Background: Researchers use lung cancer death rates (rates) as an index of the cumulative burdens of smoking. That index lacks direct validation and calibration. So this study directly validates and calibrates that index against annual approximately non-lung (all-sites minus lung and stomach) rates from 1969 to 2000 in United States black men, then estimates their cancer death rate smoking-attributable fractions (SAFs).
Methods: This study uses linear regression, age-adjusted rates from http://www.seer.cancer.gov/canques, and the formula SAF = (1- ((rate in the unexposed) / (rate in the exposed))). Estimated rates in the unexposed range between the 1969 rate and the rate predicted for a population with no smoking-attributable lung cancers. Stomach and lung cancer rate SAFs were based on published cohort studies.
Results: Lung cancer death rates predicted 98% and 97% of the variances in approximately non-lung cancer death rates throughout their 1969-1990 34% rise and subsequent declines, respectively (each P < 0.0001). The findings suggest that the SAF of the all-sites cancer death rate in black men peaked at 66% in 1990.
Conclusions: Lung cancer death rates were a good index of smoke exposure for predicting approximately non-lung cancer death rates in black men. Smoking may cause most premature cancer deaths in black men.