Upper GI Bleeding
Case 1: The parents of a 3 year old who you have been following since birth for biliary atresia, call to report a "nosebleed" (epistaxis) overnight. Closer questioning discloses that what they are calling a nosebleed is simply a puddle of blood found on the pillow. Having anticipated this potential complication, you ask them to meet you in the Emergency Department. There you find him to be in no distress, with no tachycardia or diaphoresis. You can find no site of bleeding in the nose or pharynx, and you also note his ascites has disappeared and his spleen seems smaller than when you saw him last week. What's going on?
Case #1 described above illustrates the one exception to the rule in large volume bleeding. In children with varices and ascites (both arising from portal hypertension) the acute volume loss from the bleeding can be repleted by 'auto-infusion' of the ascites. Portal hypertension triggers ascites at relatively low pressures (10-12 mm Hg), and the volume depletion from bleeding results in enough reduction in the portal pressure to coax the fluid back into the circulation. The hypovolemic state accounts for the loss of the previously existing splenomegaly. These patients also illustrate that all blood loss is whole blood and that the hemoglobin and hematocrit will not fall until they are volume repleted with crystalloid or plasma. Cirrhotic patients with ascites are the only ones where the acute CBC may be a better indicator of the volume lost than vital signs at presentation.
His hemoglobin is 7 (hct 21). His INR is 1.2. He continues to be in no acute distress and he is able to go directly to endoscopy, having requested 2 units of packed RBCs to be available in the OR for emergency transfusion if needed, with more packed RBC and fresh frozen plasma standing by. A small ulcer is found in the distal esophagus, with an associated gastric varix. There are two other fully distended esophageal varices which are band ligated, and while sclerotherapy is considered for the gastric varix (the banding is impossible to accurately apply in this location) you elect to watch as it appears to be thrombosed and plans are made to return for a repeat endoscopic inspection and treatment as needed in a week or two. He tolerates the procedure well with no complications, and after talking with his parents, you call his transplant specialists to update them on his situation.
Gastrointestinal bleeding covers a wide topic, and is best managed by subdividing it into smaller and smaller entities. But even before making the first obvious decision as to whether it's "upper" or "lower" GI bleeding, the first step is to make a rough assessment as to how MUCH bleeding is going on. In pediatrics, the best single vital sign for assessing acute volume depletion is the heart rate rather than the blood pressure, since infants, children and adolescents have a huge reserve capacity for increasing cardiac output by increasing the heart rate. Thus, the blood pressure begins to fall only in late shock. Orthostatic change in the heart rate is a useful sign (only occasionally unreliable), since a difference of 10% or more may indicate substantial acute volume depletion. Another sign to look for are cool extremities, often with a relatively sharp demarcation between cool and normal skin temperature, as an indication of peripheral vasoconstriction. These signs are applicable to acute volume depletion from any cause (such as vomiting and diarrhea) and not just to acute bleeding. As with any pediatric life support issue, the first steps in assessment and management are to verify airway, breathing, and circulation (ABC). Acute volume depletion requires rapid volume replacement and determination of the source of loss.
When there is a significant difference between the degree of volume loss from either end and the apparent normal state of the intravascular volume, the next step is to verify the material as blood and that it is indeed coming from the patient. Here are some anecdotal examples: 1) A child whose bright red diarrheal stools are closer in color to the red-orange of the food dye of the fruit punch that he has been guzzling during his acute gastroenteritis than to any blood color you've ever seen (i.e., the "bloody" diarrhea is really fruit punch). 2) The robust, demanding, overfed infant whose mother has mastitis and whose hematemesis is actually coming from his mother's blood loss (i.e., swallowing mother's blood from a bleeding nipple) rather than his own. 3) The newborn whose mother had placenta previa and a particularly bloody delivery whose hematemesis again is ingested maternal blood and not his own. The first example often is resolvable during the initial phone conversation. The third can be identified reliably by performance of an Apt test for fetal vs adult hemoglobin, but the second often requires careful history taking and examination to exclude intrinsic GI bleeding.
Converse consideration must also be given to those who present with signs of acute intravascular volume depletion (especially impending shock) and NO signs of bleeding discernible externally. The GI tract can easily hide a loss of blood amounting to a substantial portion of the total intravascular volume. Processes with high rates of bleeding, particularly if beyond the stomach (making them less prone to hematemesis) must be considered. Such lesions include (but are not limited to) duodenal ulceration with arterial bleeding, duodenal varices or varices at small bowel anastomotic sites (in the case of children with surgical hepatoportoenterostomies), small bowel vasculitides, hemobilia (bleeding in the biliary tract), and enteric duplications (including Meckel's diverticula) with secretory mucosa and secretory products that lead to ulceration.
The next step is to verify that the bleeding is indeed coming from the GI tract, in part to determine whether the following algorithms will be applied, but also as reassurance that the volume of any recurrent bleeding is not likely to be high or is likely to be easily managed. Epistaxis is a common cause for moderate volume bleeding (some epistaxis blood is swallowed and then vomited as hematemesis) and is more common than peptic ulceration as a cause of hematemesis and melena. Always check the anterior portion of the nasal septum for evidence of blood and ulceration indicative of bleeding. Application of direct pressure for 5 minutes (by the clock) to allow for good clot adherence and retraction (no peeking or the clot will lift off and the bleeding will resume) during any recurrent bleeding episode, is usually all that is necessary. Cautery typically causes more problems than it solves. Topical antibiotics can be used to treat nasal impetigo if that is the cause of the epistaxis. Dental and oral bleeding typically is smaller in volume and is usually identifiable on close inspection. Anal fissuring is technically a "GI" lesion, but the risk of rebleeding is low and its generally benign nature differentiates it from the lesions below. Here, the bleeding is typically bright red, and can be painless (though more often associated with pruritus at the anus or cramping prior to the passage of a larger caliber formed stool). Passage of clots is possible if the fissure extends internal to the internal anal sphincter. But the differentiating hallmark is any continued dripping of blood into the toilet after the formed stool has passed (and the internal anal sphincter has closed) or the persistence of bright red blood on the toilet paper for more than 2-3 wipes. Either of these indicates the presence of the lesion outside the internal (but potentially proximal to the external) anal sphincter. Good inspection requires gentle separation of the buttocks, best attained in the decubitus position with the knees up against the chest. The goal is to open the external anal sphincter, often attainable by having the patient take a slow deep inhalation. The fissure is identifiable as the red or white-based linear ulcer, usually anteriorly or posteriorly positioned. Specific attention should be paid to whether the margins appear undermined (traumatic fissures have simple vertical margins, while the anal lesions of Crohn's disease and the vasculitides are caused by the undermining of the subcutaneous supportive structures).
Once it is verified the patient is indeed losing blood from their GI tract, but is stable, attention turns to specific diagnosis. Here again subdividing the possibilities is helpful, with the next step being the semi-artificial conceptual one of "upper" vs "lower" tract bleeding. With the provisos of the preceding paragraphs accounted for, hematemesis is a reasonable indicator of an UGI source. Conversion to "coffee grounds" is not necessarily indicative of a gastric source since any blood placed in acid for even a few seconds converts to acid hematin (brown color), and conversely if the acid is neutralized or is otherwise not present for any reason, the blood will not convert (and will be red). On the other hand, melena is only an indicator of bleeding in an area bathed in acid, and while this usually indicates the upper GI tract, other sites of acid production such as a Meckel's diverticulum (with acid secreting ectopic gastric mucosa) or occasionally acid fermentation in the right colon can trigger the same chemical conversion. And conversely, the absence of melena in the bleeding does not preclude an UGI site, since brisk bleeding from an arterial source in the duodenum, plus the rapid transit times of infants and younger children, and their lower acid secretion rates can result in passage of blood per anus that is still nearly bright red.
Quantification of the bleeding seen is of some help diagnostically, as is the presence or absence of associated symptoms and signs. Further stratification by age helps establish probabilities:
Newborns are not old enough (not enough time) to develop full peptic ulceration. Coagulopathy due to liver cirrhosis is possible, but rare. Neonates under the intense physiologic stresses that would place them in an NICU setting typically develop gastric erosions rather than deeper gastric ulceration or they can hemorrhage from DIC. Even vitamin K deficiency of infancy takes time to develop.
Infants have had time to develop some, but not all of the ills that befall older children. Again mechanical trauma to (or other lesions of) oral structures will result in small to medium amounts of hematemesis and are far more common than the erosions of severe reflux esophagitis or frank ulceration of the stomach or duodenum. The latter lesions are usually suspected by signs of dyspepsia, including (but not limited to) crying, irritability centering around feedings, colic, drooling, eructation (belching), and vomiting. The preverbal infant and young child often will not be able to adequately indicate the associated pain, and peptic disease is therefore more often not identified until there is significant vomiting or even frank hematemesis. Allergic gastroenteritis more commonly first appears in the first few months of infancy though eosinophilic gastritis can be identified at any age. Though more commonly presenting as failure to thrive, allergic gastroenteritis can occasionally present with hematemesis or other evidence of upper (and lower) GI tract bleeding.
In toddlers most of the above processes continue, but their added mobility increases the possibilities for mechanical issues, including ingested caustic agents (lye, tile cleaner, electric dishwasher detergent) or acids, and foreign bodies. A toddler presenting with signs of partial esophageal obstruction (inability or unwillingness to swallow solids) and intermittent hematemesis warrants a radiographic evaluation to establish the absence of a lodged foreign body and if no clear history is obtainable regarding of the duration of the lodgment, caution is to be exercised in its removal, since penetration of or even embedding in the wall of the esophagus or adjacent structures (e.g., vena cava or aorta) is not uncommon. Ingestion of button batteries is another special consideration in the case of foreign bodies since the lithium ones can retain sufficient charge as to cause significant mucosal burn even when they appear to be dead, while the mercury, silver, and alkaline button batteries usually do not.
In older children, a Mallory-Weiss tear (esophageal tear) or an erosion caused by prolapse of a portion of the gastric cardia (typically along the lesser curvature) is far more common than nonspecific gastritis, erosion (esophageal or gastric), or frank ulceration as a cause of UGI bleeding. Such bleeding usually follows a period of protracted and usually forceful vomiting, and is usually relatively limited in volume (a few teaspoons), but it can be profuse. A rectal examination with stool which is negative for occult blood helps verify the observed short duration of the process. Esophageal erosion typically is preceded by complaints identifiable as peptic in origin in pediatrics and is rarely present in a pain free setting, which requires an extensive burn. In the case of the older child or adolescent with this presentation, consideration needs to be given to NSAID-induced gastropathy arising from self-medication. The lesions are typically erosions arising from the disruption of mucosal cytoprotection due to the broader inhibition of prostaglandin synthetase. The incidence is low compared to the widespread use of these medications and the increase in the absolute rate of this complication reflects the increasing use of these medications in the over-the-counter setting.
Upper GI bleeding etiologies by age:
Frank peptic ulceration remains uncommon as a source of bleeding during most of childhood and adolescence. Helicobacter pylori can play a role, and by itself, can cause bleeding from gastritis, though far more commonly it presents as a non-bleeding distinctive nodular gastropathy with preference for the antrum. While breath testing using labeled urea is identified as showing promise in establishing the diagnosis, the North American Society of Pediatric Gastroenterology and Nutrition's position is that endoscopy remains the only reliable means of establishing and refuting its presence, and that blood testing, due to poor specificity, is of little utility (1). In that regard, the low probability of Helicobacter gastritis in this age group even in the face of proven duodenal ulceration means that for each child appropriately identified by serologic testing, there will be several falsely labeled, and extension of this testing into evaluation of those who simply present with hematemesis or even just pain raises the likelihood of identifying false positives.
However, in true hemorrhage from the upper GI tract (defined as bleeding sufficient to require volume repletion with blood), peptic ulceration, variceal bleeding, and DIC are the most common causes. Of these, variceal bleeding is the most frightening, since blood loss rates can approach total blood volume within an hour or less. Varices arise from portal hypertension, which in turn arises most commonly from cirrhosis (e.g., biliary atresia as in case #1), but can also arise from extrahepatic obstruction of the portal vein. They most commonly present in the distal esophagus but may also be found in the gastric cardia or in the duodenum, where they can be far more difficult to treat. They represent an enlargement of the submucosal venules as a means of rerouting the blood flow from the portal to systemic venous circulation (a porto-systemic shunt) and the degree of portal hypertension required to establish the shunt is only minimally higher than that which would produce splenomegaly or ascites, making these physical findings important in the evaluation of the hemorrhaging patient. Identifying patients at risk for this prior to hemorrhage is far preferable, and a routine search for these findings should be undertaken at every office visit of any patient being followed for a process that can lead to cirrhosis (such as biliary atresia in the case presented). The bleeding is typically painless, as the vessels are superficial to the muscular layers and the erosion that starts the bleeding is therefore particularly shallow. The initial (or "herald") bleed may be surprisingly small as the vein rapidly collapses and clots, but the dislodgement of that clot can be followed by particularly voluminous bleeding. Endoscopic examination for diagnosis and treatment either by banding or injection of sclerosing agents is the preferred acute management once volume repletion has made sedation or anesthesia possible, and while a trans-jugular intrahepatic portovenous shunt placement (TIPS procedure) may be palliative, the only "cure" is resolution of the underlying cirrhosis (by transplantation) for those with intrahepatic disease conditions. Those with cavernous transformation of the portal vein typically will create other intraabdominal shunts in locations which do not bleed so profusely if they can be carried into the second half of the first decade.
The myriad causes of upper GI bleeding prompt usage of a layered strategy to the diagnostic process. With the above history parameters and physical findings indicating that: 1) the patient is (at least temporarily) hemodynamically stable, 2) no clear site of bleeding external to the GI tract as an alternate source to explain the bleeding, 3) no known lodged esophageal foreign body, 4) medium to moderate bleeding in whom the presumed risk of later hemorrhage is worth giving serious consideration; the next step is placement of a naso-gastric catheter to assess the volume of blood that has not been regurgitated, and more importantly, to identify (or refute) any ongoing bleeding. A large-bore tube is recommended, as clots may need to be removed, and the orogastric (rather than nasogastric) route may facilitate evacuation. Gastric lavage with normal saline at body or room temperature is more comfortable. Icing of the saline is not required and in smaller patients, may produce significant hypothermia. Gastric sampling and lavage may be omitted if variceal bleeding is suspected and endoscopy is already planned, but the hemorrhaging patient typically otherwise deserves both procedures. The gastric sampling and lavage serves to identify the patient who has indeed bled briskly from the upper GI tract or who may still be bleeding, and who will need to go on to immediate endoscopic examination. In contrast, the patient whose bleeding is suspected as coming from a non-intrinsic or otherwise low-risk source does not require immediate endoscopy. A negative gastric aspirate in a patient who is otherwise only suspected of an upper GI bleeding site because of passage of melena and who has no hematemesis would prompt a search for alternate sites of acid-associated bleeding such as a Meckel's diverticulum.
Gastric aspiration will only rarely be falsely negative. This would typically occur in patients who are bleeding from a deep ulcer in the duodenal bulb with sufficient pylorospasm to prevent any regurgitation of blood into the stomach. These patients can easily be mistaken to have a Meckel's diverticulum with acid-secreting ectopic gastric mucosa and secondary ulceration as they too may present with volume depletion and melena but no hematemesis, and if suspicion is otherwise high (primarily due to tenderness to epigastric palpation) and/or larger volumes of bleeding, endoscopy is warranted anyway.
Endoscopy offers both diagnostic and therapeutic advantages, and typically is much more sensitive than radiographic evaluation of the upper GI tract as most hemorrhaging lesions are still superficial. Radiographic contrast studies are dependent on identification of a moderate-sized ulcer crater, and lack sensitivity in identifying risks for rebleeding such as a visible vessel or adherent clot. Even varices are difficult to identify radiographically, and endoscopy offers the ability to intervene to reduce the risk of rebleeding via variceal band ligation or intravascular injection of a sclerosing agent. Likewise, ulcers and other lesions at significant risk for resumption of hemorrhage may be addressed thermally or chemically through the endoscope.
Peptic ulcer disease is most accurately diagnosed by endoscopy. Referral to an endoscopist (usually a gastroenterologist) facilitates diagnosis and treatment since treatment regimens which consist of antacids, histamine-2 blockers (cimetidine, ranitidine, etc.), proton pump inhibitors (e.g., omeprazole) and antibiotics for H. pylori create numerous therapeutic option combinations. Optimal therapeutic decision making for pediatric patients with peptic ulcer disease is best left to gastroenterologists who are most familiar with the most recent studies and recommendations.
If endoscopy fails to identify the bleeding lesion, further investigation of the hemorrhaging patient includes radionuclide scanning and angiography. Radiographic intervention can include selective intraarterial embolization as well as the TIPS (trans-jugular intrahepatic portovenous shunt) procedure, but a full discussion of this is beyond the scope of this chapter in basic diagnosis, as the choices are typically guided by the pediatric gastroenterologist involved in the endoscopic procedure above (2,3,4,5).
Lower GI bleeding
Case 2: The parents of an otherwise robust 3 year old boy call with a frantic report of bright red bleeding per anus. Pain is denied, as are fever, malaise, or rash. The toilet bowl seems filled with blood and clots, but the anus wipes clean with one swipe and no further blood is seen. You are able to convince them to come to the office instead of heading for the ER. In the office, the child is in no distress and wonders what all the fuss is about. His vital signs are normal for age, and physical examination shows no abnormalities, including external inspection of the anus with the child in the knee-chest position on his left side to enable full exposure of the anus down to the internal anal sphincter. What do you do next?
The patient who presents with bleeding only from the anus produces a separate (but overlapping) diagnostic tree. The presence of melena hints at an upper GI tract origin, but is simply indicative of passage of the blood through acid, which can arise in the presence of acid secreting ectopic gastric mucosa in any enteric duplication, of which the most common is a Meckel's diverticulum, or occasionally if the cecum is particularly acidic due to fermentation. Conversely the absence of melena does not exclude an upper GI tract origin if the transit time is sufficiently short or the acid is otherwise neutralized or not present and the bleeding is brisk enough.
As discussed in the patient who presents with hematemesis, the initial evaluation centers around rapid estimation of the volume of blood lost and the risk of ongoing or recurrent bleeding. Again the problem is best stratified into ages and rates of loss.
As with upper GI bleeding, the neonate has fewer diagnostic possibilities since most problems take a while to develop fully. While any infection (bacterial or viral) can lead to sufficient mucosal inflammation ranging from punctate bleeding to gross hemorrhage and DIC, necrotizing enterocolitis is almost unique to the neonatal period. In other respects, resembling ischemic injury in the older child or adult, the process in the neonate does more commonly include submucosal pneumatosis, implying compromise of the mucosal barrier. It usually presents with other signs of intestinal obstruction, partial or complete, and bleeding is typically one of the lesser findings, and is most commonly occult. It presents more commonly in the severely premature, but can afflict term infants who have a preceding clinical problem that predisposes them to bowel ischemia (such as polycythemia or birth asphyxia). Refer to the chapter on necrotizing enterocolitis.
Allergic enteropathy is more typically a problem of the young infant, as the inflammatory process is acquired and requires time to set up. It typically presents before 2 months of age with either occult or gross bleeding, and typically is accompanied by failure to thrive and/or a moderate degree of mucus in the stool to suggest widespread mucosal irritation. Involvement of the proximal GI tract (the stomach and small bowel) more commonly results in slow weight gain, signs of poor enteric motility (vomiting) and other signs of gastric stasis and protein losing enteropathy. Involvement of the colon is associated with mucusy stools laced with blood. In the latter, a Wright stain may be helpful only if it shows sheets of eosinophils, but a firm diagnosis rests on mucosal biopsy showing widespread nests of eosinophils in the submucosa rather than the scattered eosinophilia seen in more nonspecific inflammation. A clinical diagnosis may be made by rapid and complete resolution of the symptoms by elimination of the offending protein either by a return to exclusive breast feeding or substitution of a properly hydrolyzed formula (e.g., Nutramigen, Pregestimil, Alimentum). A switch to an alternate allergenic protein source (soy, goat's milk, etc.) during a period of sensitization may result in further reactivity and only true hypoallergenic feedings are to be allowed. Widespread reports, laboratory research, and personal experience indicate that while generally hypoallergenic, maternal breast milk may contain identifiable fragments of cow's milk protein from the maternal diet in quantities sufficient to trigger a reaction. Personal experience suggests the quantity needed in the maternal diet is substantial, and typically lies outside routine dietary parameters, however maternal exclusion of dairy products may be undertaken in the case of stubbornly persistent (and typically low-grade) inflammation. If allergic enteropathy (gastroenteropathy or colitis) is encountered, firm exclusion of the offending protein is to be undertaken for the entire first year of life in hopes of eliminating the clone of sensitized lymphocytes. This involves reading the ingredient panel of every item the child will eat, looking for "non-fat dairy solids" or "non-dairy" creamers (which contain powdered milk protein). If a soy allergy is present, the prohibition shifts to soy, including soy sauce and tofu. Typically the exclusion is not complete, and if (repeated) inadvertent exposure shows no sign of reaction, the restrictions can be lifted. But recurrent reactions can be severe if of the acute hypersensitivity (type I) variety. This can result in sufficient vomiting and diarrhea to cause significant volume depletion, and if uncertain as to the residual reactivity, a formal staged dietary challenge with nursing support (i.e., this may need to be done as an inpatient or in an observation unit) may be needed at one year of age before the offending food may be safely reintroduced in quantity.
Another cause for minor bleeding per anus that is unique to infancy is nodular lymphoid hyperplasia. It typically presents with punctate bleeding best characterized as streaks of blood with small streaks of mucus in otherwise normal stool in an otherwise thriving infant. The number of streaks and the amount of blood do not vary with fecal texture. This compares to infection, allergy or other more generalized inflammatory processes of the distal bowel where loose stool indicates inflammation, and therefore goes hand in hand with more mucus and blood. The only time the bleeding disappears in nodular lymphoid hyperplasia is in the face of liquid stools, in which case the streaks of mucus and blood are dissolved in the diarrhea but can be found by occult blood testing. Nodular lymphoid hyperplasia can readily be identified by proctoscopic examination which typically demonstrates a rectum that is studded with submucosal nodes measuring 2 mm across with central ulceration. The bleeding comes from the ulceration and the intervening mucosa is completely normal in appearance, explaining the disparity between the texture of the stool, the amount of bleeding and the normal growth of most of these infants. This permits exclusion of allergy and infection as possible causes since these typically cause more widespread inflammation, visible in the rectum of infants presenting with visible blood and mucus in the stool. Nodular lymphoid hyperplasia is a benign, self-limited process associated with the age-appropriate hypertrophy of the lymphatic tissue of the enteric submucosa. In some infants, the central portion of the overlying mucosa undergoes punctate ulceration. The cause of the process remains unknown. What is known is that the process normally becomes dormant during the latter half of the first year as the nodes regress in size (and activity), and though there may be occult blood found in the stool for the remainder of the first year, there is little likelihood of anemia and no association of any later enteric disease process. As such, my usual recommendation is to continue with routine feedings, introducing solid foods at the usual times as the process is not allergic in origin. The hemoglobin may be checked slightly more frequently than your usual schedule for age, and iron supplementation should be started only if it drops significantly. Inability to keep up with iron loss is atypical enough to warrant reassessment of the original diagnosis.
Toddlers in turn have a cause for chronic occult bleeding and often times severe anemia which is unique to their age range in overconsumption of milk. The lactoferrin of cow's milk has an extremely high affinity for iron, higher than anything in the human iron transport system, and it typically is not saturated in milk as routinely sold (but it is saturated in infant formula). Lack of an alternate iron source and excessive intake of milk can result in severe iron depletion, and as iron is also required for maintenance of gut mucosal integrity, the process accelerates as the iron stores fall. Hemoglobins of 4.0 gm/dl are not uncommon in this setting, and if tested, the stools may be positive for occult blood. The diagnosis is made by a detailed diet history, and verification of the extremely low iron stores. The cure is effected by a return to a truly regular diet for age with reasonable milk intake and a sufficient source of iron, though an occasional patient will require transfusion.
Processes that can cause small but visible quantities of bleeding at any age are dominated by infection, though anal fissures are even more common, as described previously. The most common worrisome organisms include Campylobacter, Salmonella, Shigella, and enterotoxigenic E coli, but other routine pathogenic bacteria and protozoans can be acquired from contaminated food and standing water sources. The first 4 are routinely included in culture screens for enteric pathogens (the rest are not). Campylobacter can cause a severe colitis which is identified more often now that CT scanning is the preferred method for identification of appendicitis in the patient presenting with crampy abdominal pain. The degree of thickening of the submucosa and muscular layers can be mistaken for transmural thickening indicative of Crohn's disease, with the pain and bloody diarrhea adding to this diagnostic possibility. But a normal ESR in the proper clinical picture suggests a pause in the rush to colonoscopy and therapy until the culture results are available, since the process clears rapidly with erythromycin treatment. Shigella also warrants antibiotic therapy if found, and while treatment of Salmonella may raise the risk of producing a chronic carrier state, since most carriers arise from colonization of the gallbladder, cautious treatment with an agent concentrated in bile (such as trimethoprim/sulfamethoxazole) if the organism is sensitive may be warranted in the patient with ongoing or severe symptoms. The one organism whose treatment with antibiotics or antispasmodics is to be avoided is enterotoxigenic E. coli (such as E. coli O157). Use of these agents can produce enough enterotoxin release as to trigger Hemolytic Uremic Syndrome. Antibiotics should be held until the offending bacteria is positively identified, and even over the counter antispasmodic agents are to be avoided.
The immune suppressed patient presents a particular challenge, for in addition to the above agents one needs to consider atypical organisms such as cytomegalovirus and Mycobacterium avium intracellulare (MAI), which can produce severe bleeding with transmural lesions scattered throughout the bowel and colon. Epstein-Barr virus can cause lymphoproliferative disease with chronic low-grade blood loss and more of a protein losing enteropathy picture. Typhlitis also occurs in the patient recovering from neutropenia as the new granulocytes are preferentially directed toward the inflamed cecum. These processes are rare in the immune competent patient.
Enterocolitis due to Hirschsprung's disease can occur both prior to and after surgical repair. The latter instance is indicative of stricture at the anastomotic site and recurrence of functional obstruction. As a stasis phenomenon it can also be seen in those with ileorectal pouches and other anastomosis, and while it can cause bleeding, it usually presents with explosive, foul diarrhea. For evaluation of the infant presenting with enterocolitis as the first manifestation of Hirschsprung's disease, see the chapter on constipation. Hirschsprung's disease itself is covered in a separate chapter.
Other processes that can cause moderate bleeding volumes, usually as part of a broader clinical picture include general obstructive processes such as intussusception, volvulus, and other mechanical issues that can cause focal bowel ischemia. They usually present with other signs of obstruction, typically with an acute onset of crampy abdominal pain that cycles every 10 to 60 minutes as the major migrating motor complex passes through the obstructed segment. Waiting for the passage of currant jelly stool (bloody stool) before considering intussusception in the differential diagnosis is to be discouraged since this is a late finding. Early radiographic evaluation with plain films is to be encouraged. In fact, the possibility of intussusception should be considered when any type of blood in the stool is encountered.
In patients presenting similar to the above, but with lesser signs of obstruction, consideration should be given to vasculitis, far more commonly due to anaphylactoid (or Henoch-Schonlein) purpura than to Systemic Lupus Erythematosus. The typical presentation is dominated by crampy pain with a usually minor bleeding component. In the case of anaphylactoid purpura, the platelet count may be high, but the ESR is typically in the normal range, in contrast to the elevation seen in SLE or to EBV-induced lymphoproliferative disease, both of which can mimic the radiologic picture of HSP. Treatment with corticosteroids is discouraged until these entities and lymphoma or leukemia are more definitively ruled out.
Inflammatory bowel disease (IBD) typically presents with a history of chronic but nonspecific signs. Poor weight gain and especially linear growth can be noted as much as 6 months before onset of cramping and bleeding, though there are hyper-acute variants of ulcerative colitis. These entities are covered in detail in a separate IBD chapter but for the purposes of this discussion, IBD can produce anything from occult bleeding to florid bloody diarrhea. In ulcerative colitis, the blood and stool texture are inversely related, with both mucusy diarrhea and bleeding being indicators of inflammation. Crohn's disease, more commonly affecting portions of the GI tract other than the rectum, can present with unremarkable stools, but will also produce mucusy diarrhea if the distal colon is involved. A quick check of the oral mucosa and anus for the undermined vasculitic lesions seen in either site can be diagnostic of Crohn's disease. A detailed family history searching for other versions of autoimmune disease can be supportive diagnostically, for while it is rare to have another with IBD in the family, there is often a strong family history of other autoimmune manifestations.
And finally, among the (relatively) common causes of colonic bleeding, polyps are to be considered whenever there is a report of painless bleeding of apparently moderate volume. Solitary juvenile polyps are the most common, and typically do not become large enough to cause bleeding before the end of the second year. As hamartomas, they are extremely vascular but have no sensory tissue and bear essentially no neoplastic risk as long as they are indeed solitary. The familial polyposis syndromes produce diffuse adenomatous polyposis, resulting in studding of the mucosa with often nearly confluent polyps, all of roughly the same size. These carry a significant neoplastic risk, and were until recently, an indication for early colonic resection, but experience with NSAIDs (particularly sulindac) in adults has been extended to children in causing regression of the visible lesions. It remains to be seen if this significantly reduces the long-term neoplastic risk, but it seems to permit a reduction from the every-other-year colonoscopy surveillance often undertaken in the second decade. The diagnosis of polyps (single or multiple) starts with the history of painless bright red bleeding, generally without anemia despite a protracted history, and no anal fissure on inspection. Digital rectal examination is usually diagnostic as most solitary polyps arise within the last 2 inches of the rectum, and the familial adenomatous polyposis syndromes result in many small polyps within reach. Gardner's syndrome is associated with unusual retinal pigmentation in affected individuals and may also present with osteomas, though these usually are more prominent in the second decade while the small but visible bleeding usually appears by the middle of the first decade. Peutz-Jeghers syndrome is associated with pigmentation in unusual sites (buccal mucosa, the webbing between fingers and toes) but like the other multiple hamartomatous polyp syndromes, PJ typically does not present with bleeding. Therapy for isolated polyps is endoscopic removal and for multiple polyps is endoscopic sampling to establish a diagnosis. Waiting for a polyp to autoinfarct will not permit specific identification as to type, and the presence of more than 3 polyps, even with a "juvenile" type histology, is still associated with a higher risk of eventual colon cancer. Recent advances in genetic screening in the diagnosis and management planning of the familial adenomatous polyposis syndromes in pediatrics was recently discussed in detail in reference #6.
Lower GI bleeding etiologies by age
In summary, identification of "lower" GI bleeding is even more dependent on the history and characterization of the bleeding than that from the upper GI tract, and can be confounded by "upper" GI sources. Even with hemorrhage, patients rarely become significantly volume depleted on an acute basis and in most instances there is enough time to perform appropriate testing, including culture, in a sequential manner. Many times, the workup of the crampy patient with modest bleeding in loose mucusy stools involves a quick survey of inflammatory markers and a 2 to 3 day wait for the culture results from the rectal swab. A rectal swab has a superior yield over culture of stool material because the center of the lumen (i.e., stool material) typically contains dead or less viable organisms, while the viable enteric pathogens are closer to the mucosa and are more readily sampled by brushing the rectal wall. If the ESR is low in the face of thrombocytosis and a history of crampy pain, particularly in a patient with any dermal lesions, a small bowel series may demonstrate the characteristic "thumb-printing" of localized mucosal edema typical of HSP and enable early administration of corticosteroids. Otherwise, if the ESR is high and the growth parameters are low, once the culture results are found to be negative, the next step is colonoscopy to look for lesions indicative of IBD. On the other hand, finding a solitary polyp on initial examination permits a relaxed scheduling on a more elective basis both for the physician and the family.
For case #2, our patient's presentation with painless bleeding of apparent moderate volume yet without signs of significant volume depletion is indicative of a polyp. Rectal examination finds a single 1 cm pedunculated polyp 2 cm from the anal verge. Colonoscopy is scheduled, as the prep for even proctoscopy for polypectomy requires stringent removal of all stool to prevent a short-circuit current and an unintended burn, and a search is to be made for further polyps. The gastroenterologist reports no others are seen and the polyp in question has been easily removed by an electrocautery snare. Histologic analysis verifies a juvenile polyp, and no followup is planned.
1. You are called to the nursery where you are shown a burp cloth with loose clots of regurgitated blood. The newborn in question is sleeping quietly, with completely normal vital signs and no sign of tenderness or other bleeding when examined. You recall his mother presented with placenta previa. What do you do next?
2. At a two month well baby visit, his parents bring in a diaper double-bagged because of the foul odor. The stool is tarry and tests positive for occult blood, but the child appears particularly robust, having gone from a birth weight of 7 pounds 1 ounce to his current weight of 12 pounds 10 ounces. He is somewhat fussy and demanding of feedings, and his mother complains of getting no rest as she has to feed him hourly. Recently, her left breast has become quite sore and there is intense pain when he nipples. On examination, the infant is colicky, but there is no abdominal tenderness and his vital signs are also within normal limits with no adjunct signs of intravascular volume depletion. What is going on?
3. 3 year old presents with melena but no hematemesis, and no abdominal pain. How do you evaluate him?
4. Melena is usually indicative of upper GI bleeding. Indicate how this can sometimes be due to lower GI bleeding.
5. Red blood per rectum is usually indicative of lower GI bleeding. Indicate how this can sometimes be due to upper GI bleeding.
6. A 14 year old female has yet to show secondary sexual development which you have always attributed to excessive involvement with the school track team. However in the last 6 months her finishing times on the mile (her favorite event) have steadily lengthened from second best in the state to this week's race where she could not finish. She presents today complaining of loose stools, streaked with blood. How do you work up her illness?
7. A 3 year old boy presents to the emergency department passing bright red blood per anus. He is diaphoretic and tachycardic (120 supine, 140 upright) and complains of generalized abdominal pain. You are unable to localize tenderness but are comfortable that there is no rebound tenderness and he is not at risk of perforation. Placement of an NG tube to lavage his stomach is negative. By the time you have given enough crystalloid to replete his blood volume, his hemoglobin has dropped to 7 grams. Since his summer physical 2 months ago had included a hemoglobin of 12, you realize he has indeed lost a substantial portion of his blood volume over a short period of time. He is admitted to the hospital, where over the next two days as you wait for the stool culture results. He requires 250 cc transfusions daily to maintain his hemoglobin and you realize that the brisk bleeding continues. The stools remain bright red. What do you do next?
1. Gold BJ, et al. Helicobacter pylori Infection in Children: Recommendations for Diagnosis and Treatment. J Pediatr Gastroenterol Nutr 2000;31:490-497.
2. Fox VL. Gastrointestinal Bleeding in Infancy and Childhood. Gastroenterol Clin North Am 2000;29(1):27-66.
3. Arain Z, Rossi TM. Gastrointestinal Bleeding in Children: an Overview of Conditions Requiring Nonoperative Management, in Seminars. Pediatr Surg 1999;8(4):172-180.
4. Racadio JM, et al. Imaging and Radiological Interventional Techniques for Gastrointestinal Bleeding in Children. Sem Pediatr Surg 1999;8(4):181-192.
5. McKiernan PJ. Treatment of Variceal Bleeding. Gastrointestinal Endoscopy Clin North Am 2001;11(4):789-812.
6. Corredor J, Wambach M, Barnard J. Gastrointestinal Polyps in Children: Advances in Molecular Genetics, Diagnosis, and Management. J Pediatr 2001;138(5):621-628.
Answers to questions
1. A modified Apt test can be done. Take the loose clots and suspend them in a minimal amount of tap water (you need a visibly pink supernatant composed of free hemoglobin, hence the tap water to lyse the cells). Centrifuge the cells and to 5 cc of pink supernatant add 1 cc of 1% sodium hydroxide. Read in two minutes: adult hemoglobin turns yellow or brown, fetal hemoglobin remains pink. If the supernatant turns yellow, the blood is mother's, and every one can relax.
2. This infant has no sign that the bleeding originates with him, as bleeding sufficient to produce melena should leave him quite shocky. The history gives every sign that he has induced a mastitis (and nipple bleeding) in his mother, and she is able to compensate for the several ounces of blood loss that produced the melanotic stool. You counsel her on proper feeding and handling techniques to keep the infant satisfied without having to overfeed, and have his mother avoid feeding on the affected side until the inflammation subsides. At followup in a week, all are smiling.
3. ABC's first. He shows no sign of acute intravascular volume depletion, but looks a little pale and turns out to be mildly anemic, indicating a longer standing problem. Next, place an NG tube to look for upper GI bleeding but you find no evidence of this. Now what? There is evidence of bleeding in an area bathed in acid, but it is not the stomach (or the duodenum). If he is hemodynamically stable, you have time to pretreat with a histamine-2 receptor blocker to improve the yield of a Meckel's scan looking for ectopic gastric mucosa. This finds a hot spot in the lower mid-abdomen which the technician assures you is not tracer in the bladder. You contact your pediatric surgeon for minimally invasive removal of a presumptive Meckel's diverticulum with acid-secreting ectopic gastric mucosa.
4. The black color is due to blood exposure to acid. Acid fermentation can take place in the cecum. If this occurs and the transit time is relatively slow, bleeding in this area can present as melena. Bleeding from a Meckel's can also result in acid exposure in the lower GI tract.
5. The acid level in the stomach is low (possibly due to antacids and H2 blockers) and/or the bowel transit time is very rapid. Also, the bleeding may originate from the duodenum which does not expose the blood to acid if the pylorus is tight or the level of stomach acid is low.
6. The history has all the hallmarks of inflammatory bowel disease, but still the common things are more common. The physical examination shows no weight loss (but little net gain over the year), and she has a mild temperature elevation (100.5 degrees) and tachycardia (105) but no specific findings in the abdomen other than a mild increase in the amount of fluid and gas palpable in the small bowel and colon. Along with the CBC and ESR, you obtain a rectal swab for stool culture. There is no anemia, but the WBC count is slightly elevated and the ESR is 6. You are puzzled until the stool culture results return 2 days later, positive for Campylobacter. You call to discuss the results and find her new puppy had been ill the week before (dogs can both harbor and become ill from this organism), and the poor race performance actually arose because she was getting fed up with her coach (her father) and had been wanting to quit. Since she is still out of school with the cramping and diarrhea, you start her on erythromycin, offer to act as a go-between on the issue of changing sports, and annotate her chart to remind yourself to monitor for other signs of depression in the future.
7. As the negative gastric aspirates over the last 2 days indicate no UGI source, you prep him for colonoscopy to look for a lower GI bleeding site. GoLYTELY is used in hopes of diluting the bleeding as blood rapidly absorbs all light even in a thin film, and you anticipate much suctioning and lavage which will markedly extend the time for the procedure. As he will be under anesthesia anyway, you also obtain consent for EGD for completeness' sake. At endoscopy, the EGD study finds the pylorus is tightly shut as there is a large duodenal ulcer (not a simple erosion) with a visible vessel (an indicator of high risk of recurrent bleeding). With this you joyfully cancel the colonoscopy as being unnecessary, and chalk up the experience as a reminder that rapid transit times and the low acid production of early childhood can sometimes prevent the blood from encountering enough acid to turn to acid hematin or melena. Indeed, the higher the volume lost, the more acid is needed and the less likely the reaction. Unfortunately, as the finding was a therapeutic surprise, you are unprepared to address the ulcer in any invasive manner (sclerotherapy, heater probe, etc.) and have to return the patient to intensive care on an IV histamine receptor blocker and carafate and sufficient antacid to keep the pH of the gastric contents, measured every hour, above 6.5 (and well above the 4.5 activation level of pepsin). Preparations are made to return with the proper equipment the next day if he continues bleeding, only to find the bleeding stops with the procedure (and the drop in splanchnic pressures encountered under anesthesia), the current measures are more than sufficient, and no further transfusions are required. The patient makes a rapid and full recovery, with no recurrence in over 5 years (based on actual personal experience).