Published online before print February 28, 2006, 10.1148/radiol.2391050610
(Radiology 2006;239:139-148.)
© RSNA, 2006
Causes of Errors in Polyp Detection at Air-Contrast Barium Enema Examination1
William M. Thompson, MD,
William L. Foster, MD,
Erik K. Paulson, MD,
Donna Niedzwiecki, PhD,
Vincent H. S. Low, MD,
Lori B. Fulford, PhD,
Bob W. Broomer, BS,
Linda Sanders, BS and
Don C. Rockey, MD
1 From the Departments of Radiology (W.M.T., E.K.P.), Medicine (L.B.F., B.W.B., L.S., D.C.R.), and Biostatistics and Bioinformatics (D.N.), Duke University Medical Center, Box 3808, Durham, NC 27710; Department of Radiology, Durham Veterans Administration Hospital, Durham, NC (W.L.F.); and Department of Radiology, Sir Charles Gairdner Hospital, Nedlands, Western Australia (V.H.S.L.). From the 2004 RSNA Annual Meeting. Received April 12, 2005; revision requested June 16; revision received July 19; final version accepted August 2.
Address correspondence to W.M.T. (e-mail: thomp132{at}mc.duke.edu).
 |
ABSTRACT
|
|---|
Purpose: To retrospectively determine the cause of errors in air-contrast barium enema (ACBE) examination for detection of polyps 6 mm or larger.
Materials and Methods: The study had institutional review board approval. Informed consent was waived for this HIPAA-compliant study. ABCE findings in 41 subjects with 56 missed polyps were evaluated by two radiologists to determine if the cause of errors was perceptual or technical. A comparison was made between total number of polyps in the proximal and distal colon and those missed at each location (Fisher exact test). The 288 ACBE examinations were assessed on a scale of 0–4 (0, excellent; 4, very poor) for six colonic segments (paired t test).
Results: Of 17 polyps 1 cm or larger not detected in 15 subjects, 11 (65%) were missed because of technical errors and six (35%) because of perceptual errors. Eight (72%) technical and four (67%) perceptual errors occurred proximal to the splenic flexure. One 3.5-cm cecal carcinoma was not diagnosed prospectively (perceptual error). Of 39 6–9-mm polyps not detected in 26 subjects, 35 (90%) were missed because of technical errors and four (10%) because of perceptual errors. Eighty percent of technical and 75% of perceptual errors were in the proximal colon. When the proportion of polyps in the proximal and distal colon was compared, 22 (63%) of 35 polyps in the distal colon and 15 (26%) of 58 in the proximal colon were detected (P = .0009). There were no detectable differences in the quality of studies in subjects whose polyps were detected and subjects whose polyps were missed (P > .05).
Conclusion: Technical errors were more common than perceptual errors. The majority of missed polyps were in the proximal colon. Detection rates of polypoid lesions might increase if the quality of ACBE examination can be improved, especially in the proximal colon.
© RSNA, 2006
|
INTRODUCTION
|
|---|
While the recent widespread implementation and use of optical colonoscopy has greatly contributed to an ongoing decrease in the performance of air-contrast barium enema (ACBE) examination, some practices still use ACBE to detect polyps and colorectal cancer. Findings of some studies indicate that the sensitivity of ACBE examination in detecting polyps 1 cm and larger ranges from 85% to 95% (1–10). However, widely quoted results from the National Polyp Study indicated that ACBE examination had a sensitivity of only 48% for detection of polyps 1 cm and larger (11). In addition, the recent completion of the much larger multicenter Colon Imaging Study, in which each subject underwent ACBE examination, computed tomographic (CT) colonography, and optical colonoscopy, likewise demonstrated that ACBE examination had a sensitivity of 48% for detection of polyps 1 cm and larger (12). The causes of the poor sensitivity reported in these studies are of general interest to all physicians who are charged with colorectal screening or evaluation of the colon in dealing with colorectal polyps. Findings of many studies (13–18) indicate that perceptual errors (polyps present at retrospective review) are more common than technical errors (polyps not present at retrospective review).
The Colon Imaging Study was designed to quantify the quality of the ACBE examination (12). On the basis of previous experience and previous reports in the literature, we hypothesized that most of the errors would be perceptual rather than technical and occur in the distal colon. Thus, the purpose of our study was to retrospectively determine the cause of errors in ACBE examination for detection of polyps 6 mm or larger.
|
MATERIALS AND METHODS
|
|---|
Patient Population and Study Demographics
Between 2001 and 2004, 614 subjects completed a multicenter prospective trial of colon imaging tests (12). The study compared ACBE examination, CT colonography, and optical colonoscopy. Two hundred eighty-eight participants completed the study at our institutions. Details of the study design have been published (12). In brief, entry criteria into the study included a family history of colorectal neoplasia, hematochezia, iron deficiency anemia, and/or positive fecal occult blood. All subjects underwent ACBE examination on 1 day. Within 2 weeks, CT colonography and optical colonoscopy were performed on the same day. A consensus colon view (by using all three tests) was used as the reference standard for polyp detection. If a consensus could not be reached, all three tests were repeated. Polyp size was determined by using a ruler for the ACBE examination, without any correction for magnification, and by using electronic calipers for CT colonography. Polyp size was determined by means of comparison with the size measured with open biopsy forceps at the time of optical colonoscopy. Polyps that were within one contiguous segment and polyps measuring more than or equal to 50% of the diameter reported at optical colonoscopy were considered a match to the results of the other two comparative tests. The prospective polyp detection rates for both sizes of polyps were tabulated from the case report forms. Polyps identified at colonoscopy were removed and submitted for pathologic analysis.
The prospective trial had institutional review board approval and informed consent and was compliant with the Health Insurance Portability and Accountability Act. Our retrospective study also had institutional review board approval, with waiver of informed consent, and was compliant with Health Insurance Portability and Accountability Act.
Preparation for ACBE examination was made by using two methods. At Duke University Medical Center, subjects received magnesium citrate (16.4 g; E-Z-Em, Westbury, NY) at 5:00 PM the day before the examination, followed by 4-mg bisacodyl tablets (E-Z-Em) at 7:00 PM, and a 10-mg bisacodyl suppository at least 2 hours prior to the ACBE examination. Subjects were instructed to maintain a clear liquid diet the day before the study and were asked to drink 6–8 fl oz (180–240 mL) of water during the day just before the study. At the Durham Veterans Administration Hospital, subjects received a 2-day preparation that consisted of a clear liquid diet for 2 days before the study. At 11:00 AM 2 days before the ACBE examination, subjects took 4-mg bisacodyl tablets, followed by magnesium citrate (16.4 g) at 8:00 AM the next morning (24 hours prior to the examination), and four more 5-mg bisacodyl tablets at 7:00 PM the night before ACBE examination.
ACBE examinations were performed by using a high-density barium suspension (100% wt/vol; E-Z-Em). After administration of the barium and distention of the colon with room air, spot radiographs were obtained of all specific colon segments. In addition, overhead radiographs were obtained in prone, prone 35° angled, supine, left and right lateral decubitus, and left lateral positions without the rectal tube. Glucagon was not used routinely. The ACBE technique at Durham Veterans Administration Hospital was slightly different, since the left and right decubitus radiographs were the first two overhead radiographs obtained, whereas at Duke University Hospital they were obtained just before the lateral rectal radiograph with the rectal tube removed was obtained.
Polyp Characteristics
At optical colonoscopy, 31 polyps 1 cm or larger were measured in 28 (10%) of 288 subjects (Table 1). Seventeen (55%) were not detected at ACBE examination. Four (13%) of the 31 were malignant and three (75%) of these cancers were detected prospectively. All four lesions were larger than other polyps in the study and ranged in size from 3.5 to 6.0 cm. Of the 16 remaining polyps not detected with ACBE examination, 13 were adenomas, two were hyperplastic polyps, and one was hamartoma. Sixty-two polyps 6–9 mm in size were proved in 40 (14%) of the 288 subjects. Thirty-nine (63%) of these polyps were not detected at ACBE examination and none were malignant. Of the 6–9-mm polyps not detected, 32 were adenomas, six were hyperplastic, and one was a lipoma.
Polyp Evaluation
The colon segment of the missed and detected polyps was recorded on the basis of case report forms. Two experienced radiologists (W.M.T. and W.L.F with 34 and 25 years of experience, respectively) jointly evaluated findings of all the ACBE examinations in 41 subjects (35 men and six women; mean age, 57 years; range, 22–80 years) in whom 56 polyps were missed to determine the sources of error. At the time of the review, the radiologists were aware of the location and size of the polyps.
Polyps that could be identified in retrospect were classified as perceptual errors. These errors were further subdivided as follows: polyps missed in the barium pool, mucosal surface of the polyp en face, contour defect due to missed polyp, poor coating, possible distraction from another detected polyp, poor distention, and other possible causes of missed polyps (12). If the polyp could not be identified at retrospective review, it was considered a technical error. We then classified technical errors as follows: excessive barium pool in the involved segment where the polyp was located, inadequate luminal distention, inadequate coating of the mucosa, over- or underexposure of film, polyp obscured by ileal reflux, and other possible causes of errors. The location of the missed and detected polyps was determined in 68 subjects by using case report forms.
Quality of Bowel Preparation
The quality of bowel preparation was determined by attending radiologists (W.L.F., V.H.S.L., W.M.T.) responsible for interpreting the findings of the ACBE examination. Their years of experience ranged from 2 to 30 years. The quality of each ACBE examination of the 288 subjects was rated on a scale of 0–4, with 0 indicating excellent quality and 4 indicating very poor quality. The five criteria used for the assessment of quality were stool in the lumen, stool in the mucosa, distention, ability to assess for diverticula, and ability to assess for polyps. Six segments of the colon (rectum, sigmoid colon, descending colon, transverse colon, ascending colon, and cecum) were evaluated separately. By assuming an underlying continuous distribution, the mean scores were computed for each of the criteria according to segment and overall. The data were tabulated for each of the five criteria for the six segments and as a total mean score for each colonic segment. A comparison of the quality bowel preparation was performed for subjects with detected polyps and those with missed polyps, as well as for subjects with polyps in proximal versus distal colon.
Statistical Analysis
For the assessment of the quality of bowel preparation, the proximal and distal mean scores were computed within subjects for all 288 subjects and were compared by using the paired t test. A Bonferonni correction was used to adjust for the five comparisons. A P value of .01 or less was considered to indicate a significant difference. Approximate estimates and 95% confidence intervals (CIs) were constructed for mean quality scores for subjects with polyps of sizes greater than or equal to 1 cm that were detected and missed at ACBE examination and those with lesions in the proximal versus the distal colon. The proportions of lesions detected in the proximal and distal colon were compared by using the Fisher exact test. P < .05 was considered to indicate a statistically significant difference. All P values are two sided.
|
RESULTS
|
|---|
Polyp Characteristics and Location and Types of Errors
The 17 undetected polyps 1 cm or larger had a mean size of 1.5 cm (range, 1–3 cm) compared with a mean of 2.1 cm (range, 1.0–6.0-cm polyps, the 6-cm lesion was a large rectal cancer) for the 14 detected polyps. The largest missed polyp was a 3.5-cm cecal carcinoma. Of five (29%) missed distal polyps, none were in the rectum (Tables 2, 3), three were located in the sigmoid, and two were located in the descending colon. Twelve (71%) missed polyps were located in the proximal colon: three in the transverse colon, five in the ascending colon, and four in the cecum. Eight (73%) of 11 technical errors occurred in the proximal colon, and four (67%) of six perceptual errors occurred in the proximal colon (Table 2).
For two of 26 subjects, radiographs with polyps 6–9 mm could not be located for review. One of these subjects had two missed polyps. The 39 6–9-mm missed polyps had a mean size of 7 mm. Of the eight missed distal polyps, only one was located in the rectum (Table 3). The majority of the proximal 31 (79%) polyps missed were in the transverse (n = 15) and ascending (n = 14) colon (Tables 2, 3). Of 35 technical errors, 28 (80%) occurred in the proximal colon; three (75%) of four perceptual errors also occurred in the proximal colon (Table 2).
There were more technical errors (n = 46, 82%) than perceptual errors (n = 10, 18%) when both groups were considered (Table 4). Fifty-eight (62%) of the 93 polyps 6 mm and larger were in the proximal colon and 35 (38%) were in the distal colon (Table 4). However, when the two sizes of missed polyps were combined, 43 (77%) of the 56 were located in the proximal colon: six in the cecum, 19 in the ascending colon, and 18 in the transverse colon (Table 3).
The proportion of polyps detected in the proximal and distal colon was compared. Twenty-two (63%) of 35 polyps in the distal colon were detected, while 15 (26%) of 58 polyps in the proximal colon were detected (Fisher exact test, P = .0009) (Table 4).
Categories of Errors
Perceptual errors.—There were 13 perceptual errors in six subjects with polyps 1 cm and larger, and five subjects had more than one cause of error (Table 5). The most common error was a failure to identify the polyp in the barium pool (Fig 1). Three polyps visualized only en face were missed (Fig 2). A contour defect was overlooked in one 3.5-cm cecal adenocarcinoma (Fig 3). Poor coating of the colon led to errors (Fig 4). In one case, adjacent diverticula were present (Fig 5). There were eight perceptual errors in four subjects with polyps 6–9 mm in size, and three subjects had more than one cause of perceptual error (Table 5). Two polyps were missed en face, and two polyps were not coated by barium.
View larger version (151K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1: Perceptual error: polyp in the barium pool. Anteroposterior spot radiograph of cecum from ACBE examination shows 9–10-mm polyp (arrow) in the barium pool not recognized during initial assessment. At optical colonoscopy, this lesion measured 15 mm.
|
|
View larger version (171K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2: Perceptual error: polyp visualized en face. Left posterior oblique radiograph from ACBE examination shows 15-mm pedunculated polyp (arrow) en face, which was not detected initially.
|
|
View larger version (124K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3a: Perceptual error: cancer visualized en face and contour defect. (a) Anteroposterior radiograph of cecum shows contour defect (arrows) due to 3.0-cm cecal adenocarcinoma, which was not recognized at initial examination. (b) Slight left posterior oblique radiograph of cecum shows 3.0-cm adenocarcinoma (arrows), which is difficult to recognize en face.
|
|
View larger version (136K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3b: Perceptual error: cancer visualized en face and contour defect. (a) Anteroposterior radiograph of cecum shows contour defect (arrows) due to 3.0-cm cecal adenocarcinoma, which was not recognized at initial examination. (b) Slight left posterior oblique radiograph of cecum shows 3.0-cm adenocarcinoma (arrows), which is difficult to recognize en face.
|
|
View larger version (122K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4: Perceptual and technical errors: poor coating. Left posterior oblique radiograph of cecum that contained three 15–20-mm polyps; one perceptual error due to poor coating (arrow). The other two polyps were not identified on any radiographs.
|
|
View larger version (149K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5: Perceptual error: distraction due to diverticulosis. Coned-down anteroposterior view from radiograph of descending colon shows 10-mm polyp (arrow) adjacent to colonic diverticuli.
|
|
Technical errors.—There were 21 technical errors in 11 subjects with polyps 1 cm and larger, and nine subjects had more than one cause of technical error (Table 6). The most common error (n = 9) was poor distention (Fig 6). Other explanations for technical errors included gross diverticulosis (Fig 7), gross stool, a cardiac pacemaker battery pack obscuring a splenic flexure polyp, and a polyp posterior and adjacent to the ileocecal valve (Fig 8). There were 44 technical errors in 20 subjects with 6–9-mm polyps, and all 21 subjects had more than one cause of error. The most common cause of a technical error was diverticulosis obscuring polyps (n = 12). Other causes included poor distention (n = 9), excessive barium (n = 8), excessive stool (n = 7), and excessive redundancy (n = 3).
View larger version (134K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 6a: Technical error: poor distention. (a) Anteroposterior radiograph and (b) anteroposterior left lateral decubitus view from ACBE examination with poor distention. A 15-mm polyp was removed from the descending colon, which could not be seen on any radiographs.
|
|
View larger version (146K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 6b: Technical error: poor distention. (a) Anteroposterior radiograph and (b) anteroposterior left lateral decubitus view from ACBE examination with poor distention. A 15-mm polyp was removed from the descending colon, which could not be seen on any radiographs.
|
|
View larger version (168K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 7a: Technical error: marked diverticulosis. (a) Overhead anteroposterior radiograph from ACBE examination demonstrates diffuse diverticulosis obscuring 15-mm descending colon polyp. (b) Coned-down anteroposterior views of descending colon in same subject show marked diverticulosis.
|
|
View larger version (124K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 7b: Technical error: marked diverticulosis. (a) Overhead anteroposterior radiograph from ACBE examination demonstrates diffuse diverticulosis obscuring 15-mm descending colon polyp. (b) Coned-down anteroposterior views of descending colon in same subject show marked diverticulosis.
|
|
View larger version (160K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 8a: Technical error: stool and a possible polyp on ileocecal value. (a) Anteroposterior radiograph from ACBE examination of cecum shows stool (arrows) obscuring 15-mm polyp. (b) Anteroposterior radiograph shows asymmetry of ileocecal valve (arrow) but no obvious other polyps. A 15-mm polyp was removed from the cecum behind a large fold, presumably the ileocecal valve, but was not proved with follow-up ACBE examination.
|
|
View larger version (147K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 8b: Technical error: stool and a possible polyp on ileocecal value. (a) Anteroposterior radiograph from ACBE examination of cecum shows stool (arrows) obscuring 15-mm polyp. (b) Anteroposterior radiograph shows asymmetry of ileocecal valve (arrow) but no obvious other polyps. A 15-mm polyp was removed from the cecum behind a large fold, presumably the ileocecal valve, but was not proved with follow-up ACBE examination.
|
|
Quality of Bowel Preparation
At one ACBE examination, there was excessive stool and a 3-cm hamartoma was missed in the transverse colon. The overall mean scores for the five criteria in the six segments in 288 patients ranged from a low of 0.29 (stool in the mucosa in the rectum) to a high of 1.08 (ability to assess for a polyp in the sigmoid) (Table 7). Adjusting for multiple comparisons (n = 288), there were significant differences in the quality of the overall bowel preparation between the proximal and distal colon for the evaluation of stool in the lumen, stool in the mucosa, and distention (Table 7). In particular, the quality of bowel preparation appeared to be better in the rectum for all five criteria than in the other locations (ascending colon, as well as descending and sigmoid colon). No significant differences were detected in the ability to assess for diverticula and polyps.
We wondered if there was any difference in the quality of bowel preparation in subjects whose polyps were detected compared with subjects whose polyps were missed. The total mean score for the bowel preparation in the 15 subjects with 17 missed polyps 1 cm and larger was 0.74 (95% CI: 0.42, 1.06). This score did not differ from that in 13 subjects with 14 polyps detected (average score, 0.61; 95% CI: 0.37, 0.84). Further, we asked if there was any difference in the quality of the preparation among subjects with lesions in the proximal versus the distal colon. We found no difference in the quality of bowel preparation when comparing 15 subjects with 16 polyps in the proximal colon and 13 subjects with 15 polyps in the distal colon (Table 8).
Twenty-six subjects had 6–9-mm polyps that were not detected at ACBE examination. Among these 26 subjects, the mean score for the quality of bowel preparation was 0.55 (95% CI: 0.40, 0.70). Similarly, among 19 subjects in whom polyps were detected, the mean score for the quality of bowel preparation was 0.39 (95% CI: 0.29, 0.49). For five subjects, the polyps were both detected and missed and are included in the estimation of both 95% CIs.
|
DISCUSSION
|
|---|
There were two striking findings in our study. First, the majority of the missed polyps were located in the proximal colon. This was especially true for polyps that were missed because of technical errors. This is interesting, as there were no apparent differences in the quality of the bowel preparations in the proximal versus the distal colon in the 56 subjects with missed polyps 6 mm or larger. However, when the proximal and the distal colon were compared in the entire group of 288 subjects, there were some substantial differences in the preparation quality, such as stool in the lumen, stool in the mucosa, and distention. The mean scores comparing all five criteria for the proximal and distal colon were not significantly different except for the rectum, which did have a better mean score and better scores for all five individual criteria than did the proximal colon segments. This may explain why only one of the missed 56 polyps was in the rectum.
Authors of other studies (1,3,13–17) have reported more missed polyps in the distal colon and have called for special attention to areas of redundancy, such as the sigmoid. Nonetheless, other studies have reported a high proportion of missed polyps in the proximal colon. Brady et al (17) reported that 11 (61%) of 18 overlooked cancers were proximal to the transverse colon. Bolin et al (19) described a relatively large number of right-sided missed polyps that were believed to have been caused by problems with cleansing and poor mucosal coating. They also suggested that the large luminal diameter and deep haustral folds in the ascending colon might obscure polyps. Other studies have shown that most missed tumors were located in the cecum and ascending colon and in rectosigmoid colon (10,11,19,20). Winawer et al (11) reported the lowest confidence levels for an ACBE examination in these two regions because of redundancy and diverticulosis in the sigmoid colon and the presence of stool and poor mucosal coating in the ascending colon and cecum.
We do not have a clear explanation for the difference in our study compared with some of the previous reports, but we question whether the reason is due to the fact that more polyps were encountered in the proximal than distal colon. In addition, the issues of colon distention and deep-haustral folds typical of the ascending colon, as emphasized by Bolin et al (19), may be important. Regardless, our findings suggest that radiologists performing ACBE examinations should pay closer attention to appropriate distention and coating of the proximal colon than perhaps they have in the past.
The second major interesting finding was that technical errors were more common than perceptual errors. It is noteworthy that Ott et al (13) and Kelvin et al (14) classified their missed colorectal carcinomas slightly differently than we did. They used a purely perceptive error (definitely seen in retrospect) and a combination of perceptive and technical errors (polyps poorly seen owing to technical factors). Kelvin et al (14) also included a category of perceptual errors of interpretation (polyps mistaken for something else) and purely technical errors (polyp invisible). We chose to classify the missed polyps as purely perceptual (definitely seen in retrospect) and purely technical (polyp not definitely seen in retrospect) errors. Brady et al (17) reported results similar to ours in terms of subdividing the misses into purely perceptual and technical. Polyps missed in the barium pool were the most common type of error in our study and in the literature (13,16). Missed polyps en face and overlooked contour defects along the bowel wall were also important causes of perceptual errors. This is particularly true for cancers (13,16). We found poor distention to be the major cause of our technical errors; this is in contrast to both Kelvin et al (14) and Brady et al (17), who found that excessive stool was the most common cause of technical errors. Poor coating was the next most common cause of errors in our series and was the third most common type of error encountered by Kelvin et al (14). We did not encounter any missed polyps because of interpretative errors except perhaps for one subject who had a prominent ileocecal valve that appeared to obscure a 15-mm adenomatous polyp. However, we had no follow-up ACBE examination to confirm this.
Diverticulosis has been previously indicated as a cause of both technical and perceptual errors. Likewise, in this study, we similarly found that diverticula were an important variable in detecting lesions with ACBE examination. This is no different than many previous reports (5,6,11): Diverticula obscuring polyps has been well known for many years by radiologists performing ACBE examinations (6).
It is interesting to note that we only had two cases of technical errors in the 1-cm and larger polyps due to excessive and undrained barium around the polyp. This was the major cause of missed colorectal cancers reported by Kelvin et al (14). Obviously, by decreasing the number of technical errors the sensitivity of ACBE examination would be improved. The quality of ACBE examinations might be improved by producing optimal distention and improved coating of the proximal colon.
The importance of polyps 6–9 mm in size remains undecided. It has been stated that these polyps do not need to be removed since they rarely contain dysplasia or cancer (21–23). However, many of these lesions contain adenomatous histologic abnormalities, and approximately 1%–2% contain advanced dysplasia and/or malignancy (24). Because of this, we chose to evaluate the polyps in two separate groups and then combine the results. Thus, we were able to present data on errors for 1-cm and larger polyps, as well as for those 6–9 mm in size. It should be pointed out that there were no cancers in our 62 polyps 6–9 mm and in the total study population of 158 6–9-mm polyps (12), which is consistent with data from another study (25).
There were a number of limitations in our study. First, the subjects came from two hospitals, which had slightly different bowel preparations and slightly different ACBE techniques. Also, we did not evaluate the role of fluoroscopy during the performance of ACBE examinations. This could have led to some bias in technical aspects of study performance.
Second, the quality of the ACBE examination was evaluated by a number of radiologists, all of whom had varying degrees of experience. Although we used a numeric scale of 0–4, which included a written description of the associated features, this type of assessment could have been subject to a variety of biases. It is likely that there would be a good interobserver agreement between readers for the 0 and 4 ratings, but it is unknown how much variability exists for ratings 1–3.
Third, the study was performed in a retrospective manner with two experienced radiologists who knew the segments where the lesions were missed at the ACBE examination. A blinded individual review might have identified specific perceptual misses that are problematic for both the very experienced and the inexperienced radiologist.
Fourth, the size of the polyps for the ACBE examination was determined by using a ruler without compensating for magnification. In addition, since measuring polyps at the time of colonoscopy can be difficult, it is possible that size differences could have contributed to bias in the assessment of a true polyp size. This is not surprising since we know there were differences in the size of polyps seen at both ACBE examination and optical colonoscopy.
Our results have a number of important implications. If perceptual error rates are eliminated, previous reports indicate that the majority of colon cancers can be diagnosed on the basis of ACBE findings (6,9,10,14). In our series, all four cancers were readily detectable, so our detection rate for cancer should have been 100%. If we had eliminated perceptual errors for polyps 1 cm or larger, the detection rate would have increased to 65% (20 of 31 polyps). Some reviewers have emphasized that double reading of ACBE studies can reduce perceptual errors (2,12,14–16), while findings in an another study (26) in which double reading was used of more than 1000 consecutive barium enema examinations (822 double- and 181 single-contrast examinations) showed no improvement in the sensitivity for polyp detection and produced a higher false-positive rate. Thus, double reading of barium enema examinations did not improve sensitivity.
A further important implication is that if technical errors can be overcome, the sensitivity of an ACBE examination can be increased considerably. Since the majority of our technical errors were in the proximal colon, we have started obtaining decubitus radiographs first to try to improve distention, especially in the proximal colon. As mentioned previously, proximal technical errors might be decreased with increased attention to the proximal colon by using more spot radiographs in more positions in an attempt to overcome redundancy and poor coating. The patients may need increased maneuvers such as more insufflation and more rotations to improve both distention and mucosal coating. While we have no data to determine if the aforementioned changes will make any difference in our technical errors, we have instituted the aforementioned changes in our performance of ACBE examinations.
In conclusion, the majority of errors (82%) for all polyps were technical, and the majority of all missed polyps (78%) were in the proximal colon. These data suggest that radiologists might improve polyp detection rates at ACBE examinations by paying greater attention to technical aspects of an ACBE examination, especially in the proximal colon.
|
ADVANCES IN KNOWLEDGE
|
|---|
- Significantly more polyps are missed because of technical rather than perceptual errors.
- More missed polyps were in the proximal than in the distal colon, which was especially true for the missed polyps caused by technical errors.
- Polyp detection rates might improve if the quality of ACBE examinations can be improved, especially in the ascending colon.
|
FOOTNOTES
|
|---|
Abbreviations: ACBE = air-contrast barium enema • CI = confidence interval
Authors stated no financial relationship to disclose.
Author contributions: Guarantor of integrity of entire study, W.M.T.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; manuscript final version approval, all authors; literature research, W.M.T.; clinical studies, W.M.T.; statistical analysis, D.N.; and manuscript editing, W.M.T., W.L.F., E.K.P., D.N., V.H.S.L., B.W.B., D.C.R.
|
References
|
|---|
- Leinicke JL, Dodds WJ, Hogan WJ, Stewart ET. A comparison of colonoscopy and roentgenography for detecting polypoid lesions of the colon. Gastrointest Radiol 1977;2:125–128.[CrossRef][Medline]
- Johnson CD, Carlson HC, Taylor WF, Weiland LP. Barium enemas of carcinoma of the colon: sensitivity of double- and single-contrast studies. AJR Am J Roentgenol 1983;140:1143–1149.[Abstract/Free Full Text]
- Thoeni RF, Menuck L. Comparison of barium enema and colonoscopy in the detection of small colonic polyps. Radiology 1977;124:631–635.[Abstract]
- Ott DJ, Chen YM, Gefland DE, et al. Single-contrast vs double-contrast barium enema in the detection of colonic polyps. AJR Am J Roentgenol 1986;146:993–996.[Abstract/Free Full Text]
- Rex DK, Lehman GA, Lappas JC, Miller RE. Sensitivity of double-contrast barium study for left-colon polyps. Radiology 1986;158:69–72.[Abstract/Free Full Text]
- Levine MS, Rubesin SE. Diagnosis of colorectal neoplasms on double-contrast barium enemas: interpretive aspects. Semin Roentgenol 2000;35:357–369.[CrossRef][Medline]
- Saito Y, Slezak P, Rubio C. The diagnostic value of combining flexible sigmoidoscopy and double-contrast barium enema as a one-stage procedure. Gastrointest Radiol 1989;14:357–359.[CrossRef][Medline]
- Ott DJ. Accuracy of double-contrast barium enema in diagnosing colorectal polyps and cancer. Semin Roentgenol 2000;35:333–341.[CrossRef][Medline]
- Connolly DJ, Traill ZC, Reid HS, Copley SJ, Nolan DJ. The double contrast barium enema: a retrospective single center audit of the detection of colorectal carcinomas. Clin Radiol 2002;57:29–32.[CrossRef][Medline]
- de Zwart IM, Griffioen G, Shaw MP, Lamers CB, DeRoos A. Barium enema and endoscopy for the detection of colorectal neoplasia: sensitivity, specificity, complications and its determinants. Clin Radiol 2001;56:401–409.[CrossRef][Medline]
- Winawer SJ, Stewart ET, Zauber AG, et al. A comparison of colonoscopy and double-contrast barium enema for surveillance after polypectomy. N Engl J Med 2000;342:1766–1772.[Abstract/Free Full Text]
- Rockey DC, Paulson EK, Niedzwiecki D, et al. Analysis of air contrast barium enema, computed tomographic colonography, and colonoscopy: prospective comparison. Lancet 2005;365:305–311.[Medline]
- Ott DJ, Gefland DW, Ramquist NA. Causes of error in gastrointestinal radiology. Gastrointest Radiol 1980;5:99–105.[CrossRef][Medline]
- Kelvin FM, Gardiner R, Vas W, Stevenson GW. Colorectal carcinoma missed on double contrast barium study: a problem in perceptual. AJR Am J Roentgenol 1981;137:307–313.[Abstract/Free Full Text]
- Kewenter J, Jensen J, Boijsen M, Lycke G, Tylen U. Perceptual errors with double-contrast enema after a positive guaiac test. Gastrointest Radiol 1987;12:79–82.[CrossRef][Medline]
- Markus JB, Somers S, O'Malley BP, Stevenson GW. Double-contrast barium enema studies: effect of multiple reading on perceptual error. Radiology 1990;175:155–156.[Abstract/Free Full Text]
- Brady AP, Stevenson GW, Stevenson H. Colorectal cancer overlooked at barium enema examination and colonoscopy: a continuing perceptual problem. Radiology 1994;192:373–378.[Abstract/Free Full Text]
- McDonald S, Lyall P, Israel L, Coates R, Frizelle F. Why barium enemas fail to identify colorectal cancers. ANZ J Surg 2001;71:631–633.[CrossRef][Medline]
- Bolin S, Franzen L, Nilsson E, Sjodahl R. Carcinoma of the colon and rectum: tumors missed by radiologic examination in 61 patients. Cancer 1988;61:1999–2008.[CrossRef][Medline]
- Cady B, Stone MD, Wayne J. Continuing trends in the prevalence of right-sided lesions among colorectal carcinoma. Arch Surg 1993;128:505–509.[Abstract]
- Muto T, Bussey HJ, Moroson BC. The evaluation of cancer of the colon and rectum. Cancer 1975;36:2251–2270.[Medline]
- Bond JH. Clinical relevance of the small colorectal polyp. Endoscopy 2001;33:454–457.[CrossRef][Medline]
- Lieberman DA, Weiss DG, Bond JH, Ahnen DJ, Garewal H, Chejfec F. Use of colonoscopy to screen asymptomatic adults for colorectal cancer. N Engl J Med 2000;343:162–168.[Abstract/Free Full Text]
- Rex DK. Barium studies/virtual colonoscopy: the gastroenterologist's perspective. Gastrointest Endosc 2002;55(7 suppl):S33–S36. [Published correction appears in Gastrointest Endosc 2002;56:324.]
- Pickhardt PJ, Choi JR, Hwang I, et al. Computed tomographic virtual colonoscopy to screen for colorectal neoplasia in asymptomatic adults. N Engl J Med 2003;349:2191–2200.[Abstract/Free Full Text]
- Canon CL, Smith JK, Morgan DE, et al. Double reading of barium enemas: is it necessary? AJR Am J Roentgenol 2003;181:1607–1610.
This article has been cited by other articles:
|
|
|
|
 
H. Singh, S. Sethi, M. Raber, and L. A. Petersen
Errors in Cancer Diagnosis: Current Understanding and Future Directions
J. Clin. Oncol.,
November 1, 2007;
25(31):
5009 - 5018.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
T. Doshi, D. Rusinak, R. A. Halvorsen, D. C. Rockey, K. Suzuki, and A. H. Dachman
CT Colonography: False-Negative Interpretations
Radiology,
July 1, 2007;
244(1):
165 - 173.
[Abstract]
[Full Text]
[PDF]
|
|
|
TOP
ABSTRACT
INTRODUCTION
