Microbiome associations and interventions in gynecologic cancer treatment — screening logbook, update-search records and recomputation code

Record-level screening logbook, update-search records, reference cross-walk and recomputation code accompanying the manuscript "Microbiome associations and interventions in gynecologic cancer treatment: a structured clinical evidence synthesis and a focused systematic review with meta-analysis of randomized oral probiotic trials during pelvic radiotherapy" (Şen S, Gülgel Şen F, 2026). Screening logbook: 25,314 frozen bibliographic records (13,518 master archive compiled from PubMed, Europe PMC, Scopus, HAL, TR Dizin, SciELO, CENTRAL and CDSR — the Scopus and CENTRAL material consists of earlier exports held in the archive, as new authenticated Scopus and Web of Science searches could not be completed; 11,794 supplementary candidates from PubMed, Europe PMC, OpenAlex and ClinicalTrials.gov searches run on 26 September 2026; 2 manual citation records). Every record carries both reviewers' independent title/abstract decisions (R1 Serhat Şen; R2 Feyza Gülgel Şen), basis, reason code, evidence category, rationale and source excerpt, the adjudicated screening action, and the post-screening duplicate/report-family audit (148 duplicate report records; 25,166 canonical records; 193 identities flagged). Inter-rater agreement: Cohen's kappa 0.821 (95% bootstrap 0.811–0.830) for the master continuation and 0.775 (0.764–0.787) for the supplementary cohort. Screening excluded 15,625 records and retained 9,689. "Advance" is a screening action, not final inclusion; full-text eligibility across the whole archive was not completed. Update search: 52 PubMed records retrieved on 27 September 2026 with the exact query and record-level classification. Reference cross-walk: the 70 manuscript references mapped to their screening-log records and manuscript roles. Code: S18_recomputation.py reproduces every pooled estimate (DerSimonian–Laird, REML, fixed-effect, modified Hartung–Knapp, leave-one-out, missing-outcome bounds, test-based I² interval); primary pool RR 0.669 (95% CI 0.392–1.142; modified Hartung–Knapp 0.189–2.375). Sheet and column names in the logbook are Turkish; DATA_DICTIONARY.md provides the English key. Search counts across sources overlap and must not be summed. Version 1.1.0 removes a placeholder line left over from workbook-template construction in the status sheet; the placeholder described a hypothetical workflow and did not describe the conduct of this review, in which all title/abstract assessments and adjudication were performed by the two named authors. No decision, denominator, agreement statistic or pooled estimate was affected (see CHANGELOG.md). Data and documentation: CC BY 4.0; code: MIT License.