Might smoking cost 3 billion lives and quadrillions of dollars this century?

Background. Officially, smoking cost 440,000 US lives in 2004 and may cost 1 billion lives globally this century. Yet new, stronger cohort and time-series evidence suggests that much US and global smoking-attributable mortality (SAM) has been overlooked. So I re-evaluated SAM globally.

Methods. I assessed cigarette/lung cancer, lung cancer/all cause, and cigarette/all cause mortality associations using Cancer Prevention Study-II (CPS2) cigarette strata on which present official SAM estimates are based, CPS2 education strata with more reliable lung cancer rates, and the methodologically improved Nurses Health Study (Nurses). I then assessed: 1) US time-series and CPS2 lung cancer/all cause mortality associations by gender and 2) national and global SAM where SAM= 1-unexposeds' death rate/observed national death rate.

Results. Cigarette/lung cancer, lung cancer/all cause, and cigarette/all cause mortality associations across cigarettes/day strata were of mostly borderline statistical significance in CPS2 yet very strong (p<0.000, correlation coefficient (r)>0.99) in Nurses. Those strong Nurse lung cancer/all cause mortality associations (slope 0.081 (95% confidence interval (CI) 0.08-0.09, r=1.00) were corroborated by equally strong associations at ages 65-69 years: across education strata in CPS2 males (slope 5.12 (CI 4.44-5.81); and in the US general population from 1992-2005. Those lung cancer/all cause mortality associations suggest 2001 US ages 35+ years SAM rate fractions of 68% (SR 57-73%) in males and 55% (SR 0-100%) in females, about thrice official estimates.

Conclusions. Lung cancer/all cause mortality associations across Nurses, CPS2 education strata, and recent years suggest far higher national and global SAM fractions and costs than are officially reported.