Analysis configuration
Load a previous session does not produce results
Upgrade old data does not produce results
exclude does not produce results
Network characteristics
Treatments
All studies
With selected studies excluded
Treatment pairs
All studies
With selected studies excluded
Numbers on the line indicate the number of trials conducted for the comparison. The shaded areas indicate there exist multi-arm trials between the comparisons.
The size of the nodes and thickness of edges represent the number of studies that examined a treatment and compared two given treatments respectively.
Treatments are ranked from best to worst along the leading diagonal. Above the leading diagonal are estimates from pairwise meta-analyses, below the leading diagonal are estimates from network meta-analyses
Relative treatment effects in ranked order for all studies
Relative treatment effects in ranked order with selected studies excluded
Assessment of inconsistency for all studies
Assessment of inconsistency with selected studies excluded
In contrast to the 'comparison of all treatment pairs' tab in the frequentist NMA results, this table only contains the estimates from the network meta analysis, i.e. does not contain estimates from pairwise meta-analysis which only contains direct evidence. If you would like to obtain the pairwise meta-analysis results, please use the Nodesplit model module
Treatment effects for all studies: comparison of all treatment pairs.
Treatment effects with selected studies excluded: comparison of all treatment pairs.
If you export and include the Litmus Rank-O-Gram or the Radial SUCRA plot in your work, please cite it as: Nevill CR, Cooper NJ, Sutton AJ, A multifaceted graphical display, including treatment ranking, was developed to aid interpretation of network meta-analysis, Journal of Clinical Epidemiology (2023)
Relative effects
Ranking results
Summary of evidence
Relative effects
Ranking results
Summary of evidence
Results details for all studies
Empirical mean and standard deviation for each variable, plus standard error of the mean:
Quantiles for each variable:
Results details excluding selected studies
Empirical mean and standard deviation for each variable, plus standard error of the mean:
Quantiles for each variable:
Residual deviance from NMA model and UME inconsistency model for all studies
Residual deviance from NMA model and UME inconsistency model excluding selected studies
This plot represents each data point's contribution to the residual deviance for the NMA with consistency (horizontal axis) and the unrelated mean effect (ume) inconsistency models (vertical axis) along with the line of equality. The points on the equality line means there is no improvement in model fit when using the inconsistency model, suggesting that there is no evidence of inconsistency. Points above the equality line means they have a smaller residual deviance for the consistency model indicating a better fit in the NMA consistency model and points below the equality line means they have a better fit in the ume inconsistency model. Please note that the unrelated mean effects model may not handle multi-arm trials correctly. (Further reading: Dias S, Ades AE, Welton NJ, Jansen JP, Sutton AJ. Network meta-anlaysis for decision-making. Chapter 3 Model fit, model comparison and outlier detection. @2018 John Wiley & Sons Ltd.)
Per-arm residual deviance for all studies
Per-arm residual deviance excluding selected studies
This stem plot represents the posterior residual deviance per study arm. The total number of stems equals the total number of data points in the network meta analysis. Going from left to right, the alternating symbols on the stems indicate the different studies. Each stem corresponds to the residual deviance ($dev.ab) associated with each arm in each study. The smaller residual deviance (the shorter stem), dev.ab, the better model fit for each data point. You can identify which stem corresponds to which study arm by hovering on the stem symbols. (Further reading: Dias S, Ades AE, Welton NJ, Jansen JP, Sutton AJ. Network meta-anlaysis for decision-making. Chapter 3 Model fit, model comparison and outlier detection. @2018 John Wiley & Sons Ltd.)
Leverage plot for all studies
Leverage plot excluding selected studies
This leverage plot shows the average leverage across the arms for each study ({sum($lev.ab)}/{number of arms} for each study) versus the square root of the average residual deviance across the arms for each study (sqrt({sum($dev.ab)}/{number of arms}) for each study). The leverage for each data point, is calculated as the posterior mean of the residual deviance, minus the deviance at the posterior mean of the fitted values. The leverage plot may be used to identify influential and/or poorly fitting studies and can be used to check how each study is affecting the overall model fit and DIC. Curves of the form x2 + y = c, c = 1, 2, 3, ., where x represents square root of residual deviance, and y represents the leverage, are marked on the plot. Points lying on such parabolas each contribute an amount c to the DIC (Spiegelhalter et al., 2002). Points that lie outside the line with c = 3 can generally be identified as contributing to the model's poor fit. Points with a high leverage are influential, which means that they have a strong influence on the model parameters that generate their fitted values. (Further reading: Dias S, Ades AE, Welton NJ, Jansen JP, Sutton AJ. Network meta-anlaysis for decision-making. Chapter 3 Model fit, model comparison and outlier detection. @2018 John Wiley & Sons Ltd. Spiegelhalter et al. (2002) Bayesian measures of model complexity and fit. J. R. Statist. Soc.B 64, Part4, pp.583-639)
Gelman convergence assessment plots for all studies
Trace plots for all studies
Posterior density plots for all studies
Gelman convergence assessment plots excluding selected studies
Trace plots excluding selected studies
Posterior density plots excluding selected studies
MCMC characteristics
Prior distributions
Note: Normal distributions are parameterized here as N(mean, variance) and in the JAGS code as N(mean, precision).
Note: For help on the scaled t-distribution see the JAGS manual.
Model codes for analysis of all studies
DownloadInitial values
Download initial values for chain 1 Download initial values for chain 2 Download initial values for chain 3 Download initial values for chain 4Download simulated data
Download data from chain 1 Download data from chain 2 Download data from chain 3 Download data from chain 4This graph was adapted from Graphs of study contributions and covariate distributions for network meta-regression , Sarah Donegan, Sofia Dias, Catrin Tudur-Smith, Valeria Marinho, Nicky J Welton, Res Syn Meth , 2018; 9 :243-260. DOI: 10.1002/jrsm.1292
Treatment effects for all studies: comparison of all treatment pairs
If you export and include the Litmus Rank-O-Gram or the Radial SUCRA plot in your work, please cite it as: Nevill CR, Cooper NJ, Sutton AJ, A multifaceted graphical display, including treatment ranking, was developed to aid interpretation of network meta-analysis, Journal of Clinical Epidemiology (2023)
Relative effects
Ranking results
Summary of evidence
Nodesplit does not produce results
Results details for baseline model
Empirical mean and standard deviation for each variable, plus standard error of the mean:
Quantiles for each variable:
Gelman convergence assessment plots for all studies
Trace plots for all studies
Posterior density plots for all studies
The bnma package does not currently include unrelated-mean-effects meta-regression models, therefore the consistency vs UME graph that is displayed in the deviance report module of the Bayesian network meta-analysis section is not available here.
Per-arm residual deviance for covariate model
This stem plot represents the posterior residual deviance per study arm. The total number of stems equals the total number of data points in the network meta analysis. Going from left to right, the alternating symbols on the stems indicate the different studies. Each stem corresponds to the residual deviance ($dev.ab) associated with each arm in each study. The smaller residual deviance (the shorter stem), dev.ab, the better model fit for each data point. You can identify which stem corresponds to which study arm by hovering on the stem symbols. (Further reading: Dias S, Ades AE, Welton NJ, Jansen JP, Sutton AJ. Network meta-anlaysis for decision-making. Chapter 3 Model fit, model comparison and outlier detection. @2018 John Wiley & Sons Ltd.)
Leverage plot for covariate model
This leverage plot shows the average leverage across the arms for each study ({sum($lev.ab)}/{number of arms} for each study) versus the square root of the average residual deviance across the arms for each study (sqrt({sum($dev.ab)}/{number of arms}) for each study). The leverage for each data point, is calculated as the posterior mean of the residual deviance, minus the deviance at the posterior mean of the fitted values. The leverage plot may be used to identify influential and/or poorly fitting studies and can be used to check how each study is affecting the overall model fit and DIC. Curves of the form x2 + y = c, c = 1, 2, 3, ., where x represents square root of residual deviance, and y represents the leverage, are marked on the plot. Points lying on such parabolas each contribute an amount c to the DIC (Spiegelhalter et al., 2002). Points that lie outside the line with c = 3 can generally be identified as contributing to the model's poor fit. Points with a high leverage are influential, which means that they have a strong influence on the model parameters that generate their fitted values. (Further reading: Dias S, Ades AE, Welton NJ, Jansen JP, Sutton AJ. Network meta-anlaysis for decision-making. Chapter 3 Model fit, model comparison and outlier detection. @2018 John Wiley & Sons Ltd. Spiegelhalter et al. (2002) Bayesian measures of model complexity and fit. J. R. Statist. Soc.B 64, Part4, pp.583-639)
MCMC characteristics
Prior distributions
Note: Normal distributions are parameterized here as N(mean, variance) and in the JAGS code as N(mean, precision).
Note: For help on the scaled t-distribution see the JAGS manual.
Model codes for analysis of all studies
DownloadInitial values
Download initial values for chain 1 Download initial values for chain 2 Download initial values for chain 3 Download initial values for chain 4Download simulated data
Download data from chain 1 Download data from chain 2 Download data from chain 3 Download data from chain 4This graph was adapted from Graphs of study contributions and covariate distributions for network meta-regression , Sarah Donegan, Sofia Dias, Catrin Tudur-Smith, Valeria Marinho, Nicky J Welton, Res Syn Meth , 2018; 9 :243-260. DOI: 10.1002/jrsm.1292
Treatment effects for all studies: comparison of all treatment pairs
If you export and include the Litmus Rank-O-Gram or the Radial SUCRA plot in your work, please cite it as: Nevill CR, Cooper NJ, Sutton AJ, A multifaceted graphical display, including treatment ranking, was developed to aid interpretation of network meta-analysis, Journal of Clinical Epidemiology (2023)
Relative effects
Ranking results
Summary of evidence
Nodesplit does not produce results
Results details for covariate model
Empirical mean and standard deviation for each variable, plus standard error of the mean:
Quantiles for each variable:
Gelman convergence assessment plots for all studies
Trace plots for all studies
Posterior density plots for all studies
The gemtc package does not currently include unrelated-mean-effects meta-regression models, therefore the consistency vs UME graph that is displayed in the deviance report module of the Bayesian network meta-analysis section is not available here.
Per-arm residual deviance for covariate model
This stem plot represents the posterior residual deviance per study arm. The total number of stems equals the total number of data points in the network meta analysis. Going from left to right, the alternating symbols on the stems indicate the different studies. Each stem corresponds to the residual deviance ($dev.ab) associated with each arm in each study. The smaller residual deviance (the shorter stem), dev.ab, the better model fit for each data point. You can identify which stem corresponds to which study arm by hovering on the stem symbols. (Further reading: Dias S, Ades AE, Welton NJ, Jansen JP, Sutton AJ. Network meta-anlaysis for decision-making. Chapter 3 Model fit, model comparison and outlier detection. @2018 John Wiley & Sons Ltd.)
Leverage plot for covariate model
This leverage plot shows the average leverage across the arms for each study ({sum($lev.ab)}/{number of arms} for each study) versus the square root of the average residual deviance across the arms for each study (sqrt({sum($dev.ab)}/{number of arms}) for each study). The leverage for each data point, is calculated as the posterior mean of the residual deviance, minus the deviance at the posterior mean of the fitted values. The leverage plot may be used to identify influential and/or poorly fitting studies and can be used to check how each study is affecting the overall model fit and DIC. Curves of the form x2 + y = c, c = 1, 2, 3, ., where x represents square root of residual deviance, and y represents the leverage, are marked on the plot. Points lying on such parabolas each contribute an amount c to the DIC (Spiegelhalter et al., 2002). Points that lie outside the line with c = 3 can generally be identified as contributing to the model's poor fit. Points with a high leverage are influential, which means that they have a strong influence on the model parameters that generate their fitted values. (Further reading: Dias S, Ades AE, Welton NJ, Jansen JP, Sutton AJ. Network meta-anlaysis for decision-making. Chapter 3 Model fit, model comparison and outlier detection. @2018 John Wiley & Sons Ltd. Spiegelhalter et al. (2002) Bayesian measures of model complexity and fit. J. R. Statist. Soc.B 64, Part4, pp.583-639)
MCMC characteristics
Prior distributions
Note: Normal distributions are parameterized here as N(mean, variance) and in the JAGS code as N(mean, precision).
Note: For help on the scaled t-distribution see the JAGS manual.
Model codes for analysis of all studies
DownloadInitial values
Download initial values for chain 1 Download initial values for chain 2 Download initial values for chain 3 Download initial values for chain 4Download simulated data
Download data from chain 1 Download data from chain 2 Download data from chain 3 Download data from chain 4Session code does not produce results
Reproduce environment does not produce results
Reference Packages does not produce results
CINeMA export does not produce results
Section: Export
ORIENTATION
Scientific practice increasingly emphasizes documentation and reproducibility. MetaInsight promotes documentation by allowing users to download information that includes sources of input data, methodological decisions, and results. One option for the documentation (see Module: Session code) is a file that can be re-run in R to reproduce the analyses. Many intermediate and advanced users of R likely will find this file useful as a template for modification. To ensure reproducibility requires the exact same versions of R and dependent packages and these can be downloaded and used to install the same packages later (Module: Reproduce environment). Additionally, MetaInsight provides citations of the particular R packages (and their versions) used in a given analysis (Module: Reference Packages).
Module: Session code
BACKGROUND
Via the Session code module, you can download files that document the analyses run in a given MetaInsight session (including executable code that can reproduce them).
This functionality supports reproducible science.
IMPLEMENTATION
Here you can download documented code that corresponds to the analyses run in the current session of MetaInsight. Two formats are available for download - the .qmd format
is an executable R script that will reproduce the analysis when run in an R session; the .html format is a report which can be opened in a web browser which contains all of the
code chunks used in the analysis. You can choose to include the outputs generated in the analysis in the .html report by toggling the Include outputs? switch. If you would like
a permanent record of your analysis, but don’t intend to change it in the future the .html format, including the outputs is best. If you would like to alter your analysis in the
future then the use the .qmd format. If you download the .qmd you can also produce the .html by using quarto::quarto_render()
The MetaInsight session code .qmd file is composed of a chain of code chunks with module functions that are for internal use in MetaInsight. Each of these functions corresponds to a single module that the user ran during the session. Users can modify their analysis, for example by adding new studies or removing different studies and rerun their analysis. Open the .qmd in RStudio, click on “Run” in the upper-right corner, and run chunk by chunk or all at once.
You can also choose to exclude modules that you have run by clicking the Exclude modules? button and deselecting modules from the menu. Note that the setup modules cannot be removed and for the Bayesian, Baseline risk and Covariate analyses, if you exclude the model it will also remove all the other modules. If you run other modules after pressing the Exclude modules? button, you must press it again to update the available modules.
Module: Reproduce environment
BACKGROUND
In order to ensure that an analysis is reproducible, it is necessary to record the exact versions of all software dependencies used in an analysis. This module allows you to download a list of all packages used by MetaInsight which can be used to restore those versions in the future.
IMPLEMENTATION
The module uses renv::snapshot() to produce a .lock file which can passed to renv::restore() to reinstall the same package versions on your own machine. Once you have downloaded this file, the Session Code module will be updated to include a chunk which reads the file before rerunning the analysis.
REFERENCES
Sandve GK, Nekrutenko A, Taylor J, Hovig E (2013) Ten Simple Rules for Reproducible Computational Research. PLoS Comput Biol 9(10): e1003285. https://doi.org/10.1371/journal.pcbi.1003285
Module: Reference Packages
BACKGROUND
shinyscholar (or apps created using it) provide citations of the particular R packages (and their versions) used in a given analysis (Module: Reference Packages). The citation of all packages used both promotes documentation and gives credit to the developers of the packages with which the app is built. Dovetailing with the modular nature of shinyscholar, such citation should increase the incentive for researchers to formalize their code into R packages on CRAN and join the app’s community to integrate them into future releases of the software.
IMPLEMENTATION
Users can download a list of references for the R packages used in the analyses. This module utilizes RefManageR and knitcitations (McLean 2020; Boettiger 2021). The list can be downloaded as a .pdf, HTML, or .doc file.
REFERENCES
Boettiger, C. (2021). knitcitations: Citations for ‘Knitr’ Markdown Files. R package version 1.0.12. CRAN
McLean, M.W. (2020). RefManageR: Straightforward ‘BibTeX’ and ‘BibLaTeX’ Bibliography Management. R package version 1.3.0. CRAN
Module: CINeMA export
BACKGROUND
CINeMA (Confidence in Network Meta-Analysis) is a method of assessing confidence in the findings of a network meta-analysis.
The easiest way to apply the method is to use the accompanying online application, which is also called CINeMA cinema.med.auth.gr.
Previously, in order to use the CINeMA app, the meta-analysis model had to be applied within the app itself.
This module allows the results from an NMA fit in MetaInsight to be downloaded, in a file that can then be uploaded into CINeMA.
Currently, only the frequentist and Bayesian NMA models are supported; results from meta-regression or baseline risk meta-regression cannot be exported.
IMPLEMENTATION
Data can only be exported when the uploaded dataset contains risk of bias data, specifically the rob and indirectness columns.
There are four options available for the download, corresponding to the frequentist or Bayesian NMA, and to the full uploaded dataset or a smaller dataset with studies excluded by the user.
The CINeMA method requires treatment effect statistics for direct and indirect evidence separately, as well as overall.
It also requires estimates of the contribution each study makes towards the treatment effect estimates.
Due to computational complexity in the former, and undeveloped theory in the latter, MetaInsight always calculates these values from the frequentist model.
For the purpose of the CINeMA method these frequentist approximations suffice.
REFERENCES
Nikolakopoulou A, Higgins JPT, Papakonstantinou T, Chaimani A, Del Giovane C, Egger M, et al. (2020) CINeMA: An approach for
assessing confidence in the results of a network meta-analysis. PLoS Med 17(4): e1003082. 10.1371/journal.pmed.1003082
Welcome to v7 of MetaInsight featuring a redesigned interface, improved plots, automatically updating models and downloadable reports. If you are a returning user, click 'See what has changed' in the sidebar for a guided tour. For a limited time, v6 is available at: https://crsu.shinyapps.io/MetaInsight_V6/ . Please send any feedback to apps@crsu.org.uk.
What is MetaInsight?
Network meta-analysis (NMA) has been increasingly adopted in evidence-based medicine to compare multiple interventions and address the questions such as ‘which intervention is the ‘best’ overall?’. Currently, NMA is primarily conducted in statistical packages such as WinBUGs, R and STATA, and the software coding can be difficult for non-statisticians and hinders the progress of carrying out systematic reviews containing NMA.
MetaInsight is a tool that conducts NMA via the web requiring no specialist software for the user to install but leveraging established analysis routines (specifically the bnma, gemtc and netmeta packages in R). The tool is interactive and uses an intuitive ‘point and click’ interface and presents results in visually intuitive and appealing ways. It is hoped that this tool will assist those in conducting NMA who do not have expert statistical programming skills, and, in turn, increase the relevance of published meta-analyses, and in the long term contribute to improved healthcare decision making as a result.
Users wishing to analyse large treatment networks or fit complex network meta-analysis models should seek advice from technical experts.
If you use the app please cite it as:
Owen, RK, Bradbury, N, Xin, Y, Cooper, N, Sutton, A. MetaInsight: An interactive web-based tool for analyzing, interrogating, and visualizing network meta-analyses using R-shiny and netmeta. Res Syn Meth. 2019; 10: 569-581.
Attributes of MetaInsight
This version of MetaInsight was built using shinyscholar and has these attributes:
- accessible: lowers barriers to conducting complex meta-analysis by providing an intuitive graphical user interface
- open: the code is free to use and modify (GPL 3.0) and available on GitHub
- expandable: users can author and contribute modules that enable new methodological options
- flexible: options for user uploads and downloads of results
- interactive: includes an embedded sortable
{DF}data tables, and visualizations of results - instructive: features guidance text that educates users about theoretical and analytical aspects of each step in the workflow
- reproducible: users can download a
{quarto}.qmd file that when run reproduces the analysis, and also save sessions and load them later - reliable: modules and their underlying functions are tested using
{testthat}and{shinytest2}
Contact us
Please email us with any other questions. If you encounter any errors with using the app, please check the troubleshooting page first before contacting us.
Disclaimer
THE SOFTWARE IS PROVIDED AS IS, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
Funding and Support Acknowledgement
MetaInsight is part of the Complex Reviews Synthesis Unit (CRSU) suite of evidence synthesis apps. The development of these apps was (majority) funded and overseen by the Evidence Synthesis Group @ CRSU (NIHR153934). Further details of other funders and support, current and past, can be found on our GitHub page. The views expressed are those of the author(s) and not necessarily those of the NIHR or the Department of Health and Social Care.
More information about the UK NIHR Complex Reviews Synthesis Unit (CRSU) can be found on our website
Developers
Developers are listed in alphabetical order:
Naomi Bradbury is a Research Associate in the Biostatistics Research Group and developed the initial version of MetaInsight.
Ryan Field is a Research Software Engineer within the Shared Research Facilities at the University of Glasgow. He has a background in computer science, software engineering and public health, and a focus on FAIR and reproducible data. With interests in research software, data analysis, evidence synthesis, web application development and statistical analysis.
Tom Morris is a Research Associate in the Biostatistics Research Group. He is a mathematician and clinical trials statistician by background, and now mostly works in meta-analysis.
Clareece Nevill is a Research Associate in the Biostatistics Research Group, where she spends half of her time as part of an NIHR Evidence Synthesis Group and the other half on her PhD. She developed the Ranking panel in the Bayesian analysis.
Janion Nevill is a Research Software Engineer in the School of Medical Sciences, Public Health and Epidemiology at the University of Leicester. He has been responsible for introducing best practices to the development team, including modularising and testing code. He has also been involved in long-term maintenance and improvements to the app.
Rhiannon Owen developed the initial version of MetaInsight.
Simon Smart received his PhD in Plant Sciences from the University of Cambridge and is currently a Software Developer in the Department of Population Health Sciences at the University of Leicester. He developed shinyscholar for producing reproducible apps and converted MetaInsight to the framework.
Yiqiao Xin developed the initial version of MetaInsight.
Managers
The managers have guided and supervised the developers:
Nicola Cooper is a Professor of Healthcare Evaluation Research at the University of Leicester and a member of the Complex Reviews Synthesis Unit. Her primary research interest is in the interface and integration of medical statistics and health economics.
Alex Sutton is a Professor of medical statistics with a current interest in developing methods for data visualisation interactive digital web content to facilitate statistical analyses and using synthesis to inform the design of future studies. He has worked in evidence synthesis (including meta-analysis and network meta-analysis) for over 25 years and authored over 200 publications; a combination of methodological work and substantive applications.
Overview of MetaInsight
This information is designed to orient users and developers to the MetaInsight interface. Two guided tours are also available - one for new users to MetaInsight and one for users of previous versions of MetaInsight.
Sections and modules
MetaInsight is composed of Sections – discrete steps in the workflow. Navigate through the sections by clicking on the names in the top navigation panel. You must use the Setup section to either load your own data or example data, but afterwards you can choose to either use a single section or all of them.
Within each section, there are various major options that can be run. These are the modules. Selecting a module opens the control panel to make decisions and run the module’s functionalities. In some sections, some modules have to be run before you can move on to later steps - for example in the Setup section, both the Load data and Configure analysis modules must be used, but most can be run independently and if you try to run a module prematurely, error messages will tell you which module needs to be run in order to proceed.
Log window and visualization panel
Analyses performed will be detailed in the log window. This is also where error messages appear.
After running a module, outputs appear in the Results window and switching between modules will reload the outputs for each module.
Guidance texts
Many scientific analyses require the researcher to make decisions and the guidance text provides information for the user about what the modules do and how to use them.
As the user proceeds through the workflow, the relevant guidance texts can be found to the right of the Results tab in the Visualization panel.
If more support is needed, the Support tab in the navigation bar at the top provides links to the MetaInsight Issues page and email.
Saving and reproducing results
Users can stop an analysis and restart it later, by saving the workflow progress as an RDS file, found in the Save tab. This file can be loaded into MetaInsight later using the Reload previous session module in the Setup section to restart the analysis where the user left off.
Additionally, MetaInsight allows the user to download their results. After each step of analysis (i.e. after running each module), the results for that particular step may be downloaded from the control panel of the module.
This version of MetaInsight is fully reproducible meaning that you can rerun the analysis outside of the app and produce the same results. This is useful for archiving and validation. To download the session code use the Reproduce section. This includes the option of a quarto file that can be opened and rerun in R, or an HTML file which can be opened in web browsers and contains a tab of results for each section of the app.
Old user guide
Download User Guide This user guide is based on version 3 of MetaInsight. Some elements of the app have been changed since the guide was originally produced.Reproducing NICE Technical Support Documents for Evidence Synthesis
The University of Sheffield host a series of NICE Technical Support Documents on evidence synthesis: https://www.sheffield.ac.uk/nice-dsu/tsds/evidence-synthesis Some of the models described in TSD2 and TSD3 can be carried out in MetaInsight. A guide has been written with instructions to reproduce the relevant analyses, which can be downloaded here. The guide is based on MetaInsight version 6.0.0.
Video tutorials
These tutorials were made for older versions of MetaInsight, with a different user interface, but may still be of use.
Tutorial for MetaInsight v4.0.0 produced for ESMARConf 2023
Treatment Ranking Demo
A short demo video of how to use the ranking module in Bayesian, Baselink Risk and Covariate analyses.
Cochrane Training Webinar
These videos were recorded live in 2019 as part of the Cochrane Training network meta-analysis learning live webinar series. They are intended for people who are interested in undertaking a network meta-analysis using MetaInsight.
MetaInsight: Background, introduction, demonstration, limitations, and future plans
Latest updates
The full update history for MetaInsight is available on GitHub
Minor update (17 June 2026 v7.1.1)
- Fix bug where incorrect numbers of study arms was reported below network plots
Major update (6 June 2026 v7.1.0)
- Add ability to export results to CINeMA
Minor update (26 May 2026 v7.0.2)
- Various minor adjustments to meet CRAN guidelines
Minor update (14 May 2026 v7.0.1)
- Code for generating gelman plots has been copied over from {coda}
Major update (11 May 2026 v7.0.0)
- The interface has been redesigned but all functionality has been retained
- Analyses are now reproducible outside of the app - once you have conducted an analysis, you can download a document which reproduces the analysis.
- Analyses can be saved and loaded at any point by downloading a file which you can upload at a later date to restore the app
- Plot downloads have been improved - choose your preferred format in one location for all downloads, and downloads now match what is shown in the app more consistently
- All models update automatically when parameters are changed
- The app is now available as a package on CRAN and can be installed using
install.packages("metainsight")and then opened usinglibrary(metainsight)andrun_metainsight()
Minor update (27 May 2025 v6.4.0)
- Added trace and posterior density plots to Bayesian output
- Added MCMC details such as prior distributions and number of iterations
- Set seeds for Bayesian models so that results are reproducible
Major update (10 July 2024 v6.0.0)
- Meta-regression has been added. One covariate is allowed, which can be a new continuous or binary variable, or baseline risk. Two new graphs are available for meta-regression. The first displays the covariate values grouped by treatment and study. The second plots the covariate against relative treatment effects, with credible regions and study-level contributions.