Initial assessment

An initial assessment was undertaken with the aim of stabilising John’s condition and preventing further deterioration. The initial assessment also gathered basic clinical information which supports formulation of a differential diagnosis. Due to the high incidence of mortality associated with GI bleeding, early intervention is crucial to stabilise patients and identify a cause.

In John’s case, a rapid airway, breathing, circulation, disability and exposure (ABCDE) assessment was carried out in the resuscitation room to ensure that life-threatening issues such as hypoxia, hypotension or reduced consciousness were identified and addressed immediately. John was talking in full sentences when required, which indicated that his airway was clear. His breathing assessment revealed a respiratory rate of 20 breaths per minute, 95% oxygen saturation on air, no evidence of cyanosis and equal chest sounds with optimal air entry. Critically unwell patients should be commenced on 15L oxygen through a non-rebreather mask while an assessment is conducted (Elsayed et al 2017, Japp and Robertson 2018, Chapman et al 2019). Although there were no concerns about John’s airway or respiratory system, he was started on 15L oxygen to prevent deterioration.

Repeat manual radial pulse taken to confirm John’s peripheral perfusion identified that he had a rapid but weak pulse, cardiac monitoring confirmed a heart rate of 118 beats per minute and there were normal heart sounds. Central capillary refill time, which is a quick assessment of a patient’s peripheral perfusion if cardiac output is suspended or reduced (hypovolaemia), was three seconds indicating a slight delay in perfusion. John was also slightly hypotensive with a blood pressure (BP) of 105/60mmHg, which raised concern that his circulatory system was affected by an underlying condition.

Tachycardia and hypotension in patients with upper GP bleeding are indicative of a 15% blood volume loss (Elsayed 2017). Physiological changes in heart rate and BP are compensatory responses aimed at maintaining cardiac output and tissue perfusion, and if untreated can result in hypovolaemic shock (Huether et al 2020). Tachycardia with BP lower than 120mmHg systolic, particularly in patients aged over 60 years, indicates hypovolemic shock (Leppert and Kelly 2021) which is associated with increased mortality (Stanley and Laine 2019). John’s tachycardia and hypotension suggested he was decompensating and was in a shocked state.

Intravenous (IV) access was gained using two large bore cannulae, one in each antecubital fossa. Stanley and Laine (2019) suggested that central venous access is preferable for critically unwell patients. However, Siau et al (2017) noted that intraosseous access could be more useful due to its potential for rapid insertion, while the Resuscitation Council UK (2021) advocates using intraosseous access in critically unwell patients when IV access has failed.

In John’s case, a full set of bloods were collected at the time of cannula insertion, including full blood count, urea and electrolyte, clotting factors and group and save and cross-match (Ramrakha et al 2016, Japp and Robertson 2018). Venous blood gas was taken to enable rapid identification of electrolyte imbalance, which can occur in patients with significant blood loss, and to determine haemoglobin (Hb) levels (Chapman et al 2019). These measures supported planning of the initial treatment and management.

At this point in John’s assessment, GI bleeding was the most likely differential diagnosis, however a secondary assessment (see below) was used to confirm diagnosis.

Initial treatment and management

John required IV fluids due to his shocked state, therefore an initial bolus of 1L sodium chloride 0.9% (a crystalloid solution) was administered. Early fluid resuscitation of haemodynamically compromised patients with upper GI bleed can decrease mortality significantly (Baradarian et al 2004), but there is debate in the literature about whether colloid or crystalloid solutions should be used (Perel et al 2018). Crystalloids comprise aqueous solutions of mineral salts (electrolytes) and other small, water-soluble molecules and are isotonic to human plasma and close to the composition of blood in the circulatory system. Colloids are larger molecule, gelatinous solutions with a high osmotic pressure in the blood, which means that the molecules remain in the intravascular space for longer, potentially affecting the platelet count, reducing clotting time and increasing the risk of pulmonary and peripheral oedema (Smith 2017). The multi-society consensus care bundle for early clinical management of acute upper GI bleeding, led by the British Society of Gastroenterology (BSG) (Siau et al 2020), suggests using crystalloid solution initially, however it also states that there are no trials involving this patient group comparing the different fluids, only studies of the rate at which fluid is administered.

Elsayed et al (2017) suggested using isotonic crystalloids such as Hartmann’s solution rather than using colloids if patients do not respond to other types of crystalloids, due to the associated risks mentioned above. Stanley and Laine (2019) noted that although there is no evidence of a difference in mortality rates when patients are treated with either colloid or crystalloids, crystalloids are safer for this patient group and colloids should only be used when crystalloids are ineffective, for example when patients are in profound shock. The European Society of Gastrointestinal Endoscopy (ESGE) guidelines for endoscopic diagnosis and management of nonvariceal upper gastrointestinal haemorrhage (Gralnek et al 2021) advocate administration of 500mL crystalloid as an initial bolus at a rate of less than 15 minutes, while the BSG consensus care bundle (Siau 2020) recommends adherence to the NICE (2016) guideline on management of acute upper GI bleeding in people aged over 16 years. However, the NICE (2016) guideline is ambiguous about fluid type, stating that either colloids or crystalloids can be used and does not recommend a timeframe for administration.

John’s Hb on venous blood gas was 83g/L and he was not actively bleeding. This was significant in relation to activating the trust’s local massive haemorrhage protocol, which only requires activation if a patient is actively bleeding, there is a loss of more than 150ml/min of blood, the patient is in clinical shock or has had 4L blood loss in 24 hours. The protocol also states that a patient’s Hb should be in a target range of 80-100 g/L. Because John did not meet these criteria, the massive haemorrhage protocol was not activated.

The point at which to activate a massive haemorrhage protocol varies, with some based on the amount of blood loss and others based on the patient’s Hb level. For example, the Scottish Intercollegiate Guidelines (2008) suggest that transfusion with red blood cells should be initiated when there is a 30% loss in a patient’s circulating volume, while NICE (2016) recommends commencing transfusion when a patient’s Hb falls below 70g/L. The BSG consensus care bundle (Siau 2020) and the ESGE guidelines (Gralneck et al 2021) recommend that if a patient is haemodynamically unstable and their Hb is below 70g/L, then infusion of red blood cells should be considered. There are other factors to take into account when making this decision, for example red blood cell transfusion in patients with cardiovascular disease could induce acute coronary syndrome (Docherty et al 2016). However, the ESGE (Gralnek et al 2021) and the BSG (Siau et al 2020) state that the risks of using red blood cell replacement are associated with the speed of infusion rather than volume and threshold for initiation.

Following management of John’s circulatory system, a reassessment of his airway, breathing and circulation was undertaken which identified that his condition had stabilised. Airway and breathing were unchanged, and although John had a heart rate of 115 beats per minute, his systolic BP had increased to 110mmHg.

The ABCDE assessment was completed by observing and recording alterations in John’s neurological function (disability) and exposure. John was alert and oriented throughout which suggested his circulatory volume was sufficient to support cerebral perfusion and maintain brain function. Temperature is also important because platelet formation can be affected and blood clotting reduced in hypothermic patients (temperature below 35.0°C) (Wyatt et al 2016). John’s temperature was 36.2°C.

Secondary assessment

A secondary assessment of John’s condition was conducted using a medical model of history taking, with the aim of gathering further information to confirm the diagnosis of upper GI bleeding, or to form differential diagnoses. The results of the secondary assessment are shown in Table 1 (Bickley 2017, Thomas and Monaghan 2019).

John’s history supported the working diagnosis of upper GI bleeding. Differential diagnoses included oesophageal varices, however this was dismissed because John’s haematemesis was ‘coffee ground’ in colour and patients with variceal bleeds often vomit large amounts of fresh red blood (Ramrakha et al 2016, Wyatt et al 2016, Japp and Robertson 2018). A Mallory-Weiss tear was also considered unlikely because patients with this condition usually present with streaks of blood in vomit and it tends to occur following prolonged retching, possibly caused by gastritis (Thomas and Monaghan 2019). Coagulation disorders such as thrombocytopenia or the use of anticoagulant medicines were not identified in John’s history or examination (Wyatt et al 2016).

John had some ‘red flags’ in his medical history, including previous gastric ulcer, previous excess alcohol use and regular use of nonsteroidal anti-inflammatory drug (NSAIDs), namely naproxen. NSAIDs are cyclooxygenase-1 (COX1) and COX2 drugs that reduce prostaglandin secretion and its cytoprotective effect on the mucosa, which in turn increases the susceptibility of the lining of the stomach to ulceration (Drini 2017). The combination of excessive use of alcohol, which is known to increase irritability and inflammation of the stomach by increasing acid production, and taking NSAIDs increases the risk of upper GI bleeding by around 30% (Japp and Robertson 2018). In John’s case, the evidence reinforced a diagnosis of upper GI bleeding.

To complete a full assessment, a rectal examination was required to identify evidence of melaena (Chapman et al 2019). A rectal examination can support diagnosis of GI bleeding and identify a risk of mortality as the presence of faecal occult blood and its colour can determine if the bleed is active or if the blood is residual (Wyatt et al 2016). John’s rectal examination identified dark tarry stools which reinforced the diagnosis of upper GI bleeding.

Endoscopy

NICE (2016) guidelines recommend that the primary intervention for patients with GI bleeding should be endoscopy conducted as quickly as possible to confirm diagnosis and instigate rapid intervention. ESGE and BGS guidance, however, suggests that when patients are stable the risk of re-bleeding – termed ‘stigmata’ and noted by clinical features such as clots, visible blood vessels or spurting vessels – should be assessed before an endoscopy is carried out (Siau et al 2020, Gralnek et al 2021).

This assessment can be conducted using predictive scoring tools such as the Glasgow-Blatchford scoring tool (Blatchford et al 1997) (Table 2), or the Rockall scoring tool (Rockall et al 1995) (Table 3), which assist in identifying the need for endoscopy and the timeframe for undertaking it. The Glasgow-Blatchford scoring tool predicts the need for rapid endoscopic intervention and mortality risk, while the Rockall scoring tool predicts mortality risk only (Chapman et al 2019, Stanley and Laine 2019). The Rockall scoring tool uses two mortality-rate prediction tools – pre-endoscopy and post-endoscopy. The pre-endoscopy score assists identification of which patients require endoscopy to prevent mortality. This was used to identify whether John required in-patient endoscopy or if his score was low enough for an outpatient endoscopy to be arranged.

The Glasgow-Blatchford and Rockall scoring tools are recommended by NICE (2016), the ESGE (Gralnek et al 2021) and the BSG (Siau et al 2020) for use with all patients with upper GI bleeding. Both tools were used to assess John, using a smartphone application MDcalc. The Rockall tool scores are calculated by adding the points awarded for each question which are then stratified to give a risk-specific percentage.

The score range is between 0-12. Scores below 3 are considered low risk and scores above 3 are considered high risk (Rockall et al 1995). John’s score for predicted re-bleeding was calculated by adding the points awarded for age, shock index and co-morbidity (total=4) which when stratified indicated a re-bleed risk of 24% and mortality risk of 11%. The Glasgow-Blatchford tool, which gives a score range of 0 to 23, with scores of 0 considered low risk and scores above 0 considered high risk for upper GI bleeding, indicated John had a high risk of re-bleeding and therefore required endoscopy (Blatchford et al 1997). The Glasgow-Blatchford scoring tool has a sensitivity of 98.6%, a specificity of 34.6%, a negative prediction value of 56%, and positive values of 96.6% for prediction of risk at initial assessment and mortality, which reinforces the tool’s efficacy (Stanley et al 2017).

Treatment plan

A treatment plan for John was formed based on his Rockall and Glasgow-Blatchford scores (Rockall et al 1995, Blatchford et al 1997). Using a PPI was discussed. There is debate in the literature and guidelines on the use of PPI pre-endoscopy for patients with GI bleeding. The ESGE guidelines (Gralnek et al 2021) suggest using a high dose PPI before undertaking endoscopy in patients like John with established peptic ulcer disease. However, this contradicts NICE (2016) guidelines which recommend withholding PPI until after endoscopy because PPI increases the gastric pH and inhibits fibrinolysis. The ESGE guidelines (Granlek et al 2021) suggest that PPI increases the stability of any ulceration, however NICE (2016) argues that this stability may distort endoscopic intervention and notes that PPI is not licenced for this use.

A Cochrane review of PPI treatment initiated before endoscopic diagnosis in patients with upper GI bleeding (Sreedharan et al 2010) identified that PPIs can reduce the prevalence of high-risk endoscopy stigmata in peptic ulcer-related bleeding, but showed little evidence to support their use before endoscopy in lower-risk patients. This is supported by the ESGE guidelines (Granlek et al 2021), which suggest that pre-endoscopy PPI does not affect patient outcomes in relation to future bleeding and mortality. These guidelines also suggest that primary use of PPI could be more cost effective than undertaking endoscopy in patients with known peptic ulcer disease (Gralnek et al 2021). As the NICE (2016) guidelines do not support the use of PPIs they were not administered to John and an inpatient endoscopy to assess his GI bleeding was arranged.

Some studies and systematic reviews have examined the use of tranexamic acid (TXA) for treatment of upper GI bleeding (Brenner et al 2019, Burke et al 2021). TXA is a antifibrinolytic agent that inhibits the breakdown of fibrin by plasmin thus reducing the need for blood transfusion (Brenner et al 2019, Burke et al 2021) and has been used to manage major bleeding in trauma and post-partum patients for several years (Brenner et al 2019). Brenner et al (2019) conducted the HALT-IT trial, an international, randomised, double-blind, placebo-controlled trial of TXA in 12,000 adults with acute upper or lower GI bleeding, and concluded that although there is evidence for its use in active upper GI bleeding there is little evidence that it affects mortality rates. NICE (2016), ESGE (Gralnek et al 2020) and BSG (Siau et al 2020) acknowledge this and therefore do not include TXA in their guidelines. John was not administered TXA.

Overview

John’s assessment was completed quicky to stabilise his condition and consider differential diagnoses. Initially John was shocked and haemodynamically unstable and using the ABCDE assessment supported management of potentially life-threatening parameters and prevention of deterioration.

John’s Hb was above the threshold for transfusion therefore sodium chloride 0.9% rather than red blood cells was administered. History taking identified that John had peptic ulcer disease and had had previous GI bleeding. Using a PPI was discussed, however the NICE (2016) guidelines do not support this therefore endoscopy and admission to hospital were arranged based on the Rockall and Glasgow-Blatchford prediction tools (Rockall et al 1995, Blatchford et al 1997). John was admitted to hospital and had an inpatient endoscopy the following day which identified that the source of bleeding was most likely stomach ulceration but that the ulcer was no longer actively bleeding and was already healing. John’s condition continued to improve and he was discharged from hospital 48 hours after his admission with an arrangement for outpatient follow up with the gastroenterology team.