Endoscopy is the primary modality for managing gastrointestinal (GI) bleeding. It can be useful for various causes of upper and lower GI bleeding. Postoperative GI bleeding (PGIB) is a rare complication that is often self-limiting and is thus managed conservatively. Nevertheless, it should still be monitored because of the possibility and potential consequences of uncontrolled postoperative hemorrhage. PGIB is associated with prolonged hospital stay and increased morbidity and mortality, particularly in critically ill patients.1-3 Severe postoperative hemorrhage can also lead to a poor overall five-year survival rate in patients with gastric cancer.4

Conservative management with blood transfusion, angiographic embolization, endoscopic intervention, and surgery are the therapeutic options for PGIB. Although reoperation is considered the mainstay of treatment for postoperative bleeding, it has been reported that patients who have undergone reoperation have more postoperative complications and a higher mortality rate.5-7 In addition to massive bleeding, injuries from prolonged hemostatic procedures, accompanied by hypothermia, coagulopathy, and acidosis, can result in death.4 Given the high morbidity and mortality associated with reoperation, minimally invasive procedures, such as angiographic interventions and endoscopic treatment, may be preferred for the diagnosis and treatment of PGIB, as an alternative to reoperation.4,8 Endoscopic management can be used as a first-line treatment for postoperative intraluminal bleeding. While there is limited concrete evidence regarding complications from endoscopic management, accumulating evidence supports the efficacy and safety of this procedure, even in the early postoperative period.

The clinical presentation of postoperative bleeding ranges from an asymptomatic decrease in hemoglobin level to overt signs of hemorrhage and hemodynamic instability. Hematemesis is a common symptom after GI surgery, which may be related to a small gastric pouch with limited gastric reservoir after surgery.9 Hematochezia is the main manifestation of GI bleeding after colorectal surgery.10 The source of bleeding can be intraluminal or extraluminal. A significant proportion of hemorrhage is extraluminal. Additionally, combined intraluminal and extraluminal bleeding is also common.11 Intraluminal GI bleeding can be caused by a ruptured pseudoaneurysm secondary to an anastomotic leak after pancreaticoduodenectomy.12

The suture line is a major cause of intraluminal GI bleeding. In addition, stress-related mucosal damage can cause upper GI bleeding (UGIB), and conditions such as gastritis, duodenitis, gastric and duodenal ulcers, can be caused by colorectal or non-GI surgeries, such as cardiothoracic or vascular surgery. A Mallory-Weiss tear may occur as a result of postoperative nausea and vomiting.13 Marginal ulcers related to delayed postoperative bleeding may develop from mucosal ischemia caused by perfusion defects, anastomotic tension, or the suture material.14 The bleeding focus can be a pouch, a contiguous small intestine, an excluded limb including the stomach, duodenum, and bypassed small intestine, or damaged vascular structure resulting in intra and extraluminal bleeding.15 Common hepatic artery and its branches, splenic artery, peripancreatic artery, and other abdominal arteries are reported as a focus of postoperative bleeding.4,16

The incidence of PGIB varied according to the type of surgery performed (Table 1).8-11,16-26 Major GI surgery is associated with a relatively high incidence of PGIB. In a population-based study conducted in Korea, the incidence rate of PGIB was 1.92% after conducting a major GI surgery.2 In particular, pancreatic surgery has a relatively high incidence of PGIB and mortality rates compared to other types of GI surgery.17 In addition, PGIB can result from non-GI surgery. Patients undergoing major cardiovascular (1.89%), head and neck (0.71%), genitourinary (0.54%), or orthopedic (0.45%) surgeries also develop PGIB.2

After gastrectomy, PGIB can occur in less than 3% of patients who have undergo surgery.1 The reported incidence of PGIB after bariatric surgery is between 0.5% and 5.8%,18,27 and the overall mortality rate is <1%.28 In a retrospective study on laparoscopic Roux-en-Y gastric bypass (RYGB), postoperative bleeding occurred in 1.5% (652/43,280) of patients. PGIB was related to a longer hospital stay (four vs. two days) and a higher mortality rate (1.38% vs. 0.15%). In addition, 25.3% of the patients underwent reoperation, and 14.9% underwent endoscopy for GI bleeding.1

Postpancreatectomy hemorrhage, which occurs after pancreatic surgery, has an incidence that range from 3% to 16% (weighted mean: 5%). Moreover, it is responsible for 21% of mortality.29 In cases of colorectal surgery, 0.3% to 6.5% of patients experienced PGIB.19,30 Notably, UGIB can occur in colorectal surgery. In a retrospective study including 2,514,228 patients who underwent colorectal resection, 0.5% developed postoperative UGIB, which was related to a higher mortality rate compared to non-bleeding patients (odds ratio [OR], 3.57; 95% confidence interval [CI], 3.40–3.75; p<0.01).3

Several factors affect the risk of PGIB development. Increased age and comorbidities have been associated with an increased risk of postoperative bleeding.2,13 In a single center study involving 5,739 patients who underwent gastrectomy for gastric cancer, male sex (hazard ratio [HR], 2.25; 95% CI, 1.08–4.70), comorbidity (HR, 2.71; 95% CI, 1.44–5.10), and previous abdominal surgery (HR, 2.79; 95% CI, 1.32–5.87) were associated with an increased risk of PGIB.20 Wang et al.31 reported that the incidence of postoperative complications increased as the number of complications increased following gastrectomy (r=0.090, p=0.014). Diabetes mellitus (adjusted OR [aOR], 1.34; 95% CI, 1.24–1.45), hypertension (aOR, 1.31; 95% CI 1.20–1.42), chronic liver disease (aOR, 1.54; 95% CI, 1.38–1.73), congestive heart failure (aOR, 1.78; 95% CI, 1.60–1.98), and peptic ulcer disease (aOR, 1.20; 95% CI, 1.11–1.30) were also associated with an increased risk of PGIB in a population-based study conducted in Korea. However, non-steroidal anti-inflammatory drug (NSAID) use was not associated with an increased risk of PGIB.2 A recent meta-analysis revealed that the use of NSAIDs in the perioperative period did not increase the risk of postoperative bleeding complications.32 In addition, preoperative administration of anticoagulants was reported to be a risk factor for PGIB after RYGB.1,33 Moreover, preoperative antiplatelet/anticoagulation therapy was reported as a risk factor for bleeding after esophagectomy.34

Surgery type was also associated with the risk of developing PGIB. Palliative setting operations have been associated with a higher risk of PGIB in patients undergoing gastric cancer surgery (HR, 4.14; 95% CI, 1.44–11.89).20 Particularly in colorectal surgery, cancer surgery (OR, 1.59; 95% CI, 1.47–1.71; p<0.01), total colectomy (OR, 4.19; CI 3.07–5.72), and transverse colectomy (OR, 3.30; 95% CI, 2.49–4.37) have been identified as factors that result in higher rates of postoperative UGIB.3 In addition, a lower tumor location in the rectum is an independent risk factor for postoperative anastomotic bleeding (risk ratio, 4.78; p=0.041) in patients undergoing lower anterior resection.30 End-to-side ileocolic anastomosis is related to anastomotic bleeding compared to other techniques of ileocolic anastomosis.21 However, surgical reconstruction methods do not seem to have a clinical impact on patients undergoing pancreatoduodenectomy.35 Regarding suture technique, there was no statistically significant difference in the rate of bleeding between hand-suture and stapling techniques in a recent meta-analysis of 26 studies.36

Endoscopy for PGIB enables locating the bleeding focus and providing hemostasis simultaneously. Moreover, endoscopy can be done to estimate the risk of rebleeding.37,38 Moreover, it has been associated with less costs and hospital stays caused by reoperation.30 However, there are concerns about the complications induced by endoscopy in the early postoperative period. Indeed, the anastomosis starts to weaken in the first two postoperative days and approaches maximal strength at approximately four weeks after anastomotic creation. During this period, the submucosal layer provides most of the strength of the anastomosis39,40; hence, the anastomosis is fragile during this period. There is concern that air insufflation, local endoscope trauma, and torque during endoscopic procedures can increase mechanical tension on the anastomosis, thereby resulting in anastomotic leakage or perforation.30,41,42 However, it has been suggested that early postoperative endoscopy is generally safe. Park et al.43 reported that intraoperative endoscopy performed immediately after surgery was safe, with no complications related to endoscopy. Another study found that endoscopy, performed within 30 days of the index operation (median [interquartile range], 22 [16–26] days) resulted in no complications related to endoscopy, such as anastomotic perforation or disruptions, after RYGB.44

In many cases, conservative management is sufficient for treating GI bleeding. However, in patients with uncontrolled postoperative bleeding, therapeutic interventions should be considered to reduce morbidity and mortality.27,45 Endoscopic management is safe and is more advantageous over the surgical approach in terms of cost savings and reduced length of hospital stay. Moreover, repeat endoscopy can be done as a secondary intervention because of its minimally invasive nature, efficacy, and non-requirement of general anesthesia.46

The hemostatic devices and techniques used for postoperative bleeding did not differ from those used for nonoperative GI bleeding. The options include hemoclipping, epinephrine injection, hemospray, electrocoagulation, and argon plasma coagulation. In general, epinephrine injection monotherapy is considered less effective than combined therapy with epinephrine injection and mechanical hemostasis.47 However, there are complexities to consider in patients with postoperative bleeding due to anatomical changes after surgery and comorbid patient status. The success of endoscopic management was closely related to the type of surgery performed (Table 2).8,9,12,16-19,21-23,25,26,42 Assessment and identification of the bleeding location can be more difficult in patients who have undergone pancreaticoduodenectomy than in those who have undergone gastrectomy or colectomy.8,11,22 After gastrectomy, the small remaining stomach volume may limit the space for endoscopic manipulation, and endoscopy may fail to reach the anastomotic site because of blood clots obstructing the intestinal lumen or the length of the bypassed limb.20,23,48

Hemoclips are efficacious and durable for anastomotic bleeding, anastomotic leaks, and perforations. It can result in minimal tissue injury.49 However, anatomical changes after surgery can make the application of hemoclips technically more difficult. The exact facing of the lesion and deployment of the hemoclip to the bleeding site can be problematic in some cases. In addition, the technical success of endoscopic management can be affected by the location of the bleeding at the anastomotic ring. In a recent study, hemostasis on the posterior wall side of the anastomotic ring was associated with the lowest hemostasis success rate compared with lesions on the anterior wall side (50% vs. 100%, respectively).23 Moreover, despite the superiority of combined hemostatic modalities for bleeding control,50 it may not be possible to use it for postoperative bleeding. Injection monotherapy has been used to control bleeding with acceptable success rate.8,18,51 In addition, irrigation with an epinephrine solution can be used to control a small amount of bleeding.30 However, epinephrine injection monotherapy can possibly be associated with rebleeding, especially in the presence of arterial spurting. Thermal therapy at the staple line and anastomosis site requires caution because of the risk of tissue injury, which can lead to perforation.52

Although there is insufficient evidence, an over-the-scope clip (OTSC) can be used to control postoperative bleeding. OTSC provide a higher compression pressure and can capture a large volume of tissue. OTSC treatment is superior to standard therapy in recurrent peptic ulcer bleeding.53 In addition to this, in a recent randomized controlled trial, the initial OTSC treatment was efficacious in patients with severe UGIB as primary therapy.54 In a case report, it was reported that OTSC was successfully applied to the gastroduodenal anastomotic bleeding three weeks after the creation of Billroth I anastomosis.55 Moreover, hemospray can be a useful option for hemostasis of postoperative bleeding. Hemospray provides hemostasis by sealing injured blood vessels and by activating platelets and the intrinsic coagulation pathway.56 In a randomized controlled trial, hemospray had a similar efficacy as that of combination therapy of hemoclip and epinephrine.57 In addition, hemospray was effective in achieving immediate hemostasis in GI bleeding in a recent meta-analysis. The overall pooled clinical success rate after the application of hemospray was 92%.58 Moreover, hemospray can be used as a primary or combination therapy. The potential advantage of hemospray over other modalities is that it can be used for obscured bleeding sites59 or in lesions that are difficult to conduct an en face therapeutic positioning due to angulations or narrow lumens, such as gastrojejunostomy or duodenoduodenal anastomosis.60 However, overall pooled early rebleeding rates after application of hemospray is quite high.58 Another novel modality for hemostasis is endoscopic suturing. In one case report, a massive hemorrhage due to a marginal ulcer 10 days after RYGB was successfully managed with endoscopic sutures without any complications.61

The administration of proton pump inhibitor (PPI) is recommended as an adjuvant to endoscopic hemostasis in the management of UGIB.62 When the bleeding site is located in the gastric pouch or gastroenteric anastomosis, endoscopic treatment, combined with intravenous PPIs, should be considered. Although it is not recommended to use pre-endoscopic PPIs, they reduce the risk of stress-induced GI bleeding in critically ill patients.63 Additionally, a recent systematic review and meta-analysis revealed the benefit of prophylactic administration of PPIs in reducing marginal ulceration after gastric bypass surgery (OR, 0.5; 95% CI, 0.28–0.90; p=0.02).64

Angiography can be used to identify the source of bleeding. Additionally, it can be used for the provision of therapeutic intervention. Angiography is the preferred method for both the diagnosis and treatment of patients with suspected pseudoaneurysm bleeding. Angiographic intervention may be considered in cases of endoscopic treatment failure or recurrent bleeding.65 Furthermore, a recent systematic review reported the superiority of interventional radiology over laparotomy in terms of mortality (22% vs. 47%, p=0.02) with no statistically significant difference in achieving hemostasis (76% vs. 80%, p=0.35). However, surgery is a viable treatment option in situations where interventional radiology is unavailable and in patients who cannot be resuscitated.66

PGIB rarely occurred. Nevertheless, if it does occur, it is associated with high morbidity and mortality rates. In patients with postoperative comorbidities, endoscopic intervention should be considered as a hemostatic strategy. Endoscopy enables localization of the bleeding focus and hemostasis. In addition, postoperative endoscopy is safe and efficacious in most cases, except for biliary-pancreatic surgeries. Hemoclipping is an effective and durable method for achieving hemostasis, even in the postoperative setting. However, the application of the endoscopic method and treatment efficacy are affected by the postoperative anatomy and bleeding location. Thus, endoscopists should know the type of surgery and time of onset to estimate the postulated bleeding focus. In cases where hemostasis is not achieved via endoscopic management, other treatment modalities, such as angiographic intervention or surgery, should be considered.