Diagnostic testing

The strongest predictors of an upper gastrointestinal bleed are black stool, age <50 years, and blood urea nitrogen/creatinine ratio 30 or more.^[8][9] The diagnosis of upper gastrointestinal bleeding is assumed when hematemesis (vomiting of blood) is observed.

A nasogastric aspirate can help determine the location (source) of bleeding and help understand the best initial diagnostic and treatment plan. Nasogastric aspirate has a sensitivity of 42%, specificity 91%, negative predictive value 64%, positive predictive value 92% and overall accuracy of 66% in differentiating upper gastrointestinal bleeding from bleeding distal to the ligament of Treitz.^[8] A positive aspirate is more helpful than a negative aspirate (If the aspirate is positive, an upper gastrointestinal bleed is likely; if the aspirate is negative, the source of a gastrointestinal bleed is probably, but not certainly, lower). A smaller study found a sensitivity of 79% and specificity of 55%, somewhat opposite results from Witting.^[10] The accuracy of the aspirate is improved by using the Gastroccult test.

Determining whether blood is in gastric contents, either vomited or aspirated specimens, may be a challenge when determining the source of the hemorrhage. Slide tests are based on orthotolidine (Hematest reagent tablets and Bili-Labstix) or guaiac (Hemoccult and Gastroccult). There is some evidence that orthotolidine-based tests more sensitive than specific, the Hemoccult test's sensitivity reduced by the acidic environment; and the Gastroccult test be the most accurate.^[11] The sensitivity, specificity, positive predictive value, and negative predictive value have been reported as follows:^[10]

Holman used simulated gastric specimens and found the Hemoccult test to have significant problems with non-specificity and false-positive results, whereas the Gastroccult test was very accurate.^[12] Holman found that by 120 seconds after the developer was applied, the Hemoccult test was positive on all control samples.

A scoring system called the Glasgow-Blatchford bleeding score found 16% of people presenting with upper gastrointestinal bleed had Glasgow-Blatchford score of "0", considered low. Among these people there were no deaths or interventions needed and they were able to be effectively treated in an outpatient setting.^{[13] [14]}

Score is equal to "0" if the following are all present:

1. Hemoglobin level >12.9 g/dL (men) or >11.9 g/dL (women)
2. Systolic blood pressure >109 mm Hg
3. Pulse <100/minute
4. Blood urea nitrogen level <18.2 mg/dL
5. No melena or syncope
6. No past or present liver disease or heart failure

Bayesian calculation

The predictive values cited are based on the prevalences of upper gastrointestinal bleeding in the corresponding studies. A clinical calculator can be used to generate predictive values for other prevalences.

Peptic ulcers

Based on evidence from people with other health problems crystalloid and colloids are believed to be equivalent for peptic ulcer bleeding.^[15] In people with a confirmed peptic ulcer, proton pump inhibitors do not reduce death rates, later bleeding events, or need for surgery.^[18] They may decrease signs of bleeding at endoscopy however.^[18] In those with less severe disease and where endoscopy is rapidly available, they are of less immediate clinical importance.^[16] Tranexamic acid might be effective to reduce mortality, but the evidence for this is weak.^[15][19] But the evidence is promising.^[20] Somatostatin and octreotide while recommended for variceal bleeding have not been found to be of general use for non-variceal bleeds.^[15]

Variceal bleeding

For initial fluid replacement colloids or albumin is preferred in people with cirrhosis.^[15] Medications typically includes octreotide or if not available vasopressin and nitroglycerin to reduce portal pressures.^[21] This is typically in addition to endoscopic banding or sclerotherapy for the varices.^[21] If this is sufficient then beta blockers and nitrates may be used for the prevention of re-bleeding.^[21] If bleeding continues then balloon tamponade with a Sengstaken-Blakemore tube or Minnesota tube may be used in an attempt to mechanically compress the varices.^[21] This may then be followed by a transjugular intrahepatic portosystemic shunt.^[21]

Blood products

If large amounts of pack red blood cells are used additional platelets and fresh frozen plasma should be administered to prevent coagulopathies.^[15] Some evidence supports holding off on blood transfusions in those who have a hemoglobin greater than 7 to 8 g/dL and only moderate bleeding.^[15][22] If the INR is greater than 1.5 to 1.8 correction with fresh frozen plasma, prothrombin complex may decrease mortality.^[15]

Procedures

The above ulcer seen after endoscopic clipping

Upper endoscopy within 24 hours is the recommended treatment.^[15][23] The benefits versus risks of placing a nasogastric tube in those with upper gastrointestinal bleeding are not well known.^[15] Prokinetic agents to empty thee stomach such as erythromycin before endoscopy can decrease the amount of blood in the stomach and thus improve the operators view.^[15] This erythromycin treatment may lead to a small decrease in the need for a blood transfusion, but the overall balance of how effective erythromycin is compared to potential risks is not clear.^[15][23] Proton pump inhibitors, if they have not been started earlier, are recommended in those in whom high risk signs for bleeding are found.^[15] It is also recommended that people with high risk signs are kept in hospital for at least 72 hours.^[15] Blood transfusions are not generally recommended to correct anemia, but blood transfusions are recommended if the person is not stable (cardiovascular system instability).^[22] Oral iron can be used, but this can lead to problems with compliance, tolerance, darkening stools which may mask evidence of rebleeding and tends to be slow, especially if used in conjunction with proton pump inhibitors. Parenteral Iron is increasingly used in these cases to improve patient outcomes and void blood usage.