Presentation (CTI)

OBJECTIVES: With the EU-mandated JCA increasing demands for comparative effectiveness evidence—particularly where direct evidence is limited—Matching-Adjusted Indirect Comparisons (MAICs) are likely to play a key role. The method proposed by Signorovitch (SignMAIC) in 2010 remains the industry standard for estimating individual patient weights (PWs) in MAICs. The PolyMAIC method was proposed in 2022 as a potential alternative. Our aim was to undertake a comprehensive comparison of these two weighting methods.
METHODS: A simulation study was conducted involving 2,000 scenarios over a range of individual patient data (IPD) sample sizes, levels of overlap in baseline characteristics (HTD vs. comparator trial), distribution types, and correlations. The performance of the two weighting methods was assessed using the effective sample size percentage (ESS %) and the maximum PW.
RESULTS: PolyMAIC matched target statistics at least as closely as SignMAIC in all scenarios. The PolyMAIC method retained a greater proportion of information from the IPD, where the mean ESS % was 2.8% higher (62.3 vs. 59.5). The maximum PW was on average 1.8 lower (4.7 vs. 6.5) for PolyMAIC vs SignMAIC. In the subset of scenarios where SignMAIC achieved an ESS between 20% and 80%, the mean ESS % was 3.0% higher for PolyMAIC (56.4 vs. 53.3).
CONCLUSIONS: The PolyMAIC approach consistently outperformed the Signorovitch method across a broad range of realistic matching scenarios. The performance gains were greater in the more challenging scenarios.