Relationship between total mortality and cholesterol reduction as found by meta-regression analysis of randomized...

All randomized trials of cholesterol lowering except those without information on total mortality or cholesterol reduction were included in this meta-analysis. Its purpose was to review the relationship between log odds ratio of total mortality in active vs. control group and degree of cholesterol reduction, adjusting for appropriate covariates such as type of trial (single or multifactor), risk level in the control group, and type of treatment. Analysis revealed that risk level in the control group was particularly important to adjust for because it correlated both to degree of cholesterol reduction (r = 0.48) and to log OR of total mortality (r = -0.38, single-factor trials) and thus confounded the cholesterol mortality association. After adjustment, a borderline significant association was found between degree of absolute cholesterol reduction and total mortality log OR. Type of treatments also seemed to be associated with log OR, mostly through excess risk in hormone-treated patients. There was no statistically significant indication that fibrate trials had inferior results as compared to other drug trials. Meta-analysis estimates of global treatment effects can be misleading in the guidance for creating appropriate recommendations regarding cholesterol reduction. More emphasis must be put on how cholesterol is reduced and how much.