Practice Rational Imaging

Acute lower gastrointestinal haemorrhage

BMJ 2009; 339 doi: (Published 17 November 2009) Cite this as: BMJ 2009;339:b4156
  1. Andrew J Edwards, radiologist,
  2. Giles F Maskell, radiologist
  1. 1Royal Cornwall Hospital, Truro TR1 3LJ
  1. Correspondence to: G F Maskell giles.maskell{at}

    In most patients acute lower gastrointestinal bleeding resolves with conservative management, but when bleeding is especially severe, more invasive investigations and treatment may be needed

    Learning points

    • In most patients acute lower gastrointestinal bleeding resolves with conservative management

    • In patients with haemodynamic instability accurate identification of the source of bleeding is required to guide therapeutic intervention

    • Multidetector row computed tomography angiography is emerging as a fast, accurate, and non-invasive method of identifying the source of haemorrhage

    • Patients with massive or recurrent lower gastrointestinal bleeding are often elderly with substantial comorbidity. Accurate localisation can reduce mortality, although this remains high in this group

    An 81 year old woman presented to hospital with a 12 hour history of passing copious blood per rectum. She gave a history of ischaemic heart disease with previous coronary stent insertions and chronic obstructive pulmonary disease. She was taking regular aspirin but was not taking an anticoagulant. She had no other relevant medical history.

    On examination she was pale with cool extremities. Her pulse was thready (very fine and scarcely perceptible) at 100 beats/min, and her blood pressure was 100/76 mm Hg. Examination of the abdomen was normal but active rectal bleeding was noted.

    What is the next investigation?

    Although the history is highly suggestive of bleeding from the lower gastrointestinal tract, about 10-15% of patients presenting in this way are bleeding from a source in the upper gastrointestinal tract.1 Upper gastrointestinal endoscopy should therefore be considered in the first instance. If the upper gastrointestinal tract has been excluded as the source, the bleeding is most likely to originate from the colon. The table gives the prevalence of the most common causes of acute colonic haemorrhage in patients aged over 50. In most (85-90%) patients in this age group with acute lower gastrointestinal haemorrhage, the bleeding will stop with conservative management.2

    Major causes of lower gastrointestinal haemorrhage and proportion of cases attributable to each cause (prevalence)1

    View this table:

    In patients with haemorrhage of sufficient severity to cause haemodynamic instability either at presentation or after resuscitation, colonic investigation is indicated to localise the site and cause of the bleeding and to guide definitive treatment.1 Mortality is high in this group and may be up to 33% in those requiring surgery.3 4


    Colonoscopy is a well established, widely available, and accurate technique for diagnosing the cause of rectal bleeding. Accuracy in the acute setting may be reduced owing to the technical challenges of an unprepared colon and an unstable patient. Nevertheless, when clinicians are highly experienced, high levels of success have been reported both in the identification of the source of haemorrhage and in using haemostatic techniques to control bleeding.5 The small risk of perforation is increased when therapeutic techniques are used, and the requirement for sedation also increases risks.

    Digital subtraction angiography

    Digital subtraction angiography is usually performed via a femoral arterial approach. To visualise the arterial supply to the liver and lower gastrointestinal tract, shaped catheters are used to insert cannulas selectively into the inferior mesenteric artery, the superior mesenteric artery, and the celiac axis and angiography is performed. Superselective catheterisation of branch vessels is performed to enable maximal opacification of the distal vascular arcades. The examination may be performed under local anaesthetic and carries only minimal risks to the patient.

    The technique may be tailored to identify active extravasation of contrast into the bowel lumen to guide further management. More subtle findings can be neovascularisation resulting from a tumour and early venous filling secondary to angiodysplasia. Hepatic angiography is important if haemobilia is suggested from upper gastrointestinal endoscopy.

    In animal models angiography can detect bleeding at flow rates of as little as 0.5 ml/min,6 but most authors estimate that the flow rates detectable in vivo are a little higher.

    Despite this, digital subtraction angiography is often negative, even in cases of severe haemorrhage, because of the intermittent nature and unpredictable timing of arterial bleeding. Reported sensitivity ranges from 10% to 40%.7 8 Nevertheless, if active haemorrhage is detected selective embolisation can be performed, which can obviate the need for major surgery—particularly important in elderly people, who often have substantial comorbidity.

    Multidetector row computed tomography angiography

    The superior speed of image acquisition of the scanners used for multidetector row computed tomography allows the entire abdomen to be scanned within five to six seconds. This allows accurate depiction of the mesenteric vasculature with excellent resolution and is also fast enough to detect active extravasation of contrast into the bowel lumen.9 Kuhle and Sheiman found that the superior contrast resolution of computed tomograms compared with digital subtraction angiograms results in a higher sensitivity in the detection of active haemorrhage6; they found that computed tomography was able to detect bleeding with a rate of 0.4 ml/min.

    Multidetector row computed tomography angiography is non-invasive, fast, and readily available in most acute hospitals in the United Kingdom. It requires intravenous iodinated contrast and has a similar radiation dose to digital subtraction angiography. The detection of a bleeding point is most likely when haemorrhage is occurring at the time of examination. The prognostic value of a negative scan is currently unknown.

    If active bleeding is demonstrated, targeted therapy with either surgery or selective embolisation can be performed.

    Nuclear scintigraphy

    Technetium labelled red blood cell scintigraphy has been studied extensively as a means of localising the source of gastrointestinal haemorrhage. This is a non-invasive examination in which the patient’s red cells are labelled with an intravenous injection of radionuclide and images obtained over about an hour using a gamma camera. No major associated risks have been reported.

    Although nuclear scintigraphy is reasonably effective in confirming the presence of ongoing haemorrhage, with reported sensitivity of up to 50%, the exact site and cause of bleeding are not reliably identified and scintigraphy influences clinical management in only a minority of cases.10


    While being actively resuscitated, the patient was transferred for computed tomography angiography, which showed contrast extravasation from a site in the proximal transverse colon (figure). Diverticulosis was seen at this site, but no other cause for bleeding was seen.


    Computed tomography angiograms (axial, top; coronal, bottom) of the abdomen showing pooling of contrast material (arrows) in the proximal transverse colon, indicating the site of haemorrhage

    The patient continued to bleed and remained unstable. After discussion between surgeons and radiologists, and in view of the lack of immediate availability of interventional radiology, she was taken to theatre and right hemicolectomy was performed. Examination of the resected specimen confirmed haemorrhage as a result of diverticular disease. Although she initially made a good recovery with no further blood loss, she died 10 days later from postoperative pneumonia and multiorgan failure.


    Cite this as: BMJ 2009;339:b4156


    • This series provides an update on the best use of different imaging methods for common or important clinical presentations. The series advisers are Fergus Gleeson, consultant radiologist, Churchill Hospital, Oxford, and Kamini Patel, consultant radiologist, Homerton University Hospital, London.

    • Contributors: GFM had the original idea, and AJE selected the patient. Both authors searched the literature and wrote the paper. GFM is the guarantor.

    • Competing interests: None declared.

    • Provenance and peer review: Commissioned; externally peer reviewed.

    • Patient consent not required (patient anonymised, dead, or hypothetical).


    View Abstract

    Sign in

    Log in through your institution