|Home | About | Journals | Submit | Contact Us | Français|
M. Dennekamp1, G. Benke1, J. Cui1, A. DelMonaco1, A. W. Musk2, N. de Klerk2, L. Fritschi3, M. R. Sim1, M. J. Abramson4. 1Centre for Occupational and Environmental Health, Monash University; 2School of Population Health, University of Western Australia; 3Western Australian Institute for Medical Research; 4Clinical Epidemiology Unit, Monash University
ObjectivesTo investigate the incidence of respiratory symptoms and changes in pulmonary function in relation to occupational exposures in two aluminium smelters in Victoria, Australia.
MethodsThe study population consisted of employees who started employment between 1995 and the end of 2003 at two aluminium smelter sites in Victoria, Australia. Participation involved an initial assessment at the start of employment and annual follow‐up assessments for the duration of employment at the smelters. These assessments involved an interviewer‐administered questionnaire which included questions about cough, shortness of breath, wheeze, chest tightness and rhinitis. In addition lung function was assessed by measuring forced expiratory volume in 1 s (FEV1) and forced vital capacity (FVC). Methacholine challenge testing was performed to identify bronchial hyper‐responsiveness (BHR). The occupational exposures investigated in this study were fluoride, sulphur dioxide (SO2), inspirable dust, benzene soluble fraction (BSF) and oil mist. The exposure levels to these substances were calculated for each participant in the study using their job history information and routine personal air monitoring data. The random effects model was used to investigate the association between occupational exposures and health outcomes.
ResultsIn total 446 employees participated in the study. Of the five symptoms, only wheeze and chest tightness, the two symptoms most closely related to asthma, showed associations with exposures at the sites. Workers who were exposed to fluoride, inspirable dust, SO2 and BSF were more likely to report wheeze. Chest tightness was associated with exposure to dust, SO2 and BSF. Many of the exposures were highly correlated, but it is most likely that fluoride and SO2, known respiratory irritants, were responsible for the symptom effects observed. Fluoride, inspirable dust and SO2 were the most important agents having an effect on lung function. A significant reduction in the FEV1/FVC ratio (an indicator of airflow limitation), and declines in both FEV1 and FVC over time were associated with these exposures. BHR was also related to exposure to fluoride, SO2, inspirable dust and BSF.
ConclusionSome occupational exposures in aluminium smelters are associated with increased wheeze and chest tightness and small decrements in lung function and BHR.
Key wordsaluminium smelter; pulmonary function; respiratory symptoms
E. Viragh. Tg‐Mures University of Medicine and Pharmacy
ObjectivesThis epidemiological field study was conducted to determine the prevalence of chronic bronchitis in female persons exposed to hydrocarbons in the petrochemical industry and the surrounding environment.
MethodsAir monitoring in all workplaces and around the factory of aliphatic (AH) and polycyclic aromatic hydrocarbons (PAH) was carried out during a 7‐year period. 147 females aged 38.2 (SD 6.1) years and with a mean exposure time of 14.5 (SD 1.8) years from different workplaces were examined. The control group also contained 147 females living around the factory. Clinical examinations, a bronchitis questionnaire and ventilatory tests were performed. The bronchitis questionnaire focused on the frequency of chronic bronchitis in relation to professional (hydrocarbons) and non‐professional (former respiratory diseases, age, smoking) factors. The χ2 test was done to find out the importance of cigarette smoking to chronic bronchitis onset. Linear regression analyses were performed to determine the relationship between the levels of exposure to hydrocarbons and the frequency of chronic bronchitis in studied females.
ResultsBitumen petroleum and its refinery products contain a large number of AH and PAH. Most of the values were above the maximum admissible concentration (MAC) during the studied period in the refinery and the surrounding area. 19.7% of exposed female workers and 8.7% of professional non‐exposed females suffered from chronic bronchitis. Approximately 25% of females were smokers in both groups. The results of the χ2 test in both groups have shown that cigarette smoking was also statistically significant in chronic bronchitis inducement. No important functional pulmonary modifications were observed. Linear regression analyses have shown a close, positive correlation between the levels of exposure to hydrocarbons and frequencies of chronic bronchitis only in the professional exposed female workers (r=6.95).
ConclusionThese findings underlined the irritative effect of hydrocarbons on the respiratory system. Cigarette smoking, the existence of former respiratory diseases, and age have also their own roles in inducing chronic bronchitis. It is advised that levels of hydrocarbons in all workplaces (consecutively around the factory) are reduced, and individual health protection equipment is correctly used.
Key wordsexposure to hydrocarbons; prevalence of chronic bronchitis; female workers
V. H. Arrandale1, M. Koehoorn2, Y. C. MacNab2, H. Dimich‐Ward2, S. M. Kennedy2. 1University of Toronto; 2University of British Columbia
ObjectivesChanges in respiratory symptoms over time are rarely studied as an outcome in occupational epidemiology. The purpose of this study was to investigate trajectories of dyspnea change over time using a newer statistical model.
MethodsData from a longitudinal study of marine workers was utilised (825 men, 101 women). Subjects attended two to four visits over a 12‐year period. SAS Proc Traj, a finite mixture model, was used to investigate multiple trajectories of dyspnea change over time and the factors associated with membership in different trajectory groups.
ResultsTwo trajectories, or distinct patterns of dyspnea change over time, provided the best fit to the data. Group 1, 70% of the cohort, had a constantly low probability of reporting dyspnea (<0.10); group 2, 30% of the cohort, had a steadily increasing probability for reporting dyspnea over time (0.40–0.80). Age, sex, childhood asthma, historical asbestos exposure, smoking, exposure to irritants and pulmonary function (FEV1% predicted) were offered into the model as predictors of trajectory membership. The best fit multivariable model indicated that older subjects (p<0.0001) and current smokers (p=0.007) were significantly more likely to belong to group 2 (increasing dyspnea). Subjects with current exposure to respiratory irritants (p<0.0001) were more likely to belong to group 1 (low probability of dyspnea). After adjusting for level of lung function, being female became a significant predictor of membership in group 2 (p=0.03), but the effect of current smoking was no longer statistically significant (p=0.09). Subjects with lower levels of lung function were more likely to belong to group 2 (p<0.0001). The effect of age and exposure (asbestos and respiratory irritants) was unchanged after adjusting for level of lung function.
ConclusionResults suggest two patterns of dyspnea change over time in the population. Older age, being female and lower lung function were associated with the increasing trajectory of dyspnea symptoms. Researchers should continue to explore the possibility that multiple trajectories of respiratory symptom change exist. Further research may identify associations between personal characteristics, symptom trajectories and disease outcomes with the potential to improve occupational disease surveillance.
Key wordsrespiratory symptoms; longitudinal studies; occupational lung disease
A. Tabaku1, S. Bala2, Z. ElizanaPetrela1. 1Public Health Institute; 2University Hospital for Pulmonary Disease
ObjectivesThe aim of this study was to demonstrate a link between air pollution of an occupational area and the possibility of developing COPD, as well as to assess the prevalence and severity of COPD, and the prevalence of its symptoms.
MethodsWe have identified and assessed all risk factors (markers of exposure: TSP, PM10, sulfur dioxide, carbon monoxide, etc) and also have measured pulmonary function (markers of health effects: FVC, FEV1, FEV1/FVC ratio, PEF and FEF25–75) in 400 subjects working in Albanian heavy industry. A standardised questionnaire was used to collect data on smoking habit, socioeconomic status, past history of pulmonary diseases, current respiratory symptoms, education, etc. We have used the SPSS 12 package for statistical processing of the results.
ResultsData obtained have shown that the concentrations of markers of exposure exceed the TLV values. The relative risk for developing COPD was computed as a whole RR of 2.79 (95% CI 1.92 to 3.28) and after stratifying based on exposure levels, RR 1.47 (95% CI 0.84 to 2.25) up to RR 5.51 (95% CI 3.98 to 6.73). The prevalence of COPD was high and varied from 5.85% to 42.37%, and its severity was ranged from mild to very severe, and from 5.26% to 8.74% of the workers were at risk. The prevalence of symptoms in this study ranged from 12.02% to 46.30%.
ConclusionThe overall conclusion was that the prevalence of COPD and its symptoms are high, and that there exists a well established link between air pollution of occupational areas and the possibility of developing COPD in Albanian exposed workers.
Key wordsoccupational exposure; COPD; PM10
T. Meijster1, N. Warren2, D. Heederik3, E. Tielemans4. 1TNO‐IRAS; 2Health and Safety Laboratory; 3Institute for Risk Assessment Sciences, Utrecht University; 4Business Unit Quality and Safety, TNO Quality of Life
ObjectivesThe population dynamic health model presented here simulates a population of workers longitudinally through time and tracks the development of sensitisation and respiratory symptoms in each worker related to their exposure.
MethodsThe model has three components: (1) a basic population model describing the length of a worker's career in the bakery sector and the influx of new workers; (2) a multi‐stage disease model describing the evolution of sensitisation and respiratory symptoms in each worker over time (this includes definition of dose–response models and probability estimates relating the risk of developing sensitisation and symptoms to individual exposure, for atopics and non‐atopics); and (3) input distributions of population occupational exposure to flour dust and allergens. The disease state for each worker is modelled using a separate discrete time Markov chain, updated yearly on the basis of the simulated exposures for that individual. A Bayesian analysis of data from an epidemiological study among bakers, supplemented with information from several other studies, has provided estimates of the yearly transition probabilities between disease states. This information was used to set up the basic model. Sensitivity analysis provides information on the most influential model parameters and uncertainties associated with the model.
ResultsThe described approach allows us to study the development of diseases and transitions between disease states over time in relation to determinants of disease including allergen exposure. Furthermore, it enables more realistic modelling of the health impacts of different intervention strategies at the workplace (eg, changes in exposure take several years to impact on ill‐health and often occur as a continual trend). Reduction of exposure during extreme situations (ie, the tail of the exposure distribution) may have a different impact on health in a population compared to an overall shift of the exposure distribution. These differences can be studies in a population dynamic model approach.
ConclusionEventually the model will be refined as necessary with information from recent cross‐sectional studies and will eventually be used to evaluate the long‐term effect of a range of interventions on ill‐health in bakery workers in the UK and Netherlands.
Key wordsdynamic population model; health impact assessment; respiratory health effects
E. Batistatou1, R. McNamee1, M. Van Tongeren2. 1University of Manchester; 2Institute of Occupational Medicine
ObjectivesExposure measurement error leads to bias in assessing exposure effects. The grouping approach used in occupational epidemiology, in which subjects are grouped into several a priori defined groups, may reduce bias by assigning the group‐mean to the subjects in a group, at the expense of increased variability of the estimates. An alternative approach to bias correction is to use instrumental variables (IV). Our aim was to compare the IV method to various grouping approaches in terms of bias.
MethodsHypothetical studies to estimate a regression coefficient were simulated in which response Y and two independent, normally distributed, error‐prone exposure measures W1, W2, with W=X+E, were measured on 1000 subjects; exposure reliability r varied from 0.5 to 0.8. The true relationship between Y and true exposure X was assumed to be Y=4−0.1X+d. For the pure IV method, W2 was used as an instrument in the regression analysis of Y on W1. In the grouping methods, individual mean exposure Wm was first calculated, then roughly equal‐sized subject groups were formed by ranking on Wm and the regression of Y on group means of Wm was found. Bias was measured by per cent attenuation of the regression coefficient b, across simulations. Additional simulations with log‐normal distributed exposure were performed. Mixed IV/group mean approaches were also applied. The IV and group mean method, with groups based on job and factory, were also used to estimate the effect of exposure on FEV1 in phase 3 of the European carbon black respiratory health study.
ResultsThere was little evidence of bias in the IV method: on average b=−0.1025, regardless of r, with standard errors from 0.0141–0.0184 depending on r. Attenuation using the grouping methods varied with the number of groups and r. The greatest attenuation (b=−0.0678, SE 0.0103) was derived with the largest number of groups (n=128) and r=0.5 and the least biased when r=0.8 and there were two groups (b=−0.1016, SE 0.0140). Mixed IV/group mean methods improved estimates only slightly. Results from the carbon black study and the log‐normal distributed exposure simulations will also be presented.
ConclusionInstrumental variables provide unbiased estimates regardless of exposure reliability.
Key wordsexposure measurement error; grouping approach; instrumental variables