Table 5. Odds ratios (95% confidence intervals) from multivariable models for colorectal cancer risk in categories of alcohol intake as assessed by food diaries and FFQs among participants with both measuresa.
|
Alcohol intake (g per day)
|
||||||||
|---|---|---|---|---|---|---|---|---|
| Non-drinkers | >0–<5 g per day | 5–<15 g per day | 15–<30 g per day | 30–<45 g per day | ⩾45 g per day | P for trend | P trend for drinkers | |
| Food diaries | ||||||||
| No. of all participants | 646 | 477 | 510 | 371 | 165 | 136 | ||
| Colorectal cancer cases | 149 | 100 | 100 | 75 | 38 | 34 | ||
| Multivariable model 1b | 1.18 (0.88–1.60) | 1.00 | 0.91 (0.66–1.26) | 0.92 (0.65–1.30) | 1.08 (0.68–1.70) | 1.24 (0.76–2.04) | 0.97 | 0.60 |
| Multivariable model 2c | 1.38 (1.00–1.91) | 1.00 | 0.90 (0.63–1.29) | 0.98 (0.67–1.42) | 1.20 (0.74–1.95) | 1.32 (0.79–2.22) | 0.84 | 0.25 |
| FFQs | ||||||||
| No. of all participants | 372 | 867 | 662 | 226 | 100 | 78 | ||
| Colorectal cancer cases | 84 | 171 | 150 | 46 | 26 | 19 | ||
| Multivariable model 1b | 1.43 (1.04–1.97) | 1.00 | 1.22 (0.94–1.58) | 1.16 (0.79–1.72) | 1.36 (0.81–2.28) | 1.40 (0.79–2.49) | 0.12 | 0.09 |
| Multivariable model 2c | 1.33 (0.96–1.86) | 1.00 | 1.16 (0.87–1.53) | 1.07 (0.71–1.61) | 1.18 (0.68–2.03) | 1.30 (0.72–2.38) | 0.36 | 0.07 |
Abbreviation: FFQ=food frequency questionnaire.
Conditional logistic regression analyses were restricted to participants who completed both the FFQ and the food diary (496 cases and 1809 controls). Owing to missing information in FFQ data, models were not adjusted for intakes of energy, red meat and processed meat. Adjusting for these variables in models using diary information did not alter the results. P-values for trend were drawn from tests for trend by modelling alcohol intake as a continuous variable in a conditional logistic regression analysis, whereas P-values for trend for drinkers were drawn from tests for trend only from non-zero alcohol drinkers.
Adjusted for age, weight, height, smoking status, social class, intakes of fibre and folate adjusted in the main model (496 cases and 1809 controls).
Adjusted for age, weight, height, physical activity, educational level, smoking status, social class, intakes of fibre and folate in the sensitivity analyses restricted to individuals with complete covariate information (442 cases and 1701 controls).