NEGATIVE BIAS IN EXPOSURE-RESPONSE TRENDS IN OCCUPATIONAL STUDIES - MODELING THE HEALTHY WORKER SURVIVOR EFFECT

Many occupational studies analyze trends between cumulative exposure a nd mortality, The authors show that such trends are, in general, negat ively confounded by employment status, Mortality rates for workers who leave work (''inactive'' workers) are higher than for active workers because some workers leave because they are ill. The percentage of ina ctive relative to active person-time is higher in low categories of cu mulative exposure, causing employment status to act as a negative conf ounder of exposure-response trends (the opposite occurs for time-since -hire). We illustrate these phenomena using 10 ''negative'' mortality studies, in which adjustment for employment status removes false trend s, However, adjustment for employment status will lead to biased estim ates when it acts as an intermediate variable between cumulative expos ure and death, as occurs directly when exposure causes a disabling dis ease that, in turn, causes death or indirectly when exposure causes wo rkers to leave work, The authors illustrate this problem using simulat ed follow-up data for leaving, disease incidence, and mortality, In th e null case in which cumulative exposure affects neither disease incid ence (or mortality) nor leaving rates, employment status indeed acts a s a negative confounder of exposure-response trends, and traditional a djustment eliminates this confounding. However, when cumulative exposu re affects disease incidence or rates of leaving, adjustment for emplo yment status will not be adequate, Employment status falls under the g eneral rubric of variables that are simultaneously confounders and int ermediate variables.