| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1
Department of Radiology, University of Miami School of Medicine, Jackson
Memorial Medical Center, West Wing 279, 1611 N.W. 12th Ave., Miami, FL
33136.
2
Department of Radiology, The Johns Hopkins Medical Institutions, 600 N. Wolfe
St., Baltimore, MD 21287.
3
Nemours Children's Clinic, 807 Nira St., P. O. Box 5720, Jacksonville, FL
32247.
4
Department of Radiology, Baptist Hospital, 8900 N. Kendall Dr., Miami, FL
33176.
Received September 20, 2000;
accepted after revision February 7, 2001.
Address correspondence to A. A. Jabra.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. The CT scans of 30 children with surgically proven small-bowel obstruction, 22 children with ileus, and 29 children who served as controls were retrospectively reviewed by two of four interpreters who were unaware of the children's final diagnoses. Causes of obstruction in the patients included 19 adhesions, six cases of volvulus, five intussusceptions, four strictures, and two cases each of internal hernia and abscess. Eight obstructions had multiple causes. The CT scans were evaluated for the presence of small-bowel obstruction using a scale with five degrees of confidence. In cases of discrepancy of more than one level of certainty, a third interpreter was consulted. Criteria for small-bowel obstruction included a discrepancy in caliber between the proximal dilated and the more distal small bowels or generalized small-bowel dilatation (>2.5 cm) in the presence of a collapsed colon. An interpreter's rating that an obstruction was either present or probable was considered a positive finding; a rating indicating that the interpreter was not sure whether an obstruction was present or believed that an obstruction was not probable or saw normal anatomic structures was considered a negative finding for small-bowel obstruction. The cause and level of obstruction also were recorded.
RESULTS. There were 26 true-positive (87%) and four false-negative (13%) interpretations for small-bowel obstruction. Among the interpretations of scans of patients with ileus, 68% were true-negative and 32% were false-positive interpretations for small-bowel obstruction. Among the control group, there were no false-positive readings. Sensitivity of CT was 87%, specificity was 86%, and accuracy was 86%. In the scans of children 2 years and younger, CT had a sensitivity of 100% and specificity of 0%. Of the patients with surgically confirmed levels of obstruction, the correct level of obstruction was described by both interpreters in 12 (86%) of 14 scans. The causes of obstruction were correctly identified in 14 (47%) of 30 scans.
CONCLUSION. CT is both sensitive and specific for use in diagnosing small-bowel obstruction in children, especially in children older than 2 years.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The five of the CT scans were performed on a 9800 HiSpeed Advantage scanner (General Electric Medical Systems, Milwaukee, WI), three were performed on a Somaton DR3 scanner (Siemens, Erlangen, Germany), and two scans were performed on the DRH Siemens scanner. The remaining 71 scans were performed on the DRH+ Siemens scanner. Scan parameters included 140-280 mA, 120-125 kVp, and 0.7-3.0 sec of scanning time. Of the 81 scans performed, 39 of the scans were helical. Collimation was 5-8 mm with data obtained at 5- to 10-mm intervals.
Both the usage and type of contrast medium varied, with some children receiving an oral contrast agent; some, an IV contrast agent; some, both; and some, none. All 29 control scans were obtained using IV contrast medium, and 28 children in the control group also received oral contrast medium. Thirteen of 22 scans in the ileus group received IV contrast medium, and 18 received oral contrast medium. Of the 30 patients with obstruction, 15 had IV contrast medium, and 24 received oral contrast medium.
The CT scans of all 81 children were randomized, and each scan was retrospectively and independently reviewed by a random combination of two of four interpreters. The interpreters were unaware of any patient's final diagnosis. The scans were evaluated for the presence or absence of obstruction on a scale with five degrees of confidence—obstruction present, obstruction probable, not sure, obstruction not probable, and normal study. In the case of a discrepancy of more than one level of certainty between the two initial interpreters, a third interpreter was consulted so that a consensus could be reached.
The criteria used for diagnosing a small-bowel obstruction have been previously described and include a discrepancy in caliber between the dilated proximal and the more distal small bowels [1,2,3] (Fig. 1) or the presence of a dilated small bowel with a collapsed colon, particularly the ascending colon [1,2,3] (Fig. 2). A dilated small bowel was defined as one having a diameter larger than 2.5 cm for children older than 1 year; for children younger than 1 year, the presence of a dilated small bowel was subjectively determined by the interpreters. A rating of "obstruction present" or "obstruction probable" was considered a positive finding for small-bowel obstruction. A rating of "not sure," "not probable," or "normal study" was considered a negative finding for small-bowel obstruction. The interpreters were asked to identify the cause of obstruction. If no cause of obstruction was identified, a diagnosis of adhesions was inferred. The interpreters were also asked to determine the level of small-bowel obstruction—proximal (jejunal) or distal (distal jejunal or ileal).
|
|
The interpreters' diagnostic confidence ratings were used to calculate receiver operating characteristic (ROC) curves for the detection of small-bowel obstruction. The ROCFIT program [8] was used to fit ROC curves to the confidence rating data.
Proportions were compared using the Fisher's exact test as calculated by a software program (Stata, version 6.0; Stata, College Station, TX). A p value of less than 0.05 was considered statistically significant.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
|
|
Sixteen of the 30 patients with small-bowel obstructions had the levels of obstruction described in the operative reports. Fourteen obstructions were distal, and two were proximal. In the 16 patients with surgically reported levels of obstruction, 14 were correctly diagnosed with obstructions by the interpreters, and two were given false-negative findings. Of the 14 patients correctly diagnosed with obstructions, the level of obstruction was correctly identified by both interpreters in 12 (86%). The correct level of obstruction was identified by at least one interpreter in all 14 patients. One additional patient was diagnosed with a volvulus involving most of the bowel. In this patient, both interpreters reported the level of obstruction to be distal.
Because the differentiation between ileus and small-bowel obstruction is often the clinically relevant question, a ROC curve for the detection of small-bowel obstruction (Fig. 5) was calculated from a subset of the data (52 cases), which includes all patients except those in the control group. The area under the ROC curve was 0.84, with a standard deviation of 0.06.
|
Within the subset of 52 children, 10 children were 2 years or younger. In all these children, a small-bowel obstruction was detected on CT scans, although three had ileus rather than an obstruction. Therefore, the specificity for the detection of small-bowel obstructions on the scans of these children was 0%, and the sensitivity was 100%. The remaining 42 children in the group with a disease were older than 2 years. For these children, the specificity of CT was 79% (15 of 19 patients) and the sensitivity was 83% (19 of 23 patients). A comparison of the two age groups shows a statistically significant difference in CT specificity (p = 0.02) but no difference in CT sensitivity (p = 0.5).
The scans for four of the 30 patients with obstructions were judged as false-negative for obstruction. These included two patients with adhesions, one with an internal hernia, and one with a volvulus around an omphalomesenteric duct. In the remaining 26 patients, the interpreters identified no cause for the obstruction in 16, including nine with isolated adhesions. The interpreters proposed likely causes for obstruction in 10 of the patients, including two with multiple causes. Of these, the interpreters correctly identified three of five intussusceptions, one of three cases of volvulus, and one abscess. The interpreters reported incorrect causes for the obstruction in the remaining five children. None of the four strictures were identified (one patient had both a stricture and Crohn's disease). Neither of the two patients with an internal hernia was identified. Of 19 patients with adhesions, the findings on the scans of two were judged as false-negative for small-bowel obstruction. Seven had both adhesions and other causes for the obstruction. Of the remaining 10 patients with isolated adhesions, the scan of one was misinterpreted as showing an intussusception; nine patients were correctly diagnosed as having a small-bowel obstruction with no apparent cause for the obstruction. Thus, the scans of these nine patients were correctly diagnosed by inference as obstruction associated with adhesions. Overall, the interpreters identified the correct cause for the obstruction in 14 (47%) of 30 scans, including intussusception (3/5), abscess (1/2), volvulus (1/6), and adhesions (9/12 isolated adhesions).
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
To our knowledge, the sensitivity of CT in revealing small-bowel obstructions in children has not been previously reported, nor has the success of CT in determining the correct cause of obstruction been previously reported in a study in which the reviewers were unaware of the final diagnosis. In our study, the sensitivity of CT in revealing small-bowel obstruction was determined to be 87%, its specificity was 86%, and its accuracy was 86%. In adult series, the sensitivity, specificity, and accuracy values were 63-94%, 78-100%, and 65-95%, respectively [2,3,4,5]. The lower range of values was reported in studies with a preponderance of patients with low-grade obstruction, and the higher range of values was reported in studies with a preponderance of patients with high-grade obstruction. The sensitivity of CT in studies with partial small-bowel obstruction ranged from 48-100% [4, 5].
In our study, we did not characterize the patients as having low- versus high-grade obstruction, a distinction that was generally not made in the discharge diagnosis of or operative notes on the patients. The results of our study fall within the range of results reported for adults in the literature. In our study, most of the scans depicting ileus and small-bowel obstruction were not obtained on helical scanners. The use of helical and multidetectors scanners will likely yield technically better scans and probably improved results.
In our study, we included a group of 22 patients with ileus because this diagnosis is likely the most difficult to differentiate from small-bowel obstruction, both in clinical presentations and on radiographs. If the group with ileus is excluded, the specificity of CT in revealing small-bowel obstruction increases to 100%. If the control group is excluded and only the small-bowel obstruction and the ileus groups are included, then the specificity of CT in revealing small-bowel obstruction decreases to 68%, and the accuracy decreases to 79% (Fig. 5), values that are still within the range reported in adult series. In our study, there were four false-negative and seven false-positive interpretations for small-bowel obstruction, which affect the rates of sensitivity and specificity, respectively.
False-Negative Interpretations
One child with an internal hernia showed no bowel dilatation on CT, likely
because of an intermittent obstruction that was not present at the time the
scan was obtained. Another patient showed a transition between mildly dilated
proximal and nondilated distal bowel. The distal bowel wall was thickened, and
an ileus related to inflammation rather than obstruction was considered to be
the cause. A third patient's scan showed air distention rather than increased
fluid content in the dilated loops. A bowel with a typical hairpin appearance
of a closed-loop obstruction (the bowel trapped by adhesions) was missed
(Fig. 3). Another patient's
scan showed a borderline dilated distal small bowel with mildly thickened
walls. The ascending colon was filled with stool. This patient had a volvulus
involving the last 60 cm of the ileum around an omphalomesenteric duct.
False-Positive Interpretations
The CT scans of three children clearly met the CT criteria for bowel
obstruction. In these children, the interpreters considered the CT scans as
showing positive results for obstruction. These children improved with
conservative treatment, and their final discharge diagnosis was ileus rather
than obstruction. The readings were therefore classified as false-positive.
These patients raise a question about the nonsurgical diagnosis of small-bowel
obstruction and the accuracy of the clinical diagnosis of ileus versus
obstruction when symptoms improve without
surgïcal intervention. We know that obstruction
caused by adhesions can improve or resolve with conservative treatment. In a
study by Donckier et al. [10],
28 (76%) of 37 patients thought to have simple (no strangulation) adhesive
small-bowel obstruction were successfully treated conservatively. In another
study, 35% of patients with adhesions were treated conservatively
[18]. We believe that these
three children were true-positive for obstruction rather than false-positive
and that their obstructions improved with conservative treatment. If these
three patients are considered as among the true-positive interpretations for
obstruction, then the specificity as well as the sensitivity in this study
improves. The study population would then consist of 33 children with
obstruction, 19 with ileus, and 29 controls. The sensitivity increases from
87% to 88%, and the specificity increases from 86% to 92%.
Level of Obstruction
The level of obstruction was correctly identified by both interpreters in
most children (86%) with levels of obstruction documented in the operative
records. CT is reported to be successful in determining the level of
obstruction in up to 97% of cases in adult series
[5]. The level of obstruction
was described in the operative records for only 57% of our children and for
most (88%) of these children, the level of obstruction was described as
distal. This distal location may reflect the select nature of our small-bowel
obstruction group; distal obstruction may be more difficult to differentiate
from ileus on conventional radiographs and could thus result in referral to CT
for the unexplained and indeterminate clinical and radiographic findings.
Cause of Obstruction
In our study, the correct cause of obstruction was identified in a limited
number of patients. Five had specific causes, and nine were correctly presumed
to have adhesions (because no cause for the obstruction was evident on the CT
scans). Considering all 30 cases with small-bowel obstruction (including four
false-negative readings), the correct cause was therefore made in only 47% of
the patients. Excluding the false-negative readings, the correct cause of
obstruction was made in 14 (54%) of 26 children. In adult series, the correct
causes of obstruction have been reported as being detected in 76-90% of cases
[2,3,5,7].
Diagnoses were missed for two of five patients with intussusception in our study, though the correct diagnosis of small-bowel obstruction was made. Neuroblastoma, noted as an abdominal mass, was suggested as the cause of obstruction by both interpreters in an 18-month-old child with postoperative intussusception. The second case of intussusception was missed in a 5-month-old infant with ileoileal intussusception. The three cases in which the intussusception was correctly identified were older children (3, 17, and 18 years old). The difficulty in identifying the two missed intussusceptions is likely a reflection of the younger age of the patients. The younger children are physically smaller, which could make the intussusception area less obvious. Another reason for the difficulty with the diagnosis is the relative paucity of mesenteric fat, which is usually identified between the intussusceptum and the intussusception and is virtually diagnostic. In one series, 14 (88%) of 16 adult intussusceptions showed this fatty area at CT [19]. In another adult study including CT scans and a small number of MR imaging studies, 28 (68%) of 41 intussusceptions showed the fat on imaging [20]. No reports are available describing the frequency of visualizing fat within an intussusception on CT scans in children. The areas of intussusception in the two missed patients were obvious in retrospect but were smaller, with the fat density in the intussusception appearing on the scans more like the density of water than that of fat (Fig. 6); no Hounsfield unit density measurements were obtained. This finding is likely related to volume averaging of the small amounts of fat with the adjacent soft tissues. In addition, the neuroblastoma may have satisfied search phenomenon, distracting from the correct diagnosis in the case of one child. The difficulty in diagnosing intussusception in the young child may also explain the statistically significant lower specificity of differentiating small-bowel obstruction from ileus in children younger than 2 years old.
|
Volvulus was detected correctly in only one of six patients and by only one interpreter. The scan of a second child with volvulus showed a whirl pattern that was misinterpreted as an intussusception by both interpreters (Fig. 7). Familiarity with this pattern in volvulus would improve the diagnostic accuracy [10,17,21]. In the remaining four children, the volvulus did not have either the whirl sign or a second pattern typically seen as radial distribution of dilated bowel around engorged mesenteric vessels converging towards the point of obstruction [9,22]. One of these four children had an additional ileoileal intussusception diagnosed correctly by one of two interpreters.
|
Both internal hernias were missed. One of two internal hernias was in an unusual retropsoas location and was obvious when reviewed retrospectively. There was no bowel dilatation at the time of scanning, and the scan was considered to show false-negative results for small-bowel obstruction. This patient has been described in detail by Benson and Strauch [23].
When no cause for the obstruction is identified at CT in a patient with a history of abdominal surgery, adhesions are usually assumed to be the cause [2]. Of the 16 patients for whom our interpreters identified no cause for obstruction, only nine (56%) had isolated adhesions. Adhesions, however, were correctly inferred as the diagnosis in nine (75%) of 12 children with isolated adhesions.
In our study, the interpreters were not requested to determine whether closed-loop obstruction or strangulation (ischemia) was present. This area has received much attention in the recent adult literature [9,11,13,14,22,24,25]. Our interpreters did not identify the cause of obstruction in five patients with closed-loop obstructions (four with segmental volvulus and one with an entrapped ileal loop by adhesions) (Fig. 3). To assess the accuracy of CT in these diagnoses in children, a separate study performed after familiarizing interpreters with the CT appearance of closed-loop obstruction and strangulation obstruction may be needed. In addition, standardizing the imaging technique in a prospective study would lead to more reliable results.
CT has a high sensitivity, specificity, and accuracy in revealing small-bowel obstructions in children, especially in children older than 2 years. The level of obstruction was correctly diagnosed in most patients. In this study, CT had limited success in helping the interpreters to determine the cause of small-bowel obstruction. Familiarity with the appearance of obstructive lesions on CT images and the use of helical and multidetector scanners will likely yield better results. A prospective study with a standardized imaging technique is needed for more accurate results.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  J. C. Leonidas, A. A. Jabra, and J. Eng CT of Small-Bowel Obstruction Am. J. Roentgenol., April 1, 2002; 178 (4): 1030 - 1031. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
