| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 All authors: Department of Radiology, University of Pittsburgh Medical Center-Presbyterian Hospital, 200 Lothrop St., Pittsburgh, PA 15213.
Received April 3, 2000;
accepted after revision June 27, 2000.
Presented at the annual meeting of the American Roentgen Ray Society, San
Francisco, April-May 1998.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Abdominal CT scans of 112 patients with blunt abdominal trauma were prospectively and retrospectively reviewed. Fifty patients had proven bowel injuries (with or without other visceral injuries), whereas 62 patients had no bowel injury and comprised the comparison or control group. Thirty-one of the 62 patients in the comparison group had surgical proof of abdominal but not bowel or mesenteric injuries. The retrospective review of the 112 CT scans was performed randomly and individually by nine radiologists unaware of the diagnosis, including three faculty abdominal radiologists, three senior residents in training, and three junior residents in training. Individual performance and group performance were evaluated by receiver operating characteristic analysis, and interobserver agreement was tested. Individual CT signs as relevant predictors of bowel injury were identified by logistic regression.
RESULTS. Relevant predictors of bowel injury included mesenteric infiltration, bowel wall thickening, extravasation of vascular or enteric contrast agent, and the presence free air. In the retrospective blinded review, CT showed good to excellent interobserver reliability for individual CT signs as well as for diagnosis of bowel and visceral injuries. Faculty radiologists tended to diagnose injuries with greater accuracy and confidence, but they showed significantly better performance than residents only in diagnosing duodenal perforation. For the prospective CT diagnosis of bowel injury, CT had a sensitivity of 64%, an accuracy of 82%, and a specificity of 97%.
CONCLUSION. Bowel injuries are challenging to diagnose on CT. Radiologists with various levels of experience and expertise can achieve accurate and reproducible results using a variety of CT criteria.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CT is regarded as highly accurate in depicting injuries to solid abdominal organs. However, the accuracy of CT in revealing bowel and mesenteric injuries for diagnosis remains controversial, with some investigators reporting almost no accurate diagnoses with CT [11, 12] and with others regarding CT as accurate and useful [13,14,15,16,17,18,19,20,21]. For the past 10 years, the CT signs of bowel and mesenteric injury have been described in various retrospective reviews and have been subjected to some limited analyses of diagnostic performance [18,19,20].
Among the possible explanations for the widely divergent reports of the performance of CT in this setting are variations in CT equipment or techniques, levels of training or expertise among the CT interpreters, and the utility of individual or combinations of CT signs of bowel or mesenteric injury. The goals of our study were to determine the relevant and reliable CT signs of bowel injury, to determine the overall performance of CT in detection of bowel and mesenteric injury, to establish the interobserver reliability of CT, and to establish the effect of the training level of radiologists on this performance.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The 112 patients included 73 males and 39 females ranging in age from 14 to 86 years (mean, 38 years; median, 33 years). The 50 patients with proven bowel or mesenteric injuries had confirmation by laparotomy (n = 48) or barium study of the upper gastrointestinal tract (2 patients with duodenal hematoma).
The control or comparison group comprised 31 patients with findings at laparotomy that showed various abdominal visceral injuries and bleeding but that excluded bowel or mesenteric injury and 31 patients who did not undergo surgery but who had no abdominal complaints or physical findings at least 2 weeks after discharge from the hospital. Although individual cases included in the comparison group were selected randomly, we did make a deliberate effort to include CT studies with negative (normal) posttrauma findings and a representative group of CT studies that showed various solid visceral injuries to simulate closely the clinical challenge of CT interpretation in cases of trauma. We did not attempt to replicate the distribution of normal versus specific visceral injures. Because bowel injury is relatively uncommon, the blinded observers would have been required to interpret and score hundreds of control CT scans, which was thought to be unreasonable. Among the 50 patients with proven bowel or mesenteric injuries, 22 had no additional abdominal visceral injuries. The other 28 patients had bowel and other injuries, including injuries to the spleen (n = 17), liver (n = 9), kidney (n = 3), pancreas (n = 2), and urinary bladder (n = 2). The group without bowel injury who had surgical exploration included six patients in whom no abdominal injuries were found and 25 who had an isolated visceral injury or combination of visceral injuries, including injuries to the spleen (n = 18), liver (n = 6), kidney (n = 2), and pancreas (n = 2). The group of patients treated nonoperatively included 20 patients with no abdominal injury and an isolated injury or combined injuries including injuries to the liver (n = 7), spleen (n = 4), kidney (n = 1), and adrenal gland (n = 1).
All patients underwent abdominopelvic CT while being evaluated in the emergency department. Both IV and oral—enteric contrast agents were used routinely. Approximately 95% of all studies were performed with both IV and oral contrast media; no study was excluded for technical reasons or because IV or oral contrast material had not been used. The IV contrast agent was either iothalamate meglumine or ioversol (Conray 60 or Optiray 320, respectively; Mallinckrodt Medical, St. Louis, MO) administered at a volume of 2 mL/kg to a maximum of 150 mL. The IV contrast material was administered by power injection (Medrad, Pittsburgh, PA) at a rate of 2.5 mL/sec. The oral-enteric contrast agent was diatrizoate meglumine and sodium (Gastroview; Mallinckrodt Medical) diluted with tap water to a ratio of 1:40. Alert patients drank 450 mL of this 2.5% solution, whereas patients with altered mental status or inability to control their airway had the same volume of this solution administered via a nasogastric tube. The enteric contrast material was given as soon as the decision was made to perform CT evaluation and was administered, usually approximately 30 min before CT, by a nurse in the emergency department. Approximately half of both the study and control patients were evaluated on conventional (nonhelical) CT scanners, and half underwent helical CT (HiLight Advantage and HiSpeed Advantage, respectively; General Electric Medical Systems, Milwaukee WI). For all patients, 7-mm-thick sections were obtained from above the diaphragm to the symphysis pubis. For cooperative patients, clusters of nonhelical scans were obtained during breath-holding, and helical scans of the upper abdomen were obtained during a single breath-hold; the pelvis was scanned during a second series (pitch of 1.5).
All 112 CT examinations were retrospectively, randomly, and individually reviewed by nine observers who were unaware of all clinical information except the history of blunt abdominal trauma. The observers knew that we were investigating the CT diagnosis of bowel trauma, but they had no information about the number of bowel trauma cases included among the 112 studies. The CT observers consisted of three faculty abdominal radiologists, three senior radiology residents (third or fourth year of training), and three junior residents (second-year trainees with at least 2 months of a dedicated abdominal CT rotation). All residents had experience monitoring CT scans of trauma patients while on-call in the hospital for evening and night duty.
The observers were given no special training but were made aware of existing published reports on CT findings that were thought to be indicative of bowel or mesenteric injury. On individual data-entry sheets, the observers were asked to report their confidence in diagnosing or excluding bowel or mesenteric injury on a scale from 0 (definitely no injury) to 5 (definite injury). On a similar scale, the observers evaluated specific CT findings including bowel wall thickening, abnormal bowel wall enhancement, mesenteric infiltration, extravasation of enteric or vascular contrast medium, extraluminal air (gas), ileus, and free or localized intraperitoneal blood. Observers also recorded diagnoses of injuries to other abdominal viscera (liver, spleen, pancreas, or kidney).
Individual performance and group performance were evaluated by standard receiver operator characteristic analysis (MRMC [multiple readers multiple conditions] type) [22]. Interobserver agreement (reliability) was assessed by item analysis with Cronbach's alpha, where an alpha value of 0.8 or greater is considered reliable [23]. Individual CT signs as relevant predictors of bowel injury were identified by logistic regression with the generation of regression coefficients.
We also reviewed the official CT interpretations rendered at the time of the study in the emergency department or the radiology department. Each CT scan had been reviewed by a resident in training and one of eight faculty abdominal radiologists.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
= 0.82-0.94)
except for ileus ( = 0.71).
|
Logistic regression analysis suggests that mesenteric infiltration, bowel wall thickening, and free intraperitoneal air are relevant predictors of bowel injury (Appendix and Fig. 1A,1B,1C). Mesenteric infiltration was seen not only in 68% of the patients with bowel injury, but also in 23% of the patients with other abdominal injuries. Bowel wall thickening was seen in 60% of the cases of bowel injuries and was seen in only 8% of the cases of other injuries. Extravasation of enteric contrast agent and extraluminal air (gas) were seen only in cases of transmural bowel laceration (14% and 32% of bowel trauma cases, respectively), except for several cases of free air being present after diagnostic peritoneal lavage. Extravasation of mesenteric vascular contrast medium was also seen only in proven bowel or mesenteric injuries and was seen in eight cases (16%) (Figs. 1A,1B,1C,2A,2B,3A,3B). Free peritoneal fluid, abnormal bowel wall enhancement, and ileus were less effective predictors. It should be noted that the receiver operator characteristic methodology was applied to a continuous data set. To report a CT sign as "present" or "absent" is to choose a binary analysis dividing the observers' responses as a sign of being "present" (toward the right side of the confidence scale) or "absent" (toward the left side).
|
|
|
|
|
|
|
Logistic regression also suggests that faculty-level observers use CT signs of associated retroperitoneal injuries, such as pancreatic or renal lacerations, to improve the sensitivity of CT for the diagnosis of bowel injury.
The performance of CT for detection of bowel injury was evaluated with receiver operator characteristic analysis (Table 2). CT was highly accurate in the diagnosis of duodenal and small-bowel perforation (transmural lacerations). As a group, faculty radiologists tended to achieve greater accuracy than the residents. The junior and senior residents had comparable performance, although individual junior residents achieved higher accuracy than individual faculty. The only statistically significant difference among the groups was that faculty radiologists diagnosed duodenal perforation more accurately than either group of residents (Fig. 4). Performance of all three groups of radiologists was lower for the detection of stomach and colon injuries (area under the receiver operator characteristic curve [Az] < 0.70) and for the detection of hematomas of the jejunum and ileum (Az < 0.70).
|
|
Review of the CT interpretations rendered at the time of the original examination revealed the following data among all 112 cases. The number of true-positive cases was 32 (for diagnosis of bowel or mesenteric injury); false-negative cases, 18; false-positive cases, two; and true-negative cases, 60. This resulted in a sensitivity of 64%, a specificity of 97%, and an overall accuracy of 82% for the prospective interpretations.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The CT equipment and techniques used in our study are similar to those used in other investigations reported over the past 15 years. Approximately half of our studies were acquired on conventional axial and half on helical CT equipment. Although we noted no significant difference in our results related to helical versus nonhelical scanners, the ability to acquire helical CT sections through the entire abdomen and pelvis during one or two breath-holds and during peak vascular opacification likely yields optimal results. The ability to eliminate motion artifacts and to depict active extravasation of vascular contrast medium are particular advantages of helical CT.
Like radiologists at many other trauma centers, we have always used oral contrast medium in the performance of abdominal CT in trauma patients. Although others have claimed that this practice is unnecessary and potentially dangerous [11], we believe these claims have little validity. In multiple studies specifically seeking evidence of aspiration pneumonitis related to the use of oral contrast medium in the trauma setting, no documented adverse events were found in more than 1000 adult and 50 pediatric patients [24,25,26]. Although we identified extravasation of oral contrast medium in only seven (14%) of 50 patients, it is a highly specific sign of transmural bowel laceration. Moreover, we believe that opacification of the bowel makes mural and interloop hematomas more easily recognized (Fig. 1A,1B,1C).
Extravasation of mesenteric vascular contrast medium is also a relatively uncommon but specific sign of significant mesenteric or bowel injury. We detected this sign of active hemorrhage on CT examinations in 16% (8/50) of the patients with bowel or mesenteric injuries and in none of the trauma patients with other injuries (Figs. 1A,1B,1C,2A,2B,3A,3B). Dowe et al. [18] reported similar observations in their investigation of 27 patients with surgically proven mesenteric injuries, 24 of whom (89%) had mesenteric vascular contrast extravasation, a mesenteric hematoma or infiltration associated with bowel wall thickening, or both. CT diagnosis of active mesenteric hemorrhage requires rapid IV injection of a contrast bolus (>2.5 mL/sec) to achieve sustained adequate enhancement of blood vessels. Active mesenteric hemorrhage makes surgical intervention necessary because of the potential for vascular compromise of the bowel, the danger of continued bleeding, and the likelihood of acute bowel wall injury [18, 27].
Variation in practice and techniques also influences another specific sign of bowel injury, the presence of extraluminal gas. Our observers detected extraluminal gas in 32% of the patients with bowel injuries (Fig. 2A,2B). With rare exceptions, this sign in the setting of blunt abdominal trauma mandates surgery. Although we had several cases without bowel injury in which observers detected or suspected free air, all were cases preceded by diagnostic peritoneal lavage, which is known to result in free fluid and air in the peritoneal cavity. Fortunately, the use of diagnostic peritoneal lavage is diminishing, and this procedure has virtually no role in the hemodynamically stable patient for whom CT evaluation is being considered. Although Hamilton et al. [28] reported several patients with blunt abdominal trauma with free abdominal air in the absence of bowel injury, in our judgment many of their cases represented misinterpretation of pneumothorax for pneumoperitoneum.
No individual CT sign can be considered both sensitive and specific for bowel or mesenteric injury. Even before testing the predictive value of these signs, we wanted to establish whether radiologists with various levels of experience could achieve reproducible results. As shown in Table 1, interobserver reliability was good to excellent for numerous individual CT signs of bowel injury and for the diagnosis of bowel or solid-organ injury. Among eight individual CT criteria for bowel injury, only the presence of ileus failed to meet test standards of reliability.
We were encouraged and somewhat surprised that radiology residents in training achieved diagnostic accuracy almost equaling that of their faculty supervisors. Some of the residents had only 2 years of training and 2 months of dedicated body CT rotations. Of course, proper instruction in trauma CT interpretation is essential, and all residents in our radiology training program receive didactic and written instructional materials directed toward this goal. The residents who participated in this investigation received no special training or coaching. Faculty-level abdominal radiologists tended to diagnose bowel and mesenteric injuries with greater accuracy and confidence than either group of residents. However, only for duodenal perforation did the faculty record significantly improved accuracy.
If pathognomonic signs of bowel trauma are detected in a minority of cases, how do radiologists make accurate diagnoses? Clearly, radiologists rely on a combination of CT findings, each having variable sensitivity and specificity. Faculty radiologists even use clues such as associated pancreatic or renal lacerations to recognize nearby bowel injuries.
Mesenteric infiltration and free intraperitoneal fluid are the most frequent CT signs associated with bowel trauma [13,14,15,16,17,18,19,20,21]. Free fluid without an obvious source (e.g., liver or spleen laceration) has long been recognized as an important clue to bowel or mesenteric injury [13,14,15,16,17,18,19,20,21, 29,30,31,32,33]. Experienced observers have learned to disregard small collections of isolated pelvic fluid in women of reproductive age, but any other fluid collection in the setting of trauma warrants close scrutiny [29,30,31,32,33]. Intraperitoneal fluid collections that are especially predictive of bowel or mesenteric injury are those in the mesentery or bowel wall itself. The sentinel clot sign describes relatively highly attenuating and heterogeneous fluid (clot) that tends to accumulate near the site of injury [34]. Interloop or triangular fluid collections or streaky, hazy infiltration of the mesentery are also highly suggestive of bowel trauma [15, 29,30,31,32,33] (Figs. 1A,1B,1C,2A,2B,3A,3B). Unfortunately, in our instruction to our observers, we did not require them to specify the location or appearance of intraperitoneal fluid collections, limiting our ability to analyze the reliability of several individual signs of mesenteric infiltration or hematoma. Our observers detected mesenteric infiltration in 68% of the cases of bowel injury, and 23% of the cases of other injuries. Other investigators report higher sensitivity and specificity for this sign but differ somewhat in its precise definition [13, 17,18,19,20,21, 30]. Rizzo et al. [17] found mesenteric infiltration or bleeding in 24 (86%) of 28 patients with bowel or mesenteric injury. Hagiwara et al. [30] reported that streaky mesenteric infiltration was 69% sensitive and 100% specific for the diagnosis of intestinal rupture. However, Breen et al. [19] found that isolated mesenteric streaking was a less useful sign (sensitivity of 77%, specificity of 44%). Several investigators have noted that hepatic and splenic lacerations rarely result in mesenteric collection of blood [11,12,13,14,15,16,17,18,19,20,21, 30].
Bowel wall thickening, as an isolated finding, is of little value in diagnosing bowel injury, except in diagnosing an obvious substantial mural hematoma. Other causes of diffuse and focal bowel wall thickening are well known including shock bowel, the pattern of diffuse intestinal submucosal edema and mucosal enhancement due to the hypovolemia or reperfusion process [35]. Nevertheless, bowel wall thickening is a common (60% of our 50 patients) and important ancillary finding. Breen et al. [19] calculated this sign as having a sensitivity of 50% and a specificity of 84% on the basis of 12 patients with bowel injury. Bowel wall thickening with mesenteric infiltration or active hemorrhage is a particularly common and important combination and was recognized in 89% of the patients who had surgical confirmation of bowel or mesenteric injuries reported by Dowe et al. [18].
Although we did not include these signs in our investigation, other clinical and radiologic signs are highly predictive of associated bowel injury, including traumatic disruption of abdominal wall muscles (traumatic hernia) and seat-belt ecchymosis or bruising of the abdominal wall, particularly in children [36, 37].
Accurate CT diagnosis of bowel injury is increasingly important because most children and adults who are hemodynamically stable are now treated nonoperatively, even those with CT evidence of hemoperitoneum and solid-viscera injuries [3, 7]. However, transmural bowel lacerations, mesenteric injuries with active extravasation (bleeding), and some other bowel and mesenteric injuries still require surgery [8,9,10, 18].
Our investigation is limited to a degree by its retrospective nature, which we believe to be unavoidable because of the relative rarity of bowel injuries and the need to collect a sufficient number of cases to achieve statistical significance. However, we believe that most biases were minimized because we included a large number of all types of abdominal injuries both in the study group and the comparison group and because we tested nine radiologists, each of whom was unaware of the diagnosis and the number of bowel trauma cases. We also report the results of the prospective interpretation of these 112 CT examinations, in which the diagnosis of bowel injury was made with an accuracy of 82%, a sensitivity of 64%, and a specificity of 97%.
Other potential criticisms of our investigation concern other possible biases. Many patients who had bowel injuries did not have CT evaluation and were not, therefore, included in the study because they had clinically obvious and severe injuries requiring urgent surgery. It would be impractical and unethical to subject these patients to CT, and so our population is biased toward less severe bowel injuries, although these patients with less severe injuries reflect the potential candidates for CT. Whether inclusion of patients with more severe bowel injuries would have influenced our results is unknown.
Our CT observers may have been biased by knowing they were participating in a study of radiologists' performance in detecting bowel and other abdominal injuries on CT and by having been encouraged to review pertinent trauma CT literature. However, no other special preparation or coaching was provided. Moreover, the receiver operator characteristic methodology that we used allowed us to analyze the trade-off between the sensitivity and specificity of CT and to detect any tendency toward overinterpreting that might have resulted from interpreter bias.
We conclude that bowel and mesenteric injuries remain a challenge to diagnose; nevertheless, radiologists with various levels of experience and expertise can achieve accurate and reproducible results using a variety of CT criteria.
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  M. Atri, J. M. Hanson, L. Grinblat, N. Brofman, T. Chughtai, and G. Tomlinson Surgically Important Bowel and/or Mesenteric Injury in Blunt Trauma: Accuracy of Multidetector CT for Evaluation Radiology, November 1, 2008; 249(2): 524 - 533. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Whitfield and J. Garner Beyond splenectomy -- options for the management of splenic trauma Trauma, October 1, 2008; 10(4): 247 - 259. [Abstract] [PDF]
|
|
|
|
|
|
C. Whitfield and J.P. Garner The early management of gunshot wounds Part II: the abdomen, extremities and special situations Trauma, January 1, 2007; 9(1): 47 - 71. [Abstract] [PDF]
|
|
|
|
|
|
J. W. Stuhlfaut, J. A. Soto, B. C. Lucey, A. Ulrich, N. K. Rathlev, P. A. Burke, and E. F. Hirsch Blunt Abdominal Trauma: Performance of CT without Oral Contrast Material Radiology, December 1, 2004; 233(3): 689 - 694. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Shanmuganathan, S. E. Mirvis, W. C. Chiu, K. L. Killeen, G. J. F. Hogan, and T. M. Scalea Penetrating Torso Trauma: Triple-Contrast Helical CT in Peritoneal Violation and Organ Injury--A Prospective Study in 200 Patients Radiology, June 1, 2004; 231(3): 775 - 784. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. B. Sirlin, M. A. Brown, O. A. Andrade-Barreto, R. Deutsch, D. A. Fortlage, D. B. Hoyt, and G. Casola Blunt Abdominal Trauma: Clinical Value of Negative Screening US Scans Radiology, March 1, 2004; 230(3): 661 - 668. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Wittenberg, M. G. Harisinghani, K. Jhaveri, J. Varghese, and P. R. Mueller Algorithmic Approach to CT Diagnosis of the Abnormal Bowel Wall RadioGraphics, September 1, 2002; 22(5): 1093 - 1107. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. Shanmuganathan, S. E. Mirvis, W. C. Chiu, K. L. Killeen, and T. M. Scalea Triple-Contrast Helical CT in Penetrating Torso Trauma: A Prospective Study to Determine Peritoneal Violation and the Need for Laparotomy Am. J. Roentgenol., December 1, 2001; 177(6): 1247 - 1256. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
