Welcome to the fifth newsletter for the Cochrane Statistical Methods Group
Issue 5 February 2026
Webinars for LMICs
We are pleased to share the successful conclusion of the SMG’s webinar series on Statistical Analysis in Systematic Reviews.
Developed mainly to support authors and trainers from low- and middle-income countries (LMICs), this six-part series aimed to strengthen practical understanding of the statistical approaches used in systematic reviews. Across the sessions, participants were introduced to core concepts and analytical techniques, progressing through the essential stages of a systematic review from planning to analysis.
Series highlights:
Defining a review question and applying the framework for synthesis: This session introduced participants to the framework for defining clear, structured synthesis questions in a systematic review and determining which studies contribute to each synthesis. It also covered the InSynQ (Intervention Synthesis Questions) checklist to support planning and reporting, helping authors ensure their review questions are precise and well-structured.
Dichotomous and continuous outcomes: This session examined the main effect measures for dichotomous and continuous outcomes, including risk ratio, odds ratio, risk difference, mean difference, and standardized mean difference. It also addressed practical challenges that arise when studies report outcomes using different formats.
Introduction to meta-analysis: This session explained the purpose and principles of meta-analysis, illustrating how findings from multiple studies are combined and presented. It provided guidance on interpreting forest plots within Cochrane reviews.
Exploring heterogeneity: This session focused on methods for identifying, assessing, and addressing heterogeneity in meta-analyses. It discussed considerations from planning through to reporting, emphasizing the importance of understanding variability across studies.
Analysing other outcomes and study designs: This method introduced the generic inverse variance method and covered the analysis of additional outcome types such as ordinal, count, and time-to-event data. It also addressed methodological considerations for synthesizing evidence from more complex study designs, including multi-arm, cluster-randomised, cross-over, and non-randomised studies.
Synthesizing and presenting results when meta-analysis is not possible: This session outlined structured and transparent approaches to synthesis when meta-analysis cannot be conducted. It presented narrative and visual methods that support meaningful interpretation across studies despite reporting limitations.
For more details about the courses, please click here.
Additional methodology webinars
In partnership with Cochrane Training, we hosted two webinars, equipping participants with practical tools and approaches to support clearer interpretation and robust use of evidence in decision-making. Summaries of both sessions are provided below.
Upcoming webinars
This webinar explores models that relax this assumption and their ability to uncover underlying data structures, such as asymmetry and clustering, that may be obscured under the normal model. While summary estimates remain largely unaffected, these models are valuable exploratory tools in seemingly non-normal data.
Past webinars
This webinar introduced an online platform designed to streamline network meta-analysis and make its findings more accessible and understandable. The tool offers interactive features to simplify analysis and support interpretation for clinicians, policy makers, and researchers.
This webinar introduced the principles of Bayesian meta-analysis and highlighted its practical advantages, including improved handling of uncertainty, incomplete data, and diverse sources of evidence. It demonstrated how Bayesian methods can enhance the rigor and reliability of evidence-based decision making.
Please do send us suggestions of presentations and topics for future learning events.
RevMan updates
In September 2022, Cochrane endorsed the implementation of new statistical methods for random-effects meta-analysis to address limitations of commonly used approaches such as the DerSimonian-Laird estimator and Wald-type confidence intervals, particularly in meta-analyses with few studies. The updated methods – including restricted maximum likelihood (REML) estimator for heterogeneity, Q-profile confidence interval method for heterogeneity, Hartung–Knapp–Sidik–Jonkman (HKSJ) confidence interval method for summary effect, updated calculation of the I² statistic, and prediction intervals have been fully implemented in RevMan and released on January 23, 2025.
Authors can now use these methods in Cochrane reviews and updates. The underlying document on the statistical algorithms in RevMan have been updated and will be published as a Supplementary File to Chapter 10 of the Cochrane Handbook (McKenzie JE, Veroniki AA, van Valkenhoef G, Higgins JPT, Deeks JJ. Statistical algorithms in Review Manager. October 2025 version; to appear).
Publications
- Hamilton DG, McKenzie JE, Nguyen P-Y, et al. Evaluation of the replicability of systematic reviews with meta-analyses of the effects of health interventions. Research Synthesis Methods. Published online 2026:1-19. doi:10.1017/rsm.2025.10064
- Karahalios A, White IR, Turner SL, Salanti G, Herbison GP, Veroniki AA, Nikolakopoulou A, McKenzie JE. An investigation of the impact of using contrast-based and arm-based synthesis models for network meta-analysis. Research Synthesis Methods. 2025;1-19. doi:10.1017/rsm.2025.18.
- Nguyen PY, McKenzie JE, Alqaidoom Z, Hamilton DG, Moher D, Page MJ. Reproducibility of meta-analytic results in systematic reviews of interventions: meta-research study. BMJ Med. 2025 Nov 30;4(1):e002024. doi: 10.1136/bmjmed-2025-002024. PMID: 41368128; PMCID: PMC12684188.
- Otalike EG, Clarke M, Akhter F, Veroniki AA, Kandala NB, Gagnier JJ. Methodological Guidance for Individual Participant Data Meta-Analyses: A Systematic Review. J Clin Epidemiol. 2025.190:112089. doi: 10.1016/j.jclinepi.2025.112089
- Veroniki AA, Tricco A, Rangira D, McKenzie J, Li T, Straus S, Smith M, Catala F, Wolfe D, Nincic V, Konstantinidis M, Franco J, Tovey D, Thirugnanasampanthar S, Dourka J, Warren R, Wells G, Stevens A, Hutton B. Updating the PRISMA reporting guideline for network meta-analysis: a scoping review. J Clin Epidemiol. 2025. 188:111985. doi: 10.1016/j.jclinepi.2025.111985
- Siemens W, Borenstein M, Evrenoglou T, Meerpohl JJ, Schwarzer G. Beyond prediction intervals in meta-analysis: reporting the expected proportion of comparable studies with clinically relevant benefit or harm. BMC Med Res Methodol. 2025 Dec 7;25(1):275. doi: 10.1186/s12874-025-02733-9. PMID: 41354774; PMCID: PMC12696960
- Evrenoglou T, Nikolakopoulou A, Schwarzer G, Rücker G, Chaimani A. Producing treatment hierarchies in network meta-analysis using probabilistic models and treatment-choice criteria. Research Synthesis Methods. Published online 2026:1-20. doi:10.1017/rsm.2026.10071
Rare events working group
- The Convenors of the SMG have established a Rare Events Working Group to summarise the evidence on statistical approaches for handling rare events in meta-analysis, from which, recommendations for methods to be implemented in RevMan will be formed. We will be seeking input from the research synthesis community via the SMGlist. If you would like to contribute to the working group, please get in touch with the Convenors (emails listed below).
Contact us
Convenors:
Joanne McKenzie, Australia
Mark Simmonds, UK
Areti Angeliki (Argie) Veroniki, USA
Theodoros Evrenoglou, Germany
Joseph Santos, Switzerland
Lizzie Korevaar, Australia
For details about how to post to the SMGlist, or how to unsubscribe from the list, please go to: https://lists.cochrane.org/mailman/listinfo/smglist