Secondary Logo

Both Primary and Secondary Abdominal Compartment Syndrome can be Predicted Early and are Harbingers of Multiple Organ Failure

Balogh, Zsolt MD; McKinley, Bruce A. PhD; Holcomb, John B. MD; Miller, Charles C. PhD; Cocanour, Christine S. MD; Kozar, Rosemary A. MD, PhD; Valdivia, Alicia RN; Ware, Drue N. MD; Moore, Frederick A. MD

The Journal of Trauma: Injury, Infection, and Critical Care: May 2003 - Volume 54 - Issue 5 - p 848-861
doi: 10.1097/01.TA.0000070166.29649.F3
ORIGINAL ARTICLES
Free

Background  Primary (1°) abdominal compartment syndrome (ACS) is a known complication of damage control. Recently secondary (2°) ACS has been reported in patients without abdominal injury who require aggressive resuscitation. The purpose of this study was to compare the epidemiology of 1° and 2° ACS and develop early prediction models in a high-risk cohort who were treated in a similar fashion.

Methods  Major torso trauma patients underwent standardized resuscitation and had prospective data collected including occurrence of ACS, demographics, ISS, urinary bladder pressure, gastric tonometry (GAPCO2 = gastric regional CO2 minus end tidal CO2), laboratory, respiratory, and hemodynamic data. With 1° and 2° ACS as endpoints, variables were tested by uni- and multivariate logistic analysis (MLA).

Results  From 188 study patients during the 44-month period, 26 (14%) developed ACS—11 (6%) were 1° ACS and 15 (8%) 2° ACS. 1° and 2° ACS had similar demographics, shock, and injury severity. Significant univariate differences included: time to decompression from ICU admit (600 ± 112 vs. 360 ± 48 min), Emergency Department (ED) crystalloid (4 ± 1 vs. 7 ± 1 L), preICU crystalloid (8 ± 1 vs. 12 ± 1L), ED blood administration (2 ± 1 vs. 6 ± 1 U), GAPCO2 (24 ± 3 vs. 36 ± 3 mmHg), requiring pelvic embolization (9 vs. 47%), and emergency operation (82% vs. 40%). Early predictors identified by MLA of 1° ACS included hemoglobin concentration, GAPCO2, temperature, and base deficit; and for 2° ACS they included crystalloid, urinary output, and GAPCO2. The areas under the receiver-operator characteristic curves calculated upon ICU admission are 1° = 0.977 and 2° = 0.983. 1° and 2° ACS patients had similar poor outcomes compared with nonACS patients including ventilator days (1° = 13 ± 3 vs. 2° = 14 ± 3 vs. nonACS = 8 ± 2), multiple organ failure (55% vs. 53% vs. 12%), and mortality (64% vs. 53% vs. 17%).

Conclusion 1° and 2° ACS have similar demographics, injury severity, time to decompression from hospital admit, and bad outcome. 2° ACS is an earlier ICU event preceded by more crystalloid administration. With appropriate monitoring both could be accurately predicted upon ICU admission.

From the Department of Surgery (Z.B., B.A.M., J.B.H., C.S.C., R.A.K., D.N.W., F.A.M.), the Department of Cardiothoracic and Vascular Surgery (C.C.M.), and the Shock Trauma ICU (A.V.), Memorial Hermann Hospital, University of Texas at Houston Medical School, Houston, Texas.

Submitted for publication July 10, 2002.

Accepted for publication February 7, 2003.

Supported by NIMGS grants P50 38529–11 and U54 GM62119–01A1.

Presented at the 61st Annual Meeting of The American Association for the Surgery of Trauma, September 26–28, 2002, Lake Buena Vista, Florida.

Address for reprints: Frederick A. Moore, MD, FACS, Department of Surgery, UT-Houston Medical School, 6431 Fannin, Suite 4.264, Houston, TX 77030, email: frederick.a.moore@uth.tmc.edu.

Postinjury abdominal compartment syndrome (ACS) has consistently been reported to have a high mortality ranging from 25–75 percent. 1–7 In the present era of damage control surgery, the potential for ACS has increased, because of the increased survival of critically injured patients and the techniques applied, such as abdominal packs and crystalloid infusions. Over the past decade numerous reports have linked intra-abdominal hypertension (IAH) to the abnormal physiology that characterizes ACS. 7–11 However, studies based either on heterogeneous or extremely homogeneous populations have biased the determination of both the incidence and the outcome of the syndrome. Publications to date have failed to identify the independent risk factors for ACS and to build a prediction models for the syndrome. Furthermore, postinjury ACS has at least two clinical manifestations: primary ACS (1° ACS), where abdominal injury is present, and secondary ACS (2° ACS), where no intra-peritoneal injury is detected. 4,5,12 It is unknown whether these subgroups have similar epidemiology, predictors, and outcome. For better understanding and prevention of ACS it is essential to describe the syndrome and the epidemiology of its subgroups. Additionally, early prediction permits prevention or timely treatment (to date surgical decompression) before organ failure occurs at which point the outcome is unequivocally poor. 3 Therefore, the purpose of this study was to analyze high-risk trauma patients admitted to a single trauma center and resuscitated in a standardized fashion, to describe the epidemiology, and to identify the early predictors of postinjury 1° and 2° ACS.

Back to Top | Article Outline

MATERIALS AND METHODS

Study Population and Standardized Treatment

During a 44-month period ending August 2001, 188 major torso trauma patients who met specific criteria (see below) underwent standardized resuscitation and comprised the study population. This represents 8 percent of the 2,258 Shock Trauma Intensive Care Unit (STICU) admissions and 1 percent of the 16,376 trauma admissions to Memorial Hermann Hospital (MHH) in Houston, Texas. MHH is the lead regional Level-I Trauma Center which serves Trauma Service Area - Q, a nine county region of the upper Gulf Coast of Texas with a population of roughly 4 million. After transport by local EMS or our air ambulance service, patients are initially evaluated and managed by the Advanced Trauma Life Support (ATLS) protocols. 13 After the required emergent interventions, major torso trauma patients were admitted to the STICU. A pulmonary artery catheter and gastric tonometer were presumptively placed in high-risk patients and standardized resuscitation protocol emphasizing volume loading and hemoglobin replacement to maintain systemic O2 delivery for the first 24 ICU hours was initiated. The protocol has been described previously. 14 In brief, the risk for multiple organ failure (MOF) and need for shock resuscitation is objectively defined by: (1) major torso trauma (including flail chest, ≥2 abdominal injuries, major vascular injury, complex pelvic fracture, ≥2 long bone fractures); (2) early arterial base deficit (BD) ≥6 mEq/L; and (3) need for transfusions of ≥6 units packed red blood cell (PRBC) during the first 12 hours, or a trauma victim of age ≥65 years with any two of the previous criteria. Each patient is assessed with these criteria by the attending trauma surgeon. Patients with these injury criteria who also have incurred significant brain injury (defined as Glasgow Coma Scale score ≤8 in the ICU and brain CT scan abnormalities) are not resuscitated by protocol unless the patient’s brain injury has been assessed by the attending neurosurgeon to be at low risk of worsening cerebral edema with volume loading. The protocol constitutes a hierarchy of five sequentially applied therapies (with intervention thresholds) to achieve DO2I ≥600 mL/min-m2 (DO2I ≥500 mL/min-m2 since January 2001) for 24 hours as the protocol goal: (1) lactated Ringer’s solution (pulmonary capillary wedge pressure [PCWP] <15 mm Hg); (2) PRBC transfusions (hemoglobin [Hb] <10 g/dL); (3) Starling curve to optimize cardiac index (CI)-PCWP relationship (Hb ≥10 g/dL, PCWP ≥15 mm Hg, and DO2I <600 mL/min-m2); (4) inotropes (CI-PCWP optimized and DO2I <600 mL/min-m2); and (5) vasopressors (MAP <65 mm Hg). During the protocol, urinary bladder pressure is monitored upon ICU admission and every 4 hours unless the attending trauma surgeon orders more frequent monitoring. Abdominal decompression was performed based on the attending trauma surgeon decision by midline laparotomy and abdominal exploration. After damage control laparotomy or abdominal decompression a Bogota-bag was the method of choice for temporary abdominal closure. If the fascial closure was not achieved at the time of the first re-exploration, vacuum-assisted wound closure technique was applied to facilitate fascial re-approximation. 15

Back to Top | Article Outline

Primary Outcome

Intra-abdominal pressure was determined indirectly by measuring urinary bladder pressure (UBP) using the method described by Krone et al. 16

ACS was defined if abdominal decompression was performed based on the attending trauma surgeon decision in a patient with UBP higher than 25 mm Hg (Grade III and IV IAH) with progressive organ dysfunction (urinary output <1 mL/kg/h or Pao2/FiO2 <150 mm Hg or PAP ≥45 cm H2O or CI <3 L/min/m2) despite resuscitation, which improved after decompression. 9 1° ACS was defined if any intraperitoneal injury(ies) were recorded. 2° ACS was defined in the absence of intra-peritoneal injury. 4

Back to Top | Article Outline

Collected Variables

The study was approved by the University of Texas Health Science Center at Houston’s (UTHSCH) Institutional Review Board and was conducted according to the guidelines of the UTHSCH Committee for the Protection of Human Subjects.

Based on previous studies and local expert opinion the possible ACS risk factors were collected and maintained in a computer database. The collected variables were the following:

  1. Demographics and injury severity [age, gender, mechanism, injury severity score (ISS), 17 new injury severity score (NISS), 18 abdominal trauma index (ATI) 19].
  2. Emergency department (ED) variables [admission systolic blood pressure (SBP), time spent at the ED (ED time), base deficit obtained before ICU admission (preICU BD), ED crystalloid administration, ED packed red blood cell (PRBC) transfusion].
  3. Urgent operative interventions and pelvic embolization before ICU admission
  4. Variables available at the time of ICU admission [time from ED admission to ICU admission (time to ICU), preICU crystalloid infusions, preICU PRBC transfusions, temperature, coagulation results, hemoglobin concentration (Hb), BD, lactate concentration, UBP, gastric tonometry data [gastric mucosal CO2 (PgCO2) minus end tidal CO2 = CO2 gap (GAPCO2)], pulmonary artery catheter data [pulmonary capillary wedge pressure (PCWP), cardiac index (CI), systemic vascular resistance index (SVRI)], central venous pressure (CVP), mixed venous O2 saturation (SvO2), respiratory parameters [peak airway pressure (PAP), static pulmonary compliance (CST)], urine output (UO) during the first ICU hour.
  5. The variables listed in 4 were monitored and recorded throughout the first 24 hours of the ICU stay.
  6. The 1° and 2° ACS patients pre- and postdecompression ICU variables (pre and postdecompression 4 hours averages for each variable), the times to decompression from ED and ICU admission, time to achieve abdominal closure, abdominal abscesses.
  7. Outcome variables [days on the ventilator, ICU length of stay (ICU LOS), hospital LOS (LOS), multiple organ failure (MOF), 20 mortality]. While for the definition of ACS the acute organ dysfunction criteria were used (see Primary Outcome section), MOF was defined by Denver MOF score, which only considers organ dysfunction after 48 hours.
Back to Top | Article Outline

Statistical Analysis

Univariate analysis was performed by using Student’s t test or Wilcoxon rank-sum test for continuous data and χ2 test for categorical data. Results are expressed as mean ± SEM. A p-value less than 0.05 was considered significant.

The risk factors for 1° and 2° ACS were identified by multiple logistic regression analysis. The first pass was done using stepwise model selection to determine which combinations of variables had the strongest statistical associations. Variables that were significant by stepwise selection or that swapped repeatedly in and out of the stepwise models were evaluated again by themselves and by using best-subsets selection. Influence of the variable combinations on the overall model log-likelihood score as well as on the stability of individual variable Wald χ2 estimates was observed manually during final model building. Two models were derived using sequential sets of patient data collected over the time until the end of the first hour of ICU admission: (1) ED model (data available upon leaving the ED plus demographics and injury severity); (2) ICU admission model (data available upon the end of first ICU hour). UBP as part of the present clinical definition of ACS was not included any of the models. Separate models were derived in both time windows for 1° ACS, 2° ACS and all ACS. For the 1° and 2° ACS models the control groups included the other ACS subgroup (1° ACS patients were in the control group for the 2° ACS model and vice versa) in addition to nonACS patients. The purpose of performing these various analyses was to identify consistent risk factors for 1° ACS, 2° ACS, and all ACS. The fit of the models was assessed by Hosmer-Lemeshow (H-L) goodness of fit statistics. Receiver operating characteristic (ROC) curves was produced for each model and areas under the curves were calculated to evaluate model prediction.

ACS was tested by logistic regression analysis to determine whether this complication is a predictor for the adverse outcomes such as MOF and death. All computations were performed using SAS version 8.02 (SAS Institute, Inc., Cary, NC) running under Windows 2000 (Microsoft Corporation, Redmond, WA ).

Back to Top | Article Outline

RESULTS

Epidemiology of ACS

Of the 188 study patients, 146 (76%) were male, mean age was 39 ± 2 years, injury mechanism was blunt in 160 patients (85%), and mean ISS was 28 ± 2. Twenty-six (14%) patients developed ACS, eleven (6%) had 1° ACS and fifteen (8%) had 2° ACS. Table 1 compares the demographics, injury characteristics, and outcomes of patients with 1° ACS versus 2° ACS versus nonACS patients. The groups had the same age, gender distribution, and high incidence of blunt mechanism. All study patients presented in shock, but the ACS groups had lower ED systolic blood pressure (ED SBP) and received more units of packed red blood cells (PRBC’s) during the first 12 hours. Of note, 85 percent (160/188) of the study cohort had at least one urgent operation or interventional radiologic embolization before ICU admission. Since severe brain injury patients were excluded, the GCS scores were high and AIS head were low. The 1° ACS group had more severe abdominal injuries reflected by higher ATI and abdominal AIS scores. The AIS extremity score was higher in the 2° ACS group because more of these patients had major pelvic fractures. 1° and 2° ACS had similar poor outcomes compared with nonACS in regards to ventilator days, multiple organ failure, and mortality.

Table 1

Table 1

Figure 1 compares hospital times until decompression. 1° and 2° ACS patients spent the same period of time in the hospital before they were decompressed (13.7 ± 2 and 12.2 ± 2 hours) but 1° ACS patients spent less time in the ED (0.9 ± 0.1 vs. 3 ± 0.3 hours). The times spent between the ED and ICU were similar (2.8 ± 0.3 vs. 3.2 ± 0.5 hours). Overall, 78 percent of the study patients required urgent operations and 17 percent required urgent interventional radiologic angio-embolization. 1° and 2° ACS patients had different interventions before ICU admission. Nine (82%) of the 1° ACS patients had urgent operations (nine damage control laparotomy, two thoracotomies, and one lower limb revascularization), only two patients (18%) did not undergo urgent laparotomy (nonoperative management of severe liver ruptures), and only one (9%) had pelvic embolization before ICU admission. Only six (40%) of the 2° ACS patients had urgent operations (three thoracotomies, one above the elbow amputation, one femoral artery and vein repair, one pelvic external fixator), but seven (47%) of them required pelvic embolization. Two (13%) of the 2° ACS patients had neither urgent operations nor pelvic embolization. These patients had multiple closed long bone fractures and pelvic fractures. Note on Figure 2 that the 1° ACS and nonACS groups had a similar distribution of urgent operations and embolizations. Figure 3 depicts the time to decompression from the admission to the ICU. 2° ACS occurred earlier (6 ± 1 hours) than the 1° ACS (10 ± 2 hours). All ACS patients were decompressed within their first 24 hours of ICU stay. Figure 4 demonstrates the UBP of the study population during the first 24 hours. The UBP of the 2° ACS patients peaks earlier and then decreased as abdomens were being decompressed, while 1° ACS patients’ UBP reached its maximum 6 hours later. Note the relatively high (∼15 ± 1 mm Hg) UBP of the nonACS group.

FIG 1

FIG 1

FIG 2

FIG 2

FIG 3

FIG 3

FIG 4

FIG 4

Figures 5, 6, and 7 summarize the crystalloid and blood transfusion in the first 24 hours. Over the first 24 hours, the 2° ACS patients received more crystalloid fluid than the 1° ACS patients, who in turn, received more than nonACS patients. These same differences are seen in the ED and ICU, but not in the IR/OR suites. As can be seen in Figure 5B, the rate of infusion in the ED was higher in primary ACS, while the rate in the ICU was higher for secondary ACS.

FIG 5

FIG 5

FIG 6

FIG 6

FIG 7

FIG 7

Figure 6A shows, that both 1° and 2° ACS patients received more PRBC transfusions during their first 24 hours and on the ICU than nonACS patients. Over the first 24 hours, 1° and 2° ACS patients received the same amount of transfusions, but the 2° ACS group had more transfusions in the ED with the same transfusion rate (Fig. 6B) compared with the 1° ACS group. During the urgent interventional radiologic and surgical interventions (IR/OR), the 1° ACS patients received more blood transfusions with a higher rate. In the ICU, 1° and 2° ACS patient transfusion volumes were similarly higher than the nonACS group.

Figure 7 compares the crystalloid/PRBC ratios. The only time interval during the first 24 hours hospital stay when the crystalloid/PRBC ratios were different among groups was the IR/OR period. The highest ratio was in the 2° ACS group and the lowest was in the 1° ACS group (Fig. 7). Again, this period in the nonACS and 1° ACS groups are dominated by operative interventions in the OR, and in the 2° ACS group, this period is dominated by pelvic embolization in the interventional radiology.

Table 2 summarizes the ACS patients’ physiologic response to decompression (The 1° and 2° ACS patients had the similar response, therefore they are presented together as ACS). The surgical decompression improved the physiology of both 1° and 2° ACS patients regardless of their outcome. The UBP, SVRI, PgCO2, GAPCO2, BD, and PAP decreased, while the urinary output, MAP, CI, SvO2, arterial pH, static pulmonary compliance, and the Pao2/FiO2 ratio increased. Figure 8 depicts the two variables which were different after decompression between survivors and nonsurvivors. Survivors had a better urinary output response than nonsurvivors (Fig. 8A). The cardiac index was the only variable that improved only among survivors but not among those who died (Fig. 8B).

Table 2

Table 2

FIG 8

FIG 8

Table 3 compares the 45 damage control patients with open abdomens in the nonACS group (28%) with the 1° and 2° ACS patients in terms of times to definitive closure and the incidence of abdominal abscesses. The fascial closure of open abdomens was achieved earlier if the underlying cause was 2° ACS (3 ± 0.8 days) than when the cause was 1° ACS (9 ± 2 days) or damage control in the nonACS group (7 ± 2), with less number of OR trips in the 2° ACS group than in the nonACS and 1° ACS groups. Fascial closure was achieved in all surviving 2° ACS patients. One 1° ACS patient and three nonACS damage control patients required plastic surgical interventions to treat their abdominal wall defect. Abdominal abscesses were more frequent in the 1° ACS group (18%) than in the nonACS (5%) and 2° ACS (7%) groups.

Table 3

Table 3

Back to Top | Article Outline

Prediction Model for ACS

The univariate comparisons of ED and ICU variables of the study population, whichwere used for the prediction models, are listed in Table 4 and the results ofmultivariate analysis aresummarized in Tables 5–7. Table 5 shows the ED model for predictionof ACS. At this very early time point (<3 hours after admission) the multiple logistic regression analysis identified high crystalloid volume and low SBP as independent risk factors for all ACS. The independent risk factors for 1° ACS are high crystalloid volume and short ED time (requiring urgent damage control); those for 2° ACS are high crystalloid volume, more than 3 units of PRBC transfusions, and the lack of the need for urgent surgery. Table 6 shows the prediction model developed from the data available within the first hour of ICU admission (≤6 hours from hospital admission). The independent predictors of ACS in general were high GAPCO2, high crystalloid volume, low urinary output, low Hb and low CI. Those for 1° ACS were low temperature, high GAPCO2, low Hb, and high BD; for 2° ACS were high GAPCO2, high crystalloid volume, and low urinary output. Table 7 shows that all predictive models had a very good ROC area results. The predictive models developed at the ED for both 1° ACS and 2° ACS have higher ROC areas than the model built up for the prediction of ACS in general. The ICU prediction models had better ROC results than the ED models.

Table 4

Table 4

Table 5

Table 5

Table 6

Table 6

Table 7

Table 7

ACS in overall is a predictor of both MOF (odds ratio = 9.2; 95% confidence intervals, 3.8–22.8;p < 0.0001) and mortality (odds ratio = 8.4; 95% confidence intervals, 3.5–20.6;p < 0.0001).

Back to Top | Article Outline

DISCUSSION

Postinjury 1° ACS became evident with the widespread application of damage control laparotomy. 21–23 More recently it was recognized that ACS can also occur in cases where only extraperitoneal injuries are present (2° ACS). 4,5,12 The early case studies and reviews characterized 1° and 2° ACS by impaired renal function, high ventilator pressures, and poor cardiac function, which are reversed after surgical decompression of the abdomen. 7,24 Despite good physiologic response to decompression, the reported outcome of ACS is consistently very poor. 11 Recognizing that primary fascial closure after damage control was frequently associated with ACS, preventive “Bogota Bag” open abdomen closure has been recommended. 22 Several authors have suggested that routine UBP monitoring could lead to earlier decompression and better outcome. 1,7–9 Not surprisingly, patients whose ACS was recognized during the first 24 hours had lower mortality than those who were decompressed after 48 hours. 4,5 In contrast, we have previously reported that when early decompression (<12 hours from admission) was unvaryingly performed, there is no difference in time to decompression between patients who survived versus those who died. 12 The utilization of damage control techniques acutely saves the lives of severely injured patients with critically impaired physiology, but we now have to face the unsolved complication of ACS. Despite the liberal use of UBP measurements and early decompression, the ACS remains a difficult challenge. Believing that the inadequate epidemiologic characterization and the failure to identify the syndrome’s risk factors are the key points in the recent years’ stagnation in the ACS prevention and treatment, we focused our efforts on providing the necessary epidemiology, early prediction models, and outcome determinants of the ACS and its subgroups.

The exact incidence of ACS is difficult to determine, and it is dependent on both the definition of the ACS (numerator) and the patient population used for the actual study (denominator). Some authors used mixed (trauma, general surgery, burns) populations, which makes it difficult to compare their findings with the postinjury cohorts. The incidence of ACS in our high-risk shock/trauma ICU resuscitation cohort was 14 percent. If all patients admitted to the shock trauma ICU are used as the denominator, the incidence of ACS is 1.2 percent. In terms of the 1° ACS, our results are comparable with those of Meldrum et al. 7 Fourteen percent incidence of ACS in a group of ICU admitted patients with an ISS >15 who had undergone emergency laparotomy and ISS >1515” used twice in this sentence -- meaning not understood. Please reword for clarity.‘. Ertel et al. 25 had similar inclusion criteria (abdominal or pelvis AIS >3 and/or laparotomy) but used only clinical assessment with no UBP to diagnose ACS in the majority of patients, and they reported a 5.5 percent incidence. Study populations including only damage control laparotomy patients report higher (33–36%) incidence of ACS. 1,2 We identified nine 1° ACS patients and 45 nonACS patients among the 54 damage control patients. This 17 percent (9/54) incidence of 1° ACS among damage control patients is lower than the previous studies. This may be attributable to the fact that our damage control patients had their abdominal fascia left open (i.e., underwent Bogota bag closure). Only four patients (4/54, 7%) had towel clip closure. Our results are comparable with Offner et al.’s. 2 For all damage control patients their incidence was 33 percent, but the incidence in those patients with fascial closure was 80 percent compared with the subset of Bogota bag closure patients incidence of 17 percent. Less data are available to establish the incidence of the 2° ACS, which is more elusive and recognized only in the last 5 years. Maxwell et al. 4 reported six 2° ACS patients from 1216 ICU admissions (0.5%), while we had 15 patients from 2258 Shock Trauma ICU admissions (0.7%). To date, there is no data comparing the relative incidence of 1° ACS and 2° ACS in the same institution among patients treated in a standardized fashion. According to our results, 1° and 2° ACS had very similar incidence (6% vs. 8%) in this high-risk major-torso trauma group.

Severe hemorrhagic shock was uniformly present in the whole cohort (Table 1), but both 1° and 2° ACS patients had a lower initial SBP and higher transfusion requirement than the nonACS group. All patients had high base deficit on hospital arrival (∼10 mEq/L), but nonACS patients responded well to the initial resuscitation, and by the time of ICU admission, their BD had improved (4 mEq/L), contrary to the 1° and 2° ACS groups (still ∼10 mEq/L) (Table 4). Hemorrhagic shock is consistently reported in the literature as a prerequisite of ACS. Although ACS can happen in patients with moderate injuries, 5 ACS patients typically have a high ISS. In the present study there was no difference among the groups in terms of injury severity. The injury pattern, however, was rather different (Table 1); the typical 1° ACS patient had severe abdominal injuries, while the classic 2° ACS patient had severe pelvic fractures or major extraperitoneal vascular injuries with multiple long bone fractures. Because of these differences in injury pattern, the 1° and 2° ACS patients’ hospital course was also different (Fig. 2). 1° ACS patients had a very short ED course because of the obvious need to go to the operating room for abdominal hemorrhage control. The OR intervention uniformly included quick damage control procedures because of the bloody “vicious cycle” physiology, 23 while they received massive transfusions. Only two 1° ACS patients did not have primary damage control laparotomy. In these two cases, 1° ACS presented as a failure of nonoperative management of blunt liver injuries. 2° ACS patients typically had more prolonged diagnostic work-up, and most frequently, they required pelvic embolization in the interventional radiology suite.

ACS patients received more crystalloid infusions and PRBC transfusions during their first 24 hours of hospital stay (Fig. 5 and 6). 2° ACS patients received more crystalloid, but not more blood compared with 1° ACS patients. The most remarkable difference between 1° and 2° ACS patients was during their interventional period (operating room and/or interventional radiology) when the 2° ACS group’s crystalloid/PBRC ratio was 1.92 ± 0.3 L/unit while the 1° ACS patients’ ratio was only 0.55 ± 0.1 L/unit (Fig. 7). The possible role of crystalloid overload in the pathogenesis of 2° ACS has been suggested, but until now crystalloid volume had not been shown to be an independent risk factor for ACS. 4,5,12,26 Our data showed that ≥3 L of crystalloid infusion in the ED predicts both 1° and 2° ACS. More than 7.5 L of crystalloid infusion before ICU admission strongly predicts 2° ACS. Both 1° and 2° ACS can be accurately predicted at the time of ICU admission (Table 7, ROC results). Some characteristics of the two syndromes are different (injury pattern, fluid resuscitation, preICU course), and they have distinct predictors. The risk factors for 1° ACS are the clinical markers of the “bloody vicious cycle,” while 2° ACS occurs in patients with low urinary output despite massive crystalloid resuscitation in those who have a high GAPCO2. Our data confirm the previous observation that gastric tonometry may be an important monitoring tool for ACS. 1,27,28

Due to the abundant monitoring inherent to our resuscitative process, ACS patients were diagnosed and treated relatively early in their ICU course (Fig. 3). While there was no difference between 1° and 2° ACS times to decompression from hospital admission, 2° ACS occurred earlier after ICU admission. This could be because 1° ACS patients’ abdomens were predominantly open, and therefore, they needed longer times to develop the critical UBP level that is known to cause ACS symptoms. The other fact is that 2° ACS patients had longer pre-ICU courses where the resuscitation was less controlled compared with the ICU.

An interesting additional finding is that the UBP of the nonACS patients (Fig. 4) was relatively high (≥15 mm Hg) and, according to Burch’s classification, represents Grade II intra-abdominal hypertension. 9 The normal UBP in a mixed group of nontrauma hospitalized patients was 6.5 mm Hg (range, 0.2–16.2 mm Hg), 29 however the UBP of critically ill patients is reported to be higher. 30 Recent basic science research suggests that sequential hemorrhagic shock and abdominal compartment syndrome resulted in more severe inflammatory response than either insult alone. 31,32 Critically ill injured patients who require massive fluid resuscitation may not tolerate even lower levels of intra-abdominal pressure. The possible adverse effect on the outcome of this elevated intra-abdominal pressure without meeting the ACS criteria needs to be evaluated.

In our cohort both ACS subgroups had the same good physiologic response to decompression (Table 2). The uniform improvement in hemodynamic and respiratory parameters is well described in previous studies. 1–12 More recently it was pointed out that the good response to decompression is not necessarily associated with better outcome. 11–12 We identified only two variables that predict better outcome: cardiac index and urinary output improvement were better among survived patients than those who died. Note that an hourly urinary output of 100 mL can be an alarming sign for impaired renal function, although this is far above the generally accepted value of 0.5 mL/kg/h.

Fascial closure of open abdomens was achieved earlier and with less number of operations in the 2° ACS group compared with the 1° ACS and nonACS damage control groups (Table 3). The 2° ACS patients had no abdominal injuries and appeared to resolve their intestinal edema more quickly. While time to fascial closure was not pre-defined outcome of this study, our patients on average achieved fascial closure 7 days earlier than was reported in previous series. 33 The possible beneficial effect of the vacuum assisted closure in the prevention of fascial retraction needs further evaluation. 15

The outcome of ACS patients measured by ventilator days, the incidence of MOF and mortality was significantly worse than that of the nonACS patients’ with similar demographics, shock, and injury severity (Table 1Outcomes). In terms of outcome the ACS patients were similar regardless of the type of ACS. Several papers suggested the possible connection between ACS and poor outcome intuitively, 2,12,26,33,34 but our present study has proven statistically that ACS is a predictor for both MOF and mortality based on logistic regression analysis.

Studies in the last 15 years, including our cohort, have failed to document a convincing improvement in the outcome of ACS despite earlier decompression and the liberal use of the temporary open abdomen techniques. This suggests that efforts directed at prevention will be more fruitful than efforts directed at early recognition and decompression. The multiple logistic regression analysis showed that high-risk patients are identifiable very early during resuscitation. 1° ACS patients develop the “bloody vicious cycle” physiology at which point abbreviated surgical intervention is the only way to save their lives. Packing the abdomen is an efficient tool to achieve hemorrhage control, but the space occupied by packs, the edematous bowel, and the recurrent bleeding after optimization of the circulation are important factors in 1° ACS. Prevention requires the avoidance of tight packing, the application of large Bogota bags, the elimination of hypothermia, and re-evaluation for possible bleeders are essential. In the future, novel hemorrhage control techniques without significant space occupying nature could have an important role. 35 While the incidence of 1° ACS can be decreased by alternative damage control techniques, trauma surgeons will continue to face this complication in patients with life threatening abdominal injuries. In the nonoperative management of abdominal solid organ injuries 1° ACS, until most recently, meant the failure of the nonoperative management and mandated exploration. Novel case reports allude to decompression that can be done without laparotomy by draining the peritoneal cavity. 36

2° ACS is predictable in patients with shock and massive crystalloid resuscitation without intra-peritoneal injuries. The standard of care crystalloid fluid based resuscitation seems to be effective in the vast majority of this high-risk cohort, but not among patients who are prone to 2° ACS. The application of alternative resuscitation fluids, such as hypertonic saline and colloids, should be considered in this subgroup. The analysis of fluid resuscitation before ICU admission. Figure 7 shows the uncontrolled nature of resuscitation in terms of PRBC/crystalloid ratio of the 2° ACS patients. To prevent this, pre-ICU needs to be better controlled. Of note, 73 percent (11/15) of the 2° ACS patients had major pelvic fracture, and 47 percent (7/15) of them required urgent angiographic embolization prompted by hemodynamic instability. The interventional radiologic embolization of the pelvic arterial bleeding is a valuable adjunct, but bleeding from the broken cancellous bony surfaces and from the retroperitoneal venous plexuses is more common. In these types of bleedings reduction of the pelvic ring and compression between the oozing cancellous bony surfaces are essential. The minimally invasive urgent pelvic stabilization techniques need to be applied in selected cases. The continuous oozing from these fractures and veins can lead to cyclic crystalloid resuscitation/re-bleeding, which may cause transient hemodilution, decreased capillary oncotic pressure, and intestinal edema. As described earlier, the increased pelvic content with the intestinal edema are key elements of the mechanism of the 2° ACS. 12

In summary, based on postoperatively collected data from a high-risk group of patients who were resuscitated in a similar manner, we conclude that ACS is a frequent early complication and is associated with massive volume loading. 1° and 2° ACS patients have different injury characteristics and undergo different preICU resuscitation, but both can be predicted as early as 6 hours after hospital admission with adequate monitoring. The outcome of ACS remains very poor and despite earlier recognition mortality did not improve. Further efforts should focus on prevention of the syndrome, most likely by: (1) standardizing and better monitoring the pre-ICU phase of the resuscitation; (2) alternative resuscitation strategies; and (3) use of novel hemorrhage control techniques in patients who are at high-risk to develop ACS.

Back to Top | Article Outline

DISCUSSION

Dr. Patrick Reilly (Philadelphia, Pennsylvania): Dr. Balogh and the group from Houston have presented data from their prospectively collected database which includes 188 major torso trauma patients undergoing a standardized resuscitation in the ICU at Memorial Herman Hospital.

From this patient population, 26 patients were identified who developed abdominal compartment syndrome.

Most of these patients were identified and treated within 14 hours of hospital presentation. Despite a heightened level of awareness and an increased urgency for prompt treatment of ACS, mortality in the group remained nearly 60 percent.

An enormous amount of information about these patients is provided, and rigorous statistical analysis ultimately leads to a predictive model for the development of abdominal compartment syndrome.

It should be noted these 26 patients compromised only 1.2 percent of all ICU admissions and a much smaller percentage of all admitted patients during this study period.

Still, they represent an important population with tremendous basic science and clinical relevance.

It is, therefore, not surprising that a recent Medline search of the key words “abdominal compartment syndrome” yielded 158 citations, most of which have been published in the last 5 years.

Unfortunately, the overwhelming majority of these citations are case reports, case series, and small clinical studies that have attempted to characterize the population in question.

So, with the addition of this manuscript, are we better off? I think the answer is clearly, yes, and clearly, no.

The epidemiology of this patient population has been eloquently described. A potentially helpful predictive model, taking into account degree or depth of shock and volume of resuscitation fluid, among other variables, is created that could be employed after resuscitation in the emergency department or ICU.

Many of these results confirm my own, and most likely most of the audience’s, clinical sense that sicker patients, especially those who receive significant crystalloid resuscitations, are at an increased risk for ACS.

It’s always nice to have your clinical suspicions confirmed by scientific review, and in that sense, I am better off.

However, I’m left with the nearly 60 percent mortality in the ACS group presented here—this from a Herman Hospital team that has written extensively about this subject and clearly is clinically “keyed in” to the phenomenon of abdominal hypertension and abdominal compartment syndrome.

I have a few questions for the authors. First, you arbitrarily divide your study patients into primary and secondary ACS groups based on whether abdominal injury is present.

However, the remaining nonACS patients are lumped together into one control group regardless of whether they have abdominal injury.

Have you examined your data comparing ACS patients with abdominal injuries to nonACS patients with abdominal injury and performed a similar analysis for those patients without abdominal injury?

Perhaps other variables could be identified that are associated with the development of ACS and that are lost in the current analysis.

Second, patients in the study were standardized by the uniform resuscitation guidelines that were followed in the ICU; however, in your manuscript, you mentioned the relatively “uncontrolled” resuscitation that occurred in some patients before their arrival in the ICU.

Crystalloid volumes were quite significant and crystalloid packed red blood cell ratios high. A number of studies have identified an association between significant crystalloid resuscitations and ACS.

We also have found a relationship in damage control patients between these same crystalloid resuscitations and an increased incidence of complications associated with hollow viscus injury repairs as well as a decreased ability to obtain delay primary fascial closure during the initial hospitalization.

How do you plan to change the resuscitation philosophy at Memorial Herman Hospital in this patient population at risk?

Do you think taking your resuscitation guideline out of the ICU and to the patient will impact the incidence of abdominal compartment syndrome?

Is there a role for a colloid resuscitation in these select patients? How useful do you really think your predictive model will ultimately be?

The ED model is less sensitive, although rapidly available. The ICU model is more sensitive but relies on some numbers such as crystalloid resuscitation greater than 7.5 liters that might only be obtained after the horse is already out of the barn.

The majority of your damage control patients already are left open rather than having fascial or even skin closure.

How will your predictive model really change your clinical practice? Have you preemptively opened an abdomen with a bladder pressure less than 25 or other clinical signs of abdominal compartment syndrome based solely on your model?

I greatly enjoyed this presentation, and the manuscript that accompanied it is well written. I commend the Houston group on their past efforts and look forward to their future projects, both at the basic science bench and the bedside.

Hopefully, in the not too distant future we can all build on this current project and come up with a resuscitative and/or operative strategy to prevent the development of this highly lethal complication.

I would like to thank the Association and the Program Committee for the opportunity to discuss this paper.

Dr. Zsolt Balogh (Houston, Texas): Thank you very much for your questions and comments, Dr. Reilly. I’ll try to answer all of them.

First of all, your first question was about the study design. We attempted to design and perform the most comprehensive clinical ACS study on post-injury ACS, and we spent months on thinking how to allocate our patients in study groups.

Originally, we had four groups. The first group was a primary ACS, and the control was the other resuscitation patients with abdominal injuries.

The third group was secondary ACS, and their controls were those resuscitation patients who had no abdominal injuries.

This resulted in four groups and less patients in each group, higher standard errors, and less statistical power.

This analysis didn’t help us to find more predictors than what we included in the manuscript. The other thing that this manuscript, together with epidemiology, prediction, outcome, and all the decompression responses, otherwise quite lengthy, and further groups would cause further expansion and even confusion.

The second question was about the less-controlled preICU resuscitation. It’s a very good observation from our manuscript.

Early on, our resuscitation protocol was criticized that probably the ICU protocol causes and the resuscitation protocol causes the abdominal compartment syndromes.

Our very first observation was that ACS occurs very early during the ICU course, so there is virtually no time to over resuscitate these patients on the ICU.

However, the pre-ICU resuscitation is less controlled and, as was clear from the presentation, is driven by the vital signs such as blood pressure and heart rate.

We already started to implement or modify the resuscitation protocol to the pre-ICU setting and possibly prevent abdominal compartment syndrome during the early hospital course.

The question about the use of alternative fluid resuscitation is another good one. I think crystalloid and blood resuscitation is presently the standard of care in the United States trauma centers.

At first, we wanted to know what we had. As Dr. Reilly alluded, most of the studies are case reports, retrospective reviews, and reviews of the reviews, and that’s why we wanted to perform an epidemiologic study and find out how the crystalloid volume affects the outcome and the development of ACS.

Based on these results, those patients who are at high risk of ACS, alternative resuscitation either with colloids or hypertonic saline may be warranted.

We can tell you that on that 85 percent of the shock resuscitation patients, the crystalloid and blood resuscitation seems to work.

If you can predict the 15 persons who are at high risk at ACS, you can actually prevent them. The third question was about the use of our prediction model in the future.

Yes, that’s true. A prediction model is absolutely worthless if it’s not implemented into the clinical practice.

So, we already started to use this model in the clinical practice. As you see, there are distinct predictors for primary and secondary ACS.

The primary ACS predictors are indicators of the bloody vicious cycle that can help us to leave these abdomens open and have a higher level of suspicion to decompress these patients earlier.

For the secondary ACS patients, there are two important things. One is the crystalloid limits that we implement into the resuscitation protocol as an alarming sign.

If the patient starts to reach this amount of crystalloid, an alternative resuscitation may start. The other thing is the 150 milliliters per hour urine output may be adequate in other patients, but in these aggressively resuscitated patients, that output may be low—an alarming sign of renal failure.

Yes, we do decompress patients with bladder pressure lower than 25, as we already started, because bladder pressure is not a very sophisticated method, and it cannot be treated as an absolute number.

If the physiology worsens and the predictors show that these patients are on the tract to develop ACS, they should be decompressed or treated as an ACS.

Dr. Steven R. Shackford (Burlington, Vermont): This is a very nicely presented paper, and I congratulate you on the delivery.

I would like to focus, just for a second, on that secondary abdominal compartment group with the large crystalloid resuscitation.

Now, did I understand you to say that most of the resuscitation of those patients was carried out before they got to the ICU—is that right?

Dr. Zsolt Balogh (Houston, Texas): Yes, when they reached the ICU, they were decompressed between 3 to 4 hours.

So most of the resuscitation occurred in the pre-ICU setting in the diagnostic evaluation.

Dr. Steven R. Shackford (Burlington, Vermont): So, what were the end points of resuscitation? I saw your ICU resuscitation protocol began with the first step of being driven by a relatively arbitrary or target DO2 or oxygen delivery, but down in the ER, or your resuscitation space, you couldn’t do that because you probably didn’t have a pulmonary artery catheter in and couldn’t obtain a cardiac output to calculate the DO2, correct?

Dr. Zsolt Balogh (Houston, Texas): Yes.

Dr. Steven R. Shackford (Burlington, Vermont): So, what were the end points of resuscitation in that group? What were you using to titrate: the amount of crystalloid you were giving those patients?

Dr. Zsolt Balogh (Houston, Texas): Our resuscitation protocol, which will extend to the pre-ICU setting, will be a modified one, and it will be driven by the CVP and the SvO2.

According to the primary reports, SvO2 can be a very good predictor for these adverse outcomes.

Dr. Steven R. Shackford (Burlington, Vermont): That’s a central –

Dr. Zsolt Balogh (Houston, Texas): Yes, it’s a central venous line.

Dr. Dennis Wang (Washington, D.C.): I just want to caution you not to rise to the bait of Dr. Reilly in actually implementing releasing abdominal compartment syndrome based on your predictive model, because some of the issues have not been addressed specifically—things such as why the patient died.

Is this from multi-organ failure due to resuscitation, or was resuscitation failure not specifically addressed?

Also, you may have to look at this compounding factor: is your resuscitation pre-ICU, or is it the actual resuscitation protocol that lead to your morbidity and mortality?

These are things that have to be addressed before you go ahead and actually start opening abdomens, before you reach the compartment pressure greater than 25.

Dr. Zsolt Balogh (closing): Thank you very much. The patients are not decompressed according to the prediction model, but they are decompressed at this time according to the measured organ dysfunction parameters such as cardiac index, respiratory, and renal failure. I think we answered the question in the presentation that the less controlled pre-ICU resuscitation and the physiology which leads to damage control surgery are important predictors of postinjury primary and secondary ACS. Once these patients reach the ICU, the syndrome is already there, and there is virtually no time to over-resuscitate them by the ICU resuscitation protocol.

Back to Top | Article Outline

REFERENCES

1. Ivatury RR, Porter JM, Simon RJ, et al. Intra-abdominal hypertension after life-threatening penetrating abdominal trauma: prophylaxis, incidence, and clinical relevance of gastric mucosal pH and abdominal compartment syndrome. J Trauma. 1998; 44: 1016–1023.
2. Offner PJ, de Souza AL, Moore EE, et al. Avoidance of abdominal compartment syndrome in damage-control laparotomy after trauma. Arch Surg. 2001; 136: 676–681.
3. Morris JA, Eddy VA, Blinman TA, et al. The staged celiotomy for trauma: Issues in unpacking and reconstruction. Ann Surg. 1993; 217: 576–586.
4. Maxwell RA, Fabian TC, Croce MA, Davis KA. Secondary abdominal compartment syndrome: an underappreciated manifestation of severe hemorrhagic shock. J Trauma. 1999; 47: 995–999.
5. Kopelman T, Harris C, Miller R, Arrillaga A. Abdominal compartment syndrome in patients with isolated extraperitoneal injuries. J Trauma. 2000; 49: 744–749.
6. Hong JJ, Cohn SM, Perez JM, et al. Prospective study of the incidence and outcome of the abdominal compartment syndrome. Br J Surg. 2002; 89: 591–596.
7. Meldrum DR, Moore FA, Moore EE, et al. Prospective characterization and selective management of the abdominal compartment syndrome. Am J Surg. 1997; 174: 667–672.
8. Ivatury RR, Diebel LD, Porter JM, et al. Intra-abdominal hypertension and the abdominal compartment syndrome. Surg Clin North Am. 1997; 77: 783–800.
9. Burch JM, Moore EE, Moore FA, et al. The abdominal compartment syndrome. Surg Clin North Am. 1996; 76: 833–842.
10. Schein M, Wittmann DH, Aprahamian CC, et al. The abdominal Compartment syndrome: The physiological consequences of elevated intra-abdominal pressure. J Am Coll Surg. 1995; 180: 745–753.
11. Saggi BH, Surgerman HJ, Ivatury RR, et al. Abdominal compartment syndrome. J Trauma. 1998; 45: 597–609.
12. Balogh Z, McKinley BA, Cocanour CS, et al. Secondary Abdominal compartment syndrome: An elusive complication of traumatic shock resuscitation. Am J Surg. (in press).
13. American College of Surgeons, Committee on Trauma. Advanced Trauma Life Support Program for Doctors, 6th ed. Chicago: American College of Surgeons, 1997.
14. McKinley BA, Marvin RG, Cocanour CS, et al. Blunt trauma resuscitation: the old can respond. Arch Surg. 2000; 135: 688–693.
15. Garner GB, Ware DN, Cocanour CS, et al. Vacuum-assisted wound closure provides early fascial reapproximation in trauma patients with open abdomens. Am J Surg. 201;182:630–638.
16. Krone IL, Harman PK, Nolan SP. The measurement of intra-abdominal pressure as a criterion for exploration. Ann Surg. 1984; 199: 28–30.
17. Baker SP, O’Neill B, Haddon W Jr, et al. The injury severity score: a method for describing patients with multiple injuries and evaluating emergency care. J Trauma. 1974; 14: 187–196.
18. Osler T, Baker SP, Long W. A modification of the injury severity score that both improves accuracy and simplifies scoring. J Trauma. 1997; 43: 922–925.
19. Borlase BC, Moore EE, Moore FA. The abdominal trauma index - a critical reassessment and validation. J Trauma. 1990; 30: 1340–1344.
20. Moore FA, Sauaia A, Moore EE, et al. Postinjury multiple organ failure: a bimodal phenomenon. J Trauma. 1996; 40: 501–512.
21. Stone HH, Storm PR, Mullins RJ. Management of the major coagulopathy with onset during laparotomy. Ann Surg. 1983; 197: 532–535.
22. Burch JM, Ortiz VB, Richardson RJ, et al. Abbreviated laparotomy and planned reoperation for critically injured patients. Ann Surg. 1992; 215: 476–484.
23. Moore EE. Staged laparotomy for the hypothermia, acidosis and coagulopathy syndrome. Am J Surg. 1996; 172: 405–410.
24. Cullen DJ, Coyle JP, Teplick R, Long MC. Cardiovascular, pulmonary and renal effects of massively increased intra-abdominal pressure in critically ill patients. Crit Care Med. 1989; 17: 118–121.
25. Ertel W, Oberholzer A, Platz A, et al. Incidence and clinical pattern of the abdominal compartment syndrome after “damage control” laparotomy in 311 patients with severe abdominal and/or pelvic trauma. Crit Care Med. 2000; 28: 1747–1753.
26. Biffl WL, Moore EE, Burch JM, et al. Secondary abdominal compartment syndrome is a highly lethal event. Am J Surgery. 2001; 182: 645–648.
27. Sugrue M, Jones F, Lee A, et al. Intraabdominal pressure and gastric intramucosal pH: Is there an association? World J Surg. 1996; 20: 988–991.
28. Engum SA, Kogon B, Jensen E, et al. Gastric tonometry and direct intraabdominal pressure monitoring in abdominal compartment syndrome. J Pediatric Surg. 2002; 37: 214–218.
29. Sanchez NC, Tenofsky PL, Dort JM, et al. What is normal intra-abdominal pressure? Am Surg. 2001; 67: 243–248.
30. Sugrue M, Buist MD, Hourihan F, et al. Prospective study of intra-abdominal hypertension and renal function after laparotomy. Br J Surg. 1995; 82: 235–238.
31. Oda J, Ivatury RR, Blocher CR, et al. Amplified cytokine response and lung injury by sequential hemorrhagic shock and abdominal compartment syndrome in a laboratory model of ischemia-reperfusion. J Trauma. 2002; 52: 625–631.
32. Rezende-Neto J, Moore EE, Masuno T, et al. The abdominal compartment syndrome (ACS) as a second insult during systemic PMN priming provokes acute lung injury (ALI). Supplement to SHOCK. 2002; 17: 25.
33. Raeburn CD, Moore EE, Biffl WL, et al. The abdominal compartment syndrome is a morbid complication of postinjury damage control surgery. Am J Surg. 201;182:542–546.
34. Sugerman HJ, Bloomfield GL, Saggi BW. Multisystem organ failure secondary to increased intraabdominal pressure. Infection. 1999; 27: 61–66.
35. Holcomb JB, Pusateri AE, Hess JR, et al. Implications of new fibrin sealant technology for trauma surgery. Surg Clin North Am. 1997; 77: 943–952.
36. Yang EY, Marder SR, Hastings G, Knudson MM. The abdominal compartment syndrome complicating nonoperative management of major blunt liver injuries: recognition and treatment using multimodality therapy. J Trauma. 2002; 52: 982–986.
Keywords:

Intraabdominal hypertension; Abdominal compartment syndrome; Multiple organ failure; Gastric tonometry

© 2003 Lippincott Williams & Wilkins, Inc.