Predictors of Mortality in End-stage Renal Disease Patients Presenting with Upper Gastrointestinal Bleeding

Abstract

Introduction

Upper gastrointestinal bleeding (UGIB) is a life-threatening medical condition. Patients with end-stage renal disease (ESRD) are at an increased risk of bleeding and mortality. The etiology of UGIB in ESRD patients includes peptic ulcer disease, angiodysplasia, and the administration of anticoagulants. Therefore, we aimed to identify the factors predicting mortality in ESRD patients presenting with UGIB.

Materials and methods

This prospective cohort study was carried out at the department of hepatogastroenterology, Sindh Institute of Urology and Transplantation (SIUT), from February 2022 to December 2023. For this study, 209 patients with ESRD on hemodialysis presenting with symptoms of NVUGIB were enrolled. Vital signs, laboratory parameters, and endoscopic findings were noted. The outcome was measured in terms of 1-month mortality. Univariate followed by multivariate logistic regression analysis was performed to identify the independent predictors of mortality in the ESRD population presenting with upper GI bleeding.

Results

Out of 209 subjects, 92 died within 1 month. On multivariate analysis, diagnosis of diabetes was significant at p = 0.01, odds ratio (OR): 4.6; hypotension, p = 0.001, OR: 2.37; tachycardia, p ≤ 0.001, OR: 1.7; Forrest class Ia ulcers, p ≤ 0.001, OR: 7.14; increased International Normalized Ratio (INR), p = 0.009, OR: 1.2; low platelet count, p = 0.002, OR: 1.9; and dual endoscopic therapy need, p ≤ 0.001, OR: 44.9.

Conclusion

This study has underlined that diabetes, hemodynamic instability, certain endoscopic findings, and the need for dual therapy are significant predictors of mortality in ESRD patients with UGIB. Aggressive management of such factors, if possible with early identification, significantly improves the outcome in this high-risk population.

Introduction

Upper gastrointestinal bleeding (UGIB) is one of the most common medical emergencies throughout the world, mainly when it has an acute onset, with an incidence of about 80 per 1,00,000 persons and is associated with a mortality rate of approximately 2–10%.^1–4 Recently, many strategies have been introduced, mainly focusing on the distribution of the patients into high-risk and low-risk categories, which not only help in making clinical decisions like timing of endoscopy and duration of stay in hospital in these patients but also aid in the level of care provided to these patients, respectively.^5,6

Chronic kidney disease (CKD) affects about 700 million people worldwide, with an increased risk of progressing toward end-stage renal disease (ESRD).⁷ Upper GI bleeding in ESRD can be multifactorial, including peptic ulcer disease, angiodysplasia, use of antiplatelet medications or unfractionated heparin and other anticoagulants, uremic platelet dysfunction, and erosive ulcers.⁸ Among these causes, the most common etiology of upper GI bleed in ESRD is peptic ulcer disease.⁹

In the general population, there are several noninvasive risk assessment scores and tools that are available for nonvariceal UGIB (NVUGIB), including the Glasgow Blatchford Score (GBS) and AIMS65 for the prediction of mortality and risk of rebleeding in patients with UGIB, with the main focus on mortality as a primary outcome.^10,11

The latest multicentered prospective study compared the current risk scores and discovered that GBS has the best role in predicting not only the need for intervention but also the risk of mortality in patients with UGIB.¹² However, the ability of this score in predicting mortality in UGIB was substandard [area under the receiver operating curve (AUROC): 0.74]. The age, blood test comorbidities (ABC) score proposed by Laursen et al. showed good accuracy in predicting 1-month mortality in patients with acute GI bleed.¹³

In patients with ESRD presenting with upper GI bleeding, data are scarce regarding the role of endoscopic intervention and the risk of in-hospital mortality in this population. Laeeq et al. proposed the need for endoscopic intervention in ESRD patients with GI bleeding having a GBS of more than 14 in the Pakistani population.¹⁴

Hence, there is a need to identify the factors that accurately predict mortality in patients with NVUGIB during the hospital stay in the hemodialysis population. Identifying these parameters will not only help decrease the mortality associated with upper GI bleed in the ESRD population but will also decrease the morbidity and improve the quality of life in this population.

Therefore, we aimed to identify the factors predicting mortality in ESRD patients presenting with UGIB.

Operational Definitions

Nonvariceal Upper GI Bleed¹⁵

It was defined as the presence of the origin of bleeding proximal to the ligament of Treitz, confirmed on endoscopy with one of the following findings: Vomiting of fresh blood, melena, or a decrease in hemoglobin (Hb) levels of ≥2 gm/dL from a prior examination.

Rebleeding15

It was defined as GI hemorrhage with vomiting of fresh blood, melena, or hemodynamic instability, and confirmed as recurrent bleeding episodes from the same source on endoscopy.

End-stage Renal Disease16

According to the KDIGO 2012 clinical practice guideline, ESRD is defined as a glomerular filtration rate (GFR) less than 15 mL/min for more than 3 months.

Materials and Methods

Study Design and Study Duration

After approval from the ethical review committee (ERC) SIUT-ERC-2021/A-1096 (Approval 1096), Sindh Institute of Urology and Transplantation (SIUT), this prospective cohort study was conducted at the Department of Hepatogastroenterology, SIUT, Karachi, from February 1, 2022, to December 31, 2023. All the patients of either gender and aged >18 years with ESRD on hemodialysis and presenting with NVUGIB (as per operational definition) were enrolled in this study. However, those patients with age <18 years or history of underlying chronic liver disease or those with history of bleeding either from tumor or after endoscopic resection, those with a history of ischemic heart disease or taking anticoagulants, or those refusing to participate were excluded from the study.

Sample Size

The sample size was 209.

Sample Technique

A nonprobability consecutive sampling technique was used.

Data Collection Procedure

Informed written consent was obtained from all the patients fulfilling the inclusion criteria before enrollment in the study. Vital signs, including blood pressure and pulse rate, were recorded at the time of admission. The presence of syncope, hypotension, and tachycardia at the time of presentation was also noted. Laboratory investigations, including complete blood count and renal function tests, were also noted. Each patient underwent upper GI endoscopy (within 24 hours), and the endoscopic findings were noted. These patients were then followed for survival at 1 month after the episode of GI bleeding.

Statistical Analysis

Statistical Package for the Social Sciences (SPSS) version 24.0 was utilized for the data entry and analysis. Mean and standard deviation (SD) were computed for the continuous variables, including age, Hb, platelet count, and other lab parameters, while the frequencies and percentages were used for the expression of the categorical variables, including gender and previous history of upper GI bleeding. The outcome was measured in terms of the presence or absence of mortality at 1 month. Chi-square test was applied for the comparison of categorical variables in predicting 1-month mortality risk in patients with GI bleeding in the hemodialysis population, while Student's t-test was used for the comparison of continuous variables. A p-value ≤ 0.05 was considered statistically significant.

Variables with statistical significance on univariate analysis then underwent multivariate logistic regression analysis to identify the independent predictors of mortality in ESRD patients presenting with upper GI bleeding.

Results

This study involved 209 participants, with a male predominance [121 (57.9%)]. Baseline characteristics are shown in Table 1. The most common comorbidities included ischemic heart disease [79 (37.8%)], hypertension [74 (35.4%)], and diabetes [56 (26.8%)]. Endoscopic findings revealed that out of 209 patients, 199 (95.2%) patients had duodenal ulcers, with various classifications indicating severity. Presenting symptoms such as tachycardia and hypotension were noted in 104 (49.8%) and 71 (34%) patients, respectively. Most of the patients [120 (57.4%)] received single therapy during endoscopic intervention, such as hemoclip application or adrenaline injection, while 89 (42.6%) patients required both endoscopic interventions in one setting for the control of bleeding. 1-month mortality was observed in 92 (44%) patients.

Table 1.

Baseline characteristics of the ESRD population presenting with upper GI bleeding (n = 209)

Study population	n (%)
Gender
Male	121 (57.9)
Female	88 (42.1)
Comorbidities
Diabetes	56 (26.8)
Hypertension	74 (35.4)
Ischemic heart disease	79 (37.8)
Endoscopic findings
Gastric erosions	71 (34)
Duodenal ulcers	199 (95.2)
Forrest class Ia	36 (17.2)
Forrest class Ib	39 (18.7)
Forrest class IIa	33 (15.8)
Forrest class IIb	36 (17.2)
Forrest class IIc	30 (14.4)
Forrest class III	25 (12)
Presenting signs and symptoms
Hypotension	71 (34)
Syncope	2 (1)
Tachycardia	104 (49.8)
Type of endoscopic treatment received
Single therapy	120 (57.4)
Dual therapy	89 (42.6)
	Mean ± SD
Mean age (years ± SD)	47 ± 12.7
Total leucocyte count (×109/L)	14.2 ± 6.4
Platelet count (×109/L)	175 ± 101
Serum urea (mg/dL)	142.7 ± 112.9
Serum creatinine (mg/dL)	4.3 ± 2.9
Outcome
Survived	117 (56)
Expired	92 (44)

ESRD, end-stage renal disease; GI, gastrointestinal; SD, standard deviation

Comparison between survivors and nonsurvivors highlighted several significant predictors of mortality, including the male gender (p = 0.002), diabetes (p ≤ 0.001), ischemic heart disease (p ≤ 0.001), hypotension (p ≤ 0.001), tachycardia (p ≤ 0.001), and specific endoscopic findings like gastric erosions and Forrest class Ia and IIa ulcers (Table 2). However, on multivariate Cox-regression analysis, diabetes (p = 0.01) [odds ratio (OR): 4.6], hypotension (p = 0.001) (OR: 2.37), tachycardia (p ≤ 0.001) (OR: 1.7), Forrest class Ia ulcers on endoscopy (p ≤ 0.001) (OR: 7.14), increased International Normalized Ratio (INR) (p = 0.009) (OR: 1.2), decreased platelet count (p = 0.002) (OR: 1.9), and the administration of dual endoscopic therapies (p ≤ 0.001) (OR: 44.9) were the independent predictors of mortality (Table 3).

Table 2.

Comparison of different variables in predicting mortality in upper GI bleeding in hemodialysis patients

Variable	Mortality		p-value
	Nonsurvivors (n = 92) N (%)	Survivors (n = 117) N (%)
Gender
Males	64 (69.5)	57 (48.7)	0.002
Females	28 (30.5)	60 (51.3)
Hypertension
Yes	26 (28.3)	48 (41.1)	0.055
No	66 (71.7)	69 (58.9)
Diabetes
Yes	52 (56.5)	4 (3.4)	≤0.001
No	40 (43.5)	113 (96.6)
Ischemic heart disease
Yes	58 (63)	21 (17.9)	≤0.001
No	34 (37)	96 (82.1)
Heart failure
Yes	20 (21.8)	24 (20.5)	0.829
No	72 (78.2)	93 (79.5)
Presence of syncope
Yes	2 (2.2)	0 (0)	0.109
No	90 (97.8)	117 (100)
Presence of hypotension
Yes	56 (60.8)	15 (12.8)	≤0.001
No	36 (39.2)	102 (87.2)
Presence of tachycardia
Yes	64 (69.5)	40 (34.2)	≤0.001
No	28 (30.5)	77 (65.8)
Gastric erosions
Yes	41 (75)	30 (24.7)	0.001
No	51 (25)	87 (75.3)
Gastric ulcers
Forrest class Ia	30 (32.6)	6 (5)	≤0.001
	62 (67.4)	111 (95)
Forrest class Ib	25 (27.2)	14 (12)	0.005
	67 (72.8)	103 (88)
Forrest class IIa	27 (29.9)	6 (5)	≤0.001
	65 (70.1)	111 (95)
Forrest class IIb	14 (15)	22 (18.8)	0.496
	78 (85)	95 (81.2)
Forrest class IIc	18 (19.6)	12 (10.3)	0.57
	74 (80.4)	105 (89.7)
Forrest class III	10 (10.9)	15 (13)	0.666
	82 (89.1)	102 (87)
Type of endoscopic treatment received
Single	23 (25)	66 (56.4)	<0.001
Dual	59 (75)	51 (33.6)
Hb (gm/dL)	8.4 ± 2.3	10.1 ± 1.6	<0.001
Total leucocyte count (×109/L)	15.3 ± 7.1	13.3 ± 5.6	0.08
Platelet count (×109/L)	195 ± 101	160 ± 99	0.03
INR	1.4 ± 0.6	1.2 ± 0.2	0.021
Serum urea (mg/dL)	162 ± 109	124 ± 113	0.021

Hb, hemoglobin. Bold values indicated that significant factors on univariate

Table 3.

Multivariate Cox-regression analysis showing independent predictors of mortality in the hemodialysis population presenting with upper GI bleeding

Variables	p-value	Odds ratio	CI (95%)
			Lower limit	Upper limit
Gender	0.677	1.468	0.241	8.9
Diabetes	0.01	4.6	1.23	6.84
Ischemic heart disease	0.119	0.180	0.021	1.555
Hypotension	0.001	2.37	1.925	3.457
Tachycardia	≤0.001	1.7	0.92	2.249
Forrest class Ia	≤0.001	7.14	3.20	9.12
Forrest class IB	0.142	0.29	0.026	1.692
Forrest class IIa	0.314	3.2	0.335	30.3
Hemoglobin	0.085	0.95	0.798	1.643
INR	0.009	1.2	0.81	1.77
Platelets	0.002	1.981	0.969	2.493
Serum urea	0.487	1.003	0.995	1.011
Administration of dual therapies on endoscopy	≤0.001	44.9	39.1	48.9

CI, confidence interval. Bold values indicates the multivariate analysis

Discussion

Upper GI bleeding is a significant complication resulting in increased mortality in the hemodialysis population. Predictors of mortality among NVUGIB were investigated in ESRD patients on hemodialysis in the present study. Our study, therefore, underlines the crucial factors associated with increased mortality that could be of immense help in clinical decision-making to achieve improved patient outcomes.

Upper gastrointestinal bleeding is considered a well-recognized medical emergency in the general population, with an incidence of 80/1,00,000 persons and a mortality rate ranging from 2 to 10%.^1–4,17 In contrast, the ESRD population presenting with UGIB has substantially higher mortality rates.¹⁸ We observed a 44% mortality rate within 1 month in our study, significantly higher than reported rates in the general population.

This disparity underlines the vulnerability of ESRD patients to poor outcomes when confronted with UGIB. Indeed, other studies dealing with the population of patients with ESRD also reported that UGIB is multifactorial, and impaired ability to manage bleeding as a consequence of uremic platelet dysfunction, anticoagulant use, and comorbid conditions contribute to higher mortality.⁸

Indeed, endoscopy plays a closely related role in the management of UGIB by not only identifying the source of bleeding but also offering a scope for therapeutic intervention. Among our patients, 95.2% had duodenal ulcers, with many of them falling into Forrest class Ia and IIa. The presence of Forrest class Ia ulcers was found to be a strong predictor of mortality with an OR of 7.14 on multivariate analysis. This finding is in agreement with previous studies that relate Forrest Ia ulcers to an increased risk for rebleeding and mortality, especially among ESRD patients, in whom impaired renal function accentuates hemodynamic instability and bleeding diathesis.^19,20

The presence of comorbidities is another factor that influences the mortality rates in the ESRD population presenting with UGIB. We noticed a significant association of higher mortality rates in patients with a history of comorbidities like diabetes mellitus. Diabetes is a concern since it has been shown to adversely affect wound healing and increase peptic ulcer disease severity.²¹

Various studies have shown that the presence of comorbidities in the ESRD population was associated with high mortality in patients with UGIB. Laursen et al. discussed how different comorbid conditions, especially cardiovascular disease, were more strongly associated with increased mortality risks among patients with UGIB and notably those with renal dysfunction.¹³

The other crucial mortality predictor was hemodynamic instability at the time of presentation. In our study, the presence of signs of hemodynamic instability like hypotension (OR: 2.37) (p = 0.001) and tachycardia (OR: 1.7) (p ≤ 0.001) on admission was strongly suggestive of poor outcome in the ESRD population with UGIB. These findings were comparable to the previous studies, which stated that hemodynamic instability is a major driver of mortality in UGIB patients. A recent study by Lee et al. on patients with CKD with NVUGIB showed that hypotension and tachycardia, upon presentation, were significantly associated with in-hospital mortality. These findings suggest that early recognition and aggressive management of hemodynamic instability are critical for improving outcomes in this high-risk population.²² End-stage renal disease patients are more sensitive to fluid overload, anemia, and cardiovascular dysfunction, which makes them more prone to hemodynamic fluctuations. They are, therefore, notoriously more sensitive to the injurious consequences of an acute bleed.

The increased INR (OR: 1.2) (p = 0.009) and decreased platelet count (OR: 1.9) (p = 0.002) were independent predictors of mortality in our study. Our results indicated coagulation abnormality in ESRD patients, complicating the management of UGIB. Uremic platelet dysfunction is a well-recognized complication of ESRD, characterized by defective platelet aggregation and prolonged bleeding times. These patients are, therefore, at an increased risk of severe, especially protracted bleeding, while the frequent use of anticoagulation in this population provides a further contribution to higher mortality rates.⁸ Studies have shown that the patients who had UGIB and an elevated INR showed increased rebleeding and mortality rates, especially in conditions of renal dysfunction.²³ The coexistence of platelet dysfunction with impaired coagulation also demands management strategies tailored to the particular condition in order to limit life-threatening bleeding episodes in patients with ESRD.

In our study, we observed higher mortality rates among the patients receiving dual endoscopic therapies (OR: 44.9) (p ≤ 0.001). Although an endoscopic intervention is of importance during the active bleeding stage through control, the need for multiple therapies could indicate a more serious form of bleeding or failure of the initial treatments, both factors associated with higher mortalities.¹⁴

One of the major limitations of our study was the single-center design, which may limit the generalizability of the findings to other healthcare settings. Secondly, the study focuses on short-term (in-hospital and 1 month) mortality, and therefore, the patients were not followed for long-term consequences after an episode of UGIB in the ESRD population, such as quality of life and admission with rebleeding. Third, we did not analyze the effect of nutritional status and frequency of hemodialysis on UGIB in the hemodialysis population.

Conclusion

This study identified the main predictors of short-term mortality in the ESRD population presenting with UGIB, such as the presence of comorbidities like diabetes, hemodynamic instability, coagulation abnormalities, endoscopic findings of bleeding ulcers, and the need for more than one endoscopic intervention. These findings stress the need for early detection and aggressive treatment of these risk factors to provide optimal outcomes. The high mortality rate in this study further demonstrates how vulnerable this population with ESRD is to poor outcomes once UGIB is presented and emphasizes the need for approaches to tailor-made therapies. Future research should be directed at long-term outcomes and the effect of nutritional status and the frequency of dialysis on the management of UGIB in this high-risk population.

Ethical Approval

Approved under SIUT-ERC-2021/A-1096 (Approval 1096).

Data Availability Declaration

The data supporting the findings of this study are available from the corresponding author upon reasonable request.

Acknowledgment

The corresponding author would like to acknowledge the support of the clinical, administrative, and researches involved in facilitating this study.

Authors’ Contributions

All authors contributed substantially to the conception and design of the study, data collection, data analysis, interpretation of results, manuscript drafting, and critical revision of the manuscript. All authors reviewed and approved the final version of the manuscript and agree to be accountable for the accuracy and integrity of the work.

Orcid

Raja Taha Yaseen Khan https://orcid.org/0000-0002-5504-5084