Purpose of review: We offer an in-depth discussion of the time-varying confounding and selection bias mechanisms that give rise to the healthy worker survivor effect (HWSE).
Recent findings: In this update of an earlier review, we distinguish between the mechanisms collectively known as the HWSE and the statistical bias that can result. This discussion highlights the importance of identifying both the target parameter and the target population for any research question in occupational epidemiology. Target parameters can correspond to hypothetical workplace interventions; we explore whether these target parameters' true values reflect the etiologic effect of an exposure on an outcome or the potential impact of enforcing an exposure limit in a more realistic setting. If a cohort includes workers hired before the start of follow-up, HWSE mechanisms can limit the transportability of the estimates to other target populations. We summarize recent publications that applied g-methods to control for the HWSE, focusing on their target parameters, target populations, and hypothetical interventions.
Keywords: G-methods; Healthy worker survivor bias; Occupational epidemiology; Selection bias; Time-varying confounding.