Inclusion in trial does not imply efficacy

Efficacy variation within a trial can be > than variation between trial and non-trial patients

Need rigorous calibration eval

Arbitrary threshold vs. indiv. decision making

Contrast w/ attributable risk due to LDL, ARR if LDL  -> target LDL
Treat if ARR > 0.02
Need to note interaction between LDL and age

Authors did not handle missing baseline data well (just excluded whole case)