) 0.081 (*) 4. ln(hair As), drinking water < 1 As mgL 1 Intercept 4.392 (***) ln(drinking water As) 0.213 (*) Gender (male) 0.905 (***) Home-grown vegetables (never) 0.975 (**) Home-grown vegetables (potted only) 0.546 2 This article is licensed under a Adjusted R ¼ 0.33 Home-grown vegetables (seasonally) 0.343 Open Access Article. Published on 19 April 2016. Downloaded 18/05/2016 16:41:47. toenail As concentration. Specic varieties of seafood were not reported high soil As concentrations in the study region51 and, signicant. The model (Model 3) was re-performed with the although values in local vegetables themselves have been found omission of non-signicant covariables and the results are at relatively low concentrations,55 the ingestion of soil particles presented in Table 5. A negative association was observed adhered to vegetables is a possible exposure pathway. between hair As concentrations and never eating home-grown vegetables. The results of this model (Model 4), with non- 4. Conclusions signicant covariables omitted, are presented in Table 5. The positive association between seafood consumption and This study is the largest investigation of long-term exposure to toenail As concentrations and the negative association between As in drinking water in the UK to-date and conrms the pres- home-grown veg consumption and hair As concentrations are of ence of prolonged exposure to inorganic As from drinking water plausible validity. Although seafood derived arsenic species of householders with PWS in Cornwall, UK. The temporal such as arsenobetaine are primarily excreted via urine,54 sea- stability of As concentrations in PWS suggests that, for this food also contains arsenosugars and arsenolipids which are particular region, measurements of As taken in the present are metabolised into methylarsonate and dimethylarsinate, both of strong predictors of past levels of exposure dating back at least which have been measured in small quantities in human 31 months. Arsenic concentrations measured in toenails and toenails.40 In the present study, drinking water exposure was the hair were useful in assessing prolonged exposure to As from primary focus of the investigation, hence, speciation analysis PWS, in agreement with numerous previous studies. Analysis of was not performed. On the basis of these ndings, future washing solutions built on the ndings of Button et al. (2009)40 studies considering dietary sources in low drinking water in that the washing procedure employed here was effective in exposure groups should consider speciation analysis to ensure removing exogenous contamination from a large sample set. meaningful results. The negative effect of not eating home- Both toenail and hair biomarkers were susceptible to the grown vegetables on hair As concentration is consistent with inuence of covariables on As concentrations. Although useful

572 | Environ. Sci.: Processes Impacts,2016,18,562–574 This journal is © The Royal Society of Chemistry 2016 View Article Online Paper Environmental Science: Processes & Impacts

in assessing prolonged exposures to As from drinking water, drinking water supplies and the impact of domestic other factors, such as diet, predominate where As concentra- treatment systems on water quality, Environmental tions in drinking water are low e.g. <1 mgL 1. A large degree of Geochemistry and Health, 2016, DOI: 10.1007/s10653-016- variation in toenail and hair biomarkers was still unaccounted 9798-0. for in this study, with exposure to soil and dust highly possible 12 Private Water Supplies Regulations 2009, Applying in explanations in a region of well-documented elevated environ- England and coming into force on 1st of January 2010, mental As. Investigation into the signicance of other exposure http://www.legislation.gov.uk/uksi/2009/3101/introduction/ routes will be the focus of future research. made. 13 WHO, Arsenic in drinking-water – Background document for Acknowledgements development of WHO Guidelines for Drinking-water Quality, World Health Organisation, WHO/SDE/WSH/03.04/75/Rev/ The authors are grateful for the contributions of Andrew Dunne 1, 2011. and Dr Andrew Marriott during the eld work campaign, Dr 14 D. R. S. Middleton, M. J. Watts, E. M. Hamilton, E. L. Ander, Louise Ander for help with constructing the eld questionnaire R. M. Close, K. S. Exley, H. Crabbe, G. S. Leonardi, T. Fletcher database, Amanda Gardner for laboratory work and the efforts and D. A. Polya, Urinary arsenic proles reveal substantial of Amy Rimell and Dr Mike Studden in logistical and manage- exposures to inorganic arsenic from private drinking water ment operations. Funding for this research was provided by the supplies in Cornwall, UK, Sci. Rep., 2016, in press. Natural Environment Research Council (NERC) via a University 15 J. P. Buchet, R. Lauwerys and H. Roels, Int. Arch. Occup. of Manchester/British Geological Survey (BGS) University Environ. Health, 1981, 48,71–79. Funding Initiative (BUFI) PhD studentship (Contract No. GA/ 16 DWI, Drinking water 2014, Private water supplies in 125/017, BUFI Ref: S204.2) and the Centre for Environmental England, A report by the Chief Inspector of Drinking Geochemistry, BGS. The participation of the 215 volunteers in Water, July 2015, 2015.

Creative Commons Attribution 3.0 Unported Licence. the wider study is also gratefully acknowledged. More infor- 17 J. G. Thundiyil, Y. Yuan, A. H. Smith and C. Steinmaus, mation can be found at: http://www.bgs.ac.uk/sciencefacilities/ Environ. Res., 2007, 104, 367–373. laboratories/geochemistry/igf/Biomonitoring/arsenicSW.html. 18 C. M. Steinmaus, Y. Yuan and A. H. Smith, Environ. Res., 2005, 99, 164–168. Notes and references 19 M. J. Slotnick, J. R. Meliker and J. O. Nriagu, Sci. Total Environ., 2006, 369,42–50. 1 IARC., IARC Monogr. Eval. Carcinog. Risks Hum., 2012, 20 F. Frost, D. Franke, K. Pierson, L. Woodruff, B. Raasina, 100C, 41–85. R. Davis and J. Davies, Environ. Geochem. Health, 1993, 15, 2 C. J. Chen, Y. C. Chuang, T. M. Lin and H. Y. Wu, Cancer Res., 209–214.

This article is licensed under a 1985, 45, 5895–5899. 21 A. G. Gault, H. A. L. Rowland, J. M. Charnock, R. A. Wogelius, 3 G. Marshall, C. Ferreccio, Y. Yuan, M. N. Bates, C. Steinmaus, I. Gomez-Morilla, S. Vong, M. Leng, S. Samreth, S. Selvin, J. Liaw and A. H. Smith, J. Natl. Cancer Inst., 2007, M. L. Sampson and D. A. Polya, Sci. Total Environ., 2008,

Open Access Article. Published on 19 April 2016. Downloaded 18/05/2016 16:41:47. 99, 920–928. 393, 168–176. 4 J. O'Reilly, M. Watts, R. Shaw, A. Marcilla and N. Ward, 22 M. R. Karagas, J. S. Morris, J. E. Weiss, V. Spate, C. Baskett Environ. Geochem. Health, 2010, 32, 491–515. and E. R. Greenberg, Cancer Epidemiol., Biomarkers Prev., 5 M. Watts, J. O'Reilly, A. Marcilla, R. Shaw and N. Ward, 1996, 5, 849–852. Environ. Geochem. Health, 2010, 32, 479–490. 23 M. R. Karagas, T. D. Tosteson, J. Blum, B. Klaue, J. E. Weiss, 6 D. Chakraborti, M. M. Rahman, B. Das, M. Murrill, S. Dey, V. Stannard, V. Spate and J. S. Morris, Am. J. Epidemiol., 2000, S. Chandra Mukherjee, R. K. Dhar, B. K. Biswas, 152,84–90. U. K. Chowdhury and S. Roy, Water Res., 2010, 44, 5789–5802. 24 M. Y. Zhijie, T. J. Dummer, A. Adams, J. D. Murimboh and 7 D. Chakraborti, B. Das, M. M. Rahman, U. K. Chowdhury, L. Parker, J. Exposure Sci. Environ. Epidemiol., 2014, 24, B. Biswas, A. Goswami, B. Nayak, A. Pal, M. K. Sengupta 135–144. and S. Ahamed, Mol. Nutr. Food Res., 2009, 53, 542–551. 25 M. R. Karagas, C. X. Le, S. Morris, J. Blum, X. Lu, V. Spate, 8 NRC, Arsenic in drinking water, National Research Council. M. Carey, V. Stannard, B. Klaue and T. Tosteson, Subcommittee on Arsenic in Drinking Water, National International Journal of Occupational Medicine and Academies Press, 1999. Environmental Health, 2001, 14, 171–175. 9 A. L. Lindberg, W. Goessler, E. Gurzau, K. Koppova, 26 A. L. Hinwood, M. R. Sim, D. Jolley, N. de Klerk, P. Rudnai, R. Kumar, T. Fletcher, G. Leonardi, K. Slotova E. B. Bastone, J. Gerostamoulos and O. H. Drummer, and E. Gheorghiu, J. Environ. Monit., 2006, 8, 203–208. Environ. Health Perspect., 2003, 111, 187. 10 D. Jovanovic, B. Jakovljevi´c, Z. Raˇsi´c-Milutinovi´c, 27 L. E. Beane Freeman, L. K. Dennis, C. F. Lynch, P. S. Thorne K. Paunovi´c, G. Pekovi´c and T. Knezevi´c, Environ. Res., and C. L. Just, Am. J. Epidemiol., 2004, 160, 679–687. 2011, 111, 315–318. 28 J. E. Heck, A. S. Andrew, T. Onega, J. R. Rigas, B. P. Jackson, 11 E. L. Ander, M. J. Watts, P. L. Smedley, E. M. Hamilton, M. R. Karagas and E. J. Duell, Environ. Health Perspect., 2009, R. Close, H. Crabbe, T. Fletcher, A. Rimell, M. Studden and 117, 1718. G. Leonardi, Variability in the chemistry of private

This journal is © The Royal Society of Chemistry 2016 Environ. Sci.: Processes Impacts,2016,18,562–574 | 573 View Article Online Environmental Science: Processes & Impacts Paper

29 M. J. Slotnick and J. O. Nriagu, Environ. Res., 2006, 102, 125– 43 R Core Team, R: A Language and Environment for Statistical 139. Computing, R Foundation for Statistical Computing, 30 D. K. Harkins and A. S. Susten, Environ. Health Perspect., Vienna, Austria, 2015, http://www.R-project.org. 2003, 111, 576. 44 M. J. Slotnick, J. R. Meliker, G. A. AvRuskin, D. Ghosh and 31 J. Roberge, A. T. Abalos, J. M. Skinner, M. Kopplin and J. O. Nriagu, J. Toxicol. Environ. Health, Part A, 2007, 70, R. B. Harris, Am. J. Environ. Sci., 2009, 5, 688–694. 148–158. 32 J. Xue, V. Zartarian, S.-W. Wang, S. V. Liu and 45 Z. Rivera-Nu´nez,˜ J. R. Meliker, J. D. Meeker, M. J. Slotnick P. Georgopoulos, Environ. Health Perspect., 2010, 118, 345. and J. O. Nriagu, J. Exposure Sci. Environ. Epidemiol., 2011, 33 K. L. Cottingham, R. Karimi, J. F. Gruber, M. S. Zens, 22, 182–190. V. Sayarath, C. L. Folt, T. Punshon, J. Morris and 46 M. L. Kile, E. A. Houseman, E. Rodrigues, T. J. Smith, M. R. Karagas, Nutr. J., 2013, 12,1. Q. Quamruzzaman, M. Rahman, G. Mahiuddin, L. Su and 34 P. Fleckman, in Nails: Therapy, Diagnosis, Surgery, ed. R. K. D. C. Christiani, Cancer Epidemiol., Biomarkers Prev., 2005, Scher and C. R. Daniel, WB Saunders Co, Philadelphia, 14, 2419–2426. 1997, ch. 4, p. 37. 47 D. F. Peach and D. W. Lane, Environ. Geochem. Health, 1998, 35 M. R. Harkey, Forensic Sci. Int., 1993, 63,9–18. 20, 231–237. 36 N. Orentreich, J. Markofsky and J. H. Vogelman, J. Invest. 48 R. N. Ratnaike, Postgrad. Med. J., 2003, 79, 391–396. Dermatol., 1979, 73, 126–130. 49 K. Chojnacka, H. Gorecka´ and H. Gorecki,´ Environ. Toxicol. 37 I. M. Kempson and W. M. Skinner, Biol. Trace Elem. Res., Pharmacol., 2006, 22,52–57. 2012, 150,10–14. 50 B. K. Mandal, Y. Ogra and K. T. Suzuki, Toxicol. Appl. 38 K. Orloff, K. Mistry and S. Metcalf, J. Toxicol. Environ. Health, Pharmacol., 2003, 189,73–83. Part B, 2009, 12, 509–524. 51 P. Mitchell and D. Barr, Environ. Geochem. Health, 1995, 17, 39 P. Favaro, P. Bode and E. De Nadai Fernandes, J. Radioanal. 57–82. – –

Creative Commons Attribution 3.0 Unported Licence. Nucl. Chem., 2005, 264,61 65. 52 A. Saad and M. A. Hassanien, Environ. Int., 2001, 27, 471 40 M. Button, G. R. Jenkin, C. F. Harrington and M. J. Watts, J. 478. Environ. Monit., 2009, 11, 610–617. 53 M. Wolfsperger, G. Hauser, W. Go¨bler and 41 M. Esteban, B. K. Schindler, J. A. Jim´enez-Guerrero, C. Schlagenhaufen, Sci. Total Environ., 1994, 156, 235–242. H. M. Koch, J. Angerer, T. C. Rivas, M. Rosado, S. Gomez,´ 54 A. Navas-Acien, K. A. Francesconi, E. K. Silbergeld and L. Casteleyn and M. Kolossa-Gehring, Environ. Res., 2014, E. Guallar, Environ. Res., 2011, 111, 110–118. 141,24–30. 55 G. Norton, C. Deacon, A. Mestrot, J. Feldmann, P. Jenkins, 42 E. M. Hamilton, T. S. Barlow, C. J. Gowing and M. J. Watts, C. Baskaran and A. A. Meharg, Environ. Sci. Technol., 2013, Microchem. J., 2015, 123, 131–138. 6164–6172. This article is licensed under a Open Access Article. Published on 19 April 2016. Downloaded 18/05/2016 16:41:47.

574 | Environ. Sci.: Processes Impacts,2016,18,562–574 This journal is © The Royal Society of Chemistry 2016 Electronic Supplementary Material (ESI) for Environmental Science: Processes & Impacts. This journal is © The Royal Society of Chemistry 2016

Figure S1 Washing procedure administered to toenail and hair samples. For toenails, the decanted supernatants of steps 1-5 were pooled to form the initial rinse fraction. The supernatant from step 6 was retained separately to form the final rinse fraction. Both were analysed for total As by ICP-MS.

575 Figure S2 Determination of the minimum mass required for the analysis of toenail and hair samples. Triplicate samples of certified reference material (CRM) NC DC 73347 Human Hair were digested and analysed at decreasing mass increments. The criteria used to establish a cut-off mass were: measurement within certified limits (a); mean recovery (red axis) and precision (relative standard deviation (RSD)) (b).

576

CHAPTER 6

Publication:

Arsenic in residential soil and household dust: human exposure assessment criteria and the influence of historical mining. Manuscript in preparation

Journal:

Environmental Science: Processes and Impacts (target journal)

Status as of thesis submission:

Manuscript in preparation

Supplementary information:

Follows publication (p. 38)

95

Arsenic in residential soil and household dust: human exposure assessment criteria and the influence of historical mining

a, b, c b b Daniel R. S Middleton , Michael J. Watts , Darren J. Beriro , Elliott M.

Hamiltonb, Giovanni S. Leonardic, Tony Fletcherc, Rebecca M. Closec and David A.

Polyaa aSchool of Earth, Atmospheric and Environmental Sciences & Williamson Research

Centre for Molecular Environmental Science, University of Manchester, Manchester,

UK, M13 9PL bInorganic Geochemistry, Centre for Environmental Geochemistry, British

Geological Survey, Nottingham, UK, NG12 5GG cEnvironmental Change Department, Centre for Radiation, Chemicals and

Environmental Hazards (CRCE), Public Health England, Chilton, Didcot,

Oxfordshire, UK, OX11 0RQ

Abstract

Exposure to arsenic (As) via residential soil and dust is a global health concern, particularly in regions affected by mining or with elevated concentrations present in underlying geology. Cornwall in south west England is one such area. Residential soil

(n=127) and household dust (n=99) samples were collected from across Cornwall as

1 part of a wider study assessing exposure to environmental As. Samples were analysed for total (soil and dust) and human ingestion bioaccessible (soils with home-grown produce) As. Arsenic concentrations ranged from 12 to 992 mg kg-1 in soil and 3 to

1,079 mg kg-1 in dust and were significantly elevated in regions affected by metalliferous mineralisation. Sixty-nine percent of soils exceeded the 37 mg kg-1 UK

Category 4 Screening Level (C4SL), a generic assessment criteria for As in residential soils which assumes 100 % oral bioaccessibility. The proportion of exceedance was reduced to 17 % and 10 % when the bioaccessibility parameter in the CLEA model was adjusted to the study 25th (13 % bioaccessibility) and 75th (23 % bioaccessibility) percentiles, respectively. Proximity to former As mining locations was a significant predictor of soil As concentration, after adjustment for generalised parent material lithology and spatial correlation. The results of this investigation reiterate previous findings of widespread elevated As in the region. Overall, the study findings highlight the value of bioaccessibility measurements and their role in addressing conservative generic assessment criteria. The importance of proximity to former mines regarding potential exposure to soil As was highlighted.

Keywords: arsenic, residential, soil, dust, mining, exposure

Abbreviations

1. Introduction

Chronic exposure to environmental inorganic arsenic (As) is a recognised risk-factor of numerous cancerous and non-cancerous human health effects (IARC 2012; WHO

2001). Globally, the most significant non-occupational exposure pathway is the ingestion of contaminated groundwater, notable examples include Bangladesh

(Chakraborti et al. 2010) and West Bengal (Chakraborti et al. 2009). Other sources

2 include contaminated food, soil and dust (Thornton 1996a). The latter two media form the focus of this paper.

Arsenic exposure via soil and dust can occur as a result of ingestion, inhalation and dermal absorption (Environment Agency 2009). Specific pathways include: inhalation of soil dust; direct soil ingestion or dermal contact; plant uptake and subsequent ingestion; ingestion of soil adhered to vegetables and direct indoor dust ingestion or dermal contact (Environment Agency 2009). Depending on the human ‘receptor’ in question, the relative importance of these pathways vary. For example, children <6 years old are more likely to be exposed to soil/dust due to the frequency of hand-to- mouth behaviour and thus accidental ingestion (Rieuwerts et al. 2006). Gardeners and home-grown vegetable consumers are also likely to come into contact with soil more frequently (Environment Agency 2009).

The Category 4 Screening Levels (C4SLs) (CL:AIRE 2014) are, health-based, generic assessment criteria (GAC) for inorganic and organic substances in soil for England and Wales. The C4SLs for total As are 37 and 40 mg kg-1 for residential properties with and without home-grown produce, respectively. The C4SLs represent the concentration of a substance in soil that would result in an average daily exposure

(ADE) equal to the lowest level of toxicological concern (LLTC) (CL:AIRE 2014).

For As, the ADE is 0.3 µg per kg of bodyweight per day and aligns with the WHO drinking water guidance value (WHO 2011) of 10 As µg L-1. The residential C4SLs are conservative and make many generic assumptions, one of which is that As uptake from soil is equal to intake i.e. 100 % bioavailable. The CLEA model permits the user to add a site specific relative bioavailable fraction (RBA) to replace the generic setting of 100 %.

3

The Normal Background Concentrations (NBCs) of contaminants (including As) in soils provide a non-health based value for comparison and reflect the observed variation in concentration attributable to underlying geology and diffuse pollution

(Ander et al. 2013). These are defined statistically as the upper 95 % confidence limit

(UCL) of the 95th percentile. Where As in soil exceeds the NBC, there is evidence to suggest that factors in addition to underlying geology and diffuse pollution are contributing to its concentration (i.e. point-source pollution). Derivations are made for given domains where soils exhibit significantly elevated contaminant concentrations

(Ander et al. 2013). The English NBCs for As are 32, 290 and 220 mg kg-1 for principal, mineralised and ironstone domains, respectively (Ander et al. 2013). Figure

1 shows the distribution of the different NBC domains for As across England.

Figure 1 Mapped domains used for the derivation of normal background concentrations (NBCs) of As in English soils. Soils from the Cornwall study area are categorised in either the principal or mineralised domain. Compiled using ESRI ArcMap 10.1. Contains British Geological Survey materials © NERC 2016. Contains Ordnance Survey data © Crown Copyright and database rights 2016.

4

Under normal circumstances, soil parent material is the dominant determinant of As and other elemental concentrations (Rawlins et al. 2003). Anthropogenic activities such as mining and mineral processing can lead to further enrichment of As in soils and household dusts (Thornton 1996b). Being a constituent of many sulphide minerals, notably arsenopyrite (FeAsS), As contamination can result from the mining of numerous associated metalliferous ores (Abrahams and Thornton 1987). Regional examples of mining related As contamination include: gold (Au) mining in many parts of Africa (e.g. artisanal mining in Ghana (Asante et al. 2007; Dartey et al. 2013)),

South America (e.g. Nicaragua (Wickre et al. 2004)) and Oceania (e.g. Australia

(Hinwood et al. 2004; Pearce et al. 2012)); copper (Cu) mining in South America (e.g.

Chile (Flynn et al. 2002)) and Europe (e.g. France (Fillol et al. 2010) and Portugal

(Candeias et al. 2014)) and tin (Sn) and As (as well as Cu) mining in south west

England, UK (Thornton 1996b). Many of these examples also provide evidence of human exposure such as elevated biomarker As concentrations (urine (Dartey et al.

2013; Fillol et al. 2010; Hinwood et al. 2004), hair (Hinwood et al. 2003) and toe/fingernails (Button et al. 2009a; Wickre et al. 2004)) and epidemiological evidence of human health risk (Pearce et al. 2012). Reportedly (Murcott 2012), at least 74 countries worldwide are affected by non-coal mining related As contamination, making the investigation of human exposure to mining-related As contamination an issue of global importance.

Cornwall in south west England, is an area of elevated environmental As. Although concentrations in this highly mineralised region would be expected to be naturally elevated, a history of extensive mining and mineral processing, predominantly of Sn,

Cu and As, has resulted in widespread anthropogenic contamination (Mitchell and

5

Barr 1995). It was estimated (Abrahams and Thornton 1987) that an area of 722 km2

(8 % of the region) is contaminated with As to some extent (>110 As mg kg-1) in south west England. Contamination may arise from mine tailings of which As concentrations vary depending on ore grade, processing efficiency and the economic cut-off point at which the ore was worth processing (Mitchell and Barr, 1995). Measured As concentrations in tailings have been reported at up to 20 % As (Klinck et al. 2005).

Cornwall has many such tailings heaps and the inability of all but the hardiest of plant species to inhabit them leaves them susceptible to wind erosion. One report (Hamilton

2000) describes a 100 m plume of dust when a former As mining site, Devon Great

Consols (DGC), was used as a car racing circuit, a potential source of airborne exposure for local residents. Given Cornwall’s former (mid-19th century) status as the world’s leading As exporter, the region is highly appropriate for investigating the transport and fate of As in the environment and the implications for human exposure and public health.

Studies conducted in south west England have reported elevated As concentrations in residential soils (i.e. >100 mg kg-1 ) (Culbard and Johnson 1984; Rieuwerts et al. 2006;

Xu and Thornton 1985), home/locally-grown vegetables (i.e. relative to control areas

) (Norton et al. 2013; Xu and Thornton 1985), household dust (i.e. >100 mg kg-1 )

(Farago et al. 1999; Rieuwerts et al. 2006), private drinking water supplies (i.e. >10

As µg L-1) (Ander et al. 2016) and human biomarkers (e.g. relative to control volunteers or correlated with environmental concentrations) such as urine (Johnson and Farmer 1989; Kavanagh et al. 1997; Kavanagh et al. 1998; Middleton et al.

2016a), toenails (Button et al. 2009a; Middleton et al. 2016b) and hair (Peach and

Lane 1998). Particular concern has been raised regarding the exposure of infants in

6 the area, with modelled As intakes as high as 2.4 and 3.5 µg kg-1 bw day-1 for dust and soil, respectively (Rieuwerts et al. 2006).

Many of the studies discussed have reported elevated As concentrations in media collected at former mining areas relative to control locations (Button et al. 2009a;

Kavanagh et al. 1998; Norton et al. 2013; Peach and Lane 1998). The relationship between environmental As concentrations (e.g. in soil and dust) and proximity to former mines has not been investigated before on a large (e.g. county-wide) scale. One study (Barnes et al. 2006) investigated ambient air particulate (≤ 10 µm in diameter -

PM10) As concentrations in relation to proximity to, and surrounding density of former mining sites and found no significant correlation.

The present study was part of a wider investigation of human exposure to As in

Cornwall which included biomonitoring (Middleton et al. 2016a; Middleton et al.

2016b) and environmental sampling (Ander et al. 2016), comprising of water, soil and dust. It forms the largest non-occupational focus on environmental As exposure in the

UK to-date. This paper aimed to investigate potential human exposure to As via residential soil and dust and assess the role that historical arsenic mining has played in its distribution across the county of Cornwall in south west England. The aim was addressed with the following objectives:

(i) Measure total As concentrations in residential soils (with and without home-grown produce) and household dust.

(ii) Measure the bioaccessible As concentration in soils used for home-grown produce.

(iii) Derive site specific assessment criteria (SSAC) using the CLEA model and C4SL settings, adjusting for specific soil bioaccessible fractions.

7

(iv) Investigate the relationship between proximity to mining sites and residential soil

As concentrations.

2. Materials and methods

2.1. Ethical approval and household selection

Ethical approval for the overall study was granted by the University of Manchester

Research Ethics Committee (Ref 13068) and the NHS Health Research Authority

National Research Ethics Committee (NRES) (Ref 13/EE/0234). Sampling units consisted of households selected for having a private water supply as per their participation in a wider study of As exposure in south west England (Middleton et al.

2016a; Middleton et al. 2016b). All householders provided written informed consent.

2.2. Sample collection

The soil collection protocol was based on the British Geological Survey’s

Geochemical Baselines Survey of the Environment (G-BASE) methodology (Johnson and Breward 2004). Topsoil (15 cm) samples were collected from vegetable patches

(‘vegetable patch soils’). Where no vegetable patch was present, other uncovered patches of land (‘garden soils’) were used. All samples were stored in Kraft sample bags. Householders were asked to provide information on any modifications made to their residential soils such as the presence of any imported soil or application of compost or manure. Composite indoor dust samples were collected by emptying the contents of the household vacuum cleaner. Information on whether or not there were pets at the property was obtained.

8

2.3. Reagents and standards

Deionised water (DIW) with a resistivity of 18.2 MΩ (Millipore, UK) was used throughout. Nitric (HNO3), hydrochloric (HCl), hydrofluoric (HF) and perchloric

(HClO4) acids and hydrogen peroxide (H2O2) were of Romil-SpA™ super purity grade

(Romil, UK). Arsenic calibration solutions for dust analysis were from a 1000 mg L-1

PrimAg® grade solution (Romil, UK). Arsenic QC standards (25 µg L-1) were prepared from a multi-element solution with As at 20 mg L-1 (Ultra Scientific, USA).

Tellurium (Te) and germanium (Ge) instrument internal standards were prepared from a PlasmaCAL 10,000 mg L-1 solution (SCP Science, Canada) and a Fluka Analytical

1,000 mg L-1 solution (Sigma-Aldrich, USA), respectively. Reagents used during the bioaccessibility protocol were identical to those reported elsewhere (Hamilton et al.

2015).

2.4. Sample preparation and dissolution of soil and dust

Soil samples were oven-dried at 40 °C before being disaggregated and sieved to <2 mm. From this <2 mm fraction, samples for total elemental analysis were ground in an agate ball mill. Pressed sample pellets were prepared using 10 g of sample and 2.5 g of binder wax (PANalytical, UK) for analysis by X-ray fluorescence spectrometry

(XRFS). Vegetable patch soils were further sieved (<250 µm) for bioaccessibility testing. Dust samples were sieved (<250 µm) and weighed (0.25 g) into PFA vials

(Savillex, USA) for digestion on a graphite hot block. Three millilitres of 5 % v/v

HNO3 + 3 mL of 50 % v/v HNO3 were added to vials and dried down overnight at 80

°C. Samples were cooled before adding 2 mL of concentrated HNO3 + 2.5 mL of concentrated HF + 1 mL of concentrated HClO4. The hot block was then set to the following heating program: 8 hrs at 80 °C; 2 hrs at 100 °C; 1 hr at 120 °C; 3 hrs at 140

9

°C; 4 hrs at 160 °C and cooled to 50 °C, before adding 2.5 mL of 50 % v/v HNO3, leaving for 30 mins at 50 °C and cooling down to 30 °C. Ten millilitres of DIW + 2.5 mL of concentrated H2O2 were added and left for 15 mins at 30 °C. Digests were transferred into 30 mL LDPE bottles (Nalgene, USA) and made up to 25 mL with 10 mL of DIW.

Bioaccessible As concentrations in soils were determined using the Bioaccessibility

Research Group of Europe (BARGE) Unified Bioaccessibility Method (UBM) (Denys et al. 2012) following a protocol reported elsewhere (Hamilton et al. 2015).

Bioaccessibility was determined by the UBM method in both the stomach and intestinal compartments of the simulated gastrointestinal tract. The highest concentration from the two compartments was selected as the bioaccessible concentration in this study and is a common approach (Appleton et al. 2012). The bioaccessible fraction (BAF) was calculated as the bioaccessible As concentration as a percentage of the total As concentration.

2.5. Elemental analyses of soil and dust

Soil total As concentrations were measured by an Axios Advanced wavelength- dispersive XRFS instrument (PANalytical, Nottingham, UK). Dust and bioaccessibility digests were diluted ×40 and ×100, respectively with 1 % v/v HNO3

+ 0.5 % v/v HCl and total As concentrations were determined by inductively coupled plasma mass spectrometry (ICP-MS) (Agilent 7500cx series) using previously reported operating conditions (Hamilton et al. 2015). A three-point calibration with concentrations at 1, 10 and 100 µg L-1 and helium (He) collision cell mode was used for total As determination. A multi-element internal standard was introduced via a T- piece and Ge was used to correct for As signal drift. Doubly-charged 150Nd++ and

10

150Sm++ interferences on As were corrected using single element standards at 100 µg

L-1 and the application of a correction factor as described previously (Hamilton et al.

2015).

2.6. Quality assurance/quality control

Seven (4 %) field duplicate soil samples were collected from different auger points at the same location. For each dissolution of 50 dust samples (two in total), four certified reference materials (CRMs), five digestion duplicates and five reagent blanks were digested. The CRMs used were National Institute of Standards and Technology

(NIST) 2584 Indoor Dust (2×0.25g per batch) and NIST 2711a Montana II Soil

(2×0.25g per batch). Pearson correlation between soil field duplicate total As concentrations was 0.996 (n=7) with a mean difference of 8 % (geometric mean (GM):

5 %; range: 1-21 %). The Pearson correlation between dust digestion duplicate total

As concentrations was 0.98 (n=10) with a mean difference of 13 % (GM: 5 %; range:

1-69 %). The mean recovery for NIST 2584 digested with dust samples was 93±3 %

(n=4) and 101±4 % (n=4) for NIST 2711a.

2.7. Spatial and historical mining variables

Mapping and spatial analysis was performed using ArcMap version 10.1 (ESRI,

USA). Generalised lithology was obtained from the BGS Soil Parent Material

Database (Lawley 2009). Metalliferous mineralisation classification (1 km squares shapefile) was produced by BGS using a dataset originally compiled by Ove Arup

(Arup 1990). This was used during the development of As NBCs for English soils

(Fig. 1) and defines the mineralised domain. A spatial dataset of historical As mining sites was compiled using a BGS publication (Dines 1956) containing details of former mining sites across south-west England and the BGS BRITPITS database (Cameron

11

2013), containing entries of active and inactive mineral workings across the UK.

Those listed in Dines’ publication as having produced As were located in BRITPITS.

Ninety three percent of the sites listed by Dines were obtainable from BRITPITS and a further 5 % were located via Google Maps. Where multiple points were present in

BRITPITS for the same name, all points were extracted if they were in the expected location. Household proximities to Cu, Sn and As mines from BRITPITS and then to the refined As-specific records were calculated using the ArcMap ‘near’ tool.

2.8. Assessment criteria derivation and exposure calculations

The RBA parameter in the CLEA software (version 1.071) (Environment Agency

2009) was changed from 100 % to represent the findings of this study to derive SSAC.

This value was compared with existing C4SLs. The RBAs selected were the 25th (P25) and 75th (P75) percentiles of soil BAF from this study, used to provide two scenarios to broadly represent the study population. Full details on the generic exposure and toxicological parameters can be found in the As C4SL report (CL:AIRE 2014).

Average daily exposures (ADE) were calculated to show the potential ADE for a given occupant at each household using the measured soil concentrations using the following equation:

ADE = Cs × (LLTC/AC), [1]

-1 where Cs is the total As concentration measured in residential soil (µg kg ); LLTC is

0.3 µg kg-1 day-1 (the C4SL As LLTC – residential with home grown produce) and

AC is the As C4SL (residential with home grown produce) or the SSAC ( adjusted for bioaccessibility).

12

2.9. Statistical analysis

Statistical analyses and plotting were performed in the R programming environment version 3.2.2 (R Core Team 2013) (base package unless otherwise specified).

Exploratory data analysis was performed to determine distributions and inform test selection. Concentration data in soil and dust were found to be positively skewed leading to the calculation of geometric means (GMs) in addition to arithmetic means.

For the same reason, data were natural log (ln)-transformed prior to statistical analyses. Unequal variances (determined by F tests) and unequal group sizes led to the use of Welch’s test to compare concentration data (ln-transformed) between different groups (e.g. NBC domains). Pearson correlation coefficients (rp) were calculated

(including p-values and 95 % confidence intervals) to test relationships between, for example, soil total As concentrations and those in dust (ln-transformed). The significance between strength in correlations was determined using Williams’ test in the ‘psych’ package (Revelle 2014). Linear regression was used to investigate the relationship between soil As concentration and mine proximity (ln-transformed variables). To test for independence, spatial correlation of residuals was investigated with variogram analysis using the ‘sp’ and ‘gstat’ packages (Pebesma and Bivand

2004, 2005). Generalised linear modelling was used to normalise for spatial correlation using the ‘nlme’ package (Pinheiro et al. 2016).

3. Results and discussion

3.1. Soil total and bioaccessible As concentrations

Summary statistics for total As in vegetable patch, garden, mineralised and principal

NBC domain soils are shown in Table 1 with bioaccessible As concentration and BAF for vegetable patch soils.

13

Table 1 Summary statistics for total As, bioaccessible As and As bioaccessible fraction (BAF) for various sample groups and NBC domains. Sample type or NBC n Min Max Arithmetic Geometric domain mean mean Total As (mg kg-1) All 127 12 992 89 57 All vegetable patch 68 12 992 94 58 All garden 59 16 474 82 55 All principal 56 16 436 55 40 Principal vegetable patch 32 16 436 65 45 Principal garden 24 17 106 41 34 Principal modifieda 13 22 436 96 59 Principal unmodifieda 19 16 146 44 38 All mineralised 71 12 992 115 74 Mineralised vegetable 36 12 992 120 73 patch Mineralised garden 35 16 474 111 75 Mineralised modifieda 13 40 992 171 107 Mineralised unmodifieda 23 12 395 91 59 Bioaccessible As (mg kg-1) Vegetable patch 68 2 87 15 10 Bioaccessible fraction (BAF) (%) All vegetable patch 68 3 57 19 17 aModifications apply to vegetable patch soils only.

One hundred and twenty nine households were visited for this study. Residential soil was collected from 127 (98 %) households, 68 of which (54 %) were using the soil for home-grown produce (vegetable patch soils) and the remaining 59 (46 %) were garden soils. Soils (vegetable patch and garden) from 56 (44 %) households were from the principal NBC domain and 71 (56 %) were from the mineralised NBC domain.

Twenty-six (38 %) of the vegetable patches had been modified in some way (e.g. addition of imported soil, compost or manure). Welch’s tests found no significant difference between As concentrations in vegetable patch and garden soils within NBC domains (principal: p=0.12; mineralised: p=0.88). Higher total As concentrations were measured in modified vegetable patch soils compared to unmodified ones but the difference was not significant (principal: p=0.12; mineralised: p=0.06). It is noted that these analyses were constrained by small group sizes. Unsurprisingly, soils from the

14 mineralised domain were found to contain significantly higher total As concentrations than those from the principal domain (GM: 74 versus 40 As mg kg-1). No significant difference was found in the BAF of vegetable patch soils between different NBC domains (p=0.33) or soil modifications (p=0.07).

These findings are consistent with previously reported high As concentrations in residential soils in south west England. Those measured in 1,154 topsoil samples collected for the G-BASE (Johnson and Breward 2004) south west England (Devon and Cornwall) campaign ranged from 5 to 1,949 mg kg-1 (mean: 50 mg kg-1) (BGS

2016). Culbard and Johnson (1984) reported a range of 119-1,130 mg kg-1 in garden soils collected from the former mining area of Camborne and . Farago and

Kavanagh (1999) reported concentrations of 120-1,695 mg kg-1 from gardens in the former mining area of Gunnislake and 345-52,600 mg kg-1 at the Devon Great Consols

(DGC) mine. Xu and Thornton (1985) reported a range of 144-892 mg kg-1 in soils used for home-grown vegetable production in the former mining areas of Hayle,

Camborne and Godolphin.

The BAF of vegetable patch soils in the present study (mean: 19 %; P25: 13 %; P75:

23 %) are comparable with previously reported As bioaccessibility in soils from across the UK (2 to 68 % (Appleton et al. 2012)) and south west England (10-20 %

(Rieuwerts et al. (2006); 16 % (single measurement) (Cave et al. 2002); 10 to 34 %

(mean: 19 %) (Button et al. 2009b)). Whilst BAF estimates have been found to be higher in mining areas (15 %) relative to other mineralised soils with no previous history of mining activity (9 %), the overall BAF of As in soils remains low (Palumbo-

Roe and Klinck 2007). These findings show that the C4SLs can be overly conservative. This observation is important because it has been reported that the assumption of 100 % RBA in risk assessment can lead to unnecessary remediation of

15 potentially contaminated land and potential blight for homeowners that live within such areas (Nathanail 2009; Nathanail and Smith 2007).

3.2. Dust total As concentrations

Dust samples were collected from 99 (77 %) households. Summary statistics for total

As concentrations are presented in Table 2.

Table 2 Summary statistics of total As in composite indoor dust (vacuum cleaner) samples. NBC domain n Min Max Arithmetic Geometric mean mean Total As (mg kg-1) All 99 3 1,078 84 41 Principal 40 5 903 54 28 Mineralised 59 3 1,078 104 54

Previous studies have measured total As concentrations in household dust samples in south west England and the findings presented in this paper are within a similar range.

Farago and Kavanagh (1999) reported As concentrations of 24 to 3,740 and 33 to

1,160 mg kg-1 in two separate mining locations. Culbard and Johnson (1984) reported a lower concentration range of 1 to 330 mg kg-1 in a former mining village, as did

Rieuwerts et al. (2006) (43-486 mg kg-1). These concentrations can be considered elevated in that they are much higher than concentrations from a previously studied non-mining area in Cornwall (e.g. 1.7-29 mg kg-1 (Rieuwerts et al. 2006)). A survey of household vacuum cleaner dusts (n=1025) conducted in Canada reported lower concentrations (range: 0.1-153 mg kg-1; GM: 7.7; P95: 40.6) (Rasmussen et al. 2013) than the present study, indicating that many households in Cornwall have elevated As in dust relative to a nationally representative background.

16

Dust from households in the mineralised domain contained significantly (p=0.001) higher As concentrations than those in the principal domain. A weak, but significant

(p<0.01), Pearson correlation was observed between residential soil and dust As concentrations (rp=0.26; 95 % CI=0.07, 0.44). The median ratio of dust/soil total As concentration was 0.62, broadly comparable to the soil-to-dust transport factor of 0.5 used in the CLEA model (Environment Agency 2009). Previous studies have failed to detect a significant correlation between soil and dust As concentrations (e.g. Rieuwerts et al. 2006), whereas others have found weak (0.40), but significant, correlations

(Keegan and Hong 2002). Of the 97 households where both soil and dust was collected, 48 reported having pets in the house and 29 reported having no pets. For the remaining 20 households, pets were reportedly kept outside or results were ambiguous. A significant (p=0.01), weak correlation (rp=0.35; 95 % CI=0.08, 0.58) was found between soil and dust As concentrations in households with pets. There was no significant (p=0.55) correlation between soil and dust As concentrations in households without pets (rp=0.12; 95 % CI=-0.26, 0.46). Chemical elements and other toxicants have the potential to be tracked indoors by pets and on the surfaces of footwear (CL:AIRE 2014; Lioy et al. 2002). While these findings provide evidence of this mechanism, the difference between correlations was not found to be significant

(p=0.31) using Williams’ test. It is evident that household specific factors affect indoor household As concentrations. Factors such as the number of householders, their occupation and climatic conditions have previously been associated with indoor As and other element concentrations (Meyer et al. 1999), but were not investigated in the present study.

17

3.3. Generic assessment criteria (C4SL) and normal background concentrations

Soil concentration exceedances in relation to the C4SL and domain specific NBCs are summarised in Table 3. The C4SLs of 37 mg kg-1 (with home-grown produce) and 40 mg kg-1 (without home-grown produce) apply to vegetable patch and garden soils, respectively.

Table 3 Exceedances of the C4SL GAC for total As in residential soils and the NBCs of total As in English soils. Results are presented for different sample types and NBC domains. Sample type or NBC n Value n exceeding % exceeding domain (As mg kg-1) C4SL (with home-grown produce) All vegetable patch 68 48 71 Principal 32 37 20 63 Mineralised 36 28 78 C4SL (without home-grown produce) All garden 59 39 66 Principal 24 40 9 38 Mineralised 35 30 86 Principal NBC All Principal 56 32 29 52 Mineralised NBC All Mineralised 71 290 6 8

A high proportion of households across the study region exceeded the As C4SL value for soils with home-grown produce (71 %) and without (66 %). This proportion was expectedly higher in the mineralised domain, especially for the group without home grown produce. The NBC for the mineralised domain (290 mg kg-1) is almost 10-times the value of the C4SL, indicating that a large number of households across the UK are also likely to be in exceedance. Only 8 % of households in the mineralised domain exceeded the NBC, whereas 52 % exceeded the NBC in the principal domain. Given the widespread legacy of mining in Cornwall, many of the concentrations measured may have been above the NBC due to soil As inputs extending beyond diffuse

18 pollution. Additionally, it is possible that the spatial resolution of the NBC domains may have resulted in the misclassification of households in the study area. For example, households categorised as principal domain that reside on localised, unmapped mineralisation.

3.4. Bioaccessibility-adjusted assessment criteria

Figure 2 shows the distribution of the soil As BAF across both domains and the P25

(13 %) and P75 (23 %) BAFs used to update the RBA parameter in the CLEA model to generate less conservative SSAC that are more representative of the soils collected in the present study. With the exception of RBA, all other parameters in the CLEA model were left unchanged. The land-use scenario applied was residential with consumption of home-grown produce, which assumes As exposure of a 0-6 year old female receptor. This is conservative since most households in the present study were not always occupied by young children (16 % of households with occupants <16 years old) and 46 % do not grow their own produce. Notwithstanding, this degree of conservativism was considered acceptable because of the uncertainty over future occupancy rates and associated land-use and the need to generalise in the absence of a site-specific risk assessment.

19

Figure 2 Histogram of soil As bioaccessible fractions (BAF) and the 25th (P25) and 75th (P75) percentiles. These values were selected to update the relative bioavailability (RBA) parameter in the CLEA model to derive SSAC.

Estimates of soil As BAF from this study (mean: 19 %; P25: 13 %; P75: 23 %) show that the generic assumption of 100 % RBA used to derive the C4SLs using the CLEA model is over-conservative. The results from Welch’s test found no significant difference (p=0.33) between the BAF estimates from both NBC domains. For this reason, two BAF values (P25 and P75) were selected to use as the site specific RBA values in the derivation of SSACs, the same for both domains. The P25 and P75 were chosen because these values are representative of a large proportion of the homes in the study area. A representative range for the study area was selected because: (i) there was a weak correlation between BAF and bioaccessible concentration (rp=0.29; p=0.02; 95 % CI=0.06, 0.50); (ii) there was a strong correlation between the total and bioaccessible concentration (rp=0.82; p<0.001; 95 % CI=0.73, 0.89) and (iii) soils with 20 a higher BAF (i.e. >30 %) had lower concentrations of total As (i.e. ~50 mg kg-1) and thus were less likely to exceed assessment criteria.

The SSAC derived using 13 % (SSAC13) and 23 % (SSAC23) RBA were 190 and

129 mg kg-1, respectively. Equation 1 was used to calculate ADEs for the 100, 23 and

13 % RBA parameters. Individual residential soils are plotted against the C4SL and both of the RBA-adjusted SSACs for both the principal and mineralised domains in

Fig. 3A. Figure 3B presents the findings as they equate to ADE relative to the LLTC of 0.3 µg kg-1 day-1 and the total household soil As concentration.

Figure 3 Total As concentrations (A) for individual residential soils in relation to the C4SL GAC and the two derived SSAC. Modelled ADEs (B) using the three RBA scenarios (100, 23 and 13 %) in relation to the LLTC of 0.3 µg kg-1 day-1. Findings are presented for both principal and mineralised domain soils.

Average daily exposure exceedances of the LLTC were reduced following adjustment of exposure parameters with 23 % (17 % of households >SSAC23) and 13 % BAF (10

% of households >SSAC13) in comparison to the 69 % of households exceeding the

21

C4SL. Only two households in the principal domain exceeded SSAC13 in comparison to 11 in the mineralised domain. This implied that, whilst the derivation of an SSAC for this study resulted in a large reduction in the number of household exceedances, a small number of residential soil As concentrations, particularly in the mineralised domain, still equated to ADEs above the LLTC. Rieuwerts et al. (2006) used a similar approach to derive ADEs using bioaccessibility data and reported that 0.3 µg kg-1 day-

1 was exceeded by 75 % of households in a former mining area. Whilst it is acknowledged that these estimates are still conservative, they are consistent with previously raised concerns (Mitchell and Barr 1995; Rieuwerts et al. 2006) that infants and small children may be particularly vulnerable to As exposure in particularly elevated spatial locations such as mineralisation and in proximity to former mining sites.

Further work is needed to quantify exposure of the adult population via pathways such as home-grown vegetable consumption in the study region. Studies conducted elsewhere have found this to be a substantial pathway of exposure. One study

(Ramirez-Andreotta et al. 2013) conducted in Arizona, investigated gardening and home-grown vegetable consumption at properties in the vicinity of a former mining and smelting site. It was reported that garden soils and home-grown vegetables accounted for 16 and 7 % of daily intake, respectively. The authors used correlations between As concentrations in soils and vegetables to derive maximum allowable concentrations in soils to limit excess cancer risk to 10-6 (i.e. one in a million). These estimates ranged from 1.56 mg kg-1 to 12.4 mg kg-1 for different vegetable families, of which 100 and 99 % of soils in the present study exceeded, respectively. A detailed assessment of home-grown vegetables consumption was not conducted in the present study, but the findings from Arizona (Ramirez-Andreotta et al. 2013), and that lower

22 toenail As concentrations were associated with never consuming any type of home- grown produce in this Cornwall cohort (Middleton et al. 2016b), make this a topic worthy of further investigation. Nevertheless, with a local population of approximately

40, 000 0-6 year olds (ONS 2011) and a widespread elevation of environmental As, the estimates in the present paper have relevance to the wider area of Cornwall, as well as comparable locations elsewhere in the world.

3.5. Spatial influences on residential As concentrations

A dataset containing the names and locations of 103 As-producing mines in Cornwall was generated from BRITPITS and Dines’ publication. These mines, in addition to other Cu and Sn mines and mineralisation in relation to households are plotted in Fig.

4. The names of individual mines, and the digitalised mining districts (Dines 1956) are shown in Fig. S1. Geometric mean household distances to all mining sites were

4.4 and 1.2 km for the principal and mineralised domains, respectively and 7.1 and 3.2 km for As-specific mines. Due to the inter-domain differences and the significantly higher soil As concentrations in the mineralised domain, regression analysis was performed on separate domains. Two households in each domain shared the same GPS location. Both pairs were within 6 km of a mining site and, as they were not spatially independent observations, one of each pair was randomly excluded.

23

E

E E E E Launceston " E E #* #* E #*#* E #*")#*") #* #*")#*#*#*#* E E #*#*#*#*#*")#*#*#* E #* #*#*#*#* #*")#* #* #* E E #* #* #*#*#*")#* #*")#*#*#*#*#*#*#* #*#* #* E ") #*")") #*#*#*#*#* #*")#*")#*")#*")#*")#*")")E #*")#* #* #*")#*")#*")")#*#*#*")#*") #*") #*#* E #* #* #*E EBodminE E #*")#*#* " #*") #* E E E #*E E E E #*#*E#* E E #*") E E E#*") " #* E E #* E E " #*#*#*#*#* E E #* E E E EEE E E E #*#* #*")#*") E " #*") #*") #*#* #*") E E #*")#*") #*")#*")#*") EE E E #*") #*E") #*E E") " Redruth#*")#*") #*")#* Camborne #*#*#*") #*") ")E E #*#*#*#*#*#*#*#*")#*")#*#*")#*#*")")#*") E E E #*#*#*")#*")#*")#*#*#*")#*#*#*#*")#*#*") #*")#*")#* E #*#* E #*#* #*")")#*#*")#*")#*#*")#*#*")#*#*#* #*#*#*")#* #*E E E #* " ")#*#*#*#*#*#*#* E E#*#* #* #*")#*")#*")#*#*#*#*#*#*#*#*")#*") E#* ± #*#*#*#* #* E #* #*#* #* #*")#*#*")#*#*") #*#* E #* E#*E#*#*#*#*#*#* #* #*")#*")#*") #* #*#* E #*#*#*#*E#* #* E#*")#*#* #*") #* EE E EE E#*#* 0 5 10 20 30 40 #*#*")#*#*#*E#* #* #*")#*#* E#*#*#* E #*")#*")#* ") E#* #*#* #*#*#* " Kilometers #*E#*#*")#* #*")#*#*#*#*#*#*E #*#*#* #* #*#*#*#*#*#*E #*#*#*")#*#*#*#*")#*#* E#*#*#* #*E#* #*#*#* #* #*#*E " #*#*#*#*#*#*#*#*#*#*") #*#*#*#*#*E#*#*#*#*#*#*#*#*E E E #*#*#*E #*#*#*#*#*#*") #* #*#*")#*#* #*#*E#*#*#* #* E Falmouth ") #* #*#*#*E#*#*")#*E#* #*") EEEE#* As-specific mining sites #*#* EE E #* #* E E As-specific + Sn and Cu mining sites E E E E Households (sampling units) #*") #* #* Mineralised domain

Figure 4 The spatial distribution of As-specific former mining sites, located from BRITPITS and Dines (1956), in relation to sampled households. The mineralised NBC domain and additional Cu and Sn mines are also plotted. Compiled using ESRI ArcMap 10.1. Contains British Geological Survey materials © NERC 2016. Contains Ordnance Survey data © Crown Copyright and database rights 2016.

Several linear models were initially tested for log-residential As concentrations as a function of log-mining proximity. These models showed that distance to all mines was not a significant predictor of total soil As in the mineralised domain (r2=0.03; p=0.15), however, it was significantly (p=0.01), but weakly (r2=0.11) inversely associated with soil As in the principal domain. Distance to As-specific mines was significantly

(p<0.001) inversely related (r2=0.28) with total soil As in the mineralised domain but not in the principal domain (r2<0.01; p=0.65). Distance to mine (all or As-specific) was not a significant predictor of dust As concentration in either domain.

24

The inverse relationship between soil As concentration and As-specific mine proximity in the mineralised domain (Model-1) was chosen for further investigation.

It was necessary to investigate potentially confounding variation that could result in a non-causal relationship between distance to mine and soil As concentration. The locations of the mines are dictated by the economic ore grade at given locations (Dines

1956), determined by local geochemistry. Soil element concentrations are subject to geochemical controls with parent material (29 %) and lithostratigraphy (36 %) shown to account for As concentration variance (Rawlins et al. 2003). Whilst it is not the aim of this paper to investigate the specific geochemical controls on As concentrations, it was necessary to incorporate the effect of natural soil forming processes in conjunction with the anthropogenic input under investigation. Generalised soil parent material lithology (see Lawley 2009 for classifications) was used to adjust the regression model as well as an interaction term for the relationship between mine proximity and lithology. Although less significant, mine proximity remained a significant (p=0.01) predictor of soil As concentration. Granite (p<0.001) and sandstone (p=0.047) had significant negative effects on soil As concentration and the interaction between mine proximity and both granite (p=0.01) and metamudstone (p=0.03) both had significant positive effects. The adjusted r-squared of the lithology adjusted model (Model-1A) was 0.54 in comparison to 0.28 pre-adjustment.

Due to the spatial nature of the data under investigation, points occupying nearby locations in space with similar mining proximities and soil As concentrations had the potential to violate model independence. As an additional, independent validation of the relationship between soil As concentration and mine proximity, the Model-1 residuals were investigated for spatial correlation using a variogram. Residuals exhibited a spatial correlation up to approximately 8 km and, therefore a spatial

25 correlation structure was added to a generalised least squares (GLS) model (Model-

1B) of soil As concentration against mine proximity. Several correlation structures

(see Militino 2010) were tested and, using Akaike’s information criterion (AIC), a spherical structure yielded the best fit and was significantly better than the GLS without the addition of the correlation structure (AIC: 159 versus 168; ANOVA p=0.002). The Model-1B residuals exhibited no spatial correlation following normalisation with the spherical structure and soil As concentration remained significantly inversely correlated with mine proximity. The results of Model-1, Model-

1A and Model-1B are displayed in Table 4.

Table 4 Regression model results for total As concentration in residential soil (mineralised domain) as a function of proximity to As-specific former mining sites. (***), (**) and (*) denote significance to <0.001, <0.01 and <0.05, respectively. Model Terms Coefficient p-value 1 r2=0.28 Intercept 4.95 (***) <0.001 ln(As-specific mine proximity) -0.54 (***) <0.001 1A Adjusted r2=0.54 Intercept 7.35 (***) <0.001 ln( As-specific mine proximity ) -1.34 (*) 0.01 Clay-Silt-Sand-Gravel -0.13 0.93 Granite -3.82 (**) 0.006 Metamudstone-Pelite -3.78 0.24 Metamudstone -2.21 0.090 Metamudstone-Metasandstone -2.34 0.08 Microgranite -12.79 0.17 Mudstone-Sandstone -2.09 (*) 0.048 Sand -1.28 0.19 Schist -1.40 0.11 ln(Mine proximity)*Silt-Sand-Gravel -0.70 0.36 ln(Mine proximity)*Granite 1.51 (*) 0.01 ln(Mine proximity)*Metamudstone-Pelite 1.50 0.34 ln(Mine proximity)*Metamudstone 1.17 (*) 0.03 ln(Mine proximity)*Metamudstone-Metasandstone 0.72 0.20 ln(Mine proximity)*Microgranite 5.71 0.22 ln(Mine proximity)*Mudstone-Sandstone NA NA ln(Mine proximity)*Sand NA NA ln(Mine proximity)*Schist NA NA 1B Intercept 4.89 (***) <0.001 ln( As-specific mine proximity ) -0.47 (***) <0.001

26

The widespread As contamination resulting from the extensive mining operations in

Cornwall’s past have been widely reported (Abrahams and Thornton 1987; Hamilton

2000; Mitchell and Barr 1995; Thornton 1996b) as well as high concentrations at specific, heavily contaminated locations (Camm et al. 2004; Dybowska et al. 2006).

The utility of the As-specific refined mining dataset generated in this study highlighted the importance of individual mining site characteristics in how residential As, and other elemental contamination is distributed in the study region. Although this relationship cannot be confirmed as causal based on the analysis presented in this paper, the data may help local authorities to prioritise certain areas for investigation.

A limitation of the data used in the present paper is acknowledged, in that site-specific variables were not available to include in analyses. Sites require investigation to quantify the levels of contamination present at a given site (e.g. Klinck et al. 2005) and the spread of contamination around former workings (e.g. Camm et al. 2004).

4. Conclusions

This study was the largest of its kind to be conducted at residential properties in the region to date and has confirmed widespread elevated concentrations of As in soil and dust in Cornwall, south west England. A high proportion of households exceeded the

C4SL for As in residential soils. The human ingestion bioaccessibility data quantified for this study enabled the derivation of SSAC that were less conservative than the

C4SL. The number of household exceedances of SSAC were substantially reduced in comparison to the C4SL. A small number of households, particularly in mineralised areas, remained in exceedance. Further investigation is warranted to assess the exposure of the local population, particularly small children and home-grown vegetable consumers, to As in residential soil and dust. The compilation of an As- specific historical mining dataset enabled the identification of a suggested relationship

27 between residential soil As concentration and proximity to historical As workings. The findings presented in this paper could be used by land use professionals and policy makers in the study region, and other locations across the globe that are affected by mining-related elemental contamination, to identify priority areas of concern and mitigate human exposure.

Acknowledgments

The authors gratefully acknowledge the contributions of Dr Mark Cave for statistical and scientific review of the manuscript. Andrew Dunne and Dr Andrew Marriott are thanked for their participation in field work and Dr Louise Ander for help with constructing the field database. Joshua Coe is thanked for contributions to laboratory analysis. Dr Helen Crabbe, Dr Karen Exley, Amy Rimell and Dr Mike Studden are thanked for their contributions to the wider project. Funding for this research was provided by the Natural Environment Research Council (NERC) via a University of

Manchester/British Geological Survey (BGS) University Funding Initiative (BUFI)

PhD studentship (Contract No. GA/125/017, BUFI Ref: S204.2) and the Centre for

Environmental Geochemistry, BGS. The participation of the 215 volunteers in the wider study is also gratefully acknowledged.

28

References

Abrahams P, Thornton I. 1987. Distribution and extent of land contaminated by arsenic and associated metals in mining regions of southwest England. Institution of Mining and Metallurgy Transactions Section B Applied Earth Science 96.

Ander E, Watts M, Smedley P, Hamilton E, Close R, Crabbe H, et al. 2016. Variability in the chemistry of private drinking water supplies and the impact of domestic treatment systems on water quality. Environmental Geochemistry and Health:1-20.

Ander EL, Johnson CC, Cave MR, Palumbo-Roe B, Nathanail CP, Lark RM. 2013. Methodology for the determination of normal background concentrations of contaminants in English soil. Science of the Total Environment 454:604-618.

Appleton J, Cave M, Wragg J. 2012. Anthropogenic and geogenic impacts on arsenic bioaccessibility in UK topsoils. Science of the Total Environment 435:21-29.

Arup O. 1990. Mining instability in Britain. Department of environment contract. Hardcopy maps digitised and adapted by the British Geological Survey.

Asante KA, Agusa T, Subramanian A, Ansa-Asare OD, Biney CA, Tanabe S. 2007. Contamination status of arsenic and other trace elements in drinking water and residents from Tarkwa, a historic mining township in Ghana. Chemosphere 66:1513- 1522.

Barnes J, Ledbrooke A, Parsons B, Salter L, Unit AQ. 2006. Monitoring of ambient air particulate arsenic concentrations at nine sites in Cornwall.

BGS. 2016. G-BASE regional geochemistry map for arsenic (As) in soils in south- west England. 2016. British Geological Survey, Nottinghamshire, UK. Available: http://www.bgs.ac.uk/gbase/gbasesw.html.

29

Button M, Jenkin GR, Harrington CF, Watts MJ. 2009a. Human toenails as a biomarker of exposure to elevated environmental arsenic. Journal of Environmental Monitoring 11:610-617.

Button M, Watts MJ, Cave MR, Harrington CF, Jenkin GT. 2009b. Earthworms and in vitro physiologically-based extraction tests: Complementary tools for a holistic approach towards understanding risk at arsenic-contaminated sites. Environmental Geochemistry and Health 31:273-282.

Cameron D. 2013. User guide for the BRITPITS GIS dataset. British Geological Survey, Nottingham. Available: http://nora.nerc.ac.uk/503964/. Accessed: 20/02/16.

Camm GS, Glass HJ, Bryce DW, Butcher AR. 2004. Characterisation of a mining- related arsenic-contaminated site, Cornwall, UK. Journal of Geochemical Exploration 82:1-15.

Candeias C, Melo R, Ávila PF, da Silva EF, Salgueiro AR, Teixeira JP. 2014. Heavy metal pollution in mine–soil–plant system in San Francisco de Assis–Panasqueira mine (Portugal). Applied Geochemistry 44:12-26.

Cave M, Wragg J, Palumbo B, Klinck B, Mcleod K. Physico-chemical controls on the bioaccessibility of arsenic in UK soils. In: Proceedings of the EGS General Assembly Conference Abstracts, 2002, Vol. 27, 3908.

Chakraborti D, Das B, Rahman MM, Chowdhury UK, Biswas B, Goswami A, et al. 2009. Status of groundwater arsenic contamination in the state of West Bengal, India: A 20‐year study report. Molecular Nutrition & Food Research 53:542-551.

Chakraborti D, Rahman MM, Das B, Murrill M, Dey S, Chandra Mukherjee S, et al. 2010. Status of groundwater arsenic contamination in Bangladesh: A 14-year study report. Water Research 44:5789-5802.

30

CL:AIRE. 2014. SP1010 – Development of Category 4 Screening Levels for assessment of land affected by contamination final project report (revision 2). Contaminated Land: Applications in Real Environments (CL:AIRE). Available: http://randd.defra.gov.uk/Default.aspx?Module=More&Location=None&ProjectID= 18341.

Culbard E, Johnson L. 1984. Elevated arsenic concentrations in house dusts located in a mineralised area of south-west England: Implication for human health. Trace Substances in Environmental Health XVIII. University of Missouri, Columbia: 311- 319.

Dartey E, Sarpong K, Darko G, Acheampong-Marfo M. 2013. Urinary arsenic and mercury levels in artisanal miners in some communities in the Obuasi municipality of Ghana. Journal of Environmental Chemistry and Ecotoxicology 5:113-118.

Denys S, Caboche J, Tack K, Rychen G, Wragg J, Cave M, et al. 2012. In vivo validation of the unified barge method to assess the bioaccessibility of arsenic, antimony, cadmium, and lead in soils. Environmental Science & Technology 46:6252- 6260.

Dines HG. 1956. The metalliferous mining region of south-west England: HM Stationery Office.

Dybowska A, Farago M, Valsami-Jones E, Thornton I. 2006. Operationally defined associations of arsenic and copper from soil and mine waste in south-west England. Chemical Speciation and Bioavailability 17:147-160.

Farago M, Kavanagh P, Ármannsson H. 1999. High arsenic-containing soils in SW England and human exposure assessment. In: Proceedings of the Geochemistry of the earth's surface Proceedings of the 5th International Symposium, Reykjavik, Iceland, 15-20 August 1999. AA Balkema, 181-184.

31

Fillol C, Dor F, Labat L, Boltz P, Le Bouard J, Mantey K, et al. 2010. Urinary arsenic concentrations and speciation in residents living in an area with naturally contaminated soils. Science of The Total Environment 408:1190.

Flynn HC, Mc Mahon V, Diaz GC, Demergasso CS, Corbisier P, Meharg AA, et al. 2002. Assessment of bioavailable arsenic and copper in soils and sediments from the Antofagasta region of northern Chile. Science of the Total Environment 286:51-59.

Hamilton EI. 2000. Environmental variables in a holistic evaluation of land contaminated by historic mine wastes: A study of multi-element mine wastes in west Devon, England using arsenic as an element of potential concern to human health. Science of The Total Environment 249:171-221.

Hamilton EM, Barlow TS, Gowing CJ, Watts MJ. 2015. Bioaccessibility performance data for fifty-seven elements in guidance material BGS 102. Microchemical Journal 123:131-138.

Hinwood AL, Sim MR, Jolley D, de Klerk N, Bastone EB, Gerostamoulos J, et al. 2003. Hair and toenail arsenic concentrations of residents living in areas with high environmental arsenic concentrations. Environmental Health Perspectives 111:187.

Hinwood AL, Sim MR, Jolley D, de Klerk N, Bastone EB, Gerostamoulos J, et al. 2004. Exposure to inorganic arsenic in soil increases urinary inorganic arsenic concentrations of residents living in old mining areas. Environmental Geochemistry and Health 26:27-36.

IARC. 2012. Arsenic, metals, fibers and dusts. IARC Monogr Eval Carcinog Risks Hum 100C:41-85. International Agency for Research on Cancer, Lyon, France.

Environment Agency. 2009. Updated technical background to the CLEA model. Available: https://www.gov.uk/government/publications/updated-technical- background-to-the-clea-model.

32

Johnson C, Breward N. 2004. G-BASE: Geochemical Baseline Survey of the Environment. British Geological Survey, Nottingham, UK. (CR/04/016N) (Unpublished). Available: http://nora.nerc.ac.uk/509527/

Johnson LR, Farmer JG. 1989. Urinary arsenic concentrations and speciation in Cornwall residents. Environmental Geochemistry and Health 11:39-44.

Kavanagh P, Farago ME, Thornton I, Elliott P, Goessler W, Irgolic KJ. 1997. Urinary arsenic concentration in a high arsenic area of south west England. Occupational and Environmental Medicine 54:840.

Kavanagh P, Farago M, Thornton I, Goessler W, Kuehnelt D, Schlagenhaufen C, et al. 1998. Urinary arsenic species in Devon and Cornwall residents, UK. A pilot study. Analyst 123:27-29.

Keegan T, Hong B. 2002. Assessment of environmental arsenic levels in Prievidza district. Journal of Exposure Analysis & Environmental Epidemiology 12.

Klinck B, Palumbo-Roe B, Cave M, Wragg J. 2005. Arsenic dispersal and bioaccessibility in mine contaminated soils: A case study from an abandoned arsenic mine in Devon, UK. (RR/04/003). British Geological Survey, Nottingham, UK. Available: http://nora.nerc.ac.uk/3681/.

Lawley R. 2009. The soil-parent material database: a user guide. British Geological Survey, Nottingham, UK. (OR/08/034) (Unpublished). Available: http://nora.nerc.ac.uk/8048/.

Lioy PJ, Freeman NC, Millette JR. 2002. Dust: A metric for use in residential and building exposure assessment and source characterization. Environmental Health Perspectives 110:969.

Meyer I, Heinrich J, Lippold U. 1999. Factors affecting lead, cadmium, and arsenic levels in house dust in a smelter town in eastern Germany. Environmental Research 81:32-44.

33

Middleton D, Watts M, Hamilton E, Ander E, Close R, Exley K, et al. 2016a. Urinary arsenic profiles reveal substantial exposures to inorganic arsenic from private drinking water supplies in Cornwall, UK. Scientific Reports 6, 25656.

Middleton D, Watts M, Hamilton E, Fletcher T, Leonardi G, Close R, et al. 2016b. Prolonged exposure to arsenic in uk private water supplies: Toenail, hair and drinking water concentrations. Environmental Science: Processes & Impacts. 18, 562-574.

Militino AF. 2010. Mixed effects models and extensions in ecology with R. Journal of the Royal Statistical Society: Series A (Statistics in Society) 173:938-939.

Mitchell P, Barr D. 1995. The nature and significance of public exposure to arsenic: A review of its relevance to south west England. Environmental Geochemistry and Health 17:57-82.

Murcott S. 2012. Arsenic contamination in the world: An international sourcebook 2012. Water Intelligence Online 11:9781780400396.

Nathanail C. 2009. Professional practice note: Reviewing human health risk assessment reports invoking contaminant oral bioavailability measurements or estimates. Chartered Institute for Environment and Health.

Nathanail CP, Smith R. 2007. Incorporating bioaccessibility in detailed quantitative human health risk assessments. Journal of Environmental Science and Health Part A 42:1193-1202.

Norton G, Deacon C, Mestrot A, Feldmann J, Jenkins P, Baskaran C, et al. 2013. Arsenic speciation and localization in horticultural produce grown in a historically impacted mining region. Environmental Science & Technology 47:6164-6172.

34

ONS. 2011. Office for National Statistics, 2011 census: Aggregate data (England and Wales). UK data service census support. Downloaded from: http://infuse.ukdataservice.ac.uk. This information is licensed under the terms of the open government licence [http://www.Nationalarchives.Gov.Uk/doc/open- government-licence/version/2].

Palumbo-Roe B, Klinck B. 2007. Bioaccessibility of arsenic in mine waste- contaminated soils: A case study from an abandoned arsenic mine in SW England (UK). Journal of Environmental Science and Health Part A 42:1251-1261.

Peach DF, Lane DW. 1998. A preliminary study of geographic influence on arsenic concentrations in human hair. Environmental Geochemistry and Health 20:231-237.

Pearce DC, Dowling K, Sim MR. 2012. Cancer incidence and soil arsenic exposure in a historical gold mining area in Victoria, Australia: A geospatial analysis. Journal of Exposure Science and Environmental Epidemiology 22:248-257.

Pebesma EJ, R.S. Bivand. 2004. Multivariable geostatistics in S: The gstat package. Computers & Geosciences, 30: 683-691.

Pebesma EJ, R.S. Bivand. 2005. Classes and methods for spatial data in R. R news 5 (2), http://cran.R-project.Org/doc/rnews/.

Pinheiro J BD, DebRoy S, Sarkar D and R Core Team. 2016. Nlme: Linear and nonlinear mixed effects models. R package version 3.1-126, .

R Core Team. 2013. R: A language and environment for statistical computing, R foundation for statistical computing, Vienna, Austria. www.R-project.Org.

Ramirez-Andreotta MD, Brusseau ML, Beamer P, Maier RM. 2013. Home gardening near a mining site in an arsenic-endemic region of Arizona: Assessing arsenic exposure dose and risk via ingestion of home garden vegetables, soils, and water. Science of the Total Environment 454:373-382.

35

Rasmussen PE, Levesque C, Chénier M, Gardner HD, Jones-Otazo H, Petrovic S. 2013. Canadian house dust study: Population-based concentrations, loads and loading rates of arsenic, cadmium, chromium, copper, nickel, lead, and zinc inside urban homes. Science of The Total Environment 443:520-529.

Rawlins BG, Webster R, Lister TR. 2003. The influence of parent material on topsoil geochemistry in eastern England. Earth Surface Processes and Landforms 28:1389- 1409.

Revelle W. 2014. Psych: Procedures for psychological, psychometric, and personality research, Northwestern University, Evanston, Illinois, http://cran.R- project.Org/package=psych.

Rieuwerts JS, Searle P, Buck R. 2006. Bioaccessible arsenic in the home environment in southwest England. Science of The Total Environment 371:89-98.

Thornton I. 1996a. Sources and pathways of arsenic in the geochemical environment: Health implications. Geological Society, London, Special Publications 113:153-161.

Thornton I. 1996b. Impacts of mining on the environment; some local, regional and global issues. Applied Geochemistry 11:355-361.

WHO. 2001. Environmental health criteria 224: Arsenic and arsenic compounds. World Helath Organisation, Geneva.

WHO. 2011. Arsenic in drinking-water - background document for development of WHO guidelines for drinking-water quality, World Helath Organisation, who/sde/wsh/03.04/75/rev/1.

Wickre JB, Folt CL, Sturup S, Karagas MR. 2004. Environmental exposure and fingernail analysis of arsenic and mercury in children and adults in a nicaraguan gold mining community. Archives of Environmental Health: An International Journal 59:400-409.

36

Xu J, Thornton I. 1985. Arsenic in garden soils and vegetable crops in Cornwall, England: Implications for human health. Environmental Geochemistry and Health 7:131-133.

37

E

E E E E

E E CALLINGTON & TAVISTOCK DISTRICT E E (!(! E (! E E 80 (!84 E 82(! E WADEBRIDGE DISTRICT E 97(! (!90 E !92 !95!(! (!(!(! 94(!((!(!(! E 93!( (!(!(! 96(!(!(!(! ( !((!(!(! 101 85 DISTRICT 102( (!91E(!(! (!100 (! E 98 89 E E 88 87 (! E 81(!(! 99 E E 86 EE E E E E E E 76(! E E (!83 E E E E E E E E E E EE EE ST. AUSTELL DISTRICT E E ST. AGNES (! (! EE 71 70 DISTRICT 78(! (!! 73 ( 79(!(! (!!77 74 (!(! E E ((! E 72 E E EE (!75(! CAMBORNE, E E 24 REDRUTH & 66(! E(! (! 25(! 47 50(!(! ST. DAY DISTRICT 55 44(!! E E 5138 (! (!(( (!35 (!40(!45 64(!67(! E 39 ! 69 57!(!(!(! E 58 56(!54 49(! 43 53(! (! ! ( 46 E E 59(! (! (! 34( 41 (!(! 36(! 62 61 ! E E E 60 (! (! (! (! (30 48E E !(! E GWINEAR 37!( 63 28 68 E ((!29(! (!(!23 ST. IVES 26 ! 31 E DISTRICT (!27(32 65 DISTRICT 33 (! E 21(! 52 E ! (! 42 EE E E 19( 20 E (! ± E EE 22 (! EE E ! 2 ST. JUST E 4( E (! E E (! E (!!(! (! 11 WENDRON & E( 1 DISTRICT 7 (!18 E (! 17 FALMOUTH DISTRICT 3 E 12(!1316(! E E 0 5 10 20 30 40 E (! E E E E 10 !8 9 E E (! ! E Kilometers ( (!15 (! EE 14 E E 6 EE E E E EEE MOUNT'S BAY DISTRICT E E E E Households (sampling units) E (!5 Mining districts (Dines 1956)

Devon county

!( Mining sites 1:Botallack:Sn 27:Camborne Vean:Cu 53:Poldice:Cu 78:Polmear:Cu 2:Levant:Cu 28:Carn Brea and Tincroft:Cu 54:Poldice, East and Agar:Sn 79:St. Austell Consols:Sn 3:Owles:Sn 29:Carn Camborne:Cu 55:Poldice, West:Sn 80:Treore:As 4:Spearn Moor:Sn 30:Clifford Amalgamated:Cu 56:Pool, North:Cu 81:Ambrose Lake:Sn 5:Fenwick:Sn 31:Comford:Cu 57:Prosper:Sn 82:Lemarne:As 6:Fortune, Great (Breage):Sn 32:Condurrow:Cu 58:Roskear, South:Cu 83:Trelawny:Pb 7:Fortune, Great (Ludgvan):Sn 33:Condurrow, South:Sn 58:Roskear, North:Cu 84:Vincent:Sn 8:Great Western:Sn 34:Consolidated Mines:Cu 59:Seton:Cu 85:Arthur:Cu 9:Great Work:Sn 35:Creegbrawse and Penkeevil:Cu 60:Seton, West:Cu 86:Bedford Consols:Cu 10:Hampton:Sn 36:Crofty, South:Cu 61:St. Day United:Cu 87:Brothers:Ag 11:Leedstown Consols:Sn 37:Dolcoath:Cu 62:Tehidy:Cu 88:Callington Consols:Sn 12: Mines:Cu 38:Downs, Great North:Cu 63:Tincroft:Cu 89:Drakewalls:Sn 13:Prosper United:Cu 39:Emily Henrietta:Cu 64:Tolgullow United:Sn 90:Duchy Great Consols:Cu 14:St. Aubyn and Grylla:Sn 40:Falmouth and Sperries:As 65:Tresavean:Cu 91:Florence:Sn 15:Sydney Godolphin:Sn 41:Falmouth Consolidated:Sn 66:Treskerby, North:Cu 92:Gunnislake Clitters:Cu 16:Tindene:Sn 42:Forest:As 67:Unity:Sn 93:Hingston Down:Cu 17:Tregembo:Sn 43:Gorland:Cu 68:United Mines:Cu 94:Holmbush:Cu 18:Trevarthian Downs:Sn 44:Hallenbeagle:Cu 69:Unity Wood:Cu 95:Kelly Bray:Cu 19:Rosewarne and Herland United:Cu 45:Jane:Sn 70:Anna:Zn 96:Kit Hill United:Sn 20:Rosewarne, New West:Sn 46:Jane, West:S (sulphur) 71:Blue Hills:Sn 97:Martha:Cu 21:Rosewarne, North:Cu 47:Killifreth:Sn 72:Chiverton, Great South:Pb 98:Newton:As 22:Unity Consols:Cu 48:Nangiles:S (sulphur) 73:Chiverton, New:Zn 99:Okeltor:Cu 23:Roskrow United:S (sulphur) 49:Pedn-and-Drea:Sn 74:Kitty:Sn 100:Prince of Wales:Sn 24:Boscawen:Cu 50:Peevor:Sn 75:Tywarnhayle:Cu 101:Prince of Wales, West:Sn 25:Busy:Cu 51:Peevor, West:Sn 76:Castle-an-Dinas:W 102:Redmoor:Cu 26:Camborne Consols:Cu 52:Pendarves United:Cu 77:Polgooth, South:Sn

Figure S1 The spatial distribution of As-specific former mining sites, located from BRITPITS and Dines (1956), in relation to sampled households. The former mining districts were digitalised from the paper maps provided in Dines (1956). Mining sites listed in bold type are those reported to have produced >2,000 tonnes of As. Compiled using ESRI ArcMap 10.1. 38

CHAPTER 7: SUMMARY AND CONCLUSIONS The aim of this work was to assess human exposure to environmental As from PWS in Cornwall in south west England and investigate exposure from additional sources (i.e. residential soil and dust). This was achieved by conducting a multi-agency, cross-sectional biomonitoring study, on a scale not yet undertaken in the UK. The participation of 215 volunteers from 129 households led to the collection of three biological media, six environmental media, and data from an exposure assessment questionnaire. Throughout the project, efforts were made to use the most up-to-date biomonitoring methodologies and assess and improve their performance. Major findings, together with implications for potential impact, recommendations for further work and conclusions, are summarised in this chapter.

7.1. Key findings and further work 7.1.1. Arsenic exposure from private water supplies Paired drinking water and urine samples of 207 volunteers from 127 households were used to assess recent/ongoing As exposure from PWS in the study population (Chapter 3). Twelve percent (n=15) of households exceeded the UK PCV and WHO guidance value of 10 As µg L-1. This equated to 10 % (n=21) of volunteers. Whilst over-representing the underlying population exceedance rate of 5 % (Ander et al. 2016), the recruitment strategy successfully captured volunteers in the upper exposure range. The age distribution of the study group was also not representative of the underlying target population. This was possibly due to a greater number of older age groups being at home when telephone calls were made. It is noted that the sampling bias in this study rendered this particular data inappropriate for assessing population exposure. Population exposure to As and other elements from PWS in Cornwall and the wider UK is a focus of ongoing research (Crabbe et al. 2016).

Urinary As speciation was necessary to deduct the contribution of non- or less-toxic AB to reveal exposures to inorganic As in drinking water and isolate this source from dietary As intake. Hydration adjustment, using osmolality as the metric of urinary dilution, increased the strength in correlation between drinking water and urinary As concentrations- more so than creatinine, which yielded no increase in correlation strength. These findings confirmed that As exposure from PWS was present in the study population. Where As concentrations in drinking water increased above 1 µg L-1, drinking water was the dominant exposure route.

96

The variation in urinary AsIMM concentrations where drinking water contained <1 As µg L-1 were not explained by drinking water As concentrations. Possible alternative exposure routes that have been previously established in Cornwall, such as soil (Button et al. 2009a) and dust (Rieuwerts et al. 2006), may have contributed. Furthermore, the inclusion of DMA in the employed exposure metric – AsIMM – may have resulted in exposure misclassification. The presence of DMA in urine, although a metabolite of inorganic As, also occurs from the metabolism of arsenosugars and arsenolipids in seafood (Navas-Acien et al. 2011). Despite requesting volunteers to refrain from seafood consumption, AB was present in 73 % of the urine samples collected in the present study. It is likely that a proportion of the DMA measured in urine cannot be attributed to exposure to inorganic environmental As. It has been suggested (Hata et al. 2016) that DMA should be completely excluded from the urinary As biomarker (i.e. the sum of AsIII, AsV and MA). It may be possible, using data from the present study, to model urinary DMA concentrations as a function of the other urinary As species to attribute the proportion representative of environmental inorganic As exposures. This will be a matter of further research.

The findings that volunteers were exposed to As from PWS in the study population has implications beyond Cornwall and south west England. There are currently 6,474 single domestic PWS registered with the DWI in England, the owners of which are not obliged to monitor water quality or install treatment systems (DWI 2015). The true number of PWS is likely much greater as many go unregistered. Arsenic is not the only analyte of concern. Exceedances of the PCV were also reported for nitrate (11 %), manganese (12 %) and aluminium (7 %) in PWS surveyed for the precursor to the present study (Ander et al. 2016).

7.1.2. The performance of urinary hydration adjustments Prompted by the findings in Chapter 3 - osmolality outperforming creatinine as a urinary As adjustment factor - and the inconsistencies in the scientific literature, data from the US NHANES survey (2009-10 and 2011-12) were used to comparatively assess a range of previously employed adjustment methods (Chapter 4). In particular, the proposal (Hays et al. 2015) that urinary analyte excretion rates (ERs) using urinary flow rates (UFR) provide a superior method of adjusting for urinary hydration status, was investigated. Two independent criteria were employed: Criterion A: the correlation between adjusted analyte concentrations and UFR (weak

97

correlations desired to demonstrate that hydration variation was removed) and Criterion B: the correlation between urinary and blood analyte concentrations (strong correlations desired to demonstrate that internal dose was more accurately reflected). It was found that, not only was ER calculation not superior to other methods, but was inherently biased as it directly incorporated hydration variation into analyte (AsIMM, Pb, Cd and I) results due to its mathematical application. Coefficients (Araki’s b values) (Araki et al. 1986) were derived to improve the use of UFR data for hydration adjustment. Araki’s b values describe the slope of analyte concentrations as a function of UFR, vary from one analyte to the next (Sata et al. 1996) and are assumed to be linear by alternative adjustment methods (Vij and Howell 1998). Overall, osmolality adjustment out performed all of modified UFR, creatinine, ER and bodyweight adjusted ER adjustment factors. The application of UFR-based methods were hindered by (i) the inaccuracy of UFR calculations reliant on self- reported previous urinary void times and (ii) the inability to derive Araki’s b values on repeat urine samples where the only variable is UFR and on a demographic- specific basis.

Whilst it was not possible to develop a ‘gold-standard’ urinary adjustment method, a theoretical gold-standard approach was highlighted. For urinary hydration adjustment factors to be successful in their application, the following requirements must be met: (i) the metric employed needs to be an accurate measure of urinary dilution; (ii) the mathematical application employed needs to account for the relationship between the specific analyte concentration and changing hydration (i.e. Araki’s b value) and (iii) adjustment factors need to account for the demographic characteristics of the population on which they are being applied. The interpretation of biomonitoring data is an area that has been highlighted as deficient in comparison to the advances made in analytical techniques (Clewell et al. 2008). Progress made in this area and the findings of the present study may prove valuable to the wider field of urinary biomonitoring given its range of clinical, nutritional, forensic and environmental medical applications for host of analytes extending beyond inorganic As.

Further work in this area will involve the analysis of the Cornwall urinary As data to compare the performance of creatinine, specific gravity and osmolality in relation to an additional assessment criterion: the correlation between urinary and drinking

98

water As concentrations (strong correlations indicating a better representation of external exposure). Preliminary analysis is presented in the extended conference abstract in Appendix D.

7.1.3. Prolonged arsenic exposure from private water supplies Exposure to As over a longer time period (> 2.5 years) was assessed in the Cornwall PWS cohort (Chapter 5). This was achieved through the analysis of repeat drinking water As measurements –those from the initial PWS survey (Ander et al. 2016) compared to follow-up measurements during the biomonitoring phase – and the analysis of longer term toenail and hair biomarkers. Stability in drinking water As concentrations over time was observed, as was a moderate-strong correlation between drinking water As and concentrations measured in toenails and hair. Findings complemented those of recent exposure (Chapter 3), in that exposure was demonstrated over a longer timescale than the short window of exposure represented by urine samples. They also demonstrated that As measurements made in the present, may be useful in representing past concentrations. Importantly, some of the households that exceeded the UK PCV and WHO guidance value of 10 As µg L-1 in drinking water at the initial collection, continued to exceed at follow-up. Some (but not all) of these households had ample time to install treatment. This has important implications regarding the risk-perception of PWS-users or perhaps the advice strategy employed. The lower As concentrations present in the Cornwall study population do not induce visible symptoms of arsenicosis or impart any aesthetical or palatability characteristics to drinking water – As is odourless and tasteless (Jomova et al. 2011). This may be partly responsible for the lack of action on the part of householders. The incurred costs of installing treatment is also likely to be a factor.

7.1.4. Toenails and hair as viable biomarkers of arsenic exposure A useful methodological aspect of the study was the successful application of the toenail washing procedure on a large sample set (Chapter 5). The analysis of washing solutions initially employed by Button et al. (2009a) not only demonstrated the potential contribution of external sample contamination, but also provided evidence of its effective removal. The weaker correlations observed between toenail/hair As concentrations and drinking water As concentrations may indicate the limited utility of these biomarkers over urine, but may also reflect their ability to

99

detect more sporadic, less frequent exposure pathways than the regular consumption of As in drinking water. Previously reported correlations between urinary AsIMM and soil As concentrations are often lower (e.g. r2=0.13 (Wilhelm et al. 2004b); r (Pearson)=0.14 (Tsuji et al. 2005); r (Pearson)=0.25 (Hwang et al. 1997)) than those between longer-term biomarkers such as toenails versus soil As (e.g. r2=0.23 (fingernails) (Wilhelm et al. 2004b); r (Spearman)=0.50 (Hinwood et al. 2003); r (Pearson)=0.42 (Pearce et al. 2010)). Comparisons between the biomonitoring samples collected in the present study and concentrations measured in soil and dust have not yet been made yet. This will be the focus of further investigation. However, never consuming home-grown vegetables was associated with lower hair As concentrations, possibly indicative of a lesser degree of soil related exposure.

7.1.5. Potential arsenic exposure from residential soil and dust Potential exposure to As via soil and dust was investigated by comparing the As concentrations in residential soil and composite indoor dust samples to previously published findings and existing assessment criteria for As in English soil (Chapter 6). The As concentrations measured in residential soil and dust samples were expectedly elevated and compared with previously published findings from the local area (Culbard and Johnson 1984; Norton et al. 2013; Rieuwerts et al. 2006; Xu and Thornton 1985). Concentrations were higher in mineralised areas and in the vicinity of former As mining sites. The majority of residential soil samples (69 %) exceeded the English Category 4 Screening Level C4SL of 37 mg kg-1 for residential properties with home-grown produce. The conservatism of the C4SL was reduced using the Contaminated Land Exposure Assessment (CLEA) model to derive bioaccessibility-adjusted assessment criteria. The generic assumption of 100 % relative bioavailability was changed to represent the 25th (13 % bioaccessibility) and 75th (23 % bioaccessibility) percentile soil bioaccessible fractions (BAFs) measured in the Cornwall samples. These were 190 and 129 As mg kg-1 and reduced exceedances to 10 and 17 % of households, respectively. Whilst a number of households still exceeded these adjusted criteria, it is noted that all other generic assumptions in the CLEA model were upheld. Furthermore, as noted, the analysis of biomonitoring data in conjuction with additional environmental samples is a matter of further work and as yet there is no direct evidence of As intake from soil or dust in the study population. Nevertheless, the results are still of concern, particularly

100

regarding the exposure of infants and small children in the study area - a group highlighted as particularly vulnerable to As exposure (Mitchell and Barr 1995; Rieuwerts et al. 2006).

7.1.6. Ongoing analysis As noted, a number of questions remain outstanding from this research. The large quantity of samples collected for the project resulted in a large amount of data. Not all of this data could be analysed in the time allocated for this PhD project and work is ongoing. This includes the data from the rice and indoor dust wipe samples described in Chapter 2. Data have not yet been properly interrogated and are therefore not presented in this thesis. The relative importance of the alternative exposure routes in this study group (drinking water, soil and dust) will be quantified using multivariate analyses on subsets of volunteers as follows: high drinking water As/low soil As; low drinking water As/high soil As and high drinking water and soil As.

101

7.2. Conclusions Notwithstanding the further work yet to be undertaken, the study has allowed some important conclusions to me made to-date:

(i) PWS drinking water is a significant source of As exposure in south west England in a proportion (~5 %) of households. (ii) Urinary hydration adjustment is a necessary step to yielding the most meaningful results, but many inconsistencies remain in its application. Hydration adjustments are analyte- and demographic group-specific. The analyte-hydration relationship needs accounting for in adjustment factors. (iii) Exposure to As from PWS drinking water is occurring over a prolonged timescales and some householders are not seeking appropriate treatment. (iv) Toenails and hair are viable biomarkers of prolonged As exposure providing that samples are properly prepared and cleaned and the removal of exogenous contamination can be verified. These matrices are complimentary biomarkers to urine and may reflect longer term average exposure resulting from less frequent intake events. (v) Other exposure routes (i.e. soil and dust) may be less predominant in the adult study population, but further analysis is required.

102

REFERENCES

Abrahams P, Thornton I. 1987. Distribution and extent of land contaminated by arsenic and associated metals in mining regions of southwest England. Institution of Mining and Metallurgy Transactions Section B Applied Earth Science 96.

ACGIH. 2001. American Conference of Government Industrial Hygienists (ACGIH). Documentation of Biological Exposure Indices. 7th Edition. Cincinnati, Ohio: ACGIH Worldwide; 2001.

Akerstrom M, Barregard L, Lundh T, Sallsten G. 2013. The relationship between cadmium in kidney and cadmium in urine and blood in an environmentally exposed population. Toxicology and Applied Pharmacology 268:286-293.

Allen-Price ED. 1960. Uneven distribution of cancer in west Devon with particular reference to the divers water-supplies. The Lancet 275:1235-1238.

Ander E, Watts M, Smedley P, Hamilton E, Close R, Crabbe H, et al. 2016. Variability in the chemistry of private drinking water supplies and the impact of domestic treatment systems on water quality. Environmental Geochemistry and Health:1-20.

Ander EL, Johnson CC, Cave MR, Palumbo-Roe B, Nathanail CP, Lark RM. 2013. Methodology for the determination of normal background concentrations of contaminants in English soil. Science of the Total Environment 454:604-618.

Angerer J, Aylward LL, Hays SM, Heinzow B, Wilhelm M. 2011. Human biomonitoring assessment values: Approaches and data requirements. International Journal of Hygiene and Environmental Health 214:348-360.

Appleton J, Cave M, Wragg J. 2012. Anthropogenic and geogenic impacts on arsenic bioaccessibility in UK topsoils. Science of the Total Environment 435:21-29.

Araki S, Murata K, Aono H, Yanagihara S, Niinuma Y, Yamamoto R, et al. 1986. Comparison of the effects of urinary flow on adjusted and non‐adjusted excretion of heavy metals and organic substances in ‘healthy’men. Journal of Applied Toxicology 6:245-251.

103

Araki S, Sata F, Murata K. 1990. Adjustment for urinary flow rate: An improved approach to biological monitoring. International Archives of Occupational and Environmental Health 62:471-477.

Arup O. 1990. Mining instability in Britain. Department of environment contract. Hardcopy maps digitised and adapted by the British Geological Survey.

Asante KA, Agusa T, Subramanian A, Ansa-Asare OD, Biney CA, Tanabe S. 2007. Contamination status of arsenic and other trace elements in drinking water and residents from Tarkwa, a historic mining township in Ghana. Chemosphere 66:1513- 1522.

ASTM. 2002. ASTM (American Society for Testing and Materials), Report No. E 1728–02, standard practice for collection of settled dust samples using wipe sampling methods for subsequent lead determination, West Conshohocken, Pennsylvania, United States.

Aston S, Thornton I, Webb J, Milford B, Purves J. 1975. Arsenic in stream sediments and waters of south west England. Science of The Total Environment 4:347-358.

ATSDR. 2007. Agency for Toxic Substances and Disease Registry (ASTDR). 2007. Toxicological profile for arsenic. Atlanta, GA: U.S. Department of Health and Human Services, Public Health Service.

ATSDR. 2014. Support document to the 2013 priority list of hazardous substances that will be the subject of toxicological profiles, April 2014, Agency for Toxic Substances and Disease Registry, Division of Toxicology and Environmental Medicine, Atlanata, GA.

Aylward LL, Hays SM, Smolders R, Koch HM, Cocker J, Jones K, et al. 2014. Sources of variability in biomarker concentrations. Journal of Toxicology and Environmental Health, Part B 17:45-61.

Ayotte JD, Montgomery DL, Flanagan SM, Robinson KW. 2003. Arsenic in groundwater in eastern New England: Occurrence, controls, and human health implications. Environmental Science & Technology 37:2075-2083.

104

Banerjee M, Banerjee N, Bhattacharjee P, Mondal D, Lythgoe PR, Martínez M, et al. 2013. High arsenic in rice is associated with elevated genotoxic effects in humans. Scientific Reports 3.

Barber T, Wallis G. 1986. Correction of urinary mercury concentration by specific gravity, osmolality, and creatinine. Journal of Occupational and Environmental Medicine 28:354-359.

Barnes J, Ledbrooke A, Parsons B, Salter L, Unit AQ. 2006. Monitoring of ambient air particulate arsenic concentrations at nine sites in Cornwall.

Barr DB, Wilder LC, Caudill SP, Gonzalez AJ, Needham LL, Pirkle JL. 2005. Urinary creatinine concentrations in the US population: Implications for urinary biologic monitoring measurements. Environmental Health Perspectives 113:192.

Barrett JR. 2015. Urinary biomarkers as exposure surrogates: Controlling for possible bias. Environmental Health Perspectives 123:A97.

Basu A, Mitra S, Chung J, Mazumder DG, Ghosh N, Kalman D, et al. 2011. Creatinine, diet, micronutrients, and arsenic methylation in West Bengal, India. Environmental Health Perspectives 119:1308.

Baxmann AC, Ahmed MS, Marques NC, Menon VB, Pereira AB, Kirsztajn GM, et al. 2008. Influence of muscle mass and physical activity on serum and urinary creatinine and serum cystatin C. Clinical Journal of the American Society of Nephrology 3:348-354.

Beane Freeman LE, Dennis LK, Lynch CF, Thorne PS, Just CL. 2004. Toenail arsenic content and cutaneous melanoma in Iowa. American Journal of Epidemiology 160:679-687.

BGS. 2016. G-BASE regional geochemistry map for arsenic (As) in soils in south- west England. 2016. British Geological Survey, Nottinghamshire, UK. Available: http://www.bgs.ac.uk/gbase/gbasesw.html.

Brima EI, Haris PI, Jenkins RO, Polya DA, Gault AG, Harrington CF. 2006. Understanding arsenic metabolism through a comparative study of arsenic levels in

105

the urine, hair and fingernails of healthy volunteers from three unexposed ethnic groups in the United Kingdom. Toxicology and Applied Pharmacology 216:122-130.

Buchet JP, Lauwerys R, Roels H. 1981. Comparison of the urinary excretion of arsenic metabolites after a single oral dose of sodium arsenite, monomethylarsonate, or dimethylarsinate in man. International Archives of Occupational and Environmental Health 48:71-79.

Button M, Jenkin GR, Bowman KJ, Harrington CF, Brewer TS, Jones GD, et al. 2010. DNA damage in earthworms from highly contaminated soils: Assessing resistance to arsenic toxicity by use of the COMET assay. Mutation Research/Genetic Toxicology and Environmental Mutagenesis 696:95-100.

Button M, Jenkin GR, Harrington CF, Watts MJ. 2009. Arsenic biotransformation in earthworms from contaminated soils. Journal of Environmental Monitoring 11:1484- 1491.

Button M, Jenkin GR, Harrington CF, Watts MJ. 2009a. Human toenails as a biomarker of exposure to elevated environmental arsenic. Journal of Environmental Monitoring 11:610-617.

Button M, Watts MJ, Cave MR, Harrington CF, Jenkin GT. 2009b. Earthworms and in vitro physiologically-based extraction tests: Complementary tools for a holistic approach towards understanding risk at arsenic-contaminated sites. Environmental Geochemistry and Health 31:273-282.

Button M. 2009. Arsenic contaminated soils: Human exposure and environmental toxicology. Thesis submitted for the degree of doctor of philosophy at the University of Leicester, April 2009. Awarded 23 October 2009.

Calderon RL, Hudgens E, Le XC, Schreinemachers D, Thomas DJ. 1999. Excretion of arsenic in urine as a function of exposure to arsenic in drinking water. Environmental Health Perspectives 107:663.

Caldwell KL, Jones RL, Verdon CP, Jarrett JM, Caudill SP, Osterloh JD. 2008. Levels of urinary total and speciated arsenic in the US population: National Health

106

and Nutrition Examination Survey 2003–2004. Journal of Exposure Science and Environmental Epidemiology 19:59-68.

Cameron D. 2013. User guide for the BRITPITS GIS dataset. British Geological Survey, Nottingham. Available: http://nora.nerc.ac.uk/503964/.

Camm GS, Glass HJ, Bryce DW, Butcher AR. 2004. Characterisation of a mining- related arsenic-contaminated site, Cornwall, UK. Journal of Geochemical Exploration 82:1-15.

Candeias C, Melo R, Ávila PF, da Silva EF, Salgueiro AR, Teixeira JP. 2014. Heavy metal pollution in mine–soil–plant system in San Francisco de Assis–Panasqueira mine (Portugal). Applied Geochemistry 44:12-26.

Cascio C, Raab A, Jenkins RO, Feldmann J, Meharg AA, Haris PI. 2011. The impact of a rice based diet on urinary arsenic. Journal of Environmental Monitoring 13:257- 265.

Cave M, Wragg J, Palumbo B, Klinck B, Mcleod K. Physico-chemical controls on the bioaccessibility of arsenic in UK soils. In: Proceedings of the EGS General Assembly Conference Abstracts, 2002, Vol. 27, 3908.

CDC (Centers for Disease Control and Prevention). 2013. Case definition: Arsenic (inorganic) webpage. Available: http://emergency.cdc.gov/agent/arsenic/casedef.asp. [accessed: 26/04/2016].

CDC (Centers for Disease Control and Prevention). 2013. National Biomonitoring Program Lead Biomonitoring Summary Webpage. Available: http://www.cdc.gov/biomonitoring/Lead_BiomonitoringSummary.html [accessed 05 September 2015].

CDC (Centers for Disease Control and Prevention). 2013. National Biomonitoring Program Cadmium Biomonitoring Summary Webpage. Available: http://www.cdc.gov/biomonitoring/Cadmium_BiomonitoringSummary.html [accessed 05 September 2015].

107

CDC (Centers for Disease Control and Prevention). 2015. National Health and Nutrition Examination Survey Homepage. Available: http://www.cdc.gov/nchs/nhanes.htm [accessed 01 October 2015].

Chakraborti D, Das B, Rahman MM, Chowdhury UK, Biswas B, Goswami A, et al. 2009. Status of groundwater arsenic contamination in the state of West Bengal, India: A 20‐year study report. Molecular Nutrition & Food Research 53:542-551.

Chakraborti D, Rahman MM, Das B, Murrill M, Dey S, Chandra Mukherjee S, et al. 2010. Status of groundwater arsenic contamination in Bangladesh: A 14-year study report. Water Research 44:5789-5802.

Chen C-J, Chuang Y-C, Lin T-M, Wu H-Y. 1985. Malignant neoplasms among residents of a blackfoot disease-endemic area in Taiwan: High-arsenic artesian well water and cancers. Cancer Research 45:5895-5899.

Chen Y-C, Amarasiriwardena CJ, Hsueh Y-M, Christiani DC. 2002. Stability of arsenic species and insoluble arsenic in human urine. Cancer Epidemiology Biomarkers & Prevention 11:1427-1433.

Chilvers D, Peterson P. 1987. Global cycling of arsenic. Lead, mercury, cadmium and arsenic in the environment. pp. 279-301.

Chojnacka K, Górecka H, Górecki H. 2006. The effect of age, sex, smoking habit and hair color on the composition of hair. Environmental Toxicology and Pharmacology 22:52-57.

Chojnacka K, Zielińska A, Górecka H, Dobrzański Z, Górecki H. 2010. Reference values for hair minerals of Polish students. Environmental Toxicology and Pharmacology 29:314-319.

Chowdhury UK, Rahman MM, Sengupta MK, Lodh D, Chanda CR, Roy S, et al. 2003. Pattern of excretion of arsenic compounds [arsenite, arsenate, MMA (V), DMA (V)] in urine of children compared to adults from an arsenic exposed area in Bangladesh. Journal of Environmental Science and Health, Part A 38:87-113.

108

CL:AIRE. 2014. SP1010 – Development of Category 4 Screening Levels for assessment of land affected by contamination final project report (revision 2). Contaminated Land: Applications in Real Environments (CL:AIRE). Available: http://randd.defra.gov.uk/Default.aspx?Module=More&Location=None&ProjectID= 18341.

Clewell HJ, Tan YM, Campbell JL, Andersen ME. 2008. Quantitative interpretation of human biomonitoring data. Toxicology and Applied Pharmacology 231:122-133.

Clough P. 1980. Incidence of malignant melanoma of the skin in England and Wales. British Medical Journal 280:112.

Cohen SM, Arnold LL, Eldan M, Lewis AS, Beck BD. 2006. Methylated arsenicals: The implications of metabolism and carcinogenicity studies in rodents to human risk assessment. CRC Critical Reviews in Toxicology 36:99-133.

Colbourn P, Alloway BJ, Thornton I. 1975. Arsenic and heavy metals in soils associated with regional geochemical anomalies in south-west England. Science of The Total Environment 4:359-363.

Cone EJ, Caplan YH, Moser F, Robert T, Shelby MK, Black DL. 2009. Normalization of urinary drug concentrations with specific gravity and creatinine. Journal of Analytical Toxicology 33:1-7.

Cottingham KL, Karimi R, Gruber JF, Zens MS, Sayarath V, Folt CL, et al. 2013. Diet and toenail arsenic concentrations in a New Hampshire population with arsenic- containing water. Nutrition Journal 12:1.

Council. of the European Union 1998. The quality of water intended for human consumption. Council Directive 98/83/EC.

Crabbe H, Close, R., Rimmell, A., Leonardi, G., Watts, M.J., Ander, E.L., Hamilton, E.M., Middleton, D.R.S., Smedley, P.L., Gregory, M., Robjohns, S., Sepai, O., Studden, M., Polya, D.A., and Fletcher, T. 2016. Estimating the population exposed to arsenic from groundwater-sourced private drinking water supplies in Cornwall, UK. In: Bhattacharya P, Polya DA and Jovanovic D (Eds.) Best practice guide for

109

the control of arsenic in drinking water, IWA Publishing, Chapter A4, ISBN13: 9781843393856.

Culbard E, Johnson L. 1984. Elevated arsenic concentrations in house dusts located in a mineralised area of south-west England: Implication for human health. Trace Substances in Environmental Health XVIII. University of Missouri, Columbia: 311- 319.

Cullen WR. 2014. Chemical mechanism of arsenic biomethylation. Chemical Research in Toxicology 27:457-461.

Dartey E, Sarpong K, Darko G, Acheampong-Marfo M. 2013. Urinary arsenic and mercury levels in artisanal miners in some communities in the Obuasi municipality of Ghana. Journal of Environmental Chemistry and Ecotoxicology 5:113-118.

Denys S, Caboche J, Tack K, Rychen G, Wragg J, Cave M, et al. 2012. In vivo validation of the unified barge method to assess the bioaccessibility of arsenic, antimony, cadmium, and lead in soils. Environmental Science & Technology 46:6252-6260.

Dines HG. 1956. The metalliferous mining region of south-west England: HM Stationery Office.

Dobbin KK, Simon RM. 2011. Optimally splitting cases for training and testing high dimensional classifiers. BMC Medical Genomics 4:1.

Dossin O, Germain C, Braun JP. 2003. Comparison of the techniques of evaluation of urine dilution/concentration in the dog. Journal of Veterinary Medicine Series A 50:322-325.

Dragulescu AA. 2014. Xlsx: Read, write, format excel 2007 and excel 97/2000/xp/2003 files. http://CRANR-projectorg/package=xlsx.

Drahota P, Filippi M. 2009. Secondary arsenic minerals in the environment: A review. Environment International 35:1243-1255.

110

DWI. 2015. Drinking water 2014, Private Water Supplies in England, a report by the chief inspector of drinking water, July 2015. Available: http://www.dwi.gov.uk/about/annual-report/2014/index.html.

Dybowska A, Farago M, Valsami-Jones E, Thornton I. 2006. Operationally defined associations of arsenic and copper from soil and mine waste in south-west England. Chemical Speciation and Bioavailability 17:147-160.

Elghali L. 1994. Evaluation of exposure of young children to arsenic contamination in south west England: MSc thesis, University of London.

Environment Agency. 2009. Updated technical background to the CLEA model. Available: https://www.gov.uk/government/publications/updated-technical- background-to-the-clea-model.

Esteban M, Schindler BK, Jiménez-Guerrero JA, Koch HM, Angerer J, Rivas TC, et al. 2014. Mercury analysis in hair: Comparability and quality assessment within the transnational COPHES/DEMOCOPHES project. Environmental Research 141:24- 30.

European Commission. 2015. Commission regulation (EU) No 2015/1006 of 25 of June 2015 amending regulation (EC) No 1881/2006 setting maximum levels for certain contaminants in foodstuffs as regards maximum levels of inorganic arsenic in foodstuffs. Official Journal of the European Union 161:14-16.

Ewers U, Krause C, Schulz C, Wilhelm M. 1999. Reference values and human biological monitoring values for environmental toxins. International Archives of Occupational and Environmental Health 72:255-260.

Farago M, Kavanagh P, Ármannsson H. 1999. High arsenic-containing soils in SW England and human exposure assessment. In: Proceedings of the Geochemistry of the earth's surface Proceedings of the 5th International Symposium, Reykjavik, Iceland, 15-20 August 1999. AA Balkema, 181-184.

Farago M, Thornton I, Kavanagh P, Elliott P, Leonardi G. 1997. Health aspects of human exposure to high arsenic concentrations in soil in south-west England. In: Arsenic. Springer, 210-226.

111

Favaro P, Bode P, De Nadai Fernandes E. 2005. Trace elements in nail polish as a source of contamination of nail clippings when used in epidemiological studies. Journal of Radioanalytical and Nuclear Chemistry 264:61-65.

Feldmann J, Lai VW, Cullen WR, Ma M, Lu X, Le XC. 1999. Sample preparation and storage can change arsenic speciation in human urine. Clinical Chemistry 45:1988-1997.

Fillol C, Dor F, Labat L, Boltz P, Le Bouard J, Mantey K, et al. 2010. Urinary arsenic concentrations and speciation in residents living in an area with naturally contaminated soils. Science of The Total Environment 408:1190.

Fleckman P. 1997. Basic science of the nail unit. In: Nails: Therapy, Diagnosis, Surgery. Scher RK, Daniel CR (Eds) WB Saunders Co, Philadelphia.

Flynn HC, Mc Mahon V, Diaz GC, Demergasso CS, Corbisier P, Meharg AA, et al. 2002. Assessment of bioavailable arsenic and copper in soils and sediments from the Antofagasta region of northern Chile. Science of the Total Environment 286:51-59.

Francesconi KA, Kuehnelt D. 2004. Determination of arsenic species: A critical review of methods and applications, 2000–2003. Analyst 129:373-395.

Frost F, Franke D, Pierson K, Woodruff L, Raasina B, Davis R, et al. 1993. A seasonal study of arsenic in groundwater, Snohomish County, Washington, USA. Environmental Geochemistry and Health 15:209-214.

García-Esquinas E, Pollán M, Umans JG, Francesconi KA, Goessler W, Guallar E, et al. 2013. Arsenic exposure and cancer mortality in a US-based prospective cohort: The strong heart study. Cancer Epidemiology Biomarkers & Prevention 22:1944- 1953.

Gault AG, Rowland HAL, Charnock JM, Wogelius RA, Gomez-Morilla I, Vong S, et al. 2008. Arsenic in hair and nails of individuals exposed to arsenic-rich groundwaters in Kandal Province, Cambodia. Science of The Total Environment 393:168-176.

112

Gilbert-Diamond D, Cottingham KL, Gruber JF, Punshon T, Sayarath V, Gandolfi AJ, et al. 2011. Rice consumption contributes to arsenic exposure in US women. Proceedings of the National Academy of Sciences 108:20656-20660.

Gilbert-Diamond D, Li Z, Perry AE, Spencer SK, Gandolfi AJ, Karagas MR. 2013. A population-based case-control study of urinary arsenic species and squamous cell carcinoma in New Hampshire, USA. Environmental Health Perspectives 121(10):1154.

Gong Z, Lu X, Ma M, Watt C, Le XC. 2002. Arsenic speciation analysis. Talanta 58:77-96.

Haddow JE, McClain MR, Palomaki GE, Hollowell JG. 2007. Urine iodine measurements, creatinine adjustment, and thyroid deficiency in an adult United States population. The Journal of Clinical Endocrinology & Metabolism 92:1019- 1022.

Hamilton EI. 2000. Environmental variables in a holistic evaluation of land contaminated by historic mine wastes: A study of multi-element mine wastes in west Devon, England using arsenic as an element of potential concern to human health. Science of The Total Environment 249:171-221.

Hamilton EM, Barlow TS, Gowing CJ, Watts MJ. 2015. Bioaccessibility performance data for fifty-seven elements in guidance material BGS 102. Microchemical Journal 123:131-138.

Hanh HT, Kim J-Y, Bang S, Kim K-W. 2010. Sources and fate of as in the environment. Geosystem Engineering 13:35-42.

Harkey M. 1993. Anatomy and physiology of hair. Forensic Science International 63:9-18.

Harkins DK, Susten AS. 2003. Hair analysis: Exploring the state of the science. Environmental Health Perspectives 111:576.

Hata A, Kurosawa H, Endo Y, Yamanaka K, Fujitani N, Endo G. 2016. A biological indicator of inorganic arsenic exposure using the sum of urinary inorganic arsenic

113

and monomethylarsonic acid concentrations. Journal of Occupational Health (Advanced publication). Available: http://doi.org/10.1539/joh.15-0241-OA.

Hays S, Becker R, Leung H, Aylward L, Pyatt D. 2007. Biomonitoring Equivalents: A screening approach for interpreting biomonitoring results from a public health risk perspective. Regulatory Toxicology and Pharmacology 47:96-109.

Hays SM, Aylward LL, Blount BC. 2015. Variation in urinary flow rates according to demographic characteristics and body mass index in NHANES: Potential confounding of associations between health outcomes and urinary biomarker concentrations. Environmental Health Perspectives 123:293-300.

Hays SM, Aylward LL, Gagné M, Nong A, Krishnan K. 2010. Biomonitoring equivalents for inorganic arsenic. Regulatory Toxicology and Pharmacology 58:1-9.

Hays SM, Aylward LL. 2009. Using biomonitoring equivalents to interpret human biomonitoring data in a public health risk context. Journal of Applied Toxicology 29:275-288.

Heck JE, Andrew AS, Onega T, Rigas JR, Jackson BP, Karagas MR, et al. 2009. Lung cancer in a US population with low to moderate arsenic exposure. Environmental Health Perspectives 117:1718.

Henderson R, Hobbie J, Landrigan P, Mattisoti D, Perera F, Pfttaer E, et al. 1987. Biological markers in environmental health research. Environmental Health Perspectives 7:3-9.

Hindmarsh JT. 2002. Caveats in hair analysis in chronic arsenic poisoning. Clinical Biochemistry 35:1-11.

Hinwood AL, Sim MR, Jolley D, de Klerk N, Bastone EB, Gerostamoulos J, et al. 2003. Hair and toenail arsenic concentrations of residents living in areas with high environmental arsenic concentrations. Environmental Health Perspectives 111:187.

Hinwood AL, Sim MR, Jolley D, de Klerk N, Bastone EB, Gerostamoulos J, et al. 2004. Exposure to inorganic arsenic in soil increases urinary inorganic arsenic

114

concentrations of residents living in old mining areas. Environmental Geochemistry and Health 26:27-36.

Hopenhayn-Rich C, Biggs ML, Fuchs A, Bergoglio R, Tello EE, Nicolli H, et al. 1996. Bladder cancer mortality associated with arsenic in drinking water in Argentina. Epidemiology 7:117-124.

Hughes MF, Beck BD, Chen Y, Lewis AS, Thomas DJ. 2011. Arsenic exposure and toxicology: A historical perspective. Toxicological Sciences 123:305-332.

Hughes MF. 2002. Arsenic toxicity and potential mechanisms of action. Toxicology Letters 133:1-16.

Hwang Y-H, Bornschein RL, Grote J, Menrath W, Roda S. 1997. Urinary arsenic excretion as a biomarker of arsenic exposure in children. Archives of Environmental Health: An International Journal 52:139-147.

IARC. 2012. Arsenic, metals, fibers and dusts. IARC Monogr Eval Carcinog Risks Hum 100C:41-85. International Agency for Research on Cancer, Lyon, France.

Imran S, Eva G, Christopher S, Flynn E, Henner D. 2010. Is specific gravity a good estimate of urine osmolality? Journal of Clinical Laboratory Analysis 24:426-430.

Jaffe M. 1886. Concerning the precipitate produced in normal urine by picric acid and a new reaction of creatinine. Z Physiol Chem 10:391-400.

Johnson C, Breward N. 2004. G-BASE: Geochemical Baseline Survey of the Environment. British Geological Survey, Nottingham, UK. (CR/04/016N) (Unpublished). Available: http://nora.nerc.ac.uk/509527/

Johnson LR, Farmer JG. 1989. Urinary arsenic concentrations and speciation in Cornwall residents. Environmental Geochemistry and Health 11:39-44.

Jomova K, Jenisova Z, Feszterova M, Baros S, Liska J, Hudecova D, et al. 2011. Arsenic: Toxicity, oxidative stress and human disease. Journal of Applied Toxicology 31:95-107.

115

Jovanovic D, Jakovljević B, Rašić-Milutinović Z, Paunović K, Peković G, Knezević T. 2011. Arsenic occurrence in drinking water supply systems in ten municipalities in Vojvodina region, Serbia. Environmental Research 111:315-318.

Kaise T, Watanabe S, Itoh K. 1985. The acute toxicity of arsenobetaine. Chemosphere 14:1327-1332.

Kapaj S, Peterson H, Liber K, Bhattacharya P. 2006. Human health effects from chronic arsenic poisoning–a review. Journal of Environmental Science and Health Part A 41:2399-2428.

Karagas MR, Le CX, Morris S, Blum J, Lu X, Spate V, et al. 2001. Markers of low level arsenic exposure for evaluating human cancer risks in a US population. International Journal of Occupational Medicine and Environmental Health 14:171- 175.

Karagas MR, Morris JS, Weiss JE, Spate V, Baskett C, Greenberg ER. 1996. Toenail samples as an indicator of drinking water arsenic exposure. Cancer Epidemiology Biomarkers & Prevention 5:849-852.

Karagas MR, Tosteson TD, Blum J, Klaue B, Weiss JE, Stannard V, et al. 2000. Measurement of low levels of arsenic exposure: A comparison of water and toenail concentrations. American Journal of Epidemiology 152:84-90.

Karagas MR, Tosteson TD, Morris JS, Demidenko E, Mott LA, Heaney J, et al. 2004. Incidence of transitional cell carcinoma of the bladder and arsenic exposure in New Hampshire. Cancer Causes & Control 15:465-472.

Kavanagh P, Farago M, Thornton I, Goessler W, Kuehnelt D, Schlagenhaufen C, et al. 1998. Urinary arsenic species in Devon and Cornwall residents, UK. A pilot study. Analyst 123:27-29.

Kavanagh P, Farago ME, Thornton I, Elliott P, Goessler W, Irgolic KJ. 1997. Urinary arsenic concentration in a high arsenic area of south west England. Occupational and Environmental Medicine 54:840.

116

Keegan T, Hong B. 2002. Assessment of environmental arsenic levels in Prievidza district. Journal of Exposure Analysis & Environmental Epidemiology 12.

Kempson IM, Skinner WM. 2012. A comparison of washing methods for hair mineral analysis: Internal versus external effects. Biological Trace Element Research 150:10-14.

Kile ML, Houseman EA, Rodrigues E, Smith TJ, Quamruzzaman Q, Rahman M, et al. 2005. Toenail arsenic concentrations, GSTT1 gene polymorphisms, and arsenic exposure from drinking water. Cancer Epidemiology Biomarkers & Prevention 14:2419-2426.

Kim Y, Lee B-K. 2011. Association between urinary arsenic and diabetes mellitus in the Korean general population according to KNHANES 2008. Science of the Total Environment 409:4054-4062.

Kligerman AD, Doerr CL, Tennant AH, Harrington‐Brock K, Allen JW, Winkfield E, et al. 2003. Methylated trivalent arsenicals as candidate ultimate genotoxic forms of arsenic: Induction of chromosomal mutations but not gene mutations. Environmental and Molecular Mutagenesis 42:192-205.

Klinck B, Palumbo-Roe B, Cave M, Wragg J. 2005. Arsenic dispersal and bioaccessibility in mine contaminated soils: A case study from an abandoned arsenic mine in Devon, UK. (RR/04/003). British Geological Survey, Nottingham, UK. Available: http://nora.nerc.ac.uk/3681/.

Klumpp DW, Peterson PJ. 1979. Arsenic and other trace elements in the waters and organisms of an estuary in SW England. Environmental Pollution (1970) 19:11-20.

Kuhn M. 2015. Caret: Classification and regression training. http://cran.R- project.Org/package=caret.

LaKind JS, Aylward LL, Brunk C, DiZio S, Dourson M, Goldstein DA, et al. 2008. Guidelines for the communication of Biomonitoring Equivalents: Report from the Biomonitoring Equivalents expert workshop. Regulatory Toxicology and Pharmacology 51:S16-S26.

117

Lawley R. 2009. The soil-parent material database: a user guide. British Geological Survey, Nottingham, UK. (OR/08/034) (Unpublished). Available: http://nora.nerc.ac.uk/8048/.

Leech S, Penney M. 1987. Correlation of specific gravity and osmolality of urine in neonates and adults. Archives of Disease in Childhood 62:671-673.

Leese E, Morton J, Tan E, Gardiner PH, Carolan VA. 2014. µLC–ICP-MS determinations of unexposed UK urinary arsenic speciation reference values. Journal of Analytical Toxicology 38:24-30.

Leonardi G, Elliott P, Thornton I. Arsenic contamination and bladder cancer: An ecological study. In: Proceedings of the Ann Conf Internat Soc Environmental Epidemiology and International Society for Exposure Analysis Abstract PO71 Epidemiology, 1995, Vol. 6.

Leonardi G, Vahter M, Clemens F, Goessler W, Gurzau E, Hemminki K, et al. 2012. Inorganic arsenic and basal cell carcinoma in areas of Hungary, Romania, and Slovakia: A case–control study. Environmental Health Perspectives 120:721.

Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF, Feldman HI, et al. 2009. A new equation to estimate glomerular filtration rate. Annals Internal Medicine 150:604-612.

Levine L, Fahy JP. 1945. Evaluation of urinary lead concentrations. I. The significance of the specific gravity. Journal of Industrial Hygiene Toxicology 27:217-223.

Lindberg A-L, Goessler W, Gurzau E, Koppova K, Rudnai P, Kumar R, et al. 2006. Arsenic exposure in Hungary, Romania and Slovakia. Journal of Environmental Monitoring 8:203-208.

Lioy PJ, Freeman NC, Millette JR. 2002. Dust: A metric for use in residential and building exposure assessment and source characterization. Environmental Health Perspectives 110:969.

118

Mandal BK, Ogra Y, Suzuki KT. 2003. Speciation of arsenic in human nail and hair from arsenic-affected area by HPLC-Inductively Coupled Argon Plasma Mass Spectrometry. Toxicology and Applied Pharmacology 189:73-83.

Mandal BK, Suzuki KT. 2002. Arsenic round the world: A review. Talanta 58:201- 235.

Marchiset-Ferlay N, Savanovitch C, Sauvant-Rochat M-P. 2012. What is the best biomarker to assess arsenic exposure via drinking water? Environment International 39:150-171.

Marshall G, Ferreccio C, Yuan Y, Bates MN, Steinmaus C, Selvin S, et al. 2007. Fifty-year study of lung and bladder cancer mortality in Chile related to arsenic in drinking water. Journal of the National Cancer Institute 99:920-928.

Mayersohn M, Conrad KA, Achari R. 1983. The influence of a cooked meat meal on creatinine plasma concentration and creatinine clearance. British Journal of Clinical Pharmacology 15:227-230.

McDonald LT, Rasmussen PE, Chénier M, Levesque C. 2011. Extending wipe sampling methodologies to elements other than lead. Journal of Environmental Monitoring 13:377-383.

Meek M, Boobis A, Crofton K, Heinemeyer G, Raaij M, Vickers C. 2011. Risk assessment of combined exposure to multiple chemicals: A WHO/IPCS framework. Regulatory Toxicology and Pharmacology: RTP.

Meharg A, Williams P, Deacon C, Norton G, Hossain M, Louhing D, et al. 2014. Urinary excretion of arsenic following rice consumption. Environmental Pollution 194:181-187.

Metcalf SW, Orloff KG. 2004. Biomarkers of exposure in community settings. Journal of Toxicology and Environmental Health, Part A 67:715-726.

Meyer I, Heinrich J, Lippold U. 1999. Factors affecting lead, cadmium, and arsenic levels in house dust in a smelter town in eastern Germany. Environmental Research 81:32-44.

119

Michaud DS, Wright ME, Cantor KP, Taylor PR, Virtamo J, Albanes D. 2004. Arsenic concentrations in prediagnostic toenails and the risk of bladder cancer in a cohort study of male smokers. American Journal of Epidemiology 160:853-859.

Middleton D, Watts M, Hamilton E, Ander E, Close R, Exley K, et al. 2016a. Urinary arsenic profiles reveal substantial exposures to inorganic arsenic from private drinking water supplies in Cornwall, UK. Scientific Reports 6, 25656.

Middleton D, Watts M, Hamilton E, Fletcher T, Leonardi G, Close R, et al. 2016b. Prolonged exposure to arsenic in uk private water supplies: Toenail, hair and drinking water concentrations. Environmental Science: Processes & Impacts. 18, 562-574.

Miekeley N, Carneiro MD, da Silveira CP. 1998. How reliable are human hair reference intervals for trace elements? Science of the Total Environment 218:9-17.

Militino AF. 2010. Mixed effects models and extensions in ecology with R. Journal of the Royal Statistical Society: Series A (Statistics in Society) 173:938-939.

Mitchell P, Barr D. 1995. The nature and significance of public exposure to arsenic: A review of its relevance to south west England. Environmental Geochemistry and Health 17:57-82.

Moeller KE, Lee KC, Kissack JC. Urine drug screening: Practical guide for clinicians. In: Proceedings of the Mayo Clinic Proceedings, 2008, Elsevier. Vol. 83: 66-76.

Morgan MS. 1997. The Biological Exposure Indices: A key component in protecting workers from toxic chemicals. Environmental Health Perspectives 105:105.

Morton J, Carolan VA, Gardiner PH. 2002. Removal of exogenously bound elements from human hair by various washing procedures and determination by Inductively Coupled Plasma Mass Spectrometry. Analytica Chimica Acta 455:23-34.

Murcott S. 2012. Arsenic contamination in the world: An international sourcebook 2012. Water Intelligence Online 11:9781780400396.

120

Nathanail C. 2009. Professional practice note: Reviewing human health risk assessment reports invoking contaminant oral bioavailability measurements or estimates. Chartered Institute for Environment and Health.

Nathanail CP, Smith R. 2007. Incorporating bioaccessibility in detailed quantitative human health risk assessments. Journal of Environmental Science and Health Part A 42:1193-1202.

Navas-Acien A, Francesconi KA, Silbergeld EK, Guallar E. 2011. Seafood intake and urine concentrations of total arsenic, dimethylarsinate and arsenobetaine in the US population. Environmental Research 111:110-118.

Navas-Acien A, Silbergeld EK, Pastor-Barriuso R, Guallar E. 2008. Arsenic exposure and prevalence of type 2 diabetes in US adults. Jama 300:814-822.

Nermell B, Lindberg A-L, Rahman M, Berglund M, Åke Persson L, El Arifeen S, et al. 2008. Urinary arsenic concentration adjustment factors and malnutrition. Environmental Research 106:212-218.

Nermell B, Lindberg AL, Rahman M, Berglund M, Åke Persson L, El Arifeen S, et al. 2008. Urinary arsenic concentration adjustment factors and malnutrition. Environmental Research 106:212-218.

Nichol I, Thornton I, Webb J, Fletcher W, Horsnail R, Khaleelee J. 1971. Regional geochemical reconnaissance of part of Devon and north Dornwall: HM Stationery Office.

Norton G, Deacon C, Mestrot A, Feldmann J, Jenkins P, Baskaran C, et al. 2013. Arsenic speciation and localization in horticultural produce grown in a historically impacted mining region. Environmental Science & Technology 47:6164-6172.

NRC. 1999. Arsenic in drinking water. National Research Council . Subcommittee on arsenic in drinking water. National academies press

O’Reilly J, Watts M, Shaw R, Marcilla A, Ward N. 2010. Arsenic contamination of natural waters in San Juan and La Pampa, Argentina. Environmental Geochemistry and Health 32:491-515.

121

O’Reilly J. 2010. Arsenic speciation in environmental and biological samples from Argentina: Relationship between natural and anthropogenic levels and human health status. A thesis submitted to the University of Surrey for the degree of doctor of philosophy.

ONS. 2011. Office for National Statistics, 2011 census: Aggregate data (England and Wales). UK data service census support. Downloaded from: http://infuse.ukdataservice.ac.uk. This information is licensed under the terms of the open government licence [http://www.Nationalarchives.Gov.Uk/doc/open- government-licence/version/2].

Orentreich N, Markofsky J, Vogelman JH. 1979. The effect of aging on the rate of linear nail growth. Journal of Investigative Dermatology 73:126-130.

Orloff K, Mistry K, Metcalf S. 2009. Biomonitoring for environmental exposures to arsenic. Journal of Toxicology and Environmental Health, Part B 12:509-524.

Palumbo-Roe B, Klinck B. 2007. Bioaccessibility of arsenic in mine waste- contaminated soils: A case study from an abandoned arsenic mine in SW England (UK). Journal of Environmental Science and Health Part A 42:1251-1261.

Peach DF, Lane DW. 1998. A preliminary study of geographic influence on arsenic concentrations in human hair. Environmental Geochemistry and Health 20:231-237.

Pearce DC, Dowling K, Gerson AR, Sim MR, Sutton SR, Newville M, et al. 2010. Arsenic microdistribution and speciation in toenail clippings of children living in a historic gold mining area. Science of the Total Environment 408:2590-2599.

Pearce DC, Dowling K, Sim MR. 2012. Cancer incidence and soil arsenic exposure in a historical gold mining area in Victoria, Australia: A geospatial analysis. Journal of Exposure Science and Environmental Epidemiology 22:248-257.

Pebesma EJ, R.S. Bivand. 2004. Multivariable geostatistics in S: The gstat package. Computers & Geosciences, 30: 683-691.

Pebesma EJ, R.S. Bivand. 2005. Classes and methods for spatial data in R. R News 5 (2). http://cran.R-project.Org/doc/rnews/.

122

Peters SC, Blum JD, Klaue B, Karagas MR. 1999. Arsenic occurrence in New Hampshire drinking water. Environmental Science and Technology 33:1328-1333.

Pinheiro J BD, DebRoy S, Sarkar D and R Core Team. 2016. Nlme: Linear and nonlinear mixed effects models. R package version 3.1-126, .

Polya DA, Middleton, DRS 2016. Arsenic in drinking water: Sources & human exposure routes. In: Bhattacharya P, Polya DA and Jovanovic D (Eds.) Best practice guide for the control of arsenic in drinking water, IWA publishing, Chapter 1, ISBN13: 9781843393856.

Private Water Supplies Regulations 2009. Applying in England and coming into force on 1st of January 2010. http://www.legislation.gov.uk/uksi/2009/3101/introduction/made.

R Core Team. 2013. R: A language and environment for statistical computing, R foundation for statistical computing, Vienna, Austria. www.R-project.Org.

Raab A, Feldmann J. 2005. Arsenic speciation in hair extracts. Analytical and Bioanalytical Chemistry 381:332-338.

Ramirez-Andreotta MD, Brusseau ML, Beamer P, Maier RM. 2013. Home gardening near a mining site in an arsenic-endemic region of Arizona: Assessing arsenic exposure dose and risk via ingestion of home garden vegetables, soils, and water. Science of the Total Environment 454:373-382.

Rasmussen PE, Levesque C, Chénier M, Gardner HD, Jones-Otazo H, Petrovic S. 2013. Canadian house dust study: Population-based concentrations, loads and loading rates of arsenic, cadmium, chromium, copper, nickel, lead, and zinc inside urban homes. Science of The Total Environment 443:520-529.

Ratnaike RN. 2003. Acute and chronic arsenic toxicity. Postgraduate Medical Journal 79:391-396.

123

Rawlins B, O’Donnell K, Ingham M. 2003. Geochemical survey of the Tamar catchment (south-west England). British Geological Survey Report (CR/03/027), Keyworth, Nottingham. Available: http://nora.nerc.ac.uk/8497/.

Rawlins BG, Webster R, Lister TR. 2003. The influence of parent material on topsoil geochemistry in eastern England. Earth Surface Processes and Landforms 28:1389- 1409.

Revelle W. 2014. Psych: Procedures for psychological, psychometric, and personality research, Northwestern University, Evanston, Illinois, http://cran.R- project.Org/package=psych.

Richardson P. 1992. Mines of Dartmoor and the Tamar valley after 1913: Northern Mine Research Society.

Rieuwerts JS, Searle P, Buck R. 2006. Bioaccessible arsenic in the home environment in southwest England. Science of The Total Environment 371:89-98.

Rivera-Núñez Z, Meliker JR, Linder AM, Nriagu JO. 2010. Reliability of spot urine samples in assessing arsenic exposure. International Journal of Hygiene and Environmental Health 213:259-264.

Rivera-Núñez Z, Meliker JR, Meeker JD, Slotnick MJ, Nriagu JO. 2011. Urinary arsenic species, toenail arsenic, and arsenic intake estimates in a Michigan population with low levels of arsenic in drinking water. Journal of Exposure Science and Environmental Epidemiology 22:182-190.

Roberge J, Abalos AT, Skinner JM, Kopplin M, Harris RB. 2009. Presence of arsenic in commercial beverages. American Journal of Environmental Sciences 5:688-694.

Saad A, Hassanien MA. 2001. Assessment of arsenic level in the hair of the nonoccupational Egyptian population: Pilot study. Environment International 27:471-478.

124

Sage C. 1994. Environmental arsenic in England and Wales: The assessment and management of risk. Unpublished MSc. Thesis, Centre for Environmental Technology, Imperial College of Science, Technology and Medicine, London.

Salmela S, Vuori E, Kilpiö JO. 1981. The effect of washing procedures on trace element content of human hair. Analytica Chimica Acta 125:131-137.

Samanta G, Chowdhury TR, Mandal BK, Biswas BK, Chowdhury UK, Basu GK, et al. 1999. Flow Injection Hydride Generation Atomic Absorption Spectrometry for determination of arsenic in water and biological samples from arsenic-affected districts of West Bengal, India, and Bangladesh. Microchemical Journal 62:174-191.

Saoudi A, Zeghnoun A, Bidondo M-L, Garnier R, Cirimele V, Persoons R, et al. 2012. Urinary arsenic levels in the French adult population: The French national nutrition and health study, 2006–2007. Science of the Total Environment 433:206- 215.

Sata F, Araki S, Yokoyama K, Murata K. 1996. Adjustment of creatinine-adjusted values in urine to urinary flow rate: A study of eleven heavy metals and organic substances. International Archives of Occupational and Environmental Health 68:64-68.

Sata F, Araki S. 1996. Adjustment of creatinine-adjusted value to urine flow rate in lead workers. Archives of Environmental Health: An International Journal 51(4):329-33.

Sauvé JF, Lévesque M, Huard M, Drolet D, Lavoué J, Tardif R, et al. 2015. Creatinine and specific gravity normalization in biological monitoring of occupational exposures. Journal of Occupational and Environmental Hygiene 12:123-129.

Schulz C, Angerer J, Ewers U, Heudorf U, Wilhelm M, Agency HBCotGFE. 2009. Revised and new reference values for environmental pollutants in urine or blood of children in Germany derived from the German Environmental Survey on children 2003-2006 (GerES IV). International Journal of Hygiene and Environmental Health 212:637-647.

125

Schulz C, Wilhelm M, Heudorf U, Kolossa-Gehring M. 2011. Update of the reference and HBM values derived by the German Human Biomonitoring Commission. International Journal of Hygiene and Environmental Health 215:26- 35.

Schwartz GJ, Muñoz A, Schneider MF, Mak RH, Kaskel F, Warady BA, et al. 2009. New equations to estimate GFR in children with CKD. Journal of the American Society of Nephrology 20:629-637.

Senofonte O, Violante N, Caroli S. 2000. Assessment of reference values for elements in human hair of urban schoolboys. Journal of Trace Elements in Medicine and Biology 14:6-13.

Shaper A, Straf M, Perrin R. 1979. Epidemiology for geochemists [and discussion]. Philosophical Transactions of the Royal Society of London B, Biological Sciences 288:127-136.

Signes‐Pastor AJ, Al‐Rmalli SW, Jenkins RO, Carbonell‐Barrachina ÁA, Haris PI. 2012. Arsenic bioaccessibility in cooked rice as affected by arsenic in cooking water. Journal of Food Science 77:T201-T206.

Slotnick MJ, Meliker JR, AvRuskin GA, Ghosh D, Nriagu JO. 2007. Toenails as a biomarker of inorganic arsenic intake from drinking water and foods. Journal of Toxicology and Environmental Health, Part A 70:148-158.

Slotnick MJ, Meliker JR, Nriagu JO. 2006. Effects of time and point-of-use devices on arsenic levels in southeastern Michigan drinking water, USA. Science of The Total Environment 369:42-50.

Slotnick MJ, Nriagu JO. 2006. Validity of human nails as a biomarker of arsenic and selenium exposure: A review. Environmental Research 102:125-139.

Smedley P, Kinniburgh D. 2002. A review of the source, behaviour and distribution of arsenic in natural waters. Applied Geochemistry 17:517-568.

Smedley PA, D. 2004. Baseline report series. 16, the granites of south-west England. Environment Agency, 47pp. (CR/04/255N). Avilable: http://nora.nerc.ac.uk/3547/.

126

Smith AH, Lingas EO, Rahman M. 2000. Contamination of drinking-water by arsenic in Bangladesh: A public health emergency. Bulletin of the World Health Organization 78:1093-1103.

Sobus JR, DeWoskin RS, Tan Y-M, Pleil JD, Phillips MB, George BJ, et al. 2015. Uses of NHANES biomarker data for chemical risk assessment: Trends, challenges, and opportunities. Environmental Health Perspectives 123:919-927.

Sorahan T, Pang D, Esmen N, Sadhra S. 2008. Urinary concentrations of toxic substances: An assessment of alternative approaches to adjusting for specific gravity. Journal of Occupational and Environmental Hygiene 5:721-723.

Spayd SE, Robson MG, Buckley BT. 2015. Whole-house arsenic water treatment provided more effective arsenic exposure reduction than point-of-use water treatment at New Jersey homes with arsenic in well water. Science of The Total Environment 505:1361-1369.

Sriwana T, van Bergen MJ, Sumarti S, de Hoog JCM, van Os BJH, Wahyuningsih R, et al. 1998. Volcanogenic pollution by acid water discharges along Ciwidey river, west Java (Indonesia). Journal of Geochemical Exploration 62:161-182.

St-Amand A, Werry K, Aylward LL, Hays SM, Nong A. 2014. Screening of population level biomonitoring data from the Canadian Health Measures Survey in a risk-based context. Toxicology Letters 231:126-134.

Steinmaus C, Carrigan K, Kalman D, Atallah R, Yuan Y, Smith AH. 2005. Dietary intake and arsenic methylation in a US population. Environmental Health Perspectives 113:1153.

Steinmaus CM, Yuan Y, Smith AH. 2005. The temporal stability of arsenic concentrations in well water in western Nevada. Environmental Research 99:164- 168.

Sugár É, Tatár E, Záray G, Mihucz VG. 2012. Field separation–based speciation analysis of inorganic arsenic in public well water in Hungary. Microchemical Journal 107:131-5.

127

Tellez-Plaza M, Navas-Acien A, Crainiceanu CM, Guallar E. 2008. Cadmium exposure and hypertension in the 1999-2004 National Health and Nutrition Examination Survey (NHANES). Environmental Health Perspectives: 51-56.

Thoresby P, Thornton I. 1979. Heavy metals and arsenic in soil, pasture herbage and barley in some mineralised areas in Britain: Significance to animal and human health. Trace Substances in Environmental Health 13:93-103.

Thornton I. 1996a. Sources and pathways of arsenic in the geochemical environment: Health implications. Geological Society, London, Special Publications 113:153-161.

Thornton I. 1996b. Impacts of mining on the environment; some local, regional and global issues. Applied Geochemistry 11:355-361.

Thundiyil JG, Yuan Y, Smith AH, Steinmaus C. 2007. Seasonal variation of arsenic concentration in wells in Nevada. Environmental Research 104:367-373.

Tseng C-H. 2009. A review on environmental factors regulating arsenic methylation in humans. Toxicology and Applied Pharmacology 235:338-350.

Tseng W-P. 1977. Effects and dose-response relationships of skin cancer and blackfoot disease with arsenic. Environmental Health Perspectives 19:109.

Tsuji JS, Van Kerkhove MD, Kaetzel RS, Scrafford CG, Mink PJ, Barraj LM, et al. 2005. Evaluation of exposure to arsenic in residential soil. Environmental Health Perspectives:1735-1740.

Tyson J. 2013. The determination of arsenic compounds: A critical review. ISRN Analytical Chemistry 2013.

Vahter M. 2002. Mechanisms of arsenic biotransformation. Toxicology 181– 182:211-217.

Vaughan DJ. 2006. Arsenic. Elements 2:71-75.

Venables WNR, Ripley BD. 2002. Modern applied statistics with S. Fourth edition. Springer, New York. ISBN 0-387-95457-0.

128

Vij HS, Howell S. 1998. Improving the specific gravity adjustment method for assessing urinary concentrations of toxic substances. American Industrial Hygiene Association 59:375-380.

Wang QQ, Thomas DJ, Naranmandura H. 2015. Importance of being thiomethylated: Formation, fate, and effects of methylated thioarsenicals. Chemical Research in Toxicology 28:281-289.

Warnes GR. 2014. Sasxport: Read and write SAS xport files. http://CRANR- projectorg/package=SASxport.

Water Industry Act 1991. http://www.legislation.gov.uk/ukpga/1991/56/contents.

Water Supply (Water Quality) Regulations 2000 (amendment) Regulations 2007. Coming into force on 22nd December 2007. http://www.legislation.gov.uk/uksi/2007/2734/introduction/made.

Watts M, Barlow T, Button M, Sarkar S, Bhattacharya B, Alam MA, et al. 2013. Arsenic speciation in polychaetes (Annelida) and sediments from the intertidal mudflat of Sundarban mangrove wetland, India. Environmental Geochemistry and Health 35:13-25.

Watts M, Joy E, Young S, Broadley M, Chilimba A, Gibson R, et al. 2015. Iodine source apportionment in the Malawian diet. Scientific Reports 5: 15251. DOI: 10.1038/srep15251.

Watts M, O’Reilly J, Marcilla A, Shaw R, Ward N. 2010. Field based speciation of arsenic in UK and Argentinean water samples. Environmental Geochemistry and Health 32:479-490.

Watts MJ, Button M, Brewer TS, Jenkin GR, Harrington CF. 2008. Quantitative arsenic speciation in two species of earthworms from a former mine site. Journal of Environmental Monitoring 10:753-759.

Webb JS, Thornton I, Thompson M, Howarth R, Lowenstein P. 1978. The Wolfson geochemical atlas of England and Wales. Applied Geochemistry Research Group, Imperial College, London, UK.

129

Wheeler BW, Kothencz G, Pollard AS. 2013. Geography of non-melanoma skin cancer and ecological associations with environmental risk factors in England. British Journal of Cancer 109:235-241.

WHO. 2001. Environmental Health Criteria 224: Arsenic and arsenic compounds. World Helath Organisation, Geneva.

WHO. 2011. Arsenic in drinking-water - background document for development of WHO guidelines for drinking-water quality, World Helath Organisation, who/sde/wsh/03.04/75/rev/1.

Wickre JB, Folt CL, Sturup S, Karagas MR. 2004. Environmental exposure and fingernail analysis of arsenic and mercury in children and adults in a Nicaraguan gold mining community. Archives of Environmental Health: An International Journal 59:400-409.

Wilcox RR, Tian T. 2008. Comparing dependent correlations. Journal of General Psychology 135:105-112.

Wilhelm M, Ewers U, Schulz C. 2004a. Revised and new reference values for some trace elements in blood and urine for human biomonitoring in environmental medicine. International Journal of Hygiene and Environmental Health 207:69-73.

Wilhelm M, Pesch B, Wittsiepe J, Jakubis P, Miskovic P, Keegan T, et al. 2004b. Comparison of arsenic levels in fingernails with urinary as species as biomarkers of arsenic exposure in residents living close to a coal-burning power plant in Prievidza district, Slovakia. Journal of Exposure Science and Environmental Epidemiology 15:89-98.

Wilson J, Dixon S, Galke W, McLaine P. 2006. An investigation of dust lead sampling locations and children's blood lead levels. Journal of Exposure Science and Environmental Epidemiology 17:2-12.

Wolfsperger M, Hauser G, Göβler W, Schlagenhaufen C. 1994. Heavy metals in human hair samples from Austria and Italy: Influence of sex and smoking habits. Science of the Total Environment 156:235-242.

130

Xu J, Thornton I. 1985. Arsenic in garden soils and vegetable crops in Cornwall, England: Implications for human health. Environmental Geochemistry and Health 7:131-133.

Xue J, Zartarian V, Wang S-W, Liu SV, Georgopoulos P. 2010. Probabilistic modeling of dietary arsenic exposure and dose and evaluation with 2003–2004 NHANES data. Environmental Health Perspectives 118:345.

Yeh HC, Lin YS, Kuo CC, Weidemann D, Weaver V, Fadrowski J, et al. 2015. Urine osmolality in the US population: Implications for environmental biomonitoring. Environmental Research 136:482-490.

Zhijie MY, Dummer TJ, Adams A, Murimboh JD, Parker L. 2014. Relationship between drinking water and toenail arsenic concentrations among a cohort of Nova Scotians. Journal of Exposure Science and Environmental Epidemiology 24:135- 144.

131

APPENDIX A: ETHICAL APPROVAL LETTER

132

Secretary to Research Ethics Committees Room 2.004 John Owens Building Compliance and Risk Office University of Manchester Tel: 0161 275 2206/2046 Oxford Road Fax: 0161 275 5697 Manchester, M13 9PL Email: [email protected] ref: ethics/13068

Professor David Polya Professor of Environmental Geography, SEAS, Willamson Building, G27

6th June 2013

Dear Dave,

Research Ethics Committee 2 Polya, Middleton, Mitchell, Cutting, Gibson, Pan, Watts, Studden, Russell, Sauer: BAPS- biomonitoring of arsenic exposure from consumption of private water supplies in South West England (ref 13068)

I write to thank you for attending the meeting on 3rd June and to confirm that the amendments and additional information sent in your email of 4thJune satisfies the concerns of the Committee and that the project has been given a favourable ethical opinion.

This approval is effective for a period of five years and if the project continues beyond that period it must be submitted for review. It is the Committee’s practice to warn investigators that they should not depart from the agreed protocol without seeking the approval of the Committee, as any significant deviation could invalidate the insurance arrangements and constitute research misconduct.

I would be grateful if you could complete and return the attached form at the end of the project or by the end of June 2014.

Yours sincerely

Dr T P C Stibbs Secretary to the University Research Ethics Committee

133

Enclosed: Report form

UNIVERSITY OF MANCHESTER

COMMITTEE ON THE ETHICS OF RESEARCH ON HUMAN BEINGS

Progress or Completion Report Form on an Approved Project

The Committee's procedures require those responsible for projects which have been approved by the Committee to report on any of the following:

* Any incident, accident or untoward event associated with the project (Please note that if the incident constitutes an accident or dangerous occurrence, the usual Health and Safety reporting mechanism must still be used) * Any variation in the methods or procedures in the approved protocol * A termination or abandonment of the project (with reasons) * A report on completion of the project or a progress report 12 months after approval has been given.

The report should be sent to the Secretary to the Committee, Dr T P C Stibbs, Room 2.004 John Owens Building, University of Manchester, Oxford Road, Manchester M13 9PL (tel: 0161-275-2046/2206).

Project: BAPS- biomonitoring of arsenic exposure from consumption of private water supplies in South West England (ref 13068)

134

APPENDIX B: RESULTS LETTERS

135

[Name]

[Address]

[Date]

Dear [Name]

“A study of the association between arsenic consumption from private drinking water supplies and measured biological levels in the population of Cornwall, UK”

We are writing to you following the recent collection of your toenail, hair, urine and environmental samples as part of our private drinking water supplies study. The work is part of a joint study undertaken by Public Health England, the University of Manchester and the British Geological Survey. I would like to take this opportunity to thank you for taking part in this study and for providing samples. Many participants requested water results to be fed back as quickly as possible and we are now able to do so. However, the results of your biological and other environmental samples will take more time to prepare, and we will send them to you in due course.

The concentrations of several naturally occurring elements in your drinking water were measured from a sample taken from your drinking water tap. This study did not include microbiological testing for levels of bacteria and other water borne pathogens.

You will probably be aware that Cornwall is an area rich in metal deposits. Over the years copper, tin and arsenic, as well as other metals have all been mined in the county.

136

As a result we would expect to find higher than usual concentrations of these naturally occurring elements in some Cornish private drinking water supplies. However, the majority of the elements found are likely to be at very low levels and are unlikely to cause harm to health, in some instances they may even be beneficial.

Your drinking water test results

To understand what your results mean, we have provided a table of results (separate sheet enclosed) which shows your results compared with the existing Prescribed Concentrations or Values (PCV) as set out in the Private Drinking Water Regulations. This lists all the chemical elements with at least one exceedance in our surveys of Cornish private water supplies. The regulations exist to protect aesthetic characteristics (such as taste and smell) as well as public health and ensure a good quality of water.

PCVs are European and national standards that determine an acceptable level of certain elements, compounds and substances which are permitted in drinking water.

We want to make it clear that your result is only based on one sample, therefore, the result must be interpreted with caution, as there are different factors, such as seasonal variation (particularly wet or dry months) that could affect the results.

Exceedances: If your result exceeds any of the PCVs, we would advise that your water supply is tested again. This will confirm if subsequent action, such as installing water treatment, is necessary. If your results show exceedances we have provided you with a factsheet describing the possible health risks of long term exposure.

No exceedances: If your sample results indicate that there were no exceedances of the PCVs, then you should continue as normal and no further action is required based on these results.

If you would like a repeat water test to be carried out on your drinking water supply you can call Cornwall Council on 01209 616990 and ask for the private water

137

supply officer for your area, who will give you advice and details regarding retesting. There may be a charge for this.

If, after repeat sampling of your water supply, there is confirmation that elements present in your drinking water supply still exceed the PCV set out in the Regulations, there are various ways to reduce the concentration: It may be there is leaching from the plumbing system which can be fixed or you can install specialist treatment equipment such as a filtering or chemical system or you could seek an alternative supply. A treatment plant should be as close to the house as possible, to reduce the concentrations of the elements in your water supply to comply with the PCVs. Cornwall Council will be able to discuss these options with you and suggest the necessary actions that you may need to undertake.

Responsibilities

Responsibility for ensuring that a supply is of the required quality may lie with one or more people such as the home owner, the occupier, landlord, owner of the land where the supply comes from or some other person having control or management over it. Exactly who this might be in any given case can vary.

Further Information

Should you need further advice or help on your private water supply please call Cornwall Council on 01209 616990 and ask for the private water supply officer for your area.

Should you need further advice or information on the study, then please contact Public Health England; By email: [email protected] Or to speak to one of the team, phone: 01235 825042

Yours sincerely

[Signature of T.Fletcher]

Dr Tony Fletcher, Principal Investigator, Public Health England

138

The research team for the project “A study of the association between arsenic consumption from private drinking water supplies and measured biological levels in the population of Cornwall, UK” includes Daniel Middleton (PhD Researcher, University of Manchester), Professor David Polya (University of Manchester, Principal Investigator), Dr Michael Watts (British Geological Survey, Principal Investigator) and Dr Tony Fletcher (Public Health England).

139

Your result:

Sample details: Sample collected from [Name] Sample address [Address] Sample date and time [Appointment date] [Appointment time] Sample ID [Sample code] Source code [Source code] Sample point Kitchen tap

Please find your individual water test results below, in the final column (sample results). You can compare your results to the upper limit in the drinking water regulations (prescribed concentration or value, PCV) found in the third column for each reported element.

Naturally Units Recommended Sample result occurring regulation level element (PCV) Aluminium µg/L 200 137 Antimony µg/L 5 2.1 Arsenic µg/L 10 5.3 Copper µg/L 2000 130 Iron µg/L 200 198 Lead µg/L 10 20 Manganese µg/L 50 53 Nickel µg/L 20 10

Concentration units 1 mg/l (milligrams per litre) = 1000 µg/L (micrograms per litre). Sometimes the terminology “parts per million” (ppm) is used for milligrams per litre (mg/L) and “parts per billion” (ppb) for micrograms per litre (µg/L).

The PCV for lead was reduced from 25µg/L to 10µg/L on the 25/12/2013.

Explanation of terms PCV - Prescribed concentration or values are part of the Private Drinking Water Regulations 2009, these regulations are set out to protect public health and ensure a good quality of water. They are European standards that determine an acceptable level of certain elements, compounds and substances which are permitted in water.

If one of your results exceeds the Prescribed Concentrations or Values (PCV) set out we have enclosed relevant health information on that element.

Public Health England also holds data for your sample on other chemical parameters analysed on this sample.

140

[Name]

[Address]

[Date]

“A study of the association between arsenic consumption from private drinking water supplies and measured biological levels in the population of Cornwall, UK”

Dear [Name],

We are writing to you following the collection of your toenail, hair, urine and environmental samples for this study. The work was part of a joint study undertaken by Public Health England, the University of Manchester and the British Geological Survey. We would like to take this opportunity to thank you for taking part and to provide you with your results.

You will probably be aware that Cornwall is an area rich in mineral deposits in the ground. Over the years arsenic, as well as metals, have all been mined in the area. As a result, we can find higher than usual concentrations of these naturally occurring elements in Cornish private drinking water, as well as in soil and dust around the area. Additionally, food such as seafood and imported rice can contain concentrations of arsenic and other metals.

The overall aim of this study is to look at patterns of arsenic measured in both water supplies and local residents (measured in urine, hair and nails), also taking into account potential uptake from food and garden soil. We will be investigating the relationship between these measurements to understand better the sources of community exposure, especially to arsenic. We are providing you with your individual results, for your interest.

Your results You have previously received from us and Cornwall Council, the results of the water sampling testing and the levels of chemicals in your private water supply. We are now

141

enclosing the results of your biological samples, along with an interpretation. To understand what your results mean, we have provided a table in Annexe 1 (see the separate sheet enclosed) which shows your results compared to typical values we might expect to see and an explanation of what your results mean is provided.

The results from samples of soil, rice and dust from the dust wipes taken (if you gave these) will contribute to the wider research conducted by the British Geological Survey, the University of Manchester and Public health England, examining arsenic in dust and food. We are still analysing and compiling the results of soil sampling and when this is completed, we will write to you with the results from your samples.

Due to the technique used to analyse arsenic levels, other chemical parameters have also been measured in your biological samples. The data helps us understand background population levels and patterns of exposure. We have checked and for those chemicals where there are health based guidance values, no sample results exceed these values, and so we have not provided this data. Should you require any of this data or the measurements we made in the rice and dust, please contact us and we will forward them to you. These data are held securely in accordance with UK data protection legislation and approved by the University of Manchester ethics committee (ref 13068) and the NHS National Research Ethics Committee (ref 13/EE/0234).

We have now given you the results of arsenic in your biological and water samples taken from your household. Thank you once again for taking part in this research. We will publish our overall findings of the study results over the next 12 to 18 months which you will be able to access from the British Geological Survey project website: http://www.bgs.ac.uk/cornwallPWS . This website also provides further information about the study and further sources of information.

Should you need further advice, please contact Public Health England; By email: [email protected] Or to speak to one of the team, phone: 01235 825042

Yours sincerely [Signature of T. Fletcher]

142

Dr Tony Fletcher, Principle Investigator, Public Health England

The research team for the project “A study of the association between arsenic consumption from private drinking water supplies and measured biological levels in the population of Cornwall, UK” also includes Daniel Middleton (PhD Researcher, University of Manchester), Professor David Polya (University of Manchester, Principal Investigator), Dr Michael Watts (British Geological Survey, Principal Investigator)

143

Annexe 1: Biological sample results

Sample collected from [Name] Appointment date [Appointment date] House code [House code] Volunteer code [Volunteer code]

Please find your individual urine, hair, and toenail results, alongside your drinking water results. For urine you can compare your results with the current guidance value which is provided in the table. There are no guidance values for arsenic in hair and toenails, we can only tell you how they compare to the range found across the study population. The levels of arsenic measured in your urine reflects exposure to arsenic in the previous 48 hours, whereas hair and toenails reflect past exposures over several weeks and may be related to your drinking water supply or other sources.

Finding a measurable amount of arsenic in your urine, hair or toenail samples does not mean that the level of arsenic will cause an adverse health effect. However, it is good to keep these exposures as low as possible, particularly below the current guidance value. Also, we want to make it clear that your results are only based on one sample and can be affected by a range of factors.

Sample Units Guidance Range of study Sample Result collected value* population result** compared to levels guidance value Urine µg/L 35

Water µg/L 10

Hair mg/kg Not available

Toenail mg/kg Not available

*Guidance values for water from the Drinking Water Inspectorate. For urine we have used the recommended limit for these arsenic compounds developed in the USA for workers. There are no guidance values established for the general population, but this value for workers is helpful for comparison. **Insufficient sample – for some hair and nail samples there was insufficient sample for analysis.

Concentration units In urine or water µg/L (micrograms per litre) = ppb (parts per billion) In hair and toenails 1 mg/kg (milligrams per kilograms) = 1 ppm (parts per million) = 1000 µg/kg (micrograms per kilogram).

144

Very High [This text will only go to participants with high urine] Urine As explained to you during the telephone call on the 24th September 2014, High Water both your urine and water samples showed arsenic levels above the guidance values. It is very possible that the levels of arsenic measured in your urine are related to your drinking water supply, and therefore your exposure should go down as a benefit from reducing arsenic levels in your drinking water. As we had suggested before when we wrote about the water concentration: your result exceeded the arsenic Prescribed Concentration Value (PCV), and so we advise that your water supply is tested again and you consider subsequent action, such as installing water treatment.

Next steps You may want to discuss this further with your GP. If you do then please take this letter and the information leaflet provided with you if you seek further advice from your GP. We are aware that you have recently had your water quality re-tested by the council and depending on these results you may want to consider additional water treatment. Please refer to the private drinking water advice sheet enclosed. Further advice on remediation of water quality is available from Cornwall Council on 0300 1234 212. High Urine [This text will only go to participants with high urine >35 and <100] High Water Both your urine and water samples showed arsenic levels above the guidance values. It is very possible that the levels of arsenic measured in your urine are related to your drinking water supply, and therefore your exposure should go down as a benefit from reducing arsenic levels in your drinking water. As we had suggested before when we wrote about the water concentration: your result exceeded the arsenic Prescribed Concentration Value (PCV), and so we advise that your water supply is tested again and you consider subsequent action, such as installing water treatment.

Next steps Consider re-testing your water quality and installing additional water treatment. Please refer to the private drinking water advice sheet enclosed. Further advice on remediation of water quality is available from Cornwall Council on 0300 1234 212.

145

If you would like to discuss this, or anything else, further please call PHE on 01235 825042. High urine [This text will only go to participants with high urine and low water] Low water Your urine result is above the guidance value.

Analysis of your water sample showed low arsenic levels well below the guidance value.

The most likely explanation for this is that you had a higher than normal arsenic level related to dietary intake, perhaps some fish or shellfish not long before we took the samples. Most of the arsenic from these sources is in a harmless form.

If you would like to discuss this further please call PHE on 01235 825042.

Next steps As the arsenic level in your water sample was low, no immediate action on your water supply is needed. Please continue to monitor you water quality through regular testing (because levels can change over time), for arsenic and other potentially harmful constituents.

Low urine Your urine result is below the guidance value. High water Analysis of your water sample showed that the arsenic level was above the guidance value.

Next steps As your water sample was high, you have already received a letter in which we have suggested that your water quality is re-tested and you may want to consider additional water treatment. Please refer to the private drinking water advice sheet enclosed. Further advice on remediation of water quality is available from Cornwall Council on 0300 1234 212.

Low urine Your urine result is below the guidance value. Low water Analysis of your water sample showed that the arsenic level was below the guidance value. Please continue to monitor you water quality through regular testing (because levels can change over time), for arsenic and other potentially harmful constituents.

146

[Name]

[Address]

[Date]

“A study of the association between arsenic consumption from private drinking water supplies and measured biological levels in the population of Cornwall, UK”

Dear [Name],

We are writing to you following the earlier letters you received about water and biological samples collected in this study. The work was part of a joint project undertaken by Public Health England, the University of Manchester and the British Geological Survey, and we again thank you for taking part.

The overall aim of this study is to look at patterns of arsenic measured in both water supplies and local residents (measured in urine, hair and nails), but also taking into account potential uptake from food and garden soil. We are currently investigating the relationship between these measurements to understand better the sources of community exposure, especially to arsenic. This letter is letting you know the results we found in the soil samples from your garden, for your interest.

Your soil results Cornwall’s soils contain a number of minerals, particularly arsenic and lead, which are higher than the national average because of the natural mineral deposits and related historical mining/smelting activities in the county.

The sampling approach in this study involved, in most cases, taking single samples per garden or vegetable patch which were analysed to indicate the chemical elements contained in the soil. The plan for this part of the study was to present an overview of chemical elements in soil across the county, and for that we have taken one sample per house in most cases. For a detailed assessment of exposure, or risk to people, in a specific household's garden, several measurements would be required. The single

147

measurement we have taken does however provide an indication of possible exposures and following a single high value it may be worthwhile undertaking a fuller investigation. The local authority can advise: more samples could be taken to see if there are consistently high values and therefore, if any protective measures are recommended to be taken.

Your results and a comparison with the guidance values are shown in the Annexe at the end of this letter.

For information only, the soil results have been compared to guidance values used by local authorities for assessing land quality, namely the ‘Category 4 Screening Level (C4SL)’ that exist for specific chemical elements. You can access further information on the C4SL guidance value for specific chemical elements, via the project update on the BGS project web link given at the end of this letter, which will direct you to the DEFRA Department for Environment and Rural Affairs web link.

In summary, if the concentrations of chemical elements contained in your soils are higher than the guidance levels it does not necessarily mean that there will be a risk to your health or that land is contaminated, it suggests that further investigation may be justified. For a chemical to be a risk, you have to come into contact with it by breathing it in, eating it or by absorbing it through your skin. Your exposure to these elements in the soil would depend upon the amount of contact that you have with it and whether you eat fruit and vegetables which are grown in the soil.

If your results are higher than the guidance levels then you may wish to investigate this further – if this is the case we recommend you discuss this further with Cornwall Council. A contact number is provided on the results sheet. We want to make it clear that your results are based on single or a small number of samples, meaning results they must be interpreted with caution, as they may not fully represent the soil in your garden.

148

What next

We have now given you the results of arsenic in your biological, water and soil samples taken from your household. The results from all these samples, plus those from rice and dust from the dust wipes taken (if you gave these) will contribute to the wider research conducted by the research team examining patterns of mineral exposures and uptake. We will publish our overall findings of the study results over the next 12 to 18 months (aggregate results not individual measurements), which you will be able to access from the British Geological Survey project website http://www.bgs.ac.uk/sciencefacilities/laboratories/geochemistry/igf/biomonitoring/ar senicSW.html

Thank you once again for taking part in this research.

Yours sincerely [Signature of T. Fletcher]

Dr Tony Fletcher, Principle Investigator, Public Health England

The research team for the project “A study of the association between arsenic consumption from private drinking water supplies and measured biological levels in the population of Cornwall, UK” also includes Daniel Middleton (PhD Researcher, University of Manchester), Professor David Polya (University of Manchester, Principal Investigator), Dr Michael Watts (British Geological Survey, Principal Investigator).

149

Annexe: Soil sample results

As part of this study we took soil samples in most gardens to investigate the various possible sources of exposure, mainly arsenic. We will be looking at correlations between these arsenic and other measurements across the population, to help PHE understand population exposures. We are providing you with your individual results, for your interest.

The results from analysing your soil samples are listed in the tables below. Results are provided for elements for which there is a ‘Category 4 Screening Level (C4SL)’ – reference value. More elements were measured in the soil samples, but are not presented in this letter owing to the lack of a reference value for comparison. These additional results are available on request.

Sample details:

Sample collected from [Name] Address [Address] Sample date [Appointment date] House code [House code]

Your soil sample results: Two values of soil levels are presented for the elements which we can compare to reference values, including arsenic, cadmium and lead. Values are provided where a soil sample was collected from a vegetable patch used for ‘home grown’ food or a soil representative of the local area soil type. The tables below give your sample results which are compared to the relevant C4SL value for the soils tested at your property. Where ‘NA’ is given in the tables, a laboratory measurement or soil sample was not available. Where ‘

Soil arsenic: Sample Sample Units Reference Measured Arsenic description Value

1 Vegetable mg/kg 37 patch 2 Garden mg/kg 40

Soil cadmium: Sample Sample Units Reference Measured description Value Cadmium

1 Vegetable mg/kg 14 patch 2 Garden mg/kg 87

150

Soil lead: Sample Sample Units Reference Measured Lead description Value

1 Vegetable mg/kg 200 patch 2 Garden mg/kg 310

If your soil sample results are high for arsenic, cadmium or lead this does not necessarily mean that there will be an effect on your health. For a chemical to be a risk, you have to come into contact with it by breathing it in, eating it or through skin contact. Your risk of exposure to these elements from the soil will depend upon the amount of direct contact that you have with it and whether you eat fruit and vegetables which are grown in the soil.

However, if your value is high and in particular exceeds the C4SL reference value, it is recommended to take extra care to reduce your exposure (see next step). For further information on the C4SL reference value see http://randd.defra.gov.uk/Default.aspx?Module=More&Location=None&ProjectID=183 41.

Next Steps Cornwall Council are the principal regulators for contaminated land and have sole responsibility for the identification of contaminated land within your area. You can contact them about any concerns you may have. Their contact number is 0300 1234 212, please ask to speak to someone in regard to contaminated land.

Meanwhile we would offer the following temporary advice to limit any exposure to chemicals in your garden:

DO • Continue to enjoy using all of your garden space as you normally would, including grassed areas, decking, patios and other hard standings • Wear gloves when gardening • Follow normal hygiene precautions and wash hands thoroughly after working or playing in the garden and before handling food • Remove your normal gardening shoes before entering the house • Wash and peel produce grown in the soil from your garden in order to remove any soil and dust • Talk to your Council before having building works done

DON`T • Let children play with the soil or put it in their mouths • Let pets dig holes in your garden • Dig holes for building works without talking to your Council first

151

APPENDIX C: LIST OF CONFERENCE PRESENTATIONS AND OUTREACH ACTIVITIES

152

Conference presentations Middleton DRS, Watts MJ, Hamilton EM, Ander EL, Close RM, Crabbe H, Exley KE, Leonardi GS, Fletcher T and Polya DA. Adjusted urinary biomonitoring results indicate exposure to inorganic arsenic from private drinking water supplies in Cornwall, South West England. 4th UK and Ireland Exposure Science Meeting. Health and Safety Laboratory (HSL), Buxton, UK. (Poster presentation). (2016) (1st Place best poster prize). Middleton DRS. Human biomonitoring of inorganic arsenic exposure from private water supplies, Cornwall, UK. BGS Lunchtime Lecture. (2015) Middleton DRS, Watts MJ, Hamilton EM, Ander EL, Close RM, Exley KE, Crabbe H, Leonardi GS, Fletcher T and Polya DA. Urinary arsenic profiles reveal substantial exposures to inorganic arsenic from private water supplies in Cornwall, UK. 6th International Conference on Medical Geology, Portugal. (Oral presentation) (2015) Middleton DRS, Watts MJ, Hamilton EM, Close RM, Exley KE, Crabbe H, Leonardi GS, Fletcher T and Polya DA. Toenail biomarkers of exposure to inorganic arsenic from private water supplies in Cornwall, UK. 31st International Conference of the Society for Environmental Geochemistry and Health, Slovakia. (Oral presentation) (2015) Watts MJ, Middleton DRS. Joint presentation of project progress to stakeholders including UK drinking water inspectorate (DWI). Environmental Public Health Tracking Strategic Advisory Group Meeting. PHE (2015) Middleton DRS, Watts MJ, Hamilton EM, Ander EL, Close RM, Crabbe H, Exley KE, Leonardi GS, Fletcher T and Polya DA. Non-invasive biomonitoring of human arsenic exposure in Cornwall, UK. 3rd UK and Ireland Exposure Science Meeting. Imperial College, London. (Oral presentation) (2015) Middleton DRS, Watts MJ, Hamilton EM, Ander EL, Close RM, Crabbe H, Exley KE, Dunne A, Marriott A, Rimell A, Leonardi GS, Fletcher T and Polya DA. Urinary arsenic concentrations of a population of private water supply users: Cornwall, SW England. 30th International Conference of the Society for Environmental Geochemistry and Health, Newcastle, UK. (Poster presentation +2 min oral advertisement) (2014) In addition: 2 local authority (Cornwall County Council) oral presentations on project progress; oral presentations at postgraduate research events at the University of Manchester and PHE; 2 poster presentations at BGS postgraduate science festivals. Outreach activities Radio interview (Pirate FM) (2016). Radio interview (BBC Radio Cornwall) (2016). Middleton DRS. Biomonitoring studies: examples and benefits. Manchester Policy Week (public event. Invited speaker (Oral presentation and panellist) (2015)

153

News articles arising from this research project Toxic arsenic exposure discovered in Cornish private water supplies -Today Topics (11/05/2016)

Toxic arsenic exposure discovered in Cornish private water supplies -Medical News Today (11/05/2016)

Dangerously high levels of toxic arsenic found in Cornwall private water supplies -Cornish Guardian (11/05/2016)

Toxic arsenic found in Cornwall private water supplies -BBC News (10/05/2016)

Toxic arsenic exposure discovered in Cornish private water supplies -Phys.org (10/05/2016)

DRINKING WATER SCARE: Homeowners at risk from dangerously high levels of deadly arsenic -Express (10/05/2016)

154

APPENDIX D: EXTENDED ABSTRACT

155

A modified creatinine adjustment method to improve urinary biomonitoring of exposure to arsenic in drinking water

D.R.S. Middleton1,2,3, M.J. Watts2, T. Fletcher3 & D.A. Polya1 1School of Earth, Atmospheric and Environmental Sciences and Williamson Research Centre for Molecular Environmental Science, University of Manchester, Manchester, UK 2Inorganic Geochemistry, Centre for Environmental Geochemistry, British Geological Survey, Nicker Hill, Keyworth, Nottinghamshire, UK 3Centre for Radiation, Chemicals and Environmental Hazards (CRCE), Public Health England, Chilton, Didcot, Oxfordshire, UK

ABSTRACT: Spot urinary concentrations of arsenic (As) require adjustment for hydration variation. There is no consensus as to the most appropriate method. We compared the performance of creatinine (Cre), specific gravity (SG), osmolality and a modified variant of Cre adjustment on spot urinary As concentrations of 203 volunteers from the UK. Strength in correlation between drinking water total As and urinary non-arsenobetaine As was used to indicate good performance. Performance of the order: Cre (modified) > SG > osmolality > unadjusted > Cre (routine) was observed. Only Cre (modified) adjustment performed significantly better than unadjusted concentrations and both Cre (modified) and SG performed significantly better than Cre (routine). This is the first study to compare the performance of multiple hydration adjustments of urinary As on the basis of reflectance of external exposure and, importantly, allows others to apply this methodology to existing data without further measurements.

1 INTRODUCTION propose a modification of the Cre adjustment, based on previous work (Vij & Howell, 1998) as follows: Urinary biomonitoring is routine for assessing recent z exposure to arsenic (As) given its rapid excretion Asadjusted = Asunadjusted × (Creref/Crespecimen) (2) from the body (2-4 days post-ingestion) (Orloff et al., 2009). Twenty-four hour sampling is often not where: Creref is the median Cre concentration of the feasible and spot urine samples require adjustment study group and z is a coefficient, empirically de- for inter-volunteer differences in hydration i.e. uri- rived by dividing the regression slope (b) of As nary dilution (Aylward et al., 2014). The Google against UFR by that of Cre against UFR (a) hence z Scholar search term: “arsenic; urine; (one of) creat- = b/a. We test the performance of modified Cre ad- inine, specific gravity, osmolality; adjustment OR justment against alternative methods using the corre- correction” yielded 5070, 2840 and 625 results for lation between urinary As and drinking water As, creatinine (Cre), specific gravity (SG) and osmolali- with stronger correlations indicative of better per- ty, respectively. This reflects the wide use of Cre ad- formance. justment, relative to alternative adjustment methods. Creatinine adjustment has received criticism in the 2 METHODS/EXPERIMENTAL literature due to demographic and nutritionally de- rived variations in creatinine excretion (Barr et al.,

2005). Less attention has been paid to the mathemat- 2.1 Data acquisition ical application of creatinine adjustment, routinely Two-hundred and three volunteers from Cornwall, performed: UK provided paired urine and drinking water sam- ples as per their participation in a wider biomonitor- Asadjusted = Asunadjusted/Crespecimen, (1) ing study for which ethical approval was granted by the University of Manchester Research Ethics with results commonly expressed as As µg/g Cre Committee (Ref 13068) and the NHS Health Re- (The term “arsenic; "/g cre" OR "/g cr" OR "/g cre- search Authority National Research Ethics Commit- atinine" yielded 3320 results). This assumes a pro- tee (NRES) (Ref 13/EE/0234). Urinary As specia- portional change in the concentration of both As and tion was performed by HPLC-ICP-MS and AsIII + Cre in response to change in urinary flow rate AsV + MA +DMA concentrations used as the expo- (UFR). This has been disputed by findings that a sure biomarker: U-AsIMM. Urinary Cre was deter- range of urinary analytes exhibit different concentra- mined by the Jaffe method using a colorimetric liq- tion changes in response to UFR and adjustments uid assay kit, SG was determined using a digital require the inclusion of a coefficient to account for refractometer and osmolality was measured using a this relationship (Araki et al., 1990). We, therefore, cryoscopic osmometer.

156

2.2 Adjustment factor performance assessment 4 CONCLUSIONS

Urinary-AsIMM concentrations were adjusted for: Cre While acknowledging the criticisms of the use of (routine method – Equation 1); SG and osmolality Cre adjustment of biomonitoring results due to de- (Levine-Fahy method) (Levine & Fahy, 1945) and mographic differences in Cre clearance, we show in Cre (modified method - Equation 2). The z value this example that an appreciable degree of perfor- used in Equation 2 was derived numerically by fix- mance is hindered by the mathematical application ing the denominator (a) at 0.68 (Araki et al., 1990) of the Cre metric, seldom addressed to-date. These and, given that no published values for As against findings reiterate the element-specific nature of uri- UFR (b) were available, processing multiple correla- nary hydration adjustments and provide an oppor- tions for a range of b between 0 and 1.5 at incre- tunity for research groups to retrospectively adjust ments of 0.01. The value yielding the strongest cor- existing biomonitoring data using this modified ap- relation was selected and resulted in a z value of proach without the requirement of extra measure- ments. This is the first study to compare Cre, SG and 0.49. Pearson correlation coefficients (rp) between unadjusted and various adjusted U-AsIMM concentra- osmolality adjustments of urinary As concentrations tions against drinking water total As were calculat- against a performance criteria based on external en- ed. Variables were natural log-transformed prior to vironmental exposure. analysis due to positively skewed distributions. To test the significance of differences between correla- ACKNOWLEDGEMENTS tions, Williams’ tests were performed (Venables, 2002). All statistical analyses were performed in the DRSM was funded by NERC via a University of R programming environment (R Core Team, 2013). Manchester/BUFI (Centre for Environmental Geo- chemistry) PhD studentship (GA/125/017, BUFI 3 RESULTS AND DISCUSSION Ref: S204.2). We thank Elliott Hamilton, Louise Ander, Andrew Marriott and Andrew Dunne (all Performance correlations for the range of adjustment BGS) and Giovanni Leonardi, Rebecca Close, Helen factors investigated are presented in Table 1 with Crabbe, Karen Exley, Amy Rimell and Mike Stud- their significant differences relative to one-another. den (all PHE) for their involvement in the wider pro- The order of performance on the basis of numerical ject. We thank Nigel Kendall and David Gardner from the University of Nottingham Biosciences for correlations alone was Cre (modified) (rp = 0.69) > SG (r = 0.68) > osmolality (r = 0.67) > unadjusted facilitating creatinine and osmolality measurements. p p Murray Lark, Chris Milne, Mark Cave and Simon (rp = 0.65) > Cre (routine) (rp = 0.62). Williams’ test results indicated that only modified Cre adjustment Chenery from the BGS are thanked for advice. yielded a statistically stronger correlation than unad- justed concentrations and modified Cre and SG ad- REFERENCES justments yielded a statistically stronger correlations Araki, S., Sata, F., Murata, K. 1990. Adjustment for urinary than routine Cre adjustment. The failure to detect flow rate: An improved approach to biological monitoring. significantly stronger correlations between SG and Int. Arch.Occ. Env. Hea. 62: 471-477. osmolality and unadjusted concentrations may be at- Aylward, L.L., Hays, S.M., Smolders, R., Koch, H.M., Cocker, tributed to the sample size and a lack of statistical J., Jones, K., et al. 2014. Sources of variability in biomarker power. It is possible to modify SG and osmolality concentrations. J. Toxicol. Env. Heal. B 17:45-61. Barr, D.B, Wilder LC, Caudill SP, Gonzalez AJ, Needham LL, adjustments in the manner which was performed for Pirkle JL. 2005. Urinary creatinine concentrations in the US Cre adjustment (Vij & Howell, 1998) and future ef- population: Implications for urinary biologic monitoring forts may also improve their performance. measurements. Environ. Health Perspect. 113:192. Levine, L., Fahy, J.P. 1945. Evaluation of urinary lead IMM Table 1. Drinking water total As versus U-As Pearson cor- concentrations. I. The significance of the specific gravity. J. relations across the range of adjustment methods. Correlations Indus. Hyg.Toxicol. 27: 217-223. share a letter when not significantly different from one another Orloff, K., Mistry, K., Metcalf, S. 2009. Biomonitoring for based on Williams’ tests. environmental exposures to arsenic. J. Toxicol. Env. Heal. ______B 12: 509-524. ______Adjustment method rp (95 % CI) Significance_____ R Core Team. 2013. R: A language and environment for Unadjusted 0.65 (0.56, 0.72) a c statistical computing, r foundation for statistical computing, Crea 0.62 (0.53, 0.70) a Vienna, Austria. www.R-project.org. SG 0.68 (0.60, 0.75) b c Venables, W.N.R.B.D. 2002. Modern Applied Statistics with S. Osmolality 0.67 (0.58, 0.74) a b Fourth edition. Springer, New York. ISBN 0-387-95457-0. b ______Cre 0.69 (0.61, 0.75) ______b Vij, H.S., Howell, S. 1998. Improving the specific gravity a Routine method – Equation 1. adjustment method for assessing urinary concentrations of b Modified method – Equation 2, z=0.49. toxic substances. Am. Ind. Hyg. Assoc. J. 59:375-380.

157