Table 5A. Coefficients for the Model Predicting Colorectal Cancer in Men
VariablesCoefficientStandard Error
Age*
    Linear component0.09170.0106
    Nonlinear component−0.02340.0106
Diabetes0.11020.0758
Regular use of multivitamins−0.19180.0532
Family history of colon cancer0.24250.0850
Years of education*
    Linear component0.07210.0181
    Nonlinear component−0.07340.0193
Race/ethnicity
    Hawaiian0.16090.1208
    Japanese0.25350.0820
    Latino−0.13660.0925
    White−0.16730.0921
Body mass index*
    Linear component0.01800.0162
    Nonlinear component0.00900.0194
Alcoholic drinks per day*
    Linear component0.28380.0752
    Nonlinear component−1.73750.5356
Hours of moderate activity per day*
    Linear component−0.09070.0702
    Nonlinear component0.11030.1496
Regular aspirin use
    Yes, not currently−0.03230.0719
    Yes−0.20960.0655
Pack-years of smoking*
    Linear component0.00020.0059
    Nonlinear component0.01770.0170
  • * Continuous variables were modeled using restricted cubic splines with 3 knots. Therefore, continuous variables contain one coefficient for the linear component and a nonlinear component.