Coronary Arteriography in the Study of the Epidemiology of Coronary Artery Disease1

The greatest strength of coronary arteriography as an epidemiologic tool is that it allows coronary atherosclerosis to be studied rather than its clinical manifestations. Thus, the pathogenic sequence in which a normal coronary artery becomes progressively narrowed and then occluded can be examined in terms of the factors causing stenosis and/or infarction. Further refinements in the arteriographic techniques will probably occur, including improvements in the resolution of the images, additional methods for computer-assisted measurement of the extent of lesions, and possibly even the removal of the need to perform the technique by arterial puncture and direct coronary injection. If these noninvasive techniques become available, coronary arteriography may be even more available to study coronary atherosclerosis on a population basis. The validity and reproducibility of these techniques would, of course, have to be established. However, the numerous advantages of arteriography in the study of coronary atherosclerosis could then be applied to populations, allowing better definition of cases and controls, identification of subgroups of special importance (e.g., left main disease), the opportunity to study anatomic features in random samples of the population, and further studies of the mechanism by which risk factors act in the natural history of coronary atherosclerosis. Indeed, studies of the effects of risk factor modifications have already successfully used arteriography rather than clinical criteria to measure disease end points. This trend will likely continue. Until these technologic advances are realized, there remains the need to identify and control for a number of biases which heretofore have limited the inferences which can be made from cross-sectional arteriographic studies. The sampling and measurement biases found in most case-control studies may be particularly prevalent in arteriographic studies (table 6). Further work is needed to determine the presence, extent, and direction of the biases in cross-sectional studies that use present-day technology.