Risk of Post-polypectomy Bleeding After Colorectal Endoscopic Mucosal Resection in Patients with Chronic Kidney Disease: A Propensity-Matched Analysis of the US Collaborative Network

Abstract

Background

Studies evaluating the risk of post-polypectomy bleeding (PPB) after colorectal endoscopic mucosal resection (EMR) in patients with chronic kidney disease (CKD) and end-stage renal disease (ESRD) are limited.

Methods

This retrospective cohort study utilized the U.S. Collaborative Network to assess the risk of PPB after colorectal EMR in patients with CKD compared to controls. Using one-to-one propensity score matching (PSM), the primary outcome measured was PPB within 30 days after colorectal EMR. The PPB risk was further stratified by CKD severity: non-advanced CKD and advanced CKD.

Results

After PSM, each cohort included 9,196 patients. Overall, CKD was associated with increased risk of PPB following colorectal EMR (5.4% vs. 3.8%, odds ratio [OR] 1.44, 95% confidence interval [CI] 1.25–1.66, p < 0.001). The PPB risk was significantly higher in patients with advanced CKD (8.1% vs. 4%, OR 2.09, 95% CI 1.65–2.65, p < 0.001), while those with non-advanced CKD showed modest increase in risk of PPB (4.7% vs. 4%, OR 1.20, 95% CI 1.01–1.41, p = 0.03).

Conclusion

Patients with CKD had higher risk of PPB than patients without CKD. The PPB risk was notably increased in patients with advanced CKD. Optimizing patients with CKD, especially advanced CKD, before colorectal EMR and monitoring for post-procedure bleeding remains important.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10620-025-09122-8.

Introduction

Colorectal cancer is the third most commonly diagnosed cancer and the third most common cause of cancer-related death in the United States [1]. Colonoscopy is the gold standard method for screening for colorectal cancer and allows for resection of colon polyps, which reduces the incidence and mortality of colorectal cancer [2, 3]. The lifetime risk of developing chronic kidney disease (CKD) stage 3 or higher in the United States is 7–10% [4]. Patients with advanced CKD and end-stage renal disease (ESRD) are more likely to have colorectal adenomas and advanced adenomas than control patients [5, 6]. Dialysis and kidney transplantation prolong life. For patients on dialysis, the American Society of Nephrology recommends colon cancer screening only for individuals who are transplant candidates. Colonoscopy is a routine part of evaluation prior to kidney transplant [7].

Endoscopic Mucosal Resection (EMR) is the preferred technique for resecting benign large (≥ 20 mm [mm] in size) colorectal polyps. Colorectal polyps 10–19 mm in size are also frequently resected using EMR per endoscopist discretion [8, 9]. The most frequent major adverse event after colorectal EMR is post-polypectomy bleeding (PPB) [10–12]. The risk of PPB following endoscopic resection of gastrointestinal lesions has been evaluated in several cohort studies of patients with varying stages of renal disease and found that patients with advanced CKD stage ≥ 4 are more likely to suffer from gastrointestinal (GI) bleeding after endoscopy [13–15]. The safety of colorectal EMR in patients with CKD remains understudied. We have previously evaluated the safety of colorectal EMR in patients with cirrhosis, another group who typically have increased risk of bleeding with procedures [16]. We conducted a retrospective United States-based, propensity-matched cohort study to assess the risk of PPB after colorectal EMR in patients with CKD.

Methods

Study Design and Data Source

This large, population-based, retrospective cohort study was conducted using the U.S. Collaborative Network in the TriNetX platform (Cambridge, MA, USA). TriNetX is a federated multicenter research network that provides real-time access to a de-identified dataset from participating healthcare organizations'electronic health records (EHR). Clinical data is obtained directly from EHRs and supplemented by a built-in natural language processing system, which extracts relevant variables from clinical documents. Rigorous quality control is applied at the point of data extraction to ensure accuracy before inclusion in the database. The platform displays only aggregate counts and statistical summaries to protect patient privacy, keeping data de-identified at all stages, with additional anonymity safeguards, such as masking patient counts below 11. Details of the TriNetX network are described in previous studies [17, 18]. We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline [19].

Study Participants and Cohorts

A real-time search and analysis of the U.S. Collaborative Network on the TriNetX platform was conducted from its inception through April 30, 2025. The CKD cohort included all adults (≥ 18 years) with a diagnosis of CKD (stage 1–5, ESRD, and dependence on renal dialysis) who underwent colorectal EMR, identified by International Classification of Diseases-10 (ICD-10) codes for CKD and the Current Procedural Terminology (CPT) codes for renal dialysis and colorectal endoscopic mucosal resection performed via colonoscopy or flexible sigmoidoscopy. Patients who had undergone renal transplantation were excluded from the study. To ensure colorectal EMR occurred after the CKD diagnosis, TriNetX's functionality for defining index events and excluding prior outcomes was applied, including only patients who received colorectal EMR following a CKD ICD-10 code. The control cohort consisted of all adults (≥ 18 years) who underwent colorectal EMR in the same period and did not have an ICD-10 code for CKD. Details of Further details on data sources and diagnosis codes used for patient selection and outcomes (according to predefined ICD-10, CPT, and RxNorm codes) are described in Supplementary Table 1. Figure 1 demonstrates the flowchart for cohort identification in this study.

Fig. 1.

Flowchart for cohort identification in the study

Study Outcomes and Definitions

The primary outcome was the risk of PPB within 30 days of colorectal EMR between CKD and control groups. PPB was identified using ICD-10 codes for “Gastrointestinal hemorrhage, unspecified” (K92.2) or “Hemorrhage of anus and rectum” (K62.5). Secondary outcomes included endoscopic reintervention for hemostasis, blood transfusion, and intensive care unit (ICU) admissions within 30 days of the procedure.

We conducted secondary analyses to further stratify the risk of clinical outcomes after colorectal EMR by the severity of CKD (non-advanced CKD and advanced CKD) compared to the matched control group. The non-advanced CKD cohort was defined as patients with CKD stages 1, 2, and 3. The advanced CKD cohort was defined as patients with CKD stages 4 and 5, ESRD, or dependence on dialysis. Advanced CKD on dialysis specifically referred to patients with CKD who were dependent on dialysis. Of note, the secondary analyses were conducted independently of the primary analysis and did not directly stem from the same matched cohort.

Statistical Analysis

All statistical analyses were conducted using the real-time analytics capabilities of the TriNetX Live platform (TriNetX LLC, Cambridge, MA). Baseline characteristics of the study cohorts were summarized using means, standard deviations, and proportions. To ensure comparability between cohorts, one-to-one propensity score matching (PSM) was employed to balance demographic variables (age, race, ethnicity, gender), comorbidities (such as hypertension, diabetes, cirrhosis, and obesity), thrombocytopenia, and the use of anticoagulants and antiplatelets. Further details of the PSM process are provided in Supplementary Table 2.

Propensity scores for individual subjects were generated using logistic regression analysis based on the specified covariates. Matching was performed with a greedy nearest-neighbor algorithm, applying a caliper width of 0.1 pooled standard deviations, which was determined to be optimal. To minimize potential bias, TriNetX randomized the row order before matching. Following PSM, outcome risks were calculated and expressed as odds ratios (OR) with 95% confidence intervals (CIs). Statistical significance was defined as a two-sided p-value of less than 0.05.

RESULTS

Patient Baseline Characteristics

Initially, we identified 9,456 patients in the CKD cohort and 60,870 patients in the control cohort who underwent colorectal EMR during the study period. The CKD cohort was older (69.2 ± 9.7 vs. 62.1 ± 11.1, p < 0.001) and had a higher proportion of male patients (53.5% vs. 46.4%, p < 0.001), Black or African American race (15.9% vs. 8.4%, p < 0.001), type 2 diabetes mellitus (20.8% vs. 4.7%, p < 0.001), hypertension (25.7% vs. 4.7%, p < 0.001), cirrhosis (2% vs. 0.6%, p < 0.001), obesity (4% vs. 1.6%, p < 0.001), and thrombocytopenia (1.7% vs. 0.3%, p < 0.001) compared to the control cohort (Supplementary Table 2). Additionally, a greater proportion of patients in the CKD cohort were on antiplatelets: aspirin (5.7% vs. 2.1%, p < 0.001) and clopidogrel (2.1% vs. 0.5%, p < 0.001), and on anticoagulants: warfarin (1.5% vs. 0.2%, p < 0.001), apixaban (2.6% vs. 0.6%, p < 0.001), and rivaroxaban (1% vs. 0.3%, p < 0.001) (Supplementary Table 2).

After PSM for age, sex, race, ethnicity, comorbidities (hypertension, diabetes, obesity, and cirrhosis), thrombocytopenia, and medications (anticoagulants, antiplatelets, non-steroidal anti-inflammatory drugs [NSAIDs]), each cohort included 9,196 patients. All variables were balanced between the two groups after PSM. The details of PSM are shown in Supplementary Table 2.

Post-polypectomy Outcomes After Colorectal EMR in CKD Overall

After PSM, a total of 494 (5.4%) patients in the CKD cohort had PPB after colorectal EMR compared to 349 (3.8%) patients in the control cohort. The risk of PPB after colorectal EMR was higher in the CKD cohort compared to the control cohort (OR 1.44, 95% CI 1.25–1.66, p < 0.001) (Table 1). Furthermore, the endoscopic reinterventions for hemostasis (1.6% vs. 1.2%, OR 1.34, 95% CI 1.04–1.72, p = 0.02), blood transfusion (1.4% vs. 0.6%, OR 2.45, 95% CI 1.78–3.36, p < 0.001), and ICU admission (2% vs. 0.9%, OR 2.32, 95% CI 1.78–3.02, p < 0.001) were higher in the CKD cohort compared to the control cohort (Table 1).

Table 1.

Clinical outcomes within 30 days of colorectal EMR in CKD compared to matched control group

Outcomes	Cohort	Events/Total (%)	OR	95% CI	p-value
Post-polypectomy bleeding after colorectal EMR	CKD	494/9,196 (5.4)	1.44	1.25–1.66	< 0.001
	Control	349/9,196 (3.8)
Endoscopic reintervention for hemostasis	CKD	144/9,196 (1.6)	1.34	1.04–1.72	0.02
	Control	108/9,196 (1.2)
Blood transfusion	CKD	131/9,196 (1.4)	2.45	1.78–3.36	< 0.001
	Control	54/9,196 (0.6)
ICU admission	CKD	181/9,196 (2)	2.32	1.78–3.02	< 0.001
	Control	79/9,196 (0.9)

CKD chronic kidney disease, CI confidence interval, EMR endoscopic mucosal resection, ICU intensive care unit, OR odds ratio

Post-polypectomy Outcomes After Colorectal EMR in Non-advanced CKD

The non-advanced CKD cohort had higher rates of PPB (4.7% vs. 4%, OR 1.20, 95% CI 1.01–1.41, p = 0.03), blood transfusion (1% vs. 0.6%, OR 1.79, 95% CI 1.20–2.68, p = 0.004), and ICU admission (1.7% vs. 1%, OR 1.78, 95% CI 1.31–2.42, p < 0.001) following colorectal EMR compared to the matched control cohort (Table 2). However, endoscopic reintervention for hemostasis (1.3% vs. 1.2%, OR 1.05, 95% CI 0.78–1.42, p = 0.76) was similar between the two groups.

Table 2.

Clinical outcomes within 30 days of colorectal EMR in advanced and non-advanced CKD compared to matched control group

Outcomes	Cohort	Events/Total (%)	OR	95% CI	p-value
Advanced CKD (overall)
PPB after colorectal EMR	Advanced CKD	218/2,702 (8.1)	2.09	1.65–2.65	< 0.001
	Control	109/2,702 (4)
Endoscopic reintervention for hemostasis	Advanced CKD	62/2,702 (2.3)	1.79	1.18–2.72	0.01
	Control	35/2,702 (1.3)
Blood transfusion	Advanced CKD	72/2,702 (2.7)	3.34	2.06–5.39	< 0.001
	Control	22/2,702 (0.8)
ICU admission	Advanced CKD	76/2,702 (2.8)	2.58	1.68–3.95	< 0.001
	Control	30/2,702 (1.1)
Advanced CKD on dialysis only
PPB after colorectal EMR	Advanced CKD on dialysis	68/708 (9.6)	2.90	1.81–4.65	< 0.001
	Control	25/708 (3.5)
Endoscopic reintervention for hemostasis	Advanced CKD on dialysis	18/708 (4.2)	1.82	0.84–2.97	0.13
	Control	10/708 (1.4)
Blood transfusion	Advanced CKD on dialysis	30/708 (4.2)	3.09	1.50–6.37	0.001
	Control	10/708 (1.4)
ICU admission	Advanced CKD on dialysis	26/708 (3.7)	2.04	1.04–4.00	0.04
	Control	13/708 (01.8)
Non-advanced CKD
PPB after colorectal EMR	Non-advanced CKD	317/6,688 (4.7)	1.20	1.01–1.41	0.03
	Control	267/6,688 (4)
Endoscopic reintervention for hemostasis	Non-advanced CKD	87/6,688 (1.3)	1.05	0.78–1.42	0.76
	Control	83/6,688 (1.2)
Blood transfusion	Non-advanced CKD	66/6,688 (1)	1.79	1.20–2.68	0.004
	Control	37/6,688 (0.6)
ICU admission	Non-advanced CKD	113/6,688 (1.7)	1.78	1.31–2.42	< 0.001
	Control	64/6,688 (1)

CKD chronic kidney disease, CI confidence interval, EMR endoscopic mucosal resection, ICU intensive care unit, OR odds ratio, PPB post-polypectomy bleeding

Post-polypectomy Outcomes After Colorectal EMR in Advanced CKD

The advanced CKD cohort had significantly higher rates of PPB (8.1% vs. 4%, OR 2.09, 95% CI 1.65–2.65, p < 0.001), endoscopic reintervention for hemostasis (2.3% vs. 1.3%, OR 1.79, 95% CI 1.18–2.72, p = 0.01), blood transfusion (2.7% vs. 0.8%, OR 3.34, 95% CI 2.065.39, p < 0.001), and ICU admission (2.8% vs. 1.1%, OR 2.58, 95% CI 1.68–3.95, p < 0.001) following colorectal EMR compared to the matched control cohort (Table 2).

Post-polypectomy Outcomes After Colorectal EMR in Advanced CKD on Dialysis

The CKD on dialysis cohort had significantly higher rates of PPB (9.6% vs. 3.5%, OR 2.90, 95% CI 1.81–4.65, p < 0.001), blood transfusion (4.2% vs. 1.4%, OR 3.09, 95% CI 1.50–6.37, p = 0.001), and ICU admission (3.7% vs. 1.8%, OR 2.04, 95% CI 1.04–4.00, p = 0.04) following colorectal EMR compared to the matched control cohort (Table 2). However, endoscopic reintervention for hemostasis was similar between the two groups (2.5% vs. 1.4%, OR 1.82, 95% CI 0.84–3.97, p = 0.13) (Table 2).

Discussion

Our findings suggest that patients with CKD are at an increased risk of PPB following colorectal EMR compared to those without CKD. This increased risk is associated with a greater need for endoscopic interventions to control bleeding, along with increased rates of blood transfusions and ICU admissions. The risk of PPB was particularly pronounced in patients with advanced CKD, whereas those with non-advanced CKD experienced a more modest increase in risk compared to the matched control group.

CKD, especially advanced CKD and ESRD, has been identified as an independent risk factor for bleeding in several population studies [20–22]. The increased risk of PPB associated with CKD is likely multifactorial including uremia-induced impaired platelet function with subsequent endothelial dysfunction [23], presence of albuminuria [20, 21], increased risk of angiodysplasia [24], decreased clearance of von Willebrand factor [25], and use of heparin during hemodialysis [26]. Additionally, the uremic environment resulting from declining kidney function, along with changes in the gut microbiome observed in CKD, may contribute to this risk [27].

A recent retrospective study noted immediate PPB and delayed bleeding in 26.1% and 2.2% of patients with end-stage renal disease, respectively [28]. Data from a large retrospective cohort study examining colonoscopy with any polypectomy demonstrated that the risk of significant immediate PPB in patients with ESRD is higher than non-ESRD patients (4.94% vs. 1.36%, p < 0.001); however, the rate of delayed PPB was similar (0.64% vs. 0.40%, p = 0.08) [29]. Similarly, patients with advanced CKD and ESRD undergoing endoscopic submucosal resection for gastric neoplasms had a higher risk of bleeding and longer hospital stays than non-CKD patients. Despite this higher risk of bleeding (OR 6.1, 95% CI 2.7–13.6, p < 0.001), patients with ESRD were observed to respond to endoscopic hemostatic interventions similarly to a control group [13]. A study of colorectal endoscopic submucosal dissection included 118 patients with CKD stage ≥ 3 and demonstrated lower en bloc resection rates with advancing kidney disease; however, complications, including bleeding and perforation, were not different between groups. Due to a small number of patients with CKD stages 4 and 5, predictive modeling was performed, which did not show changes in complications based on decreasing glomerular filtration rates [15]. Our study findings indicate that patients with CKD, particularly those with advanced stages, have an increased risk of PPB, aligning with previous studies [14, 15].

Given the elevated bleeding risk in patients with CKD, it is critical to optimize their condition before undergoing EMR and to ensure vigilant post-procedural monitoring. Optimization should include a comprehensive assessment of renal function and, when appropriate, multidisciplinary input from nephrology or hematology. Strategies to proactively reduce bleeding risk are essential and may involve identifying gastrointestinal bleeding risk factors and adjusting the management plan for high-risk individuals—particularly by reevaluating the need for and intensity of antithrombotic therapy. Since antiplatelet and anticoagulant medications heighten bleeding risk, a careful benefit-risk analysis is warranted, especially in those on dual or triple therapy. In select patients, a less intensive antithrombotic regimen—such as reducing the number of agents or substituting less potent options like aspirin for agents such as ticagrelor—may be more appropriate. The use of nonsteroidal anti-inflammatory drugs should also be minimized due to their bleeding risk. For patients at particularly high risk of PPB, pharmacologic interventions like desmopressin may be considered. Additionally, scheduling dialysis the day before EMR can help correct uremic platelet dysfunction and further lower bleeding risk. Finally, in select cases, employing more conservative endoscopic techniques—such as cold snare EMR instead of hot EMR and the use of prophylactic clipping—may be beneficial.

This study has some limitations. By utilizing a large, de-identified database, we were unable to identify the size of lesion for which patient underwent EMR, the use of electrocautery, and if prophylactic closure was performed. Clip closure of large polyp EMR defects removed with electrocautery proximal to the splenic flexure has been shown to reduce the risk of delayed bleeding [30]. There may also be variability amongst endoscopists in clip closure [31]. The technique used for EMR also influences post-polypectomy bleeding risk. Two randomized controlled trials have found that cold EMR significantly lowered post-procedure bleeding compared to hot EMR, though it increased the risk of polyp recurrence [32, 33]. In select patients with advanced CKD, cold EMR may be preferred to minimize bleeding risk, accepting a higher risk of residual or recurrent polyp [34]. Additionally, as a retrospective cohort design, despite one-to-one PSM, this study is subject to selection and confounding biases, including factors such as polyp size and location, EMR technique (hot vs. cold), or use of prophylactic clips, all of which might influence the risk of PPB. A dedicated ICD-10 code for PPB after EMR is not available. We defined PPB following EMR as lower gastrointestinal bleeding recorded with ICD-10 codes occurring within 30 days of colorectal EMR, but this bleeding may not always be attributable to EMR. We also could not capture the timing of anticoagulant and antiplatelet discontinuation before the procedure or their resumption afterward, which could influence PPB risk following EMR. However, we assume that most patients adhered to standard guidelines for discontinuing these medications, given that EMR is an elective procedure [35]. As with any database study, there are inherent concerns regarding potential misdiagnosis, underreporting of certain variables, and inaccuracies in diagnosis entry. There may be under-reporting of complications if patients received care outside any of the health networks participating in the TriNetX, particularly for post-procedure complications. Finally, we were unable to distinguish between hemodialysis and peritoneal dialysis, as the ICD-10 codes used encompass both modalities.

Despite its limitations, our study has several notable strengths. It is a large-scale, multicenter, U.S.-based population study involving a diverse cohort, which improves the generalizability of our findings. To our knowledge, this is the first and largest study to provide a comparative analysis of PPB risk in patients with CKD undergoing colorectal EMR in comparison to a matched control group. Prior studies in this area have been limited by smaller sample sizes and lack of propensity score adjustment [36]. Furthermore, we employed rigorous one-to-one PSM, carefully balancing relevant comorbidities, demographic factors, and the use of anticoagulants and antiplatelet agents between the cohorts. These efforts enhance the robustness of our findings, which provide valuable insights to inform clinical decision-making and serve as a foundation for future research on the safety of EMR in patients with CKD.

In conclusion, patients with CKD exhibited an increased risk of post-polypectomy bleeding following colorectal EMR compared to controls. The risk was notably increased in patients with advanced CKD. Optimizing patients with CKD, especially advanced CKD, prior to colorectal EMR and monitoring for post-procedure bleeding remains important. Further prospective studies are needed to validate these findings, specifically assessing the impact of polyp size and location, exploring the role of prophylactic lesion closure, and investigating the potential benefits of cold resection in this select population.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

Conceptualization and design: AB, JJG. Manuscript drafting: AB, KE, FP, AA, FK, NS, DR, KS, and JG. Data collection: AB, KE, and IB. Statistical analysis: AB and KE. Tables: FK, AC, AA, and AB. Critical revision of manuscript: ST, KS, DS, DR, IB, and JG. Guarantor of article: JG. Final review and approval of the manuscript: All authors.

Funding

None.

Data Availability

Data was obtained from TriNetX tool at trinetx.com.

Declarations

Conflict of interest

FNP: Travel Support—Boston Scientific, DKR: Consultant- Olympus Corporation, Boston Scientific, Braintree Laboratories, Norgine, Medtronic, Acacia Pharmaceuticals; Research Support—Olympus Corporation, Medivators, Erbe USA Inc, Braintree Laboratories; Shareholder—Satisfai Health, JJG: Travel Support—Boston Scientific Corporation, Olympus Corporation, Ovesco Endoscopy AG. The authors declare no competing interests.