Table 4.
Risk prediction model using quadratic terms in smokers analogous to model 2*
| Variable | Model 2b† | P |
| Smokers in PLCO control arm (N = 33 049), OR (95% CI) | ||
| Age, per year | 1.885 (1.340 to 2.650) | <.001 |
| Age squared | 0.996 (0.993 to 0.998) | .002 |
| Education, per 1 of 7 levels change | 0.927 (0.885 to 0.970) | .001 |
| BMI, per 1 unit change | 0.971 (0.956 to 0.987) | <.001 |
| Family history of lung cancer, yes vs no | 1.560 (1.311 to 1.855) | <.001 |
| COPD, yes vs no | 1.370 (1.142 to 1.644) | .001 |
| Chest x-ray in past 3 y, per 1 of 3 levels | 1.122 (1.023 to 1.230) | .015 |
| Pack-years smoked, per 1 pack-year | 1.039 (1.031 to 1.048) | <.001 |
| Pack-years squared | 0.9998 (0.9997 to 0.9999) | <.001 |
| Smoking duration, linear, per 1 y | 1.013 (0.996 to 1.030) | .144 |
| Smoking quit-time, per 1 y | 0.941 (0.912 to 0.972) | <.001 |
| Smoking quit-time squared | 1.001 (1.000 to 1.002) | .011 |
| Smoking status, current vs former | 1.339 (1.067 to 1.682) | .012 |
| Model performance statistics | ||
| Hosmer–Lemeshow goodness of fit | .008 | |
| Nagelkerke’s R2 | 0.153 (0.152) | |
| ROC AUC/c statistic (BOC) | 0.8099 (0.8077) | |
| Calibration line | Slope (BOC) = 0.988 | |
| Intercept (BOC) = −0.030 | ||
| Mean absolute error = 0.0022 | ||
| 0.9 quantile of absolute error = 0.0047 | ||
| External validation | ROC AUC (95% CI) | |
| Validation sample, smokers only (n = 15 169) | 0.779 (0.740 to 0.819) |
*
BOC = bootstrap optimism corrected; BMI = body mass index; CI = confidence interval; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; ROC AUC = receiver operator characteristic area under the curve.
†
Bootstrap optimism corrected estimates of model performance based on 200 bootstrap resamplings.