Figure 1From: Using imputed pre-treatment cholesterol in a propensity score model to reduce confounding by indication: results from the multi-ethnic study of atherosclerosis Hazard ratios from cox proportional hazards models with multiple imputations by CVD endpoint. Back to article page