Study objectives: Numerous studies indicate that smoking is associated with poorer outcomes in patients with cancer. The aim of this study was to determine whether smoking independently predicts survival in patients with lung cancer or whether an existent effect is mediated through comorbidity and/or treatment.
Design and setting: Cox proportional hazards analysis was used to study a cohort of 1,155 patients with lung cancer diagnosed at the Henry Ford Health System between 1995 and 1998, inclusive.
Results: Adjusted for the baseline covariates, age, gender, illicit drug use, adverse symptoms, histology, and stage, the hazard ratio (HR) for smoking (current vs former/never) was 1.37 (95% confidence interval [CI], 1.18 to 1.59; p < 0.001). Adjusted for the baseline covariates and for 18 deleterious comorbidities, the HR for smoking was 1.38 (95% CI, 1.18 to 1.60; p < 0.001), indicating that the hazardous effect of smoking was not mediated through comorbidity. Current smoking was inversely associated with treatment (any surgery and/or chemotherapy and/or radiation therapy vs none) [odds ratio, 0.73; 95% CI, 0.55 to 0.98 (p = 0.03)]. Adjusted for baseline covariates, comorbidities and treatment, the HR for current smoker vs former/never was 1.26 (95% CI, 1.08 to 1.47; p = 0.003), a decline of 30.7% explained by treatment (HR for any treatment vs none, 0.40; 95% CI, 0.33 to 0.48; p < 0.001).
Conclusions: Current smoking at diagnosis is an important independent predictor of shortened lung cancer survival. That this effect was not explained by sociodemographic/exposure factors, adverse symptoms, histology, stage, comorbidity, and treatment suggests that it may be mediated through direct biological effects.