Mahmood Ahmad
Tahir Heart Institute
author@example.com

Protocol: The Fragility Atlas: Multiverse Analysis of 403 Cochrane Meta-Analyses Reveals 58% Are Sensitive to Analyst Choices

This protocol describes the evidence synthesis for The Fragility Atlas: Multiverse Analysis of 403 Cochrane, targeting reproducible estimation of Robustness score (percent agreement across specifications) in a versioned workflow. Eligible studies include Cochrane systematic reviews and randomised trials reporting the primary outcome, with no restrictions on year, language, or sample size. Searches will cover the Cochrane Library, PubMed, and Embase using structured terms, reference-list screening, and duplicate review before extraction. The primary analysis will estimate Robustness score (percent agreement across specifications) using restricted maximum likelihood random-effects meta-analysis, reporting 95 percent confidence intervals, prediction intervals, and model checks. Heterogeneity will be summarised using I-squared and tau-squared, with sensitivity analyses across estimators, exclusion scenarios, and leave-one-out patterns. Analysis code will be versioned and archived at https://github.com/mahmood726-cyber/fragility-atlas, and reporting will follow PRISMA 2020 guidance to support independent verification and reuse. Anticipated limitations include publication bias, clinical heterogeneity, sparse data in some settings, and the constraints of aggregate-level evidence synthesis.

Outside Notes

Type: protocol
Primary estimand: Robustness score (percent agreement across specifications)
App: Fragility Atlas v1.0
Code: https://github.com/mahmood726-cyber/fragility-atlas
Date: 2026-03-26
Validation: DRAFT

References

1. Walsh M, Srinathan SK, McAuley DF, et al. The statistical significance of randomized controlled trial results is frequently fragile: a case for a Fragility Index. J Clin Epidemiol. 2014;67(6):622-628.
2. Atal I, Porcher R, Boutron I, Ravaud P. The statistical significance of meta-analyses is frequently fragile: definition of a fragility index for meta-analyses. J Clin Epidemiol. 2019;111:32-40.
3. Borenstein M, Hedges LV, Higgins JPT, Rothstein HR. Introduction to Meta-Analysis. 2nd ed. Wiley; 2021.

AI Disclosure

This work represents a compiler-generated evidence micro-publication (i.e., a structured, pipeline-based synthesis output). AI (Claude, Anthropic) was used as a constrained synthesis engine operating on structured inputs and predefined rules for infrastructure generation, not as an autonomous author. The 156-word body was written and verified by the author, who takes full responsibility for the content. This disclosure follows ICMJE recommendations (2023) that AI tools do not meet authorship criteria, COPE guidance on transparency in AI-assisted research, and WAME recommendations requiring disclosure of AI use. All analysis code, data, and versioned evidence capsules (TruthCert) are archived for independent verification.