The influence of study characteristics on the healthy worker effect: a multiple regression analysis.

The so-called 'healthy worker effect' (HWE) describes a reduced mortality rate in occupational populations. From 85 occupational cohorts of workers exposed to organic solvents, we have previously found a low weighted standardized mortality ratio (SMR) for all causes (SMR = 89.5, 95 per cent confidence interval [Cl] = 89.0-90.2). Characteristics of the cohorts were examined for association with the HWE. A multiple regression analysis revealed that, for overall deaths, the closer the comparison group, the smaller the HWE (p = 0.001); the more hazardous the potential exposures, the higher the SMR (p = 0.02); the higher the social class, the greater the HWE (p = 0.02); and the higher the rates of loss to follow up, the greater the HWE (p = 0.04). These results have shown that, in addition to time-related modifiers, the comparison group, type of occupational exposure, social class and rates of loss to follow-up significantly influence the size of the so-called HWE. Differences in the HWE related to gender and race were probably attributable to different rates of loss to follow-up.