Injury death excesses in smokers: a 1990-95 United States national cohort study.

OBJECTIVES Assess injury death relative risks (RR), dose-response, and attributable fractions for current cigarette smokers (smokers) in a recent representative sample of the United States population without and with adjustment for (a) demographic and (b) additional behavioral risk factors. SETTING United States. METHODS National Health Interview Survey (NHIS) adult (ages 18+ years) interviewees from 1990 or 1991 were followed through 1995. Referents had never smoked a total of 100 cigarettes. Relative risks were estimated with Stata software's Cox proportional hazard regressions, using NHIS final weights and primary sampling units. The resulting RR and published data were used to estimate population smoking attributable fractions of injury deaths in the United States. RESULTS The crude, age-race-gender adjusted, and fully (demographic plus educational attainment, marital status, alcohol use level, and seat belt use) adjusted RRs for injury death in smokers were 1.86 (95% confidence interval (CI) 1.30 to 2.66), 1.60 (CI 1.12 to 2.29), and 1.42 (CI 0.99 to 2.05) respectively. Those RRs correspond to United States injury death smoking attributable fractions of 18%, 13%, and 9%, respectively. Those smoker/ injury death RRs each showed a significant dose response relationship (p<0.030). Smokers' unadjusted unintentional injury, motor vehicle crash, and suicide RRs were 1.87 (CI 1.22 to 2.86), 2.14 (CI 1.12 to 4.11), and 2.17 (CI 1.02 to 4.62), respectively. CONCLUSIONS Smokers in the United States have significant dose-response excesses of injury death, independent of age, race, gender, alcohol use, seat belt use, education, and marital status. This supports earlier studies suggesting that smoking may be a leading contributor to injuries and injury may be a leading burden from smoking, both nationally and globally.