Acute upper gastrointestinal bleeding: Analysis of 462 hospitalized Chinese patients in Shanghai

Abstract

While acute upper gastrointestinal bleeding (AUGIB) remains clinically critical, the etiology of bleeding and risk factors for mortality remain uncertain. This study aimed to evaluate the underlying causes of AUGIB and identify risk factors associated with fatality. A retrospective survey was conducted in a major clinical hospital in Shanghai, where inpatients diagnosed with AUGIB were meticulously collected and analyzed. This retrospective, single-center study encompassed 462 patients diagnosed with AUGIB from the emergency department of Changhai Hospital between January 1, 2019 to November 30, 2021. The analysis involved various aspects including epidemiology, gastroscopy, imaging, laboratory tests and data, clinical features, and hospital mortality. Among the 462 patients studied, 87% (400 out of 462) were afflicted with noncirrhotic hemorrhage (up to 38% with peptic ulcer), and 4 patients required digital subtraction angiography embolization (two died). The mortality rate was 12% (49/400), of which 49% (24/49) of patients died of malignant tumors. Mortality did not differ according to sex (χ² = 0. 851, P = .356) but was significantly influenced by age (χ² = 18.292, P = .001), hemoglobin levels (χ² = 22.908, P = .001), and comorbidity (χ² = 17.355, P = .001). In patients with esophageal and gastric variceal bleeding, the mortality rate was as high as 25% (16/62). Their mortality did not vary with sex (χ² = 0.318, P = .573), hemoglobin levels (χ² = 4.139, P = .247), or age (χ² = 7.674, P = .087), but was predominantly associated with hemorrhagic shock (88%, 14/16). Esophageal and gastric variceal bleeding and noncirrhotic hemorrhage patients observed significantly different mortality rates (P = .004). Peptic ulcers remained the primary cause of AUGIB. Old age, low hemoglobin levels, and comorbidities were the main risk factors contributing to mortality; nearly half of the patients died from malignant tumors. Hemorrhagic shock was independently associated with a high risk of in-hospital mortality in patients with cirrhotic hemorrhage.

1. Introduction

Upper gastrointestinal bleeding (UGIB) is a common issue affecting a large global population, with an annual incidence in Western countries ranging from approximately 80 to 150 cases per 100,000 individuals.^[1] The associated mortality rates are estimated to fall within the range of 2% to 15%,^[1] whereas the rebleeding rate is approximately 10%.^[2] A multicenter retrospective study in China, peptic ulcers remain the primary cause of nonvariceal upper gastrointestinal bleeding (NVUGIB) from 2008 to 2012.^[3] UGIB, defined as bleeding from the esophagus, stomach, or duodenum, is responsible for 50% or more of these hospitalizations.^[4] The rate of death among patients who are already hospitalized for another condition when UGIB develops is approximately 3 to 4 times as high as the rate among patients who are admitted to the hospital for UGIB.^[4] While acute upper gastrointestinal bleeding (AUGIB) mortality rates remain high, the elderly have witnessed a surge in AUGIB cases attributed to the use of nonsteroidal anti-inflammatory drugs (NSAIDs). To address these concerns, we conducted a comprehensive analysis of the clinical characteristics of 462 patients admitted to Changhai Hospital with confirmed AUGIB between January 1, 2019 to November 30, 2021. in Shanghai, China. Our primary objective was to analyze the etiology of AUGIB and associated risk factors for mortality.

2. Materials and methods

2.1. Study design and patients

This study recruited adult patients (aged ≥ 15 years) from January 1, 2019 to November 30, 2021 at Changhai Hospital. UGIB is defined as bleeding from the esophagus, stomach, or duodenum above the ligament of Treitz.^[4] Patients diagnosed with hematemesis (bright red emesis or coffee-ground emesis), hematochezia, or melena, which are indicative of UGIB.^[1] Patients with hemoptysis were excluded from the study. In cases of noncirrhotic hemorrhage, intravenous proton pump inhibitors (PPIs) are administered in the form of an 80 mg bolus followed by an 8 mg/h infusion for at least 72 hours.^[5] In case the hemoglobin level falls below 7–8 g/dL,^[6,7] or if hemorrhagic shock is present, blood transfusion is advised. In cases of esophageal and gastric variceal bleeding (EGVB), PPI is administered in combination with octreotide, somatostatin, or vasopressin.

2.2. Data collection and statistical analysis

We collected epidemiological, demographic, clinical, laboratory, therapeutic, and outcome data from patients^’ medical records. The clinical outcomes were followed up until November 30, 2021. A description of the clinical features, demographics, comorbidities, laboratory results, radiographic imaging, treatment, and clinical outcomes was also obtained. Descriptive statistical methods were employed, presenting the mean (‾χ ± SD) for normally distributed variables and median (IQR) for non-normally distributed variables. Categorical variables were expressed as counts (%). The chi-square test and Fisher exact test were used for statistical analysis and examination of risk factors for mortality, respectively. The SPSS 22 software was used for all analyses.

3. Results

In a total of 462 patients presenting with AUGIB enrolled in this study, 400 patients (87%) were diagnosed with noncirrhotic hemorrhage (group I), and 62 patients (13%) were diagnosed with EGVB (group II).

3.1. Diagnosis

Within the noncirrhotic hemorrhage group, 62% (248/400) of the patients underwent gastroscopy, of whom 152 (38%) were found to have peptic ulcers, with 11 patients (7% of the peptic ulcer cases) requiring gastroscopic hemostasis due to high-risk ulcers. Additionally, 50 patients (12% of the noncirrhotic hemorrhage group) were found to be taking oral NSAIDs and anticoagulants, while 21 patients (5% of the noncirrhotic hemorrhage group) were diagnosed with anastomositis (Table 1). Four patients with active gastrointestinal bleeding underwent digital subtraction angiography (DSA). DSA embolization was effective in 2 patients with duodenal cancer bleeding and gastric ulcer bleeding. Unfortunately, 2 patients succumbed to AUGIB and liver metastasis. Furthermore, 5 patients were diagnosed with gastric ulcers accompanied by cirrhosis, whereas 3 patients presented with other systemic malignancies.

Table 1.

Etiological distribution of 462 patients with acute upper gastrointestinal bleeding.

	Patients (n = 462)
EGVB*	62 (13%)
Noncirrhotic hemorrhage	400 (87%)
Peptic ulcer	152 (38%)
Tumors of the upper gastrointestinal tract	68 (17%)
NSAIDs,** antiplatelet drugs	50 (13%)
Anastomositis	21 (5%)
Erosive gastritis	12 (3%)
Esophagitis	13 (3%)
Mallord-Weiss syndrome	8 (2%)
Postoperative gastroscopic treatment	6 (2%)
Duodenitis	4 (1%)
Acute gastric mucosal lesions	8 (2%)
Stomach polyps	3 (1%)
Gastric telangiectasia	2 (1%)
Other	53 (13%)

^*EGVB = esophageal and gastric variceal bleeding.

^**NSAIDs = nonsteroidal anti-inflammatory drugs.

3.2. Patient characteristics

On admission, the patient presented with hematemesis and/or melena. Among 400 patients with noncirrhotic esophageal variceal bleeding, 299 (75%) were men, aged 15 to 95 years, with an average age of 65.9 years (SD = 15.9), 95% CI (64.3–67.5 years). Additionally, 290 (72.5%) patients were over 60 years of age. The median length of hospital stay was 7 days [IQR 5–11]). 27% (108) of the patients received blood transfusion. Furthermore, 224 (56%) patients had chronic diseases, including cardiovascular and cerebrovascular diseases (42% [95/224]), endocrine system disease (33% [75/224]), diabetes and malignant tumors (32% [71/224]), neurological disorders (3% [8/224]), chronic obstructive pulmonary disease (3% [7/224]), chronic renal failure (12% [27/224]), atrial fibrillation (4.9% [11/224]), hematological disease (3% [7/224]), and rheumatoid arthritis (2.7% [6/224]). Among the patient cohort, 9% (36/400) had a prior incidence of peptic ulcer bleeding, and 2.7% (11/400) had a prolonged history of alcohol consumption. Treatment outcomes revealed that 88% (351/400) of patients were discharged from the hospital, whereas 12% (49/400) of patients died (Table 2).

Table 2.

Clinical features and treatment results of noncirrhotic hemorrhage in 400 patients.

Age, years
Mean (SD)	65.9 (15.9)
Range	15–95
<40	30 (8%)
40–49	32 (8%)
50–59	48 (12%)
60–70	128 (32%)
>70	162 (41%)
Gender
Male	299 (75%)
Female	101 (25%)
Hemoglobin before transfusion (g/L)
<70	107 (27%)
70–89	85 (21%)
90–100	48 (12%)
>100	160 (40%)
Chronic medical illness
Cardiovascular and cerebrovascular diseases	95 (42%)
Endocrine system disease*	75 (33%)
Nervous system disease	8 (4%)
COPD**	7 (3%)
Malignant tumor	71 (32%)
Atrial fibrillation	11 (5%)
Hematological disease	7 (3%)
Arthritis	6 (2%)
Two or more comorbidities	224
Others
History of drinking alcohol	11 (3%)
History of peptic ulcer bleeding	36 (9%)
Clinical outcome
Discharged	351 (88%)
Died	49 (12%)

^* Mainly were diabetes.

^**COPD = chronic obstructive pulmonary disease.

3.3. Comorbidities

On admission, a significant number of the patients had impaired organ function. Out of 400 patients, 19% (76) of the patients developed severe complications, including acute myocardial infarction in 12% (9) patients, heart failure in 15% (11) patients, hemorrhagic shock in 16% (12) patients, pneumonia in 37% (28) patients, renal failure in 4% (3) patients, stroke in 7% (5) patients, multiple organ dysfunction syndrome (MODS) in 3% (2) patients, cardiac and renal failure in 4% (3) patients, and severe infection and heart failure in 1% (1) patient (Table 3).

Table 3.

Distribution of the patients’ comorbidity.

	Patients n = 76
Acute myocardial infarction	9 (12%)
Cardiac failure	11 (15%)
Hemorrhagic shock	12 (16%)
Pneumonia	28 (37%)
Renal failure	3 (4%)
Stroke	5 (7%)
MODS*	2 (3%)
Cardiac failure and renal failure	3 (4%)
Severe infection and heart failure	1 (1%)
Acute liver failure	2 (2%)

^*MODS = multiple organ dysfunction syndrome.

3.4. Risk assessment

On admission, the male mortality rate was 11% (34/299), while the female mortality rate was 15% (14/101), indicating no statistically significant difference between the sexes (χ² = 0.851, P = .356). However, when considering age groups (<40, 40–49, 50–59, 60–70, and >70 years), there was a significant statistical difference in mortality rates across these ages (χ² = 18.292, P < .001). Similarly, when examining hemoglobin levels (<70, 70–89, 90–100, and >100 g/L), there were significant statistical differences in mortality rates (χ² = 22.908, P < .001). Furthermore, when analyzing comorbidities (0 comorbidities, ≥1 comorbidity), there were significant statistical differences in the mortality rates (χ² = 17.355, P < .001) (Table 4).

Table 4.

Statistical differences in mortality rate among the 400 patients with noncirrhotic hemorrhage, grouped by gender, age, hemoglobin, and comorbidities.

n	400	Mortality (%)		Chi-squared test	P value
		No	Yes	χ 2	P
Gender				0.851	.356
Male	299	265	34 (11%)
Female	101	86	15 (15%)
Total	400	351	49 (12%)
Age, years				χ 2	P
<40	30	29	1 (3%)	18.292	.001
40–49	32	32	0
50–59	48	47	1 (2%)
60–70	128	111	17 (13%)
>70	162	132	30 (18%)
Hemoglobin after transfusion (g/L)				χ 2	P
<70	41	27	14 (34%)	22.908	.001
70–89	146	124	22 (15%)
90–100	54	51	3 (5%)
>100	159	149	10 (6%)
Comorbidities (%)				χ 2	P
0	176	168	8 (4%)	17.355	.001
≥1	224	183	41 (18%)

The most significant impact on the attributable risk factors for mortality was observed in individuals aged >70 years (OR = 13.8636, 95% CI = 1.8477–104.0224, P = .0004). Hemoglobin levels above 100 g/L were considered as the reference category, while hemoglobin levels below 70 g/L were associated with a higher risk of death (OR = 7.7259, 95% CI = 3.1124–19.1778, P < .001). Additionally, the presence of comorbidities (≥1) was associated with an increased risk of mortality (OR = 4.7049, 95% CI = 2.1439–10.3250, P < .001) (Table 5).

Table 5.

Multivariate analysis of factors associated with mortality.

Independent variable	OR value	95% Confidence interval	P value
Age, years
≤40–49	1
50–59	1.2979	0.0791–21.2959	1.0000
60–70	9.3423	1.2138–71.9084	.0079
>70	13.8636	1.8477–104.0224	.0004
Hemoglobin after transfusion (g/L)
>100	1
90–100	0.8765	0.2321–3.3100	1.0000
70–89	2.6435	1.2063–5.7931	.0148
<70	7.7259	3.1124–19.1778	<.001
Comorbidities (%)
0
≥1	4.7049	2.1439–10.3250	<.001

3.5. Mortality

By the conclusion of the study on November 30, 2021, a total of 49 (12%) patients with noncirrhotic hemorrhage had unfortunately succumbed to various causes. Among these fatalities, 24 (49%) patients were afflicted with malignant tumors, 8 (16%) patients experienced hemorrhagic shock, 4 (8%) patients suffered from acute myocardial infarction, 4 (8%) patients experienced acute heart failure, 2 (4%) patients experienced severe pneumonia respiratory failure, 2 (4%) patients developed MODS, 2 (4%) patients experienced asphyxia. Furthermore, 1 female patient over the age of 90 with an initial hemoglobin level of 40 g/L opted against receiving blood transfusions (Table 6).

Table 6.

Distribution of causes of death in 49 cases from 400 noncirrhotic hemorrhages.

Malignancy	24 (49%)
Hemorrhagic shock	8 (16%)
Acute myocardial infarction	4 (8%)
Acute heart failure	4 (8%)
Severe pneumonia respiratory failure	2 (4%)
MODS*	2 (4%)
Asphyxia	2 (4%)
Pyloric canal ulcer	1 (2%)
Intestinal obstruction	1 (2%)
Abandon blood transfusions	1 (2%)

^*MODS = multiple organ dysfunction syndrome.

On admission, 62 patients diagnosed with EGVB were included in this study; 41 patients (66%) were men, ranging from 35 to 84 years (57.4 years in average). Among the cohort, 18% (11/62) of the patients received blood transfusions, while 27% (17/62) had hepatocellular carcinoma. The treatment outcome revealed that 26% (16/62) of the patients succumbed to their condition.

3.6. Esophageal and gastric variceal bleeding

Further analysis showed that the mortality rate was 29% (16/46) in male and 19% (4/17) in female, but the difference was not statistically significant (χ² = 0.318, P = .573). The age groups (<40, 40–49, 50–59, 60–70, >70 years) revealed no significant difference in age distribution (χ² = 7.674, P = .087). Similarly, the hemoglobin level groups (<70, 70–89, 90–100, and >100 g/L) showed no statistically significant difference in hemoglobin distribution (χ² = 4.139, P = .247). Of the 16 cases that resulted in death, 14 (88%) were attributed to hemorrhagic shock, one (6%) to MODS, and the other one (6%) to hepatic encephalopathy (Table 7).

Table 7.

62 patients with esophageal and gastric variceal bleeding (Fish test).

Gender	n	Mortality			(Patients n = 62)
		No	Yes		χ 2		P value
Male	41	29	12 (29%)		0.318		.573
Female	21	17	4 (19%)
Total	62	46	16 (25%)
Age, years					χ 2		P
<40	4	4	0		7.674		.087
40–49	14	11	3 (21%)
50–59	17	14	3 (17%)
60–70	18	10	8 (44%)
>70	9	7	2 (22%)
Hemoglobin after transfusion (g/L)					χ 2		P
<70	20	17	3 (15%)		4.139		.247
70–89	16	13	3 (18%)
90–100	4	2	2 (50%)
>100	22	14	8 (36%)
Others
Hepatocellular carcinoma					17 (27%)
Death from EGVB*		(Patients, n = 16)
Hemorrhagic shock		14 (88%)
MODS**		1 (6%)
Hepatic encephalopathy		1 (6%)
Comparison of mortality		n	Discharge	Death		χ 2		P
Noncirrhotic hemorrhage		400	351	49 (12%)		8.160		.004
EGVB*		62	46	16 (25%)

^*EGVB = esophageal and gastric variceal bleeding.

^**MODS = multiple organ dysfunction syndrome.

The mortality rate for noncirrhotic hemorrhage in 400 patients was 12% (49/400), whereas the mortality rate for EGVB in 62 patients was 25% (16/62), indicating a significant statistical difference between noncirrhotic hemorrhage and cirrhotic hemorrhage (χ² = 8.16, P = .004) (Table 7).

Patient 1: A 64-year-old man with a diagnosis of gastric ulcer stage A1 and coronary heart disease, experienced cardiac arrest on day 1 of bleeding cessation. Cardiopulmonary resuscitation was successfully administered to restore autonomic rhythm and consciousness. Coronary angiography was performed without the placement of coronary stents, indicating possible coronary spasms. The patient had good prognosis.

Patient 2: A 79-year-old man with a prior gastric ulcer surgery, presented with bleeding from a pyloric channel ulcer, as observed during gastroscopy (Fig. 1). Hemostasis was attempted during gastroscopy, followed by unsuccessful DSA embolization to control bleeding. Owing to persistent active hemorrhage, surgical intervention was deemed infeasible. Subsequently, the patient experienced severe hemorrhagic shock and ultimately succumbed to the condition.

Figure 1.

Residual gastric pyloric channel ulcer with active bleeding of a 79-year-old man (patient 2).

Patient 3: A 65-year-old man with hepatitis B and oral aspirin, presented to the emergency department with coffee-like blood vomiting at approximately 3:00 am. Upon arrival, the patient had a blood pressure of 76/48 mm Hg, heart rate of 86 beats per minute, hemoglobin level of 119 g/L, platelet count of 71 × 10⁹/L, prothrombin time of 53 seconds, and an INR of 6.3. Immediate interventions included administration of a PPI, somatostatin, terlipressin, vitamin K1, and fluid resuscitation. Despite these measures, the patient experienced another episode of coffee-like vomiting. One hour after admission, the patient’s blood pressure was reevaluated to be 87/52 mm Hg. However, 3 hours later after the patient’s arrival, the patient experienced sudden respiratory cardiac arrest. It is important to note that cirrhotic hemorrhage can lead to unexpected and rapid mortality.

4. Discussion

This descriptive study provides a comprehensive analysis of the epidemiological and clinical characteristics of 462 patients admitted to the emergency department of Changhai Hospital in Shanghai, China. These findings shed light on the prevailing disease patterns in the emergency wards of Shanghai. In our study, 38% (152/400) of the patients were diagnosed with peptic ulcers, consistent with NVUGIB. Notably, peptic ulcer remains the leading cause of such bleeding in China and America.^[3,8]

Our study further elucidates the various causes of AUGIB, including peptic ulcers, upper gastrointestinal tumors, cirrhotic hemorrhages, antiplatelet drugs and NSAIDs, anastomositis, erosive gastritis, esophagitis, Mallord-Weiss syndrome, postoperative gastroscopic therapy, duodenitis, acute gastric mucosal lesions, gastric polyps, and telangiectasia of the base of the stomach. Ronald et al reported that the most common causes of UGIB include peptic ulcer disease, Mallory-Weiss syndrome, erosive gastritis, duodenitis, esophagitis, malignancy, angiodysplasia, and Dieulafoy lesion. Upper gastrointestinal endoscopy is the gold standard for diagnosis and treatment.^[9]

Cirrhotic hemorrhages pose significant risks, as they often result in hemorrhagic shock, rapid confusion, respiratory failure, and mortality within 24-hour time frame. Early interventions such as tips or gastroscopic sclerotherapy are typically employed. The administration of hemostatic drugs did not improve outcomes in cirrhotic patients with AUGIB.^[10] The mortality rate of EGVB in our study (25%) was higher than that in the Spanish study (13%),^[11] possibly due to population and methodological differences.

The use of acid suppression and vasoactive drugs has demonstrated a decrease in recurrent bleeding in both nonvariceal and variceal bleeding cases. However, in cases of nonvariceal bleeding that do not respond to conventional treatment, angiographic intervention is now favored over surgical intervention.^[12] In our study, 4 patients with active gastrointestinal bleeding underwent DSA embolization. Two patients experienced successful outcomes, specifically in cases of duodenal cancer and gastric ulcer bleeding. However, unfortunate outcomes were observed in the other 2 patients, who succumbed to pyloric canal ulcer bleeding and liver metastases. For patients with a high risk of subsequent bleeding or mortality due to AUGIB, endoscopy should be performed within 6 to 24 hours after consultation.^[13] Endoscopy is the first-line treatment, and endoscopic hemostasis can be achieved in most cases, even in ulcers >2 cm. Surgery is indicated in patients in whom arterial bleeding cannot be controlled by endoscopy. Angiographic embolization is an optimized option.^[14] The mortality rate of UGIB is high because affected patients often have serious preexisting illnesses.^[15] Consistent with our results, the cause of death was strongly associated with comorbid diseases. Our study revealed that the primary causes of mortality in patients with AUGIB were malignant tumors, cirrhosis hemorrhage, hemorrhagic shock, acute myocardial infarction, acute heart failure, severe pneumonia, respiratory failure, MODS, and pyloric canal ulcer. The mortality rate was 12% (49/400) in our study, consistent with a European study (around 10%)^[16] and an international multicenter study (2%–15%),^[2] which was much higher than that reported in another study in China (1.71%) where only 3.53% (105/2977) of patients were admitted to an emergency ward, and only 14.41% (429/2977) of patients had comorbid conditions.^[3] Jairath demonstrated that wide differences in reported NVUGIB mortality rates are attributable to the different methodologies adopted and populations studied.^[17]

Initial stabilization and resuscitation of patients with UGIB are imperative, and minimally invasive techniques have surpassed surgery as the treatment of choice in most cases.^[18] Thromboelastography provides a more comprehensive global coagulation assessment than routine tests and may avoid unnecessary blood component transfusions in EGVG and NVUGIB.^[19] In an international randomized, double-blind, placebo-controlled trial, it was found that tranexamic acid did not reduce death from gastrointestinal bleeding. Thus advised tranexamic acid should not be used to treat gastrointestinal bleeding.^[20]

Despite normal blood pressure, the case of sudden onset of cold sweats and the presence of severe anemia should be considered a potential massive hemorrhage in the gastrointestinal tract. Specific complications have been actively addressed to enhance the survival rates. For instance, in the case of acute myocardial infarction, prompt percutaneous coronary intervention is recommended, whereas acute heart failure necessitates reduced fluid intake and diuretic administration.

In recent years, during gastroscopy procedures conducted on patients with gastrointestinal tract issues, we observed that gastric cardia tumors and gastrointestinal stromal tumors were not unusual. In the case of noncirrhotic hemorrhage, addressing the associated complications may improve survival rates to some extent. Nonetheless, it has been proposed that PPIs do not improve clinical outcomes such as further bleeding, surgery, or death.^[21]

Although the incidence of cirrhotic hemorrhage has declined in recent years, it would be advantageous to include a larger number of patients from multiple hospitals in Shanghai and other cities in China to obtain more comprehensive knowledge. Nonetheless, it is important to note that the data of 462 patients in this study pertain specifically to the clinical characteristics of AUGIB in Changhai Hospital, one of the major clinical hospitals in Shanghai, China.

In summary, our study indicates that peptic ulcers remain the primary etiology of AUGIB, followed by malignant tumors and EGVB. The risk factors for mortality in cases of noncirrhotic hemorrhage include old age, low hemoglobin levels, and comorbidities. Notably, nearly half of the patients who succumbed to their condition had malignant tumors. In cases of EGVB, death is predominantly associated with hemorrhagic shock.

Acknowledgments

Gratitude is owed to Dr Zhou You-Lang and Dr Chen Qi for statistical analysis, and to Prof Zhang Rong-Jia and Prof Paul Hsieh for revising the manuscript.

Author contributions

Conceptualization: Jian-Ping Zhou, Mei-tang Wang.

Data curation: Jian-Ping Zhou, Min Wu.

Formal analysis: Jian-Ping Zhou.

Funding acquisition: This research received no external funding.

Investigation: Jian-Ping Zhou, Min Wu, Mei-tang Wang.

Methodology: Mei-tang Wang.

Project administration: Mei-tang Wang.

Resources: Jian-Ping Zhou.

Software: Jian-Ping Zhou.

Supervision: Min Wu.

Validation: Min Wu.

Visualization: Jian-Ping Zhou.

Writing – original draft: Jian-Ping Zhou.

Writing – review & editing: Jian-Ping Zhou, Min Wu, Mei-tang Wang.