Research Group: Evidence Synthesis Methods

The "Evidence Synthesis Methods" research group focuses on methods for synthesizing evidence from studies about the efficacy and safety of healthcare interventions. We develop, advance, apply and disseminate methodology for pairwise and network meta-analysis. Our work includes research on methods to address publication bias, the role of non-randomized studies in evidence synthesis, multivariate meta-analysis models and methods to synthesize data about rare safety outcomes. 

Our team has long-standing expertise on network meta-analysis. Traditional meta-analytical methods commonly used for synthesizing evidence from clinical trials are limited in comparing two interventions. However, for any given condition there is usually a plethora of alternative treatment options; this situation has motivated the development of network meta-analysis techniques, which extent usual pairwise meta-analysis for the case when more than two treatments are being compared in the available trials.

Group leader

Group members

  1. Evidence-based planning of clinical research (EBAR)
    (funded by EU research grant MSCA-IF-703254)
    This project aims to develop a theoretical framework for efficient and sustainable research planning, with the potential to direct resource allocation after considering the existing evidence.

  2. Enhancing methods for evaluating the comparative safety of medical interventions
    (funded by the Swiss National Science Foundation)
    This project aims to advance the methods for synthesizing evidence from randomised trials on the safety of interventions by developing and exploring meta-analytical models for correlated rare events and network meta-analysis of adverse events.

  3. Comparative effectiveness and safety of disease modifying drugs in early treatment of multiple sclerosis
    (funded by the Swiss Multiple Sclerosis Society)
    This project aims to answer a) What happens when people who receive a diagnosis of multiple sclerosis decide to start treatment with a disease-modifying drug? b) Which disease modifying drugs have the best efficacy-safety profile?

Selected Publications

  1. Nikolakopoulou A, Mavridis D, Egger M, Salanti G. Continuously updated network meta-analysis and statistical monitoring for timely decision-making. Stat Methods Med Res. 2016 (Epub ahead of print)
  2. Nikolakopoulou A, Mavridis D, Salanti G. Planning future studies based on the precision of network meta-analysis results. Stat Med. 2016 Mar 30;35(7):978-1000
  3. 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 Jun 2;162(11):777-84.
  4. Efthimiou O, Debray T, Valkenhoef G, Trelle S, Panayidou K, Moons KGM, Reitsma JB, Shang A, Salanti G. GetReal in network meta‐analysis: a review of the methodology. Res Synth Methods 2016 Sep;7(3):236-63
  5. Efthimiou O, Mavridis D, Riley RD, Cipriani A, Salanti G. Joint synthesis of multiple correlated outcomes in networks of interventions. Biostatistics. 2014 Jul 2. pii: kxu030.
  6. Mavridis D, Welton NJ, Sutton A, Salanti G. A selection model to account for publication bias in a full network meta-analysis. Stat Med. 2014 30;33(30):5399-412
  7. Salanti G, Del Giovane C, Chaimani A, Caldwell DM, Higgins JP. Evaluating the quality of evidence from a network meta-analysis. PLoS One. 2014 Jul 3;9(7):e99682.
  8. Efthimiou O, Mavridis D, Cipriani A, Leucht S, Bagos P, Salanti G. An approach for modelling multiple correlated outcomes in a network of interventions using odds ratios. Stat Med. 2014 Jun 15;33(13):2275-87.
  9. Chaimani A, Higgins JP, Mavridis D, Spyridonos P, Salanti G. Graphical tools for network meta-analysis in STATA. PLoS One. 2013 Oct 3;8(10):e76654.
  10. Cipriani A, Higgins JP, Geddes JR, Salanti G. Conceptual and technical challenges in network meta-analysis. Ann Intern Med. 2013 Jul 16;159(2):130-7.
  11. Filippini G, Del Giovane C, Vacchi L, D'Amico R, Di Pietrantonj C, Beecher D,Salanti G. Immunomodulators and immunosuppressants for multiple sclerosis: a network meta-analysis. Cochrane Database Syst Rev. 2013 Jun 6;(6):CD008933.
  12. Mavridis D, Sutton A, Cipriani A, Salanti G. A fully Bayesian application of the Copas selection model for publication bias extended to network meta-analysis. Stat Med. 2013;32(1):51-66.
  13. Salanti G. Indirect and mixed-treatment comparison, network, or multiple-treatments meta-analysis: many names, many benefits, many concerns for the next generation evidence synthesis tool. Res Synth Meth. 2012 3 (2): 80.
  14. Chaimani A, Salanti G. Using network meta-analysis to evaluate the existence of small-study effects in a network of interventions. Res Synth Meth. 2012 Jun;3(2):161-76.
  15. Mavridis D, Salanti G. A practical introduction to multivariate meta-analysis. Stat Meth Med Res 1012 2:133- 58
  16. Salanti G, Ades AE, Ioannidis JP Graphical methods and numerical summaries for presenting results from multiple-treatments meta-analysis: an overview and tutorial J Clin Epidemiol. 2011 Feb;64(2):163-71
  17. Salanti G, Higgins JP, Ades A, Ioannidis JP. Evaluation of networks of randomized trials. Stat Methods Med Res. 2008 (3):279-301.

Confidence in Network Meta-Analysis: How to evaluate study limitations (theory)

This video explains how to evaluate the impact of study limitations (risk of bias) in the results of network meta-analysis.

Confidence in Network Meta-Analysis: How to evaluate study limitations (practical)

This video explains how to use the web-application CINeMA to evaluate the impact of study limitations (risk of bias) in the results of network meta-analysis.

Evidence-based designing of clinical trials using Living Network Meta-analysis

This is a talk presented at the 24th Cochrane Colloquium in 2016 in Seoul as part of the Methods Symposium. It presents, partly, work done within the EU funded project EBAR (MSCA-IF-703254)

A 10 minutes introduction to Network Meta analysis

Network meta-analysis (NMA) has emerged as the new evidence synthesis tool. Clinical papers that use NMA are increasingly published in the medical literature. This is a 10-minutes non-technical introduction to the concept of indirect comparison and NMA.

Combining randomised and non-randomised evidence in network meta-analysis

Observational studies convey valuable information about the effectiveness of interventions in real-life clinical practice and there is a growing interest for methods to include non-randomized evidence in the decision-making process. We then present three alternative methods that allow the inclusion of observational studies in an NMA of RCTs: the design-adjusted synthesis, the use of observational evidence as prior information and the use of three-level hierarchical models.

Network meta-anlysis (NMA)

Mathematical models

Individual Patient Data IPD meta analysis - Matthias Egger

Systematic reviews, meta analysis and real world evidence - Matthias Egger

Policy makers and guideline developers face challenges in evaluating the quality of evidence from systematic reviews with multiple interventions. We previously developed a framework to judge the confidence that can be placed in results obtained from a network meta-analysis (NMA) based on the GRADE domains: study limitations, indirectness, inconsistency, imprecision and publication bias. The framework combines judgments about direct evidence with their statistical contribution to network meta-analysis results, enabling evaluation of the credibility of NMA treatment effects and treatment rankings. However, the process is cumbersome and time-consuming for large networks.

Our user-friendly web application CINeMA (Confidence In Network Meta-Analysis) will greatly simplify the evaluation of NMA results, guiding reviewers through a structured process, with semi-automation of several steps that should decrease workload considerably. Only study outcome data and study-level risk of bias assessments are required as input; then CINeMA produces graphical and numerical summaries of NMA output, indicating likely judgments for the five credibility domains.

A ‘proof-of-concept’ version of CINeMA is available in

Salanti G, Del Giovane C, Chaimani A, Caldwell DM, Higgins JP. Evaluating the quality of evidence from a network meta-analysis. PLoS One. 2014 Jul 3;9(7):e99682.

Working Papers on CINeMA


In these videos from a Cochrane Learning Live webinar, Georgia Salanti and Theodore Papakonstantinou present the CINeMA (Confidence in Network Meta-analysis) framework and web aplication developed to judge the confidence that can be placed in results obtained from a network meta-analysis by adapting and extending the GRADE domains (study limitations, inconsistency, indirectness, imprecision and publication bias).

Below you will find the videos covering:

Part 1: Introduction to CINeMA framework
Part 2: Within-study bias, indirectness
Part 3: Imprecison, heterogeneity/incoherence, reporting bias
Part 4: Upcoming developments of CINeMA app, questions and answers

According to a recent UN report, the vast majority of mental health needs remain unaddressed because of the lack of investment in relevant research and urges for considering mental health actions as part of a national response to COVID19. The chances of mounting a successful response to a pandemic greatly depend on the speed and accuracy of the available information, and hence collecting high-quality data on the mental health effects of the COVID-19 pandemic is an immediate priority.

In this project, we aim to provide reliable large-scale evidence about mental health during the COVID19 pandemic and examine how the changes in mental health state of the societies depend on the lockdown measures put in place worldwide.

More specifically, they will answer the questions:

  • What is the prevalence of mental health problems in the general population and subpopulations worldwide during the COVID-19 pandemic?
  • How are mental health problems associated with a) characteristics of the pandemic b) the extent and intensity of measures to contain the pandemic? And which population characteristics (e.g. sex, age, comorbidities, cultural characteristics) modify these prevalences.