Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Surveillance of asthma in relation to work among Canada's adult population Garzia, Nichole Andrea 2008

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

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
24-ubc_2008_spring_garzia_nichole_andrea.pdf [ 4.47MB ]
Metadata
JSON: 24-1.0066676.json
JSON-LD: 24-1.0066676-ld.json
RDF/XML (Pretty): 24-1.0066676-rdf.xml
RDF/JSON: 24-1.0066676-rdf.json
Turtle: 24-1.0066676-turtle.txt
N-Triples: 24-1.0066676-rdf-ntriples.txt
Original Record: 24-1.0066676-source.json
Full Text
24-1.0066676-fulltext.txt
Citation
24-1.0066676.ris

Full Text

SURVEILLANCE OF ASTHMA IN RELATION TO WORK AMONG CANADA'S ADULT POPULATION by Nichole Andrea Garzia B.A., University of Washington, 2005 A THESIS SUBMITTED IN PARTIAL FULLFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in The Faculty of Graduate Studies (Occupational and Environmental Hygiene) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) April 2008 Nichole Andrea Garzia, 2008 Abstract Work-related asthma surveillance is needed to improve management of occupational exposures, clinical recognition/diagnosis, and worker compensation policies. This work investigated asthma in relation to work by evaluating the utility of existing Canadian surveillance data in providing useful information about the burden of work-related asthma; estimating the burden of work-related asthma among Canada's adult population; and evaluating the effect of job risk on asthma after considering other potential risk factors for asthma. The working population formed samples from two Statistics Canada surveillance programs: Canadian Community Health Survey (CCHS), 2002/03 Cycle 2.1 and National Population Health Survey (NPHS), Longitudinal Component (1994/95-2002/03). Both surveys enquired about health professional-diagnosed asthma; NPHS additionally asked age at time of diagnosis, so adult-onset versus childhood-onset asthma was determined. Both surveys enquired about current job held; corresponding job codes were linked to an asthma-specific job exposure matrix to judge job risk for occupational asthma. CCHS only provided current job information, in contrast, NPHS longitudinal data was used to determine job held at time of asthma-onset. Statistical measures for asthma in relation to job risk were estimated. CCHS results were likely biased by the healthy worker effect, as it showed the opposite effect of job risk on asthma than the NPHS; higher asthma prevalence was shown for NPHS men and women in high risk jobs. NPHS results indicated a large burden of adult- onset asthma among men (19,000) and childhood-onset asthma among women (17,000) attributed to working in high risk jobs for occupational asthma. Using NPHS, adjusted and crude prevalence odds ratio estimates were compared to further assess effect of job risk on asthma. For adult-onset asthma, there was no difference between estimates (men: 1.8, women: 1.1); for childhood-onset asthma, adjusted estimates were larger than crude, respectively (men: 1.3 v 1.2, women: 2.0 v 1.7). ii Age of asthma-onset and job held at time of asthma-onset is necessary surveillance information for estimating work-related asthma. There may be increased risk of work "caused" asthma among men and work "exacerbated" asthma among women in high risk jobs. Considering other risk factors for asthma did not reduce effect of job risk on asthma. iii Table of Contents Abstract^ ii Table of Contents^ iv List of Tables vi List of Figures^ viii Disclaimer ix Acknowledgements^ x Dedication^ xi Co-Authorship Statement^ xii Chapter 1. Introduction 1 Occupational Exposures^  1 Surveillance of Work-related asthma^ 3 Research Objectives^  5 Bibliography 6 Chapter 2. The burden of work-related asthma in Canada: an evaluation of data from two national surveillance programs^ 9 Introduction^ 9 Methods  10 Results^  14 Discussion  19 Bibliography^ 24 Chapter 3. Asthma in Relation to Job Risk Group: Adjusting for Other Potential Risk Factors^ 27 Introduction 27 Methods^ 28 iv Results^ 32 Discussion 43 Bibliography^ 46 Chapter 4. Discussion^ 49 Summary of Key Messages 49 Strengths and Limitations^ 53 Future Perspective 56 Bibliography^ 61 Appendices 64 Appendix A: Crude Estimates of Prevalence Odds Ratios (POR)^ 64 Appendix B: Variable Descriptions^ 66 Appendix C: Bivariate Associations between Predictor Variables^  70 Appendix D: Multivariable Model Results for Childhood-Onset Asthma among Men and Women (household income included as a predictor)^  91 Appendix E: Recoding of the Asthma-Specific Job Exposure Matrix^ 93 v List of Tables Table 2. 1 Descriptive statistics for Canadian Community Health Survey (CCHS) and National Population Health Survey (NPHS) study samples^ 14 Table 2. 2 National Population Health Survey (NPHS) study sample: The burden of work-related asthma in Canada as estimated with PAR% and the Attributable Number (AN) of prevalent asthma cases associated with HR jobs (stratified by sex and age of asthma-onset). ^ 19 Table 3. 1 National Population Health Survey (NPHS) sample by demographic characteristics for men and women^ 32 Table 3.2 Estimates of asthma outcome prevalence (%) by potential predictor variables ^ 35 Table 3. 3 Final multivariable logistic regression results: POR estimates for associations between job risk group and asthma, adjusted for other predictors and stratified by gender^ 40 Table A. 1 Prevalence Odds Ratio: Adult-Onset Asthma among Men^ 64 Table A. 2 Prevalence Odds Ratio: Adult-Onset Asthma among Women 64 Table A. 3 Prevalence Odds Ratio: Childhood-Onset Asthma among Men^ 65 Table A. 4 Prevalence Odds Ratio: Childhood-Onset Asthma among Women 65 Table B. 1 Description of asthma outcome and predictor variables considered defined for multivariate logistic regression analyses (supplement of Chapter 3) 66 Table C. 1 Adult-onset asthma among women: association between Age and Body Mass Index 70 Table C. 2 Adult-onset asthma among women: association between Job Risk Group and Age   71 Table C. 3 Adult-onset asthma among women: association between Job Risk Group and Body Mass Index^ 72 Table C. 4 Childhood-Onset Asthma among Men: association between Age and Household Income^ 73 Table C. 5 Childhood-Onset Asthma among Men: association between Age and Body Mass Index^ 74 Table C. 6 Childhood-Onset Asthma among Men: association between Job Risk Group and Household Income^ 75 Table C. 7 Childhood-Onset Asthma among Men: association between Household Income and Race^ 76 Table C. 8 Childhood-Onset Asthma among Men: association between Household Income and Body Mass Index^ 77 Table C. 9 Childhood-Onset Asthma among Men: association between Race and Body Mass Index^ 78 vi Table C. 10 Childhood-Onset Asthma among Women: association between Age and Household Income^ 79 Table C. 11 Childhood-Onset Asthma among Women: association between Age and Body Mass Index^ 80 Table C. 12 Childhood-Onset Asthma among Women: association between Age and Smoking Status^ 81 Table C. 13 Childhood-Onset Asthma among Women: association between Education and Household Income^ 82 Table C. 14 Childhood-Onset Asthma among Women: association between Education and Smoking Status^ 83 Table C. 15 Childhood-Onset Asthma among Women: association between Job Risk Group and Education^ 84 Table C. 16 Childhood-Onset Asthma among Women: association between Job Risk Group and Age^ 85 Table C. 17 Childhood-Onset Asthma among Women: association between Job Risk Group and Household Income^ 86 Table C. 18 Childhood-Onset Asthma among Women: association between Job Risk Group and Body Mass Index^ 87 Table C. 19 Childhood-Onset Asthma among Women: association between Age and Education88 Table C. 20 Childhood-Onset Asthma among Women: association between Household Income and Smoking Status^ 89 Table C. 21 Childhood-Onset Asthma among Women: association between Body Mass Index and Smoking Status^ 90 Table D. 1 Multivariable logistic regression results (prevalence odds ratios) for childhood-onset asthma among men and women, considering household income as a predictor variable ^ 91 vii List of Figures Figure 2.1 Canadian Community Health Survey (CCHS) versus National Population Health Survey (NPHS): current asthma prevalence reported in 2002/03, by job risk exposure group and gender.^  15 Figure 2.2 Population Health Survey (NPHS) study sample: asthma prevalence by age of asthma- onset, job risk group and gender.^  16 Figure 2.3 National Population Health Survey (NPHS) study sample: asthma prevalence in relation to two HR exposure categories, by age of asthma-onset. ^  17 Figure 2.4 National Population Health Survey (NPHS) study sample: adult-onset asthma prevalence by all HR exposure categories and combined HR exposure categories^ 18 viii Disclaimer It is required that the following statement be provided when using Statistics Canada data for research: "The research and analysis are based on data from Statistics Canada and the opinions expressed do not represent the views of Statistics Canada." ix Acknowledgements I would like to thank Dr. Susan Kennedy, my thesis supervisor. Susan's expertise taught me several valuable lessons about research, and her guidance helped me to grow and become confident in my work and abilities as a researcher. It was a privilege to have the opportunity to work with her. I would also like to thank Dr. Mieke Koehoorn and Dr. Paul Demers for their expertise, support, and contribution to my thesis work. I was fortunate to have a committee so strong in knowledge and experience. I would like to acknowledge the British Columbia Research Data Centre, where I accessed the Statistics Canada data, with a further extension of gratitude to Lee Grenon for his guidance in using the Centre and its sources. I would also like to acknowledge AllerGen NCE Inc for their funding support. Lastly, I'd like to thank my parents and my husband for all of their love and support. My parents have instilled in me the importance of education and dedication to work. My husband's faith in my capabilities has provided me with the drive and determination to continuously push myself forward. Mom, dad and Chris — thank you. IX syy,9 "puvgsny 'Cut oL Co-Authorship Statement The research, methodology, analyses and written documents for this thesis were developed and performed by me, and in consultation with Dr. Susan Kennedy. Dr. Susan Kennedy is the author of the asthma-specific job exposure matrix, which was a major component to the analyses of this work. I recoded the asthma-specific job exposure matrix for the purpose of this work, and received guidance from Susan Kennedy on its use, application, and interpretation. I prepared the manuscripts and received feedback from all co-authors: Dr. Mieke Koehoorn, Dr. Susan Kennedy, and Dr. Paul Demers. xii Chapter I. Introduction The prevalence of asthma has been increasing over time among populations of industrialized nations (1) . Among adults, a large proportion of this asthma increase is attributed to workplace exposures in the form of work-related asthma. Work-related asthma imposes a heavy personal and social burden. Those who suffer from work-related asthma are risking their health, jobs and quality of life (2 ' 3) due to misdiagnosis, improper treatment and continued exposure to an offending agent (4) . Canada's public health care system suffers greatly from this situation as well, because individuals with uncontrolled work-related asthma require more medical assistance and are more likely to be hospitalized (5), even compared to asthmatics whose asthma is not work-related (6) . Work-related asthma (WRA) can be either "caused" or "exacerbated" by workplace exposures. Work "caused" asthma is referred to as occupational asthma, resulting from an allergic (with latency) or an irritant-induced mechanism (7 ' 8) . A formal and accepted definition of occupational asthma states that the condition is "characterized by variable airflow limitation and/or airway hyperresponsiveness due to causes and conditions attributable to a particular occupational environment and not stimuli encountered outside the workplace" (9) . Work-exacerbated asthma, also known as work-aggravated asthma, is a "pre-existing asthma that is persistent since childhood or reactivated in adulthood" X10) due to workplace exposures. Occupational Exposures Types of Occupational Exposures More than 350 workplace exposures have been related to work-exacerbated asthma, including "chemicals, mixtures, and processes", and more than 250 of these exposures have been identified as occupational causes of asthma (1 ' 6) . These workplace exposures are classified based on their molecular weights (i.e. as high or low molecular weight agents). High molecular weight agents (> 5 kilodaltons) are primarily proteins (11) such as animal, 1 fish/shellfish, and flour-associated antigens (12) . High molecular weight agents cause asthma through an immunologic, IgE-dependent mechanism (8) . Low molecular weight agents (< 5 kilodaltons), such as isocyanates and acid anhydrides (11, 12), can also cause asthma through an immunologic, IgE-dependent mechanism. However, the mechanisms by which low molecular weight agents cause asthma is not well understood, as some mechanisms of these agents are still unknown (8) . There is often a higher risk of developing occupational asthma after exposure to low molecular weight agents compared to high molecular weight agents, because the time needed for sensitization to occur is shorter for low molecular weight agents (13) . Awareness, Prevention and Management Currently, there exists little awareness about high risk exposures in the workplace and among physicians (13) . Physicians rarely inquire about occupational history or consider the possibility of work-related asthma among their patients (13). However, for prevention of work-related asthma to be effective, workplaces must appropriately manage exposures that can "cause" or "exacerbate" asthma, and physicians must consider the possibility of work-related asthma among all adults presenting with asthma (13) . Workplaces and physicians should be aware of all known and suspected exposures and occupations where work-related asthma is a possibility. In addition, "exposures levels at which immunological sensitization is unexpected to occur", known as permissible exposure levels, should be known for all high risk agents (9, 13) and adherence to these levels should be mandatory for all applicable working environments. Permissible exposure levels cannot control all work-related asthma, however, as they would not be effective for individuals at risk of work "exacerbated" asthma or pre-diagnosed with occupational asthma (9) . Workplaces and physicians must also be aware that different forms of work-related asthma require different workplace management and treatment (14) . Occupational asthma is a permanent condition, but severity of asthma is related to whether or not exposure to the offending agent continues (13) . Both early diagnosis and removal from the offending agent can keep asthma from worsening (3) . It's estimated that 75% of workers diagnosed with occupational asthma will continue to have "permanent bronchial hyperresponsiveness" (9) , especially if re-exposed to any level of the original offending 2 agent (4) . Workers with occupational asthma may have to change jobs if exposure to the offending agent cannot be terminated (4) . Work-exacerbated asthma can result in increased severity and "chronic impairment" with continued exposure to an offending agent (15) ; early recognition for removal or exposure reduction to the agent is necessary (4, 15) Workers with work-exacerbated asthma usually do not need to leave or change their job, as long as appropriate treatment is provided and workplace control measures are implemented to reduce exposure to a minimum if complete removal is not an option (4) . Personal protective equipment (e.g. masks and respirators (13)), engineering controls, and minimized duration and intensity of exposure are essential workplace interventions for work-exacerbated asthma. Surveillance of Work-related asthma Current Surveillance: how much asthma is work-related? Current estimates of asthma attributable to workplace exposures vary widely based on the type of work-related asthma and population under investigation, as well as the methodology used for defining asthma and exposure. A meta-analysis based on estimates from several studies concluded that "about 1 in 10 cases of adult asthma, including new onset disease and reactivation of pre-existing asthma" is attributable to occupational exposures (10) . Among the studies considered in this meta-analysis by Blanc, P.D. et al (1999), only one study accounted for the different types of work-related asthma in their estimates; 19% of the asthma cases were estimated to be "re-activation" of asthma (i.e. work-exacerbated asthma), and 26% of asthma cases were estimated to be "new-onset asthma" (i.e. occupational asthma) (101 . A study that examined risk of occupational asthma in relation to occupation and reported exposures, estimated that "18.2% of cases of adult- onset asthma [among Canada's] urban population" were associated with "high risk occupations/industries and exposures" (16) . A case-control study of a Swedish population estimated the attributable fraction of adult asthma "attributed to occupational exposures" to be 11%, after adjusting for other potential risk factors (17) . A literature review performed by the American Thoracic Society, suggests that 15% of asthma is "likely to be 3 work-related" (18) . Essentially, current surveillance provides estimates suggesting that a 743, , ,10,16,18)range of 10% to 18% of adult asthma may be work-related ( Limitations of Current Surveillance In general, there are few population-based studies that have examined the effect of work on asthma (19) . Among the studies that are available, there are several gaps in the knowledge they provide about population risks of work-related asthma. For instance, most population risk estimates focus only on occupational asthma (7) , when "there may be much greater morbidity and productivity loss associated with exacerbations of pre-existing asthma" (18) . Future estimates of work-related asthma must consider both occupational asthma and work-exacerbated asthma to fully understand the contribution of work on asthma in a population, and for incorporation of proper workplace management and intervention for the different forms of work-related asthma (14) . It is difficult for population-based studies to provide accurate estimates of work-related asthma, because many are based on cross-sectional data that have limited information on exposure and disease (19). Studies that rely on self-reported exposure can be influenced by recall bias, and studies that rely solely on job titles are often criticized for using "poor indicator[s] of exposure" (19) . Population-based studies should use advanced exposure assessment methods to reduce bias and improve validity of estimates for work-related asthma. An asthma-specific job exposure matrix has been developed by Susan M. Kennedy to classify jobs based on exposure to "known risk factors" for occupational asthma (12) . Unlike other exposure assessment methods available, this is an exposure assessment tool with two powerful components: it specifically identifies exposures for asthma based on what is known about workplace causes of asthma, and it includes a "verification step" for expert judgment of exposure (12) . The job exposure matrix includes "specific" high molecular weight and low molecular weight agents, as well as mixed environment or agent exposures (e.g. metal working fluids) associated with occupational asthma (12) . This is a robust exposure assessment tool, linking job codes that have a high probability of exposure to known risk factors for occupational asthma (12) . 4 Significance of Work-Related Asthma Surveillance Surveillance is a primary step for reducing and controlling work-related asthma in a population. It is needed to identify where the burden of work-related asthma exists among a population, through identification of at-risk groups and occupational exposures. Occupations exposures can vary between populations (13), so surveillance must be population specific. The role of surveillance is intended to improve awareness in a population and among key stakeholders. Awareness will allow for efficient control of workplace exposures and proper diagnosis of asthmatic patients. In addition, the knowledge provided about the burden of work-related asthma in a population will allow future research to appropriately focus in areas of need, i.e. specific at-risk groups. Research Objectives There are three main objectives for this thesis research that incorporate what is known and what is needed in the current realm of work-related asthma surveillance. This research will focus on work-related asthma in Canada; however, there are many lessons and strengths associated with this research that make it a strong addition to the field of work- related asthma in general. The three objectives for this thesis research are: 1. To evaluate the utility of existing Canadian surveillance data to provide information on the burden of work-related asthma; 2. To identify occupational exposure groups at high risk for work-related asthma; and 3. To evaluate the effect of job risk on asthma after considering other potential risk factors for asthma. 5 Bibliography 1. Petsonk EL. Work-Related Asthma and Implications for the General Public. Environ Health Perspect. 2002 Aug;110 Suppl 4:569-72. 2. Blanc PD, Trupin L, Eisner M, Earnest G, Katz PP, Israel L, et al. The Work Impact of Asthma and Rhinitis: Findings from a Population-Based Survey. J Clin Epidemiol. 2001 Jun;54(6):610-8. 3. Tarlo SM, Liss GM. Prevention of Occupational Asthma - Practical Implications for Occupational Physicians. Occupational Medicine. 2005;55(8):588-94. 4. Youakim S. Work-Related Asthma. Am Fam Physician. 2001 Dec 1;64(11):1839-48. 5. Liss GM, Tarlo SM, Macfarlane Y, Yeung KS. Hospitalization among Workers Compensated for Occupational Asthma. Am J Respir Crit Care Med. 2000 Jul;162(1):112- 8. 6. Breton CV, Zhang Z, Hunt PR, Pechter E, Davis L. Characteristics of Work Related Asthma: Results from a Population Based Survey. Occup Environ Med. 2006 Jun;63(6):411-5. 7. Malo JL. Future Advances in Work-Related Asthma and the Impact on Occupational Health. Occupational Medicine. 2005;55(8):606-11. 8. Chan-Yeung M, Malo JL. Aetiological Agents in Occupational Asthma. Eur Respir J. 1994 Feb;7(2):346-71. 6 9. Malo JL, Chan-Yeung M. Occupational Asthma. J Allergy Clin Immunol. 2001 Sep;108(3):317-28. 10. Blanc PD, Toren K. How Much Adult Asthma can be Attributed to Occupational Factors? Am J Med. 1999 Dec;107(6):580-7. 11. Toren K, Brisman J, Olin AC, Blanc PD. Asthma on the Job: Work-Related Factors in New-Onset Asthma and in Exacerbations of Pre-Existing Asthma. Respir Med. 2000 Jun;94(6):529-35. 12.Kennedy SM, Le Moual N, Choudat D, Kauffmann F. Development of an Asthma Specific Job Exposure Matrix and its Application in the Epidemiological Study of Genetics and Environment in Asthma (EGEA). Occup Environ Med. 2000 Sep;57(9):635- 41. 13. Chan-Yeung M, Malo JL, Tarlo SM, Bernstein L, Gautrin D, Mapp C, et al. Proceedings of the First Jack Pepys Occupational Asthma Symposium. Am J Respir Crit Care Med. 2003 Feb 1;167(3):450-71. 14. Liss GM, Tarlo SM. Work Related Asthma. Occup Environ Med. 2002 Aug;59(8):503-4. 15. Goe SK, Henneberger PK, Reilly MJ, Rosenman KD, Schill DP, Valiante D, et al. A Descriptive Study of Work Aggravated Asthma. Occup Environ Med. 2004 Jun;61(6):512-7. 16. Johnson AR, Dimich-Ward HD, Manfreda J, Becklake MR, Ernst P, Sears MR, et al. Occupational Asthma in Adults in Six Canadian Communities. Am J Respir Crit Care Med. 2000 Dec;162(6):2058-62. 7 17. Toren K, Balder B, Brisman J, Lindholm N, Lowhagen 0, Palmqvist M, et al. The Risk of Asthma in Relation to Occupational Exposures: A Case-Control Study from a Swedish City. Eur Respir J. 1999 Mar;13(3):496-501. 18. Balmes J, Becklake M, Blanc P, Henneberger P, Kreiss K, Mapp C, et al. American Thoracic Society Statement: Occupational Contribution to the Burden of Airway Disease. Am J Respir Crit Care Med. 2003 Mar 1;167(5):787-97. 19. Le Moual N, Kennedy SM, Kauffmann F. Occupational Exposures and Asthma in 14,000 Adults from the General Population. Am J Epidemiol. 2004 Dec 1;160(11):1108- 16. 8 Chapter 2. The burden of work-related asthma in Canada: an evaluation of data from two national surveillance programs Introduction The goal of this study was to evaluate the utility of existing Canadian surveillance data to provide information on the burden of work-related asthma, and to identify occupational exposure groups at high risk for asthma related to work. More than 250 agents have been identified as causing work-related asthma (1-3), and they contribute greatly to the burden of asthma among adults in developed nations (1 ' 3 ' 4) . As well, new substances are continuously introduced into the workplace before understanding their full effects on human health. Work-related asthma (WRA) includes both Occupational Asthma (OA) defined as asthma "due to causes and conditions that are attributable to a particular occupational environment and not to stimuli encountered outside the workplace" (5) and Work-Exacerbated Asthma (WEA), defined as the exacerbation or aggravation of pre-existing childhood asthma or asthma that "reactivates in adulthood" due to specific workplace exposure(s) (6) . It is important to know how OA and WEA individually contribute to the overall burden of WRA because "the treatment and intervention implications are quite different for [these] different types of WRA" (7) . Several studies using different methods for defining work-related asthma and high risk exposures have estimated from 10% to 18% of all adult asthma may be attributable to work exposures (population attributable risk or PAR) (6, 8-12). However, population-based studies that estimate the PAR usually do not differentiate between the types of WRA or investigate the types of high risk exposures contributing to these estimates of burden. An integrative understanding of both the type of WRA workers suffer from (OA versus A version of this chapter will be submitted for publication. Garzia NA, Koehoorn M, Demers PA, Kennedy SM. The burden of work-related asthma in Canada: an evaluation of data from two national surveillance programs. 9 WEA) (7) and the type of exposure (e.g. high molecular weight versus low molecular weight agents) are necessary for a complete understanding of the burden of WRA and for targeting high risk groups for prevention and intervention. For this investigation, we used data from two Canadian surveillance programs with different asthma and work information; the Canadian Community Health Survey (CCHS) and the National Population Health Survey (NPHS), Longitudinal Household Component (13, 14) This study is the first that we are aware of to estimate the burden of WRA using Canadian survey data, while considering both the type of WRA and the type of high risk exposure. This research has important implications for alleviating the personal and social burden (i.e. public health care) of this disease in Canada, and for planning future asthma surveillance programs. Methods Surveillance Data and Survey Samples Both CCHS and NPHS were developed and conducted by Statistics Canada. Both are telephone surveys, with sophisticated random sampling designs and extensive quality 13, 14) .control procedures. Response rates were approximately 81% for both surveys (s^The estimates derived from them are intended to be representative across Canada's population, excluding "persons living on Indian Reserves, Crown Lands, Residents of health institutions, full-time members of the Canadian Forces Bases and some remote areas in Ontario and Quebec" (13) . The CCHS is a cross-sectional survey that was initiated in 2000; it is continuously conducted on a two-year cycle (14). For this analysis, we used data from Cycle 2.1 (2002/03). The NPHS, Longitudinal Household Component, is a longitudinal survey with multiple survey cycles; data from the first five cycles were used for this study because the fifth cycle was conducted in the same year as CCHS (1994/95 to 2002/03). For this analysis, inclusion criteria for both surveys were being between 15 and 65 years of age (inclusive) and working full-time (> 30 hours per week) in 2002/03. For NPHS, all 10 respondents were required to have full responses to survey cycles 1 (1994/95) and 5 (2002/03). Asthma Outcomes In both surveys, the Statistics Canada interviewer provided respondents with a preamble to prepare them for a series of questions about chronic conditions "that have lasted or are expected to last 6 months or more and that have been diagnosed by a health professional", and then respondents were asked "do you have asthma?" X15,16)  For this study, those who responded "yes" to having health professional-diagnosed asthma in 2002/03 were classified as current asthmatics for 2002/03. In the NPHS longitudinal survey, respondents were further asked either "when were you diagnosed with this asthma?" or "how old were you when this asthma was diagnosed?" (l5) and this information was used to classify current asthmatics as having either childhood-onset (diagnosed before 18 years of age) or adult-onset (diagnosed at or after 18 years of age) asthma. Exposure Assessment Both surveys enquired about the current job held by respondents and classified this job using the Standard Occupational Classification (SOC) 1991 (17) . For CCHS, exposure estimation could only be determined for the currently held job. In contrast, NPHS supplied 5 survey cycles worth of information on the job held at the time of each survey. For respondents with adult-onset asthma, this information was used, together with data on the age asthma-onset*, to identify the job held at time of asthma-onset. Such identification was possible for most (84%) adult-onset asthmatics. For all non-asthmatics and childhood-onset asthmatics whose asthma persisted into adulthood, the current job held in 2002/03 was used for exposure estimation. The identified job code (either current job or job at time of asthma-onset) was then linked to an asthma-specific job exposure matrix (JEM) (18) designed to classify each person as * Reported age of diagnosis was used to approximate age of asthma-onset in order to decipher between childhood-onset and adult-onset asthma. 11 having either a high risk (HR) or low risk (LR) job with respect to the development of occupational asthma. As the published version of this JEM was based on a different job coding system (18), it was necessary to create a new version of the JEM for this study, using SOC-1991 codes (further details in Appendix E). This asthma-specific JEM classifies a job as being at HR for OA if there is a high probability of exposure to any one of 17 agents or combinations of agents in the matrix for the majority of people working in that job (18) . The matrix 'agents' are grouped as 'high molecular weight (HMW) agents' (9 specified agents, e.g. animal, fish/shellfish, and flour-associated antigens), low molecular weight (LMW) agents' (5 specified agents, e.g. highly reactive chemicals, metal and metal fume antigens, and reactive cleaning/disinfectant products) and 'mixed environments' (3 specified environments, e.g. metal working fluid exposures, textile production and agricultural antigens) (18) . Combined exposure categories were also considered for analyses; these included workers in HR jobs where the asthma-specific JEM identified exposure to more than one agent from different HR exposure categories (e.g. a worker exposed to both a HMW agent and a LMW agent). Only workers in HR jobs could have exposure to an agent in one of the HR exposure categories or combined exposure categories, all other workers were judged to be in LR jobs and considered unexposed. Statistical Analyses SAS, version 9.1 (SAS Institute Inc., Cary, NC, USA) was used for analyses. Prevalence estimates were generated and weighted using survey weights developed by Statistics Canada, which consider individual probability of selection and adjustment for survey lad4)^•cycle non-response (13, ^Chi-square tests of association for 2-way contingency tables were performed to compare asthma prevalence estimates between exposure groups; p- values <0.05 indicated a statistical association between asthma and exposure. A benefit of using these national surveillance programs is that they provide estimates intended to reflect Canada's population, therefore, we calculated the attributable number of adults in Canada that may be suffering from WRA. The attributable number represents 12 an estimate of the number of prevalent asthma cases attributable to working in HR jobs (19) . Separate estimates for age of asthma-onset and gender were calculated. Attributable Number = Ne(PcPu) Where Ne= number of exposed persons (e.g. women working in HR jobs); Pe prevalence among exposed persons; Pu= prevalence of unexposed persons. A common epidemiological measure for estimating the burden of disease related to a particular exposure is the population attributable risk percent (PAR%) (9) . We calculated the PAR% using the "exposure approach", or the fraction of exposed population to total population, and the estimated crude Relative Risk (RR) of asthma that is associated with exposure (9) . Estimates of PAR% were calculated by age of asthma-onset and gender. PAR% = [p (RR — 1)/p (RR —1) + 1] x 100 Where p = NANI+No]; N1 = exposed population, No = unexposed population (9) . 13 Results Description of Study Samples Table 2. 1 Descriptive statistics for Canadian Community Health Survey (CCHS) and National Population Health Survey (NPHS) study samples Descriptive estimate CCHS NPHS Men Women Men Women Unweighted Sample Size (n) (*Weighted Sample Size) 30,962 (7,358,626) 25,255 (5,286,969) 3,045 (7, 412, 489) 2,662 (5,353,399) Age (years) Mean 39.9 39.3 40.3 39.2 Standard Deviation 11.9 10.9 12.3 10.5 Range 15-65 15-65 15-65 15-65 Self-Reported Asthma Prevalence (%) in 2002/03 6.0 9.4 5.9 8.7 Sample working in HR jobs (%) 13.9 15.9 13.3 14.7 *Weights were developed by Statistics Canada for each survey respondent; they were applied to provide representative estimates across most of Canada's population (13) . The CCHS and NPHS samples were similar, as shown in Table 2.1. However, reported asthma prevalence for 2002/03 was higher among women compared to men in both samples, especially among women in the CCHS sample where the highest prevalence of exposed workers was also observed. Comparison of findings: CCHS and NPHS The prevalence of current asthma from each of the 2 surveys, stratified by job risk group and gender, is shown in Figure 2.1. As evident in the figure, the apparent association 14 ■ High Risk m Low Risk 12.0 — 10.0 cv 8.0 —^* C.) CD 6.0 -To i_w 4.0 — CL 2.0 — 0.0 Women I Men I Women CCHS^I^NPHS Men * between job risk and asthma was opposite when comparing the 2 surveys, for both men and women. For CCHS, using current job only, the prevalence of current asthma was higher among men and women working in LR jobs compared to HR jobs. For NPHS, using job held at time of asthma-onset for adult-onset asthmatics and current job for childhood-onset asthmatics, the prevalence was higher among men and women in HR jobs compared to LR jobs. This analysis was redone including respondents with part-time employment (<30 hours per week) in 2002/03 as well, but the findings did not differ. Figure 2.1 Canadian Community Health Survey (CCHS) versus National Population Health Survey (NPHS): current asthma prevalence reported in 2002/03, by job risk exposure group and gender. *p<0.05 for difference between high risk and low risk job exposure groups Further exploration of job risk and asthma using NPHS data For the NPHS survey we were also able to separately assess the effect of job risk group on adult-onset and childhood-onset asthma, by gender (Figure 2.2). An increased prevalence of current asthma for adult-onset and childhood-onset was observed for both men and women working in HR jobs. The greatest difference in prevalence between job risk groups was seen for adult-onset asthma among men, and childhood-onset asthma among women. 15 Childhood-onset 6.0 5.0 e• 4.0 L 1.0 0.0 Figure 2.2 Population Health Survey (NPHS) study sample: asthma prevalence by age of asthma-onset, job risk group and gender. *p<0.05 for difference between high risk and low risk job exposure groups HR exposure categories were investigated in more detail to better understand the asthma prevalence observed among men and women working in HR jobs. Figure 2.3 shows current asthma prevalence by adult-onset and childhood-onset status stratified by two HR exposure categories (HMW agents and LMW agents). Further stratification by gender was not possible for this analysis due to sample size and Statistics Canada disclosure constraints. Results show a significant association between adult-onset asthma and exposure to HMW agents and between both adult-onset and childhood-onset asthma and exposure to LMW agents. 16 ■ Exposed • Not Exposed 9.0^** 8.0 7.0 ;;:' 6.0 a, 5.0 2 4.0 S1) 3.0 > 2.0 1.0 0.0 HMW I LMW I HMW^LMW Adult-onset^Childhood-onset Figure 2.3 National Population Health Survey (NPHS) study sample: asthma prevalence in relation to two HR exposure categories, by age of asthma-onset. *p<0.05, **p<0.0001 for difference between exposed and not exposed groups Adult-onset asthma was further assessed across all three HR exposure categories and two combined HR exposure categories (exposure to a HMW and LMW agent, and exposure to HMW agent and a mixed environment) as shown in Figure 2.4. The prevalence of adult- onset asthma was significantly greater across nearly all HR exposure categories when compared to unexposed workers, with the highest prevalence in jobs with exposure to multiple risk categories. 17 14.0 - 12.0 - R.' 10.0 - --- a) 8.0 - * ** ** ** 6.0 a 4.0 2.0 - 0. 0 Not^Mixed LMW HMW HMW + HMW + Exposed^Environ. LMW Mixed Environ. Exposure Figure 2.4 National Population Health Survey (NPHS) study sample: adult-onset asthma prevalence by all HR exposure categories and combined HR exposure categories. *p<0.05, **p<0.0001 for difference between exposed groups and not exposed group Although Statistics Canada privacy policies prohibit listing of asthma prevalence estimates by individual job titles (due to small sample size), it is possible to provide examples of the high risk jobs for which the highest asthma prevalence estimates were observed. For adult-onset asthma these included: mining, oil and gas; art, culture, recreation and sport; fishing and fish processing; agriculture; other manufacturing (plastics, rubber, furniture, boats, upholstery); nurses and nursing aides; metal trades; and domestic and personal services. For childhood-onset asthma, high prevalence was seen for: health technologists; nurses and nursing aides; domestic and personal services; construction; and metal trades. Among all of these jobs listed above, women dominated the following few: nurses and nursing aides, domestic and personal services, and health technologists. Finally, the burden of work-related asthma in Canada using the estimated attributable number of prevalent asthma cases related to working in HR jobs and the PAR% are presented in Table 2.2. These estimates suggest that the burden of work-related asthma is 18 most pronounced for men with adult-onset asthma and for women with childhood-onset asthma, likely OA and WEA respectively. Table 2. 2 National Population Health Survey (NPHS) study sample: The burden of work-related asthma in Canada as estimated with PAR% and the Attributable Number (AN) of prevalent asthma cases associated with HR jobs (stratified by sex and age of asthma-onset). Burden of Work-Related Asthma Men (N — 7.4 M) Women (N — 5.4M) PAR % (AN) PAR % (AN) Adult-Onset Asthma (Caused by Work Exposures?) 8.9 (18,816) 1.0 (2,530) Childhood-Onset Asthma (Exacerbated by Work Exposures?) 2.4 (5,416) 8.6 (17,276) Discussion This study evaluated the utility of existing Canadian surveillance data, collected by Statistics Canada, in providing information on the burden of WRA. Our results showed that a surveillance program such as the CCHS, with data only on current asthma and current job, does not provide the necessary retrospective asthma information (i.e. time of onset) and work information (job held at time of asthma-onset) for estimating WRA, given that the well established association between work and asthma (1, 2, 5-9, 11, 12, 20) was not seen in the analysis of CCHS data. In comparison, with the NPHS survey data, because it was possible to reconstruct the approximate age of asthma-onset, and the job held at that time for most respondents, we were able to estimate the burden of WRA and 19 consider the contribution that the type of WRA and HR exposures may have on these estimates. Our results suggest a large burden of OA among men and a large burden of WEA among women in Canada, and overall, this study suggests that the burden of work "exacerbated" asthma is just as significant as the burden of work "caused" asthma in Canada. In addition, our results provide evidence that workers exposed to multiple HR exposures are at even greater risk for WRA compared to all other workers, given the significant risk differences seen for adult-onset asthma among this exposure group. All findings in this study reinforce the importance of having information on age of asthma-onset and job held at time of asthma-onset provided by Canadian surveillance programs. Canadian Surveillance Data for WRA Research Collection of the necessary surveillance information is critical for population-based research of WRA, as shown by comparisons between CCHS and NPHS in this study. The results from the CCHS were likely biased by the healthy worker effect, partially because it was cross-sectional in design, but mainly due to the lack of necessary surveillance information. The longitudinal nature of the NPHS data did allow us to determine job held at time of asthma-onset using previous survey cycles. When results were compared between surveys, the analyses performed on the CCHS sample did not show the effect of job risk on asthma prevalence as the NPHS analysis did, most likely because we were not able to stratify by age of asthma-onset or link the appropriate jobs to the asthma-specific JEM for adult-onset asthmatics. Regardless of the study design of the survey (i.e. cross- sectional or longitudinal), it is important that the necessary asthma and work information be collected, which includes at least the age of asthma-onset and job held at time of asthma-onset (21) . 20 Burden of WRA among Canada's Adult Population The estimated number of prevalent asthma cases associated with HR jobs from this analysis indicates that in a population of 13 million adults, more than 44,000 asthmatics in any given year in Canada may suffer from asthma that is either "caused" or "exacerbated" by workplace exposures. More specifically, this suggests that nearly 19,000 men suffer from OA and about 17,000 women suffer from WEA. PAR estimates of WRA allow us to compare our findings to those from other studies. For men the estimated PAR's were 8.9% for OA and 2.4% for WEA, and for women, 1.0% and 8.6%, respectively. Estimates of PAR from other studies typically range from 10% to 18 (6, 8-12) for men and women combined. The estimated PAR for OA among men and WEA among women (i.e. the types of WRA that appear to contribute the most to these estimates) fall just below the lower range of PAR's reported in other studies, likely because our estimates are stratified by gender and age of asthma-onset and certain forms of bias introduced by our methodology may have resulted in an underestimation of burden. Our estimates of WRA are very informative because they highlight a clear gender difference related to the type of WRA contributing to the burden. This gender difference in WRA is not completely unexpected, as this finding may reflect the gender dynamic of the healthy worker effect as discussed in a recent paper by Le Moual N, et al (2007) stating that "a stronger healthy worker hire effect for men and a stronger healthy worker survivor effect among women has been reported" (21) . With a strong healthy worker hire effect among men, we would expect that men with childhood-onset asthma would be less likely to be hired into HR jobs. And with a strong healthy worker survivor effect among women, we would expect many adult-onset asthmatics to leave HR jobs. However, why there are so many women with pre-existing asthma working in HR jobs is unclear; other unexplored factors likely play a role in this finding. The type of HR exposure also contributed to estimates of WRA. Results suggest an increased risk for adult-onset asthma among workers exposed to more than one agent from different HR exposure categories (i.e. combined HR exposure) compared to workers exposed to one HR exposure category, and especially compared to unexposed workers. Results also showed that exposure to the LMW agents was a significant risk factor for 21 adult-onset and childhood-onset asthma, however, exposure to HMW agents was a significant risk factor only for adult-onset asthma. One possibility for why we observed an effect of HMW agents for adult-onset asthma but not for childhood-onset asthma may be because more is known about HMW agents (5, 12) and the occupations in which exposure to these agents can occur are better known, compared to LMW agents. As a result, it's likely that individuals with childhood-onset asthma, who are likely more conscious of their exposures, would be able to avoid occupations with exposure to HMW agents. Overall, our estimates are based on valid, robust exposure assessment methods; we identified appropriate jobs given a workers' age of asthma-onset and we applied an asthma-specific JEM to determine job risk. In comparison, other studies have used the 'longest job held' rather than job held at time of asthma-onset (3) and self-reported exposures (22) for methods of exposure assessment. All survey studies are affected by recall bias and the healthy worker effect to some extent, but minimizing these biases will provide more valid estimates. Limitations There are some specific limitations in this study. The healthy worker effect is one form of bias that may have affected findings from CCHS and NPHS analyses. The cross-sectional CCHS data provided insufficient surveillance information to determine age of asthma- onset and job held at time of asthma-onset, likely introducing a healthy worker effect bias from inadequate temporal exposure information relevant to disease onset. In addition, study sample criteria for both the NPHS and CCHS included only currently employed survey respondents, which may have introduced bias from both the healthy worker hire effect and the healthy worker survivor effect (21) . Findings from the CCHS and NPHS may also be affected by non-differential misclassification of exposure. The SOC-1991 job codes, used to link to the asthma-specific JEM to determine exposure risk groups, were pre-coded by Statistics Canada using self-reported text responses to "job title" and "job tasks". We were denied access to the self-reported text information, preventing us from using the 'verification' of exposure estimates step, as recommended by the asthma- specific JEM method (18) . Second, although there are many benefits to using an asthma- specific JEM for estimating the burden of WRA, it judges jobs as being HR or LR based on exposure to agents known to "cause" asthma, not "exacerbate" asthma. Therefore, our 22 WEA estimates, determined using childhood-onset asthma in combination with job risk as a proxy, may also be affected by non-differential misclassification of exposure. Both biases, the healthy worker effect and non-differential classification of exposure, generally push results toward the null (21, 23) , implying that our estimates of burden are likely underestimated (esp. WEA estimates). Conclusions and Future Direction In order for Canadian surveillance data to provide useful information on the burden of WRA, future surveillance programs must at least collect information on age of asthma- onset and job held at time of asthma-onset. This study has shown that the burden of work "exacerbated" asthma is just as significant as work "caused" asthma, and that as many as 44,000 asthmatics in any given year in Canada suffer from asthma that is in some way attributed to their working exposures. Sufficient survey data is needed to fully understand and monitor WRA in Canada, and the large estimates of burden exhibit a need for intervention and improved awareness among the leading stakeholders (workplace policy- makers, physicians, public health officials, researchers, etc.). Future studies estimating the burden of WRA must consider the type of WRA and the type of exposures that contribute to the burden among the population under study, because these factors have important implications for WRA awareness, management and prevention. 23 Bibliography 1. Petsonk EL. Work-Related Asthma and Implications for the General Public. Environ Health Perspect. 2002 Aug;110 Suppl 4:569-72. 2. Breton CV, Zhang Z, Hunt PR, Pechter E, Davis L. Characteristics of Work Related Asthma: Results from a Population Based Survey. Occup Environ Med. 2006 Jun;63(6):411-5. 3. Arif AA, Whitehead LW, Delclos GL, Tortolero SR, Lee ES. Prevalence and Risk Factors of Work Related Asthma by Industry among United States Workers: Data from the Third National Health and Nutrition Examination Survey (1988-94). Occup Environ Med. 2002 Aug;59(8):505-11. 4. Goe SK, Henneberger PK, Reilly MJ, Rosenman KD, Schill DP, Valiante D, et al. A Descriptive Study of Work Aggravated Asthma. Occup Environ Med. 2004 Jun;61(6):512-7. 5. Chan-Yeung M, Malo JL. Aetiological Agents in Occupational Asthma. Eur Respir J. 1994 Feb;7(2):346-71. 6. Blanc PD, Toren K. How Much Adult Asthma can be Attributed to Occupational Factors? Am J Med. 1999 Dec;107(6):580-7. 7. Liss GM, Tarlo SM. Work Related Asthma. Occup Environ Med. 2002 Aug;59(8):503- 4. 24 8. Tarlo SM, Liss GM. Prevention of Occupational Asthma - Practical Implications for Occupational Physicians. Occupational Medicine. 2005;55(8):588-94. 9. Balmes J, Becklake M, Blanc P, Henneberger P, Kreiss K, Mapp C, et al. American Thoracic Society Statement: Occupational Contribution to the Burden of Airway Disease. Am J Respir Crit Care Med. 2003 Mar 1;167(5):787-97. 10. Johnson AR, Dimich-Ward HD, Manfreda J, Becklake MR, Ernst P, Sears MR, et al. Occupational Asthma in Adults in Six Canadian Communities. Am J Respir Crit Care Med. 2000 Dec;162(6):2058-62. 11. Youakim S. Work-Related Asthma. Am Fam Physician. 2001 Dec 1;64(11):1839-48. 12. Malo JL. Future Advances in Work-Related Asthma and the Impact on Occupational Health. Occupational Medicine. 2005;55(8):606-11. 13. Statistics Canada. National Population Health Survey Household Component Cycle 5 (2002-2003), Longitudinal Documentation. In press November 24, 2004. 14. Statistics Canada. Canadian Community Health Survey 2003: User Guide for Public use Microdata File. In press January 2005. 15. Statistics Canada. National Population Health Survey, Household Component Cycle 5(2002-2003) Questionnaire. In press November 2004. 16. Statistics Canada. Canadian Community Health Survey (CCHS) Questionnaire for Cycle 2.1. In press 2005. 17. Standard Occupational Classification (SOC) 1991 July 3, 2006. Available from: http://www.statcan.ca/english/Subjects/Standard/soc/1991/soc91-index.htm. 25 18. Kennedy SM, Le Moual N, Choudat D, Kauffmann F. Development of an Asthma Specific Job Exposure Matrix and its Application in the Epidemiological Study of Genetics and Environment in Asthma (EGEA). Occup Environ Med. 2000 Sep;57(9):635- 41. 19. Last JM, editor. Dictionary of Epidemiology. fourth ed. New York, New York: Oxford University Press, Inc.; 2001. 20. Chan-Yeung M, Malo JL, Tarlo SM, Bernstein L, Gautrin D, Mapp C, et al. Proceedings of the First Jack Pepys Occupational Asthma Symposium. Am J Respir Crit Care Med. 2003 Feb 1;167(3):450-71. 21. Le Moual N, Kauffmann F, Eisen EA, Kennedy SM. The Healthy Worker Effect in Asthma: Work may Cause Asthma, but Asthma may also Influence Work. Am J Respir Crit Care Med. 2007 Sep 13. 22. Caldeira RD, Bettiol H, Barbieri MA, Terra-Filho J, Garcia CA, Vianna EO. Prevalence and Risk Factors for Work Related Asthma in Young Adults. Occup Environ Med. 2006 Oct;63(10):694-9. 23. Rothman KJ. Epidemiology: An Introduction. United States of America: Oxford University Press; 2002. 26 Chapter 3. Asthma in Relation to Job Risk Group: Adjusting for Other Potential Risk Factors2 Introduction A large burden of asthma in relation to job risk was presented in Chapter 2, with age of asthma-onset and gender playing a major role in the overall findings. The main goal of this chapter was to perform adjusted analyses to evaluate the effect of job risk group on asthma, after considering other potential risk factors for asthma. Adjusted analyses were performed separately by age of asthma-onset and gender, using data from the National Population Health Survey (NPHS), Longitudinal Household Component (1994/95- 2002/03). Adjusted analyses were not performed using the Canadian Community Health Survey (CCHS) Cycle 2.1 (2002/03), because as shown in Chapter 2, sufficient asthma and work information was not provided in the CCHS to approximate age of asthma-onset or job held at time of asthma-onset. Several risk factors that have been identified in the literature as being potentially associated with asthma were considered: age, socio-economic status (SES), race, body mass index, and smoking status. Studies have shown that adults with a lower SES have an increased risk for asthma and a lower health status overall (1 ' 2) . Two indicators were used to adjust for SES in the multivariable models, highest attained education level and household income, as has been done in previous analyses of the Canadian survey populations (2) . Postmenopausal hormones may be associated with new-onset asthma (3 ' 4) and increased asthma severity (5) among adult women, so the risk of asthma between women of approximate pre- and post-menopausal ages was compared. Specific race/ethnic groups have been identified as having higher rates of asthma (e.g. Blacks, Natives, and South Asians) compared to others in diverse population studies (6 ' 7) . However, 2 A version of this chapter will be submitted for publication. Garzia NA, Koehoorn M, Demers PA, Kennedy SM. Asthma in Relation to Job Risk group: adjusting for other potential risk factors. 27 distinguishing the role of race versus SES can be difficult when investigating risk factors for asthma. A concern for increased body mass index on the effects of chronic disease have evolved with the high rates of obesity in developed nations, suggesting an increased asthma incidence among overweight and obese men and women (8-10) . And lastly, smoking status was considered as a potential risk factor since previous studies have found that it may be "an important determinant of some types of occupational asthma but not others" (11) . Methods Population & Data The dataset used for the current study was the NPHS, Longitudinal Household Component, using the same sample criteria as described in Chapter 2 (respondents between 15-65 years of age in 2002/03, who provided a full response to survey cycles 1 (1994/95) and 5 (2002/03), and who reported working full-time during the 2002/03 year). The sample population identified with these criteria represented approximately 12,765,888 men and women in Canada. Approximately 5% of the sample population was excluded from the analyses because information on their occupation or age of asthma-onset was not provided. For analyses of adult-onset asthma, respondents with childhood-onset asthma were excluded. Data Analysis Data analyses were carried out using SAS, version 9.1 (SAS Institute Inc., Cary, NC, USA). Bivariate Analyses Bivariate relationships between categorical and continuous variables were explored by calculating measures of central tendency and spread of a continuous variable for each category (mean, median, standard deviation), and by evaluating box-plots between 28 categories. Bivariate relationships were assessed between all categorical variables, using cross-tabulations and chi-square tests of association to test the null hypothesis that there is no association between variables (rejected when p-value<0.05 for chi-square statistic). Associations between predictors were considered to be strong if statistically significant and subjective evaluation of cross-tabulations also suggested a relationship between the variables. These bivariate results were used for predictor selection in the multivariable model. Multivariable Analyses Adjusted risk estimates were derived from four multivariable logistic regression models with adult-onset asthma among men, adult-onset asthma among women, childhood-onset asthma among men, and childhood-onset asthma among women as the outcomes. The prevalence odds ratio (POR) was used as the measure of association for the four multivariable models. These adjusted POR estimates were used for comparison to crude POR estimates (Appendix A) to address the purpose of the study, which was to evaluate the effect of job risk group on asthma after adjusting for other potential risk factors. Adult-onset asthmatics and childhood-onset asthmatics were identified within the sample, using the same methodology explained in Chapter 2. The primary predictor of interest was job risk group, where jobs were judged as being high risk (HR) or low risk (LR) for occupational asthma by an asthma-specific job exposure matrix as described in Chapter 2 (12) . The other potential predictor variables, based on self-reported information from survey cycle 5 (2002/03), included: age, education level, household income, race, body mass index, and smoking status. Age was considered as a continuous variable and as a categorical variable; grouped by age decades for men and women, and grouped into approximate pre- and post-menopausal ages (15-49 vs. 50-65) for women. Socio- economic status was controlled for using highest attained education level as a personal indicator and household income as a household indicator. Education was adjusted for age by breaking down self-reported years of education into quintiles, and applying a previously "standardized version number of years of education" (2) to each quintile; this process was performed for each age group, by 5-year intervals (e.g. 15-19 years old, 20-24 years old, etc.). Household income was grouped and adjusted for number of household 29 members by Statistics Canada. Race could only be used as a two-level variable, white versus all other races, due to small samples sizes of the other race categories. Body mass index information was not available for the sample 15-17 years of age, because the methods for determining body mass index for individuals less than 18 years of age are based on growth curves. Instead, self-reported height and weight was used to determine body mass index classification according to the United States Centers for Disease Control and Prevention growth charts for boys and girls (13) . Self-reported responses about smoking were grouped into smoking status categories by Statistics Canada. Refer to Appendix B for further information on the categories of predictors and their referent groups. Statistical Tests To evaluate the significance of a fitted model, the null hypothesis that all variable coefficients in the model are equal to zero was tested using the Likelihood Ratio Test, rejected at p<0.05 level of significance (14) . The Wald test was used to test significance of the individual predictors in the multivariable model; the null hypothesis (coefficient is equal to zero) was rejected at p<0.05 level of significance (14) . Predictor Selection Potential confounding and collinearity was assessed by examining the bivariate associations among all potential predictors and between each potential predictor and the 4 outcomes. Predictors used in the final multivariable model were included based on the results of these bivariate analyses. A full model, including all predictors that were statistically associated with the outcome at the bivariate level for each of the 4 outcomes, was the first model fitted. With the exception of the primary predictor (job risk group), all other predictors that were not associated with the outcome in the multivariable model were excluded from the final model presented. Predictors strongly associated with the primary predictor of job risk group and with the outcome, based on bivariate analyses (e.g. potential confounders), were removed from the model to evaluate their effect on the coefficient size of the primary predictor. Instances where inclusion of a confounder or collinearly-related predictor in the model changed the size of the primary predictor coefficient (i.e. changing the effect of the primary predictors' association with the outcome), the other predictor was excluded from the final model. However, if removal of 30 a potential confounder or collinearly-related predictor from the model caused only minor changes in the size of the primary predictor coefficient, and was itself significantly associated with the outcome, the predictor was retained in the final model. This same method was used to evaluate strong associations between predictors (not involving the primary predictor) for inclusion in the final model. Estimate Interpretation From the final multivariable logistic regression model, the coefficient estimates of categorical predictors were interpreted as the change in odds of the likelihood of the outcome for categories of the predictor, when all other variables were held constant (14) . For continuous predictor variables, the coefficient estimates were interpreted as the change in the odds of the likelihood of the outcome for a 1-unit increase in the predictor variable, when all other variables were held constant (14) . Given the purpose of this study, I was most interested in the interpretation of the estimated coefficient associated with the primary predictor, job risk group. Weighted Estimates Sampling weights were created and calculated by Statistics Canada for each individual selected for the survey. A sample weight is the product of two components, a design weight and a non-response adjustment factor (15) . The design weight is a respondents' probability of inclusion in the survey sample, dependent on survey design and adjusted for certain characteristics (sex, age, and province) (15) ; it corresponds to the number of individuals in Canada's general population that each respondent is intended to represent for the point in time the sample was selected (i.e. 1994/95) (15) . The non-response adjustment factors controls for non-response from sampled respondents in each survey cycle. All analyses described in this chapter have been weighted using these sample weights. However, to avoid inflated standard errors (narrow confidence intervals) but retain representativeness, a standardized weight for each of the four models was created, using the ratio of total un-weighted subpopulation size to total weighted subpopulation size. All estimates in the results section, except for specified un-weighted sample sizes (n), are based on the standardized weight. 31 Results Population Demographics by Gender Table 3. 1 National Population Health Survey (NPHS) sample by demographic characteristics for men and women Demographic Variable by Category Men (%) Un-weighted n= 3045 (Weighted N= 7,412,489) Women (%) Un-weighted n= 2662 (Weighted N= 5,353,399) Total Weighted (N) Job Risk High risk jobs 13.3 14.7 1,773,163 Group Low risk jobs 86.7 85.3 11,000,000 Age 15-24 years 12.4 12.6 1,591,991 (decades) 25-34 years 19.6 21.8 2,624,960 35-44 years 29.0 32.1 3,865,542 45-54 years 26.7 24.9 3,311,451 55-65 years 12.4 8.5 1,371,944 Age 15-49 years 4n/a 80.3 4,297,040 (pre- v post- menopausal) 50-65 years 4n/a 19.7 1,056,359 Years of Lowest quintile, <20% range 8.8 5.5 945,589 Attained 2"d lowest quintile,20-40% range 18.7 18.4 2,367,895 Education 3rd qu intile, 40-60% range 20.4 18.2 2,488,032 4 th quintile, 60-80% range 21.0 22.2 2,740,571 Highest quintile, >80% range 31.1 35.8 4,214,983 Missing information <I% <I% 32 Demographic Variable by Category Men (%) Un-weighted n= 3045 (Weighted N= 7,412,489) Women (%) Un-weighted n= 2662 (Weighted N= 5,353,399) Total Weighted (N) Household Lowest household income group 3.1 4.4 435,089 Income Middle household income group 13.5 14.9 1,690,913 Highest household income group 83.5 80.7 9,900,712 Missing information <6% <6%. Race White 88.5 87.2 11,200,000 Black 2.3 2.7 317,406 East Asian (Korean, Filipino, Japanese, Chinese) 3.6 4.0 484,973 Aboriginal to North America 0.6 0.8 86,013 South and South East Asian 3.1 3.4 412,256 Other/Mixed (Arab/West Asian, Latin American, Multi-race) 1.8 1.9 234,775 Missing Information <0.5% <0.5% Body Mass Normal or underweight 35.1 56.6 5,508,693 Index Overweight 45.7 27.5 4,789,516 Obese 19.2 15.9 2,240,639 Missing information 0.5% 3.5% Smoking Non-smoker 29.6 35.4 4,072,780 Status Former smoker 41.7 35.8 4,993,282 Current smoker 28.7 28.8 3,652,518 Missing information <1% <1% 4 Variable created for analyses with women only. 33 The demographic distribution of the NPHS sample, shown separately by gender, is available in Table 3.1. Most men and women in this sample were middle-aged (35 - 44 years old), with the highest level of education attained (top 20% range for all respondents, adjusted for age), and the highest annual household income level (adjusted for number of household members). 34 Bivariate Analyses Table 3. 2 Estimates of asthma outcome prevalence (%) by potential predictor variables Predictor Predictor Category Adult-onset Asthma Childhood-onset Asthma Men Women Men Women Prevalence (%) Chi- square p-value Prevalence (%) Chi- square p-value Prevalence (%) Chi-square p-value Prevalence (%) Chi- square p-value Job Risk Group Low risk 2.68% 0.03 5.05% 0.8 2.95% 0.6 3.42% 0.03 3.50% 5.61%High risk 4.66% 5.39% Age (decades) 15-24 years 2.45% collapsed 0.2 1.85% 0.005 10.00% <0.0001 13.34% <0.0001 25-34 years 4.28% 3.26% 4.85% 3.13% 2.28%35-44 years 2.54% 5.58% 0.62% collapsed 0.80% collapsed 45-54 years 3.98% 5.27% 55-65 years 2.81% 9.03% Age (pre/post menopausal) 15-49 years 4n/a 4.85% 0.3 4n/a 'not included 50-65 years 6.07% Years of Attained Education Low (<20% range) 3.55% 0.3 'not included 3.78% 0.7 7.14% 0.004 2.64% 2.91%Low-mid (20-40% range) 1.94% 3.55%Middle (40-60% range) 2.22% 6.07% 2.44% 3.10%Mid-high (60-80% range) 3.49% 3.11% 2.87%High (>80% range) 3.47% Predictor Predictor Category Adult-onset Asthma Childhood-onset Asthma Men Women Men Women Prevalence (%) Chi- square p-value Prevalence (%) Chi-square p-value Prevalence (%) Chi-square p-value Prevalence (%) Chi-square p-value Household Income Low not included 3.29% 0.5 12.90% <0.0001 8.00% <0.00012.34% 831%4.27%Middle 2.49% 2.53%High 5.41% Race White 'not included 'not included 2.55% <0.0001 'not included All Other Races 6.75% Body Mass Index Normal or Underweight 3.05% 0.4 4.28% 0.02 4.37% 0.002 4.26% 0.05Overweight 2.58% 5.50% 1.88% 2.15% Obese 3.68% 7.72% 3.39% 3.53% Smoking Status Non-smoker 3.84% 0.2 4.86% 0.4 3.65% 0.2 3.25% 0.002Former smoker 2.65% 5.91% 2.37% 2.60% Current smoker 2.49% 4.40% 3.39% 5.79% 4 Variable created for analyses with women only. ♦ Variable not included in analyses because sample size of one or more categories was too small when stratified by the outcome for this subsample. The bivariate results in which age was considered as a continuous variable against the outcome of interest and other potential predictors were not reported because age as a categorical variable was more informative for all analyses. Associations between the Asthma Outcome and Predictor Variables Adult-Onset Asthma Among men, only the primary predictor (job risk group) was statistically associated with adult-onset asthma (p=0.03). Evaluation of the cross-tabulation, suggested that adult-onset asthma was nearly twice as prevalent among men working in HR jobs compared to men working in LR jobs (Table 3.2). Household income and race were not considered in bivariate analyses, because sample sizes were too small in one or more categories of these variables when stratified by the outcome and it was not possible to collapse their categories further. Men aged 15-34 were collapsed into one age group for all bivariate and multivariable analyses, also due to small sample sizes when stratified by the outcome. Among women, two predictors, age (p=0.005) and body mass index (p=0.02), were statistically associated with adult-onset asthma (Table 3.2). The prevalence estimates indicate an increase in the outcome with increasing age, and with increasing body mass index. The primary predictor was not statistically associated with adult-onset among women, although prevalence was higher among women working in HR jobs. Education and race could not be considered in analyses due to small sample sizes after stratification by the outcome variable. Childhood-Onset Asthma Among men, childhood-onset asthma was not statistically associated with the primary predictor. Several other predictors were statistically associated with the childhood-onset asthma among men (Table 3.2), including: age (p<0.0001), household income (p<0.0001), race (p<0.0001) and body mass index (p=0.002). Cross-tabulations suggest that the prevalence of childhood-onset asthma was higher among the men in the youngest age group (15-24 years old) and the lowest household income group; prevalence was also higher among "all other" races (non-white race) and among men classified as "normal or 37 underweight" and "obese" compared to "overweight". Men aged 45-65 years were collapsed into one age group due to small sample sizes when stratified by the outcome. Among women, the primary predictor, job risk group, was statistically associated with the childhood-onset asthma (p=0.03) with a higher prevalence among women working in HR jobs compared to LR jobs. And as shown in Table 3.2, the outcome was statistically associated with age (p<0.0001), education (p=0.004), household income (p<0.0001), body mass index (p=0.05) and smoking status (p=0.002). The prevalence of childhood-onset asthma was highest among women aged 15-24 years old (13.3%) and those in the lowest education groups (although prevalence varied across other education groups). The prevalence was also highest among women with body mass index classified as "normal or underweight" and among "current" smokers. Women between 45-65 years old were collapsed into one age group due to small sample sizes. Race and the other categorical age variable (for stratifying women into post- and pre-menopausal groups) could not be considered in the analyses due to small sample sizes when stratified by the outcome. Associations between Predictor Variables All bivariate relationships between predictors were explored. Results of these analyses were only reported for a strong association of predictors considered in the full model. Cross-tabulations for all potentially strong associations between predictors can be seen in Appendix C. Adult -Onset Asthma Although there were strong bivariate relationships observed between predictors, none of them were of importance for the multivariate model of adult-onset asthma among men, since only the primary predictor was associated with the outcome variable. Among women, bivariate results suggested strong relationships between the primary predictor of job risk group and two other predictors, age (p<0.0001) and body mass index (p=0.024). Age and body mass index were also strongly associated with each other (p<0.0001). Evaluation of the cross-tabulations of these strong associations indicated that the prevalence of exposed workers (HR job group) was highest among women aged 45-54 38 years and the prevalence of exposed workers was higher among women classified as "overweight" and "obese"; and women in the older age categories were much more likely to have a body mass index classification of "overweight" or "obese". Childhood-Onset Asthma Among men, a strong association between the primary predictor and household income (p=0.0006) was observed. There were several potentially strong associations between the other predictors, where household income was associated with age (p<0.0001), body mass index (p<0.0001), and race (p<0.0001); body mass index was associated with age (p<0.0001) and race (p=0.03). Evaluation of the cross-tabulations indicated the following relationships were very strong: the prevalence of exposed workers was lower for men with the highest household income; the likelihood of being in the highest household income category increased with increasing age categories and was more likely among the "white" race; and the likelihood of being "overweight" or "obese" was much greater for men in the older age categories. Among women, there were several strong associations between the primary predictor and other predictors, including: age (p=0.004), education (p<0.0001), household income (p=0.013) and body mass index (p=0.016). Cross-tabulations suggested that women in the older age categories (35-65 years old) were more likely to be working in HR jobs; women in the higher education levels were much less likely to be working in HR jobs (34% prevalence of exposure in lowest education group versus 9.5% prevalence in the highest education group); and women with the highest household income and a body mass index classification of "normal or underweight" had the lowest prevalence of exposed workers. Strong associations that were observed between the other predictors in bivariate analyses include: household income and age (p<0.0001), household income and education (p<0.0001), household income and smoking status (p<0.0001), age and body mass index (p<0.0001), age and smoking status (p<0.0001), and education and smoking status (p<0.0001). The cross-tabulations suggested that women in the older age categories were more likely to be in the highest household income category; women in the higher education categories were more likely to be in the highest household income category; women in the highest household income category were more likely to be a "former smoker" rather than a "current smoker"; women in the older age categories were more 39 likely to "overweight" or "obese"; women in the older age categories were more likely to be "former smokers" or "non-smokers"; and women in the high education categories were more likely to be "former smokers" or "non-smokers". A few other bivariate relationships were statistically associated, however, based on evaluation of cross-tabulation results, these relationships were judged to not be very strong overall (Appendix C). Multivariable Analyses Table 3. 3 Final multivariable logistic regression results: POR estimates for associations between job risk group and asthma, adjusted for other predictors and stratified by gender Adult-Onset Asthma Childhood-Onset Asthma MEN WOMEN MEN WOMEN POR [95% CI] POR [95% CI] POR [95% CI] POR [95% CI] Weighted N= Weighted N= Weighted N= Weighted N= Predictor 7,188,137 4,981,196 7,396,793 5,353,399 Predictor Category Job Risk Low risk Group High risk 1.78 1.07 1.27 1.99 [1.05, 3.01] [0.65, 1.76] [0.70, 2.29] [1.20, 3.29] Age 15-24 years 0.36 3.57 6.78 (decades) [0.15, 0.89] [2.14, 5.95] [3.91, 11.75] 25-34 years 0.86 1.09 2.28 [0.51, 1.44] [0.60, 1.97] [1.27, 4.11] 35-44 years ns 45-54 years 0.94 [0.60, 1.49] 0.20 0.34 55-65 years 1.66 [0.09, 0.44] [0.14, 0.81] [0.96, 2.88] 40 Predictor Predictor Category Adult-Onset Asthma Childhood-Onset Asthma MEN POR [95% CI] Weighted N= 7,188,137 WOMEN POR [95% CI] Weighted N= 4,981,196 MEN POR [95% CI] Weighted N= 7,396,793 WOMEN POR [95% CI] Weighted N= 5,353,399 Race White • not included •not included not included All Other Races 2.85 [1.74, 4.68] Body Mass Index Normal or Underweight Ans *ns "ns Overweight 1.15 [0.75, 1.76] Obese 1.67 [1.06, 2.64] Model Likelihood Ratio Test, x2 Statistic (p-value) 4.12 (p=0.04) 19.05 (p=0.008) 89.85 (p<0.0001) 97.81 (p<0.0001) ♦ Variable not included in analyses because sample size of one or more categories was too small when stratified by the outcome for this subsample. 4. Variable not considered significant for inclusion in final model based on analyses. Adult-Onset Asthma For men, the final model only included the primary predictor, job risk group, because none of the other potential risk factors considered were associated with adult-onset asthma (Table 3.3). For women, the final model included job risk group, age and body mass index as predictors (Table 3.3). Although both age and body mass index were strongly associated 41 with job risk group, models excluding age and body mass index, individually and simultaneously, were performed to evaluate changes in the coefficient of job risk group. Based on evaluation of the models, only slight changes in the coefficient size of the primary predictor were observed when each of these other predictors was removed from the model. There were also no significant changes in the coefficient sizes of age when body mass index was removed, and vice versa. It was concluded that age and body mass index did not confound the overall relationship between job risk group and the outcome, and more was explained about adult-onset asthma in women by leaving age and body mass index in the final model. Childhood -Onset Asthma The final model for men included job risk group, as well as age and race (Table 3.3). Household income was not included in the final model, because household income and the primary predictor (job risk group) may be collinearly associated with childhood-onset asthma. When household income was removed from the model, the coefficient of job risk group changed; although household income may be a predictor for childhood-onset asthma, there was concern that including it in the final model would effect the contribution of the primary predictor on the outcome. A model including household income as a predictor is provided in Appendix D. Body mass index was also not included in the final model, because after removing household income, with which body mass index was associated, it was no longer significantly associated with the outcome. The final model for women included job risk group and age (Table 3.3). Body mass index, education and smoking status were excluded from the final model because they were not significantly associated with the outcome in the multivariable models. Both age and household income were strongly associated with the primary predictor, however, they had very different effects on job risk group in the model. When household income was included in the model, the coefficient for job risk group was changed and not significant in the model, compared to when household income was removed. For the same reasons acknowledged above for men, household income was excluded from the final model for women, on the basis of potential collinearity. A model that includes household income as a predictor for childhood-onset asthma among women is supplied in Appendix D. 42 Removing age from the model only caused a slight change in the coefficient size of the primary predictor; therefore, it was decided to keep it in the model because it helped to explain more about the outcome and it was still significant in the multivariable model. Discussion Among men but not women, job risk group was significantly associated with adult-onset asthma. The adjusted POR estimates of adult-onset asthma associated with job risk group among men and women were 1.78 and 1.07, respectively. In other words, the odds of adult-onset asthma was 78% greater among men working in HR jobs compared to men working in LR jobs; and 7% higher among women working in HR jobs compared to women working in LR jobs. Even after considering other risk factors for adult-onset asthma among men and women, the POR associated with job risk group did not change from the crude estimates (Appendix A). In the final adjusted models, job risk group was the only predictor associated with adult-onset asthma among men, whereas age and body mass index were the only predictors associated with adult-onset asthma among women. Among women but not men, job risk group was significantly associated with childhood- onset asthma. The adjusted POR estimates for childhood-onset asthma associated with job risk group among men and women were 1.27 and 2.0, respectively. The odds of childhood-onset asthma were 27% greater among men working in HR jobs compared to LR jobs, and 100% greater among women working in HR jobs compared to LR jobs. The adjusted estimates were higher compared to the crude estimates for men and women with unadjusted PORs of 1.19 and 1.68 (Appendix A), respectively. Therefore, the effect of job risk group on childhood-onset asthma was greater after accounting for other potential risk factors among men and women. In the final models, age was significantly associated with childhood-onset asthma for both men and women, and race for men only. Unlike the models for adult-onset asthma, other potential risk factors were needed in the models of childhood-onset asthma in order to more accurately estimate the odds of disease 43 associated with job risk group. One possible explanation for this may be related to the asthma-specific job exposure matrix, the tool used to judge job risk groups (12) . The asthma-specific job exposure matrix is intended to judge jobs as being high risk or low risk for occupational asthma, not for work exacerbated asthma. Therefore, it will more accurately judge the job risk for adult-onset asthma than for childhood-onset asthma. In the context of these analyses, this may explain why other risk factors needed to be adjusted for in analyses of childhood-onset asthma for men and women to more accurately estimate the effect of job risk group on childhood-onset asthma, whereas for adult-onset asthma in men and women, considering other risk factors did not effect the POR associated with job risk group. There are some specific limitations to acknowledge. First, due to the restrictions associated with the release of Statistics Canada data results, estimates for some of the risk factors could not be reported. For instance, in the models of childhood-onset asthma, race was included as a predictor in the final model for men, whereas for women, race could not even be considered because of the sample size restriction. However, it's unlikely that this would change any of the final messages from the evaluation of job risk group on adult and childhood asthma. Second, in the analyses of adult-onset asthma, the measures for the non job predictors were based on information reported in 2002/03, instead of being based on information at time of asthma-onset, as was the case for job risk group. This would not have affected the models for childhood-onset asthma because all measures of all predictors were based on the 2002/03 data. The reason for this was due to changes in the way that Statistics Canada asked questions or grouped responses to certain questions (e.g. education) between the survey cycles. So, to be consistent, data from 2002/03 was used to determine the measures for the predictors considered for adjusted effects in all of the models. As a result, it is possible that the measures of some of these predictors were different in 2002/03 compared to time of asthma-onset for some individuals. However, the impact of this limitation on the overall results was likely small, because the time frame of the survey was not very long (1994/95 to 2002/03), and the age of asthma-onset could have occurred during any one of these years. In conclusion, this analysis showed that taking other potential risk factors for asthma into account did not reduce the estimate of the effect of job risk on asthma. In fact, the effect 44 of job risk group on childhood-onset asthma increased after considering other potential risk factors. However, this was not the case for adult-onset asthma, and it was likely a result of the method used for assessing exposures, which was intended to identify the job risk for occupational asthma and not for work-exacerbated asthma. And as suggested by Chapter 2 findings, even after considering other risk factors, occupational exposures were associated with a burden of adult-onset asthma among men and a burden of childhood- onset asthma among women. 45 Bibliography 1. Basagana X, Sunyer J, Kogevinas M, Zock JP, Duran-Tauleria E, Jarvis D, et al. Socioeconomic Status and Asthma Prevalence in Young Adults: The European Community Respiratory Health Survey. Am J Epidemiol. 2004 Jul 15;160(2):178-88. 2. Prus SG. A Life Course Perspective on the Relationship between Socio-Economic Status and Health: Testing the Divergence Hypothesis. Can J Aging. 2004;23 Suppl 1:S145-53. 3. Barr RG, Wentowski CC, Grodstein F, Somers SC, Stampfer MJ, Schwartz J, et al. Prospective Study of Postmenopausal Hormone use and Newly Diagnosed Asthma and Chronic Obstructive Pulmonary Disease. Arch Intern Med. 2004 Feb 23;164(4):379-86. 4. Troisi RJ, Speizer FE, Willett WC, Trichopoulos D, Rosner B. Menopause, Postmenopausal Estrogen Preparations, and the Risk of Adult-Onset Asthma. A Prospective Cohort Study. Am J Respir Crit Care Med. 1995 Oct;152(4 Pt 1):1183-8. 5. Siroux V, Oryszczyn MP, Varraso R, Le Moual N, Bousquet J, Charpin D, et al. Environmental Factors for Asthma Severity and Allergy: Results from the EGEA Study. Rev Mal Respir. 2007 May;24(5):599-608. 6. Netuveli G, Hurwitz B, Sheikh A. Ethnic Variations in Incidence of Asthma Episodes in England & Wales: National Study of 502,482 Patients in Primary Care. Respir Res. 2005 Oct 21;6:120. 46 7. Meng YY, Babey SH, Hastert TA, Brown ER. California's Racial and Ethnic Minorities More Adversely Affected by Asthma. Policy Brief UCLA Cent Health Policy Res. 2007 Feb;(PB2007-3)(PB2007-3):1 -7. 8. Guerra S, Sherrill DL, Bobadilla A, Martinez FD, Barbee RA. The Relation of Body Mass Index to Asthma, Chronic Bronchitis, and Emphysema. Chest. 2002 Oct; 122(4): 1256-63. 9. Beuther DA, Sutherland ER. Overweight, Obesity, and Incident Asthma: A Meta- Analysis of Prospective Epidemiologic Studies. Am J Respir Crit Care Med. 2007 Apr 1 ;175(7):661-6. 10. Shore SA. Obesity and Asthma: Cause for Concern. Curr Opin Pharmacol. 2006 Jun;6(3):230-6. 11. Chan-Yeung M, Malo JL. Aetiological Agents in Occupational Asthma. Eur Respir J. 1994 Feb;7(2):346-71. 12.Kennedy SM, Le Moual N, Choudat D, Kauffmann F. Development of an Asthma Specific Job Exposure Matrix and its Application in the Epidemiological Study of Genetics and Environment in Asthma (EGEA). Occup Environ Med. 2000 Sep;57(9):635- 41. 13. Body Mass Index. 2007 May 22, 2007. Available from: http://www.cdc.gov/nccdphp/dnpa/bmi/. 14. Vittinghoff E, Glidden DV, Shiboski SC, McCulloch CE. Regression Methods in Biostatistics: Linear, Logistic, Survival, and Repeated Measures Models. United States of America: Springer; 2005. 47 15. Statistics Canada. National Population Health Survey Household Component Cycle 5 (2002-2003), Longitudinal Documentation. In press November 24, 2004. 48 Chapter 4. Discussion Three objectives were investigated for this thesis research: to evaluate the utility of existing Canadian surveillance data to provide useful information on the burden of work- related asthma; to identify occupational exposure groups at high risk for work-related asthma; and to evaluate the effect of job risk on asthma after considering other potential risk factors for asthma. This chapter will discuss what was learned from the investigations of these objectives, the main strengths and limitations associated with this research, as well as provide perspective on the needs and significance of future surveillance on work- related asthma. This thesis research contributes several important messages for current and future population-based research on work-related asthma. Summary of Key Messages Surveillance Information From the investigation on the utility of existing Canadian surveillance data for providing useful information on the burden of work-related asthma, it was learned that information on age of asthma-onset and job held at time of asthma-onset is necessary. Two Canadian surveillance programs were evaluated: the Canadian Community Health Survey (CCHS) (1) and the National Population Health Survey (NPHS), Longitudinal Component (2) . Canadian Community Health Survey (CCHS) The CCHS surveillance program was not useful in providing information on the burden of work-related asthma, because it only provided information on "current" asthma status and "current" job. The insufficient asthma and work information did not allow us to appropriately assess exposure relevant to asthma-onset, because it was unknown when the 49 asthma was diagnosed by a health professional (i.e. age of asthma-onset) and what the exposure was at the time of asthma-onset (i.e. job held at time of asthma-onset). Cross- sectional data is often criticized for producing biased estimates of work-related asthma (3) , mainly due to limited information it typically provides for exposure assessment (4, 5). Although the CCHS is cross-sectional in design, additional asthma and work questions that provide not only cross-sectional information, but that are designed to more appropriately determine exposure relative to asthma-onset, would improve the utility of this surveillance program and minimize forms of bias that cross-sectional studies suffer from (e.g. healthy worker effect bias (5)). However, due to the information limitations of the CCHS, it was not useful for this research, and the burden of work-related asthma could not be properly investigated with this surveillance program. National Population Health Survey (NPHS) The information provided in the NPHS data was useful for the purposes of this thesis research. The NPHS surveillance program asked respondents about age at time of health professional-diagnosed asthma, and the longitudinal nature of the NPHS data allowed us to use subsequent survey cycles to determine the job held at time of asthma-onset for most adult-onset asthmatics. Although the NPHS data did provide useful information on the burden of work-related asthma, it would have been beneficial for surveillance research if respondents who reported "yes" to having health professional diagnosed asthma were also asked: "what was your job at the time that your asthma symptoms developed?". If respondents who reported having asthma were asked this type of question, then the exposure information used for analyses would have been more accurate and available for more, if not all, respondents. A longitudinal surveillance program is not necessary for the surveillance of work-related asthma, although it was useful for this research since a specific question about job exposure was not asked, such as the one I suggested above, in the NPHS. Appropriate information on exposure relevant to the time of disease onset is necessary for surveillance of work-related asthma, which includes at least, age of asthma- onset and job held at time of asthma-onset. 50 Burden of Work-related asthma: Occupational Exposure Groups The NPHS data was used to identify occupational exposure groups at high risk for work- related asthma. These exposure groups were identified by considering age of asthma-onset in relation to job risk group (i.e. high risk versus low risk jobs) and type of high risk exposure (e.g. high molecular weight agents); both of these exposures were based on judgment by the asthma-specific job exposure matrix (6) . Based on these investigations of occupational exposure groups at risk for work-related asthma, the following key messages were learned: the burden of work exacerbated asthma is just as significant as the burden of work "caused" asthma; workers are at increased risk of work-related asthma if they work in high risk jobs, and in particular, where there is exposure to multiple high risk agents. Gender Differences of Work-related asthma An interesting gender dynamic was observed when asthma was investigated by job risk group, and analyses were stratified by age of asthma-onset and gender. Among men, the statistical difference between job risk groups was observed for adult-onset asthma, but for women, the statistical difference was observed for childhood-onset asthma. There is no available literature, of which I am aware, that could provide a precise reason for this observation between the genders. However, based on what is known about my research methods, and differences in the types of work that men tend to dominated compared to women, there are some postulations that could be made about this observation. Reflecting on Figure 2.2 (National Population Health Survey (NPHS) study sample: asthma prevalence by age of asthma-onset, job risk group and gender') in Chapter 2, helps to break down this discussion by considering three separate questions: Among men, why is there a significant difference in the prevalence of adult-onset asthma between job risk groups, but not for childhood-onset asthma? Among women, why is there a significant difference in childhood-onset asthma prevalence between job risk groups? And among women, why is there a high prevalence of adult-onset asthma in both job risk groups (high risk and low risk job groups)? First of all, among men there was a statistical difference between job risk groups for adult- onset asthma, but not for childhood-onset asthma. One possible reason is that men with childhood-onset may be more likely to self-select into jobs that are low risk for asthma. 51 Previous research indicates that men are more likely to exhibit the "healthy worker hire effect" (5), the phenomena in which "healthier subjects at lower risk of disease tend to be employed preferentially" (5) . In other words, men with pre-existing asthma are less likely to be employed into the jobs that are high risk for asthma. Men may also be more likely to self-select into lower risk jobs because the public is more aware about asthma that is related to "typical" industrial jobs (e.g. Mining, Oil & Gas, and Fishing) that tend to be dominated by men. Among women, a significant difference in childhood-onset asthma prevalence between job risk groups may be related to a "lack of awareness" of the asthma risk associated with jobs that women are more likely to dominate. Many women do not work in the "typical" industrial jobs among which there is more awareness about asthma risks. Instead, women are more likely to work as health technologists or in personal and domestic service jobs, both of which are judged to be high risk for occupational asthma based on the asthma- specific job exposure matrix (6) . Jobs such as these are not likely to be associated with asthma, making it difficult for women to self select into lower risk jobs. High prevalence of adult-onset asthma among women working in both high risk and low risk jobs may be related to a possible non-differential misclassification of exposure by the asthma-specific job exposure matrix. The job risk as determined by the asthma-specific job exposure matrix is likely to be more accurate for judging asthma risk associated with the "typical" industrial jobs that are dominated by men. This is because the job exposure matrix determines job risks associated with "known risk factors" for occupational asthma (6) , and more is known about asthma in these industrial jobs. Therefore, it's possible that some of the jobs that women report working may be high risk for asthma, rather than low risk, but more research on work-related asthma is needed to identify known risk factors related to these other jobs that are dominated by women. Adjusting for Other Risk Factors of Asthma The effect of job risk group on asthma prevalence was evaluated by considering other potential risk factors using multivariable logistic regression, and fitting four separate models (adult-onset asthma among men, adult-onset asthma among women, childhood- 52 onset asthma among men, and childhood-onset asthma among women). The evaluation was based on comparison between the adjusted prevalence odds ratio (POR) estimates of job risk group for each outcome of interest and crude POR estimates. It was concluded for this investigation that adjustment of other risk factors was necessary for estimating the odds of childhood-onset asthma in relation to job risk group for men and women, but adjustment of risk factors was not necessary for estimating the odds of adult-onset asthma in relation to job risk group for men and women. Adjustment of other risk factors associated with childhood-onset asthma in men and women may be related to the purpose of the asthma-specific job exposure matrix. The asthma-specific job exposure matrix judges job risk based on exposure to "known risk factors" for occupational asthma (i.e. adult-onset asthma) (6) , not for work-exacerbated asthma (i.e. childhood-onset asthma). Strengths and Limitations Limitations Sources of Bias There is reason to suspect the potential for bias by the healthy worker effect (HWE) in this research. Cross-sectional studies are typically susceptible to the HWE (5), as clearly exemplified in this research with the CCHS that included information only on "current" asthma status and "current" exposure. Estimates from both the CCHS and NPHS may be affected by the HWE, because only "actively" employed persons (5 ' 7) were considered in the sample population for these cross-sectional analyses. By only including the "actively" employed, the association between asthma and work may be underestimated (5 ' 7) . The likelihood that our estimates account for the "less healthy" workers is decreased, because these individuals may have already left the work force or changed their jobs (5 ' 7) . Non-differential misclassification of exposure is the second form of bias that may have impacted the results of this research. There are two potential sources for this bias: the first 53 is the pre-coded job titles by Statistics Canada, and the second is the asthma-specific job exposure matrix. Based on self-reported responses to survey questions asked about "job title" and "job tasks" in both the CCHS and NPHS, Statistics Canada coded the information into standardized occupational codes. However, without access to both the text information and the occupational codes, the "verification" step of the asthma-specific job exposure matrix that further evaluates the exposure associated with a job code, could not be applied (6) . This "verification" step has the potential to reduce misclassification and provide more accurate estimates of exposure (6) . The asthma-specific job exposure matrix is the second source that may have contributed to non-differential misclassification of exposure, because it is intended to judge jobs based on exposure to "known risk factors" for occupational asthma (i.e. adult-onset asthma) (6) . As a result, the job risk groups that were associated with childhood-onset asthma may suffer from non-differential misclassification of exposure. Also, the original asthma-specific job exposure matrix used the International Standard Classification of Occupations (ISCO), but Statistics Canada coded occupations using the Standard Occupational Classification system (SOC-1991). Therefore, I spent several months re-coding the asthma-specific job exposure matrix to the SOC-1991 codes (see Appendix E for further detail), and this process may have additionally contributed to the potential of this bias. Non-differential bias typically forces results toward the null hypothesis of "no effect" (8), which may have caused the estimates for the association between asthma and exposure to be underestimated. Statistics Canada Privacy Protected Data There were some specific limitations associated with this thesis research as a result of using the privacy protected Statistics Canada data (i.e. not public-use data). First of all, using the privacy protected data delayed the process of this research, causing me to reduce the research questions I had originally planned to investigate. This is because the data can only be used during the hours of the Statistics Canada Research Data Centre (data portal for the privacy protected data), and all analyses must be performed twice and documented for review before any result can be disclosed for use. The analyses need to be performed twice because only weighted estimates can be released from the Research Data Centre, but all unweighted estimates used to determine the weighted estimates must be documented to show a minimum unweighted sample size (refer to the 'Methods' Section of Chapter 3 for 54 further detail on weighted estimates). The minimum unweighted sample size leads into the second limitation; because of the minimum sample size requirement, analyses of the types of high risk exposures associated with asthma could not be stratified by both age of asthma-onset and gender. Also, some potential risk factors for adjusted analyses performed in Chapter 3, could not be considered in instances where the sample sizes were too small when stratified by another variable. Strengths There are limited number population-based studies on work-related asthma (5 ' 9' 10)  This research is particularly important, not only because it is one of few population-based studies, but also because our findings can improve the current and future surveillance of work-related asthma in Canada. This research has several strengths associated with it that other population-based studies are lacking. First of all, national Statistics Canada surveillance programs were used for this research, and the estimates derived from these surveys are intended to reflect the true estimates across Canada's population (1 ' 2) . These surveys have high response rates, approximately 81% (1,2), and great quality control (e.g. sampling methods and standardized methods for survey interviews). The methods used for assessing asthma in relation to work in this research, stand apart from many other population-based studies. For instance, exposure assessment was based on the asthma-specific job exposure matrix, a robust and standard method that favors "specificity over sensitivity" (6) , rather than exposure based solely expert judgment or self- reported exposures to risk factors (3 ' 1042) . Second, job held at time of asthma-onset was used for adult-onset asthmatics to determine exposure, whereas other studies have used the "longest job held" as the exposure related to asthma (3) ; a method that can be heavily influenced by the HWE bias (5) . A third strength of this is research is that the burden estimates consider the contribution of both forms of work-related asthma (work- exacerbated and occupational asthma) and types of high risk exposures (e.g. high- molecular weight agents versus low-molecular weight agents), two important factors that few studies have attempted to investigate (4) . Other population-based studies have 55 considered work-related asthma in general or have focused only on one form, such as occupational asthma (9, 10, 13)  The significance of considering these two factors in the burden estimates of work-related asthma are reinforced when I recall attention to the key messages about occupational exposure groups: work-exacerbated asthma is just as significant as work "caused" asthma, and workers are at increased risk of work-related asthma if they work in high risk jobs where there is exposure to multiple high risk agents. Future Perspective Needs for Future Surveillance Information Canadian surveillance programs must include information on age of asthma-onset and job held at time of asthma-onset, at least, for the surveillance of work-related asthma to be possible. It would not be difficult for these current, sophisticated Statistics Canada surveillance programs to add a question that such as, "what was your job at the time that your asthma symptoms developed?" of all respondents who report "yes" to the already included question, "do you have asthma?" (2) . It would also be helpful if a question was included to ask respondents "what age did you develop your asthma symptoms?", to allow for a more accurate approximation of age of asthma-onset. Although it may seem common knowledge, the following statement must be reinforced: "a prerequisite for a causal relationship between exposure and asthma is that the exposure has occurred before the start or before the substantial aggravation of the condition" (14) . Studies that do not hold up to the requirement of this statement are prone to bias in their analyses of disease and exposure. Significance of Work-Related Asthma Surveillance The surveillance of work-related asthma is necessary. Asthma is both a personal and social burden in Canada. As discussed in chapter 2, our findings suggest that in a population of 13 million adults, more than 44,000 asthmatic men and women in any given year in Canada may be suffering from asthma that is either "caused" or "exacerbated" by workplace exposures. In addition, occupational exposure in relation to work-exacerbated 56 asthma is needed, because it contributes to the burden of work-related asthma as much as occupational asthma. Work-related asthma can affect "work productivity" (15) and financial/socio-economic stability (16) . Health care and economic costs are significantly impacted by asthma severity and uncontrolled asthma (17 ' 18) . This impact is very important for the efficiency of Canada's public health care system. Improved surveillance can reduce the personal and social burden of work-related asthma with identification of at-risk groups and appropriate intervention that are specific for different types of high risk exposures and different types of work-related asthma (19) . Recommendations: ideal surveillance questions for work-related asthma Health Canada and Statistics Canada have collaborated in the past on the development and conduction of the 1996 Asthma Supplement (20) . However, this supplement was only used once, and the questions concerning risk factors did not ask specifically about the effect of work on asthma. The Asthma Supplement did ask more in depth questions about age of onset of asthma symptoms. Although I already mentioned the most necessary questions to be added to Canadian surveillance programs in above sections, I have provided an outline of ideal questions for the surveillance of work-related asthma. These questions, following, consider age of asthma-onset and the questions to appropriately identify the job exposure of concern for occupational asthma and work-exacerbated asthma. Recommended Asthma and Work Questions: 1. "Do you have health professional-diagnosed asthma?" (2) If "yes", then ask question 3. If "no", then ask question 2. 2. Have you ever had asthma? If "yes" then ask question 3. If "no" then no more questions about asthma and work. 57 3. "How old were you when a doctor first diagnosed your asthma?" (2) Move to question 4 for all responses. 4. "How old were you when you had your first asthma symptoms?" (20) If age response is before 15 years of age, then ask question 5. If age response is at or after 15 years of age, then ask question 13. If age of first asthma symptoms is younger than 15 years (reported in question 3) then move to question 5; if age of first asthma symptoms is at or after age 15 years, then move to question 13. 5. Did your asthma ever disappear? If "yes" then ask question 6. If "no" then ask question 9. 6. Has your asthma come back? If "yes" then ask question 7. If "no" then ask question 9. 7. Were you working when your asthma came back? If "yes" then ask question 8. If "no" then ask question 9. 58 8. What was your job at the time that your asthma came back? What were your job tasks? After response, ask question 9. 9. Are you working now? If "yes" then ask question 10. If "no" then ask question 11. 10. What is your current job and what are your current job tasks? After response ask question 12. 11. Is the reason that you don't work, related to your asthma? After response, no more questions. 12. "Is your asthma ever worse at work?" No more questions. If older than 15 years of age at time of first asthma symptoms, then ask question 13: 13. Were you working when you developed your asthma symptoms? If "yes" then ask question 14. If "no" then ask question 15. 59 14. What was your job at the time that your asthma symptoms started? What were your job tasks? After response ask question 15 15.Are you working now? If "yes" then ask question 16. If "no" then ask question 17. 16. What is your current job and current job tasks? After response, ask question 17. 17. "Is your asthma ever worse at work?" [END] 60 Bibliography 1. Statistics Canada. Canadian Community Health Survey 2003: User Guide for Public use Microdata File. In press January 2005. 2. Statistics Canada. National Population Health Survey Household Component Cycle 5 (2002-2003), Longitudinal Documentation. In press November 24, 2004. 3. Arif AA, Whitehead LW, Delclos GL, Tortolero SR, Lee ES. Prevalence and Risk Factors of Work Related Asthma by Industry among United States Workers: Data from the Third National Health and Nutrition Examination Survey (1988-94). Occup Environ Med. 2002 Aug;59(8):505-11. 4. Le Moual N, Kennedy SM, Kauffmann F. Occupational Exposures and Asthma in 14,000 Adults from the General Population. Am J Epidemiol. 2004 Dec 1;160(11):1108- 16. 5. Le Moual N, Kauffmann F, Eisen EA, Kennedy SM. The Healthy Worker Effect in Asthma: Work may Cause Asthma, but Asthma may also Influence Work. Am J Respir Crit Care Med. 2007 Sep 13. 6. Kennedy SM, Le Moual N, Choudat D, Kauffmann F. Development of an Asthma Specific Job Exposure Matrix and its Application in the Epidemiological Study of Genetics and Environment in Asthma (EGEA). Occup Environ Med. 2000 Sep;57(9):635- 41. 61 7. Checkoway H, Pearce N, Kriebel D. Selecting Appropriate Study Designs to Address Specific Research Questions in Occupational Epidemiology. Occup Environ Med. 2007 Sep;64(9):633-8. 8. Rothman KJ. Epidemiology: An Introduction. United States of America: Oxford University Press; 2002. 9. Malo JL. Future Advances in Work-Related Asthma and the Impact on Occupational Health. Occupational Medicine. 2005;55(8):606-11. 10. Toren K, Balder B, Brisman J, Lindholm N, Lowhagen 0, Palmqvist M, et al. The Risk of Asthma in Relation to Occupational Exposures: A Case-Control Study from a Swedish City. Eur Respir J. 1999 Mar;13(3):496-501. 11. Caldeira RD, Bettiol H, Barbieri MA, Terra-Filho J, Garcia CA, Vianna EO. Prevalence and Risk Factors for Work Related Asthma in Young Adults. Occup Environ Med. 2006 Oct;63(10):694-9. 12. Saarinen K, Karjalainen A, Martikainen R, Uitti J, Tammilehto L, Klaukka T, et al. Prevalence of Work-Aggravated Symptoms in Clinically Established Asthma. Eur Respir J. 2003 Aug;22(2):305-9. 13. Johnson AR, Dimich-Ward HD, Manfreda J, Becklake MR, Ernst P, Sears MR, et al. Occupational Asthma in Adults in Six Canadian Communities. Am J Respir Crit Care Med. 2000 Dec;162(6):2058-62. 14. Toren K, Brisman J, Olin AC, Blanc PD. Asthma on the Job: Work-Related Factors in New-Onset Asthma and in Exacerbations of Pre-Existing Asthma. Respir Med. 2000 Jun;94(6):529-35. 62 15. Blanc PD, Trupin L, Eisner M, Earnest G, Katz PP, Israel L, et al. The Work Impact of Asthma and Rhinitis: Findings from a Population-Based Survey. J Clin Epidemiol. 2001 Jun;54(6):610-8. 16.Ameille J, Pairon JC, Bayeux MC, Brochard P, Choudat D, Conso F, et al. Consequences of Occupational Asthma on Employment and Financial Status: A Follow- Up Study. Eur Respir J. 1997 Jan;10(1):55-8. 17. Barnes PJ, Jonsson B, Klim JB. The Costs of Asthma. Eur Respir J. 1996 Apr;9(4):636-42. 18. Cisternas MG, Blanc PD, Yen IH, Katz PP, Earnest G, Eisner MD, et al. A Comprehensive Study of the Direct and Indirect Costs of Adult Asthma. J Allergy Clin Immunol. 2003 Jun;111(6):1212-8. 19. Liss GM, Tarlo SM. Work Related Asthma. Occup Environ Med. 2002 Aug;59(8):503-4. 20. Environment and Workplace Health: Inventory of Federal, Provincial and Territorial Environmental and Occupational Health Data Sources and Surveillance Activities. 2004. Available from: http://www.hc-sc.gc.ca/ewh- semt/pubs/eval/inventory-repertoire/nphs_e.html. 63 Appendices Appendix A: Crude Estimates of Prevalence Odds Ratios (POR) Where, POR= odds of disease in exposed / odds of disease in unexposed Table A. 1 Prevalence Odds Ratio: Adult-Onset Asthma among Men High Risk Job Group (Exposed) Low Risk Job Group (Unexposed) Total Adult-onset Asthma (disease) 44313 166941 211254 No Asthma (no disease) 905963 6070921 6976883 Total 950276 6237861 7188137 Estimated POR (weighted) = 1.78 Table A. 2 Prevalence Odds Ratio: Adult-Onset Asthma among Women High Risk Job Group (Exposed) Low Risk Job Group (Unexposed) Total Adult-onset Asthma (disease) 40094 222771 262865 No Asthma (no disease) 704152 4186155 4890307 Total 744246 4408927 5153172 Estimated crude POR (weighted) = 1.07 64 Table A. 3 Prevalence Odds Ratio: Childhood-Onset Asthma among Men High Risk Job Group (Exposed) Low Risk Job Group (Unexposed) Total Childhood- onset Asthma (disease) 34430 189922 224352 No Asthma (no disease) 950276 6237861 7188137 Total 984706 6427783 7412489 Estimated POR (weighted) = 1.19 Table A. 4 Prevalence Odds Ratio: Childhood-Onset Asthma among Women High Risk Job Group (Exposed) Low Risk Job Group (Unexposed) Total Childhood- onset Asthma (disease) 44212 156016 200227 No Asthma (no disease) 744246 4408927 5153172 Total 788457 4564942 5353399 Estimated crude POR (weighted) = 1.68 65 Appendix B: Variable Descriptions Table B. 1 Description of asthma outcome and predictor variables considered defined for multivariate logistic regression analyses (supplement of Chapter 3) Variable Type ariableV Categories (if applicable) Category Interpretation Job Risk Group Categorical Predictor (primary predictor of interest) High Risk Jobs Versus Low Risk Jobs (Referent Group) Age Continuous Predictor 15-65 years old in 2002/03 Age (by decades) Categorical Predictor 1St age category 15<----age<24 years old in 2002/03 2nd age category 25<=age<34 years old in 2002/03 3 rdi^age category (Referent group) 35<=age<44 years old in 2002/03 4 th age category 45<=age<54 years old in 2002/03 5 th age category 55<----age <=65 years old in 2002/03 (note: this group includes one extra age) 66 Age (approximate post- versus pre- menopausal ages) Categorical Predictor Postmenopausal women, 50<=age<=65 years old in 2002/03 Versus Premenopausal, 15<=age<50 years old in 2002/03 (Referent Group) Education (years attained) Categorical Predictor Low: 1st Quintile Lowest quintile (<20%) for years of education as adjusted for age Low-Middle: 2" Quintile Education as adjusted by age is between 20% to <40% range of population (2nd quintile) Middle: 3 rd Quintile (Referent Group) Education as adjusted by age is between 40% and <60% range of population (3rd quintile) Middle-High: 4 th Quintile Represents the 4th (60% to <80%) quintile of years education adjusted for age High: 5 th Quintile Represents highest years of education adjusted for age, the 5th quintile (>=80%) Household Income Categorical Predictor Lowest Income Category Income is low: 1-2 persons & $10- 14,999 OR 3-4 persons & $10-19,999 OR 5+ persons & $15-29,999 OR any income lower than these criteria Middle Income Category Income is middle: 1-2 persons & $15-29,999 OR 3-4persons & $20- 39,999 OR 5+ persons & $30- 59,999 67 Highest Income Category (Referent Group) Income is high: 1-2 persons & $30-59,999 OR 3-4 persons & $40-79,999 OR 5+ persons & 60- 79,999 OR household income than these criteria Race Categorical Predictor "all other races": self-report any other race other than white ("black, korean, filipino, japanese, chinese, aboriginal, arab/west asian, latin american, south asian, SE asian, or multirace") Versus "white race": self-reported race as "white" Body Mass Index (BMI) Categorical Predictor Normal or Underweight (Referent Group) Normal: if 18-65 years old and BMI < 25; If 15-17 years old and 5%<=BMI<85% Overweight Overweight: if 18-65 years old and 25<=BMR30; if 15- 17 years old and 85%<=bmi<95% Obese Obese: if 18-65 years old and BMI>30; if 15-17 years old BMI>=95% 68 Smoking Status Categorical Predictor Non Smoker (Referent Group) Self—report is 'never smoked' Former Smoker Self-report is 'former daily smoker' or 'former occasional smoker' Current Smoker Self-report is 'daily smoker', 'occasional smoker & former daily smoker' or 'always occasional smoker' Adult-onset asthma Dichotomous, Outcome Variable Report having asthma in 2002/03 and age of asthma-onset is >=18 years of age Versus All non-asthmatics (do not include childhood-onset asthmatics, as they are not considered at-risk for adult-onset if have childhood- onset asthma) Childhood- onset asthma Dichotomous, Outcome Variable Report having asthma in 2002/03 and age of asthma-onset is <18 years of age Versus All others - all persons 15-65 in 2002/03 who are non-asthmatics or adult-onset asthmatic; adult- onset asthmatics at risk for childhood-onset asthma 69 Appendix C: Bivariate Associations between Predictor Variables Adult-onset asthma among women Table C. 1 Adult-onset asthma among women: association between Age and Body Mass Index Frequency Percent Row Pct Col Pct Table of Age by BMI Age BMI Total Normal/ Underweight Overweight Obese 15-24 427852 89979 47088 564919 years 8.59 1.81 0.95 11.34 75.74 15.93 8.34 15.3 6.47 5.93 25-34 686944 209068 125796 1021807 years 13.79 4.2 2.53 20.51 67.23 20.46 12.31 24.56 15.03 15.85 35-44 905049 451999 276986 1634035 years 18.17 9.07 5.56 32.8 55.39 27.66 16.95 32.36 32.5 34.9 45-54 607788 451016 247946 1306750 years 12.2 9.05 4.98 26.23 46.51 34.51 18.97 21.73 32.43 31.24 55-65 169236 188707 95741 453685 years 3.4 3.79 1.92 9.11 37.3 41.59 21.1 6.05 13.57 12.06 Total 2796869 1390769 793558 4981196 56.15 27.92 15.93 100 Statistic DF Value Prob Chi-Square 8 128.0257 <.0001 Weighted Sample Size = 4981196 70 Table C. 2 Adult-onset asthma among women: association between Job Risk Group and Age Frequency Table of Job Risk Group by Age Percent Job Age Total Row Pct Risk 15-24 25-34 35-44 45-54 55-65 Col Pct Group years years years years years Low 511968 999698 1429003 1061546 406711 4408927 Risk 9.94 19.4 27.73 20.6 7.89 85.56 11.61 22.67 32.41 24.08 9.22 87.7 89.87 85.07 80.37 89.09 High 71794 112687 250760 259223 49781 744246 Risk 1.39 2.19 4.87 5.03 0.97 14.44 9.65 15.14 33.69 34.83 6.69 12.3 10.13 14.93 19.63 10.91 Total 583762 1112385 1679763 1320770 456492 5153172 11.33 21.59 32.6 25.63 8.86 100 Statistic DF Value Prob Chi- 4 26.2137 <.0001 Square Weighted Sample Size = 5153172 71 Table C. 3 Adult-onset asthma among women: association between Job Risk Group and Body Mass Index Frequency Percent Row Pct Col Pct Table of Job Risk Group by Body Mass Index Job Risk Group Body Mass Index Total Normal/ Underweight Overweight Obese Low Risk 2438297 1152686 669203 4260186 48.95 23.14 13.43 85.53 57.23 27.06 15.71 87.18 82.88 84.33 High Risk 358573 238083 124354 721010 7.2 4.78 2.5 14.47 49.73 33.02 17.25 12.82 17.12 15.67 Total 2796869 1390769 793558 4981196 56.15 27.92 15.93 100 Statistic DF Value Prob Chi- 2 7.4545 0.0241 Square Weighted Sample Size = 4981196 72 Childhood-onset asthma among men Table C. 4 Childhood-Onset Asthma among Men: association between Age and Household Income Frequency Percent Table of Age by Household Income Age Household Income Total Row Pct Low Middle High Col Pct 15-24 years 73051 151390 545134 769574 1.05 2.17 7.82 11.04 9.49 19.67 70.84 34.02 16.15 9.37 25-34 years 26571 219352 1127738 1373661 0.38 3.15 16.18 19.71 1.93 15.97 82.1 12.37 23.4 19.38 35-44 years 33571 293496 1751727 2078794 0.48 4.21 25.13 29.82 1.61 14.12 84.27 15.63 31.31 30.11 *45-65 years 81559 273134 2393753 2748446 1.17 3.92 34.34 39.43 2.97 9.94 87.09 37.98 29.14 41.14 Total 214752 937371 5818352 6970475 3.08 13.45 83.47 100 Statistic DF Value Prob Chi-Square 6 81.0094 <.0001 Weighted Sample Size = 6970475 *Collapsed oldest two decades due to small samples sizes 73 Table C. 5 Childhood-Onset Asthma among Men: association between Age and Body Mass Index Frequency Table of Age by Body Mass Index Percent Age Body Mass Index Total Row Pct Normal/ Overweight Obese Col Pct Underweight 15-24 529004 258819 117389 905212 years 7.17 3.51 1.59 12.28 58.44 28.59 12.97 20.45 7.68 8.28 25-34 577549 623465 249006 1450020 years 7.83 8.46 3.38 19.67 39.83 43 17.17 22.32 18.51 17.56 35-44 691712 1004102 447315 2143129 years 9.38 13.62 6.07 29.07 32.28 46.85 20.87 26.74 29.81 31.54 *45-65 789017 1481760 604313 2875090 years 10.7 20.1 8.2 38.99 27.44 51.54 21.02 30.5 43.99 42.62 Total 2587282 3368146 1418023 7373452 35.09 45.68 19.23 100 Statistic DF Value Prob Chi - 6 129.2001 <.0001 Square Weighted Sample Size = 7373452 *Collapsed oldest two decades due to small samples sizes 74 Table C. 6 Childhood-Onset Asthma among Men: association between Job Risk Group and Household Income Frequency Table of Job Risk Group by Household Income Percent Job Risk Household Income Total Row Pct Group Low Middle High Col Pct Low Risk 177553 760382 5115034 6052969 2.55 10.91 73.38 86.84 2.93 12.56 84.5 82.68 81.12 87.91 High Risk 37200 176989 703318 917507 0.53 2.54 10.09 13.16 4.05 19.29 76.66 17.32 18.88 12.09 Total 214752 937371 5818352 6970475 3.08 13.45 83.47 100 Statistic DF Value Prob Chi-Square 2 14.7682 0.0006 Weighted Sample Size = 6970475 75 Table C. 7 Childhood-Onset Asthma among Men: association between Household Income and Race Frequency Percent Table of Household Income by Race Row Pct Household Race Total Col Pct Income All Other White Races Low 50335 164284 214619 0.72 2.36 3.09 23.45 76.55 6.41 2.66 Middle 146774 787850 934624 2.11 11.33 13.44 15.7 84.3 18.7 12.77 High 587984 5217553 5805536 8.45 75.02 83.48 10.13 89.87 74.89 84.57 Total 785093 6169686 6954779 11.29 88.71 100 Statistic DF Value Prob Chi-Square 2 23.7108 <.0001 Weighted Sample Size = 6954779 76 Table C. 8 Childhood-Onset Asthma among Men: association between Household Income and Body Mass Index Frequency Table of Household Income by Body Mass Index Percent Household Body Mass Index Total Row Pct Income Normal/ Overweight Obese Col Pct Underweight Low 85952 76042 52758 214752 1.24 1.1 0.76 3.09 40.02 35.41 24.57 3.61 2.36 3.95 Middle 417739 363814 147264 928817 6.02 5.24 2.12 13.38 44.98 39.17 15.86 17.56 11.28 11.03 High 1875604 2785146 1135381 5796131 27.03 40.13 16.36 83.52 32.36 48.05 19.59 78.83 86.36 85.02 Total 2379296 3225001 1335403 6939700 34.29 46.47 19.24 100 Statistic DF Value Prob Chi-Square 4 27.8439 <.0001 Weighted Sample Size = 6939700 77 Table C. 9 Childhood-Onset Asthma among Men: association between Race and Body Mass Index Frequency Table of Race by Body Mass Index Percent Race Body Mass Index Total Row Pct Normal/ Overweight Obese Col Pct Underweight All Other 347795 366537 132940 847272 Races 4.73 4.98 1.81 11.52 41.05 43.26 15.69 13.46 10.92 9.39 White 2236740 2991372 1282371 6510483 30.4 40.66 17.43 88.48 34.36 45.95 19.7 86.54 89.08 90.61 Total 2584535 3357910 1415311 7357755 35.13 45.64 19.24 100 Statistic DF Value Prob Chi- 2 6.9843 0.0304 Square Weighted Sample Size = 7357755 78 Childhood-onset asthma among women Table C. 10 Childhood-Onset Asthma among Women: association between Age and Household Income Frequency Percent Table of Age by Household Income Age Household Income Total Row Pct Low Middle High Col Pct 15-24 years 62513 117349 428023 607885 1.24 2.32 8.47 12.02 10.28 19.3 70.41 28.37 15.57 10.48 25-34 years 45513 226699 851883 1124095 0.9 4.48 16.85 22.23 4.05 20.17 75.78 20.66 30.08 20.87 35-44 years 69228 231281 1332149 1632658 1.37 4.57 26.35 32.29 4.24 14.17 81.59 31.42 30.69 32.63 *45-65 years 43082 178213 1470305 1691600 0.85 3.52 29.08 33.46 2.55 10.54 86.92 19.55 23.65 36.02 Total 220337 753542 4082361 5056239 4.36 14.9 80.74 100 Statistic DF Value Prob Chi-Square 6 65.968 <.0001 Weighted Samp e Size = 5056239 *Collapsed oldest two decades due to small samples sizes 79 Table C. 11 Childhood-Onset Asthma among Women: association between Age and Body Mass Index Frequency Percent Row Pct Col Pct Table of Age by Body Mass Index Age Body Mass Index Total Normal/ Underweight Overweight Obese 15-24 489441 104476 52602 646519 years 9.48 2.02 1.02 12.52 75.7 16.16 8.14 16.75 7.35 6.39 25-34 720528 216437 136836 1073801 years 13.95 4.19 2.65 20.79 67.1 20.16 12.74 24.66 15.23 16.63 35-44 923571 459656 286999 1670227 years 17.88 8.9 5.56 32.33 55.3 27.52 17.18 31.61 32.34 34.89 *45-65 787870 640801 346178 1774850 years 15.25 12.41 6.7 34.36 44.39 36.1 19.5 26.97 45.08 42.08 Total 2921411 1421370 822616 5165397 56.56 27.52 15.93 100 Statistic DF Value Prob Chi-Square 6 127.4676 <.0001 Weighted Sample Size = 5165397 *Collapsed oldest two decades due to small samples sizes 80 Table C. 12 Childhood-Onset Asthma among Women: association between Age and Smoking Status Frequency Table of Age by Smoking Status Percent Age Smoking status Total Row Pct Non-smoker Former Current Col Pct Smoker Smoker 15-24 years 185511 176952 311183 673646 3.47 3.31 5.83 12.61 27.54 26.27 46.19 9.82 9.24 20.23 25-34 years 410956 372155 385140 1168251 7.69 6.97 7.21 21.87 35.18 31.86 32.97 21.75 19.44 25.04 35-44 years 587632 654509 476090 1718230 11 12.25 8.91 32.16 34.2 38.09 27.71 31.1 34.19 30.95 *45-65 years 705674 710679 365695 1782048 13.21 13.3 6.85 33.36 39.6 39.88 20.52 37.34 37.12 23.78 Total 1889773 1914295 1538108 5342176 35.37 35.83 28.79 100 Statistic DF Value Prob Chi-Square 6 87.1604 <.0001 Weighted Sample Size = 5342176 *Collapsed oldest two decades due to small samples sizes 81 Table C. 13 Childhood-Onset Asthma among Women: association between Education and Household Income Frequency Percent Table of Education by Household Income Education Household Income Total Row Pct Low Middle High Col Pct Low 29584 54329 179300 263212 (<20% range) 0.59 1.07 3.55 5.21 11.24 20.64 68.12 13.43 7.22 4.39 Low-Mid 66041 170162 704606 940809 (20-40% 1.31 3.37 13.94 18.61 range) 7.02 18.09 74.89 29.97 22.62 17.26 Middle (40- 50871 165436 678630 894937 60% range) 1.01 3.27 13.42 17.7 5.68 18.49 75.83 23.09 21.99 16.62 Mid-High (60- 28815 196913 911591 1137319 80% range) 0.57 3.9 18.03 22.5 2.53 17.31 80.15 13.08 26.17 22.33 High (>80% 45026 165535 1608233 1818795 range) 0.89 3.27 31.81 35.98 2.48 9.1 88.42 20.44 22 39.39 Total 220337 752375 4082361 5055072 4.36 14.88 80.76 100 Statistic DF Value Prob Chi-Square 8 80.3941 <.0001 Weighted Sample Size = 5055072 82 Table C. 14 Childhood-Onset Asthma among Women: association between Education and Smoking Status Frequency Table of Education by Smoking Status Percent Education Smoking Status Total Row Pct Non- Former Current Col Pct smoker Smoker Smoker Low 77211 76351 138418 291980 (<20% range) 1.45 1.43 2.59 5.47 26.44 26.15 47.41 4.09 4 9 Low-Mid 278673 335560 367160 981392 (20-40% 5.22 6.29 6.88 18.39 range) 28.4 34.19 37.41 14.76 17.56 23.87 Middle (40- 284546 368584 320579 973709 60% range) 5.33 6.91 6.01 18.24 29.22 37.85 32.92 15.07 19.29 20.84 Mid-High (60- 383898 447693 356874 1188465 80% range) 7.19 8.39 6.69 22.27 32.3 37.67 30.03 20.33 23.43 23.2 High (>80% 864276 682542 355077 1901896 range) 16.19 12.79 6.65 35.63 45.44 35.89 18.67 45.76 35.72 23.09 Total 1888605 1910730 1538108 5337443 35.38 35.8 28.82 100 Statistic DF Value Prob Chi-Square 8 115.965 <.0001 Weighted Sample Size = 5337443 83 Table C. 15 Childhood-Onset Asthma among Women: association between Job Risk Group and Education Frequency Percent Table of Job Risk Group by Education Job Risk Education Total Row Pct Group Low Low- Middle Mid- High Col Pct (<20% Mid (20- (40-60% High (>80% range) 40% range) (60-80% range) range) range) Low Risk 191838 842574 791109 1006003 1732252 4563775 3.59 15.75 14.79 18.81 32.39 85.33 4.2 18.46 17.33 22.04 37.96 65.7 85.85 81.25 84.65 90.55 High Risk 100142 138819 182601 182462 180868 784892 1.87 2.6 3.41 3.41 3.38 14.67 12.76 17.69 23.26 23.25 23.04 34.3 14.15 18.75 15.35 9.45 Total 291980 981392 973709 1188465 1913119 5348666 5.46 18.35 18.2 22.22 35.77 100 Statistic DF Value Prob Chi- 4 72.1146 <.0001 Square Weighted Sample Size = 5348666 84 Table C. 16 Childhood-Onset Asthma among Women: association between Job Risk Group and Age Frequency Percent Row Pct Table of Job Risk Group by Age Job Risk Group Age Total Col Pct 15-24 25-34 35-44 *45-65 years years years years Low Risk 583826 1045648 1452795 1482672 4564942 10.91 19.53 27.14 27.7 85.27 12.79 22.91 31.83 32.48 86.67 89.44 84.51 82.75 High Risk 89820 123433 266200 309004 788457 1.68 2.31 4.97 5.77 14.73 11.39 15.65 33.76 39.19 13.33 10.56 15.49 17.25 Total 673646 1169081 1718996 1791677 5353399 12.58 21.84 32.11 33.47 100 Statistic DF Value Prob Chi -Square 3 13.4583 0.0037 Weighted Sample Size = 5353399 *Collapsed oldest two decades due to small samples sizes 85 Table C. 17 Childhood-Onset Asthma among Women: association between Job Risk Group and Household Income Frequency Percent Table of Job Risk Group by Household Income Job Risk Household Income Total Row Pct Group Low Middle High Col Pct Low Risk 183804 610763 3538792 4333360 3.64 12.08 69.99 85.7 4.24 14.09 81.66 83.42 81.05 86.68 High Risk 36532 142779 543568 722879 0.72 2.82 10.75 14.3 5.05 19.75 75.19 16.58 18.95 13.32 Total 220337 753542 4082361 5056239 4.36 14.9 80.74 100 Statistic DF Value Prob Chi-Square 2 8.6779 0.0131 Weighted Sample Size = 5056239 86 Table C. 18 Childhood-Onset Asthma among Women: association between Job Risk Group and Body Mass Index Frequency Table of Job Risk Group by Body Mass Index Percent Job Risk Body Mass Index Total Row Pct Group Normal/ Overweight Obese Col Pct Underweight Low Risk 2544978 1176029 689373 4410379 49.27 22.77 13.35 85.38 57.7 26.67 15.63 87.11 82.74 83.8 High Risk 376433 245341 133243 755017 7.29 4.75 2.58 14.62 49.86 32.49 17.65 12.89 17.26 16.2 Total^2921411 1421370 822616 5165397 56.56 27.52 15.93 100 Statistic DF Value Prob Chi- 2 8.2679 0.016 Square Weighted Sample Size = 5165397 87 Table C. 19 Childhood-Onset Asthma among Women: association between Age and Education Frequency Table of Age by Education Percent Age Education Total Row Pct Low Low-Mid Middle Mid- High Col Pct (<20% (20-40% (40-60% High (>80% range) range) range) (60-80% range) range) 15-24 36283 89157 238777 49283 260147 673646 years 0.68 1.67 4.46 0.92 4.86 12.59 5.39 13.23 35.45 7.32 38.62 12.43 9.08 24.52 4.15 13.6 25-34 46358 180046 177881 396312 367317 1167913 years 0.87 3.37 3.33 7.41 6.87 21.84 3.97 15.42 15.23 33.93 31.45 15.88 18.35 18.27 33.35 19.2 35-44 118879 370601 273682 340742 611526 1715430 years 2.22 6.93 5.12 6.37 11.43 32.07 6.93 21.6 15.95 19.86 35.65 40.71 37.76 28.11 28.67 31.96 *45-65 90461 341588 283370 402128 674129 1791677 years 1.69 6.39 5.3 7.52 12.6 33.5 5.05 19.07 15.82 22.44 37.63 30.98 34.81 29.1 33.84 35.24 Total 291980 981392 973709 1188465 1913119 5348666 5.46 18.35 18.2 22.22 35.77 100 Statistic DF Value Prob Chi- 12 157.5228 <.0001 Square Weighted Sample Size = 5348666 *Collapsed oldest two decades due to small samples sizes 88 Table C. 20 Childhood-Onset Asthma among Women: association between Household Income and Smoking Status Table of Household Income by Smoking Status Percent Household Smoking status Total Row Pct Income Non- Former Current Col Pct Smoker Smoker Smoker Low 86431 43577 90329 220337 1.71 0.86 1.79 4.37 39.23 19.78 41 4.95 2.35 6.25 Middle 230947 231471 291123 753542 4.58 4.59 5.77 14.94 30.65 30.72 38.63 13.22 12.5 20.13 High 1430022 1576578 1064536 4071137 28.35 31.25 21.1 80.7 35.13 38.73 26.15 81.84 85.15 73.62 Total 1747400 1851626 1445989 5045015 34.64 36.7 28.66 100 Statistic DF Value Prob Chi-Square 4 40.2712 <.0001 Weighted Sample Size = 5045015 89 Table C. 21 Childhood-Onset Asthma among Women: association between Body Mass Index and Smoking Status Frequency Table of Body Mass Index by Smoking Status Percent Body Mass Smoking status Total Row Pct Index Non- Former Current Col Pct Smoker Smoker Smoker Normal/ 1017063 945429 952916 2915408 Underweight 19.73 18.34 18.49 56.56 34.89 32.43 32.69 55.61 51.59 63.85 Overweight 535726 554356 326067 1416149 10.39 10.76 6.33 27.48 37.83 39.15 23.02 29.29 30.25 21.85 Obese 276278 332949 213390 822616 5.36 6.46 4.14 15.96 33.59 40.47 25.94 15.1 18.17 14.3 Total 1829066 1832734 1492372 5154173 35.49 35.56 28.95 100 Statistic DF Value Prob Chi-Square 4 27.7792 <.0001 Weighted Sample Size = 5154173 90 Appendix D: Multivariable Model Results for Childhood-Onset Asthma among Men and Women (household income included as a predictor) Table D. 1 Multivariable logistic regression results (prevalence odds ratios) for childhood-onset asthma among men and women, considering household income as a predictor variable Childhood-Onset Asthma MEN WOMEN POR [95% CI] POR [95% CI] Weighted n= Weighted n= 7,396,793 5,353,399 Predictor Predictor Category Job Risk Group Low Risk _ — High Risk 1.50 1.62 [0.81, 2.76] [0.93, 2.85] Age (decades) 15-24 years 2.86 7.22 [1.62, 5.05] [3.96, 13.18] 25-34 years 1.08 2.52 [0.59, 2.01] [1.34, 4.72] 35-44 years 45-54 years 0.20 0.42 [0.09, 0.46] [0.17, 1.04] 55-65 years 91 Household Income Low 3.59 [1/74, 7.43] 2.11 [0.97, 4.58] Middle 0.70 [0.34, 1.44] 2.75 [1.71, 4.42] High _ ____ Race White _ Not included in final model All Other Races 2.596 [1.50, 4.48] Model Likelihood Ratio Test, z2 Statistic (p-value) 81.83 (p<0.0001) 109.04 (p<0.0001) Table D.1 has been provided as a supplement to Chapter 3 to show the multivariate logistic regression results for childhood-onset asthma among men and women when household income is included as a predictor in the model. Please refer to the 'Results' section of Chapter 3 to place this table into the context of this thesis. 92 Appendix E: Recoding of the Asthma-Specific Job Exposure Matrix The asthma-specific job exposure matrix (JEM) was the key component for exposure assessment throughout this thesis work. The JEM developed by Susan M. Kennedy utilized the International Labour Organisation's (ILO) International Standard Classification of Occupations (ISCO-88) system for job codes2 . However, the job codes provided by the Statistics Canada surveillance programs used for this thesis work were based on the Standard Occupational Classification (SOC) 1991 system 3 . As a result, in order to use the JEM for the exposure assessment aspect of this thesis work (i.e. for determining job risk groups, as described in 'Methods' section of Chapter 2), it was necessary to recode the JEM to the SOC 1991 system. There are no available cross-walks for translating between the ISCO-88 and SOC 1991 job codes. As a result, each ISCO-88 code was individually matched to an SOC 1991 code based on the job code description and job examples provided for each job code. The job codes, descriptions and job examples used for matching were found on the ILO website and Statistics Canada website for the ISCO-88 and SOC 1991 job codes, respectively. The recoding of the JEM was a lengthy process, requiring approximately two months of my time. Job codes for which there was uncertainty in the match from the ISCO-88 to an SOC 1991 coding system were reviewed with Susan M. Kennedy, before a final match was made. ' Kennedy SM, et al. OEM. 57;635-641 (2000) 2 International Standard Classification of Occupations (ISCO-88). United States Bureau of Statistics: International Labour Organisation. http://www.ilo.org/public/english/bureau/stat/isco/isco88/major.htm . Updated: September 18, 2004. 3 Standard Occupational Classification (SOC) 1991 — Canada. Statistics Canada. http://www.statcan.ca/english/Subjects/Standard/soc/1991/soc91-menu.htm . Date Modified: June 17, 2003. 93

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}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            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.24.1-0066676/manifest

Comment

Related Items