Cancer Rates Not Explained by Smoking: A County-Level Analysis

Background: At least 12 types of cancer are caused by cigarette smoking, and smoking cessation is the most effective known prevention strategy. Environmental and occupational carcinogens receive less emphasis in prevention messaging which typically focuses on reducing smoking and other “lifestyle” factors. To inform decision making on priorities in cancer prevention, we evaluated the potential impact of smoking cessation on cancer rates. Methods: Using cancer incidence data from 612 counties in the SEER database, and county-level smoking prevalences, we investigated the impact of smoking cessation on incidence for 12 smoking-related cancer types, 2006—2016. A multilevel mixed-effects regression model quantified the association between county-level smoking prevalence and cancer incidence, adjusting for age, gender and variability over time and among counties. We simulated complete smoking cessation and estimated the effects on county-level cancer rates. Findings: Assuming complete smoking cessation, the overall incidence of cancers for the 12 types fell by 39·8%. And, while the actual smoking-related incidence rates from 2006-2016 showed a modest but measurable decline (annual percent change (APC) = -0·8%, 95% CI = -1·0 to -0·5%), after smoking elimination, the trend in the remaining 60% of cancers was stable (APC = -0·1%, 95% CI = -0·4 to +0·1%). Not all counties were predicted to benefit equally from smoking elimination, and cancer rates would fall less than 10% in some counties. Interpretation: About 60% of tumors of the12 smoking-related cancer types may not be preventable by smoking cessation alone, and apart from declines in tobacco consumption, there is little evidence of improvement in the incidence trend over time. Comprehensive cancer prevention should address the full range of modifiable causes, including not only personal behaviors, but also environmental and occupational carcinogens. Funding Statement: Dr. Hoppin’s and Ms. Jacobs' work was supported in part by a grant from the Heinz Endowments. Declaration of Interests: The authors declare no competing interests. Ethics Approval Statement: Because all data were publicly available and not personally identifiable, Institutional Review Board approval was not required.