Table 2.

Regression coefficients from GEE model for associations of FEV₁ (milliliters) with endotoxin exposures among cotton workers evaluated at successive 5-year survey intervals over 25-year follow-up.^a

	All cotton workers	Actively employed	Retired
	Estimate (95% CI)	Estimate (95% CI)	Estimate (95% CI)
Years since cessation of exposure	6.02 (0.82 to 11.21)*	NA	14.53 (6.35 to 22.71)*
Past endotoxin exposure (high)b	16.72 (−42.04 to 75.49)	14.12 (−41.77 to 70.02)	−132.64 (−281.11 to 15.83)
Past endotoxin exposure (medium)b	53.82 (10.68 to 96.96)*	39.40 (1.04 to 77.77)*	−153.56 (−312.33 to 5.21)
Recent endotoxin exposure (1,000 EU/m3-years)c	0.77 (−0.50 to 2.04)	−0.07 (−1.51 to 1.37)	3.32 (−0.91 to 7.54)

NA, not applicable.

^aEstimated from GEE models using all available data from 447 cotton workers. All variables except sex and height are treated as time-dependent variables. Models were adjusted for age, mean height of the first three surveys, sex, and smoking status.

^bPast endotoxin exposure was defined as cumulative exposure to endotoxin (EU/m³-years) before the start of the survey interval in which FEV₁ was measured but was modeled with two binary variables comparing high with low and medium with low levels. Tertile percent cutoffs of past endotoxin exposure at each survey was used to defined low, medium, and high levels. For workers hired < 5 years before the baseline survey, past endotoxin exposure before 1981 was zero.

^cRecent endotoxin exposure was defined as cumulative exposure within the 5 years prior to each FEV₁ measurement and was modeled continuously; estimates indicate the change in FEV₁ level associated with a 1,000 EU/m³-years increase in endotoxin exposure.

^*p < 0.05.