Risk Scores in Acute Lower Gastrointestinal Bleeding: Current Evidence and Clinical Applications

Abstract

Lower gastrointestinal bleeding (LGIB) is a frequent and potentially life-threatening clinical condition. Over the past two decades, several prognostic scoring systems have been developed to stratify risk and guide the management of patients with LGIB. This comprehensive review aims to summarize and compare the current evidence on the utility, accuracy, and limitations of key LGIB scoring systems, including the Glasgow-Blatchford Score (GBS), AIMS65, ABC score, Oakland score, SALGIB, CHAMPS, and Rockall score. We conducted a structured literature review of studies evaluating these scores in adult patients with LGIB. For each scoring system, we analyzed its origin, components, intended use, and predictive performance regarding clinical outcomes such as severe bleeding, transfusion requirement, in-hospital mortality, rebleeding, and safe discharge. Comparative analyses of diagnostic accuracy were extracted where available. Our findings indicate that while no single score offers comprehensive predictive accuracy across all outcomes, certain tools are particularly effective for specific endpoints. The Oakland and GBS scores are useful for identifying patients at low risk who may be managed safely as outpatients. The ABC and CHAMPS scores demonstrate superior performance in predicting mortality, especially in elderly or comorbid populations. SALGIB, a newer score developed in Vietnam, shows promising performance for early triage but requires further validation. The Rockall score, although historically valuable in upper GI bleeding, offers limited applicability in LGIB due to its reliance on post-endoscopic findings. In conclusion, multiple prognostic tools are now available to support early decision-making in LGIB. Their optimal use requires understanding their strengths, limitations, and appropriate clinical contexts. Integrating these scores into routine practice, along with clinical judgment, can enhance patient outcomes and resource allocation.

1. Introduction

Lower gastrointestinal bleeding (LGIB) is a frequent gastrointestinal emergency encountered in both medical and surgical practice. Epidemiological studies have indicated that the incidence of LGIB (approximately 1.26 per 1000 person-years) has surpassed that of upper gastrointestinal hemorrhage (UGIB, approximately 0.94 per 1000 person-years) in recent years [1]. The incidence and complication rates of LGIB have been rising, particularly among the elderly, contributing to longer hospital stays and more complex management requirements [1,2,3]. Although most LGIB episodes are self-limited, a significant subset of patients experience severe bleeding that necessitates blood transfusions, endoscopic or radiologic interventions, surgery, or can even result in death. This has heightened the importance of diligent LGIB management and risk assessment in clinical practice.

Early prediction of disease severity and patient prognosis is crucial for guiding management decisions. Prompt and appropriate intervention for high-risk patients can reduce complications and mortality, while safely managing low-risk cases conservatively or as outpatients can avoid unnecessary procedures and hospitalizations. In recent years, numerous prognostic scoring systems have been studied and applied to stratify risk and predict clinical outcomes in patients with LGIB. These tools provide an evidence-based framework to support clinical decision-making by identifying patients at higher risk for adverse outcomes (such as severe bleeding, rebleeding, need for intervention, or death) and those at lower risk who might be candidates for early discharge.

However, no single scoring system has proven sufficiently accurate for predicting all important in-hospital outcomes of LGIB (e.g., severe bleeding, transfusion need, rebleeding, intervention, or mortality). Each score has its own strengths and limitations. This review summarizes the current diagnostic approaches to LGIB and provides an overview of the prognostic scoring systems being used in clinical practice for LGIB, highlighting their performance, utility, and limitations in guiding management.

2. Materials and Methods

2.1. Literature Search Strategy

This review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. A comprehensive literature search was performed using PubMed, Scopus, and Web of Science databases for articles published up to April 2025. Search terms included “lower gastrointestinal bleeding”, “risk score”, “prognostic model”, “Glasgow-Blatchford”, “Oakland score”, “AIMS65”, “ABC score”, “CHAMPS”, “SALGIB”, and “Rockall score”.

2.2. Inclusion and Exclusion Criteria

We included original clinical studies (prospective or retrospective), systematic reviews, and meta-analyses that reported on the development, validation, or comparative performance of risk scores in adult patients with acute LGIB. Studies focused solely on upper GI bleeding, pediatric populations, or non-acute bleeding episodes were excluded.

2.3. Data Extraction and Synthesis

For each included study, we extracted data on study design, population characteristics, scoring system evaluated, outcomes predicted (e.g., transfusion, mortality, rebleeding), and diagnostic performance metrics (e.g., area under the ROC curve [AUROC], sensitivity, specificity). The scores were then summarized descriptively, highlighting key components, intended applications, and validated endpoints. Where comparative studies were available, we discussed relative strengths and limitations.

This review does not involve new human participants and did not require ethical approval.

3. Diagnostic Approaches to LGIB

LGIB is conventionally defined as bleeding originating from a source distal to the ligament of Treitz in the gastrointestinal tract. The clinical presentation is typically hematochezia (passage of bright red or maroon blood per rectum). Most patients with LGIB present with red or maroon-colored blood in stool; melena is less common and usually suggests an upper GI source, except in cases of a right colon or small bowel bleed where transit is slow. Notably, about 10–15% of apparent LGIB cases are actually due to an upper GI hemorrhage that is brisk enough to manifest as hematochezia; in such instances, hemodynamic instability often accompanies the bleeding [4].

Severity Definitions: Severe LGIB is typically defined as persistent bleeding within the first 24 h and/or rebleeding after 24 h of stability, accompanied by a drop in hematocrit > 20% or the need for transfusion of ≥2 units of packed red blood cells. Rebleeding is defined as renewed bleeding after initial stabilization (after 24 h), with similar criteria of hematocrit drop > 20% or requiring ≥2 units transfused [5].

Colonoscopy: Most patients with suspected LGIB should undergo colonoscopy for diagnostic and potential therapeutic purposes. According to the American College of Gastroenterology (ACG) clinical guideline updates in 2023, colonoscopy is recommended as the first-line diagnostic procedure in hemodynamically stable patients with acute LGIB. Colonoscopy permits direct visualization of the colonic mucosa, localization of the bleeding source, biopsy of lesions, and endoscopic hemostatic therapy if needed, all in a single session. In practice, colonoscopy for LGIB is often performed during the index hospitalization after adequate resuscitation and bowel preparation. Notably, there is no high-quality evidence that urgent colonoscopy (within 24 h of presentation) improves rebleeding or mortality rates compared to routine inpatient colonoscopy [6].

Computed Tomography Angiography (CTA): In hemodynamically unstable patients or when rapid localization of bleeding is needed, CTA is a valuable diagnostic tool. CTA can accurately identify the site of active bleeding and is widely available, with the advantage of not requiring bowel preparation. Reported sensitivity and specificity of CTA for localizing LGIB are approximately 90% and 92%, respectively. A positive CTA can guide urgent intervention (angiography or surgery) to the bleeding site [6].

Angiography and Interventional Radiology: Conventional mesenteric angiography is both a diagnostic and therapeutic modality in LGIB. Angiography can detect active arterial bleeding (usually requiring a bleeding rate >0.5 mL/min) and enables therapeutic embolization of the bleeding vessel. This approach is typically reserved for cases of ongoing bleeding where endoscopy is not feasible or has been unsuccessful. While angiographic embolization can control hemorrhage, it requires active bleeding for detection and carries risks of bowel ischemia and other complications [6].

In summary, the diagnostic evaluation of LGIB involves a combination of endoscopic and radiologic approaches. Hemodynamically stable patients are generally investigated with colonoscopy, whereas unstable patients or those with massive bleeding may benefit from rapid localization via CTA and possible therapeutic angiographic intervention. These diagnostic measures, coupled with clinical assessment, set the stage for risk stratification using prognostic scores, which can inform further management.

4. Overview of Prognostic Scoring Systems

Numerous scoring systems have been explored for predicting outcomes in LGIB. Many of these scores were initially developed for upper GI bleeding and later validated or applied to LGIB, while some have been designed specifically for LGIB. Broadly, most LGIB risk scores are applied before endoscopy (pre-endoscopic), utilizing clinical and laboratory parameters available at presentation. This allows for early risk stratification even before diagnostic procedures. By contrast, the Rockall score incorporates endoscopic findings (in its complete form) and is considered a post-endoscopic risk score. Below, we review the key prognostic scoring systems for LGIB, including their origin, components, and performance.

4.1. Glasgow-Blatchford Score (GBS)

The Glasgow-Blatchford Score (GBS) was originally developed in 2000 by Blatchford et al. for upper GI hemorrhage. It is a pre-endoscopy score that uses only clinical and laboratory criteria to identify patients at risk of requiring treatment (transfusion or intervention) for GI bleeding. Although designed for UGIB, GBS has been studied and applied in LGIB with comparable effectiveness [7].

Components: GBS consists of simple parameters measured at presentation: heart rate, systolic blood pressure, blood urea nitrogen, hemoglobin, and clinical markers of bleeding or comorbidity (melena, syncope, hepatic disease, cardiac failure). Each factor is assigned a point value (0–6 points each), for a maximum total score of 23 points. Higher scores indicate greater risk.

Table 1. Glasgow-Blatchford Score (GBS) [8].

Parameter	Points	Risk Category a
Heart rate ≥ 100/min	1	Low risk: GBS 0–1 High risk: GBS ≥ 5
Systolic BP (mmHg)
100–109	1
90–99	2
<90	3
Blood urea (mmol/L)
6.5–7.9	2
8–9.9	3
10–24.9	4
≥25	6
Hemoglobin (male) (g/dL)
12.0–12.9	1
10.0–11.9	3
<10	6
Hemoglobin (female) (g/dL)
10.0–11.9	1
<10	6
Melena	1
Syncope	2
Hepatic disease	2
Heart failure	2
Total score (max)	23

^a GBS 0–1 indicates low risk (patients often suitable for outpatient management), whereas a score ≥ 5 suggests high risk of requiring intervention or transfusion.

outlines the GBS scoring system and risk interpretation [8].

Clinical Performance: A 2018 study by Ur-Rahman et al. found that GBS predicted the need for intervention and in-hospital mortality in LGIB with an area under the ROC curve (AUROC) of approximately 0.76, similar to its performance in UGIB [5]. In that study, 39.9% of high-risk LGIB patients (GBS above a certain threshold) had a primary adverse outcome, compared to only 7.1% of low-risk patients. GBS has demonstrated good sensitivity for detecting patients who will require blood transfusion or have severe bleeding (AUROC approximately 0.83 for severe LGIB) [9]. Importantly, a low GBS (e.g., 0 or 1) has a high negative predictive value for adverse events, which can identify patients safe for early discharge. Indeed, Oakland et al. (2017) reported that GBS was effective in identifying LGIB patients suitable for outpatient management (the AUROC was approximately 0.80 for predicting safe discharge) [9].

Limitations: The GBS has only modest specificity for predicting outcomes like mortality and rebleeding in LGIB. Its positive predictive value for adverse events is relatively low [8]. For example, while very useful to “rule out” high-risk cases, a high GBS alone does not definitively “rule in” the need for intervention due to low specificity. Some studies have found that GBS is no better than dedicated LGIB scores (e.g., SALGIB or Oakland) for predicting transfusion need. Notably, GBS does not incorporate findings of colonoscopy and has no capability to predict the need for endoscopic therapy in LGIB (AUROC approximately 0.50 for predicting hemostatic intervention). Thus, while GBS is very useful for initial triage (especially to identify low-risk patients), it should be used in conjunction with clinical judgment for definitive management decisions. It is worth noting that the inclusion of melena as a criterion may reduce GBS specificity in LGIB, as melena is typically associated with upper GI bleeding and is less commonly seen in lower tract sources.

4.2. AIMS65 Score

The AIMS65 score is a pre-endoscopic mortality risk score originally developed for upper GI bleeding. “AIMS65” is an acronym for its five parameters: Albumin < 3.0 g/dL, INR > 1.5, altered Mental status, Systolic BP ≤ 90 mmHg, and Age ≥ 65 years. Each criterion scores 1 point, for a total of 0–5 points (

Table 2. AIMS65 scoring system [9].

Criterion	Points	Risk Category b
Albumin < 3.0 g/dL	1	Low risk: 0–1 points High risk: ≥2 points
INR > 1.5	1
Altered mental status	1
Systolic BP ≤ 90 mmHg	1
Age ≥ 65 years	1
Total score (max)	5

^b A score of 0 or 1 indicates low risk, whereas ≥ 2 points suggests high risk of in-hospital mortality.

).

Development: AIMS65 was derived and validated by Saltzman et al. (2011) using data from 29,222 UGIB hospital admissions in the United States, with the aim of predicting in-hospital mortality, length of stay, and cost [10]. Subsequent studies have evaluated AIMS65 in LGIB populations. Oakland et al. (2017) reported that AIMS65 also predicts mortality in LGIB, with an AUROC around 0.78 [9].

Clinical Application in LGIB: In LGIB, AIMS65 has been shown to stratify mortality risk. Oakland et al. (2017) [9] found AIMS65 to be the best predictor of 30-day mortality among several scores (AUROC 0.78 for mortality in LGIB, outperforming GBS and Rockall). However, AIMS65 is not as useful for other outcomes in LGIB. It has poor predictive value for rebleeding or the need for transfusion (AUROC approximately 0.63 for both), and it does not predict the need for intervention (endoscopic or surgical).

Strengths: AIMS65 is simple and uses readily available clinical data (aside from albumin). Being a pre-endoscopy score, it can be calculated on admission to quickly flag high-risk patients who have a significantly elevated mortality risk. Among UGIB/LGIB scores, AIMS65 has been highlighted for its mortality prediction—one study noted it was the most accurate for mortality compared to GBS and Rockall in LGIB.

Limitations: One practical limitation of using AIMS65 on presentation is that serum albumin is not always measured routinely on immediate admission. If the albumin is not available early, it limits the calculation of the score. Moreover, AIMS65 was designed specifically to predict mortality—it deliberately does not attempt to identify low-risk patients for safe discharge. In fact, it is not recommended for identifying patients who can be managed as outpatients. Its poor performance in predicting rebleeding or transfusion needs means it should be supplemented with other assessments for those outcomes [11].

4.3. ABC Score

The ABC score is a more recently developed pre-endoscopy risk score that combines Age, Blood tests, and Comorbidities to predict mortality in GI bleeding. Laursen et al. (2021) derived the ABC score from an international multicenter cohort of UGIB and LGIB patients, aiming to improve mortality prediction. Uniquely, ABC was created using both upper and lower GI bleed data, making it broadly applicable [12].

Components: The ABC score includes eight factors assessed at admission (

Table 3. ABC score components and risk strata [12].

Factor	Points	30-Day Mortality Risk c
Age (years)		Low risk: ≤ 3 (mortality approximately 0.6%) Medium risk: 4–7 (mortality approximately 6.3%) High risk: ≥ 8 (mortality approximately 18%)
60–74	1
≥75	2
Blood urea nitrogen > 10 mmol/L	1
Albumin < 30 g/L	1
Creatinine (µmol/L)
100–150	1
>150	2
Comorbidities
Altered mental status	2
Cirrhosis	2
Metastatic cancer	4
ASA class III (severe systemic disease)	1
ASA class ≥ IV (life-threatening disease)	3
Total score range	0–17

^c Risk categories for 30-day mortality are based on the ABC score: ≤3 (low), 4–7 (intermediate), ≥8 (high). All albumin values now consistently appear in g/L.

): patient age, several laboratory values (blood urea, hemoglobin/hematocrit, albumin, creatinine), and comorbid conditions (altered mental status, cirrhosis, metastatic cancer, and an ASA physical status score for overall health). Each component is weighted 1–6 points. The total score stratifies patients into low, intermediate, or high risk for 30-day mortality.

Predictive Performance: The ABC score has demonstrated excellent discrimination for mortality in LGIB. In the original study, ABC had an AUROC of approximately 0.84 for 30-day mortality in LGIB, outperforming AIMS65 (0.75), GBS (0.74), and Oakland (0.69). ABC was also able to predict both bleeding-related and non-bleeding-related 30-day mortality with high accuracy (combined outcome: AUROC approximately 0.85). These findings suggest that ABC is a robust tool for the early identification of patients at high risk of death from LGIB [12].

Strengths: As a pre-endoscopic score, ABC allows early risk stratification. It incorporates a broad view of patient status by including comorbidities (Charlson Comorbidity Index via ASA class) and lab markers of organ perfusion/function (BUN, creatinine, albumin) in addition to age. This comprehensive approach likely contributes to its strong mortality prediction [12]. Clinically, an elevated ABC score on admission can prompt aggressive management and intensive monitoring for high-risk patients.

Limitations: The trade-off for ABC’s accuracy is increased complexity. It requires collecting eight variables, including some (albumin, ASA class) that may not be immediately available or routinely assessed in all settings. This may limit its ease of use compared to simpler scores. Additionally, ABC is focused on mortality prediction. Studies have found it less accurate for predicting the need for hemostatic intervention (AUROC approximately 0.68) or rebleeding within 7 days (AUROC approximately 0.63). Thus, while excellent for mortality risk stratification, the ABC score is not a standalone predictor for other clinical endpoints like rebleeding or intervention [13].

4.4. Oakland Score

The Oakland score is a specialized risk score derived specifically for LGIB. Oakland et al. (2017) [9] developed this score using data from 2336 patients with acute LGIB in the UK, aiming to identify patients at a low risk for adverse outcomes who could be safely managed without hospitalization. Adverse outcomes were defined as rebleeding, need for transfusion or intervention, re-hospitalization, or death within 28 days. The Oakland score ranges from 0 to 35 points, with higher scores indicating greater risk of adverse outcome [9].

Components: The Oakland score is calculated from seven variables measured on presentation: age, sex, prior hospital admission for LGIB, findings on digital rectal exam (presence of blood), heart rate, systolic blood pressure, and hemoglobin level. These factors reflect patient demographics, hemodynamic status, and bleeding severity. The detailed point assignment is shown in

Table 4. Oakland score components [14].

Factor	Points
Age (years)
<40	0
40–69	1
≥70	2
Sex
Female	0
Male	1
Previous LGIB admission
No	0
Yes	1
Rectal exam
No blood	0
Blood present	1
Hemoglobin (g/L)
≥160	0
130–159	2
110–129	8
90–109	13
70–89	17
<70	22
Heart rate (min)
<70	0
70–89	1
90–109	2
≥110	3
Systolic BP (mmHg)
<90	5
90–109	4
110–129	3
130–159	2
≥160	0
Total score range:	0–35

.

Clinical Use and Cutoffs: The Oakland score’s primary use is to identify low-risk LGIB patients who can be safely discharged. In the original UK study, an Oakland score ≤ 8 was proposed as a cutoff for “safe discharge,” which corresponded to a 95% sensitivity for no adverse outcomes. A subsequent validation in a Vietnamese cohort suggested a slightly higher cutoff (≤10) to achieve approximately 97% sensitivity for the absence of adverse events [5]. In practice, patients with low Oakland scores (≤8–10) and a stable condition can often be managed as outpatients.

Performance: The Oakland score has been shown to outperform some traditional scores for certain outcomes in LGIB. For example, in an Asian validation study, Oakland better predicted the need for blood transfusion than GBS (AUROC 0.93 vs. 0.87). Oakland was also the single best predictor of severe LGIB (defined by significant bleeding or the need for intervention) among the scores tested (AUROC approximately 0.90, compared to 0.83 for GBS and 0.87 for SALGIB). It demonstrated moderate accuracy for mortality prediction, with an AUROC of approximately 0.63 for in-hospital death. Importantly, the Oakland score was specifically highlighted by the ACG in 2023 as a recommended tool to stratify LGIB patients and guide management decisions, such as early discharge for low-risk cases [14].

Advantages: The Oakland score is tailored for LGIB and uses simple clinical criteria and basic lab results that are routinely available. Being a pre-endoscopy score, it does not require colonoscopic findings. It has demonstrated strong ability to predict which patients will not have adverse outcomes, thus supporting safe discharge decisions. It also correlates well with the severity of bleed—higher scores reflect higher risk of transfusion or severe bleeding.

Limitations: Despite its strengths in certain areas, the Oakland score is less robust for predicting some outcomes. Studies indicate that it has limited accuracy in predicting the need for urgent hemostatic intervention (endoscopy or surgery) in LGIB (AUROC approximately 0.52) [5]. Its performance in predicting rebleeding during hospitalization is also low (AUROC approximately 0.46). Additionally, while useful for identifying low-risk patients, a very high Oakland score does not pinpoint exactly which type of intervention a patient might need—it simply signals high overall risk. As with all scores, clinical judgment is required, especially for intermediate-risk patients [14].

Clinical note: In practice, an Oakland score cut-off (e.g., ≤8) can be used alongside clinical criteria to select patients for outpatient management. However, factors like patient frailty, ongoing bleeding, or unreliable follow-up might still warrant admission despite a low score. The higher cutoff (≤10) validated in the Vietnamese population likely reflects population differences in baseline hemoglobin, comorbid burden, and local thresholds for intervention or admission.

4.5. SALGIB Score

The SALGIB score (Severe Acute Lower Gastrointestinal Bleeding score) is a prognostic tool developed specifically to predict the risk of severe outcomes in LGIB. Quach et al. derived the SALGIB score from a cohort of 357 LGIB patients in Vietnam and then validated it on a separate multicenter cohort of 324 LGIB patients. The primary aim was to predict “severe LGIB,” generally corresponding to persistent or major bleeding requiring significant intervention [15].

Components: SALGIB is a simple pre-endoscopic score comprising four variables measured at presentation: heart rate, systolic blood pressure, hematocrit, and platelet count. Each factor is assigned points, with a higher total indicating higher risk. The scoring schema is given in

Table 5. SALGIB score components and risk interpretation [5].

Factor	Points	Risk Category
Heart rate		Low risk: SALGIB < 2 High risk: SALGIB ≥ 3
≥100/min	1
<100/min	0
Systolic BP (mmHg)
<100 mmHg	3
≥100 mmHg	0
Hematocrit (%)
30–34.9	1
25–29.9	3
<25	5
≥35	0
Platelet count
≤150 × 103/µL	1
>150 × 103/µL	0
Total score range	0–10

. The maximum SALGIB score is 10 points [5].

Performance: The SALGIB score demonstrated high accuracy in identifying severe LGIB in the development and validation cohorts. In the initial study, SALGIB predicted severe bleeding with AUROC 0.91 (derivation) and 0.86 (validation). Patients with SALGIB < 2 had very low rates of severe bleeding (3.7% in the first cohort, 1.2% in the validation cohort). This indicates that a SALGIB score below 2 reliably signifies low risk of a severe bleeding course. Comparatively, SALGIB performed on par with the Oakland score in predicting adverse LGIB outcomes. For instance, SALGIB and Oakland had equivalent efficacy in predicting composite adverse outcomes (rebleeding, intervention, etc.) in Quach et al.’s study. SALGIB was particularly strong in predicting the need for transfusion (AUROC approximately 0.91) and identifying patients who would develop hemodynamic instability or severe bleeding. It also showed good discrimination for in-hospital mortality (AUROC approximately 0.82) [5].

Advantages: SALGIB is a pre-endoscopy score that uses only four readily available parameters (vital signs and basic labs). This makes it quick and convenient for initial triage. Its simplicity (fewer criteria than Oakland) can facilitate easy adoption in busy clinical settings. SALGIB’s strong performance in predicting severe bleeding and transfusion need means it can alert clinicians early to patients who are likely to need aggressive resuscitation or intervention. In comparative studies, SALGIB’s ability to predict severe LGIB was similar to the best existing scores, but with the benefit of simplicity.

Limitations: Subsequent research has highlighted some limitations. A 2024 study by Li et al. found that SALGIB was poor at predicting rebleeding (AUROC approximately 0.64) and only modest in predicting the need for hemostatic intervention (AUROC approximately 0.71). The original SALGIB study by Quach et al. also noted that SALGIB did not effectively predict the need for endoscopic or surgical intervention (AUROC approximately 0.53 in their cohort). Additionally, there were insufficient data to determine SALGIB’s ability to predict safe discharge (i.e., identify patients who could be managed as outpatients). As a relatively new score, SALGIB might require further validation in diverse populations outside of Vietnam to confirm its generalizability [16].

4.6. CHAMPS Score

The CHAMPS score is a six-factor pre-endoscopic score initially developed for upper GI bleeding but recently studied in LGIB. The acronym CHAMPS reflects its components: Charlson Comorbidity Index ≥ 2, in-Hospital onset, Albumin < 2.5 g/dL, altered Mental status, Performance status (ECOG) ≥ 2, Steroid use. Matsuhashi et al. first derived CHAMPS in 2021 on a cohort of 2205 UGIB patients in Japan to predict in-hospital mortality [17,18]. Tajika et al. (2022) applied CHAMPS to 387 LGIB cases and validated its efficacy in the lower GI context [11].

Components: Each of the six CHAMPS criteria is assigned 1 point if present (

Table 6. CHAMPS score components and mortality risk [11].

Criterion	Points	Predicted In-Hospital Mortality d
Charlson Comorbidity Index ≥ 2	1	Low risk: ≤1 (mortality approximately 1.8%) Medium risk: 2–3 (mortality approximately 15.8%) High risk: ≥4 (mortality approximately 37.1%)
Bleeding onset as inpatient (hospitalized)	1
Albumin < 2.5 g/dL	1
Altered mental status	1
ECOG performance status ≥ 2	1
Current corticosteroid use	1
Total score range	0–6

^d Risk categories are based on Tajika et al.’s LGIB data.

). Notably, CHAMPS is unique in including in-hospital onset of bleeding as a risk factor (distinguishing patients who bleed during hospitalization for another reason). It also employs two indices requiring clarification: the Eastern Cooperative Oncology Group (ECOG) [19] performance status and the Charlson Comorbidity Index (CCI) [20]. In CHAMPS, ECOG ≥ 2 (indicating at least some limitation in self-care; ECOG ranges from 0 to 5 with higher = worse functional status) scores 1 point, and Charlson Comorbidity Index ≥ 2 (indicating significant comorbid burden) scores 1 point. Albumin < 2.5 g/dL, altered mental status, and chronic steroid use are the other variables, each contributing 1 point.

Predictive Performance: In the LGIB study by Tajika et al. (2022) [11], CHAMPS demonstrated excellent performance for predicting in-hospital mortality. It achieved an AUROC of 0.80 for mortality, which was significantly higher than other scores like GBS, Rockall, ABC, or AIMS65 in that comparison. CHAMPS was the most accurate score for identifying high-risk LGIB patients (those who would die or have major complications) among the five scores tested. Moreover, CHAMPS is applicable to both outpatient-onset and inpatient-onset LGIB, a distinction many other scores do not consider. An attractive finding was that low-risk CHAMPS patients (score 0–1) had a mortality rate of approximately 1.8%, suggesting that the score could help identify patients who might be discharged early or managed conservatively if other factors allow.

Advantages: CHAMPS uses only clinical and common lab factors, and all criteria are assessed at the time of admission (pre-endoscopy). It is novel in accounting for bleeding that starts during hospitalization, which often portends worse outcomes due to patient frailty or comorbidities. The inclusion of ECOG performance status and Charlson Index brings a comprehensive assessment of a patient’s general health and comorbid burden, aspects highly relevant to outcome but not captured by purely bleeding-focused scores. CHAMPS can stratify LGIB patients whether their bleed began in the community or in the hospital, making it versatile [11]. Additionally, CHAMPS has been noted to potentially identify patients safe for early discharge when the score is low.

Limitations: The CHAMPS score’s complexity is a consideration—performance status (ECOG) and Charlson Index require calculation or familiarity, which might not be as instant as checking a lab value. However, these indices are straightforward once known: ECOG ≥ 2 roughly means the patient cannot carry out any work activities and is up and about <50% of the day; CCI ≥ 2 means at least two mild comorbidities or one moderately severe comorbidity (e.g., diabetes with end-organ damage, metastatic cancer, etc.). In Tajika’s study, CHAMPS did not significantly outperform other scores for predicting rebleeding, with an AUROC of 0.67 (compared to approximately 0.57–0.68 for the other scores). Thus, like other scores, CHAMPS is not very useful for foreseeing recurrent bleeding. Its main value lies in mortality and overall risk stratification. As CHAMPS was only recently applied to LGIB, further validation in different populations would be beneficial [11].

4.7. Rockall Score

The Rockall score is a well-known risk scoring system initially designed for UGIB, incorporating endoscopic findings into risk assessment. Important: The Rockall score comes in two forms: a clinical Rockall (using only pre-endoscopy variables) and a complete Rockall (including endoscopic diagnosis and stigmata of hemorrhage) [21]. Here we discuss the complete Rockall score (RS) as it was described by Rockall et al. in 1996, though its utility in LGIB is limited [9].

Components: The complete Rockall score has five components (

Table 7. Complete Rockall score parameters [21].

Parameter	Score 0	Score 1	Score 2	Score 3
Age (years)	<60	60–79	≥80	–
Shock (hemodynamics)	None (BP ≥ 100 and HR < 100)	Tachycardia (HR > 100)	Hypotension (BP < 100 mmHg)	–
Comorbidities (ASA class proxy)	None		Moderate disease (CHF, ischemic heart disease, etc.)	Severe disease (renal failure, liver failure, metastatic cancer)
Endoscopic diagnosis e	Mallory–Weiss tear, no lesion, or minor lesion not bleeding	Peptic ulcer, erosive disease, esophagitis	GI malignancy	–
Endoscopic stigmata	None or dark spot on ulcer base	Blood in upper GI tract, adherent clot, visible vessel	Active bleeding (spurting or oozing) f	–
Maximum total score	11

^e Originally defined for UGIB sources; not applicable to LGIB, since colonoscopic diagnoses differ. ^f (Active bleeding is not explicitly listed in the text snippet above but is part of Rockall’s original scoring for stigmata.)

): age, presence of shock (defined by heart rate and blood pressure), comorbidities, endoscopic diagnosis, and endoscopic stigmata of recent hemorrhage. Each component is 0–3 points; the maximum score is 11. Note that the endoscopic criteria (diagnosis and stigmata) are specific to upper GI bleeding lesions (e.g., Mallory–Weiss tear, peptic ulcer, etc.), so in LGIB their analog would be findings on colonoscopy (which are different). Therefore, formally, the Rockall score is not validated for LGIB. However, it is occasionally referenced for comparison.

Performance in LGIB: The Rockall score has limited applicability in LGIB because it requires endoscopic findings relevant to UGIB (e.g., peptic ulcer stigmata). If one were to apply the Rockall concept to LGIB, only the pre-endoscopy portion (age, shock, comorbidities) could be used prior to colonoscopy, and colonoscopic findings do not map onto Rockall’s UGIB categories. In a study by Oakland et al. (2017) on LGIB, the authors calculated Rockall scores for comparison and found that the Rockall score had only moderate accuracy for predicting mortality (AUROC approximately 0.75) and was inferior to LGIB-specific or other pre-endoscopy scores for most outcomes. Rockall was poor at predicting rebleeding, need for intervention, or safe discharge in LGIB (AUROCs generally 0.52–0.64 for these outcomes) [9].

Strengths: In UGIB, the Rockall score is well-established for stratifying risk of mortality and rebleeding, combining clinical and endoscopic data. In LGIB, however, its only potential use would be after colonoscopy by analogy—but even then, the scoring system does not align well with colonic lesions. If considering just the clinical Rockall (age/shock/comorbidity), those factors are indeed important: advanced age, hemodynamic instability, and significant comorbidities are universally recognized as risk factors for worse outcomes. In that sense, the essence of Rockall reinforces what other scores also capture [9].

Limitations: Rockall’s need for endoscopic information renders it not practical for initial triage of LGIB. By the time colonoscopy is performed, management plans are often already in motion. Moreover, as noted, Rockall does not effectively predict key LGIB outcomes aside from mortality. Current guidelines do not recommend using the Rockall score for LGIB patients [9]. It is primarily of historical interest and as a benchmark for newer scoring systems.

5. Clinical Application of Scoring Systems

Effective management of LGIB hinges on early identification of patients who require aggressive intervention versus those who can be managed conservatively. Although significant advances have been made in endoscopic techniques to diagnose and treat LGIB, the condition can still result in serious complications or death if not promptly addressed. Risk scoring systems provide an objective method to assess the severity of bleeding and the likelihood of adverse outcomes, thereby aiding clinical decisions such as level of care (e.g., ICU vs. ward), timing of interventions, and discharge planning.

An ideal LGIB risk score would be one that is easy to use, accurate, and reliable, so that no high-risk patient is overlooked (“missed”). In practice, clinicians may use a combination of scores to guide different aspects of management (

Table 8. Scoring systems for prognostication in lower gastrointestinal bleeding (LGIB).

Scoring System	Application	Advantages	Limitations
Glasgow-Blatchford	- Applied in patients with LGIB at initial presentation outside hospital. - Pre-endoscopic score. - Comprises 6 admission criteria.	- Simple parameters including clinical signs and routine labs. - No need for colonoscopy results. - Good prediction of severe LGIB and transfusion needs. - Good predictor of safe discharge.	- Moderate prediction of in-hospital mortality and rebleeding. - Low specificity and positive predictive value. - Does not predict need for hemostatic intervention.
AIMS65	- Applied in LGIB presenting outside hospital. - Pre-endoscopic score. - Comprises 5 admission criteria.	- Relatively simple evaluation. - No need for colonoscopy results. - Moderate predictor of mortality.	- Albumin not routinely available. - Does not predict rebleeding, transfusion, or need for intervention. - Not recommended for identifying low-risk patients for outpatient management.
ABC	- Applied in LGIB at initial presentation. - Pre-endoscopic score. - Comprises 8 admission criteria.	- Does not require colonoscopy results. - Good predictor of 30-day mortality.	- Relatively complex with non-routine laboratory tests. - Does not predict rebleeding or hemostatic intervention.
Oakland	- Applied in LGIB presenting outside hospital. - Pre-endoscopic score. - Comprises 7 admission criteria.	- Simple criteria. - No need for colonoscopy results. - Good predictor of severe LGIB, transfusion needs, and mortality. - Good predictor of safe discharge.	- Does not predict rebleeding or hemostatic intervention.
SALGIB	- Applied in LGIB at initial presentation. - Pre-endoscopic score. - Comprises 4 admission criteria.	- Simple evaluation. - No need for colonoscopy results. - Good prediction of severe LGIB, transfusion need, and in-hospital mortality.	- Poor prediction of rebleeding and intervention. - Predictive value for safe discharge remains inconclusive.
CHAMPS *	- Applied in both inpatient and outpatient LGIB. - Pre-endoscopic score. - Comprises 6 admission criteria.	- Utilizes clinical and laboratory criteria. - No need for colonoscopy results. - Good predictor of mortality risk.	- Relatively complex criteria. - Limited value in predicting rebleeding.
Rockall	- Applied in LGIB after hospital admission. - Post-endoscopic score. - Includes 3 clinical and 2 endoscopic criteria.	- Moderate predictor of mortality.	- Poor predictor of rebleeding, transfusion requirement, and hemostatic intervention.

* CHAMPS score mortality risk categories: low risk (≤1), medium risk (2–3), high risk (≥4); based on Tajika et al. 2022 [11].

):

Predicting Need for Urgent Intervention: To date, no scoring system has shown high accuracy in predicting the necessity of urgent endoscopic or surgical intervention for LGIB. All available scores have only modest performance in this area. Therefore, clinical judgment (e.g., ongoing bleeding, hemodynamic instability, transfusion requirement) remains paramount in deciding which patients need emergent colonoscopy or angiography.

Identifying Low-Risk Patients for Safe Discharge: Scores like GBS and Oakland have been particularly useful for identifying patients at low risk of adverse outcomes who may not require hospitalization. GBS ≤ 1 (and certainly 0) has been associated with very low rates of intervention or mortality, supporting outpatient management in UGIB, and similarly, in LGIB settings, this can be considered. The Oakland score (≤8 in UK, ≤10 in VN cohorts) has been prospectively studied as a criterion for safe discharge; its use can significantly reduce unnecessary admissions while maintaining patient safety. The CHAMPS score may also help in this regard—a CHAMPS of 0 or 1 identifies a group with <2% mortality, suggesting that truly low-risk LGIB patients could be discharged early if bleeding has ceased. In practice, if a patient’s bleeding has stopped and their risk score is low (and they have adequate support and follow-up), outpatient management can be contemplated.

Predicting Mortality: For assessing the risk of death, the newer scores have an edge. CHAMPS, SALGIB, and ABC have all shown superior discrimination for mortality (with AUROCs around 0.8 or higher). For example, CHAMPS and SALGIB were among the best predictors of mortality in comparative analyses. Clinicians dealing with an elderly LGIB patient with multiple comorbidities might find the CHAMPS score particularly relevant, as it directly accounts for comorbidity and functional status, highlighting those at high risk of dying. AIMS65, GBS, and Oakland have only moderate accuracy for mortality prediction (AUROCs approximately 0.63–0.75); therefore, while a high score on these tools may raise concern, clinicians should exercise caution—in critical prognostication, the patient’s overall condition and consideration of higher-performing scores are essential.

Predicting Severe Bleeding and Transfusion Requirement: GBS, Oakland, and SALGIB have demonstrated good ability to predict which patients will have severe bleeding courses or require blood transfusions. GBS includes blood urea and hemoglobin, capturing the effect of bleeding on volume status, which correlates with transfusion need. Oakland and SALGIB explicitly factor in hemodynamics and hemoglobin/hematocrit, directly targeting severe hemorrhage risk. If a patient has a high Oakland or SALGIB score on arrival, clinicians should be prepared for potential massive bleeding—for instance, having blood products ready and considering early intervention if feasible.

Each scoring system has particular strengths; thus, an approach some centers take is to calculate multiple scores for a given patient to obtain a more complete picture. However, using many scores can be impractical. It is reasonable to adopt one well-validated score locally for routine use (e.g., GBS or Oakland for triage) and be aware of others’ insights (e.g., know that albumin < 3 or an older patient on steroids is high risk, even if not fully captured by GBS).

It is also critical to emphasize that risk scores complement but do not replace clinical judgment. Guidelines (such as ACG 2023 [6]) endorse the use of a risk stratification tool like the Oakland score to guide management strategies. Still, scores are aids; they cannot account for every nuance of a patient’s condition. For example, a patient might formally score low risk but have persistent bleeding—obviously, they should be admitted and managed aggressively despite the score. Conversely, a patient might score high due to age or comorbidities but have a clear identified bleeding source that has stopped—clinicians might individualize decisions in such cases.

In summary, prognostic scores in LGIB are valuable for stratifying patients into risk categories, thereby informing decisions on level of monitoring, timing of colonoscopy, need for intervention, and disposition (admit vs. discharge). The combined use of a risk score with careful clinical evaluation and close patient monitoring offers the best approach to improving LGIB outcomes. As research continues, these scoring systems may be refined or new ones developed, but the multifaceted nature of LGIB outcomes suggests that no single score will ever be 100% predictive. A prudent strategy is to understand what each score offers and use them as part of a holistic clinical assessment.

6. Conclusions

Early risk stratification of patients with acute lower gastrointestinal bleeding is becoming an integral part of management, with the goal of reducing mortality and complications. Numerous scoring systems have been developed or adapted for LGIB in pursuit of this goal. However, so far no score can be considered a “gold standard” for predicting all outcomes in LGIB. Each scoring system has certain advantages and limitations, and none can accurately predict every outcome of interest (severe bleeding, mortality, rebleeding, need for transfusion, need for intervention) on its own.

In practice, applying risk scores for early prognostication in LGIB is very useful, but clinicians should not rely on any single score in isolation. It is essential to recognize the capabilities and shortcomings of each scoring system. For instance, some scores are better at identifying low-risk patients, while others excel at predicting mortality. Clinicians should integrate score results with fundamental clinical assessments—considering the patient’s comorbid conditions and baseline risk factors, and observing their clinical course and response to initial treatment. Careful monitoring and re-evaluation remain crucial, as a patient’s status can evolve rapidly.

In conclusion, risk scoring systems are valuable tools that enhance data-driven decision-making in LGIB. When used appropriately, they contribute to improved patient care by guiding resource allocation (such as ICU admission or urgent intervention) and identifying patients who might avoid unnecessary hospitalization. Ongoing research and validation studies will continue to refine these tools. Among available tools, the Oakland and SALGIB scores are particularly effective in predicting severe bleeding and transfusion requirements. The ABC and CHAMPS scores demonstrate strong performance for mortality risk stratification, especially in elderly or comorbid patients. For identifying low-risk patients suitable for discharge, the GBS and Oakland scores have proven utility. Recognizing these distinctions allows for more targeted application of each score in clinical decision-making. Ultimately, the optimal approach is to integrate these tools with clinical judgment and ongoing reassessment to tailor management strategies and improve outcomes in LGIB.