A network meta-analysis (NMA), also known as a mixed-treatment comparison or indirect comparison, is a statistical method that allows for the simultaneous comparison of multiple treatments—even when no direct head-to-head studies exist between all treatments. By evaluating direct and indirect evidence from a network of studies at the same time, the network meta-analysis provides rankings and relative effect estimates for all compared therapies. It is thus a central tool in evidence-based medicine, Health Technology Assessment (HTA), and benefit assessment according to Section 35a SGB V in Germany, where indirect comparisons are explicitly permitted in the absence of direct evidence.
Methodological Foundations
Network meta-analysis extends classic pairwise meta-analysis by allowing for the inclusion of indirect comparisons. An indirect comparison between treatment A and treatment C becomes possible if both treatments have each been directly tested against a common reference treatment B. The combination of direct and indirect evidence takes place in a statistical model that requires the so-called transitivity assumption: the compared studies must be comparable in terms of clinical and methodological moderators. The statistical model is typically implemented as a Bayesian or frequentist mixed model. Bayesian approaches allow for the direct calculation of probabilities for treatment rankings and are particularly common in complex networks with many treatments. Frequentist approaches based on mixed regression models offer an alternative and are easier to replicate, as they do not require prior distributions. In regulatory practice, both approaches are accepted provided they are transparently documented.
Areas of Application and Use in the HTA Context
Network meta-analyses are used in various contexts:
- Health Technology Assessment (HTA): Reimbursement authorities such as the G-BA in Germany, NICE in the UK, or HAS in France increasingly require NMAs as a basis for benefit assessments when direct comparative evidence is lacking.
- Guideline Development: Medical professional societies use NMAs to create evidence-based therapy recommendations for entire classes of substances. The European Society of Cardiology, the German Cancer Society, and other professional associations regularly base their therapy algorithms on NMA results.
- Price Negotiations and Market Access: Pharmaceutical companies submit NMAs to demonstrate added benefit compared to the appropriate comparator therapy.
- Clinical Trial Planning: NMAs can summarize the state of information for an indication and help identify the optimal comparator therapy for a new study. Furthermore, they make it possible to identify gaps in the evidence and plan priority head-to-head studies.
Quality Requirements and Limitations
The validity of a network meta-analysis depends decisively on the quality of the included primary studies and the fulfillment of the transitivity assumption. A systematic literature search according to PRISMA guidelines is a prerequisite. The risk of inconsistency—i.e., deviations between direct and indirect evidence—must be statistically tested and reported. Frequently used quality instruments include the GRADE methodology for assessing the quality of evidence and the PRISMA-NMA reporting statement for transparent reporting. Important limitations include susceptibility to publication bias, lack of individual patient data, and heterogeneous endpoint definitions between the included studies. To minimize publication bias, the literature search should be carried out systematically in several databases and through manual searches in study registries such as ClinicalTrials.gov or the EU Clinical Trials Register.
Relevance for clinical trials
Network meta-analyses are now an indispensable tool for the benefit assessment and market access of innovative medicinal products. Particularly when only a single-arm study or a limited study program is available, a well-documented NMA can provide indirect evidence of added benefit compared to the appropriate comparator therapy. Full-service CROs like mediconomics support sponsors in the planning, execution, and regulatory-compliant reporting of network meta-analyses for HTA submissions to the G-BA, NICE, and other reimbursement authorities. This also includes the preparation of NMA results for early benefit assessment according to Section 35a SGB V as well as preparation for inquiries from the G-BA regarding methodology.
Frequently Asked Questions (FAQ)
What is the difference between a pairwise meta-analysis and a network meta-analysis?
A pairwise meta-analysis always compares only two treatments directly with each other and summarizes the results of all studies that have investigated this comparison. A network meta-analysis allows for the simultaneous comparison of more than two treatments by combining direct and indirect evidence. It provides a complete ranking of all compared therapies as well as effect estimates for all treatment pairs.
What is the transitivity assumption and why is it important?
The transitivity assumption states that the patients, study designs, and outcome definitions in all studies in the network are so similar that an indirect comparison is meaningful. If this assumption is violated—for example, because studies come from different eras with very different patient populations—the indirect comparisons may be biased. Verification of the transitivity assumption is therefore mandatory and must be documented in the NMA methodology protocol.
Does the G-BA accept network meta-analyses as a basis for evidence?
Yes, under certain conditions. The G-BA accepts indirect comparisons and NMAs if no direct study comparison with the appropriate comparator therapy is available and the NMA meets the PRISMA-NMA methodological requirements. However, the quality of evidence for indirect comparisons is generally rated lower than direct randomized data, which can affect the classification of the added benefit.
Regulatory References
- PRISMA-NMA: Preferred Reporting Items for Systematic Reviews and Meta-Analyses including NMAs (2015)
- EMA Guideline on the Investigation of Subgroups in Confirmatory Clinical Trials (EMA/CHMP/539146/2013)
- G-BA Rules of Procedure Section 35a SGB V: Requirements for indirect comparisons and NMAs
- NICE Decision Support Unit Technical Support Document 2: Evidence Synthesis for NMA (2011, updated)
- ICH E9(R1) – Addendum on Estimands (2019): Consistency of meta-analysis methods with estimands