1. The Journal of TRAUMA Injury, Infection, and Critical Care
A Ten-Year Retrospective Review: Does Pyloric Exclusion
Improve Clinical Outcome After Penetrating Duodenal and
Combined Pancreaticoduodenal Injuries?
Mark J. Seamon, MD, Paola G. Pieri, MD, Carol A. Fisher, BA, John Gaughan, PhD, Thomas A. Santora, MD,
Abhijit S. Pathak, MD, Kevin M. Bradley, MD, and Amy J. Goldberg, MD
Objectives: We sought to determine injuries, postoperative complications, length Conclusion: In our study population,
whether the performance of pyloric exclu- of hospital stay, and mortality. the performance of pyloric exclusion for
sion during repair of penetrating advanced Results: Fifteen of 29 patients were penetrating advanced duodenal injury
duodenal injuries prevents postoperative managed without pyloric exclusion and 14 and combined pancreatic and duodenal
duodenal fistulas and improves clinical with exclusion. Both groups were similar injuries did not improve clinical outcome.
outcome. with respect to age, sex, mechanism, in- The trend toward a greater overall com-
Methods: A retrospective chart re- jury grade, ISS, hemodynamic stability, plication rate, pancreatic fistula rate, and
view of patients from 1995 to 2004 with the presence of vascular injury, associated increased length of hospital stay in the
penetrating duodenal injuries >grade II abdominal injuries, and mortality rates. A pyloric exclusion group suggests that sim-
and all combined pancreaticoduodenal in- trend toward a higher overall complica- ple repair without pyloric exclusion is
juries was performed. Patients managed tion rate (71% vs. 33%), pancreatic fistula both adequate and safe for most penetrat-
either without or with pyloric exclusion rate (40% vs. 0%), and length of hospital ing duodenal injuries.
were compared on the basis of age, sex, stay (24.3 days vs. 13.5 days) was evident Key Words: Duodenal injury, Duode-
mechanism, injury grade, Injury Severity in the pyloric exclusion group. No duode- nal fistula, Pyloric exclusion.
Score (ISS), hemodynamic stability, the nal fistula was detected in either patient
presence of vascular injury or associated group.
J Trauma. 2007;62:829 – 833.
lthough penetrating mechanisms account for the major- Operative attempts that focus on the prevention of fistu-
ity of duodenal injuries, previous reports estimate that las by protecting the new surgical repair add operative time,
duodenal injuries comprise less than 5% of all abdom- alter the gastrointestinal anatomy, and carry an appreciable
inal injuries.1 Given the infrequency of penetrating duodenal morbidity. Although several authors have attempted to define
injuries, there is no clear consensus for their treatment. In- which advanced duodenal injuries may require these more
novative procedures such as duodenal “diverticulization”, sophisticated procedures, these studies failed to compare out-
pyloric exclusion, and “triple tube” drainage, have been de- comes after these more complex procedures with simple
veloped to both repair duodenal wounds and divert gastroin- repair alone.1– 4 We sought to determine whether the addition
testinal secretions to prevent duodenal fistulas. With up to 10 of pyloric exclusion to the repair of advanced penetrating
L of ingested food and enzymatically active digestive secre- duodenal and combined pancreaticoduodenal injuries pre-
tions passing through the duodenum each day, the most vents duodenal fistula formation and results in improved
serious complication related to the surgical repair of duodenal clinical outcomes.
wounds is the duodenal fistula.
PATIENTS AND METHODS
Submitted for publication July 11, 2006. After Institutional Review Board approval, we per-
Accepted for publication January 8, 2007. formed a retrospective chart review of all patients admitted to
Copyright © 2007 by Lippincott Williams & Wilkins, Inc. a Level I trauma center between January 1995 and December
From the Department of Surgery (M.J.S., C.A.F., T.A.S., A.S.P.,
K.M.B., A.J.G.), Temple University School of Medicine, Philadelphia, Penn- 2004, which revealed 54 patients with duodenal injuries.
sylvania; Department of Surgery, Division of Surgical Critical Care (P.G.P.), Twenty-five patients were excluded (10 with blunt injuries, 6
University of Maryland Medical Center, Baltimore, Maryland; and the De- who died within 48 h of massive associated injuries, 6 with
partment of Physiology and Biostatistics (J.G.), Temple University School of grade I duodenal injuries without pancreatic involvement that
Medicine, Philadelphia, Pennsylvania. did not require operative repair, 3 who underwent more
Presented as a poster at the 19th Annual Meeting of the Eastern Associ-
ation for the Surgery of Trauma, January 10 –14, 2006, Orlando, Florida. extensive procedures including 2 duodenal diverticulariza-
Address for reprints: Mark J. Seamon, MD, Department of Surgery, tions, and 1 Whipple’s procedure). The remaining 29 patients
Temple University Hospital, 3401 North Broad St., Philadelphia, PA 19104; had penetrating duodenal injuries grade II by American
email: email@example.com. Association for the Surgery of Trauma (AAST) criteria5 or
DOI: 10.1097/TA.0b013e318033a790 combined penetrating pancreaticoduodenal injuries.
Volume 62 • Number 4 829
2. The Journal of TRAUMA Injury, Infection, and Critical Care
Injuries were diagnosed and graded during laparotomy.
Operative repair was dictated by surgeon preference. Duode-
nal wound repairs were hand-sewn, one or two layer closures.
Pyloric exclusions were stapled distal to the pylorus and
included decompressive retrograde duodenostomy tubes.
Gastrojejunostomies were either hand-sewn or stapled.
Patients repaired without pyloric exclusion were com-
pared with patients repaired with pyloric exclusion and de-
scriptive statistics with the appropriate post hoc analysis
applied ( 2, Fisher exact test, t test). A p value less than 0.05
was considered statistically significant. Recorded and ana-
lyzed data included age, sex, mechanism of injury, grade of
injury, Injury Severity Score (ISS; calculated using Abbrevi-
ated Injury Scales [AIS] for head, face, chest, abdomen,
Fig. 1. Eighty-one associated injuries were found in 29 patients
extremities, and external; the three most severely injured
(mean 2.8/patient). The liver was the most frequently associated
body regions were each squared (x2) and then summed to
organ injured (19/29, 66%), but vascular (14/29, 48%), pancreatic
yield the ISS), evidence of hypovolemic shock (systolic pressure
(13/29, 45%), colon (10/29, 34%), small bowel (7/29, 24%), and
90 mm Hg), major vascular injuries, associated abdominal
stomach (7/29, 24%) injuries were still common. Gallbladder (6/29,
injuries, postoperative complications, fistula formation, length of
21%), kidney (4/29, 14%), and spleen (1/29, 3%) were also injured.
hospital stay, and mortality (survival 48 h).
RESULTS marily comprised gun shot victims (no PE, 67% vs. PE, 87%,
The study population was 100% male with a mean age of p 0.169). Although a statistical difference in duodenal
29 years (range, 19 – 69 years). All had penetrating injuries, injury grade between groups was not identified, a trend to-
of which 23 of 29 (79%) suffered gun shot wounds, and 6 of ward greater injury severity was noted in the pyloric exclu-
29 (21%) were stabbed. Seventeen of 29 (59%) patients had sion group (no PE, 2.5 0.7 vs. PE, 3.0 0.6, p 0.064).
multiple duodenal injuries— 46 duodenal wounds were dis- The single patient in the study population with a grade I
covered in total. The majority of wounds sustained (30 of 46, duodenal injury had an associated pancreatic injury and was
65%) were to the second portion of the duodenum (Table 1). repaired without the protection of a pyloric exclusion. Nine
The remaining injuries were distributed anatomically as fol- patients (31%) suffered grade II duodenal injuries, of which
lows: first portion, 4 injuries (9%); third portion, 9 injuries six were repaired without pyloric exclusion and three repaired
(20%); fourth portion, 3 injuries (7%). None of these wounds with pyloric exclusion. Sixteen of the 29 patients (55%) had
involved the ampullary complex. All patients in this study grade III injuries. Eight of this group of 16 with grade III
suffered injuries to abdominal organs other than the duode- injuries were primarily repaired without exclusion, and an-
num. In total, 81 associated injuries were identified in these other eight were treated with pyloric exclusion. Three suf-
29 patients (mean 2.8 associated injuries per patient). The fered severe grade IV duodenal trauma, all of which were
liver was the most frequently injured associated organ (19 of combined pancreaticoduodenal injuries and all treated with
29, 66%), but major vascular (14 of 29, 48%), pancreatic (13 pyloric exclusion (Table 2).
of 29, 45%), colon (10 of 29, 34%), small bowel (7 of 29, Thirteen of the 29 patients suffered combined pancreati-
24%), stomach (7 of 29, 24%), and gallbladder (6 of 29, 21%) coduodenal injuries. Ten of these 13 (77%) patients underwent
injuries were also common. Kidney and splenic injuries were pyloric exclusion, whereas 3 (23%) patients had primary repair
less frequent (Fig. 1). alone. Patients suffering combined pancreaticoduodenal injuries
Patients repaired without pyloric exclusion (no PE, n (n 13) had similar pancreatic injury grades between compar-
15) and with pyloric exclusion (PE, n 14) were similar with ison groups (no PE 2.7 2.1 vs. PE 2.0 0.9, p 0.799). ISS
respect to age, sex, and injury mechanism. Both groups pri- scores, a global indicator of injury severity in patients with
multiple injuries, were also similar between groups (no PE,
18.4 8.3 vs. PE, 23.2 17.5, p 0.631). Shock (no PE, 33%
Table 1 Duodenal Injury Location (n 46) vs. PE, 21%, p 0.682) and vascular injury (no PE, 53% vs. PE,
Location No. Wounds (%) 43%, p 0.847) were similarly present in both patient groups.
After confirming that our two surgical treatment groups
D1 4 (9)
D2 30 (65) had similar demographics and clinical characteristics, clinical
D3 9 (20) outcomes were compared (Table 3). Measured parameters
D4 3 (7) included overall complication rate, the development of duo-
Forty-six wounds were discovered in 29 patients. The majority of denal and pancreatic fistulas, length of hospital stay, and
these (65%) were contained in the second portion of the duodenum. mortality. Five of 15 (33%) patients who were repaired with-
830 April 2007
3. Pyloric Exclusion for Penetrating Duodenal Injuries
Table 2 Demographics and Clinical Characteristics
No Exclusion Exclusion
(n 15) (n 14)
Age (yr) 29.9 7.7* 28.5 12.9 0.280†
Sex (male) 15 (100%) 14 (100%) 1.0‡
Gun shot wounds 10 (67%) 13 (87%) 0.169§
Duodenal Injury Grade 2.5 0.6 3.0 0.7 0.064†
I 1 (7%) 0%
II 6 (40%) 3 (21%)
III 8 (53%) 8 (57%)
IV 0% 3 (21%)
Pancreatic Injury Grade 2.7 2.1 (n 3) 2.0 0.9 (n 10) 0.799†
Injury Severity Score 18.4 8.3 23.2 17.5 0.631†
Shock 5 (33%) 3 (21%) 0.682§
Vascular injury 8 (53%) 6 (43%) 0.847‡
Associated injury 15 (100%) 14 (100%) 1.0§
Patients were statistically similar with respect to age, sex, mechanism, injury grade, injury severity, the presence of shock, vascular injury,
and other associated injuries.
* Mean SD.
Mann-Whitney ranked sum test.
Fisher exact test.
out exclusion had postoperative complications during their with pancreatic injuries (40%, p 0.497) developed pancreatic
hospital course. Of these five patients with postoperative com- fistulas. Length of hospital stay was greater in the PE group (no
plications, four (80%) had evidence of hemorrhagic shock at PE, 13.5 7.7 days vs. PE, 24.3 19.7 days, p 0.087).
admission. Complications included pneumonia (3), adult respi- In-hospital mortality rates were similar in both populations [no
ratory distress syndrome (ARDS; 1), acute renal failure (1), PE, 1/15 (7%) vs. PE, 3/14 (21%), p 0.273].
intra-abdominal abscess (1), and urosepsis (1). None of these 15
(0%) patients suffered a duodenal or pancreatic fistula. Ten of 14
(71%) patients with pyloric exclusion had postoperative com- DISCUSSION
plications consisting of sepsis (5), ARDS (4), pancreatic fistulas Significant controversy exists regarding the best opera-
(4), ventilator dependent respiratory failure (3), acute renal fail- tive treatment for duodenal injuries. Throughout the years,
ure (3), small bowel obstruction (2), wound infections (2), pneu- surgeons have developed several innovative procedures to
monia (2), and intra-abdominal abscesses (1). Only three of both repair the wounded duodenum and prevent fistulization
these 10 (30%) patients with postoperative complications had from repair breakdown. The first method of suture line pro-
evidence of hemorrhagic shock at arrival. Although none of tection was the “triple tube ostomy” described by Stone.6,7 In
these 14 (0%) patients suffered a duodenal fistula, four patients this straightforward surgical technique, a gastrostomy tube
and two separate jejunal tubes are placed. The proximal
jejunal tube is threaded in a retrograde fashion into the duo-
Table 3 Postoperative Complications and Outcome denum to decompress the suture line, whereas the distal tube
No Exclusion Exclusion is placed as standard jejunal feeding access. Despite its tech-
(n 15) (n 14) nical simplicity and encouraging initial results, reports from
Complications (5/15) 33% (10/14) 71% 0.093‡ others have failed to show improved outcomes with the
Duodenal fistula (0/15) 0% (0/14) 0% 1.0§ technique.8,9 Procedures for complete diversion of the gas-
Pancreatic fistula (0/3) 0% (4/10) 40% 0.497§ trointestinal stream were soon developed. Berne and Dono-
Length of stay (d) 13.5 7.7* 24.3 19.7 0.087†
van excluded repairs by “diverticulizing” the duodenum.10,11
Mortality (1/15) 7% (3/14) 21% 0.273§
This procedure originally consisted of a duodenal repair, vagot-
A trend toward a higher overall complication rate (71%) and omy, antrectomy, gastrojejunostomy, tube duodenostomy, and T
hospital length of stay (24.3 d) was evident in the pyloric exclusion
group. Although no patient repaired without pyloric exclusion devel- tube biliary drainage. Although effective in diverting enzymatic
oped a pancreatic fistula, 40% (4 of 10) of patients with pyloric secretions, the procedure is complex, time consuming, and re-
exclusion later formed pancreatic fistulas. No patient in either group sects normal tissue in young, often healthy patients. Today
leaked from their duodenal repair. Mortality rates were statistically diverticulization is seldom performed and has largely been re-
similar between groups. placed by the simpler pyloric exclusion.
* Mean SD.
Mann-Whitney ranked sum test. First described by Vaughan in 1977, the pyloric exclu-
test. sion consists of a duodenal repair, over-sewing the pylorus
Fisher exact test. through a gastrotomy, and gastrojejunostomy.12,13 At present,
Volume 62 • Number 4 831
4. The Journal of TRAUMA Injury, Infection, and Critical Care
the pyloric exclusion is often performed by applying a non- sumption that the pyloric exclusion adjunct prevents
cutting stapler immediately distal to the pylorus, further sim- fistulas.12–15,21 With a trend toward a greater complication
plifying the procedure. Despite its technical simplicity and rate and length of hospital stay in the pyloric exclusion group,
swiftness, the procedure permanently alters the gastrointesti- we have shown that patients repaired without pyloric exclu-
nal tract with the construction of the gastrojejunostomy. Al- sion have similar clinical outcomes when compared with
though most reports indicate that the pylorus reopens within those repaired with pyloric exclusion. Although no compli-
3 weeks in most of patients, the pyloric exclusion remains an cation was directly attributable to the procedure itself, two
ulcerogenic operation. Postpyloric exclusion marginal ulcer- early postoperative partial small bowel obstructions were
ation incidence ranges from 0% to 33% in numerous reports, observed in the pyloric exclusion group. Both resolved with
with most studies demonstrating marginal ulcers in approxi- conservative management. Furthermore, four pancreatic fis-
mately 10% of patients who underwent surveillance tulas were observed after pyloric exclusion in those with
endoscopy.12–15 Postoperative anastomotic leaks and small combined injury. This 40% pancreatic fistula rate is similar to
bowel obstructions have also been described.3,14,15 previous reports describing fistula rates of 33% to 45% after
Concern for these potential morbidities has led several pancreatic injury.14 –17,21,22
authors to attempt to define which duodenal injuries may be Despite our findings, we acknowledge several limitations
repaired with simple suture techniques and which injuries of this study. This was an inpatient study, and thus no long-
may require more sophisticated procedures such as the pylo- term follow-up data were reviewed. As a result, many of the
ric exclusion.2– 4,8 –9,12–14,16 –19 Snyder classified duodenal in- reported complications of the pyloric exclusion procedure
juries as either mild or severe (missile injury, damage to such as marginal ulceration or bowel obstruction may have
75% of the wall circumference, involvement of the first or been overlooked. Furthermore, without a single duodenal
second portion of the duodenum, injury to repair interval 24 fistula, we were unable to examine the clinical outcomes of
h, and common bile duct injury) and advocated complex patients with duodenal fistulas when repaired without and
repairs for these advanced injuries.4 Adhering to these crite- with pyloric exclusion. We postulate that the protection of the
ria, 93% of our population had severe injuries, though pa- pyloric exclusion may benefit those patients with breakdown
tients repaired without pyloric exclusion fared equally or of the duodenal repair because of increased spontaneous
better than those with pyloric exclusion in each of our mea- closure rates, decreased fistula output, and improved nutri-
sured clinical parameters. tion. Lastly, the small sample size, despite a 10-year retro-
AAST grading may be employed to classify duodenal in- spective review of a busy, Level I trauma center, is a result of
jury severity, but injury grade alone may not be an important the infrequency of injury to the well protected duodenum.
factor when deciding to perform a pyloric exclusion.5,20 In our Although the study population is small, the population is
study population, a trend toward greater duodenal injury severity homogeneous, consisting of only penetrating, advanced duo-
was evident in the pyloric exclusion group although statistical denal or combined pancreaticoduodenal injuries. Only a
significance was not reached. Although numerous authors have large, randomized, prospective trial comparing patients with
described the use of the pyloric exclusion for more “severe” duodenal injuries repaired without and with pyloric exclusion
duodenal injuries, Timaran reported that duodenal injury grade will be able to demonstrate the superiority of one technique
is not predictive of either duodenal fistula or mortality. After over the other. Given the rarity of the penetrating duodenal
analyzing several risk factors including shock (SBP 90 mm injury and the innumerable clinical variables, a randomized
Hg), mechanism of injury, Abdominal Trauma Index, and Du- prospective study seems unlikely.
odenal Injury Score, preoperative or intraoperative hypotension In conclusion, the performance of pyloric exclusion for
proved to be the most important predictor of overall complica- penetrating advanced duodenal injury and combined pancre-
tions, duodenal fistulae, and mortality.20 In our series, shock was atic and duodenal injuries was not associated with either
more common in the group without pyloric exclusion, although duodenal fistula formation or improved clinical outcomes in
statistical significance was not reached. Despite the preponder- our study patient population. The observed trend toward
ance of shock in the primary repair-alone group, clinical out- increased rate of pancreatic fistula, overall complications, and
come was equivalent to those who underwent pyloric exclusion. length of hospitalization suggests that the addition of pyloric
In a meta-analysis, Asensio reviewed 15 clinical series exclusion to the repair of duodenal injuries provides no added
containing 1,408 patients with duodenal injuries who under- benefit. On the basis of our findings and those of others, we
went various surgical repairs and found an overall duodenal recommend primary repair without pyloric exclusion for most
fistula rate of 6.6%.1 In our study population, no patient of the penetrating duodenal injuries in hemodynamically sta-
developed a duodenal fistula regardless of surgical treatment. ble patients.
To our knowledge, no previous study has compared patients
repaired without pyloric exclusion to those compared with the
adjunctive procedure. Prior reports have argued in favor of REFERENCES
pyloric exclusion based on low overall duodenal fistulization 1. Asensio JA, Feliciano DV, Britt LD, et al. Management of duodenal
rates in heterogeneous study groups and the unproven as- injuries. Curr Probl Surg. 1993;30:1023–1093.
832 April 2007
5. Pyloric Exclusion for Penetrating Duodenal Injuries
2. Ivatury RR, Nassoura ZE, Simon RJ, et al. Complex duodenal 13. Martin TD, Feliciano DV, Mattox KL, et al. Severe duodenal
injuries. Surg Clin North Am. 1996;76:797– 812. injuries: treatment with pyloric exclusion and gastrojejunostomy.
3. Carrillo EH, Richardson DJ, Miller FB. Evolution in the Arch Surg. 1983;118:631– 635.
management of duodenal injuries. J Trauma. 1996;40:1037–1046. 14. Feliciano DV, Martin TD, Cruse PA, et al. Management of
4. Snyder WH III, Weigelt JA, Watkins WL, et al. The surgical combined pancreaticoduodenal injuries. Ann Surg. 1987;205:673–
management of duodenal trauma. Arch Surg. 1980;115:422– 429. 680.
5. Moore EE, Cogbill TH, Malangoni MA, et al. Organ injury scaling. 15. Buck JR, Sorensen VJ, Fath JJ, et al. Severe pancreatico-duodenal
II. Pancreas, duodenum, small bowel, colon and rectum. J Trauma. injuries: the effectiveness of pyloric exclusion with vagotomy. Am
1990;30:1427–1429. Surg. 1992;58:557–561.
6. Stone HH, Garoni WJ. Experiences in the management of duodenal 16. Mansour MA, Moore JB, Moore EE, et al. Conservative management
wounds. South Med J. 1966;59:864 – 867. of combined pancreaticoduodenal injuries. Am J Surg. 1989;158:531–
7. Stone HH, Fabian TC. Management of duodenal wounds. J Trauma. 535.
1979;19:334 –339. 17. Flynn WJ, Cryer HG, Richardson JD. Reappraisal of pancreatic and
8. Ivatury RR, Nallathambi M, Gaudino J, et al. Penetrating duodenal duodenal injury management based on injury severity. Arch Surg.
injuries: analysis of 100 consecutive cases. Am J Surg. 1985;2:153– 1990;125:1539 –1541.
158. 18. Nassoura ZE, Ivatury RR, Simon RJ, et al. A prospective reappraisal
9. Cogbill TH, Moore EE, Feliciano DV, et al. Conservative of primary repair of penetrating duodenal injuries. Am Surg. 1994;
management of duodenal trauma: a multicenter perspective. 60:35–39.
J Trauma. 1990;30:1469 –1475. 19. Velmahos GC, Kamel E, Chan LS, et al. Complex repair for the
10. Berne CJ, Donovan AJ, White EJ, et al. Duodenal “diverticulization” management of duodenal injuries. Am Surg. 1999;65:972–975.
for duodenal and pancreatic injury. Am J Surg. 1974;127:503–507. 20. Timaran CH, Martinez O, Ospina JA. Prognostic factors and
11. Berne CJ, Donovan AJ, Hagen WE. Combined duodenal pancreatic management of civilian penetrating duodenal trauma. J Trauma.
trauma: the role of end-to-side gastrojejunostomy. Arch Surg. 1968; 1999;47:330 –335.
96:712–722. 21. Graham JM, Mattox KL, Vaughan GD III, et al. Combined
12. Vaughan GD III, Frazier OH, Graham DY, et al. The use of pyloric pancreaticoduodenal injuries. J trauma. 1979;19:340 –346.
exclusion in the management of severe duodenal injuries. Am J Surg. 22. Wynn M, Hill DM, Miller DR, et al. Management of pancreatic and
1977;134:785–790. duodenal trauma. Am J Surg. 1985;150:327–332.
Volume 62 • Number 4 833