Fistulas are abnormal communications between two epithelialized surfaces. An intestinal fistula is an abnormal anatomic connection between a part (or multiple parts) of the intestinal lumen and the lumen of another epithelialized structure or the skin. Intestinal fistula includes many clinical entities. [1, 2] Because fistulas are widely defined, they are generally classified by anatomic, physiologic, and etiologic methods, all of which have treatment implications. 
In 450 BC, the ancient Indian surgeon Sushruta wrote the Sushruta Samhita, which described major abdominal operative procedures, including resection and reanastomosis of intestinal segments for repair of external intestinal perforations or enterocutaneous fistulas. An ingenious method for suturing the severed ends of intestine was described, in which the cut ends were apposed and a specific type of black ant was made to bite the apposed ends. Once the ants' pincers had closed, their heads were severed. Rigor mortis held the pincers in place, keeping the cut ends of the intestine in apposition. The ant heads and pincers, being organic matter, were digested in due course, not unlike the catgut used in present-day surgery.
In 350 BC, the ancient Greek surgeon Praxagoras created an enterocutaneous fistula to relieve intestinal obstruction. This technique was referenced in Western medical literature in 1899 by Werth, who advocated creation of an enterostomy to treat intestinal obstruction. In 1906, at a surgical society meeting in Chicago, John Finney mentioned enterostomy as a surgical option to treat severe postoperative ileus recalcitrant to other management.
In 1597, Fabricius Hildanus, a German surgeon, reported a patient with a Richter hernia who developed intestinal gangrene and formed an enterocutaneous fistula. The fistula continued to discharge intestinal contents for a period of approximately 2 months, then spontaneously healed. Hildanus was so amazed by this that he quoted several prominent townspeople as witnesses to this "miraculous cure" of an intestinal wound.
As a general rule in the treatment of intestinal fistulas, medical treatment and stabilization precede attempts at surgical intervention. In patients with all forms of enteric fistulas, sepsis is a major cause of mortality and must be treated aggressively. Surgical treatment is reserved for patients whose fistulas do not resolve with medical and nonsurgical therapy. Aortoenteric fistulas, which mandate emergency surgery when diagnosed, are an exception.
The aim of surgical intervention is to restore gastrointestinal (GI) tract continuity, as well as to repair and restore function to the other involved structures. One surgical procedure may not suffice; staged surgical procedures may be required. Treatment shou;d be individualized on the basis of the patient's overall medical condition and radiologic and intraoperative findings.
The pathophysiology of all forms of small-bowel fistulas is related to the exposure of nonintestinal tissue to intestinal contents because of the fistula. The intestinal bacterial flora leads to contamination and eventual development of sepsis. The local effect of intestinal fluid can be damaging or corrosive to the nonintestinal tissue, leading to breakdown, erosions, and loss of normal organ or organ system function.
A small intestinal fistula can be classified according to the anatomic structures involved, the etiology of the disease process leading to its formation, and its physiologic output (primarily for enterocutaneous fistulas). Anatomic classifications define the sites of fistula origin, drainage point, and whether the fistula is internal or external. Physiologic classifications rely on fistula output in a 24-hour period. Etiologic classifications (eg, malignancy, inflammatory bowel disease, or radiation) define the associated disease entity leading to the development of the fistula.
Each type of classification system carries specific implications regarding the likelihood of spontaneous closure, prognosis, operative timing, and nonoperative care planning. These classification schemes are not exclusive; if possible, all three methods should be used to classify each fistula.
The etiology of small intestinal fistulas is important for determining the subsequent treatment. The common mechanisms of intestinal fistula formation are outlined below.
Operative trauma is the most common cause of enterocutaneous fistula formation. Inadvertent enterotomies  and leakage from intestinal anastomoses result in leakage of intestinal contents with abscess formation. The abscess erodes through the abdominal wall, commonly at the surgical incision site or drainage site. This results in communication of the intestinal lumen with the skin surface, forming an enterocutaneous fistula (see the image below).
Intestinal anastomoses are susceptible to partial or complete dehiscence in the presence of impaired blood supply to the area, systemic hypotension, anastomotic suture line tension, perianastomotic infection, and diseased bowel segment anastomosis.
Exposure of the bowel to prosthetic mesh or a large abdominal defect can lead to wall erosion, resulting in enterocutaneous fistula. Intraperitoneal drainage tubes can erode into the intestinal lumen, leading to enterocutaneous fistula formation.
Penetration of the intestinal wall from a foreign body (eg, an ingested metallic object or a fish bone) can lead to enteroenteric fistula formation because of erosion into adjacent bowel loops. Fistulas do not commonly form this way. Similarly, penetrating trauma (ie, stab wound) rarely results in enterocutaneous or enteroenteric fistula formation. Nephroenteric fistula formation because of penetrating flank trauma is slightly more common.
Intestinal infections that erode through the wall cause an abscess and may lead to fistula formation between the intestine and an adjacent viscus, a solid organ, or the exterior of the body. Amebiasis, actinomycosis, tuberculosis, Salmonella, coccidiomycosis, and cryptosporidiosis can all result in periluminal abscesses and fistulas.
A solid-organ abscess, such as an amebic hepatic abscess, can erode into small bowel loops. Similarly, rupture of a perinephric abscess can lead to nephroenteric fistula formation. Diverticular and appendiceal abscesses can also lead to enteroenteric or enterocutaneous fistula formation. Appendicocutaneous fistulas are uncommon and occur most often after percutaneous drainage of an appendiceal abscess. In patients with Crohn enteritis, fistulas that occur in the right lower quadrant after an appendectomy usually arise because of the involved terminal ileum adhering to the healing abdominal incision. In these instances, the fistula rarely arises from the appendiceal stump.
Crohn disease leads to ulceration and chronic transmural inflammation of the intestinal wall. The serosa of a healthy viscus adheres to the diseased intestine. Adjacent bowel loops, bladder, colon, and vagina are commonly involved. Inflammation gradually progresses to microabscess formation and internal perforation in the ulcerated areas. The ulcerated areas penetrate through the bowel wall into the adjacent involved structure, leading to fistula formation. Enteroenteric, enterovesical, enterovaginal, and perineal fistulas develop frequently in patients with Crohn disease. Ulcerated bowel wall perforation may also lead to interloop abscess formation. The abscess may erode into adjacent bowel loops, resulting in fistula formation.
Radiation injury and malignancy
Long-term radiation injury to the intestine leads to ischemic changes in the intestinal wall. Erosions and dense adhesions between bowel loops develop, which can result in enteroenteric fistula formation. Similarly, degeneration of malignant tumors of the intestine or solid abdominal structures can lead to erosion into adjacent bowel loops, leading to fistulas.
Complete failure of the omphalomesenteric duct to obliterate results in an enterocutaneous fistula at the umbilicus (see the image below). This is a rare congenital form of enterocutaneous fistula. The appearance of feculent material at the umbilicus suggests the diagnosis, and surgical resection of the patent duct is performed.
Approximately 80-90% of all small intestinal fistulas occur because of operative intervention. Approximately 50% of small intestinal fistulas form because of inadvertent enterotomies in patients in which no intestinal anastomoses were performed. The remaining 50% are related to complete or partial disruption of intestinal anastomotic suture lines.
Approximately 10-20% of all small-bowel fistulas arise spontaneously in association with inflammatory processes, malignancy, radiation therapy, and infectious diseases. Of these 10-20%, Crohn disease accounts for 5-50%, cancer for 2-15%, peptic ulcer disease for 3-5%, pancreatitis for 3-10%, radiation therapy for 2-5%, and infections for 2-5%.
Surgical procedures that are commonly associated with postoperative fistula formation include reoperative procedures that require extensive lysis of adhesions, trauma surgery, mesh repair of ventral hernias, laparoscopic procedures, and surgery for cancer.
Crohn disease is the leading cause of spontaneous small intestinal fistulas, accounting for more than 50% of cases. Small intestinal fistulas develop in 20-40% of all patients with Crohn enteritis; half of these are enterocutaneous, and the remainder are internal fistulas to other abdominal viscera or organs.
Multiple factors predict spontaneous fistula closure and the mortality rate associated with intestinal fistulas. These factors guide the decision to institute conservative or surgical intervention. (See the image below.)
As recently as the early 20th century, the mortality associated with enterocutaneous fistulas was quoted to be as high as 70-100%. This was attributable primarily to sepsis, severe electrolyte and fluid imbalances, and malnourishment. With the advent of parenteral nutritional support and aggressive treatment of sepsis, the mortality and morbidity associated with fistulas decreased to 30-50%. In the current treatment of intestinal fistulas, a multidisciplinary approach has helped decrease the mortality to 15-37%.
The wide range of the reported mortality represents the heterogeneous etiology and the varied patient population in which intestinal fistulas develop. Early morbidity and mortality occur as a consequence of severe electrolyte imbalances that may develop in high-output fistulas. The mortality is primarily attributed to sepsis and multiple organ failure. This is more likely to develop in patients who are severely malnourished, have undergone radiation therapy, and have large abdominal defects and complex fistulas that are inadequately drained externally.
The spontaneous fistula closure rate is reported as 23-80%. The wide range reflects the multiple factors that affect fistula closure. The typical spontaneous closure rate is 30-35%, achieved through ensuring nutritional support and infection control. Ancillary surgical or interventional procedures in the form of drainage of abscess cavities are required in almost 20-30% of all fistulas that spontaneously close.
Of all fistulas that are likely to close spontaneously, 85-90% close within 4-6 weeks after the initiation of conservative management (ie, nutritional support, treatment of sepsis, control of fistula output). After 3 months of conservative treatment, fistulas that have not healed spontaneously will not heal. Most fistulas should be treated conservatively for 4-6 weeks. Patients with fistulas that show signs of improvement during this period may continue to be treated conservatively with the expectation of closure. Conversely, patients who show no signs of improvement should undergo surgical treatment. In patients who undergo definitive surgical closure for treatment of fistulas, the fistula recurrence rate is reported to be 13-34%.
In patients with intestinal fistulas, approximately 75-80% of all deaths can be directly attributed to sepsis and resultant multiple organ failure. The likelihood of spontaneous fistula closure in a patient with infectious complications is 16 times lower than that in a patient without associated infectious complications.
The rates of death and spontaneous fistula closure also correlate with fistula output. The chances of spontaneous closure are three times higher in patients with low-output fistulas than in those with high-output fistulas. Reports in the literature suggest mortality figures of 54-32% for patients with high-output fistulas and 26-6% for those with low-output fistulas.
The organ of fistula origin is an important predictor of spontaneous closure but is not as directly related to mortality. Jejunoileal enterocutaneous fistulas and lateral duodenal fistulas are both high-output fistulas with an associated spontaneous closure rate of only 18-20%; both are more likely to require surgical intervention for closure.
The type of nutritional support (eg, parenteral vs enteral) also affects outcome. A higher mortality, mostly secondary to sepsis, is reported in patients who receive parenteral nutrition. The rate also reflects that patients with high-output fistulas requiring parenteral nutrition are the most critically ill. Enteral nutrition is more feasible in patients with low-output fistulas and is associated with a lower mortality.
Serum albumin level is reported to be strongly predictive of both mortality and spontaneous fistula closure. One study found that an initial serum albumin level was less than 2.5 mg/dL in 55% of patients; of this 55%, those patients who had an associated complication carried a mortality of 64% and a spontaneous closure rate of 23%. Serum transferrin levels of greater than 200 mg/dL are also reported to be associated with a higher rate of fistula closure and survival.
Patients who undergo postoperative fistula treatment at the same institution where the primary surgical procedure was performed are likely to have a spontaneous closure rate of 40%. In contrast, the spontaneous closure rate for a patient who is referred from an outside institution is only about 20%. Similarly, the mortality figures for patients from the same institution versus a referred institution are reported as 36% and 26%, respectively.
Advanced patient age does not independently correlate with spontaneous closure rates. Despite this, the observed spontaneous closure rate in elderly patients is lower because of a higher mortality from associated comorbid conditions and poor nutritional and physiologic reserves.