Cardiovascular disease risk prediction factors.

To the Editor: The Emerging Risk Factors Collaboration (ERFC) reported on the addition of lipid-related markers to total cholesterol and high-density lipoprotein cholesterol (HDL-C) in the prediction of cardiovascular disease. We attempted to replicate the findings of eTable 1, which summarized the 26 studies that compared apolipoprotein B with the cholesterol markers. Many discrepancies with the primary publications were found. First, 4 eligible studies, including the Framingham Offspring Study, were not included. Second, 5 studies did not meet the entry criteria, which included approved definitions and recording of both fatal and nonfatal cardiovascular events. The Whitehall I study recorded only fatal ischemic heart disease and so should not have been eligible. Yet strokes were listed in eTable 1 even though they were not a protocol outcome. Third, in 19 of the 26 studies, the number of participants differed substantially between eTable 1 and the primary publication. In 14 studies, there were fewer participants whereas in 5 studies, there were more participants. For example, the ERFC study listed 1293 patients for the Quebec Cardiovascular Study vs the actual tally of 2072. The ERFC study listed 23 175 for the Women’s Health Study vs 15 632 in the primary report. Fourth, the values for apolipoprotein B in 4 study reports differed substantially from those given by the ERFC and neither the methods to measure apolipoprotein B nor the results were given in 4 other studies. Furthermore, in at least 2 studies, apolipoprotein B was measured many years after collection with no documentation of adequate sample preservation. Fifth, the methods and results of 11 studies have not been recorded in the public domain. We assume there are reasonable explanations for the discrepancies but, even so, the ERFC publication is difficult to reconcile with the primary material. Given the apparent deficiencies and discrepancies, the validity of the ERFC findings and their recommendations need to be reexamined. Additional details are available from the author on request.