SURGICAL MANAGEMENT OF

GASTROINTESTINAL FISTULAS 0039-6109/96 $0.00 + .20

CLASSIFICATION AND
PATHOPHYSIOLOGY OF
ENTEROCUTANEOUS FISTULAS
Scott M. Berry, MD, and Josef E. Fischer, MD, FACS

Many classification schemes have been used to define fistulas of the gastro-
intestinal (GI) tract. Anatomic, physiologic, and etiologic classification schemes
are the most commonly used. Each type of classification system carries specific
implications with regard to likelihood of spontaneous closure, prognosis, opera-
tive timing, and nonoperative care planning (Table 1). These classification
schemes are by no means exclusive of one another; indeed, it is desirable to
define each fistula by all three systems. In this way an integrated understanding
of the enterocutaneous fistula and its physiologic impact on the patient can be
achieved.
AoFtoenteric fistulas are normally not considered in an article of this title
or nature but are surgical emergencies, and suspicion of their presence in a
patient considered an operative candidate is an indication for immediate sur-
gery. Aortoenteric fistulas most commonly occur secondarily after placement of
prosthetic aortic grafts. They are of three types: (1) true graft enteric fistulas in
which one of the suture lines, usually the proximal one, communicates with the
intestinal tract; (2) a proximal suture line pseudoaneurysm that has eroded into
adjacent bowel; and (3) perigraft enteric erosions in which the midportion of the
graft erodes into adjacent bowel (Table 2). These present with profound GI
bleeding, perigraft infections, or graft thrombosis.
The pathophysiology of primary aortoenteric fistulas is virtually always
erosion of the aneurysmal or infected aorta into surrounding viscera. The re-
mainder of primary aortoenteric fistulas arise from other intestinal and gyneco-
logic processes, such as esophageal, gastric, and cervical cancer, and various
inflammatory processes (Table 3). Diagnosis is difficult and mortality high for
these particular types of fistulas.

From the Department of Surgery, University of Cincinnati, Cincinnati, Ohio

SURGICAL CLINICS OF NORTH AMERICA

VOLUME 76 NUMBER 5 * OCTOBER 1996 1009

1010 BERRY & FISCHER

Table 1. CLASSIFICATIONS OF GASTROINTESTINAL FISTULAS AND

THEIR SIGNIFICANCE
Significance
Scheme Classification Favorable Unfavorable
Anatomic Internal Esophageal, duodenal Gastric, lateral duodenal,
External stump, pancreatobiliary, ligament of Treitz, ileal,
Organ involved jejunal, small leak, tract complete disruption,
<2 cm, defect <1 cm2 epithelialization, distal
obstruction
Physiologic Output Output does not Output does not
Low <200 m u d prognosticate closure prognosticate closure
Moderate
High >500 m u d Well nourished, no sepsis, Malnourished, sepsis,
transferrin >200 mg/dL transferrin <200 mg/dL
Etiologic Disease process Appendicitis, diverticulitis, Cancer, inflammatory
postoperative bowel disease, foreign
body, radiation

CLASSIFICATION OF ENTEROCUTANEOUS FISTULAS

Fistulas are abnormal communications between two epithelialized surfaces.

Most occur after operation or instrumentation. Other causes include tumor,
inflammation, and irradiation. Classification schemes have been devised to de-
fine various aspects of a fistula and the fistula's impact on the individual.
Anatomic, physiologic, and etiologic classifications are most useful to the sur-
geon charged with the care of patients with these difficult problems. In general,
anatomic information is the first information gained. These data are most readily
obtained by contrast studies or by CT scan. We have found fistulograms to be
more helpful in defining the anatomy of the abnormal communication than
formal GI contrast studies. CT scans may show areas of abnormal bowel which
are likely sources of the enterocutaneous fistula but seldom demonstrate the
fistula tract proper. The main utility of CT scan in the early management period
is to localize undrained collections of purulence.26* 31
Anatomic information has prognostic significance with regard to spontane-
ous closure of the fistula tract.', 2, 4, 1s15, 34, 50 In addition, anatomic information
helps narrow the differential diagnosis with regard to the etiologic process
underlying fistula formation. Anatomic characteristics associated with nonheal-
ing fistulas include large adjacent abscess, intestinal discontinuity, distal obstruc-
tion, poor adjacent bowel, fistula tracts less than 2 cm in length, enteral defects
greater than 1 cm, and fistulas arising from certain bowel segments such as
stomach, lateral duodenum, ligament of Treitz, and ileum. Those anatomic
segments with more favorable closure rates include oropharyngeal, esophageal,
duodenal stump, pancreatobiliary, and jejunal. Notwithstanding previous com-

Table 2. TYPES OF SECONDARY AORTOENTERIC FISTULAS

Graft enteric fistulas Suture line erosion into adjacent bowel
Erosion of pseudoaneurysm Pseudoaneurysm erosion into adjacent bowel
Enteric erosion Midgraft erosion into adjacent bowel
 CLASSIFICATION AND PATHOPHYSIOLOGY OF ENTEROCUTANEOUS FISTULAS 1011

Table 3. ETIOLOGY OF PRIMARY AORTOENTERIC FISTULAS

Abdominal Aortic Aneurysm 92%
Atherosclerosis 85%
Infectious 15%
Nonaneurysmal Aorta 8%
Cervical cancer 2%
Colon cancer 2%
Radiotherapy, diverticulitis, duodenal ulcer, cholelithiasis, 4%
and foreign body

ments, even when anatomic factors are favorable, the ability to predict spontane-
ous closure of a fistula tract is inexact (Table 4).
Physiologic classification schemes are most useful in planning a nonopera-
tive treatment regimen. The fistula output in a 24-hour period is the most
important determinant of the physiologic impact of a fistula on a patient.
Enterocutaneous fistulas result in the external loss of fluid, minerals, trace
elements, and protein. These losses can have profound effects on the patient
and his or her eventual outcome. Accurate knowledge of fistula output is
requisite for anticipating metabolic deficits and correcting ongoing losses. Fistula
output is an independent predictor of patient death. It is not prognostic of
eventual closure, although it is true that the 24-hour output from most fistulas
decreases as a prelude to closure.', 4* 16, 50
Three different categories exist within the physiologic classification sys-
tem-low output (<200 mL/24-hour period), moderate output (200-500 mL/24-
hour period), and high output (>500 mL/24-hour period).', 4, 16, 50 These outputs
are most useful in planning nonoperative nutrition, fluid, and electrolyte man-
agement (Table 5).
Malnutrition is a prominent part of the morbidity and mortality associated
with enterocutaneous fistulas and is present in 55% to 90% of patients with
enterocutaneous fistulas. In general, patients with low-output fistulas should
receive their full resting energy expenditure, 1 to 1.5 grams of protein per
kilogram per day, with roughly 30% of calories being supplied as lipid. These
patients should receive at least a portion of their caloric needs enterally. Those
with high-output fistulas should receive 1.5 to 2.0 times their resting energy
expenditure and 1.5 to 2.5 grams of protein per kilogram per day. In addition,
these patients should receive two times the recommended daily allowance

Table 4. ANATOMIC FEATURES AFFECTING SPONTANEOUS CLOSURE RATES

Favorable Unfavorable
Continuity maintained Complete disruption
End fistula Lateral fistula
No associated abscess Associated abscess
Healthy adjacent bowel Diseased adjacent bowel
Free flow distally Distal obstruction
Esophageal Gastric
Duodenal stump Lateral duodenal
Jejunal Ileal
Tract >2 cm Tract <2 cm
Defect <1 cm Defect >1 cm
1012 BERRY & FISCHER

Table 5. PHYSIOLOGIC CLASSIFICATION AND NUTRITIONAL SIGNIFICANCE

Low output High Output
(c200 mud) (>500 mud)
Form of nutrition Enteral Usually requires some or all of
nutrition in parenteral form
Protein 1-1.5 g/kg/d 1.5-2.5 glkgld
Calories Resting energy expenditure Resting energy expenditure
times 1.5
Lipid Enteral, 20%-30% of total Parenteral, 20%-30% of total
calories calories
Vitamins RDA, 2 times RDA for vitamin C Two times RDA
Minerals Not Problematic May be difficult, magnesium,
zinc, potassium, sodium,
bicarbonate
RDA = recommended daily allowance.

(RDA) for vitamins and trace minerals, 5 to 10 times the RDA for vitamin C,
and zinc supplementation.
Fistula cause or pathophysiology is predictive of spontaneous closure and
independently predictive of patient death. Etiologic information is often the last
piece of information obtained. Those fistulas that occur within 7 to 10 days after
GI surgery are seldom an enigma. The vast majority result from anastomotic
failure secondary to tension on the anastomosis, poor blood supply, or poor
technique, with the remainder arising from unrecognized bowel injuries during
dissection or abdominal closure. Fistulas that occur late or those that occur
spontaneously are more problematic.
Spontaneous causes comprise approximately 15% to 25% of all enterocuta-
neous fistulas.I7,40, 47, 48 Radiation, inflammatory bowel disease, diverticular dis-
ease, appendicitis, ischemic bowel, erosion of indwelling tubes, perforation of
duodenal ulcers, and pancreatic and gynecologic malignancies are the most
common causes in spontaneously occurring fi~tu1as.l~. 40, 47, 48 Those that arise
from radiation-damaged intestine or from malignant intestinal lesions are un-
likely to close. Those fistulas arising from bowel involved by inflammatory
bowel disease often close, only to reopen at a later date.
The remaining 75% to 85% of enterocutaneous fistulas are postoperative.
Operations for cancer, inflammatory bowel disease, and lysis of adhesions are
the most common operations preceding enterocutaneous fistula formation. In
addition, operations for peptic ulcer disease and pancreatitis can lead to postop-
erative enterocutaneous fistula formation. Fistulas more commonly occur in
settings of emergency surgery, for which patient preparation has been poor or
when the patient is chronically malnourished. Some of the factors that predis-
pose a patient to a postoperative enterocutaneous fistula are within the control
of the surgeon; some are not. Use of healthy bowel to perform an anastomosis
well away from inflamed or diseased tissue, preoperative mechanical bowel
preparation, preoperative intraluminal or systemic antibiotics, tension-free anas-
tomoses, meticulous hemostasis, secure abdominal wall closure, maintenance of
adequate oxygen-carrying capacity in the postoperative period, and preoperative
maximization of nutritional status all lessen the risk of postoperative enterocuta-
neous fistula formation.
 CLASSIFICATION AND PATHOPHYSIOLOGY OF ENTEROCUTANEOUS FISTULAS 1013

PATHOPHYSIOLOGY OF SPECIFIC ENTEROCUTANEOUS

FISTULAS

Gastric fistulas are iatrogenic in roughly 85% of cases, with the remainder
being due to irradiation, inflammation, ischemia, and malignancy?, 11, 21, 25, 27, 38, 51
Anastomotic leak after a gastric resection for cancer occurs in 5% to 10% of
cases. Many are due to residual cancer at the suture line and consequently are
unlikely to close and carry a 50% to 75% m~rtality.~, 12* 25, 39, 52 Gastric leaks after
resections for peptic ulcer disease, antireflux procedures, and bariatric surgery
occur in roughly 1% to 3% of each type of case.3,5, 6, 12, 19, 22, 36 Most of those patients
(85%)present with intra-abdominal abscess mandating immediate drainage, and
subsequently a fistula becomes apparent. The remainder of patients (15%)have
a fistula as their primary presenting symptom.
Duodenal fistulas occur as a complication of gastric resection, duodenal
resection, biliary tract procedures, pancreatic resections, right colon operations,
and aortic and kidney operations in 85% of cases. The remainder are the result
of trauma, perforated peptic ulcers, and cancer.Io,42, 49 The overall mortality for
duodenal fistulas of all causes is roughly 30%. Duodenal fistulas occur in 3% to
5% of patients having gastric resections. The incidence has been shown to be
decreased to less than 1% by the liberal use of tube duodeno~torny.~~ End-
duodenal stump fistulas close in roughly 85% of cases, whereas lateral duodenal
fistulas close spontaneously in only 30% to 40% of cases.’, 7, 32, 34, 35

Complications of abdominal surgery are the cause of small intestinal fistulas

in 70% to 90% of cases.’,30, 43, 45, These causes include disruption of the anasto-
motic suture line, inadvertent enterotomy, or inadvertent small bowel injury at
the time of abdominal closure. Roughly half are thought to be due to anastomotic
failure, and half are due to inadvertent enterotomy. Thus, half of small bowel
fistulas occur after operations in which there has been no resection or anastomo-
sis.
Anastomoses are jeopardized by inadequate blood flow from devasculariza-
tion or systemic hypotension, tension on anastomotic suture lines, perianasto-
motic abscesses, and anastomosis performed in diseased bowel. Small bowel
defects greater than half of the bowel circumference should be treated by
resection with end-to-end anastomosis. This is especially true when the mesen-
teric border is involved and ischemia is likely. When closure of multiple defects
is necessary, resection should be considered, provided that this does not compro-
mise intestinal length to the point of short gut syndrome. Approximately 30
inches of small bowel are necessary to prevent short gut syndrome, provided
that the ileocecal valve is present.
Spontaneous causes produce the remaining 10% to 15% of small bowel
fistulas. These processes include Crohn’s disease (5% to 50%), cancer (2% to
15%), peptic ulcer disease (3% to 6%), and pancreatitis (3% to loo/,). In industrial-
ized countries, the most common cause of primary small bowel disease leading
to enterocutaneous fistula formation is Crohn’s disease.33The transmural in-
flammation that occurs with Crohn’s disease leads to adherence of the involved
bowel segment to adjacent structures. Microperforation then leads to abscess
formation and subsequent perforation into the adjacent structure. Usually this
is a loop of bowel, urinary bladder, or vagina. Less frequently it emerges
through the skin of the anterior abdominal wall. Fistulas develop in 20% to 40%
of patients with Crohn‘s disease, one half internal and one half external. Two
types of external fistulas occur with Crohn’s disease. The first type occurs after
resective therapy from the anastomotic suture line that is not grossly involved
with Crohn‘s disease. These have favorable spontaneous closure rates and thus
1014 BERRY & FISCHER

should be initially managed expectantly. The other type arises in bowel involved
with Crohn’s disease, has a less favorable spontaneous closure rate, often re-
opens upon resumption of enteral intake, and should be considered for early
surgical intervention after the fistula has closed. This interval resection provides
the best chance for complete resolution, as the sepsis in and around the bowel
wall is minimal during this time.
Colocutaneous fistulas result primarily from diverticulitis, cancer, inflam-
matory bowel disease, appendicitis, or surgical treatment of one of these dis-
eases. In addition, because of increased survival in modern times, the surgical
treatment of necrotizing pancreatitis has become a more frequently recognized
cause of colonic fistula. Inadequate resection of colon involved with diverticula
or cancer can lead to colonic fistula formation when either of those disease
processes recurs.
Radiation therapy is another major cause of colonic fistulas. Although
radiation therapy has improved long-term survival in many malignancies, a 5%
to 10% incidence of radiation-induced intestinal complications is seen weeks to
years after administration. Moreover, bowel resection and anastomosis in irradi-
ated tissues place the patient at increased risk for anastomotic breakdown and
subsequent fistula formation and should be avoided if possible.
Finally, the least common spontaneous colocutaneous fistulas arise from
the appendix. More commonly such appendicocutaneous fistulas occur after
percutaneous drainage of an appendiceal abscess. Fistulas that occur after appen-
dectomy in a patient subsequently found to have Crohn’s disease are usually
not from the appendiceal stump, but arise from the terminal ileum where the
active Crohn’s segment adheres to the healing abdominal suture line.

CLASSIFICATION AND PATHOPHYSIOLOGY OF

AORTOENTERIC FISTULAS

Prior to the advent of antibiotics, as many as one fourth of aortoenteric

fistulas were the sequelae of tuberculosis, syphilis, or infection with Salmonella,
Staphylococcus, or the mycoses (see Table 3).=,37, 41, 44, 53 With the advent of
antibiotics, these causes became rare. In recent times, primary aortoenteric fistu-
las are reported to occur in 0.1% to 0.8% of patients with aneurysms and are
due mainly to atherosclerotic aneurysms of the abdominal aorta eroding into
adjacent 37, 41, 44 Sir Astley Cooper is credited with the first description
of an aortoduodenal fistula in 1829. It was in the late 1950s that the first
description of successful aneurysmorrhaphy for aortoenteric fistula was pub-
l i ~ h e d . *In~ 1961 MacLean and Couves described extra-anatomic bypass with
Dacron for aortoenteric fistulas. The average age of patients with primary aor-
toenteric fistulas is roughly 60 years, and 75% are Back pain is present in
60%, melena in 40%, hematemesis in 40%, and a palpable abdominal mass in
50%. The average aneurysm size is 7 cm, with a range from 3 to 15 cm.8 CT is
the diagnostic procedure of choice for diagnosing primary aortoenteric fistulas.
Esophagogastroduodenoscopy rarely demonstrates the fistula but is useful to
rule out other causes of upper GI Aortography is useful in planning
the operative approach but rarely identifies the actual fistula.8,37 Fifty percent to
70% of primary aortoenteric fistulas occur into the third portion of the
duodenum,=, 44; the remainder of aortoenteric communications are to the esopha-
gus (1%to 5%), stomach YO), jejunum (20%), ileum (5%), and colon (5% to
10%) (Table 6).41,44 Patients usually present with GI bleeding. The diagnostic
triad of abdominal pain, GI bleeding, and an abdominal mass is present in only
 CLASSIFICATION AND PATHOPHYSIOLOGY OF ENTEROCUTANEOUS FISTULAS 1015

Table 6. SITE OF PRIMARY AORTOENTERIC FISTULAS

Stomach 3%
Duodenum 82%
Jejunum 9%
Colon 6%

a small percentage of patients.23,37, 44 Survival after primary aortoenteric fistula

is only 20% to 30%.
Secondary aortoenteric fistulas occur after aortic replacement with pros-
thetic grafts in 1% to 2% of 29, 37 They occur in three basic types-a
fistula from the prosthetic graft suture line to the intestine, a suture line pseu-
doaneurysm that has eroded into the intestine, and a paraprosthetic fistula in
which the body of the graft erodes into adjacent bowel (Table 7).3' Mortality
rates are 35% to 70%, with amputation rates as high as 30?'0.~~, 29 Positive blood
cultures portend a worse prognosis, one set of positive cultures raising the
mortality to 90%!1 Fifty percent to 60% of secondary fistulas occur in patients
who have undergone either emergent repair of ruptured aneurysms or multiple
aortic proced~res.3~. 41 The mean time from the most recent operation to the

development of a secondary aortoenteric fistula is 37 months, with a range from

3 months to 15 yearsjl Eighty percent to 90% are to the duodenum, 5% to 15%
to the jejunum, and 1%to 5% to the The proximal suture line is the site
of the fistula in 8O%, 10% to 15% have midgraft enteric erosion of the graft, and
5% develop from the distal anastomotic suture line (see Table 2).41Sixty percent
to 90% of patients present with hemorrhage: 20% with exsanguinating hemor-
rhage, 40% with hemodynamically significant bleeding, and 40% with a sentinel
hemorrhage.20,29, 41 Of those patients with a sentinel hemorrhage, roughly 30%
rebleed within the first day, 30% within the first week, and another 30% within
the first month after the sentinel hemorrhage.8,54 Of patients with secondary
aortoenteric fistulas, 5% to 40% have no bleeding but present with either sys-
temic sepsis or graft thrombosis.20, 29, 41 Emergent esophagogastroduodenoscopy
is diagnostic in only 10% to 25% of patients and reveals other lesions to which
the bleeding is wrongly attributed in 25%.28,29, 37, 41 CT is diagnostic in only 30%
to 60% of patients with secondary aortoenteric 29, 41 Thus, a high index

of suspicion must be maintained in a patient with GI bleeding who has under-

gone previous aortic grafting. Although the majority of the early fistulas were
associated with silk sutures, the advent and use of monofilament sutures have
not obviated this problem.
Considerable controversy still exists over the optimal method of treatment
for aortoenteric fistulas. Mortality rates remain between 40% and 50%, amputa-
tion rates are 20% to 30%, and aortic stump disruption occurs in 10% to 20% of
patients after extra-anatomic bypass and graft resection with oversewing of the

Table 7. SITES OF SECONDARY AORTOENTERIC FISTULAS

Stomach 3%
Duodenum 79%
Jejunum 9%
Ileum 4%
Colon 4 yo
Appendix and rectum 1 Yo
1016 BERRY ik FISCHER

aortic This has led many vascular surgeons to rethink therapeutic

strategies in dealing w i t h this difficult problem.

References

1. Aguirre A, Fischer JE: Intestinal fistulas. In Fischer JE (ed): Total Parenteral Nutrition.
Boston, Little, Brown & Co, 1976, pp 203-218
2. Aguirre A, Fischer JE, Welch CE: The role of surgery and hyperalimentation in the
management of gastrointestinal-cutaneous fistulae. Ann Surg 180:393-401, 1974
3. Alexandre JH, Fraioli JP, Sage M Value of water-soluble opaque media in the early
detection of fistulae following operation for hiatus hernia. J Radio1 63:115-117, 1982
4. Benson DW, Fischer JE: Fistulas. In Fischer JE (ed): Total Parenteral Nutrition, ed 2.
Boston, Little, Brown & Co, 1991, pp 253-262
5. Buckwalter JA, Herbst CA Jr: Leaks occurring after gastric bariatric operations. Surgery
1031156-160, 1988
6. Bumett UF, Read RC, Dale Morris W, et al: Management of complications of fundopli-
cation and Barrett’s esophagus. Surgery 82:521-530, 1977
7. Buzby GP, Blouin G, Colling CL, et al: Perioperative total parenteral nutrition in
surgical patients. N Engl J Med 325:525-532, 1991
8. Calligaro KD, Bergen WS, Savarese RP, et al: Primary aortoduodenal fistula due to
septic aortitis. J Cardiovasc Surg 33:192-198, 1992
9. Csendes A, Diaz JC, Burdiles P, et al: Classification and treatment of anastomotic
leakage after extended total gastrectomy in gastric carcinoma. Hepatogastroenterol
37(suppl II):174-177, 1990
10. deSa LA, Roddie ME, Williamson RCN: Fatal duodenocaval fistula resulting from a
giant peptic ulcer: Case report. Acta Chir Scand 156647450, 1990
11. De Villa V, Calvo FA, Bilbao JI, et al: Arteriodigestive fistula: A complication associated
with intraoperative and external beam radiotherapy following surgery for gastric
cancer. J Surg Oncol49:52-57, 1992
12. Escudier B, Belmiba H, Henry-Amar M, et al: Anastomotic fistula complicating total
gastrectomy and esophagogastrectomy for cancer of the stomach. Am J Surg 138:399-
402, 1979
13. Fischer JE: Enterocutaneous fistulas. In Najarian JS, Delaney JP (eds): Progress in
Gastrointestinal Surgery. Chicago, Year Book Medical Publishers, 1989, pp 377-387
14. Fischer JE: Enterocutaneous fistulas. In Norton LW, Eisman B (eds): Surgical Decision
Making, ed 2. Philadelphia, WB Saunders, 1985, pp 146-147
15. Fischer JE: The management of gastrointestinal cutaneous fistulae. Contemp Surg
29:104-108, 1986
16. Fischer JE: The pathophysiology of enterocutaneous fistulas. World J Surg 744-50,
1983
17. Galland RB, Spencer J: Radiation-induced gastrointestinal fistulae. Ann R Coll Surg
Engl 6 8 5 7 , 1986
18. Garden OJ, Dykes EH, Carter DC: Surgical and nutritional management of postopera-
tive duodenal fistulas. Dig Dis Sci 33:30-35, 1988
19. Gianello P, Baulieux J, Maillet P: Esophageal and gastric fistula after hiatal hernia
surgery. Acta Chir Belg 85:169-178, 1985
20. Haiart DC, Callam MJ, Murie JA, et al: Re-operations for late complications following
abdominal aortic operation. Br J Surg 78204-206, 1991
21. Halversen RC, Hogle HH, Richards RC: Gastric and small bowel fistulas. Am J Surg
118:968, 1969
22. Hauters P, de Canniere L, Collard JM, et a1 Gastrodiaphragmatic fistula after trans-
abdominal Nissen fundoplication. J Clin Gastroenterol 12:313-315, 1990
23. Hickey NC, Downing R, Hammer JD, et al: Abdominal aortic aneurysms complicated
by spontaneous ileocaval or duodenal fistulae. J Cardiovasc Surg 32:181-185, 1991
24. Higgins RSD, Steed DL, Zajko AB, et al: Computed tomographic scan confirmation of
paraprosthetic enteric fistula. Am J Surg 162:36-38, 1991
 CLASSIFICATION AND PATHOPHYSIOLOGY OF ENTEROCUTANEOUS FISTULAS 1017

25. Holscher AM, Schuler M, Siewert J R Surgical treatment of adenocarcinomas of the

gastroesophageal junction. Dis Esoph 1:35-50, 1988
26. Jaques P, Mauro M, Safrit H, et al: CT features of intraabdominal abscesses: Prediction
of successful percutaneous drainage. AJR 146:1041-1045, 1986
27. Kaminsky VM, Dietel M: Nutritional support in the management of external fistulas
of the alimentary tract. Br J Surg 62:100, 1975
28. Katz PO, Salas L: Less frequent causes of upper gastrointestinal bleeding Gastroenterol
Clin North Am 22:875-889, 1993
29. Kuestner LM, Reilly LM, Jicha DL, et al: Secondary aortoenteric fistulas: Contemporary
outcome with use of extra-anatomic bypass and infected graft excision. J Vasc Surg
21:184-196, 1995
30. Kuvshinoff BW, Brodish RJ, McFadden DW, et al: Serum transferrin as a prognostic
indicator of spontaneous closure and mortality in gastrointestinal cutaneous fistulas.
AM Surg 217615-623,1993
31. Lambiase RE, Deyoe L, Cronan JJ, et al: Percutaneous drainage of 335 consecutive
abscesses: Results of primary drainage with 1 year follow-up. Radiology 184:167-179,
1992
32. Levy E, Cugnenc PH, Frileux P, et al: Postoperative peritonitis due to gastric and
duodenal fistulas: Operative management by continuous intraluminal infusion and
aspiration. Report of 23 cases. Br J Surg 71:543-546, 1984
33. Lindberg E, Jarnerot G, Huitfeldt B: Smoking in Crohn’s disease: Effect on localisation
and clinical course. Gut 33:779-782, 1992
34. MacFayden VB, Dudrick SJ, Ruberg RL: Management of gastrointestinal fistulas with
parenteral hyperalimentation. Surgery 74:lOO-105, 1973
35. Malangoni MA, Madura JA, Jesseph JE: Management of lateral duodenal fistulas: A
study of 14 cases. Surgery 90:645451, 1981
36. Mansour KA, Burton HG, Miles JI, et al: Complications of intrathoracic Nissen fundo-
plication. Ann Thorac Surg 32:173-178, 1981
37. Nagy SW, Marshall JB: Aortoenteric fistulas. Postgrad Med 93:211-222, 1993
38. Nassos TP, Braasch J W External small bowel fistulas: Current treatment and results.
Surg Clin North Am 51:687, 1971
39. Olbe L, Lundell L: Intestinal function after total gastrectomy and possible consequence
of gastric replacement. World J Surg 11:713-719, 1987
40. Patrick CH, Goodin J, Fogarty J: Complication of prolonged transpyloric feeding:
Formation of an enterocutaneous fistula. J Pediatr Surg 23:1023-1024, 1988
41. Peck JJ, Eidemiller LR Aortoenteric fistulas. Arch Surg 1271191-1194, 1992
42. Pokomy WJ, Brandt ML, Harberg FJ: Major duodenal injuries in children: Diagnosis,
operative management, and outcome. J Pediatr Surg 21:613-616, 1986
43. Reber HA, Roberts C, Way L, et al: Management of external gastrointestinal fistulas.
AM Surg 188:460467, 1978
44. Rhim JA, Rubin RA, Katzka DA, et al: Primary aortoduodenal fistula: A unique
presentation of a pseudoaneurysm associated with cystic medial necrosis. Am J Gas-
troenterol 88:447-450, 1993
45. Rinsema W, Gouma DJ, von Meyenfeldt MF, et al: Primary conservative management
of external small-bowel fistulas: Changing composition of fistula series? Acta Chir
Scand 156:457462, 1990
46. Rodkey GV, Welch CE: Duodenal decompression in gastrectomy. N Engl J Med
262:498-501, 1960
47. Rubin SC, Benjamin I, Hoskins WJ, et al: Intestinal surgery in gynecologic oncology.
Gynecol Oncol3430-33, 1989
48. Schein M Free perforation of benign gastrojejunocolic and gastrocolic fistula: Report
of two cases. Dis Colon Rectum 30:705-706, 1987
49. Shorr RM, Greaney GC, Donovan AJ: Injuries of the duodenum. Am J Surg 154:93-
98, 1987
50. Soeters PB, Ebeid AM, Fischer JE: Review of 404 patients with gastrointestinal fistulas:
Impact of parenteral nutrition. Ann Surg 190:189-202, 1979
51. Tarazi R, Coutsoftides T, Steiger E, et al: Gastric and duodenal cutaneous fistulas.
World J Surg 7463-473, 1983
1018 BERRY & FISCHER

52. Troidl H, Kusche J, Vestwela KH: Pouch versus esophagojejunostomy after total
gastrectomy: A randomized clinical trial. World J Surg 11:699-712, 1987
53. Van Damme H, Belachew M, Damas P, et al: Mycotic aneurysm of the upper abdomi-
nal aorta ruptured into the stomach. Arch Surg 127478482, 1992
54. Wheeler WE, Hanks J, Raman VK: Primary aortoenteric fistulas. Am Surg 58:53-54,
1992

Address reprint requests to

Josef E. Fischer, MD, FACS
Department of Surgery
University of Cincinnati
Cincinnati, OH 45267