The Value Of Metal Clips In Transcatheter Arterial Embolization For Acute Nonvariceal Upper Gastrointestinal Bleeding

Introduction

Acute nonvariceal upper gastrointestinal bleeding refers to bleeding in the digestive tract above the ligament of Treitz and is caused by nonvariceal diseases. It also includes bleeding in the pancreatic duct or bile duct and diseases near the anastomosis after gastrojejunostomy¹. ANVUGIB is a common clinical emergency; its annual incidence rate is 19.4 ~ 57.0 per 100,000, and the fatality rate can reach 8.6%. Patients with ANVUGIB should undergo endoscopy or endoscopic treatment within 24 hours after bleeding if conditions permit. However, the rebleeding rate after endoscopic treatment is still high, reaching 15–20%². For refractory ANVUGIB with failed endoscopic hemostasis, transcatheter arterial embolization (TAE) is safe and feasible, with technical and clinical success rates of 69%-100% and 63%-67%, respectively. Due to its low invasiveness, short operation time, quick effects, and good patient compliance, TAE has been the common therapy for refractory ANVUGIB in many medical institutions, especially in high-risk patients, and has become the first-line alternative to surgical procedures^2,3.

Before embolization, angiography is required to find the bleeding site where the contrast agent overflows. However, it may not be possible to see the contrast agent spill due to venous hemorrhage or inactive bleeding of patients. Under this circumstance, physicians always take empirical embolization of the left gastric artery or gastroduodenal artery However, the blood supply of the pylorus and duodenum is complicated, and the collateral circulation between the celiac trunk and the superior mesenteric artery branches varies greatly. Empirical embolization was often ineffective. In the study of Kim et al.⁴, the rebleeding rate after TAE was up to 16% in patients with negative angiographic results. Since most ANVUGIB patients underwent endoscopy or endoscopic treatment before TAE, it has been proposed that the metal clip can be placed at the site or edge of bleeding points during endoscopy to guide embolism during TAE based on the radiopacity of the metal clip.(Figure-1)

Currently, there are just several case reports about the role of metal clips in the TAE of ANVUGIB patients marked with metal clips. No controlled studies on the value of metal clips in TAE have been reported. (For details, see Table 1: Research status of metal clips in TAE operations in patients with ANVUGIB.) These studies have preliminarily proven that marking the bleeding site with a metal clip can help locate the bleeding point and guide embolization in patients with ANVUGIB who have failed endoscopic treatment. Metal clip marking could also reduce the occurrence of complications (organ ischemia, necrosis, dislocation of coils, etc.).

Table 1

Research status of metal clips in TAE operations in patients with ANVUGIB
Author	Study design	Average age	Positive angiography rate	Technical success rate	Clinical success rate	Rebleeding rate	Reintervention rate
Eriksson et al. N = 10	prospective	74.5	40%	90%	80%	/	20%
Song et al. N = 16	retrospective	59.4	43.8%	100%	87.5%	12.5%	6.3%
Wang et al. N = 18	retrospective	63.6	55.6%	100%	94.4%	5.6%	11.1%

This retrospective study aimed to explore the value of endoscopic metal clip marking at the site of bleeding in TAE for patients with ANVUGIB. Moreover, univariate analysis and multivariate analysis were conducted to evaluate influencing factors of the clinical success rate of TAE.

Materials And Methods

Study design

We performed a retrospective review of patients with ANVUGIB in whom bleeding sites were identified endoscopically and treated with TAE at West China Hospital of Sichuan University from January 1st, 2005 to July 1st, 2021. This study was approved by the Ethics Committee on Biomedical Research of West China Hospital of Sichuan University.

Patients

The inclusion criteria for the study population were patients who had been diagnosed with ANVUGIB by clinical manifestation and endoscopy with a bleeding site found under endoscopy and treated with TAE. Exclusion criteria included patients who had biliary or pancreatic duct hemorrhage, hemodynamically unstable patients who were highly dependent on intensive care support, and digestive tract bleeding due to malignant tumors. According to the presence or absence of preplaced metal clips, they were divided into two groups.

Variables

The outcome measurements included the clinical success rate, mortality rate, rebleeding rate, need for surgery, and hemoglobin level at discharge (g/L).

Clinical success refers to a follow-up until 30 days after discharge, the bleeding stopped, and there was no need for additional endoscopic treatment, repeated embolization or invasive treatment such as surgery. Mortality was observed within 30 days after TAE.

The baseline information included sex, age, history of surgery or trauma in 30 days, the use of antithrombotics, Rockall score, hemorrhage site, intervals from hemorrhage to TAE, angiography result, blood transfusion before TAE, hemoglobin before TAE, coagulation disorders (PLT ≤ 80×109/L or international normalized ratio, INR ≥ 1.5, or activated partial thromboplastin time, APTT ≥ 45 s)⁵ and embolization materials.

Technical approach

The procedure is performed using a transfemoral cannulation route in which a 5–6 Fr arterial sheath is placed over the common femoral artery, followed by access to the abdominal trunk using various smaller caliber selective catheters to the common hepatic artery and superior mesenteric artery, respectively, as appropriate, for arteriography to delineate the anatomy and to determine contrast extravasation. If no contrast spillage is found, angiography is performed using a microcatheter cannulated into the gastroduodenal artery, left gastric artery, or splenic artery, depending on the information provided by the endoscopy about the possible location of the bleeding source. If contrast spillage is detected, embolization is performed using coils, burcrylate, gelfoam sponges, and embolic granules depending on the active bleeding vessel; if contrast spillage is still not detected and a metal clip is available, the blood supply vessel to the site is embolized according to its location, or if no metal clip is available, embolization is performed according to the endoscopic indication of the bleeding site. Technical success refers to a repeat of the responsible angiogram that did not show the responsible artery, and no collateral blood supply was formed immediately after embolization.⁶ The technical success rate in this study was 100%.

Statistical analysis

Continuous variables subjected to a normal distribution are reported as the mean, standard deviation and range, and continuous variables that were not subjected to a normal distribution are reported as the median, interquartile range and range. Categorical variables were reported as frequencies and composition ratios.

The comparison of two sets of sample means used the t test when continuous and subjected to a normal distribution, while the Mann–Whitney U test was used for nonnormally distributed data. The Pearson chi-square test or Fisher's exact probability was applied when the two sets of composition ratios were compared. For quantitative variables, the strength of association was calculated by simple binary logistic regression for binary outcomes and was reported as odds ratios (ORs) with 95% confidence intervals (CIs). The significance level was set at P ≤ 0.05, and the two-sided P values were reported. Statistical analysis was performed using IBM SPSS Statistics, version 25.0 (IBM Corp, Armonk, New York, USA).

Result

According to the inclusion criteria, a total of 172 patients were screened out, and 70 patients were excluded. Ultimately, 102 patients were included in the study. There were 73 cases in the metal clip group and 29 cases in the nonmetal clip group. For the flow chart, see Fig. 2.

There were no significant differences between the two groups in sex, age, history of surgery/trauma, history of anti-thrombotic medication, Rockall score, bleeding site, time from bleeding to TAE, angiography result, preoperative blood transfusion volume, preoperative hemoglobin, accompanying coagulation dysfunction or use of embolic materials (for details, see Table 2: Comparison of baseline data).

Table 2

Comparison of baseline data
	metal clip group n = 73	Nonmetal clip N = 29	P value
gender(male/female)	63(86.3)/10(13.7)	22(75.9)/7(24.1)	0.255^a
age(years)	61(21)[19–83]	59(46)[33–86]	0.450^d
History of surgery/trauma(yes/no)	19(26.0)/54(74.0)	3(10.3)/26(89.7)	0.141^c
History of anti-thrombotic medication(yes/no)	15(20.5)/58(79.5)	6(20.7)/23(79.3)	1.000 ^c
Rockall score	5(3)[1–9]	5(3)[1–8]	0.337^d
Bleeding site			1.000^b
cardia	3(4.1)	1(3.4)	/
Stomach	29(39.7)	12(41.4)	/
duodenum	32(43.8)	12(41.4)	/
multiple-site bleeding	4(5.5)	2(6.9)	/
no active bleeding	5(6.8)	2(6.9)	/
Intervals between bleeding to TAE(day)	6(10.5)[0–46]	12(21)[0–50]	0.081^d
Angiography result(negative/positive)	55(75.3)/18(24.7)	20(69.0)/9(31.0)	0.510^a
Preoperative blood transfusion(U)	4(8)[0-75.5]	3(6.25)[0-35.5]	0.307 ^d
Preoperative hemoglobin(g/L)	*65.5(16.7)[16–98]	*68.6(22.2)[29–114]	0.473^e
coagulopathy(yes/no)	30(41.1)/43(58.9)	10(34.5)/19(65.5)	0.539 ^a
Embolic agents			0.589^b
coils	38(52.1)	20(69.0)	/
bucrylate	7(9.5)	1(3.4)	/
PVA granules	1(1.4)	0	/
Coil and PVA granules	3(4.1)	2(6.9)	/
Coils and bucrylate	23(31.5)	6(20.7)	/
Coils and gelfoam sponge	1(1.4)	0	/

Comparison of outcome variables

Among all of the patients in this study, the technical success rate was 100%, the clinical success rate was 74.5%, the mortality rate was 14.7%, the rebleeding rate was 13.7%, and the additional surgery rate was 13.7%. No TAE-related complications occurred.

In the group comparison, the clinical success rate of the metal clip group was higher than that of the nonmetal clip group (82.2% vs 55.2%, P < 0.001), and the rebleeding rate (8.2% vs 27.6%, P = 0.039) of the metal clip group was lower than that of the nonmetal clip group. The additional surgery rate of the metal clip group was lower than that of the nonmetal clip group (11.0% vs 20.7%, P < 0.001). There were statistical significance for all of the differences. The mortality rate in the metal clip group was reduced compared with that in the nonmetal clip group (11.0% vs 24.1%), although there was no significant difference, which might be related to the limited sample size. There was no statistically significant difference in hemoglobin levels at discharge. (For details, see Table 3: Comparison of outcome measurements).

Table 3

Comparison of outcome measurements
	Metal clip group N = 73	Nonmetal clip group N = 29	P values
Clinical success(yes/no)	60(82.2)/13(17.8)	16(55.2)/13(45.8)	< 0.001^a
death(yes/no)	8(11.0)/65(89.0)	7(24.1)/22(75.9)	0.146^a
rebleeding(yes/no)	6(8.2)/67(91.8)	8(27.6)/21(72.4)	0.039^a
Additional surgery(yes/no)	8(11.0)/65(89.0)	6(20.7)/23(79.3)	< 0.001^a
Discharged hemoglobin(g/L)	*84.3(16.2)[27–125]	*85.4(27.1)[43–138]	0.360^e

Analysis of the factors influencing clinical success

A univariate analysis of the factors influencing clinical success showed that there were significant differences in Rockall score (5 vs 7, P = 0.016) and metal clip placement rate (78.9% vs 50%, P = 0.005) between the clinical success group and clinical failure group. Multivariate analysis showed that metal clip placement at the bleeding site was a protective factor for clinical success. The probability of clinical success for TAE hemostasis in patients with metal clips placed at the bleeding sites was 3.75 times higher than in those without metal clip placement (OR = 3.750, 95CI = 1.456–9.659, P = 0.004). (For details, see Table 4: Comparison of risk factors).

Table 4

Comparison of risk factors
	Clinical success N = 76	Clinical fail N = 26	P value	OR(95CI)
gender(male/female)	64(84.2)/12(15.8)	21(80.8)/5(19.2)	0.762^a	/
age(years)	60(23.25)[19–83]	58(33.25)[22–80]	0.776^d	/
History of surgery/trauma(yes/no)	15(19.7)/61(80.3)	7(26.9)/19(73.1)	0.442^a	/
History of anti-thrombotic medication(yes/no)	15(19.7)/61(80.3)	6(23.1)/20(76.9)	0.934^c	/
Angiography(negative/positive)	54(71.1)/22(28.9)	21(80.8)/5(19.2)	0.332^a	/
Metal clip (yes/no)	60(78.9)/16(21.1)	13(50)/13(50)	0.005^a	3.750(1.456–9.659)
Rockall score	5(2)[1–8]	7(3.25)[2–9]	0.016^d	/
Intervals between bleeding to TAE(day)	6(13.75)[0–46]	8(10.25)[0–50]	0.416^d	/
Preoperative blood transfusion(U)	4(7.5)[0-75.5]	4(11.5)[0-35.5]	0.601^d	/
coagulopathy(yes/no)	30(39.5)/46(60.5)	10(38.5)/16(61.5)	1.000^a	/
Enumeration data: Frequency (percentage); Measurement data: Median (quartile range) [range], * Mean (standard deviation) [range];
a: Pearson chi-square test; b: Fisher's exact probability; c: Continuous calibration chi-square test; d: Mann–Whitney U test; e: Independent sample T test.

Discussion

At present, the clinical treatment of ANVUGIB is still very challenging. Endoscopic examination is always required by these patients with hemostasis implemented if necessary. However, there is still a high rebleeding rate after endoscopic treatment. Patients who have failed endoscopic hemostasis are likely to be treated with TAE. However, it is often difficult to locate the responsible blood vessel during TAE due to inactive bleeding. Previous case reports have shown that preplaced metal clips under endoscopy can play a role in locating the responsible blood vessels during TAE. However, these studies were all case series. To date, there have been no controlled studies evaluating the effect of endoscopic metal clip marking on rebleeding rates and clinical outcomes after TAE. This study is the first controlled study to endoscopically assess the impact of metal clips placed at the bleeding site of ANVUGIB patients on the clinical effect of TAE. In previous related studies, advanced age⁷, history of trauma or invasive surgery⁸, use of antithrombotic drugs⁹, bleeding site¹⁰, longer time to TAE^5,11, red blood cell suspension ≥ 6 U infusion before TAE^12,13,14, low hemoglobin¹², coagulation dysfunction^13,14,15, different embolic materials¹² and angiography result¹⁶ were reported to affect prognosis of ANVUGIB patients after TAE treatment. In this study, these factors were comparable between the metal clip group and the nonmetal clip group.

Compared with the nonclips group, the group with metal clips had a higher clinical success rate (82.2% vs. 45.0%, P < 0.001). The metal clip group also had a lower rebleeding rate (8.2% vs 27.6%, P = 0.039) and additional surgery rate (11.0% vs 20.7%, P < 0.001). Consistent with these results, metal clip marking was also shown to improve the clinical success (OR = 3.750, 95CI = 1.456–9.659, P = 0.004) of TAE in multivariate analysis. Compared with patients without metal clips, patients marked with metal clips were 3.75 times more likely to achieve successful hemostasis by TAE. These results indicated that a metal clip placed at the bleeding site was a protective factor for the success of hemostasis by TAE. Metal clips probably play a protective role in TAE by locating bleeding vessels and guiding embolization during TAE procedures. Due to the diversity of vascular anatomy, empirical vascular embolization often fails to cover the responsible vessel and results in failure of hemostasis. In angiographically negative patients, metal clip marking was demonstrated to increase the accuracy of angioembolization by indicating the location of the bleeding lesion and suggesting the responsible vessel¹⁷. In addition, the metal clip probably improves the rate of positive angiography by shortening the distance between the location of the contrast release and the bleeding site in selective angiography². Moreover, metal clip marking could shorten the angiographic operation time⁶. Therefore, it is recommended that metal clip labeling should be performed if there is a possible TAE.

The mortality rate of the metal clip group was also lower than that of the no metal clip group, but the difference was not significant (11% vs 24.1%, p = 0.146). A meta-analysis¹⁸ including 819 patients who underwent TAE without metal clip marking reported an overall mortality rate of 28%. This was higher than the mortality rate of both the marking and nonmarking groups in this study. There were many factors that might be related to the mortality rate, such as coagulation dysfunction and interval between bleeding and TAE. In our study, the lower mortality rate might be attributed to all patients included in this study undergoing endoscopy prior to TAE, and the site of bleeding was detected endoscopically, which was helpful to guide angiography and embolization. In addition, patients who did not undergo endoscopy due to unstable vital signs of bleeding and went straight to emergency TAE, as well as patients whose bleeding site was not observed endoscopically due to massive blood retention in the gastrointestinal tract, were excluded, as these patients were often bleeding more heavily and in a more critical condition. The mortality rate in the metal clip group was less than half that in the nonclip group, but this result was not significantly different. The insignificant difference may be related to the limited sample size, which was due to the exclusion of a large number of patients who had undergone angiography only, without arterial embolization.

Many studies have discussed factors related to recurrent bleeding. Lau et al.¹⁹ defined the criteria for high risk of recurrent bleeding as ulcers ≥ 20 mm, spurting bleeding, hypotensive shock or hemoglobin < 90 g/L. Patients who fulfilled one or more of the above criteria had a risk of at least 16.7% to rebleed. Kaminskis et al.²⁰ used Forrest type Ia-IIb and Rockall score ≥ 5 as standards to define whether patients were highly likely to rebleed. These criteria are used to determine the need for TAE and to assess the need for endoscopic hemostasis. Previous studies also showed that the failure of endoscopic hemostasis was related to the exposure of arteries, ulcers and adhesive blood clots, massive bleeding or large ulcers combined with active bleeding.²

Although no consensus has been reached, the overall view of these studies is basically the same. Combined with the results of our study, we recommend that lesions with a Forrest classification of Ia-IIb and a Rockall score ≥ 5 should be labeled with metal clips when conditions permit. The metal clip label could be performed either by direct metal clip hemostasis or by marking at the edge of the lesion without hemostasis.(Fig. 3) However, metal clip marking is only applicable to patients who have undergone endoscopy before TAE and have found bleeding sites under endoscopy. It is not applicable to patients with contraindications to endoscopy, for example, heavy bleeding, hemodynamic instability requiring emergency TAE, and a large amount of blood retention in the digestive tract.

In this study, no TAE-related complications occurred in either the metal clip group or the nonmetal clip group. Arterial embolization in the GI tract above the duodenal suspensory ligament is generally considered to be very safe, as the stomach and duodenum have abundant collateral circulation; thus, serious complications such as gastrointestinal ischemia are rare.¹⁷ Metal clip hemostasis and marking also proved to be safe. In our study, no extra bleeding or other complications occurred due to the placement of the metal clip. The guidelines for the diagnosis and treatment of ANVUGIB²¹ recommend that patients with ANVUGIB should accept endoscopy or treatment within 24 hours after bleeding. The metal clip can act both as a hemostasis tool and embolization label, and the marking can be performed at the same time as the metal clip used for endoscopic hemostasis. For patients who underwent endoscopic hemostasis by electrocoagulation or other methods, a metal clip could also be placed at the edge of the lesion. It does not require additional examination. Therefore, the use of metal clips for marking is safe and convenient.

Limitations of this study are the inherent flaws of retrospective research, the operators and other factors that may affect the results cannot be controlled comprehensively; the data that can be collected are limited, such as the operation time and other important outcome indicators cannot be collected; and the bias of a single-center study is difficult to avoid.

References

1. Bai Y, Li ZS. Guidelines for the diagnosis and treatment of acute non-variceal upper gastrointestinal bleeding (2015, Nanchang, China). J Dig Dis. 2016;17(2):79-87. doi:10.1111/1751-2980.12319

2. Song JS, Kwak HS, Chung GH. Nonvariceal Upper Gastrointestinal Bleeding: the Usefulness of Rotational Angiography after Endoscopic Marking with a Metallic Clip. Korean J Radiol. 2011;12(4):473. doi:10.3348/kjr.2011.12.4.473

3. Loffroy R, Favelier S, Pottecher P, et al. Transcatheter arterial embolization for acute nonvariceal upper gastrointestinal bleeding: Indications, techniques and outcomes. Diagnostic and Interventional Imaging. 2015;96(7-8):731-744. doi:10.1016/j.diii.2015.05.002

4. Kim JH, Shin JH, Yoon HK, et al. Angiographically Negative Acute Arterial Upper and Lower Gastrointestinal Bleeding: Incidence, Predictive Factors, and Clinical Outcomes. Korean J Radiol. 2009;10(4):384. doi:10.3348/kjr.2009.10.4.384

5. Loffroy R, Rao P, Ota S, De Lin M, Kwak BK, Geschwind JF. Embolization of Acute Nonvariceal Upper Gastrointestinal Hemorrhage Resistant to Endoscopic Treatment: Results and Predictors of Recurrent Bleeding. Cardiovasc Intervent Radiol. 2010;33(6):1088-1100. doi:10.1007/s00270-010-9829-7

6. Wang Y, Jia P. The role of metallic clips in transcatheter intravascular embolization for non-variceal upper gastrointestinal bleeding cases receiving unmanageable endoscopic therapy: A retrospective cohort study. International Journal of Surgery. 2018;58:26-30. doi:10.1016/j.ijsu.2018.08.006

7. Sildiroglu O, Muasher J, Arslan B, et al. Outcomes of Patients With Acute Upper Gastrointestinal Nonvariceal Hemorrhage Referred to Interventional Radiology for Potential Embolotherapy. Journal of Clinical Gastroenterology. 2014;48(8):687-692. doi:10.1097/MCG.0000000000000181

8. Schenker MP, Duszak R, Soulen MC, et al. Upper Gastrointestinal Hemorrhage and Transcatheter Embolotherapy: Clinical and Technical Factors Impacting Success and Survival. Journal of Vascular and Interventional Radiology. 2001;12(11):1263-1271. doi:10.1016/S1051-0443(07)61549-8

9. Short- and long-term results of transcatheter embolization for massive arterial hemorrhage from gastroduodenal ulcers not controlled by endoscopic hemostasis. Accessed August 11, 2021. https://schlr.cnki.net/en/Detail/index/SJPD_01/SJPD13031100333595

10. Kim PH, Tsauo J, Shin JH, Yun SC. Transcatheter Arterial Embolization of Gastrointestinal Bleeding with N -Butyl Cyanoacrylate: A Systematic Review and Meta-Analysis of Safety and Efficacy. Journal of Vascular and Interventional Radiology. 2017;28(4):522-531.e5. doi:10.1016/j.jvir.2016.12.1220

11. Walsh R. Role of angiography and embolization for massive gastroduodenal hemorrhage. Journal of Gastrointestinal Surgery. 1999;3(1):61-66. doi:10.1016/S1091-255X(99)80010-9

12. Loffroy R, Guiu B, D’Athis P, et al. Arterial Embolotherapy for Endoscopically Unmanageable Acute Gastroduodenal Hemorrhage: Predictors of Early Rebleeding. Clinical Gastroenterology and Hepatology. 2009;7(5):515-523. doi:10.1016/j.cgh.2009.02.003

13. Aina R, Oliva VL, Therasse É, et al. Arterial Embolotherapy for Upper Gastrointestinal Hemorrhage: Outcome Assessment. Journal of Vascular and Interventional Radiology. 2001;12(2):195-200. doi:10.1016/S1051-0443(07)61825-9

14. Huang YS, Chang CC, Liou JM, Jaw FS, Liu KL. Transcatheter Arterial Embolization with N-Butyl Cyanoacrylate for Nonvariceal Upper Gastrointestinal Bleeding in Hemodynamically Unstable Patients: Results and Predictors of Clinical Outcomes. Journal of Vascular and Interventional Radiology. 2014;25(12):1850-1857. doi:10.1016/j.jvir.2014.08.005

15. Zenilmani ME. Angiographic Embolization for Gastroduodenal Hemorrhage: Safety, Efficacy, and Predictors of Outcome—Invited Critique. Arch Surg. 2008;143(5):461. doi:10.1001/archsurg.143.5.461

16. Busch M, Schneider A, Lankisch T, von Hahn T. [Acute hemorrhage in the upper gastrointestinal tract]. Internist (Berl). 2017;58(3):226-232. doi:10.1007/s00108-016-0184-x

17. Loffroy R, Rao P, Ota S, De Lin M, Kwak BK, Geschwind JF. Embolization of acute nonvariceal upper gastrointestinal hemorrhage resistant to endoscopic treatment: results and predictors of recurrent bleeding. Cardiovasc Intervent Radiol. 2010;33(6):1088-1100. doi:10.1007/s00270-010-9829-7

18. Loffroy R, Rao P, Ota S, De Lin M, Kwak BK, Geschwind JF. Embolization of Acute Nonvariceal Upper Gastrointestinal Hemorrhage Resistant to Endoscopic Treatment: Results and Predictors of Recurrent Bleeding. Cardiovasc Intervent Radiol. 2010;33(6):1088-1100. doi:10.1007/s00270-010-9829-7

19. Lau JYW, Pittayanon R, Wong KT, et al. Prophylactic angiographic embolisation after endoscopic control of bleeding to high-risk peptic ulcers: a randomised controlled trial. Gut. 2019;68(5):796-803. doi:10.1136/gutjnl-2018-316074

20. Kaminskis A, Kratovska A, Ponomarjova S, et al. Preventive transarterial embolization in upper nonvariceal gastrointestinal bleeding. World J Emerg Surg. 2017;12(1):3. doi:10.1186/s13017-016-0114-1

21. Chinese Journal of Internal Medicine, National Medical Journal of China, Chinese Journal of Digestion, Chinese Journal of Digestive Endoscopy, Chinese Digestive Endoscopist Association. [Guidelines for the diagnosis and treatment of acute non-variceal upper gastrointestinal bleeding (2018, Hangzhou)]. Zhonghua Nei Ke Za Zhi. 2019;58(3):173-180. doi:10.3760/cma.j.issn.0578-1426.2019.03.005