Skip to main content
. 2015 Apr 1;10(4):e0121783. doi: 10.1371/journal.pone.0121783

Table 3. Relative hazard of death for statin use after cancer diagnosis vs. no statin use (n = 7813).

Model HR (95%) P-value
Model 1: Unadjusted model using statin as a non-time-dependent variable 0.69 (0.66, 0.72) <0.0001
Model 2: Unadjusted model using statin as a time-dependent variable (removes immortal time bias) 0.76 (0.72, 0.80) <0.0001
Model 3: Model 2 + age, sex, race, neighborhood income adjustment 0.79 (0.75, 0.83) <0.0001
Model 4: Model 3 + stage, grade, tumor size 0.80 (0.76, 0.84) <0.0001
Model 5: Model 4 + resection, radiation and chemotherapy 0.80 (0.76, 0.84) <0.0001
Model 6: Model 5 + obesity, dyslipidemia, diabetes/IGT, chronic pancreatitis and COPD 0.79 (0.75, 0.93) <0.0001
Model 7: Modeling statin as a time-dependent variable that lags by 2 months (removes reverse causation, restricts population to >2 month survivors) 0.94 (0.89, 1.01) 0.08

HR = hazard ratio for statin use; IGT = impaired glucose tolerance, COPD = chronic obstructive pulmonary disease