Table 1.
Baseline characteristics of NHS (1976) and HPFS (1986) cohorts in model building and validation datasets.
Table 2.
Hazard ratio (95% CI) of variables associated with COPD risk in the model building data.
Fig 1.
Incidence of diagnosed COPD per 100,000 for females and males by smoking status.
The solid line is the expected incidence of diagnosed COPD from the model, and the dashed lines are its 95%CI. The dots represent the observed data.
Fig 2.
Discriminatory accuracy of models.
Receiver operating characteristic curve (ROC) and the corresponding area under the receiver operating characteristic curve (AUC) comparisons between the combined model and the pack-years only model [AUC (95% CI)]. * Significantly better than pack-years only model at p<0.05.
Fig 3.
Examples for age-specific incidence rates and relative risks of COPD.
(a) and (b) show the age-specific COPD incidence rates (per 100,000) among never smokers and current smokers. (c) and (d) show the relative risk of COPD of females vs. males among never smokers and current smokers (20 CPD). (e) and (f) show the relative risk of COPD of current smokers vs. never smokers and former smokers vs. current smokers, respectively. Smokers were assumed smoking either 20 or 40 CPD starting at age 20. Former smokers were assumed to quit smoking at age 40. Blue lines (males); Red lines (females).
Table 3.
Examples of 6-year absolute risk estimates for incidence of diagnosed COPD.
Selected scenarios include current or former smokers at age 50, 60, 70 or 80, who have smoked either 20 pack-years or 40 pack-years. The smoking duration varies by 20, 30, or 40 years. These 6-year risk estimates were calculated without adjusting for other causes of mortality.