Skip to main content

Table 4 Hazard ratios for statins from cox proportional hazards models for incident CHD

From: Using imputed pre-treatment cholesterol in a propensity score model to reduce confounding by indication: results from the multi-ethnic study of atherosclerosis

 

CHD, Hard

 

CVD, Hard

 
 

(Events = 119)

 

(Events = 190)

 

Model (N = 6035)

HR (95% CI)

p-value

HR (95% CI)

p-value

Unweighted 1

0.97 (0.60,1.59)

0.915

0.92 [0.62,1.36]

0.676

Weighted

    

  Propensity model

    

  No cholesterol

0.92 [0.54,1.57]

0.767

0.82 [0.54,1.26]

0.373

  Imputed cholesterol2

0.74 [0.38,1.42]

0.363

0.72 [0.43,1.21]

0.215

  New-user analysis (N = 4886)

0.91 [0.45,1.80] (Events = 100)

0.777

0.71 [0.38,1.30] (Events = 153)

0.265

  Sensitivity analysis3

    

  No truncating

0.59 [0.27,1.32]

0.199

0.62 [0.34,1.15]

0.130

  <.2 & >20

0.68 [0.34,1.36]

0.275

0.69 [0.40,1.17]

0.166

  <.3 & >30

0.64 [0.30,1.34]

0.238

0.66 [0.38,1.16]

0.152

  <.3 & >10

0.74 [0.39,1.42]

0.370

0.73 [0.44,1.21]

0.220

  1. 1 Adjusted for traditional risk factors: Age, gender, race/ethnicity, BMI, diabetes status, family history of heart attack, smoking status, hypertensive medications, systolic BP.
  2. 2 Multiple imputations were used for models where imputed cholesterol was used in creating the weights. Weights < .1 were set to 0.1 and weights >10 were set to 10.
  3. 3 Using non-truncated stabilized weights and various thresholds for truncating.
Back to article page

BMC Medical Research Methodology

ISSN: 1471-2288

Contact us