Purpose. We assessed the smoking/injury death association in a nationally representative cohort, with simultaneous adjustment for multiple potential demographic and behavioral confounders.
Methods. National Health Interview Survey (NHIS) interviewees from 1990-91 were followed-up through 1995. We used Stata software, Cox proportional hazards regression, primary sampling units, and final weights to perform the analysis. We stratified for gender and adjusted for demographic factors (5-year age-groups and race) and behavioral factors (education, marital status, seatbelt use, and alcohol use).
Results. The crude injury death hazard ratio (HR) in current versus non-smokers was 1.9 (95% confidence interval (CI) 1.3-2.7). The HR were 1.8 (CI 1.3-2.6) and 1.5 (CI 1.0-2.2) with adjustment for demographic and demographic plus behavioral factors, respectively. Dose-response trends were noted (p<.02). For smokers of 25+ cigarettes/day, the HR was 2.7 (CI 1.6-4.4) with no adjustments and 2.6 (CI 1.5 -4.3) and 2.0 (1.1 - 3.5) independent of demographic or demographic plus behavioral factors.
Conclusions. The smoking/injury death association is: present in a nationally representative sample; independent of numerous potential confounders; and strong in heavy smokers. This supports existing animal model, challenge-rechallenge, cohort, cotwin control, and randomized trial data suggesting that smoking likely contributes to 20,000 US and 1 million global injury deaths annually.
Learning Objectives: Quantify the statistical significance of smoking/injury associations Quantify the public health significance of smoking/injury associations Quantify the statistical significance of dose-response in smoking/injury associations
Keywords: Smoking, Injury
Presenting author's disclosure statement:
Organization/institution whose products or services will be discussed: None
I do not have any significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.