Table 4.
Odds ratios between various smoking-related variables (continuous) and lung cancer risk
| Model | Cigarette smoking variablesa | Unit | β estimate | ORb (95% CI) | DF | AIC | Δ AICc |
|---|---|---|---|---|---|---|---|
| Among men who ever smoked [N = 1 454] | |||||||
| 1 | Duration (Dur) | per year | (7.7 x 10-2) x Dur | 1.08 [1.06 - 1.10] | 24 | 1630 | - |
| 2 | Intensity (Int) | per pack | 0.84 x Log(Int) | 2.31 [1.85 - 2.89] | 24 | 1731 | -22 |
| 3* | Pack-years (PY) | per pack-year | 1.06 x Log(PY) | 2.87 [2.35 - 3.52] | 24 | 1646 | -39 |
| 4 | CSI index (CSI) | per unit | 1.31 x CSI | 3.70 [3.04 - 4.50] | 24 | 1583 | - |
| 5 | Duration (Dur) | per year | (7.2 x 10-2) x Dur | 1.07 [1.06 - 1.09] | 25 | 1602 | -8 |
| Intensity (Int) | per pack | 0.64 x Log(Int) | 1.90 [1.49 - 2.42] | ||||
| 6 | Duration (Dur) | per year | variable not retained in the final model | 25 | 1600 | -10 | |
| Intensity (Int) | per pack | 0.75 x Log(Int) | 2.12 [1.66 - 2.70] | ||||
| Time since cess. (TSC) | per year | (-7.9 x 10-2) x TSC | 0.92 [0.91 - 0.94] | ||||
| 7 | Pack-years (PY) | per pack-year | 0.75 x Log(PY) | 2.11 [1.70 - 2.61] | 25 | 1588 | -19 |
| Time since cess. (TSC) | per year | -0.4 x Log(TSC) | 0.67 [0.60 - 0.75] | ||||
| Among women who ever smoked [N = 734] | |||||||
| 1* | Duration (Dur) | per year | (8.1 x 10-2) x Dur | 1.08 [1.06 - 1.11] | 17 | 781 | - |
| 2* | Intensity (Int) | per pack | 1.22 x Log(Int) | 3.40 [2.47 - 4.66] | 17 | 792 | -33 |
| 3 | Pack-years (PY) | per year | 1.15 x Log(PY) | 3.17 [2.41 - 4.16] | 17 | 755 | -21 |
| 4 | CSI index (CSI) | per unit | 1.74 x CSI | 5.68 [4.02 - 8.00] | 17 | 735 | - |
| 5* | Duration (Dur) | per year | (6.8 x 10-2) x Dur | 1.07 [1.04 - 1.10] | 18 | 743 | -16 |
| Intensity (Int) | per pack | 1.05 x Log(Int) | 2.87 [2.02 - 4.08] | ||||
| 6* | Duration (Dur) | per year | (3.8 x 10-2) x Dur | 1.04 [1.00 - 1.08] | 19 | 739 | -16 |
| Intensity (Int) | per pack | 1.13 x Log(Int) | 3.09 [2.15 - 4.45] | ||||
| Time since cess. (TSC) | per year | (-5.0 x 10-2) x TSC | 0.95 [0.91 - 0.99] | ||||
| 7* | Pack-years (PY) | per pack.year | 0.97 x Log(PY) | 2.63 [1.97 - 3.51] | 18 | 741 | -12 |
| Time since cess. (TSC) | per year | -0.36 x Log(TSC) | 0.70 [0.58 - 0.84] | ||||
This table presents several model fits based on one or several smoking-related variables (continuous). Different functions were used to model the relations between continuous smoking metrics and the logit of the lung cancer risk, including linear and logarithmic functions models as well as fractional polynomials
* More complex models are associated with better AIC values: the β estimates being respectively:
- for Model 3 among men: 0.78 x PY0.5 - 0.03 x PY - AIC = 1638
- for Model 1 among women: (3.4 x 10-4) x Dur3 + (-8 x 10-5) x Dur3 x log(Dur) - AIC = 773
- for Model 2 among women: -2.09 x Int-0.5 - AIC = 785
- for Model 5 among women: (6.6 x 10-2) x Dur and -1.79 x Int-0.5 - AIC = 740
- for Model 6 among women: (4.1 x 10-2) x Dur, -1.9 x Int-0.5 and 0.89 x TSC-2 - AIC = 731
- for Model 7 among women: 0.41 x PY0.5 and 1.03 x TSC-2 - AIC = 737
aComputed after a discount up to 2 years when relevant
bOdds ratio [OR] estimated with logistic regression, adjusted for age, respondent type, education, ethnic group, annual income and exposure to lung carcinogens [each exposure being considered separately – cf. the method section]
cDifference in AIC criterion with linear model