Endoscopic transparent cap assisted hemostasis: A multicenter retrospective cohort study

Abstract

BACKGROUND

Non-variceal upper gastrointestinal bleeding (NVUGIB) remains a common and potentially life-threatening emergency despite advances in endoscopic therapy. Endoscopic transparent caps (CAPs) have been proposed to enhance lesion visualization and procedural stability, but their effect on clinical outcomes is uncertain.

AIM

To evaluate whether CAP use during therapeutic endoscopy influences hemostatic efficacy and short-term outcomes in NVUGIB.

METHODS

A total of 206 patients who underwent emergency endoscopic hemostasis between 2021 and 2025 at six tertiary centers in Türkiye were included. Patients were divided into CAP-assisted (n = 67) and non-CAP (n = 139) groups. Baseline characteristics, comorbidities, Glasgow-Blatchford and Rockall scores, hemostatic methods, and clinical outcomes were analyzed. Multivariable linear regression identified independent predictors of post-admission transfusion requirement and hospital length of stay (LOS).

RESULTS

Baseline demographic and clinical variables were comparable between groups. CAP was more often used in hemodynamically unstable patients. The CAP group required a greater number of hemostatic techniques (2.66 ± 1.00 vs 2.14 ± 0.87; P < 0.001), with higher use of Ankaferd Blood Stopper and argon plasma coagulation, but received fewer post-admission transfusions (14.8 % vs 33.8 %; P = 0.006) and a lower transfusion volume (0.44 ± 1.13 U vs 0.91 ± 1.31 U; P = 0.012). Mortality, intensive-care need, and LOS did not differ significantly. CAP use independently predicted reduced transfusion volume (β = -0.47, P = 0.010), whereas higher Charlson Comorbidity index and intensive care unit admission predicted longer LOS.

CONCLUSION

CAP-assisted endoscopy may improve procedural efficiency and hemostatic control in NVUGIB, reducing transfusion requirements without affecting mortality or hospitalization. Prospective randomized trials are warranted to validate these findings and clarify CAP’s role in standardized bleeding management.

Key Words: Non-variceal upper gastrointestinal bleeding; Endoscopic hemostasis; Transparent cap; Hemostatic efficacy; Transfusion requirement; Charlson Comorbidity index; Multicenter retrospective study

Core Tip: This multicenter retrospective study compared transparent cap (CAP)-assisted and standard endoscopic hemostasis in non-variceal upper gastrointestinal bleeding (NVUGIB). CAP was used more often in hemodynamically unstable patients and required a greater number of hemostatic techniques but was associated with lower post-admission transfusion rates and volumes. Mortality and hospital stay were comparable between groups. CAP assistance may enhance procedural precision and efficiency without altering systemic outcomes, representing a simple and inexpensive adjunct for improving endoscopic management of NVUGIB.

INTRODUCTION

Upper gastrointestinal bleeding is a prevalent medical emergency worldwide, characterized by hemorrhage originating from the esophagus, stomach, or duodenum. Clinically, patients most commonly present with hematemesis vomiting of fresh blood or coffee-ground material or melena, whereas hematochezia may occur in cases of severe bleeding and is usually associated with hemodynamic instability[1,2]. Several prognostic scoring systems, such as the Glasgow-Blatchford score (GBS), have been developed to predict key outcomes in upper gastrointestinal bleeding, including mortality, the need for endoscopic or other hemostatic interventions, and the risk of rebleeding[3]. Despite substantial advances in pharmacologic and endoscopic management, the mortality rate of upper gastrointestinal bleeding remains approximately 10%, largely attributable to the increasing age and comorbidity burden of affected patients[3,4]. During therapeutic esophagogastroduodenoscopy, the proximity of the endoscope tip to the target lesion and its instability can impair visualization quality and hinder access to the lesion. When intervention is performed on the target site, a phenomenon known as “red-out” in which the field of view becomes obscured by blood, pus, or tissue debris may adversely affect treatment. To restore a clear view, the lesion is often irrigated endoscopically with large volumes of saline or water; however, this may lead to clinical complications such as aspiration, or technical challenges due to the movement of intragastric particles that can obstruct the target area[5,6]. Transparent caps (CAPs) used during endoscopic procedures add a tactile dimension between the tip of the endoscope and the target lesion, facilitating visualization and precise targeting of the lesion, thereby simplifying therapeutic manipulations[7,8]. A limited number of studies in the literature have investigated the use of CAPs in gastrointestinal bleeding interventions[9-12]. However, in clinical practice, CAPs are frequently utilized, particularly in bleeding sites that are difficult to access or visualize[13]. In the present study, we hypothesized that the use of an endoscopic CAP may positively influence clinical outcomes in hemostatic interventions, and we retrospectively evaluated short-term outcomes of endoscopic procedures performed with and without cap assistance.

MATERIALS AND METHODS

Study design and population

This retrospective multicenter study included emergency endoscopic hemostatic procedures performed between 2021 and 2025 at six hospitals located in four different cities in Türkiye, all of which were capable of performing advanced endoscopic interventions. Endoscopy reports, emergency department records, and inpatient files of the patients were reviewed retrospectively. All emergency bleeding interventions were performed by gastroenterologists with a minimum of three years of experience in therapeutic endoscopy.

Separate institutional permissions were obtained from all participating hospitals, and multicenter ethical approval for the study was granted by the Ondokuz Mayıs University Ethics Committee (approval No. 2024000168-2).

The primary aim of the study was to evaluate the effectiveness of CAP use during hemostatic interventions. Therefore, patients who underwent endoscopic procedures with or without CAP assistance were included. Esophageal, gastric, and duodenal variceal bleedings were excluded because variceal hemorrhage follows distinct diagnostic and therapeutic algorithms and routinely requires the use of a ligation cap as standard care, precluding meaningful comparison with non-variceal CAP-assisted interventions. Additionally, malignancy-related bleeding, which is inherently prone to recurrence; Mallory-Weiss tears, which are mechanically induced and typically self-limited; and Dieulafoy lesions, which may only be intermittently visible during active bleeding, were excluded due to their potential to introduce diagnostic variability and confound outcome interpretation. Only patients presenting with active non-variceal bleeding (Forrest Ia or Ib) or visible-vessel ulcers (Forrest IIa) who underwent endoscopic hemostatic therapy were eligible for inclusion.

A total of 206 patients were ultimately included, comprising 139 without CAP use and 67 with CAP use. Demographic characteristics (age, sex), comorbidities, and medication use were recorded. To evaluate potential differences in outcomes related to comorbidity, Charlson Comorbidity index (CCI) scores at admission were also analyzed.

Clinical and endoscopic data collection

The severity of bleeding at admission was assessed using the GBS, while post-procedural risk was evaluated using the Rockall score. Baseline clinical parameters at presentation were also recorded individually. Endoscopy reports and hospitalization records were reviewed to document the type of endoscopic hemostatic intervention(s) performed during esophagogastroduodenoscopy, as well as the agents and dosages used during the procedure.

Endoscopic hemostatic techniques

In this retrospective study, the choice to use or not use a CAP, as well as the selection of hemostatic techniques, was determined entirely by the performing endoscopist according to real-time clinical judgment during the bleeding episode. Endoscopists based their decisions on factors such as the anatomical location of the lesion, the severity of active bleeding, the presence of blood, debris, or purulent material impairing visualization, and the anticipated technical difficulty of achieving hemostasis. Accordingly, CAP was preferentially used in situations where improved visualization or enhanced scope stability was expected to facilitate precise targeting of the bleeding source.

Endoscopic hemostasis was achieved using a combination of injection, mechanical, and thermal techniques, selected according to bleeding characteristics and procedural needs. The CAP functions as a clear distal attachment that maintains a fixed distance between the endoscope tip and the mucosa, enhances lesion exposure, reduces visual obstruction, and stabilizes the scope during therapeutic maneuvers. These features are particularly beneficial in anatomically challenging regions or in the presence of brisk bleeding.

Ankaferd Blood Stopper (ABS), a topical liquid hemostatic agent widely used in Türkiye, was applied directly to the bleeding surface. It can be applied either as a liquid onto the lesion or sprayed in an atomized form, similar to commercially available hemostatic sprays. ABS acts by forming a protein network that induces erythrocyte aggregation and promotes clot formation, showing particular effectiveness in mucosal and oozing bleedings. Adrenaline sclerotherapy was performed by injecting diluted epinephrine (1:10000 to 1:20000) around the bleeding site to achieve local vasoconstriction and tamponade. Argon plasma coagulation (APC), a non-contact thermal coagulation technique using ionized argon gas, was employed to achieve superficial coagulation in oozing lesions or diffuse mucosal bleeding areas. When mechanical therapy was indicated, hemoclips were deployed to compress and close the bleeding vessel directly, particularly in cases of focal arterial bleeding or visible vessels. Aethoxysklerol 2% (lauromacrogol), a chemical sclerosing agent, was injected into or around the bleeding site to induce localized thrombosis and fibrosis, thereby contributing to vessel obliteration. Coagrasper, a hemostatic forceps capable of delivering monopolar coagulation under direct contact, was used for targeted coagulation of persistent bleeding points or visible vessels, especially when primary hemostasis was suboptimal. Because CAP use and hemostasis modality selection were based on individualized endoscopist judgment rather than predefined criteria, we evaluated baseline bleeding severity and comorbidity burden including CCI to ensure comparability between CAP-assisted and non-CAP groups and to minimize potential bias[2,10,14-16].

Outcome measures

Post-procedural outcomes included mortality, intensive care unit (ICU) requirement, hospital length of stay (LOS), and the need for repeat endoscopy. Additionally, blood transfusion requirements within the first 24 hours and further transfusions and total blood units administered after the initial 24 hours of hospitalization were also recorded. These parameters were accepted as clinical outcomes and compared between the CAP and non-CAP groups.

Statistical analysis

All statistical analyses were performed using IBM SPSS Statistics for Windows, version 26.0 (IBM Corp., Armonk, NY, United States). Continuous variables were expressed as mean ± SD and categorical variables as n (%). The Kolmogorov-Smirnov test was used to assess the normality of data distribution. Comparisons between the CAP and non-CAP groups were performed using the Student’s t-test for normally distributed continuous variables and the Mann-Whitney U test for non-normally distributed data. χ² or Fisher’s exact test was applied for categorical variables, as appropriate. To identify independent predictors of continuous outcomes, multivariable linear regression analyses were conducted for hospital LOS and post-admission transfusion amount. Variables with a P value < 0.10 in univariate analyses were included in the multivariable models, and CAP use was forced into the model regardless of its univariate significance. Regression results were presented as β coefficients with 95% confidence intervals and corresponding P values. All P values were two-tailed, and a P value < 0.05 was considered statistically significant.

RESULTS

A total of 206 patients were included in the final analysis, comprising 67 (32.5%) in the CAP-assisted group and 139 (67.5%) in the non-CAP group.

Demographic and baseline clinical characteristics were comparable between the two groups, with no statistically significant differences in age (62.97 ± 19.72 years vs 59.82 ± 22.44 years, P = 0.348), sex distribution (male 76.7% vs 70.5%, P = 0.356), CCI (4.62 ± 3.35 vs 4.16 ± 2.86, P = 0.334), smoking status (28.6% vs 27.9%, P = 0.920), or use of nonsteroidal anti-inflammatory drugs, aspirin, clopidogrel, or oral anticoagulants (Table 1). These findings indicate a well-balanced baseline comorbidity and medication profile between cohorts.

Table 1 Baseline demographics and medication use, mean ± SD/n (%).

Variable	Non-CAP (n = 139)	CAP-assisted (n = 67)	P value
Age (years)	62.97 ± 19.72	59.82 ± 22.44	0.348
Gender (male)	102 (76.7)	43 (70.5)	0.356
Gender (female)	31 (23.3)	18 (29.5)	0.356
Charlson score	4.62 ± 3.35	4.16 ± 2.86	0.334
Smoking (yes)	38 (28.6)	17 (27.9)	0.920
NSAID (yes)	30 (22.6)	8 (13.1)	0.124
Clopidogrel (yes)	17 (12.8)	9 (14.8)	0.708
ASA (yes)	32 (24.1)	17 (27.9)	0.571
OAC (yes)	18 (13.5)	11 (18.0)	0.439

CAP: Transparent cap; OAC: Oral anticoagulant; NSAID: Nonsteroidal anti-inflammatory drugs; ASA: Aspirin.

When bleeding severity and clinical parameters were assessed (Table 2), both groups exhibited similar mean GBS (11.29 ± 3.92 vs 12.15 ± 3.54, P = 0.134) and Rockall scores (5.44 ± 2.12 vs 5.95 ± 2.05, P = 0.114). However, syncope (50.8% vs 28.6%, P = 0.003), tachycardia (67.2% vs 41.4%, P < 0.001), and hypotension (32.8% vs 66.9%, P < 0.001) were significantly more common in the CAP-assisted group, suggesting that CAP use was more frequently employed in hemodynamically unstable patients. Hemoglobin values tended to be slightly higher in the non-CAP group (P = 0.051). The duodenal bulb was the most common bleeding site in both cohorts (85.2% vs 74.4%, P = 0.196).

Table 2 Bleeding severity and clinical parameters, mean ± SD/n (%).

Variable	Non-CAP (n = 139)	CAP-assisted (n = 67)	P value
Blatchford score	11.29 ± 3.92	12.15 ± 3.54	0.134
Rockall score	5.44 ± 2.12	5.95 ± 2.05	0.114
Melena (yes)	126 (94.7)	58 (95.1)	1.000
Hematemesis (yes)	52 (39.1)	32 (52.5)	0.081
Hematochezia (yes)	6 (4.5)	4 (6.6)	0.509
Syncope (yes)	38 (28.6)	31 (50.8)	0.003
Blood pressure > 90 mmHg	89 (66.9)	20 (32.8)	< 0.001
Pulse > 100 bpm	55 (41.4)	41 (67.2)	< 0.001
HGB > 10 g/dL	34 (25.6)	8 (13.1)	0.051
INR > 1.2	32 (24.1)	12 (19.7)	0.498
Transfusion in first 24 hour (yes)	85 (63.9)	36 (59.0)	0.613
Intervention time	1.81 ± 0.71	1.98 ± 0.65	0.098
Bleeding site (bulbus)	99 (74.4)	52 (85.2)	0.196

CAP: Transparent cap; HGB: Hemoglobin; INR: International normalized ratio.

The distribution of Forrest classifications (Ia, Ib, IIa) is presented in Figure 1. CAP-assisted endoscopy was more frequently used in Forrest Ia and Ib lesions, whereas Forrest IIa ulcers were relatively more frequent in the non-CAP group.

Figure 1 Distribution of Forrest classification by transparent cap use. CAP: Transparent cap.

Endoscopic hemostatic interventions are summarized in Table 3. The use of ABS and APC was significantly higher in the CAP-assisted group (both P < 0.001), whereas hemoclip application was more frequent in the non-CAP group (P = 0.001). Aethoxysklerol was also used more commonly in the CAP group (11.5% vs 3.0%, P = 0.038), while the use and dosage of adrenaline and coagrasper were comparable between groups (P > 0.05). Notably, the mean ABS dose was significantly lower in CAP-assisted procedures (3.16 ± 1.46 mL vs 6.10 ± 2.24 mL, P < 0.001), likely due to more precise topical application under improved visualization.

Table 3 Endoscopic techniques and intervention details, mean ± SD/n (%).

Variable	Non-CAP (n = 139)	CAP-assisted (n = 67)	P value
ABS (used)	87 (65.4)	55 (90.2)	< 0.001
AST (used)	85 (63.9)	43 (70.5)	0.192
APC (used)	33 (24.8)	31 (50.8)	< 0.001
Hemoclips (used)	66 (49.6)	15 (24.6)	0.001
Aethoxysklerol 2% (lauromacrogol) (used)	4 (3.0)	7 (11.5)	0.038
Coagrasper (used)	10 (7.5)	3 (4.9)	0.758
ABS dose (mL)	6.10 ± 2.24	3.16 ± 1.46	< 0.001
AST dose (mL)	5.61 ± 1.54	5.11 ± 1.76	0.141
Aethoxyklerol dose (mL)	1.50 ± 1.00	2.79 ± 1.15	0.211
Number of hemostatic techniques	2.14 ± 0.87	2.66 ± 1.00	< 0.001
Hemostasis failure (yes)	4 (3.0)	3 (4.9)	0.545

CAP: Transparent cap; ABS: Ankaferd Blood Stopper; AST: Adrenaline sclerotherapy; APC: Argon plasma coagulation.

The mean number of hemostatic techniques per patient was higher in the CAP group (2.66 ± 1.00 vs 2.14 ± 0.87, P < 0.001), indicating a multimodal approach often adopted in cases with more severe or technically demanding bleeding. Hemostasis failure, defined as persistent bleeding requiring surgery or radiologic embolization, occurred in 3 (4.9%) CAP-assisted and 4 (3.0%) non-CAP patients (P = 0.545).

Clinical outcomes are detailed in Table 4. Patients in the CAP-assisted group required significantly fewer post-admission transfusions (14.8% vs 33.8%, P = 0.006) and had a lower total transfusion volume (0.44 ± 1.13 units vs 0.91 ± 1.31 units, P = 0.012). No statistically significant differences were observed regarding ICU admission (31.1% vs 27.8%, P = 0.635), mortality (11.5% vs 12.0%, P = 0.912), or hospital LOS (5.00 ± 2.54 days vs 5.30 ± 5.56 days, P = 0.606). Rebleeding rates during second-look endoscopy were comparable between groups (42.9% vs 46.2%, P = 0.842).

Table 4 Clinical outcomes, mean ± SD/n (%).

Variable	Non-CAP (n = 139)	CAP-assisted (n = 67)	P value
ICU admission (yes)	37 (27.8)	19 (31.1)	0.635
Mortality (yes)	16 (12.0)	7 (11.5)	0.912
Bleeding in 2nd scope (yes, if rescoped)	12 (46.2)	6 (42.9)	0.842
Hospital stay (days)	5.30 ± 5.56	5.00 ± 2.54	0.606
Post-admission transfusion (yes)	45 (33.8)	9 (14.8)	0.006
Transfusion amount (units)	0.91 ± 1.31	0.44 ± 1.13	0.012

CAP: Transparent cap; ICU: Intensive care unit.

In multivariable linear regression analyses (Table 5), CAP use was independently associated with a reduction in post-admission transfusion amount (β = -0.473, P = 0.010), but not with hospital LOS (β = -0.215, P = 0.762). Higher CCI (β = 0.338, P = 0.024) and ICU admission (β = 1.673, P = 0.034) independently predicted prolonged hospitalization. For transfusion outcomes, an initial transfusion within the first 24 hours strongly predicted subsequent transfusion requirements (β = 0.826, P < 0.001).

Table 5 Multivariable linear regression analyses.

Variable	Length of stay β (95%CI), P value	Post-admission transfusion β (95%CI), P value
Intercept	1.770 (-0.921 to 4.461), 0.196	-0.135 (-0.882 to 0.612), 0.722
CAP use	-0.215 (-1.617 to 1.186), 0.762	-0.473 (-0.832 to -0.114), 0.010
Age	0.013 (-0.031 to 0.057), 0.564	-0.000 (-0.009 to 0.008), 0.912
Charlson index	0.338 (0.045 to 0.631), 0.024
Blatchford score	0.061 (-0.128 to 0.251), 0.524	0.050 (-0.001 to 0.101), 0.055
ICU admission	1.673 (0.132 to 3.215), 0.034
HGB ≥ 10 g/dL		-0.116 (-0.601 to 0.370), 0.639
First 24 hours transfusion		0.826 (0.437 to 1.215), < 0.001

CI: Confidence interval; CAP: Transparent cap; HGB: Hemoglobin; ICU: Intensive care unit.

DISCUSSION

The success of endoscopic hemostasis largely depends on the clarity of visualization of the bleeding site and the technical precision of the procedure[17]. The use of a CAP, attached to the distal end of the scope, expands the field of view, facilitates direct targeting of the lesion, and enhances the stability of the endoscope during manipulation. CAP is particularly useful in cases where access is difficult or visualization is impaired by blood or debris[12]. However, evidence regarding the impact of CAP use on clinical outcomes including bleeding control, transfusion requirement, rebleeding, and mortality remains limited[10,18]. Most published studies have emphasized the technical advantages of CAP, such as improved visualization, optimal lesion alignment, and enhanced procedural control, while its contribution to clinically meaningful outcomes has not been clearly established[6,18,19]. The present study was designed to address this gap by comparing endoscopic hemostatic efficacy, transfusion requirements, and short-term outcomes between CAP-assisted and standard endoscopic interventions.

In this multicenter study of 206 patients, CAP-assisted endoscopy was performed in 67 patients (32.5%), while 139 patients (67.5%) underwent standard endoscopy without a cap. The two groups were comparable in terms of demographics, comorbidities, and medication use. The CCI, which quantifies overall disease burden based on weighted scores for 17 comorbid conditions, was employed to objectively evaluate comorbidity status[20]. Previous studies have demonstrated that higher CCI scores are associated with adverse outcomes in gastrointestinal bleeding, and a CCI ≥ 3 has been identified as a predictor of poor prognosis and increased mortality in non-variceal upper gastrointestinal bleeding (NVUGIB)[21-24]. The use of CCI in the present analysis ensured homogeneity between groups and provided a standardized measure of systemic disease burden, representing a methodological strength of this retrospective design.

The implementation of validated risk stratification tools, such as the GBS and the Rockall score, facilitates the early identification and safe discharge of patients at low risk for upper gastrointestinal bleeding. This approach reduces the need for endoscopic evaluation and hospitalization, optimizing healthcare resource utilization without compromising patient safety[25,26]. Both scoring systems have been widely endorsed in the literature as integral components of upper gastrointestinal bleeding management. Moreover, a GBS > 1 has demonstrated strong predictive value for the need for hospital-based interventions, endoscopic therapy, blood transfusion, and surgery[27,28]. In the present study, all included patients had a GBS greater than 1 and required urgent endoscopic intervention. Although there were no statistically significant differences in Blatchford and Rockall scores between the CAP and non-CAP groups the slightly higher mean Blatchford score in the CAP-assisted group prompted further evaluation of the individual clinical parameters contributing to bleeding severity. Examination of these components revealed significantly higher rates of syncope, hypotension, and tachycardia in the CAP group. These findings suggest that although aggregate scoring systems were similar, CAP may have been preferentially used in patients with more severe real-time bleeding or impaired visualization, reflecting endoscopist judgment at the time of intervention. In real-world endoscopic practice, endoscopists tend to use CAP preferentially in situations where bleeding appears rapid or brisk, visualization is impaired by blood or debris, or enhanced scope stability is needed to accurately target the lesion[7,8,13]. Thus, the differences in hemodynamic instability markers likely reflect the endoscopist’s clinical judgment at the time of the procedure rather than a baseline imbalance not captured fully by formal scoring systems. Hemoglobin levels were marginally higher in the non-CAP group, which may further support the interpretation that patients in the CAP group had more severe bleeding at presentation.

Distinct differences were observed in the endoscopic hemostatic techniques used between the two groups. The application of ABS and APC was significantly more frequent in the CAP group, whereas hemoclip placement was performed more often in the non-CAP group. ABS is a liquid hemostatic agent that is widely used in Türkiye for the management of upper gastrointestinal bleeding. Owing to its viscous nature, it is particularly effective in duodenal bulb lesions, where it produces visible macroscopic hemostasis within approximately 10 seconds after application[14,15]. However, electron microscopy studies have demonstrated that the hemostatic action of Ankaferd begins within less than one second after contact, inducing aggregation of all blood components particularly erythroid elements[29]. The stabilization and direct visualization provided by CAP likely prolonged the agent’s contact time with the bleeding surface and enhanced its hemostatic efficacy. Moreover, the ability to apply ABS directly onto the lesion under cap guidance, together with the prevention of fluid loss due to the cap’s containment effect, may explain why the total ABS dose was significantly lower in the CAP group. In contrast, no significant differences were found between groups in the doses of adrenaline and aethoxysklerol, which are injected directly into or around the lesion and therefore less affected by visualization or fluid dynamics. The higher rate of APC use in the CAP group may be explained by the fact that APC requires maintaining a stable distance of 3-5 mm between the probe and the tissue for optimal coagulation[30]. Procedures performed at high voltages (> 50 W) have been associated with an increased risk of perforation[31]. CAP reduces the accumulation of blood and debris, thereby maintaining a clear visual field, and allows easier control of the distance to the bleeding site factors that may enhance the overall effectiveness of APC. For these reasons, APC may have been more frequently preferred in procedures performed with CAP assistance. In situations where visualization was limited, operators may have preferred mechanical methods such as hemoclip application. In both groups, the number of hemostatic techniques used for bleeding control was generally consistent with American Gastroenterological Association recommendations, which advocate the use of at least two complementary modalities for optimal hemostasis[32]. However, the mean number of techniques was significantly higher in the CAP group (2.66 ± 1.00 vs 2.14 ± 0.87; P < 0.001). This finding suggests that lesions treated with CAP were likely more complex and technically demanding, prompting endoscopists to employ additional methods to achieve effective hemostasis, especially ABS may have been preferred as a rescue adjunctive therapy in cases of severe bleeding[16]. Differences in the frequency or dosage of particular hemostatic agents between groups may reflect the greater technical complexity of cases in which CAP was applied and endoscopist preference in real-world practice, rather than a direct pharmacologic or CAP-related effect. Conversely, as reported by Kim et al[10], the use of a CAP facilitates endoscopic accessibility in technically difficult cases but does not necessarily translate into improved clinical success. Despite the more frequent use of multiple techniques, the rate of hemostasis failure was comparable between groups (4.9% vs 3.0%; P = 0.545) and consistent with previously reported failure rates ranging between 3% and 6% in the literature[33].

In this study, blood transfusion requirements were evaluated in two distinct temporal phases. Transfusions administered within the first 24 hours after hospital admission were considered resuscitative, intended to compensate for pre-hospital blood loss and to restore hemodynamic stability. In contrast, transfusions given after the initial 24-hour period, following endoscopic intervention, were regarded as a measure of procedural hemostatic success. Previous reports have shown that minor post-procedural oozing can result in hemoglobin declines of approximately 1-2 g/dL after endoscopic therapy[34]. Similarly, a recent study evaluating transfusion patterns in gastrointestinal bleeding emphasized that early transfusions primarily reflect pre-endoscopic stabilization efforts, whereas later transfusions correlate more closely with ongoing or recurrent bleeding episodes[35]. In our cohort, the post-24-hour transfusion requirement was significantly lower in the CAP group, suggesting that CAP use contributed to more effective and durable hemostasis. By providing enhanced visualization and stability, CAP likely allowed more accurate identification of the bleeding source and more precise therapeutic application. These findings indicate that the use of the CAP enhances the technical quality and efficacy of endoscopic hemostasis, thereby significantly reducing the need for post-procedural blood transfusion, while not exerting a measurable effect on broader clinical outcomes such as ICU admission rates, mortality, or hospital LOS. The absence of significant differences in ICU requirement and hospitalization duration suggests that CAP’s principal benefit lies in procedural success rather than overall clinical trajectory. These results may be consistent with existing evidence suggesting that, in the management of upper gastrointestinal bleeding, the precision and effectiveness of the endoscopic intervention rather than solely the adjustment of transfusion thresholds may be key determinants of transfusion utilization and patient outcomes[36,37].

According to multivariable regression analysis, CAP use emerged as an independent predictor of reduced post-procedural transfusion volume, underscoring its contribution to procedural efficacy and improved hemostatic performance. In contrast, CAP use did not independently influence hospital LOS, suggesting that its effect is primarily technical rather than systemic. Both the CCI and ICU admission were identified as independent predictors of prolonged hospitalization. These findings are consistent with prior studies demonstrating that comorbidity burden and disease severity are key determinants of hospital stay duration among patients with gastrointestinal bleeding[36,38,39]. Thus, while CAP improves the technical success of hemostasis and reduces transfusion requirements, it may not significantly modify overall recovery dynamics or hospitalization length.

This study has several limitations. First, its retrospective design inherently carries the risk of selection bias and potential incomplete data capture. The precise anatomical localization of the bleeding site (e.g., posterior vs anterior wall of the duodenal bulb) was not consistently documented across centers. Because emergency endoscopy reports served as the primary data source and lacked standardized subsite classification, subsite-specific analyses and evaluation of CAP’s positional advantages could not be reliably performed. Secondly, although the study involved four centers with potential differences in operator experience, endoscopic equipment, and local practice patterns, all procedures were performed by gastroenterologists with at least three years of therapeutic endoscopy experience, which likely reduced inter-center variability; however, given the retrospective multicenter design, such variability and selection bias cannot be fully eliminated, and standardized inclusion criteria, exclusion of non-CAP centers, and multivariable adjustment were applied to mitigate these limitations. Third, the relatively modest sample size reflects the strict inclusion criteria (active NVUGIB requiring endoscopic therapy), the exclusion of centers without CAP use, and the inherently low incidence of such cases in routine clinical practice. Although the study involved six tertiary centers, two were excluded to avoid selection bias, and participating hospitals were not the sole referral centers in their regions, further limiting case volume. These factors restrict sample size and generalizability and support the need for larger, prospective multicenter studies with standardized long-term follow-up. Fourth, the analysis focused solely on short-term outcomes, including early rebleeding, post-procedural transfusion, and hospital stay. Rebleeding could only be assessed during the index hospitalization based on the need for and findings of a second endoscopy; long-term rebleeding after discharge could not be reliably obtained across centers. Similarly, mortality data reflected only deaths occurring during the initial hospitalization, as post-discharge survival time (including 30-day mortality) could not be uniformly retrieved due to variations in electronic medical records and data-protection policies. These limitations restrict the assessment of long-term clinical outcomes in this study. Finally, although the multivariable regression model included clinically relevant parameters, certain potential confounding factors such as bleeding etiology, ulcer size, or lesion morphology could not be incorporated due to incomplete documentation. Despite these limitations, the use of an objective comorbidity index (CCI), the inclusion of multiple centers, and the balanced baseline characteristics between groups strengthen the internal validity and generalizability of this study.

CONCLUSION

In this multicenter retrospective study, the use of a CAP during therapeutic endoscopy for NVUGIB may enhance procedural efficacy and reduce post-procedural transfusion requirements, without a measurable effect on hospital LOS, ICU admission, or mortality. CAP likely provides better scope stabilization and visualization, enabling more precise hemostatic interventions and facilitating the application of multiple complementary techniques, particularly in hemodynamically unstable or technically demanding cases. Multivariable analysis indicated that CAP use may independently contribute to lower post-admission transfusion volumes, supporting its potential role in improving procedural efficiency. In contrast, CCI and ICU admission remained independent predictors of prolonged hospitalization, consistent with prior evidence suggesting that comorbidity burden and disease severity are major determinants of hospital stay duration. Overall, these findings suggest that the primary advantage of CAP lies in improving the technical precision and completeness of endoscopic hemostasis, rather than altering systemic recovery parameters. CAP may represent a simple, inexpensive, and safe adjunct that could be incorporated into endoscopic practice to optimize hemostatic outcomes in NVUGIB. Future prospective, randomized studies are warranted to confirm these findings and to better define the long-term impact of CAP-assisted endoscopy on rebleeding, transfusion needs, and patient prognosis.