Efficacy and Safety of SGLT2 Inhibitors and Sotagliflozin in Heart Failure: A Systematic Review and Meta-Analysis of 59 Randomized Controlled Trials

Reproducibility archive for the systematic review and meta-analysis: Efficacy and Safety of SGLT2 Inhibitors and Sotagliflozin in Heart Failure: A Systematic Review and Meta-Analysis of 59 Randomized Controlled Trials. PROSPERO registration: CRD420251167908. Reporting standard: PRISMA 2020. Risk of bias: Cochrane RoB 2.0. Certainty of evidence: GRADE. Scope: 6,239 records identified across PubMed, Cochrane CENTRAL, ClinicalTrials.gov, and WHO ICTRP; 4,511 screened after deduplication; 1,652 full texts assessed; 114 studies included in qualitative synthesis; 59 randomized controlled trials (29,692 participants) pooled in quantitative synthesis. Random-effects models with REML estimator and Hartung-Knapp-Sidik-Jonkman confidence intervals; risk ratio (Mantel-Haenszel) as primary effect measure with hazard ratio sensitivity; mean difference for continuous outcomes; Trial Sequential Analysis (O'Brien-Fleming, RRR 15% and 20%, two-sided) for primary endpoints. Contents: raw and deduplicated search exports; full Python search/screening pipeline (phases 1-5); full R analysis pipeline (17 scripts, 00-15); 64 generated figures (forest, subgroup, leave-one-out, funnel, cumulative, RoB 2.0 composite, PRISMA flow); 21 output tables; manuscript source (Rmd) and rendered DOCX; complete Cureus Journal of Medical Science submission package (4-section Review Article body, 116-entry Vancouver-Cureus reference list with CrossRef-validated DOIs and registry-URL fallbacks, methods tables, pre-submission audit report); protocol search strategies and amendment log; PROSPERO amendments draft. What's new in v3: reference list expanded to 116 entries (17 methodological + 58 unique poolable trial papers after Vancouver deduplication + 41 citable excluded studies) with DOI/PMID validation against CrossRef and PubMed and public-URL fallback (ClinicalTrials.gov, EUCTR, Cochrane CENTRAL) for records without a resolvable DOI, so all 114 included records carry a reference number with zero orphans; PRISMA Figure 1 regenerated via the canonical PRISMA2020 R package at 2400 px native; all forest, subgroup, and sensitivity figures re-rendered with reference numbers inline in study labels; RoB 2.0 Figure 2 assembled as Panel A (traffic light) and Panel B (domain summary) at 2440 x 4515 px native; sensitivity and publication-bias/TSA figures consolidated into multi-panel composites; per-study tables carry a leading reference column; all section files verified for acronym expansion on first mention and explicit by-letter appendix citation in body text; consolidated pre-submission audit covering reference order, DOI audit, per-study reference columns, in-text media citation/placement, and risk-of-bias denominator consistency. Protocol deviations (registered in PROSPERO): search restart in March 2026 (original July 2025 searches had methodological issues); LILACS not searched (portal inaccessible during execution; scoping searches identified no exclusive LILACS-indexed records); Embase not searched (no institutional access); AI tools (Claude, Codex) disclosed and used in non-judgmental triage and structuring roles with human verification of all outputs; sotagliflozin (dual SGLT1/SGLT2 inhibitor) included under shared SGLT2 mechanism with pre-specified sensitivity analysis confirming non-material effect. Reproducibility: run Rscript scripts/R/run_all.R from the archive root to regenerate all analyses, figures, and tables. Manuscript and supplement render via rmarkdown::render(). See README.md for full instructions. Released under CC BY 4.0.