Profile and outcome of patients with upper gastrointestinal bleeding presenting to urban emergency departments of tertiary hospitals in Tanzania

Upper Gastrointestinal bleeding (UGIB) is a medico-surgical emergency. Although there has been a global decline in the mortality associated with UGIB, the incidence and mortality associated with GI bleeding remains high in limited income countries. In the United States, UGIB accounts for 300,000 admissions per year with approximately 5% mortality rate [1], whereas in several studies from sub-Saharan Africa, mortality ranges from 6 to 30% [2‐5]. It is not clear if the higher mortality seen in low and middle income countries (LIMC’s) is due to patient demographics, severity on presentation, etiology or compliance with care standards. Esophageal varices has been implicated as the most common cause of UGIB in several African studies [2‐11]. Rather than being due to alcoholic liver disease as in high income countries (HIC’s), varices in sub-Saharan Africa result from Schistosoma-related portal hypertension [5, 8, 12]. This is in contrast to HIC where erosive gastritis has been commonly implicated [1, 13, 14]. Other potential contributors to the higher mortality may be the severity of disease presentation, as there is little primary care in these settings, and many patients seek care very late in their disease.

Another potential contributor is failure to treat patients in accordance with management guidelines. Proton pump inhibitor (PPI) for those with suspected Non-variceal UGIB (NVUGIB), somatostatin analogues such as octreotide and antibiotics in suspected cases of variceal UGIB (VUGIB) and in those with clinical suspicion of liver disease, timely blood transfusion and early use of endoscopy may not occur in these settings due to lack of appropriate specialists, lack of resources or supply chain issues [15‐17]. A knowledge gap on the current UGIB management guidelines and recommendations amongst the health care providers may also contribute to inadequate management of UGIB cases. Prior studies in Tanzania studies demonstrate lack of appropriate care offered to patients with UGIB. Nearly half of the patients in one the studies did not receive endoscopic evaluation and treatment [5] whereas 47.1% of the patients who required blood transfusion, did not receive blood transfusion in another study [10].

There is little information on the presentation, etiology and management of patients presenting to EDs and this is particularly true for LIMCS, where emergency medicine is a new specialty. The presence of full capacity emergency departments at Muhimbili National Hospital (MNH) and MUHAS Academic Medical Centre (MAMC) in Dar es Salaam have provided the opportunity for early stabilization and management of patients presenting with UGIB. The aim of this study was to describe the patient profile, clinical severity and outcomes of the patients who present with UGIB to the ED.

During the period of study, there were 31,987 patient visits (30,800 at EMD-MNH and 1187 at EMD-MAMC). From these we identified 123 (0.4%) patients with upper GI bleeding, all of whom were eligible and consented to be in the study (Fig. 1). Median age was 42 [IQR 32–64] years and 87 (70.7%) were male; 87 (70.7%) were married 44 (35.8%) were self-employed. The majority of patients 77 (62.6%) had been previously seen at a lower capacity health facility and transferred to one of the two study sites. Most patients (99, 80.5%) were uninsured. (Table 3).

UGIB in this limited income setting was associated with a high mortality rate (18.7%). One third of all deaths occurred in the first 24 h. This may be a reflection of the high disease severity seen among these patients, but also could reflect that patients did not receive guideline-adherent care. In particular, while endoscopic evaluation at any point during the hospital stay was found to be protective factor against mortality, only 37.4% of the patients received UGI endoscopy and only a handful received early UGI endoscopy i.e. within 24 h of ED presentation.

Compared with studies from HIC, patients in our study were of younger age, with the median age being 42 years [1, 13, 14, 22]. We observed that age above 40 years was an independent predictor of mortality. This has also been noted in prior studies both in HIC and LMIC [1, 5, 13, 14].

The mortality observed in our study is much higher compared to developed countries [1, 13, 14]. Factors contributing to the high observed mortality rate include; higher disease severity, disposition status and gaps in management of these patients.

Our patients had a higher disease severity in comparison to those in HIC. A large proportion of our patients had a moderate to severe clinical Rockall score (i.e. score of 3–4 and > 4) as compared to a Chinese study carried out in the ED, where most (50.6%) had a low Rockall score of 0–2 [23]. Clinical Rockall score ≥ 4 was found to be an independent predictor of higher mortality. This was also observed in several other studies (18,22). Similarly, the Glasgow-Blatchford among patients in our study was much higher compared to a British study carried out between the years 2008–2009, where the majority of patients had a GBS score ≤ 2 [24]. Severe anemia (Hb < 8 g/dl) was observed in over a half of our patients while in studies conducted in HIC, less than one-third of the patients had a Hb count suggestive of severe anemia [13, 14, 25].

As part of the ED management, majority of the patients received intravenous fluids and proton-pump inhibitor (PPI). Although international guidelines such as the European Society of Gastrointestinal Endoscopy Guidelines (ESGE) suggest use of high dose PPI followed by an hourly infusion, 19 (59.4%) patients with a high clinical Rockall score of ≥ 4 received an initial high dose PPI. It is worthy to note that none of the patients received the hourly PPI infusion as recommended in the guidelines. A knowledge gap concerning the current recommendations amongst the medical care providers maybe the plausible to this shortcoming. As per the departmental protocol, PPI should be given to all patients presenting with UGIB. This may reason out to why PPI was the first choice of drug prior to undergoing endoscopy. A small proportion of the patients received octreotide, although only half of them had indication for its use based on historical features (such esophageal varices and liver diseases) and many of those with such a history did not receive this medication. The utilization was probably limited by the availability and the cost of the drug in our setting. But the fact that some received it inappropriately suggests a knowledge gap. Antibiotic administration especially in patients with prior history of liver disease and esophageal varices was low. The National Institute of Health and Care Excellence (NICE) in the UK, emphasizes prophylactic antibiotic use in patients with suspected or confirmed variceal bleed or a history suggestive of liver disease including cirrhosis [15]. Provision of antibiotics was associated with increased incidence of mortality only in univariate analysis, probably due to confounding with the presence of liver disease which is associated with a poor outcome. Finally, despite the high proportion of patients with severe anemia, less than half of the patients received transfusion. A previous study carried out at MNH, revealed difficulties in obtaining blood for transfusion, including a general unavailability of blood, scarcity of un-cross matched blood at the blood bank along with delays in obtaining blood [26]. It is worth noting that ED management of UGIB is guided by the presence of departmental protocol for various conditions including UGIB. As such, provision of certain management may be guided by the protocol.

Only a small proportion of patients were admitted to ICU. In view of the higher disease severity in our study, a larger number of patients would be expected to be admitted into the ICU. Admission to the general ward implies that these patients were unlikely to receive the necessary aggressive care for their severity of illness; thus contributing to a high mortality rate observed. In a systematic review by Chiu et al it was noted that the majority of patients with a higher severity score were admitted to the ICU [27]. ICU admission for such high-risk patients has further been recommended in couple of international guidelines [15, 16]. Although the reasons for low levels of ICU admission was not determined in this study, it is presumed to be due to the limited number of ICU beds compared with the patient volume and acuity that is seen at these national referral hospitals.

UGI Endoscopic evaluation and treatment is a major cornerstone in the management of UGIB and early UGI endoscopy is associated with reduced mortality and hospital length of stay [28, 29]. Just over a third of patients (37.4%) in our study received UGI endoscopy at any time during the hospital stay; moreover only 17.5% received early endoscopy (within 24 h) and none received emergent endoscopy. The commonly used methods of hemostatic control of bleeders in our endoscopy units were hemoclipping and argon plasma coagulation (APC). Esophageal banding is usually reserved for variceal bleeders in our study setting. It is also worth noting that esophageal binding kits are not readily available in the hospital stores. Studies in HIC show that a larger proportion of their patients receive UGI endoscopy, with most receiving early UGI endoscopies [13, 14]. While literature suggests endoscopy should be done within 24 h, we found that endoscopy at any time was associated with lower mortality. This finding was statically significant in the univariate analysis but not in the multivariate regression analysis. This may be observed due to the fact that the study was not powered to evaluate the impact of endoscopy to mortality. Further studies with a larger sample size may be useful to evaluate the impact of UGI endoscopy on mortality.

This study was conducted at two high capacity, referral EDs and so the patient population and outcomes could be different at smaller, or lower capacity facilities. However, because these EDs receive referrals from all over the country, the patients sampled likely provide a wide representation of the Tanzanian population.

Investigations were ordered at the discretion of the physician, and thus not all patients received all tests. Some laboratory variables were not obtained for all the patients, thus this may have underestimated or overestimated the significance of these variables to the outcomes under study, including the scores such as Glasgow-Blatchford score.

The sample size for the study was estimated for our overall outcome of mortality, but not necessarily for risk factors analysis, and thus some factors not found to be significant may show statistical significance in a larger study.

The mortality rate for UGIB in our setting remains substantially higher than in non-African countries. This appears to be due to the higher severity of disease in our patients, lack of ICU care, and inadequate adherence to treatment guidelines for medication and endoscopy. Some efforts may have been met with lack of resources, however further studies are needed to assess the knowledge of providers on managing patients with UGIB and familiarity with the international recommendations. Endoscopic evaluation and treatment at any point during the hospital stay still remains an independent predictor of mortality. Efforts are needed to increase the number of patients receiving endoscopic evaluation and treatment in a timely fashion in LIMC.