Invasive versus noninvasive studies of risk factors and atherosclerosis

Associations of risk factors with atherosclerosis may be assessed by either invasive methods for measuring the arterial disease, such as angiography, or noninvasive methods; these methods differ in their potential for bias. Biases associated with coronary angiography may be difficult to control in statistical analysis, either because they are unrecognized or because they are amenable to neither stratification nor multivariate analysis. Problems in control selection include the likelihood that angiography controls overrepresent related ischemic or noncoronary cardiac conditions with their own risk factor associations. Differential misclassification is more likely in the clinical setting when invasive studies are used than in a research setting involving ultrasound imaging of carotid arteries. Nondifferential misclassification, however, affects both types of studies and clouds interpretation of the comparative strength of risk factor associations with atherosclerosis assessed by the two methods. Recent angiographic studies have generally provided insufficient information to evaluate these biases. However, with proper attention to such biases, one may be able to learn much about early and late stages of atherosclerosis by comparing risk factor associations with disease measured by both coronary angiography and carotid ultrasound.