Skip to main content
. 2017 Oct 17;14(10):e1002410. doi: 10.1371/journal.pmed.1002410

Table 6. Coefficients for the CVD and severe adverse event models fit by traditional backwards selection.

Variable Hazard ratio (exponentiated coefficient) in CVD event/death model (95% CI) Hazard ratio (exponentiated coefficient) in serious adverse event model (95% CI)
Individual terms
Age 1.064 (1.048, 1.081) 1.054 (1.046, 1.062)
Female 1.177 (1.021, 1.356)
Black race
Hispanic ethnicity
Systolic blood pressure 1.002 (0.993, 1.011) 1.003 (0.999, 1.008)
Diastolic blood pressure 1.005 (0.992, 1.018)
Number of blood pressure medications 1.205 (1.102, 1.318) 1.245 (1.129, 1.374)
Current smoker 1.994 (1.532, 2.594) 1.633 (1.335, 1.998)
Former smoker
Daily aspirin use
Statin use
Serum creatinine 1.603 (1.292, 1.988) 1.835 (1.566, 2.148)
Total cholesterol 1.003 (1.001, 1.006)
High-density lipoprotein cholesterol 0.986 (0.979, 0.993)
Triglycerides
Urine microalbumin/creatinine ratio 1.001 (1.000, 1.001) 1.001 (1.001, 1.001)
Body mass index
Intensive treatment arm 2.123 (0.180, 25.063) 1.567 (1.156, 2.124)
Interaction with intensive treatment arm
Age 0.985 (0.962, 1.008)
Systolic blood pressure 1.010 (0.997, 1.024)
Diastolic blood pressure 0.981 (0.962, 1.001)
Number of blood pressure medications 0.911 (0.800, 1.038)

Because the backwards selection models failed external validation, we would not recommend them for clinical use; see main text and Tables 2 and 3 for models estimated through elastic net regularization.

CVD, cardiovascular disease.