From the 2102 randomized patients in ALERT we selected 1734 patients with a complete set of risk and adjustment factors used in the study. Cox regression analysis was used to estimate relationships between baseline values of risk factors and first occurrence of MACE. Chi square statistics, receiver operating characteristics (ROC) and net reclassification improvement (NRI) were used to compare the information value of different risk factors.
Atherogenic LC and especially non-high density cholesterol (nHDL-C) were as predictive as creatinine and nHDL-C was about as predictive as diabetes. CRP, body mass index, hypertension and glucose had less prediction ability than nHDL-C. The rank order of prediction was the same in the two treatment groups. By regression modelling the actual MACE risk reduction from 6 weeks onwards was well predicted from the difference in LC at 6 week.
LC and especially nHDL-C predicted MACE at least as good as creatinine. Diabetes was about equally good as nHDL-C to predict MACE occurrence. Inflammation had less prediction ability than the other factors. Treated levels of atherogenic LC predicted MACE risk reduction well.