O-269 Age-at-exposure and time-since-exposure in causal inference: ionizing radiation and cancer mortality in INWORKS

Introduction

Ionizing radiation is an established carcinogen. Previously, the International Nuclear Workers Study (INWORKS), a pooled cohort of 308,297 workers from the United States (US), the United Kingdom (UK), and France followed 1944–2005, yielded an estimated excess relative rate of 0.51 per Gy (90% CI = 0.23, 0.82) for cancer mortality (19,748 deaths). Prior analyses suggested potential for healthy worker survivor bias and modification by age-at-exposure and time-since-exposure.

Material and Methods

We reanalyzed INWORKS data using the parametric g-formula, which can address healthy worker survivor bias. We estimated impacts of 2 hypothetical interventions: 1) reduce historical standards of worker equivalent doses to 5 mSv/year; and, 2) exposure to constant levels of 20 mGy/year vs. 5 mGy/year while at work. This approach requires models for exposure, employment and cause-specific mortality. Mortality models allow variation in the radiation-mortality association with age-at-exposure and time-since-exposure.

Results

Conditional on prior exposure and covariates, employment in the prior year was inversely associated with mortality among French workers but positively associated in US and UK workers. Associations between dose and cancer mortality were largest for doses received after age 45, lagged 30+ years. Approximately 224 cancer deaths per 1,000 workers were expected by age 90 in the pooled cohort. Relative to historical exposure levels (10% of working-years > 5 mGy/year), we estimated that a hypothetical standard at 5 mGy/year would result in 6.9 (95% CI = -5.7, 19) fewer cancer deaths per 1,000 workers by age 90. A constant exposure at 5 mGy/year (relative to 20 mGy/year) would have resulted in 13 (95% CI = -1.1, 27) fewer deaths.

Conclusion

Our results suggest that confounding by employment status was present, suggesting healthy worker survivor bias. The importance of doses at older ages and long times-since-exposure suggest continued need to assess potential impacts of occupational ionizing radiation exposures.