World J Surg (2008) 32:436–443
DOI 10.1007/s00268-007-9304-z
Systematic Management of Postoperative Enterocutaneous
Fistulas: Factors Related to Outcomes
Jose L. Martinez Æ Enrique Luque-de-Leon Æ
Juan Mier Æ Roberto Blanco-Benavides Æ
Felipe Robledo
Published online: 5 December 2007
Ó Société Internationale de Chirurgie 2007
Abstract
Background Most enterocutaneous fistulas are postoper-
ative in origin. Sepsis, malnutrition, and hydroelectrolytic
deficit are still the most important complications to which
patients with postoperative enterocutaneous fistulas (PEF)
are exposed. Knowledge of prognostic factors related to
specific outcomes is essential for therapeutic decision-
making processes.
Methods We reviewed files of all consecutive patients
with PEF treated in our hospital during a 10-year period.
Our aim was to identify factors related to spontaneous
closure, need for operative treatment, and mortality. Uni-
variate and multivariate analyses were performed.
Results A total of 174 patients were treated. The most
frequent site of origin was the small bowel (90 patients: 48
jejunal, and 42 ileal), followed in frequency by the colon
(50 patients). Postoperative enterocutaneous fistula closure
was achieved in 151 patients (86%), being spontaneous in
65 (37%) and surgical in 86 (49%). Factors that signifi-
cantly precluded spontaneous closure were jejunal site,
multiple fistulas, sepsis, high output, and hydroelectrolytic
deficit at diagnosis or referral. Origin of PEF at our hospital
was the only factor significantly associated with sponta-
neous closure. The most frequent operative indication was
PEF persistence without sepsis. Factors significantly asso-
ciated with the need for operative treatment were high
output, jejunal site, and multiple fistulas. Closure was
J. L. Martinez (&)
E. Luque-de-Leon
J. Mier

R. Blanco-Benavides
F. Robledo
General and Gastrointestinal Surgery, Centro Médico Nacional,
Siglo XXI, Instituto Mexicano del Seguro Social (IMSS),
Cuauhtémoc #330, Col. Doctores, Deleg. Cuauhtémoc,
06725 México, D.F, México
e-mail: jlmo1968@yahoo.com
123
achieved in 84% of patients who underwent operation. A
total of 23 patients died (13%). Factors associated with
mortality were serum albumin \3.0 g/dl (at diagnosis or
referral), high output, hydroelectrolytic deficit, multiple
fistulas, jejunal site, sepsis, and a complex fistulous tract.
Conclusions In spite of advances in management of PEF,
the associated morbidity and mortality remain high.
Among several variables influencing outcome, our multi-
variate analysis disclosed high output, jejunal site, multiple
fistulas, and sepsis as independent adverse factors related to
non-spontaneous closure, need for operative treatment,
and/or death.
Enterocutaneous fistulas are one of the most complex and
challenging complications encountered in a surgical prac-
tice. Because *80% occur as a result of previous operative
procedures, the surgeon plays an important role, not only in
their treatment but also in their origin. Most surgical
patients are exposed to several complications; sepsis,
malnutrition and hydroelectrolytic deficit are the most
common, and the main causes of morbidity and mortality.
Despite important advances in medical treatment, rates
of spontaneous fistula closure have not improved signifi-
cantly. Moreover, medical and surgical innovations have
not changed morbidity rates, and have had only a moderate
impact on mortality. Prognosis depends on several factors,
such as site of origin, number of fistulas, output, patient’s
nutritional status, development of complications, and more
importantly, whether an adequate medical and surgical
program for fistula management was implemented. Our
aim in this study was to determine factors associated with
spontaneous closure, need for operative treatment, and
death in a series of consecutive patients treated in a tertiary
World J Surg (2008) 32:436–443
care facility. To cover a homogeneous group, only patients
with postoperative enterocutaneous fistulas (PEF) were
included.
Materials and methods
We retrospectively reviewed clinical records of consecu-
tive patients with PEF that were seen at our department
from January 1995 to December 2004. Postoperative en-
terocutaneous fistulas were defined as abnormal
communications that connect any portion of the gastroin-
testinal tract with the skin. Because of their different
nature, treatment, and prognosis, patients with spontane-
ous, biliary, pancreatic, internal, or perianal fistulas were
specifically excluded. Initial diagnosis was made by clini-
cal examination and confirmed with imaging studies
(fistulography, water-soluble and/or barium contrast
studies, and enteroclysis, among others) and/or
intraoperatively.
Data collection
Information retrieved from files included demographic
data, type of referral, type of initial (causative) operation
(urgent versus elective), number of fistulas, site of origin,
fistulous tract characteristics, output during 24 h, presence
of hydroelectrolytic imbalance or malnutrition at diagnosis
or referral, serum albumin, development of sepsis, and type
and duration of nutritional support. If operative treatment
was undertaken, indications and type of operation were
also recorded. Procedures performed outside the operating
room (i.e., percutaneous drainage of an abscess), were not
considered part of the operative treatment. All patients
were followed until hospital discharge or death.
Definition of variables
Dependent variables were spontaneous closure (defined as
closure of PEF without an operation), need for operative
treatment, and death. Our aim was to determine risk factors
associated with those dependent variables. The indepen-
dent variables examined included: age (\65 years-old/C 65
years-old), gender (male/female), origin (esophageal, gas-
tric, duodenal, jejunal, ileal, or colorectal), output during
24 h (\500 ml: low/C 500 ml: high), number of PEF (one:
single/[ one: multiple), type of referral (referral from
another hospital/PEF developed at our hospital), fistulous
tract (simple: short with a direct communication to the skin
surface/complex: associated abscess cavity or multiple
involved loops of bowel), type of initial operation (urgent/
437
elective), serum albumin (g/dl), presence of malnutrition
and hydroelectrolytic imbalance at diagnosis or referral,
development of sepsis during the course of disease, and use
of octreotide.
Statistical analysis
All data were collected and entered into the statistical
package program SPSS version 12.0 (SPSS Inc, Chicago
IL) on a personal computer. Unless specified otherwise,
numerical values are expressed as median (and range).
Statistical comparisons were made using Student’s t-test
for continuous variables and chi-square or Fisher’s exact
test when categorical variables were compared. All com-
parisons are two-tailed probabilities. Multiple factors were
evaluated by univariate analysis for their ability to predict
our dependent variables; significant factors were then
submitted to multivariate analysis (logistic regression) to
determine independent prognostic factors. Odds ratios for
these latter factors were calculated. Probabilities B 0.05
were considered significant.
Results
A total of 174 patients were included. Median age was 45
years (range: 16–83 years). General characteristics of
patients and PEF are shown in Table 1. Initial pathologies
and operations that created the fistula were varied and are
listed in Table 2. It is important to note that, besides pri-
mary closure of bowel perforations (n = 16), bowel
resections (of any kind) were performed in a total of 80
patients in these initial operations.
Management for all patients was based on the scheme of
4 phases proposed by Chapman and Sheldon. A total of 156
patients (90%) received nutritional support for a median of
30 days (range: 1–350 days), including enteral nutrition in
30 patients (17%), total parenteral nutrition (TPN) in 90
patients (52%), and combined in 36 patients (21%). Except
for patients with low output and/or distal PEF, most
patients were started on TPN. Indications to keep this type
of nutrition included high output, proximal PEF, and
inability to tolerate enteral nutrition (distal obstruction,
insufficient small bowel, and others). In 18 patients (10%)
the type of nutritional support was not specified. Patients
were weaned from TPN once nutritional requirements
could be achieved through the enteral route.
A suction system that was fashioned in our hospital and
used there for some years was used in selected patients
(*25%), especially those with multiple fistulas or those
that drained through large dehiscent wounds. With it,
suction is created with a rim of karaya that surrounds and
123
438 World J Surg (2008) 32:436–443

Table 1 Characteristics of patients and postoperative enterocutane- seals the edges of the dehisced wound and thus functions as
ous fistulas (n = 174) a skin protector. The wound is then covered with a plastic
Variable Number % sheet through which two tubes are placed; one is connected
to a suction device so as to evacuate fistula effluents and
Gender
the other allows entrance of air so as to prevent its collapse.
Male 105 60 For the remaining patients, ordinary methods of skin pro-
Female 69 40 tection and collection of gastrointestinal fluids were
Site of origin employed. Although we use vacuum-assisted closure (KCI,
Esophagus 6 3 San Antonio, TX) in selected patients, this system was not
Stomach 8 4 employed during the period of this study.
Duodenum 20 11 Antibiotic strategies varied during this long study per-
Jejunum 48 28 iod. The initial scheme was usually a combination of either
Ileum 42 24 a third-generation cephalosporin or a quinolone with an
Colon 50 29 antianaerobic agent, such as metronidazole or clindamycin.
Fistula output According to clinical course and results of cultures (when
High ([500 ml) 57 33 obtained), these regimens were later modified. Imipenem
Low (B500 ml) 117 67 cilastatin was frequently used as a second line agent. An-
Fistulous tract tifungals were added as needed.
Simple 159 91 Overall, PEF closure was obtained in 151 patients
Complex 15 9 (86%); a total of 65 (37%) had spontaneous closure, and
Number of fistulas the remaining 86 (49%) healed through operative means.
Multiple 32 18 Rates of total fistula closure (spontaneous and operative),
Single 142 82 as well as mortality, did not vary over time. Table 3 depicts
Initial (causative) operation these figures according to site of PEF origin within the
Urgent 124 71 gastrointestinal tract. Data regarding time intervals
Elective 50 29 between identification of the fistula and spontaneous clo-
Origin of patient sure or operative treatment for the different PEF sites is
Other hospital 134 77 shown in Table 4. It should be noted that because infor-
Same (our) hospital 40 23 mation from referring hospitals is frequently incomplete or
Sepsis inaccurate, identification of the fistula in patients referred
Yes 80 46 from other institutions was considered as the day of arrival
No 94 54 at our hospital.
Malnutrition
Yes 98 56
No 76 44
Spontaneous closure
Hydroelectrolytic imbalance
Univariate and multivariate analysis of factors related to
Yes 68 39
spontaneous closure are shown in Table 5. Jejunal site,
No 106 61
high output, multiple fistulas, and development of sepsis

Table 2 Initial pathologies and

Initial pathologies Number Initial operation Number
operations performed in 174
patients with postoperative Complicated appendicitis 25 Small bowel resection 36
enterocutaneous fistulas
Small bowel obstruction 22 Exploratory laparotomy 33
Trauma 21 Hemicolectomy 29
Complicated diverticular disease 12 Primary closure of bowel perforations 16
Cholecystitis 9 Appendectomy 12
Gunshot wound 9 Incisional (ventral) hernia repair 10
Stab wound 9 Gastrectomy 6
Complicated peptic ulcer disease 9 Cholecystectomy 6
Incisional (ventral) hernia 8 Other procedures (26) B4 each
Other diseases (30) B4 each

123
World J Surg (2008) 32:436–443 439

Table 3 Postoperative
Site of Number of Spontaneous closure Operative closure Total closure Deaths
enterocutaneous fistula
origin patients (n = 65) (n = 86) (n = 151) (n = 23)
outcomes in 174 patients
Number (%) Number (%) Number (%) Number (%)
according to location of the
fistula Esophagus 6 5 (83) 1 (17) 6 (100) 0 (0)
Stomach 8 3 (38) 4 (50) 7 (88) 1 (12)
Duodenun 20 10 (50) 9 (45) 19 (95) 1 (5)
Jejunum 48 7 (15) 27 (56) 34 (71) 14 (29)
Ileum 42 16 (38) 22 (52) 38 (90) 4 (10)
Colon 50 24 (48) 23 (46) 47 (94) 3 (6)

Table 4 Time intervals for

Site of origin Number Diagnosis— spontaneous Number Diagnosis—operative
spontaneous closure or
closure (days)b treatment (days)b
operative treatment in 167
patients with PEFa Esophagus 5 20 (12–22) 1 20
Stomach 3 10 (7–14) 4 25 (10–40)
a Duodenum 10 32 (21–40) 10 22 (2–60)
Seven patients died without
spontaneous closure or Jejunum 7 30 (20–50) 36 27 (1–150)
operative treatment and are not Ileum 16 28 (14–60) 25 25 (1–180)
included here
b Colon 24 30 (10–180) 26 38 (1–730)
Median (range)

Table 5 Factors related to

Variable Number of patients Univariate Multivariate analysis
spontaneous closure of
with spontaneous analysis
postoperative enterocutaneous
closure/total p Value OR (95% CI) p Value
fistulas in 174 patients
(univariate and multivariate Jejunum
analysis)
Yes 7/48 (15%) 0.001 0.282 (0.111–0.716) 0.008
No 58/126 (46%)
Multiple fistulas
Yes 5/32 (16%) 0.003 0.353 (0.119–1.052) 0.06
No 60/142 (42%)
Hydroelectrolytic imbalance
Yes 18/68 (26%) 0.013 0.895 (0.410–1.952) 0.78
No 47/106 (44%)
High output
Yes 10/57 (18%) 0.001 0.405 (0.174–0.941) 0.03
No 55/117 (47%)
Sepsis
Yes 21/80 (26%) 0.004 0.570 (0.275–1.181) 0.13
No 44/94 (49%)
Origin at our hospital
OR odds ratio (spontaneous Yes 20/40 (50%) 0.05 0.668 (0.303–1.472) 0.316
versus non-spontaneous closure)
No 45/134 (34%)
CI = confidence interval

were identified through univariate analysis as factors that
precluded spontaneous closure. After multivariate analysis,
multiple (versus single) PEF almost achieved statistical
significance against spontaneous closure (p = 0.06); how-
ever, statistical significance prevailed only for high output
and jejunal site. Odds ratios were calculated and chances
for spontaneous closure were 3.5 and 2.5 times lower for
jejunal and high output PEF, compared to other sites of
origin and low output PEF, respectively. The only factor
that favored spontaneous closure was that the fistula was
developed at our hospital. Often, patients with PEF that
originated at other institutions who were referred to our
institution did not experience spontaneous closure. Octre-
otide was used irregularly during our study period in a total

123
440 World J Surg (2008) 32:436–443

of 52 patients (in *50% and *25% of the patients with
high and low output PEF, respectively). This factor was
analyzed as an independent variable, but it was not asso-
ciated with any of the dependent variables (outcomes).
Surgical treatment
Overall, 102 patients (59%) required operative treatment;
PEF closure was achieved in 86 (84%) of them. Indications
for surgery in these patients were PEF persistence without
sepsis in 36 (35%), source control in 29 (28%), eversion of
mucosa in 23 (23%), distal obstruction in 10 (10%), and
other causes in 4 (4%). Jejunal site, high output, and
multiple fistulas were identified through univariate analysis
as factors that favored the need for operative treatment.
After multivariate analysis, statistical significance pre-
vailed only for multiple PEF and high output PEF. Their
odds ratios were 3 and 2.5 times higher, compared to single
PEF and low output PEF, respectively (Table 6).
Our operative plan included PEF resection with primary
anastomosis in most cases; however, this was not feasible
in every patient. The type of operations included intestinal
or colonic resection with primary anastomosis in 64
patients (63%), PEF resection with proximal enterostomy
or colostomy in 16 patients (16%), drainage of an abscess
in 12 patients (12%), and primary closure in 10 patients
(10%). Surgical closure was achieved in 59 of the 64
patients (92%) undergoing fistula resection and primary
anastomosis, 13 of 16 (81%) after resection and a diverting
ostomy, 7 of 10 (70%) after primary PEF closure, and 7 of
12 (58%) after abscess drainage.
Mortality
A total of 23 patients (13%) died; 16 of them (70%) had
been submitted to operative treatment, and 14 (61%) had
jejunal PEF. Mortality according to location of PEF origin
is shown in Table 3. Several variables were significantly
associated with mortality through univariate analysis,
which included serum albumin \3 g/dl, hydroelectrolytic
imbalance at the moment of diagnosis or referral, devel-
opment of sepsis, jejunal site, high output, complex
fistulous tract, and multiple fistulas. After multivariate
analysis, however, only jejunal site, multiple PEF, and
development of sepsis prevailed as independent significant
factors. Odds ratios for these subgroups of patients
revealed that the chances of dying were 5, 6, and 16 times
higher compared to patients with other sites of PEF, single
PEF, and those that did not develop sepsis, respectively
(Table 7). Probability of death increased with the number
of risk factors identified through univariate analysis. Mor-
tality rates for patients with 0–2, 3–4, and 5–7 factors were
2% (2 of 102 patients), 17% (9 of 54 patients), and 67% (12
of 18 patients), respectively.
Discussion
Enterocutaneous fistulas are one of the most difficult
complications that a surgeon can face [1–4]. In spite of
several advances in its management, mortality remains
high. Even in successfully treated cases, inherent morbidity
and the nature of the disease result in lengthy hospitaliza-
tions [4]. Hospitals specializing in the management of
enterocutaneous fistulas usually receive their patients from
other institutions [1, 5–7]. This type of referral improves
outcomes in terms of fistula closure, survival, and costs [6,
8]. A potential explanation is that timely identification and
management of PEF within experienced facilities compares
quite favorably to delays and referral after all attempts to
control and treat the problem have failed.
Site of origin of the fistula also plays a crucial role.
Because they have higher output, proximal fistulas behave
more aggressively than distally located ones [4, 9–13]. Risk

Table 6 Factors influencing the

Factor Number of patients Univariate analysis Multivariate analysis
need for surgical treatment in
that required surgical
174 patients with postoperative p Value OR (95% CI) p Value
treatment/total
enterocutaneous fistulas
(univariate and multivariate High output
analysis)
Yes 44/57 (77%) 0.001 2.630 (1.232–5.613) 0.012
No 58/117 (49%)
Jejunal fistula
Yes 36/48 (75%) 0.007 1.902 (0.858–4.216) 0.114
No 66/126 (52%)
Multiple fistulas
OR odds ratio (surgical Yes 26/32 (81%) 0.004 3.116 (1.171–8.290) 0.023
treatment versus nonsurgical No 76/142 (54%)
treatment)

123
World J Surg (2008) 32:436–443 441

Table 7 Factors associated

Variable Number of dead Univariate analysis Multivariate analysis
with mortality in 174 patients
patients/total
with postoperative p Value OR (95% CI) p Value
enterocutaneous fistulas
(univariate and multivariate Serum albumin \3 g/dl
analysis) Yes 22/114 (19%) 0.002 5.389 (0.487–59.594) 0.170
No 1/60 (2%)
High output
Yes 16/57 (28%) 0.001 1.479 (0.368–5.938) 0.581
No 7/117 (6%)
Hydroelectrolytic imbalance
Yes 17/68 (25%) 0.001 1.512 (0.337–6.782) 0.590
No 6/106 (6%)
Multiple fistulas
Yes 13/32 (40%) 0.001 5.881 (1.487–23.253) 0.012
No 10/142 (7%)
Sepsis
Yes 22/80 (27%) 0.001 16.59 (1.84–149.05) 0.012
No 1/94 (1%)
Jejunal fistula
Yes 14/48 (29%) 0.001 4.628 (1.167–18.346) 0.029
No 9/126 (7%)
Complex tract
Yes 5/15 (33%) 0.032 1.509 (0.205–11.095) 0.686
OR odds ratio (mortality versus No 18/159 (11%)
survival)

of hydroelectrolytic imbalance, malnutrition, and sepsis
increases in these patients [10, 14]. Moreover, probability
and timing of spontaneous closure is also related to location
of the fistula. Overall, small bowel fistulas have a lower
chance of spontaneous closure (*30%) and require longer
courses of treatment before spontaneous closure (*50 days),
compared to colonic fistulas (*85% and *35 days,
respectively) [15]. Time intervals between identification of
the fistula and spontaneous closure or operative treatment for
the different PEF sites in our series are shown in Table 4.
These intervals, however, are only accurate in those 40
patients whose PEF were originated at our hospital. Because
*80% of our patients were referred from other institutions,
no conclusions can be drawn from these data.
Spontaneous closure
Because spontaneous closure rates depend on many vari-
ables, a wide range, varying from 17% to 71% is reported
[1, 6, 8, 14, 16–24]. In our experience, 37% of PEF healed
without surgery. The known low rate of spontaneous clo-
sure for small bowel fistulas was confirmed in our study; it
occurred in only 23 of 90 patients (25%).
Several authors report a higher rate of spontaneous
closure for jejunal fistulas than for ileal fistulas (*40%
versus 25%, respectively) [12, 18, 22, 25]. This is in
contrast with our results (17% versus 38%, respectively)
and could be explained by the fact that 58% of our
patients with jejunal PEF had high output fistulas,
compared to 26% of those with ileal fistulas. In any
case, our multivariate analysis disclosed that both high
output and jejunal site, were independent significant
factors that precluded spontaneous closure. Our sponta-
neous closure rate of 48% for colonic fistulas is in
accordance with the findings of most authors [5, 9, 12,
18, 20, 22].
Source control and elimination of sepsis are essential to
promote spontaneous closure [26]. In our group, sponta-
neous closure rates for patients without sepsis was almost
double the rate of those with sepsis (49% versus 26%,
respectively); these rates, however, significantly diminish
(\10%), if closure is not achieved 1 month after sepsis
control [12, 20].
Although use of octreotide was not significantly related
to our dependent variables, no conclusions can be made
because of the number of patients and the irregular way in
which it was used in our series. Recent reviews state that,
even though management of enterocutaneous fistulas has
been aided by these hormones, results from randomized
controlled trials have not favoured including them as part
of the standard of care [27].

123
442
Surgical treatment
Although appropriate medical treatment has led to a
decline in the need for operative management [3], up to
80% of PEF patients still require it [5, 16, 19, 23, 24, 28,
29]. In our series, 59% underwent operation. The most
common indication for surgery is sepsis control (usually
abscess drainage). Perhaps because of our referral pattern,
in our patients this was the second most common cause,
second only to fistula persistence (without sepsis). Fistulas
whose anatomic factors preclude closure, such as distal
obstruction and terminal type, among others, usually
require operative treatment [17, 20, 25, 30]. Surgery should
also be advised for patients with fistula recurrence after
spontaneous or surgical closure, as well as those with
clinical deterioration [16, 17, 18, 31]. A rational conclusion
to be drawn from our study is that the group of patients
with unfavorable predictive factors such as high output,
jejunal site, and multiple fistulas, should be considered for
operative treatment after a reasonable period of conserva-
tive management. Unwarranted delays can increase the risk
of severe malnutrition, fluid and electrolyte imbalance, or
sepsis. Although early operations (especially for patients
with small bowel fistulas), has also been suggested [1, 2,
9], several other authors, have reported their best results in
terms of recurrence, morbidity, and mortality in patients
who waited the longest time [14, 27, 32]. Evenson et al.
[27] have even suggested waiting at least 4 months after the
causative operation. Putative benefits for late operations in
patients without sepsis are that the inflammatory process
will become inactive and the characteristics of intra-
abdominal adhesions will allow a safer procedure [6, 9, 11,
13, 14, 32].
Ideally, surgical alternatives for patients with PEF will
be aggressive and radical [9, 19, 30, 31]. When possible,
resection of involved loop(s) of bowel and anastomosis
should be attempted, because this approach has provided
the lowest failure rates in most series [9, 29]. This alter-
native was performed in 64 of our patients and was
successful in 92% of them.
Mortality
Mortality rates among PEF patients have decreased and
now range between 10% and 30% [8, 24, 27]; ours was
13%. Patients with small bowel fistulas have high mortality
rates (*31%) [2, 8, 9, 12, 15, 29]. Jejunal fistulas have
been reported to have double the death rates associated
with ileal fistulas [2, 6, 12, 14, 18]. Our patients with
jejunal fistulas had a mortality rate of 30%, compared to
10% for those with ileal fistulas; overall, 6 out of every 10
patients that died had jejunal fistulas. Some authors have
123
World J Surg (2008) 32:436–443
associated site (jejunum) with a high output as an expla-
nation for this higher mortality [2, 6, 12, 14, 18]; however,
our multivariate analysis established site of origin (and not
output) as an independent variable associated with
mortality.
The mortality rate in our patients with colonic fistulas
was 6%, substantially lower than the *20% reported for
most series [7, 9, 12, 17, 18, 20, 28, 29]. A potential
explanation for this is the lower incidence of high output
(14%) in our patients with colonic fistulas when compared
to other sites. High output has been associated with a bad
prognosis by most authors [8, 14, 15, 17, 23, 27, 33].
Overall, almost 30% of our patients with high output PEF
died, compared to only 6% of those with low output fis-
tulas. Association between presence of multiple (as
opposed to single) fistulas and mortality is controversial [6,
7, 14]. Our patients with multiple fistulas had a mortality
rate of 40%; multivariate analysis revealed this was an
independent factor related to mortality.
Hydroelectrolytic imbalance, malnutrition, and sepsis
are still the most common complications and most impor-
tant factors related to mortality in patients with PEF [12].
Over 40 years ago, mortality for patients with hydroelec-
trolytic imbalance was *80% [9]. Improvements in
medical therapy have significantly improved this figure [5,
18, 22, 29]; in our group, 1 of every 4 patients with hy-
droelectrolytic imbalance at diagnosis or referral died.
Similar to previous reports [14, 17, 22, 28, 29, 33] mor-
tality was higher in our malnourished patients; those with a
serum albumin \3 g/dl at diagnosis or referral had a
mortality rate of 19%, compared to 2% for all others.
Sepsis may be catastrophic. Mortality rates as high as 85%
have been reported for patients with uncontrolled sepsis [5,
8, 9, 12–15, 29, 33, 34]. In our series, chances of dying
were 16 times higher for patients with sepsis (compared to
patients without it). Indeed, the surgeon should always be
aware of its potential or real presence, and every effort
should be made to control and eliminate it.
In summary, outcomes for patients with PEF have
improved through protocolized, multidisciplinary man-
agement in tertiary care facilities. Source control and
elimination of sepsis are essential steps. Even though they
may be achieved by nonsurgical means, some patients may
require operative treatment. Enteral nutrition (either alone
or in combination) should be used whenever possible to
attain optimal functional and nutritional status, and hasten
spontaneous closure or prepare patients for definitive sur-
gery. Timing for this operation is controversial. In theory,
the presence of adverse factors, such as those found in our
study, should lead to ‘‘early’’ operations, because sponta-
neous closure is rare and potential complications during the
course of disease may be prevented. These benefits should
outweigh the risks that represent an unfit patient and/or an
World J Surg (2008) 32:436–443
‘‘unfriendly’’ abdomen. For these latter cases, and those in
whom no unfavorable factors (precluding spontaneous
closure) are found, withholding operative treatment may be
more beneficial. Once the decision to operate is made,
radical operations seem to yield the best results in terms of
success (92% in our series).
Enterocutaneous fistulas are among the most challeng-
ing problems a surgeon can face. Most cases are due to
complications or technical mishaps in abdominal and/or
gastrointestinal surgery. Improvements in the multidisci-
plinary management of these patients have significantly
decreased morbidity and mortality. However, other efforts
should be made to prevent its occurrence, to diagnose and
classify affected individuals, and to further diminish
complications and death. Advances in nutrition (glutamine,
arginine, omega-3 fatty acids, nucleotides) and wound
therapy (vacuum-assisted closure system), among others,
may represent avenues leading to these goals.
Acknowledgments The authors acknowledge all members of the
Department of General and Gastrointestinal Surgery, UMAE Hospital
de Especialidades, CMN Siglo XXI (IMSS), for allowing the inclu-
sion of their clinical experience.
References
1. Halversen RC, Hogle HH, Richards RC (1969) Gastric and small
bowel fistulas. Am J Surg 118:968–972
2. Nassos TP, Braash JW (1971) External small bowel fistulas.
Current treatment and results. Surg Clin North Am 51:687–692
3. Hollender LF, Meyer C, Avet D, et al. (1983) Postoperative fis-
tulas of the small intestine: therapeutic principles. World J Surg
7:474–480
4. Martineau P, Schwed JA, Denis R (1996) Is octreotide a new
hope for enterocutaneous and external pancreatic fistulas closure?
Am J Surg 172:386–395
5. Kuvshinoff BW, Brodish RJ, McFadden DW, et al. (1993) Serum
transferrin as a prognostic indicator of spontaneous closure and
mortality in gastrointestinal cutaneous fistulas. Ann Surg
217:615–623
6. Lévy E, Frileux P, Cugnenc PH, et al. (1989) High-output
external fistulae of the small bowel: management with continuous
enteral nutrition. Br J Surg 76:676–679
7. Schein M, Decker GAG (1991) Postoperative external alimentary
tract fistulas. Am J Surg 161:435–438
8. Campos ACL, Andrade DF, Campos GMR, et al. (1999) A
multivariate model to determine prognostic factors in gastroin-
testinal fistulas. J Am Coll Surg 188:483–490
9. Edmunds LH Jr, Williams GM, Welch CE (1960) External fistulas
arising from the gastrointestinal tract. Ann Surg 152:445–471
10. Foster CE, Lefor AT (1996) General treatment of gastrointestinal
fistulas. Recognition, stabilization, and correction of fluid and
electrolyte imbalances. Surg Clin North Am 76:1037–1052
443
11. Dudrick SJ, Maharaj AR, McKelvey AA (1999) Artificial nutri-
tion support in patients with gastrointestinal fistulas. World J
Surg 23:570–576
12. Reber HA, Roberts C, Way LW, et al. (1978) Management of
external gastrointestinal fistulas. Ann Surg 188:460–467
13. Rubelowsky J, Machiedo GW (1991) Reoperative versus con-
servative management for gastrointestinal fistulas. Surg Clin
North Am 71:147–157
14. Fazio VW, Coutsoftides T, Steiger E (1983) Factors influencing
the outcome of treatment of small bowel cutaneous fistula. World
J Surg 7:481–488
15. Berry SM, Fischer JE (1994) Enterocutaneous fistulas. Curr Prob
Surg 31:483–566
16. Fischer JE (1983) The pathophysiology of enterocutaneous fis-
tulas. World J Surg 7:446–450
17. Chapman R, Foran R, Dunphy JE (1964) Management of intes-
tinal fistulas. Am J Surg 108:157–164
18. Aguirre A, Fischer JE, Welch CE (1974) The role of surgery and
hyperalimentation in therapy of gastrointestinal-cutaneous fistu-
lae. Ann Surg 180:393–401
19. McIntyre PB, Ritchie JK, Hawley PR, et al. (1984) Management
of enterocutaneous fistulas: a review of 132 cases. Br J Surg
71:293–296
20. Rose D, Yarborough MF, Canizaro PC, et al. (1986) One hundred
and fourteen fistulas of the gastrointestinal tract treated with total
parenteral nutrition. Surg Gynecol Obstet 163:345–350
21. Prickett D, Montgomery R, Cheadle WG (1991) External fistulas
arising from the digestive tract. South Med J 84:736–739
22. MacFadyen BV Jr, Dudrick SJ, Ruberg RL (1973) Management
of gastrointestinal fistulas with parenteral hyperalimentation.
Surgery 74:100–105
23. Sitges-Serra A, Jaurrieta E, Sitges-Creus A (1982) Management
of postoperative enterocutaneous fistulas: the roles of parenteral
nutrition and surgery. Br J Surg 69:147–150
24. Draus JM Jr, Huss SA, Harty NJ, et al. (2006) Enterocutaneous
fistula: are treatments improving? Surgery 140:570–578
25. Berry SM, Fischer JE (1996) Classification and pathophysiology
of enterocutaneous fistulas. Surg Clin North Am 76:1027–1036
26. Sancho JJ, di Costanzo J, Nubiola P, et al. (1995) Randomized
double-blind placebo-controlled trial of early octreotide in
patients with postoperative enterocutaneous fistula. Br J Surg
82:638–641
27. Evenson AR, Fischer JE (2006) Current management of entero-
cutaneous fistula. J Gastrointest Surg 10:455–464
28. Sheldon GF, Gardiner BN, Way LW, et al. (1971) Management
of gastrointestinal fistulas. Surg Gynecol Obstet 133:385–389
29. Soeters PB, Ebeid AM, Fischer JE (1979) Review of 404 patients
with gastrointestinal fistulas. Ann Surg 190:189–202
30. Hill GL (1983) Operative strategy in the treatment of enterocu-
taneous fistulas. World J Surg 7:495–501
31. Lorenzo G. Beal JM (1969) Management of external small bowel
fistulas. Arch Surg 99:394–396
32. Lynch AC, Delaney CP, Senagore AJ, et al. (2004) Clinical
outcome and factors predictive of recurrence after enterocutane-
ous fistula surgery. Ann Surg 240:825–831
33. Altomare DF, Serio G, Pannarale OC, et al. (1990) Prediction of
mortality by logistic regression analysis in patients with postop-
erative enterocutaneous fistulae. Br J Surg 77:450–453
34. Hollington P, Mawdsley J, Lim W, et al. (2004) An 11-year
experience of enterocutaneous fistula. Br J Surg 91:1646–1651
123