Biases in estimating the effect of cumulative exposure in log-linear models when estimated exposure levels are assigned

Objectives Exposure-response trends in occupational studies of chronic dise ase are often modeled via log-linear models with cumulative exposure as the metric of interest. Exposure levels for most subjects are often unknown, b ut can be estimated by assigning known job-specific mean exposure levels fr om a sample of workers to all workers. Such assignment results in (nondiffe rential) measurement error of the Berkson type, which does not bias the est imate of exposure effect in linear models but can result in substantial bia s in log-linear models with dichotomous outcomes. This bias was explored in estimated exposure-response trends using cumulative exposure. Methods simulations were conducted under the assumptions that (i) exposure level is assigned to all workers based on the job-specific means from a sam ple of workers, (ii) exposure level and duration are log-normal, (iii) the true exposure-response model is log-linear for cumulative exposure, (iv) th e disease is rare, and (v) the variance of job-specific exposure level incr eases with its job-specific mean. Results Assignment of job-specific mean exposure levels from a sample of wo rkers causes an upward bias in the estimated exposure-response trend when t here is little variance in the duration of exposure but causes a downward b ias when duration has a large variance. This bias can be substantial (eg, 3 0-50%). Conclusions Berkson errors in exposure result in little bias in estimating exposure-response trends when the standard deviation of duration is approxi mately equal to its mean, which is common in many occupational studies. No bias occurs when the variance of exposure level is constant across jobs, bu t such conditions are probably uncommon.