TABLE 1. Contributions to Cardiovascular Disease Prediction in the Women’s Health Study*

VariableVariable χ2PRR per 2 SDc Statistic
+ indicates the addition of each variable separately to the model with age only, or age, SBP, and smoking only; −, the deletion of each variable separately from the full model; Ln, natural logarithms; RR, relative risk; and TC, total cholesterol. χ2 is the likelihood ratio statistic for each single variable when added to the model (top and middle) or subtracted from the full model (bottom). Ln were used to normalize distributions and improve the fit for individual predictors.
*Estimated from Cox proportional hazards models.
†RR is relative risk when each variable is included in models shown in the top and middle sections, and is relative risk in full model in the bottom section; RR compares risk across 2 SD units, except for smoking, which is yes versus no.
Effect of adding variables to model with age only
    Ln(age) only395.9<0.00014.00.70
        +Ln(SBP)148.8<0.00012.50.74
        +Current smoking121.1<0.00012.90.73
        +Ln(HDL)85.7<0.00011/2.00.73
        +Ln(TC)33.3<0.00011.60.72
        +LDL28.8<0.00011.50.71
Effect of adding variables to model with age, SBP, and smoking
    Ln(age), Ln(SBP), smoking0.76
        +Ln(HDL)45.3<0.00011/1.70.77
        +Ln(TC)21.5<0.00011.40.77
        +LDL18.6<0.00011.40.77
Effect of deleting variables from full model
    Ln(age), Ln(SBP), smoking, Ln(TC), Ln(HDL)−0.78
        −Ln(TC)39.8<0.00011.60.77
        −Ln(HDL)63.6<0.00011/1.70.77
        −Current smoking99.7<0.00012.60.76
        −Ln(SBP)114.1<0.00012.20.76
        −Ln(Age)257.4<0.00013.20.73