Background:
To evaluate the relationship between long-term exposure to cotton dust and Gram-negative bacterial endotoxin on lung function, we conducted an 11-year follow-up study of cotton textile workers in Shanghai, China. Methods: Workers at a nearby silk-thread manufacturing mill were used as a referent population. Ninety percent of the original cohort of 445 cotton and 467 silk textile workersboth active and retired -were identified for testing in the 11th year. Questionnaires and spirometric testing were performed, as well as cotton dust and endotoxin sampling at three points over the 11-year follow-up period: at baseline, at Year 5, and at Year 11. After excluding deaths and subjects on sick-leave, 84% of the original cohort had complete health and environmental data. Results: The data were reanalyzed using generalized estimating equations feedback model which allow for subject transfer over time between work areas, various exposure levels to dust and endotoxin, and FEV 1 . Cotton workers had a larger loss of FEV 1 during the first 5 years of study (-40 mls/yr) as compared with the second 6 years of follow-up (-18 mls/yr). During the same periods, the average decline among silk workers was slightly higher in the first period, but was more consistent (-30 mls/yr vs. -27 mls/yr), and these differences could not be explained by worker selection or dropout. When cumulative exposure to dust and endotoxin were estimated and used in a multivariate model (GEE) for FEV 1 loss, cumulative dust, but not endotoxin, was associated with 11-year loss in FEV 1 after adjustments for confounders. There was evidence of feedback between dust-exposure levels and FEV 1 , indicating the existence of a healthy-worker survivor effect. After accounting for a healthy-worker survivor effect, we found a significant relationship between dust exposure and FEV 1 decline. Conclusions: Our results suggest that cotton dust is more strongly associated with chronic airflow limitation than associated endotoxins. Further work is needed to clarify potential reversibility after cessation of exposure, and the relative contributions of dust, endotoxin, and