A Model-Robust G-Computation Method for Analyzing Hybrid Control Studies Without Assuming Exchangeability
Zhiwei Zhang, Peisong Han, Wei Zhang
TL;DR
This work tackles bias risk in hybrid control designs that combine an RCT with external data by introducing a model-robust GC-VS g-computation method. GC-VS uses an outcome-regression model with adaptive lasso to identify non-exchangeable components between internal and external controls, enabling selective borrowing of external information while preserving consistency even under misspecification. Theoretical results establish consistency and asymptotic normality, with oracle-property guarantees for the variable selection, and demonstrate potential efficiency gains over GC-RCT when some interaction terms are null. Simulation and real-data examples (HIV trials) show GC-VS can substantially improve precision without sacrificing validity and is straightforward to implement in standard software, enhancing the practical utility of hybrid-control analyses.
Abstract
There is growing interest in a hybrid control design for treatment evaluation, where a randomized controlled trial is augmented with external control data from a previous trial or a real world data source. The hybrid control design has the potential to improve efficiency but also carries the risk of introducing bias. The potential bias in a hybrid control study can be mitigated by adjusting for baseline covariates that are related to the control outcome. Existing methods that serve this purpose commonly assume that the internal and external control outcomes are exchangeable upon conditioning on a set of measured covariates. Possible violations of the exchangeability assumption can be addressed using a g-computation method with variable selection under a correctly specified outcome regression model. In this article, we note that a particular version of this g-computation method is protected against misspecification of the outcome regression model. This observation leads to a model-robust g-computation method that is remarkably simple and easy to implement, consistent and asymptotically normal under minimal assumptions, and able to improve efficiency by exploiting similarities between the internal and external control groups. The method is evaluated in a simulation study and illustrated using real data from HIV treatment trials.