Article Text

PDF
OC-004 Development and validation of a prognostic scoring system for lower gastrointestinal bleeding
  1. G Leggett,
  2. C Smith,
  3. JM Thomson,
  4. B Vijayan,
  5. A Fraser,
  6. JS Leeds
  1. Gastroenterology, Aberdeen Royal Infirmary, Aberdeen, UK

Abstract

Introduction Lower gastrointestinal bleeding (LGIB) is a common and heterogeneous condition, in which there is a paucity of data concerning predictors of adverse outcomes. This study aimed to identify independent risk factors for adverse outcomes in LGIB and derive prognostic scoring systems to stratify patients by risk on admission.

Method The Aberdeen bleeding unit opened in 1991 and has recorded demographics, presenting features, haemodynamic status and outcomes on all admissions in a comprehensive database. Analysis was performed on admissions due to LGIB over the period 1991 to 2005. Independent risk factors for re-bleeding, requirement for surgical intervention, and mortality at 30 days were elucidated by means of univariate and multivariate binary logistic regression analyses. Risk factors were then modelled into simple numerical prognostic scoring systems which underwent preliminary validation using the period 2015–2016 to determine their predictive performance. Threshold analysis with a score <1 was performed to assess suitability of the score for determining early discharge.

Results The derivation cohort consisted of 2385 patients admitted with LGIB over a 15 year period. Re-bleeding was experienced in 322 (13.5%), 135 (5.7%) required surgery and 129 (5.6%) died within 30 days of admission. Multivariate analysis revealed that rebleeding was associated with inpatient status at the time of initial bleed (OR 1.8), age 60–79 (OR 1.5), age >80 (OR 2.1), syncope (OR 2.3), underlying malignancy (OR 2.1), hypotension (OR 2.3) and haemoglobin <10 g/dL (OR 5.0). 30 day mortality was associated with inpatient status at the time of initial bleed (OR 3.3), age 60–79 (OR 3.3), age >80 (OR 6.0), underlying liver disease (OR 7.2), hypotension (OR 2.9) and tachycardia (OR 2.1). Modelling provided a 30 day rebleeding risk score (0–7) and mortality risk score (0–7) with area under ROC curves 0.742 and 0.802 respectively. A score of 0 reflected a re-bleeding risk of 1.1% (95% CI 0.4–2.7) and 30 day mortality of 0.0% (95% CI 0.0–0.1), whereas a score of 6 reflected a re-bleeding risk and 30 day mortality risk of 50% in both scoring systems. In the validation cohort (n=121, median age 68 years, 60 males), application of the scoring systems gave 30 day rebleeding and mortality rates that were not significantly different from those predicted. A score of 0 gave a 30 rebleeding risk of 0.0% (95% CI 0.0–9.8) and 30 day mortality risk of 0.0% (95% CI 0.0%–10.7%).

Conclusion LGIB rebleeding and mortality can be reliably predicted using these scoring systems. Patients scoring 0 may be eligible for safe early discharge and outpatient management. Further external validation and confirmation is required.

Disclosure of Interest None Declared

  • Gastrointestinal bleed
  • prognosis

Statistics from Altmetric.com

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.