Chapter 10. Abdominal Abscess and Enteric Fistulae

Definition and Etiology

Abscesses are well-defined collections of infected purulent material that are walled off from the rest of the peritoneal cavity by inflammatory adhesions, loops of intestine and their mesentery, the greater omentum, or other abdominal viscera. Abscesses may occur in the peritoneal cavity, either within or outside of abdominal viscera (extravisceral), as well as in the retroperitoneum.1 Most relevant to the surgeon are extravisceral abscesses that usually arise in one of two situations: (1) after resolution of diffuse peritonitis in which a loculated area of infection persists and evolves into an abscess and (2) after perforation of a viscus or an anastomotic breakdown that is successfully walled off by peritoneal defense mechanisms. More than 80% of intra-abdominal abscesses occur in the postoperative period, the majority of which occur after pancreaticobiliary or colorectal surgery and are usually related to anastomotic dehiscence.2,3 Occasionally, postsurgical abscesses result from infection of an intraperitoneal hematoma that develops following surgery. Less frequently, intra-abdominal abscesses are unassociated with previous surgery and are usually attributable to spontaneous inflammatory processes associated with a small, localized perforation, such as in appendicitis, diverticulitis, and Crohn's disease.3,4 Visceral abscesses are most commonly caused by hematogenous or lymphatic spread of bacteria to the organ. Retroperitoneal abscesses may be caused by several mechanisms, including perforation of the gastrointestinal (GI) tract into the retroperitoneum and hematogenous or lymphatic spread of bacteria to retroperitoneal organs, particularly the inflamed pancreas.

Pathophysiology of Abscess Formation

After bacterial contamination of the peritoneal cavity, a complex series of events is initiated that, under ideal circumstances, effects complete eradication of invading bacteria. The three major defense mechanisms in the peritoneal cavity are (1) mechanical clearance of bacteria via the diaphragmatic lymphatics, (2) phagocytosis and destruction of suspended or adherent bacteria by phagocytic cells, and (3) sequestration and walling off of bacteria coupled with delayed clearance by phagocytic cells.5 The first two mechanisms act rapidly, usually within hours. Egress of bacteria from the peritoneal cavity via the lymphatics is responsible for the early septic response due to bacteremia and initiation of the innate immune response to infection.

The initial peritoneal response to bacterial contamination is characterized by hyperemia, exudation of protein-rich fluid into the peritoneal cavity, and a marked influx of phagocytic cells. Resident peritoneal macrophages predominate early in the infection, but the rapid influx of neutrophils after a 2- to 4-hour delay makes them the predominant phagocytic cell in the peritoneal cavity for the first 48–72 hours.6 The combination of resident peritoneal cells plus the influxing into the peritoneum serves to propagate the initiation of the innate immune response, including the elaboration of inflammatory cytokines and the procoagulant response. In humans with severe intra-abdominal infection, peritoneal levels of tumor necrosis factor-alpha (TNF-α), interleukin (IL)-1, and IL-6 are higher than levels measured simultaneously in plasma.7,8 Haecker and colleagues reported that TNF-α and IL-10 levels are increased and reach 100- to 1000-fold that is observed in the plasma following appendiceal perforation. In adult patients, a correlation between the magnitude of the cytokine response and outcome in infected patients has been demonstrated in several clinical studies.9 Higher levels of circulating TNF-α and IL-6 have been recorded in patients who later die with intra-abdominal infection.7 Interestingly, elevated peritoneal persist even after systemic inflammatory response has abated. This suggests that during resolving peritonitis, there is compartmentalization of the response with local cytokine elaboration, thereby promoting local resolution of infection. Other cell types are likely important in the initiation of the local peritoneal response. Peritoneal mast cells and mesothelial lining cells have also been shown to be potent producers of a range of cytokines and procoagulants. Fibrin deposition appears to play an important role in this compartmentalization of infection, not only by incorporating large numbers of bacteria within its interstices10 but also by causing loops of intestine to adhere to each other and the omentum, thereby creating a physical barrier against dissemination. Fibrin deposition is initiated after the exudation of protein-rich fluid containing fibrinogen into the peritoneal cavity. The conversion of fibrinogen to fibrin is promoted by the elaboration of tissue factor by both mesothelial cells and stimulated peritoneal macrophages.11 In addition, generation of other inflammatory mediator molecules and components of the complement cascade (eg, C3a and C5a) further promotes the development of local inflammation. The net effect of these responses is the localization of the bacterial infection in the peritoneal cavity, wherein ultimate resolution can occur. However, a number of local factors thwart complete resolution and presumably establish the local environment for persistent infection and hence abscess formation. These include regional fibrin deposition that impedes phagocytic cell migration, factors that inhibit phagocytic cell function such as hemoglobin, particulate stool, low pH, and hypoxia. On the microbial side, polymicrobial flora of these infections as well as the near ubiquitous presence of Bacteroides fragilis and its unique capsular polysaccharide have been implicated in persistence of infection and abscess formation. Considered together, while the process of abscess formation represents a successful outcome of the peritoneal response to bacterial contamination of the peritoneal cavity, one is left with a residual infection that carries with it morbidity and potential mortality and must be actively managed.

Clinical Presentation and Diagnosis

Clinical Presentation

Diagnosis of an intra-abdominal abscess is based on clinical suspicion complemented by radiologic confirmation of the presence of the abscess. High spiking fevers, chills, tachycardia, tachypnea, and leukocytosis, associated with localized abdominal pain, anorexia, and delay in return of bowel function in the postoperative patient are the classic signs and symptoms associated with the presence of an intra-abdominal abscess. The presence of a well-localized tender mass on clinical examination is consistent with the presence of an abscess. However, there may be considerable variability in the clinical appearance of the patient with this infection, ranging from a relatively mild picture where the patient appears generally well but is “slow to recover” from his surgical procedure to those who manifest evidence of profound systemic inflammation. There may be no mass palpable on clinical examination. A number of factors may contribute to this variability, including patient factors such as age, immunocompetence, and concurrent use of antimicrobials, as well as abscess factors such location and size of the abscess and how well walled off the abscess is. For example, subphrenic abscesses can present with vague upper quadrant abdominal pain, referred shoulder pain, and occasionally hiccoughs but with no localized abdominal tenderness or palpable mass. By contrast, paracolic abscesses present with localized tenderness and may manifest as a palpable mass on abdominal examination. Pelvic abscesses may also cause local irritation of the urinary bladder causing frequency, or of the rectum resulting in diarrhea and tenesmus. Retroperitoneal collections, particularly psoas abscesses, can manifest as leg and back pain with muscular spasm and flexion deformity of the hip. In reality, with the ready availability of computed tomography (CT) scanning in most institutions, almost any deviation from the normal recovery trajectory in the postoperative period will prompt a CT scan and possible early detection of the abscess.

Diagnostic Tests

Imaging provides the definitive evidence of the presence of an intra-abdominal abscess. Abdominal plain films can be helpful in identifying air-fluid levels in the upright or decubitus positions, extraluminal gas, or a soft tissue mass displacing the bowel. In the postoperative patient, however, extraluminal gas may be present for up to 7 days. Overall, plain radiography may suggest the presence of an abscess, but other imaging modalities have essentially replaced plain films in the evaluation of intra-abdominal abscesses.

CT scanning has emerged as the radiological investigation of choice in the diagnosis of intra-abdominal abscess.12 With its ready availability, it has essentially supplanted abdominal ultrasound (US) as the main diagnostic tool in this setting, mainly because of its accuracy, but also because its functionality is not impaired in the setting of ileus, wound dressings, stomas, and the open abdomen. The accuracy of the scan is improved if contrast is used. IV contrast increases the accuracy of defining the presence of an abscess, while GI tract contrast helps to distinguish fluid-filled bowel loops from an abscess and in addition may detect the presence of an anastomotic leak. In a retrospective study that compared US and CT in diagnosing intra-abdominal abscesses, the sensitivity of US in 123 patients was 82% compared to 97% in 74 patients by CT, and the overall accuracy of US was found to be 90% versus 96% for CT.13 Criteria for identification of an abscess by CT have been well described and include identification of an area of low CT attenuation in an extraluminal location or within the parenchyma of solid abdominal organs. The density of abscesses usually falls between that of water and solid tissue.14 Other radiological signs of an abscess are mass effect that replaces or displaces normal anatomic structures, a lucent center that is not enhanced after the intravenous administration of a contrast medium, enhancing rim around the lucent center after IV contrast administration, and gas in the fluid collection. One of the major advantages of CT over US is the ability to detect abscesses in the retroperitoneum and pancreatic area. There are also some disadvantages to CT scanning. In the absence of contrast rim enhancement, gas or visible septations, CT cannot distinguish between sterile and infected fluid collections. Occasionally, there may be a solid-appearing collection that is really an abscess with a high leukocyte and protein content. Septations and other signs of loculated abscesses can often be better visualized with US than CT. Finally, CT scanning is sometimes unable to differentiate between subphrenic and pulmonic fluid, a relatively common situation in abdominal surgery.15 In these limited circumstances, US may be considered as a complement to CT imaging.

Other modalities include magnetic resonance imaging (MRI). While MRI can sometimes better delineate the extent of an abscess, particularly in relation to adjacent soft tissue structures such as muscles and major blood vessels, it does not clearly have advantages over CT scanning and its practicality may be limited in the sick surgical patient.16 One area where US and MRI may be relevant is in the investigation of the pregnant patient with abdominal pain.17 US is particularly useful when appendicitis/appendiceal abscess is suspected, and MRI may be useful when localization is less clear. The roles of radiolabelled compounds in the diagnosis of abdominal abscesses are limited at present.18

Management

The basic principles underlying the successful treatment of intra-abdominal abscesses are threefold:

1. Adequate resuscitation and support

2. Antimicrobial therapy

3. Source control/abscess drainage

Resuscitation and Support

In keeping with the variable presentation of patients with intra-abdominal abscesses, the initial approach to resuscitation and support will vary considerably. Attention to the ABCs (airway, breathing, circulation) while individualizing the intervention for each patient according to his/her deviation from normal physiology is appropriate. Particularly in the postsurgical patient, nutritional support should be considered. When feasible, oral nutrition should be given in preference to total parenteral nutrition. Some patients are able to ingest food and/or supplements by mouth, while others might require an enteral feeding tube, due to anorexia, precluding adequate ingestion of nutrients. Systematic review of the literature suggests that infectious complications and cost are reduced in critically ill patients receiving enteral nutrition compared to parenteral nutrition.19 One can presumably extrapolate to patients with intra-abdominal infection. When abscess formation occurs due to an anastomotic leak, there is a sense that this might preclude use of enteral nutrition. This concern is likely unfounded, unless there is profound ileus associated with the infection.

Antimicrobial Therapy

Considerations regarding antimicrobial use are based on the microbial flora recovered from the infections. Over the past decade, there has been increasing appreciation that there is an evolution of the flora with increasing severity of abdominal infection.20 For example, Table 10-1 shows the bacteriology of peritonitis in patients with community-acquired peritonitis and those with postoperative peritonitis. The major pathogens in community-acquired intra-abdominal infections are coliforms (esp. Escherichia coli) and anaerobes (esp. B. fragilis). As illustrated, while both are polymicrobial, postoperative peritonitis has a higher incidence of more resistant microbes. Aside from patients with postoperative peritonitis, other factors predict this shift in microbiology, including advanced age, severe physiologic derangement, immunosuppression, previous use of antibiotics, and residence in a health care institution in hospitals and nursing homes, etc. Guidelines have been developed recently by the Surgical Infection Society and the Infectious Diseases Society of America regarding the use of antimicrobial therapy in intra-abdominal infection.21 These authors have risk-stratified patients into three categories and provided recommendations for empiric antimicrobial regimens according to category. The three categories are (1) community-acquired infections of mild to moderate severity; (2) high-risk or severe community-acquired infections; and (3) health care–associated infections. Factors that dictate conversion from mild-to-moderate severity to high severity include severe physiologic derangement (eg, high Acute Physiology and Chronic Health Evaluation II [APACHE II] score), advanced age, or immunocompromised state. Table 10-2 shows the recommended agents according to this stratification. These guidelines are therefore readily applicable to decision making regarding patients coming into the hospital with abscesses, including processes such as appendiceal abscess or peridiverticular abscess. It is noteworthy that while enterococcus is frequently recovered in isolates in these infections, the evidence demonstrates no additional benefit to treating this microbe as part of empiric therapy. When possible, switchover to oral agents is appropriate. The duration of antibiotics should be 4–7 days, anticipating resolution of the clinical signs and symptoms during this period. Should there be no resolution by this time, reevaluation of the patient for the presence of persistent infection in the abdomen and elsewhere is appropriate.

Patients who present in the postsurgical period fall into the category of patients with health care–associated infection. In these patients, empiric therapy should include agents with expanded spectra against gram-negative aerobic and facultative bacilli, including meropenem, imipenem-cilastatin, doripenem, piperacillin-tazobactam, or ceftazidime or cefepime in combination with metronidazole. Table 10-3 shows the considerations regarding selection depending on local institutional microbial isolates. Empiric anti-enterococcal treatment should be given. Treatment of Candida with fluconazole when recovered from cultures and treatment of methicillin-resistant Staphylococcus aureus with vancomycin should be followed if the patient is colonized with the microbe.

Source Control

Source control is a term used to include all physical measures taken to control a focus of infection. Here we focus our discussion to abscess drainage, but adequate source control may also include debridement of necrotic tissue, surgical repair, resection, and/or exteriorization of the anatomic defect causing peritoneal contamination.22

Over the past two decades, percutaneous drainage of abscesses has become an established technique and a safe alternative to surgery. This evolution of care has not been based on a series of strong randomized trials showing equivalence or superiority of this approach. Rather, observational studies from a number of centers have shown it to be a safe effective alternative to surgical intervention, with equivalent success rates, comparable mortality (10–20%) and morbidity (~25%).23–25 Combined with other advantages of percutaneous approaches including avoidance of general anesthesia, lower costs, and the potential for fewer complications, it has now become the default approach to abscess management. Prerequisites for catheter drainage include an anatomically safe route to the abscess, a well-defined unilocular abscess cavity, concurring surgical and radiologic evaluation, and surgical backup for technical failure. Multiple abscesses, abscesses with enteric connections as seen with enterocutaneous fistulas, and the need to traverse solid viscera are not contraindications. Indeed, as the technique has evolved over several decades, the barriers to accessing unusually positioned collections have disappeared with the use of unconventional routes (transgluteal, transvaginal, transrectal) and the advent of new technologies including endoscopic US.26,27 Even the presence of septations and loculations has not precluded at least an attempt to use percutaneous drainage.28

Percutaneous drainage can be performed with US or CT guidance. CT provides for more precise identification of organs and bowel loops and is more accurate for planning of drainage route.15 Once the abscess is identified, initial diagnostic aspiration should be sent for Gram's stain and microbiological culture. The catheter used for drainage should be as small as possible for safety, yet large enough so that the tubing does not become obstructed. Most commonly used catheters range in size from 8 to 12F. With appropriate catheter placement, the abscess cavity typically decompresses and collapses. Irrigation of the catheter should be done once daily to ensure tube patency. As catheter drainage decreases, repeat CT scanning can be performed to evaluate for residual contents. If drainage increases over time or continues at a steady rate, the development of an enteric fistula must be suspected. This may not have been unexpected when the catheter was initially placed near a perianastomotic abscess or an abscess adjacent to some underlying pathological process. Potential complications of catheter placement include bacteremia, sepsis, vascular injury, enteric puncture, cutaneous fistula, or transpleural catheter placement.

Catheters should be maintained on closed drainage systems. There does not appear to be benefit to the use of suction or irrigation of these catheters, although flushing once per day with saline ensures patency. Patients should respond with defervesce of symptoms within 48 hours of catheter insertion. If they do so, a repeat CT scan is done at approximately 5–7 days to ensure shrinkage of the abscess. Criteria for removal of the drain include (1) clinical resolution of septic parameters, including patient well-being, normal temperature, and leukocyte count; (2) minimal drainage from the catheter; and (3) CT evidence of the resolution of the absence.

As noted previously, studies comparing outcomes of surgical and percutaneous drainage of intra-abdominal abscesses demonstrate comparable efficacy. In one study, patients were matched for age, abscess location, and etiology, and had similar APACHE II scores. There were no differences between percutaneous and surgical drainage in patient morbidity, mortality, or duration of hospital stay.24 Furthermore, initial percutaneous drainage of abscesses in the context of diverticular disease allowed for subsequent definitive operative resection and primary anastomosis in one rather than two operations. Another group retrospectively examined postoperative intra-abdominal abscesses after laparotomy. This study similarly demonstrated that use of either form of drainage resulted in similar cure rates for postoperative intra-abdominal abscesses.25

With clear demonstration of its efficacy when compared to surgical drainage, percutaneous drainage should be considered the preferred approach in source control of abscesses. Table 10-4 shows outcome of percutaneous drainage according to underlying pathological processes. In general, one should predict a successful outcome in patients with a single, well-defined abscess with no enteric communication. The presence of enteric communication per se does not reduce the likelihood of success as it is defined by the resolution of the infection. In a postoperative abscess, following drainage of the infection, the underlying anastomotic defect will usually close. In other settings, there may be a requirement for subsequent surgery to manage the underlying disease process such as diverticular disease or Crohn's disease. For example, in one study, approximately 75% of patients with large peridiverticular abscesses were drained percutaneously and then they went on to a single-stage sigmoid colectomy.28 Other circumstances such as fungal abscesses, infected hematomas, peripancreatic necrosis, or necrotic-infected tumor have a lower success rate for percutaneous drainage and early consideration for surgical intervention.29 CT features such as the presence of a “rind,” a sharp exterior margin, air-fluid levels, and septations do not predict outcome and therefore should not be determinants as to whether or not initial percutaneous drainage should be used.30 Finally, one should use clinical judgment as to the need for percutaneous drainage for small abscess (<5 cm diameter) such as those that might occur associated with acute diverticulitis, Crohn's disease, and interloop collections. These may well respond to antibiotics alone, and the use of percutaneous drainage may be meddlesome and potentially morbid.31

There are circumstances where percutaneous drainage should be considered contraindicated. Most important among these is the circumstance where peritoneal infection is not localized, such as in the early postsurgical period where an anastomotic leak leads to diffuse peritonitis. Abdominal CT scans performed in this scenario may demonstrate one or more discrete fluid collections. When there is diffuse peritoneal irritation on clinical examination, fluid collections distant from the anastomosis, or the presence of massive intraperitoneal air, surgical intervention is clearly indicated. Attempts to manage such situations with percutaneous interventions invariably lead to delayed definitive surgical management and adverse outcome.

Surgical Drainage

As stated previously, percutaneous drainage is the procedure of choice for the majority of intra-abdominal abscesses, with the caveats being those indicated. Specifically, when the infection is diffuse rather than localized, surgical intervention is clearly indicated. Second, when the content of the abscess is too thick for percutaneous drainage, an initial percutaneous attempt may be reasonable, but conversion to surgery early in the course is reasonable. Finally, when access is impossible, surgery is indicated. This last circumstance is increasingly rare.

The transperitoneal approach allows for examination of the entire abdominal cavity and allows for the drainage of multiple abscesses. Subphrenic abscesses and right subhepatic abscesses may also be approached by lateral abdominal incisions. Once abscess cavities are identified, they are entered and drained quickly to minimize spillage and contamination of the rest of the peritoneal cavity. The abscess cavity should then be widely opened. Specimens should be sent for Gram's stain and culture. Copious warm irrigation must be used at the end of the operation to properly cleanse the abdominal cavity. Closed-suction drains should be placed in dependent positions to reduce the risk of reaccumulation. In extremely contaminated cases, the incision may be left open and packed to prevent wound infection.

Introduction

A fistula is defined as an abnormal communication between two epithelial surfaces. Enteric fistulas may arise in a number of settings: (1) diseased bowel extending to surrounding epithelialized structures; (2) extraintestinal disease eroding into otherwise normal bowel; (3) surgical trauma to normal bowel including inadvertent or missed enterotomies; or (4) anastomotic disruption following surgery for a variety of conditions. The first two generally occur spontaneously, while the latter two occur following surgical procedures. For the surgeon, the latter two are generally more problematic, in part because they are iatrogenic complications of surgery, but also because their early management often requires treatment of the critically ill patient with sepsis.

While this chapter overviews general considerations regarding the pathophysiology and management of enteric fistulas, it focuses on postsurgical enteric fistulas, particularly fistulas to the skin, that is, enterocutaneous fistulas. In this particular patient population, the mortality rate remains high, between 3 and 22% in series dating back six decades, largely due to the frequent complications of sepsis and malnutrition (Table 10-5). Successful outcome requires a multidisciplinary team of health care workers, including surgeons, infectious disease specialists, intensivists, radiologists, nurses, enterostomal therapists, and nutrition specialists. Management of these patients must also take into account the psychosocial and emotional needs of the patient and his/her family through a prolonged and often complex treatment course.

One of the challenges in attempting to discern optimal management of these patients relates to the quality of the medical literature. Most reports are retrospective reviews of large case series emanating from referral institutions. Notwithstanding this shortcoming, these series provide general approaches to therapy, which help to guide treatment.

Classification

Fistulas involving the alimentary tract have traditionally been classified in three distinct ways: by the etiology responsible for their formation, that is, spontaneous versus postoperative, by the anatomy of the structures involved, and finally by the amount and composition of drainage from the fistula. Such distinctions may provide important prognostic information about the physiologic impact of fistulas and the likelihood that they will close without surgical intervention.

Spontaneous versus Postoperative

Enterocutaneous fistulas may be classified as either spontaneous or postoperative. Approximately three-quarters of fistulas occur in the postoperative setting, most commonly subsequent to procedures performed for malignancy, inflammatory bowel disease (IBD), or adhesive bowel obstruction.32 These fistulas become evident to the surgeon in a number of different ways: (1) They may occur in the early postoperative period as a septic complication of surgery, sometimes with catastrophic physiological deterioration. This is usually a result of uncontrolled diffuse intra-abdominal infection caused by anastomotic leakage, breakdown of enterotomy closure, or a missed enterotomy. (2) They may occur in a more delayed manner, following treatment of a postsurgical infection with percutaneous drainage of a deep abscess or opening of a superficial wound infection may reveal that an underlying connection to the GI tract as a cause. (3) They may occur very late after the surgery due to unanticipated injury to the GI tract. The development of a wound infection following use of mesh for hernia repair would fall into this category either through erosion of mesh into bowel or due to iatrogenic injury to the bowel as one attempts to debride infected mesh. Overly aggressive management of an open abdominal wound can also lead to intestinal injury and fistula formation. This complication has been reported to occur in up to 25% of patients during treatment with an open abdomen for abdominal sepsis.42

The remaining 25% of fistulas occur spontaneously, that is, without an antecedent surgical intervention. These fistulas often develop in the setting of cancer or inflammatory conditions. Fistulas occurring in the setting of malignancy or irradiation are unlikely to close without operative intervention. Inflammatory conditions such as IBD, diverticular disease, perforated ulcer disease, or ischemic bowel can result in fistula development.43 Of these, fistulas in patients with IBD are most common; these fistulas may close following a prolonged period of parenteral nutrition, only to reopen when enteral nutrition resumes.33

Anatomic Classification

Fistulas may communicate with the skin (external fistulas: entero- or colocutaneous fistulas) or other intra-abdominal or intrathoracic organs (internal fistulas). Internal fistulas that bypass only short segments of bowel may not be symptomatic; however, internal fistulas of bowel that bypass significant length of bowel or that communicate with either the bladder or vagina typically cause symptoms and become clinically evident. Identification of the anatomic site of origin of external fistulas may provide further information on the etiology and management of the fistula.

Oral, Pharyngeal, and Esophageal Fistulas.

Radical resections and reconstructions for head and neck malignancy may be complicated by postoperative fistulas in 5–25% of cases.44 Alcohol and tobacco use, poor nutrition, and preoperative chemoradiation therapy all contribute to poor wound healing and increase the risk of fistula formation. Failure of closure of the pharyngeal defect at the base of the tongue most commonly leads to fistula formation, and free microvascular flaps are the preferred method for closure. Brown and colleagues reported a significantly decreased postoperative fistula rate in patients who underwent free flap closure versus those with pedicled pectoralis flap closure, 4.5 versus 21%, respectively.45 Most esophagocutaneous fistulas result from either breakdown of the cervical anastomosis following resection of esophageal malignancy or following esophageal trauma. Less common causes of oropharyngeocutaneous or esophagocutaneous fistula include tuberculosis, laryngeal or thoracic surgery, trauma, congenital neck cysts, anterior cervical spine fusion, and foreign body perforations.46–48

Gastric Fistulas.

The most commonly reported procedure associated with gastrocutaneous fistula formation is the removal of a gastrostomy feeding tube, particularly in children. The duration of gastrostomy tube placement appears to be related to the likelihood of development of a fistula after tube removal, with nearly 90% of children developing a fistula when the tube had been in situ for more than 9 months.49 The rate of gastrocutaneous fistula following operations for nonmalignant processes such as ulcer disease, reflux disease, and obesity is between 0.5 and 3.9%.50 The recent rapid increase in the number of bariatric surgical procedures was anticipated to lead to an increase in the incidence of gastrocutaneous fistula following surgery for benign disease, as the rate of anastomotic leakage after gastric bypass surgery is 2–5%. One study has reported that approximately 10% of patients with staple line leaks go on to form chronic fistulas, making the overall rate less than 0.5%.51 Fistula formation following resection for gastric cancer remains a dreaded complication with significant mortality rates. Spontaneous gastrocutaneous fistulas are uncommon but can result from inflammation, ischemia, cancer, and radiation.

Duodenal Fistulas.

The majority of duodenocutaneous fistulas develop after distal or total gastric resections or surgery involving the duodenum or pancreas. Inadvertent injury to or intentional excision of a portion of the duodenum during surgery of the colon, aorta, kidney, or biliary tract may also result in fistula formation. Spontaneous cases resulting from trauma, malignancy, Crohn's disease, and ulcer disease account for the remaining duodenal fistulas.52,53 Prognostically, duodenal fistulas segregate into two groups: lateral duodenal fistulas and duodenal stump fistulas. Some authors have reported a decreased spontaneous closure rate with lateral duodenal fistulas when compared to that with duodenal stump fistulas.32,54

Small Bowel Fistulas.

Fistulas arising in the small bowel account for the majority of gastrointestinal-cutaneous fistulas, the majority of which (70–90%) occur in the postoperative period.33,34,55 Postoperative small bowel fistulas result from either disruption of anastomoses (either small bowel anastomoses or small bowel to colon anastomoses) or inadvertent and unrecognized injury to the bowel during dissection or closure of the abdomen. Operations for cancer, IBD, and adhesiolysis for bowel obstruction are the most common procedures antecedent to small bowel fistula formation. As noted previously, spontaneous small bowel fistulas arise from IBD, cancer, peptic ulcer disease, or pancreatitis. Crohn's disease is the most common cause of spontaneous small bowel fistula. The transmural inflammation underlying Crohn's disease may lead to adhesion of the small bowel to the abdominal wall or other abdominal structures. Microperforation may then cause abscess formation and erosion into adjacent structures or the skin. Approximately half of Crohn's fistulas are internal and half are external.56–58 Crohn's fistulas typically follow one of two courses. The first type represents fistulas that present in the early postoperative period following resection of a segment of diseased bowel. These fistulas arise in otherwise healthy bowel and follow a course similar to non-Crohn's fistulas with a significant likelihood of spontaneous closure. The other group of Crohn's fistulas arises in diseased bowel and has a low rate of spontaneous closure.

Appendiceal Fistulas.

Fistulas of appendiceal origin may result from drainage of an appendiceal abscess or post-appendectomy in a patient either without or with Crohn's disease.59,60 In the latter case, the fistula often originates from the terminal ileum, not the cecum. The inflamed ileum adheres to the abdominal wall closure and subsequently results in fistula formation.

Colonic Fistulas.

While spontaneous fistulas of the colon may result from inflammatory conditions such as diverticulitis, appendicitis and IBD, or from advanced malignancy, the majority of colocutaneous fistulas are postsurgical, usually secondary to anastomotic breakdown following colonic resection for one of these conditions. Preoperative radiation therapy reduces the risk of local recurrence and death from advanced rectal cancer and is an accepted practice.61 However, radiation therapy contributes to both spontaneous and postoperative colocutaneous fistulas. Russell and Welch authors reported a 31% incidence of breakdown of primary anastomoses performed in irradiated tissues with resulting sepsis or fistula formation.62

Physiologic Classification

Traditionally, fistulas have been classified into high-output (>500 mL/d), moderate-output (200–500 mL/d), and low-output (<200 mL/d) groups. Enterocutaneous fistulas cause the loss of fluid, minerals, trace elements, and protein, and, when improperly managed, they can result in profound irritation of the skin and subcutaneous tissues. Depending on the origin of the fistula and its anatomy, the amount of output and nature of the effluent may be estimated (Table 10-6). However, direct measurement of these parameters for an individual fistula allows for accurate replacement and an understanding of the physiologic and metabolic challenges to the patient. Classification of enterocutaneous fistulas by the volume of daily output provides information regarding mortality and has been used to predict spontaneous closure and patient outcome.32,63–65 In the classic series of Edmunds and associates, patients with high-output fistulas had a mortality rate of 54%, compared to a 16% mortality rate in the low-output group.32 More recently, Levy and colleagues reported a 50% mortality rate in patients with high-output fistulas, while those with low-output fistulas had a 26% mortality.63 Soeters and coworkers reported no association between fistula output and rate of spontaneous closure,33 while multivariate analysis by Campos and associates suggested that patients with low-output fistulas were three times more likely to achieve closure without operative intervention.65 The reason for these different closure rates most likely relates to the nature of the particular fistula, rather than the volume of output per se. If the fistula totally diverts flow, for example a pouting small bowel opening in the center of an open abdomen, it will be both high output and unlikely to close, without these two factors being causally related. By contrast, a defect at a small bowel anastomotic site with a long fistula tract and no local infection will likely be walled off by surrounding tissues and close spontaneously. These fistulas, while initially high output, will often close because of favorable local conditions. In essence, prediction of closure should be based on the local conditions, and particularly the nature of the fistula rather than the output. To the extent that the output often reflects the nature of the fistula, it will then be predictive.

Predicting Closure of Enterocutaneous Fistulas

Spontaneous closure of enterocutaneous fistulas without the need for major surgical intervention is clearly a desirable outcome for these patients. The precise probability of spontaneous closure is somewhat difficult to assess since the large series reporting management of fistulas are usually derived from specialty centers for fistula management and thus not only represent a biased sample but also reflect differences in referral practice. Thus, spontaneous closure has been reported to occur in 10–75% of patients.36,39,40,66,67 Nevertheless, a number of factors have been suggested to be predictive of failure of spontaneous closure of fistulas (Table 10-7). Some of these factors are modifiable, for example nutritional status, presence of local infection, and foreign bodies, while many do not include location, presence of an open wound, and the presence of distal obstruction. Knowledge of these factors should prove to be helpful in discussion of outcome with the patient and family members, as well as with the multidisciplinary team.

Risk Factors and Prevention of Enterocutaneous Fistulas

The majority of enterocutaneous fistulas arise in the postoperative period, often related to leakage of small bowel/colonic anastomoses or enterotomy closure. A number of factors have been associated with postsurgical enteric leaks. These can be divided into patient factors such as old age, immunosuppression, malnutrition, emergency surgery, and peritoneal contamination, and surgical factors such as emergency surgery, level of anastomosis, preoperative radiation, duration of surgery, blood loss, tension on anastomosis, inadequate blood supply to anastomosis, and technical error in suturing or stapling. Use of mechanical bowel preparation, anastomotic technique (stapled vs hand-sewn; single vs double layer), and omentoplasty has not been shown to influence anastomotic integrity. A recent meta-analysis in 2008 of 13 trials and 4601 patients showed no difference in the anastomotic leak rate when a mechanical bowel preparation was used compared to when it was not used in elective colon resection.68

Clearly, optimization of modifiable factors will serve to reduce anastomotic leak. In the elective setting, operations may be delayed to allow for normalization of nutritional parameters, thus optimizing wound healing and immune function. In emergency operations, the luxury of optimizing nutritional status preoperatively is not possible. Instead, emphasis should be on adequate resuscitation and restoration of circulating volume, normalization of hemodynamics, and use of appropriate antibiotic therapy.

Once a patient has been optimized preoperatively, attention is then turned to operative techniques to minimize the development of a fistula. Performance of anastomoses in healthy, well-perfused bowel without tension provides the best chance for healing. Testing of the rectal and sigmoid anastomoses with intraoperative air insufflations has been shown to reduce “radiologic” leak rate through guiding placement of additional sutures as needed.69 Careful hemostasis to avoid postoperative hematoma formation will decrease the risk of abscess, while inadvertent enterotomies and serosal injuries should be identified and repaired. A recent meta-analysis based on three randomized trials showed that omentoplasty to buttress a colonic anastomosis did not reduce the rate of postoperative radiological leaks, alter mortality or change the need for reoperation.70 However, while omentoplasty per se does not reduce the probability of anastomotic leakage, interposition of an omental flap to separate the anastomosis from the abdominal incision may lessen the probability of injuring the bowel during closure or of an enterocutaneous fistula should anastomotic leakage occur. A recent study pooling the data from five European randomized clinical trials studying rectal cancer care demonstrated that diverting stomas reduced the rate of symptomatic anastomotic leaks and improved overall survival but had no effect on cancer-specific survival.71 The differential survival was primarily attributable to early postoperative mortality. Proximal diverting colostomy or ileostomy may allow sufficient anastomotic healing prior to suture-line challenge by luminal contents.

Approach to Management

An organized treatment approach is of paramount importance to ensuring the optimal patient outcome. Table 10-4 lists overall mortality of patients presenting with enterocutaneous fistulas from a number of reports dating back six decades. Overall, the more recent studies appear to be associated with a lesser mortality rate, presumably a result of improvements in imaging, fluid resuscitation, antibiotic management, and intensive care support. However, the ultimate goals in treating patients with enterocutaneous fistulas are closure of the fistula with abdominal wall closure and return to baseline functioning level. Evenson and Fischer72 outlined five distinct phases of management that can be used to guide care of this patient population. These phases are discussed in detail and also summarized in Table 10-8.

Phase 1: Recognition and Stabilization

Identification and Resuscitation.

As noted in the Introduction, the clinical presentation of patients with enterocutaneous fistulas depends on the underlying pathophysiological process. Invariably, the patient who develops a postoperative enterocutaneous fistula will do well clinically for the first few days after operation. Within the first week, however, the patient may suffer delayed return of bowel function, as well as fever and leukocytosis, together suggestive of intra-abdominal infection. This setting will usually prompt a request for an abdominal CT scan that demonstrates a perianastomotic abscess. Percutaneous drainage for therapeutic management of the abscess will serve to confirm anastomotic disruption, either immediately or a few days later when there is evidence of enteric content. Occasionally, erythema of the wound develops and opening the wound reveals purulent drainage that is soon followed by enteric contents. In both these circumstances, the peritoneal host defenses have successfully walled off and contained infection. By contrast, in some patients, diffuse peritoneal contamination arising from a leaking anastomosis or enterotomy causes profound and rapid deterioration of the patient with diffuse abdominal tenderness, evidence of organ dysfunction, and hemodynamic instability. Usually, these patients exhibit signs of organ dysfunction in the days prior to their catastrophic deterioration, including reduced level of consciousness, tachycardia, and mild renal impairment. The diagnosis then becomes clear and management shifts from routine postoperative care to the management of a potentially critically ill patient. As with all critically ill patients, attention should turn to management of the ABCs. The patient with a localized collection or one that has necessitated into the wound can usually be managed on the ward, while the patient with a more significant septic response may require transfer to an intensive care unit (ICU) setting. In both scenarios, restoration of intravascular volume usually crystalloid is appropriate with or without inotropic support as determined by physiologic monitoring. A recent Cochrane Database Systematic Review showed no difference in outcome in critically ill patients managed with crystalloid versus colloid and therefore recommended crystalloid as the preferable resuscitation fluid.73 The initiation of broad-spectrum antibiotic therapy should occur early and be directed toward the most likely pathogens involved. Patients with postoperative peritonitis have increased probability of having multiresistant microorganisms and should receive broader-spectrum antibiotics. The consensus guidelines published by the Surgical Infection Society/Infectious Diseases Society of America address antimicrobial options for these severe health care–associated infections21 (see Tables 10-2 and 10-3).

Control of Sepsis.

Uncontrolled infection with the development of a septic response and the concomitant fluid imbalance and malnutrition are the leading causes of mortality in modern series of enterocutaneous fistulas. The leakage of enteric contents outside of the bowel lumen may lead to a localized abscess or to generalized peritonitis. Percutaneous management of localized abscesses accompanied by appropriate antibiotic therapy and supportive measure is usually sufficient to resolve infection in this subgroup. Diffuse peritoneal infection represents a much greater management challenge. In general, the generalized nature of the infection precludes successful therapy with percutaneous drainage and therefore an operative approach is indicated. Particularly in the early postoperative period, the surgeon should be wary of attempting to treat multiple intra-abdominal fluid collections observed on CT scan with percutaneous drains, when surgical intervention is required for definitive management.

Surgical Approach.

The goals of operative management of peritonitis are to eliminate the source of contamination, reduce the bacterial inoculum, and prevent recurrent or persistent infection. The operative technique used to control contamination depends on the location and the nature of the pathological condition in the GI tract.43 For patients progressing to diffuse peritonitis in the early postoperative period, the abdomen is usually reentered through the previous incision with the discovery of pus and enteric content. After aspiration of the fluid, an exploration to find the source of contamination is warranted. Anastomotic dehiscence/enterotomy should generally be managed by exteriorization of the affected bowel. Whether this is performed via a single stoma site or with separate stomas (ie, end stoma plus mucous fistula) depends on the specific scenario. Obviously, if one is able to exteriorize the intestinal defect, the likelihood of a postoperative enteric fistula is markedly reduced. It is attractive to hope that a surgically repaired enterotomy or leaking enterotomy might heal primarily, given the obvious simplicity of the procedure. However, this is rarely successful in the setting of diffuse peritoneal infection, and therefore this approach is not recommended. Reoperation after this misjudgment is fraught with potential difficulty, in that the surgeon is faced with the need to reoperate on the patient in the early postoperative period. This laparotomy is invariably more difficult, often associated with bleeding, further enterotomies, and a bowel that is extremely difficult to exteriorize. Under these circumstances, there should be consideration of a proximal defunctioning stoma if technically feasible. These cases are frequently the ones associated with inability to close the abdominal wall.

A number of anatomical circumstances may also preclude exteriorization of a leaking anastomosis. The principle of “defunction and drain” is appropriately applied in this setting. Most important among these is the rectal or sigmoid anastomosis where the distal end can be neither exteriorized nor closed. Unless the anastomosis is greater than 50% disrupted, it is reasonable to defunction with an ileostomy or a colostomy upstream and drain the site of the hookup. This approach is preferred as it increases the probability of future restoration of the GI tract. This is particularly true of leaks below the peritoneal reflection.74 If the anastomosis is almost completely disrupted, the surgeon is obliged to perform an end stoma and drain the pelvis, as the preserved anastomosis would stricture and preclude later stoma closure.

Control of Fistula Drainage and Skin Care.

Concurrent with drainage of sepsis, a plan to control fistula drainage and provide local skin care will prevent continued irritation of the surrounding skin and abdominal wall structures. Obviously, fistulas created following percutaneous drainage of abscesses are usually well managed by the drain itself. Indeed, the drainage of a local infection is frequently sufficient to permit closure of the fistula. For small low-output fistulas, dry dressing may suffice. In less controlled circumstances, particularly in the setting of the open abdomen, control of the effluent is not straightforward and must be managed aggressively. A skilled enterostomal therapist can often provide useful insight into these issues and should work in concert with a dedicated nursing team.75 The goals of therapy are to protect the skin, accurately monitor output, and minimize patient anxiety over effluent control. Use of a drainable wound pouch that is tailored to the size of the open wound is effective. This is often combined with some of colloid paste to protect skin and have an improved base on which to secure the stoma. Vacuum-assisted closure devices have been reported to aid in the care of these complicated wounds, including the promotion of closure. For example, Wainstein and coauthors reported promising results after reviewing their 10-year experience with it. In this study, fistula output was profoundly suppressed soon after commencing use of the device and spontaneous closure was achieved in 46% of patients. The use of a vacuum-assisted device was also found to reduce the frequency of wound dressing changes and improve dermatitis in all cases.76 These findings are consistent with most surgeons' anecdotal experience with vacuum treatment. Some authors have reported a small number of patients developing new enteric fistulas with the vacuum device. Therefore, some judgment is required in patient selection; presumably stabilized patients with some granulation overlying the exposed bowel may be appropriate.77,78

Reduction in Fistula Output.

While fistula output does not correlate with the rate of spontaneous closure, reduction in fistula drainage may facilitate wound management and decrease the time to closure. Further, reduced output enhances the ease of fluid and electrolyte management and may make local wound care easier. In the absence of obstruction, prolonged nasogastric drainage is not indicated and may even contribute to morbidity in the form of patient discomfort, impaired pulmonary toilet, alar necrosis, sinusitis or otitis media, and late esophageal stricture. Measures to decrease the volume of enteric secretions include administration of histamine antagonists or proton pump inhibitors. Reduction in acid secretion will also aid in the prevention of gastric and duodenal ulceration as well as decrease the stimulation of pancreatic secretion. Antimotility agents such as loperamide and codeine may also be effective.

As inhibitors of the secretion of many GI hormones, somatostatin, and octreotide were postulated to promote nonoperative closure of enterocutaneous fistulas. As recently reviewed, these agents did not accomplish this, although the data suggest that fistula output is reduced and time to spontaneous closure is lessened.79 This effect is more pronounced with somatostatin infusion than with its longer-acting analogue, octreotide. Infliximab, a monoclonal antibody to TNF-α, has been shown to be beneficial in inflammatory and fistulizing IBD.80 In a randomized trial of patients with chronic fistulas (duration >3 months), administration of infliximab resulted in a significantly increased rate of closure of all fistulas when compared to placebo.80 Some evidence suggests a role for infliximab in treatment of fistulas complicating IBD and its use has been reported to promote healing of persistent fistulae even in non-IBD patients.81 A number of other approaches to managing fistula output and promoting closure have been reported. These include endoscopic injection of fibrin glue into identified fistula openings,82 radiologically guided percutaneous Gelfoam embolization of the enteric opening,83 and the insertion of an absorbable fistula plug using a combination of percutaneous and endoscopic approaches.84 All three involve the “plugging” of the opening with a biological material, presumably with the expectation of tipping the local conditions toward healing. These low-morbidity techniques may therefore be considered as adjuvant considerations for fistula management. One would speculate that their greatest efficacy would be in the setting of a long tract, without epithelialization and with low output. Recently, endoscopic insertion of a silicone-covered stent across the fistula opening related to gastrojejunal leak following gastric bypass surgery has been described as a means of allowing early feeding and promoting fistula closure.51 One well-documented and potentially morbid complication of the stent use is its downstream migration with obstruction and erosion of the intestine. Clearly, no consensus regarding use of this approach has been achieved, given the small patient numbers described.

Nutritional Support.

Provision of nutritional support and time may be all that are necessary for spontaneous healing of enterocutaneous fistulas. Alternatively, should operative intervention be required, normalization of nutritional parameters will optimize patients in preparation for their surgery. Malnutrition, identified by Edmunds in 1960 as a major contributor to mortality in these patients, may be present in 55–90% of patients with enterocutaneous fistulas.33 Patients with postoperative enterocutaneous fistulas are often malnourished due to a combination of poor enteral intake, the hypercatabolic septic state, and the loss of protein-rich enteral contents through the fistula and via the open abdominal wall. The optimal route of nutrition in the management of enterocutaneous fistulas has not been critically studied. Parenteral nutrition has long been the cornerstone of support for patients with enterocutaneous fistulas.33,85–87 This, in part, is related to the fear that early enteral feeds will exacerbate the fistula through increasing output and also that enteral feeds may not be an adequate form of nutritional support. Parenteral nutrition can be commenced once sepsis has been controlled and appropriate intravenous access has been established. Transition to partial or total enteral nutrition has been advocated in recent reports to prevent atrophy of GI mucosa as well as support the immunologic and hormonal functions of the gut and liver. Additionally, parenteral nutrition is expensive and requires dedicated nursing care to prevent undue morbidity and mortality from line insertion, catheter sepsis, and metabolic complications. Thus, attempting enteral feeding is appropriate in most fistula patients. As achieving goal rates of enteral feeding may take several days, patients are often maintained on parenteral nutrition as tube feedings are advanced. Enteral feeding may occur per os or via feeding tubes placed nasogastrically or nasoenterically. Enteral support typically requires 4 ft of small intestine and is contraindicated in the presence of distal obstruction. Drainage from the fistula may be expected to increase with the commencement of enteral feeding, although this does not uniformly occur and is often dependent on fistula location and size of the fistula defect; however, spontaneous closure may still occur, often preceded by a decrease in fistula output. When parenteral and enteral nutrition are both options, the latter is preferred. It is far less expensive, safer, and is easier to administer (particularly if the intent is to manage the patient as an outpatient). A meta-analysis by Gramlich et al19 indicated that ICU patients receiving enteral feeds have a lesser infection rate than those receiving parenteral feeds.

In patients with high-output proximal fistulas, it has been suggested to provide enteral nutrition by a technique called fistuloclysis. In fistuloclysis, an enteral feeding tube is placed directly into the matured high-output fistula.88 Teubner et al reported on their experience using fistuloclysis in 12 patients before reconstructive surgery.89 Eleven of twelve patients were able to discontinue parenteral support and nutritional status was maintained until surgery in nine patients (19–422 days) and for at least 9 months in the two patients who did not undergo operative intervention.89 Of note, surgeons in this study also reported improved bowel caliber, thickness, and ability to hold sutures in patients who had received enteral nutrition.89 Other measures such as the use of recombinant human growth hormone (rGH) on fistula patients have been examined. While able to promote intestinal mucosal epithelial cell proliferation; increase levels of total proteins, albumin, fibronectin, and prealbumin; and transfer and reduce nitrogen excretion, its clinical role has not been clearly defined.90

Psychological Support.

Patients who develop postoperative enterocutaneous fistulas require considerable psychological support. They have sustained a major complication of surgery and are frequently faced with prolonged postoperative stay, excessive abdominal discomfort, and potentially one or more additional surgical interventions. In aggregate, all of these factors lead to psychological distress for patient and their families and should be addressed once the acute disease is dealt with.

Phase 2: Investigation

Once the patient has been stabilized with control of sepsis and commencement of nutritional support, early radiological investigation may be of value. Abdominal CT scanning with GI contrast with help to discern whether there is residual local infection that requires drainage, to localize the level of the fistula and the amount of contrast flowing beyond the defect, and occasionally whether there is distal obstruction. Fistulograms down drainage tracts will elucidate the length, course, and relationships of the fistula tract. If the fistula is spontaneous, the nature of the local pathological process from which the fistula arises may be determined. In the setting where the mucosal bud of the fistula is readily observed in the center of an open abdomen, aside from a CT scan to rule out distant infection, little further early imaging is required. Because patients with enterocutaneous fistulas are frequently referred to larger centers for management, it is essential that all notes, particularly operative notes, be obtained from the referring hospital. Personal communication with the surgeon may further elucidate other factors in the patient's disease that are not readily evident from the notes.

Phase 3: Decision

Spontaneous closure of fistulas restores intestinal continuity and allows resumption of oral nutrition. As noted previously, the rate of spontaneous closure varies considerably from series to series, with an average of approximately one-third of patients. This wide range likely represents patient selection in the various series, and in particular whether the series emanates from a referral center where the patient population tends to be more complex. A number of factors predict spontaneous closure. These are listed in Table 10-7. One might consider two case scenarios to illustrate these points. A long, narrow fistula tract originating from a small leak in a colonic anastomosis with no evidence of distal obstruction and a well-drained perianastomotic abscess is almost certain to close spontaneously. By contrast, a small bowel defect revealing itself as a mucosal bud in the middle of an open abdomen is unlikely to heal as the tract is short and epithelialized, in essence mimicking a stoma. Fistulas associated with IBD often close with nonoperative management only to reopen upon resumption of enteral nutrition. These fistulas should be formally resected once closed to prevent recurrence. Fistulas in the setting of malignancy or irradiated bowel are particularly resistant to closure and would suggest the need for earlier operative intervention.

Most authors agree that once resuscitation, wound care, and nutritional support are established, 90–95% of fistulas that will spontaneously close typically do so within 4–8 weeks of the original operation.25,85 In the absence of closure, there should be consideration of surgical closure. Like any surgical procedure, weighing of the risk and benefits of surgical intervention is critical prior to proceeding to operation. This is particularly relevant in this patient population where the surgical procedure is a major one and has a finite risk of recurrence. Some patients are perfectly well, are tolerating a regular diet, and have fistula effluent that is trivial in volume and requires only coverage with dry gauze. The potential risks of a major operation in this type of patient might outweigh the ultimate benefit. The timing of elective operative intervention for fistulas that are unlikely to or fail to close is extremely important. Early operation is only indicated to control sepsis not amenable to percutaneous intervention. These early procedures are typically limited to drainage of infected fluid collections and drainage, defunctioning, or exteriorization of the defect.

There is some controversy in the literature as to how long one should wait before attempting definitive elective closure of enterocutaneous fistulae. Very early closure appears to be contraindicated because the patient condition is generally not optimized. Further, from a technical standpoint, adhesions tend to be dense and vascular, therefore rendering the procedure difficult. In one retrospective study, Keck et al observed that operative difficulty and denser adhesions leading to inadvertent enteromies were more common when patients were taken to surgery for reversal of a Hartmann's procedure before 15 weeks compared to after.91 Poor outcome when surgery is performed in the 2-week to 3-month window has been report by several groups.38,92 At least two reports suggest that a very long delay before definitive surgery (>36 weeks) might adversely affect outcome.41,93 It is generally recommended that definitive surgery be considered in the window of 3–6 months after the patient is stabilized from the initial recovery from the procedure that lead to the fistula formation. Various factors will influence where, in this interval, surgery is performed. Patient factors such as nutritional status, ease of managing the fistula, and family support may influence decision making. Some authors talk about the “soft” abdomen and prolapse of the fistula as being a valuable clinical signs that peritoneal conditions are reasonable to proceed with surgery.37 On occasion, there is intense pressure from the patient and family to reoperate and “fix” the fistula during this early period. This approach should be resisted.

Phase 4: Definitive Management

Operations repairing enterocutaneous fistulas may be complex and often lengthy. In addition to repairing the fistula, many of these patients require complex abdominal wall closures. Before definitive management, the patient should have achieved optimal nutritional parameters and be free of all signs of sepsis. Through careful management of fistula drainage, a well-healed abdominal wall without inflammation should be present.

Consent.

As for all operations, the patient should be fully apprised of the nature of the procedure and its potential for complications. Connolly and colleagues reported a very high incidence of complications following intestinal reconstructive surgery (82.5% of procedures) when one considered postoperative nosocomial infections including surgical site infections, respiratory infections, and central line sepsis together with postsurgical myocardial dysfunction, GI bleeding, and deep vein thrombosis.94 In discussions with patients and their families, the unique difficulty of these procedures should be raised, pointing out the potential for adhesions and therefore inadvertent injury and excessive bleeding. The fistula recurrence rate is also significant with reported rates up to 33% (see Table 10-5), depending on the individual circumstance. The patient and relevant family members should know that the procedure may be prolonged and may require an ICU stay in the postoperative period. Some of the anxiety of the patient may be related to mistrust of physicians in general following a previously complicated operation. Clearly, the sensitive nature of reoperation for prior complications requires a strong physician-patient relationship to minimize patient anxiety prior to the planned procedure.

Patient Preparation.

It is critically important for the operating surgeon to fully understand the nature of the prior surgeries. Reviewing the previous operative notes as well as speaking with the original surgeon will consolidate one's knowledge of the initial pathological process and the precise anatomy to be corrected in the reoperative setting. One should also be very liberal about using preoperative contrast imaging or endoscopy to completely define the anatomy. In the hypothetical case of reoperation after a colonic anastomotic dehiscence, the need for definition of the anatomy varies according to the initial source control procedure. A prior operation consisting of exteriorization of an end colostomy with nearby mucus fistula or exteriorization of the disrupted anastomosis is a circumstance where investigation is probably unnecessary. In preparation for closure of a Hartmann's procedure, the rectal stump should be routinely investigated by endoscopy. This may help with planning of the operation as well as locating the stump at surgery. Closure of a defunctioning ileostomy or colostomy should also be preceded by investigation of the downstream anastomosis. This is intended to rule out the presence of a stricture or persistent defect at that site, both of which would alter surgical approach. Finally, contrast studies are essential when complex fistulas exist and are to be treated by reoperative surgery.

The general principles related to preparation for any surgery should be applied to reoperation. These would include optimization of the general medical status of the patient, administration of subcutaneous heparin and/or other antithrombotic strategies, and initiation of measures aimed at reducing postoperative infectious complications. Orthograde intestinal lavage by mouth as well as distally via the defunctioned limb has been recommended for mechanical preparation of the bowel. However, the evidence underlying this recommendation is limited and, in fact, recent studies show that mechanical bowel preparation for elective colon surgeon does not improve outcome and may have some deleterious effects.69 Our practice is to forego the use of mechanical prep unless reconstruction involves passage of stapling device transanally. Clearance of inspissated mucus in the rectal stump with an enema may facilitate advancement of the stapler proximally. Finally, prophylactic intravenous antibiotics with broad-spectrum coverage of both facultative gram-negative enterics as well as anaerobic bacteria are indicated. Consideration of coverage of resistant microbes should be made.21

Operative Intervention.

Patients should be positioned to permit optimal exposure to the field of surgery, to take into account potential requirements for extension of the operative field, and to facilitate optimal reconstruction of the GI tract and/or drainage of the operative field. In the majority of situations, the supine position is adequate. Concomitant lithotomy positioning is often helpful, particularly when reconstruction involves the left colon or rectum, where transanal access for endoscopy or stapling may be useful. When reoperation involves the upper GI tract, left lateral decubitus positioning will allow an initial thoracoabdominal incision or extension of an abdominal incision into the chest.

Careful planning of the location and type of incision are mandatory prior to making the initial incision. It is preferable to enter the peritoneal cavity through a previously unoperated area of the abdominal wall, thereby avoiding the areas where the most intense adhesions would be expected, that is, beneath the previous abdominal wall incision and in the region of the abdomen where the inflammation might have been the most severe. Inadvertent enterotomy is relatively common during reoperation, occurring in approximately 20% of patients, and is associated with a higher rate of postoperative complication and a longer postoperative hospital stay.95 In addition, it is a frustrating beginning to an often long and tedious operation.

The use of the midline incision, beginning with entry either cephalad or caudad to this initial incision through an unoperated field is the most common approach to reentering the abdomen. This approach provides broad access to the peritoneal cavity with opportunity for extension and is also readily closed. Other approaches may include unilateral or bilateral subcostal incisions, transverse incisions, flank incisions, or thoracoabdominal incisions. In general, these should be considered when a specific area of the abdomen is operated on, because they generally afford less access to the overall peritoneal cavity. When placing new incisions, care should be taken not to render intervening tissue bridges ischemic. This might occur when a midline incision is placed adjacent to a previous paramedian incision. It is preferable to use the previous paramedian incision with extension into the midline above or below. When the fistula opening is in the center of a reepithelialized section of the abdomen with no underlying fascia/muscle, one should preferably enter the abdomen as described above, either cephalad and caudad to the previously operated area. When this is not possible, one should consider placing the initial incision along the line of the fascial edge, rather than though the reepithelialized portion. In the latter operative field, the skin may be very adherent to the underlying bowel, therefore increasing the chance of bowel injury. This is particularly true when there is retained mesh, which may have contributed to fistula formation in the first place.

Upon entering the peritoneal cavity, adhesions between the anterior abdominal wall and the underlying omentum and bowel must be released. By 3–6 months following the initial surgery, adhesions are generally relatively filmy and readily divided using scissor or cautery dissection. Gentle traction on the bowel with countertraction on the abdominal wall will facilitate exposure of the appropriate tissue plane for division. A similar approach is appropriate for dense adhesions, with some surgeons preferring knife dissection. During this dissection, it may be necessary to leave patches of abdominal wall (peritoneum with or without fascia) or even mesh adherent to bowel to avoid enterotomy. It is also noteworthy that enterotomies may be caused by traction on the bowel due to retraction on the abdominal wall. Clearance of the fascial edges along both sides of the entire incision is necessary to achieve adequate and safe closure of the abdominal wall.

Having successfully entered the abdominal cavity, one faces varying degrees of interloop adhesions. The degree to which these must be lysed depends on the particular operation to be performed. When one is operating on the colon for the purpose of stoma closure or reestablishment of colonic continuity, there is generally little need to exhaustively take down small bowel adhesions. The fact that the patient has been tolerating a normal diet preoperatively provides ample evidence that the small bowel adhesions are not of physiological significance. While not having to lyse all adhesions, it is necessary, however, to free small bowel loops from their attachments to the colon so that the latter might be adequately mobilized to permit easy closure or anastomosis. When operating to close a small bowel stoma or to correct an enterocutaneous fistula, when possible, one should consider more comprehensive lysis of adhesions, along the entire length of the small bowel, but in particular the distal small bowel. The presence of a stoma or fistula may serve to defunction a distal small bowel adhesive obstruction prior to surgery and may therefore preclude its recognition. The presence of a distal obstruction following upstream anastomosis could prove catastrophic in the postoperative period.

Adhesiolysis varies considerably in its degree of difficulty. Even when the reoperation is appropriately delayed from the initial operative procedure and vascularized adhesions are no longer present, the number and density of residual fibrous adhesions may still be significant and represent a significant technical challenge. As described for opening the peritoneal cavity, good lighting of the operative field, excellent surgical assistance, and a dose of patience are absolute requirements for this part of the operation. Two experienced surgeons working together facilitates adhesiolysis. During lysis of adhesions, one should also be wary of encountering previous anastomoses. Adhesions may be particularly tenacious in these areas, particularly when the prior anastomosis was performed using a stapled technique. For side-to-side functional end-to-end stapled anastomoses, the crotch of the anastomosis may be mistaken for intense adhesions. Failure to recognize this may result in inadvertent enterotomy and the attendant increased morbidity.

When surgery has been timed appropriately, one usually finds that the dissection distant from the fistula to be reasonably straightforward. As one approaches the fistula site, it becomes increasingly tedious with multiple adherent loops of bowel. We recommend that the fistula be addressed relatively late in the dissection, after most of the small bowel has been mobilized. This minimizes inadvertent injury to loops of bowel uninvolved in the fistula.

Several of the large case reviews address surgical technique and risk of recurrence.37,38,40,41 In general, it appears to be preferable to locally resect the segment of small bowel bearing the fistula rather than simply closing the intestinal opening. This may represent a biased finding since the instances where simple closure was used correlated with the finding of an abdomen with impossibly dense adhesions, therefore precluding mobilization and resection. Under these latter circumstances, one might consider the addition of a temporary proximal defunctioning stoma.

In the elective surgical setting, stapled anastomoses have been shown to be equivalent to hand-sewn anastomoses in terms of anastomotic dehiscence.96 By contrast, for closure of enterocutaneous fistulas, hand-sewn appears to be the preferred approach to performing the anastomosis following resection. Whether single layer versus two layers of sutures or running versus interrupted stitching should be used has not been systematically addressed. Frequently, the chronically defunctioned bowel is atrophic, line-walled, and stiff. Under these circumstances, the stapling devices are unable to accommodate the pathological nature of this bowel, where hand sewing can better accommodate differences in size, thickness, and compliance of the intestine.

Wrapping of the anastomosis with omentum has been examined as a means of preventing anastomotic leakage but has not proven to be effective.71 However, placement of a flap of omentum between the fresh anastomosis and the abdominal wall closure may minimize recurrence of fistulization. Some have advocated the placement of a decompressive gastrostomy and/or the placement of a feeding jejunostomy, both of which may aid in the postoperative care of patients undergoing procedures of this scale.

As the cumulative experience with complex laparoscopic procedures has increased, several groups have reported laparoscopic approaches to enteric and enterocutaneous fistulas.97–102 The largest of these series reported 73 procedures in 72 patients, 20% of which were enterocutaneous fistulas.101 The authors reported a mean operative time of 199 minutes with a 4.1% conversion rate.101 Because surgical procedures for the management of enteric fistulas are generally complex ones, a laparoscopic approach would seem appropriate only in the hands of a skilled and experienced laparoscopic surgeon and only in selected circumstances.

Abdominal Wall Closure.

After the fistula has been appropriately managed, one is left with closure of the abdominal wall. The complexity of this aspect of the operation varies depending on the preoperative state of the abdominal wall. Closure may be straightforward when the enterocutaneous fistula is along a previous drain tract or through necessitation of an abscess through an abdominal wound. By contrast, when the prior patient management involved an open abdomen approach with the fistula draining from the center of the wound, patients may present with large ventral hernias that are not amenable to simple fascial closure. In advance of surgery, it is essential that the surgeon consider management of abdominal wall a significant part of the procedure and reflect upon the various surgical options. Included in these preoperative deliberations should be the proactive involvement of a plastic surgeon to aid in the assessment of options and to potentially prepare him/her for involvement in the operation. Table 10-9 outlines the various approaches. Prior to beginning abdominal wall closure, it is desirable to debride/remove any residual infected foci, including chronically infected suture material and previously placed infected mesh. One should also attempt to position the intestinal anastomosis away from the closure and, if possible, to interpose omentum between the anastomosis and the abdominal wall. Finally, it is generally considered that, in the setting of GI surgery where there is contamination of the surgical field, the use of nonabsorbable permanent mesh is contraindicated as it is associated with an increased risk of infection and refistulization.103

When no defect or a small defect in the fascia exists, primary closure is usually achievable although there may be some mild tension on the closure. This is, in part, related to the stiffness of the abdominal wall attendant with repeated abdominal surgery. In these circumstances, relaxing incisions placed in the aponeurosis of the external oblique muscle approximately 2 cm lateral to the edge of the rectus muscle may minimize any tension. Polydioxanone, a slowly absorbable monofilament suture material, appears preferable as it is equivalent to nonabsorbable monofilament suture in terms of recurrent hernias but has less wound pain and sinus formation.104 Various closure techniques have been proposed when primary fascial closure is not possible.103,105–107 There has been increasing enthusiasm regarding the use of the component separation technique as a means of achieving abdominal wall closure without prosthetic material.105–107 In brief, this approach involves the separation of the external oblique and internal oblique muscles bilaterally plus division of the posterior rectus fascia. Together, these accomplish approximately 12-, 22-, and 10-cm advancement of the upper, middle, and lower thirds of the abdomen, respectively.105 This approach has been reported for abdominal wall closure after trauma surgery, in patients with sepsis managed with the open abdomen and in patients with enterocutaneous fistulas.

Wind and colleagues examined the application of this technique in the presence of a contaminated abdominal wall defect, including during closure of an enterocutaneous fistula and/or stoma.106 This study reported the feasibility of this approach in terms of achieving abdominal wall closure but noted considerable morbidity, including wound seromas, wound infections, and hematomas as well as recurrent abdominal wall hernias in approximately 22% of patients. Recurrence of the enterocutaneous fistula occurred in 25% of patients. In a small percentage of patients, the use of absorbable mesh was combined with the component separation technique, because the advancement of the abdominal wall alone was not sufficient to cover the defect.

Finally, absorbable prosthetics may be considered for management of the defect. Synthetic meshes such as polyglactin effect good initial coverage but have the anticipated long-term consequence of incisional hernia formation.94 As an alternative, biological prostheses including porcine collagen mesh and acellular dermal matrix have been suggested with the potential advantage of increased resistance to infection and reduced late incisional hernias. These outcomes have not been uniformly achieved when used in the treatment of fascial defects following repair of enterocutaneous fistulas.94,108

In summary, management of the abdominal wall following reoperative surgery in these patients may be a considerable challenge. The major objective is to prevent recurrent fistula formation and minimize postoperative infection. Prevention of late ventral hernia formation is a secondary goal. Involvement of a surgical team with expertise in the options, including the use of the component separation technique, would appear to broaden the clinical options for the patient.

Phase 5: Postsurgical Phase

The postoperative period can be divided into two parts: the early postsurgical recovery period and the later rehabilitation and convalescence phase. The former of these periods can be somewhat complex as postoperative complications are frequent, with up to 80% of patients having one or more complications.94 In particular, these patients have a significant incidence of postoperative infection, both at the surgical site and at distant sites including lung and central venous lines. As shown in Table 10-5, the incidence of recurrent fistulization following surgery is considerable and is associated with prolonged hospital stays and repeat admissions to the ICU as well as repeat interventions. Brenner et al reported that recurrence of the enterocutaneous fistula in the postoperative period was the strongest predictor of mortality, invariably due to the development of overwhelming sepsis and organ failure.41 Mortality is related to the presence of preoperative comorbidities.109 Short of death, the recurrence of enterocutaneous fistula following surgery represents a major complication. Among those who survive this recurrence, only 50–66% go on to further surgery and successful closure, while the remainder live with a chronic fistula.38,41 A number of factors predict recurrence (Table 10-10).

By the time their fistulas have been surgically closed, these patients have often been undergoing medical care, usually both as inpatients and outpatients for several months following the initial development of their enterocutaneous fistulas. By the end of this period, which may have included prolonged in-hospital stays, multiple surgical and radiological interventions, frequent visits to health care facilities as outpatients, and an overriding focus on their medical disability, patients are invariably physically deconditioned and emotionally fatigued. The impact on the long-term quality of life, as measured by objective questionnaires, even in those treated, continues to be lower than matched controls especially if there is a concurrent medical illness.110 Physical and occupational therapists play a role throughout each patient's hospitalization, but their efforts become even more important during the healing phase as the focus shifts to reintroducing the patient to normal activities of daily living. Involvement of case management staff early in the patient's course will identify obstacles to the patient's successful reintroduction to an active lifestyle, while use of psychiatric consultation-liaison services will identify and address issues of depression and adaptive disorders. Finally, active involvement by the senior surgeon responsible for the patient's care to ensure clear communication to the patient and the family during what is invariably a prolonged convalescence and rehabilitation period is essential. Optimally, this physician-patient relationship would have begun early in the patient's illness and would continue through till complete recovery occurs.

Conclusion

Enteric fistulas, occurring spontaneously or in the postsurgical period, represent a significant management challenge. This chapter has focused predominantly on the postsurgical enterocutaneous fistulas, which may result in both morbidity and occasionally mortality for the patient. The care in these patients may be complex and has led to the establishment of specialized intestinal failure units, aimed at optimizing outcome. General principles of care include (1) early recognition and stabilization of patients with fistulas combined with control of sepsis and provision of nutritional support; (2) investigation of the anatomic and etiological characteristics of each fistula, thus providing information about the likelihood of spontaneous closure or need for operative management; (3) decision making regarding the approach to management, including the involvement of a multidisciplinary team, will provide the best possibility of resolution of the fistula; (4) definitive surgical therapy in a controlled setting; and (5) postoperative care including physical rehabilitation and emotional support, which together help the patient return to their premorbid condition.