NETWORK META-ANALYSIS, HEOR AND MARKET ACCESS
Network Meta-analysis, HEOR and Market Access Evidence Synthesis
Adapt rigorous evidence synthesis to health technology assessment, payer, reimbursement, and comparative-effectiveness decisions. Plan systematic literature reviews, indirect comparisons, and network meta-analysis around decision-relevant populations, comparators, outcomes, and submission requirements.
Evidence and current methodological guidance reviewed through 9 August 2026.
In HTA and HEOR, the statistical method should follow the decision problem rather than lead it. A network meta-analysis can be technically elegant and still be irrelevant if it compares the wrong treatments, represents the wrong population, uses outcomes at the wrong time point, or rests on cross-trial differences that make indirect inference implausible. The first task is therefore to define the decision that the evidence must inform, then ask what evidence architecture can answer it credibly.1,2,3
This is what distinguishes the methodological territory of health technology assessment and health economics and outcomes research from a page about one statistical technique. HTA is a decision process that appraises health technologies for defined populations and comparators. HEOR is a broader research field that includes clinical, economic, outcomes and real-world evidence. Evidence synthesis is one of the methods that supplies those decisions and analyses. Market access may use the resulting evidence, but it is a decision and organizational context rather than a synthesis method in its own right.1,15
The decision problem comes before the evidence model
A conventional research question can be academically coherent while still missing the decision that an HTA body, payer, guideline group or health system actually faces. NICE formalizes a scope and decision problem that specifies the population, intervention, relevant comparators, outcomes and care pathway. EU Joint Clinical Assessment adds another complication: the assessment scope can contain several PICOs reflecting Member State needs, and the dossier must address them appropriately.2,3
That changes the role of PICO. PICO is not merely a search-framework mnemonic here. The population determines whether trial participants are transferable to the decision population. The comparator determines which treatment links are relevant. Outcome definitions and time points determine whether studies can be synthesized. Treatment line, previous therapy, disease severity and other effect modifiers can determine whether an indirect comparison is credible at all. A difference that looks minor when drafting eligibility criteria can become a structural problem once the evidence network is assembled.
Direct, indirect and network evidence are different architectures
When head-to-head randomized trials directly address the decision-relevant comparison, indirect methods may add little. PBAC guidance, for example, states that an indirect comparison is not usually required when direct randomized trials compare the proposed medicine with the main comparator. But direct evidence is not automatically adequate merely because it exists. Its population, comparator, outcome definitions, follow-up and treatment setting still have to match the decision problem.11
When several trials make the same direct comparison, pairwise meta-analysis can synthesize those effects. When treatments have not been compared head-to-head but share a common comparator, an anchored indirect treatment comparison can preserve the randomized contrasts within each trial, provided cross-trial comparability is plausible. Network meta-analysis extends this structure across multiple treatments and can combine direct and indirect evidence in a connected network. The gain is broader relative-effect estimation; the price is a larger set of assumptions that must remain credible across the network.5,6
Population-adjusted indirect comparisons enter for a different reason. Standard aggregate-data ITC and NMA require sufficient comparability of effect-modifier distributions across trials. If individual participant data are available for at least part of the evidence base, methods such as matching-adjusted indirect comparison (MAIC), simulated treatment comparison (STC), and multilevel network meta-regression (ML-NMR) can attempt to adjust for observed cross-trial differences. They do not remove the need for a defensible estimand, adequate overlap, correctly identified effect modifiers or assumptions about unmeasured differences.7,8
Feasibility is a methodological assessment, not a software check
Before choosing an indirect method, map the evidence. Which interventions form nodes? Which trials connect them? Are the comparator definitions genuinely equivalent? Are outcomes measured on compatible scales and at compatible time points? Which characteristics are plausible treatment-effect modifiers, and how are they distributed across comparisons? A network diagram is useful because it exposes structure, but connectivity by itself says nothing about whether the indirect comparison is valid.5,6
Transitivity is the central conceptual assumption behind valid randomized indirect comparison. In practical terms, the trials should be sufficiently comparable with respect to effect modifiers that the relative treatment effects could be exchanged across the comparisons relevant to the network. Consistency, or coherence, concerns agreement between direct and indirect estimates where both are available. Heterogeneity concerns variation in effects among studies making the same comparison. These ideas are related, but they are not substitutes for each other: a non-significant inconsistency test cannot establish clinical comparability, and statistical homogeneity cannot prove transitivity.5,9
Connected is structural
A common comparator creates a route for randomized indirect evidence. It does not show that populations, treatment definitions, outcomes or effect modifiers are exchangeable.
Comparable is clinical and methodological
Similarity has to be argued from the trial evidence and the decision context. Statistical diagnostics can reveal problems, but they cannot supply missing clinical justification.
Adjustment is conditional
MAIC, STC and related methods can address observed population differences under assumptions. They do not guarantee that unmeasured effect modification has been controlled.
No estimate is a legitimate result
If the evidence cannot support a credible direct, anchored or appropriately adjusted comparison, describing the evidence gap may be more defensible than forcing a numerical estimate.
HTA rules change; the reasoning principles are more durable
Jurisdiction matters because the target comparator and evidence package are not abstract. The EU HTA Regulation has applied since 12 January 2025, with Joint Clinical Assessments now operating for defined categories of medicines and a phased expansion. Current HTACG guidance explicitly addresses direct and indirect comparisons, and its 2025 Q&A states that anchored methods should be preferred when an appropriate common comparator is available. NICE begins from a defined scope and decision problem. PBAC asks submissions to identify relevant trials, map possible indirect links and justify exclusions that may affect transitivity. These requirements differ in detail, but all reinforce the same principle: synthesis methods are judged against a target decision rather than in isolation.1,2,3,11
The page therefore uses jurisdiction examples to make methodological consequences concrete, not to create a country-by-country submission manual. Regulatory and HTA procedures will continue to evolve. The permanent framework is the reasoning that survives those changes: specify the decision, identify the relevant evidence, test feasibility, choose the least assumption-dependent credible comparison, assess uncertainty, and state what the evidence can and cannot support.
Use the resources below at the point where the broad architecture becomes a specific task
The current resource set begins with NMA assumptions, indirect-comparison appraisal and HEOR/HTA literature-review planning. Those are starting points, not the permanent boundary of this page; future resources can extend into population adjustment, evidence feasibility, certainty assessment and jurisdiction-specific planning without changing the hub’s identity.
HTA, HEOR & COMPARATIVE EVIDENCE RESOURCES
Apply comparative-evidence methods to an HTA or HEOR decision
Move from the broad HTA and HEOR evidence architecture into three decision-focused tasks: appraising whether a network meta-analysis is credible, assessing an indirect treatment comparison, and planning a systematic literature review around a real HTA or HEOR decision problem.
Practical comparative-evidence resources for HEOR and market access
Use these resources to examine the credibility of comparative evidence and to configure evidence-generation work around the population, comparators, evidence structure, jurisdiction, and downstream decision the analysis must support.
Network meta-analysis assumptions checklist
Appraise an evidence network before interpreting pooled relative effects. Record network connectivity, clinically important treatment-effect modifiers, transitivity, heterogeneity, direct–indirect agreement, and unresolved assumptions requiring specialist review.
Indirect treatment comparison appraisal checklist and evidence table
Appraise an indirect comparison by documenting its evidence structure, common-comparator integrity, cross-trial population differences, outcome and estimand compatibility, population adjustment where applicable, and residual sources of uncertainty.
HEOR systematic literature review protocol template
Configure a decision-linked evidence program for HTA or HEOR. Map the decision context and comparators, manage multiple PICOs and evidence streams, preserve study and data-cut lineage, plan evidence updates, and capture information needed for later comparative-effectiveness and economic analyses.
Network meta-analysis assumptions: transitivity, consistency and heterogeneity
Indirect treatment comparisons explained: common comparators, assumptions and credibility
How to plan a systematic literature review for HEOR and HTA
Population adjustment can address observed differences, not abolish uncertainty
Population-adjusted indirect comparison is often introduced as the next step when conventional aggregate-data ITC or NMA is threatened by cross-trial differences in effect modifiers. That description is useful only if its limits remain visible. MAIC reweights individual-level data from one trial so selected baseline characteristics match aggregate characteristics reported for another trial. STC uses outcome regression to estimate treatment outcomes in a target population. ML-NMR can combine individual and aggregate information across a network and estimate treatment effects for specified target populations under a model for effect modification.7,8
The anchored-versus-unanchored distinction matters more than the software name. When a valid common comparator exists, anchored population adjustment retains the randomized relative effects within the trials and focuses adjustment on treatment-effect modifiers. When no randomized anchor exists, an unanchored comparison relies on much stronger assumptions because absolute outcomes must also be made comparable. Prognostic variables as well as effect modifiers become consequential, and residual confounding can remain even after extensive adjustment.7,8
Effective sample size is therefore a diagnostic rather than a pass/fail score. Strong MAIC weights can reduce the effective amount of information and signal limited overlap between the individual-level and target populations. A severe reduction can make the estimate imprecise and sensitive to a small number of heavily weighted participants. It does not, by itself, prove that the populations are biologically incomparable. Nor does a comfortable effective sample size prove that all relevant effect modifiers have been observed.
Credibility has more than one layer
For NMA, a model fit statistic is not a complete credibility assessment. Reviewers have to consider risk of bias in the contributing studies, reporting bias, indirectness, imprecision, heterogeneity, transitivity and incoherence. GRADE guidance for NMA and the CINeMA framework approach these judgments differently. GRADE provides a broader certainty framework that has been extended for direct, indirect and network estimates. CINeMA operationalizes confidence assessment across six domains and uses contribution information to help connect problems in the direct evidence to specific network estimates.9,10
Neither framework removes judgment. CINeMA’s contribution matrix makes the flow of evidence more explicit, but decisions about within-study bias, indirectness, imprecision, heterogeneity and incoherence still depend on thresholds and substantive interpretation. Published comparisons have found that CINeMA and GRADE can disagree, which is a reason to state the framework and judgments transparently rather than treating the two labels as interchangeable.10
Evidence synthesis supplies economic decisions; it does not replace economic modeling
HTA and HEOR projects often need several evidence streams. Comparative-effectiveness estimates may feed a cost-effectiveness model. Epidemiological synthesis may inform population size or baseline risks. Utility evidence may help parameterize health states. Resource-use and cost reviews may support inputs that are strongly jurisdiction-dependent. Evidence synthesis can improve the traceability of those inputs, but the validity of an economic model also depends on its structure, assumptions, perspective, time horizon and uncertainty analysis. Those are different methodological tasks.
| Evidence stream | What synthesis can provide | What remains a separate decision/modeling task |
|---|---|---|
| Comparative clinical effects | Relative treatment effects, uncertainty, subgroup or network estimates where defensible | How effects enter the economic model, extrapolation and structural assumptions |
| Baseline risk / epidemiology | Incidence, prevalence, event risks, natural-history evidence | Choice of target population, transition structure and long-term extrapolation |
| Health-state utility | Relevant preference-based estimates and their context | Selection of model health states, mapping assumptions and jurisdictional tariff decisions |
| Resource use and costs | Published utilization and cost evidence, with setting and currency context | Price year, local costing, inflation, perspective and scenario construction |
This boundary also prevents duplication elsewhere in the Academy. Patient-reported outcome measurement and instrument quality belong with the PRO/COSMIN methodology resources. Generic meta-analysis belongs with the quantitative synthesis resources. This page focuses on the point where those evidence streams become part of a decision-relevant HTA or HEOR evidence program.
Universal principles meet jurisdiction-specific evidence requirements
HTA agencies do not apply one identical evidence rulebook. The EU JCA process focuses on relative clinical effectiveness and safety at the joint level while pricing, reimbursement and economic evaluation remain national responsibilities. NICE defines a scope and decision problem, and its Decision Support Unit has developed influential technical support on indirect and population-adjusted comparisons. Germany’s benefit-assessment process centers on the appropriate comparator therapy. Australia’s PBAC explicitly accommodates indirect comparisons through a common reference when direct randomized comparisons are unavailable. Canada’s national HTA organization now operates as Canada’s Drug Agency (CDA-AMC) and publishes current quantitative methods guidance. France’s HAS evaluates indirect evidence within its own clinical-value framework.1,3,11,12,13
| Decision context | What this framework takes from it | What should not be generalized |
|---|---|---|
| EU Joint Clinical Assessment | Several decision-relevant PICOs may need to be addressed; current guidance prefers anchored indirect comparisons when an appropriate randomized network exists. | The EU procedural dossier and certainty language are not universal HTA rules. |
| NICE | Scope and decision problem precede evidence modeling; population adjustment requires explicit target-population and assumption reasoning. | NICE-specific process and economic-evaluation conventions do not automatically apply elsewhere. |
| Germany — IQWiG / G-BA | Comparator relevance can determine whether an otherwise sophisticated evidence synthesis answers the assessment question. | The German legally defined comparator process is jurisdiction-specific. |
| Australia — PBAC | Network mapping, common-reference indirect comparisons and transitivity reasoning are explicit parts of submission guidance. | PBAC submission structure is not a universal reporting standard. |
| Canada — CDA-AMC | Current quantitative methods guidance gives a contemporary HTA context for ITC appraisal and evidence uncertainty. | Canadian reimbursement processes and evidence categories should not be treated as global defaults. |
| France — HAS | Comparative evidence is interpreted against clinical value and the available therapeutic alternatives. | French rating consequences and committee doctrine are specific to France. |
Common errors are usually errors of reasoning before they are errors of calculation
Starting with NMA because the software can run it
First establish whether the target treatments form a decision-relevant, clinically coherent network. Feasibility is not proven by a successful model fit.
Treating connectivity as transitivity
A common comparator links trials structurally. It does not make the distributions of treatment-effect modifiers comparable.
Calling consistency a clinical validity test
Agreement between direct and indirect estimates is informative, but statistical non-detection of incoherence cannot prove the underlying cross-trial assumptions.
Escalating to an unanchored method automatically
Disconnected evidence increases the assumption burden. Sometimes the correct conclusion is that no credible quantitative comparison can be made.
Reading treatment rankings without effect estimates
Ranks can look decisive even when treatment effects are close or uncertain. Magnitude, uncertainty and certainty of evidence should accompany any hierarchy.
Assuming population adjustment guarantees transportability
Adjustment addresses specified observed differences under a model. The target estimand, overlap, unmeasured effect modification and effect-measure properties still matter.
Frequently asked questions
Is network meta-analysis always preferable to pairwise meta-analysis?
No. NMA is useful when the decision requires comparisons across several treatments and the network and assumptions support valid indirect or mixed evidence. If the decision is a single well-supported direct comparison, pairwise synthesis may answer it with fewer assumptions.
Does a connected network prove transitivity?
No. Connectivity shows that a mathematical route for indirect comparison exists. Transitivity is a clinical and methodological assumption about comparability across the trials, especially the distribution of treatment-effect modifiers.
Are transitivity and consistency the same thing?
No. Transitivity is the underlying assumption that permits valid indirect inference. Consistency or coherence concerns agreement between direct and indirect estimates where both are available. Observed consistency can support—but cannot prove—transitivity.
When should MAIC or STC be considered?
They may be considered when individual participant data are available for part of the evidence base and observed cross-trial differences threaten a conventional indirect comparison. Anchored and unanchored analyses have different assumptions; the latter require much stronger control of prognostic and effect-modifying differences.
Does a low effective sample size invalidate a MAIC?
Not automatically. A low effective sample size is a warning that weighting has concentrated information in fewer effective observations, often reflecting poor overlap and increased imprecision. It should be interpreted with balance diagnostics, the target population, model assumptions and sensitivity analyses.
Which is the current NMA reporting guideline?
PRISMA-NMA 2015 remains the current published PRISMA extension for network meta-analysis. An update is in development; the 2025 scoping review identified candidate items for subsequent consensus work, not a finalized replacement checklist.
Is CINeMA a replacement for GRADE?
No universal replacement relationship should be assumed. Both are used to assess confidence or certainty in network evidence, with different operational approaches. State which framework was used and report the judgments transparently.
Does HTA evidence synthesis include economic modeling?
This page treats economic modeling as a downstream decision-analysis task. Evidence synthesis can supply comparative effects, baseline risks, utilities, epidemiology, resource-use and cost evidence, but model structure and cost-effectiveness analysis require separate methods.
Related methodology
A defensible comparison answers the decision before it answers the model
The durable question in HTA and HEOR is not “Which advanced method can we run?” It is “What comparison does the decision require, and what evidence can support that comparison without hiding the assumptions?” Direct evidence, indirect treatment comparison, network meta-analysis and population-adjusted methods are different responses to that problem.
Good evidence synthesis makes the chain inspectable: target decision, evidence search, network or comparison structure, effect modifiers, chosen estimand, model assumptions, uncertainty and applicability. When that chain breaks, more modeling does not restore the missing evidence. Sometimes the most useful conclusion is a clearly described evidence gap.
Get the Resource Infrastructure Free Forever
One sign-up grants lifetime access to every template, workbook, and methodology guide on the MetaSyn Resource Infrastructure, current and future. As new resources release, you receive them by email automatically. No spam, no marketing noise, methodology updates from a working meta-analyst, only when there’s something real to share.
References
- European Commission. Implementation of Regulation (EU) 2021/2282 on health technology assessment. Current implementation information; Regulation applies from 12 January 2025. European Commission HTA implementation.
- Member State Coordination Group on Health Technology Assessment. Questions and answers on general methodological and procedural issues for joint clinical assessments. 28 November 2025. HTACG JCA methodological Q&A.
- National Institute for Health and Care Excellence. NICE technology appraisal and highly specialised technologies guidance: the manual (PMG36). Current online manual. NICE PMG36.
- Hutton B, Salanti G, Caldwell DM, et al. The PRISMA Extension Statement for Reporting of Systematic Reviews Incorporating Network Meta-analyses of Health Care Interventions: checklist and explanations. Ann Intern Med. 2015;162(11):777-784. https://doi.org/10.7326/M14-2385
- Jansen JP, Fleurence R, Devine B, et al. Interpreting indirect treatment comparisons and network meta-analysis for health-care decision making: report of the ISPOR Task Force on Indirect Treatment Comparisons Good Research Practices: Part 1. Value Health. 2011;14(4):417-428. PubMed.
- Hoaglin DC, Hawkins N, Jansen JP, et al. Conducting indirect-treatment-comparison and network-meta-analysis studies: report of the ISPOR Task Force on Indirect Treatment Comparisons Good Research Practices: Part 2. Value Health. 2011;14(4):429-437. PubMed.
- Phillippo DM, Ades AE, Dias S, Palmer S, Abrams KR, Welton NJ. Methods for population-adjusted indirect comparisons in health technology appraisal. Med Decis Making. 2018;38(2):200-211. https://doi.org/10.1177/0272989X17725740
- Phillippo DM, Dias S, Elsada A, Ades AE, Welton NJ. Population adjustment methods for indirect comparisons: a review of National Institute for Health and Care Excellence technology appraisals. Int J Technol Assess Health Care. 2019;35(3):221-228. PubMed.
- Puhan MA, Schünemann HJ, Murad MH, et al. A GRADE Working Group approach for rating the quality of treatment effect estimates from network meta-analysis. BMJ. 2014;349:g5630. https://doi.org/10.1136/bmj.g5630
- Nikolakopoulou A, Higgins JPT, Papakonstantinou T, et al. CINeMA: an approach for assessing confidence in the results of a network meta-analysis. PLoS Med. 2020;17(4):e1003082. https://doi.org/10.1371/journal.pmed.1003082
- Pharmaceutical Benefits Advisory Committee. PBAC Guidelines: Section 2 — Clinical evaluation; Appendix 3 — Identify relevant trials. Australian Government. PBAC indirect-comparison guidance.
- Canada’s Drug Agency (CDA-AMC). Methods Guide for Health Technology Assessment. Current quantitative methods guidance. CDA-AMC Methods Guide.
- Haute Autorité de Santé. Doctrine de la Commission de la Transparence. Current public doctrine page. HAS Transparency Committee doctrine.
- Veroniki AA, Tricco AC, Rangira D, et al. Updating the PRISMA reporting guideline for network meta-analysis: a scoping review. J Clin Epidemiol. 2025;188:111985. https://doi.org/10.1016/j.jclinepi.2025.111985
- Jansen JP, Trikalinos T, Cappelleri JC, et al. Indirect treatment comparison/network meta-analysis study questionnaire to assess relevance and credibility to inform health care decision making: an ISPOR-AMCP-NPC Good Practice Task Force report. Value Health. 2014;17(2):157-173. PubMed.