Long-term outcomes and prognostic factors in older adults aged 65 years and over with aneurysmal subarachnoid haemorrhage, stratified by age band

Protocol, data, screening records and analysis code for a systematic review and meta-analysis of long-term outcomes (assessed at 12 months or later) and prognostic factors in patients aged 65 years and over with aneurysmal subarachnoid haemorrhage, stratified by age band and by admission severity, with a meta-regression on calendar era. HOW THIS PROJECT IS ARRANGED. The protocol versions are placed at the top level of this project so that the amendment history can be read without downloading anything. Everything else is in OSF�

�開_0816.1.zip, which holds the same protocol files under 00_protocol/ together with nine further folders. README.md and MANIFEST_sha256.txt are also at the top level: the first is a guide to those folders, the second a SHA-256 checksum of every file. CURRENT PROTOCOL VERSION: 2.1, dated 16 August 2026 (protocol_v2.1_20260816.docx). Versions 1.3 and 1.5 remain as originally deposited on 22 and 31 July 2026. Versions 1.6 to 2.0 were written on the dates recorded in each document's version history and were deposited on 17 August 2026 together with the rest of this package; this is stated in README_protocol_versions.md rather than left for the upload date to imply. Version 2.1 carries the full version history. WHAT THIS DEPOSIT CONTAINS. Everything needed to check the review other than the full texts of the included studies, which are under copyright and cannot be redistributed. Screening was assisted by a large language model; the model ordered and classified records and was never permitted to exclude a record on its own, and every exclusion carries a recorded reason. The screening records are what that process produced, so the assistance can be checked against its output. SCREENING AND VERIFICATION. Eligibility at full text was assessed independently by two reviewers in both database arms. In the PubMed arm, all 59 reports examined in full text; Cohen's kappa 0.38, 18 disagreements, all adjudicated, none changed the included set. In the Ichushi-Web arm, all 55 reports examined in full text, blinded and presented in randomised order; Cohen's kappa 0.73, five disagreements, all adjudicated, none changed the included set. Title and abstract screening was by a single reviewer with liberal exclusion rules and a tracked set of 63 known relevant reports that survived every pass. SEARCHES. PubMed and Ichushi-Web were searched and screened. Cochrane CENTRAL was searched but its records could not be exported without a Cochrane sign-in; the portion drawn from PubMed had already been screened, and the trial-registry portion was covered by a direct search of ClinicalTrials.gov on 16 August 2026, which returned no eligible study. Forward and backward citation tracking of the 23 included studies indexed in PubMed, performed the same day, returned 151 records not retrieved by the original search, none of which was eligible. Embase and the Web of Science were not searched, the review team not holding an entitlement to them; this is reported among the limitations. THIS REVIEW IS NOT REGISTERED IN PROSPERO. By the time registration was sought, the electronic searches were complete and a first-pass extraction had begun, so the review no longer met PROSPERO's requirement that registration precede the start of data extraction. This protocol is deposited instead. It was timestamped before the two-reviewer screening, the two-reviewer data extraction, the risk-of-bias assessment and any analysis were carried out, and no pooled estimate, subgroup analysis, meta-regression or test for small-study effects had been computed at the date of first deposit. REPORTING follows PRISMA 2020 and MOOSE. Risk of bias: QUIPS, with GRADE adapted for prognostic factor research. LICENCE. Data and records CC0 1.0 Universal; code MIT.