Advanced Methods for Matching-Adjusted Indirect Comparison (MAIC)

PURPOSE: Anchored covariate-adjusted indirect treatment comparisons (ITCs) inform reimbursement decisions when (1) there are no head-to-head trials between the treatments of interest, (2) there is a common comparator arm shared by two studies, and (3) there are patient-level data limitations. Matching-adjusted indirect comparison (MAIC) is currently the most widely used method in HTA but is not always possible. We begin with an overview of newly proposed MAIC weighting schemes and approaches for assessing the numerical feasibility of conducting an MAIC. We then consider the "two-stage MAIC", an extension to MAIC which has been recently proposed and shown to yield improved precision and efficiency, while maintaining similarly low levels of bias when assumptions are met. We review the two-stage MAIC and how it has recently been applied to achieve greater precision. Finally, certain approaches in causal inference are known to be “doubly-robust” meaning that they will provide unbiased estimates so long as either the trial allocation model or the outcome model is correctly specified. However, doubly-robust methods for ITCs have not yet been proposed. Here we consider the potential of “augmented” MAIC for doubly-robust estimation in anchored ITCs.

DESCRIPTION: Workshop attendees will review the main advantages and limitations of MAIC and be introduced to newly proposed methods. The audience will be asked (real-time polling) to consider if these “advanced” methods should be more widely considered in HTA . Ms. Cope will chair the session and introduce the main issues involved with existing MAIC methods and summarize recent work on alternative weighting schemes and feasibility assessment (20 min). Dr. Remiro-Azócar will summarize his recent work on the two-stage MAIC method (15 min) and Dr. Campbell will consider the potential of “augmented” methods for doubly-robust estimation (15 min). To conclude, an audience discussion will consider the potential and priorities for MAIC methods research (10 min).