|Home | About | Journals | Submit | Contact Us | Français|
E. C. McCanlies1, D. Fekedulegn1, L. Charles1, W. Sanderson2, L. Croen3, I. Hertz‐Picciotta4. 1National Institute for Occupational Safety and Health; 2University of Iowa; 3Kaiser Permanente; 4University of California‐Davis
ObjectivesObtaining accurate occupational exposure information can be a challenge. Here we compared agreement of an occupational exposure assessment by an industrial hygienist panel with self‐reported exposures from the Childhood Autism Risks from Genetics and Environment (CHARGE) study.
MethodsMothers of 249 children (ages 2–5) diagnosed with autism, mental retardation or developmental delay, and children from the general population were interviewed via telephone about the parents' occupation around the time of pregnancy with the study child. Fathers were not interviewed. Three industrial hygienists independently assessed likelihood of occupational exposure based on parents' listed industry, occupation, job type and responsibilities. Self‐report and expert‐assessed exposure data were obtained for approximately 190 mothers and 208 fathers. Prevalence of exposure to various agents was determined. The κ statistic and positive predictive value were used as measures of concordance between the self‐report and industrial hygiene data.
ResultsThe least and most prevalent reported exposures for fathers were ethylene oxide exposure (2.5%) and disinfectant exposure (26.9%). The κ statistics for these were −0.02 and 0.22, respectively. Industrial hygienists' prevalence estimates for fathers varied from 0.96% (anaesthetic gas) to 43.0% (solvents). The κ statistics for these were −0.02 and 0.45, respectively. The exposure with the highest positive predictive value was solvents (86%) and the highest overall κ statistic was 0.45 (metal dust or fumes, PCBs and solvents). The least prevalent exposure self‐reported by the mothers was ethylene oxide (0.54%) and the most prevalent exposure was disinfectants (38.3%). The κ statistics for these were 0.12 and 0.22, respectively. Disinfectant exposure was also the most prevalent exposure the industrial hygienists reported for the mothers (10.6%), but they reported PCBs as occurring least often (0.0%). The exposure with the highest positive predictive value was ethylene oxide (100%) and the highest overall κ statistic was 0.64 (x ray/radioactive material).
ConclusionThe agreement between the self‐report and expertly assessed occupational exposure data is relatively low. Exposure estimates based on industrial hygiene review have been found to be more accurate than self‐report data. This would indicate that if these data were to be used in future studies, using the expert exposure assessment may provide more accurate results.
Key wordsexposure assessment; self‐report; expert‐assessed
B. Pesch1, I. Gross1, T. Weiss1, M. Gebel1, R. Van Gelder2, D. Dahmann3, T. Bruening1. 1BGFA; 2BGIA; 3IGF
ObjectivesThe International Agency for Research on Cancer (IARC) classified nickel compounds as human carcinogens (group 1) due to increased lung cancer mortality in nickel refineries. Metallic nickel and welding were classified as possibly carcinogenic to humans (group 2B). To estimate the lung cancer risks of nickel we provide an approach to assess occupational exposure by nickel compound and setting in population‐based studies.
MethodsThe approach to assess occupational exposure combines a nickel‐specific job exposure matrix (JEM) with additional exposure information from supplemental questionnaires (SQs). We assembled available JEMs, workplace measurements and other information on nickel exposure in various settings such as refining and welding.
ResultsThe proposed nickel JEM is structured by compound, setting (industry/job), time and region. To quantify exposure levels, occupational hygiene measurements from the German MEGA database and other data have been analysed. Models of historical workplaces and expert ratings on the use of protective measures and other exposure determinants add information along the time axis. The job axis refers to the typical technological processes by industry and/or job title. To quantify a worker's exposure, this job axis has to be linked to the occupational history via job and industry data from the questionnaire using common economic classifications (ISCO 1968, ISIC 1968). Additional determinants of exposure levels such as protective measures can be retrieved from SQs to refine a worker's exposure estimate with weight factors.
ConclusionSeveral JEMs have been developed to assess exposure to nickel but do not differ by nickel compound. Here, we consider the chemical form and solubility with respect to its impact on carcinogenicity. For a quantitative JEM, appropriately determined nickel concentrations should be assembled, for example in the alveolar fraction of particles sampled with personal devices. A limitation of JEMs is the use of average exposure levels in jobs and industries. The proposed approach combines a measurement‐based JEM with additional exposure determinants retrieved from SQs to improve the assessment of occupational exposure to nickel in lung cancer research.
Key wordsexposure assessment; nickel; lung cancer
R. Vermeulen1, L. Portengen1, J. Coble2. 1Institute for Risk Assessment Sciences, Utrecht University; 2National Cancer Institute, Rockville, US
ObjectivesOccupational exposure assessment for case‐control studies is often based on the linkage of expert assessments or job exposure matrices to individual job information. Only limited use has been made of quantitative exposure data in case‐control studies for chronic health effects. Recent developments in the establishment of (national) exposure databases provides a unique, albeit challenging, opportunity to incorporate quantitative estimates of exposure in an efficient and transparent manner. We present here the experiences gained in the conduct of benzene exposure assessment for an industry based case‐cohort and a population based case‐control study.
MethodsThe case‐cohort study is a multi‐centre study of benzene and cancer risk nested within the NCI/CDC cohort. Approximately 3000 subjects working at more than 500 sites were included in this study. Exposure data is being extracted from study factories and governmental records. The constructed benzene exposure database will comprise nearly 9000 stationary benzene measurements collected between 1955 and 2004. The Shanghai Women's Health Study is a population based study of approximately 75000 women. To aid the exposure assessment in this study, exposure measurements have been extracted from files of the Shanghai CDC. The Shanghai database consists of nearly 70000 stationary benzene measurements collected between 1953 and 2000.
ResultsAnalyses of variance elucidating the variability at an industry, factory or job level could inform decisions about approaches to exposure assessment in population based case‐control studies. If exposure variability were largely determined by “factory”, then exposure measurements and other relevant determinants need to be collected at the factory level. In contrast, if exposure were determined largely by the job, one would focus on ascertaining information at this level. However, ascertaining information at the factory level from individuals in the study or their next of kin, might be challenging if the determinants are not directly related to the person's job but rather to workplace characteristics.
ConclusionLarge exposure databases can provide useful data to inform exposure assessment strategies and could be used to develop tools (ie, company‐ or job‐specific questionnaires) for case‐control studies.
Key wordscase‐control; benzene; exposure assessment
H. Kromhout1, L. Lillienberg2, J‐P. Zock3. 1IRAS UU; 2Sahlgrenska University; 3CREE IMIM
ObjectivesAssessing exposure for epidemiological studies trying to arrive at quantitative exposure–response relationships has become the norm rather than the exception. Until now quantitative exposure assessment has been restricted to industry‐based studies. Only very few community‐based studies have ever explored the feasibility to quantitatively assess exposure for their study subjects, mainly because of the multitudes of workplaces involved and lack of accessibility to or measurement data from these workplaces. Alternative methods using algorithms derived from exposure measurement databases might be an approach for getting around these limiting factors.
MethodsWithin the follow‐up study of the multi‐centre European Community Respiratory Health Survey we used structured questionnaires focusing on several jobs and exposures. Each of the seven modules was less than two pages long and tried to ask for information on a priori identified determinants of exposure.
ResultsOnly for the welding module were we able to estimate (semi‐)quantitatively cumulative exposure to welding fumes during the years of follow‐up. For the other modules quantitative approaches were impossible because they were either dealing with several different agents or we lacked readily available measurement data.
ConclusionWe are convinced that structured short questionnaires focusing on main exposure affecting factors in combination with exposure algorithms based on accurate statistical or physical models for certain occupational exposures will allow for quantitative exposure–response analyses even in community‐based studies. However, it will only be possible for a limited amount of well‐studied agents. Furthermore, one might argue that quantitative exposure–response relationships should preferably and more appropriately be studied in industry‐based studies. Community‐based studies could then be used to generate hypotheses or to roughly estimate population attributable risks in the community at large.
Key wordsexposure assessment; community‐based studies; questionnaires
L. Fritschi1, T. Sadkowsky2, M. Friesen3, J. Girschik1, D. Glass3, G. Benke3. 1Western Australian Institute for Medical Research; 2Research IT Online Pty Ltd; 3Centre for Occupational and Environmental Health, Monash University
ObjectivesAssessing occupational exposure in retrospective, community‐based case‐control studies is difficult as often there are no measurement data. The expert assessment method, developed by Jack Siemiatycki and his team in Montreal, is considered the most accurate way to collect this information but it is a very time consuming and expensive process.
MethodsWe have developed a web‐based software application called “OccIDEAS” which manages and automates aspects of the expert assessment of occupational exposures within epidemiological studies. OccIDEAS is a comprehensive and very flexible system, which provides an extensive range of functionalities and can be tailored easily according to the user's needs to focus on selected exposures. It has been designed to manage all facets of the exposure assessment workflow, including: selection and retrieving of appropriate questions; interviewer/interviewee interface; completed questionnaire data storage; extraction of questionnaire data; questionnaire data viewing interface; exposure assessment interface; and exposure data storage.
ResultsIn particular, OccIDEAS brings a novel approach to occupational agent exposure assessment, combining the experts' knowledge with sophisticated computer algorithms designed to allow the program to learn from experience. Thus we can automate some of the assessments and the proportion of assessments which can be automated increases with time. The automatic assessments decrease the repetitive part of the experts' work, allowing them to spend more time on assessing the more challenging and unusual jobs. Experienced occupational researchers can develop their own questionnaires and their own assessments using the OccIDEAS structure or we can provide a full occupational assessment service for those researchers without occupational epidemiological expertise. The whole system will make the expert assessment of exposures more efficient and consistent and less costly.
ConclusionThis novel application is being trialled in studies now. In this presentation we will outline the features and the potential uses of OccIDEAS.
Key wordsexposure assessment; case‐control; community studies