Matching-Adjusted Indirect Comparison (MAIC)
MAIC is a statistical method used to compare treatments when they haven't been directly tested against each other in the same clinical trial. It adjusts for differences in patient characteristics across studies by using detailed individual patient data from one trial and matching it to summary data from another, making the comparison more fair and relevant for a specific patient population.
At a glance
Use when
Comparing treatments across trials with different patient populations, especially when individual patient data is available for one but not all trials, and when effect-modifying variables differ between studies.
Avoid when
Key effect modifiers are unknown or unreported, individual patient data is of poor quality, or when conducting unanchored indirect comparisons without a common comparator.
Inputs
Individual patient data (IPD) from one trial, aggregate data (AD) from one or more external trials, including means and distributions of effect-modifying covariates in the target population.
Outputs
Adjusted treatment effect estimates (e.g., hazard ratios, odds ratios) comparing interventions in a common target population, with associated uncertainty intervals.
How it works
Matching-Adjusted Indirect Comparison (MAIC) is a population-adjusted indirect comparison method that uses individual patient data (IPD) from one trial and aggregate data (AD) from another to estimate relative treatment effects in a common target population. It weights the IPD to match the covariate distribution of the AD population, thereby adjusting for imbalances in effect-modifying variables. MAIC relies on the assumption of no unmeasured effect modification and is typically used in anchored indirect comparisons (with a common comparator). The method employs inverse propensity score weighting to achieve balance, enabling valid indirect treatment comparisons in health technology assessment where direct head-to-head trials are unavailable.
- HTA domains
- Clinical Effectiveness
- Assumptions
- All effect-modifying variables are measured and included in the model; no unmeasured confounding; correct specification of the weighting model; availability of individual patient data from at least one trial; and exchangeability of trials conditional on observed covariates.
- Strengths
- Enables indirect comparisons in the absence of direct head-to-head trials,Adjusts for differences in patient populations using individual patient data,Improves validity of comparisons in health technology assessment submissions,Provides transparent and reproducible weighting-based adjustment,Widely accepted by reimbursement agencies like NICE when applied appropriately
- Limitations
- Relies heavily on the availability and quality of individual patient data,Vulnerable to bias if key effect modifiers are unmeasured or poorly reported,Increased uncertainty due to weighting can lead to wider confidence intervals,Assumptions are difficult to validate empirically,Results may not be generalizable if the target population differs substantially from trial populations
- Also known as
- MAIC
Questions this answers
- › How does a treatment perform compared to a competing therapy in a specific target population when no direct trial exists?
- › What is the relative effectiveness of two treatments after adjusting for differences in patient characteristics across trials?
- › Can we make a fair comparison between treatments when only one has individual patient data available?
- › How can we account for effect-modifying variables in indirect treatment comparisons?
- › Is a new treatment likely to be more effective than an existing alternative in a real-world population with different baseline characteristics?
- › What are the expected outcomes when comparing treatments across trials with non-overlapping populations?
References & sources
Related methods
Similar by meaning
- Parametric G-computation for indirect treatment comparison
- Population-adjusted indirect comparisons for NICE submissions (NICE DSU TSD 18)
- Conducting Indirect-Treatment-Comparison and Network-Meta-Analysis Studies
- multinma R package
- ISPOR-AMCP-NPC Questionnaire for Indirect Treatment Comparisons/Network Meta-Analysis
Beta record. Generated from the primary source via AI extraction and independent audit, pending final human review.