Free scoping read. A written assessment of your manuscript in two business days, at no cost.

Start one
Contents

AcademyMeta Analysis

Guide

Network Meta-Analysis and Indirect Treatment Comparison in the US: A Guide for Comprehensive Meta-Analysis (CMA) Software Users

Network meta-analysis and indirect treatment comparison let you compare treatments never tested head to head. This guide shows what Comprehensive Meta-Analysis (CMA) software can and cannot do, with a step-by-step workflow and a worked example.

Dr. CharlieSenior Medical WriterPublishedReading time5 minutesSources5
  • Meta-analysis
  • Network meta-analysis
  • Statistics
  • Software

At a glanceMeta Analysis

ITC
Compares two treatments through a shared comparator
NMA
Combines direct and indirect evidence across many treatments
CMA software
Pairwise analyses; Bucher ITC by manual calculation
Reporting
PRISMA-NMA for full networks

Talk to a Pubrica expertDiscuss your project

Quick answer

  • An adjusted indirect comparison estimates A vs C from A vs B and C vs B trials that share comparator B.
  • Comprehensive Meta-Analysis (CMA) software runs the two pairwise analyses; you combine them by hand: d(AC) = d(AB) − d(CB).
  • CMA does not document a dedicated NMA module, so use specialist software such as R’s netmeta for closed networks, and confirm with the vendor.
  • Check transitivity and consistency, report confidence intervals, and follow PRISMA-NMA for full networks.

Why indirect comparisons matter

Head-to-head trials rarely cover every treatment a US clinician or payer must compare. Network meta-analysis (NMA) and indirect treatment comparison (ITC) fill that gap by linking trials through a common comparator, so two drugs that were each tested against placebo, for example, can still be compared with each other. This guide explains the core methods, shows where Comprehensive Meta-Analysis (CMA) software fits, and walks through a practical workflow for researchers preparing evidence for US decision-makers. Throughout, CMA means the commercial meta-analysis program, not cost-minimisation analysis.

Indirect evidence in US decision-making

US formulary committees, health systems and manufacturers often need relative effectiveness estimates for drugs that have never been tested against each other. To help these readers, a joint task force of ISPOR, the Academy of Managed Care Pharmacy (AMCP) and the National Pharmaceutical Council (NPC) developed a 26-item questionnaire for judging whether an ITC or NMA is relevant and credible enough to inform health care decisions1. Its relevance questions ask whether the results apply to the decision-maker’s setting; its credibility questions cover the evidence base, analysis methods, reporting, interpretation and conflicts of interest1.

In practice, a reviewer reading your meta-analysis asks two things: do the included trials resemble the patients you want to treat, and are the statistical methods transparent? A well-reported ITC answers both. A poorly justified one is easy to set aside.

Choosing the right method

Method choice matters as much as software choice. The simplest option is the adjusted indirect comparison described by Bucher and colleagues, in which two treatments that share one common comparator are compared through their effects against that comparator. Because the comparison works on within-trial differences, it preserves the randomization of each trial, unlike naive comparisons of single treatment arms2. Full NMA extends the idea to many treatments at once, combining direct and indirect evidence across a connected network3.

When trial populations differ in effect modifiers, standard methods can mislead. Population-adjusted approaches such as matching-adjusted indirect comparison (MAIC) and simulated treatment comparison use individual patient data from a subset of trials to adjust for those differences4. Comparisons anchored on a common comparator arm rest on weaker assumptions than unanchored ones, which are widely regarded as infeasible4.

Table 1. Indirect comparison methods and what CMA can do

MethodBest forCMA support
Bucher adjusted ITCTwo treatments, one common comparatorPartial (pairwise analyses in CMA, then a manual calculation)
Network meta-analysisMultiple treatments, closed loopsNo dedicated module documented; use specialist software
MAIC and other population-adjusted methodsDiffering trial populationsNo; needs individual patient data and specialist software

Need help with an indirect comparison or network meta-analysis?

Pubrica’s biostatisticians and medical writers design evidence networks, run and check indirect comparisons, and prepare PRISMA-NMA-compliant reports for journals and US decision-makers.

Where CMA software fits

If your team already licenses CMA, it is a convenient starting point. CMA is built for pairwise meta-analysis. Its published features include a spreadsheet-style data interface, effect-size calculation from many data formats, meta-regression and publication-bias tools, and recent versions report prediction intervals; check your version’s documentation for exactly what it provides. Its feature descriptions do not mention a dedicated NMA module, so confirm with the vendor before you plan an analysis around it.

CMA suits the classic Bucher comparison well: you run the two pairwise analyses in CMA and combine the results by hand2. Networks with several treatments or closed loops need dedicated tools, such as the netmeta package in R or Bayesian software, which are designed to estimate and check consistency across the whole network3. If you are unsure which design you have, our meta-analysis type decision tree can help.

Step-by-step adjusted ITC in CMA

The workflow below covers the common case of treatment A and treatment C, each tested in randomized trials against a shared comparator B, with no trial comparing A and C directly. The evidence network is small enough to describe in a table.

Table 2. The evidence network in this example: three treatments linked through comparator B

ComparisonEvidenceHow it is estimated
A vs BRandomized trialsPairwise meta-analysis in CMA
C vs BRandomized trialsPairwise meta-analysis in CMA
A vs CNo head-to-head trialsIndirect comparison through B

Illustrative example. Suppose the A vs B analysis gives a pooled log odds ratio, ln(OR), of −0.40 (SE 0.12) and the C vs B analysis gives −0.25 (SE 0.15). The indirect A vs C effect is −0.40 − (−0.25) = −0.15. Its standard error is √(0.12² + 0.15²) = √0.0369 = 0.19. Back-transformed, that is an OR of 0.86 with a 95% confidence interval of 0.59 to 1.25. The interval crosses 1, so this indirect comparison shows no clear difference between A and C. Notice that the indirect SE is larger than either pairwise SE: indirect estimates add variances, so they are always less precise than the direct evidence they come from2. These numbers are invented for teaching and do not describe real drugs.

Adjusted ITC workflow

  • Define the PICO and pre-register the protocol, for example on PROSPERO, naming the comparator, outcomes and effect measure.
  • Extract arm-level data from randomized trials only, ideally in duplicate.
  • Run two random-effects analyses in CMA (A vs B and C vs B) using the same effect measure, such as the log odds ratio.
  • Record each pooled effect and its standard error (SE).
  • Calculate the indirect effect, d(AC) = d(AB) − d(CB), with SE = √(SE²(AB) + SE²(CB)), then back-transform the estimate and its confidence interval.
  • Review heterogeneity (I², prediction interval) and publication bias in each pairwise analysis before reporting.

Interpreting and reporting results

Report the indirect estimate with its confidence interval, not just a p-value. A non-significant result is not proof of equivalence, especially when the interval is as wide as in the example above. State the effect measure, the model (fixed or random effects), heterogeneity statistics and every assumption you relied on, and keep a decision log so others can reproduce your results.

Where a closed loop exists, compare direct and indirect estimates. For a full network, follow PRISMA-NMA, the 32-item extension of the PRISMA statement for systematic reviews that incorporate network meta-analyses5. Be cautious with treatment rankings: rankings built from point estimates alone can look more certain than the underlying evidence supports, particularly when the network contains small trials. The questionnaire from the ISPOR-AMCP-NPC task force is a useful final check, because it mirrors what a critical reader will ask1.

Assumptions and pitfalls

Every indirect comparison stands or falls on its assumptions. NMA relies on agreement between direct and indirect evidence; when that agreement fails, the network estimates can mislead, so consistency checks are essential rather than optional3. Standard aggregate-data methods also assume that effect-modifying variables are distributed similarly across trials4. If a network has closed loops or several treatments, plan for specialist software and a biostatistician from the start rather than forcing CMA to do the job.

Check these assumptions

Transitivity: trials must be similar in populations, doses, comparators and outcome definitions, or the common comparator does not link like with like4. Consistency: compare direct and indirect estimates wherever both exist3. Precision: indirect estimates add variances, so expect wider intervals than direct ones2. Transparency: document the search, selection and every modelling choice5.

Pubrica team verdict

Comprehensive Meta-Analysis (CMA) software is a practical, transparent tool for simple adjusted indirect comparisons, but full network meta-analyses and population-adjusted methods need specialist software. Explicit assumptions, transparent reporting and honest uncertainty matter more than the program you use.

Frequently asked questions

CMA is built for pairwise meta-analysis, and its published feature descriptions do not mention a dedicated network meta-analysis module. You can use it for a simple Bucher adjusted indirect comparison by combining two pairwise results by hand. For closed networks, use specialist software and confirm current features with the vendor.

An indirect treatment comparison usually estimates one comparison, such as A vs C, through a single common comparator. A network meta-analysis analyses many treatments together and combines direct and indirect evidence across the whole network.

The variance of an indirect estimate is the sum of the variances of the two pairwise estimates, so it is always less precise than either one. A wide interval that crosses the line of no effect is not evidence that the treatments are equivalent.

Transitivity is the assumption that the trials being linked are similar enough, in populations, doses, comparators and outcome definitions, for the common comparator to connect like with like. If effect modifiers differ between trials, the indirect estimate can be biased.

PRISMA-NMA, the extension of the PRISMA statement for systematic reviews incorporating network meta-analyses, published in 2015 with a 32-item checklist.

Sources

Sources were checked on 19 November 2026. Requirements and guidance may change after that date; confirm the current version with the issuing body.

  1. 1Jansen JP, Trikalinos T, Cappelleri JC, Daw J, Andes S, Eldessouki R, 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):157-73. https://doi.org/10.1016/j.jval.2014.01.004
  2. 2Bucher HC, Guyatt GH, Griffith LE, Walter SD. The results of direct and indirect treatment comparisons in meta-analysis of randomized controlled trials. J Clin Epidemiol. 1997;50(6):683-91. https://doi.org/10.1016/s0895-4356(97)00049-8
  3. 3Efthimiou O, Debray TP, van Valkenhoef G, Trelle S, Panayidou K, Moons KG, et al. GetReal in network meta-analysis: a review of the methodology. Res Synth Methods. 2016;7(3):236-63. https://doi.org/10.1002/jrsm.1195
  4. 4Phillippo 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):200-11. https://doi.org/10.1177/0272989X17725740
  5. 5Hutton B, Salanti G, Caldwell DM, Chaimani A, Schmid CH, Cameron C, 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):777-84. https://doi.org/10.7326/M14-2385

network meta-analysisindirect treatment comparisonComprehensive Meta-Analysis (CMA) softwareCMA softwareadjusted indirect comparisonBucher method

How to cite this article

Pick your referencing style and copy it. Your university or target journal may vary the punctuation, so check its guide if it has one.

Dr. Charlie. (2026, November 19). Network Meta-Analysis and Indirect Treatment Comparison in the US: A Guide for Comprehensive Meta-Analysis (CMA) Software Users. Pubrica. https://pubrica.com/academy/meta-analysis/network-meta-analysis-indirect-treatment-comparison-cma/

This article is written to be learned from. Cite it if you are referring to what it says; if you are using the evidence behind it, cite the sources listed above instead.

Get a free quote

Regulations

Check your project against this rule

  1. 1Your work
  2. 2Your details
  3. 3Send
What do you need?

Material you send is held under signed confidentiality, with access limited to the assigned team and encrypted transfer. It is not used to train anything and is not passed to third-party tools. Ask for the agreement before you send anything you would rather cover first.