Time-zero kidney biopsy and deceased-donor transplant outcomes: analysis package (code and data)

Résumé fourni par la source

Version 3.0.1 (3 September 2026) repairs the public input contract of version 3.0.0 without changing the selected effects or any statistical result. The distributed data/analysis_input.csv now carries the four public keys required by the scripts (effect_id, cell_id, cohort_id, source_row_id) alongside their internal provenance fields, so ./code/run_all.sh data/analysis_input.csv OUTPUT_DIR runs as documented. The package adds the direct outputs of a clean run (results/executable_run/) and a validator that checks the article-facing result files against them (15 file pairs, 49,723 values, tolerance 1e-10); independent recomputation passed 12 of 12 pooled models (maximum absolute difference 4.26e-14). The previous version, 3.0.0, remains available at https://doi.org/10.5281/zenodo.21908980. Version 3.0.0 is the complete analysis package (code and data) supporting the systematic review and meta-analysis "Time-Zero Kidney Biopsy and Deceased-Donor Transplant Outcomes: A Systematic Review and Meta-Analysis of Prognostic Value, Prediction, and Clinical Utility". Relative to version 2.0.0 (code-only), this version adds the machine-readable extraction table (2,063 effect rows), the selected-effect input (211 effects), the effect-selection ledger (551 candidates), the screening decision ledger (25,746 records), the generated model outputs (12 pooled models, 216 sensitivity routes, 65 leave-one-cohort-out analyses), the certainty assessments (12 GRADE bodies), and a data dictionary. The code implements restricted-maximum-likelihood random-effects models with Knapp-Hartung inference; prediction intervals for analyses with at least five cohorts; principal and exploratory pooling thresholds; prespecified sensitivity routes; adjustment subgroup and moderator rules; leave-one-cohort-out analyses; exact PICOTS strata; influence and small-study analyses; and illustrative absolute-risk translations for eligible delayed-graft-function odds ratios. No participant-level data are included. The MIT license applies to the contents of code/; the record as a whole is released under CC BY 4.0. Registration: PROSPERO CRD420261435405. Previous version (code-only): 10.5281/zenodo.21879259. Concept DOI: 10.5281/zenodo.21812528.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.