OBJECTIVES: Matching-adjusted indirect comparisons (MAICs) typically compare two treatments and consider two randomized controlled trials (RCTs)–one RCT with individual participant data (IPD) and another with summary-level data. However, for many therapeutic areas, there may be multiple RCTs that warrant consideration in an MAIC, and there is limited guidance available for these situations.

METHODS: We conducted a scoping search to identify case studies where MAICs involving multiple RCTs were required. We ran MAIC analyses using NICE DSU Technical Support Document 18 code to assess approaches for conducting MAICs involving multiple RCTs. We considered two approaches:1) conduct multiple independent MAICs using IPD for the treatment of interest in tandem with multiple RCTs for the comparator, followed by pooling of the MAICs using traditional meta-analysis; 2) conduct one MAIC for the treatment of interest versus the comparator, wherein the MAIC was based on a weighted average of characteristics and eligibility criteria across the set of relevant comparator RCTs. We assessed the advantages and disadvantages of each approach.

RESULTS: We identified several examples (e.g., psoriasis, rheumatoid arthritis) in the literature where MAICs involving multiple RCTs were required. Our analyses determined that the meta-analytic approach pooling multiple independent MAICs allows for examination of consistency in findings between MAICs across individual studies; however, meta-analysis of these comparisons double counts patients from the IPD dataset, thereby overestimating precision. The second approach (one MAIC) approach overcomes the difficulty of double counting, but limits the MAIC to characteristics and eligibility criteria that are common across multiple RCTs and does not allow decision-makers to assess alignment of MAIC findings across comparator RCTs.

CONCLUSIONS: Advantages and disadvantages are associated with methodological approaches for conducting MAICs involving multiple RCTs. Both methodological approaches should be considered when conducting MAICs involving multiple RCTs, given they complement each other.