Measurement errors in standard visual analysis of coronary angiograms: consequences on clinical trials.