Performance of MAIC Under Covariate Imbalance: Insights from Real World Data and Simulation
SNDS
poster
Auteurs : A. Brouquet, F.Raguideau, E.Herquelot
Date de publication : 1 septembre 2026
Objectifs
The objective is to illustrate the use of MAIC on treatment comparison and to evaluate the impact of variables selected for MAIC adjustment.
Conclusion
MAIC enables indirect comparisons using individual and aggregated data, but including variables with large treatment imbalances markedly reduces ESS and may lead to overestimation of treatment effects, reduced precision, and poorer HR coverage. When key variables show large imbalance, alternative external data source or designs should be considered.