Estimation of healthy worker survivor bias among middle-aged populations in Korea

Abstract

Background Healthy worker survivor bias (HWSB) skews health outcome studies by favouring healthier employed individuals. While advanced statistical methods exist, their application in Korea has been limited due to insufficient occupational and mortality data. This study quantifies HWSB due to employment status changes (HWSB-ES) using Korea's National Health Insurance Service (NHIS) database.Methods This retrospective cohort study analysed NHIS data to assess HWSB-ES in individuals aged 30-59 years who maintained consistent insurance types from 2008 to 2010. The primary outcome, all-cause mortality, was tracked until December 2022. Insurance type determined employment status, with industry details collected for employees. Landmark analysis (origin: 2011; current: 2012-21) estimated HWSB-ES by assessing mortality risk attenuation in fixed and dynamic cohorts, stratified by age, sex, and landmark periods (1-10 years for short-term; 1-7 years for long-term).Results After exclusions, 18 192 989 participants were included (median age: 44 years; 49.05% male). HWSB-ES was more pronounced in female, dynamic cohorts, and longer landmark periods. Importantly, the effect of HWSB-ES intensified with age but showed a smaller long-term attenuation compared to the short-term effect. Short-term HWSB-ES attenuated mortality risk by 25%-30% in male and 36%-39% in female. Long-term attenuation was lower, at 7%-15% in male and 12%-18% in female.Conclusions The quantified HWSB-ES results provide critical national-level estimates for adjustment, especially in female and older cohorts, to prevent the underestimation of adverse health effects in occupational research.