Has Evolution in Awareness of Guidelines for Institution of Damage Control Improved Outcome in the Management of the Posttraumatic Open Abdomen? FREE

Recognition that the triad of hypothermia, acidosis, and coagulopathy invariably leads to mortality if uninterrupted led Stone et al¹ to describe the "bail-out" approach, including initial abandonment of laparotomy, intra-abdominal packing, correction of coagulopathy, and deferred surgical repairs. Further refinement of this approach by Rotondo et al² resulted in damage control, a multiphase approach designed to correct the patient's physiological derangements before returning the patient to the operating room for definitive surgery. These 2 studies^1- 2 ushered in the era of staged surgical procedures for the management of severely injured patients.

Other investigators^3- 7 have since attempted to define models that include subjective and objective parameters as guidelines to define timing for performance of damage control. Temperature, pH, number of units of blood transfused, estimated blood losses, intraoperatively measured lactic acid levels, and coagulation studies have all been incorporated into these models,^3- 7 although statistical validation has proved difficult. Clearly, the triad of hypothermia, acidosis, and coagulopathy, the "bloody vicious cycle,"⁴ remains a formidable enemy to both patients and trauma surgeons.

Data from our institution⁷ based on 548 patients who sustained massive blood losses and required damage control described and validated a model that used easily followed intraoperative objective parameters to serve as suggested guidelines for institution of damage control. On the basis of these data, we advocate institution of damage control before reaching these limits.⁷ The purpose of this study was to review our institutional experience with these parameters as predictors of outcome and to evaluate if evolution in the awareness of guidelines for early institution of damage control improves outcomes mortality-wise and morbidity-wise in 2 different patient groups: those admitted prior to vs those who underwent damage control following awareness of these suggested guidelines.

During a 104-month period (November 1992 to July 2001) all patients admitted to the Los Angeles County and University of Southern California Medical Center, Los Angeles, an American College of Surgeons–verified level I trauma center, who underwent damage control that resulted in a posttraumatic open abdomen were reviewed and data entered onto a collection sheet. Institutional review board approval was obtained. These patients were divided in 2 groups: patients admitted between November 1992 and December 1998 (group 1), retrospectively studied, who underwent damage control based on less defined guidelines, such as the individual surgeon's perception of the need for this procedure, vs patients admitted between January 1999 and July 2001 (group 2), prospectively studied, who required damage control based on the awareness of these suggested guidelines.

Parameters included in the suggested guidelines used to indicate damage control for patients in group 2 included the following: operating room (OR) temperature of 34°C or less, pH of 7.2 or less, serum bicarbonate level of 15 mEq/L or less, transfusion volumes of 4000 mL or less of packed red blood cells (PRBCs), total blood replacement of 5000 mL or less (if a combination of both whole blood and PRBCs were used), and total OR fluid replacement inclusive of crystalloids, blood, and blood products of 12 000 mL or less. Similarly, patients sustaining injuries known to be predictors of poor outcome, such as thoracic vascular, abdominal vascular, and hepatic injuries; those requiring either emergency department thoracotomy or OR thoracotomy; or patients developing intraoperative complications, such as coagulopathy or dysrhythmias, also underwent damage control. The following are the suggested guidelines for instituting damage control.

  Sustained Hypotension Acidosis pH less than or equal to 7.2 Serum bicarbonate level less than or equal to 15 mEq/L Hypothermia Temperature less than or equal to 34°C Coagulopathy, clinically observed Transfusion volumes PRBCs less than or equal to 4000 mL Total blood replacement less than or equal to 5000 mL (if combination of whole blood and PRBCs is used) Total OR fluid replacement less than or equal to 12 000 mL (crystalloids, blood, and blood products) Injuries associated with poor outcomes Thoracic vascular injuries Abdominal vascular injuries Complex hepatic injuries requiring packing (American Association for the Surgery of Trauma–Organ Injury Scale grades IV-V) Patients requiring emergency department thoracotomy Patients requiring OR thoracotomy

Damage control included some or all of the following procedures: hepatic packing, temporary hollow viscus closures or rapid stapled resections, drainage of selected organ injuries, rapid splenectomy or nephrectomy, abdominal packing, and temporary abdominal wall closure with a prosthetic device, usually a 3-L intravenous bag or vacuum pack.

Data collected included demographics, age, mechanism of injury, admission vital signs, emergency department procedures, including thoracotomy, Revised Trauma Score (RTS), and Injury Severity Score (ISS). Other data included surgical findings and indications for damage control, including number and types of organs injured, intraoperative estimated blood loss (EBL), volume of intraoperative fluids, blood, and blood products administered, and the presence of intraoperative complications, such as sustained hypotension, coagulopathy, hypothermia, and acidosis.

In addition, operative times from the beginning to the conclusion of the damage control procedure were tracked and collected. Postoperative complications, length of time that patients remained with an open abdomen, and surgical intensive care unit (SICU) and hospital length of stay were also recorded.

The main outcome measure of the study was survival. Univariate analysis was used to compare the demographic and clinical characteristics between patients in group 1 and group 2 and between survivors and nonsurvivors. The Fisher 2-sided exact test or the χ² test with Yates correction was used for categorical variables, and the 2-sample t test or the Wilcoxon 2-sample test was used for continuous variables. The factors that were significant at P<.20 by these tests were entered into a stepwise logistic regression analysis to identify independent predictors of outcome. All analyses were performed with SAS statistical software, version 8.2 (SAS Institute Inc, Cary, NC). Data are presented as mean ± SD.

During this 104-month study (November 1992 to July 2001), there were 139 patients who required immediate surgical intervention and underwent staged surgical procedures and damage control. They included 127 men (91%) and 12 women (9%): 100 (72%) admitted with penetrating injuries, 86 (86%) admitted with sustained gunshot wounds, and 14 (14%) admitted with sustained stab wounds. Thirty-nine patients (28%) were admitted secondary to blunt trauma, of which 20 (52%) sustained injuries due to motor vehicle crashes, 8 (20%) were pedestrians struck by vehicles, 6 (15%) sustained injuries due to unintentional falls, and 5 (13%) were admitted secondary to other mechanisms of injury.

Admission blood pressure was 113 ± 36 mm Hg. Admission heart rate was 107.2 ± 28.7/min. The RTS was 6.81 ± 1.63 (range, 0-7.84). The ISS was 24.0 ± 12.9 (range, 16-75), indicating a severely injured patient population (Table 1). No patients required emergency department thoracotomy. All patients underwent immediate surgical intervention. Indications included acute hemoperitoneum and hemodynamic instability. Twenty-one patients (15%) required OR thoracotomy and/or median sternotomy; 14 (67%) survived.

Indications for institution of damage control resulting in a posttraumatic open abdomen included sustained hypotension, acidosis, hypothermia, clinical coagulopathy, the need to pack hepatic injuries and/or the abdominal cavity, and the surgeon's decision that a staged surgical procedure was necessary to save the patient's life. Patients in group 2 were also selected for damage control based on awareness of the previously described guidelines with its objective parameters. In addition, 58 patients (42%) developed massive bowel edema as denoted by increases in the width of the bowel between 2 and 3 times its normal size, and 31 (22%) manifested signs of an acute abdominal compartment syndrome with increased peak airway pressures and decreased urinary output when attempts at abdominal wall closure were made.

The mean EBL was 4764 ± 5349 mL (range, 250-25 000 mL). The mean intraoperative fluid replacement with crystalloids, blood, and blood products was 22 034 mL (Table 2). One hundred one patients (73%) experienced a total of 228 complications, for a mean of 2.26 complications per patient. Sixty-eight (49%) experienced sustained hypotension, 56 (40%) experienced coagulopathy, 53 (38%) developed acidosis, and 51 (37%) were hypothermic. There were a total of 403 associated injuries: 308 abdominal (76%), 46 abdominal vascular (11%), 21 musculoskeletal (5%), 18 thoracic (5%), 7 central nervous system (2%), and 3 large associated burns (1%), for a mean of 2.9 associated injuries per patient.

Of the 139 patients, 34 (25%) died as a result of their injuries. Factors associated with outcome included age (P = .01), RTS (P = .003), Glasgow Coma Scale score (P<.001), and respiratory rate (P = .02). The length of stay in SICU for all patients (n = 139) was 19.8 ± 18.6 days, and their hospital stay was 31.3 ± 28.2 days, significantly different between survivors and nonsurvivors (Table 3). A large number of patients experienced significant postoperative complications: 57 patients (41%) developed extra-abdominal infections, 50 (36%) experienced either single and/or multiple systems organ failure, 24 (17%) had intra-abdominal abscesses, and 18 (13%) developed gastrointestinal tract fistulas. Using the significant risk factors identified in the univariate analysis, stepwise logistic regression identified 3 independent significant factors associated with survival: age of 55 years or younger (P<.001), absence of organ failure (P = .04), and absence of infection (P = .03) (Table 4).

Patients were compared between the 2 study periods. Group 1 (n = 86) consisted of patients admitted between November 1992 and December 1998, studied retrospectively, and group 2 (n = 53) consisted of patients admitted between January 1999 and July 2001, studied prospectively. Analysis revealed no differences between their ages, admission vital signs, physiological state on presentation (RTS), or anatomic degree of injury (ISS). Similarly, there were no differences in mortality between the 2 groups (21 [24%] of 86 patients in group 1 died compared with 13 [24%] of 53 patients in group 2; P = .85).

Moreover, patients in group 1 spent a longer time in the OR (mean, 4.09 ± 1.99 hours; range, 0.35-9.53 hours) vs those in group 2 (mean, 2.34 ± 1.50 hours; range, 0.3-6.2 hours; P<.001). In addition, there were statistically significant differences between the 2 groups in the number of units of PRBCs (P = .002) and units of fresh frozen plasma (P = .042) transfused intraoperatively and their EBL (P = .002) (Table 5).

A significant number of patients experienced intraoperative and postoperative complications in both groups. However, patients in group 2 sustained less hypothermia (P = .07) and bowel edema (P = .01) than patients in group 1. Postoperatively, patients in group 2 developed fewer extra-abdominal infections (P = .002) and fewer intra-abdominal abscesses identified either during reinterventions or by computed tomographic scan (P = .02) (Table 6). Of the 65 surviving patients in group 1, only 14 (22%) were able to have their abdomens closed (6.2 ± 12.7 days). Of the 40 surviving patients in group 2, 37 patients (93%) underwent abdominal wall closure (6.2 ± 3.5 days) (P<.001).

Profound shock or massive injuries responsible for large blood losses quickly initiate the cycle of hypothermia, acidosis, and coagulopathy,^7- 13 described by Moore⁴ as the "bloody vicious cycle." A fourth component of this cycle is dysrhythmias, which usually herald the patient's death.^7,14- 16 Recognizing that the "bloody vicious cycle" must be interrupted, Stone et al¹ described the "bail-out" approach, ushering in the era of staged surgical procedures for the management of severely injured patients. This approach was later refined by Rotondo et al,² with the goals of returning the patient to the OR after all physiological derangements have been corrected.

The hallmark work by Stone et al¹ did not, however, describe any predictive parameters intraoperatively to select patients to undergo interruption of surgical procedures other than observation of clinical coagulopathy. Phillips et al¹⁷ identified patients who sustained massive transfusion exceeding 2 times their estimated blood volume as patients at risk for the development of organ failure, focusing on the volume of blood transfusion of 25 U and derangements of the coagulation system.

In a series of 200 patients treated with unorthodox techniques to interrupt laparotomy and the triad of hypothermia, acidosis, and coagulopathy, Burch et al³ proposed a model to predict 48-hour survival based on clinical and laboratory parameters, including core temperature of 32°C or less, pH of 7.09 or less, and a mean volume of PRBC transfusions of 22 U. Burch et al³ postulated that this model could predict 48-hour survival in critically injured patients based on a linear regression model that identified red cell transfusion rates of approximately 12.5 U/h and pH. Furthermore, they advanced the concept of abbreviated laparotomy as a rational approach to an apparently hopeless situation. Sharp and LoCicero¹⁸ focused their efforts on a series of 39 patients of which 31 sustained massive hepatic injury and proposed a model consisting of objective parameters such as a temperature of 33°C or less, a pH of 7.18 or less, prothrombin time of 16 seconds or less, partial prothrombin time of 50 seconds or less, and a mean transfusion volume of 10 units or more of blood to indicate early packing.

Rotondo et al² described a multiphase approach to the management of exsanguinating patients who sustained abdominal injury based on 46 patients but could not identify any statistical differences between the 22 patients who underwent definitive laparotomy vs the 24 patients who underwent damage control laparotomy. The authors then identified a maximum injury subset consisting of 22 patients, of which 9 were subjected to definitive laparotomy vs 13 who underwent damage control laparotomy. In this group of patients, the survival rate in the damage control group was 77% vs 11% in the definitive laparotomy group.² Thus, they concluded that damage control was a promising approach to increase survival in exsanguinating patients. The authors did not identify any objective parameters during the intraoperative phase of damage control other than "identifying coagulopathy in the judgment of the senior surgeon."² Recently, Johnson et al¹⁹ confirmed the original findings and recommendations of Rotondo and coauthors based on a series of 24 patients from the same institution.

Morris et al⁵ described a series of 107 patients who underwent staged celiotomy with abdominal packing and focused on the indications and timing of reconstruction, criteria for emergency return to the OR, complications after reconstruction, and the abdominal compartment syndrome, proposing to proceed with damage control celiotomy early in the operation based on patient temperature of less than 35°C, acid base status consisting of a base deficit less than 14, and the presence of medical bleeding.

Moore⁴ described factors predictive of a severe coagulopathic state, describing progressive coagulopathy as the most compelling reason for staged laparotomy. In this model, he suggested factors predictive of a severe coagulopathic state, including intraoperatively measured prothrombin time and partial prothrombin time greater than 2 times normal, massive and rapid blood transfusion exceeding 10 U in 4 hours, persistent cellular shock defined as an oxygen consumption index of less than 110 mL/min per M², lactic acid level of greater than 5 mmol/L, pH of less than 7.2, base deficit greater than 14, and core hypothermia with a temperature less than 34°C. Subsequently, Cosgriff et al⁶ postulated that the ability to predict the onset of coagulopathy, as one of the most important components of the "bloody vicious cycle," would have significant decision making implications with regard to the institution of damage control. This predictive model for life-threatening coagulopathy included a systolic blood pressure of less than 70 mm Hg, temperature of less than 34°C, pH of less than 7.10, and ISS of 25 or higher.

Clearly, no single model can accurately predict the timing for institution of damage control. A pH of less than 7.09 or 7.10 or a core temperature of less than 33°C may indicate that the "bloody vicious cycle"⁴ is too far advanced to be interrupted. Similarly, intraoperatively measured prothrombin time, partial prothrombin time, fibrinogen levels, and lactic acid levels are difficult to obtain; the results take too long to be returned, and these laboratory studies are often unavailable in the ORs of America's busiest trauma centers to serve as indications for the institution of damage control, whereas the ISS proposed in some models⁴ is clearly not a usable parameter intraoperatively.

Recently, Asensio and colleagues⁷ proposed a model consisting of easily followed, objective intraoperative measures as predictors of outcome based on 548 patients admitted to their institution who met the following criteria: EBL of 2000 mL or more during a trauma operation, transfusion of 1500 mL or more of PRBCs during their initial resuscitation, or diagnosis of exsanguination. In these series, patients had a mean OR pH of 7.15 and mean OR temperature of 34.3°C and in addition received an average of 14 165 mL of crystalloids, blood, and blood products. Parameters in their model that indicated damage control included the following: OR temperature of 34°C or less, pH of 7.2 or less, serum bicarbonate level of 15 mEq/L or less, transfusion volume of 4000 mL or less of PRBCs, total blood replacement of 5000 mL or less if both PRBCs and whole blood are used, and total OR fluid replacement, including crystalloids, blood, and blood products, of 12 000 mL or less.

All of these predictors of outcome were statistically validated and considered as the absolute upper limits that would be acceptable before institution of damage control. The authors strived to institute damage control much earlier before reaching these limits. On the basis of their logistic regression analysis, OR replacement of 4000 mL or more of PRBCs, absence of need for emergency department or OR thoracotomy, and absence of abdominal vascular injury were identified as independent factors of outcome for survival.

The most important goal of early institution of damage control is survival of the patient.^1- 7,16 These patients are then returned to the OR at a later date when physiological derangements, such as acidosis, hypothermia, and coagulopathy, have been corrected.^1- 7,19 Frequently, survival of these patients leads to the posttraumatic open abdomen as a logical extension of their damage control procedure. The management of the posttraumatic open abdomen is challenging, because these patients continue to lose significant amounts of fluid and heat through their open abdomens, which are often covered by laparotomy packs and a plastic intravenous bag. This exposes them to the development of gastrointestinal tract fistulas, which further complicates their fluid management.

On the basis of our previously reported data,⁷ we believe that we have validated a model with relatively simple objective parameters that can be followed intraoperatively to serve as guidelines for when to institute damage control. However, we wanted to further explore the validity of these suggested guidelines to evaluate whether they would result in improvement for any of the known factors predictive of outcome in the "bloody vicious cycle"⁴ (eg, acidosis, hypothermia, coagulopathy, fluid replacement, bowel edema, a decrease in time spent in the OR for damage control, and ability to definitively close a posttraumatic open abdomen after damage control).

For these purposes, we chose to compare a group of patients admitted before the awareness of suggested and validated guidelines for damage control vs another group of patients who underwent damage control based on these guidelines, recognizing the inherent flaws with the use of retrospective historical controls. These 2 groups were comparable with regard to their physiologic state on presentation and anatomic degree of injury; however, statistically significant differences were noted in patients subjected to damage control based on the use of objectively measured parameters included in the suggested guidelines with regard to their EBL, number of units of PRBCs and fresh frozen plasma transfused, and their length of stay in the SICU and hospital. This, we believe, further validates the predictive factors incorporated in our guidelines.

Patients in the second group incurred less hypothermia and developed fewer infections, intra-abdominal abscesses, and fistulas. These patients were also noted to have less bowel edema, although this is a subjective parameter that is difficult to quantitate. We believe that these results are related to less time spent in the OR before the institution of damage control. Of greater significance is the higher percentage of patients in group 2 (93%) who were able to undergo definitive abdominal wall closure during their hospital stay compared with those patients in group 2 (22%). Although no significant independent predictors of mortality were identified by linear regression analysis between the groups, this may simply indicate that we could not alter outcome based on the high injury severity of this patient population or that the sample of patients studied may not have been sufficiently large to detect differences. Nevertheless, a mortality of approximately 25% for both groups is significant and corroborates these patients' injury severity.

On the basis of these data, we strongly recommend close monitoring of intraoperative predictors of outcome as validated within our guidelines and recommend following our model for institution of damage control as early as possible and definitely before reaching the upper limits of these parameters, which include pH of 7.2 or less, temperature of 34°C or less, serum bicarbonate level of 15 mEq/L or less, transfusion volumes of 4000 mL or less of PRBCs, total blood replacement of 5000 mL or less if both PRBCs and whole blood are used, and total OR fluid replacement, including crystalloids, blood, and blood products, of 12 000 mL or less.⁷ We cannot overemphasize the need for the earliest possible interruption of the initial surgical procedure, especially in patients who have EBLs that approximate 5000 mL and those sustaining injuries that are well known to cause exsanguination.^7- 16

Institution of damage control implies immediate control of life-threatening hemorrhage, placement of chest tubes, thoracic packing if needed, closure of the skin if the chest has been opened, hepatic packing, temporary duodenal and hollow viscus closures or rapid stapled resections, drainage of pancreatic injuries, rapid resection of pancreatic injuries if present to the left of the superior mesenteric artery with staplers, rapid splenectomy and nephrectomy or occlusion of their vascular pedicles with a vascular clamp left in situ, uses of intraluminal shunts, and judicious abdominal packing with temporary abdominal wall closures.⁷

We recognize that there is significant research yet to be done to reach a greater understanding of the cellular and subcellular mechanisms triggered by profound shock, exsanguination, acidosis, hypothermia, and coagulopathy. Further research is also needed to investigate and develop newer resuscitation fluids and to identify gene patterns activated in profound shock. With awareness of these guidelines,⁷ we have been able to objectively and statistically validate our model, detect improvements in some predictors of outcome, decrease time spent in the OR for damage control, and improve the time for closure of the posttraumatic open abdomen. However, we were not able to decrease mortality for these patients. Therefore, the ongoing challenge is to continue to identify better predictors of outcome, improved means of resuscitation, greater understanding of the physiological derangements incurred by these patients, and most importantly better timing to institute damage control. Only then can we begin to reduce the high mortality experienced by these patients.

Corresponding author: Juan A. Asensio, MD, Department of Surgery, Division of Trauma and Critical Care, LAC and USC Medical Center, 1200 N State St, Room 10-750, Los Angeles, CA 90033-4525 (e-mail: asensio@usc.edu).

Accepted for publication September 23, 2003.