Risk factors and outcomes of upper gastrointestinal bleeding in hospitalized patients in a tertiary care hospital

Objective: To determine different aetiologies and outcomes of upper gastrointestinal bleeding in hospitalised patients. Method: The retrospective cohort study was conducted at the Aga Khan University Hospital, Karachi, and comprised data from December 2019 to April 2021 related to adult patients of either gender with non-gastrointestinal illnesses who developed bleeding at least 24 hours after admission. Data was reviewed for clinical characteristics, cause of bleeding and clinical outcome. Data was analysed using SPSS 23. Results: Among 47,337 hospitalised patients, upper gastrointestinal bleeding was identified in 147(0.3%); 98 (66.7%) males and 49 (33.3%) females. The overall mean age was 62.73±14.81 years (range 20-95 years). Of the total, 125(85%) presented with overt bleeding and 22(15%) with a drop in haemoglobin level. There were 34(23%) patients on aspirin, 36(24%) on dual anti-platelets, 41(28%) on therapeutic anticoagulation, and 81(55%) on prophylactic anticoagulation. There were 7(5%) patients having a history of non-steroidal anti-inflammatory drugs (NSAIDs), and 12(8%) were on steroids. In terms of associated medical conditions, pneumonia, stroke, and acute coronary syndrome were commonly seen with frequency of 29.9%, 8.1% and 7.4% respectively. Overall, 36(24.4%) patients underwent endoscopy, 8(5.4%) had therapeutic measures to control bleeding, 14(9.5%) had bleeding for >48 hours, 89(60.5%) were stepped up to special care. Mortality was seen in 36(24.5%) cases. Conclusion: Hospital�acquired gastrointestinal bleeding was found to be uncommon, and there were several risk factors for such bleeding events.


Introduction
Upper gastrointestinal (GI) bleeding, defined as bleeding arising from the oesophagus, stomach or duodenum 1 , is an important complication in hospitalised patients.These bleeding events significantly increase mortality in patients during hospital stay.The incidence of hospitalacquired GI bleeding varies from 0.17% to 5% 2 .The common causes of such bleeding are ulceration and stress-related mucosal disease 3,4 .Clinically important stress ulcers occur in ∼1% of patients, but its severity and outcome are important issues 5 .
The reason behind such ulcers is impaired blood flow to gastric mucosa due to haemodynamic instability caused by either systemic causes, like hypotension, or local causes, like low visceral blood flow in mechanically ventilated patients.Causes also include anticoagulation therapy, renal insufficiency, burns, neurological insults like stroke and postoperative stress 6 .GI bleeding in noncritical patients in hospital is uncommon and carries a low risk of morbidity and mortality.
Stress ulcer prophylaxis (SUP) are acid-suppressive medications, such as histamine-2 (H2) receptor antagonists, sucralfate, or proton pump inhibitors (PPIs) 7 .Studies have demonstrated that these medications reduce the incidence of clinically significant nosocomial GI bleeding in hospitalised patients, both in and outside of the intensive care unit (ICU), with relative risk reduction ranging from 29% to 61% 2 .SUP is recommended in international guidelines and considered a standard of care in the ICU setting in critically ill patients, but guidelines recommend against its routine use in patients outside of the ICU 8 .However, there may be certain subsets of non-critically ill patients in whom the risk of nosocomial GI bleeding is high enough that prophylactic use of acid-suppressive medication may be warranted 9 .Indications for initiating SUP vary considerably.These inconsistencies in the initiation of SUP may be explained by sparse research data and variable recommendations.They are prescribed in an indiscriminate fashion in patients admitted to hospitals, suggesting that physician preference dictates the practice.
Studies in the ICU setting investigating risk factors for nosocomial GI bleeding have consistently identified mechanical ventilation and coagulopathy as significant independent predictors 10 , both of which confer high enough risk to warrant prophylactic acid-suppressive medication in this particular patient population 11 .Whether similar risk factors exist in non-critically ill patients has not been well examined.Such information is crucial to aid clinicians and to provide a better understanding regarding more appropriate use of acidsuppressive medication.PPIs remain the mainstay as the chemoprophylaxis, but the cost and adverse effect of nosocomial diarrhoea remain a limitation 12,13 .Patients with sufficiently low risk of bleeding may not need prophylaxis.To the best of our knowledge, a systematic study on GI bleeding occurring in hospitalised patients admitted for non-GI disorders has not been conducted.The current study was planned to fill the gap by determining different aetiologies and outcomes of upper GI bleeding in hospitalised patients.

Materials and Methods
The retrospective cohort study was conducted at the Aga Khan University Hospital (AKUH), Karachi, and comprised patient data from December 2019 to April 2021.After approval from the institutional ethics review committee, data was retrieved related to adult patients of either gender with non-GI illnesses who developed upper GI bleeding at least 24 hours after admission.Data was excluded for patients aged <18 years, patients with documented GI bleeding complaint at the time of the index admission, patients with lower GI bleed, and those presenting with haematochezia during hospitalisation.
Demographic data was extracted by chart review.Data on comorbid conditions, including hypertension (HTN), diabetes mellitus (DM), dyslipidaemia, arthritis, liver cirrhosis, congestive heart failure (CHF), coronary artery disease (CAD), cerebrovascular accidents (CVAs) and chronic kidney disease (CKD), was also recorded.Prior medication usage history, presenting complaint, and non-GI diagnosis from the initial hospitalisation were noted.Daily progress notes were reviewed to determine the event of upper GI bleeding, significant laboratory investigations in the form of haemoglobin (Hb) level, activated prothrombin time (PT), platelet (Plt), partial thromboplastin time (PTT), medical management, endoscopic evaluation, need for advanced techniques to control bleeding (therapeutic endoscopy or angiography with embolisation) were also noted.The end point of the study was to identify mortality rate in the targetted patient population.Medical records were reviewed to identify patients who had overt GI bleeding or a significant drop in Hb level during hospitalisation.Overt manifestations of bleeding were in the form of haematemesis or melena.Patients with drop in Hb level had nasogastric tube lavage positive for coffee-ground aspirate or clotted or altered blood.
Upper GI endoscopy had been done by trained personnel in the Gastroenterology Unit.Endoscopic findings were categorised by the nature of the visualised lesions, and if multiple lesions were noted, the endoscope technician's impression of the most likely bleeding site was used to define the source of bleeding.Forrest classification was used to evaluate the ulcers 14 .Findings include spurting haemorrhage (Forrest Ia), oozing haemorrhage (Forrest Ib), a nonbleeding visible vessel (Forrest IIa), an adherent clot (Forrest IIb), a pigmented spot on ulcer base (Forrest IIc) and a clean ulcer base (Forrest III).
Data was analysed using SPSS 23.Data was presented as frequencies and percentages for categorical variables and as mean ± standard deviation (SD) for continuous variables.
Majority of the patients 144(97.9%)were on intravenous (IV) omeprazole 40mg once daily as prophylaxis since the time of admission.After the development of upper GI bleed, 9(6.1%) patients remained on once daily dosage, while 39(26.5%)were managed with 40mg 12 hourly and 99(67.3%)were given infusion at 8mg per hour.
Of the patients with ulcerated mass, 1(50%) patient had it infiltrating into the duodenum which, on imaging, was found to be peri-ampullary mass.Besides, 1(50%) patient had a gastric mass which was later found on biopsy to gastrointestinal stromal tumour (GIST).

Discussion
The current study aimed at identifying the frequency and aetiology of nosocomial GI bleeding in patients with non-GI illness on admission.It looked at various parameters, including age, gender, comorbidities, risk factors, primary complaint, presentation of bleeding, primary care provided, disease severity, treatment modalities, duration of bleeding and patient outcome.
Mean age of the patients who developed nosocomial GI bleeding was 62.78 years, and males (66.7%) were more likely to develop nosocomial GI bleeding compared to females (33.3%).The findings were comparable to a study conducted by Herzig et al., who found that the incidence of GI bleeding was greater in those aged 60 and above, and the male gender was more likely to be affected 9 .
The most significant primary complaints in the current study were dyspnoea (33.3%), fever (23.1%) and drowsiness (14.3%).These primary complaints can be explained by the fact that approximately one-fourth of the participants were seeking care for pneumonia mainly caused by COVID-19.Occurrence of GI bleeding in such patients has been reported by recent studies 15 .
The leading presentation was overt GI bleed; haematemesis (40.1%) and melena (33.3%).Since in the current study only nosocomial bleeding was included, which is acute, a decrease in Hb level due to occult bleeding was observed in fewer patients.
In terms of primary care service provided, majority of the patients were treated by the department of internal medicine (75%), followed by oncology (6.8 %) and cardiology (5.4%).
Primary disease severity in the study warranted SCU admissions in 60.5% of patients, ward admission in 32.7%, and ICU admissions in 6.8% of patients.This goes to show that disease severity did not predetermine the likelihood of a nosocomial GI bleed in the patients.
In patients who developed ulcers, gastric ulcers were more common than duodenal.The least common were oesophageal ulcers.Among those with gastric and duodenal ulcers, the highest percentage had Forrest class III ulcers.
The most common treatment modality offered to the
Figure: Severity of the primary disease.ICU: Intensive care unit, SCU: Special care unit.