Assessing the Internal Healthy Worker Effect Using Survival Analysis

• Activity Number: 342
• Type: Contributed
• Date/Time: Wednesday, August 6, 2003 : 9:00 AM to 10:50 AM
• Sponsor: Section on Health Policy Statistics
• Author(s): Jacques Baillargeon*+ and Gwen Baillargeon
• Companies: University of Texas and University of Texas
• Keywords: occupational epidemiology ; public health ; selection bias ; survival analysis ; healthy worker effect ; epidemiologic methods

Abstract

The healthy worker effect (HWE) is a form of selection bias that occurs when an occupational cohort is compared with the general population. To avoid this problem, investigators often employ a method of internal comparisons. This consists of making mortality/disease comparisons of employees from the same work site across different levels of a given variable (e.g. cumulative exposure). Internal comparisons can be conducted via directly adjusted rates or forms of multivariate analysis such as Cox's proportional hazards modeling. Internal comparisons on the basis of exposure are particularly informative when the occupational cohort is sufficiently large and is relatively comparable across levels of the study variable. Internal comparisons, however, are not entirely free from the biases related to the HWE. If employees are selected on the basis of factors associated with health into subgroups that serve as the basis for comparison (e.g. exposure) then a form of internal comparison bias may occur. The present study, relying on Cox's proportional hazards modeling, reported substantial selection bias according to race, age at hire, occupational class and duration of employment.