Brown CVR, Velmahos GC, Neville AL, Rhee P, Salim A, Sangthong B, Demetriades D. Hemodynamically “Stable” Patients With Peritonitis After Penetrating Abdominal TraumaIdentifying Those Who Are Bleeding. Arch Surg. 2005;140(8):767-772. doi:10.1001/archsurg.140.8.767
Despite initial presentation, hemodynamically stable patients with penetrating abdominal trauma may have significant ongoing hemorrhage and major intra-abdominal injuries requiring emergent surgical intervention.
Cohort analytic study.
Academic, level I trauma center.
One hundred thirty-nine consecutive hemodynamically stable patients with penetrating abdominal trauma in whom peritonitis was the sole indication for laparotomy.
Main Outcome Measures
The primary outcome was the amount of blood initially found at laparotomy. Secondary outcomes included additional intraoperative blood loss, intraoperative hypotension, transfusion, fluid, and vasopressor requirement; need for admission to the intensive care unit and mechanical ventilation; complications; survivor length of stay in the hospital and intensive care unit; and mortality.
The admission systolic blood pressure (mean ± SD, 131 ± 22 mm Hg) and heart rate (mean ± SD, 91 ± 22 beats/min) were normal. Median time from peritonitis to incision was 40 minutes. Ninety-seven percent of patients had intra-abdominal injury, including 81%, hollow visceral; 36%, solid organ; and 11%, vascular injury. Though most patients had less than 750 mL3 of blood found initially at laparotomy, there were 11% with 750 to 1500 mL3 and 7% with 1500 mL3 or more. Intraoperative hypotension (25%) and blood transfusion (39%) were common. Postoperatively, 40% of patients required intensive care (78% of them requiring mechanical ventilation) and 19% required additional transfusion within 24 hours. Complications occurred in 25% of patients, with intra-abdominal abscess (12%) and wound infection (7%) being the most common. Three patients died, 2 of exsanguination and 1 of multisystem organ failure.
Following penetrating abdominal trauma, peritonitis should be a trigger for emergent operation regardless of vital signs, because hemodynamic “stability” does not reliably exclude significant hemorrhage. Vascular injury, subsequent hypotension, blood transfusion, and complicated postoperative course are common in this population.
Selective nonoperative management of most solid organ injuries1- 3 and many stab wounds (SWs) to the abdomen4 is accepted and practiced at most trauma centers in this country. In addition, our institution has adopted selective nonoperative management of gunshot wounds (GSWs) to the abdomen.5- 7 Two absolute contraindications to selective nonoperative management of penetrating abdominal trauma are hemodynamic instability and diffuse abdominal tenderness indicating peritonitis. Hemodynamically unstable patients with penetrating abdominal trauma are presumed to have active hemorrhage and are emergently taken to the operating room for laparotomy. Though also taken for urgent laparotomy, hemodynamically stable patients with penetrating abdominal trauma with peritonitis are usually assumed to have a hollow visceral perforation and may or may not have significant intra-abdominal hemorrhage. Our hypothesis was that despite initial presentation, hemodynamically stable patients with peritonitis following penetrating abdominal trauma may have significant ongoing hemorrhage and catastrophic intra-abdominal injuries requiring emergent surgical intervention.
A prospective, observational study was carried out at the Los Angeles County/University of Southern California Medical Center in Los Angeles from January 2003 through December 2004. All hemodynamically stable (defined as prehospital and emergency department systolic blood pressure ≥90 mm Hg) patients with penetrating abdominal trauma were eligible for enrollment. By protocol, paramedics in Los Angeles county “scoop and run” penetrating trauma victims, obtain intravenous access in the ambulance, and do not give fluid to hemodynamically stable patients. Patients who underwent exploratory laparotomy, with peritonitis as the only indication for operation, were included. Demographic variables collected on each patient included age, sex, mechanism of injury (SW or GSW), admission systolic blood pressure and heart rate, hemoglobin level drawn at admission, results of admission focused assessment with sonography for trauma (FAST) examination, Injury Severity Score, intra-abdominal and associated extra-abdominal injuries, and time from admission to incision.
The primary outcome was the amount of blood initially found at laparotomy. Both the primary operating surgeon and the first assistant were independently queried, blinded to the other’s response, as to the amount of blood found immediately on entering the abdomen, and these 2 values were averaged to determine the final amount. The amount of free intraperitoneal blood initially found at laparotomy was used to stratify patients by class of hemorrhage (class I, <750 mL3; class II, 750-1500 mL3; classes III and IV, >1500 mL3) as per the Advanced Trauma Life Support for Doctors Instructor Course Manual.8 Secondary outcomes included additional intraoperative blood loss (obtained from anesthetic record), intraoperative hypotension (defined as systolic blood pressure<90 mm Hg continuously for more than 5 minutes), packed red blood cells transfused during operation and the first 24 hours of admission, intraoperative fluids received, vasopressor requirement, need for admission to the intensive care unit (ICU) and mechanical ventilation, complications, survivor length of stay in the hospital and ICU, and mortality. Subgroup analysis was performed based on mechanism of injury.
Statistical analysis was performed using SPSS version 12 (SPSS Inc, Chicago, Ill) and Microsoft Excel 2002 (Microsoft Corporation, Redmond, Wash). Values are reported as mean ± SD, median (range), or as raw percentages where applicable. Categorical variables were compared using χ2 or Fisher exact tests. Continuous variables were compared using independent samples t test, 1-way analysis of variance, or Wilcoxon test as appropriate. Statistical significance was considered at the level of P<.05 for all comparisons. The local institutional review board approved this study.
There were 139 patients who met inclusion criteria during the 2-year study period. The population was 94% male (n = 130), aged mean ± SD 29 ± 10 years, with 100 (72%) and 39 (28%) patients sustaining GSW and SW, respectively. The mean ± SD prehospital time, including scene time and transport time, was 25 ± 10 minutes. Patients were hemodynamically stable (mean ± SD systolic blood pressure, 131 ± 22 mm Hg and heart rate, 91 ± 22 beats/min), had a normal initial hemoglobin level (13 ± 2 g/dL), and had positive FAST results 31% (n = 43) of the time, prior to laparotomy. Median time from admission until initial skin incision was 40 minutes. Overall, 97% (n = 135) of patients had an intra-abdominal injury and 45% (n = 62) had associated extra-abdominal injury, with a mean ± SD Injury Severity Score of 14 ± 8 (Table 1).
Intraoperatively, 25% (n = 35) of patients developed hypotension, 39% (n = 54) received blood transfusion (median, 4 units [range, 1-26 units]), and 6% (n = 8) required vasopressors. Postoperatively, 40% (n = 55) of patients were admitted to the ICU and stayed in the ICU for 3 days and the hospital for 7 days after admission; 78% (n = 43) of these patients required mechanical ventilation. Complications were common, occurring in 25% (n = 35) of patients, with intra-abdominal abscess and wound infection being most frequent (Table 2). Three patients (2%) died as a result of their injuries, 2 of intraoperative exsanguination and 1 of multisystem organ failure 3 days after injury.
Of the 139 patients, 82% (n = 114, group 1) had less than 750 mL3, 11% (n = 15, group 2) had 750 to 1500 mL3, and 7% (n = 10, groups 3 and 4) had more than 1500 mL3 of free intraperitoneal blood initially found during laparotomy. The demographic and admission characteristics of each group are presented in Table 3. Though there was no difference in the rate of hollow visceral injury (group 1, 82% [n = 93]; group 2, 67% [n = 10]; groups 3 and 4, 90% [n = 9]; P = .29), groups 2 and 3 and 4 more often had solid organ (group 1, 30% [n = 34]; group 2, 60% [n = 9]; groups 3 and 4, 70% [n = 7]; P<.01) and vascular (group 1, 8% [n = 9]; group 2, 20% [n = 3]; groups 3 and 4, 30% [n = 3]; P = .05) injury.
Increasing amounts of free blood initially found at laparotomy were associated with intraoperative hypotension, additional blood loss, resuscitation, blood transfusion, and vasopressor requirement, as well as worse postoperative outcomes (Table 4). Complications were similar among groups, with the exceptions of more enterocutaneous fistulae (group 1, 2% [n = 2]; group 2, 13% [n = 2]; groups 3 and 4, 20% [n = 2]; P<.01) and a trend toward more intra-abdominal abscesses (group 1, 9% [n = 10]; group 2, 20% [n = 3]; groups 3 and 4, 30% [n = 3]; P = .07). One patient (1%) in group 1 died of multisystem organ failure, while 1 patient each in groups 2 (7%) and 3 and 4 (10%) exsanguinated on the operating table.
The majority of patients sustained a GSW (72%) rather than a SW (28%), and though admission systolic blood pressure (mean ± SD, GSW, 133 ± 21 mm Hg vs SW, 126 ± 23 mm Hg; P = .07), heart rate (mean ± SD, GSW, 92 ± 23 beats/min vs SW, 91 ± 19 beats/min; P = .88), and hemoglobin level (mean ± SD, GSW, 14 ± 2 g/dL vs SW, 14 ± 2 g/dL; P = .96) were similar, patients with GSW were younger (mean ± SD, GSW, 26 ± 9 years vs SW, 36 ± 11 years; P<.01), more often had an associated extremity injury (GSW, 29% vs SW, 10%; P = .02), and had a higher Injury Severity Score (mean ± SD, GSW, 15 ± 8 vs SW, 11 ± 7; P = .01). The majority of patients with GSW and SW sustained intra-abdominal injuries (GSW, 98% vs SW, 95%; P = .31), and many had associated extra-abdominal injuries (GSW, 49% vs SW, 33%; P = .09). After GSW vs SW, there was no difference in solid organ injury (GSW, 36% vs SW, 36%; P = .99), but patients with GSW had more hollow visceral injuries (GSW, 85% vs SW, 69%; P = .04). In particular, patients with GSW sustained more injuries to the small bowel (GSW, 54% vs SW, 33%; P = .03) and colon (GSW, 50% vs SW, 26%; P<.01) but fewer gastric injuries (GSW, 12% vs SW, 26%; P = .05). Additionally, patients with GSW sustained more than a 2-fold higher incidence of intra-abdominal vascular injury (GSW, 13% [n = 13] vs SW, 5% [n = 2]; P = .23), with 87% (13/15) of vascular injuries occurring after GSW.
Intraoperatively, only patients with GSW were found to have a class III or IV hemorrhage at the time of laparotomy (GSW, 10% vs SW, 0%, P = .06) and had more additional intraoperative blood loss (median, GSW, 500 mL3 vs SW, 300 mL3; P<.01), received more intravenous crystalloid (mean ± SD, GSW, 5 ± 3 L vs SW, 3 ± 1 L; P<.01), and more often received blood transfusions (GSW, 46% vs SW, 21%; P<.01). In addition, patients with GSW had a more complicated postoperative course (Table 5). Moreover, all 3 deaths in the population were secondary to a GSW.
The current study prospectively investigates a hemodynamically stable population of patients with diffuse peritonitis secondary to penetrating abdominal trauma. The vast majority (97%) of patients had intra-abdominal injury identified at laparotomy, with an additional 45% sustaining injury to an extra-abdominal region. As would be expected, 81% of patients had perforation of hollow viscera, which could clearly account for diffuse peritonitis. However, the remaining 19% of patients had no visceral perforation, with only solid organ injury or free intraperitoneal blood as a source for peritoneal irritation. In fact, 36% of patients had solid organ injury, with liver (25%) being the most commonly injured. There were also 15 patients (11%) with intra-abdominal vascular injury, which included major vasculature structures such as the iliac vessels (5 patients), inferior vena cava (4 patients), and superior mesenteric vessels (3 patients).
Despite potentially exsanguinating and lethal intra-abdominal injuries, all patients in this population were hemodynamically stable on presentation, not providing early clinical clues as to the severity of injury or ongoing hemorrhage. Though most patients (82%) were found to have less than 750 mL3 of free blood initially at laparotomy, 18% had 750 mL3 or more and 7% had 1500 mL3 or more. By classic teaching, this amount of hemorrhage would be expected to cause hypotension in patients.8 Unfortunately, patients may be seen early after injury or have enough physiologic reserve to maintain a “normal” blood pressure despite massive bleeding. With increasing amounts of free blood found initially at laparotomy, patients in the current study were more likely to have intraoperative hypotension and additional hemorrhage and require blood transfusion, greater fluid resuscitation, and vasopressors. In addition, patients with significant bleeding more often required ICU admission, mechanical ventilation, and postoperative blood transfusion and had more complications and longer hospital stays. Though it is intuitive that patients who bleed more have worse outcomes (eg, sicker patients do worse), the difficulty rests in preoperatively distinguishing patients with significant intra-abdominal injury and hemorrhage.
All patients in this series were hemodynamically stable, making the initial vital signs unreliable to clearly identify bleeding patients. Though statistical differences existed between patients based on class of hemorrhage, patients with class III or IV hemorrhage still maintained a “normal” blood pressure (mean ± SD, 122 ± 17 mm Hg). Though these patients were tachycardic (mean ± SD, 111 ± 37 beats/min), elevated heart rate in the trauma bay is often attributed to pain, agitation, or exogenous substances. Additionally, patients with class II hemorrhage, who would be expected to be more tachycardic, actually had a lower heart rate than those with class I hemorrhage (mean ± SD, 83 ± 21 beats/min vs 91 ± 19 beats/min, respectively). Many authors have previously described the unreliable nature of the heart rate and blood pressure in trauma patients,9- 11 most describing a relative bradycardia associated with hypotension after trauma. Most recently, Victorino et al12 examined 14 325 trauma patients, of whom 489 were hypotensive. They found that heart rate was neither sensitive nor specific for identifying patients with hypotension, concluding that tachycardia was not a reliable sign of hypotension after trauma. Though our study did not specifically address this issue, patients with class III and IV hemorrhage, who would be expected to manifest hypotension, did not. Furthermore, patients with a class II bleed, who should be tachycardic, were not.
Other diagnostic tools that are often used to evaluate acutely injured patients include hemoglobin (or hematocrit) level and the FAST examination. A hematocrit level drawn immediately in the emergency department has been evaluated for both blunt and penetrating trauma. Snyder13 evaluated 524 trauma patients with an initial spun hematocrit level and found that a level of 35% or less more often required an operation. Similarly, Paradis et al14 studied the utility of an initial hematocrit level in a penetrating trauma population. They found that a lower hematocrit level correlated with either significant thoracic or abdominal injury. We found a similar association, as patients with significant hemorrhage found at laparotomy had a lower initial hemoglobin level (11 g/dL). However, this modest drop does not seem to accurately represent someone who has lost 30% to 40% of their blood volume. In addition, patients with class II hemorrhage had a relatively normal hemoglobin level (13 g/dL) on admission to the emergency department. Patients may be seen early after injury, sometimes with minimal resuscitation, and have not yet had time to equilibrate their hemoglobin concentration. Though the FAST examination is used extensively in patients with blunt trauma, its application in penetrating trauma is less well understood. Udobi et al15 found FAST to be quite specific (94%) in patients with penetrating abdominal trauma. However, they found a sensitivity of only 46%, with an average of 775 mL3 (range, 150-2000 mL3) of free intraperitoneal blood found during laparotomy in patients with negative FAST results. The FAST examination was unreliable in our population as well. Patients with 750 mL3 or more of free intraperitoneal blood found at laparotomy only had positive FAST results just over half the time, making negative FAST results useless in ruling out significant intra-abdominal hemorrhage after penetrating abdominal trauma.
Though both mechanisms of injury had an extremely high incidence of intra-abdominal injury (SW, 95% and GSW, 98%), patients who sustained a GSW had more blood lost before and during laparotomy and required more resuscitation with fluid and blood. In addition, 87% of the vascular injuries in this series were secondary to a GSW. The rate of intra-abdominal injury following a GSW depends on location, with a 61% and 26% incidence of intra-abdominal injury requiring therapy for anterior and posterior GSW, respectively.7 Based on the current series, the presence of peritonitis following a GSW is marker for a near 100% rate of intra-abdominal injury. In particular, these patients sustained hollow visceral and solid organ injuries in 85% and 36% of cases. In addition, despite hemodynamic stability, patients with GSW had a 13% incidence of intra-abdominal vascular injury and were found to have 1500 mL3 or more of free blood initially at laparotomy in 10% of cases.
Patients who present with peritonitis as a result of penetrating abdominal trauma, despite hemodynamic “stability,” should prompt emergent operation, because significant intra-abdominal bleeding cannot be reliably identified preoperatively. Vascular injury, subsequent hypotension, blood transfusion, and complicated postoperative course are common in this population. In particular, patients who have a GSW are more likely to sustain vascular injury and have significant hemorrhage found during laparotomy.
Correspondence: Carlos V. R. Brown, MD, Los Angeles County/University of Southern California Medical Center, 1200 N State St, 6341, Los Angeles, CA 90033 (firstname.lastname@example.org).
Accepted for Publication: April 19, 2005.
Previous Presentation: This paper was presented at the 76th Annual Meeting of the Pacific Coast Surgical Association; February 20, 2005; Dana Point, Calif; and is published after peer review and revision. The discussions that follow this article are based on the originally submitted manuscript and not the revised manuscript.
Michael J. Sise, MD, San Diego, Calif: Congratulations, Dr Brown, on a thoughtful study of a very relevant clinical problem. Once again, the group at LA County/USC [Los Angeles County/University of Southern California Medical Center] has taken a clinical suspicion, studied it thoughtfully, and produced evidence that we can use in managing our patients with penetrating abdominal trauma. All of us in trauma and critical care owe a debt of gratitude to your group for its continued excellence in clinical research. While running one of the busiest trauma centers in the country, you continue to ask the pertinent questions we all need answered. Strong work!
The results of this study demonstrate the high-risk nature of penetrating abdominal trauma. In an age of increasing pressure to avoid operation, this series should serve as a cautionary tale. Few would have predicted the high incidence of major hemorrhage in this group of hemodynamically stable patients with peritonitis as the only indication for operation. Many of us have been surprised to encounter significant blood loss in this same setting. You've confirmed a hunch that many trauma surgeons have had about this group of patients. These results should create a sense of urgency in getting to the operating room with these injured patients. We should also have an intense sense of unease about those patients we choose to treat nonoperatively following penetrating abdominal trauma. By falling into the trap of deciding not to operate and then resisting a change of plans, trauma surgeons place many patients at risk for untoward events. This study refocuses our attention on the high-risk nature of any penetrating abdominal injury.
There are a number of interesting observations in this study that deserve emphasis. FAST [focused assessment with sonography for trauma] proved particularly unreliable in this group of patients, with a sensitivity of only 46%. Given the amount of hemorrhage encountered, this is surprising. Heart rate also did not provide any clues to the degree of hemorrhage. However, initial hematocrit proved helpful in predicting bleeding. Gunshot wounds were highly morbid, producing all of the class III and IV hemorrhage. Stab wounds were less likely to produce hemorrhage and had a predictably low rate of vascular injury at 5%.
Please comment on the disappointing results with FAST in this group of patients. Any thoughts on why it was so insensitive? Other centers tout its reliability in detecting intra-abdominal hemorrhage. What wisdom does your group have to share on the use of FAST in penetrating abdominal trauma? Do you continue to perform FAST in this setting?
Although our group at our level I urban trauma center has encountered many of the same outcomes you report in this series, I would like to better understand this clinical problem. There were 139 patients in your 2-year series. What percentage was this group of all patients requiring laparotomy for penetrating abdominal trauma at your center? How many patients went directly to the operating room for hemodynamic instability? How many were treated nonoperatively initially and then became unstable and required operation? And, finally, how many developed delayed signs of peritoneal irritation? We need to have a sense of how often we can expect to see these patients with peritonitis and unsuspected hemorrhage. Although you may not be able to answer these questions today, it would be helpful to address them in your final manuscript.
Once again, congratulations on a thoughtful study that adds to our understanding of these injuries and better prepares us to successfully manage them.
Richard J. Mullins, MD, Portland, Ore: Commander Brown, your patients’ mean admission hemoglobins were 11. This lower value implies that your patients were infused with intravenous fluid prior to arrival. One randomized controlled trial published in a peer-reviewed journal supports the concept that aggressive resuscitation of patients prior to incision that controls hemorrhage increases the risk of death from bleeding. What is the policy of the prehospital care providers and yourselves in the emergency department in terms of resuscitating these patients? I am wondering, did hemorrhage occur to some patients after they arrived in your trauma center because they were overresuscitated prior to surgery to control hemorrhage?
Steven N. Parks, MD, Fresno, Calif: These are patients with peritonitis who are going to the operating room. This study was done to determine when they go to the operating room. The message that we have here is that many of them are bleeding a lot more than you suspect when they have peritonitis. Delaying their trip to the operating room to treat them like other patients with peritonitis (like appendicitis or other inflammatory conditions) is foolhardy because many of them are bleeding.
I'm not surprised that the FAST exam was not reliable, because if you remember, the highly reliable FAST exam is in the patient with shock where the exam is done to find the bleeding. Then it becomes a more reliable test.
The message today is that these patients need to not be delayed on their way to the operating room even with normal vital signs.
Jan K. Horn, MD, San Francisco, Calif: I enjoyed your discussion and paper very much. I think that your institution has in the past given us criteria and recommendations for the nonoperative management of gunshot wounds. The question that I have for you is, after operating on all of these patients with peritonitis, did you have the feeling that any of them really didn't need the surgery? In other words, did you find nonbleeding liver injuries or other nonhemorrhaging injuries of the mesentery where you could have actually avoided operation?
Dr Demetriades: First I want to start with Dr Steve Parks' question. I fully agree with you, and I will repeat the reasons for doing this study. First, we often noticed that in penetrating abdominal injuries, whenever the blood pressure was normal, the residents, the nurses, the anesthesiologists, everybody took it easy, and there was no sense of urgency. The second reason was the suggestion by some of our colleagues that patients with penetrating injuries and peritonitis who have a normal blood pressure can wait on the operating room board like the rest of the nontrauma peritonitis patients. I think our message is very clear. In penetrating trauma and peritonitis, once you decide that the patient needs an operation, the name of the game is time. You do not walk. You run.
Dr Sise, your comments are excellent. First, I want to say a few things about the FAST exam. In our experience it has been very disappointing. I can tell you that we were one of the very first trauma centers in the country to introduce FAST. We have 2 machines, one in the emergency department and a second one in the surgical admission ward. All patients coming through the resuscitation room of the emergency department undergo a FAST exam. So we have a lot of experience with that. We have regular in-service courses for everybody. We always compare the FAST results with the operative or CT [computed tomographic] scan findings as part of our performance improvement process. The results are really very poor. If the FAST is positive, especially in a hemodynamically unstable patient, it means a lot. You can rely on that. However, if the FAST is negative, it means little or nothing.
The heart rate and the blood pressure are notoriously unreliable. There are many reasons for that. (1) The prehospital time may be short. If you have a short prehospital time, the patient might have a major cardiovascular injury, and the initial blood pressure may be normal. (2) Often we deal with young, athletic patients. They might lose a fair amount of blood before they become hypotensive. (3) Very often our population is on illicit drugs—cocaine, amphetamines—and we know that these illicit drugs might maintain the blood pressure at falsely normal levels. (4) Elderly people who are often on β-blockers may not manifest with tachycardia despite significant blood loss, or the normal systolic blood pressure for a 75-year-old gentleman is 170. Now he comes to the emergency department with a blood pressure of 100, and the inexperienced surgeon thinks that is a normal blood pressure; in reality the patient is in severe hypotension. So there are many reasons for the unreliability of the initial vital signs.
Dr Mullins, you made a comment about the fluid resuscitation. In Los Angeles County, the policy is scoop and run. The paramedics do not even put an intravenous line in at the scene; they try to insert a line in the ambulance. But you mustn't be surprised about the low initial hemoglobin. We know that we do not need crystalloid administration in order to have a drop of the initial hemoglobin. We know that in the presence of bleeding there is mobilization of extravascular fluid into the vessels, which dilutes the blood and causes a low hemoglobin. We have seen this many times. Within 15 minutes after injury and without fluid resuscitation, the first hemoglobin may be low.
Dr Horn, the incidence of nontherapeutic operations in this series was only 3%.