Gastrointestinal bleeding (GIB) represents a prevalent reason for hospital admissions due to medical emergencies, and it is associated with significant morbidity, mortality, and extensive utilization of hospital resources.

The incidence of GIB varies widely across studies, with rates ranging from 15 to 172 per 100.000 person-years for the upper gastrointestinal bleeding (UGIB) and from 20.5 to 87 per 100 000 person-years for the lower gastrointestinal bleeding (LGIB) [1]. Over the last decade, the pattern of GIB leading to hospitalization has shifted, with UGIB events declining while LGIB events have slightly increased, especially among the elderly population. These divergent trends in GIB incidence are mostly attributed to the increased use of medications reducing the UGIB risk (e.g. proton pump inhibitors and Helicobacter pylori eradication therapies), and the rising use of antithrombotic drugs, alongside the lack of effective preventive strategies for LGIB [2].

The most common cause of UGIB is peptic ulcer bleeding, followed by gastroduodenal or oesophageal erosive disease, variceal bleeding, Mallory Weiss tears, malignancy and vascular lesions. For LGIB, diverticular disease accounts for over 20 % of hospitalization and represents the most common cause of bleeding originating distally to the ileocecal valve. Anorectal conditions, specifically haemorrhoids, are the second prevalent cause of bleeding in hospitalized patients. Other causes of LGIB include various forms of colitis (ischemic, inflammatory, infectious), post-polypectomy bleeding, vascular abnormalities (e.g., angioectasias, Dieulafoy's lesion), colorectal neoplasia, rectal/stercoral/NSAID-induced ulcers, and radiation proctopathy. Notably, no identifiable cause is found in 3–19 % and up to 23 % of patients admitted for acute upper and lower GIB, respectively [3].

The outcome of gastrointestinal bleeding can be highly variable. While bleeding stops spontaneously in the majority of patients at their presentation, there remains a subgroup of patients who continue to bleed or develop recurrent bleeding. Despite advancements in patient management and endoscopic treatment, the in-hospital fatality rate of GIB events remains significant, ranging from 2 % to 15 % for UGIB and from 0.5 % to 8 % for LGIB [4].

Assessing risk during the initial patient evaluation is a crucial step in the clinical process aimed at distinguishing patients who need hospitalization, which are at risk of unfavourable outcomes and therefore require prompt interventions and intensive care, and those who can be suitable for safely discharged for outpatient care. This process is often complex, depending on several factors such as the severity of the GIB at presentation, the cause of bleeding and, not least, the patient's age and comorbidities that greatly influence the bleeding outcome.

To facilitate the clinician's decision-making process, several clinical prediction models, or risk scores, have been developed. Typically, such tools ought to be easy to calculate and should include a limited number of easily obtainable clinical variables. Furthermore, they should exhibit high accuracy in predicting clinically relevant outcomes and, most importantly, they should add value compared to the clinician's mere intuition and judgement (“gut feeling”).

This article will describe the existing risk score systems for GI bleeding and will review their strengths, weakness, limitations for use in clinical practice, and potential future use and development.