Presentation (CTI)

OBJECTIVES: In Health Technology Assessment (HTA), comparative clinical evidence is required for a new intervention. In the EU Joint Clinical Assessment, a large number of comparators of interest and limited availability of head-to-head trials necessitate the submission of indirect treatment comparisons (ITCs). In the case of dissimilar study populations, population-adjusted indirect comparisons (PAICs) adjust for differences in effect modifiers, aiming to improve the validity of the comparison. When individual patient data is available for at least one study in a connected network, EU methodological guidelines recommend various PAIC methods to estimate treatment effects — matching-adjusted indirect comparison (MAIC), simulated treatment comparison (STC), and multilevel network meta-regression (ML-NMR). However, the question remains of which method to choose in a given data situation.
METHODS: We conducted a simulation study to evaluate the performance of PAIC methods in scenarios relevant to HTA submissions. Each simulation represented an ITC between two studies, with the unadjusted Bucher method included as a reference case. Across the simulation scenarios, we varied the degree of baseline differences between populations, the strength of treatment effect modification, the true relative treatment effect, and the study sample size. Methods were evaluated in terms of bias, empirical coverage of 95% confidence intervals, and statistical power.
RESULTS: The simulations provide a structured comparison of PAIC methods relative to the unadjusted ITC, highlighting the impact of baseline imbalances, effect modification, and sample size on performance. Our results indicate a substantial reduction in bias with any method, while statistical power is below typical standards in many conditions. Detailed findings on the trade-offs between bias reduction, coverage, and power across scenarios are presented.
CONCLUSIONS: No single PAIC suits every situation. This study offers practical guidance for the use of PAICs in HTA applications, helping to inform methodological choices in the presence of imbalanced effect modifiers across studies.