Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Retrospective pesticide exposure assessment for studying multiple myeloma risk for farm work in British… Garzia, Nichole A. 2019

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2019_may_garzia_nichole.pdf [ 1.79MB ]
Metadata
JSON: 24-1.0378062.json
JSON-LD: 24-1.0378062-ld.json
RDF/XML (Pretty): 24-1.0378062-rdf.xml
RDF/JSON: 24-1.0378062-rdf.json
Turtle: 24-1.0378062-turtle.txt
N-Triples: 24-1.0378062-rdf-ntriples.txt
Original Record: 24-1.0378062-source.json
Full Text
24-1.0378062-fulltext.txt
Citation
24-1.0378062.ris

Full Text

   Retrospective Pesticide Exposure Assessment for Studying Multiple Myeloma Risk for Farm Work in British Columbia, Canada  by  Nichole A. Garzia  B.A. University of Washington, 2005 MSc, University of British Columbia, 2008   A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY  in  The Faculty of Graduate and Postdoctoral Studies (Population and Public Health)   THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  April 2019  © Nichole A. Garzia, 2019   ii  The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, the dissertation entitled: Retrospective pesticide exposure assessment for studying multiple myeloma risk for farm work in British Columbia, Canada  submitted by Nichole A. Garzia  in partial fulfillment of the requirements for the degree of Doctor of Philosophy In Population and Public Health  Examining Committee: Dr. John Spinelli Supervisor  Dr. Kay Teschke Supervisory Committee Member  Dr. Carolyn Gotay Supervisory Committee Member Dr. Hugh Davies University Examiner Dr. Murray Isman University Examiner  Additional Supervisory Committee Members:  Supervisory Committee Member  Supervisory Committee Member   iii  Abstract  Multiple myeloma (MM) is one of the cancer types that research has shown to be in excess among farm workers. Pesticides have been highly suspected as a risk factor, and much of the research aimed at understanding MM etiology in farm workers has focused on pesticides. However, certain methodological challenges related to the retrospective assessment of pesticide exposure have contributed to weak and at times, inconsistent, epidemiologic evidence.  The aims of this research were to collect and understand the usefulness of exposure data from the literature for future exposure assessment, to develop a method for estimating cumulative pesticide exposure levels in farm workers when combined with self-reported information, and to apply this method to evaluate the relationship between pesticide exposure and MM risk among a sample of farm workers in British Columbia (BC), Canada.  The first study used a systematic approach to collect and evaluate the literature to identify studies that provided quantitative dermal pesticide exposure information on farm workers. The data were extracted and evaluated for usefulness for exposure assessment development. The second study involved the development of a pesticide exposure algorithm, and its application to farm data previously collected as part of an MM case-control study conducted in BC. The results of this algorithm were compared to the results of two other methods, one of which was a previously developed and evaluated algorithm from a well-known prospective cohort of pesticide applicators. The third study was an epidemiologic analysis to evaluate the relationship between pesticide exposure and MM among farm workers identified in the BC case-control study. Pesticide exposure was assessed using dichotomous, simple metrics of exposure as well as the two algorithms from the second study to estimate cumulative pesticide exposure. Overall, the research from this dissertation contributed new knowledge and provided additional evidence to support existing knowledge on the methodological issues surrounding the study of pesticide exposure in relation to MM among farm workers. Additionally, it provided a new pesticide exposure algorithm that can be further evaluated and improved upon regarding its ability to accurately assess exposure among farm workers using self-reported historical exposure information.      iv  Lay Summary  Compared to the general population, multiple myeloma (MM) rates have been shown to be higher in farm workers. Pesticides are a primary exposure of concern. However, studying pesticides in relation to cancer is difficult because farm workers experience multiple different exposures, and many studies rely on detailed information that subjects are asked to recall from the past.  The research focus was to understand how studies identified from the literature for having measured pesticide exposure in farm workers could be used to improve methods applied for epidemiology, where researchers aim to estimate MM risk but do not have measured exposure data for the study population. Using the measured exposure data from the identified studies, a method was developed and used to examine how pesticide exposure is related to MM risk for farm work using a study of people who were versus were not diagnosed with MM in British Columbia (BC), Canada.                 v  Preface  Chapters 2-4 of this dissertation have been prepared as individual manuscripts that have been or will be submitted to peer-reviewed journals for publication. The research in this dissertation has been designed, conducted, analyzed and written in preparation for publication by the candidate, Nichole A. Garzia, and in consultation with the supervisory committee: Drs. John Spinelli, Kay Teschke and Carolyn Gotay.  Below, the contributions of the candidate (NA Garzia) and co-authors have been outlined for each manuscript-based research chapter (2-4). Chapters 3 and 4 involved the secondary use of epidemiologic data that has been approved by the University of British Columbia – British Columbia Cancer Agency Research Ethics Board (Certificate: H10-02876). Chapter 2: A version of this chapter has been published, “Garzia NA, Spinelli JJ, Gotay CC, Teschke K. (2018) Literature review: dermal monitoring data for pesticide exposure assessment of farm workers, Journal of Agromedicine, 23:3, 187-214, doi: 10.1080/1059924X.2018.1448734”. NA Garzia conducted the literature review (i.e., literature search, screening and information/data extraction), analyses and wrote the manuscript.  The initial idea of this literature review came from discussions between NA Garzia and Dr. Kay Teschke, with additional advice from Dr. Patricia Stewart. Dr. Kay Teschke provided guidance on the interpretation of the results. All co-authors provided feedback on multiple revisions of this manuscript prior to publication. Chapter 3: A version has been written with the intent to submit for peer-review publication, “Garzia NA, Spinelli JJ and Teschke K. A comparison of methods for estimating pesticide exposure for farm jobs identified from a multiple myeloma case-control study.” Previously cleaned occupational data were organized and provided by two researchers of the BC Cancer Agency, Agnes Lai and Zenaida Abanto. NA Garzia led the design and development of the exposure assessment algorithm that was based on the data and information collected in the literature review of Chapter 2. Drs. John Spinelli and Kay Teschke provided feedback on the development of the algorithm. NA Garzia further cleaned the data and applied this algorithm, along with two other methods, for the comparative analysis of this research study. NA Garzia wrote the manuscript. Dr. John Spinelli provided statistical guidance and Dr. Kay Teschke provided interpretation guidance. All co-authors provided feedback for revision of the manuscript.  Chapter 4: A version has been written with the intent to submit for peer-review publication, “Garzia NA, Teschke K, Gotay CC, Spinelli JJ. Agricultural pesticide exposure and risk of multiple myeloma in a population-based case-control study in British Columbia, Canada”.  Previously cleaned epidemiologic data were organized and provided by two researchers of the BC Cancer Agency, Agnes Lai and Zenaida Abanto, for the epidemiologic analyses of this study. NA Garzia applied all exposure assessment   vi  methods, conducted the statistical analyses and wrote the manuscript. Dr. John Spinelli provided statistical guidance. Drs. John Spinelli and Kay Teschke provided interpretation guidance that led to an additional analysis that was conducted by NA Garzia, and both provided feedback for revision of the manuscript. Dr. Carolyn Gotay also reviewed the manuscript for comprehension, clarity and presentation.                 vii  Table of Contents  Abstract ..................................................................................................................................................... iii Lay Summary ............................................................................................................................................ iv Preface ........................................................................................................................................................ v Table of Contents .................................................................................................................................... vii List of Tables .............................................................................................................................................. x List of Figures .......................................................................................................................................... xii List of Abbreviations .............................................................................................................................. xiii Acknowledgements ................................................................................................................................ xiv Dedication ............................................................................................................................................... xiv  Chapter 1: Introduction and background ................................................................................................ 1 1.1 Farm Workers .............................................................................................................................. 1 1.2 Multiple Myeloma ......................................................................................................................... 2 1.2.1 Rates and Risk Factors ....................................................................................................... 2 1.2.2 Epidemiologic Overview: pesticide exposure in farm work and multiple myeloma ............. 3 1.3 Characterizing the Elements of the Research Problem .............................................................. 6 1.3.1 Exposure Variation in Farm Work ........................................................................................ 7 1.3.2 Ascertainment of Historical Pesticide Exposure Information ............................................... 8 1.3.3 Exposure Misclassification................................................................................................... 9 1.4 Research Objectives.................................................................................................................. 10  Chapter 2: Literature Review: dermal monitoring data for pesticide exposure assessment of farm workers ..................................................................................................................................................... 12 2.1 Introduction ................................................................................................................................ 12 2.2  Methods ..................................................................................................................................... 13 2.2.1 Literature Search ............................................................................................................... 13 2.2.2  Article Screening and Selection ......................................................................................... 13 2.2.3 Study Detail Collection ....................................................................................................... 15 2.2.4 Summary of Dermal Monitoring Data from Farm Literature .............................................. 15 2.3 Results ....................................................................................................................................... 16 2.3.1 Overview of the Qualifying Farm Literature ....................................................................... 16 2.3.2 Types and Sources of Study Variation .............................................................................. 16   viii  2.3.3 Dermal Exposure Comparisons within Studies: 1. Body Part-Specific Exposure Levels .. 29 2.3.4 Dermal Exposure Comparisons within Studies: 2. Total Body Exposure Levels .............. 44 2.3.5 Additional Work Factors for Potential Dermal Exposure in Farm Workers ....................... 47 2.3.6 Uses of the Existing Dermal Monitoring Data for Occupational Exposure Assessment ... 48 2.4 Discussion ................................................................................................................................. 49 2.5  Conclusion ................................................................................................................................. 52  Chapter 3: A comparison of methods for estimating pesticide exposure for farm jobs identified from a multiple myeloma case-control study ....................................................................................... 53 3.1 Introduction ................................................................................................................................ 53 3.2 Methods ..................................................................................................................................... 54 3.2.1 Multiple Myeloma Case-Control Study .............................................................................. 54 3.2.2 Pesticide Exposure Assessment Methods ........................................................................ 54 3.3  Results ....................................................................................................................................... 59 3.3.1  Pesticide Exposure Assessment ....................................................................................... 63 3.3.2 Comparison of Exposure Results for Algorithm-based Method 1 versus Method 2 ......... 65 3.4 Discussion ................................................................................................................................. 69 3.5 Conclusion ................................................................................................................................. 73  Chapter 4: Agricultural pesticide exposure and risk of multiple myeloma in a population-based case-control study in British Columbia, Canada ................................................................................. 75 4.1 Introduction ................................................................................................................................ 75 4.2 Methods ..................................................................................................................................... 76 4.2.1 Study Population ................................................................................................................ 76 4.2.2 Farm Workers .................................................................................................................... 76 4.2.3 Exposure Assessment ....................................................................................................... 79 4.2.3.1 Cumulative Pesticide Exposure ..................................................................................... 79 4.2.4 Exposure Metrics ............................................................................................................... 81 4.2.5 Statistical Analysis ............................................................................................................. 82 4.3 Results ....................................................................................................................................... 83 4.3.1 Pesticide Exposure Analyses ............................................................................................ 87 4.3.1.1 Dichotomous (Ever/Never) Pesticide Exposure ............................................................ 87 4.3.1.2 Cumulative Pesticide Exposure ..................................................................................... 88 4.4 Discussion ................................................................................................................................. 91   ix  4.4.1 Key Findings ...................................................................................................................... 91 4.4.2 Comparisons of Findings to Other Studies ........................................................................ 92 4.4.3 Limitations and Strengths .................................................................................................. 94 4.5 Conclusion ................................................................................................................................. 95  Chapter 5: Conclusion ............................................................................................................................ 96 5.1 Summary of Findings in Light of the Current State of Knowledge ............................................ 96 5.1.1 Literature Review: dermal monitoring data for pesticide exposure assessment of farm workers……………………………………………………………………………………………………….96 5.1.2 A comparison of methods for estimating pesticide exposure for farm jobs identified from a multiple myeloma case-control study ................................................................................................ 97 5.1.3 Agricultural pesticide exposure and risk of multiple myeloma in a population-based case-control study in British Columbia, Canada ........................................................................................ 99 5.2  Overall Strengths and Limitations ........................................................................................... 101 5.2.1 Strengths ......................................................................................................................... 101 5.2.2 Limitations ........................................................................................................................ 102 5.3 Recommendations for Future Research Directions ................................................................ 103 5.4 Concluding Comments ............................................................................................................ 105  References ............................................................................................................................................. 106  Appendix A ............................................................................................................................................. 113 A.1 Copy of the farm questionnaire ............................................................................................... 113 Appendix B ............................................................................................................................................. 125 B.1 Additional details on how the algorithm was applied for exposure assessment ..................... 125 B.2 Additional details on how the algorithm was developed .......................................................... 126        x  List of Tables  Table 2.1: Study and dermal sampling details of all qualifying literature review farm studies (n=31), by farm job studied. ........................................................................................................................................ 18  Table 2.2: Summary of exposure measure types (dosimeter residue, estimated potential dermal exposure, estimated percent of total dermal exposure) and body part dermal data from all qualifying literature review farm studies (n=31), by farm job studied. ....................................................................... 30  Table 2.3: Within study comparison of total body potential dermal exposure to pesticides by job task (spray vs. mix-load) and application method (tractor mounted airblast spray vs. aerial spray). ............... 45  Table 2.4: Within study comparison of total body potential dermal exposure by farm job task (mix-load-spray vs. spray vs. thinning vs. harvest) and pesticide formulation (dust vs. wettable powder). .............. 46  Table 2.5: Work factors reported in farm studies to potentially influence dermal pesticide exposure, by farm job. ..................................................................................................................................................... 47  Table 2.6: Identifying the uses and non-uses of existing dermal monitoring data from this literature review of farm studies (n=31), and the corresponding implications for future pesticide exposure assessment in occupational epidemiology. ....................................................................................................................... 48  Table 3.1: Agricultural Health Study ‘general’ algorithm scoring system using updated weights from Coble et al. 2011 and the multiple myeloma case-control farm module questions used to assign weights for farm jobs. .............................................................................................................................................. 56  Table 3.2: Pesticide Handling Task Weights for the Potential Dermal Exposure Algorithm ..................... 58  Table 3.3: Task-specific Body Area Exposure Levels (mg/hr) for the Potential Dermal Exposure Algorithm .................................................................................................................................................... 58  Table 3.4: Farm job characteristics by self-reported pesticide use on farm for Groups 1-3. .................... 62  Table 3.5: Comparison of pesticide exposure assessment method characteristics to summarize the advantages and disadvantages of each method. ...................................................................................... 69   xi   Table 4.1: Summary of the imputation methods used for farm job observations with missing data (total n= 59 farm jobs). ............................................................................................................................................. 81  Table 4.2: Demographic characteristics for multiple myeloma case-control study sample (N=773). ....... 84  Table 4.3: Farm and pesticide use characteristics among subjects in the multiple myeloma case-control study (N=773). ........................................................................................................................................... 85  Table 4.4: Odds ratios (OR) and corresponding 95% confidence intervals (CI) for multiple myeloma (MM) in relation to agricultural pesticide exposure as determined by five dichotomous exposure metrics, among subjects in a multiple myeloma case-control study conducted in British Columbia, Canada. ...... 88  Table 4.5: Odds ratios (OR) and corresponding 95% confidence intervals (CI) for multiple myeloma (MM) in relation to agricultural pesticide exposure as determined by two cumulative exposure metrics and pesticide group (any pesticide, herbicides, insecticides) among subjects in a multiple myeloma case-control study conducted in British Columbia, Canada. .............................................................................. 90  Table 4.6: The effect of PPE use as a modifier of the relationship between cumulative potential dermal exposure and multiple myeloma is shown via Odds Ratios (ORs) for ‘PPE not used’ versus ‘when PPE used’, by pesticide group (any pesticide, herbicides, and insecticides) and exposure level group. ......... 91  Table B.1: Study data used for the derivation of the task-specific algorithm weights for Operators, i.e., farm workers who personally handled pesticides to perform both the ‘mix-load’ and ‘apply’ tasks. ....... 128  Table B.2: Study data used to derive the body area-specific weight of the potential dermal exposure algorithm, representing exposure levels (mg/hour) by pesticide handling task (a. mix-load task; b. apply task). ........................................................................................................................................................ 130         xii  List of Figures  Figure 3.1: Multiple Myeloma Case-Control Study: Total Farm Job Counts Eligible for Exposure Assessment based on Pesticide Use Variables. ....................................................................................... 61  Figure 3.2: Distribution of farm job exposure values (n=103) as determined with Method 1, AHS Algorithm. ................................................................................................................................................... 65  Figure 3.3: Distribution of farm job exposure values (n=103) as determined with Method 2, Potential Dermal Exposure Algorithm (mg/hr). ......................................................................................................... 66  Figure 3.4: The relationship between farm job (n=46) exposure values determined using Method 1 (AHS Intensity Algorithm) versus Method 2 (Potential Dermal Exposure Algorithm). ........................................ 67  Figure 4.1: Multiple Myeloma Case-Control Study: Subject and Farm Job Counts Assessed for Cumulative Pesticide Exposure ............................................................................................................... 788                xiii  List of Abbreviations  AHS  Agricultural Health Study BC  British Columbia CI  Confidence Interval DDT  dichloro-diphenyl-trichloroethane MCPA  4-chloro-2-methylphenoxy acetic acid MGUS  monoclonal gammopathy of undetermined significance MM  multiple myeloma OR  Odds Ratio PHED  Pesticide Handlers’ Exposure Database PPE  personal protective equipment RR  Relative Risk U.S.  United States U.S. EPA United States Environmental Protection Agency 2,4-D  2,4-Dichlorophenoxyacetic acid           xiv  Acknowledgements  I would like to extend a deep gratitude to my supervisory committee members, Dr. John Spinelli and Dr. Kay Teschke for their support, mentorship and dedication to my success throughout this dissertation. I have learned so much from them both as it relates to this field and to research in general, and I am very grateful to have had them as mentors. I would also like to thank my committee member, Dr. Carolyn Gotay for her support and always reminding me to think back to the core importance of the research. I would like to acknowledge Agnes Lai and Zenaida Abanto at BC Cancer for their data support and help in general during my time there. Since most of this research was accomplished at a distance from the University, I would also like to acknowledge Emily Van Gulik at the School of Population and Public Health for her extra support in helping me with administrative tasks that I could not easily do from my location. Lastly, I’d like to thank my family for the help they all contributed so that I could do this research. To my husband, especially, who provided me with constant support and stayed up many nights with me as I worked into the early morning hours. Lastly, thank you to the little boy who entered our lives during this time for his cuddles and patience while I finished this work.                xv    For Chris and Teo, I could not have done this without your love and inspiration.                   1 Chapter 1: Introduction and background  1.1 Farm Workers  Farm workers are considered a “unique” population because their morbidity and mortality patterns tend to differ from those of the general population.1 They are often considered to be healthier due to higher levels of physical activity from performing farm tasks, and having lower smoking and drinking rates.1,2 A decreased risk for many cancers,, such as lung and bladder cancer, 3 has been consistently observed among farm workers, however, the opposite (an increased incidence and mortality risk ) has been consistently observed for specific cancer types, such as non-Hodgkin lymphoma, multiple myeloma, and skin, stomach, brain and connective tissue cancer.4 Research has focused on attempting to understand the exposures in the farm environment that contribute to these cancer patterns in farm workers. Pesticides have been one of the most highly suspected environmental risk factors for hematologic malignancies among farm workers.5 Malignancies of this type affect the blood and lymphatic systems where the cancer begins in the bone marrow or cells of the immune system, and include: lymphoma, leukemia and multiple myeloma.  Hypotheses surrounding the carcinogenic potential of certain pesticides have biological plausibility supported by animal toxicology studies,6 as well as knowledge about the “known carcinogenic effect of organic chemicals on hematopoietic cells” from in vitro studies.7  However, researchers evaluating pesticide-cancer relationships are challenged by the fact that farm workers engage in multiple tasks and experience multiple exposures as part of their work,3 and this can create serious methodological challenges for epidemiology. The overall evidence for the effects of pesticide exposure on specific types of cancer has been inconsistent,8 and “until we can fully characterize these pesticide-cancer relationships [related to farm work], we cannot commit to developing cancer initiatives to prevent them”.4 To move research forward effectively, it’s important to understand what these specific challenges are in the field of pesticide-cancer epidemiology and how future studies can better address them. This dissertation focused on pesticide exposure in farm workers and its association to multiple myeloma (MM). Compared to other hematologic cancers, MM has poor prognosis and a low survival rate.8,9  Overall, the etiology of MM is not well understood.10    2 1.2 Multiple Myeloma 1.2.1 Rates and Risk Factors  When the white blood cells of the bone marrow, known as plasma or plasma B cells, begin to divide uncontrollably in the bone marrow, it can lead to the development of MM.11 MM is almost always preceded by a non-cancerous condition known as monoclonal gammopathy of undetermined significance (MGUS), in which cancer has not developed yet, but “abnormal” activity has begun in the plasma cells.11 MM is the third most common lymphohematopoietic cancer following non-Hodgkin lymphoma and leukemia,9  and contributes less than 1% of all new cancer diagnoses globally.12  In Canada, the annual percent change in age-standardized incidence rates for the period of 1998 to 2007 was 0.4% for men and 0.0% (stable) for women.9 The estimated lifetime risk of developing MM is 1 in 131 for Canadian men and 1 in 157 for Canadian women.9  The median age for diagnosis is 62 for men and 61 years old for women.13 Age standardized incidence rates are the highest in Western Europe, North America and Australasia, ranging from approximately 4.6 to 5.8 per 100,000 persons as of 2016.14 Although MM is a rare tumour and treatments have improved since the 1990’s,13,14 the disease is still not curable and it has a poor prognosis and low survival rate.9, 15 Depending on the treatment and overall health and age of the patient, median survival ranges from 3-7 years after diagnosis.13 In Canada, approximately 42% of those diagnosed with this cancer will survive 5 years.11  Several risk factors have been investigated to improve understanding of MM etiology, ranging from demographics, medical history and lifestyle factors to occupational and environmental exposures. Demographic risk factors that are associated with MM include: increasing age, sex (men) and race (black Americans).9,10,12  In fact, based on 2003 to 2007 statistics, the U.S. black population had the highest incidence rate of MM in the world.12 In terms of medical history and lifestyle factors, family history of MM (a genetic component has been related to MM and MGUS), infectious agents (AIDS and the bacterial infections: pneumonia, meningitis and septicemia) as well as excess body weight (possibility related to metabolic hormone functioning) have shown supporting evidence regarding an increased risk of MM.12 However, evidence has been less consistent for individuals with a history of autoimmune conditions and atopic disease.12  There is little to no evidence supporting an increased MM risk with diet, over-the-counter medication, smoking or alcohol consumption. 12 Occupational and environmental risk factors that have shown positive but often weak, non-significant associations with MM, include benzene exposure and certain occupations, namely: agriculture, cleaning workers, machinists and metal processors, carpenter, food and beverage service workers, and bakers.12   3 Inconsistent or little evidence supports exposure to ionizing radiation and chlorinated solvents, specifically trichloroethylene (used in metal degreasing and dry cleaning), as risk factors for MM.12 In terms of farm occupations, multiple hypotheses over the years have been suspected as contributing excess risks of MM, the primary ones being: pesticides,7,16 zoonotic viruses prevalent in chicken farming,7,16 and history of allergies.17  1.2.2 Epidemiologic Overview: pesticide exposure in farm work and multiple myeloma  Early research on farm workers “focused on characterizing cancer patterns”, but in recent years, the research has been focused on “[identifying] factors that might account for the cancer patterns observed”4. With regards to MM specifically, an increased risk in relation to farming has been observed for several years now.4,7,18-24 This association has been consistently reported across the decades and for many different countries.18,22 This association was first reported by Milham (1971)18,19 using a proportional mortality study in which cause of death and occupation was provided on death certifications obtained in Washington and Oregon State, where Milham (1971) reported a “statistically significant association between farming and death” from MM.19  Following the finding by Milham (1971), additional studies in the 1980’s investigated the relationship between farming and MM as well, but at this time did not investigate the specific farm exposures of concern. A case-control study by Gallagher et al. (1983) was conducted in BC, Canada and confirmed an increased risk of MM among subjects with a history of agricultural work that was performed for one year or more.17 Data pertaining to farm work, such as farm type, were not collected in this study.17 In Sweden, a study conducted by McLaughlin et al. (1988) also confirmed the association between farming and MM.25 In this study, employment history data was linked with cancer incidence data for the years 1961-1979, and McLaughlin reported a “significant 20% excess risk for [MM among] subjects who worked in the farming industry”.25 Specific farm exposures were not evaluated by McLaughlin et al. (1998) either, and it was noted that the “exposures contributing to the risk among farmers [were] unclear”.25  In the 1990’s the research began to dig a bit deeper into the relationship between farming and MM, focusing on main hypotheses, such as pesticide exposure in farm workers. For example, a population-based case-control study conducted in Sweden by Eriksson et al. (1992) collected  job history and exposure information using questionnaires to evaluate ‘any pesticide exposure’ and exposure to specific pesticide chemicals and/or classes in relation to MM.18 Eriksson et al. (1992) reported significant increased risks for MM among farmers in general with a Relative Risk (RR) =1.68 (90% CI=1.23-2.33); exposure to ‘any pesticide’ with an RR=1.55 (90% CI=1.15-2.26); exposure to the pesticide chemical   4 class of phenoxyacetic herbicides with an RR=1.15 (90% CI=1.15-4.66); and exposure to the organochlorine insecticide called dichloro-diphenyl-trichloroethane, commonly known as DDT, with an RR=1.75 (90% CI=1.19-2.64).18 Eriksson et al. (1992) confirmed previous findings regarding the association between farming and MM, and provided new information about potential pesticide exposures that may be linked with the increased risk of MM in farm workers.18 Nanni et al. (1998) conducted a case-control study in Italy, also using questionnaires for the collection of exposure information through in-person interviews.20 The excess risk of MM among farmers was again confirmed, but the association was not statistically significant with an Odds Ratio (OR) =1.31 (95% CI=0.62-2.74).20 Analyses focused on specific pesticide classes identified an excess risk for chlorinated insecticide exposure only in farm workers with an OR=1.6 (95% CI=1.1-2.4).20  Two meta-analyses that considered pesticide exposure in farm workers were performed between the late 1990’s and early 2000’s. The first was conducted by Khuder and Mutgi (1997) and included 32 studies (approximately 60% were case-control, and the remaining were mortality or cohort studies)from multiple locations (United States, Canada, Italy, Sweden, Denmark, New Zealand, Australia, France, Ireland, England and Wales) that were published between 1981 and 1996 to evaluate the relationship between farming and MM.7 Among the 32 studies of this meta-analysis, only three of them reported a negative association between farming and MM, while one study found no association.7 Three separate meta-analyses were performed as part of this study; the first included all 32 studies and reported an RR estimator of 1.23 (95% CI=1.14-1.32) between farming and MM.7 The second meta-analysis included data for female farmers from nine studies and resulted in an RR estimator of 1.23 (95% CI=1.17-1.29), and the third analysis included five studies that represented farmers in central United States and resulted in an RR estimator of 1.38 (95% CI=1.27-1.51).7 Specific farm exposures were not evaluated as part of this meta-analysis, as few studies had begun to investigate beyond the farm-MM relationship at this time. Approximately 10 years later, Perrotta et al. (2008) published a meta-analysis based on 28 case-control studies published since 1970.22 Perrotta et al. (2008) covered a wider publication range, but there was an overlap of 11 case-control studies that were included in this meta-analysis and the above discussed meta-analysis by Khuder and Mutgi (1997); neither meta-analysis reported the exposure years that were covered by the studies. Perrotta et al. (2008) reported that exposure was most commonly determined from detailed occupational history information and job-specific questionnaires, while a few studies used expert judgment from industrial hygienists to assess exposure based on occupational history information.22 Exposure to pesticides was most commonly based on exposure to any pesticide as an entire group, and a few studies evaluated more specific pesticide exposures such as herbicides, chlorophenols and DDT.22 Perrotta et al. (2008) estimated a pooled OR =1.39 (95% CI= 1.18-1.65) for farming and MM; a pooled OR = 1.47 (95%CI =1.11-1.94) for general pesticide exposure and MM; a pooled OR =1.69 (95% CI= 1.01-1.83) for herbicide exposure and MM, and a pooled OR=2.19 (95% CI=1.30-2.95) for DDT exposure and MM.22 Additionally, for farmers who worked more than ten years on a farm  in relation to MM, a pooled OR of 1.87 was estimated (95% CI=1.15-3.16).22 In summary,   5 Perrotta et al. (2008) reported consistent evidence supporting a relationship between farm work and MM from studies conducted in several countries during the last several decades.22 Regarding pesticide exposure, there is evidence of an excess MM risk for ‘ever exposed to any pesticides’, and for specific pesticide chemicals/chemical classes, the following were most commonly linked with MM: DDT, chlorophenols and phenoxyacetic herbicides.22  The Agricultural Health Study (AHS), a well-known prospective cohort study of pesticide applicators and their spouses in Iowa and North Carolina began in 1993 to study cancer and other health outcomes among pesticide applicators.26 Several analyses have been performed on this study over the years for multiple health effects. Based on the AHS analyses, possible associations between MM and these specific herbicides have been identified and require further evaluation: alachlor, atrazine, and glyphosate.4  A case-control study of men in multiple Canadian provinces, diagnosed between 1991 and 1994,27 was performed by Kachuri et al. 2013 to evaluate the relationship between lifetime use of multiple pesticides and MM risk.24 As described by Kachuri et al. (2013), it is common for farmers to be exposed to multiple pesticides during their lifetime because “multiple pesticides could be used simultaneously or during the same growing season”, but exposure to multiple pesticides had not been explored yet.24 The exposure information used was originally obtained using a postal questionnaire and telephone interview for those who reported at least 10 hours per year of pesticide use.24 This study was not limited to farm workers, but approximately half of the cases and controls reported having ever lived or worked on a farm.24 Pesticides were grouped based on their type, chemical class and carcinogenic potential.24 Two metrics for multiple pesticide exposure were applied for analyses, ‘number of pesticides used’ and ‘days per year of pesticide use’, and exposure to some individual pesticides were also examined.  Excess MM risk was observed for: men who reported use of at least one carbamate pesticide (OR=1.94, 95% CI=1.05-4.66), at least one phenoxy herbicide (OR=1.56, 95% CI=1.09-2.25), and three or more organochlorine insecticides (OR=2.21, 95% CI=1.05-4.66).24 Significant excess risks of MM were observed for carbaryl exposure (OR=2.71, 95% CI=1.47-5.00) and captan exposure (OR=2.96, 95% CI-1.40-6.24).24 In regards to multiple pesticide exposure using ‘days per year of pesticide use’, mecoprop used for more than 2 days/year was significantly associated with an excess MM risk (OR=2.15, 95% CI=1.03-4.48). This study identified some commonly used pesticides that may be contributing to excess risks of MM and was the first to evaluate multiple pesticide exposure, but exposure was not limited to farm work.24 More recently, Presutti et al. (2016) conducted a pooled analysis using data from three case-control studies from the United States and Canada, where exposure information was collected by questionnaire. Three different metrics were used to assess exposure to specific pesticide chemicals: ever/never use, duration of use, and cumulative lifetime days.23 This study also included, but was not limited to, analyses of farm workers only. The main finding was a significant excess risks of MM in relation to the ‘ever use’ of   6 carbaryl, captan and DDT pesticides.23 Although some specific pesticides were identified as possible risk factors for MM, Presutti et al. (2016) reported that “the role of pesticide in the etiology of MM remains unclear”.23 The Canadian case-control data included in this pooled study was the same data used in the Kachuri et al. (2013) study mentioned above. Therefore, there was overlap in the evidence regarding specific pesticides found to be associated with an increase MM risk (carbaryl, captan); these findings were based on the same metric (ever/never use). As the above evidence exhibits, farm work has been consistently associated with MM, but the evidence regarding pesticide exposure has been less consistent. The specific pesticides associated with MM risk have varied, although phenoxy herbicides and organochlorine insecticides (especially DDT) have been commonly reported.22 They are or have also been among some of the most commonly used pesticides.  What is most evident from the above research is that most of the evidence comes from retrospective studies, and much of the exposure information in these studies is based on self-reports to questionnaires. Understanding the role that retrospective study designs and questionnaire data contribute to the current state of evidence can help provide insight with regards to improving future research. In addition, many of the associations for farming and pesticide exposure in relation to MM were either non-significant excess risks or marginally significant excess risks. This may be due to the fact that it was common for all farm workers to be considered as the ‘exposed’ group, rather than focusing on specific farm jobs or incorporating other farm variables that can influence exposure. Less commonly have “attempts been made in epidemiologic studies to quantify internal dose” and characterize the factors that influence exposure.28 Moving away from the use of simple surrogates and incorporating exposure determinants to assist with estimating exposure intensity will help reduce exposure misclassification. Exposure assessment based on these simple surrogates alone is likely to introduce non-differential error that pushes risk measures toward the ‘null’ of no association.29  1.3 Characterizing the Elements of the Research Problem  To improve consistency of the human evidence regarding pesticide-cancer relationships related to work on a farm, it’s important to better understand the main methodological challenges of this research field, and how they are related to each other. ‘Exposure variation’ and the ‘ascertainment of historical pesticide exposure information’ are serious challenges in this research field, both of which have the ability to increase the magnitude of exposure misclassification in an epidemiologic study. Exposure misclassification is a type of bias that can have profound effects on the resulting measures of association.   7 1.3.1 Exposure Variation in Farm Work  Farming is one of the most diverse industries regarding the range of possible exposure scenarios;30 there is a “wide range” of possible exposures and “they can vary considerably from farm to farm depending upon the type of farm operation”.4 For example, the different types of exposure hazards that farm workers may experience during work are extensive, from chemical (pesticides, fertilizers, solvents), biological (zoonotic viruses, mycotoxins), and physical hazards (heat, noise, vibration, solar radiation) to engine fuels and exhaust, among others.4 Each type of exposure hazard varies within itself as well. For example, each pesticide is its own unique mixture of chemicals that varies in terms of its mode of action, physical, chemical and toxicological properties.31   Second, there is a significant amount of exposure variation related to the farm jobs performed, which means that exposure assessment based on farm job alone is rarely adequate.32 Within farm jobs, there are several factors that contribute to potential exposure variation, such as the type of farm operation (e.g., field crop farm versus animal farm), the types of equipment used, and the frequency of performing certain tasks, not to mention that some farm workers are only seasonal.30 Understanding the farm jobs performed can be a good way to identify workers who may be exposed to pesticides due to personally handling them (e.g., applicators) compared to those who may be exposed more indirectly, such as through treated crops during re-entry field work (e.g., crop harvesters).  Lastly, the routes of exposure that contribute most significantly to internal dose, and how that dose can vary based on individual factors such as with personal protective equipment (PPE) use, engineering controls, and work practices and behaviours, is also important for characterizing exposure variation in farm workers. Pesticide exposure can occur by many routes (dermal, inhalation and ingestion), but the dermal route has been identified as a significant contributor of the internal dosedose, particularly for pesticide handlers.33,34  There are many types of engineering controls (e.g., enclosed cabs on tractors) and personal protective equipment that can be used to try to mitigate exposure levels, but whether or not they are used and/or accessible can vary greatly from one worker to another. This provides only a brief overview of the extent of exposure variation in farm work. Having access to reliable information that will allow researchers to characterize a modest extent of this variation is a separate challenge.      8 1.3.2 Ascertainment of Historical Pesticide Exposure Information  Case-control studies are the most feasible and commonly used study design for rare outcomes such as MM. However, the existing exposure heterogeneity among farm workers becomes additionally complicated by the fact that exposure information is often obtained historically, usually from self-reports to questionnaires,28,35 as noted above. Questionnaires typically collect limited occupational information that may only include “job title, type of employer, and the days the jobs were held, supplemented sometimes by work activities”.35 Even when specific pesticide exposure is assessed, the intensity of exposure is not usually quantified.28 This limited extent of historical occupational data makes it challenging to accurately assess exposure to pesticides. Alternative sources of data, such as quantitative measures (e.g., biological samples or pesticide residue measurements) can be useful for quantifying exposure intensity in case-controls studies, but they are not reliable for studying outcomes with a long latency because many commonly used pesticides have very short half-lives in the body. For example, the half-life of herbicide 2,4-D in urine samples is between 12 and 72 hours.36  Since retrospective studies are often reliant on self-reported information about pesticide exposure, it is essential for these studies to incorporate questions whose responses have been shown to be valid and reliable to obtain the necessary individual-level exposure details. Stewart et al. (1998) made suggestions for improving questionnaires for this exact purpose, by suggesting the use of job-specific questionnaires that begin with generic occupational information and appropriately migrate into specific questionnaires (e.g., task-based) based on responses to generic questions.35 Furthermore, Stewart et al. (1998) also suggested the use of standardized questionnaires that can be used across multiple studies in order to improve comparability of data across studies.29 Nieuwenhiujsen (2005) provided a detailed overview of important considerations regarding questionnaire development, from questionnaire length to types of questions and wording used in the questionnaires, as well as administration of the questionnaires for obtaining “good answers” to necessary questions needed for epidemiologic studies.37 In essence, questionnaires can be a powerful and reasonable way to obtain valuable information about exposure from a large number of subjects, however, more effort needs to be placed on the development and administration of questionnaires because wording, and even the types of categories defined as responses to a question, can influence the quality of responses received, if any at all.37 Although improvements to questionnaires can improve the quality of information obtained from study subjects, so as to more accurately assess exposure to pesticides, some challenges still remain. For example, job-specific questionnaires may involve more detailed questions about the job practices and tasks performed, but the ability to recall detailed historical information by subjects can be problematic. For example, farm workers being asked to recall specific pesticide information (e.g., the chemical or trade names of products used on the farm), may not be able to remember these details depending on their knowledge of this information, the length of time their farm job was held, the type of farm job and   9 tasks they performed, and perhaps the duration of time that has passed since the job was held. Responses to detailed questions like these may favour subjects who have a reason to recall this information more readily (e.g., cases), subjects who worked in farm jobs that involved purchasing pesticides, or who worked in the farm job for a much longer duration of time or more recently. As a result, recall differences such as these could lead to either under-reporting and/or reporting bias.37 Blair et al. (2002) conducted repeat interviews with more than 4,000 pesticide applicators from the AHS to evaluate the reliability of reporting pesticide use information, as well as other types of information.38 These interviews were one year apart and the information reported was tested for agreement, the following pesticide-related variables tested high for agreement: ever/never use of specific pesticides (reporting reliability did not differ by type of pesticide chemical or class of chemicals) and application practices.38 Agreement was lower for these pesticide-related variables: duration, frequency, or decade of first use for specific pesticides.38 Blair et al. (2002) acknowledged that agreement regarding specific pesticides “decreased as the amount of detail sought increased, such as number of years a person applied specific pesticides instead of ever/never use of specific pesticides”.38  As Stewart et al. (1998) and Nieuwenhuijsen (2005) have acknowledged more planning needs to be done at the questionnaire development and administration phase of studies that rely on self-reported information for exposure assessment. As Nieuwenhuijsen (2005) states the importance of this concept very clearly, “asking a question is one thing, getting a good answer is another”.37 Additionally, many farm workers are migrant and questionnaires need to be translated to reflect the languages of the study sample, so that information is not lost based on language barriers.   1.3.3 Exposure Misclassification  The two matters discussed above (‘exposure variation’ and ‘ascertainment of historical pesticide exposure information’) have the potential to contribute to an increased level of exposure misclassification in epidemiologic studies. In general, “exposure misclassification probably occurs in nearly every study”; however, the bigger problem is in understanding the potential direction of effect and magnitude of error resulting from misclassification because even small amounts of misclassification can have substantial effects on risk estimates and “lead to an interpretation of no effect”.29  When it comes to case-control studies, where exposure information is obtained after the disease has been diagnosed, then misclassification of exposure is likely to be differential (different for cases versus controls).29  Differential exposure misclassification can bias the epidemiologic risk estimates either away or toward the “null” of no association.29  In contrast, non-differential exposure misclassification biases the risk estimates toward the “null” hypothesis of ‘no association’.29  The goal is to minimize exposure   10 misclassification as much as possible by developing more precise exposure assessment methods to obtain more reliable measures of association.26  Kromhout and Heederik (2005) suggest incorporating exposure determinants into epidemiologic studies of farm workers as essential for improving exposure assessment methods, given the extent of exposure variation that exists for this occupation.30 Exposure determinants are the factors related to exposure.39 In terms of farm work, these factors may include, but are not limited to: jobs, tasks, farm type, specific pesticides used, size and location of farm, equipment used, engineering controls used, and personal protective equipment used. Different farm jobs are likely to have different exposure determinants; for example, ‘application method’ may only be an exposure determinant for farm workers who personally apply pesticides but not for crop harvesters. Once exposure determinants are identified and characterized for different farm jobs, then “knowledge of these determinants can be used in questionnaires for retrospective exposure assessment”.30 This brings us back to Stewart et al. (1998) suggestion about job-specific questionnaires,35 as this is the only way to ask relevant job-related questions for each subject. When it comes to epidemiologic studies of pesticide-cancer relationships, the lack of accurate exposure assessments has been “a major weakness”.28 Moving forward, it’s important to understand and avoid the limitations of the current literature, so as to begin building a more solid and consistent body of human evidence regarding pesticide exposure and MM related to farm work.  1.4 Research Objectives  In light of the research challenges presented above and the current state of knowledge regarding the relationship between pesticide exposure and MM related to farm work, the theme of this research was to focus on providing reliable estimates of pesticide exposure among farm workers in order to evaluate the epidemiologic association between pesticide exposure and MM. Throughout this dissertation, the individual goals of each chapter embodied the importance of the exposure assessment process. Specifically, this dissertation included the collection and review of pesticide exposure data (Chapter 2), the development and application (Chapter 3) of a pesticide exposure assessment algorithm, and an epidemiologic analysis (Chapter 4) of the relationship between pesticide exposure and MM.  The specific objectives were to:  1. Conduct a systematic review to determine what types of information the existing dermal pesticide monitoring data can provide for future pesticide exposure assessment in occupational epidemiology;   11 2. Compare farm job exposure estimates between three different exposure assessment methods, using farm job data from a population-based case-control study of Multiple Myeloma; and  3. Evaluate the relationship between agricultural pesticide exposure and MM, using multiple exposure assessment methods and data from a population-based case-control study of multiple myeloma.               12 Chapter 2: Literature Review: dermal monitoring data for pesticide exposure assessment of farm workers  2.1 Introduction  The goal of pesticide exposure assessment for occupational epidemiology is to accurately classify a workers’ exposure, but when direct exposure measurements of subjects are not feasible, this can be a challenging task.40 The majority of epidemiological studies that have evaluated pesticide-response relationships are population-based retrospective case-control studies and exposure assessment is often reliant on the information collected as part of the study (e.g., job histories, self-reported exposure information).41,42 This presents a challenge for pesticide exposure assessment because jobs may not be indicative of exposure, and self-reported exposure may not be reliable and can be prone to recall bias.42 As a result, pesticide exposure assessments for retrospective case-control studies need to be improved. There are some important elements to consider, such as the exposure route that contributes most significantly to internal dose and investigation of exposure determinants to assist in more accurate exposure assessment.42,43 Regarding pesticides, dermal exposure has been established as a primary route that significantly contributes to internal dose among workers, particularly pesticide handlers who mix, load and/or apply pesticides.33,34 Since the development of the passive dosimetry patch technique of Durham and Wolfe (1962),44 a relatively easy and inexpensive method for monitoring individual body part exposure, a number of studies have measured dermal exposure in workers handling pesticides.45 However, it’s not common for dermal monitoring data to be used beyond these individual studies, and still, “few epidemiologic studies have assessed dermal exposures”.43 Ideally, existing dermal monitoring data could be combined across studies to develop “more sophisticated exposure assessments” for epidemiology that aim to quantify exposure rather than assign exposure using simple surrogates (e.g. job title alone).32   Regarding determinants of pesticide exposure, such information “can be quantitatively associated with dermal exposure and might be used to group workers into high- and low- exposure categories”.42 Few epidemiological studies have done this, but understanding exposure determinants can strengthen the “validity of exposures assessed by questionnaires”41 by helping investigators identify the workers truly exposed compared to those who are not exposed or experience very low levels of exposure.46  The use of simple surrogates (e.g., job title alone) that designate every farm worker as “exposed” is not only incorrect, but impedes a researcher’s ability to identify a true association.32,47 Farming is a broad industry that includes husbandry of many animal species and growing many crop types. The use of pesticides or not varies a great deal between farm types.  Even where pesticides are used, the types of pesticides used and the application methods vary. Each job consists of different tasks and other characteristics that   13 can influence how the worker may be exposed to a pesticide, e.g., an applicator spraying pesticide using a backpack sprayer or a field worker picking berries two hours after a pesticide was applied. Beyond the work environment, there are also many differences between individual pesticides; each has its own set of physical, chemical and biological properties that can influence the amount of dermal uptake at contact, along with its own toxicity profile if absorption occurs.33  If a pesticide is absorbed into the skin, the rate of absorption depends on the body part with which contact occurs.33,48 There are many factors to consider when assessing a worker’s exposure to pesticides; therefore, a solid understanding of the most influential factors and how they are related to dermal pesticide exposure is important for accurate exposure assessment. Given the importance of the dermal route and the need for more information regarding work factors that can influence pesticide exposure, the aim of this study was to conduct a systematic review to determine what types of information the existing dermal pesticide monitoring data could provide for future pesticide exposure assessment in occupational epidemiology.  2.2  Methods 2.2.1 Literature Search  A literature search was performed using eight online databases: Medline (OvidSP), AGRICOLA, EMBASE, Web of Science, BIOSIS, PubMed, CAB Direct and NIOSHTIC-2. The search included literature published before January 1, 2015; except BIOSIS, which included literature published through May 1, 2011 (note: BIOSIS results were similar to PubMed and Web of Science Core Collections, so the earlier search end date unlikely affected the literature identified). The search terms (i.e., MeSH/keywords/search words) included these descriptors and their synonyms: “dermal” (synonym “skin”), “pesticide”, “occupation” (synonym “work”), and “exposure”. Each search term created a separate search string connected with ‘AND’ operators; synonyms were connected by the ‘OR’ operator. All identified articles were imported into RefWorks®, an online research management tool.   2.2.2  Article Screening and Selection  To determine study eligibility, two screening phases were performed. During phase 1, titles and abstracts (full articles when necessary) were evaluated for basic criteria:  Article written in English;   14  Study location in U.S., Canada, Austrailia or Western Europe – to limit type of farming and climates covered (if not clear, location of first author was used);  Occupational study (lab simulated and quasi-experimental studies not included);  Pesticides were studied;  Study was original source of the data;  Dermal exposure was evaluated;   Study population was 18 years or older; and  Study was published before January 1, 2015. In phase 2, articles that fulfilled all basic criteria were screened for the following:   Study must have dermal data;   Dermal data must include “over” clothing or on uncovered skin measurements, and   50% or more of the body parts were measured for dermal exposure. To determine if “50% or more of the body parts” were measured, the body was classified into eight areas (below) based on typical parts measured in the literature. Measurement of four or more of the body areas meant this criterion was met; it ensured each study measured a sufficient number of body parts for exposure comparisons.  1. Face, head, neck and/or shoulders; 2. Upper arms (both sides); 3. Chest, front trunk or abdomen; 4. Back trunk;  5. Forearms and/or wrists (both sides); 6. Hands (both sides); 7. Upper legs (both sides); and 8. Lower legs and/or feet (both sides). Due to variation in work clothing and personal protective equipment (PPE) used between workers and studies, only dermal measurements taken over clothing or on uncovered skin were included in this study.  Many studies also measured under worker clothing, however, these measurements reflect dermal exposure as a result of penetration that varies based on the type of clothing/PPE. Since only over clothing or on uncovered skin dermal measurements were included all dermal data and exposure references throughout this review represent “potential” dermal exposure.      15  2.2.3 Study Detail Collection  Once all qualifying studies were identified, each was assigned to an industry group as determined by the occupational information provided, thus grouping studies that evaluated similar “workplaces”. The following information was recorded from all qualifying studies into a Microsoft Excel® Spreadsheet, by assigned industry group:  Author;  Title;  Study location (country);  Industry studied;  Jobs studied and corresponding work tasks (including methods and equipment used);  Farm type (agriculture only)  Number and sex of study subjects;  Dermal sampling method for body;  Dermal sampling method for hands;  All body parts measured; and  Pesticide(s) studied. Based on the reported job titles and task information from qualifying studies, a standardized job list was created for each industry. Job title standardization was necessary because job titles were used interchangeably between studies (e.g., “farmer”, “operator” or “applicator”) and did not always correspond to the same job tasks across studies. Task-defined job groups allowed exposure information to be more appropriately compared.   2.2.4 Summary of Dermal Monitoring Data from Farm Literature  Due to variation observed when recording study details for all industries, this review only focused on studies classified under the Farm Industry, which had the most diverse job coverage and greatest number of studies overall. For all qualifying Farm Industry studies, potential dermal exposure data were recorded by body part and for total body (when provided) in a Microsoft Excel® Spreadsheet.  Dermal monitoring data were categorized by type of measure: 1) dosimeter residue (concentration measured on sampling media), 2) estimated dermal exposure (dosimeter residue of a body part/length of   16 time exposed multiplied by corresponding body part surface area),49 or 3) percent of total body exposure ([estimated body part-specific dermal exposure/estimated total body exposure]*100). These data were examined to determine what useful information they could provide for pesticide exposure assessment.   2.3 Results 2.3.1 Overview of the Qualifying Farm Literature  The literature search resulted in 2,923 unique studies. Approximately 7% (206 studies) met Phase 1 screening criteria; 2,717 studies were excluded (location, language, non-occupational, simulated study, dermal exposure not measured, etc.). Approximately 40% (81 studies) met Phase 2 screening criteria; 125 studies were excluded (no dermal data, measurements not taken over clothing, lack of details regarding body parts measured, too few body parts measured). The 81 studies were classified by Industry: Farming (n=31), Greenhouse (n=28), Residential/Commercial Pest Control Services (n=8), Nursery (n=5), Forestry (n=3), Lawn Care (n=3), Pest Container Recycling/Formulation (n=2), Aircraft Pest Control (n=1). Although Farm and Greenhouse Industries are agricultural, these industries were distinguished for this review because their work settings are very different. Greenhouses are enclosed climate controlled environments; this influences how work is performed, which may influence how and the extent to which dermal pesticide exposure occurs.  The 31 qualifying farm studies were summarized by task-defined farm job. Many of them evaluated dermal exposure for more than one farm job (43 farm jobs studied among the 31 farm studies): Operators (mix, load and apply pesticides) were evaluated in 15 studies, Applicators (apply pesticides only) in 12 studies, Mixer-loaders (mix and load pesticides only) in 9 studies, Field Workers (perform any combination of re-entry tasks, e.g., harvest, thin, crop scout to identify pests, etc.) in 6 studies, and Flaggers (stand in field to mark flight path for aerial applications) in 1 study. Two studies presented dermal data as an aggregated value across body parts (hands provided separately), although they measured a sufficient number of body parts as required by the screening criteria.50,51   2.3.2 Types and Sources of Study Variation  A significant amount of variation was observed between studies, and it was classified into two types: 1) variation in study focus and reporting, and 2) variation in exposure levels. The variation in study focus and reporting posed the most difficulty for this review because it indicated that the dermal data from each   17 study represented different work scenarios and types of quantitative exposure, even for the same farm jobs. Therefore, it was determined that dermal data could not be reliably compared between studies. Sources of variation in study focus included: pesticide type, farm type, application/work method, equipment used, and body parts measured. Study specific details are presented in Table 2.1 by farm job.   Several different pesticide active ingredients were evaluated and contributed to the variation in focus.  Operators were studied in relation to 21 active ingredients (7 herbicides, 3 fungicides, 11 insecticides) across 15 studies.50-64 Applicators were studied in relation to 13 active ingredients (3 herbicides, 2 fungicides, 6 insecticides, 2 miticides) across 12 studies.56,65-75 Mixer-loaders were studied in relation to 11 active ingredients (2 herbicides, 1 fungicide, 6 insecticides, 2 miticides) across 9 studies.63,65-68,70,72,76,77  Field workers were studied in relation to 6 active ingredients (1 herbicide, 2 fungicides, 3 insecticides) across 6 studies.49,53,56,57,77,78  In the single flagger study, 1 herbicide was studied.79  It was common for a study to evaluate exposure to more than one active ingredient.  Many farm types were evaluated, especially for operators, from fruit orchards and vineyards to several field crops and one animal farm.  In the study of flaggers,79 the crop was cotton. The remaining three job groups (applicators, mixer-loaders and field workers) were mostly studied in fruit orchards, vineyards and field crops. The most common farm types overall included grains (wheat, corn),51,55,59,61,69,78 citrus,65,67,71,72,75 grape vineyards,56,74 and cotton.68,76,79 Farm type information was not provided for 3 studies.62,70,77  Application/work methods and types of equipment also contributed to variation in study focus. These are important exposure factors because they influence how a worker may be exposed to pesticides. The application method (along with other variables, e.g., crop type) determined the spray equipment that was used, which could influence where and how much workers were dermally exposed. For operators and applicators, tractor-mounted or -drawn sprayers were the most commonly used spray equipment, which varied significantly in design (open or enclosed cab, open or closed windows, air conditioning or air filtration or neither, sprayer mounted in front, behind or at an elevated level, and high pressure, low pressure or airblast spray).51-57,60-63,65-67,70-72,75  Backpack spray was used in 5 studies of operators and applicators,50,62,69,73,74 aerial application was used in 1 applicator study,68 pour-on method (for animal farming) was used in 1 operator study,50 and manual seed treatment in a drill box was used in 1 operator study.59  For mixer-loaders, there were two methods for mixing pesticide formulation, a conventional open-pour system or a closed-system; the former was most common and used in all mixer-loader studies. One study evaluated both open- and closed-systems.70 For field workers, work method concerned how they were exposed to crops/foliage as they performed tasks: 5 studies evaluated hand harvesting;49,53,56,57,77 1 study evaluated crop scouting;78 2 studies involved thinning;56,57 and 1 study evaluated field workers in a variety of tasks (bending branches, summer pruning, sorting and transporting).57 Work methods for flaggers did not vary.79   18 Table 2.1: Study and dermal sampling details of all qualifying literature review farm studies (n=31), by farm job studied.  Author (Pub Year)  Study Title  Study Location  Pesticide Active Ingredients Studied (functional group)  Farm Type  Application/ Work Methods and/or Equipment   Dermal Sampling Method  Number of Workers Studied  Sex of Workers  Body Parts Sampled (“over” clothing samples only)  Body  Hands  OPERATOR STUDIES (n=15)  Winterlin W. et al. (1984)53 Worker Re-entry Studies for Captan Applied to Strawberries in California  California, USA Captan (fungicide) Strawberry Tractor mounted boom spray Patch Sampling Gloves (latex) 1 Operator Not reported Chest, Back, Forearms, Thighs, Shins, Hands Senior P. et al. (1992)54 Determination of potential dermal exposure during application of crop protection products by boom spraying  United Kingdom Not specified (insecticide) Spring Oilseed Rape Tractor mounted boom spray (cab vs. no cab) Patch    Sampling Gloves 3 Operators Not reported Head, Upper Chest, Abdomen, Hands, Arms, Thighs, Lower Legs Lonsway J. et al. (1997)52 Dermal and respiratory exposure of mixers/sprayers to acephate, methamidophos, and endosulfan during tobacco production  Kentucky, USA Acephate (insecticide), Endosulfan (insecticide), Metamidophos (insecticide) Tobacco Tractor mounted boom spray and Open air highboy crop sprayer Patch Sampling Gloves 5 Operators Not reported Face V- and Back of Neck,  Chest, Back, Upper Arms, Forearms, Hands, Thighs, Lower Legs Lebailly P, et al (2009)55 Exposure to pesticides in open-field farming in France France Isoproturon (herbicide) Wheat, barley Rear mounted sprayer, Trailer Sprayer Whole Body Dosimetry Sampling Gloves (cotton) 39 Operators Male Chest, Back, Upper Arms, Forearms, Hands, Thighs, Lower Legs  Winterlin W. et al. (1986)56 Worker Reentry into Captan-treated Grape Fields in California, USA Captan (fungicide) Grapes Elevated Spray Boom (operated from inside cab; Patch Sampling Gloves (cotton) 5 Operators Not reported Chest, Back, Forearms, Hands, Thighs, Shins   19  Author (Pub Year)  Study Title  Study Location  Pesticide Active Ingredients Studied (functional group)  Farm Type  Application/ Work Methods and/or Equipment   Dermal Sampling Method  Number of Workers Studied  Sex of Workers  Body Parts Sampled (“over” clothing samples only)  Body  Hands California  boom reaches over and to the sides of vines)  Stewart P. et al. (1999)50 Exposure of farmers to phosmet, a swine insecticide Iowa, USA Phosmet (insecticide) Animal - swine Backpack sprayer, low and high pressure sprayers, pour-on method Patch Sampling Gloves 10 Operators Not reported Head (if hat worn), Chest, Back,  Lower Arm, Hands (if no gloves, n=5), Upper Inner/Outer, Upper, Lower Legs Cock, J. (1998)57 Exposure to captan in fruit growing Netherlands Captan (fungicide) Fruit Orchards (mostly apples) Airblast sprayer pulled by tractor (cabin vs. no cabin) Patch Hand Rinse 94 Operators Male Forehead, V-/Back of Neck, Back, Forearm (left side if right handed), wrists Vitali M. et al. (2009)51 Operative modalities and exposure to pesticides during open field treatments among a group of agricultural subcontractors Italy Dicamba (herbicide), Alachlor (herbicide), Terbuthylazine (herbicide), Azinphos-methyl (insecticide), dimethoate (insecticide) Sugar Beets, Maize Trailer mounted spray tank with booms at rear Patch Hand Wash 10 Operators Male Head, Back of Neck, Upper Chest, Back, Upper Right Arm, Left Forearm, Hands (if gloves not worn, n=6), Upper right Leg, Left Calf, Right Foot  Vercruysse F. et al. (1999)58 Exposure assessment of professional pesticide users during treatment of potato fields Belgium Chlorothalonil (fungicide), Fluazinam (fungicide), Primicarb (insecticide)  Potato Fields Tractor mounted hydraulic boom sprayer (closed cabin with carbon filter vs. half-open cab)  Patch Sampling Gloves (cotton) 4 exposure scenarios (unknown number of workers) Not reported Head, Back/Front of Neck, Back, Chest/Stomach, Upper and Lower Arms, Hands, Thighs, Shins    20  Author (Pub Year)  Study Title  Study Location  Pesticide Active Ingredients Studied (functional group)  Farm Type  Application/ Work Methods and/or Equipment   Dermal Sampling Method  Number of Workers Studied  Sex of Workers  Body Parts Sampled (“over” clothing samples only)  Body  Hands Fenske R. et  al. (1990)59 Worker exposure and protective clothing performance during manual seed treatment with Lindane South Dakota, USA Lindane (insecticide) Winter Wheat Manual Treatment - drill box (half full with seed) mixed with pesticide using a stick Patch Hand Wash; exclude, protective nitrile gloves worn 4 Operators Male Chest, Back, Forearms, Upper Arms, Upper Legs, Lower Legs Dubelman S. et al. (1982)64 Operator exposure measurements during application of herbicide Diallate North Dakota, USA Diallate (herbicide) Sugar Beets Mix-Load: conventional vs. closed-system; Application: tractor mounted 30-ft spray boom (cab vs. no cab) Patch Sampling Gloves (cotton), protective gloves worn over top during closed-system tank fills only. Don’t know (27 exposure trials) Not reported Head, Forehead, Shoulder, Chest, Back (arms, thighs and ankles for closed-system tank filling) Yeung, P. et al. (1998)60 Exposure of air blast applicators to ethyl parathion and methyl parathion in orchards: A comparison of Australian conditions to overseas predictive exposure models  Victoria, Australia Ethyl Parathion (insecticide), Methyl Parathion (insecticide) Fruit Orchards Tractor Mounted High Pressure Sprayers and Electrostatic Sprayers (air conditioned cabin vs. open tractor) Patch Sampling Gloves (cotton) – some workers wore PVC gloves over sampling gloves  16 Operators Not reported Head, Chest, Back, Shoulders, Upper Arms, Forearms, Hands (if gloves not worn), Thighs, Shins Protano C. et al. (2009)61 Performance of different work clothing types for reducing skin exposure to pesticides during open field treatment Italy Azinphos-methyl (insecticide), Terbutylazine (herbicide), Alachlor (herbicide), Dimethoate (insecticide), Dicamba (herbicide) Sugar Beet, Maize Tractor with trailer-mounted spray tanks and booms at rear Patch Not measured 10 Operators Not reported Head, Back of Neck, Upper Chest, Shoulders, Upper Right Arm, Left Forearm, Right Thigh, Left Calf, Instep     21  Author (Pub Year)  Study Title  Study Location  Pesticide Active Ingredients Studied (functional group)  Farm Type  Application/ Work Methods and/or Equipment   Dermal Sampling Method  Number of Workers Studied  Sex of Workers  Body Parts Sampled (“over” clothing samples only)  Body  Hands Abbott I.M. et al. (1987)62 Worker exposure to a herbicide applied with ground sprayers in the United Kingdom United Kingdom 2,4-D (herbicide) Don’t know Tractor-drawn and mounted hydraulic sprayers;  Tractor-mounted controlled droplet applicator; Knapsack sprayer with boom and single lance  Whole Body/ Sample Clothing Sampling Gloves 30 Trials (6 for each of the 5 equipment types) Not reported Head, Body Front, Body Back, Hands, Arms, Legs, Feet Chester G. et al (1986)63 Biological monitoring of a herbicide applied through backpack and vehicle sprayers Canada Fluazifop-butyl (herbicide) Flax, Potatoes, Sunflowers Tractor mounted sprayers: tractor-drawn field sprayer, “spra-coupe”, and “floater”  Whole Body (Sampling Overalls) Exclude, PVC protective gloves worn during mix-load  10 Operators Not reported Front Trunk, Back Trunk, Arms, Upper Legs, Lower Legs  APPLICATORS (n=12 studies)  Nigg H.N. et al. (1986)65 Dicofol exposure to Florida citrus applicators: effects of protective clothing Florida, USA Dicofol (miticide) Citrus groves Airblast sprayer pulled by canopied tractors Patch  Hand Rinse 2 Applicators Not reported Chest, Back, Shoulders, Upper Arms, Forearms (when PPE not worn on hot days), Hands, Thighs, Shins Wojeck G.A., et al. (1982)66 Worker exposure to arsenic in Florida grapefruit spray operations Florida, USA Lead Arsenate (insecticide) Grapefruit Tractor drawn airblast sprayer (under canopy – open cab vs. cab enclosed with wire mesh)  Patch Sampling Gloves (cotton) 3 Applicators Not reported Chest, Back, Shoulders, Forearms, Hands, Thighs   22  Author (Pub Year)  Study Title  Study Location  Pesticide Active Ingredients Studied (functional group)  Farm Type  Application/ Work Methods and/or Equipment   Dermal Sampling Method  Number of Workers Studied  Sex of Workers  Body Parts Sampled (“over” clothing samples only)  Body  Hands Wojeck G.A., et al. (1983)75 Worker exposure to paraquat and diquat Florida, USA Paraquat (herbicide) Tomatoes and Citrus Tractor mounted boom spray for tomatoes (cab vs. no cab); Tractor drawn shielded boom spray for citrus (open, no canopy)  Patch Hand Rinse 13 Applicators Male Chest, Back, Shoulders, Forearms, Hands, Thighs, Shins Nigg H.N., et al. (1983)67 Exposure of spray applicators and mixer-loaders to chlorobenzilate miticide in Florida citrus groves  Florida, USA Chlorobenzilate (miticide) Citrus Groves Airblast sprayer pulled by canopied tractor Patch  Glove Wash (excluded because gloves worn)  4 Applicators Not reported Chest, Back, Shoulders, Wrists, Shins (excluded forearms and shins due to patch location) Chester G., et al. (1987)68 Worker exposure to, and absorption of, cypermethrin during aerial application of an ‘ultra low volume’ formulation to cotton  Mississippi, USA Cypermethrin (insecticide) Cotton Aerial application Whole Body Dosimetry/Sampling Garments  Sampling Gloves 2 (Pilot) Applicators Not reported Head, Front Trunk, Upper Arms, Forearms, Hands, Above Knees, Below Knees Lengerich S.K., et al. (1989)69    Near real-time monitoring of potential dermal exposure during backpack herbicide spraying USA A.I. not specified (herbicide) Corn, Soy Backpack Sprayer with a high pressure hose to spray boom (6 nozzles) attached to the spray tank     Water Sensitive Paper Strips (turn blue when exposed to moisture) None 1 Applicator Male Chest, Waist, and the right side and full circumference (due to boom location) of: forearm, mid-thigh, mid-calf, and ankle.    23  Author (Pub Year)  Study Title  Study Location  Pesticide Active Ingredients Studied (functional group)  Farm Type  Application/ Work Methods and/or Equipment   Dermal Sampling Method  Number of Workers Studied  Sex of Workers  Body Parts Sampled (“over” clothing samples only)  Body  Hands Sanderson W., et al. (1995)70 Exposure of commercial pesticide applicators to the herbicide alachlor Ohio, USA (based on author) Alachlor (herbicide) Don’t know Four or wide-wheeled flotation vehicles mounted with long spray boom (enclosed cabs)  Patch Glove Wash; exclude, results not separated by Job  20 Applicators Male Head, Chest, Back, Upper Arms, Thighs Carman G., et al. (1982)71 Pesticide applicator exposure to insecticides during treatment of citrus trees with oscillating boom and airblast units California, USA Parathion (insecticide), Dimethoate (insecticide) Citrus Groves Spray Rig,  oscillating airblast boom (open tractor vs. cab open windows; cab closed windows vs. cab closed windows/ filtered air) Gauze Sponges None 2 Applicators  Not reported Chest, Back, Shoulders, Upper Arms, Lower Arms, Upper Legs Wojeck G.A., et al. (1981)72 Worker exposure to ethion in Florida citrus Florida, USA Ethion (insecticide) Citrus Groves Tractor-drawn airblast sprayer (open canopy)   Patch Sampling Gloves 9 Applicators Male Head/Neck, Chest, Back, Arms, Hands, Legs and Feet  Simpson G., et al. (1965)73 Exposure to parathion: dermal and inhalation exposure to parathion while spraying tomato bushes with a knap-sack mister  New South Wales, Australia Parathion (insecticide), Demeton methyl (insecticide) Tomatoes Knap-sack sprayer Filter Papers (similar to Patch Technique) None 1 Applicator (12 trials) Not reported Head, Chest, Back, Shoulders, Wrist, Forearm Tsakirakis A.N., et al. (2014)74 Dermal and inhalation exposure of operators during fungicide application in vineyards: evaluation of Viotia, Greece Penconazole (fungicide) Grapevines/ Vineyards Hand-held single nozzle spray gun connected to tractor tank Whole Body Dosimetry (Coveralls: 50/50 cotton/polyester vs. 100% cotton)  Sampling Gloves 5 Applicators Not reported “jacket” (upper body), “pants” (legs), “cap” (head), “gloves” (hands)   24  Author (Pub Year)  Study Title  Study Location  Pesticide Active Ingredients Studied (functional group)  Farm Type  Application/ Work Methods and/or Equipment   Dermal Sampling Method  Number of Workers Studied  Sex of Workers  Body Parts Sampled (“over” clothing samples only)  Body  Hands coverall performance 2014  Winterlin W., et al. (1986)56   Worker re-entry into Captan-treated grape fields in California California, USA Captan (fungicide) Grapes Spray rig with elevated spray boom (stretching above and to sides of vines), closed cab  Patch Sampling Gloves (cotton) 2 Applicators Not reported Chest, Back, Forearms, Hands, Thighs, Shins  MIXER-LOADER STUDIES (n=9)  Nigg H.N., et al. (1986)65 Dicofol exposure to Florida citrus applicators: effects of protective clothing  Florida, USA Dicofol (miticide) Citrus Groves Mix and Load Patch Hand Wash 2 Mixer-loaders Not reported Chest, Back, Shoulders, Upper Arms, Hands, Thighs, Shins Wojeck G.A., et al. (1982)66 Worker exposure to arsenic in Florida grapefruit spray operations Florida, USA Lead Arsenate (insecticide) Grapefruit Trees Mix and Load Patch  Sampling Gloves (cotton); exclude, protective gloves worn and results not provided by Job 2 Mixer-loaders Male Chest, Back, Shoulders, Wrists (also exclude shins because patches placed beneath rubber apron) Nigg H.N., et al. (1983)67 Exposure of spray applicators and mixer-loaders to chlorobenzilate miticide in Florida citrus groves Florida, USA Chlorobenzilate (miticide) Citrus Groves Mix and Load Patch Glove Wash 2 Mixer-loaders Not reported Chest, Back, Shoulders, Forearms, Wrists (thigh and shin pads excluded because placed underneath rubber apron)   25  Author (Pub Year)  Study Title  Study Location  Pesticide Active Ingredients Studied (functional group)  Farm Type  Application/ Work Methods and/or Equipment   Dermal Sampling Method  Number of Workers Studied  Sex of Workers  Body Parts Sampled (“over” clothing samples only)  Body  Hands Chester G., et al. (1987)68 Worker exposure to, and absorption of, cypermethrin during aerial application of an ‘ultra low volume’ formulation to cotton  Mississippi, USA Cypermethrin (insecticide) Cotton Mix and Load Whole Body Dosimetry (sampling overalls, hoods, socks) Sampling Gloves  2 Mixer-loaders Not reported Head, Back Trunk, Upper Arms, Forearms, “Boots”  Sanderson W., et al. (1995)70 Exposure of commercial pesticide applicators to the herbicide alachlor Ohio, USA (based on author) Alachlor (herbicide) Don’t know Drive trucks mounted with large tanks, transport, mix and load (closed system or open-pour system)  Patch Glove wash; exclude, results not separated by Job.  7 Mixer-loaders Male Head (“cap”), Chest, Back, Upper Arms, Thighs Wojeck G.A., et al. (1981)72 Worker exposure to ethion in Florida citrus Florida, USA Ethion (insecticide) Citrus Groves Mix and Load (pump spray mixture through hose from supply tank to spray tank), and drive supply truck to field treatment area  Patch Sampling Gloves 8 Mixer-loaders Male Head/Neck, Chest, Back, Upper Arms, Thighs Kiefer M., et al. (1996)76 Health Hazard Evaluation Report 95-0248-2562, Dirty Bird Inc. Grady, Arkansas    Arkansas, USA Methyl Parathion (insecticide), Acephate (insecticide), Profenofos (insecticide) Rice, Cotton Mix and Load Patch Sampling Gloves (excluded because protective gloves worn) 2 Mixer-loaders Not reported Chest, Stomach, Forearms, Thighs, Shins Everhart L., et al. (1982)77 Potential benlate fungicide exposure during Florida, USA Benlate (fungicide) Bush beans, Pole beans, Mix and Load for Aerial Application  Patch Sampling Gloves (cotton) 10 Mixer-loaders Not reported Face, Chest, Back of Neck, Forearm,   26  Author (Pub Year)  Study Title  Study Location  Pesticide Active Ingredients Studied (functional group)  Farm Type  Application/ Work Methods and/or Equipment   Dermal Sampling Method  Number of Workers Studied  Sex of Workers  Body Parts Sampled (“over” clothing samples only)  Body  Hands mixer-loader operations, crop harvest, and home use  Celery, Avocado  Hands Chester G., et al. (1986)63 Biological monitoring of a herbicide applied through backpack and vehicle sprayers  Western Canada Fluazifop-butyl (herbicide) Flax, Potatoes, Sunflowers Mix and Load Sampling Clothing  Sampling Gloves (cotton) 3 Mixer-loaders Not reported Head, “Body”, Upper arms, Lower arms, Hands, Legs, Feet  FIELD WORKER STUDIES (n=6)  Herman N.D., et al. (1985)49 Hand harvester exposure to maleic hydrazide (MH) in flue-cured tobacco  North Carolina, USA Maleic Hydrazide (herbicide) Tobacco Hand Harvest Patch Hand Rinse 4 Field workers Not reported Back, Stomach (Left and Right), Forearms, Hands, Thighs Kamble S., et al. (1992)78 Field worker exposure to selected insecticides applied to corn via center-pivot irrigation Nebraska USA Chlorpyrifos (insecticide), Carbaryl (insecticide), Permethrin (insecticide)  Corn Crop Scouting (check for insect activity) Patch Sampling Gloves (cotton) 3 Field workers Not reported Head, Face V-/Back of neck, Front/Back Trunk, Forearms, Upper Arms, Hands, Lower Legs, Thighs  Winterlin W., et al. (1984)53 Worker re-entry studies for Captan applied to strawberries in California  California, USA Captan (fungicide) Strawberries Hand Harvest Patch Sampling Gloves (latex) 12 Field workers Not reported Chest, Back, Forearms, Hands (residue only), Thighs, Shins  Everhart L.P., et al. (1982)77 Potential benlate fungicide exposure during mixer-loader operations, crop California, USA (for harvester sample) Benlate (fungicide) Strawberries Hand Harvest Patch Sampling Gloves (cotton) 3 Field workers Female Face, Chest, Forearms, Hands, Upper Thighs, Lower Legs   27  Author (Pub Year)  Study Title  Study Location  Pesticide Active Ingredients Studied (functional group)  Farm Type  Application/ Work Methods and/or Equipment   Dermal Sampling Method  Number of Workers Studied  Sex of Workers  Body Parts Sampled (“over” clothing samples only)  Body  Hands harvest, and home use  Winterlin W., et  al. (1986)56 Worker reentry into Captan-treated grape fields in California  California, USA Captan (fungicide) Grapes Harvest and/or Thinning Patch Sampling Gloves (cotton) 44 Field workers Not reported Chest, Back, Forearms, Hands, Thighs, Shins Cock J., et al. (1998)57 Exposure to Captan in fruit growing Netherlands Captan (fungicide) Fruit Orchards (mostly apple) Bending Branches, Thinning Fruit, Summer Pruning, Harvesting, Sorting and Transporting  Patch Hand Rinse 94 Field workers Not reported Forehead, Sternal Area, Forearm (left side for right handed persons), Wrists, Hands  FLAGGER STUDY (n=1)  Chester G., et al. (1984)79 Occupational exposure and drift hazard during aerial application of paraquat to cotton California, USA Paraquat (herbicide) Cotton Flag for aerial application Patch Sampling Gloves (cotton) 2 Flaggers Female Forehead, Neck, Abdomen, Shoulders, Forearm (left), Hands, Thigh (one side), Lower Leg (one side)        28  The variation in pesticides, farm types, and application/work methods studied might be useful if the quantitative exposure reporting between studies was standard, but it was not. Sources of variation in reporting included: type of exposure measure (dosimeter residue, estimated dermal exposure, percent of total body exposure), the units accompanying these measures, and how they were presented in the studies also varied (i.e., exposure time interval represented, stratification by other work variables). The extent of reporting variation can be illustrated with the operator studies (n=15 studies). Considering the types of exposure measures, eight operator studies (50%) provided ‘estimated dermal exposure’ measures,51,53,54,56,57,59-61 three studies provided ‘percent of total dermal exposure’ measures,55,62,63 and four studies provided ‘dosimeter residue’ measures.50,52,58,64,75 The problem of varying types of exposure measures is that they don’t represent the same aspect of exposure; i.e., ‘dosimeter residue’ measures have not been extrapolated to the corresponding body part surface area and only represent the residue found on the sampling medium of the respective body part (unlike ‘estimated dermal exposure’ and ‘percent of total dermal exposure’). Among studies that did extrapolate dermal samples to the entire measured body part, the surface areas used also varied between studies; there are “many models of skin surface areas for body regions”60 available (e.g. 1100 cm2 versus 1700 cm2 surface areas that have been used for the head)48,60. The total body size that each model was based upon (e.g. average adult sized male) and how the body parts were broken down and/or combined have contributed to the differences between surface area models that have been used to calculate body part specific dermal exposure. Additionally, a range of different units accompanied the data. Studies that provided ‘estimated dermal exposure’ measures used these units: mg/person, mg of Active Ingredient/Operation, mg/hour, mg/person/hour, mg/m2/hour, mg.  Studies that provided ‘dosimeter residue’ measures used these units: µg/hour, µg/cm2hour, µg/kg of Active Ingredient, and µg/cm2.  Data presentation also contributed to variation in reporting. Across operator studies,  exposure data represented different durations of exposure: hourly,50,52,54,56,57,60,62,75 8-hour work day,53 the average duration (minutes) it took workers to complete the task,51,54,55,59,61,64 and for other studies it was unclear.58,63 Lastly, the number of variables by which the dermal data were stratified, also varied. Operator studies stratified the data by one or more of these variables: operator task (most common),52,54,55,62-64 pesticide active ingredient,51,58,75 type of equipment,62,75 pesticide formulation,56 type of operation,56 mix-load, and application methods.64  Study-specific information pertaining to these sources of variation is shown in Tables 2.1 and 2.2. Operators were used as an example to highlight the sources and extent of reporting variation across studies. Standardization in reporting could greatly improve the possibility of combining and comparing dermal exposure data.     29 2.3.3 Dermal Exposure Comparisons within Studies: 1. Body Part-Specific Exposure Levels  Since comparing dermal measurement data across studies was not possible, the data were examined within studies to identify potential exposure patterns by body part, farm job and other work factors. We then checked whether there were similar patterns in other studies. Within each study, exposure levels were compared across all measured body parts to determine the body part with the highest measured exposure and to see if results were similar for other studies of the same farm job (Table 2.3). More specifically, the body part with the highest average potential dermal exposure relative to the other body parts measured in the same study was identified (noted in Table 2.2); if the data were stratified, they were presented as a range across the stratification variable in Table 2.2. When the data were stratified by multiple variables, the average exposure level was determined for each body part and for each stratification variable to determine that which had the highest average potential dermal exposure value overall, even if the data were not presented by all stratification variables in Table 2.2.                30 Table 2.2: Summary of exposure measure types (dosimeter residue, estimated potential dermal exposure, estimated percent of total dermal exposure) and body part dermal data from all qualifying literature review farm studies (n=31), by farm job studied.   Study    Exposure Measure Type and Details  Units  Upper Body  Arms  Legs  Hands    Face/ Head, Neck, Shoulders  Chest/ Front Trunk  Back  Upper Arm  Forearm /Wrist  Thigh  Lower Leg/ Feet  OPERATOR STUDIES (n=15)  Winterlin W. et al. (1984)53 Dosimeter Residue, 2-hour exposure period  µg/cm2 -- 1.599 1.468 -- 2.214 4.423 1.827 1.513 Estimated Dermal Exposure, 8-hour work day  mg person-1 -- 9.824 9.016 8.000 (“Sleeve”) 61.14a 18.94 --  Senior P. et al. (1992)54 Estimated Dermal Exposure, Range of Mean Exposure for Mix-Load Task  mg A.I./ operation <0.01-0.77 <0.01 <0.01 <0.01-0.30 (Right), 0.03-0.15 (Left) <0.01-0.04 (Right), <0.01 (Left) <0.01-0.13 (Right), <0.01-0.13 (Left) 0.02-0.7 (Right) 0.25-1.75a (Left)  Estimated Dermal Exposure, Range of Mean Exposure for Spray Task  mg A.I./hr  0.03-0.06 (Head) <0.01 <0.01-0.03 (Right), 0.03-0.15 (Left) 0.03-0.39 (Right), <0.01-0.13 (Left)  <0.01-0.27 (Right), <0.01-0.01 (Left)  <0.01-0.02 (Right) <0.01-0.01 (Left) <0.01 mg (Right) <0.01-0.01 (Left) Estimated Dermal Exposure, Range of Mean  Exposure for Washing Equipment Task  mg A.I./ Operation <0.01-0.58 (Head)   <0.01 <0.01 0.02-0.15 (Right), <0.01-0.05 (Left)   <0.01-0.06 (Right), <0.01-0.03 (Left)  <0.01 (Right), <0.01-0.06 (Left) <0.01-0.02 (Right), <0.01-0.02 (Left) Lonsway J. et al. (1997)52 Dosimeter Residue, Range of Mean Residues for Mix-Load Task and 3 Insecticides  mg/hr 0.0-0.1 0.2-0.5 0.1-0.3 0.1-0.2 0.1-0.7 0.3-1.1 0.1-0.7 1.1-133.5a Dosimeter Residue, Range of Mean mg/hr 0.3-25.4 (Face V-0.3-18.6 0.0-0.5 0.1-0.4 0.1-1.6 0.7-2.4 0.1-1.0 0.0-40.1a    31  Study    Exposure Measure Type and Details  Units  Upper Body  Arms  Legs  Hands    Face/ Head, Neck, Shoulders  Chest/ Front Trunk  Back  Upper Arm  Forearm /Wrist  Thigh  Lower Leg/ Feet Residues for Spray Task and 3 Insecticides Neck) 0.0-12.9 (Back of Neck)  Lebailly P., et al. (2009)55 Dosimeter Residue, Mean Residue for Mix-Load Task and 2 Types of Spray Equipment (Rear-mount and Trailer Spray)   mg -- 6.2 1.2 1.6 12.3 3.5 7.0 46 % Total Dermal Exposure, Mean % for Mix-Load Task, and 2 Types of Spray Equipment (Rear-mount and Trailer Spray)  % -- 7 2 2 14 3 8 64a Dosimeter Residue, Mean Residue for Spray Task and  2 Types of Spray Equipment (Rear-mount and Trailer Spray)  mg -- 2.5 1.0 2.1 2.8 0.9 2.8 24.6 % Total Dermal Exposure, Mean % for Spray Task and  2 Types of Spray Equipment (Rear-mount and Trailer Spray)  % -- 8 5 7 10 3 10 57a Winterlin W., et al. (1986)56 Dosimeter Residues,  Mean Residue for Dust Formulation/ Thinning Operation   µg/cm2 -- 6.66 4.36 -- 50.00 15.30 3.45 41.77   32  Study    Exposure Measure Type and Details  Units  Upper Body  Arms  Legs  Hands    Face/ Head, Neck, Shoulders  Chest/ Front Trunk  Back  Upper Arm  Forearm /Wrist  Thigh  Lower Leg/ Feet Estimated Dermal Exposure,  Dust Formulation/ Thinning Operation  mg/person/hr  -- 4.13 3.00 -- 19.03 44.65a  5.78 31.60  Dosimeter Residues, Mean Residue for Wettable Powder/Thinning Operation  µg/cm2 -- 0.55 0.17 -- 2.40 2.08 0.83 20.34 Estimated Dermal Exposure, Wettable Powder/Thinning Operation  mg/person/hr  -- 1.29 0.29 -- 1.93 19.25a 3.63 8.33 Dosimeter Residues, Mean Residue for Wettable Powder/ Harvest Operation  µg/cm2 -- 0.76 1.29 -- 1.19 2.09a 0.79 1.51 Stewart P., et al. (1999)50 Dosimeter Residues, Mean Residue for 4 Application Methods (backpack, high pressure spray, low pressure spray, pour-on)  µg/hr -- 23.0 – 502.0 (Range of mean residues combined for body parts: chest, back, lower arm, thigh, lower legs) 2904.0a (no gloves) Cock J., et al. (1998)57 Estimated Dermal Exposure, Mean across all Operators mg/m2/hr 6.2 (Forehead), 2.1 (Neck) -- -- -- 22.0a (Lower Arm),  10.30 (wrists) -- -- -- Vitali M., et al. (2009)51 Estimated Dermal Exposure Range across all Operators (n=5) who did not wear gloves µg 39.44 – 646.76 (Range of mean estimated exposure for combined body parts: head, back of neck, upper chest, back, upper right arm, left forearm) 1700.00 – 4600.00a Vercruysse F., et al. (1999)58  Dosimeter Residue Range for Application task only,4 Trials and µg kg-1 A.I. 0.78-63.98 (Head & Neck) 1.39-130.18 (Chest/ Stomach) 1.39-181.14 1.04-38.9 0.39-13.25 5.49-150.96   3.11-48.65 3.90-225.87a   33  Study    Exposure Measure Type and Details  Units  Upper Body  Arms  Legs  Hands    Face/ Head, Neck, Shoulders  Chest/ Front Trunk  Back  Upper Arm  Forearm /Wrist  Thigh  Lower Leg/ Feet 3 pesticides    Fenske R., et al. (1990)59  Dosimeter Residue, Mean Residue  µg/cm2 (%) -- 0.9 0.70 1.52 14.67 8.89  0.56 -- Estimated Dermal Exposure  (% Total Dermal Exposure), Mean Values  mg -- 3.21 (5.1) 2.48 (3.9) 4.43 (7.0) 17.75  (28.1) 33.96a  (53.8) 1.34 (2.1) -- Dubelman S., et al. (1982)64 Dosimeter Residue, Mean for Conventional Tank Fill-Mix Across Replications  µg/cm2 0.15 (Head), 0.67 (Forehead), 0.06 (Shoulders) 0.19 0.07 No Sample Take (Other Body = average for thigh, forearm, bicep, ankle) 71.20a Dosimeter Residue, Range of Mean Residues for Closed Tank Fill-Mix for 3 Operations  µg/cm2 <0.01 (Head), <0.01 (Forehead), <0.01 (Shoulders)  <0.005 <0.005 0.005 – 0.013 (Other Body = average for thigh, forearm, bicep, ankle) -- Dosimeter Residue, Range of Mean Residues for 3 Application Methods (boom, harrow, disc)  µg/cm2  0.03-0.06 (Head), 0.05-0.13 (Forehead), 0.06-0.16 (Shoulders)  0.04 – 0.20 0.03 – 0.16 No Sample Take (Other Body = average for thigh, forearm, bicep, ankle) 0.11 – 0.60a   34  Study    Exposure Measure Type and Details  Units  Upper Body  Arms  Legs  Hands    Face/ Head, Neck, Shoulders  Chest/ Front Trunk  Back  Upper Arm  Forearm /Wrist  Thigh  Lower Leg/ Feet Yeung P., et al. (1998)60 Estimated Dermal Exposure, Mean Values µg/hr 197 (Head & Neck), 159 (Shoulders) 1126 (Chest) 2303 (Hips) 117 150 295 5533 1497 (Lower Legs) 707 (Feet) 80,354a (did not wear PVC gloves n=5) Protano C., et al. (2009)61 Estimated Dermal Exposure, Range for All Pesticides and Workers (not wear gloves, n=6)      µg -- 5.0-223.7a 2.5-60.4 2.9-104.8 1.6-9.7 1.6-62.3 1.7-29.3 (Left Calf) 0.0-39.3 (Instep) -- Abbott I.M., et al. (1987)62 % of Total Dermal Exposure, Range Across Operators for Mix-Load Task (Tractor Drawn/ Mounted Sprayer)    % -- 0.3-3.3 (“Body” front) 0.1-0.3 (“Body” back) 1.1-5.0 (Right), 0.8-2.9 (Left) 0.2-2.9 (Right), 0.4-1.7 (Left) 0.1-9.5 (Right Leg), 0.4-1.1 (Left Leg), 0.1-0.2 (Right Foot), 0.1-0.3 (Left Foot)  39.4-62.9a  (Right),  27.0-49.9a  (Left) % of Total Dermal Exposure, Range Across Operators for Mix-Load Task (Knapsack Sprayer) % -- 3.5-3.6 (“Body” front) 1.9-2.0 (“Body” back) 1.2 (Right Arm), 0.7-2.6 (Left Arm) 2.2-2.4 (Right), 0.04-2.8 (Left) 1.5-3.3 (Right Leg), 2.9-3.1 (Left Leg), 0-1.9 (Right Foot), 0.2-0.7 (Left Foot)  46.3-50.9a  (Right),   31.0-34.3a  (Left) % of Total Dermal Exposure, Range % 0.7-1.1 (Head) 1.3-3.4 (“Body” 0.6-2.0 (“Body” 1.6-4.1 (Right), 1.0-3.3 (Right) 0.6-7.7 (Right Leg), 30.7-35.0a  (Right),    35  Study    Exposure Measure Type and Details  Units  Upper Body  Arms  Legs  Hands    Face/ Head, Neck, Shoulders  Chest/ Front Trunk  Back  Upper Arm  Forearm /Wrist  Thigh  Lower Leg/ Feet Across Operators for Spray Task with Tractor Mounted/ Drawn Sprayer front) back) 1.3-8.8 (Left) 2.0-2.9 (Left) 1.1-7.0 (Left Leg), 0.4-6.8 (Right Foot), 0.1-4.6 (Left Foot)  27.6-51.8a  (Left) % of Total Dermal Exposure, Range Across Operators for Spray Task with Knapsack Sprayer      % 0.1-1.8 (Head) 0.6-1.8 (“Body” front) 1.7-2.1 (“Body” back) 0.3-0.5 (Right), 0.4-0.5 (Left) 2.7-3.4 (Right), 3.9-4.2 (Left) 27.3-42.4 (Right Leg), 24.3-39.1 (Left Leg), 0.9-1.5 (Right Foot), 1.3-1.4 (Left Foot)  2.4-10.3 (Right), 2.0-23.0 (Left) Chester G., et al. (1986)63  % of Total Dermal Exposure, Mean Exposure for Mix-Load and Spray Tasks  % 1 (Head) 33a (“Body”) 2 7 29 (“Leg”), 1  (Foot) --  APPLICATOR STUDIES (n=12)  Nigg H.N., et al. (1986)65  % of Total Dermal Exposure, Mean for 2 Applicators % 3 12 5 7 9 42a 22 -- Wojeck, G.A., et al. (1982)66 Estimated Dermal Exposure, Mean for All Applicators mg arsenic/hr 1.0 (Head & Neck)  3.8 2.5 10.0 (Arms) 23.6 (legs & feet) 28.3a % of Total Dermal Exposure,  Mean for All Applicators  % 1.0 (Head & Neck) 5 4 14 (Arms) 34 (legs & feet) 41a Wojeck, G.A. et al. Dosimeter Residue, Range of Mean µg/cm2hr 0.00-2.96 (Shoulders) 0.00-10.49 0.00-9.37 -- 5.15-8.03 0.00-9.87 0.00-8.25 1.96-46.30a   36  Study    Exposure Measure Type and Details  Units  Upper Body  Arms  Legs  Hands    Face/ Head, Neck, Shoulders  Chest/ Front Trunk  Back  Upper Arm  Forearm /Wrist  Thigh  Lower Leg/ Feet (1983)75 Residues for Boom Application (3 Tractor Types) of Paraquat to Tomatoes Dosimter Residue, Range of Mean Residues for Boom Application (2 locations) of Paraquat to Citrus  µg/cm2 hr 0.01-1.04 (Shoulders) 0.08-0.13 0.04-0.11 -- 0.46-1.03 0.52-1.03 0.34-0.43 8.28-15.48a Nigg H.N., et al. (1983)67  Dosimeter Residue, Mean Residue For All Applicators  µg/cm2hr 1.68 (Shoulders) 1.13 0.47 -- 3.82a (Wrists) -- 1.46  -- Chester G., et al. (1987)68 Dosimeter Residue, Range for 2 Pilot Applicators (6 Trials per Applicator) µg/sample <0.50-1.12 (“Hood”) <1.00-3.22 <1.00-121 <0.50-13.6 (Right), <0.50-0.62 (Left) <0.50-5.84 (Right), <0.50-22.4 (Left) <1.00-3.44 (Right), <1.00-6.30 (Left) <1.00-1.34 (Right Leg), <1.00-5.80 (Left Leg) <0.05-10.7 (Right Sock),  <0.05-16.9 (Left Sock) <1.00-101a (Right), 5.07-58.5a (Left)   Lengerich S.K., et al. (1989)69  % of Total Dermal Exposure, Mean for Low Boom % -- 0.7 (Waist), 0.5 (Chest) -- -- 1.1 2.9 80.4a (ankle), 14.4a (lower leg) -- % of Total Dermal Exposure Mean for High Boom % -- 1.1 (Waist), 0.7 (Chest)  -- -- 0.9 6.6 74.9a (ankle), 15.8a (lower leg)  -- Sanderson W., et al. (1995)70  Dosimeter Residue, Spray Boom in Front (n=1)  µg/cm2 -- 1.24 1.36 2.84 (Arms) 29.87a -- -- Dosimeter Residue, Geometric Mean for Spray Boom at Rear (n=19, except for “Cap” where n=10)  µg/cm2 0.11 (“Cap”) 0.20 0.13 0.42 (Arms) 0.85a -- --   37  Study    Exposure Measure Type and Details  Units  Upper Body  Arms  Legs  Hands    Face/ Head, Neck, Shoulders  Chest/ Front Trunk  Back  Upper Arm  Forearm /Wrist  Thigh  Lower Leg/ Feet Carman G., et al. (1982)71  Dosimeter Residueb ,  Range for Oscillating Boom and Open Tractor (Parathion) µg/cm2/hr 4.1-8.9 (Left Shoulder), 2.3-3.4 (Right Shoulder)  1.0-1.3 1.7-3.9 3.4-12 (Left), 1.2 (right) 5.0 -11 (Left), 2.0-4.7 (Right) 2.1-18 (Left), 2.8-2.9 (Right)  -- -- Dosimeter Residueb Range for Oscillating Boom and Open Windows (Parathion)  µg/cm2/hr 2.8 (Left Shoulder), <0.01-0.16 (Right Shoulder)  0.32-0.71 0.14 8.8  (Left), 1.2  (Right) 3.4-10 (Left), 0.04-0.71 (Right)  0.9-7.2 (Left), 0.57-1.9 (Right)  -- -- Dosimeter Residueb, Airblast Spray for Open Tractor (Parathion, Wettable Powder)  µg/cm2/hr 10 (Left Shoulder), 7.9 (Right Shoulder)  4.0 2.3 9.5 (Left), 9.5 (Right) 8.7 (Left), 7.9 (Right) 9.5 (Left), 6.3 (Right) -- -- Dosimeter Residueb, Airblast Spray for Open Tractor (Emulsifier Concentrate)  µg/cm2/hr 10 (Left Shoulder), 7.1 (Right Shoulder) 14 (Left, Front) 3.4 6.1 (Left), 9.6 (Right) 17 (Left), 6.4 (Right) 11 (Left), 6.3 (Right) -- -- Dosimeter Residueb, Airblast Spray for Tractor with Cab and Open Windows (Dimethoate Emulsifier)   µg/cm2hr 18 (Left Shoulder), 0.52 (Right Shoulder)   2.1 1.8 23 (Left), 0.29 (Right) 11 (Left), 2.9 (Right) 4.6 (Left), 1.2 (Right) -- -- Wojeck G.A., et al. (1981)72  Estimated Dermal Exposure, Range of Means for 2 Locations mg ethion/hr 2.4-2701.6 10.4-375.8 4.1-14.3 7.8-29.6 (Arms) 81.8-109.4 (legs and feet) 180.7-7,239.3a Simpson G., et al. (1965)73 Dosimeter Residue,  Mean of 2 Weather Conditions (calm, windy) µg/100 cm2/hr 319 (Shoulder), 350 (“hat”, head) 421 494 261 (Arm) 300 (wrists) -- -- --   38  Study    Exposure Measure Type and Details  Units  Upper Body  Arms  Legs  Hands    Face/ Head, Neck, Shoulders  Chest/ Front Trunk  Back  Upper Arm  Forearm /Wrist  Thigh  Lower Leg/ Feet Estimated Dermal Exposure, Mean of 2 Weather Conditions (calm, windy)  µg/hr 2314 (face), 524 (back of neck)  631  (front/ chest) -- -- 3,158 -- -- 2,460 Tsakirakis A.N., et al. (2014)74  Dosimeter Residue,  Range for 5 Applicators (Coverall Type A, Resist Spills); Note: a.s. = active substance  mg a.s./kg a.s. applied 0.51-1.20 (“Cap”) 21.5-64.3 (“Outer Jacket”) -- -- -- 14.4-238a (“Outer Pants”) 3.99-20.8 (“Outer Gloves”) Dosimeter Residue, Range for 5 Applicators (Coverall Type B, Cotton)  mg a.s./kg a.s. applied 0.48-3.63 (“Cap”) 9.76-112 (“Outer Jacket”) -- -- -- 25.8-237a (“Outer Pants”) 6.61-25.6. (“Outer Gloves”) Winterlin W., et al. (1986)56 Estimated Dermal Exposure, Harvest Operation (Captan as Wettable Powder)  mg/person/hr exposure µg -- 0.08 1.20 0.93 (“Sleeve”) 3.91a 0.52 0.94 (“Glove”)    MIXER-LOADER STUDIES (n=9)  Nigg H.N., et al. (1986)65  % of Total Dermal Exposure, Mean for 2 Mixer-loaders  % 1 4 3 1 4 39 48b -- Wojeck G.A., et al. (1982)66  Estimated Dermal Exposure, Mean  mg arsenic/hr 0.2 (Head & Neck) 0.7 0.3 9.9 (Arms) 40.7 (Legs & Feet) 57.1a % of Total Dermal Exposure, Mean   % <1 (Head & Neck) <1 <1 9 (Arms) 37 (Legs & Feet) 52a NIgg H.N., et al. (1983)67  Dosimeter Residue, Mean  µg/cm2 hr 0.15 (Shoulders) 0.19 0.09 -- 0.45a (Wrists) -- -- --   39  Study    Exposure Measure Type and Details  Units  Upper Body  Arms  Legs  Hands    Face/ Head, Neck, Shoulders  Chest/ Front Trunk  Back  Upper Arm  Forearm /Wrist  Thigh  Lower Leg/ Feet Chester G., et al. (1987)68 Dosimeter Residue, Range for 2 Mixer-loaders (3 Samples per Mixer-Loader) µg/ sample 2.21-60.3 (“Hood”) 3.99-674a 5.07-274  2.00-21.4 (Right), 1.62-45.6 (Left) 4.62-156 (Right), 9.64-187 (Left) 1.75-54.0 (Right), 3.06-159 (Left) 1.38-36.4 (Right leg), 5.48-22.1 (Left leg) 6.21 (Right “Boot”), 8.75 (Left “Boot”) 3.94-43.1 (Right Sock) 2.71-32.4 (Left Sock)  2.27-232  (Right) 4.22-106 (Left) Sanderson W., et al. (1995)70  Dosimeter Residue,  Geometric Mean  µg/cm2 0.17 (“Cap”) 0.25 0.06 0.12 (Arms) 1.69a -- -- Wojeck G.A., et al. (1981)72 Estimated Dermal Exposure,  Mean for Location 1 (where denim coveralls worn and laundered daily)  mg ethion/hr 3.90 (Head & Neck) 3.20 43.00 8.40 (Arms) 46.30 (Legs & Feet) 2199.00a Estimated Dermal Exposure, Mean for Location 2 (where long/short-sleeve shirts & long trousers worn multiple times before laundering)  mg ethion/hr 0.20 (Head & Neck) 1.00 0.80 2.80 (Arms) 25.20 (Legs & Feet) 252.80a dKiefer M., et al. (1996)76  Dosimeter Residue, Range for 2 Mixer-loaders (methyl parathion; Aug 15-16, 1995; 2 sampling days each) µg/cm2/hr -- 0.006-0.16 (Stomach) 0.005-0.16 (Left Chest),  0.003-0.01 (Right Chest)  -- -- 0.02-2.50 (Right), 0.01-0.05 (Left) NDc – 0.04 (Right) 0.05-0.84a (Left) 0.03 (Right) 0.01 (Left)   --   40  Study    Exposure Measure Type and Details  Units  Upper Body  Arms  Legs  Hands    Face/ Head, Neck, Shoulders  Chest/ Front Trunk  Back  Upper Arm  Forearm /Wrist  Thigh  Lower Leg/ Feet Dosimeter Residue, Range for 2 Mixer-loaders (Acephate Concentrations Aug 15-16; 1995; 2 sampling days each)    µg/cm2/hr -- 0.03 - 0.11 (Stomach) NDc - 0.10 (Left Chest), NDc - 0.12 (Right Chest) -- -- 0.02-0.31 (Right), NDc - 0.31 (Left) 0.03-0.71a (Right) 0.05-0.73a (Left) 0.01 (Right), NDc (Left) -- Dosimeter Residue, Range for 2 Mixer-loaders (Profenofos; Aug 15-16, 1995; 2 sampling days each)  µg/cm2/hr -- NDc - 0.004 (Stomach) NDc – 0.002  (Left Chest), NDc – 0.003 (Right Chest)  -- -- NDc – 0.004 (Right) NDc – 0.005 (Left) 0.001-0.002 (Right) 0.001-0.05a (Left) 0.004 (Right), 0.003 (Left) -- Everhart L., et al. (1982)77  Dosimeter Residue,  Range for 10 Mixer-loaders (10 Trials) mg /body area <0.02-7.8 (Face) <0.01-0.71 <0.01-0.42 -- 0.75-38 -- -- 2.3-45a (not sampled for 2 trials, n=8) Chester G., et al. (1986)63 % of Total Dermal Exposure, Mean for 3 Mixer-loaders  % 2 (Head) 26 (Body) -- 6.4 12.7 42.2a (Leg), 2.9 (Foot)    --  FIELD WORKER STUDIES (n=6)  Herman N.D., et al. (1985)49 Estimated Dermal Exposure, Range of Means for 2 Exposure Periods µg/hr 15-21 (Back of Neck), 27-77 (Front of Neck) 430-1100 (Left Chest & Stomach), 200-720 (Right Chest & Stomach)   470-680 -- 290-870 510-2100a -- 32-430   41  Study    Exposure Measure Type and Details  Units  Upper Body  Arms  Legs  Hands    Face/ Head, Neck, Shoulders  Chest/ Front Trunk  Back  Upper Arm  Forearm /Wrist  Thigh  Lower Leg/ Feet Kamble S., et al. (1992)78 Estimated Dermal Exposure, Range of Means for All Pesticides  (2 hour re-entry interval) µg/hr 2.1-12.4 (Head), 0.0-5.8 (Face & V-Neck), 0.0-0.9 (Back of Neck)  0.0-3.9 0.0-7.5 0.0-50.7 3.7-135.5  0.0-18.6 0.0-57.0 58.4-118.2a,e  Estimated Dermal Exposure, Range of Means for All Pesticides (4 hour re-entry interval)  µg/hr 0.4-14.4 (Head), 0.0-6.6 (Face & V-Neck), 0.0-3.2 (Back of Neck)  0.0-3.8 0.0-8.6 0.2-8.2 2.0-40.2 0.0-20.5 0.0-59.7 37.5-79.9a Estimated Dermal Exposure, Range of Means for All Pesticides  (8 hour re-entry interval) µg/hr 0.1-2.5 (Head), 0.0-1.9 (Face & V-Neck), 0.0 (Back of Neck)  0.0-4.1 0.0 0.0-0.6 0.0-25.4 0.0-2.3 0.0-8.7 13.9-42.2a Estimated Dermal Exposure, Range of Means for All Pesticides  (24 hour re-entry interval)  µg/hr 0.0-2.3 (Head), 0.0-0.2 (Face & V-Neck), 0.0 (Back of Neck) 0.0-0.7 0.0 0.0-0.3 0.0-16.9 0.0-1.1 0.0-0.4 9.9-31.5a Estimated Dermal Exposure, Range of Means for All Pesticides  (48 hour re-entry interval) µg/hr 0.0-1.2 (Head), 0.0 (Face/V-Neck), 0.0 (Back of Neck)  0.0 0.0 0.0 0.0-0.1 0.0 0.0-1.5 5.1-14.8a   42  Study    Exposure Measure Type and Details  Units  Upper Body  Arms  Legs  Hands    Face/ Head, Neck, Shoulders  Chest/ Front Trunk  Back  Upper Arm  Forearm /Wrist  Thigh  Lower Leg/ Feet Winterlin W., et al. (1984)53  Dosimeter Residue,  Mean  (4 hour exposure period)  µg/cm2 -- 0.837 0.624 5.909 3.708 1.936 2.198 Estimated Dermal Exposure, Mean  (8 hour, 3 days post application)  mg person-1 -- 2.568 1.92 15.20 25.63a 10.04 -- Everhart L.P., et al. (1982)77  Dosimeter Residue, Range across Workers (Hand Harvest)  µg <1-1 (Face) <1-1.2 -- -- 8.6-21 <1-2.6 <1-2.3 6,885-14,688a Estimated Dermal Exposure, Range across Workers  mg <0.02 (Face) <0.01-0.01  -- -- 0.40-0.99 <0.09-0.22 <0.09-0.21 6.9-15a Winterlin W., et al. (1986)56 Dosimeter Residue, Range of Means across Workers and Operation Types (dust/thinning, wettable powder/ thinning, wettable powder/ harvest)  µg/cm2 -- 1.52-10.6 0.53-32.8 2.28-37.9 2.16-23.8 0.81-5.25 11.1-46.6 Estimated Dermal Exposure, Range of Means across Workers and Operation Type (dust/thinning, wettable powder/ thinning, wettable powder/ harvest)  mg/person/hr exposure µg -- 0.48-3.24 0.17-10.09 -- 0.59-9.76 1.49-16.31a 0.42-2.75 2.28-9.97 Cock J., et al. (1998)57 Estimated Dermal Exposure, Range of Means across Tasks mg/m2/hr 0.16-0.45 (Forehead) 0.06-0.27 -- -- 1.55-3.06 (Arm), 1.96-6.16 (Wrist) -- -- 16.2-37.5a   43  Study    Exposure Measure Type and Details  Units  Upper Body  Arms  Legs  Hands    Face/ Head, Neck, Shoulders  Chest/ Front Trunk  Back  Upper Arm  Forearm /Wrist  Thigh  Lower Leg/ Feet (thinning, bending/tying-up, pruning, harvesting)    FLAGGER STUDY (n=1)  Chester G., et al. (1984)79 Estimated Dermal Exposure, Range across 2 Trials (n=1 Flagger) mg 0.04-1.40 (Head), 0.001-0.44 (V-Neck), 0.01-0.33 (Shoulders), 0.004-0.03 (Back of Neck) 0.001-10.89a   0.21-0.44 -- 0.04-0.84 0.08-10.8  0.02-4.00 0.02-0.40 a Body part with highest average “potential” dermal exposure as determined from study data;  b A single body part with the highest average “potential” dermal exposure could not be determined for this exposure measure, and only some of the dermal data are presented here as a result; c Not Detected; d Data pertaining to only 3 of the 8 different pesticides were presented for this study because “potential” exposure values were mostly “not detected” for body parts measured of the pestic ides not shown here; e Hands had the highest average exposure across pesticides and re-entry levels for the same pesticide, although data are shown here by re-entry interval for this study.     44 For operators, the hands had the highest average potential dermal exposure levels in more than half (9/15) of the studies,50-52,54,55,58,60,62,64 the thighs had the highest levels in 3/15 studies,53,56,59 the trunk area (front and back) had the highest levels in 2/15 studies,61,63 and forearms had the highest in 1/15 studies.57 For applicators, legs (thighs to feet) had the highest exposure levels in 5/12 studies,56,65,69,70,74 hands in 4/12 studies,66,68,72,75 and head-shoulder region and wrists were each the highest in 1/12 studies.67,73 In one applicator study, the body part with the highest average potential dermal exposure level could not be determined because dermal data were stratified by at least 5 variables;71 only data corresponding to the highest residue levels was presented in Table 2.2. For mixer-loaders, the legs (thighs to feet)  had the highest exposure levels  relative to other measured body parts in 4/9 studies,63,65,70,76 hands in 3/9 studies,66,72,77 front trunk in 1/9 studies,68 and wrists in 1/9 studies.67 For field workers, the hands 57,77,78 and  thighs both had the highest measured levels for 3/6 studies.49,53,56 For flaggers, the chest received the highest exposure relative to other measured body parts, closely followed by the thighs.79 For all farm jobs, except flaggers, the hands and legs (thighs to feet) most commonly had the highest average potential dermal exposure relative to other measured body parts.  2.3.4 Dermal Exposure Comparisons within Studies: 2. Total Body Exposure Levels  The second type of within study comparisons were performed on studies that evaluated more than one farm job and provided total body potential dermal exposure, as this is most important for epidemiology; total body exposures were compared across farm jobs/tasks and any other work variables for which exposure was evaluated (Tables 2.3 and 2.4).           45 Table 2.3: Within study comparison of total body potential dermal exposure to pesticides by job task (spray vs. mix-load) and application method (tractor mounted airblast spray vs. aerial spray). Study  Applicator (SPRAY)  Mixer-loader (MIX and LOAD) Within Study Exposure Pattern   Tractor Mounted Airblast Spray  Aerial Spray Tractor Mounted Airblast Spray Aerial Spray Nigg et al. (1986) 11.5a mg/hr  3.7a mg/hr  Spray > Mix-Load 24.7a mg/hr  7.5a mg/hr    Chester et al. (1987)   0.13 mg/hr  1.32 mg/hr Spray < Mix-Load   Wojeck et al. (1981)  1972.5 mg/hr  1799.0 mg/hr  Spray > Mix-Load   Among six studies that measured dermal exposure for applicators and mixer-loaders,65-68,70,72 three of them provided total body potential dermal exposure estimates,65,68,72 presented in Table 2.3 by farm job/task and application method with a qualitative within study description of the observed exposure pattern. In Nigg et al. (1986) and Wojeck et al. (1981), total exposure was greater for the spray task (tractor-mounted airblast) than it was for the mix-load task;65,72 the opposite exposure pattern was observed for aerial spray, mix-load > spray.68 The applicators and mixer-loaders in the Nigg et al. (1986) and Wojeck et al. (1981) studies worked on citrus grove farms, whereas the workers in Chester et al. (1987) worked on a cotton farm. The pesticides studied were different in all three studies (Table 2.3).        46 Table 2.4: Within study comparison of total body potential dermal exposure by farm job task (mix-load-spray vs. spray vs. thinning vs. harvest) and pesticide formulation (dust vs. wettable powder). Study Operator (MIX-LOAD-SPRAY)  Applicator (SPRAY)  Field worker (RE-ENTRY)  Within Study Exposure Pattern  Dust Wettable Powder Wettable Powder  Thinning (Dust formulation applied before task)  Thinning (Wettable Powder formulation applied before task)  Harvesta (Wettable Powder formulation applied before task) Farm Job Tasks Pesticide Formulation  Winterlin et al. (1986)  108.19 mg/person/hr exposure µg  34.72 mg/person/hr exposure µg  8.30 mg/person/hr exposure µg  5.43 mg/person/hr exposure µg  15.81 mg/person/hr exposure µg  52.12 mg/person/hr exposure µg        Harvest > Mix-Load-Spray > Thinning > Spray (for Wettable Powder Formulation)  Mix-Load-Spray > Thinning (for Dust Formulation)   Dust > Wettable Powder (for Operators)    Wettable Powder > Dust (for Field workers) a Harvest operation performed by hand  Two additional studies provided total body potential dermal exposure for more than one farm job. In the first study by Winterlin et al. (1986),56 three farm jobs were evaluated (operators, applicators, field workers), allowing for within study comparison of multiple tasks (mix-load-spray vs. spray. vs. thinning vs. harvesting) and two pesticide formulations (dust vs. wettable powder) for Captan (Table 2.4).56 Comparing total body potential dermal exposure across tasks for wettable powder formulation, field workers who harvested after pesticide was applied had 1.5 times greater exposure than operators who mixed, loaded and sprayed; almost 3.5 times greater than field workers who thinned crops after pesticide was applied; and more than 6 times greater than applicators who sprayed. In contrast, for dust formulation, operators who mixed, loaded and sprayed had approximately 20 times greater total body exposure than field workers who thinned crops after pesticide was applied. Comparing total body potential dermal exposure across pesticide formulations and within each task, dust resulted in higher exposure (> 3x) compared to wettable powder for operators (mix-load-spray); whereas for  field workers, exposure was approximately 3-10 times greater when wettable powder had been applied rather than dust formulation.  The operators and applicators in this study operated an elevated spray rig from inside a cab for grape farming (Table 2.1).56 In the second study by Everhart et al. (1982),77 total body potential dermal exposure was measured for mixer-loaders (for aerial application to beans, celery, avocadoes) and field workers (hand harvesting strawberries).  Total body exposure for mixer-loaders was more than 2 times that of field workers (26 mg   47 versus 12 mg, respectively),77 which suggested the mix-load task resulted in greater potential dermal exposure than the hand harvest task performed by field workers in this study.  2.3.5 Additional Work Factors for Potential Dermal Exposure in Farm Workers  Additional work factors reported in the farm studies as having influenced potential dermal exposure in farm workers are outlined in Table 2.5 by farm job and reported exposure pattern; details were provided if the exposure pattern could not be described. These work factors could not be further examined due to study comparison limitations, but they may play an important role for exposure assessment.   Table 2.5: Work factors reported in farm studies to potentially influence dermal pesticide exposure, by farm job.  Farm Job  Work Factor  Qualitative Exposure Pattern / Details about Work Factors   Operators  Task  Mix-load > Spray52,54,56,58 Equipment Open Tractor Cab > Enclosed Tractor Cab > Enclosed Tractor Cab with AC filtration51,55,56,59  Application Method Knapsack Boom Spray > Tractor Mounted Spray63    Applicators  Equipment  Tractor Mounted Spray with Elevated Boom (≥ 48 cm above ground) > Tractor Mounted Spray without Elevated Boom70,72   No Tractor Cab/Open Cab > Enclosed Tractor Cab72   Weather Conditions Windy > Calm74    Mixer-loaders  Work Method  Open-pour system > Closed- pour system71    Pesticide Amount Higher [Pesticide Concentration] in Tank > Lower [Pesticide Concentration] in Tank67  Work Habits Leaning against contaminated equipment, and wiping hands on pant legs after splash71   Field workers  Re-entry Time (time between crops being sprayed and workers entering the field)  Narrow Re-entry (2 hours) Time Window  > Wider Re-entry Time Window78   Duration of Exposure   2+ Hours of Exposure > less than 2 Hours of Exposure78  Work Habits Body Positioning (e.g., kneeling between rows of crops and reaching over plants exposes arms, thighs and chest more significantly)53,57    Flaggers  None to report  __     48  2.3.6 Uses of the Existing Dermal Monitoring Data for Occupational Exposure Assessment  Considering the variation between studies of this review and the limited data comparisons, Table 2.6 summarizes the uses and non-uses of the dermal monitoring data for pesticide exposure assessment in retrospective occupational epidemiology of farm workers. Table 2.6: identifying the uses and non-uses of existing dermal monitoring data from this literature review of farm studies (n=31), and the corresponding implications for future pesticide exposure assessment in occupational epidemiology.  Uses versus Non-Uses of Existing Dermal Monitoring Data from this Literature Review of Farm Studies (n=31)  Potential Dermal Exposure Data from Farm Studies (n=31) Implications for Pesticide Exposure Assessment in Occupational Epidemiology of Farm Workers   Uses  Identify specific body parts that commonly receive the highest levels of potential dermal exposure for different farm jobs.   Consider how increased exposure to certain body parts may affect pesticide absorption; useful for epidemiology if known where and how often a worker was typically exposed to a pesticide on the body. This information could be cross-referenced with self-reports about body parts that come into contact with pesticides during work.   Determine how (qualitatively and quantitatively) potential dermal exposure levels may vary by certain work factors: farm jobs/tasks, application method and pesticide formulation.  Identify other work factors that may increase potential dermal exposure levels, by farm job, and qualitatively identify exposure ranking in some cases.  Consider how the different work factors reported in this study (e.g., job, tasks, application method, and pesticide formulation) may impact dermal pesticide exposure. Compare these work factors to patterns found in other exposure studies, and consider collecting this information in the epidemiological study for use in an exposure assessment.   Non-uses  Cannot compare or combine quantitative measurement data between studies.   More systematic reporting of quantitative measures of dermal exposure is needed to identify and confirm potential determinants of dermal exposure.  Cannot evaluate the influence of work factors fully (e.g., all types of application methods) or at all in regards to total body potential dermal exposure levels due to variation that limited study comparisons.  Determinants of dermal pesticide exposure need to be used to develop exposure assessments that can be applied to case-control studies.       49 2.4 Discussion  Accurate pesticide exposure assessment is needed for retrospective epidemiological studies that rely on questionnaire data. This review considered existing dermal monitoring data from the literature to identify patterns and work factors that may influence dermal pesticide exposure in farm workers. The most important message was the significant amount of variation between studies that measured exposure for the same farm job, classified here as: variation in study focus and reporting, and variation in exposure levels; the former created the most difficulty and limited data comparisons to within studies only. Future exposure studies need to improve standardization for reporting dermal measurements to improve our ability to combine and compare exposure data across studies. We have two recommendations: (1) provide the raw measured data for each, specific body part, and (2) provide the characteristics of each set of values, including: units, exposure time, body part, pesticide, job, task and control characteristics of the work scenario. These recommendations should be easy for authors to implement. As an example of a study from this literature review that did report the exposure data as recommended here, Fenske et al. (1990) provided residue, exposure and % of total dermal exposure for each body part measured, along with the necessary details about these data, including the sample time that defined each ‘work period’, units, the specific body part surface area values used for calculating exposure, the tasks performed by the workers, the specific pesticide used, etc.59 Since the necessary details were provided for the dermal data by Fenske et al. (1990), it would have been possible to calculate between the different types of exposure measures (residue, exposure or % total exposure) had only one of them been reported. A clear understanding of the work scenario that these data represent makes it easier to understand the context in which these data can be compared to other studies as well. This review provided information about work factors that should be considered for dermal pesticide exposure assessment for epidemiology, and where further research is needed to determine more completely if and how these factors are determinants of exposure. In the first within study comparisons, potential dermal exposure levels were compared across all body parts measured to identify that which had the highest exposure relative to other measured body parts; these results were examined to see if other studies evaluating the same farm job had similar results and suggested that certain body parts may experience higher levels of potential dermal exposure. However, limitations in study comparisons did not allow for the potential determinants to be evaluated. Hands and legs were the body areas that most commonly had the highest potential dermal exposure levels for all farm jobs, except flaggers. Future research is needed to identify determinants of exposure to specific body parts, as this information could be used in quantitative approaches of exposure assessment. In studies that measured total potential dermal exposure for multiple farm jobs (n=5 studies), within study comparisons of total body exposure levels were examined across farm jobs/tasks and two additional work factors: application method and pesticide formulation. Results suggested total body exposure levels vary for different farm jobs   50 (operators, applicators, mixer-loaders, and field workers), job tasks (mix-load-spray, spray only, mix-load only, harvest, thinning), application methods (tractor mounted airblast, aerial) and pesticide formulations (wettable powder, dust). The following exposure patterns were observed: applicators had higher exposure than mixer-loaders65,72 when tractor mounted airblast spray was used, whereas the opposite pattern was observed for aerial application;68 operators had higher exposure than field workers (thinning) for wettable powder and dust formulations of Captan, but field workers (hand harvesting) had the highest levels and applicators the lowest levels overall when wettable powder Captan was used;56 and mixer-loaders had higher exposure than  field workers (hand harvesting) when aerial application of the fungicide Benlate was used.77 Unlike the body part exposure comparisons above, total body exposure results have a more obvious role in exposure assessment because they suggested actual exposure patterns with regards to some work factors; such information could be combined with questionnaire data about job, task, and application method to improve exposure classification. However, these exposure patterns require further investigation, since they are limited by the number and types of reliable study comparisons. For example, the exposure patterns observed in Winterlin et al. (1986) should only be used with caution (Table 2.4) given that they are based on a single study of 5 farm workers. For use in exposure classification, the observed exposure patterns should be compared to those observed in other studies. Additional work factors to be considered in future investigations are outlined in Table 2.5.  The Agricultural Health Study (AHS), which was not part of the 31 studies of this literature review, is a prospective cohort study of farm applicators and their spouses in the United States (Iowa, North Carolina); it has been a landmark study for pesticide exposure and health outcomes.80 The AHS used an algorithm to assess exposure intensity among pesticide applicators.80 Two versions of the algorithm were developed, a “general algorithm” consisting of basic variables (mixing/loading status, application method, equipment repair status, and PPE use) and a “detailed algorithm” with several additional variables that are more specific about work tasks/habits (mixing system, tractor with enclosed cab and/or charcoal filter, frequency of washing equipment, frequency of glove replacement, personal hygiene and changing clothes after a spill). 80  Each algorithm variable was assigned a weight based on expert judgement and external data to derive lifetime chemical-specific pesticide exposure for applicators. This AHS exposure assessment method has been evaluated against biomonitoring data for a few pesticides,40 however its reliability for use in a case-control study is not well known. Such an application may be difficult because case-control studies are unlikely to have collected the many variables included in the “detailed algorithm”.  In addition, “the reliability and validity of data” on farmers’ self-reported historical pesticide use has raised concerns.81 Pesticide use and application practices are usually well recalled by farmers “even years after the event”,38,81 but farmers’ ability to recall information about “duration, frequency, or decade of first use of specific pesticides” may be less reliable. 38 Given potential issues with use of the “detailed algorithm”, the “basic algorithm” may be the only option transferable from the AHS to retrospective case-control studies and should be evaluated. It is useful to compare findings from our review to components of the AHS algorithms. Application method and pesticide handling task (mix-load   51 status) were included in the AHS algorithms, and our review identified exposure patterns related to these factors as well. Pesticide formulation was not included in the AHS algorithm, but according to our review it may play an important role. The AHS algorithms did not include all potentially exposed farm jobs, for example, they omitted field workers and flaggers. In population-based studies, exposure estimates are needed for all exposed workers. PPE use was included in the AHS algorithms, but it was not examined in our review due to the nature of the dermal data used.  The Pesticide Handler’s Exposure Database (PHED) is a well-known generic exposure database, developed by the U.S. Environmental Protection Agency, Health Canada, and American Crop Protection Association in the 1990s; it provides exposure estimates for certain pesticide handler scenarios that are not chemical-specific.82 The PHED was used to assist with AHS algorithm development, but as stated by Dosemeci (2002) et al. “there is concern about its relevance to actual exposure situations because of the controlled, almost experimental, conditions under which the application occurs” in the studies from which the PHED is based (mostly from pesticide manufacturing companies).80 The PHED and AHS algorithms have their own strengths and limitations, however, the aim of this review was not to challenge either of these resources but rather provide a summary of the potential uses of dermal data and information from the existing literature for improving retrospective pesticide exposure assessment for multiple farm jobs. There are important limitations to be acknowledged for this review. Although detailed methods and screening criteria were used to collect studies from the existing literature, they were not evaluated using a quality grading scale; the reason for this is because there are no such scales appropriate for evaluating the dermal pesticide exposure literature and there were already a limited number of studies that qualified for this review.  Second, this review only considered dermal data based on over clothing and uncovered skin measurements; this was due to variation in clothing/PPE use between workers and studies, a lack of clothing/PPE reporting in many studies, and issues of pesticide penetration through different types of materials. There are several dermal monitoring studies that have measured under clothing and they may be able to provide important information regarding dermal pesticide exposure protection for farm workers. Also, it should be acknowledged that most of the farm studies date back to the 1990’s and 1980’s (4 studies published after 2008 and 1 study published before 1980) and changes in farm practices since this time may not be reflected in measured exposure values.  Data from older studies can be useful for assessing exposures in epidemiological studies of diseases of long latency, but are not as useful for studies of acute effects or for risk assessments of current farm employment. Finally, this review only included studies classified in the Farm Industry; however, the recommendations provided for standardizing dermal measurement reporting can be applied to all industries. More studies that measure dermal exposure to field workers and flaggers are needed; although these workers do not personally handle the pesticides, it’s important to understand and quantify their exposure levels since they are exposed.     52 2.5  Conclusion  Improved pesticide exposure assessment methods that can be applied to case-control study designs are needed, and it will require further investigation of potential determinants of dermal pesticide exposure, such as those highlighted in this review (jobs/tasks, pesticide formulation, application methods). Future epidemiology studies should collect information on the work factors summarized in this review by incorporating appropriate questions in questionnaires/interviews to help improve exposure classification.41,46 Most importantly, future exposure monitoring studies should standardize reporting procedures, as suggested above, to allow data to be compared and combined across studies for improved information on dermal pesticide exposure patterns for all farm jobs.                   53 Chapter 3: A comparison of methods for estimating pesticide exposure for farm jobs identified from a multiple myeloma case-control study  3.1 Introduction  Agricultural exposure to pesticides has long been a focus of study due to the observed elevated risks of certain types of cancers in this population, such as “cancer of the lip, skin, brain, lymphatic and hematopoietic system and soft-tissue sarcoma”,26 compared to the general population. However, epidemiologic research has struggled with “variation in disease risk estimates associated with occupational pesticide exposure [that] may be due to variation in exposure classification”.46 The majority of epidemiologic studies that have evaluated pesticide-cancer relationships are population-based retrospective case-control studies, where exposure assessments have often relied on self-reported occupational histories and exposures.41,42 However, there are two main issues with regards to self-reported exposures: 1) they typically “represent ‘usual’ exposures” and do not account for any time variability in exposure, and 2) they can be prone to recall bias resulting from case vs. control status.41 Additionally, it’s common for exposure histories to be limited to basic information, such as job title. When exposure assessment methods are based on job title alone, it assumes the same exposure for all individuals who worked in a specific job. This is problematic because it does not allow for specific agents to be identified as risk factors41 and it can increase the magnitude of exposure misclassification, a bias that can underestimate the risks associated with exposure and result in “diluted exposure-response gradients” in epidemiologic analyses.83 The objective of the current study was to compare farm job exposure estimates between three different exposure assessment methods, using farm job data from a population-based case-control study of multiple myeloma. The three exposure assessment methods varied in regards to the type of exposure assignment (i.e., quantitative ranked score vs. estimated exposure level vs. dichotomous exposure assignment), the questionnaire data that were used to apply these methods, and the way in which the variables were combined.  One of these methods was an algorithm developed for use in the Agricultural Health Study (AHS), a well-known prospective cohort study of pesticide applicators, 80 described in more detail below. This algorithm has been previously applied and tested for validity against other sources of information, such as biomonitoring data (urinary concentrations) and “simplistic determinants often used in epidemiologic studies (e.g., days of use and acres treated)”. 26 Given this, the AHS algorithm was considered an important comparison method for this study.   54 3.2 Methods 3.2.1 Multiple Myeloma Case-Control Study  A case-control study of multiple myeloma in British Columbia (BC), Canada was used to identify the farm sample for this study.  Multiple myeloma cases (n=393; aged 35-79) diagnosed between 2009 and 2013 were recruited from the BC Cancer Registry. Controls (n=376) were recruited from the Client Registry of the BC Ministry of Health that includes 95% of the provincial population. Controls were randomly selected and frequency matched to the cases on sex and age at time of diagnosis. Questionnaires were either administered by mail or by telephone (computer assisted) using trained interviewers who were able to speak languages of the major ethnic minorities in BC (Cantonese, Mandarin, Punjabi and Tagalog).  Subjects were asked to provide their residential (year, address, type of habitat, primary water source) and occupational (time period, industry, company name and location, and job title) history timelines. Based on the occupational history, subjects who indicated they lived on a farm and/or worked in agriculture, gardening, parks, golf courses or forestry were provided with a farm module questionnaire. This farm module questionnaire asked about:  farm details (type, farm, size, location, pesticide use), the jobs and tasks performed (description, apply pesticides, mix and/or load pesticides, wash and/or repair equipment, duration and frequency of task involvement) and work practices used (application method, crops treated, pesticide products, targeted pests, use of personal protective equipment, and body contact with pesticides during tasks). Based on the self-reported occupational history information, a total of 158 farm jobs were identified for this study. Jobs in which the subject classified the farm/workplace as ‘greenhouse’ (n=9), ‘garden/parks/lawn’ (n=2) or ‘base camp’ (n=1) were excluded. These exclusions were to ensure farm workplaces were as similar as possible in this analysis. Additionally, exposure assessment method 2 (described below), was developed using external data that did not include exposure monitoring data for greenhouse workers or other types of workplaces. All exposure estimation was provided at the farm job level.  3.2.2 Pesticide Exposure Assessment Methods  Three pesticide exposure assessment methods were applied to as many farm jobs as possible in the identified sample (n=158 jobs) for this study.  Method 1: Agricultural Health Study “General” Algorithm for Ranking Pesticide Exposure Intensity  As part of the Agricultural Health Study (AHS), a large prospective cohort of pesticide applicators in North Carolina and Iowa, two versions of an algorithm (‘General’ Algorithm vs. ‘Detailed’ Algorithm) were   55 developed to rank pesticide exposure intensity in applicators.80  The farm module questions used for the farm sample identified for this study correlated with the variables used in the ‘General’ Algorithm (mixing status, application method, equipment repair status and PPE use),80 so this version of the AHS algorithm selected for assessing farm job pesticide exposure in the current study. The algorithm is intended to estimate chemical-specific exposure intensity, and can be multiplied by time-related information (duration and frequency of corresponding chemical use) to provide lifetime cumulative chemical-specific exposure in applicators.80  The algorithm is shown in Equation 1. The algorithm weights were developed by comparing exposure monitoring results (dermal, inhalation, and biological) from 100 literature studies between and within algorithm variables.80 The dermal monitoring exposure data only included those that came from “pseudo-skin” measures (e.g., pads or patches, special clothing, coveralls, caps and gloves) and fluorescent tracer techniques.80  Data from the Pesticide Handlers’ Exposure Database (PHED) was used as supplementary information to “refine relative comparisons between application methods and various types of protective equipment”.80 This algorithm has since been evaluated against biomonitoring data and passive dosimetry measurement data; some of the variable weights were updated to reflect the results of the evaluation studies performed on subsets of the AHS applicator sample.40 We used the updated weights of Coble et al. (2011).40  The updates to the AHS algorithm were as follows: increase of some application method weights (air blast, boom on tractor, broadcast application, personally applied to seed and banded/directed spray for liquid); the addition of application methods (garden hose, hand held squeeze or squirt bottle, watering can/sprinkling can, hand spreader or push spreader, planter box); the reduction of the MIX weights; all weights for all three task variables were up-scaled by a factor of 10 to more easily allow for Coble et al. (2011) to rank the exposure differences between application methods and tasks; and lastly, PPE reduction factor for chemical resistant glove use was increased.40 Table 3.1 presents the scoring method used for applying the AHS algorithm (Equation 1) to quantitatively rank pesticide exposure for the farm job sample, as well as the corresponding farm module questions used from the case-control study.   Equation 1. Agricultural Health Study (AHS) Algorithm Intensity Level Rank = (MIX + APPLY + REPAIR) * PPE 80        56 Table 3.1: Agricultural Health Study ‘general’ algorithm scoring system using updated weights from Coble et al. 2011 and the multiple myeloma case-control farm module questions used to assign weights for farm jobs (1PPE weight details are summarized at the end of this table).  Algorithm Variable  Algorithm Variable Details  Weight  Multiple Myeloma Case-Control Farm Module Questions  MIX  Did not mix  0  “Did you mix and/or load any of the following yourself: herbicides, fungicides, insecticides?”  “How often did you mix and/or load pesticides yourself?” (options: never, sometimes, often or always)    Mix < 50% of time  20   Mix > 50% of time  50  REPAIR  No  0  “How often did you wash and/or repair spraying and mixing equipment yourself?” (options: never sometimes, often or always)    Yes  20  APPLY  Air blast  150  “Did you personally apply any one of the following items on this job: herbicides, insecticides, fungicides?”  “How did you apply the product?”   Hand spray Mist blower or fogger Fog or mist animals Greenhouse sprayer Pour fumigant from bucket Powder duster  90   Backpack sprayer  80   Dust animals Pour on animals  70   Garden hose Hand held squeeze or squirt bottle Watering can/sprinkling can  50  Soil injected or drilled Spray over rows Boom on tractor Broadcast application Personally applied to seed 40   Banded/directed spray (liquid)  30    Banded application (granular) Gas canister Hang pest strips in barn In-furrow Incorporated Inject animals Seed treatment Hand spreader or push spreader Planter box  20   Aerial   10 1PPE reduction factors were updated in Coble et al. 2011 as well, i.e., 60% reduction factor if chemical resistant or rubber gloves were worn, and a 10% reduction (up to a maximum of 30%) for each of the following that were worn: cartridge respirator, Tyvek coveralls, face shields/ goggles, chemical resistant boots, apron, other.40    57 Method 2: Potential Dermal Exposure Algorithm A systematic review of the dermal pesticide exposure monitoring literature was previously conducted to determine the usefulness of dermal data form 31 studies for exposure assessment (Chapter 2).84 Some of the data collected for this literature review was used to develop a new, novel algorithm (Equation 2) to estimate potential dermal exposure in farm workers when combined with self-reported answers to farm module questions about personally mixing-loading and/or personally applying pesticides for farm job, as well as the location(s) on the body (head/face, body, arms and hands, legs/feet) where pesticides usually came into contact with the worker during each reported task. Unlike the algorithm for method 1, this algorithm was developed from dermal data only because the dermal exposure route has been identified as the primary route that contributes most significantly to internal dose in workers exposed to pesticides.33,34  Dermal exposure uptake can vary depending on the body area where contact occurs33 which can be influenced by the pesticide handling task(s) performed as well. Instead of the relatively ranked exposure provided by method 1, this algorithm provided an estimated level of potential dermal exposure (mg/hr) for each farm job.  Equation 2. Potential Dermal Exposure Algorithm Potential Dermal Exposure (mg/hr) = MIX-LOAD (HeadMIX-LOAD + BodyMIX-LOAD + Arms/HandsMIX-LOAD + Legs/FeetMIX-LOAD) + APPLY (HeadAPPLY + BodyAPPLY + Arms/HandsAPPLY + Legs/FeetAPPLY)  The weights associated with the pesticide handling task variables, MIX-LOAD and APPLY, represent the relative proportion of time spent performing the task and are dependent on self-reported responses to whether a farm job included the personal involvement of: none, either one, or both of these tasks (Table 3.2). For farm jobs where subjects reported only personally mixing and/or loading pesticides, the task weight for ‘MIX-LOAD’ was set to 1.0 and the task weight for ‘APPLY’ set to 0.0, as it would be expected that body areas reported to have usually come into contact with the pesticide during task performance, would be due to this task alone (and vice versa for farm jobs where subjects reported personally applying pesticides only). However, for farm jobs where subjects reported personally performing both of these tasks, the contribution of the body area exposure resulting from the performance of each task is multiplied by the average proportion of time that each task was likely performed for these jobs, i.e., 0.2 for MIX-LOAD task weight and 0.8 for APPLY task weight. The average proportion of time spent on each task was determined from the literature review, specifically those studies that measured dermal exposure in Operators who performed both of these tasks. Of the 31 farm studies in this literature review, 16 measured dermal exposure among Operators, but only five provided information on the duration of time   58 spent performing each task; these five studies were used to determine the average proportion of time spent mixing/loading versus applying pesticides. Table 3.2 shows the task weight values (Equation 2).  Table 3.2: Pesticide Handling Task Weights for the Potential Dermal Exposure Algorithm  Weights Task MIX-LOAD APPLY Mix-Load Only 1.0 0.0 Apply Only 0.0 1.0 Mix-Load and Apply 0.2 0.8 None 0.0 0.0  Task-specific body area exposure values were also included in the algorithm for method 2, and were broken down into four major areas: head (or face), body, arms/hands, and legs/feet. Considering all studies from the literature review, those that provided the following information were used to determine the median average body area exposure (mg/hr) for each task in the algorithm: 1. potential dermal data by body part for either one or both pesticide handling tasks (if provided for both tasks, the data needed to be provided separately for each task), and 2. potential dermal residue or exposure data that could be converted to mg/hour units. Ten of the 31 studies provided this type of dermal data for the mixing-loading task, and 15 of the 31 studies provided this data for the application task. Using these studies, the median average body area exposure level (mg/hr) was determined for each task and used to represent the corresponding body area exposure variable in the algorithm (Table 3.3). The assignment of these values (Table 3.3) was based on self-reported responses to farm job specific questions that asked subjects to mark “the parts of the body that usually came into contact with pesticides” for each of these tasks that was performed. Table 3.3: Task-specific Body Area Exposure Levels (mg/hr) for Potential Dermal Exposure Algorithm     Task-specific Exposure Level (mg/hr) Body Area MIX-LOAD APPLY HeadMIX-LOAD/APPLY 0.56 1.0 BodyMIX-LOAD/APPLY 1.0 5.01 Arms/HandsMIX-LOAD/APPLY 58.26 12.56 Legs/FeetMIX-LOAD/APPLY 1.57 4.43     59 Method 3: Simple Surrogate Approach The goal of this final method was to assign farm jobs as “exposed” or “unexposed” based on self-reported information about farm pesticide use only. Farm jobs in which the subject reported “no” to the question, “Were pesticides to kill weeds, insects, fungus or moulds, or rodents such as rats applied on this farm/workplace?” were assigned to the unexposed job group. Whereas, farm jobs in which the subjects reported “yes” to this same question were assigned to the exposed job group. This method did not result in any exposure intensity estimation, but it was included because pesticide exposure has been frequently assessed this way for epidemiological purposes making it an important method for comparison.   3.3  Results  There were a total of 3,212 jobs indicated by 779 subjects, and among these, 158 (5%) were farm jobs reported by 122 (16%) subjects. This study focused on these farm jobs; Figure 3.1 shows their breakdown by self-reported response about pesticide use on the farm (Groups 1-3) and responses to questions shown in Appendix A about the personal handling of pesticides (apply, mix and/or load, wash and/or repair pesticide equipment) performed for the job (Group 1, A-C). This information was used to determine whether the farm jobs were eligible for pesticide exposure assessment.  The farm module question was used to determine Groups 1-3 (Figure 3.1): “Were pesticides to kill weeds, insects, fungus or moulds, or rodents such as rats applied on this farm/ workplace?” (Appendix A, Question 8). Farm jobs could not be assigned exposure by any of the three methods when subjects did not respond to this question or responded “don’t know”, n=33 (21%) farm jobs (Group 3, Figure 3.1).  To distinguish between farm jobs that involved different combinations of pesticide handling tasks (Groups 1A-1C, Figure 3.1), the following farm module questions were used: “did you personally apply any of the following on items on this job: herbicides, insecticides, fungicides?” (Appendix A, Question 9), “did you mix and/or load any of the following yourself: herbicides, insecticides, fungicides?” (Appendix A, Question 14), and “how often did you wash and/or repair spraying and mixing equipment yourself?” (Appendix A, Question 21). Farm jobs could not be assigned exposure using methods 1 or 2 when subjects did not respond or responded “don’t know” to one or more of these pesticide handling task questions, n=22 (14%) farm jobs (Group 1D, Figure 3.1). The primary reason that farm jobs were classified in Group 1D, was due to non-responses about personally applying pesticides (n=14, 64%), followed by farm jobs for which subjects did not respond to either question that asked about personally mixing and/or loading pesticides or personally washing and/or repairing equipment for pesticides.    60 Table 3.4 compares farm job characteristics across Groups 1, 2 and 3 from Figure 3.1. Using age at the time of job end, all farm jobs were classified into age groups (child: 0-11 years old, teen: 11-17 years old, and adult 18 years or older). For both Group 2 and Group 3 (Figure 3.1), close to half of the farm jobs ended before the subjects were adults (Table 3.4); this differs from Group 1, where 82% of the subjects were adults at the time of job end. In terms of the length of time the farm job was held, the largest proportion for each group was for farm jobs held for more than 10 years. Across all three groups, the majority of farm jobs were held by subjects that lived on the farm, regardless if they also identified themselves as an owner/operator or worker of the farm, but Groups 2 and 3 had higher proportions reporting only living on a farm (i.e., without being an owner/operator or worker). The other noticeable difference between the groups was for farm type reported. For Group 3, 52% of farm jobs took place on “animal only” farms, which compares to only 15 and 32% for Groups 1 and 2, respectively.   A total of 103 farm jobs (46 farm jobs of Groups 1A-1C, and 57 farm jobs of Group 2) were assessed for pesticide exposure using the algorithm-based methods 1 and 2; due to incomplete responses, the above discussed 55 farm jobs (34%) were excluded, i.e., Group 3 and Group 1D farm jobs.  For method 3, however, since less detailed information was required, a total of 125 (all of Group 1 and 2) farm jobs were assessed for pesticide exposure. The 57 farm jobs that  reported “no” to pesticide use on the farm (Group 2), were considered to be unexposed to pesticides for all methods, and were assigned the following exposures: Method 1, Intensity Score = 0; Method 2: Level = 0 mg/hr; and Method 3: unexposed. Methods 1 and 2 were applied to estimate exposure for the remaining 46 farm jobs that reported ‘yes’ to farm pesticide use and provided full responses about personal involvement in the three pesticide handling tasks (Groups 1A-C).    61  Figure 3.1: Multiple Myeloma Case-Control Study: Total Farm Job Counts Eligible for Exposure Assessment based on Pesticide Use Variables.  3,212 Total Jobs in Multiple Myeloma Case-Control Study 158 Farm Jobs Group 2: Reported 'NO' to  pesticide use on farm, n = 57 Group 3: No mention of pesticide use on farm, n = 33   Group 1: Reported 'YES' to pesticide use on farm, n = 68  Group 1A: Apply, mix-load and wash-repair,  n= 25   Group 1B: Apply and Mix-Load,  n = 12  Group 1C: Apply only, n = 9 Group 1D: No mention of one or more tasks being performed for farm job, n = 22  Group 1-3: Based on self-reported use of any pesticides on farm. Group 1A – 1D: based on self-reported tasks personally performed for farm job (apply, mix-load, wash-repair).   62 Table 3.4: Farm job characteristics by self-reported pesticide use on farm for Groups 1-3. Farm Variables Group 1: Reported ‘YES’ to pesticide use on farm, n (%) Group 2: Reported ‘NO’ to pesticide use on farm, n (%)  Group 3: No mention of pesticide use on farm, n (%)   Age at time of job end  Child (0-11 years)  3 (4.4)  9 (15.8)  7 (21.2) Teen (11-17 years) 9 (13.2) 14 (24.6) 6 (18.2) Adult (18 and older) 56 (82.4) 30 (52.6) 19 (57.6) Unknowna 0 (0.0) 4 (7.0) 1 (3.0)   Total n (%)  68 (100) 57 (100) 33 (100)   Length of Time Farm Job Held (years)  < 5 YRS  11 (16.2)  19 (33.3)  6 (18.2) 5-10 YRS 20 (29.4) 8 (14.0) 11 (33.3) > 10 YRS 32 (47.1) 25 (43.9) 14 (42.4) Unknowna 5 (7.4) 5 (8.8) 2 (6.1)   Total n (%)  68 (100) 57 (100) 33 (100)   Description of work on farm  Lived only  21 (30.9)  40 (70.2)  17 (51.5) Owner/Operator only 3 (4.4) 1 (1.8) 0 (0.0) Worker Only 2 (2.9) 2 (3.5) 3 (9.1) Lived and Owner/Operator 9 (13.2) 2 (3.5) 4 (12.1) Lived and Worker 22 (32.4) 10 (17.5) 5 (15.2) Owner/Operator  and Worker 1 (1.5) 0 (0.0) 0 (0.0) Lived, Owner/Operator and Worker 9 (13.2) 2 (3.5) 2 (6.1) Unknownb 1 (1.5) 0 (0.0) 2 (6.1)   Total n (%) 68 (100) 57 (100) 33 (100)   Farm Type  Crop   3 (4.4)  3 (5.3)  1 (3.0) Fruit 8 (11.8) 6 (10.5) 1 (3.0) Animal  10 (14.7) 18 (31.6) 17 (51.5) Combined Crop & Animal 14 (20.6) 9 (15.8) 1 (3.0) Combined Fruit & Animal 6 (8.8) 6 (10.5) 2 (6.1) Combined Crop & Fruit 3 (4.4) 1 (1.8) 0 (0.0) Combined Crop, Fruit & Animal 21 (30.9) 11 (19.3) 6 (18.2) Other (e.g., forage, hobby farm) 1 (1.5) 2 (3.5) 0 (0.0) Unknownc 2 (2.9) 1 (1.8) 5 (15.2)   Total n (%)  68 (100) 57 (100) 33 (100)  a Farm records classified as ‘unknown’ if no response to  ‘age’ at time when farm record ended. b  Farm records classified as ‘unknown’ if no response provided for describing farm work; one farm record (under Group 3) reported ‘no’ to all farm module possible responses for describing farm work. c Farm records classified as ‘unknown’ when there was no response or the response was “don’t know”.        63 3.3.1  Pesticide Exposure Assessment  Method 1: Agricultural Health Study (AHS) Algorithm for Pesticide Exposure Intensity When applying the AHS algorithm, certain circumstances required modifications to determine an individual weight or the final intensity score for a farm job. First, if a single farm job consisted of more than one record, which was common (typically due to the use of more than one pesticide and/or application method), then an AHS intensity score was derived for each record and then averaged across all records for a single farm job average intensity score.  To determine the APPLY weight for farm jobs in which subjects reported personally applying pesticides, the AHS application methods (Table 3.1) needed to be matched to those reported in response to the farm module questions of the case-control study. There were differences in the application method options between these two sources. For the most part, the application methods commonly reported were backpack spray, boom on tractor, hand spray or air blast spray, which were easily matched and included in both the multiple myeloma study and the AHS algorithm. Some less common application methods included self-reported specified responses such as: “by hand then by auger”, “mixed with grain” or “mechanized sprayer”. Although not common, when there was uncertainty in the application method match between the two sources, the average APPLY weight across all application methods was assigned to the farm job (average APPLY weight = 51).  To determine the MIX weight for farm jobs in which subjects reported ‘yes’ to personally mixing and/or loading pesticides, the frequency at which this task was performed was also used (Table 3.1). In the Multiple Myeloma study questionnaire, subjects were asked to characterize their frequency of performing this task using the following options: never, sometimes, often, or always.  Since the AHS algorithm assigned the MIX weight based on these more quantitative options (Table 3.1),  never mix, < 50% of time mixed, or 50+ % of time mixed,40,80 the frequency definitions were correlated as follows: if a subject reported ‘no’ to mixing and/or loading or ‘never’ to frequency of mixing and/or loading, the farm job was assigned a MIX weight of ‘0’ (never mix); if a subject reported ‘yes’ to mixing and/or loading and ‘sometimes’ to frequency of mixing and/or loading, the farm job was assigned a MIX weight of ‘3’ (< 50% of time mixed); and if a subject reported ‘yes’ to mixing and/or loading and ‘often’ or ‘always’ to frequency of mixing and/or loading, the farm job was assigned a MIX weight of ‘9’ (50+ % of time mixed). Among the 46 farm jobs that responded to all three questions that asked about pesticide handling tasks (Group 1A-C) , 12 of these farm jobs did not mention the frequency of mixing and/or loading pesticides although the subjects reported ‘yes’ to performing this task. To maintain these 12 farm jobs in the analysis, since they provided all other necessary exposure history details, they were assigned a MIX weight of ‘3’ (< 50% of time mixed); this could underestimate the total AHS intensity score for these jobs, but it is the most conservative approach.     64 The REPAIR weight assignment was straightforward, as it only relied on whether or not a subject reported “yes” or “no” to performing this task and was applied accordingly (Table 3.1).  The final variable of the AHS algorithm represents the type of personal protective equipment that was typically worn, i.e., the PPE reduction variable (Equation 1). This variable was not included when determining intensity scores in this farm job sample because this paper aims to compare job exposure between methods and PPE was not included in the algorithm for Method 2.   Method 2: Potential Dermal Exposure Algorithm  This method was very straight forward to apply to the farm job data. The same 46 farm jobs were assessed for pesticide exposure with this method, using the questions that asked about personally applying pesticides and personally mixing and/or loading pesticides for farm jobs, as well as corresponding questions that asked “when [applying or mixing/loading] pesticides, what parts of your body usually came into contact with the pesticides?”. The optional body area responses for this question included all of the following that applied: head and/or face, arms and hands, body, legs/feet, lungs and respiratory tract, digestive tract, or none (note: lungs and respiratory tract and digestive tract were not included in algorithm since they represent alternate routes of exposure). For each of the two main pesticide handling tasks that were reported as being personally performed for a farm job, the corresponding task-related body areas marked as “yes” were included in the algorithm. Similar to method 1, for farm jobs in which there were multiple records per farm job with different reported information, the exposure level (mg/hr) was estimated for each record and averaged across all records for a final average exposure level, but this was not common for this method.  In addition to the 57 farm jobs that reported “no” to farm pesticide use (Group 2) that were assigned a exposure level of 0 mg/hr, there were an additional 6 farm jobs assigned this level (0 mg/hr) because “none” was reported when asked which body areas came into contact with pesticides during the performance of these pesticide handling tasks.  Method 3: Simple Surrogate Approach  Among the 158 farm jobs in the dataset, 57 (Group 2, 36%) were classified as unexposed and 68 (Group 1, 43%) were classified as exposed in regards to pesticides. As shown in Figure 3.1, there were 33 (Group 3, 21%) farm jobs for which there was no mention of pesticide use, so these jobs could not be assessed using method 3. Unlike methods 1 and 2, this method was less prone to issues of non-  65 response data and the farm jobs in Group 1D (Figure 3.1) were not excluded since task data was not necessary for the application of this method.   3.3.2  Comparison of Exposure Results for Algorithm-based Method 1 versus Method 2  Among the 103 farm jobs (Group 1A-C and Group 2) evaluated for pesticide exposure using methods 1 and 2, there were at least 57 jobs assigned an exposure value of ‘0’ since the subjects of these jobs reported “no” to pesticide use on the farm (Figures 3.2 and 3.3).  The mean exposure values for all103 farm jobs were a ranked intensity score of 50.0 for method 1 and 8.0 mg/hr for method 2.  Figure 3.2: Distribution of farm job exposure values (n=103) as determined with Method 1, AHS Algorithm.     0102030405060Number of Farm Jobs Method 1: AHS Intensity Scores AHS Intensity Algorithm Scores for Farm Jobs (n=103)   66 Figure 3.3: Distribution of farm job exposure values (n=103) as determined with Method 2, Potential Dermal Exposure Algorithm (mg/hr).     Figures 3.2 and 3.3 show the frequency distributions of estimated exposure for all 103 farm jobs assessed using methods 1 and 2, respectively. For method 2, there were more farm jobs assigned an exposure value of ‘0’ than in method 1, and most of the farm jobs fall within four defined quantitative groups: <6 mg/hr, 10-14 mg/hr, 20-28 mg/hr and 30-32 mg/hr. With method 1, the exposure values above ‘0’ were most heavily assigned in the range of ranked scores between 60 and 160.   010203040506070Number of Farm Jobs Method 2: Potential Dermal Exposure Level (mg/hr) Potential Dermal Exposure Level (mg/hr) for Farm Jobs (n=103)   67 Figure 3.4: The relationship between farm job (n=46) exposure values determined using Method 1 (AHS Intensity Algorithm) versus Method 2 (Potential Dermal Exposure Algorithm).   A scatter plot (Figure 3.4) and the Pearson correlation coefficient were used to examine the relationship between farm job exposure values (n=46) determined by methods 1 and 2, excluding the farm jobs that had an exposure value of ‘0’ for both methods (n=57, Group 2 of Figure 3.1). The resulting correlation coefficient was close to zero (r = -0.099, p-value = 0.514), indicating no relationship between the farm job exposure values estimated by method 1 versus 2. The circled data points in the scatterplot (Figure 3.4) represent six farm jobs for which the exposure estimates assigned by the two methods were extremely different. For two of these farm jobs, the exposure values estimated using the method 2 potential dermal exposure algorithm were much higher compared to the exposure values estimated using the method 1 AHS algorithm, whereas the opposite situation occurred for the remaining four jobs (i.e., exposure estimates using method 2 were much lower than those estimated using method 1). To understand what contributed to these differences, the individual variables of each algorithm were examined for these six farm jobs. For the four farm jobs with much higher exposure intensity estimates using method 1, this was due to the APPLY and MIX scores that were assigned to these jobs based on the self-reported information about application method/pesticide type and frequency of mixing-loading pesticides. They 051015202530350 20 40 60 80 100 120 140 160 180 200Method 2: potential dermal exposure level (mg/hr) Method 1: AHS intensity ranked score AHS intensity ranked score vs. potential dermal exposure level (mg/hr) for farm jobs, n=46 Median (method 2) = 22.7 mg/hr Median (method 1) = 111.5 Pearson Correlation Coefficient = -0.099     P-value = 0.514   68 were assigned APPLY scores of 90 and MIX scores of 50, which are some of the highest scores that can be assigned for this method (Table 3.1).  However, these four farm jobs had much lower exposure estimates using method 2 (equal to or close to 0 mg/hr), because this algorithm used body area exposure level variables that were assigned based on self-reports about body areas that usually came into contact with pesticides during each task. For one of these farm jobs, the subject reported the ‘head’ only as coming into contact with pesticides, which has the smallest assigned exposure level compared to all other body areas across both tasks (Table 3.3). For the second farm job, the subject did not report any body areas coming into contact with pesticides during performance of these tasks.  For the two farm jobs that were estimated to have much higher exposure intensity using the method 2 algorithm compared to method 1, this was because the subjects reported all areas (head, arms/hands, body and legs/feet) as usually coming into contact with pesticides during pesticide handling tasks. However, based on the variables and questionnaire responses needed for the method 1 algorithm, the assigned APPLY (≤ 50) and MIX (< 50) weights were low (see Table 3.1) for these two farm jobs.  Table 3.5 provides a visual comparison and summary of the main characteristics of each exposure assessment method of this study. As shown, all three methods are reliant upon self-reported information about exposure, however, method 3 only requires basic information about farm pesticide use for its application and this method does not estimate exposure intensity. Methods 1 and 2 use a combination of different variables and more detailed information about past job exposures to estimate exposure intensity, but only method 1 has been previously evaluated against other methods in attempt to validate the algorithm.            69 Table 3.5: Comparison of pesticide exposure assessment method characteristics to summarize the advantages and disadvantages of each method. Characteristics of the Three Pesticide Exposure Assessment Methods Method 1: AHS Intensity Algorithm Method 2: Potential Dermal Exposure Algorithm Method 3: Simple Surrogate (farm pesticide use)  Based on external measurement data      Includes multiple work variables that can influence exposure      Method is used with self-reported information about past farm job exposures      Previously applied and evaluated against other methods (urine and passive dosimetry data)      Quantifies pesticide exposure intensity      Exposure is quantified as an estimated dose level (mg/hr)      Can be applied to assess lifetime exposure      Can be applied to assess pesticide-specific exposures  a a   Issues of non-response is lower, due to fewer variables needed for method     aThis characteristic only applies if the data collected for the algorithm is pesticide-specific.   3.4 Discussion  Self-reported exposure data from a multiple myeloma case-control study was used to apply and compare the results of three pesticide exposure assessment methods to 103 farm jobs. The AHS algorithm (method 1) was previously developed and evaluated using a prospective cohort of applicators and was considered an important comparison method for the current study.40,80 The potential dermal exposure   70 algorithm (method 2) was a newly developed method from a recent systematic literature review of dermal monitoring data that correlated specific body area exposure estimates with pesticide handling tasks (apply and mix-load).84 The Simple Surrogate Approach (method 3) was based solely on questionnaire data reported for a single exposure factor that identified exposed/unexposed groups rather than estimating exposure intensity. This method was included because many epidemiologic studies have used simple surrogates, such as job title alone to assign exposure.26  There were two key findings from this study. First, due to the ease of applying the Simple Surrogate Approach, because it was reliant upon the least amount of self-reported information, this was the only method that could classify exposed people who did not report information on pesticide handling tasks (Group 1D). Overall, this method was the least prone to observation loss (Table 3.5). Second, the comparison of exposure intensity results from the two algorithm-based methods showed no correlation when applied to this farm job sample, which may have resulted from differences in: (1) the exposure monitoring literature that was used to develop each algorithm, and (2) the aspects of pesticide exposure that each algorithm included, and the questionnaire data that was used to create the required variables.  In comparing the exposure monitoring literature used to develop each algorithm-based method, it’s important to note that Dosemeci et al. (2002) provided limited information about the literature used for the AHS algorithm. We were unable to find documentation about the studies from which exposure monitoring data were extracted, and exactly how these data were combined to derive the algorithm weights in light of the variation that may have existed between studies. In Chapter 2, the systematic literature review that evaluated the usefulness of dermal exposure monitoring data for pesticides, a significant amount of variation in the way data were reported between studies was found to limit data comparability for exposure assessment.84 Dosemeci et al. (2002) reported that the literature was the primary source of information for the development of the AHS algorithm, but the main details provided were that they “extracted exposure data from more than 100 available published articles that had numerous measurements of pesticide exposures in relation to mixing, application or work practices in agricultural settings”.80 Data was collected for three exposure routes (dermal, inhalation and internal), but dermal measurement data were mostly used for algorithm weight development.80,85 However, with this information alone it is difficult to know the types of farm working populations the AHS algorithm may best suited for estimating exposure intensity, e.g., in terms of the pesticides, formulations, application methods or types of equipment the data best represents. Additionally, we do not know if and how variation between studies was managed and/or if it could have influenced the algorithm in any way. The potential dermal exposure algorithm was developed from dermal pesticide monitoring data obtained from the literature as well, as part of a systematic literature review of 31 studies.84 Due to issues of variation between studies that limited data comparability, data were extracted from fewer than half of these studies to develop the algorithm weights. 84 Conditions were established to determine which   71 studies from this literature review could be used to estimate the median exposure value for each body area. For example, the dermal data needed to be reported by body part and separately for one or both tasks (so as to be combined into these larger body areas: head/face, body, arms/hands, legs/feet), and we needed sufficient information to convert the dermal data, which may have been provided as residue or exposure estimates, into units of mg/hour (e.g., needed exposure duration information that corresponded with all samples).84 The dermal pesticide monitoring data used for this algorithm originated from studies of farm workers who applied pesticides from each functional use group (e.g., Herbicides: Paraquat, Diallate, and 2,4-D; Fungicides: Captan, Chlorothalonil, and Fluazinam; and Insecticides: Primicarb, Endosulfan, and Acephate) and who used multiple application methods (53% boom spray, 33% airblast, 6% handheld and 6% a combination of the three).84  The boom spray and airblast data included a range of vehicles, from tractors with no cabs, open cabs, cabs with open vs. closed windows, and closed cabs with carbon filtration.84 Regarding the application of this algorithm to the case-control data used for the current study, most of the subjects who reported “yes” to personally applying pesticides during their farm job reported the same application methods noted above (approx. 70%), i.e., boom spray, airblast and handheld spray. For the remaining farm jobs, subjects reported alternate application methods, such as: pour on method, aerial spray, backpack spray, and hand spreader or push spreader that are not well represented by the exposure data used to develop this algorithm. However, this information does show that the dominant pesticide application methods reported by the study population were well represented by the studies used to develop the weights of this algorithm.  In terms of the different variables of each algorithm, the AHS algorithm focused on job-specific variables (e.g., tasks, application method, and pesticide type) whereas the potential dermal exposure algorithm included job-specific (task) and personal-level variables (body areas that usually came in contact with pesticides during each task). Previous evaluations of the AHS algorithm lend some evidence of validity for a few of the individual variables in their ability to accurately rank-order exposure intensity for pesticide applicators.36 In a study by Coble et al. (2005), the AHS algorithm was applied to a sample of pesticide applicators and the results were compared to urine samples of the applicators who recently applied herbicides 2,4-D and/or MCPA, using Spearman rank correlations.36 In this Coble et al. (2005)study, the APPLY and MIX weights could not be validated since nearly all applicators used the same application method  (tractor with spray boom) and reported personally mixing pesticides before application (See Table 3.1 for variable description and weight values).36 Although there was some variation in the reporting of the pesticide equipment repair task (REPAIR weight), Coble et al. (2005) reported that it was simply a dichotomous variable that increased total exposure if the task was performed.36 Therefore, the only algorithm variable that really contributed to variation in the exposure scores was personal protective equipment (PPE).36 Coble et al. (2005) reported that the “algorithm scores, based mostly on PPE use, provided reasonably valid estimates of exposure intensity for these applicators” since the results showed an increasing trend in the urine concentration across three, increasing exposure groupings of the algorithm scores for both herbicides assessed (2,4-D and MCPA).36 However, the Spearman rank   72 correlations were only statistically significant for 2,4-D and not for MCPA.36 Although this study provided some validating evidence for the AHS algorithm, it was limited to the PPE weight and the specific pesticides that were measured. Another study by Thomas et al. (2010) evaluated the AHS algorithm using a subset of the AHS pesticide applicator cohort sample, and collected urinary biomarker measurements for herbicide 2,4-D and insecticide chlorpyrifos, as well as passive dosimetry measurements (dermal patch for body, hand wipes and personal air samples).85 In this study, there was more variation with regards to the application methods used for 2,4-D and chlorpyrifos (broadcast spray, hand spray, directed spray, in-furrow granular, broadcast/directed liquid spray).85 With regards to 2,4-D applicators, there were significantly positive correlations (all P-values < 0.03) between the exposure rank scores and all direct measurements (urine biomarker measurements, dermal patch, hand wipe and air samples).85 For chlorpyrifos applicators, the algorithm-based intensity scores were significantly, positively correlated with dermal patch results.85 However, an evaluation of this metric by Hines et al. (2008) using an AHS sample of private orchard applicators who used the insecticide Captan, reported that the exposure intensity rank score “did not predict air, hand rinse or urinary” biomarker exposures.86 Therefore Hines et al. (2011) provided algorithm adjustment recommendations that were used in combination with results from other studies to update some of the algorithm weights.40,87  The updated algorithm was used in the current study with the specific weights outlined in detail by Coble et al. (2011). 40 Given the multiple evaluations of the AHS algorithm, it was used as a comparison method in this study. As noted in Hines et al. (2011), “to a large extent, the exposure scenario and population of interest dictate the exposure determinants found for a group of workers”.87 Therefore, the AHS algorithm may rank exposure intensity with a reasonable level of accuracy for the AHS sample of pesticide applicators, but its accuracy when used with other study samples is less understood. Further evaluation of the AHS algorithm is still needed, particularly with regards to its application to retrospective studies where self-reported historical information would be used to apply the algorithm. Overall, the AHS algorithm has been evaluated to show some validity evidence. Compared to the potential dermal exposure algorithm, it is comprised of a greater number of variables that may be predictive of exposure. The potential dermal exposure algorithm used a different set of variables for estimating exposure intensity, and although the dermal exposure route has been established as the most significant contributing route to internal dose33,34 and rate of dermal uptake varies by body part,33,48 less is known about using body part exposures as determinants in a questionnaire-based metric of exposure intensity. Finally, the variables of each algorithm are applied to self-reported information, and this leads us to consider how differences in the questionnaire data may have contributed to the findings of this study. The reliability of self-reported data can heavily influence the results because they are susceptible to reporting error, especially when subjects are asked to recall historical information.  So how reliably do farmers report information about their farm work? Blair et al. (2002) evaluated the reliability of self-reported ‘use of specific pesticides’ (ever/never mix or apply) and ‘application methods’ using repeated applicator interviews that were conducted 1-year apart, and reported that agreement was high (70-90%)   73 for these types of information.38 However, agreement between interviews was lower (generally 50-60%) for ‘duration’ (years mixed or applied), ‘frequency’ (days per year), and the ‘decade first applied’.38 Although agreement was high for application method and specific pesticide use information according to Blair et al. (2002), the interviews were only one year apart, but subjects in case-control studies are usually asked to recall information that extends far beyond a single year. Acquavella et al. (2006) also reported a need for additional information on the reporting reliability of duration and frequency of specific pesticide use.88 Additionally, they reported a noticeable difference in the reporting of ‘frequency of equipment repair’ and ‘use of personal protective equipment’ between trained observers and the farmers of the study sample.88 Overall there is a need for more information regarding the validity and reliability of self-reported farm work and practices. In terms of the potential dermal exposure algorithm, we could not find information about how well subjects report about body areas coming into contact with pesticides. Future studies focused on how these personal-level data are reported by subjects could provide information regarding the accuracy of the potential dermal exposure algorithm at estimating exposure intensity. Overall, the different questionnaire data upon which each algorithm depended could have contributed to the differences in exposure results observed in this study. The primary strength of this study is that it has applied and compared three different methods for assessing pesticide exposure among farm workers, and in doing so, the main strengths and weaknesses of each method have been outlined (Table 3.5). In addition, the key components to consider when developing, applying and describing an algorithm-based method that uses external exposure measurement data in combination with questionnaire data have been discussed. Lastly, this study developed a new algorithm that provides exposure estimates in units of mg/hour, which makes it easier to compare exposure estimates across studies. This method can be further evaluated to determine its ability to accurately estimate pesticide exposure intensity levels among farm workers. The primary weakness of this study relates to the small number of farm jobs that could be assessed for exposure and the large number of observations with incomplete information.   3.5 Conclusion  The results for the application of the simple surrogate approach in assessing farm job pesticide exposure in this study were as expected; it was easy to implement and less susceptible to observation loss compared to the two algorithm-based methods in this study. However, these results were expected because it’s an approach similar to those that have been commonly used in the past, but unfortunately, it does not attempt to measure exposure intensity and does not adequately address exposure variation. There were no expectations with regards to the results of the two algorithm-based methods, since these approaches have been used less frequently because epidemiological studies rarely collect the necessary   74 information to apply them. The result of no correlation between the exposure estimates of these two algorithms was both interesting and logical given differences in the exposure monitoring data used for their development, the variables included in each algorithm and the fact that they were defined based on different types of self-reported information. In essence, these methods likely measure different aspects of pesticide exposure. Moving forward, although the AHS algorithm has some supporting validity evidence,36,85 particularly when applied to the AHS pesticide applicator sample, further evaluations that focus on this methods ability to accurately assess exposure in case-control studies would be useful for understanding the context of its application. The potential dermal exposure algorithm requires much more evaluation, such as in studies comparing its exposure intensity estimates against biomonitoring and dermal dosimetry data. In general, the identification of the most important determinants of pesticide exposure in farmers, and characterizing the validity and reliability of self-reported information regarding such determinants is needed for this research field.   These exposure assessment methods will be used in an epidemiologic analysis based on this same case-control data to evaluate the relationship between cumulative farm job pesticide exposure and multiple myeloma.                 75 Chapter 4: Agricultural pesticide exposure and risk of multiple myeloma in a population-based case-control study in British Columbia, Canada  4.1 Introduction  Farming has been known to be associated with increased incidence of Multiple Myeloma for some time, with much of the research focused on exposure to pesticides. However, the epidemiological evidence has been inconsistent and there remains uncertainty regarding the specific farm exposures and the level of risk that are associated with this cancer.22 In fact, evaluating agricultural pesticide exposure in relation to any cancer type faces several challenges due to the significant variation in job factors that influence exposure – from the tasks performed and the frequency of performing those tasks to the specific types and combinations of chemicals that farm workers are exposed to, among other factors.30 As discussed in Kromhout and Heederik (2005) “measurement error in agricultural exposures can be expected to be substantial”, given the extent of exposure variation in this industry.30 For example, a commercial applicator may spray several types of chemicals throughout the year, a farm operator may spray one or two chemicals a few times a year, and a seasonal worker may not spray at all but may work on several farms (have potential to be exposed to various chemicals) for one season of the year.  An advantage of the case-control study design is the ability to compare the exposures of the farm workers to those of the general population, as this comparison can “result in considerable contrast between persons exposed and unexposed to agricultural exposures”.30 This contrast is more challenging to obtain in a cohort study of only agricultural workers. However, exposure misclassification remains a significant challenge, and as Blair et al. (2007) discuss, it’s likely to be present to some level in every epidemiologic study and even “relatively small errors can have sizable effects” on risk estimates.29 Some suggestions that may help to reduce the degree of exposure misclassification include using more information on determinants of agricultural exposures to “predict exposure more reliably” and to focus on “identifying persons with high exposures, since their outcome determines the slope of the exposure-response relation”. 30 Furthermore, considering the potential effect of exposure misclassification on study results when they are interpreted is also important when it comes to occupational epidemiology.29 Too often, confounding is considered to be “an explanation for positive findings without providing any information that the very specific conditions required for it to occur actually do”, whereas the likelihood of occurrence and the subsequent magnitude of effect that exposure misclassification can have on study results, is greater overall in occupational epidemiology.29  The aim of this study was to use data from a population-based case-control study of multiple myeloma that collected detailed information about farm work to evaluate the relationship between occupational   76 pesticide exposure and MM, using multiple exposure assessment methods. The two main methods were both algorithm-based, combining exposure determinants and external measurement data with self-reported farm job information, to estimate cumulative pesticide exposure in farm workers.  Five additional metrics were used to assign dichotomous exposure based on self-reported information about individual variables that were used as exposure surrogates.   4.2 Methods 4.2.1 Study Population  Epidemiologic data from the population-based MM case-control study (n = 773) based in British Columbia, Canada, were used to evaluate the relationship between agricultural pesticide exposure and Multiple Myeloma. The BC Cancer Registry was used to recruit cases (n=398) based on enrolment as newly diagnosed MM patients aged 35-79 years, between 2009 and 2013. Controls (n=375) were frequency matched on age and sex, and then randomly selected from the British Columbia Ministry of Health Registry; this is the population-based health insurance plan for BC residents, which contains identifying information. Exclusions for this study included subjects diagnosed with monoclonal gammopathy of undetermined significance (MGUS), which is a known risk factor for MM.  Questionnaires were either administered by mail or by telephone (computer assisted) using trained interviewers who spoke English and the languages of the major ethnic minorities in BC (Cantonese, Mandarin, Punjabi and Tagalog).  Demographic data were collected from these subjects, including family history of MM, age, sex, education level, ethnicity, and body mass index. In addition, self-reported information provided to a Family History and Residence/Occupation Questionnaire was used to identify subjects who performed farm work, specifically subjects who indicated having worked in agriculture, gardening, parks, golf courses or forestry in their occupational history. These subjects were provided with a follow-up questionnaire that collected details about the farm site (type of farm, size, location, general use of pesticides) and the farm job (tasks performed, handling of pesticides, types of pesticides, application methods, use of personal protective equipment (PPE), and usual contact of pesticides with body areas), see Appendix A.  4.2.2 Farm Workers  Among the 773 subjects, 124 (16% of the study sample) were given the farm module questionnaire. A total of 170 farm jobs were recorded among these 124 subjects, as some subjects held more than one   77 farm job.  Among these were six subjects who held one or more farm jobs described as ‘greenhouse’ or ‘garden’ work. Although the exposure assessment methods used in this paper may not be as accurate at estimating pesticide exposure for these jobs, they were included in the analyses since pesticide use was reported and the subjects should not be considered unexposed for these jobs. Note, these farm jobs were not included in the analyses for Chapter 3, due to differences in the aims of the studies; this, as well as the exclusion of MGUS subjects accounts for the discrepancy in subject and farm counts reported in Chapters 3 and 4.  Figure 4.1 shows the breakdown of the study sample into farm working groups based on self-reported information about farm work, farm pesticide use and personal involvement in pesticide handling, as this information was pertinent for the exposure assessment methods used in below analyses. Groups 1-3 are defined based on self-reported information about general pesticide use on the farm. Subjects in Group 1 (who responded affirmatively to farm pesticide use) were further broken down into Groups 1A-E based on their self-reported involvement in pesticide handling tasks (e.g., apply, mix-load). Subjects in Group 2 and Groups 1A-C provided ‘complete’ data for exposure assessment. However, subjects in Group 3, 1D and 1E were identified as providing ‘incomplete’ data, due to non-response (or “don’t know” responses) to pesticide use information; exposure assessment for these subjects with missing information is described below.               78 Figure 4.1: Multiple Myeloma Case-Control Study: Subject and Farm Job Counts Assessed for Cumulative Pesticide Exposure1  N = 773 Subjects (3,212 All Jobs) Farm Sample, n = 124 Subjects (170 Farm Jobs) Group 2. NEVER worked on a farm where pesticides were used:  n = 44 Subjects  (58 Farm Jobs) Group 3. UKNOWN if ever worked on a farm where pesticides were used: n = 28 Subjects (34 Farm Jobs )  Group 1. EVER worked on farm where pesticides were used: n = 52 Subjects (78 Farm Jobs)  Group 1A. Applied, mixed-loaded, and washed-repaired at farm job:  n = 22 Subjects  (32 Farm Jobs) Group 1B. Applied and mixed-loaded at farm job: n = 9 Subjects (12 Farm Jobs)  Group 1C. Applied only at farm job: n = 4 Subjects   (9 Farm Jobs) Group 1D. No response to  some tasks: n = 3 Subjects (7 Farm Jobs)  Group 1E. No response to any task: n = 14 Subjects (18 Farm Jobs)  Groups 1-3: Based on self-reported use of any pesticides on farm. Groups 1A – 1E: based on self-reported tasks personally performed for farm job (apply, mix-load, wash-repair). 1 The farm sample (n=124) identified from the Multiple Myeloma Study includes subjects who held more than one farm job, and at times, there were differences in the characteristics (pesticide use, tasks, etc.) reported across farm jobs held by the same subject. Therefore, subject counts of the defined groups in Figure 4.1 reflect a subject-level farm job summary (subject group assignment considered information reported across all farm jobs). The farm job counts for each group reflect what the subject specifically reported for the farm job.   79 4.2.3 Exposure Assessment  Two algorithm-based methods were used to estimate exposure intensity to pesticides for each farm job held by a subject; these methods have been described previously.84 Briefly, the first method was based on the application of an updated version of the Agricultural Health Study (AHS) Algorithm40 that used self-reported information about personally applying pesticides (including application method), and personally mixing pesticides and/or repairing equipment, as well as the general use of personal protective equipment, to relatively rank exposure intensity using an estimated ‘score’.40,80  Specific details regarding this updated AHS algorithm method are reported by Coble et al. (2011).  A second algorithm-based method, called the Potential Dermal Exposure Algorithm, was developed from a recent literature review of dermal exposure data.84 This algorithm used self-reported information about personal involvement in applying and mixing-loading pesticides in the farm job, as well as the body areas (head, arms, body, legs) that were reported as commonly coming into contact with pesticides during the performance of these tasks, to estimate the level of pesticide exposure intensity (mg/hour). Pesticide exposure intensity was estimated for all subjects and their reported farm job(s) for which ‘complete’ exposure data was provided (n = 79 subjects, 111 farm jobs), Groups 1A-C and Group 2 in Figure 4.1. 4.2.3.1 Cumulative Pesticide Exposure Once farm job-level exposure intensities were estimated using the algorithms, cumulative pesticide exposure across all farm jobs was determined for each subject. There were two main components needed to do this, and they included the estimated exposure intensity (described above) and the exposure time variables (frequency and duration) for all farm jobs held by a single subject. Equations 1 and 2 show how these two components were combined for the AHS and Potential Dermal Exposure Assessment methods, respectively, for all subjects who provided ‘complete’ exposure data, n= 79 subjects (111 farm jobs). Regarding the ‘frequency’ time variable, the farm module questionnaire permitted responses in the form of a specific number of days per year, or selection of a frequency category (< 10, 10-39, 40-69, 70-99, > 99 days per year). Using the responses that were provided in the form of a specific number of days per year, a mean frequency was determined for each of the above listed categories and used as the frequency value for subjects who reported a frequency category (rather than specified number of days per year) for their farm job.  For subjects who provided ‘incomplete’ exposure data, n= 45 subjects (59 farm jobs) Group 1D, 1E and Group 3 in Figure 4.1, imputation approaches were performed that varied by exposure assessment method, the component of the cumulative exposure algorithm for which the data were missing (exposure intensity or time-related variables), and the farm group assignment of the job (Figure 4.1). In general, the   80 observed control mean impute method89 was used to assign the necessary exposure information needed to estimate cumulative pesticide exposure for these subjects, combined with the partial data provided by subjects (e.g., self-reported information about one of the job tasks) to apply each of the above described algorithm methods. Table 4.1 summarizes the imputation methods used for the farm jobs held by these subjects, by the type of missing data and farm job group assignment.  Subjects who did not report a farm job were not provided with the farming questionnaire, and therefore, were automatically classified as unexposed and assigned a cumulative exposure value of ‘0’ for both algorithm-based methods. Equation 1. Cumulative Farm Job-specific Cumulative Exposure using the Agricultural Health Study (AHS) Intensity Algorithm (summed across all farm jobs held by subject): 1,2Cumulative AHS Farm Job Intensity (Score Days) =  [(ScoreMIX + ScoreAPPLY + ScoreREPAIR) * (FrequencyAPPLY(days/year) * DurationAPPLY(years))] * ScorePPE  Where, Duration = number of years applied pesticides for farm job Frequency = number of days of pesticide applications per year for farm job  1Cumulative exposure, as explained in Dosemeci et al. 2002 for the AHS study, is intended to represent a specific chemical; however, in this study, it will represent farm job; 2 Final values were divided by 365 days to convert units to Score Years.  Equation 2. Cumulative Farm Job-specific Cumulative Pesticide Exposure using the Potential Dermal Exposure Intensity Algorithm (summed across all farm jobs held by subject): 1,2Cumulative Farm Job Potential Dermal Exposure Intensity (mg/hr*days) = [MIXLOAD * (Head/FaceMIXLOAD(mg/hr) + BodyMIXLOAD(mg/hr) + Arms/HandsMIXLOAD(mg/hr) + Legs/FeetMIXLOAD(mg/hr)) + APPLY * (Head/FaceAPPLY(mg/hr) + BodyAPPLY(mg/hr) + Arms/HandsAPPLY(mg/hr) + Legs/FeetAPPLY(mg/hr))] * (FrequencyAPPLY(days/year) * DurationAPPLY(years))  Where, Duration = number of years applied pesticides for farm job Frequency = number of days of pesticide applications per year for farm job  1PPE is not included as a variable in the above method, due to the nature of the exposure data used that the algorithm is based upon (over clothing or on uncovered skin dermal measurements, see Chapters 2 and 3); however a PPE variable was considered in the regression model to examine its effect on MM risk; 2 Final values were divided by 365 days to convert units to mg/hr*years.    81 Table 4.1: Summary of the imputation methods used for farm job observations with missing data (total n= 59 farm jobs)1.   Farm Job Group with ‘Incomplete’ Data  PESTICIDE EXPOSURE INTENSITY VARIABLES   TIME VARIABLES  Agricultural Health Study & Potential Dermal Exposure Algorithms (Equations 1 and 2) Agricultural Health Study Algorithm (Equation 1) Potential Dermal Exposure Algorithm (Equation 2)  Type of Data Missing   Imputed Value  Type of Data Missing  Imputed Value  Type of Data Missing  Imputed Value  Groups 1D and 1E  (n=25 farm jobs)  All three tasks (Group 1E)  Control mean intensity score2   Both tasks  Control mean intensity level (mg/hr)2  Frequency of pesticide application   Mean frequency value (days/year)2  Some tasks (Group 1D)  Control mean task-specific score2  One task  Task-specific control mean intensity level (mg/hr)2  Duration of pesticide application  Average proportion of time (years) in farm job spent applying pesticides1, multiplied by number of years farm job was held  PPE use for all tasks Mean PPE reduction factor2   n/a n/a    Group 3  (n=34 farm jobs)  Farm pesticide use data and task data  Control mean intensity score3  Farm pesticide use and both tasks  Control mean intensity level (mg/hr)3  Frequency of pesticide application   Mean frequency value (days/year)2  PPE use for all tasks  Mean PPE reduction factor2  n/a  n/a  Duration of pesticide application  Average proportion of time (years) in farm job spent applying pesticides2, multiplied by number of years farm jobs was held  1 Note, some cells are denoted as ‘n/a’ because the PPE variable was not included as a parameter in the Potential Dermal Exposure Algorithm (Equation 2), so imputation methods did not need to be addressed regarding PPE for this algorithm method only; 2Based on all farm job classified in Groups 1A-C (n=53 farm jobs), Figure 4.1; 3Based on all farm jobs classified in Groups 1A-C and Group 2 (n=87 farm jobs), Figure 4.1.  4.2.4 Exposure Metrics  A total of 7 exposure metrics were used in this study to evaluate agricultural pesticide exposure in relation to MM.  Two of the metrics were defined as cumulative agricultural pesticide exposures, and they involved the use of the ‘Cumulative AHS Exposure Intensity’ and the ‘Cumulative Potential Dermal Exposure Intensity’ algorithms noted above (Equations 1 and 2). For these cumulative metrics, analyses were based on pesticide group (any pesticide, herbicides, and insecticides). There were too few subjects   82 who reported use of fungicides, so this pesticide group was not evaluated. For each cumulative analysis, exposure categories (low-medium exposure and medium-high exposure) were determined based on the median control exposure estimates that were greater than ‘0’.    The remaining five metrics were defined based on the following variables as surrogates for dichotomous exposure classification (ever/never): worked on a farm, worked on a farm where pesticides were used, personally used herbicides at a farm job, personally used insecticides at a farm job, and personal protective equipment (PPE) use for applying and/or mixing-loading pesticides.  4.2.5 Statistical Analysis  SAS Version 9.4 was used for analyses. Odds ratios (OR) and 95% confidence intervals were determined using unconditional logistic regression to evaluate the risk of MM in relation to estimated pesticide exposure, as determined by each exposure metric. The following demographic variables were considered as potential confounders in the regression models: sex, age, ethnicity, education, family history of MM, and body mass index (BMI). For each exposure metric, a full regression model was initially fit with all potential confounders, and the change in estimate approach was used to delete variables with the smallest change in the estimated exposure effect until all remaining variables in the model had at least a 10% change in the estimated exposure effect. A final model was fit for each exposure. Subjects who did not provide complete demographic data for the adjustment variables in each regression model were not included in the final analysis. For cumulative exposure, the algorithm for the AHS exposure metric included PPE use as part of the exposure estimation (Equation 1); however, this was not the case for the potential dermal exposure metric (Equation 2). Given this and the fact that only farm workers who were exposed could have used PPE, we examined PPE use as an effect modifier for the association of cumulative potential dermal pesticide exposure and MM by determining the ORs for farm workers who reported never wearing PPE when applying or mixing-loading pesticides at a farm job versus farm workers who reported wearing any type of PPE (overalls/coveralls, goggles, gloves, mask, full face shield, respirator/gas mask, boots and/or hat) when performing one of these tasks at a farm job.  The referent exposure for both cumulative metrics were subjects with an estimated exposure of ‘0’, which included subjects who never held a farm job (n= 649) and subjects who worked at a farm job where pesticides were not used (n=44, Group 2, Figure 4.1). For the potential dermal exposure metric, there were 4 subjects who worked at farm jobs and personally handled pesticides (Group 1, Figure 4.1) who were additionally included in the referent exposure group because their estimated cumulative potential dermal exposure was ‘0’ when the algorithm was applied (Equation 2).    83 For the following dichotomous metrics, the referent exposure was defined as above (i.e., non-farm workers and subjects who worked at a farm job where pesticides were not used, Group 2):  ‘ever worked on a farm where pesticides were used’, ‘ever personally applied and/or mixed-loaded herbicides at a farm job’ and ‘ever personally applied and/or mixed-loaded insecticides at a farm job’. For the ‘PPE use for applying and/or mixing-loading pesticides’ dichotomous metric, the referent exposure group additionally included subjects who were eligible for the PPE questions in Group 1A-D and responded “no” to PPE use for all farm jobs held (n=14). For the ‘ever worked on a farm’ dichotomous metric, the referent exposure simply included subjects who never held a farm job (n=649). Although farm job exposures were imputed for subjects in Groups 3 (n=28) and Group 1E (n=14) of Figure 4.1, as shown in Table 4.1, these subjects were excluded from all cumulative analyses and all dichotomous exposure analyses for which the necessary data were not complete. Specifically, Group 3 was excluded from all analyses except for the ‘ever worked on a farm’ metric, and Group 1E was excluded from all analyses except for ‘ever worked on a farm’ and ‘ever worked on a farm where pesticides were used’ metric. These subjects were excluded from analyses in which they did not provide complete data in order to more accurately estimate the OR; i.e., for the exposure group to represent as accurately as possible the “truly exposed” and for the unexposed group to represent as accurately as possible the “truly unexposed”. The three subjects in Group 1D (Figure 4.1) were included in all analyses, although they did not provide complete data to all task questions and their cumulative exposure estimates were partially based on imputed values (Table 4.1), because they provided an affirmative response to personally performing at least one of the three tasks and could be confidently classified as exposed. There were also two subjects in Group 1A who held farm jobs that were classified in different farm jobs groups; 1 subject held a farm job classified in Group 1A and another classified in Group 1D, while the second subject held a farm job classified in Group 1A and another classified in Group 3. For this first subject, cumulative exposure estimates were maintained, although they were partially imputed (Table 4.1) due to missing data for the farm job in Group 1D. For the second subject, the cumulative exposure estimate was adjusted to only reflect exposure to the farm job classified in Group 1A (i.e., exposure for farm job in Group 3 was set = 0) in this study to be consistent with the exclusion of subjects who held farm jobs classified in Group 3.  4.3 Results  A total of 773 subjects, 398 cases and 375 controls, were part of MM case-control study and they were frequency matched by sex and age group.  Table 4.2 shows case and control counts by demographic variables considered for adjustment in the regression analyses. The majority of subjects were ‘White’, followed by ‘Asian’ ethnicity. More than 50% of the cases and controls achieved at least 2 years of post-  84 secondary education, the majority of subjects did not have a family history of MM, and the greatest proportion of cases and controls were classified as ‘overweight’.   Table 4.2: Demographic characteristics for multiple myeloma case-control study sample (N=773).  Demographic Characteristic   MM Cases (N=398)   Controls (N=375)  Sex    Male  235  213 Female  163 162  Age Categories (years)    <45  11 4 46-50  24 10 51-55  45 25 56-60  58 57 61-65  82 84 66-70  76 83 71-75  51 67 >75  51 45  Ethnicity    White  313 330 Asian  37 24 Hispanic/Black/Other  39 19 Unknown  9 2  Education    Less than Secondary School   64 52 Secondary School Completed   75 86 2 years post-Secondary School  128 106 4 years or more of post-Secondary School  121 130 Unknown  10 1  Family History of Multiple Myeloma    Yes  14 7 No 366 358   85   Unknown  18 10  Body Mass Index (BMI)    Underweight or Healthy Weight: <25.00 89 74  Overweight/Pre-Obese: 25.00-29.99 186 171  Obese Class I: 30.00-34.99 77 90  Obese Class II or III: ≥ 35 31 39  Unknown 15 1   There were 124 subjects identified as having held at least one farm job, with a greater proportion of these subjects being controls (Table 4.3). Table 4.3 shows similar counts of cases and controls by many of the farm characteristic variables. Some differences that stand out include: non-response observations were more common among control subjects for all variables in which an ‘unknown’ category is shown in Table 4.3 (i.e., ever worked on a farm where pesticides were used, ever personally used pesticides at a farm job, and ever used PPE when applying and/or mixing-loading), fungicides were the pesticides type used the least among both cases and controls, a greater number of cases reported using phenoxy herbicides and organophosphate insecticides compared to controls, control subjects were more likely to have worn PPE when applying pesticides and cases were more likely to have worn PPE when mixing-loading pesticides. But overall, more cases reported ‘no’ to PPE use, while a greater number of controls reported ‘yes’ to PPE use of any type.  Table 4.3: Farm and pesticide use characteristics among subjects in the multiple myeloma case-control study (N=773).  Farm Work and Pesticide Use Characteristics   Cases N=398  Controls N=375  Ever Worked on a Farm Yes    54  70 No    344 305  Ever Worked on a Farm Where Pesticides were Used  Yes   24 28  No    19 25 Unknown  11 17      86  Ever Personally Used Pesticides at a Farm Job  Yes   19 19    Pesticide Use Types Ever Used at Farm Job1   Herbicides  16 13   Insecticides  18 14   Fungicides  7 4    Pesticide Chemical Classes Ever Used at Farm Job1   Phenoxy Herbicides  (2.4-D, MCPA) 8 5  Phosphonoglycine Herbicide (Glyphosate) 4 6  Organochlorine Insecticides (DDT, Deldrin, Lindane) 6 6  Organophosphate Insecticides  (Malathion, Diazinon, Guthion) 5 2   No    0   0 Unknown   5 9  Ever Used Personal Protective Equipment (PPE) when Personally Handling Pesticides   Yes  10 13    Farm Job Tasks2   Applying  8 13   Mixing-Loading  10 7  No    8 6 Unknown   6 9 1The subject counts in the ‘pesticide use type’ and ‘pesticide chemical class’ categories will sum beyond the total case (n=19) and control (n=19) counts that reported having ever personally using pesticides at farm job(s) because many subjects used pesticides of more than one ‘use type’ and/or ‘chemical class’; 2 Note that the majority of subjects that reported personally having used pesticides, were involved in both applying and mixing-loading tasks.      87 4.3.1 Pesticide Exposure Analyses 4.3.1.1 Dichotomous (Ever/Never) Pesticide Exposure The odds ratios, representing the association between MM and self-reported exposure as defined by the five dichotomous exposure metrics, are presented in Table 4.4. Based on the change in estimate approach for variable selection, the models for the first and fourth exposures shown in Table 4.4 were not adjusted for any of the potential confounders, whereas the models representing the other dichotomous exposures were adjusted for either family history of MM or education level. None of the associations were statistically significant. The metrics that represented ‘farm work’ or ‘working on a farm where pesticides were used’ had negative associations with MM, opposite to expectation. The only two positive, but weak associations were observed for ‘personally used herbicides at farm job’ (OR= 1.07, 95% CI= 0.50-2.27) and ‘personally used insecticides at farm job’ (OR= 1.17, 95% CI= 0.57-2.39), but again, these were not statistically significant. For a subject to have been identified as having personally used herbicides or insecticides at a farm job, they only needed to report personal use of a single herbicide or insecticide chemical for one farm job. There were not enough subjects to perform analyses for fungicide use or any of the pesticide chemical classes shown in Table 4.3.  In regards to the PPE use metric (‘PPE use for applying and/or mixing-loading at farm job’), subjects who ever used PPE had a 34% lower odds of MM when compared to subjects who did not use PPE, were not eligible for the PPE questions (not farm workers), or did not work on farms where pesticides were used (OR= 0.66, 95% CI= 0.29-1.54).         88 Table 4.4: Odds ratios (OR) and corresponding 95% confidence intervals (CI) for multiple myeloma (MM) in relation to agricultural pesticide exposure as determined by five dichotomous exposure metrics, among subjects in the case-control study conducted in British Columbia, Canada.  Dichotomous Exposure Metrics MM Cases  Controls  OR 95% CI  Worked on a Farm4     Never1 344 305 1.00 -- Ever 54 70 0.68 0.46 – 1.01 Worked on a Farm Where Pesticides Were Used5     Never2 345 322 1.00 -- Ever 24 26 0.80 0.44 – 1.43 Personally Used Herbicides at Farm Job6     Never2 353 329 1.00 -- Ever 16 13 1.07 0.50 – 2.27 Personally Used Insecticides at Farm Job4     Never2 363 330 1.00 -- Ever 18 14 1.17 0.57 – 2.39 Personal Protective Equipment (PPE) use for Applying and/or Mixing-Loading Pesticides at Farm Job6     Never3 361 335 1.00 -- Ever 10 13 0.66 0.29 – 1.54 1 Referent exposure group includes all subjects who did not hold any farm jobs; 2 Referent exposure group includes all subjects who did not hold any farm jobs as well as subjects who held farm jobs, but reported “no” to working on a farm where pesticides were used (Group 2, Figure 4.1); 3 Referent exposure group includes all subjects who were not eligible for questions about PPE use (non-farm workers and farm workers who worked on farms where pesticides were not used, Group 2), as well as subjects who personally handled pesticides and reported “no” to PPE use questions for all farm jobs;   4 Not adjusted for any potential covariates; 5Adjusted for family history of Multiple Myeloma; 6 Adjusted for education level.  4.3.1.2 Cumulative Pesticide Exposure Cumulative exposure analyses are presented by metric and by type of pesticide (‘any pesticide use’, ‘herbicide use’ and ‘insecticide use’) in Table 4.5. The potential dermal exposure metric typically resulted in larger ORs across the pesticide analyses, compared to the AHS exposure metric (Table 4.5). There were no statistically significant associations for the two cumulative metrics, even when broken down by pesticide use group. The largest OR was for ‘herbicide use’, estimated using the cumulative potential dermal exposure method (Equation 2); this OR indicated a 2.4 times greater odds of MM in subjects classified with ‘medium-high’ cumulative herbicide exposure compared to the unexposed. For this same metric, the ORs for ‘insecticide use’ exposure were positive for both exposure groups (‘low-medium’ and   89 ‘medium-high’) when compared to the unexposed. However, as mentioned above, the algorithm used for the potential dermal exposure metric did not include PPE in the exposure estimation process, whereas the algorithm for the AHS metric did (Equations 1, 2).  To assess the effect of PPE with the potential dermal exposure metric, a PPE use variable (defined as ever use/never use) was added to the regression models. The ORs for when ‘PPE was not used’ and when ‘PPE was used’ were determined for each pesticide analysis and exposure level (Table 4.6). In comparing these ORs, the most profound effects were for ‘herbicide use’ (Table 4.6). The associations between herbicide exposure and MM were strongly positive when PPE was not used, significantly so at the medium-high exposure level. There was a clear difference in effect when PPE was used: the ORs indicated non-significant associations, either negative (low-medium exposure) or weakly positive (medium-high exposure). There were similar but less dramatic differences in MM associations with and without PPE for any pesticide use and insecticide use.             90 Table 4.5: Odds ratios (OR) and corresponding 95% confidence intervals (CI) for multiple myeloma (MM) in relation to agricultural pesticide exposure as determined by two cumulative exposure metrics and pesticide group (any pesticide, herbicides, insecticides) among subjects in the case-control study conducted in British Columbia, Canada. Cumulative Exposure Metrics MM Cases  Controls  OR 95% CI  AHS Exposure Intensity Level (score years)1  Any Pesticide Use2     0  (unexposed) 336 321 1.00 -- >0 – 18.4 (low-medium exposure) 13 9 1.23 0.51 – 2.95 >18.4 – 336.0   (medium –high exposure) 6 9 0.51 0.18 – 1.50  Herbicide Use3     0  (unexposed) 353 329 1.00 -- >0 - 5.9 (low-medium exposure) 3 6 0.46 0.12 – 1.88 >5.9 – 260.5   (medium –high exposure) 13 7 1.57 0.61 – 4.01  Insecticide Use3     0  (unexposed) 353 329 1.00 -- >0 - 30.7 (low-medium exposure) 13 7 1.66 0.65 – 4.22 >30.7 - 336.0  (medium- high exposure)  5 7 0.61 0.19 – 1.95  Potential Dermal Exposure (mg/hr*years)4  Any Pesticide Use5     0  (unexposed) 345 326 1.00 -- > 0 – 4.3  (low-medium exposure) 10 7 1.26 0.47 – 3.38 > 4.3 – 47.2  (medium-high exposure) 9 7 1.12 0.41 – 3.09  Herbicide Use6     0  (unexposed) 330 324 1.00 -- > 0 – 1.0  (low-medium exposure) 3 5 0.50 0.11 – 2.20 > 1.0 – 39.5  (medium-high exposure) 13 5 2.37 0.82 – 6.85  Insecticide Use5     0  (unexposed) 345 323 1.00 -- > 0 – 4.0 (low-medium exposure) 10 6 1.41 0.50 – 3.99 >4.0 – 47.2 (medium-high exposure) 8 6 1.26 0.43 – 3.68 1 Referent exposure includes all non-farmworkers and all subjects who worked on a farm but reported “no” to pesticide use on farm (Group 2, Figure 4.1); 2Adjusted for education level and family history of Multiple Myeloma; 3 Adjusted for education level; 4 Referent exposure includes non-farmworkers, subjects who worked on a farm where no pesticides were used (Group 2, Figure 4.1), as well as 4 farm working control subjects whose estimated potential dermal exposure level was ‘0’; 5 Adjusted for family history of Multiple Myeloma;6 Adjusted for age, education level, family history of Multiple Myeloma, and body mass index.   91 Table 4.6: The effect of PPE use as a modifier of the relationship between cumulative potential dermal exposure and multiple myeloma is shown using the Odds Ratios (ORs) for ‘PPE not used’ and ‘when PPE used’ compared to the unexposed, by pesticide group (any pesticide, herbicides, and insecticides) and exposure level group.   Potential Dermal Exposure (mg/hr*years) by Pesticide Group and Exposure Level Group1   OR when PPE not used  (95% CI) OR when PPE used (95% CI) Any Pesticide Use   Low-Medium Exposure  (>0 – 4.3) 2.05 (0.52 – 8.11) 0.86 (0.29 – 2.56) Medium-High Exposure (>4.3 - 47.2)  1.77 (0.51 – 6.17) 0.74 (0.22 – 2.46) Herbicide Use1   Low-Medium Exposure (>0 – 1.0) 3.28  (0.26 – 41.6) 0.38 (0.08 – 1.85) Medium-High Exposure (>1.0 – 39.5) 11.7 (1.28 – 106*) 1.35 (0.41 – 4.43) Insecticide Use   Low-Medium Exposure (>0 – 4.0) 1.47 (0.37 – 5.77) 1.12 (0.34 – 3.74) Medium-High Exposure (>4.0 – 47.2) 1.48 (0.38 – 5.69) 1.13 (0.33 - 3.85) *p-value < 0.05 1 The effect modification for PPE use was significant at p<0.01  4.4 Discussion 4.4.1 Key Findings  This study evaluated the relationship between agricultural pesticide exposure and Multiple Myeloma using multiple metrics for exposure. For the dichotomous metrics, neither ‘farm work’ nor ‘working on a farm where pesticides were used’ were positively associated with MM. However, non-significant positive associations were observed for the dichotomous metrics that identified subjects who ‘personally used herbicides’ and subjects who ‘personally used insecticides’. Regarding cumulative exposure analyses, the AHS exposure intensity metric did show some positive associations (‘low-medium’ exposure to any pesticides, ‘low-medium’ exposure to insecticides, and ‘medium-high’ exposure to herbicides), but none were statistically significant. For the potential dermal exposure metric, most of the associations were positive, although none were statistically significant until PPE use was included as an effect modifier in the models. When PPE was not used, a statistically significant association was observed for ‘medium-high’ herbicide exposure (OR=11.7, 95% CI: 1.28-106). The effect modification of PPE on the association between medium-high herbicide exposure and MM was strong; when PPE was not used by   92 farm workers exposed to herbicides, there was a 12 fold increase in MM risk compared to less than a 2 fold increase in MM risk when PPE was used.   4.4.2 Comparisons of Findings to Other Studies  The results in this study for ‘any pesticide use’ (Table 4.5) were predominantly positive non-significant associations; the ORs were larger when broken down by pesticide group (herbicides, insecticides). It is important to conduct pesticide-specific analyses when possible, since  grouping all pesticides together when assessing exposure, as was done in the ‘any pesticide use’ analyses, is likely to lead to exposure misclassification that will affect the size and direction of the measures of association. In a study by Kachuri et al. (2013) evaluating pesticide-MM relationships, exposure was assessed using two different exposure metrics (‘number of pesticides used’ and ‘days per year of pesticide use’) as a proxy for lifetime use of multiple pesticides, and the positive results were only statistically significant when pesticides were grouped by type.24 Kachuri et al. (2013) grouped pesticides by carcinogenic potential, chemical class, as well as by individual active ingredient for analyses when possible. If we had larger samples, it would have been ideal to at least examine the associations by chemical class. In Table 4.3, along with the case-control counts by pesticide use group, we showed subject counts by commonly used pesticide chemical classes. The most commonly reported active ingredients for herbicides (2,4-D and MCPA, which are both phenoxy herbicides; glyphosate, which is a phosphonoglycide herbicide) and insecticides (Malathion, Diazinon and Guthion for organophosphates; DDT, Dieldrin and Lindane for organochlorines) were also noted in Table 4.3. In the pesticide-specific analyses performed by Kachuri et al. (2013), excess risks were observed among Canadian men who reported using at least one carbamate insecticide (similar mode of action to organophosphates, but chemically different), at least one phenoxy herbicide, and 3 or more organochlorine insecticides.24 Although not consistent across all exposure level groups, the positive associations that were observed in our study for cumulative exposure to herbicides and insecticides using both metrics (Table 4.5), may be due to the dominating use of the above specified pesticides (i.e., phenoxy herbicides, organochlorine and organophosphate insecticides) as reported in Kachuri et al. (2013). In a study by Presutti et al. (2016), three case-control studies from the U.S. and Canada were pooled to evaluate the association between pesticide use and MM among men. In this pooled analysis, three pesticide metrics were used:  ‘dichotomous’ (ever/never used), ‘duration of use’ and ‘cumulative lifetime days’.23 The main findings included an increased risk of MM for ‘ever use’ of carbaryl, captan and DDT.23 DDT was the most commonly reported organochlorine in the current study as well, and its use may have contributed to the resulting positive association observed for most analyses performed by ‘insecticide use’, regardless of the metric used. Neither Kachuri et al. (2013) or Presutti et al. (2016) incorporated PPE use into the exposure metrics or analyses. Published data from   93 case-control studies that were included in a systematic literature review by Perotta et al. (2008) reported similar findings to those described above where excess risks of MM were observed for DDT, chlorophenols and phenoxy-acetic acids.22  Our study observed a negative association between farm work and MM (Table 4.4), based on the dichotomous metric (ever/never farm work), but this is contrary to findings from other case-control studies. The systematic review by Perotta et al. (2008) reports that epidemiologic studies over the last few decades have shown “consistent evidence” regarding the association between farm work and MM, but that this is not a strong association according to a pooled OR of 1.39 (95% CI=1.18-1.65) from more recently published case-control studies that aimed to “identify possible agricultural exposures”. 22 Perrotta et al. (2008) identified the following contributing factors as rationale for the positive, but weak association between farm work and MM: “study design, adjustment for level of education in the individual studies estimates and the different farming techniques and pesticide use around the world and across the decades”.22 “Across the decades” is an interesting and potentially important factor to consider with regards to exposure. Information about exposure years is rarely reported in epidemiologic studies, but could provide additional information useful for exposure assessment given changes in agricultural and regulatory practices, as well as differences in the years of use for individual pesticides. In this study, the range of exposure years of the farm sample were 1931-2014 for herbicide use and 1938-2014 for insecticide use. There are certain to be factors related to changes in the farm industry that are likely contributing to exposure variation across nearly 9 decades of farm work represented by this study sample. For example, regarding registration information for some of the specific pesticides used by subjects in this study: 2,4-D was first registered for use in Canada in 1946 and continues to be used today;90 DDT (along with other organochlorine insecticides) was heavily used for agriculture in the mid-1940s to late 1960s and then all of its uses were banned in Canada in 1985;91 and glyphosate (the active ingredient in the well-known herbicide product called Roundup® was first registered for use in Canada in 1976 and remains registered for use today.92 This provides an example of the importance of understanding the “decades” of exposure represented by the study population as many changes can occur across a wide range of exposure years, i.e., insecticide exposure in 1930 is not the same as insecticide exposure in 1960 simply based on chemicals available for use. This can influence how an exposure determinant may need to be characterized, and in this case, highlights the importance of performing chemical-specific analyses. Additionally, considering the potential influence of latency on the exposure-response relationship may also be important.      94 4.4.3 Limitations and Strengths  Overall, this study was limited by its small sample of farm workers, which likely influenced the confidence intervals above (Tables 4.4 to 4.6) and did not allow for more specific pesticide exposure analyses (i.e., chemical class or active ingredient) to be performed.  Approximately 30% of the farm workers in this study did not provide responses to exposure-related questions that were needed for applying the exposure assessment methods (Groups 3 and 1E), among whom 40% were cases and 60% were controls. Differential reporting bias may have played a role, possibly due to the length and detail level of the questionnaires, but we do not think that it was a primary reason for non-response rates in this study. There were no proxy responses, and further efforts were not made to obtain missing information from the study subjects at the time of the study. Complete data for a greater proportion of the farm sample would have provided more reliable estimates and consequently, a better representation of the exposure variation. Whether it would have yielded different results is unknown, but certainly possible. In addition, many analyses focused on ‘any pesticide use’ and it’s likely that these results were heavily influenced by non-differential exposure misclassification, given that not all pesticides are likely to cause cancer and there was a wide variety of pesticide chemicals reported across the farm workers. Lastly, the PPE variable used in the model with the cumulative potential dermal exposure metric was only broadly defined based on “ever” use of any type of PPE versus “never” use of any PPE; it did not consider the type of PPE used (e.g., chemical resistant gloves or boots) or whether PPE was used for all farm jobs held by a subject. However, this PPE variable still showed to have a strong effect on the exposure-response relationship for this analysis, which leads to the possibility it may have been indicative of something else, e.g., possibly a marker representing the “era” of exposure given the wide range in exposure years of the study population. Among the farm workers who reported handling pesticides and across all decades of farm work represented by this sample, we calculated the proportion of workers in each decade who reported using PPE (1930/40, 1940/50, 1950/60, etc..). We found that 1947 was the first year that any worker who handled pesticides reported using PPE, and those who worked prior to 1947 began working on a farm and handling pesticides during their childhood which may have also influenced PPE use. From 1950 to1990, the proportion of farm workers who worked at a farm job handling pesticides and reported using PPE, consistently increased across the decades for this study sample: 5 out of 15 farm workers for 1950-60 up to 16 out of 21 farm workers for 1980-90 ( this was the highest decade-specific proportion for PPE use that was observed). The proportions began to decrease for the decades that followed 1990. There may be some evidence here that the PPE use variable could be indicative of the decade of exposure or possibly age at time of exposure. Future retrospective analyses should consider both of these factors when evaluating PPE use among a study sample in which the years of exposure extend far into the past, as in this study.     95 4.5 Conclusion  Contrary to expectations, a negative association between farm work and MM was observed in this study. The main cumulative exposure analyses (Table 4.5) resulted in some non-significant positive associations between pesticide exposure and MM, especially for exposure determined by the potential dermal exposure metric. Analyses broken down by ‘herbicide use’ and ‘insecticide use’ provided slightly stronger measures of association for the potential dermal exposure metric, but they remained non-significant.  Unlike the AHS exposure intensity metric, ‘PPE use’ was not included as an algorithm parameter for cumulative exposure estimation using the potential dermal exposure metric. However, when ‘PPE use’ was included in the regression model for cumulative analysis using the potential dermal exposure metric, it showed to have a strong effect on the exposure-response relationship. This effect was particularly strong for the association between herbicide use and MM when PPE was not used by farm workers. The fact that PPE showed to have a strong effect on this association, although it was very simply defined as ever/never use, suggests that it may be indicative of something else among the farm workers (e.g., “era” of exposure, worker behaviour, level of training, etc.) whose exposure years spanned across 9 decades. Future studies should examine the effect of PPE use as a determinant of exposure for comparison to the findings in this study. The fact that PPE was not incorporated in the potential dermal exposure algorithm is what allowed us to easily evaluate its effect on the exposure-response relationship, whereas this was not possible with the AHS exposure intensity algorithm since PPE was included in the algorithm itself. The inclusion of multiple determinants within a single metric makes it difficult to see the effect of an individual factor, making it beneficial to perform analyses by the determinants themselves.                                                                                                                                                          96 Chapter 5: Conclusion  The theme of this dissertation is the need for improved exposure assessment methods to study pesticide-cancer relationships related to farm work, particularly for a specific cancer, multiple myeloma (MM). The research led to new information for direct improvements in this field, from the development of the exposure assessment method to its application in retrospective epidemiology. Additionally, a new exposure assessment method was developed by this research that can be evaluated and used in future studies to estimate pesticide exposure intensity levels for farm work in relation to any type of cancer.  The main findings of this dissertation are summarized for each of its component studies (Chapters 2-4), and in context of the current state of knowledge.  5.1 Summary of Findings in Light of the Current State of Knowledge 5.1.1 Literature Review: dermal monitoring data for pesticide exposure assessment of farm workers  The first research objective was to conduct a systematic review to determine the types of information the existing pesticide exposure monitoring literature could provide for exposure assessment in occupational epidemiology. Cancer studies are typically retrospective in design because the disease is rare. Such studies often rely on simple surrogates to assign exposure status (e.g., job title), but these have been shown to be “poor indicators” .32 It would be ideal to have quantitative measures of exposure for each study subject, but in retrospective studies, one of the primary reasons this is difficult or impossible is the long latency period for MM and other cancers. Direct measurements (e.g., biomonitoring samples) taken of subjects at time of study may not reflect long ago pesticide exposures that contributed to disease onset, since many commonly used pesticides have short half-lives36. Therefore, the rationale behind this research objective was to understand if and how reliably dermal measurements from the literature could be combined to provide information on determinants of pesticide exposure and to derive body area-specific estimates to be used for exposure assessment in retrospective studies of farm workers. The primary finding was the need for standardized dermal data reporting procedures across the exposure monitoring literature. This review suggested that future studies provide the raw measured data for each body part, and consistently report details about how these data were measured and what they represent, including: units, exposure time, body parts measured, pesticide(s) used, specific jobs and tasks performed, and any other work factors that characterize the exposure scenario and may influence dermal pesticide exposure (e.g. equipment used). Such standardization will allow for more reliable   97 dermal data combining and comparison between studies, and therefore, the development of more accurate quantitative exposure assessment methods for retrospective epidemiology.84 . However, as a result of the data reporting variation in the current literature, dermal data could only be combined for a subset of studies of this review to estimate quantitative dermal pesticide exposure levels for some farm jobs. . More specifically, dermal data from fewer than half of the studies were combined to develop an algorithm to estimate average total body exposure levels for pesticide handlers; this algorithm was then applied for the second research objective (Chapter 3). Study comparison limitations did not permit the evaluation of exposure determinants, nor did it allow for the algorithm to include non-pesticide handling farm jobs. Although not evaluated, this review did identify some potential determinants by farm job(see Chapter 2, Table 4.5 for details), such as ‘task’, ‘equipment type’ and ‘application method’ for farm workers involved in personally applying and mixing-loading pesticides.   The exposure monitoring literature is a foundation upon which we can improve our understanding about determinants of pesticide exposure and carry forward quantitative measurements to improve exposure assessment methods for retrospective epidemiology, so the findings presented here should be considered for future studies on this topic.  5.1.2 A comparison of methods for estimating pesticide exposure for farm jobs identified from a multiple myeloma case-control study  For the second research objective, farm job exposure estimates were compared across three different methods, using data from a population-based case-control study of MM. As discussed in Chapter 3, these included a simple surrogate method (self-reports of farm pesticide use) and two quantitative algorithms that provided estimates of exposure intensity, the updated Agricultural Health Study (AHS) ‘general’ algorithm40,80 and a potential dermal exposure algorithm developed from studies of the Chapter 2 literature review. The exposure determinants of the AHS ‘general’ algorithm had been previously applied and evaluated against different types of exposure measures (e.g. biological and passive dosimetry measures) that provided some validity evidence in regard to its ability to accurately rank-order pesticide exposure intensity for pesticide applicators39,40 (See Chapter 3 for algorithm details). For this reason, the AHS ‘general’ algorithm was considered an important comparison method by which to evaluate the newly developed potential dermal exposure algorithm developed for this work. However, no correlation between these algorithms was found when their farm job exposure estimates were compared, which suggested that they were measuring different aspects of pesticide exposure.  Details regarding some of the validity and reliability evidence of the AHS ‘general’ algorithm will not be repeated here, because these details can be found in the Chapter 3. Instead this section is focused on   98 the validity and reliability regarding the exposure determinants and the self-reported data of the potential dermal exposure algorithm to understand its future applications, limitations and what further information can be used to improve this method. As shown in the algorithm itself in Chapter 3, the main exposure determinants included pesticide handling tasks (apply and mix-load) that were based on self-reported information from study subjects about whether one or both of these tasks was performed, and additionally defined by a time proportion weight (further details in Chapter 3).  Pesticide exposure has been shown to be related to these handling tasks,80,93 and these self-reported occupational history details are likely reported well by farm workers since they would have been directly involved in performing the task. In fact, asking directly about performing these tasks is likely to be more reliable than using job title since job terminology is not always consistent and may not clearly indicate the performance of these tasks. In retrospective studies, if there is a long duration between the time of study and the time the farm job was held and/or if a job was held for a short duration, the validity and reliability of this self-reported task information may be reduced.41 However, it’s unlikely that the validity and reliability of self-reported task information was compromised by these factors much, because as explained by Blair and Zahm (1993), farmers tend to recall detailed information about their tasks, such as specific pesticide chemicals used, reasonably well and regardless of disease status.81 The task-specific body area exposure variables of this algorithm were defined based on self-reports about body areas that usually came into contact with pesticides during the performance of either one or both pesticide handling tasks (see Chapter 3 for details and Appendix A, questions 13 and 18). However, there exists little information about the validity and reliability of these body area exposure variables as determinants of exposure, as well as validity and reliability of the self-reported information upon which they were defined.. As mentioned in Chapter 2, we know that the dermal route contributes the most to internal dose33,34 and that absorption rates vary by body part,33 so it’s reasonable to consider using body area-specific exposure estimates in an algorithm to estimate total body exposure levels (mg/hr). However, it’s important to note acknowledge that the self-reported information used to define these body area exposure variables represented “usual” exposure rather than variation in body area contact over time, which can be expected with self-reported exposure information. 41 Additionally, these types of questions are more likely to elicit a subjective response, and consequently are more susceptible to recall bias as compared to questions about the types of tasks performed.41. It would be reasonable to assume that cases may be more likely to recall usual contact with pesticides compared to controls, and this could lead to differential bias that could push risk estimates away from the null hypothesis, resulting in false-positive associations in an epidemiologic analysis.81 Details regarding the dermal exposure monitoring data that was used to develop the weights of the task-specific body area estimates can be found in Chapter 3. These task-specific body area weights can easily be refined with new data from applicable studies.     99 Application of this method to future studies should be mindful of the potential lack of validity and reliability regarding the self-reported information used to define the body area exposure determinants of this algorithm. However, this algorithm may provide more accurate estimates of exposure if applied to epidemiologic studies that have evaluated the validity and reliability of self-reported information about the body areas that usually come into contact with pesticides during handling tasks. In regards to the case-control data used in this dissertation, the method best suited for accurately estimating pesticide exposure intensity remains unknown given the differences between the two algorithms compared (see Chapter 3 for details) and the lack of evidence regarding the validity and reliability of the potential dermal exposure algorithm.  5.1.3 Agricultural pesticide exposure and risk of multiple myeloma in a population-based case-control study in British Columbia, Canada  The final objective was to conduct an epidemiologic analysis to evaluate the relationship between pesticide exposure and MM among farm workers identified from a case-control study, and using multiple metrics for exposure. Contrary to expectations, farm work was not associated with MM, which is inconsistent with previous findings reported in the literature.4,22  In this study the simple metric ‘farm work’ (ever/never) was determined using self-reports to “did you ever live on a farm, or work in agriculture, gardening, parks, golf courses or forestry?”. The activities reported among those who responded “yes” to this question were examined, and the majority reported working on a farm (many also lived on the farm) and very few subjects reported other non-farm activities (see Chapter 4 for details). However, it’s possible that this sample was more heterogeneous in terms of the farm work performed compared to the samples assessed in other studies. For example, the evidence regarding pesticide-cancer relationships for farm workers in the AHS is based on pesticide applicators only.26 It is also possible that the years of exposure, which spanned almost nine decades for this case-control sample, may have contributed to an obscured association due to heterogeneity in farm work and practices over time. It is important to interpret results of epidemiologic studies in light of the ‘years of exposure’ covered, so that interpretations can consider changes in exposure, agricultural practices and policies for the same period of time.  Although farm work defined by a self-reported yes/no response alone did not have a positive association with MM. However, positive associations were observed using cumulative exposure metrics based on the two algorithms compared in Chapter 3 (i.e., the AHS ‘general’ algorithm and the potential dermal exposure algorithm), especially when pesticides were broken down by type of use group (herbicides and insecticides).  Although small sample sizes did not permit analyses by type of pesticide chemical or chemical families, the herbicides and insecticides most frequently reported in the study sample of this   100 research (e.g., specific chemicals from the phenoxyacetic acid herbicide and organochlorine insecticide families) corresponded with those that have been found in previous studies (see Chapter 1) to suggest an excess risk of MM among farm workers.18,20,22 The most interesting finding from this study was the modifying effect of PPE use on the association between pesticide exposure and MM, as estimated by the potential dermal exposure intensity metric (see Chapter 4). This modifying effect was particularly strong with regards to the association between ‘herbicide use’ and MM risk, indicating a significant excess risk in MM among farm workers who did not use any type of PPE and were classified in the highest pesticide exposure group, OR=11.7 (95% CI: 1.28-106). In contrast, farm workers who did report wearing PPE and who were also classified in the highest pesticide exposure group were estimated to have a non-significant excess MM, OR=1.35 (95% CI: 0.41-4.43). What was most surprising about this finding was the magnitude of the OR, even though PPE was very crudely defined as ‘ever/never’ use of any type of PPE for farm work. It is possible that PPE use was representing something other than what we intended. Given that all subjects who worked in farm jobs prior to 1947 reported “no” to using any PPE and all of these same subjects had worked on farms since childhood, it’s possible that the PPE use variable was indicative of the “era” of exposure. PPE use may not have been common, available for use or accessible to farm workers. Additionally, it may have been more common in the 1930s and 1940s for children to engage in pesticide handling activities during childhood. However, only a small number of subjects held farm jobs prior to 1947, so it was difficult to establish for certain what this PPE use variable may have represented beyond what we intended, such as the ‘era’ of exposure.  However, it is also possible that the observed effect of PPE use in Chapter 4 did represent a real effect of PPE on the association between herbicide use and MM risk. Validation evidence of the AHS ‘general’ algorithm by Coble et al. (2005), as discussed in Chapter 3, reported that the “algorithm intensity scores based primarily on the use of PPE provided a reasonably valid measure of exposure intensity”, particularly for the herbicide 2,4-D.36 The PPE variable of the AHS ‘general’ algorithm was not evaluated alone against urinary measurement data for herbicides 2,4-D and MCPA in the sample of pesticide applicators by Coble et al. (2002), but since there was little to no variation regarding the scores of the other algorithm variables, the final intensity scores were primarily a function of variation in self-reported PPE use.36 The details regarding the scoring system of this algorithm can be found in Chapter 3 or in Dosemeci et al. (2002) and Coble et al. (2011), but in brief, the PPE variable acts as an exposure reduction factor determined based on self-reported information.40,80 Therefore, subjects who reported “no” to wearing any PPE were assigned a reduction factor of 1.0 (i.e., no reduction to the total exposure as estimated by the other variables) and those who reported wearing any PPE were assigned a reduction factor of < 1.0, depending on the combination of PPE types worn.40,80 In essence, the validation study by Coble et al. (2005) showed that subjects who did not use PPE had higher exposure   101 levels, as determined by the AHS algorithm and the urinary samples. This lends credence to the finding of this research that no use of PPE resulted in higher exposure, and therefore, likely a higher risk.   5.2  Overall Strengths and Limitations  5.2.1 Strengths  In terms of the overall strengths of this research, Chapter 2 provided insight regarding the current state of dermal pesticide exposure monitoring literature. The findings provided both direction and suggestions for future dermal exposure monitoring studies to allow for comparing and combining data across multiple studies and for improving our understanding of exposure determinants. The pesticide exposure monitoring literature can provide the necessary information to characterize the exposure variation among farm samples, improve exposure assessment methods for retrospective studies reliant on external quantitative measures, and ultimately reduce exposure misclassification for more reliable estimates of association.   Although validity and reliability information is needed to fully understand if the potential dermal exposure algorithm can accurately estimate exposure levels among farm workers involved in applying and/or mixing-loading pesticides, this research has provided a new algorithm that provides actual exposure level estimates (mg/hour) and is believed to be the first to incorporate body area exposure estimates as determinants of exposure within an algorithm-based method. The AHS exposure intensity algorithm results in relative ranks of exposure, which makes it more difficult to extend the results outside of the study for exposure comparison.  This research independently evaluated the effect of PPE use on the relationship between cumulative pesticide exposure as estimated by the potential dermal exposure algorithm, and the risk of MM. Independent evaluation of potential determinants of exposure is valuable to understand the potential impact of their effect on the exposure-response relationship, which is difficult to assess with an algorithm that contains multiple, potential determinants of exposure as seen in the differing epidemiological results where PPE was included vs. not within the exposure algorithm. The fact that a PPE use variable was not included in the potential dermal exposure algorithm is what allowed us to easily evaluate its effect. In addition, the exposure years were reported in the epidemiologic analysis of Chapter 4, which is rarely done. This allowed for the possible indication of “era” of exposure by the PPE use variable.     102 5.2.2 Limitations  There are some important overall limitations to this research to be considered along with the findings reported above. First, in addition to the fact that the potential dermal exposure algorithm requires more information regarding its validity and reliability, particularly in terms body area exposure variables and the corresponding self-reported information to define these potential determinants, the following limitations should also be acknowledged: (1) other potentially important exposure determinants that are not included in this algorithm, and (2) duration and frequency information are not well reported. To address this first limitation, there are many other potential exposure determinants that can influence pesticide exposure in farm workers, such as application method, pesticide active ingredient, farm type, quantity of active ingredient, equipment used, other control measures besides PPE use (e.g., enclosed tractor cabs) that have been identified for other exposure assessments.80,93 To evaluate the reliability of self-reported information about the duration and frequency of specific pesticide use, Blair et al. (2002), administered questionnaires 1 year apart to a sample of AHS pesticide applicators.38 Blair et al. (2002) compared the self-reported information from these two questionnaires and determined agreement was low (50-60%) for self-reported frequency and duration of specific pesticide use 38 However, a study by Hoppin et al. (2002) used different methods to test the accuracy of calendar ‘time’ related information among AHS pesticide applicators.94 In this study, self-reported information about ‘decade of first use’ and ‘total years of use’ were evaluated using a logic check approach, using pesticide registration information from the United States Environmental Protection Agency among other sources. Hoppin et al. (2002) reported that less than 1% of the subjects overestimated their responses to these two questions based on the comparison of information.94 However, as reported by Hoppin et al. (2002), this analysis did not validate the accuracy of this information, but it “suggests that participants [provided] plausible information regarding their pesticide use”.94 However, regardless of the findings from these two studies, the recall time was much longer for the retrospective study data used for this dissertation, and this is a factor that has been reported to negatively affect the validity and reliability of self-reported information.41 Limitations regarding the epidemiologic data used for this dissertation must also be acknowledged. First, there was a significant amount of missing information on farm variables, due to non-response, particularly as the level of detail being questioned increased. There are many possible reasons as to why responses were low, such as inability to recall the information, the length of the questionnaire, or the clarity of the questions. This issue was discussed in more detail in Chapter 4, but in the context of the findings of this dissertation overall, the proportion of non-responses could have influenced results associated with the exposure intensity estimates from both algorithms in Chapter 3 and epidemiologic analyses reported in Chapter 4. In addition, the overall sample size of farm workers was small in this study and the power was low; this limited sub-analyses that could be performed (e.g., specific pesticide use) and may have reduced chances of detecting important associations. The inability to control for   103 some potential confounders was another limitation associated with the epidemiologic data, for example benzene exposure,95 which could have influenced epidemiologic findings. However, Blair et al. (2006) compared the impact of confounding versus exposure misclassification in occupational epidemiology, stating that “substantial confounding [is] rare in occupational epidemiology”, whereas, exposure misclassification is a bigger concern because it “probably occurs in all studies”.29 This brings the discussion to exposure misclassification as a limitation of this research. Much discussion has focused on the validity and reliability of exposure determinants and the self-reported information used to characterize the exposure determinants because they provide insight on the potential for exposure misclassification. Focusing on the algorithm-based methods that were the  primary means for exposure assessment in this dissertation, although exposure misclassification likely exists, misclassification is more likely to be in regards to the ‘level of exposure’  estimated by either/both algorithms, rather than with respect to whether a subject was considered unexposed  versus exposed. Subjects who did not provide complete responses to necessary farm jobs variables for applying the algorithms were not included in the analyses and proxy respondents were not used in this study, so these factors would not have contributed to exposure misclassification. However, it’s possible that exposure misclassification occurred as a result of reporting error with regards to ‘farm pesticide use’ because subjects who worked on a farm and did not handle pesticides, may not have been aware of general pesticide use on the farm. However, most farm workers in this study reported applying and mix-loading pesticides, so if this was an issue, it was likely a small contribution to exposure misclassification.   5.3 Recommendations for Future Research Directions  Based on the findings from this research, there are some important suggestions for future research in this field. First, in regards to the exposure monitoring literature, in order to improve reporting standardization across the literature and allow the quantitative data to be easily compared across several studies to provide insight on exposure determinants and to be used for exposure assessment development, it would be ideal for a database to be developed for this purpose. Ideally the database would include the standardized reporting suggestions made as part of this research. There is another such database, known as the Pesticide Handlers’ Exposure Database of the United States Environmental Protection Agency, which includes quantitative exposure data for pesticide handlers and flaggers, but these data are limited to specific potential exposure determinants (exposure route, some PPE use data, task, application method and pesticide formulation) with mean exposure values provided by various exposure scenarios.96 Additionally, the data used in this database does not come from real world literature, but rather comes from careful studies conducted by pesticide companies as a condition of registering their product.80 Therefore, it would be best to have a database that includes data and   104 information from the literature to collect standardized details needed for data combining, but does not limit the information that can be reported in regards to the exposure scenario studied. This would provide an accessible resource for researchers to easily leverage quantitative measurements from the literature to explore exposure determinants, develop exposure assessment methods, as well as to identify gaps where more data and information is needed with regards to farm worker pesticide exposure. For example, as shown by the studies included in the literature review of Chapter 1, there may be an under-representation of exposure monitoring data for non-pesticide handling farm workers (e.g. field workers). It is important to identify and characterize exposure determinants for each type of farm job.  Such a database would clearly require a lot of planning and work to design and maintain, but it could yield a valuable resource for exposure assessment in the context of retrospective epidemiology studies of pesticides and cancer in particular.  Epidemiologists need to increase efforts at the questionnaire design phase to improve the quality of self-reported information about exposure and necessary job details. This would require epidemiologists to understand the exposure determinants for which self-reported information should be collected on from the questionnaire. It would also require epidemiologists to develop and administer questions carefully to obtain complete and quality responses (both in terms of what questions need to be asked and how they should be asked). Lastly, efforts should be made to follow-up with subjects where information is incomplete.35,37 Such efforts would allow for questionnaire-based metrics to more accurately assess exposure in retrospective studies. Compared to cohort studies, case-control studies have the advantage of “considerable [exposure] contrast” because they include subjects from the general population that are not farm workers,30,41 but consequently, they can suffer from low exposure prevalence as noted in this study. Therefore, when possible, future epidemiologic studies should focus on conducting studies in areas where there is a higher prevalence of farm workers; 41 labour statistics could provide useful guidance for this purpose. Lastly, although not a suggestion for future research, it’s important to acknowledge that the development and advancement of new technologies will apply to future research in this area. For example, real-time exposure monitoring (e.g., wearable devices) and the collection of real-time information about exposure, actual practices and behaviours (e.g., drones and smartphones) will likely replace current exposure monitoring methods and supplement, improve or possibly replace questionnaires. For retrospective studies, information from these new technologies could reduce the level of detail that subjects are asked to recall or could change the types of information that are collected from questionnaires. There will still remain a need for knowledge on the determinants of exposure in order to develop and implement pesticide exposure reduction strategies for farm work.     105 5.4 Concluding Comments  A case-control study is the most feasible study design for evaluating pesticide exposure related to farm work in relation to MM due to the rareness of the disease. However exposure assessments for retrospective studies need to improve. This research has provided some recommendations for the exposure monitoring literature and has highlighted potential concerns about the questionnaire-based metrics used for this research. Overall, questionnaire-based metrics that use quantitative measurements from the literature have the potential to be a valuable means for exposure assessment if efforts are made to standardize data reporting across the exposure monitoring literature; these efforts will assist with exposure determinant characterization. Additionally, more efforts need to be placed on evaluating the validity and reliability of self-reported exposure information, as this information can be used to improve questionnaires (development and administration) and characterize the accuracy of questionnaire-based exposure assessment methods.                  106 References  1. Kachuri L, Harris MA, MacLeod JS, Tjepkema M, Peters PA, Demers PA. Cancer risks in a population-based study of 70,570 agricultural worker: Results from the Canadian census health and environment cohort (CanCHEC). BMC Cancer. 2017;17(1):343.  doi: 10.1186/s12885-017-3346-x. 2. Koutros S, Alavanja MC, Lubin JH, et al. An update of cancer incidence in the agricultural health study. J Occup Environ Med. 2010;52(11):1098-105. doi: 10.1097/JOM.0b013e3181f72b7c. 3. Blair A, Malkder H, Cantor K, Burmeister L, Wiklund K. Cancer among farmers: A review. Scandinavian Journal of Work, Environment and Health. 1985;11(6):397-407. 4. Blair A, Freeman LB. Epidemiologic studies in agricultural populations: Observations and future directions. J Agromed. 2009;14(2):125-131. doi: 10.1080/10599240902779436. 5. Pearce N, Reif JS. Epidemiologic studies of cancer in agricultural workers. American Journal of Industrial Medicine. 1990;18(2):133-148. 6. Weichenthal S., Moase C., Chan P. A review of pesticide exposure and cancer incidence in the agricultural health study cohort. Environmental Health Perspectives. 2010;118(8):1117-1125. doi:  10.1289/ehp.0901731. 7. Khuder SA, Mutgi AB. Meta-analyses of multiple myeloma and farming. American Journal of Industrial Medicine. 1997;32(5):510-516. https://doi.org/10.1002/(SICI)1097-0274(199711)32:5<510::AID-AJIM11>3.0.CO;2-5. 8. Alavanja MC, Bonner MR. Occupational pesticide exposures and cancer risk: A review. J Toxicol Environ Health B Crit Rev. 2012;15(4):238-263. doi: 10.1080/10937404.2012.632358. 9. Xie L, Semenciw R, Mery L. Cancer incidence in Canada: Trends and projections (1983-2032), surveillance and epidemiology division, centre for chronic disease prevention, Public Health Agency of Canada. Health Promotion and Chronic Disease Prevention in Canada: Research, Policy and Practice. 2015;35(1):2-186. 10. Perrotta C, Kleefeld S, Staines A, et al. Multiple myeloma and occupation: A pooled analysis by the international multiple myeloma consortium. Cancer Epidemiology. 2013;37(3):300-305. doi: 10.1016/j.canep.2013.01.008. 11. Canadian Cancer Society. Multiple myeloma: What is multiple myeloma? http://www.cancer.ca/en/cancer-information/cancer-type/multiple-myeloma/multiple-myeloma/?region=on. Updated 2018. Accessed 10/05, 2018. 12. Purdue MP, Hofman JN, Brown EE, Vachon CM. Multiple myeloma. In: Cancer epidemiology and prevention. Oxford Scholarship Online; 2017:797-813. 10.1093/oso/9780190238667.001.0001. 13. Raab MS, Podar K, Breitkreutz I, Richardson PG, Anderson KC. Multiple myeloma. Lancet. 2009;374(9686):P324-339. doi: 10.1016/S0140-6736(09)60221-X. 14. Cowan AJ, Allen C., Barac A, Basaleem H, et al. Global burden of multile myeloma: A systematic analysis for the global burden of diease study 2016. JAMA Oncol. 2018;4(9):1221-1227. doi:10.1001/jamaoncol.2018.2128. 15. Leukemia and Lymphoma Society of Canada. Blood cancer in Canada: Facts and stats 2016. 2016.   107 16. Cantor KP, Blair A. Farming and mortality from multiple myeloma: A case-control study with the use of death certificates. J Natl Cancer Inst. 1984;72(2):251-255. 17. Gallagher RP, Spinelli JJ, Elwood JM, Skippen DH. Allergies and agricultural exposure as risk factors for multiple myeloma. British Journal of Cancer. 1983;48(6):853-857. 18. Eriksson M, Karlsson M. Occupational and other environmental factors and multiple myeloma: A population based case-control study. Brit J Indust Med. 1992;49(2):95-103. 19. Milham S. Leukemia and multiple myeloma in farmers. American Journal of Epidemiology. 1971;94(4):307-310. 20. Nanni O, Falcini F, Buiatti E, et al. Multiple myeloma and work in agriculture: Results of a case-control study in Florli, Italy. Cancer Causes and Control. 1998;9(3):277-283. 21. Mehri M, Raynal H, Cahuzac E, Vinson F, Cravedi JP, Gamet-Payrastre L. Occupational exposure to pesticides and risk of hematopoietic cancers: Meta-analysis of case-control studies. Cancer Causes Control. 2007;18(10):1209-1226. doi:10.1007/s10552-007-9061-1. 22. Perrotta C, STaines A, Cocco P. Multiple myeloma and farming. A systematic review of 30 years of research. where next? J Occup Med Toxicol. 2008;3(27). doi: 10.1186/1745-6673-3-27. 23. Presutti R, Harris SA, Kachuri L, et al. Pesticide exposures and the risk of multiple myeloma in men: An analysis of the north american pooled project. Int J Cancer. 2016;139(8):1703-1714.  doi: 10.1002/ijc.30218. 24. Kachuri L, Demers PA, Blair A, et al. Multiple pesticide exposures and the risk of multiple myeloma in Canadian men. International Journal of Cancer. 2013;133(8):1846-1858. doi: 10.1002/ijc.28191. 25. McLaughlin JK, Malker HS, Linet MS, et al. Multiple myeloma and occupation in Sweden. Arch Environ Health. 1988;43(1):7-10. doi:10.1080/00039896.1988.9934365. 26. Beane Freeman L. Evaluation of agricultural exposures: The agricultural health study and the agricultural cohort consortium. Rev Environ Health. 2009;24(4):311-318. 27. Pahwa P, Karunanayake CP, Dosman JA, Spinelli JJ, McDuffie HH, McLaughlin JR. Multiple myeloma and exposure to pesticides: A Canadian case-control study. J Agromedicine. 2012;17(1):40-50. doi: 10.1080/1059924X.2012.632339. 28. Arbuckle TE, Burnett R, Cole D, et al. Predictors of herbicide exposure in farm applicators. Int Arch Occup Environ Health. 2002;75(6):406-414. doi:10.1007/s00420-002-0323-7. 29. Blair A., Stewart P., Lubin J.H., Forastiere F. Methodological issues regarding confounding and exposure misclassification in epidemiologic studies of occupational exposures. American Journal of Industrial Medicine. 2007;50(3):199-207. doi:10.1002/ajim.20281. 30. Kromhout H, Heederik D. Effects of errors in the measurement of agricultural exposures. Scand J Work Environ Health. 2005;31(Suppl 1, Discussion 5-7):33-38. 31. Hoppin JA, Adgate JL, Eberhart M, Nishioka M, Ryan PB. Environmental exposure assessment of pesticides in farmworker homes. Environ Health Perspect. 2006;114(6):929-935. doi: 10.1289/ehp.8530.     108 32. MacFarlane E, Glass D, Fritschi L. Is farm-related job title an adequate surrogate for pesticide exposure in occupational cancer epidemiology? Occup Environ Med. 2009;66(8):497-501. doi: 10.1136/oem.2008.041566. 33. MacFarlane E, Carey R, Keegel T, El-Zaemay S, Fritschi L. Dermal exposure associated with occupational end use of pesticides and the role of protective measures. Saf Health Work. 2013;4(3):136-141. Accessed 20131007. doi: http://dx.doi.org/10.1016/j.shaw.2013.07.004. 34. Vermeulen R, Stewart P, Kromhout H. Dermal exposure assessment in occupational epidemiologic research. Scand J Work Environ Health. 2002;28(6):371-385. doi: 10.5271/sjweh.689. 35. Stewart PA, Stewart WF, Siemiatycki J, Heineman EF, Dosemeci M. Questionnaires for collecting detailed occupational information for community-based case-control studies. Am Ind Hyg Assoc J. 1998;59(1):39-44. doi:10.1080/15428119891010325. 36. Coble J, Arbuckle T, Lee W, Alavanja M, Dosemeci M. The validation of a pesticide exposure algorithm using biological monitoring results. Journal of Occupational and Environmental Hygiene. 2005;2(3):194-201. doi:10.1080/15459620590923343. 37. Nieuwenhuijsen MJ. Design of exposure questionnaires for epidemiologic studies. Occupational and Environmental Medicine. 2005;62(4):272-280. 38. Blair A, Tarone R, Sandler D, et al. Reliability of reporting on life-style and agricultural factors by a sample of participants in the agricultural health study from iowa. Epidemiology. 2002;13(1):94-95-99. doi: http://dx.doi.org/10.1016/S1047-2797(00)00113-7. 39. Burstyn I, Teschke K. Studying the determinants of exposure: A review of methods. Am Ind Hyg Assoc J. 1999;60(1):57-72. doi:10.1080/00028899908984423. 40. Coble J, Thomas KW, Hines CJ, et al. An updated algorithm for estimation of pesticide exposure intensity in the agricultural health study. Int J Environ Res Public Health. 2011;8(12):4608-4622. doi: 10.3390/ijerph8124608. 41. Teschke K, Olshan AF, Daniels JL, et al. Occupational exposure assessment in case-control studies: Opportunities for improvement. Occup Environ Med. 2002;59(9):575-594. doi: 10.1136/oem.59.9.575. 42. Heederik D, Teschke K. Exposure assessment for pesticides in epidemiological studies. In: Franklin CA, Worgan JP, eds. Occupational and residential exposure assessment for pesticides. John Wiley & Sons, Ltd.; 2005:245-272. doi: 10.1002/0470012218.ch7. 43. Stewart P. Challenges to retrospective exposure assessment. Scand J Work Environ Health. 1999;25(6):505-510. doi: 10.5271/sjweh.473. 44. Durham WF, Wolfe HR. Measurement of the exposure of workers to pesticides. Bull World Health Organ. 1962;26:75-91. 45. Fenske RA, Day EW Jr. Assessment of exposure for pesticide handlers in agricultural, residential and institutional environments. In: Franklin CA, Worgan JP, eds. Occupational and residential exposure assessment for pesticides. John Wiley & Sons, Ltd; 2005:13. doi: 10.1002/0470012218.ch1. 46. Daniels JL, Olshan AF, Teschke K, Hertz-Picciotto I, Savitz DA, Blatt J. Comparison of assessment methods for pesticide exposure in a case-control interview study. Am J Epidemiol. 2001;153(12):1227-1232. doi: https://doi.org/10.1093/aje/153.12.1227.   109 47. Stewart PA, Prince JK, Colt JS, Ward MH. A method for assessing occupational pesticide exposures of farmworkers. Am J Ind Med. 2001;40(5):561-570. https://doi.org/10.1002/ajim.1122. 48. McLaughlin T. Review of dermal absorption. NSCEP. 1984;EPA/600/8-84/033 (NTIS PB85170694). https://nepis.epa.gov/Exe/ZyNET.exe/9100RNVF.txt?ZyActionD=ZyDocument&Client=EPA&Index=1981%20Thru%201985&Docs=&Query=&Time=&EndTime=&SearchMethod=1&TocRestrict=n&Toc=&TocEntry=&QField=&QFieldYear=&QFieldMonth=&QFieldDay=&UseQField=&IntQFieldOp=0&ExtQFieldOp=0&XmlQuery=&File=D%3A%5CZYFILES%5CINDEX%20DATA%5C81THRU85%5CTXT%5C00000019%5C9100RNVF.txt&User=ANONYMOUS&Password=anonymous&SortMethod=h%7C-&MaximumDocuments=1&FuzzyDegree=0&ImageQuality=r75g8/r75g8/x150y150g16/i425&Display=hpfr&DefSeekPage=x&SearchBack=ZyActionL&Back=ZyActionS&BackDesc=Results%20page&MaximumPages=1&ZyEntry=2. 49. Herman ND, Hunt TW, Sheets TJ. Hand harvester exposure to maleic hydrazide (MH) in flue-cured tobacco. Bull Environ Contam Toxicol. 1985;34(4):469-475. doi: https://doi.org/10.1007/BF01609762. 50. Stewart PA, Fears T, Kross B, Ogilvie L, Blair A. Exposure of farmers to phosmet, a swine insecticide. Scand J Work Environ Health. 1999;25(1):33-38. doi: 10.5271/sjweh.380. 51. Vitali M, Protano C, Monte AD, Ensabella F, Guidotti M. Operative modalities and exposure to pesticides during open field treatments among a group of agricultural subcontractors. Arch Environ Contam Toxicol. 2009;57(1):193-202. doi: 10.1007/s00244-008-9225-3. 52. Lonsway JA, Byers ME, Dowla HA, Panemangalore M, Antonious GF. Dermal and respiratory exposure of mixers/sprayers to acephate, methamidophos, and endosulfan during tobacco production. Bull Environ Contam Toxicol. 1997;59(2):179-186. doi: https://doi.org/10.1007/s001289900462. 53. Winterlin WL, Kilgore WW, Mourer CR, Schoen SR. Worker reentry studies for captan applied to strawberries in california. J Agric Food Chem. 1984;32(3):664-672. doi: 10.1021/jf00123a059. 54. Senior PL, Lavers A. Determination of potential dermal exposure during application of crop protection products by boom spraying. Ann Occup Hyg. 1992;36(6):589-599. doi: https://doi.org/10.1093/annhyg/36.6.589. 55. Lebailly P, Bouchart V, Baldi I, et al. Exposure to pesticides in open-field farming in france. Ann Occup Hyg. 2009;53(1):69-81. doi: https://doi.org/10.1093/annhyg/men072. 56. Winterlin WL, Kilgore WW, Mourer CR, Hall G, Hodapp D. Worker reentry into captan-treated grape fields in california. Arch Environ Contam Toxicol. 1986;15(3):301-312. doi: https://doi.org/10.1007/BF01061107. 57. Cock JD, Heederik D, Kromhout H, et al. Exposure to captan in fruit growing. Am Ind Hyg Assoc J. 1998;59(3):158-165. doi: 10.1080/15428119891010415. 58. Vercruysse F, Drieghe S, Steurbaut W, Dejonckheere W. Exposure assessment of professional pesticide users during treatment of potato fields. Pestic Sci. 1999;55(4):467-473. doi: 10.1002/(SICI)1096-9063(199904)55:43.0.CO;2-#. 59. Fenske RA, Blacker AM, Hamburger SJ, Simon GS. Worker exposure and protective clothing performance during manual seed treatment with lindane. Arch Environ Contam Toxicol. 1990;19(2):190-196. doi: https://doi.org/10.1007/BF01056086.   110 60. Yeung P, Cantrell P, Murdoch C, et al. Exposure of air blast applicators to ethyl parathion and methyl parathion in orchards: A comparison of australian conditions to overseas predictive exposure models. Appl Occup Environ Hyg. 1998;13(4):242-251. doi: http://dx.doi.org/10.1080/1047322X.1998.10390075. 61. Protano C, Guidotti M, Vitali M. Performance of different work clothing types for reducing skin exposure to pesticides during open field treatment. Bull Environ Contam Toxicol. 2009;83(1):115-119. doi: 10.1007/s00128-009-9753-1. 62. Abbott IM, Bonsall JL, Chester G. Worker exposure to a herbicide applied with ground sprayers in the united kingdom. Am Ind Hyg Assoc J. 1987;48(2):167-175. doi: 10.1080/15298668791384571. 63. Chester G, Hart TB. Biological monitoring of a herbicide applied through backpack and vehicle sprayers. Toxicol Lett. 1986;33(1-3):137-149. doi: https://doi.org/10.1016/0378-4274(86)90078-0. 64. Dubelman S, Lauer R, Arras DD, Adams SA. Operator exposure measurements during application of the herbicide diallate. J Agric Food Chem. 1982;30(3):528-532. doi: 10.1021/jf00111a030. 65. Nigg HN, Stamper JH, Queen RM. Dicofol exposure to florida citrus applicators: Effects of protective clothing. Arch Environ Contam Toxicol. 1986;15(1):121-134. doi: https://doi.org/10.1007/BF01055257. 66. Wojeck GA, Nigg HN, Braman RS. Worker exposure to arsenic in florida grapefruit spray operations. Arch Environ Contam Toxicol. 1982;11(6):661-667. doi: https://doi.org/10.1007/BF01059152. 67. Nigg HN, Stamper JH. Exposure of spray applicators and mixer-loaders to chlorobenzilate miticide in florida citrus groves. Arch Environ Contam Toxicol. 1983;12(4):477-482. doi: https://doi.org/10.1007/BF01057592. 68. Chester G, Hatfield LD, Hart TB. Worker exposure to, and absorption of, cypermethrin during aerial application of an 'ultra low volume' formulation to cotton. Arch Environ Contam Toxicol. 1987;16(1):69-78. doi: https://doi.org/10.1007/BF01055361. 69. Lengerich SK, Burroughs FG. Near real-time monitoring of potential dermal exposure during backpack herbicide spraying. Am Ind Hyg Assoc J. 1989;50(10):516-519. doi: http://dx.doi.org/10.1080/15298668991375092. 70. Sanderson WT, Ringenburg V, Biagini R. Exposure of commercial pesticide applicators to the herbicide alachlor. Am Ind Hyg Assoc J. 1995;56(9):890-897. doi: 10.1080/15428119591016520. 71. Carman GE, Iwata Y, Pappas JL, O'Neal JR, Gunther FA. Pesticide applicator exposure to insecticides during treatment of citrus trees with oscillating boom and airblast units. Arch Environ Contam Toxicol. 1982;11(6):651-659. doi: 10.1007/BF01059151. 72. Wojeck GA, Nigg HN, Stamper JH, Bradway DE. Worker exposure to ethion in florida citrus. Arch Environ Contam Toxicol. 1981;10(6):725-735. doi: https://doi.org/10.1007/BF01054856. 73. Simpson GR, Beck A. Exposure to parathion. dermal and inhalation exposure to parathion while spraying tomato bushes with a knap-sack mister. Arch Environ Health. 1965;11(6):784-786. 74. Tsakirakis AN, Kasiotis KM, Charistou AN, et al. Dermal & inhalation exposure of operators during fungicide application in vineyards. evaluation of coverall performance. Sci Total Environ. 2014;470-471:282-289. www.elsevier.com/locate/scitotenv. Accessed 20141009. doi: 10.1016/j.scitotenv.2013.09.021.   111 75. Wojeck GA, Price JF, Nigg HN, Stamper JH. Worker exposure to paraquat and diquat. Arch Environ Contam Toxicol. 1983;12(1):65-70. doi: https://doi.org/10.1007/BF01055003. 76. Kiefer M, Lenhart SW, Tubbs RL, Mattorano D. HHE report no. HETA-95-0248-2562, dirty bird, inc., grady, arkansas. NIOSH. 1996:46. 77. Everhart LP, Holt RF. Potential benlate fungicide exposure during mixer/loader operations, crop harvest, and home use. J Agric Food Chem. 1982;30(2):222-227. doi: 10.1021/jf00110a003. 78. Kamble ST, Byers ME, Witkowski JF, Ogg CL, Echtenkamp GW. Field worker exposure to selected insecticides applied to corn via center-pivot irrigation. J Econ Entomol. 1992;85(3):974-980. 79. Chester G, Ward RJ. Occupational exposure and drift hazard during aerial application of paraquat to cotton. Arch Environ Contam Toxicol. 1984;13(5):551-563. doi: https://doi.org/10.1007/BF01056333. 80. Dosemeci M, Alavanja MCR, Rowland AS, et al. A quantitative approach for estimating exposure to pesticides in the agricultural health study. Ann Occup Hyg. 2002;46(2):245-260. 81. Blair A, Zahm SH. Patterns of pesticide use among farmers: Implications for epidemiologic research. Epidemiology. 1993;4(11):55-56-62. 82. Beauvis S, Powell S, Zhao W. Surrogate handler exposure estimates for use in assessments by the california department of pesticide regulation. 2007;HS-1826. http://www.cdpr.ca.gov/docs/whs/pdf/hs1826.pdf. Accessed August 10, 2016. 83. Blair A ZS. Methodologic issues in exposure assessment for case-control studies of cancer and herbicides. American Journal of Industrial Medicine. 1990;18(3):285-293. doi: 10.1002/ajim.4700180308. 84. Garzia NA, Spinelli JJ, Gotay CC, Teschke K. Literature review: Dermal monitoring data for pesticide exposure assessment of farm workers. J Agromedicine. 2018;23(3):187-214. doi: 10.1080/1059924X.2018.1448734. 85. Thomas KW, Dosemeci M, Coble JB, et al. Assessment of a pesticide exposure intensity algorithm in the agricultural health study. Journal of Exposure Science and Environmental Epidemiology. 2010;20(6):559-569.  doi: 10.1038/jes.2009.54.  86. Hines CJ, Deddens JA, Jaycox LB, Andrews RN, Striley CAF, Alavanja MCR. Captan exposure and evaluation of a pesticide exposure algorithm among orchard pesticide applicators in the agricultural health study. Ann Occup Hyg. 2008;52(3):153-166. doi: 10.1093/annhyg/men001. 87. Hines CJ, Deddens JA, Coble J, Kamel F, Alavanja MC. Determinants of captan air and dermal exposures among orchard pesticide applicators in the agricultural health study. Ann Occup Hyg. 2011. doi: 10.1093/annhyg/mer008.  88. Acquavella JF, Alexander BH, Mandel JS, Burns CJ, Gustin C. Exposure misclassification in studies of agricultural pesticides: Insights from biomonitoring. Epidemiology. 2006;17(1):69-74. 89. Weinberg CR, Moledor ES, Umbach DM, Sandler DP. Imputation for exposure histories with gaps, under an excess relative risk model. Epidemiology. 1996;7(5):490-497. 90. Government of Canada. Questions and answers - final decision on the re-evaluation of 2,4-D. Government of Canada Web site. https://www.canada.ca/en/health-canada/services/consumer-product-safety/pesticides-pest-  112 management/public/protecting-your-health-environment/questions-answers-final-decision-evaluation-2-4-d.html#registered. Published 2009-01-01. Updated 2009. Accessed 10/09, 2018. 91. Canadian Council of Ministers of the Environment. Canadian soil quality guidelines for the protection of environmental and human health: DDT (TOTAL). Canadian Environmental Quality Guidelines. 1999. http://ceqg-rcqe.ccme.ca/download/en/265?redir=1539142436. Accessed 2018-10-09. 92. CAREX Canada. Glyphosate. Surveillance of environmental and occupational exposures for cancer prevention Web site. https://www.carexcanada.ca/en/glyphosate/. Published 2018-07. Updated 2018. Accessed 10/09, 2018. 93. Baldi I, Lebailly P, Rondeau V, et al. Levels and determinants of pesticide exposure in operators involved in treatment of vineyards: Results of the PESTEXPO study. J Expo Sci Environ Epidemiol. 2012;22(6):593-600. doi: 10.1038/jes.2012.82. 94. Hoppin JA, Yucel F, Dosemeci M, Sandler DP. Accuracy of self-reported pesticide use duration information from licensed pesticide applicators in the agricultural health study. J Expo Anal Environ Epidemiol. 2002;12(5):313-318. doi:10.1038/sj.jea.7500232. 95. Louis LM, Lerro CC, Friesen MC, et al. A prospective study of cancer risk among agricultural health study farm spouses association with personal use of organochlorine insecticides. Environ Health. 2017;16(1):95. doi: 10.1186/s12940-017-0298-1. 96. Nielsen A, Curry P, Leighton T. The pesticide handlers exposure database (PHED): Regulatory overview. In: Curry PB, Iyengar S, Maloney PA, Maroni M, eds. Methods of pesticide exposure assessment. Volume 19 ed. NATO - Challenges of Modern Society; 1995:89-90-93.               113 Appendix A  A.1 Copy of the farm questionnaire  A copy of the farm questionnaire that was used for assessing pesticide exposure for this research is provided below. This questionnaire was developed for the Multiple Myeloma and Monoclonal Gammopathy of Undetermined Significance Study conducted by BC Cancer.   Occupation Calendar / Work History This calendar is part of a research study looking at environmental, lifestyle and genetic factors and their role in the development of Multiple Myeloma (MM) and Monoclonal Gammopathy of Undetermined Significance (MGUS).     These questions are for those who have lived on a farm, or work in agriculture, gardening, parks, golf courses and forestry.      Job No.  Time Period  Type of  Industry,  Business or  Service  Company  Name and  Location  Job Title  Start  End  eg  June 2010  Dec 2010  farming  DOW,  Vancouver BC  Owner  1            2            3            4            5                    114  Farming Module 1. For this period, how would you describe your work on the farm/workplace?  (CHECK  ALL  THAT APPLY)   Lived on a farm   Farmer (owner/operator)   Worker    Applicator   Harvester   Machine operator   Picker   Graders/sorters   Trucker/tractor driver   Others (specify) _________________   Other (specify) _____________________    2 Where did you work during this time period?     Country  Province/State  City          3 How would you classify your farm/workplace (CHECK  ALL THAT APPLY)     Field crops  (If yes, go to Field crop farming)  Yes      No       DK   Fruits and vegetables  (If yes, go to fruit & vegetable farming)  Yes      No       DK   Animal or poultry farm (If yes, go to animal farming)  Yes      No       DK   Greenhouse (If yes, go to greenhouse)  Yes      No       DK   Gardens / Parks / Lawns    Yes      No       DK   Golf courses  Yes      No       DK   Forestry  Yes      No       DK   Other (specify) ____________________________   Yes      No       DK       4 [In a farm], what was the size of the farm?  ________________ Acres/Hectares (1 acre=0.40 hectare, 1 hectare=2.47 acres)    5 [Field crop farming] What were the main crops grown on this farm/workplace?  (CHECK  ALL THAT APPLY)      115  Wheat   Canola   Soybean    Oilseed(except soybean farming)   Corn    Other grains (specify) _____________________  Legumes e.g. dry peas, beans, lentils.   Hay or seed pasture   Other field crops (specify)  ____________________    6 [Fruit and vegetable farming] What were the main crops grown on this farm/workplace?  (CHECK  ALL THAT APPLY)     Vegetables and melon (specify) _________________   Fruit and tree nut (specify) _______________________   Berries (specify) ____________________    7 [Greenhouse] What were the main crops grown in this greenhouse? (CHECK  ALL THAT APPLY)     Greenhouse vegetables   Other greenhouse products (specify) _____________________   Flowers e.g. cut flowers, potted, bedding plants   Nursery, e.g. trees, shrubs grown out of doors   Mushrooms   Honey or bees   Maple syrup products   Others (specify) _____________________________    8 Were pesticides to kill weeds, insects, fungus or moulds, or rodents such as rats applied on this farm/workplace?    Herbicides  Yes      No       DK   Insecticides  Yes      No       DK   Fungicides  Yes      No       DK    Work Practices     Questions 10-25 comprise three types of work (application, mixing and loading pesticides, cleaning and repairing) you might involve when you worked on a farm/workplace. Skip the section if you did not involve in that type of work.       116 Application    9  Did you personally apply any of the following items on this job?       Herbicides      Yes      No       DK   Insecticides    Yes      No       DK   Fungicides      Yes      No       DK     ** If Yes/DK to any one of the three questions, answer the protection section.   ** If No, skip the protection section.  ** Proceed to Questions 26-31 for those who have worked on an animal and poultry farm.                    117 11.       What crop was treated?  What was the product name or  active ingredient,  e.g. 2,4-D?              What pest did this product target?  How  did  you  apply  the  product? (write the number)  1. Hand held sprayers  2. Backpack sprayer  3. Boom sprayers  4. Airblast  5. Applicators for solid formulations (granular or dust applicators)  6. Aerial sprayers (aircraft)  7. Fumigation   8. Foggers  9. Chemigation (sprinkler, flood, furrow)  10. Other (specify)  _________  How many  years did you apply this pesticide?    In an average year, how many  days did you use this pesticide?       Herbicides   Insecticides  Fungicides   Don’t know          _____  years  _____ days/ yr   OR   < 10 days   10-39 days   40-69days   70-99 days   >99 days       Herbicides   Insecticides   Fungicides   Don’t know          _____  years  _____ days/ yr   OR   < 10 days   10-39 days   40-69days   70-99 days   >99 days    118     Herbicides   Insecticides   Fungicides   Don’t know          _____  years  _____ days/ yr   OR   < 10 days   10-39 days   40-69days   70-99 days   >99 days       Herbicides   Insecticides   Fungicides   Don’t know          _____  years  _____ days/ yr   OR   < 10 days   10-39 days   40-69days   70-99 days   >99 days     Herbicides   Insecticides   Fungicides   Don’t know          _____  years  _____ days/ yr   OR   < 10 days   10-39 days   40-69days   70-99 days   >99 days       Herbicides   Insecticides   Fungicides   Don’t know          _____  years  _____ days/ yr   OR   < 10 days   10-39 days   40-69days   70-99 days   >99 days      119 12 What types of protective equipment did you use when you applied pesticides?     Never use protective equipment   Overalls/coveralls   Goggles   Gloves    Chemical resistant gloves   Fabric/leather gloves   Mask   Dust/disposable mask   Full face shield   Cartridge respirator, gas mask   Boots   Hat   DK    13 When applying pesticides, what parts of your body usually came in contact with the pesticides? (Mark all apply)     Head and/or face   Arms and hands   Body   Legs/feet   Lungs and respiratory tract (from breathing fumes)   Digestive tract (from ingesting/swallowing)   None      Mixing and Loading pesticides    14 Did you mix and/or load any of the following yourself?    Herbicides   Yes      No       DK   Insecticides   Yes      No       DK   Fungicides   Yes      No       DK     ** If Yes to any one of the three questions, answer the protection section.   ** If No, skip the protection section.           120 15 What pesticide product(s) or active ingredient(s) did you mix and/or load?  i)_______ii)______ iii)______ iv)______    16 How often did you mix and/or load pesticides yourself?     Never   Sometimes   Often   Always    17 What types of protective equipment did you usually use when you mixed and/or loaded pesticides? (Mark all that apply)     Never use protective equipment   Overalls/coveralls   Goggles   Gloves    Chemical resistant gloves   Fabric/leather gloves   Mask   Dust/disposable mask   Full face shield   Cartridge respirator, gas mask   Boots   Hat   DK    18 When mixing and/or loading pesticides, what parts of your body usually came in contact with the pesticides? (Mark all apply)     Head and/or face   Arms and hands   Body   Legs/feet   Lungs and respiratory tract (from breathing fumes)   Digestive tract (from ingesting/swallowing)   None    19 After mixing and/or loading pesticides, when did you usually change into clean work clothes?      121  Right away   Within 1-3 hours   At the end of that work day   After several days   Always use disposable outer clothing   Other (specify) _____________   DK    20 After mixing and/or loading pesticides, when did you usually wash yourself?     Right away   Within 1-3 hours   At the end of that work day   After several days   Other (specify) _____________   DK    Wash and repair equipment for pesticides    21 How often did you wash and/or repair spraying and mixing equipment yourself?     Never   Sometimes   Often   Always    22 What types of protective equipment did you usually use when you washed and/or repaired spraying and mixing equipment? (Mark all that apply)     Never use protective equipment   Overalls/coveralls   Goggles   Gloves    Chemical resistant gloves   Fabric/leather gloves   Mask   Dust/disposable mask   Full face shield   Cartridge respirator, gas mask   Boots   Hat   DK    122   23 When washing and/or repairing the equipment, what parts of your body usually came in contact with the pesticides? (Mark all apply)     Head and/or face   Arms and hands   Body   Legs/feet   Lungs and respiratory tract (from breathing fumes)   Digestive tract (from ingesting/swallowing)   None    24 After washing and/or repairing the equipment, when did you usually change into clean work clothes?     Right away   Within 1-3 hours   At the end of that work day   After several days   Always use disposable outer clothing   Other (specify) _____________   DK    25 After washing and/or repairing the equipment, when did you usually wash yourself?     Right away   Within 1-3 hours   At the end of that work day   After several days   Other (specify) _____________   DK         Questions 26-31 are for those who have worked on an animal and poultry farm.    26 [Animal or poultry farming] What were the main animals on this farm/workplace?  (CHECK ALL THAT APPLY)     Daily cattle and milk production   Beef cattle ranching and farming   Hog and pig farming   Chicken and egg production    123  Other poultry (specify) _______________   Horse and other equine production   Other animal (specify) ______________________    27 [Animal or poultry farming] Did any infectious epidemics occur among the animals at the farm during your employment there?     Yes (complete table below)    Epidemic  Year                 No    DK    28 [Animal farm ONLY] How often were you involved in feeding animals?   Never   Sometimes   Often   Always    29 [Animal and poultry farm ONLY] Were you involved in treating the animals for parasites?     Yes   How many days a year usually did you treat animals?  ________ days/year  OR   Less than 9 days   10-39 days   40-69 days   70-99 days   More than 100 days   No    DK     30. [Animal poultry farm ONLY] How often were animals slaughtered on the farm?     Never (skip next question)   Sometimes    124  Often   Always      31. [Animal poultry farm ONLY] Did you personally slaughter the animals?   Yes   How frequent? __________ times per day/week/month/year   No   DK                        125 Appendix B   Provided here is additional documentation on the application and development of the potential dermal exposure algorithm of this dissertation.   B.1 Additional details on how the algorithm was applied for exposure assessment  Potential Dermal Exposure Algorithm  Intensity (mg/hr) = MIX-LOAD (Head/FaceML + BodyML + Arms/HandsML + Legs/FeetML) + APPLY (Head/FaceAPP + BodyAPP + Arms/HandsAPP + Legs/FeetAPP)  STEP 1: Assign the task-specific weights (MIX-LOAD and APPLY) based on self-reported responses to: “did you personally apply any of the following items on this job: herbicides (yes/no/don’t know), insecticides (yes/no/don’t know), fungicides (yes/no/don’t know)? “and “did you mix and/or load any of the following yourself: herbicides (yes/no/don’t know), insecticides (yes/no/don’t know), fungicides (yes/no/don’t know)?” as shown in Appendix A. The responses were assigned task-specific weights as follows:  Subject self-reports “yes” to personally MIXING and/or LOADING any pesticide group and “no” to personally APPLYING all pesticide groups, then: MIX-LOAD weight = 1 and APPLY weight = 0;   Subject self-reports “no” to personally MIXING and/or LOADING all pesticide groups and “yes” to personally APPLYING any pesticide group, then: MIX-LOAD weight = 0 and APPLY weight = 1;   Subject self-reports “yes” to personally MIXING and/or LOADING any pesticide group and “yes” to personally APPLYING any pesticide group, then: MIX-LOAD weight = 0.2 and APPLY weight = 0.8;   Subject self-reports “no” to personally MIXING and/or LOADING pesticides and “no” to personally APPLYING pesticides, then: MIX-LOAD weight = 0 and APPLY weight = 0.    126 STEP 2: Assign body area-specific weights for each task based on self-reports to “when (applying / mixing and/or loading pesticides), what parts of your body usually came in contact with the pesticides” (mark all that apply: head and/or face, arms and hands, body, legs/feet), as shown in Appendix A.  For each of these four body areas that was not marked by the subject for the correspond task performed, the body area was set = 0.  For each of these four body areas that was marked as “usually coming into contact with the pesticides”, the corresponding exposure levels were assigned for each task performed: For the MIX-LOAD task, body area exposure levels are assigned as: [Head/FaceML] = 0.56 mg/hr [BodyML] = 1.0 mg/hr [Arms/HandsML] = 58.26 mg/hr [Legs/FeetML] = 1.57 mg/hr For the APPLY task, body area exposure scores are assigned as: [Head/FaceAPP] = 1.0 mg/hr [BodyAPP] = 5.01 mg/hr [Arms/HandsAPP] =12.65 mg/hr  [Legs/FeetAPP] = 4.43 mg/hr  B.2 Additional details on how the algorithm was developed   As described in Chapter 3, an algorithm was developed  as part of this research to estimate farm worker pesticide exposure intensity by combining quantitative exposure information from the literatures with self-reports about pesticide handling tasks (apply and mix-load) and body areas commonly coming into contact with pesticides during these tasks. The task--specific weights and corresponding body area exposure levels derived from the quantitative information are described in detail below.     127 Task-Specific Weights The task weight corresponds to the duration of time spent performing each of the two pesticide handling tasks based on whether a worker was a mixer-loader (mix-load only), an applicator (apply only) or an operator (mix-load and apply). For mixer-loaders, the task weight for ‘mix-load’ was set to 1.0 and the task weight for ‘apply’ was set to 0.0 because it’s expected that their dermal exposures are due predominantly to their mixing-loading task only. The opposite task weight assignment was applied to applicators, i.e., ‘apply’ was set to 1.0 and ‘mix-load’ was set to 0.0 since these farm workers self-reported only performing the pesticide application task.  For Operators, in which both tasks were performed, it was less clear the duration of time spent on each task. A total of n=16 studies from the literature review (Chapter 2) were evaluated for determining the task-specific weights for Operators that could represent an estimate of the time spent performing each task. Among these, only 5 studies provided data on duration of time spent performing each task;  data from these 5 studies was recorded to determine the median proportion of time that Operators spent performing the ‘mix-load’ versus the ‘apply’ tasks during a work period (Table B.1). The median proportion of time the Operators in these 5 studies spent performing these tasks was 0.20 for ‘mix-load’ and 0.80 for ‘apply’, as shown in Table B.1. Therefore, 0.20 (‘mix-load’) and 0.80 (‘apply’) were set as the task-specific weights for Operators.            128 Table B.1: Study data used for the derivation of the task-specific algorithm weights for Operators, i.e., farm workers who personally handled pesticides to perform both the ‘mix-load’ and ‘apply’ tasks.   Study  Task Duration in Study (minutes)   Mean Task Duration in Study (minutes)  Proportion of Total Time Spent Performing Task in Study*  Mix-Load Apply Mix-Load Apply Mix-Load Apply  Senior et al. (1992)  30  43  30.00  29.60  0.5  0.5 -- 23 -- 23   Lebailly et al. (2009)  37  137  28.70  102.6  0.2  0.8 34 107 15 64   Stewart et al. (1999)   102  1320  102.0  1320  0.1  0.9  Vercruysse et al. (1999)  21  23  11.70  29.25  0.3  0.7 7 19 7 20 -- 55  Dubelman et al. (1982)  2.8  42.6  13.35  45.45  0.2  0.8 17.3 42.6 17.0 69.3 16.3 69.3 -- 27.6 -- 27.6 -- 42.3 -- 42.3   Median proportion of total time spent performing each task across all 5 studies   0.2  0.8 *Total time was determined for each study by adding the average time spent on each task; proportions were rounded to nearest tenth place.  Task-Specific Body Area Exposure Level Weights Body parts were combined into four major areas to coordinate with the self-reported options available in the farm questionnaire (Appendix A), and they include: head/face, body, arms and hands, legs/feet. Considering all studies from the literature review presented in Chapter 2, those that met the following conditions were used to determine median average body area exposure levels for each task: 1. potential dermal data were provided by body part for either one or both pesticide handling tasks (note: if data provided for both tasks, they needed to be provided separately); and 2. potential dermal residue or   129 exposure data could be converted to mg/hour units. There were 10 of 31 literature review studies that provided this type of dermal data for the ‘mix-load’ task, and 15 of 31 studies that provided it for the ‘apply’ task. Using these studies, the median body area exposure level (mg/hr) was determined for each task and used to represent the corresponding body area exposure variable in the above algorithm. Below are the rules that were consistently applied to the collected dermal data: 1. If residue data was provided by body part and task, the data was extrapolated to total body part using the EPA body part surface areas as presented in Vercruysse et al. (1999), since the body part breakdown best represented the data collected.  2. For each study, once the data were averaged across all subjects/trials (if not already done in the study) and converted to mg/hour units, all measured body parts were combined into the four body areas for this algorithm. 3. If dermal data measurements were recorded as <0.01, these values were set to 0.01 for purposes of calculating mean body area exposure values; and when a body part was noted as “none detected’, these measurements were set to 0 for mean calculations.  4. For studies where residue or exposure data were not provided by hour or any unit of time, the study was examined for duration of task information; if found, the average duration of task was calculated and used to determine an hourly rate.    130 Table B.2: Study data used to derive the body part-specific weight of the potential dermal exposure algorithm, representing exposure levels (mg/hour) by pesticide handling task (a. mix-load task; b. apply task).   a. Mix-Load Task Study Units of Exposure Data (conversion details) Study-Specific Average Dermal Exposure/Residue Data by Body Parts Measured   Averaged Exposure Levels for the Four Main Body Areas of the Algorithm (mg/hour)  head, face, neck chest, front of trunk and/or shoulders back upper arms lower arms/ wrists thighs lower legs/feet hands head/ face body arms and hands legs/ feet Senior et al. (1992) mg of active ingredient / 30 min Operation (multiplied by 2.0 for hourly exposure)  0.56 0.02 (chest), 0.02 (abdomen) 0.02 0.26 0.52 0.06 0.24 2.66 0.56 0.06 3.18 0.3 Lonsway et al. (1997) mg/hr 0.03 (back of neck), 0.1 (face and v-neck) 0.33 0.17 0.13 0.3 0.77 0.3 57.83 0.13 0.5 58.26 1.07 Lebailly et al. (2009) mg/30 min operation (multiplied by 2.0 for hourly exposure)  Not Measured 12.4 2.4 3.2 24.6 7.0 14.0 92.0 Not measured 14.8 119.8 21.0 Dubelman et al. (1982)  *Conventional **Closed-system   *µg/cm2 (residue data: multiplied by 20.0 for hourly residue, extrapolated using body part surface areas, and converted to mg) 0.15 (head), 0.67 (forehead) 0.19 0.07 -- -- -- -- 71.2 21.32 18.64 1167.7 (hands only) Not measured   131 **µg/cm2 (residue data: multiplied by 20.0 for hourly residue, extrapolated using body part surface areas, and converted to mg)  0.005 (head), 0.005 (forehead) 0.005 0.005  0.009 (average of samples from thigh, forearm, bicep, ankle) 0.06 0.05 0.12 0.31 0.14 Nigg et al. (1986) µg/cm2 (extrapolated using body part surface areas, and converted to mg)  -- 0.06 0.05 0.06 0.17 0.63 0.71 -- Not measured 0.40 0.38 (arms only) 4.10  Wojeck et al. (1982)  mg/hour 0.2 0.7 0.3 9.9 40.7 51.7 0.2 1.0 67.0 40.7 NIgg et al. (1983) µg/cm2 (extrapolated using body part surface areas, and converted to mg)  -- 0.19 (chest), 0.15 (shoulders) 0.09 -- 0.45 -- 0.66 -- Not measured 1.01 0.54 (lower arms only) 1.57 (lower legs only)  Wojeck et al. (1981)  mg/hour 3.0 (head and neck) 2.7 32.5 12.2 41.0 1712.8 3.0 35.2 1725.0 41.0 Kiefer et al. (1996) µg/cm2 (extrapolated using body part surface areas, and converted to mg)  -- 421 (front), 319 (shoulders) 494 261 300 0.14 0.01 -- Not measured 0.14 (front only) 0.13 (lower arms only) 0.55 Everhart et al. (1982) µg/cm2 (multiplied by 19 for hourly residue, extrapolated using body 69.4 39.6 23.5 -- 193.9 -- -- 12201.8 1.32 1.20 235.5 (lower arms & hands) Not measured   132 part surface areas, and converted to mg)   Median Dermal Exposure Across Studies (mg/hour) 0.56 1.0 58.26  1.57    b. Apply Task Study Units of Exposure Data (conversion details) Study-Specific Average Dermal Exposure/Residue Data by Body Parts Measured   Averaged Exposure Levels for the Four Main Body Areas of the Algorithm (mg/hour)  head, face, neck chest, front of trunk and/or shoulders back upper arms lower arms/ wrists thighs lower legs/feet hands head/ face body arms and hands legs/ feet Senior et al. (1992) mg/hour 0.45 0.01 (chest), 0.01 (abdomen) 0.02 0.27 0.15 0.02 0.02 0.45 0.04 0.29 0.17 Lonsway et al. (1997) mg/hour 8.8 (face + v-neck), 4.3 (back of neck)  6.6 0.2 0.3 0.7 1.37 0.63 13.5 13.1 6.8 14.5 2.0 Wojeck et al. (1983) µg/cm2 (extrapolated using body part surface areas, and converted to mg)  -- 2.14 (trunk), 0.81 (shoulders) 2.06 4.46 2.39 2.0 15.08 Not measured 5.01 19.5 4.39 Lebailly et al. (2009) mg (multiplied by 0.58 for hourly exposure)  -- 2.5 1.0 2.1 2.8 0.9 2.8 24.6 Not measured 2.03 17.1 2.14  WInterlin et al. (1986)  mg/hour -- 0.08 1.2 0.93 3.91 0.52 0.94 Not Measured 1.28 1.87 4.43   133 Vercruysseet al. (1999)  µg/kg AI (multiplied residue by 2.0 for hourly exposure and converted to mg mg/kg AI/hour  24.69 46.04 78.17 23.98 7.61 52.45 31.9 108.5 0.05 0.25 0.29 0.16 Dubelman et al. (1982) µg/ cm2 (multiplied by 1.3 for hourly, then extrapolated using body part surface areas and converted to mg)  0.045 (head), 0.093 (forehead) 0.12 0.09 -- -- -- -- 0.27 0.23 0.99 0.27 (hands only) Not measured Abbott et al. (1987) % exposure  (multiplied each value by averaged total body exposure)  -- 2.46 0.90 4.02 2.94 5.64 84.4 Not measured 2.76 72.7 6.98 Nigg et al. (1986) µg/ cm2 (extrapolated using body surface areas and converted to mg)  -- 0.64 (front), 0.62 (shoulders) 0.26 1.08 1.06 1.94 0.92 -- Not measured 3.28 4.42 (arms only) 9.60  Wojeck et al. (1982)  mg/hour 1.0  3.8 2.5 10.0 23.6 28.3 1.0 6.3 38.3 23.6 Nigg et al. (1983)  µg/ cm2 (extrapolated using body surface areas and converted to mg)   -- 1.13 (front), 1.68 (shoulders) 0.47 -- 3.82 -- 1.46 -- Not measured 5.93 4.62 (lower arms only) 3.47 (lower legs only)   134 Sanderson et al. (1995) µg/ cm2 (2.9 to convert to hourly residue, extrapolated using body surface areas and converted to mg)  -- 1.24 1.36 2.84 29.9 -- Did not include since most wore gloves Not measured 18.46 23.4 (upper and lower arms only) 29.9 (upper  legs only) Carman et al. (1982) µg/ cm2 /hour (extrapolated using body surface areas and converted to mg)  3.3 1.04 0.61 4.40 1.90 3.55 -- -- Not measured 6.36 15.1 (arms only) 13.6 (upper legs only) Wojeck et al. (1981) mg/hour 24.4 91.6 6.30 11.6 88.2 1749.3 24.4 97.9 Excluded as outlier 88.2 Simpson et al. (1965) µg/100cm2/ hour (extrapolated using body part surface areas, then converted to mg/hour)  365  421 (front), 319 (shoulders) 494 261 300 -- -- -- 4.75 33.0 11.20 (arms only) Not measured    Median Dermal Exposure Across Studies (mg/hour)  1.0 5.01 12.65 4.43  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0378062/manifest

Comment

Related Items