Dr S. Tanaka, Department of Endoscopy, Hiroshima University Hospital, 1-2-3 Kasumi, Minami-ku, Hiroshima, 734-8551, Japan. E-mail: firstname.lastname@example.org
Background Lower gastrointestinal bleeding is a frequent cause of hospitalization, but diagnostic methods for this condition are not fully established. Transabdominal ultrasound is a widely accepted diagnostic tool in bowel diseases.
Aim To evaluate the usefulness of transabdominal ultrasound for lower gastrointestinal bleeding.
Methods We reviewed the medical records of consecutive patients who underwent transabdominal ultrasound as the first diagnostic procedure for acute haematochezia during the period June 1999 to June 2004. The study group comprised 111 patients and all underwent colonoscopy thereafter. Detection and diagnosis of lower gastrointestinal bleeding by ultrasonographic examination were evaluated by comparing the ultrasound diagnosis with the colonoscopic findings and final diagnosis.
Results The bleeding site was localized by colonoscopy in 90 of the 111 patients (81%). The bleeding site was localized by ultrasound in 59 of the 90 patients (66%). When the bleeding site was in the rectum, ultrasonographic detectability was 30% (10/33); ultrasonographic detectability was 82–100% when the bleeding site was elsewhere. Rectal bleeding and diverticular bleeding were difficult to diagnose by ultrasound, but for the other diseases, diagnosis by ultrasonographic examination was possible in 91–100% of cases.
Conclusions Ultrasonographic examination may be an effective screening method for lower gastrointestinal bleeding.
Gastrointestinal (GI) bleeding is a frequent cause of doctor consultations and hospital admissions. In upper GI bleeding, endoscopy has been established as the first-line diagnostic tool, and many therapeutic modalities have been developed.1–4 Methods of diagnosing lower GI bleeding include nuclear scintigraphy, mesenteric angiography and colonoscopy, but a single standard method has not been established because each has inherent advantages and disadvantages.5–11
The sonomorphologic appearance of bowel wall thickening in patients with acute or chronic disorders of the gut was recently evaluated for its value in the diagnosis of inflammatory bowel disease, ischaemic colitis, infectious colitis and malignant bowel tumours, and other bowel diseases.12–23 Transabdominal ultrasound (US) is a non-invasive and repeatable imaging study that can be performed easily without bowel preparation. The purpose of this study was to evaluate the usefulness of US for diagnosing lower GI bleeding.
Materials and methods
Patient population and study location
We reviewed the medical records of consecutive patients who visited Hiroshima University Hospital or Miyoshi General Hospital and underwent US as the first diagnostic procedure for acute haematochezia during the period June 1999 to June 2004. Patients with only occult blood in the stools were excluded. The study group comprised 53 men and 58 women with a mean age of 58 years (age range, 18–96 years). All patients underwent colonoscopy after US, and the interval between US and colonoscopy was less than 1 week. This study was approved by the ethics committee of our hospital and informed consent was obtained from all patients.
Sonograms were obtained with a real-time ultrasound apparatus (SSA-390A, 770A; Toshiba Medical Systems, Tokyo, Japan) and 3.5–3.75 MHz convex or 8 MHz linear transducers. Three ultrasound operators (T.Y., N.M., J.H.) with more than 2 years of experience performing US examinations of the gut were blinded to the results of other imaging procedures but had general information about the indications for US examination. The US examinations were performed with patients in the supine position. There were no specific preparations for the examination.
In each case, the entire large bowel and terminal ileum were imaged in the longitudinal and transverse planes. Examination of the large intestine begins in the right upper quadrant with identification of the ascending colon, recognized by its predictable position and its haustra. The examination proceeds along the ascending colon to the right lower quadrant, where the caecum is identified as a blind-ending loop of the large intestine; the terminal ileum is then identified. Once the right lower quadrant is adequately assessed, the ascending colon is followed to the transverse and descending portions of the colon. The sigmoid colon is followed into the pelvis, and the rectum is visualized through the distended urinary bladder. When the urinary bladder is empty, it is difficult to visualize the rectum.
An intestinal wall thickness of 4 mm or more was considered pathological, and was evaluated with respect to location, the presence of wall stratification and perigut features. Because each bowel disease is associated with differential features12–23 (Figure 1a–e), criteria developed on the basis of these findings (Table 1) were used to diagnose lower GI bleeding.
Preserved or lost stratification Focal destruction of stratification
Normal perigut fat
Normal perigut fat
Thickening of perigut fat Enlarged lymph node
Thickening of perigut fat Enlarged lymph node Abscesses, fistula
Normal perigut fat
In cases of massive bleeding, after US examination, urgent colonoscopy was performed without bowel preparation after the haemodynamics had stabilized. If necessary, endoscopic haemostasis was performed. In cases of scant haematochezia, colonoscopy was performed after bowel preparation. In cases of urgent colonoscopy, the examination was discontinued when the bleeding source was identified. In cases of non-urgent colonoscopy, the goal was to reach the caecum and enter the terminal ileum. If the bleeding source could not be detected by urgent colonoscopy, full colonoscopy with patient preparation was performed on another day.
We identified the bleeding source as a lesion with active bleeding, a visible vessel or an adherent clot. Lesions such as a friable tumour, colitis, discrete ulcer and severe haemorrhoid were also identified as sources of bleeding. When blood was absent in the lumen and no lesion was identified, we considered that no bleeding source had been identified.
We evaluated the ability of US to detect lower GI bleeding and diagnostic utility of US findings in cases of lower GI bleeding. The final diagnoses were established on the basis of colonoscopic, clinical and pathological findings and stool cultures. Detectability was evaluated by comparing US findings with colonoscopic findings, and the diagnostic utility was evaluated by comparing US diagnosis with the final diagnosis.
Statistical analysis was performed with the use of the Scheffe test for comparison of the detectability of lower GI bleeding by US according to locations. Differences were considered significant at P < 0.05.
The bleeding source was located in the large intestine or terminal ileum by colonoscopy in 90 of 111 patients (81%). The bleeding source was not detected by colonoscopy in 21 patients (19%). Small bowel bleeding was diagnosed in two of the patients in whom no bleeding source was detected by colonoscopy. A diagnosis of small bowel bleeding is generally made when fresh blood is found in the terminal ileum or coming through the ileocaecal valve and results of upper GI endoscopy are negative. The definite diagnoses in the 90 patients mentioned above included ischaemic colitis (n = 23), ulcerative colitis (n = 12), haemorrhoid (n =11), colorectal cancer (n = 7), diverticular bleeding (n = 7), infectious colitis (n = 7), Crohn's disease (n =5), colitis (unknown cause) (n = 5), rectal ulcer (n =4), proctitis (n = 3) and other bowel diseases (n = 6) (Table 2). Only three patients (two with diverticular bleeding and one with rectal ulcer) required urgent colonoscopic treatment to control the bleeding (with the use of a metal clip), and no surgery was performed.
Table 2. Final diagnoses for 111 patients with lower gastrointestinal bleeding, made at colonoscopy
The bleeding site was localized by US in 59 of the 90 patients. Sensitivity, specificity, positive predictive values and negative predictive values of US in detection of the bleeding site were 66% (59/90), 100% (21//21), 100% (59/59) and 40% (21/52) respectively (Table 3). Of the remaining 31 patients in whom US was unable to detect the bleeding site, 10 had rectal bleeding. Sensitivity of US in detection of the bleeding site was 30% (10/33) in the rectum and 82–100% in other locations. Thus, rectal bleeding was significantly more difficult to detect by US compared with bleeding in the other locations (P < 0.05) (Figure 2). Sensitivity of US in detection of the bleeding site in the colon was 86% (49/57).
Table 3. Detectability of lower gastrointestinal bleeding by ultrasound (US)
Bleeding source diagnosed
Bleeding source not known
Sensitivity 66% (95% CI: 0.55–0.75)
Specificity 100% (95% CI: 0.87–1.0)
Positive predictive value 100% (95% CI: 0.95–1.0)
Negative predictive value 40% (95% CI: 0.28–0.55)
Detectable by US
Undetectable by US
Diagnosis of lower GI bleeding
In 15 of the total 111 patients (nine with ulcerative colitis, four with Crohn's disease, one with Bechet's disease, one with eosinophilic enterocolitis) the clinical diagnosis was known at the time of US. These patients were excluded from the analysis, thus we evaluated the diagnostic utility of US in 96 patients with lower GI bleeding. Rectal bleeding and diverticular bleeding were difficult to diagnose by US, but differential diagnosis of bleeding in other lower GI sites by US was highly accurate (91–100%) (Figure 3).
This study was performed to evaluate the usefulness of US for diagnosis of lower GI bleeding. No such evaluation has been reported, although many studies have indicated that colonoscopy, angiography and scintigraphy can be used conventionally to localize lower GI bleeding.5–11 Angiography can be performed in patients with active lower GI bleeding and can be used as a therapeutic tool for lower GI bleeding, but it is invasive, and its sensitivity is low.5–9 For an angiogram to be considered positive, arterial bleeding of 0.5–1 mL/min is necessary,5, 7 and angiography is not without complications.5, 6, 9 Nuclear scintigraphy is a non-invasive technique. It is not more sensitive in detecting slow rates of bleeding but its sensitivity is not as good.5, 7, 10 A major disadvantage of scintigraphy is that it can be used to localize the bleeding to an area of the abdomen but cannot be used to identify the specific location.10 Colonoscopy as an early investigative modality offers both diagnostic and therapeutic potential for colonic sources of bleeding. It is more sensitive and specific than angiography and scintigraphy.5 Urgent colonoscopy provides a final diagnosis of colonic lesions in 74–90% of patients.10 However, colonoscopy should be aborted if the patient becomes haemodynamically unstable or if the bleeding is so severe that identification of a precise origin is impossible. Moreover, severe active colonic inflammation or ischaemia is a contraindication to complete colonoscopy because of the risk of perforation and severe pain.7
Transabdominal ultrasound is now considered a valuable method for assessment of bowel diseases in association with various clinical conditions. Inflammatory bowel disorders, such as ulcerative colitis, Crohn's disease, diverticulitis, infectious colitis and ischaemic colitis, produce mucosal inflammation, erosions and ulcer of various intensities in different bowel areas.12–23 Because it is radiation free, non-invasive and cost-effective, we commonly perform US in patients with abdominal complaints. A great advantage of US is that it can be performed at any time without bowel preparation, and the colon is easily depicted by sonography.17, 21 Therefore, we evaluated the usefulness of US for lower GI bleeding in the present study.
In our study, detectability of the bleeding site by US was poor in the rectum (30%), but good in other locations (82–100%). We believe that rectal bleeding was difficult to detect by US because of the penetration limits imposed by transducers and because there were some patients whose urinary bladder was not full. Sensitivity of US in the detection of bleeding sites in the colon was high (86%). In the light of this high sensitivity and other advantages, US could be the first choice of examination for detecting the site of bleeding in the colon.
In our analysis of the diagnostic utility of US, we found that rectal bleeding and diverticular bleeding were difficult to diagnose by US. The US features of colonic diverticulitis have been well described, and this condition is often easily diagnosed with US when the inflamed diverticulum is observed.13, 17, 22 However, diverticulitis does not always accompany diverticular bleeding. US shows no abnormalities in cases of diverticular bleeding without inflammation. Therefore, we were not able to diagnose diverticular bleeding by US. In cases of other bowel diseases, US was highly sensitive for a differential diagnosis.
In our series, only 3 of 111 patients (3%) required urgent colonoscopic treatment to control the bleeding. In other studies, endoscopic therapy was not applied in the majority of patients with lower GI bleeding.5, 6 This is in accordance with the findings in our series and suggests that urgent colonoscopy is not always required in lower GI bleeding. Most lower GI bleeding due to colonic tumours and inflammatory bowel disease, such as ulcerative colitis, Crohn's disease, infectious colitis or ischaemic colitis, requires no colonoscopic treatment. In our study, differential diagnosis of these bowel diseases was achieved by US with high sensitivity. Thus, performing US as a screening method could make it possible to better determine patients in need of urgent colonoscopy. Moreover, performance of proctosigmoidoscopy, rather than colonoscopy, is useful for patients in whom the site of bleeding cannot be detected by US because sources of bleeding that cannot be detected by US are often in the rectum. Bowel preparation for colonoscopy requires several hours,7, 10 but proctosigmoidoscopy can be performed easily with a glycerin enema alone.
Interpretation of our findings is limited, first because diverticular bleeding was present in only 7 of 111 patients in our study (6%). Diverticular bleeding is the most common cause of lower GI bleeding in Western countries,6, 7 but in Japan, the most common cause is inflammation of the colon, such as ischaemic or ulcerative colitis.24 This was evident in our study population, so we believe that our results reflect clinical practice situation in Japan. Second, the accuracy of US depends upon the experience of the examiner. In our study, examiners had more than 2 years of experience performing US examinations of the gut. It is often difficult to perform US examination in obese patients, but there was no case in which US could not be performed because of obesity in our study.
In summary, our results indicate that US may be an effective screening method for lower GI bleeding. US has many advantages and shows good sensitivity in the diagnosis of lower GI bleeding. The diagnostic utility of US is such that it is a good first procedure for cases of lower GI bleeding.