Bradley MJ, DuBose JJ, Scalea TM, Holcomb JB, Shrestha B, Okoye O, Inaba K, Bee TK, Fabian TC, Whelan JF, Ivatury RR, for the AAST Open Abdomen Study Group. Independent Predictors of Enteric Fistula and Abdominal Sepsis After Damage Control LaparotomyResults From the Prospective AAST Open Abdomen Registry. JAMA Surg. 2013;148(10):947-955. doi:10.1001/jamasurg.2013.2514
Enterocutaneous fistula (ECF), enteroatmospheric fistula (EAF), and intra-abdominal sepsis/abscess (IAS) are major challenges for surgeons caring for patients undergoing damage control laparotomy after trauma.
To determine independent predictors of ECF, EAF, or IAS in patients undergoing damage control laparotomy after trauma, using the AAST Open Abdomen Registry.
The AAST Open Abdomen registry of patients with an open abdomen following damage control laparotomy was used to identify patients who developed ECF, EAF, or IAS and to compare these patients with those without these complications. Univariate analyses were performed to compare these groups of patients. Variables from univariate analyses differing at P < .20 were entered into a stepwise logistic regression model to identify independent risk factors for ECF, EAF, or IAS.
Fourteen level I trauma centers.
A total of 517 patients with an open abdomen following damage control laparotomy.
Main Outcomes and Measures
Complication of ECF, EAF, or IAS.
More patients in the ECF/EAF/IAS group than in the group without these complications underwent bowel resection (63 of 111 patients [57%] vs 133 of 406 patients [33%]; P < .001). Within the first 48 hours after surgery, the ECF/EAF/IAS group received more colloids (P < .03) and total fluids (P < .03) than did the group without these complications. The ECF/EAF/IAS group underwent almost twice as many abdominal reexplorations as did the group without these complications (mean [SD] number, 4.1 [4.1] vs 2.2 [3.4]; P < .001). After multivariate analysis, the independent predictors of ECF/EAF/IAS were a large bowel resection (adjusted odds ratio [AOR], 3.56 [95% CI, 1.88-6.76]; P < .001), a total fluid intake at 48 hours of between 5 and 10 L (AOR, 2.11 [95% CI, 1.15-3.88]; P = .02) or more than 10 L (AOR, 1.93 [95% CI, 1.04-3.57]; P = .04), and an increasing number of reexplorations (AOR, 1.14 [95% CI, 1.06-1.21]; P < .001).
Conclusions and Relevance
Large bowel resection, large-volume fluid resuscitation, and an increasing number of abdominal reexplorations were statistically significant predictors of ECF, EAF, or IAS in patients with an open abdomen after damage control laparotomy.
Open abdominal (OA) management has emerged as a common component in the treatment of severe abdominal trauma and complex surgical pathologies.1,2 As experience with this practice has evolved, however, so has an increasing appreciation of the potential complications associated with the use of OA management. Enterocutaneous fistula (ECF), intra-abdominal sepsis/abscess (IAS), deep soft tissue infection, and ventral incisional hernia are a few of the challenges surgeons face with regard to caring for these patients. An ECF, which is often referred to as an enteroatmospheric fistula (EAF) in the setting of an open abdomen, is one of the most devastating of posttraumatic complications in this patient population. With an incidence of 4.5% to 25%,3- 7 EAFs are associated with considerable morbidity and mortality.8 An EAF combined with an IAS represents one of the most significant management dilemmas in this population.
To date, however, the risk factors for EAF and IAS following OA management has not been well elucidated. The primary aim of the present study was to use the American Association for the Surgery of Trauma (AAST) Open Abdomen Registry9 to determine the independent predictors for the occurrence of EAF or IAS in patients undergoing damage control laparotomy after trauma. It is our hope that, by identifying these risk factors, additional study can be focused on mitigating their effect on subsequent outcomes among patients.
Our study used the AAST Open Abdomen registry, which was created through a prospective observational multi-institutional study sponsored by the AAST. The study protocol was approved by the AAST Multi-Institution Trials Committee, and each participating center obtained approval from its own institutional review board. Patients with an open abdomen following damage control laparotomy were prospectively enrolled over a 2-year period from 2010 through 2011. The inclusion criterion was nonclosure of fascia following the initial trauma laparotomy. Patients younger than 18 years of age and pregnant patients were excluded from the present study.
Participating centers on the AAST multicenter study portal securely uploaded data. These data were subsequently collected at the end of the study period and analyzed by the principal investigators. All patients included in the present study had basic demographic data, admission laboratory values, intraoperative details, injury patterns, fluid use (both intraoperatively and postoperatively for the first 48 hours), ventilator settings, and OA management documented. All variables were defined using a data dictionary specifically designed for the study to promote uniformity of data collection. These included the definition of intraoperative acidosis as any pH measurement of less than 7.35 intraoperatively and the definition of intraoperative hypothermia as a recorded temperature below 35.0°C at any point during the initial operation.
The incidence of EAF and IAS in the AAST Open Abdomen Registry has been previously reported.9 Enteric fistula was defined as any leakage of the enteric contents from the lumen of the gastrointestinal tract (excluding created ostomies). They were termed enterocutaneous (ie, ECF) when communicating to an epithelialized external opening and enteroatmospheric (ie, EAF) when draining freely into the open abdomen field itself (the most common occurrence in the setting of an open abdomen). An IAS was defined as a nonfreely draining collection associated with such enteric leakage sources. The primary aim of the present study was to use the AAST Open Abdomen Registry to determine the independent predictors for the occurrence of these complications in patients undergoing damage control laparotomy after trauma. This secondary end point was a priori defined in the design of the original AAST Open Abdomen Study.9
Patients who developed an ECF, EAF, or IAS were compared with those who did not. Univariate analyses were performed to compare these 2 groups of patients. The student t test was used to compare continuous variables, and the Pearson χ2 test or the Fisher exact test was used to compare proportions. Two-tailed comparisons were used in all cases when available. Variables from the univariate analysis differing at P < .20, as well as clinically important variables, were entered into a stepwise logistic regression model to identify independent risk factors for the development of ECF, EAF, or IAS.
A total of 517 patients from the 14 level I trauma centers participating in our study were included. The mean (SD) age of the patients enrolled was 39 (17) years, with 20% of this population 55 years of age or older. The majority of the patients were male (79%) and were predominantly victims of blunt trauma (61%). Severe abdominal injury (Abbreviated Injury Score of ≥3) was a common condition in this population (84%) and was associated with a global injury burden (ie, 85% of these patients had a Injury Severity Score of ≥15). The mean (SD) time to successful definitive fascial closure was 4.48 (7.94) days. In comparison, patients whose primary fascial closure was unsuccessful had a mean (SD) time to attempt at closure (ie, initial visceral coverage) of 11.45 (13.97) days (P < .001).
Patients were divided into 2 subpopulations: those who had a documented ECF, EAF, or IAS after damage control laparotomy and those who did not (Table 1). Among the various demographic and clinical characteristics compared between these 2 groups, only a head injury with an Abbreviated Injury Score of 3 or greater was significantly different, with half as many patients in the ECF/EAF/IAS group sustaining severe head injury compared with patients in the group without complications (95 of 334 of patients without complications [28%] vs 14 of 92 of patients with [15%]; P = .01). There was no statistically significant difference in laboratory test results at admission or in intraoperative management between the 2 groups (Table 2). Sixty percent of the patients were operated on within the first 2 hours of injury, and 69% were left with an open abdomen for damage control laparotomy. Acidosis was the most common of the “deadly triad” indicators cited for damage control laparotomy, specifically referenced in 70% of patients (Table 3). Perioperative antibiotics were administered to most patients (88%), and only a minority of patients (14%) had an estimated blood loss of 5 L or greater. The majority of patients received less than 5 L of crystalloids (75%) and less than 5 L of intraoperative blood products (74%).
Among the operative interventions (Table 4), univariate analysis revealed that patients in the ECF/EAF/IAS group required more bowel resections than did patients in the group without complications (63 of 111 patients with complications [57%] vs 133 of 406 patients without [33%]; P < .001) and were more commonly left with bowel in discontinuity after the initial surgery (44 of 111 patients with complications [40%] vs 76 of 406 patients without [19%]; P < .001). In the first 24 hours after surgery, significantly more patients in the ECF/EAF/IAS group than in the group of patients without complications received higher volumes of crystalloid (5-10 L or >10 L; P = .02). Within the first 48 hours after surgery, the ECF/EAF/IAS group received more colloids (mean [SD] volume, 1937.4 [4751.8] L vs 1091.8 [3089.6] L; P < .03) and total fluids (mean [SD] volume, 10 081.6 [9580.7] L vs 7866.5 [9394.9] L; P < .03) than did the group of patients without complications (Table 5). The ECF/EAF/IAS group also underwent almost twice as many abdominal reexplorations as did the group of patients without complications (mean [SD] number, 4.1 [4.1] vs 2.2 [3.4]; P < .001).
After multivariate analysis, we found that the independent predictors of ECF/EAF/IAS development were large bowel resection (adjusted odds ratio [AOR], 3.56 [95% CI, 1.88-6.76]; P < .001), total fluid intake at 48 hours between 5 and 10 L (AOR, 2.11 [95% CI, 1.15-3.88]; P = .02) or greater than 10 L (AOR, 1.93 [95% CI, 1.04-3.57]; P = .04), and number of reexplorations (AOR, 1.14 [95% CI, 1.06-1.21]; P < .001) (Table 6).
The development of fistula after trauma can lead to a host of complications, including malnutrition, fluid losses, electrolyte abnormalities, and complex wound care issues.10 As a result, fistulas are associated with considerable morbidity and mortality. In a study of more than 2000 patients undergoing laparotomy for trauma, Teixeira et al8 found that the patients who developed a fistula had a significantly increased length of intensive care unit stay (mean [SD] duration, 28.5 [30.5] days vs 7.6 [9.3] days; P = .004) and hospital stay (82.1 [100.8] days vs 16.2 [17.3] days; P < .001) compared with patients who did not. In addition, the mean hospital charges were significantly higher in those patients with postlaparotomy fistulas than in those without ($539 309 vs $126 996; P < .001). In another review of 2224 patients requiring laparotomy after trauma over a 10-year time frame, Fischer and colleagues11 found a mortality rate of 14% after an average 59-day intensive care unit length of stay for those patients who developed an ECF. These investigations highlight the significant burden associated with the development of fistula after trauma.
Despite the adverse sequelae associated with fistulas, little is known about the risk factors for their development, specifically in the setting of OA management. To our knowledge, our study represents the largest series to date identifying the independent predictors of ECF/EAF/IAS in OA management after damage control laparotomy. Our data show that large bowel resection, large volume resuscitation, and an increasing number of reexplorations were statistically significant independent predictors for the development of a fistula in an open abdomen after trauma.
Multiple studies have suggested that the method of anastomosis does not likely influence the rate of anastomotic complications. A large multicenter prospective study of 297 patients sponsored by the AAST and conducted by Demtriades et al12 compared the incidence of colon-related complications between patients with a hand-sewn anastomosis and patients with a stapled anastomosis after colon resection following penetrating trauma. These investigators12 found that there was no difference in the incidence of anastomotic leak or fistula between the 2 groups of patients.
Although current data suggest that the type of anastomosis has not been shown to be a risk factor for the development of fistula, we found that the location of resection (specifically the large bowel) can be a significant risk factor. The aforementioned large study by Teixeira et al8 had similar findings. This group of investigators8 determined that 89% of patients with an ECF had a hollow viscus organ injury and that 69% of these patients had sustained a colonic injury. Our data did not account for the specific location of the colonic resection, and therefore we were unable to relate the specific location of the colon injury to the subsequent risk for fistula. Current literature, however, supports that the risk for complications after colonic resection for injury (abscess, leak, or fistula) is not dependent on the specific location of the original colonic injury.13
Damage control resuscitation (DCR) has emerged as a common component in trauma management, focusing on permissive hypotension, minimizing crystalloids, and maintaining close blood product transfusion ratios.14,15 Several studies have demonstrated a survival advantage after civilian trauma when practices minimizing the use of crystalloids in resuscitation are used. A retrospective cohort study by Cotton et al16 of 390 patients who underwent a damage control laparotomy that compared patients who underwent DCR with a group of patients who were resuscitated with traditional (pre-DCR) practices found a statistically significant increase in 30-day survival in the DCR group (AOR, 2.50 [95% CI, 1.10-5.58]; P = .03). The DCR group also received less crystalloids during the immediate perioperative period than did the pre-DCR group (5 L vs 14 L; P < .05).
In a DCR survival study of 124 patients, Duchesne and colleagues17 found a survival benefit for DCR (AOR, 0.19 [95% CI, 0.05-0.33]; P = .005), with less crystalloids given intraoperatively to the DCR group than to the pre-DCR group (4.7 L vs 14.2 L; P = .009). Focusing on large-volume crystalloid infusion and the colon anastomotic leak rate, a retrospective review by Schnuriger et al18 of 123 patients who underwent colonic anastomosis following trauma identified large-volume resuscitation of greater than 10.5 L in the first 72 hours as an independent predictor of anastomotic leakage (AOR, 5.26 [95% CI, 1.14-24.39]; P = .03). Our study did not account for blood product ratios, but there were no statistically significant differences between our study groups with regard to the amount of blood loss or blood administered. However, there was a significant difference in the amount of crystalloids given in the immediate postoperative period, with more crystalloid infused in the ECF/EAF/IAS group than in the group without complications.
Minimizing bowel manipulation and local trauma to an edematous, friable bowel may decrease the rate of fistula development.19,20 Likewise, early primary fascial closure has been associated with lower complication rates. Miller et al5 performed a retrospective study of 344 trauma patients requiring open abdomens after initial damage control surgery. They found a statistically significant difference in the rate of fistula development between the delayed primary facial closure group and the temporizing fascial closure group (3% vs 30%; P < .001). In addition, they found that the percentage of all complications, including EAF and abdominal sepsis, was significantly increased in patients closed after 8 days vs those patients closed prior to 8 days from initial surgery (52% vs 12%; P < .01). In a similar retrospective review of more than 200 trauma patients managed with an open abdomen, Burlew and colleagues21 found that, compared with patients obtaining fascial closure earlier, patients obtaining fascial closure beyond 5 days were 4 times more likely to develop an anastomotic leak (3% vs 12%; P = .02). These data are consistent with our findings that multiple reexplorations represent significant risk for the subsequent development of ECF, EAF, or IAS following OA management in the trauma setting.
A variety of techniques have been reported in the literature for temporary abdominal closure following damage control laparotomy. Some of the most commonly used techniques in the modern era include vacuum-assisted closure (VAC [ie, the V.A.C. System; Kinetics Concepts, Inc]), other negative pressure therapy dressings, the Wittman patch, or a Bogata silastic bag. A systematic review by Boele van Hensbroek et al22 of 51 articles and more than 3000 patients reported the success rates of fascial closure and the morbidity and mortality rates associated with various closure techniques. In their review, Boele van Hensbroek et al22 determined that the Wittman path (90%), dynamic retention sutures (85%), and the VAC dressing (60%) were the methods with the highest fascial closure rate. The lowest mortality rates were associated with the Wittman patch (17%), the VAC dressing (18%), and dynamic retention sutures (23%), whereas the lowest fistula rates were associated with the Bogata silastic dressing (0%), the Wittman patch (2%), and the VAC dressing (2.9%). The current Eastern Association for the Surgery of Trauma recommendations for temporary abdominal closure, based on level II data, suggest using the Bogata bag, the Wittman patch, or the Vacuum pack.23 The authors of these recommendations also recommend against the use of permanent mesh owing to the high fistula rates associated with its use.23
The use of the VAC has increased with OA management, and case series have even reported the use of negative pressure therapy dressing as a successful option for managing EAFs in open abdomens.24 Other reports, however, have cautioned its use in the setting of an open abdomen because it has been suggested as a possible contributor to the development of fistulas. In a prospective randomized trial of 51 patients comparing VAC with polyglactin mesh, Bee et al25 found a fistula rate of 21% in the VAC group. In a report of 29 patients undergoing VAC management of open abdomens, Rao and colleagues26 identified 6 patients (21%) who developed a fistula during VAC treatment. More than 90% of the patients in our registry received VAC treatment after their initial damage control laparotomy for trauma. We were not, however, able to discern if the use of vacuum therapy was associated with a subsequent risk factor for ECF, EAF, or IAS in the study population.
Although we used a prospective design, our study had important limitations. Our design could not control for several potential confounding variables. Correction for variability in practice among the 14 trauma centers may not be adequately captured in our analysis and represents one of several important limitations inherent in our study design. In addition, the cause-and-effect relationship between fistula development and the duration of OA management and the timing/use of various closure techniques could not be precisely determined by our design; a patient’s abdomen may have remained open because he or she developed a fistula, or the fistula may have developed because of a prolonged open abdomen. This design shortcoming hindered our ability to make any meaningful conclusions regarding the precise effect of closure or the technique used for closure on the subsequent development of ECF, EAF, or IAS. Our data also did not account for the time frame of fistula development and did not capture whether the fistula was specifically attributed to anastomotic breakdown, although available data suggest that the majority of fistulas may be associated with anastomotic leak.27 Additional parameters that might have a potential effect on the development of a fistula and infectious complications, including diabetes, obesity, and antecedent immunosuppressive medications, were not included comprehensively in analysis. Our failure to scrutinize the potential effect of these variables on resulting complications should be considered in extrapolating our results to patients with these comorbidities. Finally, because of the observational design, no control groups were used for comparison in our study, which would have been ideal.
The management of trauma patients with open abdomens continues to be challenging and complex, especially in the event of fistula development. Further studies are warranted to optimize the prevention, treatment, and outcomes of patients with fistulas in open abdomens after damage control laparotomy. Studies focusing on resuscitation strategies and improvements in techniques of early fascial closure are likely to lead to superior patient outcomes.
Accepted for Publication: February 13, 2013.
Corresponding Author: Joseph J. DuBose, MD, Division of Trauma and Acute Care Surgery, University of Maryland Medical System, R. Adams Cowley Shock Trauma Center, 22 S Greene St, T5R46, Baltimore, MD, 21201 (email@example.com).
Published Online: August 21, 2013. doi:10.1001/jamasurg.2013.2514.
Author Contributions: Study concept and design: DuBose, Scalea, Holcomb, Inaba, Ivatury.
Acquisition of data: Bradley, DuBose, Shrestha, Okoye, Inaba, Bee, Fabian, Whelan, Ivatury.
Analysis and interpretation of data: Bradley, DuBose, Okoye, Inaba.
Drafting of the manuscript: Bradley, DuBose, Okoye.
Critical revision of the manuscript for important intellectual content: Bradley, DuBose, Scalea, Holcomb, Shrestha, Inaba, Bee, Fabian, Whelan, Ivatury.
Statistical analysis: Bradley, DuBose, Okoye.
Administrative, technical, or material support: Bradley, DuBose, Scalea.
Study supervision: DuBose, Scalea, Holcomb.
Conflict of Interest Disclosures: None reported.
Group Information: The AAST Open Abdomen Study Group members were Binod Shrestha, MD, and John B. Holcomb, MD, University of Texas Houston Medical Center, Houston; Kenji Inaba, MD, Obi Okoye, MD, and Agathoklis Konstantinidis, MD, Los Angeles County and University of Southern California Hospital; Thomas M. Scalea, MD, Jay Menaker, MD, and Joseph J. DuBose, MD, R. Adams Cowley Shock Trauma Center, University of Maryland Medical Center, Baltimore; James F. Whelan, MD, Rao R. Ivatury, MD, and Stephanie R. Goldberg, MD, Virginia Commonwealth University, Richmond; Martin D. Zielinski, MD, and Donald Jenkins, MD, Mayo Clinic Trauma Centers, Rochester, New York; Stephen Rowe, MD, Darrell Alley, MD, John Berne, MD, and LaDonna Allen, RN, East Texas Medical Center, Tyler; Paola G. Pieri, MD, and Starre Haney, RN, MS, Maricopa Integrated Health System, Phoenix, Arizona; Jeffrey A. Claridge, MD, and Katherine Kelly, MD, MetroHealth Medical Center, Cleveland, Ohio; Tiffany K. Bee, MD, and Timothy C. Fabian, MD, University of Tennessee Health Science Center, Memphis; Raul Coimbra, MD, PhD, and Jay Doucet, MD, University of California San Diego School of Medicine; Ben Coopwood, MD, David Keith, MD, and Carlos Brown, MD, University of Texas Southwestern–Austin, University Medical Center Brackenridge; James M. Haan, MD, and Jeanette Ward, BA, Via Christi Hospital, St Francis Campus, Wichita, Kansas; Stuart M. Leon, MD, Evert Eriksson, MD, and Debbie Couillard, RN, BSN, The Medical University of Southern Carolina, Charleston; and Marc A. de Moya, MD, and Gwendolyn M. van der Wilden, MSc, Massachusetts General Hospital, Boston.
Additional Contributions: The AAST Open Abdomen Study group would like to sincerely thank the following individuals, without whom this study could not have been completed: Meghan Spencer, Rebecca Grace Lopez, MS4, Jinfeng Han, BSN, Judith S. Katzen, MS, RN, Terry Curry, RN, Sadia Ali, MPH, and Mary Waage.
Correction: This article was corrected on February 5, 2014, for a misspelling of the surname of an AAST Open Abdomen Study Group member in the Group Information section.