A115 A SYSTEMATIC REVIEW AND META-ANALYSIS OF LOWER GASTROINTESTINAL BLEEDING RISK SCORES TO PREDICT ADVERSE OUTCOMES

Abstract Background Acute lower gastrointestinal bleeding (LGIB) is a common reason for emergency hospitalization. In most patients, bleeding resolves spontaneously but some cases can be fatal. Risk prediction scores can be useful in risk stratifying patients with LGIB at the time of presentation although the most discriminative LGIB risk score is unknown. Aims To perform a systematic review and meta-analysis comparing LGIB risk prediction scores. Methods Following the PRISMA statement, a systematic search for relevant publications after 1990 was conducted in Ovid Medline, EMBASE, Web of Science and CENTRAL electronic databases. We also searched published conference abstracts over the past 5 years. Studies with a primary aim of deriving or validating a LGIB risk score were included. Title and abstracts were reviewed by two independent reviewers followed by full text review and data extraction by both reviewers. Diagnostic classification data for combinations of risk score and clinical outcome were meta-analyzed using a hierarchical summary receiver operator characteristic curve (ROC) model, allowing for random-effects by study, and fixed-effect of the risk score thresholds to influence both sensitivity and specificity. Area under the summary ROC were estimated from model parameters for the pre-specified LGIB risk score thresholds-of-interest. Results Our search identified 2,331 citations for review, of which 100 remained after the title and abstract screen, and 18 ultimately met criteria for inclusion in the meta-analysis after full text review. From these, we identified 21 risk prediction scores for LGIB, although only four had sufficient number of papers to meta-analyze (Oakland, Strate, NOBLADS, and BLEED score). For the outcome safe discharge from hospital, the Oakland score had an area under the receiver operating characteristics curve (AUROC) of 85.5% (95% CI: 82.1%, 88.3%). For the outcome major bleeding, the Oakland score had an AUROC of 78.9% (95% CI: 75.1%, 82.2%); the Strate score had an AUROC of 74.4% (95% CI: 70.4%, 78.0%); the NOBLADS score had an AUROC of 60.3% (95% CI: 55.9%, 64.5%); and the BLEED score had an AUROC of 65.6% (95% CI: 61.4%, 69.7%). For the outcome, need for hemostasis, the Oakland had an AUROC of 99.0% (95% CI: 97.7%, 99.6%); the Strate score had an AUROC of 82.1% (95% CI: 78.5%, 85.2%); the NOBLADS score had an AUROC of 23.9% (95% CI: 20.3%, 27.8%). For the outcome, need for transfusion, the Oakland score had an AUROC of 99.0% (95% CI: 97.7%, 99.6%); the NOBLADS score had an AUROC of 87.7% (95% CI: 84.5%, 90.3%). Conclusions The Oakland score was the most discriminative risk prediction model for safe discharge from hospital, major bleeding, need for hemostasis, and need for transfusion. Funding Agencies None

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.