J Gastrointest Surg (2012) 16:156164 DOI 10.

1007/s11605-011-1703-7

2011 SSAT PLENARY PRESENTATION

Factors Predictive of Recurrence and Mortality after Surgical Repair of Enterocutaneous Fistula
Jose L. Martinez & Enrique Luque-de-Len & Guillermo Ballinas-Oseguera & Jos D. Mendez & Marco A. Jurez-Oropeza & Ruben Romn-Ramos

Received: 9 May 2011 / Accepted: 14 September 2011 / Published online: 15 October 2011 # 2011 The Society for Surgery of the Alimentary Tract

Abstract Many enterocutaneous fistulas (ECF) require operative treatment. Despite recent advances, rates of recurrence have not changed substantially. This study aims to determine factors associated with recurrence and mortality in patients submitted to surgical repair of ECF. Consecutive patients submitted to surgical repair of ECF during a 5-year period were studied. Several patient, disease, and operative variables were assessed as factors related to recurrence and mortality through univariate and multivariate analysis. There were 35 male and 36 female patients. Median age was 52 years (range, 1781). ECF recurred in 22 patients (31%), 18 of them (82%) eventually closed with medical and/or surgical treatment. Univariate analyses disclosed noncolonic ECF origin (p=0.04), high output (p=0.001), and nonresective surgical options (p=0.02) as risk factors for recurrence; the latter two remained significant after multivariate analyses. A total of 14 patients died (20%). Univariate analyses revealed risk factors for mortality at diagnosis or referral including malnutrition (p=0.03), sepsis (p=0.004), fluid and electrolyte imbalance (p=0.001), and serum albumin <3 g/dl (p= 0.02). Other significant variables were interval from last abdominal operation to ECF operative treatment 20 weeks (p= 0.03), preoperative serum albumin <3 g/dl (p=0.001), and age 55 years (p=0.03); the latter two remained significant after multivariate analyses. Interestingly, recurrence after surgical treatment was not associated with mortality (p=0.75). Among several studied variables, recurrence was only independently associated with high output and type of surgical treatment (operations not involving resection of ECF). Interestingly, once ECF recurred its management was as successful as non-recurrent fistulas in our series. Mortality was associated to previously-reported bad prognostic factors at diagnosis or referral.

DDW 2011. SSAT Plenary Session V. May 9th, 2011. Chicago, IL, USA. J. L. Martinez (*) : E. Luque-de-Len : G. Ballinas-Oseguera Department of General and Gastrointestinal Surgery, UMAE Hospital de EspecialidadesCentro Mdico Nacional Siglo XXI (IMSS), Av.Cuauhtmoc 330, 3er piso, Colonia Doctores, Delegacin Cuauhtmoc, Mxico, DF CP 06725, Mexico e-mail: jlmo1968@yahoo.com J. D. Mendez Unidad de Investigacin Mdica en Enfermedades Metablicas, UMAE Hospital de Especialidades, Centro Mdico Nacional Siglo XXI (IMSS), Mxico City, Mexico M. A. Jurez-Oropeza Facultad de Medicina, Universidad Nacional Autnoma de Mxico, Mxico City, Mexico R. Romn-Ramos Divisin Ciencias Biolgicas y de la Salud, Universidad Autnoma Metropolitana-Iztapalapa, Mxico City, Mexico

J. L. Martinez Programa de Doctorado de Ciencias Biolgicas y de la Salud, Universidad Autonoma Metropolitana-Iztapalapa, Mxico City, Mexico

J Gastrointest Surg (2012) 16:156164

157

Keywords Intestinal fistula . Intestinal fistula surgery . Intestinal fistula recurrence . Enterocutaneous fistula . Surgical complications

Introduction Management of enterocutaneous fistulas (ECF) is a complex and challenging problem commonly encountered by the general and specialized surgeon. Advances in parenteral nutrition (PN), radiological diagnosis, and treatment of intraabdominal abscesses and antibiotics among others have led to a decrease in mortality rates which were previously reported at 40% to the current 10%.13 When initially diagnosed most patients already suffer at least one of the frequent complications related to ECF such as sepsis, malnutrition, and fluid and electrolyte imbalance; these are the main causes of morbidity and mortality. Medical therapy aims to treat these complications and favor spontaneous closure. However, this is achieved in only 2550% of cases making surgical treatment necessary in most patients.4 Determining the best timing of surgery is a difficult decision mostly based on individual surgeons criteria. The goal is to withhold operative treatment until the patient has achieved resolution of all (or most) related complications and also until the abdominal cavity has become a less hostile environment. Recurrence of ECF after surgical closure has a devastating effect to both the patient and surgical team. Our aim in this study was to determine factors associated with recurrence and mortality in patients submitted to surgical repair of ECF.

Management for all patients was based on the scheme of four phases proposed by Chapman5 and Sheldon.6 In brief, upon diagnosis or referral, sepsis was ruled out; when present, source control measures were initiated. Utilization and choice of antibiotics was decided in a case-to-case basis. In most patients, ordinary methods for skin protection and collection of gastrointestinal fluids were utilized. Excluding patients with low output and/or distal ECF, most patients were started on PN. Indications to keep this type of nutritional support included high output, proximal ECF, and those unable to tolerate enteral nutrition (i.e., patients with distal obstruction, insufficient small bowel, and others). Medical (conservative) management continued until ECF spontaneous closure was achieved or a decision to undertake operative treatment was decided upon. Data Collection Information collected included demographic data, type of referral, initial pathologies, type of initial (causative) operation, number of abdominal operations performed related to origin and management of ECF, history of open abdomen management, number of ECF, site of origin, fistulous tract characteristics, presence of enteroatmospheric fistulas, output during 24 h, presence of fluid and electrolyte imbalance, malnutrition or sepsis at diagnosis or referral, serum albumin, development of sepsis during course of disease, and type and duration of nutritional support. Regarding operative treatment, indications, type of operations, and interval between last abdominal operation and attempt at surgical closure were also collected. In case of ECF recurrence, its management was recorded. Procedures performed outside the operating room (i.e., percutaneous drainage of an abscess), were not considered as part of this operative treatment. All patients were followed up until hospital discharge or death. After discharge, patients were followed-up at regular intervals in an ambulatory basis. Definition of Variables Dependent variables were: (a) ECF recurrence, defined as leakage of luminal (intestinal) contents from the surgical wound(s), after operative treatment; it was diagnosed clinically, and sometimes corroborated through imaging (contrast) studies and/or operatively, and (b) operative mortality, defined as death occurring during postoperative hospitalization or during the first 30 postoperative days in those patients that were discharged from the hospital. Our aim was to determine risk factors that were associated to them. These factors (independent variables) included age (<55/55 years old), gender (male/female), site of origin (esophageal, gastric, duodenal, jejunal, ileal, or colorectal), output during 24 h (<500 ml, low/500 ml, high), number of ECF (1, single/>1, multiple), type of

Materials and Methods We have been prospectively collecting data of all consecutive patients with ECF that are managed in our service. For the purpose of our study, we analyzed only those consecutive patients submitted to surgical repair during a 5year period (January 2005December 2009). A total of 123 patients were diagnosed and treated during the study period. Spontaneous closure was achieved in 45 (37%). Another seven patients (5%) died before spontaneous closure was achieved or surgical closure attempted. Thus, 71 (58%) of them were submitted to operative treatment during the course of their illness and constitute our study group. ECF were defined as abnormal communications that connect any portion of the gastrointestinal tract with the skin. Due to their different nature, treatment, and prognosis, patients with biliary, pancreatic, internal, or perianal fistulas were specifically excluded. Initial diagnosis was made by clinical examination and confirmed with imaging studies (fistulography, water-soluble and/or barium contrast studies, enteroclysis, etc.) and/or intraoperatively.

158

J Gastrointest Surg (2012) 16:156164 Table 1 Initial pathologies leading to development of ECF in 71 patients that eventually required surgical closure of their ECF Initial pathology Bowel obstruction Colostomy status Complicated appendicitis Trauma (firearm/blunt) Ileostomy status Complicated diverticular disease Hernia Others (24) ECF enterocutaneous fistula n=71 13 8 6 5 (3/2) 4 3 3 2 each

referral (referral from another hospital/ECF 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), presence of intestinal continuity (yes, lateral/no, terminal), type of initial (causative) operation (urgent/elective), history of open abdomen management (yes/no), drainage directly through a defect without overlying soft tissue (enteroatmospheric fistulas; yes/no), serum albumin both at diagnosis (or referral) and preoperative (g/dl), presence of malnutrition and fluid and electrolyte imbalance at diagnosis or referral (yes/ no), presence of sepsis at diagnosis (or referral) or development during the course of disease (yes/no), interval between last abdominal operation and our surgical attempt at ECF closure (weeks), and type of surgical treatment (ECF resection/other types of nonresective procedures). Statistical Analysis All data were collected and entered in a computerized Microsoft Excel database (Microsoft, Redmond, WA, USA). Analyses were performed with the statistical package program SPSS version 16 (SPSS, Chicago, IL, USA). Unless specified otherwise, numerical values are expressed as median (and range). To determine factors associated to ECF recurrence and mortality, Students t test was used to perform comparisons between continuous variables and Pearson chi-square test or Fischers exact test for categorical variables. All comparisons are two-tailed probabilities. Significance was determined at the 95% confidence interval (95% CI, p0.05). To determine independent risk factors, significant variables found through univariate analyses were then submitted to multivariate analyses (logistic regression). Odds ratios were calculated for these factors.

presented with complications related to the ECF which included malnutrition in 41 (58%), fluid and electrolyte imbalance in 24 (34%), and sepsis in 19 (27%). Six further patients (8%) developed sepsis during the course of disease. Initial medical management was established in all cases. A total of 69 (97%) received nutritional support which consisted of total parenteral nutrition (TPN) in 57, both parenteral and enteral in 10, and enteral only in 2. TPN was utilized for a median of 60 days (range, 5358). Octreotide was used in 22 (31%) patients for a median of 21 days (range, 1042). Surgical treatment (operative closure) was successful in 57 patients (80%). Median interval between last abdominal operation and attempt at ECF surgical closure was 63 days (range, 5979). Table 3 divides total number of patients from our original group of 123 according to location of ECF and lists need for operative closure, interval between last abdominal operation and attempt at surgical closure, recurrence after (attempted) surgical treatment and final figures of operative closure. This last column integrates operative closure at the first attempt and closure after treatment (medical or surgical) of ECF recurrence.

Results There were 36 females (51%) and 35 males (49%) with a median age of 52 years (range, 1781); 58 (82%) were referred from other hospitals. ECF origin was postoperative in 69 patients (97%), and spontaneous (Crohns disease and complicated diverticular disease) in 2 (3%). These latter, along with initial pathologies that led to operations which in turn originated ECF are listed in Table 1. Causative operations are included in Table 2. These were performed in an urgent basis in 40 patients (58%). ECF were located in the jejunum, ileum, colon, stomach, and duodenum in 26, 19, 19, 4, and 3 patients, respectively. Output was determined to be high in 39 patients (55%). A total of 27 (38%) were classified as enteroatmospheric fistulas, most of them (26) had a history of management with an open abdomen. At admission, several patients
Table 2 Initial operations leading to development of ECF in 69 patients that eventually required surgical closure of their ECF Initial operations Bowel resection Colostomy closure Adhesiolysis Appendectomy Ileostomy closure Ventral hernioplasty Oopherectomy Laparotomy (for trauma) Others (18) ECF enterocutaneous fistula n=69 12 8 7 4 4 4 3 3 2 each

J Gastrointest Surg (2012) 16:156164 Table 3 Original group of patients with ECF according to site and then submitted to operative closure (study group) Site of ECF Surgical closure of ECF study group (n) 4 3 26 19 19 71 Interval between last abdominal operation and surgical closure of ECF (daysa) 15 39 65 50.5 69.5 63 (593) (1248) (8368) (5798) (9979) (5979) ECF recurrence after (attempted) surgical closure (%) 2 (50) 0 (0) 10 (38) 8 (42) 2 (11) 22 (31)

159

Final closure (%) 3 1 20 15 18 57 (75) (33) (77) (79) (95) (80)

Stomach (n=11) Duodenum (n=8) Jejunum (n=36) Ileum (n=34) Colon (n=34) Total

Also depicted are interval between operations, recurrence rate and final closure rate ECF enterocutaneous fistula
a

Median (range)

There were several operative indications (Table 4) and procedures (Table 5) performed in order to treat those patients who did not achieve spontaneous closure of their ECF (our study group). Persistence without sepsis was the most frequent indication. If factors precluding spontaneous closure (such as mucosal eversion, distal occlusion, and terminal ECF) are added to the latter indication, only 17 patients (24%) were operated due to sepsis. In accordance to these figures, resection of the ECF with either primary anastomosis or ostomy (terminal or diverting) creation was feasible in 58 patients (82%). There was a correlation between operative indications and type of surgery performed, which displayed a trend toward avoiding anastomosis in patients with sepsis; primary anastomosis was attempted in only 1 out of 17 patients with sepsis, compared to 37 out of 54 patients without it. ECF Recurrence ECF recurred in 22 patients (31%) after attempts at operative closure. Medical (conservative) measures were established in nine, obtaining eventual closure in seven of them. The other 13 patients were submitted again to operative treatment, achieving closure in 11 of them. Thus, final ECF closure was attained in 18 of the 22 patients (82%) with ECF recurrence. ECF recurred in all four patients in whom direct repairoversew was attempted. Despite the small sample, this recurrence rate was significantly higher to that presented by patients submitted to
Table 4 Surgical indication in 71 patients with ECF

ECF resection and primary anastomosis or ostomy (terminal or diverting) creation (p<0.02). The benefit of including resection of ECF in the operative treatment is further demonstrated by comparing its recurrence rate to that of all operations in which ECF were not resected, which was also significantly higher (p<0.02). Univariate and multivariate analysis of factors related to ECF recurrence after surgical closure are shown in Table 6. Aside from type of surgery (see above), the former only identified noncolonic ECF site (p<0.04) and high output (p< 0.01) as factors that favored recurrence among all variables that were studied. After multivariate analysis, both high output and type of surgery (operations not including resection of ECF) were found to be independent factors related to recurrence with a risk over four times higher in each. Although not statistically significant, enteroatmospheric fistulas (p=0.06) and a history of open abdomen management (p=0.06) were factors that clearly showed a tendency to increase the risk of recurrence. Mortality A total of 14 patients (20%) died. Five of them had ECF recurrence after operative closure. Causes included abdominal sepsis (n=6), pulmonary sepsis (n=6), pulmonary thromboembolism (n=1), and acute myocardial infarction (n=1).
Table 5 Surgical procedures performed in 71 patients with ECF Surgical procedure n 37 21 4 4 3 2

Surgical indication Persistence without sepsis Sepsis Mucosal eversin Distal occlusion Terminal ECF

n 38 17 12 3 1

ECF enterocutaneous fistula, terminal ECF fistula without intestinal continuity

Resection and primary anastomosis Resection and ostomy Direct repairoversew Drainage of abscesses Small bowel bypass Tube placement ECF enterocutaneous fistula

160

J Gastrointest Surg (2012) 16:156164

Table 6 Factors associated with ECF recurrence in 71 patients submitted to operative closure of their fistula No recurrence (n=49) Age 55 Gender Male Female ECF site Stomach Duodenum Jejunum Ileum Colon History of open abdomen management Urgent operation Origin of patient (at our hospital) Complex ECF tract Multiple fistulas High output Malnutrition Sepsis Fluid and electrolyte imbalance Enteroatmospheric fistula Serum albumin <3 g/dl Preoperative use of TPN Interval between operations 20 weeks Preoperative serum albumin <3 g/dl Type of surgery Fistula nonresection Fistula resection Recurrence (n=22) Univariate analysis p Value 0.29 0.12 Multivariate analysis p Value Odds ratio (95% CI)

21 29 20 2 3 16 11 17 14 27 13 8 11 20 26 16 14 15 23 45 21 15 5 44

6 8 14 2 0 10 8 2 12 13 3 8 7 19 15 9 10 12 10 21 18 8 8 14

0.58 0.54 0.30 0.25 0.04 0.06 0.80 0.35 0.79 0.55 0.001 0.30 0.59 0.18 0.06 1.00 1.00 0.55 0.78 0.017

0.109

0.029

4.27 (1.1515.7)

0.039

4.55 (1.0819.2)

ECF enterocutaneous fistula, CI confidence interval, TPN total parenteral nutrition

Regarding mortality, we found several variables that were significantly associated in this group of operated patients, which included age 55 years (p<0.03), sepsis at admission (p< 0.01), malnutrition (p<0.05), fluid and electrolyte imbalance (p<0.01), serum albumin <3 g/dl (p< 0.05), interval between last abdominal operation and our surgical attempt at ECF closure 20 weeks (p<0.05), and preoperative serum albumin <3 g/dl (p<0.01). After multivariate analysis however, only age 55 years and preoperative serum albumin <3 g/dl prevailed as independent significant factors (Table 7). Interestingly, ECF recurrence was not associated with mortality.

Discussion In spite of several advances in the medical treatment of patients with ECF, almost 3080% of them eventually

require an operation during their management.4,7 In a previous study, we found proximal (jejunal) location, high output and multiple fistulas as factors associated with the need of operative treatment.1 Source control and definitive closure of ECF are the common goals of surgical therapy. While the former is the usual cause of early operations, the latter is indicated in patients with ECF persistence despite optimal medical treatment.7 Several factors that have been found to preclude spontaneous closure and favor ECF persistence include terminal fistulas, short fistulous tracts, fistula opening 1 cm2, distal occlusion, and high output.1,8,9 Moreover, enteroatmospheric fistulas which are those with mucosal exposure that drain directly to the skin or through a granulating wound, are associated with practically no possibility of spontaneous closure.911 One of the goals of medical treatment is to prepare patients for an eventual operation should it be required. Ideally prior to being operated, patients should be in the

J Gastrointest Surg (2012) 16:156164 Table 7 Factors associated with mortality in 71 patients with ECF submitted to operative closure Survivors (n=57) Age 55 Gender Male Female ECF site Stomach Duodenum Jejunum Ileum Colon History of open abdomen management Urgent operation Origin of patient (at our hospital) Complex tract Multiple fistulas High output Malnutrition Sepsis Fluid and electrolyte imbalance Enteroatmospheric fistula Serum albumin <3 g/dl Preoperative use of TPN Operation at 20 weeks Preoperative serum albumin <3 g/dl ECF recurrence 18 33 24 3 1 21 14 18 20 32 11 12 13 30 29 15 13 21 22 5 39 12 17 Death (n=14) 9 4 10 1 2 5 5 1 6 8 5 4 5 9 12 10 11 6 11 14 13 11 5 1.00 0.10 1.00 0.50 0.09 0.75 1.00 0.28 0.72 0.32 0.55 0.03 0.004 0.001 0.76 0.02 0.57 0.03 0.001 0.75 Univariate analysis p value 0.03 0.07 Multivariate analysis p value 0.043 Odds ratio (95% CI)

161

9.6 (1.487.0)

0.252 0.761 0.153 0.295 0.450 0.034

23.2 (1.2425.8)

ECF enterocutaneous fistula, CI confidence interval, TPN total parenteral nutrition

best possible conditions, free of septic complications and with adequate serum albumin and other biochemical parameters.4 The most frequent indication for operative treatment in our patients was ECF persistence in spite of having achieved control of sepsis. Clearly, those patients with ECF and abdominal sepsis that cannot be resolved with medical and/or radiological treatment(s), must be operated promptly in order to achieve source control.12,13 Median interval between the last abdominal operation and that performed in our unit was 17 days for those with abdominal sepsis and 76 days for those without it. Establishing the ideal timing to operate patients with ECF persistence in spite of optimal medical treatment, without abdominal sepsis and who may have factors that do not favor spontaneous closure, is indeed a more controversial issue. Recurrence The 31% recurrence rate found in our study is within the range reported by several authors.1418 Our analyses

disclosed that high output ECF have over four times more chances of recurring, and also that colonic ECF have a significantly less recurrence rate when compared to all other locations; this latter finding corroborates previous reports.16 Among surgical alternatives for definitive closure of ECF, the best results have been obtained with resection and primary anastomosis.13,15,19,20 Other options include resection and proximal ostomy,15,21 and resection with primary anastomosis and proximal (diverting) ostomy.4,15 Our results disclosed that ECF recurred almost five times more when the procedure did not include resection of the fistula. In fact, we, as others,15,16,19 found direct repairoversew of ECF led to higher recurrence rates; all ECF in whom it was attempted, recurred. It is important to emphasize that these four patients had enteroatmospheric fistulas (three had a history of management with an open abdomen). Several local operative techniques have been described in order to favor closure of these type of ECF such as direct repairoversew, cutaneous flaps, or biological meshes;

162

J Gastrointest Surg (2012) 16:156164

however, success rates only range from 40% to 60%.20,22,23 Moreover, in those cases where the technique fails, the defect of the new fistula tends to be larger than that of the original opening.23,24 Due to the fact that the inflammatory response in these type of fistulas is an ongoing process, a longer waiting interval of at least 6 12 months is recommended (instead of the 36 month period suggested for other ECF), prior to attempting definitive surgical closure.24 The interval between the last abdominal operation and the operative attempt to definitively close the ECF has been reported to be related to chances of recurrence. While some refer lower rates for intervals 12 weeks,15 others have found higher rates for waiting periods of 36 weeks;16 our results do not support one or the other. Our findings display a tendency toward higher recurrence rates for patients with a history of open abdomen management; this latter is frequently used for patients with severe abdominal sepsis and in whom multiple reentries to the abdominal cavity are required. Multiple fistula openings are common in these cases. Closure of the abdominal wall plays an important role in preventing recurrence. Ideally, it should be accomplished with the patients own tissues, when necessary even with the aid of a plastic and reconstructive surgeon in order to perform the component separation technique which has proven to be an adequate option in the management of these types of patients.17,25 In cases in which this is not feasible, in spite of having a higher incidence of incisional hernias, the use of biologic prosthesis (meshes) is preferred to those with polipropilene or even PTFE which are known to cause more ECF.21,25 Other factors related to recurrence include presence of abdominal sepsis at the moment of the operation19 and type of anastomosis (higher rates for mechanical than manual).16 Prognosis of patients with recurrent ECF in our series was as good as those with nonrecurrent ECF. Definitive closure was achieved in a total of 18 out of 22 (82%) patients (nine of them with conservative treatment). These figures compare favorably with other series who report closure rates of recurrent ECF (developed after surgical treatment) of 50%.1416 ECF recurrence after attempts at surgical closure represents a devastating event not only to the patient, but also to the surgical team who usually invest time and effort in order to bring these patients in the best possible conditions to the operating theater. It may lead to new (abdominal) septic events, which will require invasive procedures and/or operations for its control and may end up with a hostile abdominal environment. These patients require lengthy hospital stays and the best clinical advice is to consider recurrence a new ECF and treat it accordingly, with initial medical management and the usual operative timing.

Mortality Factors that have traditionally been associated with mortality are those directly related to complications of ECF such as sepsis, malnutrition, and fluid and electrolyte imbalance; these findings were no different in our own series. In relation to timing of surgery, we found that patients that were operated upon before week 20 had a significantly higher mortality rate than those operated afterwards. The longer interval of medical treatment prior to operative intervention prepares and delivers fitter patients with better general conditions to the operating room. Early operations in patients who are responding to source control and conservative measures seem ill advised. As was previously mentioned, it is important to note however, that patients with abdominal sepsis do need significantly earlier operations (than those without sepsis) in order to achieve source control. Thus, patients conditions (and not only surgeons choices), add into the equation in surgical decision making. The fact that patients with lower preoperative serum albumin had higher mortality rates further supports these findings. Interestingly, mortality rate for our patients with enteroatmospheric fistulas was 22%, which again compares favorably with the 50% reported in most series.2628 Overall, sepsis continues to be the most important factor related with mortality. Thus, any measure to prevent its development or control its effects should be attempted.

Conclusion Most patients with ECF require surgical treatment to achieve its resolution. Performing such operations in patients with the best possible physical conditions is an essential step in order to achieve successful results. Our findings indicate that patients with high output, small bowel, enteroatmospheric fistula, and/or with a history of open abdomen management have the highest chances of recurrence after operative closure attempts. Thus, preparation of this type of patients must be carefully accomplished prior to surgery. If recurrence develops, it should be considered as a new ECF and treated accordingly. This can lead to an excellent prognosis, which may be similar to that of the original ECF. Ideally, operative intervention should be performed in patients with optimal conditions, without evidence of complications related to ECF, especially sepsis. Our study demonstrated that presence of this latter was associated with higher mortality rates; interestingly, we did not find an association between ECF recurrence and increased deaths. Timing of surgery also plays a vital role; waiting a longer period seems to convey several benefits such as improving fluid and electrolyte

J Gastrointest Surg (2012) 16:156164

163 20. Reber HA, Roberts C, Way LW, Dunphy JE. Management of external gastrointestinal fistulas. Ann Surg 1978;188:460467 21. Jamshidi R, Schecter WP. Biological dressings for the management of enteric fistulas in the open abdomen: a preliminary report. Arch Surg 2007;142:793796 22. Fischer JE. The importance of reconstruction of the abdominal wall after gastrointestinal fistula closure. Am J Surg 2009;197:131132 23. Sarfeh IJ, Jakowatz JG. Surgical treatment of enteric bud fistulas in contaminated wounds. A riskless extraperitoneal method using split-thickness skin grafts. Arch Surg 1992;127:10271030 24. Schecter WP, Hirshberg A, Chang DS, Harris HW, Napolitano LM, Wexner SD, Dudrick SJ. Enteric fistulas: principles of management. J Am Coll Surg 2009;209:484491 25. Connolly PT, Teubner A, Lees NP, Anderson ID, Scott NA, Carlson GL. Outcome of reconstructive surgery for intestinal fistula in the open abdomen. Ann Surg 2008;247:440444. 26. Mastboom WJ, Kuypers HH, Schoots FJ, Wobbes T. Small-bowel perforation complicating the open treatment of generalized peritonitis. Arch Surg 1989;124:689692. 27. Sriussadaporn S, Sriussadaporn S, Kritayakirana K, Pak-art R. Operative management of small bowel fistulae associated with open abdomen. Asian J Surg 2006;29:17 28. Schein M, Decker GA. Gastrointestinal fistulas associated with large abdominal wall defects: experience with 43 patients. Br J Surg 1990;77:97100.

imbalance and malnutrition (which were related to mortality in our study), and finding a friendlier abdominal cavity, among others.
Acknowledgments Dr. Jose L. Martnez is a CONACYT (Mxico) Doctoral Fellow (Registration 224708) at Universidad Autnoma Metropolitana-Iztapalapa.

References
1. Martinez JL, Luque-de-Leon E, Mier J, Blanco-Benavides R, Robledo F. Systematic management of postoperative enterocutaneous fistulas: factors related to outcomes. World J Surg 2008;32:436443 2. Draus JM Jr, Huss SA, Harty NJ, Cheadle WG, Larson GM. Enterocutaneous fistula: are treatments improving? Surgery 2006;140:570576 3. Wainstein DE, Fernandez E, Gonzalez D, Chara O, Berkowski D. Treatment of high-output enterocutaneous fistulas with a vacuumcompaction device. A ten-year experience. World J Surg 2008;32:430435 4. Visschers RG, Olde Damink SW, Winkens B, Soeters PB, van Gemert WG. Treatment strategies in 135 consecutive patients with enterocutaneous fistulas. World J Surg 2008;32:445453 5. Chapman R, Foran R, Dunphy JE. Management of intestinal fistulas. Am J Surg 1964;108:157164 6. Sheldon GF, Gardiner BN, Way LW, et al. Management of gastrointestinal fistulas. Surg Gynecol Obstet 1971;133:385389 7. Soeters PB, Ebeid AM, Fischer JE. Review of 404 patients with gastrointestinal fistulas. Ann Surg 1979;190:189202 8. Fischer JE. The pathophysiology of enterocutaneous fistulas. World J Surg 1983;7:446450 9. Dudrick SJ, Maharaj AR, McKelvey AA. Artificial nutrition support in patients with gastrointestinal fistulas. World J Surg 1999;23:570576 10. Joyce MR, Dietz DW. Management of complex gastrointestinal fistula. Curr Probl Surg 2009;46:384430 11. Schecter WP, Ivatury RR, Rotondo MF, Hirshberg A. Open abdomen after trauma and abdominal sepsis: a strategy for management. J Am Coll Surg 2006;203:390396 12. Schein M. Postoperative small bowel leak. Br J Surg 1999;86:979980 13. Hill GL, Operative strategy in the treatment of enterocutaneous fistulas. World J Surg 1983;7:495501 14. Hollington P, Mawdsley J, Lim W, Gabe SM, Forbes A, Windsor AJ. An 11-year experience of enterocutaneous fistula. Br J Surg 2004;91:16461651 15. Lynch AC, Delaney CP, Senagore AJ, Connor JT, Remzi FH, Fazio VW. Clinical outcome and factors predictive of recurrence after enterocutaneous fistula surgery. Ann Surg 2004;240:825831. 16. Brenner M, Clayton JL, Tillou A, Hiatt JR, Cryer HG. Risk factors for recurrence after repair of enterocutaneous fistula. Arch Surg 2009;144:500505 17. Wind J, van Koperen PJ, Slors JF, Bemelman WA.Single-stage closure of enterocutaneous fistula and stomas in the presence of large abdominal wall defects using the components separation technique. Am J Surg. 2009;197:249. 18. Gyorki DE, Brooks CE, Gett R, Woods RJ, Johnston M, Keck JO, Mackay JR, Heriot AG.Enterocutaneous fistula: a single-centre experience .ANZ J Surg. 2010;80:17881. 19. McIntyre PB, Ritchie JK, Hawley PR, Bartram CI, Lennard-Jones JE. Management of enterocutaneous fistulas: a review of 132 cases. Br J Surg 1984;71:293296

Discussant
Dr. Stuart G. Marcus, MD (Bridgeport, CT, USA): I would like to commend Drs. Martinez and Luque-de-Leon on their dedication to this difficult population of patients and also for presenting a large amount of data in a clear and concise manner. These results, which are among the best reported in the literature, reflect excellent decision making and a clear understanding of the need to withhold surgery until the patient is in the best possible physical condition. The outstanding management and results of an average of one new patient approximately every 2 weeks over a 5-year period set a benchmark for others to meet in the surgical care of enterocutaneous fistulas. I have several questions for the authors: 1. You mention that 22 patients, or 31%, were treated with Octreotide. This was not factored into your analysis. Can you tell us what effect, if any, Octreotide had on ECF recurrence and mortality after surgery? 2. You clearly point out the importance of restoring nutrition, controlling sepsis and correcting electrolyte imbalance to achieve your excellent results. Another factor also considered important in patients with ECF is maintaining skin integrity. Can you tell us some lessons learned in this area? 3. In your manuscript, you conclude that sepsis continues to be the most important factor related to mortality. Seventeen of your patients were operated on due to sepsis and 12 died with sepsis, yet on multivariate analysis sepsis is not found to be a significant factor. How do you explain this discrepancy? I would like to thank the authors and the SSAT for the opportunity to discuss this paper. Closing Discussant Drs. Jose L. Martinez and Enrique Luque-de-Leon: 1. We did analyze Octreotide use as an independent variable and did not find any relationship with recurrence or mortality. The nonprotocolized form of its use (in regards to indications, duration, dosage, etc.) made us decide not to include it in our final presentation and manuscript. We are planning to standardize its use and include it in future protocols with specific patient populations.

164 2. Skin care is undoubtedly one of the cornerstones in the management of these patients. Since more than 80% of our patient population are referred from other hospitals, we have been faced with a wide range of skin derangements and problems. Our stoma care unit was established in 1998, and has been a great asset not only for patients with stomas, but also for those with ECF and those managed with an open-abdomen. In general, once ECF control is achieved, skin integrity is maintained with transparent films without alcohol (i.e., proshell, adapt and stomadhesive) or karaya. Treatment of those referred patients with diverse skin lesions starts with its classification (discoloration erosion tissue growth, DET); hydrocolloid and alginate wound dressings and powder as well as transparent films are part of the armamentarium for their management. 3. As shown in our tables, our analysis was based on patients that had sepsis at admission or referral (19) and those that developed it

J Gastrointest Surg (2012) 16:156164 prior to surgical treatment (6); 10 of these 25 patients died for p values of 0.004 (univariate) and 0.761 (multivariate). As Dr. Stuart points out, there are however other ways to look at the numbers. For example, sepsis was the surgical indication in 17 patients (which are part of the 25 patients mentioned above). The fate of these 17 patients included death in 8 due to abdominal sepsis (4), pulmonary sepsis (3), and pulmonary thromboembolism (PTE; 1). The results are similar if the analysis is based in these numbers (p values of 0.003 and 0.57, respectively). Perhaps a greater sample would confirm sepsis as an independent factor related to mortality. Overall, 14 patients died; causes included abdominal sepsis (6), pulmonary sepsis (6), PTE (1, who had sepsis during course of disease) and acute myocardial infarction (1). Thus, only this latter patients dismissal was completely unrelated to sepsis. This only confirms presence or development of sepsis as a major factor related to outcome.