Article type
Year
Abstract
Background: Multiple randomized clinical trials have demonstrated that lowering low density lipoprotein (LDL) cholesterol reduces cardio- vascular (CV) outcomes (death, myocardial infarction (MI), stroke). Meta-analyses to date have focused on treatment comparisons with respect to single outcomes, not addressing the potential competing risks among these correlated events to compare multiple treatments across multiple outcomes.
Objectives: To assess the effect modification of changes in LDL and other lipid measures on a network of treatment effects applied to a set of correlated outcome measures using multiple treatments meta-analysis (MTM).
Methods: Data from 23 large randomized controlled trials comparing lipid lowering agents, statins, fibrates and conventional treatment at various doses and with different amounts of follow up were combined. Each trial reported some or all of 6 mutually exclusive outcomes: fatal or non-fatal MI, fatal or non-fatal stroke, or other fatal or non-fatal events. We applied a new Bayesian multinomial multiple treatments meta-regression model to estimate the effect of the different treatments on the different outcomes at different lipid levels. This model could account for the missing information on some outcomes in some trials.
Results: When LDL cholesterol was reduced by 30 mg/dl, the average risk reduction was 26% for MI, 18% for stroke and 10% for death. The reductions were further reduced for each outcome by 5% for every additional 10mg/dl. Smaller effects were found for other lipid measures. Reductions were greatest for high dose statins.
Conclusions: Meta-analysis of large statin trials demonstrates that statins effectively reduce different CV events at different rates and that these rates depend on the amount of lipid reduction achieved in the trials and on the doses employed.
Objectives: To assess the effect modification of changes in LDL and other lipid measures on a network of treatment effects applied to a set of correlated outcome measures using multiple treatments meta-analysis (MTM).
Methods: Data from 23 large randomized controlled trials comparing lipid lowering agents, statins, fibrates and conventional treatment at various doses and with different amounts of follow up were combined. Each trial reported some or all of 6 mutually exclusive outcomes: fatal or non-fatal MI, fatal or non-fatal stroke, or other fatal or non-fatal events. We applied a new Bayesian multinomial multiple treatments meta-regression model to estimate the effect of the different treatments on the different outcomes at different lipid levels. This model could account for the missing information on some outcomes in some trials.
Results: When LDL cholesterol was reduced by 30 mg/dl, the average risk reduction was 26% for MI, 18% for stroke and 10% for death. The reductions were further reduced for each outcome by 5% for every additional 10mg/dl. Smaller effects were found for other lipid measures. Reductions were greatest for high dose statins.
Conclusions: Meta-analysis of large statin trials demonstrates that statins effectively reduce different CV events at different rates and that these rates depend on the amount of lipid reduction achieved in the trials and on the doses employed.