We used logistic regression models to estimate the COPD odds ratios for each TTFC category, with TTFC >60 minutes as the referent group. Models were adjusted for demographic variables including age at PLCO questionnaire (continuous), gender, race (white non-Hispanic, other), and education (≤high school, some college, ≥college). We also adjusted for smoking-related variables to assess the additional risk that TTFC conferred; these variables included cigarettes per day smoked (≤5 cigarettes per day, 6–19 cigarettes per day, 1 pack per day, >1 pack per day), years smoked during a lifetime (continuous), pack-years (continuous), age at smoking initiation (<17, ≥17 years), and prior diagnosis of lung cancer (yes/no). Further adjustment for study center (10 locations) did not appreciably change results (Table A in S1 File), thus to avoid over-fitting the model, study center is not included in the main models. We also explored the possibility that lung cancer diagnoses prior to the follow-up questionnaire biased our estimates by excluding these individuals from our models in a sensitivity analysis. Lastly, we conducted a sensitivity analysis among only participants with complete follow-up questionnaire data
