Risk factors for rebleeding in acute lower gastrointestinal bleeding: a systematic review and meta-analysis

Abstract

Objective

To investigate the potential risk factors for rebleeding following acute lower gastrointestinal bleeding (ALGIB).

Methods

A comprehensive literature review was performed utilizing the PubMed, Embase, Web of Science, and Cochrane Library databases, encompassing articles published up to May 10, 2024. Two researchers independently performed literature screening, data extraction, and risk of bias assessment to ensure methodological rigor and minimize potential biases. The meta-analysis was performed using R (version 4.3.2) and Stata 17.0 software.

Results

A total of 2799 studies were retrieved, of which 29 were ultimately included in the analysis, comprising 23 case–control studies and 6 cohort studies, involving 14,069 patients with ALGIB, of whom 1930 experienced rebleeding. The pooled results of multivariable meta-analysis indicated that age ≥ 70 years, blood transfusion requirements, shock at presentation, hypertension, comorbid chronic kidney disease (CKD), hyperlipidemia, NSAIDs use, active bleeding on colonoscopy, colonic diverticular bleeding, and non-aspirin antiplatelet treatment are risk factors for rebleeding after ALGIB. In addition, the results of univariable analysis also showed that steroids, aspirin, and anticoagulant drug use are associated factors for rebleeding.

Conclusion

Based on available evidence, several factors, including age ≥ 70 years, blood transfusion requirements, shock at presentation, hypertension, CKD, hyperlipidemia, NSAIDs use, active bleeding on colonoscopy, colonic diverticular bleeding, and non-aspirin antiplatelet treatment, may be associated with an increased risk of rebleeding after ALGIB. Early identification of these risk factors and implementation of individualized interventions may help improve patient outcomes.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00384-026-05172-0.

Introduction

ALGIB refers to bleeding originating from any site distal to the ileocecal valve, including the rectum and anal canal, accounting for approximately 15–40% of all gastrointestinal hemorrhage, and is a common gastrointestinal emergency that requires urgent hospitalization, and the common etiologies encompass colonic diverticulosis, ischemic intestinal disease, postoperative bleeding from colonic lesions, neoplastic conditions, inflammatory bowel disease, infections, radiation-induced colitis, vascular anomalies, rectal ulcers, and hemorrhoids [1–3]. The incidence is 21–87 per 100,000 annually. Additionally, the incidence of ALGIB is positively correlated with age and shows an overall upward trend [4, 5]. With the extension of human life expectancy, ALGIB is expected to become a serious health problem in the future [6].

A high rebleeding rate (11–26%) is the most severe complication of ALGIB; patients with rebleeding often require repeated blood transfusions, examinations, and hospitalizations, leading to increased medical costs, decreased quality of life, and even death [7, 8]. Therefore, identifying the risk factors for rebleeding in ALGIB is essential for formulating strategies to reduce the occurrence of rebleeding. Evidence to date indicates that alcohol drinking, smoking, taking low-dose aspirin, and antiplatelet medications are recognized risk factors for colonic diverticular hemorrhage, and hypertension, diabetes, and atrial fibrillation are strongly linked to bleeding from ischemic bowel disease [1, 9]. Research on the risk factors for rebleeding following an initial bleeding event still requires further exploration. Moreover, lower gastrointestinal bleeding disorders frequently coexist in some patients. As studies progress and factors like aging, metabolic disruptions, and comorbid conditions increase, chronic kidney disease and lipid abnormalities are now considered to be linked to lower gastrointestinal bleeding and the risk of rebleeding [10–12]. Consequently, this investigation seeks to elucidate the risk factors associated with rebleeding in individuals suffering from ALGIB by employing a meta-analysis and systematic review, thereby identifying high-risk patients for rebleeding early and intervening promptly to improve the clinical outcomes and prognosis of these patients.

Materials and methods

The protocol for our investigation was formulated in accordance with the guidelines set forth by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Protocols (PRISMA-P). The systematic review was reported in accordance with the PRISMA guidelines [13]. Additionally, the review protocol was prospectively registered on PROSPERO under the registration number CRD42024567552.

Search strategy

A comprehensive search was conducted in PubMed, Embase, Web of Science, and the Cochrane Library for studies on risk factors for rebleeding in ALGIB from the inception of each database up to May 10, 2024. The search utilized a combination of MeSH terms and free-text words, including “gastrointestinal hemorrhage,” “rebleeding,” and “risk factors.” Supplementary Table S1 shows the detailed search strategy.

Inclusion and exclusion criteria

Inclusion criteria are as follows: (1) adhering to the diagnostic criteria for ALGIB [1, 7]; (2) studies report exposure factors of rebleeding following ALGIB; and (3) the primary outcomes included univariable or multivariable analyses of risk factors. For studies containing both types of analyses, the multivariable results were prioritized, and the maximally adjusted effect estimates (adjusted covariates are shown in Supplementary Table S2) were extracted. If only univariable analyses were available, raw data were extracted to calculate the effect estimates.

Exclusion criteria are as follows: exclude conference abstracts, meta-analyses, study protocols, correspondence letters, duplicate publications, systematic reviews, research in which the complete text is inaccessible, studies lacking available data, and experiments conducted on animals.

Literature screening and data extraction

Two independent reviewers were responsible for screening the literature and extracting data independently. They reviewed the titles, abstracts, and full texts. For easily judged articles, direct screening was done; for contentious studies, we consulted experts, downloaded, and fully read the papers. The screening process was strictly adhered to the inclusion and exclusion criteria. Relevant indicators were extracted from the studies, and cross-verification was performed to ensure consistency. The primary data extracted included the first author’s name, year of publication, country, study design, sample size for both non-rebleeding and rebleeding groups, gender, age, specific etiology, definition of rebleeding (including diagnostic method and timing), and risk factors categorized as modifiable and unmodifiable, along with their corresponding effect estimates and 95% confidence intervals (CIs).

Quality assessment

The Newcastle-Ottawa Scale (NOS) [14] was employed to evaluate the quality of both case–control and cohort studies, concentrating on three critical dimensions: selection of the study population (4 points), comparability between groups (2 points), and assessment of exposure or outcomes (3 points). This scale comprises a maximum score of 9 points, where scores of ≤ 4 reflect low quality, scores ranging from 5 to 6 represent moderate quality, and scores of ≥ 7 signify high quality. In instances where disagreements occurred between the two evaluators during the assessment, they engaged in discussions to achieve a consensus or sought the input of a third party to resolve the conflict.

Certainty of evidence assessment

The certainty of evidence pertaining to each risk factor for each outcome was assessed using the GRADE (Grading of Recommendations Assessment, Development, and Evaluation) methodology, which considers the following core domains: risk of bias, indirectness, inconsistency, imprecision, and publication bias. We generated detailed evidence profiles and categorized the overall certainty of evidence into four hierarchical levels (high, moderate, low, and very low), with classifications determined by the ratings assigned to each individual domain [15].

Statistical analysis

Statistical analysis was conducted utilizing the R (version 4.3.2) and Stata 17.0 software. In the univariable analyses, the frequencies of all categorical risk factors were extracted, and the effect size OR and its 95% CI were calculated. For continuous data, the mean and standard deviation of the corresponding variables in the non-rebleeding and rebleeding groups of the original studies were extracted, and the effect size weighted mean difference (WMD) and its 95% CI were calculated. In the multivariable analyses, the maximum adjusted ORs and their 95% CIs from the multivariable regression analysis of the included studies were extracted. Studies reporting ORs and those reporting hazard ratios (HRs) were synthesized separately, and pooled estimates were calculated according to effect measure and study design [16, 17]. Based on the results of the heterogeneity test (Cochran’s Q test) and the I², the appropriate model was chosen to calculate the combined OR value [18]. A random-effects model was adopted in instances where the I² ≥ 50% or P_{heterogeneity} < 0.1; otherwise, a common effect model was applied. Sensitivity analysis was conducted utilizing a stepwise exclusion approach to confirm the robustness of the studies. To evaluate publication bias, funnel plots alongside Egger’s test were implemented for all associated factors. If asymmetry is observed in the funnel plot, the trim-and-fill method will be used to explore its potential causes and assess its impact on the meta-analysis results. Additionally, subgroup analyses were performed to assess whether the effects of potential risk factors varied across different study designs, etiologies, and definitions of rebleeding, provided that sufficient data were available (at least two studies per subgroup). Statistical significance was determined at a threshold of P < 0.05.

Results

Literature retrieval process and outcomes

Searches were conducted in the PubMed, Embase, Cochrane Library, and Web of Science databases. Manual searches were also performed on the included studies, relevant guidelines, and references of review articles, initially yielding 2799 articles. After removing duplicate entries, 1944 articles remained. An initial screening based on titles and abstracts resulted in 138 articles. After full-text reading, 29 articles were ultimately included. The specific search flowchart is shown in Fig. 1.

Fig. 1.

PRISMA flow diagram of the study process. PRISMA, Preferred Reporting Items for Systematic review and Meta-analysis

Basic characteristics of included literature

Of the 29 included studies [19–47], 23 were case–control studies [19, 24, 26, 28, 32, 35, 38, 44, 46, 47] and 6 were cohort studies [25, 27, 33, 34, 37, 45], with most of the 22 studies originating from Japan. The studies included 14,069 cases of ALGIB, with 1930 experiencing rebleeding after ALGIB, and the age range of participants was from 33 to 102 years. Twelve unique factors reported by at least three studies were analyzed in our meta-analysis. The specific characteristics of the literature are shown in Table 1. The 29 included studies were assessed using the NOS quality evaluation, with scores between 7 and 9, reflecting a generally high quality of the included research. Detailed quality assessments are presented in Supplemental Tables S3 and S4.

Table 1.

Basic features of included literature

Study	Year	Study design	Country	Sample size	Gender (male)	Mean age (years)	No. of rebleeding	Cause of hemorrhage	Definition of rebleeding
Aoki et al.	2015	Case–control	Japan	342	197	66.70	84	a, c, d, g, e, f, h, i, j, k	Recurrence of hematochezia
Aoki et al.	2024	Case–control	Japan	2699	1825	NA	510	a, i, h, j, k	Recurrence of hematochezia
Doi et al.	2020	Case–control	Japan	128	79	71.25	29	a	Recurrence of hematochezia or hemoglobin drop
Fujino et al.	2013	Case–control	Japan	90	60	66.50	24	a	Endoscopic confirmation
Gallois et al.	2017	Case–control	French	365	184	83.60	70	a	Endoscopic confirmation
Gilshtein et al.	2016	Case–control	Israel	104	55	75.56	34	a	Recurrence of hematochezia
Gonai et al.	2022	Retrospective cohort	Japan	370	219	76.00	34	a	Recurrence of hematochezia
Hreinsson et al.	2019	Case–control	Iceland	319	166	64.00	53	a, c, e, d, f	Endoscopic confirmation
Ikeya et al.	2015	Retrospective cohort	Japan	101	74	64.70	15	a	Endoscopic confirmation
Joaquim et al.	2017	Case–control	Portugal	74	46	75.70	9	a	Hemoglobin drop
Kawanishi et al.	2018	Case–control	Japan	93	72	66.00	20	a	Endoscopic confirmation
Kitagawa et al.	2019	Case–control	Japan	144	87	73.80	17	a	Endoscopic confirmation
Matsumoto et al.	2013	Case–control	Japan	40	19	76.60	20	b	Recurrence of hematochezia, endoscopic confirmation
Muramatsu et al.	2024	Case–control	Japan	149	82	75.10	35	b	Recurrence of hematochezia, endoscopic confirmation
Nagata et al.	2015	Prospective cohort	Japan	132	87	70.00	39	a	Recurrence of hematochezia
Niikura et al.	2012	Retrospective cohort	Japan	72	50	72.00	27	a	Recurrence of hematochezia
Nishikawa et al.	2013	Case–control	Japan	172	116	70.10	69	a	Recurrence of hematochezia
Nishimura et al.	2018	Case–control	Japan	120	63	74.90	36	b	Recurrence of hematochezia, hemoglobin drop
Radaelli et al.	2021	Retrospective cohort	Italy	1198	626	78.00	105	a, g, i, h	Recurrence of hematochezia
Rodríguez et al.	2021	Case–control	Spain	548	372	67.70	47	h	Recurrence of hematochezia, hemoglobin drop, endoscopic confirmation, angiographic embolization
Sato et al.	2021	Case–control	Japan	519	256	73.10	191	a	Recurrence of hematochezia
Sato et al.	2023	Case–control	Japan	5048	3039	68.90	157	a, c, f, i, j, k, l, g	Recurrence of hematochezia, hemoglobin drop
Sugiyama et al.	2024	Case–control	Japan	282	200	64.12	25	a	Recurrence of hematochezia
Taki et al.	2017	Case–control	Japan	71	42	71.09	15	a	Endoscopic confirmation
Tanaka et al.	2022	Case–control	Japan	159	89	69.50	46	e	Recurrence of hematochezia
Watanabe et al.	2014	Case–control	Japan	151	94	70.50	52	a	Recurrence of hematochezia, hemoglobin drop
Watanabe et al.	2023	Retrospective cohort	Japan	218	139	71.80	71	a	Recurrence of hematochezia, hemoglobin drop
Yamauchi et al.	2021	Case–control	Japan	142	118	75.00	34	a	Recurrence of hematochezia, hemoglobin drop
You et al.	2023	Case–control	Korea	219	164	68.00	62	a	Recurrence of hematochezia, hemoglobin drop, endoscopic confirmation

IBD inflammatory bowel disease, a colonic diverticular, b acute hemorrhagic rectal ulcer, c ischemic colitis, d colorectal cancer, e hematochezia, f IBD, g infectious colitis, h post-polypectomy, i rectal ulcer, j angioectasia, k radiation colitis, l rectal varices

Multivariable meta-analysis results

Age ≥ 70 years

Three studies [33, 35, 40] mentioned age ≥ 70 years. Given that two cohort studies reported HRs and one case–control study reported an OR. The pooled estimate from the two cohort studies, using a common-effect model based on the results of the Cochran’s Q test (I² = 4.0%, P = 0.31), indicated that age ≥ 70 years was significantly associated with a higher risk of rebleeding (HR = 1.56, 95% CI 1.01–2.43). In contrast, the single case–control study showed a similar direction of association, but the result was not statistically significant (OR = 1.23, 95% CI 0.85–1.86). These results are presented in Table 2 and Fig. 2.

Table 2.

Multivariate analysis

Risk factors	No. of study	heterogeneity		OR/HR (95% CI)	P for pooled OR/HR	P for Egger test
		I2 (%)	P
Age ≥ 70 years	3
Cohort studies	2	4.0	0.307	1.56 (1.01, 2.43)*	0.046	NA
Case–control study	1	NA	NA	1.23 (0.85, 1.86)	NA	NA
Active bleeding on colonoscopy	4	0.0	0.976	3.23 (1.68, 6.21)	< 0.001	0.741
Colonic diverticular bleeding	3	0.0	0.485	2.12 (1.73, 2.59)	< 0.001	0.650
CKD	5	0.0	0.463	2.10 (1.53, 2.88)	< 0.001	0.946
Endoscopic treatment	5
Cohort studies	3	76.6	0.014	0.71 (0.27, 1.84)*	0.479	0.676
Case–control study	2	37.8	0.205	0.81 (0.50, 1.33)	0.403	NA
Gender(male)	5	30.8	0.216	1.36 (0.99, 1.85)	0.055	0.077
Hypertension	5
Cohort studies	3	21.1	0.282	1.91 (1.18, 3.07)*	0.008	0.471
Case–control study	2	48.4	0.121	3.72 (1.54, 8.96)	0.004	NA
Hyperlipidemia	3	0.0	0.964	3.29 (1.79, 6.04)	< 0.001	0.133
NSAIDs	6
Cohort studies	4	38.8	0.179	2.38 (1.67, 3.39)*	< 0.001	0.899
Case–control study	2	0.0	0.697	2.21 (1.36, 3.59)	0.002	NA
Non-aspirin antiplatelet drugs	3	0.0	0.584	1.98 (1.42, 2.75)*	< 0.001	0.266
Requiring blood transfusion	4	0.0	0.773	2.94 (1.54, 5.61)	0.001	0.963
Shock at presentation	6	17.5	0.300	2.01 (1.48, 2.75)	< 0.001	0.061

OR odds ratio, CKD chronic kidney disease, NSAIDs nonsteroidal anti-inflammatory drugs, NA not applicable

^*Pooled HR

Fig. 2.

Forest plot representing the risk of age ≥ 70 years in rebleeding of ALGIB in multivariable analysis. ALGIB, acute lower gastrointestinal bleeding

Blood transfusion requirements

Four case–control studies [22, 30–32] investigated blood transfusion requirements as a potential risk factor. Cochran’s Q test showed no significant variation among these studies and subgroups (I² = 0.0%, all P > 0.05), allowing for the use of a common effect model. The results indicate that blood transfusion requirements are a significant risk factor for rebleeding in ALGIB (OR = 2.93, 95% CI (1.54, 5.59), P = 0.001). Subgroup analysis based on etiology revealed a nominally stronger association between blood transfusion requirements and rebleeding risk in patients with anorectal conditions (OR = 3.48, 95% CI 1.55–7.79) compared to those with diverticular disease (OR = 2.16, 95% CI 0.74–6.35). However, the difference between subgroups did not reach statistical significance (P = 0.490), indicating insufficient evidence to conclude that the effect of transfusion on rebleeding differs substantially by etiology, as presented in Table 2 and Fig. 3.

Fig. 3.

Forest plot representing the risk of blood transfusion requirements in rebleeding of ALGIB in multivariable analysis. ALGIB, acute lower gastrointestinal bleeding

Shock at presentation

Six case–control studies [22, 30, 32, 36, 37, 39] investigated shock at presentation (systolic blood pressure ≤ 90 mmHg, heart rate > 120 beats per minute, or shock index (SI) ≥ 1) as a potential risk factor. A common effect model was employed for the analysis due to significant heterogeneity among these studies (I² = 18.0%, P = 0.30). The results indicate that shock at presentation significantly increases the risk of rebleeding in ALGIB [OR = 2.01, 95% CI (1.48, 2.75), P < 0.001], as illustrated in Table 2 and Fig. 4.

Fig. 4.

Forest plot representing the risk of shock at presentation in rebleeding of ALGIB in multivariable analysis. ALGIB, acute lower gastrointestinal bleeding

Hypertension

Five studies [22, 33–35, 44], including three cohort studies and two case–control studies, investigated hypertension as a potential risk factor. The pooled analysis of the cohort studies, performed using a common-effect model in view of the low heterogeneity identified by Cochran’s Q test (I² = 21.1%, P = 0.282), showed that hypertension was significantly associated with an increased risk of rebleeding (HR = 1.91, 95% CI 1.18–3.07). Similarly, the pooled analysis of the case–control studies, also conducted using a common-effect model, demonstrated a significant positive association between hypertension and rebleeding risk (OR = 3.72, 95% CI 1.54–8.96). Collectively, the findings from both study designs consistently support hypertension as a significant risk factor for rebleeding in ALGIB, as shown in Table 2 and Fig. 5.

Fig. 5.

Forest plot representing the risk of hypertension in rebleeding of ALGIB in multivariable analysis. ALGIB, acute lower gastrointestinal bleeding

Chronic kidney disease

Five case–control studies [29, 32, 33, 40, 43] identified chronic kidney disease (CKD) as a risk factor for rebleeding in ALGIB. Cochran’s Q test (I² = 0.0%, P = 0.46) supported the use of a common effect model for the analysis. The analysis demonstrated that CKD significantly increases the risk of rebleeding in ALGIB (OR = 2.10, 95% CI (1.53, 2.88), P < 0.001). Subgroup analysis by etiology indicated a stronger association in patients with diverticular disease (OR = 2.39, 95% CI 1.69–3.40), whereas no significant association was observed in those with anorectal conditions (OR = 1.19, 95% CI 0.58–2.47) in statistics. The subgroup comparison did not show a statistically significant difference across different etiologies (P = 0.092), and the effect direction was consistent across studies, as illustrated in Table 2 and Fig. 6.

Fig. 6.

Forest plot representing the risk of CKD in rebleeding of ALGIB in multivariable analysis. CKD, chronic kidney disease; ALGIB, acute lower gastrointestinal bleeding

Hyperlipidemia

Three case–control studies [27, 30, 44] mentioned hyperlipidemia. A common effect model was employed for analysis due to Cochran’s Q test results (I² = 0.0%, P = 0.96). The analysis results suggest that hyperlipidemia is a risk factor for rebleeding in ALGIB, with statistically significant results (OR = 3.29, 95% CI (1.79, 6.04), P < 0.001), as illustrated in Table 2 and Fig. 7.

Fig. 7.

Forest plot representing the risk of hyperlipidemia in rebleeding of ALGIB in multivariable analysis. ALGIB, acute lower gastrointestinal bleeding

Nonsteroidal anti-inflammatory drugs use

Six studies [19, 22, 33–35, 40] including four cohort studies and two case–control studies evaluated NSAID use as a potential risk factor for rebleeding in ALGIB. The pooled analysis of the cohort studies, conducted using a common effect model in view of the low heterogeneity based on Cochran’s Q test results (I² = 38.8%, P = 0.179), showed that NSAID use was significantly associated with an increased risk of rebleeding (HR = 2.38, 95% CI 1.67–3.39). Similarly, the pooled analysis of the case–control studies, also based on a common effect model, demonstrated a significant positive association between NSAID use and rebleeding risk (OR = 2.21, 95% CI 1.36–3.59). Overall, the findings from both cohort and case–control studies consistently support NSAID use as a significant risk factor for rebleeding in ALGIB, as shown in Table 2 and Fig. 8.

Fig. 8.

Forest plot representing the risk of NSAIDs in rebleeding of ALGIB in multivariable analysis. NSAIDs, nonsteroidal anti-inflammatory drugs; ALGIB, acute lower gastrointestinal bleeding

Active bleeding on colonoscopy

Four case–control studies [22, 27, 29, 30] mentioned active bleeding on colonoscopy. A common effect model was employed for analysis due to Cochran’s Q test results (I² = 0.0%, P = 0.98). The analysis results suggest that active bleeding on colonoscopy is a risk factor for rebleeding in ALGIB, with statistically significant differences (OR = 3.23, 95% CI (1.68, 6.21), P < 0.001). In the subgroup analysis based on the definition of rebleeding, the association between this risk factor and an increased risk of rebleeding was not influenced by the diagnostic method used, as shown in Table 2 and Fig. 9.

Fig. 9.

Forest plot representing the risk of active bleeding on colonoscopy in rebleeding of ALGIB in multivariable analysis. ALGIB, acute lower gastrointestinal bleeding

Colonic diverticular bleeding

Three case–control studies [20, 39, 43] investigated colonic diverticular bleeding as an underlying cause and an unmodifiable risk factor associated with a higher risk of rebleeding. Cochran’s Q test showed moderate variation among these studies (I² = 0.0%, P = 0.49), but a common effect model was still applied for the analysis. The findings revealed that diverticular bleeding significantly increases the risk of rebleeding in ALGIB (OR = 2.12, 95% CI (1.73, 2.59), P < 0.001), as presented in Table 2 and Fig. 10.

Fig. 10.

Forest plot representing the risk of colonic diverticular bleeding in rebleeding of ALGIB in multivariable analysis. ALGIB, acute lower gastrointestinal bleeding

Non-aspirin antiplatelet drugs

Three cohort studies [19, 21, 39] mentioned non-aspirin antiplatelet drugs. A common effect model was employed for analysis due to Cochran’s Q test results (I² = 0.0%, P = 0.58). The analysis results suggest that non-aspirin antiplatelet drugs are a risk factor for rebleeding in ALGIB, with statistically significant differences (HR = 1.98, 95% CI (1.42, 2.75), P < 0.001), as presented in Table 2 and Fig. 11.

Fig. 11.

Forest plot representing the risk of non-aspirin antiplatelet drugs in rebleeding of ALGIB in multivariable analysis. ALGIB, acute lower gastrointestinal bleeding

Other multivariable meta-analysis results

Gender and endoscopic treatment did not show statistically significant differences in multivariable analysis regarding their correlation with rebleeding in ALGIB. A subgroup analysis based on study design was also performed for endoscopic therapy, a related factor associated with substantial heterogeneity (I² = 60.7%, P = 0.038) in the pooled results, but no source of heterogeneity was identified. Nevertheless, consistent findings were observed, and subsequent sensitivity analyses supported the robustness of the results, as presented in Table 2 and Supplemental Figures S1-2 and S29-30.

Univariable meta-analysis

The pooled results of the univariable meta-analyses showed that gender, age, BMI, Charlson comorbidity index (CCI), diabetes, heart failure, ischemic heart disease, cerebrovascular disease, chronic respiratory disease, chronic liver disease, platelet count at presentation, antiplatelet agent use, visible vessel, adherent clot, endoscopic treatment, and endoscopic clipping treatment were not significantly associated with rebleeding after ALGIB. Steroid use (OR = 2.07, 95% CI (1.25, 3.40), P = 0.004), hemoglobin levels at presentation (WMD =  − 0.48, 95% CI (− 0.71, − 0.25), P < 0.001), anticoagulation drug use (OR = 1.31, 95% CI (1.07, 1.59), P = 0.008), and aspirin use (OR = 1.31, 95% CI (1.05, 1.63), P = 0.017) were identified as associated factors for rebleeding, given that their significance was confined to the univariable analysis. Hyperlipidemia (OR = 1.68, 95% CI (1.28, 2.20), P < 0.001), blood transfusion requirements (OR = 1.51, 95% CI (1.22, 1.88), P < 0.001), non-aspirin antiplatelet treatment (OR = 1.69, 95% CI (1.23, 2.33), P = 0.001), colonic diverticular bleeding (OR = 2.22, 95% CI (1.75, 2.80), P < 0.001), and active bleeding on colonoscopy (OR = 1.78, 95% CI (1.35, 2.34), P < 0.001) retained their statistical significance in both univariable and multivariable analyses. Results are shown in Supplementary Table S5 and Supplementary Figures S3-14.

Publication bias

The publication bias of each risk factor was evaluated using funnel plots and Egger’s test. In the univariable analyses results, the P-value of Egger’s test for active bleeding on colonoscopy was 0.009. The funnel plot asymmetry suggested potential publication bias, with an apparent absence of small null-effect studies. The trim-and-fill method was employed to further assess the sources of funnel plot asymmetry and its implications for the meta-analysis outcomes. The analysis revealed that the estimates derived from observed (OR = 1.78, 95% CI 1.35–2.34) and observed plus imputed (OR = 1.52, 95% CI 1.15–2.01) analyses showed minimal discrepancies and a consistent direction of the pooled effect, suggesting that funnel plot asymmetry had only a negligible impact on the validity of the meta-analysis results. The corresponding results are provided in the supplementary materials (Supplementary Figure S18). Other P-values of Egger’s test for both univariable and multivariable outcome measures exceeded 0.05, indicating the absence of publication bias. Detailed statistical findings are comprehensively presented in Tables 2 and 3.

Table 3.

Results of sub-analyses (specific etiology, rebleeding definition)

Risk factors	Subgroups		No. of study	Pooled Estimate	P value	I-squared	P value
Blood transfusion requirements	Overall		4	2.93 [1.54; 5.59]	< 0.001	0.00%	0.773
	Causes of ALGIB	Diverticular bleeding	2	2.16 [0.74; 6.35]	0.160	0.00%	0.438
		Anorectal conditions	2	3.48 [1.55; 7.79]	0.003	0.00%	0.842
Combining CKD	Overall		5	2.10 [1.53; 2.88]	< 0.001	0.00%	0.463
	Causes of ALGIB	Diverticular bleeding	3	2.39 [1.69; 3.40]	< 0.001	0.00%	0.686
		Anorectal conditions	2	1.19 [0.58; 2.47]	0.634	0.00%	0.920
Active bleeding on colonoscopy	Overall		4	3.22 [1.67; 6.21]	< 0.001	0.00%	0.976
	Rebleeding definition	Endoscopic diagnosis	2	3.38 [1.29; 8.85]	0.013	0.00%	0.664
		Clinical diagnosis	2	3.09 [1.26; 7.57]	0.014	0.00%	0.971
Endoscopic treatment	Overall		4	0.79 [0.48; 1.28]	0.339	60.70%	0.038
	Study design	Cohort study	3	0.71 [0.27; 1.84]	0.480	76.60%	0.014
		Case–control study	2	0.81 [0.50; 1.33]	0.403	37.80%	0.205

CKD chronic kidney diseases, ALGIB acute lower gastrointestinal bleeding

Sensitivity analysis

As shown in Figures S19–S30, the leave-one-out meta-analyses indicated that the direction of the combined effect sizes of all relevant risk factors did not change significantly in the multivariable analysis, suggesting that the results are generally robust and reliable.

Certainty of evidence

We identified low-certainty evidence suggestive of an association between the risk of rebleeding of ALGIB and hyperlipidemia, colonoscopically confirmed active bleeding, colonic diverticular bleeding, exposure to non-aspirin antiplatelet drugs, and gender. Very low-certainty evidence indicated a potential association between rebleeding risk and endoscopic treatment as well as age ≥ 70 years. Additionally, moderate-certainty evidence supported an association between the risk of rebleeding and blood transfusion requirement, shock at presentation, hypertension, CKD, and NSAIDs use.

Discussion

This research represents the first meta-analysis to investigate the risk factors for rebleeding following ALGIB, a common digestive system emergency with a high and increasing incidence in recent years [1]. Rebleeding is the most frequent complication after initial control of ALGIB. It significantly increases healthcare costs, heightens the psychological burden on patients, reduces their quality of life, and raises mortality rates. Several studies have identified risk factors for rebleeding; however, they are mostly small-sample and single-center studies with considerable variation in results, which limits clinical evidence and treatment decisions.

Multivariable analysis in this study identified age ≥ 70 as a potential risk factor for rebleeding after ALGIB, indicating a significantly higher rebleeding rate among elderly patients. This correlation may be attributed to age-related declines in physiological function, including increased vascular sclerosis and fragility, as well as reduced intestinal mucosal barrier integrity. Furthermore, many elderly individuals have comorbid chronic conditions, such as cardiovascular and cerebrovascular diseases, and require anticoagulant or antiplatelet medications, which further increase the risk of gastrointestinal mucosal damage and rebleeding [48]. Poor nutritional status and deficiencies in trace elements among the elderly can also delay wound healing following a bleeding episode [49].

This study also identified comorbid hypertension, hyperlipidemia, and CKD as significant risk factors for rebleeding after ALGIB. Hypertension can induce changes in intestinal microvessels, such as stenosis, occlusion, and altered hemodynamics within the mucosal vasculature. These alterations may cause localized ischemia and bleeding and increase the fragility of the vascular walls, particularly in the elderly [50]. Hyperlipidemia may also contribute to rebleeding by impairing coagulation. It can cause abnormal platelet function and increased blood viscosity, potentially disrupting normal hemostasis [51]. Research indicates that hyperlipidemia can also lead to increased fragility of newly formed blood vessels by promoting abnormal angiogenesis (e.g., increased levels of vascular endothelial growth factor, VEGF) [52], making these fragile new vessels more susceptible to rupture and bleeding. CKD patients often exhibit abnormal lipid metabolism, and uremic toxins can directly impair platelet function, leading to reduced platelet adhesion and aggregation. Renal insufficiency can decrease the production or functionality of many coagulation factors, such as factors VIII and IX, thereby increasing the risk of bleeding [53]. Bleeding due to vascular underdevelopment has also been reported as a cause of recurrent gastrointestinal bleeding in CKD, but the precise mechanism remains unclear [10]. In addition, subgroup analysis showed that patients with colonic diverticula comorbid CKD had a higher risk of rebleeding compared to those with rectal lesions, although the difference was not statistically significant. This may be attributed to CKD-associated dysbiosis of the intestinal microbiota and chronic low-grade inflammation [54].

Shock at presentation or blood transfusion requirements can also elevate the risk of rebleeding after ALGIB. Hemorrhagic shock, typically resulting from severe bleeding, indicates substantial blood loss prior to admission. Severe blood loss may indicate that the bleeding source is difficult to control or more complex, and hemorrhagic shock reflects a sharp decrease in blood volume and inadequate tissue perfusion. This condition can lead to hypoxia and malnutrition of the gastrointestinal mucosa, weakening its self-repair ability and increasing the risk of rebleeding [55]. Hemorrhagic shock can provoke systemic stress and inflammatory responses, potentially causing increased vascular permeability and microcirculatory disturbances, further aggravating damage to the gastrointestinal mucosa. Moreover, in the process of correcting shock, patients typically require significant fluid resuscitation and blood transfusion. These interventions can result in blood dilution and reduced concentrations of coagulation factors, impairing normal coagulation function and increasing the risk of rebleeding [56, 57].

This study also identified that the use of NSAIDs and non-aspirin antiplatelet drugs increased the risk of rebleeding following ALGIB. NSAIDs inhibit the activity of cyclooxygenase (COX) enzyme, reducing or suppressing the production of thromboxane A2 (TXA2) and prostaglandins. These effects impair the microcirculation of the gastrointestinal mucosa and platelet aggregation [58]. Non-aspirin antiplatelet drugs, like clopidogrel, can inhibit the ADP receptors (P2Y12 receptors) on platelets, preventing platelet aggregation and clotting. Although this mechanism aids in preventing thrombosis, it notably includes adverse events like gastrointestinal bleeding [59]. Additionally, patients on antiplatelet therapy frequently have underlying diseases like cardiovascular disease; combining these drugs with other anticoagulants can have a compounded damaging effect on the gastrointestinal mucosa, associated with a higher risk of bleeding recurrence [60]. Currently, the optimal timing and strategy for reinstating antithrombotic therapy after an episode of ALGIB remain uncertain. Premature or inappropriate resumption of these medications may increase the risk of rebleeding.

Colonic diverticular bleeding is an underlying cause and unmodifiable factor associated with an increased likelihood of rebleeding after ALGIB. Colonic diverticula, which are sac-like protrusions of the colonic wall, represent the most common cause of ALGIB and have been noted in the 2023 updated ACG guideline as a risk factor for recurrent bleeding. The pathogenesis of rebleeding remains incompletely understood. Firstly, the vascular walls within the diverticula are relatively thin and susceptible to mechanical stress and fecal friction, leading to rupture and bleeding. After the bleeding stops, the vessels in these weakened areas are prone to rupture and bleed again [61]. Furthermore, chronic inflammation and fibrosis surrounding the diverticula can also contribute to increased vascular fragility. Although endoscopic therapies can achieve initial hemostasis, they do not eliminate the underlying risk of rebleeding. Some patients may require repeated endoscopic treatments or surgical interventions, and these treatments themselves may raise the risk of rebleeding [62]. When active bleeding is observed during endoscopy, it is often seen as indicative of the bleeding’s severity, such as the rupture of larger blood vessels, and controlling the bleeding is more challenging, particularly when the lesion is in a location that is difficult to access endoscopically. Even if hemostasis is successfully achieved through endoscopy, residual vascular lesions or new vascular lesions may exist, leading to rebleeding [19, 63]. In patients with bilateral colonic lesions accompanied by bleeding, the complexity of surgical or endoscopic treatment increases due to the need to manage multiple lesion sites. This not only extends the duration of the procedure but may also result in an increased incidence of postoperative complications, including inadequate bleeding control [61]. Additionally, although the use of modifiable risk factors such as steroids, aspirin, or anticoagulants showed a potential association with rebleeding risk in univariable analysis, this finding warrants further validation in larger, high-quality studies.

Those findings provide critical and instructive empirical evidence for refining and innovating upon existing prediction models and clinical guidelines. Specifically, several risk factors identified in our meta-analysis, such as advanced age, NSAIDs use, and antiplatelet therapy, align with components of the Oakland or NOBLADS scores, which are currently recommended (though supported by low-quality evidence) for risk stratification in ALGIB [64, 65]. In contrast, shock at presentation and comorbidities such as CKD or hypertension emerged as novel predictors in our pooled analysis yet remain underrepresented in existing scoring systems. This discrepancy highlights an opportunity to enhance future versions of these risk tools by incorporating additional variables to strengthen their predictive accuracy, particularly for re-bleeding. Thus, our study provides a foundational evidence base for the evolution of clinical prediction rules. Furthermore, our findings validate current guideline recommendations that identify older age, hemodynamic instability at presentation, and diverticular bleeding as risk factors for rebleeding.

This investigation presents several notable limitations. Firstly, the quantity of articles incorporated into the analysis is relatively limited, with a predominant number originating from Japan, potentially introducing significant regional variations. Secondly, owing to the constraints imposed by the limited number of studies and the range of risk factors evaluated, it was not possible to separately aggregate the effect sizes of hazard ratios (HRs) and odds ratios (ORs), which may result in certain biases within the findings. Thirdly, while specific risk factors, including glucocorticoids, anticoagulant therapy, and aspirin, demonstrate a correlation with an elevated risk of rebleeding in univariable analyses, comprehensive multivariable adjustment results are insufficient to substantiate these associations. Finally, the pooled adjusted odds ratios (aORs) were derived from studies with differing covariate adjustments, which may introduce residual confounding. These estimates should be interpreted cautiously, and prospective studies with standardized adjustment are needed to confirm the associations.

Conclusion

Based on available evidence, several potential risk factors may be associated with an increased risk of rebleeding after ALGIB. These factors comprise modifiable elements, including blood transfusion requirements, shock at presentation, hypertension, CKD, hyperlipidemia, NSAIDs use, active bleeding on colonoscopy, and non-aspirin antiplatelet treatment, along with unmodifiable risk factors such as age ≥ 70 years and colonic diverticular bleeding. Given the heterogeneity in covariate adjustment across included studies, these associations should be interpreted with caution. Clinicians may consider these indicators in early risk stratification and individualized intervention to potentially improve patient outcomes.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contribution

All authors contributed to the study conception and design. **Zhongqiao Lu** : Conceptualization, Methodology, Software, Writing- Original draft, Data curation, Visualization were performed; **Deshan Zong and Zhongde Yang** : Investigation, Writing - Original Draft, Writing - Reviewing and Editing were performed; **Bin Yue** : Conceptualization, Supervision, Project administration were performed. All authors read and approved the final manuscript.

Funding

This work was partially supported by the Yunnan Province “14th Five-Year Plan” Provincial Clinical Key Specialty Construction Project (2022–2025).

Data Availability

The data used to support the findings of this study are included within the article.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.