Characteristics of non variceal upper gastrointestinal bleeding and factors related to recurrence at Hanoi medical university hospital, Vietnam

The study aims to evaluate the characteristics of non-variceal upper gastrointestinal bleeding and identify correlating factors of recurrent bleeding. A retrospective descriptive study was conducted on non-variceal upper gastrointestinal bleeding patients admitted to Hanoi Medical University Hospital with ICD 10 code K92.2 from January 2013 to March 2017. There were 444 patients with the mean age of 49.1 (18.1). There were 69.8% of patients with co-morbidities in which 25.9% had history of upper gastrointestinal bleeding.

JOURNAL OF MEDICAL RESEARCH CHARACTERISTICS OF NON-VARICEAL UPPER GASTROINTESTINAL BLEEDING AND FACTORS RELATED TO RECURRENCE AT HANOI MEDICAL UNIVERSITY HOSPITAL, VIETNAM Nguyen Phuc Binh1, Dao Viet Hang1,2, Tran Quoc Tien2, Dao Van Long1,2 Hanoi Medical University Hospital; 2Hanoi Medical University Hospital Upper gastrointestinal bleeding is an emergency requiring immediate management and cooperation of many specialties Among the causes of upper gastrointestinal bleeding, non-variceal upper gastrointestinal bleeding has the highest percentage Initial assessment, prognosis factor classification and suitable interventions will help to reduce recurrent bleeding rate The study aims to evaluate the characteristics of non-variceal upper gastrointestinal bleeding and identify correlating factors of recurrent bleeding A retrospective descriptive study was conducted on non-variceal upper gastrointestinal bleeding patients admitted to Hanoi Medical University Hospital with ICD 10 code K92.2 from January 2013 to March 2017 There were 444 patients with the mean age of 49.1 (18.1) There were 69.8% of patients with co-morbidities in which 25.9% had history of upper gastrointestinal bleeding The median Rockall score was and the median Glasgow-Blatchford Bleeding Score (GBS) was The rate of endoscopic interventions was 48.4% in which 99.1% achieved success The rate of recurrent bleeding in hospital was 4.5% and within 30 days after discharge was 1.1% There was no difference of recurrence in the groups performed mono and combined therapies Glasgow-Blatchford Score and Rockall score had low prognosis performance for in-hospital recurrence In conclusion, the rate of recurrence both in hospital and within 30 days in non-variceal upper gastrointestinal bleeding patients was low Keywords: non-variceal upper gastrointestinal bleeding; epidemiology; recurrent bleeding; related factors I BACKGROUND causes could be various including peptic ulcer, Mallory Weiss, malignancy, vascular malfor- Upper gastrointestinal bleeding is one of mation and unidentified injuries [1] Patients the most common gastrointestinal emergen- with upper gastrointestinal bleeding may de- cies which requires urgent assessment and velop recurrent bleeding in hospital (7 - 16%) interventions with mortality rate of - 15% [1] or after discharge (8%) [5; 6] To classify The incidence of upper gastrointestinal bleed- patients based on severity when admission ing ranges from 48 to 172/100.000 adults per and detect factors that are related to bleeding year in which men and old people have a recurrence is important to follow up and make higher rate [2 - 4] Among the etiologies of prognosis [7] In Vietnam, there have not been upper gastrointestinal bleeding, as many pre- many epidemiology studies in non-variceal vious studies recorded, non-variceal bleedings upper accounted for the highest percentage The centered research at 17 major hospitals in gastrointestinal bleeding A multi- Vietnam in 2015 recorded the rate of inCorresponding author: Dao Viet Hang, Hanoi Medical Univesity Email: hangdao.fsh@gmail.com Received: 11/1/2018 Accepted: 08/11/2018 JMR 116 E3 (7) - 2018 hospital recurrent bleeding in patients with upper gastrointestinal bleeding was 5.7% [8] Therefore, we decided to conduct our study at Hanoi Medical University Hospital to report 19 JOURNAL OF MEDICAL RESEARCH characteristics, recurrence rate and factors Recurrent bleeding was diagnosed with associated to recurrence in non-variceal upper symptoms of repeated hematemesis or black gastrointestinal bleeding patients II METHODS stools, a drop of Hemoglobin ≥ 2g/dl or changes of hemodynamic status after controlling bleeding or having yellow stool Cases of The study used a retrospective method in-hospital recurrent bleeding were taken from with convenient sample size, which was con- medical records, cases of recurrent bleeding ducted at Hanoi Medical University Hospital within 30 days after discharge were collected Medical records of patients who were admitted from contacting patients or patient’s family to HMUH from January 2013 to March 2017 members by phone numbers with diagnosis of upper gastrointestinal bleed- We analyzed and demonstrated data by ing according to the criteria of ICD code being using R program Statistical analysis included t K92.2 were collected Patients with melena -test, Mann – Whitney test for categorical vari- and/or hematemesis and endoscopy showing ables and Chi-square test, Fisher test for bleeding lesions in the upper GI tract except quantitative variables Logistic regression was variceal bleeding were included in the study We excluded patients who were admitted to HMUH and diagnosed with upper gastrointestinal bleeding but discharged immediately without upper endoscopy, further intervention and treatment, patients who did not provide correct addresses and contacts or patients used for evaluating the association of recurrent bleeding and related factors Recurrent bleeding’s predictive value of different scores was demonstrated by the area under the curve (AUC) A P-value of less than 0.05 was considered significant III RESULTS refused to attend in the study Patients characteristics included demographic information (age, gender), previous upper GI bleeding, comorbid diseases, arrival time, clinical symptoms, hemodynamic status on admission and after upper endoscopy, indication of blood transfusion, gastroscopic diagnosis, gastroscopic interventions and other Demographic characteristics Our study recorded 444 cases of nonvariceal upper gastrointestinal bleeding, among those 67.3% was male The average age (SD) was 49.1 (18 08), with the eldest patient being 91 years old and the youngest being 10 treatments during hospital stay Blood transfu- History sion was selected instead of haemoglobin Based on medical records, 69.8% of cases level since all patients with significantly low had at least one comorbid disease which con- heamoglobin would receive blood transfusion sisted of cardiovascular diseases, diabetes, Patients’ risks were evaluated by the Glasgow musculoskeletal diseases and liver diseases -Blatchford Bleeding Score to assess the need 25.9% had a past history of upper gastrointes- for intervention and the Rockall score (Pre- tinal bleeding 9.23% used non-steroidal anti- endoscopic and complete) to predict the risk of inflammatory drugs (NSAID) and/or coagula- recurrent bleeding and mortality tion before admission 20 JMR 116 E3 (7) - 2018 JOURNAL OF MEDICAL RESEARCH Glasgow-Blatchford Score and Rockall Symptoms, Glasgow-Blatchford Score and Rockall score score: To stratify risk, the Glasgow-Blatchford Symptoms: Black stool was the most com- Score and the Rockall score (both pre-endos mon symptom (83.8%) Other symptoms copy and complete) were used As a result, a recorded were hematemesis (32.43%), ab- median score of Glasgow-Blatchford Score dominal pain (44.14%) and fatigue (6.08%) was (4 - 10) and 63.0% of cases had Glas- Vital signs on admission: The mean heart rate was 91.9 (18.1) beats/min There was 25.7% of patients had tachycardia (heartbeat > 100 beats/min) 8.6% of the patients had low blood pressure (defined as systolic pressure lower than 90 mmHg or diastolic pressure gow-Blatchford Score ≥ About the Rockall score, a median Rockall score was (1 - 5) with 71.2% in the high-risk group (Rockall score > 3) Pre-endoscopy Rockall score reported low percentage (57.4%) in the high-risk group and the median score was (0- 2) lower than 60 mmHg) Table Risk Score of upper gastrointestinal bleeding Rockall score (Complete) (Median (Interquantile)) (2 - 5) < points (low risk group) 26,1% - points (moderate risk group) 73,3% > points (high risk group) 0,6% Glasgow-Blatchford Score (Median (Interquantile)) (4 - 10) Glasgow-Blatchford Score < 37,0% Glasgow-Blatchford Score ≥ (50% of patients need intervention) 63,0% Upper endoscopy findings and interventions Time of endoscopy: 93,5% of cases received endoscopy within 24 hours of admission in which 34.7% had endoscopy before admission (outpatient indication before admission) The durationfrom hospital admission to upper endoscopy had a median value of (2.5 - 11) hours Bleeding etiologies: Table presents the causes of non-variceal upper gastrointestinal bleeding, which duodenal bulb ulcer and gastric ulcer were predominant causes (63.5% and 23.2%), respectively In total, 33.78% of patients had active bleeding at the site of lesions Among 387 patients with bleeding ulcers, 61.5% was in high-risk stigmata group (rebleeding rate from 2255%) (Table 3) JMR 116 E3 (7) - 2018 21 JOURNAL OF MEDICAL RESEARCH Table Causes of non-variceal upper gastrointestinal bleeding Esophagus Esophageal ulcer 8/444 (1.8%) Mallory Weiss 25/444 (5.6%) Gastric ulcer Stomach 103/444 (23.2%) Dieulafoy lesion 8/444 (1.8%) Malignancy 16/444 (3.6%) Duodenal bulb ulcer Duodenum 282/444 (63.7%) Anastomosis ulcer 15/444 (3.4%) Unidentified injury 26/444 (5.9%) Table Characteristics of peptic ulcers Forrest Classification Patients with ulcers (n = 387) High-risk stigmata group Forrest IA (Spurting bleeding) 9/387 (2.3%) Forrest IB (Oozing bleeding) 109/387 (28.2%) Forrest IIA (Non-bleeding visible vessel) Forrest IIB (Adherent clot) 62/387 (16%) 59/387 (15.2%) Low-risk stigmata group Forrest IIC (Flat spot) Forrest III (Clean base) 20/387 (5.2%) 128/387 (33.1%) Endoscopy interventions: Almost half of the cases of non-variceal upper gastrointestinal bleeding required endoscopic intervention (48.4%) with the successful rate of 99.1% patients failed interventions due to restlessness Nearly 70% of the intervened patients required only mono therapy - epinephrine injection (69.77%) The second most common method was the combination of epinephrine injection and endoscopic clips Other mono the rapies such as endoscopic clips, APC or combination methods only accounted for a small percentage with less than 10% After intervention, there was a significant decrease of heartbeats before endoscopy and after endoscopy (p < 0.001) In the group with hypotension before endoscopy interventions, there was a significant increase in the mean of heart pressure (p < 0.001) 22 JMR 116 E3 (7) - 2018 JOURNAL OF MEDICAL RESEARCH Table Changes of vital signs after endoscopic intervention Pre-endoscopy Post-endoscopy p Heart rate (beats/min) 94.0 (19.2) 88.2 (11.7) < 0.001* MAP in pre-endoscopy hypotension group (mmHg) 62.4 (9.0) 81.2 (10.3) < 0.001* MAP: Mean arterial pressure; *: Statistically significant Recurrent bleeding rate and related factors There were 20 cases (4.5%) who had recurrent bleeding during hospital stay There was no difference in recurrent bleeding rate between intervention group and non-intervention group (p = 0.755) In the intervention group, the difference of in-hospital bleeding rates between mono therapy and combined therapies groups was not statistically significant (p = 0.088) Among 272 contacted patients by phone number, there were only patients (1.1%) who had recurrent bleeding within 30 days after discharge Table In-hospital recurrent bleeding and related factors Recurrent Univariate logistic Multivariate logistic bleeding regression regression Factors Yes (20) No (424) OR (95% CI) p Age ≤ 60 16 302 0.62 (0.20, 1.89) 0.395 > 60 122 Gender Female 143 4.58 (1.05, 20.01) 0.027* Male 18 281 2.53 (0.73, 8.79) 0.13 1.58 (0.61, 4.05) 0.342 OR (95% CI) P 5.60 (1.27, 24.76) 0.023* Demographic Patient’s medical history Comorbid diseases No 131 Yes 17 293 13 316 108 Past UGIB history of No Yes JMR 116 E3 (7) - 2018 23 JOURNAL OF MEDICAL RESEARCH Recurrent bleeding Univariate logistic regression Multivariate logistic regression Factors Yes (20) No (424) OR (95% CI) p 0.95 (0.34, 2.68) 0.925 2.87 (0.91, 9.06) 0.061 OR (95% CI) P 4.83 (1.80, 12.94) 0.002* Hemodynamic status Tachycardia Heartbeat < 100 15 314 Heartbeat > 100 110 Low blood pressure No 16 390 Yes 34 No intervention 10 219 Singular intervention 210 0.63 (0.22, 1.75) 0.368 Combined intervention 35 2.50 (0.74, 8.42) 0.127 Endoscopy Intervention Red blood cell transfusion in hospital No Yes 14 272 152 15 GBS ≤6 >6 4.18 (1.57, 11.09) 0.002* 120 295 1.22 (0.43, 3.43) 0.705 158 1.46 (0.55, 3.87) 0.448 14 253 Risk score Rockall score (complete)