Abstract
PURPOSE: Lung cancer mortality among never-smokers is an often overlooked yet important cause of adult mortality. Moreover, indirect approaches for estimating smoking-attributable mortality use never-smoker lung cancer death rates to approximate smoking burden. To date, though, most studies using indirect approaches import rates from the Cancer Prevention Study II (CPS-II), which is not representative of the U.S. METHODS: We use the nationally representative 1985-2015 National Health Interview Survey-Linked Mortality Files (NHIS-LMF) to calculate lung cancer death rates among never-smokers aged 50 years or older. We then import rates from NHIS-LMF and CPS-II into the Preston-Glei-Wilmoth indirect method to determine whether smoking-attributable fractions differ. RESULTS: Never-smokers account for 16% of U.S. lung cancer deaths among women and 11% among men. Lung cancer death rates among never-smokers are higher in NHIS-LMF than CPS-II for several age groups. Smoking-attributable fractions of mortality are slightly lower with NHIS-LMF rates (19% of male deaths and 16% of female deaths) than with CPS-II rates (21% of male deaths and 17% of female deaths). CONCLUSIONS: Fractions based on nonrepresentative CPS-II data may modestly overestimate smoking-attributable mortality. Thus, indirect methods should use never-smoker lung cancer death rates from such nationally representative datasets as NHIS-LMF.