부산대학교 도서관

Background: Tobacco smoking caused an estimated 100 million deaths in the 20th century, and unless widespread cessation and prevention occur, it could cause one billion deaths in the 21st century. An estimated 50 million smokers live in Latin America, but there is little direct evidence of the effects of smoking in Latin American populations. Methods: Information from three large, prospective studies of Latin American populations was used to determine the relevance of smoking and smoking cessation to mortality. In 1996-2002, 146 556 Cuban adults aged ≥30 years were recruited into the Cuba Prospective Study, of whom 8691 died at ages 30-69 during follow-up. In 1998-2004, 159 755 Mexican adults aged ≥35 years were recruited into the Mexico City Prospective Study, of whom 19 091 died at ages 35-89 during follow-up (10 773 without major chronic disease at recruitment, and 8318 with major chronic disease). Finally, in 1997-2014, 446 531 adults aged ≥30 years were recruited into the US National Health Interview Survey, of whom 48 440 died at ages 30-89 during follow-up (3242 Mexican Americans, 2177 other Hispanics, 34 482 non-Hispanic Whites, 6978 non-Hispanic Blacks, and 1561 participants of other race/ethnicity). In each prospective study, Cox regression (adjusting for age-at-risk, sex, education, area, and alcohol consumption) was used to relate smoking habits at recruitment to mortality rate ratios (RRs). Resurveys of Cuban and Mexican participants informed 'usual' smoking behaviours. Results: In the Cuba Prospective Study, half of men and one-quarter of women were smokers at recruitment. One-third of current cigarette smokers reported having started smoking regularly in childhood (age < 15 years). Compared with never smokers, all-cause mortality RRs at ages 30-69 for Cuban cigarette smokers (mean 15 cigarettes/day) who had started at ages 5-9, 10-14, 15-19, and ≥20 years were 2.51 (95%CI 2.21-2.85), 1.83 (1.72-1.95), 1.56 (1.46-1.65) and 1.50 (1.39-1.62), respectively (overall RR for current smokers: 1.66 [1.59-1.74]). In the Mexico City Prospective study, half of men and one-quarter of women were smokers at recruitment, and one-fifth of smokers began smoking before age 15, but consumption was low (8 cigarettes/day among daily smokers). Among Mexican adults without chronic disease at recruitment, compared with never smokers, the all-cause mortality RR at ages 35-69 for smoking ≥10 cigarettes/day was 1.52 (1.36-1.70), and was similar at ages 70-89 and among men and women (overall RR for daily smokers: 1.27 [1.19-1.35]). In the US National Health Interview Survey prospective study, one-fifth of smokers began smoking before age 15, which was associated with three times the mortality rate at ages 30-69 of never smoking, with even greater risk among those starting to smoke before age 10. Compared with never smokers, the all-cause mortality RR for daily smokers was 1.39 (1.24-1.54) among Mexican Americans, 2.45 (2.20-2.72) among other Hispanics, 2.74 (2.61-2.89) among non-Hispanic Whites, 2.29 (2.19-2.41) among non-Hispanic Blacks, and 2.08 (1.84-2.35) among participants of other races. In each of these populations, smoking cessation by age ~40 avoided most of the excess risk associated with continued smoking. Conclusions: In Cubans, Mexicans, and Mexican Americans, the excess mortality risk associated with smoking was substantial, but was lower than the risk associated with smoking among Western populations. This may be due to a combination of intermittent smoking patterns in each of these populations, lower levels of consumption among Mexicans and Mexican Americans, and variability in products smoked over time. This thesis provides the first reliable, direct evidence of the relationship between smoking and mortality in Cuba and Mexico, which led to estimates of smoking-attributable mortality similar to those from the Global Burden of Disease (GBD) Study. Therefore, the GBD Study may provide the best available estimates of smoking-attributable mortality in other Latin American populations in which there is not yet direct evidence.