Upper gastrointestinal bleeding: a review Despite recent diagnostic and therapeutic ad- vances, the mortality rate for upper gastrointestinal bleeding has remained at 10% for the past three Thomas A. Broughan, M.D decades.1 Two factors are probably responsible for David P. Vogt, M.D. this persistently high mortality rate. First, this pop- ulation is composed of a larger than average pro- Department of General Surgery portion of elderly patients who have marked asso- ciated medical diseases. Second, advanced tech- niques in intensive care allow patients who are critically ill or have had multiple traumas to be supported for long periods of time; many of these patients subsequently develop upper gastrointes- tinal bleeding that may be fatal. Recent developments in the management of the patient with gastrointestinal bleeding include fiber- optic endoscopy, cimetidine therapy, gastric pH monitoring, and radiographic arterial emboliza- tion. Their clinical application is being investi- gated. Although bleeding episodes are controlled in most cases with medical management, approxi- mately 15% to 20% of patients require surgery.1 Etiology The most frequently encountered causes of upper gastrointestinal bleeding are peptic ulcer disease, acute gastric mucosal lesions, esophageal varices, and the Mallory-Weiss syndrome. Depending on 97 Downloaded from www.ccjm.org on September 24, 2021. For personal use only. All other uses require permission. f>4 Cleveland Clinic Quarterly Vol. 49, No. 1 Table 1. Comparison of causes of sure in the fourth ventriclc stimulates upper gastrointestinal bleeding in two the vagal nuclei with a resultant gastric series hypersecretion.5 The erosions resulting Detroit Toronto from stress, burns (Curling's ulcer), and (Sugawa)2 (Halmagyi)"' ulcerogenic agents are in part caused by Etiology (%) (%) the destruction of the gastric mucosal Duodenal ulcer ii 41 barrier. Loss of this barrier allows back 2 Acute erosive 42 diffusion of hydrogen ions into the mu- gastritis Gastric ulcer 18 11 cosal cells, which in turn releases vaso- Mallorv-Weiss 15 6 active amines. The injuries caused by tear the hydrogen ions and vasoactive Esophageal 5 7 amines include mucosal congestion, varices hemorrhagic infarction, mucosal slough, Other 10 33 and finally bleeding. Although a patient may have docu- the patient population considered, pre- mented esophageal varices, in about cise order of frequency varies (Table 50% of patients, the bleeding episode /).2'3 Peptic ulcer disease is responsible may be from another lesion such as an for approximately 50% of the cases of ulcer or hemorrhagic gastritis. There- upper gastrointestinal bleeding in most fore, early investigation is mandatory series. Bleeding is directly responsible since treatment of these various lesions for 40% of all deaths encountered in the differs dramatically. The prognosis for 20% of patients who bleed as a result of patients with a variceal hemorrhage is peptic ulcer disease.4 Although duo- considerably worse than for patients denal ulcers are encountered more fre- with other lesions that may cause upper quently, gastric ulcers are more virulent gastrointestinal bleeding and may ap- because of their greater associated blood proach 50% mortality. The exact mech- loss and greater tendency for bleeding anism that precipitates variceal rupture to recur. and hemorrhage is not clear. No definite The term acute gastric mucosal lesions correlation has been found between the encompasses a wide range of disorders. degree of portal hypertension and the These include stress ulcers, Cushing's risk of subsequent bleeding. ulcer, Curling's ulcer, and erosive gas- The Mallory-Weiss syndrome has tritis. Although the exact pathophysiol- been recognized more frequently with ogy may vary among these disorders, the routine use of fiberoptic endoscopy the lesion that results is the same, mu- in studying patients with upper gas- cosal ulceration that does not penetrate trointestinal hemorrhage. Atkinson et the muscularis mucosa. The gross ap- al7 have described an increased trans- pearance of these lesions may vary from mural pressure gradient in the stomach a single bleeding point to a diffuse hem- and esophagus that may occur with pro- orrhagic gastritis that involves the entire tracted vomiting or any other condition mucosal surface of the stomach. that markedly increases intra-abdomi- The pathophysiology is at least par- nal pressure. This increased pressure tially understood in some of these dis- gradient seems to focus on the area of orders. In Cushing's ulcer, associated the gastroesophageal junction and may with patients who have had neurologi- result in mucosal laceration in this re- cal trauma or surgery, increased pres- gion. Risk factors include alcohol or Downloaded from www.ccjm.org on September 24, 2021. For personal use only. All other uses require permission. Summer 1982 Gastrointestinal bleeding 99 aspirin abuse, and the presence of a Table 3. Associated diseases in patients hiatal hernia. An endoscopic study re- with upper gastrointestinal bleeding3 8 ported by Knauer observed that most Percent tears involved either the gastroesopha- Cardiovascular 31.5 geal junction or the proximal stomach. Liver 13.2 Skeletal 8.9 Clinical presentation Chest 8.0 Although a good history and physical Central nervous system 8.0 Renal 6.3 examination are mandatory in studying Psychiatric 6.0 the patient with upper gastrointestinal Multiple system failure 6.0 bleeding, they are of limited value in Malignancy 5.3 identifying the source of bleeding. Allan Hormonal 3.9 and Dykes found the history to be di- Pancreas 1.2 Septicemia 1.0 agnostically helpful in only 57% of cases. Blood dyscrasia 0.7 Twenty-five percent of patients with chronic dyspepsia do not have a bleed- ing peptic ulcer, and conversely, up to tients with critical associated medical one third of patients with a bleeding diseases are at higher risk and therefore peptic ulcer deny any previous symp- should be identified. Table 3 shows a toms.10 Although esophageal varices are series reported by Halmagyi in which the most likely source in a patient with 65% of patients had at least one associ- classic signs of chronic liver disease, ated disease and 50% had more than other lesions must be considered. A one. thorough oropharyngeal examination should be conducted to eliminate this Diagnosis area as a possible source of the bleeding. Fiberoptic endoscopy is the procedure Most patients present either with of choice in identifying the source of melena or hematemesis; several associ- bleeding in a patient with upper gas- ated symptoms and signs are shown in trointestinal hemorrhage. Sugawa et al2 Table 2. Another important reason to reported a 97% accuracy rate with emer- obtain a thorough history is that pa- gency endoscopy, compared to a 33% accuracy rate with emergency upper Table 2. Signs and symptoms of upper gastrointestinal fluoroscopy. It is impor- gastrointestinal bleeding tant to perform endoscopy early because its accuracy decreases from 90% to 33% Percent after 48 hours.9 Melena 87.0 Four early studies in the 1970s cast Hematemesis 61.6 Pain 57.2 doubt on the efficacy of emergency en- Dizziness 46.8 doscopy in decreasing morbidity and Tachycardia 28.2 mortality associated with upper gas- I Ieartburn 24.0 trointestinal bleeding.9,11-13 Although Vomiting 22.6 these studies had several errors in exper- Shock 17.9 imental design, no subsequent prospec- Anorexia 17.7 Fainting 13.5 tive, randomized study has yet clearly Weight loss 12.4 shown that emergency endoscopy as the Bloating 4.7 single variable decreases the morbidity Pyloric obstruction 3.0 and mortality of upper gastrointestinal Downloaded from www.ccjm.org on September 24, 2021. For personal use only. All other uses require permission. f>4 Cleveland Clinic Quarterly Vol. 49, No. 1 bleeding. However, Griffiths et al14 iden- line, a central venous pressure line, a tified a subset of patients who did bene- Foley catheter, and perhaps a Swan- fit from early endoscopy. These patients Ganz catheter. A patient who has suf- all had a visible vessel in the base of the fered a major bleeding episode should ulcer associated with an 86% incidence be admitted to an intensive care area of bleeding recurrence, and a 14% inci- and initially resuscitated with crystal- dence of uncontrolled hemorrhage. All loid solutions. If indicated, whole blood of these patients required surgery. De- should be given after proper blood typ- spite the unproved benefit from early ing and crossmatching has been per- endoscopy, this aggressive attitude re- formed. In addition, fresh frozen plasma mains. An emergency upper gastrointes- should also be available for patients tinal barium study has many shortcom- with liver disease or for those who will ings. The barium may obstruct the view require several units of blood. Once in- of the endoscopist, and this study fails itial evaluation and resuscitation are un- to identify most of the superficial mu- derway, arrangements should be made cosal lesions. More importantly, barium for urgent endoscopy to delineate the in the gastrointestinal tract may obscure source of bleeding. any subsequent attempts at emergency Passing either a large nasogastric tube diagnostic angiography. or an Ewald tube is helpful for many When a rapidly bleeding lesion makes reasons. Return of a bloody gastric as- accurate endoscopy impossible, angiog- pirate suggests that the bleeding point raphy is the procedure of choice. Exper- is proximal to the ligament of Treitz. imentally, a bleeding rate of at least More importantly, evacuation of clots 0.5-1.0 ml/min is necessary to demon- from the stomach facilitates endoscopy. strate a bleeding site radiographically. Thorough gastric lavage controls the In practice, however, a bleeding rate of hemorrhage in approximately two 3-4 ml/min is probably necessary. Un- thirds of cases. Although iced saline is der these circumstances, the bleeding the most frequently used irrigating so- vessel may be identified in 50% to 70% lution, Ponsky et al advocate the use of cases.15 In addition to its diagnostic of room-temperature solutions.