| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Radiology, G1-228, Academic Medical Center, Meibergdreef 9,
Amsterdam 1105 AZ, The Netherlands.
2 Department of Radiology, Rijnstate Hospital, Arnhem 6815 AD, The
Netherlands.
3 Department of Colorectal Surgery, University Medical Center Utrecht, Utrecht
3584 CX, The Netherlands.
4 Department of Clinical Epidemiology and Biostatistics, Academic Medical
Center, Amsterdam 1105 AZ, The Netherlands.
5 Department of Colorectal Surgery, Academic Hospital Maastricht, Maastricht
6229 HX, The Netherlands.
Received September 1, 2004;
revised December 6, 2004;
Supported by The Netherlands Organization for Health Research and
Development ZON MW, grant 945-01-013, 2001.
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
SUBJECTS AND METHODS. Defecography was performed in 105 consecutive fecal-incontinent patients. Observers were classified by level of experience and their findings were compared with the findings of an expert radiologist. The quality of the expert radiologist's findings was evaluated by an intraobserver agreement procedure.
RESULTS. Intraobserver agreement was good to very good except for
anismus: incomplete evacuation after 30 sec (
, 0.55) and puborectalis
impression (, 0.54). Interobserver agreement for enterocele and
rectocele was good (, 0.66 for both) and for intussusception, fair
(, 0.29). Interobserver agreement for anismus: incomplete evacuation
after 30 sec was moderate (, 0.47), and for anismus: puborectalis
impression was fair (, 0.24). Agreement in grading of enterocele and
rectocele was good (, 0.64 and 0.72, respectively) and for
intussusception, fair (, 0.39). Agreement separated by experience level
was very good for rectocele (, 0.83) and grading of rectoceles
(, 0.83) and moderate for intussusception (, 0.44) at the most
experienced level. For enterocele and grading, experience level did not
influence the reproducibility.
CONCLUSION. Reproducibility for enterocele, anterior rectocele, and severity grading is good, but for intussusception is fair to moderate. For anismus, the diagnosis of incomplete evacuation after 30 sec is more reproducible than puborectalis impression. The level of experience seems to play a role in diagnosing anterior rectocele and its grading and in diagnosing intussusception.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
|
3 times a day), overflow incontinence, proctitis, soiling,
previous ilioanal or coloanal anastomosis; and patients with an anorectal
prolapse. Although intussusception grade 3 (for definitions of grades, see
under Evaluation Criteria) or enterocele grade 3 is only seen in anorectal
prolapse and may lead to fecal incontinence
[3], patients with these
disorders were not entered in our study. These exclusion criteria were used
because this study was part of a larger cohort study concerning the diagnostic
workup in patients with fecal incontinence in relation to the two most
commonly used treatment techniques (biofeedback and anterior anal repair). For
example, incontinence caused by rectal prolapse requires a different surgical
treatment [15]. The cohort study was approved by the medical ethics committee of the participating institutions, and all patients signed informed consent forms. No additional approval or informed consent was required for this study.
Study Design
Data from 105 consecutive patients were collected, with 35 patients per
center. In only one medical center, in addition to videotape recordings,
radiographs were obtained during rest, squeeze, straining, and evacuation.
Clinical assessments were made on the basis of a standardized scoring form.
The form included the following items: Was an enterocele present? If yes,
enter grade 1, 2, or 3. Was an anterior rectocele present? If yes, enter grade
1, 2, or 3. Was an intussusception present? If yes, enter grade 1, 2, or 3. In
anismus, Was evacuation complete after 30 sec? Was puborectalis muscle
nonrelaxing during evacuation?
|
Evaluation Criteria
Enterocele was defined as a herniation of a peritoneal sac with extension
of the small bowel along the ventral rectal wall and between the rectum and
the vagina [16]
(Fig. 1). Rectocele was defined
as an outward bulge of the (usually anterior) rectal wall
[16]. The presence of an
anterior rectocele was measured perpendicularly to the extrapolated line of
the rectal wall illustrated in Figure
2, using the maximum depth of the bulge
[17]
(Fig. 2). Anterior rectoceles
were scored as follows: grade 0 = no rectocele; grade 1 = rectocele smaller
than 2 cm; grade 2 = 2 to less than 4 cm; grade 3 = rectocele 4 cm or greater.
Intussusception was defined as a circular invagination of the proximal rectal
wall during defecation [16]
(Fig. 3). Grade 1 is a minimal
infolding of part of the rectal wall or circumferential infolding that remains
entirely intrarectal. Grade 2 is scored when the leading edge of the
intussusception is intraanal, into the orifice of the anal canal.
|
Interobserver and Intraobserver Reliability
In the three academic medical centers, the first assessment of defecography
was performed by 24 observers with different levels of experience. To evaluate
if there was a difference in assessment by level of experience, all observers
were classified into experience categories. Categories of assessment were
1–10 defecography examinations with supervision; 11–50
defecography examinations with supervision; 11–50 defecography
examinations without supervision; 51–100 defecography examinations with
supervision; 51–100 defecography examinations without supervision;
101–250 defecography examinations with supervision; more than 250
defecography examinations; and more than 1,000 defecography examinations.
Supervision was provided by radiologists with experience in at least 100
defecography examinations. In subsequent assessment by the expert radiologist,
the videotapes were presented randomly. All collected defecography data were
observed independently and blinded for results. To consider reproducibility,
the second assessment of all data was made on the basis of the same
standardized scoring form. Only clinical data considered relevant were
provided: these included sex, date of birth, diagnosis of fecal incontinence,
and previous operations. The second reviewer (i.e., the expert) was not
involved in the primary interpretations. To determine the reliability of the
interpretations of this observer, an intraobserver assessment was performed
after at least 8 weeks of 30 patients (10 patients per center) who were
representative of the disease spectrum of the interobserver study. The second
reviewer was blinded to the results of the first assessment and to previous
interpretations by others, and to the selection criteria of the defecography
examination included in the intraobserver assessment study. To avoid recall
bias and effects of ordering, the videotapes were randomly presented. All
assessments were done under the same conditions: the same videotape recorder
was used (type TVR 4510, Grundig) in the same room and during the same working
hours.
Defecography Procedure
The standard procedure in all three participating centers was as follows:
One to four hours before the examination, the patient was instructed to take
an oral contrast medium (a barium sulfate suspension diluted in water
[Micropaque, Guerbet; or Polibar, EZ-EM]). In two centers, a Microlax (sodium
lauryl sulfoacetate–sodium citrate–sorbitol, Pharmacia) microenema
was administered 20–120 min before the start of the procedure. The
investigation started with the patient in the left decubital position. A thick
barium paste (200–300 mL of barium sulfate prepared by the hospital
pharmacy or Evacu-Paste [EZ-EM]) was placed in the rectum by means of an
injection pistol. In only one center, Liquid Polibar (EZ-EM), combined with a
mix of Metamucil (psyllium, Procter & Gamble), was used. For opacifying
the vagina in women, 30 mL of amidotrizoic acid 50% gel solution and barium
sulfate solution (produced by the hospital pharmacy or by EZ-EM) was
introduced with a syringe, with a pediatric enema tip attached to it, into the
vagina. The perineum was visualized by using a catheter. Subsequently, the
entire imaging table was tilted upright 90° and the patient was seated on
a radiolucent, especially developed defecography chair. The dynamics of
defecation were recorded on videotape (type TVR 4510, Grundig). Videotape
recordings (and radiographs) were made first, with the patient at rest, when
the pelvic floor muscles were shown to be completely relaxed; next, during
instructed straining, at maximum distention of the anal canal; and finally,
during squeezing at the end of defecation, when the patient was asked to
contract the pelvic floor muscles maximally. The procedure was terminated when
rectal evacuation stopped or after several attempts of straining at
defecation. The complete procedure took approximately 15 min (room time).
To evaluate whether diagnostic judgment could have been influenced by technical differences between hospitals, a scoring form was developed. The form included seven items: screen acquisition, whether analog or digital; presence of time registration; imaging quality; presence of recordings at rest, during straining, and during squeezing; use of vaginal tampon; use of contrast medium (product, application, volume); and use of catheter marked in centimeters for quantification of abnormalities, corrected for magnification.
Statistical Analysis
Intra- and interobserver agreement was quantified using Cohen's kappa
statistic for nominal data (software program SPSS [version 11.5, Statistical
Package for the Social Sciences] for Windows [Microsoft]). For ordinal data, a
weighted kappa statistic was used (software program StatXact, version 3.0 for
Windows, Cytel Software).We considered a kappa value of 0.8 satisfactory and
used a cutoff value of 0.6 because a lower kappa value was not acceptable. To
calculate the sample size to a kappa value of 0.8 (distance from kappa to
limit 0.2) with a confidence interval of 95%, a one-tailed tested minimum of
25 patients per center was needed. To increase statistical power, we chose to
include 35 patients per center. Kappa values range from 0 to 1, where
0–0.2 is considered poor, 0.2–0.4 is fair, 0.4–0.6 is
moderate, 0.6–0.8 is good, and greater than 0.8 is very good agreement
[19].
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
|
|
Intraobserver Agreement
The intraobserver agreement, calculated to assess the quality of the
findings of the expert radiologist, was very good for the assessment of
enterocele (, 0.84; 95% confidence interval [CI], 0.51–1), good
for anterior rectocele (, 0.73; CI, 0.48–0.98) and
intussusception (, 0.71; CI, 0.41–1), and moderate for anismus:
incomplete evacuation after 30 sec (, 0.55; CI, 0.27–0.83) and
anismus: nonrelaxing puborectalis muscle during evacuation (, 0.54; CI,
0.16–0.91).
The intraobserver agreement for the grading of enterocele, anterior
rectocele, and intussusception was very good (for grading of enterocele,
weighted , 0.96; CI, 0.86–1) or good (for grading of anterior
rectocele, weighted , 0.74; CI, 0.60–0.88; and for
intussusception, , 0.74; CI, 0.47–1).
Level of Experience
To evaluate whether differences existed in assessment by level of
experience, all radiologists were classified into experience categories. When
analyzing the data, we reduced the experience groups from seven to three
categories to create comparable groups in size and in prevalence of
abnormalities. Therefore, the presented analyses are derived from the
following groups: little experience, n = 39 assessments (experience,
< 100 assessments with supervision); moderate experience, n = 31
assessments (experience, < 1,000 assessments without supervision); and very
experienced, n = 35 assessments (experience, > 1,000
assessments).
Interobserver Agreement
Interobserver agreement for enterocele was good (, 0.66; CI,
0.43–0.89). The results separated by level of experience show that
experience did not play a role in diagnosing enteroceles. The interobserver
agreement for anterior rectocele was also good (, 0.66; CI,
0.51–0.81). Only when the experience of the observer was more than 1,000
did the agreement become very good (, 0.83; CI, 0.65–1). For
intussusception, the interobserver agreement was fair (, 0.29; CI,
0.11–0.46). Only when the experience rose over 1,000 did the agreement
become moderate (, 0.44; CI, 0.16–0.72). Interobserver agreement
for anismus: incomplete evacuation after 30 sec was moderate (, 0.47;
CI, 0.30–0.64). No relationship was seen between the level of experience
and the level of agreement. For nonrelaxing puborectalis muscle, the
interobserver agreement was fair (, 0.24; CI, 0.01–0.47), as well
as for the different levels of experience. Results for agreement in the
grading of enterocele and anterior rectocele were good (weighted ,
0.64; CI, 0.41–0.87; and weighted , 0.72; CI, 0.61–0.83,
respectively) and for intussusception, fair (weighted , 0.39; CI,
0.16–0.61). The results separated by observers' level of experience for
the difference in grading of anterior rectocele showed good agreement
(weighted , 0.69; CI, 0.47–0.9) at the little-experience level
versus very good agreement (weighted , 0.83; CI, 0.74–0.93) at
the most experienced level. The results for level of experience for the
differences in grading of enterocele and intussusception were not calculated
because of too-low prevalences.
Technical Differences
Concerning the influence of technical differences in diagnostic
interpretation between hospitals, the conclusions were as follows: In only one
center, clock time was registered on the videotape images, facilitating
reproducibility studies. To register whether evacuation was after 30 sec, the
reproducibility in this center was moderate (, 0.54; CI,
0.26–0.82). Centers with no registration of time on the videotape images
showed fair agreement (, 0.39; CI, 0.12–0.67; and , 0.35;
CI, 0–0.72).
In one center, the use of vaginal contrast material was limited to a small
volume. An important reason for vaginal opacification is to detect enteroceles
[20]. The reproducibility for
enteroceles in this center was moderate (, 0.44; CI, 0–0.89)
versus good (, 0.72; CI, 0.36–1) and very good (, 0.84;
CI, 0.54–1) in the other centers with an adequate volume of the vaginal
contrast medium. To measure the size of a rectocele, a catheter marked in
centimeters can be helpful. In this study, in only one center the perineum was
visualized using a catheter marked in centimeters. No statistical significance
was found between the measurement of the severity of the rectocele in the
centers without a magnification factor and the measurement in the one center
with a magnification factor. The results of this study indicate that
centimeter marking does not contribute to an increase in agreement.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Limitations
No reference standard is available for defecography. To calculate
interobserver variability, we chose an expert radiologist with a high level of
experience (> 1,000 interpretations). To increase reliability, we
calculated stability in scoring using an intraobserver procedure. This study
shows that the chosen expert radiologist is very reliable for diagnosing
abnormalities such as enterocele, anterior rectocele, and intussusception and
their severity. However, for both aspects of anismus, the reliability of our
expert radiologist appeared moderate.
This study was a cohort study concerning the diagnostic workup in patients with fecal incontinence in relation to biofeedback and anterior anal repair. Patients with a rectal prolapse reported at medical history or found at physical examination were therefore excluded. Consequently, our patient population had a low prevalence of rectal prolapse. Only one rectal prolapse was not mentioned at medical history or detected at physical examination, and this is most likely related to the fact that a rectal prolapse may be transient and difficult to reproduce [4].
To our knowledge, no data are available about the reproducibility of diagnosing enteroceles and their corresponding grading or about diagnosing anismus on defecography. Few data are available about observer agreement for puborectalis impression, anterior rectoceles and their corresponding grading, and intussusception. Concerning the latter, the data give conflicting results [7, 8]. No reproducibility data of the differences in grading intussusception are available in the literature.
In our study, we did not specifically investigate a learning curve. By classifying the observers by their level of experience, we intended to study whether a little or a lot of experience may influence reproducibility.
Enterocele
In our study, the reproducibility of enterocele and its corresponding
grading were good. The prevalence of enteroceles was low (8.6%) in our
fecal-incontinent study population. Not uncommonly, an enterocele accompanies
a deep rectal intussusception or rectal prolapse
[21]. Possibly, if we had not
excluded rectal prolapse, the prevalence of enteroceles in our study
population would have been higher.
Anterior Rectocele
Klauser et al. [7], in a
group of 100 patients with defecation disorders, reported good total
agreement, which is comparable to our data. However, no kappa values are
available. Müller-Lissner et al.
[8] reported observer agreement
with kappa values of more than 0.4 in another group of 14 selected
defecography studies from patients with disordered defecation. Exact
interobserver agreements were not given. The prevalence of anterior rectoceles
in our study group is comparable to that in other studies
[8,
17]. Small rectoceles (
2
cm) are not of clinical relevance
[9,
18,
22]. A small rectocele may be
found in 25–77% of asymptomatic control subjects
[2]. Failure to recognize the
correct grading of such abnormalities can easily lead to over- or
underdiagnosis and inappropriate treatment.
Intussusception
Our data show, for both intussusception and its corresponding grade, fair
agreement at a low prevalence (10.5%). The reasons for fair agreement are,
among others, as follows. First, our reviewers had little experience in
diagnosing intussusception. Second, examinations were interpreted incorrectly
with respect to intussusception; a minimal infolding that disappears after the
bolus has passed is probably caused by a transient prolapse of the rectal wall
and should not be considered pathologic
[4]. Third, an intussusception
may appear at the end of the evacuation; when the examination is not
completely videorecorded, it is difficult to clearly diagnose the presence or
absence of an intussusception. And fourth, there is seldom doubt about
complete extraanal prolapse (intussusception grade 3); however, lesser grades
of prolapse are more difficult to diagnose, and prolapse of the anal canal
cushions and variations of fold patterns caused by some degree of asymmetry of
the rectum during its emptying, may confuse the interpretation
[2].
When an intraanal intussusception occurs (grade 2), difficulty in diagnosis arises in the way the rectum collapses. The collapse is not uniform, and it is easy to make false assumptions from viewing its collapse in only one plane. Usually an intussusception appears when the proximal part of the rectum telescopes into the distal part or even further (extraanal). As a result, the proximal part tapers off. Suspicion of an S-form rectum occurs when the proximal part remains circular. In that case it may be possible that the proximal and distal rectum shifted parallel to each other. In the lateral view, the rectum is compressed by the pelvic contents as it empties and flattened out, so that the folds are not seen unless viewed in the anteroposterior plane [18]. In our study, two patients were suspected of having an S-form rectum as diagnosed by the second observer (expert radiologist). There was disagreement with the first interpretation in which both patients were scored with the presence of intussusception and no additional anteroposterior view was made. Therefore, this pitfall could not be confirmed on an anteroposterior view.
Anismus
Although the exact incidence of anismus is unknown, it is a finding in
constipated and outlet-obstructed patients
[23] and apparently represents
an important etiologic factor
[2]. However, Voderholzer et
al. [24] showed that
paradoxical sphincter contraction is a common finding in healthy control
subjects and in patients with chronic constipation and fecal incontinence. Our
data show prevalences of 36.2% for incomplete evacuation and 19% for
puborectalis muscle impression. In our study, both intra- and interobserver
agreements for anismus, incomplete evacuation after 30 sec, were moderate.
With regard to incomplete evacuation
[7] or incomplete rectal
emptying [6,
8,
9] in populations of
asymptomatic subjects, reports of disordered defecation and constipation in
the literature confirm the variability of our findings.
Only Pfeifer et al. [9] showed inter- and intraobserver data of puborectalis impression. However, no kappa values are available and the agreement rates were low. Our data showed a fair interobserver agreement and a moderate intraobserver agreement for puborectalis impression. Emptying after 30 sec was shown to be the best indicator of anismus, with a positive predictive value of 90% [13, 18]. Therefore, we defined anismus in our study as incomplete evacuation after 30 sec or a nonrelaxing puborectalis muscle during evacuation.
We did not differentiate the primarily functional from the anatomic causes for incomplete evacuation [23, 25]. Neither did we distinguish other causes of delayed emptying. Consequently, data of incomplete evacuation by a nonrelaxing puborectalis muscle could not be distinguished from incomplete evacuation caused by a rectocele or by other factors (e.g., patient felt very uncomfortable, could not defecate because of lack of privacy, or was not instructed adequately). The fact that the observers were not able to distinguish this has led to a wider interpretation by observers of the idea of "anismus," probably with an overestimation of its prevalence. We assume that because of the absence of explicit diagnostic criteria, moderate intraobserver agreement and fair to moderate interobserver agreement were to be expected.
Conclusion
We conclude that the reproducibility for enterocele and anterior rectocele
and their grading is good and can be valuable in the diagnostic workup. With
regard to intussusception, agreement is fair and becomes moderate only when
the observer is very experienced. In the diagnostic workup, an observer who is
highly experienced in diagnosing intussusceptions is needed. For anismus, the
diagnosis of incomplete evacuation after 30 sec is more reproducible than
puborectalis impression. The level of experience seems to play a role in
diagnosing anterior rectocele and its grading and in diagnosing
intussusception.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  M. E. Lockhart, J. R. Fielding, H. E. Richter, L. Brubaker, C. G. Salomon, W. Ye, C. M. Hakim, C. Y. Wai, A. H. Stolpen, and A. M. Weber Reproducibility of Dynamic MR Imaging Pelvic Measurements: A Multi-institutional Study Radiology, November 1, 2008; 249(2): 534 - 540. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Karasick Defecography for the Diagnosis of Abnormalities in Patients with Fecal Incontinence Am. J. Roentgenol., June 1, 2006; 186(6): E20 - E20. [Full Text] [PDF]
|
|
|
|
|
|
A. C. Dobben and J. Stoker Reply Am. J. Roentgenol., June 1, 2006; 186(6): E20 - E20. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
