Meta-analysis of disease risk associated with smoking, by gender and intensity of smoking.

BACKGROUND: The risks associated with cigarette smoking can be substantial, particularly for females. In 2000, the mortality rate for lung cancer among women was higher than that for breast cancer. OBJECTIVE: To obtain overall risk for intensity of smoking for both males and females, a meta-analysis was performed on recent studies that assessed the morbidity and mortality associated with smoking. METHODS: Using the PubMed database, a literature search was conducted for cohort and case-control studies on the effect of smoking on morbidity and mortality. Only studies that had quantified the risk of disease associated with smoking were included. Nineteen studies were selected, with data obtained on the disease affected by smoking, point estimates of risk, 95% CIs, sample size, type of study, and the number of patients of each sex. Meta-analyses were performed for low level of use, defined as 1 to 20 cigarettes per day, and for high level of use, >20 cigarettes per day. RESULTS: For low level of use, the rate ratio point estimate of 1.77 (95% CI, 1.40-2.24) for females was higher than that of 1.42 (95% CI, 1.23-1.64) for males, indicating a gender effect associated with smoking as a disease risk. The point estimate for females who smoked at high levels was 2.75 (95% CI, 2.14-3.52), well beyond the estimate of 1.95 (95% CI, 1.70-2.24) for males, indicating there was a substantial gender effect with high-level use. All point estimates for low and high levels of smoking were significant; those for each sex at high levels of smoking exceeded those found for low levels. The increase in risk from low to high levels of smoking was greater for females than for males. CONCLUSIONS: Few systems in the body were unaffected by smoking, and intensity was a risk factor for disease. Results were consistent with and strengthened previous research demonstrating an increase in overall risk with an increase in smoking intensity. In addition, gender differences were noted that may contribute to risk magnitude.