Efficacy of BCG Vaccination Against Tuberculosis: A Random-Effects Meta-Analysis of 13 Randomized Trials
BCG vaccine and tuberculosis: a meta-analysis
- Demonstration Author [UNVERIFIED]
- Department of Demonstration, Institution [UNVERIFIED]
Demonstration Author [UNVERIFIED]
Meta-analysis (systematic review of interventions)
PRISMA 2020
- Abstract: 295 words (structured: Background, Methods, Results, Conclusion)
- Body (Introduction through Discussion): 1,604 words
- Tables: 1 (Table 1)
- Figures: 3 (PRISMA flow, forest plot, funnel plot)
- References: placeholders — see note below
[UNVERIFIED — to be supplied by the senior/corresponding author]
[UNVERIFIED — none declared in this demonstration]
The trial-level data are the canonical, openly available BCG vaccine dataset
(dat.bcg, 13 randomized trials) distributed with the metafor R package. The
analysis dataset (data/dat_bcg.csv), analysis scripts, result tables, and
figures generated for this manuscript are available in the demonstration package.
This manuscript was produced as a clean-room demonstration of the medsci-skills v3.7.0 pipeline. An artificial intelligence language model, Claude (Opus 4.8, Anthropic), accessed via the Claude Code command-line interface in June 2026, was used to assist with manuscript drafting, section structuring, and internal consistency checking of reported statistics. All statistical results were generated by executing R code (metafor 4.8.0, meta 8.2.1) and were read directly from the resulting data files; the AI tool did not generate any numerical result. All content was reviewed and is the responsibility of the authors. The AI tool was not listed as an author. Generative AI was not used to create, modify, or alter any image, figure, or table.
This is a methods demonstration, not a clinical systematic review. The 13 included randomized trials and all of their event counts are the genuine published BCG vaccine trial data. The upstream record counts in the PRISMA study-selection diagram (Figure 1) are illustrative of a typical identification-and-screening cascade for this corpus and are not the yield of a new database search. In-text references are left as [UNVERIFIED] placeholders; a production manuscript would resolve these through the reference-manager workflow before submission.