Network Meta-Analysis (NMA)

Network meta-analysis (NMA) is a statistical technique that estimates the comparative effectiveness of three or more interventions within a single unified model by combining two kinds of evidence: direct evidence, from trials that compared two of the interventions head-to-head, and indirect evidence, inferred through one or more common comparators (such as a shared placebo or standard-of-care arm) that connect trials which never compared those two interventions directly. A body of evidence qualifies as an NMA, rather than a standard pairwise meta-analysis, when it (1) includes three or more interventions across the included trials, (2) links those trials into a connected network via shared comparators, and (3) produces relative-effect estimates for treatment pairs even when no trial tested that pair directly. NMA rests on an assumption called transitivity — that trials contributing indirect evidence are similar enough in patient population, setting, and other effect modifiers that comparisons through a common comparator are valid — and its statistical counterpart, consistency, which checks that direct and indirect estimates for the same pair of interventions agree.

ByCASRAI Editorial Board

· Last updated 23 Jul 2026

Ask about Network Meta-Analysis (NMA)

Answers are drawn from this dictionary entry and the rest of the CASRAI corpus, with a link to every source.

Answers are AI-generated from CASRAI’s own published pages and can be wrong, so check the linked sources before relying on one; your question is logged without personal data — never sold, never used to train a third-party model — to show us what CASRAI is missing, so please do not type personal or confidential details. How we use this

Worked examples

Is an instance

A review of treatments for a chronic condition where Drug A has been trialed against placebo, Drug B against placebo, and Drug C against Drug A, but Drug B and Drug C have never been compared head-to-head. NMA uses placebo as the common comparator to estimate the indirect B-vs-C effect within the same model as the direct comparisons, and can rank all three drugs against each other using a single, internally consistent set of estimates.

Is an instance

Health technology assessment (HTA) bodies, including the UK's National Institute for Health and Care Excellence (NICE), routinely commission network meta-analyses when appraising a new therapy against several existing treatment options for the same condition, since head-to-head trials rarely exist for every possible pair of competing treatments.

Looks similar, but isn't

Not an instance

A meta-analysis that pools trial results comparing only two interventions (e.g., Drug A vs. placebo across five trials) is a standard pairwise meta-analysis, not an NMA, regardless of whether it uses a fixed-effect or random-effects model — it has no indirect-evidence structure and involves only one comparison, not a network of interventions.

Network meta-analysis (NMA) — also called multiple-treatments meta-analysis or mixed-treatment comparison — extends conventional pairwise meta-analysis to situations where three or more interventions are of interest and not every pair of them has been tested against each other in a randomized controlled trial. Rather than running a separate pairwise meta-analysis for each pair that happens to have direct trial evidence, NMA combines all the trials into a single statistical model, using shared comparators (often placebo or a standard-of-care treatment) to bridge trials that never compared two treatments directly.

How NMA differs from pairwise meta-analysis

A standard pairwise meta-analysis, as covered in CASRAI’s Systematic Review vs. Meta-Analysis comparison, statistically combines results from two or more studies that all compared the same two interventions (or an intervention against a common control). NMA is a distinct, more advanced technique layered on top of that same underlying logic of pooling effect sizes: it requires a connected network of trials spanning three or more interventions, and it produces estimates for treatment pairs that were never directly compared in any included trial, by routing the comparison through one or more common comparators. The Cochrane Handbook for Systematic Reviews of Interventions (Chapter 11) describes NMA as generalizing pairwise meta-analysis to allow simultaneous comparison of multiple interventions within one coherent analysis, rather than a series of disconnected pairwise pools. See CASRAI’s Cochrane Handbook entry for the broader methodological framework this sits within.

Direct evidence, indirect evidence, and the network

In NMA terminology, a network is typically visualized as a diagram of nodes (interventions) connected by edges (direct head-to-head comparisons available in the trial evidence). Direct evidence comes from trials that compared two interventions against each other. Indirect evidence is derived statistically, by combining the direct evidence for A-vs-C and B-vs-C to estimate the A-vs-B effect, even though no trial ever randomized patients directly to A vs. B. A network is only analyzable as an NMA if it is connected — every intervention needs to be linked, directly or indirectly, to every other intervention through some chain of shared comparators.

Transitivity and consistency

NMA’s validity depends on an assumption called transitivity: that the trials contributing indirect evidence are similar enough in patient characteristics, dosing, follow-up duration, and other effect modifiers that it is reasonable to route a comparison through a shared comparator. The statistical check for whether this assumption holds is called consistency — testing whether direct and indirect estimates for the same treatment pair, where both exist, actually agree. A significant direct-indirect disagreement (inconsistency) signals that the network may not be safely combinable and undermines confidence in the pooled NMA estimates.

What NMA output looks like

Beyond pairwise effect estimates for every intervention pair in the network, NMA is commonly used to produce a ranking of all included interventions from most to least effective (or safest), often summarized with a ranking metric such as SUCRA (surface under the cumulative ranking curve). These rankings are a distinguishing practical output of NMA that a standard pairwise meta-analysis, comparing only two interventions, cannot produce.

Where NMA is used

NMA is widely used in evidence synthesis for clinical guideline development and health technology assessment (HTA), where decision-makers need to compare several available treatment options for the same condition but head-to-head trials do not exist for every pair. HTA bodies use NMA results, alongside direct trial evidence, to inform reimbursement and treatment-guideline decisions when a fully head-to-head evidence base is unrealistic to generate. A dedicated reporting extension, PRISMA-NMA (Hutton et al., Annals of Internal Medicine, 2015), sets out checklist items specific to reporting systematic reviews that incorporate a network meta-analysis, extending the base PRISMA reporting standard with network-specific items such as reporting the network diagram, describing how the network was assembled, and reporting any assessment of inconsistency.

Related CASRAI resources

Systematic Review vs. Meta-Analysis — the foundational comparison NMA builds on
Forest Plot
Effect Size
Cochrane Handbook for Systematic Reviews of Interventions
Research Study Types

Use in your systems

JATS XML <role> element

xml

<role vocab="credit"
      vocab-identifier="https://casrai.org/dictionary/"
      vocab-term="Network Meta-Analysis (NMA)"
      vocab-term-identifier="https://casrai.org/dictionary/term/network-meta-analysis" />

Schema.org DefinedTerm (JSON-LD)

json

{
  "@context": "https://schema.org",
  "@type": "DefinedTerm",
  "@id": "https://casrai.org/dictionary/term/network-meta-analysis",
  "name": "Network Meta-Analysis (NMA)",
  "identifier": "https://casrai.org/dictionary/term/network-meta-analysis",
  "description": "Network meta-analysis (NMA) is a statistical technique that estimates the comparative effectiveness of three or more interventions within a single unified model by combining two kinds of evidence: direct evidence, from trials that compared two of the interventions head-to-head, and indirect evidence, inferred through one or more common comparators (such as a shared placebo or standard-of-care arm) that connect trials which never compared those two interventions directly. A body of evidence qualifies as an NMA, rather than a standard pairwise meta-analysis, when it (1) includes three or more interventions across the included trials, (2) links those trials into a connected network via shared comparators, and (3) produces relative-effect estimates for treatment pairs even when no trial tested that pair directly. NMA rests on an assumption called transitivity — that trials contributing indirect evidence are similar enough in patient population, setting, and other effect modifiers that comparisons through a common comparator are valid — and its statistical counterpart, consistency, which checks that direct and indirect estimates for the same pair of interventions agree.",
  "inDefinedTermSet": "https://casrai.org/dictionary/domain/clinical-research#set",
  "url": "https://casrai.org/dictionary/term/network-meta-analysis",
  "sameAs": [],
  "license": "https://creativecommons.org/licenses/by/4.0/",
  "publisher": {
    "@id": "https://casrai.org/#organization"
  },
  "dateModified": "2026-07-23T11:41:42",
  "inLanguage": "en"
}