Commonly used methods to handle the complexities of effect modification in indirect treatment comparisons (ITCs) often rely on the shared effect modifier (SEM) assumption, that is, that the effect modifiers impact a set of treatments in the evidence network in the same way. However, it has been argued that this assumption is limiting and often unreasonable to make.

The solution, argues Senior Research Consultant, Statistics, Caitlin Daly, is Network Meta-Interpolation, a new method that takes advantage of reported subgroup analysis results — which are often available but rarely used in ITCs — to overcome this and other limitations.

What is Network Meta-Interpolation (NMI)?

In the second webinar of the four-part series on making better decisions using innovative comparative effectiveness methods, Caitlin explains that NMI uses the valuable yet often overlooked information provided by subgroup analyses to adjust for differences in the distributions of effect modifiers within a connected network of trials. It allows one to transpose the relative treatment effects of any population without requiring the SEM assumption.

As explained in Caitlin and co-authors’ recently published article, NMI is “a novel approach to balancing patient populations from various studies prior to NMA, using regression techniques to relate outcomes and effect modifiers. NMI consists of a data enrichment step, followed by two regression steps. The output is a balanced, NMA-ready AgD, evaluated at an effect modifier configuration of the researcher’s choice, which could reflect the pivotal trial or a real-world population of interest.”

NMI is carried out over three key steps: 1) enrichment of the trial data; 2) based on this enriched data, the trial-specific relative treatment effects are interpolated to a specific target population; and 3) a network meta-analysis is conducted on these interpolated relative treatment effects.

How does Network Meta-Interpolation compare to other methods?

When compared to methods such as Matching-Adjusted Indirect Comparison (MAIC), Simulated Treatment Comparison (STC), and Multi-Level Network Meta‐Regression (ML-NMR), NMI differs in several key ways:

1. MAIC, STC, and ML-NMR require individual patient data (IPD) to adjust for differences in population, while NMI only requires IPD to estimate correlations between effect modifiers if these correlations are not reliably estimated elsewhere.
2. Summary information on the effect modifiers is also required by MAIC, STC, and ML-NMR, while NMI requires the relative treatment effects within subgroups.
3. All ITC methods may be applied to anchored comparisons. And only MAIC and STC may be applied in unanchored scenarios. MAIC and STC are also limited to two-study scenarios, while all other methods may be applied to connected networks of at least two trials.
4. Finally, NMI does not depend on the SEM assumption to transport the relative treatment effects to other populations.

In a simulation study, NMI provided more accurate relative treatment effect estimates compared to NMA, network meta-regression based on aggregate data, and ML-NMR in a wide range of scenarios.

Network Meta-Interpolation is a simple, accurate, and robust option for balancing studies in an NMA, provided its data requirements are met. To learn more about this novel method, including an in-depth discussion of its uses and advantages and limitations, click here to watch “How to Select the Most Appropriate Network Meta-Analysis (NMA) Method? And Can We Improve Anchored NMAs through Network Meta-Interpolation (NMI)?”:

Share Via