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

The usual assessment of coronary angiograms by standard visual analysis was compared with the assessment by quantitative computer analysis to provide evidence of the consequences of measurement errors attributable to visual assessment in the evaluation of coronary stenoses in clinical trials.

Angiograms obtained in 151 patients enrolled in a double-blind randomized controlled trial assessing the effect of fish oil supplementation on restenosis were evaluated by both visual and quantitative methods. Proportions of patients with unsuccessful angioplasty and proportions of patients with restenosis according to visual and quantitative methods were compared by standard statistical methods.

Of the 151 patients who were considered successfully dilated by visual interpretation, 32 were not considered successfully dilated by quantitative analysis. With quantitative computer analysis, evaluation of restenosis on the 119 successfully dilated patients documented a protective effect of fish oil against restenosis (30.5% of treated patients versus 48.3% of control patients, P < 0.05). Visual evaluation of restenosis six months after angioplasty misclassified 10% of patients in each treatment group and failed to demonstrate a protective effect of fish oil against restenosis.

Misclassification by visual assessment of both success and restenosis in the present trial led to underestimating the true effect of fish oil in the prevention of restenosis. Accurate and objective methods, such as quantitative computer analysis, are thus required for evaluating the effects of interventions on coronary arteries.