In health technology assessment, existing covariate adjustment methods-such as network meta regression, matching-adjusted indirect comparison, and simulated treatment comparison often implicitly rely on the shared effect modification (SEM) assumption. When this assumption does not hold, systematic bias may arise. Network meta interpolation (NMI), by contrast, does not depend on the SEM assumption. Instead, it leverages the correlation structure of covariates, applying best linear unbiased prediction to impute unreported subgroup data, which are then used for regression analysis to obtain indirect comparison results. In this study, we conducted Monte Carlo simulations to evaluate the performance of NMI and network meta-analysis across three scenarios. The results showed that, regardless of whether the SEM assumption held, NMI produced estimates that were closer to the true values and confidence intervals with higher coverage. This method therefore offers an important complement for addressing covariate imbalance and non-shared effect modification in HTA, though caution is warranted in interpreting results due to the risks of multiple comparisons and inflated type I error.
