Table 3.

Association between usual alcohol intake before diagnosis and mortality during the follow-up period, for total alcohol and by type of alcohol, in a cohort of early stage head and neck cancer patients

Alcohol intakePerson-yearsAll-cause mortality
No. of deathsRR* (95% CI)
Total (drinks/wk)
    0 (n = 30)14531.0
    1-7 (n = 49)19761.46 (0.35, 6.14)
    8-21 (n = 60)275111.44 (0.39, 5.35)
    22-35 (n = 32)13372.36 (0.60, 9.31)
    >35 (n = 93)352354.87 (1.46, 16.27)
    P-trend<0.0001
Beer (drinks/wk)
    0 (n = 82)367141.0
    1-7 (n = 76)314141.20 (0.53, 2.73)
    8-21 (n = 52)201132.37 (1.03, 5.44)
    >21 (n = 54)219212.86 (1.34, 6.12)
    P-trend0.04
Wine (drinks/wk)
    0 (n = 160)675441.0
    1-2 (n = 43)18070.58 (0.25, 1.35)
    3-7 (n = 30)12140.44 (0.14, 1.35)
    >7 (n = 31)12570.75 (0.32, 1.76)
    P-trend0.27
Liquor (drinks/wk)
    0 (n = 91)405161.0
    1-7 (n = 61)244121.27 (0.59, 2.75)
    8-21 (n = 48)217131.04 (0.47, 2.33)
    >21 (n = 64)235212.11 (1.04, 4.28)
    P-trend0.0001
  • *Adjusted for age, sex, smoking history (pack-years, current versus former/never), educational level, BMI, race, randomization group, and other types of alcohol (where applicable). P-trend refers to the test for a linear dose-response relationship for each measure of alcohol use (total, beer, wine, and liquor separately).