maic
The maic R package helps compare treatments indirectly by adjusting for differences in patient characteristics across studies, making results more reliable and reproducible.
At a glance
Use when
Comparing treatments without direct comparative trials, especially when individual patient data is available for one intervention and only aggregate data for others.
Avoid when
When there is insufficient overlap in patient characteristics across studies or when key effect modifiers are not reported in aggregate data.
Inputs
Individual patient data from one trial, aggregate summary data from external studies, including means, standard errors, and effect estimates.
Outputs
Adjusted treatment effect estimates, weighting parameters, standard errors, and uncertainty intervals from indirect comparisons.
How it works
An R package implementing Matching-Adjusted Indirect Comparison (MAIC) methods to enable indirect treatment comparisons using individual patient data from one trial and aggregate data from another, with weighting to adjust for cross-trial differences.
- HTA domains
- Clinical Effectiveness, Costs & Economic Evaluation
- Categories
- Evidence SynthesisIndirect Comparisons
- Assumptions
- The method assumes no unmeasured effect modifiers, correct specification of the weighting model, and that overlap exists in the covariate distributions across studies.
- Strengths
- Enables use of individual patient data to improve validity of indirect comparisons; supports reproducible research through open-source R implementation; reduces bias from cross-trial differences.
- Limitations
- Relies on availability of individual patient data from at least one study; sensitive to model specification and lack of covariate overlap; cannot adjust for unreported or unmeasured variables.
- Also known as
- Matching-Adjusted Indirect Comparison R package
Questions this answers
- › How can we compare the effectiveness of two treatments when no head-to-head trials exist?
- › How can individual patient data be used to adjust for differences in baseline characteristics across studies?
- › What is the adjusted treatment effect of intervention A versus B using indirect comparison?
- › How robust are indirect comparisons after adjusting for confounding variables?
- › Can we generate unbiased estimates in network meta-analysis with limited data?
- › How can we improve reproducibility in indirect treatment comparisons?
References & sources
Related methods
Similar by meaning
Beta record. Based on the original catalogue summary; primary-source enrichment pending.