Table 2.
The HUNT Lung Cancer Model. Variables and questions to participants. Cox prediction model of lung cancer risk for 33,521 HUNT2 participants who had ever smoked⁎ and did not develop any other type of cancer in a mean follow-up time of 13·2 years.
| Variable | Questions to participants | Hazard ratio (95% CI) | P value | Beta coefficient |
|---|---|---|---|---|
| Sex | - | 1·128 (0·941–1·352) | 0·188 | 0·1205819 |
| Agea | Age at participation at screening | 0·135 (0·098–0·186) | <0·001 | -2·0020557 |
| Pack-years (log) | Estimated number of pack-years | 3·200 (2·451–4·176) | <0·001 | 1·1630181 |
| Smoking quit time, years (log) | If you previously smoked, how long has it been since you stopped? (Number of years) | 0·786 (0·705–0·876) | <0·001 | -0·2407998 |
| Body mass index (log) | BMI | 0·288 (0·153–0·539) | <0·001 | -1·2462656 |
| Cough daily, yes vs no | Do you cough daily during periods of the year? | 1·501 (1·250–1·802) | <0·001 | 0·4059355 |
| Smoke exposure, hours (log) | How long are you usually in a smoky room each day? (Number of hours) | 1·181 (1·062–1·313) | 0·002 | 0·1663201 |
| Smoking intensity per 1 cigarette increase | How many cigarettes do you or did you usually smoke daily? | 0·971 (0·951–0·991) | 0·004 | -0·0295406 |
To calculate the 16-year lung cancer risk in one person with the use of categorical variables, multiply the beta coefficient of the variable by 1 if the factor is present and by 0 if it is absent. For continuous variables other than age, multiply their value – or their log value if indicated – by the beta coefficient of the variable. For age, calculate its contribution by dividing by 100, exponentiated by the power −1, and multiply by the beta coefficient of the variable. Calculate the sum of all previously calculated beta coefficient products; this sum is represented as Xβ. To obtain the person's 16-year LC risk, calculate 1 − 0.06exp(Xβ). CI denotes confidence interval.
Age had a non-linear association with LC and was transformed as (100/Age).