Objectives In occupational epidemiology, we often rely on analytical models that look at the relationship between one exposure and one disease. However, the exposure may be related to more than one outcome at the target site (e.g., pulmonary diseases: pneumoconiosis, byssinosis, chronic obstructive pulmonary disease, or lung cancer), and one of these diseases may influence the occurrence of another. Cox proportional hazards regression models using Kaplan-Meier (KM)-based estimates may not be appropriate due to violating the model’s non-informative censoring assumption.

Method An alternative approach is explored for occupational epidemiology: competing risks and multistate model. In multistate models, subjects may contribute at-risk person-time by transitioning to multiple, sometimes competing, states. Subjects can transition to a state in which they are living with or die from the related disease (i.e., the competing risk of the outcome of interest). These models use a subdistribution hazard, in which competing events are accounted for in the survival probability and allow for additional baseline hazards for different states.

Results Competing risks and multistate models may allow researchers to build models that better reflect the biological complexity of an exposure contributing to multiple, related disease pathways simultaneously. For example, the exposure-response for the outcome of interest may be different if arriving at that outcome first through a related comorbidity than without that comorbidity.

Conclusions Although these models may improve our estimation, there are barriers to their implementation, including misclassification of disease, that will be discussed.