- Email: [email protected]
Respiratory impairment due to asbestos exposure in brake-lining workers
ARTICLE IN PRESS Letters to the Editor / Environmental Research 96 (2004) 102–108
Fish is a very healthy food and a dietary mainstay of this population; we hope that these studies will serve to further our knowledge on how to maximize the nutritional input from traditional food consumption and minimize toxic risk.
References Passos, C.J., Mergler, D., Gaspar, E., Morais, S., Lucotte, M., Larribe, F., Davidson, R., de Grosbois, S., 2003a. Eating tropical fruit reduces mercury exposure from fish consumption in the Brazilian Amazon. Environ. Res. 93, 123–130. Passos, C.J., Mergler, D., Gaspar, E., Morais, S., Lucotte, M., Larribe, F., de Grosbois, S., 2003b. Caracteriza@a˜o geral do regime alimentar de uma popula@a˜o ribeirinha na amazo˜nia basileira (General characterization of the diet of a riverside population in the Brazilian Amazon). Sau´de Meio Ambiente 4, 72–84. Morrissette, J., Takser, L., St-Amour, G., Smargiassi, A., Lafond, J., Mergler, D., 2004. Temporal variation of blood and hair mercury
105
levels in pregnancy in relation to fish consumption history in a population living along the St. Lawrence River. Environ. Res., in press. Clarkson, T.W., Strain, J.J., 2003. Nutritional factors may modify the toxic action of methyl mercury in fish-eating populations. J. Nutr. 133, 1539S–1543S.
Carlos Passos, Donna Mergler Insitute for Environmental Sciences, CINBIOSE University of Quebec in Montreal, CP 8888 succ Centreville, Montreal, Que., Canada H3C 3P8 E-mail address: [email protected] Fabrice Larribe Department of Human Genetics McGill University, Canada
doi:10.1016/j.envres.2004.04.003
Respiratory impairment due to asbestos exposure in brake-lining workers The stated purpose of the article by Erdin@ et al. (2003) was ‘‘to assess the long-term effects of chrysotile asbestos exposure on lung function and the risk of asbestos-related diseases in brake-lining production workers.’’ The authors concluded that ‘‘even in the absence of radiographic asbestosis, TL,CO and KCO may decrease after a mean 10-year duration of exposure to asbestos in brake-lining workers and this is more noticeable with cigarette burden.’’ However, because of flaws in the study design, execution, and analyses, these conclusions are not supported by the study. This was a prospective study among active workers employed by a single brake-lining production facility. Subjects completed pulmonary function tests, specifically spirometry and diffusion capacity, in 1992 and 1999. Exposed workers were employed processing asbestos brake linings; unexposed workers were employed in offices. The asbestos-exposed group was divided based on smoking status: current smokers (n ¼ 49) and nonsmokers (n ¼ 25). All unexposed workers were current smokers (n ¼ 12). There are a number of features of the study group that are unexplained and problematic. First, there were no exsmokers among the 86 subjects. Ex-smokers usually comprise a large fraction of working populations, which
suggests that a substantial number of workers were excluded from the study. Second, the number of unexposed workers was too small for statistical analyses. Third, there were no nonsmokers among the unexposed subjects, which would limit the ability to measure the independent effect of smoking on pulmonary function among subjects. Fourth, there was no mention of workers who completed testing in 1992, but who did not complete testing in 1999. It is important to know whether workers who completed follow-up testing differed at baseline from subjects who did not complete the second round of testing. These flaws in selection and/or description of the study population suggest the possibility of substantial bias. There are a number of problems with the pulmonary function tests. It was stated that ‘‘nobody had smoked at least 3 h before the testing procedure.’’ The half-life of carbon monoxide in blood while breathing room air is approximately 4 h (Rom, 1998). Since testing for diffusion capacity (i.e., transfer factor) was performed as little as 3 h after smoking, or less than one half-life, measurements in smokers were likely biased (i.e., they would appear more impaired) due to carbon monoxide backpressure. Though the authors acknowledged the possibility of a problem with carbon monoxide back-
ARTICLE IN PRESS 106
Letters to the Editor / Environmental Research 96 (2004) 102–108
pressure in diffusion measurements among smokers, they nevertheless interpret their findings as being due to asbestos. This conclusion has no scientific basis. The dramatic decline in lung function, particularly diffusion capacity, among study groups over the 7 years of follow-up was extraordinary, and suggests the presence of large, systematic measurement error. The mean change in percentage predicted (not absolute) diffusion capacity (TL,CO) was 20% among nonasbestos exposed smokers, or almost 3% per year (see Table 5). The mean age of the control group was 35.75 years at the time of follow-up, and the duration of follow-up was 7 years (see Table 1). The expected change in mean absolute diffusion capacity related to aging 7 years is 1.603 ml/mmHg/min, or only 5.4% of the baseline (less than 1% per year), based on the prediction equation for healthy male smokers published by Miller et al. (1983). If the controls were healthy male smokers, then the expected change in mean percentage predicted values should be negligible. The 20% reduction in mean percentage predicted diffusion capacity among the control group reported by Erdin@ et al. indicates that the diffusion measurements were flawed. The subjects were all active workers, so latency was the same as duration of employment, which meant that the overall mean latency of asbestos exposure was approximately 10.7 years. With such short mean latency, it would be exceedingly unlikely to observe any measurable impact of asbestos exposure on pulmonary function. The reported magnitude of loss of diffusion capacity, particularly for such a young population (i.e., mean ages less than or equal to 37 years) with such short latency is not credible, and suggests that there was something wrong with the testing procedure. The key hypothesis of the study was whether asbestos exposure among workers in a brake-lining production facility leads to an accelerated decline in lung function when adjusted for age, smoking, and possibly other covariates (e.g., other medical conditions). Remarkably,
the authors presented no statistical analyses that addressed the hypothesis of the study. In particular, there were no multivariate statistical models of change in lung function in relation to asbestos exposure with adjustment for smoking, age, and possibly other covariates (e.g., other medical conditions or obesity). The data presented in the five tables were merely descriptive statistics of the study groups with no statistical comparisons presented, not even univariate comparisons. The authors provide no basis for drawing any conclusions about the impact of asbestos exposure on lung function among workers in this study. In summary, the study population was poorly defined and suffered from unknown selection biases, the pulmonary function testing was flawed, and the statistical analyses failed to address the hypothesis of the study. For these reasons, no valid scientific conclusions can be drawn from this investigation.
References Erdin@, M., Erdin@, E., C - ok, G., Polatli, M., 2003. Respiratory impairment due to asbestos exposure in brake-lining workers. Environ. Res. 91, 151–156. Miller, A., Thornton, J.C., Warshaw, R., Anderson, H., Teirstein, A.S., Selikoff, I.J., 1983. Single breath diffusing capacity in a representative sample of the population of Michigan, a large industrial state. Am. Rev. Respir. Dis. 127, 270–277. Rom, W.N. (Ed.), 1998. Environmental & Occupational Medicine, 3rd Edition. Lippincott-Raven, Philadelphia, 659pp.
Alfred Franzblau, David H. Garabrant Enviromental Health Sciences, School of Public Health University of Michigan, 1420 Washington Heights, Rm 6023, Ann Arbor, MI 48109-2029, USA E-mail address: [email protected]
doi:10.1016/j.envres.2004.04.005
Response to ‘‘respiratory impairment due to asbestos exposure in brake-lining workers’’ We are pleased to answer the questions Franzblau and Garabrant raised about our article. This gives us the opportunity to provide more information regarding our study (Erdin@ et al., 2003).
Although pulmonary function tests were performed in 249 workers and 58 unexposed office workers in 1992, during the surveillance only 74 asbestos workers and 12 unexposed office workers who were previously examined
Fish is a very healthy food and a dietary mainstay of this population; we hope that these studies will serve to further our knowledge on how to maximize the nutritional input from traditional food consumption and minimize toxic risk.
References Passos, C.J., Mergler, D., Gaspar, E., Morais, S., Lucotte, M., Larribe, F., Davidson, R., de Grosbois, S., 2003a. Eating tropical fruit reduces mercury exposure from fish consumption in the Brazilian Amazon. Environ. Res. 93, 123–130. Passos, C.J., Mergler, D., Gaspar, E., Morais, S., Lucotte, M., Larribe, F., de Grosbois, S., 2003b. Caracteriza@a˜o geral do regime alimentar de uma popula@a˜o ribeirinha na amazo˜nia basileira (General characterization of the diet of a riverside population in the Brazilian Amazon). Sau´de Meio Ambiente 4, 72–84. Morrissette, J., Takser, L., St-Amour, G., Smargiassi, A., Lafond, J., Mergler, D., 2004. Temporal variation of blood and hair mercury
105
levels in pregnancy in relation to fish consumption history in a population living along the St. Lawrence River. Environ. Res., in press. Clarkson, T.W., Strain, J.J., 2003. Nutritional factors may modify the toxic action of methyl mercury in fish-eating populations. J. Nutr. 133, 1539S–1543S.
Carlos Passos, Donna Mergler Insitute for Environmental Sciences, CINBIOSE University of Quebec in Montreal, CP 8888 succ Centreville, Montreal, Que., Canada H3C 3P8 E-mail address: [email protected] Fabrice Larribe Department of Human Genetics McGill University, Canada
doi:10.1016/j.envres.2004.04.003
Respiratory impairment due to asbestos exposure in brake-lining workers The stated purpose of the article by Erdin@ et al. (2003) was ‘‘to assess the long-term effects of chrysotile asbestos exposure on lung function and the risk of asbestos-related diseases in brake-lining production workers.’’ The authors concluded that ‘‘even in the absence of radiographic asbestosis, TL,CO and KCO may decrease after a mean 10-year duration of exposure to asbestos in brake-lining workers and this is more noticeable with cigarette burden.’’ However, because of flaws in the study design, execution, and analyses, these conclusions are not supported by the study. This was a prospective study among active workers employed by a single brake-lining production facility. Subjects completed pulmonary function tests, specifically spirometry and diffusion capacity, in 1992 and 1999. Exposed workers were employed processing asbestos brake linings; unexposed workers were employed in offices. The asbestos-exposed group was divided based on smoking status: current smokers (n ¼ 49) and nonsmokers (n ¼ 25). All unexposed workers were current smokers (n ¼ 12). There are a number of features of the study group that are unexplained and problematic. First, there were no exsmokers among the 86 subjects. Ex-smokers usually comprise a large fraction of working populations, which
suggests that a substantial number of workers were excluded from the study. Second, the number of unexposed workers was too small for statistical analyses. Third, there were no nonsmokers among the unexposed subjects, which would limit the ability to measure the independent effect of smoking on pulmonary function among subjects. Fourth, there was no mention of workers who completed testing in 1992, but who did not complete testing in 1999. It is important to know whether workers who completed follow-up testing differed at baseline from subjects who did not complete the second round of testing. These flaws in selection and/or description of the study population suggest the possibility of substantial bias. There are a number of problems with the pulmonary function tests. It was stated that ‘‘nobody had smoked at least 3 h before the testing procedure.’’ The half-life of carbon monoxide in blood while breathing room air is approximately 4 h (Rom, 1998). Since testing for diffusion capacity (i.e., transfer factor) was performed as little as 3 h after smoking, or less than one half-life, measurements in smokers were likely biased (i.e., they would appear more impaired) due to carbon monoxide backpressure. Though the authors acknowledged the possibility of a problem with carbon monoxide back-
ARTICLE IN PRESS 106
Letters to the Editor / Environmental Research 96 (2004) 102–108
pressure in diffusion measurements among smokers, they nevertheless interpret their findings as being due to asbestos. This conclusion has no scientific basis. The dramatic decline in lung function, particularly diffusion capacity, among study groups over the 7 years of follow-up was extraordinary, and suggests the presence of large, systematic measurement error. The mean change in percentage predicted (not absolute) diffusion capacity (TL,CO) was 20% among nonasbestos exposed smokers, or almost 3% per year (see Table 5). The mean age of the control group was 35.75 years at the time of follow-up, and the duration of follow-up was 7 years (see Table 1). The expected change in mean absolute diffusion capacity related to aging 7 years is 1.603 ml/mmHg/min, or only 5.4% of the baseline (less than 1% per year), based on the prediction equation for healthy male smokers published by Miller et al. (1983). If the controls were healthy male smokers, then the expected change in mean percentage predicted values should be negligible. The 20% reduction in mean percentage predicted diffusion capacity among the control group reported by Erdin@ et al. indicates that the diffusion measurements were flawed. The subjects were all active workers, so latency was the same as duration of employment, which meant that the overall mean latency of asbestos exposure was approximately 10.7 years. With such short mean latency, it would be exceedingly unlikely to observe any measurable impact of asbestos exposure on pulmonary function. The reported magnitude of loss of diffusion capacity, particularly for such a young population (i.e., mean ages less than or equal to 37 years) with such short latency is not credible, and suggests that there was something wrong with the testing procedure. The key hypothesis of the study was whether asbestos exposure among workers in a brake-lining production facility leads to an accelerated decline in lung function when adjusted for age, smoking, and possibly other covariates (e.g., other medical conditions). Remarkably,
the authors presented no statistical analyses that addressed the hypothesis of the study. In particular, there were no multivariate statistical models of change in lung function in relation to asbestos exposure with adjustment for smoking, age, and possibly other covariates (e.g., other medical conditions or obesity). The data presented in the five tables were merely descriptive statistics of the study groups with no statistical comparisons presented, not even univariate comparisons. The authors provide no basis for drawing any conclusions about the impact of asbestos exposure on lung function among workers in this study. In summary, the study population was poorly defined and suffered from unknown selection biases, the pulmonary function testing was flawed, and the statistical analyses failed to address the hypothesis of the study. For these reasons, no valid scientific conclusions can be drawn from this investigation.
References Erdin@, M., Erdin@, E., C - ok, G., Polatli, M., 2003. Respiratory impairment due to asbestos exposure in brake-lining workers. Environ. Res. 91, 151–156. Miller, A., Thornton, J.C., Warshaw, R., Anderson, H., Teirstein, A.S., Selikoff, I.J., 1983. Single breath diffusing capacity in a representative sample of the population of Michigan, a large industrial state. Am. Rev. Respir. Dis. 127, 270–277. Rom, W.N. (Ed.), 1998. Environmental & Occupational Medicine, 3rd Edition. Lippincott-Raven, Philadelphia, 659pp.
Alfred Franzblau, David H. Garabrant Enviromental Health Sciences, School of Public Health University of Michigan, 1420 Washington Heights, Rm 6023, Ann Arbor, MI 48109-2029, USA E-mail address: [email protected]
doi:10.1016/j.envres.2004.04.005
Response to ‘‘respiratory impairment due to asbestos exposure in brake-lining workers’’ We are pleased to answer the questions Franzblau and Garabrant raised about our article. This gives us the opportunity to provide more information regarding our study (Erdin@ et al., 2003).
Although pulmonary function tests were performed in 249 workers and 58 unexposed office workers in 1992, during the surveillance only 74 asbestos workers and 12 unexposed office workers who were previously examined