Article type
Year
Abstract
Introduction: A sufficiently high control group event rate is needed to detect clinically meaningful differences in randomized control trials. Including studies with low control rates in meta-analyses (M-As) may dilute the overall treatment effect and may occasionally mislabel a useful treatment as ineffective.
Objective: To perform an assessment of the relevance of control rate in the conduct of M-A.
Methods: Cumulative M-As are performed on studies ordered by control rates, by odds ratio and risk difference methods, and by fixed and random effects models to assess the impact of studies with varying control rates on treatment efficacy. This method was applied to over 50 published M-As. A qualitative assessment was performed on the results.
Results: In most instances, treatment effects are monotonically correlated with the control rates. For example, a M-A of thrombolytic therapy in acute myocardial infarction included a heterogeneous group of studies. Studies with high control rates generally have a higher treatment efficacy when compared with studies with lower control rates. M-As subgroupcd by different levels of control rates could be used to reduce or eliminate heterogeneity of studies pooled. A weighted meta-regression of the log relative risk on the control rates could also be used.
Discussion: An evaluation of the control rate should be included in all M-As to help make specific recommendations about treatments. Some findings of inconsistencies between studies in a M-A may at least partially be explained by the differences in control rates.
Objective: To perform an assessment of the relevance of control rate in the conduct of M-A.
Methods: Cumulative M-As are performed on studies ordered by control rates, by odds ratio and risk difference methods, and by fixed and random effects models to assess the impact of studies with varying control rates on treatment efficacy. This method was applied to over 50 published M-As. A qualitative assessment was performed on the results.
Results: In most instances, treatment effects are monotonically correlated with the control rates. For example, a M-A of thrombolytic therapy in acute myocardial infarction included a heterogeneous group of studies. Studies with high control rates generally have a higher treatment efficacy when compared with studies with lower control rates. M-As subgroupcd by different levels of control rates could be used to reduce or eliminate heterogeneity of studies pooled. A weighted meta-regression of the log relative risk on the control rates could also be used.
Discussion: An evaluation of the control rate should be included in all M-As to help make specific recommendations about treatments. Some findings of inconsistencies between studies in a M-A may at least partially be explained by the differences in control rates.