Network meta-analysis, with reviewer-owned assumptions

When to use NMA

Network meta-analysis is most useful when a review compares more than two interventions and decision-makers need to understand how the options relate across both direct and indirect evidence. It is common in HTA, guideline, comparative-effectiveness, and treatment-selection questions where pairwise comparisons alone do not answer the full decision problem.

• Use NMA when the review question is explicitly comparative across multiple interventions.
• Keep the network tied to a single outcome, timepoint, and compatible effect measure.
• Do not use NMA to paper over incompatible populations, interventions, comparators, or outcome definitions.

User workflow

1. Enable NMA only for a comparative question with multiple interventions.
2. Confirm that extracted arms, comparators, and timepoints are harmonised.
3. Review treatment-node definitions, including dose, regimen, route, and combination-treatment choices.
4. Run readiness checks before interpreting the network.
5. Inspect model outputs, warnings, ranking summaries, and certainty draft rows.
6. Document assumptions and final certainty judgements before reporting.

What Axelium helps you review

Axelium keeps the NMA workflow connected to the review record. The goal is not to make treatment-network assumptions invisible; it is to make them explicit enough for reviewers, statisticians, and decision-makers to inspect.

• Treatment-node definitions and arm harmonisation choices.
• Network connectivity before model interpretation.
• Multi-arm study warnings and other readiness checks.
• Model outputs alongside the studies and extracted values that support them.
• Certainty-review rows that stay separate from treatment rankings.

Assumptions to review

• Connected network. Treatments must connect through direct or indirect comparisons. A disconnected network cannot support a single coherent NMA estimate.
• Transitivity. Studies comparing different treatment pairs should be similar enough in effect modifiers for indirect comparisons to be meaningful.
• Consistency. Direct and indirect evidence should not show unexplained conflict. Treat inconsistency warnings as prompts for methodological review, not automatic verdicts.
• Multi-arm studies. Multi-arm trials need careful checking so shared comparators and correlated evidence are handled appropriately.
• Reference treatment. Choose and document the reference node before presenting relative effects or league tables.

Measures and data readiness

NMA works best when studies report compatible comparative effect measures, timepoints, and outcome definitions. Keep the NMA measure subset narrower than the full pairwise-measure catalogue unless your protocol and data support the broader choice.

Ranking is not certainty

Outputs to inspect

• Network graph. Check whether the evidence network reflects the treatments and comparisons in the protocol.
• League table. Read relative effects with their uncertainty, not just the point estimates.
• Ranking summaries. Treat rankings as model summaries that require clinical and certainty interpretation.
• Diagnostics and warnings. Investigate inconsistency, sparse evidence, and readiness warnings before relying on results.

Certainty review

Pairwise GRADE should not be forced onto network estimates. For clinical NMA, Axelium prepares certainty-review rows that are suitable for a CINeMA-style review process, while final certainty remains a reviewer decision.

Reporting responsibly

Report the review question, eligibility criteria, treatment-node definitions, effect measure, reference treatment, network structure, model outputs, diagnostics, and certainty judgements together. If the evidence network is sparse, disconnected, or inconsistent, document those limitations rather than presenting a ranking as a stand-alone answer.