Outcome of acute upper gastrointestinal bleeding in patients with coronary artery disease: A matched case-control study.

BACKGROUND/AIM: The risk of upper gastrointestinal bleeding (UGIB) increases in patients with coronary artery disease (CAD) due to the frequent use of antiplatelets. There is some data reporting on treatment outcomes in CAD patients presenting with UGIB. We aim to determine the clinical characteristics and outcomes of UGIB in patients with CAD, compared with non-CAD patients. PATIENTS AND METHODS: We conducted a prospective multi-center cohort study (THAI UGIB-2010) that enrolled 981 consecutive hospitalized patients with acute UGIB. A matched case-control analysis using this database, which was collected from 11 tertiary referral hospitals in Thailand between January 2010 and September 2011, was performed. RESULT: Of 981 hospitalized patients with UGIB, there were 61 CAD patients and 244 gender-matched non-CAD patients (ratio 1:4). UGIB patients with CAD were significantly older, and had more frequently used antiplatelets and warfarin than in non-CAD patients. Compared with non-CAD, the CAD patients had significantly higher Glasgow-Blatchford score, full and pre-endoscopic Rockall score and full. Peptic ulcer in CAD patients was identified more often than in non-CAD patients. UGIB patients with CAD and non-CAD had similar outcomes with regard to mortality rate, re-bleeding, surgery, embolization, and packed erythrocyte transfusion. However, CAD patients had longer duration of hospital stays than non-CAD patients. Two CAD patients died from cardiac arrest after endoscopy, whereas three non-CAD patients died from pneumonia and acute renal failure during their hospitalization.
CONCLUSION: In Thailand, patients presenting with UGIB, concomitant CAD did not affect clinical outcome of treatment, compared with non-CAD patients, except for longer hospital stay.

Acute upper gastrointestinal bleeding (UGIB) is a common cause of hospital admissions with a mortality rate ranging from 4% to 14%.[12] The burden of comorbidity, including coronary artery disease (CAD) is greater among older patients. The risk of UGIB increases in patients with CAD due to frequent use of antiplatelets, including aspirin and/or clopidogrel with an overall prevalence of 1.2%–1.5% per year.[34] History of UGIB or peptic ulcer without concomitant antisecretory drugs such as proton-pump inhibitors (PPIs) or H2 receptor antagonists is an independent risk factor for UGIB in patients with CAD.[45]

Performing upper gastrointestinal (GI) endoscopy has many adverse effects on cardiopulmonary functions and can precipitate cardiovascular events, for example, cardiac arrhythmias, myocardial infarction, and ischemia, especially in patients with heart disease.[67] The incidence of GI endoscopy-related myocardial ischemia occurs in 16% of hospitalized patients with severe CAD.[8] In addition, preprocedure and postprocedure myocardial infarction occurs in 9.2% of high-risk CAD patients undergoing emergency GI endoscopy for UGIB.[9] Only a few studies have reported on the outcomes of treatment in patients with CAD presenting with acute UGIB. Several risk score systems are widely used for risk stratification of UGIB. The Glasgow–Blatchford score (GBS) identifies patients who need clinical intervention.[10] The pre-endoscopic Rockall score (RS) and full RS (including endoscopic findings) predict mortality or rebleeding in patients presenting with UGIB.[1112]

This study aims to determine the clinical characteristics, risk scores, treatment outcomes, including rebleeding, mortality, length of hospital stay, need for endoscopic interventions, surgery, and embolization in patients with known CAD who present with acute UGIB compared with patients without CAD.

PATIENTS AND METHODS

Data sources

We conducted a prospective multicentered study using the THAI UGIB-2010 database, collected in 11 tertiary referral hospitals in Thailand between January 1, 2010, and September 30, 2011. A matched case–control analysis was performed using the THAI UGIB-2010 database including 981 consecutive hospitalized patients managed for UGIB, enrolled from King Chulalongkorn Memorial Hospital, Chulalongkorn University (Bangkok), MaharatNakhonratchasima Hospital (NakhonRatchasima), Sawanpracharak Hospital (NakhonSawan), Surin hospital (Surin), MaharajNakhonsithammarat Hospital (NakhonSithammarat), HRH Princess MahaChakriSirindhorn Medical Center (Bangkok), Chonburi hospital (Chonburi), Bangkok Metropolitan Administration General hospital (Bangkok), Bangkok hospital (Bangkok), Rajavithi hospital (Bangkok), and Thammasat University hospital (PathumThani). Management of UGIB was based on the Consensus for Clinical Practice Guideline for the management of UGIB from the Thai Association for Gastrointestinal Endoscopy (TAGE). All patients admitted to the hospital underwent EGD according to a standardized examination protocol. If endoscopy was not available, patient referral for angiography or surgery was considered. Clinical management, treatment with PPI pre-endoscopy, blood transfusion, timing of endoscopy, and type of endoscopic therapy were determined by each gastroenterologist, although the majority were in accordance with the TAGE guideline.

Data collection was performed by the Clinical Research Collaboration Network (CRCN, Nonthaburi, Thailand) using standardized case record forms. Relevant demographic data were collected, including age, gender, presenting symptoms, history of alcohol consumption, comorbidities, history of NSAIDs/anticoagulant usage, hemodynamic status, laboratory parameters, and time to EGD. The following variables were recorded: Endoscopic findings, need for treatment (endoscopic therapy, packed erythrocyte transfusion, surgery, and embolization), and clinical outcomes (rebleeding and death).

The study was approved by the Institutional Review Board (IRB) of the Faculty of Medicine, Chulalongkorn University, Bangkok, Thailand (IRB number 480/51).

Case definition and identification of controls

Cases were patients who were 25 years or older, and were previously diagnosed with CAD with or without current angina symptoms. The diagnosis of CAD was based on criteria including more than 50% stenosis in the lumen of at least one coronary artery on prior coronary angiography, positive treadmill (exercise) stress test, or a history of myocardial infarction. Matched controls (non-CAD) were randomly selected from the remaining patients by the best match of 4:1 with cases based on gender. Patients qualifying in the control group were those declared without CAD history or any other comorbidity such as hypertension, diabetes, or chronic kidney disease.

Definitions

UGIB was defined as hematemesis (including coffee-ground vomiting), melena, or hematochezia.

Findings on gastroscopy that were considered high-risk features (stigmata) of upper GI bleeding included adherent clot, nonbleeding or bleeding visible vessel and varices with red color or white nipple sign.

Data analysis

We described categorical variables using number and percentage and compared groups using Pearson Chi-square test. We described continuous variables using means ± standard deviation (SD) and compared groups using the independent t-test (for normally distributed data) and Mann–Whitney test (for non-normally distributed data). Univariate and multivariate regression analyses were performed to identify prognostic factor for poor clinical outcomes. A P < 0.05 was considered statistically significant. Statistical analysis was performed using SPSS version 20.0 (IBM, NY, USA).

RESULTS

Patient characteristics

Among a total of 981 patients who presented with UGIB, 61 patients were known to have CAD (12 women, mean ± SD age = 68.9 ± 11.5 years) and 244 patients did not have CAD (50 women, mean ± SD = 55.7 ± 14.9 years). The demographic data, medical history, laboratory parameters, timing of EGD, GBS, full and pre-endoscopic RS between CAD patients compared with non-CAD are shown in Table 1. Patients with CAD were older, had more chronic kidney disease, frequently used antiplatelets and warfarin than patients without CAD. The mean GBS, full RS, and pre-endoscopic RS were significantly higher in patients with CAD than in non-CAD patients.

Table 1

Comparison of patient characteristics and clinical outcomes between acute UGIB patients with and without CAD

Causes of gastrointestinal bleeding and endoscopic findings

Endoscopic findings and treatments are shown in Table 2. Peptic ulcer bleeding was the main etiology in both groups of patients. In addition, peptic ulcers (75.4% vs 57.4%, respectively; P < 0.01), especially gastric ulcers (57.4% vs 36.1%, respectively) were more frequently found in CAD patients than in non-CAD patients. In contrast, esophageal varices were identified more in non-CAD patients than in CAD patients. High-risk stigmata on endoscopy did not significantly differ between the two groups. Requirement of endoscopic therapy was not statistically different between CAD (n = 12, 19.7%) and non-CAD groups (n = 72, 29.5%) (P > 0.05). Heater probe coagulation (9.8%) and adrenaline injection (9.8%) were the most commonly used methods for bleeding control in patients with CAD.

Table 2

Endoscopic findings and hemostasis between acute UGIB patients with and without CAD

Treatment outcome

Performing early endoscopy within 24 h was not different between patients with CAD (n = 40, 65.6%) and without CAD (n = 166, 68%) (P > 0.05). A comparison of clinical outcomes between CAD and non-CAD patients with UGIB is presented in Table 3. Rebleeding did not occur during admission or within one month in both groups of patients. Two CAD patients (3.3%) and three non-CAD patients (1.2%) died during hospitalization. The cause of death in the two patients with CAD was cardiac arrest after endoscopic therapy, whereas in non-CAD group it was pneumonia (n = 2) and acute renal failure (n = 1). Mortality rate within one month was 4.9% (n = 3) in patients with CAD and 1.6% (n = 4) in patients without CAD (P = 0.12). The length of hospital stay was significantly longer in patients with CAD (13.2 ± 48.7 days) than in non-CAD (4.4 ± 5.5 days) (P < 0.01). In addition, surgery, embolization, and mean number of units of packed erythrocyte transfused did not differ between the two.

Table 3

Outcome of treatment between acute UGIB patients with and without CAD

Prognostic factors for poor clinical outcomes

Factors that predicted poor clinical outcomes during admission, including surgery, embolization, rebleeding, and death during admission were assessed. Multivariate analysis revealed baseline hemoglobin <7 g/dL (OR = 5.0, 95%CI: 2.7–9.3, P < 0.01), hemodynamic instability (OR = 3.1, 95%CI: 1.5–6.5, P < 0.01) and high-risk stigmata on endoscopy (OR = 2.0, 95%CI: 1.0–4.0, P < 0.05) were associated with poor outcomes. CAD did not predict poor outcomes. However, CAD patients had a higher risk for hospital stays ≥5 days (OR = 2.6, 95%CI: 1.4–5.0, P < 0.01). Other factors, associated with a duration of hospital stay ≥5 days were surgery and therapeutic interventions (OR = 5.6, 95%CI: 1.3–24.4, P < 0.05), hemodynamic instability (OR = 3.3, 95%CI: 1.6–6.7, P < 0.01) and high-risk stigmata from endoscopy (OR = 2.8, 95%CI: 1.5–5.3, P < 0.01). Stepwise logistic regression analysis was performed to identify risk factors for duration of hospital stay ≥5 days in CAD patients (shown in Supplement Table 1) and found that nonpeptic ulcer bleeding (OR = 8.27, 95%CI = 2.0–33.8, P = 0.001) was an independent predictor for longer hospital stay. Half of the patients with nonpeptic ulcer bleeding had esophageal variceal bleeding. In addition, CAD patients who had high-risk endoscopic lesions, tended to have a higher risk for this outcome (P = 0.08). Early endoscopy or medications could not decrease the risk of longer hospital stay.

Risk factors for length of hospital stay ≥5 days in patients with coronary artery disease

Click here for additional data file.

DISCUSSION

The main findings of our study are: (1) In Thailand, 6.2% of patients presenting with acute UGIB had concomitant CAD; (2) Compared with non-CAD, UGIB patients with CAD had more advanced age, comorbidity, antiplatelet, and warfarin usage; (3) Peptic ulcer especially gastric ulcer was the main etiology of patients with CAD. The prevalence of peptic ulcer bleeding was significantly higher in patients with CAD than without CAD; (4) In-hospital rebleeding, mortality, surgery, embolization, and mean number of packed erythrocyte transfusion requirement were not different between UGIB patients with CAD and without CAD. However, patients with CAD were at a risk for longer hospital stay ≥5 days.

Patients with acute UGIB are more likely to be of advanced age, with 63%–65% of these patients aged ≥60 years,[113] and to have multiple comorbidities. For example, patients with CAD are at a risk for UGIB due to the frequent use of antiplatelet drugs. A multicenter survey from the United Kingdom found that 20% of patients had concomitant ischemic heart disease.[13] Our data showed a lower prevalence of CAD in Thai patients presenting with UGIB (6%). One potential difference, however, is the use of a different definition for CAD, as our current study used coronary stenosis ≥50% for the diagnosis of CAD,[14] whereas in the UK study, the definition was not documented.

Acute UGIB affects cardiac function and may precipitate myocardial infarction from hypovolemia, hemodynamic compromise, myocardial hypoperfusion, and compensated reflex tachycardia.[15] Besides, upper GI endoscopy has many adverse effects on cardiopulmonary function in patients with heart disease such as cardiac arrhythmias.[61617] In high-risk patients for CAD presenting with UGIB undergoing upper GI endoscopy, 4.6% had myocardial infarction pre-endoscopy and 4.6% postendoscopy.[9] Underlying heart disease, low blood pressure, low hemoglobin level, and persistent shock were predictors of procedure-related myocardial infarction.[9] Although both UGIB and upper GI endoscopy may contribute to cardiac events in patients with CAD, a previous study showed that clinical outcomes were not different from non-CAD. According to Tseng et al.'s. study,[18] inhospital rebleeding, mortality, surgery, embolization, and mean number of packed erythrocyte transfusion were not distinct between UGIB patients with and without CAD, consistent with results of our current study. However, our study found that patients with CAD had longer hospital stay (≥5 days) and had a higher mortality rate than non-CAD patients. Mean GBS, pre-endoscopic and full RS were higher in CAD patients when compared with non-CAD patients. Theoretically, the reason might be that CAD patients had more advanced age, and comorbidities. In CAD patients, nonpeptic ulcer bleeding, especially variceal bleeding was a predictor for longer hospital stay. The current study provided additional information about causes of death in UGIB patients with CAD, which were mainly cardiac complications. Conversely, deaths in patients without CAD were related to noncardiac organ failure or infection. Another difference between the two studies is that CAD in our study was matched with controls at a ratio of 1:4, whereas a ratio of 1:1 was used in the Tseng et al.'s study. Controls, or non-CAD patients in this study included patients without a history of CAD or other comorbidities, such as hypertension, diabetes, chronic kidney disease to exclude high-risk patients for CAD, but Tseng et al.'s included these comorbidities.

There are some limitations in this study. First, we did not evaluate ECG changes or cardiac enzyme levels before and after performing EGDs, thus the incidence of myocardial infarction or ischemia related to procedures is unknown and precluded us from calculating risk stratification scores for these cardiac events. Additionally, we could not conclude that the cause of death was associated with performing endoscopy. However, the main outcome of this study was to determine clinical outcomes of UGIB patients with a known history of CAD compared with patients without CAD. Second, the sample size was relatively small, which might lead to a statistical type 2 error and prevents us from identifying all significant differences. Third, most patients in this patient population (92.4%) received PPI infusion, which could have influenced endoscopic findings and clinical outcomes. Finally, the decision making regarding the selected endoscopic therapy and need for blood transfusions in the THAI UGIB study 2010 was subjective and varied between individual gastroenterologists.

CONCLUSIONS

This study demonstrates that clinical outcomes, including mortality, rebleeding, surgery, embolization, and packed erythrocyte transfusion requirements, are not different between UGIB patients with concomitant CAD and those without CAD. However, CAD patients had longer duration of hospital stays (≥5 days) compared with those without CAD.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.