Best Evidence Topics
  • Send this BET as an Email
  • Make a Comment on this BET

The use of a Sengstaken-blakemore tube in the management of Upper Gastintestinal haemorrhage in the Emergency Department

Three Part Question

In the Initial Emergency Management of [Adult patients with active Upper Gastrointestinal Haemorrhage secondary to Oesophageal varices and signs of haemodynamic instability] does [insertion of a Sengstaken-Blakemore tube] in addition to pharmacological treatment compared with pharmacological treatment alone] [reduce mortality and morbidity]

Clinical Scenario

An adult patient with a history of variceal bleeds presents to the Emergency Department with haematemesis associated with tachycardia and hypotension. Despite treatment with Terlipressin, Omeprazole, Tranexamic acid and blood transfusion, hypotension and tachycardia persist and the patient continues to experience haematemesis. There is approximately a thirty-minute delay before endoscopy can be undertaken. You wonder whether insertion of a Sengstaken-Blakemore tube may be beneficial in achieving haemostasis and improving haemodynamic stability.

Search Strategy

Medline (using EBSCOhost interface), 1966-01/2017 and the following search terms:
[“gastrointestinal haemorrhage” OR “gastrointestinal bleeding” OR haematemesis OR varices]
[“Sengstaken tube” OR “Sengstaken-Blakemore tube” OR “balloon tamponade” OR “Minnesota tube”]
English & Human
Publication type: All clinical trials, comparative study, controlled clinical trial, evaluation studies, meta-analysis, multicentre study, randomised controlled trial, twin study and validation study.

Google Scholar and the following keyword combinations:
each of gastrointestinal haemorrhage / gastrointestinal bleeding / haematemesis / varices
each of Sengstaken tube / balloon tamponade / Minnesota tube

Furthermore, the references of all relevant papers were scanned to identify any further pertinent studies.

Search Outcome

127 papers were identified in Medline, of which 5 were of sufficient quality and relevant to the question of this topic.
Google Scholar revealed overall 111 listings of which no additional relevant papers were identified
No further pertinent studies were found by scanning the reference lists of the relevant papers.
5 papers have been critically appraised

Relevant Paper(s)

Author, date and country Patient group Study type (level of evidence) Outcomes Key results Study Weaknesses
Pinto-Correia et al.
37 episodes of UGI bleeding secondary to varices in 33 patients 17 episodes randomised to vasopressin, 20 episodes randomised to balloon tamponade as initial intervention. In all patient episodes where the initial intervention failed to achieve haemorrhage control the alternative therapy was undertaken 2 – Individual randomised trialEfficacy of infusion of vasopressin and balloon tamponade in the initial management of bleedingVasopressin controlled haemorrhage in 65% of encounters (11/17). Remaining 6 cases, subsequent insertion of Sengstaken tube stopped haemorrhage. 2 patients in this group died from causes attributable to vasopressin. Tamponade as initial therapy controlled haemorrhage in 70% of episodes (14/26). In 5 of remaining encounters subsequent vasopressin was required. 1 remaining patient died from uncontrolled haemorrhage during procedure. In this group 2 further patients died due to complications of Sengstaken tube.Small numbers in each study group. No statistical analysis undertaken of results therefore unable to judge significant of data. Authors do discuss sample size calculation but appears to have been done after study commenced – number identified by this cited as rationale for stopping the study as unachievable. Suggests poor validity of initial study design. Vasopressin doses varied – no rationale or explanation given. Potential to significantly impact validity and reliability of results No consideration given to side effects/ complications beside death in either group. Potential impact on accuracy/ validity of findings. Group receiving initial therapy followed by balloon tamponade particularly relevant to three-part question but extremely small number
Fort, et al.
34 cirrhosis patients who experienced 47 Variceal bleeding episodes whilst on hepatology unit 23 patients randomised to nitro-glycerine and terlipressin as initial treatment and 24 randomised to receive Sengstaken tube 2 – Individual randomised trialHaemorrhage control and side effects achieved by medication versus balloon tamponade Bleeding controlled with terlipressin and nitro-glycerin in 78% of episodes (18/ 23). In remaining 5 cases 1 patient died of massive haemorrhage and 4 had subsequent effective insertion of Sengstaken tube. Sengstaken tube was effective in 79% of episodes (19/24). Of remaining 5 patients 1 died (cause not given) 1 treated with endoscopic sclerosis. .Last 3 were given terlipressin and nitroglycerin without effect, then treated surgically. Complications rates identified as low in both groups however not elaborated onSmall numbers recruited over time period. Cohorts failed significantly to reach numbers identified to achieve significance. This given as reason for stopping the study. Whilst authors recognise this in their methodology report it is not considered in the discussion of their results or conclusion. This may significantly affect the validity and reliability of conclusions for practice. Although study initially cited complications and side effects as key factor these areas are not clearly identified/ described/ analysed at any point in report. This may be argued as limiting the relevance of the study’s findings and detracting from the rigour of studies design Of limited relevance to three-part question as Sengstaken tube evaluated in isolation rather than in combination with medication
Holman and Rikkers
36 patients with endoscopically proven variceal haemorrhage 3 – Non-randomised controlled cohortNon-operative control of variceal haemorrhage. Requirement for operative procedures1 patient initially managed with Sengstaken tube and 4 with iced saline lavage. Of remaining 31 patients, all received infusion of pitressin which achieved haemorrhage control in 19 (61%). Of these 19 10 re-bled although 5 were resolved by repeat medication dose a further 3 had Sengstaken tube insertion (successful in 2 cases) and 2 required emergency surgery. In 12 cases, not initially responsive to pitressin 2 received surgery and 10 received Sengstaken tube. In all 14 patients received Sengstaken tubes it was effective in 78% (11) cases. Of three in which it was not successful 1 died and 3 required surgical interventionSmall number of subjects entered into study – no description given as to how recruited. Randomisation procedures not identified as having taken place or report that consent has been obtained. Report of results is disjointed and unclear causing difficulty in making direct comparisons between interventions. Report states statistical analysis has taken place but this is then not demonstrated in anyway. Conclusion are therefore not directly supported by data. No consideration of side effects or complications associated with interventions (other than rebleeding). Study does not identify methods that were used to follow up patients These significant methodological flaws question the validity of study design and any resulting data/ conclusions.
Panes et al.
151 episodes of oesophageal haemorrhage due to varices (confirmed by endoscopy) managed by placement of Sengstaken-Blakemore tube or Linton Nachlas tube 3 – Local non-random sampleEfficacy of balloon tamponade in achieving haemostasis at 24 hours and definitively Frequency and severity of complicationsHaemorrhage control achieved in 91.5% of episodes using Sengstaken tube at 24 hours. Linton Nachlas balloon achieved haemostasis at 24 hours in 88% Permanent haemostasis was achieved in 47.7% of cases 15 patients (10%) experienced pulmonary aspiration following tamponadeDefinition of severe complications unclear and poorly defined – may be affected by bias on part of researcher Data on complications between patients with both types of tube combined despite differences in methodology Experience of clinicians inserting tube not identified or discussed – potential to have significant bearing on rates of complication/ efficacy
Teres, et al.
108 patients admitted to the ITU with variceal bleeds on endoscopy Patients stratified by haemorrhage location (gastric/ oesophageal and degree of liver failure (low/ high). Each group then randomised resulting in 52 patients receiving balloon tamponade and 5 patients receiving pharmacological management 2 – Individual randomised trialHaemorrhage control and occurrence of side effects Comparison of early mortality (within 48 hours of enrolment) and late mortality (during hospital stay) Haemostasis efficacy 86.55 (95% CI: 72,4% to 94.4%) for tamponade & 66 % (95% CI:59.1% to 78.1%) for medication management. Early mortality was 8 patients in each interventional group. Late mortality was 16/52 in tamponade group and 17/56 in medication group. 1 patient in each group suffered potentially severe complications. Side effects requiring withdrawal of treatment occurred in 3 patients in tamponade group and 4 patients in medication groupExperience of clinicians inserting Sengstaken tubes not discussed - potentially relevant for efficacy and occurrence of complications. Study took place on ITU with clinically stable patients therefore applicability to unstable patients in other clinical settings such as Emergency Departments may be limited. Of limited relevance to three-part question as Sengstaken tube evaluated in isolation rather than in combination with medication


All the studies identified by the literature search are significantly old in research terms which may be attributable to the fact that Sengstaken-Blakemore tubes were first developed in 1950 (Sengstaken and Blakemore, 1950). However, this presents difficulties in application of these studies to current practice, since other pharmacological therapies have been added to initial management of variceal bleeding; current practices include the use of antibiotics, proton-pump inhibitors and tranexamic acid (Gluud et al, 2008) in addition to the isolated use of vasopressin and its analogues. This has the potential to limit the relevance of findings from these studies favouring the use of balloon tamponade to current clinical practice. Furthermore, four of the five trials discussed above compared the use of balloon tamponade by Sengstaken-Blakemore tube with pharmacological management in isolation, whereas the three-part question underpinning this review is based on the use of this procedure following pharmacological management. The rationale for this being pharmacological management is the current accepted ‘gold-standard’ initial practice in Emergency Departments prior to endoscopy (Scottish Intercollegiate Guidelines Network, 2008) but this means these trials may be of limited relevance to addressing the specific focus of this BestBET. However, Correia et al. (1984), and Fort et al. (1990) identify a small number of patients who had balloon tamponade following unsuccessful pharmacological management (six and four patients respectively). Whilst these cohorts are directly relevant to this review’s question, the small numbers combined with methodological weaknesses identified in the table above suggest validity of these trials design and associate findings may be compromised and application of this data to practice should be undertaken with caution. Four of the five trials analysed demonstrate various complications attributable to balloon tamponade up-to and including death. Whilst Teres et al.’s (1990) study identifies similar complication rates between balloon tamponade and pharmacological management, it is again difficult to judge the relevance of this to current practice due to developments in drug therapies used in variceal bleeding since this study was undertaken. None of the trials identify the experience or expertise of clinicians undertaking Sengstaken-Blakemore tube insertion or discuss the relevance of this to success or complication rates. Given the perceived rarity of this procedure in current clinical practice compared to the time these studies were undertaken, this may potentially be an even more significant factor in success and incidences of complications in today’s emergency care environment. However, ultimately in the clinical context of a life-threatening emergency where other options have failed or are unavailable, consideration may be given to the use of a Sengstaken-Blakemore tube, but this should be restricted to those clinicians with experience and clear competence in performing this procedure.

Clinical Bottom Line

There is insufficient high-quality or contemporary evidence that the use of Sengstaken-Blakemore tubes in the management of variceal bleeds in emergency care may improve patient outcome.


  1. Pinto-Correia, J., Martins-Alves, M., Alexandrino, P. and Silveira, J. Controlled trial of vasopressin and balloon tamponade in bleeding oesophageal varices Hepatology 1984
  2. Fort, E., Sautereau, D., Silvain, C., Ingrand, P., Pillegand, B. and Beauchant, M. A randomised trial of terlipressin plus nitroglycerin vs. balloon tamponade in the control of acute variceal haemorrhage. Hepatology 1990
  3. Holman, J. and Rikkers, L. Success of medical and surgical management of acute variceal haemorrhage. American Journal of Surgery 1980
  4. Panes, J., Teres, J., Bosch, J. and Rodes, J. Efficacy of balloon tamponade in treatment of bleeding gastric and oesophageal varices. Digestive Diseases and Sciences 1988
  5. Teres, J., Planas, R., Panes, J., Salmeron, J., Mas, A., Bosch, J., Llorente, C., Viver, J., Feu, F. and Rodes, J. Vasopressin and nitroglycerin infusion vs. oesophageal tamponade in the treatment of acute variceal bleeding: a randomised controlled trial. Hepatology 1990