Validity and consistency of matching-adjusted indirect comparison (MAIC)

Validity and consistency of matching-adjusted indirect comparison (MAIC)

BackgroundWhen new medicines are developed, researchers often want to compare them to existing treatments. However, they may not always have access to patient data from both treatment groups. One common solution is the Matching-Adjusted Indirect Comparison (MAIC) method, which adjusts patient-level data from one study to match summary data from another. This method is widely used in health technology assessments (HTA) which summarise clinical, economic, and ethical al evidence for a new health technology in order to determine whether it should be approved for use in a country.The problem is that MAIC assumes the results will be consistent and unbiased once adjustments are made, but this assumption has not been thoroughly tested in real-world settings. Most past studies have only applied MAIC in one direction (matching one treatment group to the other), rather than both ways, which would involve separate reweighting each of the trial populations to match the characteristics of the comparator population, which would allow for comparison of results and assessment of consistency. Additionally, MAIC results have rarely been compared to more robust statistical methods like propensity score weighting, which requires access to patient data from both studies. Propensity score weighting involves assigning each patient in the analysis a score based on their probability of appearing in the opposite arm, therefore giving patients with higher probabilities greater significance in the comparison. This methodology therefore provides a fairer comparison given alignment of both study populations.Study AimThis study will test the reliability of MAIC by first reproducing the results of the published MAIC which reweighted nilotinib-treated patient data to match that of dasatinib-treated patients, and comparing to this to the same comparison if the weighting had been reversed (i.e., dasatinib-treated patients were weighted to match nilotinib-treated patients). The study will therefore make use of real patient data from available studies for nilotinib and dasatinib. It will also compare MAIC results to those from propensity score weighting, which is considered a more accurate method when patient data is available for both treatments. This will help determine how much accuracy is lost when only MAIC can be used.Methods1.   Recreate a past MAIC analysis where nilotinib patient data was matched to dasatinib summary data.2.   Reverse the comparison, matching dasatinib patient data to nilotinib summary data.3.   Apply propensity score weighting using patient data from both trials to create a more balanced comparison.The study will then compare the results across these methods to see if MAIC provides consistent and reliable conclusions.Why This Matters•   Improving confidence in MAIC: If results are consistent across different approaches, this will strengthen MAIC's credibility for HTA decisions.•   Highlighting limitations: If results differ significantly, it may indicate that MAIC should be used with caution.•   Providing best practice guidance: This study will offer insights into when and how MAIC should be applied, helping researchers improve future analyses.Expected Outcomes•   A clearer understanding of whether MAIC consistently produces reliable results.•   Recommendations on how MAIC should be used in HTA and research.•   Findings that could be published in a peer-reviewed journal and contribute to a Master's dissertation.By rigorously testing MAIC against a more robust statistical method, this research will provide valuable insights into the strengths and limitations of one of the most commonly used techniques in treatment comparisons.

[{ "PostingID": 20609, "Title": "NOVARTIS-CAMN107A2303", "Description": "A phase III multi-center, open-label, randomized study of imatinib versus nilotinib in adult patients with newly diagnosed Philadelphia chromosome positive (Ph positive) chronic myelogenous leukemia in chronic phase (CML-CP)." }]