Acute lower gastrointestinal bleeding in the United Kingdom

Abstract

Introduction Lower gastrointestinal bleeding (LGIB) is a common medical emergency. Despite this, there are few data on presentations, interventions and patient outcomes. The aim of this research is to describe the epidemiology of LGIB in the United Kingdom, appraise clinical interventions and develop a risk score to optimise patient triage.

Methods A systematic review (SR) of common interventions for LGIB was undertaken. A prospective national observational study of LGIB presentations to UK hospitals during two months in 2015 was conducted, collecting data on demographics and outcomes (re-bleeding, death, re-admission). Frequency of intervention was compared to that identified in the SR. Multivariable logistic regression was used to develop a risk score to predict need for hospitalisation, and independently validated using cases collected by two additional acute hospitals using C-statistics.

Results The SR identified that colonoscopy within 24 hours of presentation was associated with a higher diagnostic yield and a reduced length of hospitalisation. In total 143/174 (82.2%) acute hospitals in the UK participated in the observational study, providing data on 2528 cases of LGIB. Most patients were elderly (median age 74 years) with major co-morbidities, including 29.4% taking antiplatelets drugs. Only 52% patients underwent inpatient investigation, 3.9% receiving colonoscopy. Median length of stay was 3 days and 85/2528 (3.4%) died during admission. Data from 2336 acute admissions were used to develop the risk score, identifying that >60% patients could avoid hospital admission. Age, gender, previous LGIB admission, blood on rectal examination, heart rate, systolic blood pressure and haemoglobin strongly discriminated safe discharge (c-statistic 0·84, 95% confidence interval 0·82 to 0·86).

Conclusion Few patients admitted with LGIB require treatment for bleeding, or come to harm. Many could safely be managed as an outpatient. We developed and validated a clinical risk score with good discriminative performance to identify these patients.