PEDIATRICS/ORIGINAL CONTRIBUTION

Identification of Children With Intra-Abdominal

Injuries After Blunt Trauma
From the Division of Emergency Medicine, Department of Internal Medicine,* Department of Radiology, Department of Pediatrics, and Department of Surgery,ll University of CaliforniaDavis School of Medicine, Sacramento, CA. Author contributions are provided at the end of this article. Received for publication June 28, 2001. Revision received December 27, 2001. Accepted for publication January 10, 2002. Presented in part at the Society for Academic Emergency Medicine annual meeting, St. Louis, MO, May 2002. Address for reprints: James F. Holmes, MD, University of CaliforniaDavis Medical Center, PSSB 2100, Division of Emergency Medicine, 2315 Stockton Boulevard, Sacramento, CA 95817-2282; 916-734-1539, fax 916-734-7950; E-mail jfholmes@ucdavis.edu. Copyright 2002 by the American College of Emergency Physicians. 0196-0644/2002/$35.00 + 0 47/1/122900 doi:10.1067/mem.2002.122900 James F. Holmes, MD* Peter E. Sokolove, MD* William E. Brant, MD Michael J. Palchak, MD* Cheryl W. Vance, MD* John T. Owings, MDll Nathan Kuppermann, MD, MPH*

See related article, p. 492, and editorial, p. 537. Study objective: We sought to determine the utility of laboratory testing after adjusting for physical examination findings in the identification of children with intra-abdominal injuries after blunt trauma. Methods: The study was a prospective observational series of children younger than 16 years old who sustained blunt trauma and were at risk for intra-abdominal injuries during a 21/2-year period at an urban Level I trauma center. Patients were examined by faculty emergency physicians and underwent standardized laboratory testing. Clinical and laboratory findings were recorded on a standardized data sheet. Intra-abdominal injury was considered present if an injury was documented to the spleen, liver, pancreas, kidney, adrenal glands, or gastrointestinal tract. We performed multiple logistic regression and binary recursive partitioning analyses to identify which physical examination findings and laboratory variables were independently associated with intra-abdominal injury. Results: Of 1,095 enrolled patients, 107 (10%, 95% confidence interval [CI] 8% to 12%) had intra-abdominal injuries. The mean age was 8.44.8 years. From both analyses, we identified 6 findings associated with intra-abdominal injury: low systolic blood pressure (adjusted odds ratio [OR] 4.1; 95% CI 1.1 to 15.2), abdominal tenderness (adjusted OR 5.8; 95% CI 3.2 to 10.4), femur fracture (adjusted OR 1.3; 95% CI 0.5 to 3.7), serum aspartate aminotransferase concentration more than 200 U/L or serum alanine aminotransferase concentration more than 125 U/L (adjusted OR 17.4; 95% CI 9.4 to 32.1), urinalysis with more than 5 RBCs per high-powered field (adjusted OR 4.8; 95% CI 2.7 to 8.4), and an initial hematocrit of less than 30% (adjusted OR 2.6; 95% CI 0.9 to 7.5). Conclusion: After adjusting for physical examination findings, laboratory testing contributes significantly to the identifi-

5 0 0

ANNALS OF EMERGENCY MEDICINE

39:5 MAY 2002

CHILDREN WITH INTRA-ABDOMINAL INJURIES AFTER TRAUMA Holmes et al

cation of children with intra-abdominal injuries after blunt trauma. [Holmes JF, Sokolove PE, Brant WE, Palchak MJ, Vance CW, Owings JT, Kuppermann N. Identification of children with intraabdominal injuries after blunt trauma. Ann Emerg Med. May 2002;39:500-509.]
INTRODUCTION

Trauma is the leading cause of death and disability in childhood, and missed abdominal injury is a leading cause of preventable morbidity and mortality.1-3 Pediatric patients suffering abdominal injury are frequently evaluated at trauma centers, but also may be seen in emergency departments and clinics without trauma center designation.4 Evaluation of the acutely injured child for intraabdominal injury may be difficult because of several factors. Although children with abdominal tenderness after blunt trauma are considered to be at high risk for intraabdominal injury,5-9 tenderness may not be detected because of decreased levels of consciousness or the inability of very young children to communicate verbally.5-11 Abdominal computed tomography (CT) is the diagnostic test of choice for identification of intra-abdominal injury in the hemodynamically stable, acutely injured child.5 This radiologic study, however, may expose the child to substantial amounts of ionizing radiation,12 requires transportation of the child out of the ED, may require pharmacologic sedation, and has associated expense. CT scanning should therefore be used selectively. Previous studies investigating the role of laboratory tests in the evaluation of pediatric trauma patients have been conflicting because some suggest laboratory testing is beneficial,8,13-15 whereas others suggest that it is unnecessary.8,16,17 These prior studies, however, did not adjust for physical examination findings when assessing the utility of laboratory tests. Therefore, the risk of intraabdominal injuries in patients with normal physical examinations but abnormal laboratory tests is unclear. The objective of this study was to determine the utility of laboratory testing after adjusting for physical examination findings in identifying children with intra-abdominal injuries after blunt trauma.
M AT E R I A L S A N D M E T H O D S

sented to a Level I trauma center between April 1996 and September 1998. These patients met trauma activation criteria at the participating center and/or were felt to be at risk for intra-abdominal injury on the basis of history or physical examination. Criteria for enrollment are listed in Figure 1. Patients transferred to the study facility after evaluation at an outside ED were excluded from the study, as were patients who presented in cardiopulmonary arrest. Patients underwent complete physical examinations by the on-duty faculty emergency physician before laboratory analysis or abdominal imaging procedures. Physical examination findings were documented on a structured data sheet after examination and before diagnostic testing. The level of consciousness was assessed using the Glasgow Coma Scale (GCS) for children 2 years and older and using the pediatric GCS for children younger than 2 years.18 A GCS score of 13 or less was considered to represent a significant decrease in the level of consciousness.19 We considered abdominal tenderness present if the child stated that palpation caused pain, if the patient grimaced on palpation, or if there was voluntary guarding. The chest examination was considered abnormal if erythema, abrasions, contusions, crepitus, tenderness, or abnormal auscultation were identified. Examination of the back was considered abnormal if abrasions, contusions, or tenderness were detected. Examination of the pelvis was considered abnormal if abrasions, tenderness, or instability were present.

Figure 1.

Study entry criteria (any one of the following). CRAMS, An outof-hospital scoring system evaluating the patients circulation, respiration, abdomen, motor responses, and speech.50 *Pediatric Trauma Score is a trauma scoring system based on the patients mental status, bony fractures, wounds, airway, and blood pressure.51
Blunt torso trauma from a significant mechanism of injury (eg, motor vehicle crash, automobile versus pedestrian, falls >10 ft) Decreased level of consciousness (GCS score of <15 or below ageappropriate behavior) in association with blunt torso trauma Blunt traumatic event with any of the following: Extremity paralysis Multiple bone fractures CRAMS score 8 Ejection from a motor vehicle Physical examination suggestive of intra-abdominal injury after blunt torso trauma Pediatric Trauma Score* 8 after blunt trauma

We conducted a prospective observational study of pediatric patients (<16 years) suffering from blunt trauma who were at risk for intra-abdominal injury and who pre-

MAY 2002

39:5

ANNALS OF EMERGENCY MEDICINE

5 0 1

CHILDREN WITH INTRA-ABDOMINAL INJURIES AFTER TRAUMA Holmes et al

Each patient underwent a standardized laboratory evaluation, which included a CBC count, alanine aminotransferase (ALT) and aspartate aminotransferase (AST) determinations, and urinalysis. Continuous laboratory variables were grouped into dichotomous categories based on prior studies and accepted clinical thresholds. An initial hematocrit less than 30% was considered abnormal for all patients regardless of age.5,17 An ALT more than 125 U/L6,8 and AST more than 200 U/L6 were considered thresholds defining abnormally increased values for these laboratory tests. A urinalysis containing more than 5 RBCs/high-powered field (hpf) was considered microscopic hematuria,8 and hematuria visible to the clinician was considered gross hematuria. We considered an intra-abdominal injury to be present if an injury was documented to the spleen, liver, pancreas, urinary tract, adrenal glands, or gastrointestinal tract during the patients ED or hospital courses. Injuries were identified by radiologic imaging, laparotomy, or at autopsy. A single faculty radiologist masked to all clinical information interpreted all abdominal CT scans. Patients were admitted to the hospital at the discretion of the trauma or pediatric surgery team in conjunction with the faculty emergency physician. Admitted patients were followed through their hospital courses to document diagnostic and therapeutic interventions. Blood transfusion and laparotomy were considered important therapeutic measures. Ethical considerations (radiation exposure,12 pharmacologic sedation, and expense) precluded us from obtaining abdominal CT scans on all patients. For this reason, patients discharged to home after ED evaluations were contacted by telephone one week after discharge. Patients without symptoms of intra-abdominal injury at the telephone follow-up were considered not to have clinically important intra-abdominal injuries.20 On completion of the study, we reviewed the trauma registry and continuing quality improvement records to evaluate whether any enrolled patients discharged from the ED or inpatient ward without the diagnosis of intra-abdominal injury later had such injuries diagnosed. We evaluated the association of 10 clinical variables with intra-abdominal injury to assess the independent contribution of laboratory testing to the detection of children with intra-abdominal injuries. In these analyses, we used multiple logistic regression and recursive partitioning methods. The variables we evaluated included initial ED physical examination findings and laboratory tests selected a priori because of the likelihood of association with intra-abdominal injury on the basis of prior re-

search.5-8,13-15,17 We limited the inclusion of predictor variables to 1 variable per 10 outcomes of interest (intraabdominal injury) in the multivariate analysis to ensure the power of the analysis.21,22 The following variables were included in the multivariate analyses: age, systolic blood pressure, abdominal tenderness, thoracic (chest or back) examination, pelvic examination, femur fracture, GCS score, urinalysis, initial hematocrit, and serum hepatic transaminase concentrations. Age was dichotomized as younger than 3 years or 3 to 16 years because of the difficulty of obtaining a reliable history and physical examination in preverbal children. The GCS score and laboratory variables were dichotomized as previously described. All 10 variables were entered into a multiple logistic regression analysis, and variables retaining a significant association (P<.05) with intra-abdominal injury were considered to be independent predictors of intraabdominal injury. We calculated the area under the receiver operating characteristic curve for the identified model and assessed the fit of the model using the Hosmer-Lemeshow goodness-of-fit test.23 To validate the logistic regression model, we performed 2 separate bootstrap procedures with 1,000 iterations each. Bootstrap resampling allows for validation of a model by randomly sampling from the database with replacement to create new databases of the same size as the original database. These databases are then used to test the model.24-26 This procedure allows for the generation of conservative estimates of confidence intervals (CIs), in addition to assessing the stability of the model. In the first bootstrap procedure, we obtained 95% biascorrected CIs of the investigated variables. In the second bootstrap procedure, we replicated the multivariate analysis on 1,000 bootstrap databases and identified the percentage of iterations of the analysis in which each predictive variable was identified as an independent predictor (P<.05) of intra-abdominal injury. An independent predictor variable was considered validated if it was selected in more than 50% of the bootstrap iterations.25 We additionally performed binary recursive partitioning with tenfold cross-validation on the same variables that were entered into the multiple logistic regression analysis. The purpose of this analysis was to explore the contribution of laboratory testing to the identification of children with intra-abdominal injuries using a different analytic technique. The use of recursive partitioning and its associated benefits as an analytic technique are described in the accompanying article regarding the predictors of thoracic injuries in children.27 Finally, we calculated the sensitivities, specificities, positive predictive

5 0 2

ANNALS OF EMERGENCY MEDICINE

39:5 MAY 2002

CHILDREN WITH INTRA-ABDOMINAL INJURIES AFTER TRAUMA Holmes et al

values, and negative predictive values for those variables retaining significance in these multivariate models. Data analysis was performed using Stata 6.0 for Windows (Stata Corporation, College Station, TX) statistical software. Answer Tree (version 2.0, SPSS Inc., Chicago, IL) statistical software was used to conduct binary recursive partitioning. The study was approved by the Human Subjects Research Committee at the participating institution.
R E S U LT S

A total of 1,095 patients were prospectively enrolled during the 28-month study period. The mean age was 8.44.8 years (range 11 days to 15.9 years) and the median Pediatric Trauma Score was 10 (range 3 to 12). One hundred ninety-five (18%) patients were younger than 3 years old. The mechanisms of injury were as follows: motor vehicle crash, 390 (36%); automobile versus pedestrian, 266 (24%); fall, 181 (17%); automobile versus bicycle, 97 (9%); fall off bicycle, 62 (6%); crush, 37 (3%); assault, 31 (3%); child abuse, 14 (1%); and other, 17 (1%). Abdominal CT scan, abdominal ultrasonography, diagnostic peritoneal lavage, laparotomy, or autopsy was performed in 664 (61%) patients. Studies obtained in these 664 patients included: abdominal CT scan in 565, abdominal ultrasonographic scan in 224, diagnostic peritoneal lavage in 23, laparotomy in 45, and autopsy in 2. One hundred ninety patients had multiple studies. Four hundred thirty-one (39%) patients did not undergo the previously mentioned methods of abdominal evaluation. One hundred two (24%) of these 431 patients were admitted to the hospital and evaluated for intra-abdominal

injury with serial physical examinations and hematocrit measurements for at least 24 hours. The remaining 329 (76%) patients were discharged to home from the ED. Telephone follow-up was obtained in 293 (89%) of these 329 patients and all were asymptomatic at the time of follow-up. There were no apparent cases of missed intraabdominal injuries among patients in this study on review of the trauma registry and continuing quality improvement records. A total of 107 (10%; 95% CI 8% to 12%) patients were determined to have intra-abdominal injuries. The mean age for patients with injuries was 7.94.8 years (range 26 days to 15.9 years) and 22 (21%) patients with injury were younger than 3 years. Injuries to the following organs were identified in the study population: liver (44), spleen (41), gastrointestinal tract (25), urinary tract (16), adrenal gland (10), and pancreas (5). Twenty-four patients (22%) had multiple injuries. Seventy-eight (73%; 95% CI 63% to 81%) of the patients with intra-abdominal injury also had intraperitoneal fluid identified by abdominal CT scan or at laparotomy. One hundred and three (96%; 95% CI 91% to 99%) of the 107 patients with intra-abdominal injuries had their injuries identified during ED evaluation or in the operating suite after direct transfer from the ED. Four patients with intra-abdominal injuries were considered to have a delay in diagnosis because their injuries

Table 2.

Multiple logistic regression model for intra-abdominal injury.

Frequency With Which Variable Was Identified Biasas an P Corrected Independent Value 95% CI* Predictor,* % 0.52.5 0.813.7 3.110.8 0.51.5 0.31.4 0.44.7 0.63.6 7.732.1 2.68.4 0.69.9 7 55 100 8 24 12 31 100 100 44

Table 1.

Physical examination findings.

IAI Present, n (%) (n=107) 11 (10) 33 (31) 62 (58) 51 (48) 16 (15) 11 (10) IAI Absent, n (%) (n=988) 15 (2) 122 (12) 291 (29) 356 (36) 113 (11) 46 (5) Difference, % (95% CI) 8 (3 to 15) 19 (10 to 27) 29 (19 to 38) 12 (2 to 12) 4 (4 to 11) 5 (0.2 to 12)

Variable

OR (95% CI)

Characteristic Low systolic blood pressure GCS score 13 Abdominal tenderness Abnormal thoracic examination Abnormal pelvis examination Femur fracture
IAI, Intra-abdominal injury.

Age <3 y 1.2 (0.52.5) .70 Low systolic blood pressure 4.1 (1.115.2) .03 Abdominal tenderness 5.8 (3.210.4) <.001 Abnormal thoracic 0.9 (0.51.5) .62 examination results Abnormal pelvic examination 0.6 (0.31.4) .23 Femur fracture 1.3 (0.53.7) .59 GCS score 13 1.7 (0.83.4) .17 ALT >125 U/L or 17.4 (9.432.1) <.001 AST >200 U/L Urinalysis >5 RBCs/hbf 4.8 (2.78.4) <.001 Initial hematocrit <30% 2.6 (0.97.5) .07
*

In bootstrap validation, as described in Materials and Methods section.

MAY 2002

39:5

ANNALS OF EMERGENCY MEDICINE

5 0 3

CHILDREN WITH INTRA-ABDOMINAL INJURIES AFTER TRAUMA Holmes et al

were identified after admission to the hospital from the ED. One of these patients had a liver injury, and the other 3 patients had gastrointestinal injuries. Physical examination findings in patients with and without intra-abdominal injuries are demonstrated in Table 1. Eight hundred and one (73%) patients had a GCS score of 15, and 139 (13%) had a GCS score of 14. Fortysix (73%; 95% CI 60% to 83%) of the 63 patients with a GCS score of 15 and intra-abdominal injury had abdominal tenderness on ED examination. Ten (91%; 95% CI 59% to 100%) of the 11 patients with a GCS score of 14 and intra-abdominal injury had abdominal tenderness. Of the 10 variables analyzed in the multiple logistic regression analysis, low systolic blood pressure, abdominal tenderness, serum ALT more than 125 U/L or AST more than 200 U/L, and hematuria more than 5 RBC/hpf were independently associated with intra-abdominal injury (Table 2). An initial hematocrit of less than 30% showed a trend (P=.07) toward association. The area under the model receiver operating characteristic curve was 0.89. The model demonstrated satisfactory goodness-of-fit, as measured by the Hosmer-Lemeshow test (P=.58). The bootstrap analysis validated the identifica-

tion of the 4 independent predictors because they were also identified as significant in more than 50% of the bootstrap iterations (Table 2). The results of the recursive partitioning analysis are demonstrated in Figure 2. In the resulting decision tree, the most important variable, increased liver transaminases, is the top node of the tree. As the tree branches, the risk of intra-abdominal injury can be further stratified for various combinations of predictor variables. This analysis identified abdominal tenderness, femur fracture, serum ALT of more than 125 U/L or AST of more than 200 U/L, an initial hematocrit of less than 30%, and hematuria of more than 5 RBC/hpf as important variables in the decision tree. The sensitivities, specificities, positive predictive values, and negative predictive values for the identified predictors of intra-abdominal injury from both the multiple logistic regression and recursive partitioning analyses are displayed in Table 3. Of the 107 patients with intra-abdominal injuries, 91 (85%; 95% CI 77% to 92%) had either physical examination findings (abdominal tenderness [62], low systolic blood pressure [11], or femur fracture [11]) suggestive of

Figure 2.

Findings of the recursive partitioning analysis. Each box presents the number of patients with and without intraabdominal injury, given the particular finding. IAI, Intra-abdominal injury.

No injury 988 (90%) IAI 107 (10%) Yes ALT >125 U/L or AST >200 U/L No

No injury 43 (46%) IAI 51 (54%) Yes

No injury 945 (94%) IAI 56 (6%) Urinalysis >5 RBCs/hpf No

No injury 100 (79%) IAI 27 (21%) Yes

No injury 845 (97%) IAI 29 (3%) Abdominal tenderness No

No injury 248 (92%) IAI 22 (8%) Yes

No injury 597 (99%) IAI 7 (1%) Hematocrit <30% No

No injury 16 (84%) IAI 3 (16%) Yes

No injury 581 (99%) IAI 4 (1%) Femur fracture No

No injury 24 (92%) IAI 2 (8%)

No injury 557 (99.6%) IAI 2 (0.4%)

5 0 4

ANNALS OF EMERGENCY MEDICINE

39:5 MAY 2002

CHILDREN WITH INTRA-ABDOMINAL INJURIES AFTER TRAUMA Holmes et al

intra-abdominal injury or a decreased level of consciousness (GCS score of 13) with unreliable physical examination of the abdomen (33). Forty-five (49%; 95% CI 39% to 60%) of these 91 patients underwent either laparotomy (39) and/or blood transfusion (18). One patient with a splenic injury underwent splenectomy after being discharged to home after his initial period of inpatient observation. Although the patient was stable during hospitalization, the patients hematocrit was subsequently found to have decreased precipitously as a result of further intra-abdominal hemorrhage. Seventyone (78%; 95% CI 68% to 86%) of these 91 patients had intraperitoneal fluid associated with their intra-abdominal injuries. Sixteen (17%) patients with intra-abdominal injuries had neither low systolic blood pressure, abdominal tenderness, nor femur fracture and had a GCS score of more than 13 (Table 4). Seven (44%; 95% CI 20% to 70%) of these 16 patients had intraperitoneal fluid associated with their intra-abdominal injuries. Two (13%; 95% CI 2% to 38%) of these 16 patients underwent specific therapy, including laparotomy (1) and blood transfusion (1). The 1 patient undergoing laparotomy presented with a GCS score of 14 and did not have abdominal tenderness. After the initial physical examination, this patient had a seizure and was subsequently intubated. To evaluate for intra-abdominal injury, a diagnostic peritoneal lavage was

Table 3.

Accuracy of the predictors* of intra-abdominal injury.

Sensitivity, Specificity, PPV, % (95% CI) % (95% CI) % (95% CI) 10 (518) 58 (4867) 10 (518) 14 (822) 50 (4060) 50 (4060) 98 (93100) 98 (9899) 71 (6873) 95 (9497) 98 (9698) 96 (9497) 89 (8791) 49 (4652) 42 (2362) 18 (1422) 19 (1032) 38 (2355) 54 (4465) 32 (2540) 17 (1420) NPV, % (95% CI) 91 (8993) 94 (9296) 91 (8992) 91 (8993) 95 (9396) 94 (9396) 99.6 (99100)

performed that was positive, and the patient was found during laparotomy to have a grade II splenic injury. Application of the 6 predictors of intra-abdominal injury identified by either the logistic regression or recursive partitioning analyses (ie, having any of the following variables: low systolic blood pressure, abdominal tenderness, femur fracture, ALT >125 U/L or AST >200 U/L, hematocrit <30%, hematuria >5 RBC/hpf) to the study population would have identified 105 (98%; 95% CI 93% to 100%) of the 107 patients with intra-abdominal injuries. The 2 patients not identified by the 6 predictors are highlighted in Table 4. A 9-year-old patient was identified with a possible renal injury after undergoing abdominal CT scanning for a decrease in hematocrit while in the ED. This patient was without abdominal, back, or flank tenderness, lacked hematuria on initial or subsequent urinalysis, and had an entirely normal renal ultrasonographic examination. The patients clinical course was uneventful. The second patient was a 12-year-old child who underwent abdominal CT scanning because of severe chest tenderness and fractured left lower ribs. He was determined to have a splenic injury that required no therapy other than observation. Using the predictive variables identified in this study, plus the GCS score, we created a diagnostic scheme for the evaluation of children with blunt torso trauma at risk for intra-abdominal injury. This diagnostic scheme was designed to conform to the usual sequence of evaluation in the ED (physical examination followed by laboratory analysis) (Figure 3). This figure also displays how accurately this algorithm would have stratified the patients in this study.
DISCUSSION

Clinical Finding Low systolic blood pressure Abdominal tenderness Femur fracture Initial hematocrit <30% ALT >125 and/or AST >200 (U/L) Urinalysis >5 RBCs/hpf Any of the above

IAI Present (N=107) IAI Absent (N=988) Presence of any 1 of the 6 predictors* Absence of all 6 predictors 105 2 482 506

PPV, Positive predictive value; NPV, negative predictive value; IAI, intra-abdominal injury. *Predictors from the logistic regression analysis or variables identified in the recursive partitioning analysis.

Identifying children with intra-abdominal injuries among those sustaining blunt torso trauma may be difficult. Missed intra-abdominal injury, however, is an important cause of morbidity in these children.2,3 In this study, 6 clinical variables were found to be predictors of intraabdominal injury: low systolic blood pressure, abdominal tenderness, femur fracture, increased serum hepatic transaminase concentrations, a low initial hematocrit, and hematuria. Using 2 different analytic techniques, we determined that laboratory tests provide important information for the identification of children with intraabdominal injuries after accounting for physical examination abnormalities. Abdominal tenderness was present in 58% of the children with intra-abdominal injuries. This figure may be

MAY 2002

39:5

ANNALS OF EMERGENCY MEDICINE

5 0 5

CHILDREN WITH INTRA-ABDOMINAL INJURIES AFTER TRAUMA Holmes et al

misleading, however, because children with decreased levels of consciousness may have unreliable abdominal examinations because of an altered ability to perceive pain.10,11 In children with a GCS score of more than 13, abdominal tenderness was present in 77% of patients with intra-abdominal injuries. Low systolic blood pressure was an independent predictor of intra-abdominal injury. Hypotension in injured children implies significant blood loss, and children with this finding should undergo rapid evaluation for all potential sources of hemorrhage, including intra-abdominal injuries. Although femur fracture was an important physical examination finding in the recursive partitioning analysis, it was not an independent predictor in the logistic regression model. In addition, femur fracture was the bottom node in the decision tree, implying that it is a less important discriminator than other variables. Its association with intra-abdominal injury may be that femur fracture is a marker for significant impact force. A GCS score of 13 or less was not found to be independently associated with the presence of intra-abdominal injury in either analyses. Patients with head injuries,

however, have an impaired ability to perceive abdominal pain.10,11 For this reason, children with decreased levels of consciousness routinely undergo abdominal evaluation with abdominal CT scanning. Our data suggest that a combination of physical examination and laboratory variables may identify those patients with decreased mental status who have intra-abdominal injuries. We would advise caution, however, in the abdominal evaluation of patients with decreased levels of consciousness and suggest a low threshold for abdominal CT scanning in these patients. Although children with low systolic blood pressure, abdominal tenderness, or femur fracture are likely to require further evaluation, the decision to perform abdominal imaging in pediatric trauma patients without these findings requires additional clinical information.5,6 The current study demonstrates that increases of serum hepatic transaminase concentrations is an important predictor of intra-abdominal injury after adjusting for physical examination findings. Increased levels of serum hepatic transaminase concentrations have previously been shown to be associated with liver injuries in pediatric trauma patients.6,13,28,29 Whether these tests conferred

Table 4.

Description of the 16 patients with intra-abdominal injuries who had neither abdominal tenderness, femur fracture, nor low systolic blood pressure, and had a GCS score of more than 13.
Hemoperitoneum No No No Yes No No Yes Yes Yes No No No Yes Yes Yes No Physical Examination Abnormalities None None None Back abrasion None Chest contusion None Abdominal tenderness developed in the ED Abrasions to pelvis None Flank/chest abrasion Chest/back tenderness/abrasion Chest tenderness/rib fractures Back tenderness Abdominal abrasion, GCS score of 14, seizure then intubated Chest/back tenderness, abdominal tenderness developed in the ED Laboratory Abnormalities Increased ALT/AST Increased ALT/AST, hematuria (40 RBCs/hpf) Increased ALT/AST Increased ALT/AST, hematuria (6 RBCs/hpf) Increased ALT/AST, hematocrit (29%) Increased ALT/AST Hematocrit=29% Hematocrit=28%, hematuria (6 RBCs/hpf) Increased ALT/AST Hematocrit decreased 8% points in the ED Increased ALT/AST Hematuria (314 RBCs/hpf) None Increased ALT/AST, hematuria (175 RBCs/hpf) Hematuria (6 RBCs/hpf) Increased ALT/AST Hospital Course Observation Observation Observation Observation Observation Observation Observation Blood transfusion Observation Observation Observation Observation Observation Observation Laparotomy Observation

Age, y 0.75 1.5 1.6 1.7 2.3 2.5 3.7 4.2 4.8 9 10 10.9 12.2 15 15 15

Mechanism MVC Fall 15 ft Auto versus ped Auto versus ped Auto versus ped MVC Fall Auto versus ped Auto versus ped MVC Auto versus bike Fall Fall MVC Auto versus ped Auto versus bike

Injury* Liver (II) Liver (I) Liver (I) Liver (I) Liver (I) Liver (I), adrenal Liver (I) Spleen (II) Liver (II) Kidney (I) Spleen (II) Kidney (II) Spleen (II) Liver (III) Spleen (II) Liver (I)

MVC, Motor vehicle crash; Auto versus ped, automobile versus pedestrian; IV, intravenous. * Injury grade is provided in parentheses after each injury.52-54 Hemoperitoneum identified by abdominal CT scan or at laparotomy. Patient did not have any of the 6 predictors of intra-abdominal injury.

5 0 6

ANNALS OF EMERGENCY MEDICINE

39:5 MAY 2002

CHILDREN WITH INTRA-ABDOMINAL INJURIES AFTER TRAUMA Holmes et al

better or additional diagnostic information for the diagnosis of intra-abdominal injury beyond that gathered from the physical examination, however, was previously unclear. In addition, an initial hematocrit lower than 30% was found to be an important predictor of intra-abdominal injury in the current study. Although it is logical that unexpected anemia would be associated with intraabdominal injury, we are aware of only one prior study that has demonstrated the marginal utility of the hematocrit above and beyond the physical examination.5 Hematuria was also associated with intra-abdominal injury in the current study. Gross hematuria has previously been shown to be associated with the presence of intra-abdominal injury in pediatric patients suffering from blunt trauma,30 and more than half of the children in the current study with gross hematuria had intraabdominal injuries (data not shown). The utility of microscopic hematuria in evaluating pediatric blunt

Figure 3.

Suggested algorithm for evaluation of children with blunt torso trauma. IAI, Intra-abdominal injury.

Blunt torso trauma

107 (9.8%) of 1,095 had IAI

Abdominal tenderness Yes or Abdominal CT scan Low systolic blood pressure 68 (18%) of 372 had IAI No ALT >125 or AST >200 U/L Yes or Abdominal CT scan Hematocrit <30% 28 (40%) of 70 had IAI No GCS score 13, Femur fracture, or Urinalysis >5 RBCs/hpf No Appropriate observation and 2 (0.4%) of 484 instructions had IAI Yes 9 (5%) of 169 had IAI Consider abdominal CT scan

Low likelihood for IAI

trauma patients, however, is controversial.6,8,31 Prior studies have suggested thresholds for microscopic hematuria predictive of intra-abdominal injury to be more than 5 RBC/hpf,8 more than 20 RBC/hpf,15 and more than 50 RBC/hpf.32 We chose more than 5 RBC/hpf as a conservative value and found this to be an independent predictor of intra-abdominal injury in both analyses. Although the presence of microscopic hematuria at the lower end of the spectrum may not mandate further abdominal evaluation in all patients, this finding should heighten the clinicians suspicion of intra-abdominal injury. The 6 predictors of intra-abdominal injury identified by these 2 analyses should serve as a useful guide when evaluating children sustaining blunt torso trauma. These predictors make clinical sense and are easy to remember. Although these predictors are not intended to supplant clinical judgment, abdominal imaging should be strongly considered for most children with these findings. Only 2 of the 107 children with intra-abdominal injuries in this study were not identified by at least 1 of these independent predictors, and 1 of these patients was likely a falsepositive imaging study. Neither of the 2 children required specific therapy other than observation. The patient with a splenic injury not identified by the 6 independent predictors had an overt injury to the overlying chest. The presence of thoracic injury has been previously suggested as an indication for abdominal CT scanning in adults.33,34 In the present study, an abnormal thoracic examination was not identified as a statistically significant predictor of intra-abdominal injury in either multivariate analysis. Direct injury to the area of the chest overlying the spleen, however, should raise the clinicians concern of an injury to this organ. The role of laboratory tests in the evaluation of pediatric trauma patients has been controversial. The laboratory tests identified as important in this study, however, are widely available and inexpensive. The importance of these selected laboratory tests in risk stratification of pediatric trauma patients for identifying children with intra-abdominal injuries is evident in both analyses. In the recursive partitioning analysis, increased liver transaminases and microscopic hematuria were the top 2 nodes of the decision tree, offering the best initial risk stratification of patients. Injuries identified by these laboratory tests may have important clinical implications. In the current study, 1 of the 16 patients with an intra-abdominal injury but without physical examination findings (low systolic blood pressure, abdominal tenderness, femur fracture, or decreased mental status) suggestive of injury underwent

MAY 2002

39:5

ANNALS OF EMERGENCY MEDICINE

5 0 7

CHILDREN WITH INTRA-ABDOMINAL INJURIES AFTER TRAUMA Holmes et al

laparotomy and another received a blood transfusion. Although laparotomy and transfusion requirements during hospitalization are important outcome variables in the study of intra-abdominal injuries, these therapeutic interventions are neither common nor well standardized. For children with solid organ injuries, the rates of laparotomy and blood transfusion are currently estimated to be less than 15% and less than 25%, respectively.35-38 Even in the absence of the need for these specific therapeutic interventions, however, we believe that identification of intra-abdominal injuries in children is important. Injuries that do not require intervention during initial hospitalization occasionally worsen and may subsequently require specific therapy after hospital discharge, as occurred in one patient in this study.39,40 In addition, the need for specific therapy is not routinely known at the time of identification of the injury. Furthermore, children with intra-abdominal injuries frequently require activity restriction and additional abdominal imaging to ensure the resolution of the injuries.41-43 Therefore, every reasonable effort should be made to detect all intra-abdominal injuries in children sustaining blunt torso trauma. There are certain limitations to this study. Thirty-nine percent of the patients did not have radiologic evaluation of the abdomen because ethical considerations prevented extensive evaluation of all patients. Patients with abnormal physical examination findings and abnormal laboratory tests were more likely to undergo further diagnostic abdominal evaluation than patients without these abnormalities, thus possibly leading to evaluation bias. Patients who did not have definitive abdominal imaging, however, still had clinical follow-up to identify potentially important missed injuries. Although clinical follow-up is an acceptable end point in both trauma44 and nontrauma20 studies when more definitive testing of all patients is not considered ethical, some patients with subclinical injuries may not have been identified. We believe this potential limitation, however, had little effect on the results or implications of the study. We did not assess interrater reliability of the criteria because of limited availability of a second faculty physician in the pediatric ED. Although the findings of the analyses were internally validated, external validation is necessary to further assess the generalizability of our findings. Finally, we did not study screening abdominal ultrasonographic examination in the evaluation of all these patients. The use of abdominal ultrasonography in pediatric trauma patients is controversial45-48 and is currently used by less than 15% of pediatric EDs.49 Therefore, rec-

ommendations for ultrasonography use would not be generalizable to all settings in which injured children are evaluated. In conclusion, after adjusting for physical examination findings, laboratory testing significantly contributes to the identification of children with intra-abdominal injuries. Low systolic blood pressure, abdominal tenderness, femur fracture, AST more than 200 U/L or ALT more than 125 U/L, an initial hematocrit less than 30%, and hematuria more than 5 RBCs/hpf are important predictors of intra-abdominal injury in children sustaining blunt torso trauma. Pediatric blunt trauma patients with any one of these findings should be considered at significant risk for intra-abdominal injury. Patients with none of these risk factors are at low risk for intra-abdominal injury.
Author contributions: JFH and PES conceived the study. JFH, PES, JTO, and NK designed the study. JFH, WEB, MJP, CWV, PES, and NK participated in data collection. JFH and NK participated in data analysis. JFH and NK participated in manuscript preparation and all authors in manuscript revision. All authors take responsibility for the paper as a whole. We thank Nicole Glaser, MD, for her critical review and helpful comments on this article.

REFERENCES
1. Bivins BA, Sacahatello CR. Diagnostic exploratory laparotomy: an outdated concept in blunt abdominal trauma. South Med J. 1979;72:969-970. 2. Muckart DJ, Thomson SR. Undetected injuries: a preventable cause of increased morbidity and mortality. Am J Surg. 1991;162:457-460. 3. Furnival RA, Woodward GA, Schunk JE. Delayed diagnosis of injury in pediatric trauma. Pediatrics. 1996;98:56-62. 4. Nelson DS, Walsh K, Fleisher GR. Spectrum and frequency of pediatric illness presenting to a general community hospital ED. Pediatrics. 1992;90:5-10. 5. Taylor GA, ODonell R, Sivit CJ, et al. Abdominal injury score: a clinical score for the assignment of risk in children after blunt trauma. Radiology. 1994;190:689-694. 6. Holmes JF, Sokolove PE, Land C, et al. Identification of intra-abdominal injuries in children hospitalized following blunt torso trauma. Acad Emerg Med. 1999;6:799-806. 7. Saladino R, Lund D, Fleisher G. The spectrum of liver and spleen injuries in children: failure of the pediatric trauma score and clinical signs to predict isolated injuries. Ann Emerg Med. 1991;20:636-640. 8. Isaacman DJ, Scarfone RJ, Kost SI, et al. Utility of routine laboratory testing for detecting intra-abdominal injury in the pediatric trauma patient. Pediatrics. 1993;92:691-694. 9. Meyer DM, Thal ER, Coln D, et al. Computed tomography in the evaluation of children with blunt abdominal trauma. Ann Surg. 1993;217:272-276. 10. Beaver BL, Colombani PM, Fal A, et al. The efficacy of computed tomography in evaluating abdominal injuries in children with major head trauma. J Pediatr Surg. 1987;22:1117-1122. 11. DuPriest RW Jr, Rodriguez A, Shatney CH. Peritoneal lavage in children and adolescents with blunt abdominal trauma. Am Surg. 1982;48:460-462. 12. Paterson A, Frush DP, Donnelly LF. Helical CT of the body: are settings adjusted for pediatric patients? AJR Am J Roentgenol. 2001;176:297-301. 13. Oldham KT, Guice KS, Kaufman RA. Blunt hepatic injury and elevated hepatic enzymes: a clinical correlation in children. J Pediatr Surg. 1984;19:457-461.

5 0 8

ANNALS OF EMERGENCY MEDICINE

39:5 MAY 2002

CHILDREN WITH INTRA-ABDOMINAL INJURIES AFTER TRAUMA Holmes et al

14. Hashmi A, Klassen T. Correlation between urinalysis and intravenous pyelography in pediatric abdominal trauma. J Emerg Med. 1995;13:255-258. 15. Lieu TA, Fleisher GR, Mahboubi S, et al. Hematuria and clinical findings as indications of intravenous pyelography in pediatric blunt renal trauma. Pediatrics. 1988;82:216-222. 16. Ford EG, Karamanoukian HL, McGrath N, et al. Emergency center laboratory evaluation of pediatric trauma victims. Am Surg. 1990;56:752-757. 17. Bryant MS, Tepas JJ, Talber JL, et al. Impact of emergency room laboratory studies on the ultimate triage and disposition of the injured child. Am Surg. 1988;54:209-211. 18. Reilly PL, Simpson DA, Sprod R, et al. Assessing the conscious level in infants and young children: a pediatric version of the Glasgow Coma Scale. Child Nerv Syst. 1988;4:30-33. 19. Livingston DH, Lavery RF, Passannante MR. Admission or observation is not necessary after a negative abdominal computed tomographic scan in patients with suspected blunt abdominal trauma: results of a prospective, multi-institutional trial. J Trauma. 1998;44:273-282. 20. Jaeschke R, Guyatt G, Sackett DL. Users guides to the medical literature. III. How to use an article about a diagnostic test. A. Are the results of the study valid? JAMA. 1994;271:389391. 21. Harrell FE, Lee KL, Mark DB. Multivariate prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med. 1996;15:361-387. 22. Concato J, Feinstein AR, Holford TR. The risk of determining risk with multivariate models. Ann Int Med. 1993;118:201-210. 23. Hosmer DW, Lemeshow S. Applied Logistic Regression. New York, NY: John Wiley & Sons Inc.; 1989. 24. Altman DG, Andersen PK. Bootstrap investigation of the stability of a Cox regression model. Stat Med. 1989;8:771-783. 25. Chen CH, George SL. The bootstrap and identification of prognostic factors via Coxs proportional hazards regression model. Stat Med. 1985;4:39-46. 26. Efron B, Tibshirani R. Statistical data analysis in the computer age. Science. 1991;253:390395. 27. Holmes JF, Sokolove PE, Brant WE, et al. A clinical decision rule for identifying children with thoracic injuries after blunt torso trauma. Ann Emerg Med. 2002;39:492-499. 28. Coant PN, Kornberg AE, Brody AS, et al. Markers for occult liver injury in cases of physical abuse in children. Pediatrics. 1992;89:274-288. 29. Hennes HM, Smith DS, Schneider K, et al. Elevated liver transaminase levels in children with blunt abdominal trauma: a predictor of liver injury. Pediatrics. 1990;86:87-90. 30. Stein JP, Kaji DM, Eastham J. Blunt renal trauma in the pediatric population: indications for radiographic evaluation. Urology. 1994;44:406-410. 31. Taylor GA, Eichelberger MR, Potter BM. Hematuria: a marker of abdominal injury in children after blunt trauma. Ann Surg. 1988;208:688-693. 32. Stalker HP, Kaufman PR, Sledge K. The significance of hematuria in children after blunt trauma. AJR Am J Roentgenol. 1990;154:569-571. 33. Mackersie RC, Tiwary AD, Shackford SR, et al. Intra-abdominal injury following blunt trauma. Arch Surg. 1989;124:809-813. 34. Grieshop NA, Jacobson LE, Gomez GA, et al. Selective use of computed tomography and diagnostic peritoneal lavage in blunt abdominal trauma. J Trauma. 1995;38:727-731. 35. Coburn MC, Pfeifer J, DeLuca FG. Nonoperative management of splenic and hepatic trauma in the multiply injured pediatric and adolescent patient. Arch Surg. 1995;130:332-338. 36. Bond SJ, Eichelberger MR, Gotschall CS, et al. Nonoperative management of blunt hepatic and splenic injuries in children. Ann Surg. 1996;223:286-289. 37. Shaft S, Gilbert JC, Carden S, et al. Risk of hemorrhage and appropriate use of blood transfusion in pediatric blunt splenic injuries. J Trauma. 1997;42:1029-1032. 38. Leone RJ Jr, Hammond JS. Nonoperative management of pediatric blunt hepatic trauma. Am Surg. 2001;67:138-142. 39. Brown RL, Irish MS, McCabe AJ. Observation of splenic trauma: when is a little too much? J Pediatr Surg. 1999;34:1124-1126.

40. Frumiento C, Sartorelli K, Vane D. Complications of splenic injuries: expansion of the nonoperative theorem. J Pediatr Surg. 2000;35:788-791. 41. Fallat M, Casale AJ. Practice patterns of pediatric surgeons caring for stable patients with traumatic solid organ injury. J Trauma. 1997;43:820-824. 42. Gandhi RR, Keller MS, Schwab CW, et al. Pediatric splenic injury: pathway to play? J Pediatr Surg. 1999;34:55-58. 43. Emery KH, Babcock DS, Borgman AS, et al. Splenic injury diagnosed with CT: US follow-up and healing rate in children and adolescents. Radiology. 1999;212:515-518. 44. Stiell IG, Wells GA, Bandemheen K, et al. The Canadian CT head rule for patients with minor head injury. Lancet. 2001;357:1391-1396. 45. Patel JC, Tepas JJ. The efficacy of focused abdominal sonography for trauma (FAST) as a screening tool in the assessment of injured children. J Pediatr Surg. 1999;34:44-47. 46. Mutabagani KH, Coley BD, Zumberge N, et al. Preliminary experience with focused abdominal sonography for trauma (FAST) in children: is it useful? J Pediatr Surg. 1999;34:48-54. 47. Benya EC, Lim-Dunham JE, Landrum O, et al. Abdominal sonography in examination of children with blunt abdominal trauma. AJR Am J Roentgenol. 2000;174:1613-1616. 48. Coley BD, Mutabagani KH, Martin LC, et al. Focused abdominal sonography for trauma (FAST) in children with blunt trauma. J Trauma. 2000;48:902-906. 49. Baka AG, Delgado CA, Simon HK. Current use and perceived utility of ultrasound for the evaluation of pediatric compared to adult trauma patients. Pediatr Emerg Care. In press. 50. Gormican SP. CRAMS scale: field triage of trauma victims. Ann Emerg Med. 1982;11:132135. 51. Tepas JJ, Mollitt DL, Talbert JL, et al. The pediatric trauma score as a predictor of injury severity in the injured child. J Pediatr Surg. 1987;22:14-18. 52. Mirvis SE, Whitley NO, Vainwright JR, et al. Blunt hepatic trauma in adults: CT-based classification and correlation with prognosis and treatment. Radiology. 1989;171:27-32. 53. Mirvis SE, Whitley NO, Gens DR. Blunt splenic trauma in adults: CT-based classification and correlation with prognosis and treatment. Radiology. 1989;171:33-39. 54. Moore EE, Shackford SR, Pachter HL, et al. Organ injury scaling: spleen, liver, and kidney. J Trauma. 1989;29:1664-1666.

MAY 2002

39:5

ANNALS OF EMERGENCY MEDICINE

5 0 9