Repeated measurements over time: statistical analysis of the angiographic outcomes in the Probucol Quantitative Regression Swedish Trial (PQRST)

Abstract

An increasing use of non-invasive techniques and arteriographies of the coronary, carotid or femoral arteries has taken place during recent years. This has prompted researchers to perform long-term randomized trials aiming at reducing atherosclerosis progression or achieving regression by cholesterol lowering in high risk populations. These trials are designed to use outcomes such as various indices of atherosclerosis which are assessed repeatedly over time. There are major decisions to be made contingent on the data structures in these trials, such as sphericity, interaction, development over time, missing data, non-normality etc. So, the statistical analysis of such data needs to be planned and specified a priori in a unified manner. This is demonstrated for the Probucol Quantitative Regression Swedish Trial (PQRST) which is designed to answer the question whether the drug probucol, given in combination with diet and cholestyramine, can delay development of atherosclerosis in hypercholesterolemic patients up to 70 years old. Change in atheroma volume is assessed by a newly developed quantitative angiographic technique based on digitized angiograms.