Article type
Year
Abstract
Background: It may not always be possible to blind participants of a randomized controlled trial (RCT) for treatment allocation. Knowledge of treatment allocation may lead to differences between treatment arms, and consequently observed differences in the outcome may not be attributable to the treatment, potentially biasing treatment effect estimates.
Objective: To extend a novel method, originally introduced in genetic research, for instrumental variable meta-analysis adjusting for bias due to unblinding of trial participants.
Methods: Using simulation studies, this novel method, Egger-IV, is introduced and compared to the performance of the 'as treated', 'intention-to-treat', and regular 'instrumental variable' estimators in various scenarios. Scenarios considered (time-varying) non-adherence, confounding, and between-study heterogeneity. The effect of treatment on a binary endpoint was quantified by means of a risk difference.
Results: In all scenarios with unblinded treatment allocation, the Egger-IV method was the least biased estimator. However, precision was lacking, and, consequently, power usually was low.
Conclusion: The Egger-IV estimator corrects for bias in meta-analyses of unblinded RCTs. Due to a lack of precision and power we suggest using this method mainly as a sensitivity analysis.
Objective: To extend a novel method, originally introduced in genetic research, for instrumental variable meta-analysis adjusting for bias due to unblinding of trial participants.
Methods: Using simulation studies, this novel method, Egger-IV, is introduced and compared to the performance of the 'as treated', 'intention-to-treat', and regular 'instrumental variable' estimators in various scenarios. Scenarios considered (time-varying) non-adherence, confounding, and between-study heterogeneity. The effect of treatment on a binary endpoint was quantified by means of a risk difference.
Results: In all scenarios with unblinded treatment allocation, the Egger-IV method was the least biased estimator. However, precision was lacking, and, consequently, power usually was low.
Conclusion: The Egger-IV estimator corrects for bias in meta-analyses of unblinded RCTs. Due to a lack of precision and power we suggest using this method mainly as a sensitivity analysis.