As health technology assessment (HTA) becomes increasingly central to healthcare decision-making, population-adjusted indirect comparisons (PAICs) have emerged as critical tools. These methods allow researchers to compare treatments across trials with varying populations, addressing situations where head-to-head comparisons are unavailable or impractical. Here, we will examine the development of PAICs, their role in network meta-analysis (NMA), and the ongoing debates around their application.
The need for population adjustments
When it comes to healthcare decision-making, one goal is to compare treatments fairly and accurately. But trials often differ in their populations—patients might vary in terms of age, severity of illness, or prior treatment experience. Without population adjustment, indirect comparisons between trials can be biased, leading to flawed treatment recommendations.
This is where PAICs come into play. By adjusting for differences in trial populations, these methods enable indirect comparisons that aim to be as fair and accurate as possible in the absence of direct head-to-head trials.
Core methods for population adjustment
There are two widely recognized methods for population-adjusted indirect comparisons:
- 1. Matching-adjusted indirect comparison (MAIC): This method uses individual patient data (IPD) from one trial to reweight patients based on baseline characteristics, aligning them with those from another trial. The method is suitable when there is good overlap between covariate distributions in the different populations.
- 2. Simulated treatment comparison (STC): In contrast to MAIC, STC uses covariate-adjustment to model the relationship between patient characteristics and outcomes, enabling treatment comparisons across different populations. Due to the model’s ability to extrapolate, this can be more useful where covariate overlap is poor.
Both of these methods can be applied in anchored or unanchored settings. Anchored comparisons use a common comparator (e.g. standard of care) between trials, and therefore only need to adjust for effect modifying covariates, whereas unanchored comparisons do not, and must adjust for effect modifying covariates and prognostic factors. This makes it more challenging to evaluate their robustness, and therefore they are less reliable. Both methods were developed for pairwise indirect comparisons only.
The rise of multi-level network meta-regression (ML-NMR)A major recent development in population adjustment is the multi-level network meta-regression (ML-NMR). This method integrates both IPD and aggregate data within a network meta-analysis framework, offering more comprehensive population adjustments than MAIC or STC alone. ML-NMR allows for adjusting multiple effect modifiers across a network of treatments, avoids aggregation bias and depending on data availability and assumptions regarding shared effect modification, produces estimates in any target population for decision-making.
Challenges in population-adjusted indirect comparisons
While PAIC methods like MAIC, STC, and ML-NMR have revolutionized health technology assessments, they are not without challenges:
- 1. Unanchored comparisons: Unanchored comparisons, where there is no common comparator between trials, are particularly prone to bias. Even with population adjustment, unobserved or missing covariates can introduce biases whose direction and magnitude is difficult to assess. Researchers have called for better ways to quantify this bias and ensure robust decision-making.
- 2. Method selection: There is no one-size-fits-all solution. Each method has its strengths and weaknesses. The most appropriate method should be determined on a case-by-case basis, considering the specific context of the analysis.
The future of population adjustment
The demand for population-adjusted analyses is likely to increase as joint clinical assessments (JCA) become the standard for health technology assessments across the European Union. JCA aims to streamline assessments across multiple countries, and PAIC methods will be vital in ensuring that treatment comparisons are fair and applicable to diverse populations.
Moving forward, further innovation in population adjustment methods is expected, particularly in the context of real-world data and personalized medicine. Advances in methods like ML-NMR are exciting, but they must be balanced with the need for transparency and simplicity in clinical practice.
Conclusion
Population-adjusted indirect comparisons are transforming health technology assessments, enabling more precise and relevant treatment comparisons across varied patient populations. As methods like MAIC, STC, and ML-NMR continue to evolve, they will play an increasingly important role in ensuring that healthcare decisions are based on robust and reliable evidence.
The key challenge for analysts and policymakers alike will be to navigate the complexities of these methods, ensuring that the right tools are used in the right contexts while avoiding over-reliance on unproven assumptions. The future of healthcare decision-making lies in balancing methodological innovation with practical, patient-centered applications.
The ConnectHEOR team is at the forefront of ITC methodologies and will be presenting multiple research pieces at the upcoming ISPOR Europe 2024. If you are curious to understand alternative approaches to MAIC, we invite you to join our presentation, “Indirect Treatment Comparison Methodology Matters: Unlocking the Essentials for Robust Analysis,” on Tuesday, 19 November, from 15:15 to 16:15 CET.
We are also showcasing a poster:
MSR108: How do STC and ML-NMR Compare in Population Adjustment for Indirect Treatment Comparisons (ITC)? Insights and Challenges (Authors: Hugo Pedder, Tushar Srivastava, Kate Ren) on 19 November, 10:30 AM to 1:30 PM CET. In this research, we have conducted an in-depth case study comparing STC and ML-NMR, revealing critical insights.Curious to learn more?
Join us for discussions, questions, and connection. For a more personal conversation, meet our ITC specialists, Kate Ren and Hugo Pedder, at ConnectHEOR Booth #1400!
