Original Research ajog.

org

OBSTETRICS
Comparing the diagnostic accuracy of 3 ultrasound
modalities for diagnosing obstetric anal sphincter
injuries
Annika Taithongchai, MBChB; Isabelle M. A. van Gruting, MD; Ingrid Volløyhaug, PhD; Linda P. Arendsen, MD;
Abdul H. Sultan, FRCOG; Ranee Thakar, FRCOG

BACKGROUND: The optimal imaging modality of obstetric anal on introital imaging and 0.70 and 0.69 on transperineal ultrasound im-
sphincter injuries needs to take into consideration convenience, avail- aging. Optimal cut-off for a significant internal anal sphincter defect was
ability, and ability to assess the sphincter morphologic condition. Endoanal
2 of 5 slices; sensitivity and specificity were 0.59 and 0.84 on introital
ultrasound imaging currently is regarded as the reference standard, but it imaging and 0.43 and 0.97 on transperineal ultrasound imaging. The area
is not widely available in obstetric units. Exoanal alternatives exist, such as under the curve for the diagnosis of external and internal anal sphincter
3-dimensional introital or transperineal ultrasound imaging, which are defects ranged from 0.70e0.74 (P<.001) for introital and transperineal
already readily available in most obstetrics and gynecology units. imaging. Positive predictive value for external and internal sphincter de-
OBJECTIVE: The primary objective was to evaluate the diagnostic fects ranged from 0.37e0.63, and negative predictive value ranged from
accuracy of 3-dimensional introital and 3-dimensional transperineal ul- 0.85e0.93 for introital and transperineal ultrasound imaging. Endoanal
trasound imaging compared with 3-dimensional endoanal ultrasound ultrasound imaging was the only modality for a defect to correlate with
imaging as the reference standard for the detection of anal sphincter symptoms; mean modified St. Mark’s score for a defect sphincter was 2.4
defects in women who sustained obstetric anal sphincter injuries. The (standard deviation, 4.1) and for an intact sphincter was 0.9 (standard
secondary objective was to correlate a diagnosis of anal sphincter defect deviation, 2.7; P<.01). Introital and transperineal ultrasound imaging
on imaging to symptoms of anal incontinence, and to assess patient were associated with less discomfort than endoanal ultrasound imaging.
discomfort that is experienced for each imaging modality. CONCLUSION: Endoanal ultrasound imaging remains the most ac-
STUDY DESIGN: A cross-sectional study was conducted of 250 curate diagnostic imaging modality. With low positive predictive values,
women who sustained obstetric anal sphincter injuries, all of whom un- introital and transperineal ultrasound imaging are not suitable for the
derwent 3-dimensional introital, transperineal, and endoanal ultrasound identification of sphincter defects; however, high negative predictive
imaging. Introital and transperineal ultrasound imaging were assessed values show a good ability to detect an intact sphincter. The optimal cut-off
with tomographic ultrasound imaging. All of the women completed a number of slices on tomographic ultrasound imaging for external and
validated modified St Mark’s Score and Visual Analogue Score for internal anal sphincters allows for standardization of a significant defect. In
discomfort. Optimal cut-off values for a significant defect on tomographic women with a history of obstetric anal sphincter injuries, introital and
ultrasound imaging were defined as those with the greatest sensitivity and transperineal ultrasound imagings are suitable to screen for an intact
specificity based on receiver operating characteristic curves with endoanal sphincter if endoanal ultrasound imaging is not available. When defects
ultrasound imaging as the reference standard. Diagnostic test charac- are found, women should then have endoanal ultrasound imaging to verify
teristics of introital and transperineal ultrasound imaging were calculated the diagnosis.
with the use of these optimal cut-offs.
RESULTS: Optimal cut-off for a significant external anal sphincter Key words: endoanal ultrasound imaging, introital ultrasound imaging,
defect was
3 of 7 slices; sensitivity and specificity were 0.65 and 0.75 obstetric anal sphincter injury (OASI), transperineal

O bstetric anal sphincter injury

(OASI) is one of the main causes
of anal incontinence because it occurs in
increasingly is associated with litigation.4
Endoanal ultrasound (EAUS) assess-
ment of the anal sphincters after OASI
research has been carried out with the
EAUS technique,1,3,10 currently regarded
as the reference standard.9,11 However, it
up to 35% of vaginal deliveries.1e3 It can has been shown to be useful, particularly requires a trained operator and expensive
impact women’s social, psychologic, and in counselling regarding mode of de- specialized equipment and is relatively
physical quality of life significantly and livery in a subsequent pregnancy.5e7 intrusive to the patient. Furthermore, it
Clinical examination is associated with may distend the muscular anatomy of the
poor detection of sphincter damage8; anal canal.12 Alternative exoanal ap-
Cite this article as: Taithongchai A, van Gruting IMA, ultrasound diagnostic accuracy is proaches include introital ultrasound
Volløyhaug I, et al. Comparing the diagnostic accuracy of better.9 imaging (IUS)12,13 and transperineal ul-
3 ultrasound modalities for diagnosing obstetric anal
There has been increasing interest in trasound imaging (TPUS),14e17 which
sphincter injuries. Am J Obstet Gynecol 2019;:.
the optimal imaging modality of OASIs, visualize the sphincter in an undisturbed
0002-9378/$36.00 taking into account convenience, avail- state. Moreover, the equipment for these
ª 2019 Elsevier Inc. All rights reserved.
https://doi.org/10.1016/j.ajog.2019.04.009 ability, and ability to assess the sphincter scans is readily available in most obstetric
morphologic condition. To date, most and gynecology units.

MONTH 2019 American Journal of Obstetrics & Gynecology 1.e1

Original Research OBSTETRICS ajog.org

anal canal, and the 3-dimensional vol-

AJOG at a Glance ume imaged the full length of the anal
Why was this study conducted? sphincter, starting proximally at the
Endoanal ultrasound imaging is regarded as the reference standard for imaging puborectalis muscle to the most distal
the anal sphincter morphologic condition. Alternatives that are more widely aspect of the subcutaneous level of the
available and accepted by patients include 3-dimensional introital and trans- external anal sphincter (EAS). IUS and
perineal ultrasound imaging. However, it is unknown whether they are accurate TPUS were performed with the GE
enough to replace endoanal ultrasound imaging. Voluson I system (GE Medical Systems,
Zipf, Austria). Both examinations were
Key findings performed at rest with the patient in the
Three-dimensional introital and transperineal ultrasound imaging provide suit- supine position. IUS was performed with
able screening tools for an intact anal sphincter but are not sensitive enough to a 3-dimensional 5e9 MHz endocavity
detect defects accurately. Onward referral for endoanal ultrasound imaging probe that was placed with low pressure
would be required if a defect is seen, because this remains the reference standard on the posterior fourchette in a vertical
and correlates best with symptoms. axis towards the anal sphincter complex.
TPUS was performed with a 3-
What does this add to what is known? dimensional 4e8.5 MHz curved array
The cut-off for an external anal sphincter defect on tomographic ultrasound abdominal probe. The probe was placed
imaging is
3 of 7 slices and for an internal anal sphincter defect is
2 of 7 slices, transversely on the perineum and in-
which provides standardization within the field for reporting and clinical use. clined to visualize the U shape of the
puborectalis muscle and angulated to
visualize the full length of the sphincter.
With ultrasound imaging advances, 3- October 2013 to August 2015. Women Both modalities had an acquisition angle
and 4-dimensional technology is also who were
18 years old and could read of 85 degrees. All ultrasound examina-
becoming increasingly popular. Advan- and understand English were eligible. tions were performed by an investigator
tages include multiplanar imaging, short The study was approved by the National (I.M.A.vG.) who is experienced in im-
examination times, and digital volume Research Ethics Service South East aging of the anal sphincter.
storage that allow for later reanalysis.16,17 London Committee (REC number 13/ The 3-dimensional image volumes of
The primary aim of this study was to LO/0232) and local research and devel- all 3 modalities were stored for off-line
evaluate the diagnostic test accuracy of opment department, IRAS project assessment. Image analysis was per-
3-dimensional IUS and 3-dimensional number 122213, and was registered in formed with the 3-dimensional viewing
TPUS compared with 3-dimensional clinicaltrials.gov (NCT 02655900). All program (version 5.19; BK Medical) for
EAUS as the reference standard for the study participants gave written informed EAUS and the 4-dimensional View
detection of anal sphincter defects in consent. software (version 10.2; GE Medical
women who sustained OASIs. The sec- Demographic data (age, body mass Systems) for IUS and TPUS by 3 in-
ondary aim was to correlate a diagnosis index, ethnicity, and parity) of each pa- dependent investigators who were
of anal sphincter defect on imaging with tient were collected. Each patient blinded to clinical and other imaging
symptoms of fecal incontinence and to completed a validated modified St. findings. Every investigator analyzed 30
assess patient discomfort that is experi- Mark’s score,18 which is a 24-point volumes of each modality; intraclass
enced for each imaging modality. scoring system for anal incontinence correlation analysis was performed to
symptoms, that accounts for fecal ur- assess agreement. After substantial
Materials and Methods gency, flatal incontinence, liquid and agreement was found, the remaining
This was a cross-sectional study of 250 solid fecal incontinence, impact on life- volumes were analyzed by a single
consecutive women who had sustained style, and the use of incontinence pads or investigator independently (A.T.
OASIs and undergone primary repairs of constipating medication. For each pa- analyzed EAUS; I.V. analyzed IUS;
the anal sphincter. They were recruited tient, all ultrasound assessments were L.P.A. analyzed TPUS).
from the perineal clinic of the tertiary performed on the same day. EAUS was The 3-dimensional EAUS volume was
urogynecology center of Croydon Uni- performed at rest with the use of the Pro- assessed by rating the sphincter complex
versity Hospital, United Kingdom. All focus 2202 or Flex-focus 500 ultrasound integrity at 3 levels starting after the U
the women were referred from within systems (BK Medical, Herlev, Denmark) shape of the puborectalis muscle: (1) the
Croydon University Hospital or the fitted with a 12e16 MHz anorectal deep level, up to where the EAS muscle
surrounding regions for assessment transducer (type 2052; focal point up to forms anteriorly in the midline, (2) the
6e12 weeks after delivery or were seen in 20 mm and focal range 5e45 mm, with superficial level, where the internal anal
a subsequent pregnancy for counselling 360-degree acquisition). With the pa- sphincter (IAS; hypoechoic) and EAS
regarding mode of delivery. Women tient lying in the left lateral position, the (hyperechoic) should be seen as com-
were recruited prospectively from probe was inserted along the axis of the plete rings, and (3) the subcutaneous

1.e2 American Journal of Obstetrics & Gynecology MONTH 2019

ajog.org OBSTETRICS Original Research

FIGURE 1
Intact anal sphincter

A, Three-dimensional endoanal ultrasound images of an intact sphincter with the external anal sphincter seen as the complete hyperechoic ring
encircling the complete hypoechoic ring of the internal anal sphincter. The puborectalis (1), deep (2), superficial (3), and subcutaneous (4) levels are
shown. B, Introital tomographic ultrasound imaging shows an intact external (slices 2e8) and internal (slices 2e6) anal sphincter. C, Transperineal
tomographic ultrasound imaging shows an intact external (slices 2e8) and internal (slices 2e6) anal sphincter.
Taithongchai et al. Ultrasound diagnosis of anal sphincter defects. Am J Obstet Gynecol 2019.

level of the EAS, where the IAS is no diagnosis. Any defect of
30 degrees of After each scan, women were asked to
longer present (Figure 1, A). partial or full thickness was measured for complete a visual analogue pain assess-
IUS and TPUS were both assessed by IAS and EAS and considered significant ment tool that ranged from 0 (no
tomographic ultrasound imaging (TUI). if present at
1 level17 (Figure 2, A). The discomfort) to 10 (severe discomfort) to
The TUI was adjusted to have 8 slices, same cut-off angle for EAS and IAS determine the discomfort of each
with the interslice interval varying ac- defect was also used for IUS and TPUS modality.
cording to individual sphincter length. for consistency in analysis (Figure 2, B The mean values for demographic
EAS (slices 2e8) and IAS (slices 2e6) and C). The EAS was evaluated both with variables were calculated. Interclass cor-
were evaluated in the same TUI. Slice 1 and without the subcutaneous level to relation analysis (absolute agreement be-
corresponds with the puborectalis level. assess whether diagnostic performance tween the mean of k raters, 2-way
Slice 2 was adjusted to be the most cra- would be affected by the inclusion of this random-effects model) between the 3
nial aspect of the EAS (deep level), where level. In addition, we looked at the deep investigators was performed for the
the muscle comes together in the level independently and calculated Norderval scores of 30 volumes for each
midline, with the superficial level ending sensitivity and specificity of IUS and imaging modality. Based on the 95%
at slice 6. Slices 7 and 8 covered the TPUS in detecting a defect at this level, confident interval of the interclass corre-
subcutaneous level (Figure 1, B and C). because this can be the most challenging lation estimate, values of <0.50 indicate
Defect sizes were measured for all 3 level to diagnose defects accurately in poor, of 0.50e0.75 indicate moderate, of
modalities with a 3-point angle, with the view of anatomic variations. 0.75e0.90 indicate good, and of >0.90
angle vertex in the middle of the anal Norderval score was calculated for all indicate excellent reliability.20,21
canal. The 3-dimensional EAUS volume 3 ultrasound modalities (Table 1), which Spearman’s rank correlation was used
was assessed in the deep, superficial, and accounted for the length, depth, and size to test correlation of Norderval scores
subcutaneous levels for defects, with of both EAS and IAS defects, with 0 be- between different imaging methods. The
manipulation of the cube in the axial, ing no defect and 7 being maximal sensitivity and specificity of IUS and
coronal, and sagittal planes to aid defect.19 TPUS was calculated with EAUS as the

MONTH 2019 American Journal of Obstetrics & Gynecology 1.e3

Original Research OBSTETRICS ajog.org

FIGURE 2
Defect anal sphincter

A, Superficial level of endoanal ultrasound imaging shows a defect in the external anal sphincter (angle) and internal anal sphincter (arrows). B, Su-
perficial level (slice 4) of introital tomographic ultrasound imaging shows a defect in the external anal sphincter (angle) and internal anal sphincter
(arrows). C, Superficial level (slice 4) of transperineal tomographic ultrasound imaging shows a defect in the external anal sphincter (angle) and internal
anal sphincter (arrows).
Taithongchai et al. Ultrasound diagnosis of anal sphincter defects. Am J Obstet Gynecol 2019.

reference standard, and receiver oper- mance was selected to define the best Sample size calculation was based on
ating characteristic curves were cut-off value for the detection of a sig- the assumption of a 30% prevalence of
created.22 The area under the receiver nificant EAS and IAS defect within a anal sphincter defects in the population
operating characteristic curve (AUC) population of known OASI. Diagnostic of interest.16 A sample size of 200 women
was calculated, in which 0.50 denotes no test characteristics for these cut-offs would provide 60 women with sphincter
clinical application as a test, 0.60e0.70 were calculated. Mann-Whitney U test defects. Sixty women with a sphincter
indicates poor, 0.70e0.80 indicates fair, was used to test the modified St. Mark’s defect would give a confidence interval
0.80e0.90 indicates good, and >0.90 Score against intact or defect sphincters (CI) of 0.50e0.75, assuming a true rate
indicates an excellent test.23 This for each imaging modality with the use of sensitivity of 0.64. One hundred and
included all levels for the EAS (slices of the new cut-off values. Mann- forty women with an intact sphincter
2e8) and IAS (slices 2e6) and subse- Whitney U test was used to assess the would provide a confidence interval of
quently excluded the subcutaneous level difference in visual analogue scores of 0.78e0.90, assuming a specificity of
of the EAS (slices 2e6). The number of discomfort for IUS and TPUS compared 0.85. Recruiting 250 women would allow
slices with the best diagnostic perfor- with EAUS. for unusable volumes for analysis or
incomplete data sets.
Statistical analysis was performed with
TABLE 1 IBM SPSS statistics software (version 23;
Norderval scoring system for anal sphincter defects19 IBM SPSS, Armonk, NY). A probability
value of <.05 was considered statistically
Score significant for all analyses.
Variable 0 1 2 3
External anal sphincter Results
Length of defect 50%
50% In total, 250 women were examined at a
Depth of defect None Partial Total and Total and median of 5 months (range, 1e137
90-degree >90-degree months) after the index (OASI) delivery,
radial extension radial extension of whom 88 were pregnant with a sub-
Internal anal sphincter sequent pregnancy at the time of exam-
Length of defect 50%
50% ination. Average age was 31.5 years
(standard deviation [SD], 4.5), mean
Depth of defect None Total and Total and
body mass index was 25.3 kg/m2 (SD, 4.7
90-degree >90-degree
radial extension radial extension kg/m2), and 183 of 248 women (74%)
Total score was calculated by the addition of the total length and depth score for both external and internal anal sphincter.
had a parity of 1. The main ethnic group
Taithongchai et al. Ultrasound diagnosis of anal sphincter defects. Am J Obstet Gynecol 2019. was white (116 women; 46%); other
ethnicities were Indian (55 women;

1.e4 American Journal of Obstetrics & Gynecology MONTH 2019

ajog.org OBSTETRICS Original Research

TABLE 2
Sensitivity and specificity per number of tomographic ultrasound imaging slices for detection of external and internal
anal sphincter defects using introital and transperineal ultrasound imaging compared with endoanal ultrasound
imaging as the reference standard using receiver operator characteristic curves
Tomographic
ultrasound Area under 95% Confidence
Variable imaging slices, n Sensitivity Specificity the curve interval P value
External anal sphincter
Without subcutaneous level
included
Introital ultrasound imaging 1 0.76 0.63 0.70 0.63e0.77 <.001
2 0.68 0.69
3 0.53 0.76
4 0.41 0.81
5 0.18 0.90
Transperineal ultrasound 1 0.69 0.63 0.68 0.61e0.76 <.001
imaging
2 0.66 0.65
3 0.64 0.69
4 0.61 0.73
5 0.54 0.78
With subcutaneous level included
Introital ultrasound imaging 1 0.82 0.61 0.74 0.66e0.81 <.001
2 0.77 0.65
3 0.65 0.75
4 0.55 0.80
5 0.34 0.86
6 0.23 0.89
7 0.13 0.93
Transperineal ultrasound 1 0.73 0.63 0.72 0.64e0.79 <.001
imaging
2 0.73 0.66
3 0.70 0.69
4 0.66 0.73
5 0.61 0.76
6 0.49 0.82
7 0.37 0.87
Internal anal sphincter
Introital ultrasound imaging 1 0.63 0.81 0.72 0.62e0.83 <.001
2 0.59 0.84
3 0.47 0.88
4 0.19 0.94
5 0.30 0.99
Taithongchai et al. Ultrasound diagnosis of anal sphincter defects. Am J Obstet Gynecol 2019. (continued)

MONTH 2019 American Journal of Obstetrics & Gynecology 1.e5

Original Research OBSTETRICS ajog.org

TABLE 2
Sensitivity and specificity per number of tomographic ultrasound imaging slices for detection of external and internal
anal sphincter defects using introital and transperineal ultrasound imaging compared with endoanal ultrasound
imaging as the reference standard using receiver operator characteristic curves (continued)
Tomographic
ultrasound Area under 95% Confidence
Variable imaging slices, n Sensitivity Specificity the curve interval P value
Transperineal ultrasound imaging 1 0.43 0.96 0.70 0.57e0.82 .001
2 0.43 0.97
3 0.39 0.98
4 0.29 0.98
5 0.21 0.99
Any external anal sphincter and/or
internal anal sphincter defect in the
deep level (slice 2)
Introital ultrasound imaging 0.36 0.84 0.60 0.52e0.69 .02
Transperineal ultrasound imaging 0.64 0.69 0.67 0.59e0.75 <.001
Taithongchai et al. Ultrasound diagnosis of anal sphincter defects. Am J Obstet Gynecol 2019.

22%), other Asian (35 women; 14%), and 0.75 on IUS and 0.70 and 0.69 on intact or defect sphincter for either IUS
black (27 women; 11%), and mixed or TPUS. Optimal cut-off for significant or TPUS: 1.12.5 vs 1.83.8 (P¼.40)
unknown ethnicity (17 women; 7%). IAS defect was
2 of 5 slices; sensi- and 1.12.6 vs 1.63.5 (P¼.17),
The interclass correlation of the Nor- tivity and specificity were 0.59 and respectively.
derval score among the 3 analysts for 30 0.84 on IUS and 0.43 and 0.97 on Discomfort scores of the imaging
volumes showed a significant correla- TPUS. The receiver operating charac- technique were documented for 238 of
tion: 0.83 (95% CI, 0.70e0.92; P<.01) teristic curves for diagnosis of EAS and 250 patients. The mean discomfort
for EAUS, 0.76 (95% CI, 0.57e0.88; IAS defects on IUS and TPUS are scores for IUS (1.01.8) and TPUS
P<.01) for IUS, and 0.86 (95% CI, presented in Figure 3, A and B. The (0.01.3) were significantly lower when
0.74e0.93; P<.01) for TPUS. AUC for EAS defects (with subcu- compared with EAUS (4.02.3; both
A defect of
30 degrees in
1 level taneous level included) on IUS was P<.001).
was present in 79 of 248 women (32%) 0.74 (95% CI, 0.66e0.81; P<.001) and
on EAUS, in 134 of 246 women (55%) on TPUS was 0.72 (95% CI, Comment:
on IUS, and on 118 of 243 women (49%) 0.64e0.79; P<.001). The AUC for IAS The study aim was to assess diagnostic
on TPUS. Two volumes were missing for defects on IUS was 0.72 (95% CI, test accuracy of 3-dimensional IUS and
different women, and not all volumes 0.62e0.83; P<.001) and on TPUS was TPUS compared with 3-dimensional
had complete data to fully assess the EAS 0.70 (95% CI, 0.57e0.82; P¼.001). EAUS as the reference standard for the
or IAS for IUS or TPUS. The mean (SD) Both IUS and TPUS had greater AUC detection of anal sphincter defects in
Norderval scores for EAUS, IUS, and for EAS defects when the subcutaneous women who sustained OASIs. Optimal
TPUS were 1.22.0, 1.81.9, and level was included, although not sta- cut-off for a significant EAS defect was
1.11.5, respectively. The correlation of tistically significant. Table 3 shows a
3 of 7 slices and for significant IAS
Norderval scores was moderate; between summary of the diagnostic test char- defect was
2 of 5 slices on TUI. Both
EAUS and IUS, it was rs¼0.42 (P<.001), acteristics of both IUS and TPUS with IUS and TPUS had AUC that showed fair
and between EAUS and TPUS, it was the use of the optimal cut-off values. ability to diagnose EAS and IAS defects.
rs¼0.47 (P<.001). Sixty-one women had anal inconti- Both had high negative predictive value,
The AUC for IUS and TPUS and the nence symptoms, of whom 30 had a which suggests good ability to identify an
sensitivities and specificities for each defect on EAUS. EAUS imaging was the intact sphincter; but low positive pre-
number of TUI slices for diagnosing only modality for a defect to correlate dictive value, which indicates poor
EAS and IAS defects are indicated in with the modified St. Mark’s Score; mean detection of sphincter defects. EAUS was
Table 2. The number of slices with the score was 2.44.1 for defect sphincter the only modality to correlate with anal
best diagnostic performance for a sig- and 0.92.7 for intact sphincter incontinence symptoms. IUS and TPUS
nificant EAS defect was
3 of 7 slices; (P<.01). There was no difference in were associated with less discomfort
sensitivity and specificity were 0.65 mean modified St. Mark’s Score between than EAUS.

1.e6 American Journal of Obstetrics & Gynecology MONTH 2019

ajog.org OBSTETRICS Original Research

The development of optimal cut-off

FIGURE 3
values for a significant EAS and IAS
Receiver operator characteristic curves
defect on TUI allows for standardized
reporting in clinical and research set-
tings. Although a cut-off of
4 of 6
slices on TUI has been validated
against symptoms in urogynecology
patients,29 we are aiming for a cut-off
to detect a sphincter defect in women
known to have OASI. We know the
majority of women with OASI will not
have symptoms until later in life, if at
all; therefore, a defect can be signifi-
cant, even if not associated with
symptoms. There has been debate
about whether the subcutaneous
component of the EAS should play a
part in defining a defect.29 We found
that its inclusion led towards improved
diagnostic performance, although not
statistically significant. The subcu-
taneous part of the EAS contributes to
a significant proportion of the
sphincter and thus should be included.
In the deep level, it was more difficult
to diagnose a defect accurately on IUS
or TPUS compared with EAUS, which
was indicated by lower AUC for this
level when isolated. This demonstrates
the poor ability of distinguishing a
defect from anatomic variation at this
level.
We believe that this is the most
A, Receiver operator characteristic curves for 3-dimensional introital tomographic ultrasound im-
adequately powered study to date to
aging (left) and 3-dimensional transperineal tomographic ultrasound imaging (right) for diagnosis of
external anal sphincter defects (with inclusion of subcutaneous level). B, Receiver operator char- compare these three 3-dimensional
acteristic curves for 3-dimensional introital tomographic ultrasound imaging (left) and 3-dimensional imaging modalities to be able to
transperineal tomographic ultrasound imaging (right) for diagnosis of internal anal sphincter defects. draw firm conclusions. We also used
IUS, introital ultrasound imaging; ROC, receiver operator characteristic; TPUS, transperineal ultrasound imaging. validated scoring systems for symp-
Taithongchai et al. Ultrasound diagnosis of anal sphincter defects. Am J Obstet Gynecol 2019. toms and scan findings. In addition,
the study population is generalizable,
and there is low risk of detection bias
because all examiners were blinded to
When first described, IUS suggested compared with 2-dimensional. The only other scan results and clinical history.
good correlation with EAUS.24 Later, a other study that compared all 3- However, the use of 3 examiners, even
larger study showed, in fact, low sensi- dimensional modalities had 55 patients; with good interclass correlation, may
tivity, with high specificity, 25 which were they substantiated that 3-dimensional have introduced bias. We acknowledge
comparable with our findings. Two- technology with TPUS improves that the quality of the scanning ma-
dimensional IUS and TPUS have been the test accuracy compared with chine for EAUS was superior to that
compared with EAUS in a large study 2-dimensional and that 3-dimensional used for IUS and TPUS. It is possible
that concluded that 2-dimensional TPUS has potential in screening26 that that accuracy could be improved with
TPUS could identify an intact is similar to other studies.27,28 With our a new generation scanner. We also
sphincter, but lacked sensitivity to detect significantly larger study, we confidently acknowledge the heterogeneity of this
defects.16 Our study found higher agree that (with AUC values of study population, because some
sensitivity values using 3-dimensional, 0.70e0.74) 3-dimensional IUS and women were pregnant. In addition,
which suggests that 3-dimensional TPUS are not suitable diagnostic tests to there was a large range in follow-up
imaging can offer improved detection substitute EAUS. time. Although presence of anal

MONTH 2019 American Journal of Obstetrics & Gynecology 1.e7

Original Research OBSTETRICS ajog.org

TABLE 3
Diagnostic test characteristics of introital and transperineal ultrasound imaging for diagnosis of external and internal
anal sphincter defects with the use of endoanal ultrasound imaging as reference standard in 250 women who
sustained obstetric anal sphincter injury
Positive Negative Positive Negative
Anal Imaging Defect,a predictive predictive likelihood likelihood
sphincter modality n/N (%) Sensitivity Specificity value value ratio ratio
External Endoanal ultrasound 73/248 (29.4) N/A N/A N/A N/A N/A N/A
imaging (N¼248)b
Introital ultrasound 80/223c (35.9) 0.65 0.75 0.50 0.86 2.60 0.47
imaging (N¼248)b
Transperineal 96/227d (42.3) 0.70 0.69 0.51 0.85 2.26 0.43
ultrasound
imaging (N¼246)b
Internal Endoanal ultrasound 34/248 (13.7) N/A N/A N/A N/A N/A N/A
imaging (N¼248)b
Introital ultrasound 52/241c (21.6) 0.59 0.84 0.63 0.93 3.69 0.49
imaging (N¼248)b
Transperineal 19/238d (8.0) 0.43 0.97 0.37 0.93 14.33 0.59
ultrasound
imaging (N¼246)b
N/A, not applicable.
a
With the use of the cut off values of
1 level for external and internal anal sphincter on endoanal ultrasound,
3/7 slices for external anal sphincter or
2/5 slices for internal anal sphincter on
introital ultrasound imaging/transperineal ultrasound imaging; b 2 volumes for different women were missing; c 22 volumes had incomplete data to assess external anal sphincter and or internal
anal sphincter fully; d 23 volumes had incomplete data to assess external anal sphincter and or internal anal sphincter fully.
Taithongchai et al. Ultrasound diagnosis of anal sphincter defects. Am J Obstet Gynecol 2019.

incontinence symptoms may change This study was carried out in a cohort low positive predictive value, women
with time and pregnancy status, these with a high prevalence of sphincter de- with defects on IUS or TPUS would need
2 confounders have no effect on fects; therefore, the negative predictive referral for EAUS to verify the
sphincter defects or morphologic value would be expected to be even diagnosis. n
condition.3,30 Therefore, because all higher in an unselected cohort of post-
scans were performed on the same day partum women. This would support Acknowledgment
for each woman, the diagnostic accu- their use to screen for an intact sphincter Thanks to Mr M. Naidu for setting up the study;
racy of each modality or correlation on labor ward immediately after de- he is employed at Croydon University Hospital, is
funded by National Health Service, and received
with symptoms should not be livery. Although likely to be highly
no compensation.
affected. accepted by patients and reduce unde-
Patient acceptability should be tected OASI, it would require widespread References
considered. As expected, the less intru- training of obstetricians, instead of 1. Sorensen M, Tetzschner T, Rasmussen OO,
sive nature of IUS and TPUS led to improving examination skills. Likely, the Bjarnesen J, Christiansen J. Sphincter rupture in
reduced discomfort. The IUS probe re- most appropriate place for these mo- childbirth. Br J Surg 1993;80:392–4.
quires pressure on the posterior four- dalities is in the antenatal setting, 2. Sultan AH, Kamm MA, Hudson CN,
Bartram CI. Third degree obstetric anal
chette; this and hence tissue proximity assessing women in subsequent preg- sphincter tears: risk factors and outcome of
could result in reduced visibility of distal nancies after OASI to advise mode of primary repair. BMJ 1994;308:887–91.
defects at the 12 o’clock position. This delivery. 3. Sultan AH, Kamm MA, Hudson CN,
may support the use of TPUS over IUS. In conclusion, 3-dimensional EAUS Thomas JM, Bartram CI. Anal sphincter disrup-
When applicability is evaluated, cost remains the most accurate method for tion during vaginal delivery. N Engl J Med
1993;329:1905–11.
and equipment availability are impor- the diagnosis of anal sphincter defects 4. Leigh RJ, Turnberg LA. Faecal incontinence:
tant. IUS and TPUS probes already are and correlats best with symptoms, hence the unvoiced symptom. Lancet 1982;1:
used widely by obstetricians and gyne- cannot be substituted by IUS or TPUS. 1349–51.
cologists and provide a cheaper alterna- High negative predictive value indicates 5. Scheer I, Thakar R, Sultan AH. Mode of de-
tive to the more specialized endoanal that, in women with a history of OASI, livery after previous obstetric anal sphincter in-
juries (OASIS): a reappraisal? Int Urogynecol J
probe. However, one must appreciate IUS and TPUS are useful for screening Pelvic Floor Dysfunct 2009;20:1095–101.
that the interpretation of all techniques an intact sphincter in situations in which 6. Jordan PA, Naidu M, Thakar R, Sultan AH.
requires training and expertise. EAUS is not available. However, with a Effect of subsequent vaginal delivery on bowel

1.e8 American Journal of Obstetrics & Gynecology MONTH 2019

ajog.org OBSTETRICS Original Research

symptoms and anorectal function in women transperineal ultrasound for detecting anal 27. Garcia-Majido JA, Palomino LG, Palacin AF,
who sustained a previous obstetric anal sphincter defects: the PREDICT study. Clin Sainz-Bueno JA. [Applicability of 3/4D trans-
sphincter injury. Int Urogynecol J 2018;29: Radiol 2011;66:597–604. perineal ultrasound for the diagnosis of anal
1579–88. 17. Dietz HP. Exoanal imaging of the anal sphincter injury during the immediate post-
7. Fitzpatrick M, Cassidy M, Barassaud ML, sphincter. J Ultrasound Med 2018;37: partum.]. Cir Cir 2017;85:80–6.
et al. Does anal sphincter injury preclude sub- 263–80. 28. Oom DM, West RL, Schouten WR,
sequent vaginal delivery? Eur J Obstet Gynecol 18. Roos AM, Sultan AH, Thakar R. St. Mark’s Steensma AB. Detection of anal sphincter
Reprod Biol 2016;198:30–4. incontinence score for assessment of anal in- defects in female patients with fecal inconti-
8. Roos AM, Abdool Z, Thakar R, Sultan AH. continence following obstetric anal sphincter nence: a comparison of 3-dimensional trans-
Predicting anal sphincter defects: the value of injuries (OASIS). Int Urogynecol J Pelvic Floor perineal ultrasound and 2-dimensional
clinical examination and manometry. Int Urogy- Dysfunct 2009;20:407–10. endoanal ultrasound. Dis Colon Rectum
necol J 2012;23:755–63. 19. Norderval S, Markskog A, Røssaak K, 2012;55:646–52.
9. Abdool Z, Sultan AH, Thakar R. Ultrasound et al. Correlation between anal sphincter 29. Guzman Rojas RA, Kamisan AI, Shek KL,
imaging of the anal sphincter complex: a review. defects and anal incontinence following ob- Dietz HP. Anal sphincter trauma and anal in-
Br J Radiol 2012;85:865–75. stetric sphincter tears: assessment using continence in urogynaecological patients. Ul-
10. Andrews V, Sultan AH, Thakar R, Jones PW. scoring systems for sonographic classifica- trasound Obstet Gynecol 2015;46:363–6.
Occult anal sphincter injuries: myth or reality? tion of defects. Ultrasound Obstet Gynecol 30. Sultan AH, Kamm MA, Hudson CN,
BJOG 2006;113:195–200. 2008;31:78. Bartram CI. Effect of pregnancy on anal
11. Sultan AH, Kamm MA, Talbot IC, 20. Koo TK, Li MY. A guideline of selecting and sphincter morphology and function. Int J Colo-
Nicholls RJ, Bartram CI. Anal endosonography reporting intraclass correlation coefficients for rectal Dis 1993;8:206–9.
for identifying external sphincter defects reliability research. J Chiropractic Medicine
confirmed histologically. Br J Surg 1994;81: 2016;15:155–63.
463–5. 21. Gisev N, Bell JS, Chen TF. Interrater agree- Author and article information
12. Sultan AH, Loder PB, Bartram CI, ment and interrater reliability: key concepts, From the Department of Obstetrics and Gynecology,
Kamm MA, Hudson CN. Vaginal endo- approaches, and applications. Res Social Adm Croydon University Hospital, London, United Kingdom
sonography: new approach to image the un- Pharm 2013;9:330–8. (Drs Taithongchai, Van Gruting, Arendsen, Sultan, and
disturbed anal sphincter. Dis Colon Rectum 22. Griner PF, Mayewski RJ, Mushlin AI, Thakar); and the Department of Clinical and Molecular
1994;37:1296–9. Greenland P. Selection and interpretation of Medicine, Norwegian University of Science and Tech-
13. Meriwether KV, Hall RJ, Leeman LM, diagnostic tests and procedures. Ann Intern nology, and the Department of Obstetrics and Gynecol-
Migliaccio L, Qualls C, Rogers RG. Anal Med 1981;94:555–600. ogy, Trondheim University Hospital, Trondheim, Norway
sphincter complex: 2D and 3D endoanal and 23. Tape TG. The area under an ROC curve. (Dr Volløyhaug).
translabial ultrasound measurement variation in Available at: http://gim.unmc.edu/dxtests/roc3. Received Dec. 19, 2018; revised March 18, 2019;
normal postpartum measurements. Int Urogy- htm. Accessed September 20, 2018. accepted April 8, 2019.
necol J 2015;26:511–7. 24. Sultan AH, Nicholls RJ, Kamm MA, Supported by the Mayday Charity Fund.
14. Valsky DV, Cohen SM, Lipscheutz M, Hudson CN, Beynon J, Bartram CI. Anal The funding agent had no involvement in study design,
Hochner-Ceilnikier D, Yagel S. Three-dimen- endosonography and correlation with in vitro collection of data, writing of the report or decision to
sional transperineal ultrasound findings associ- and in vivo anatomy. Br J Surg 1993;80: submit the article.
ated with anal incontinence after intrapartum 508–11. Clinical Trial Registration: Registered in clinicaltrials.
sphincter tears in primiparous women. Ultra- 25. Frudinger A, Bartram CI, Kamm MA. gov; NCT 02655900; https://clinicaltrials.gov/ct2/show/
sound Obstet Gynecol 2012;39:83–90. Transvaginal versus anal endosonography for NCT02655900?term¼02655900&rank¼1.
15. Weinstein MM, Pretorius DH, Jung SA, detecting damage to the anal sphincter. Am J The authors report no conflict of interest.
Hager CW, Mittal RK. Transperineal three- Roentgenol 1997;168:1435–8. Presented at the British Society of Urogynecology
dimensional ultrasound imaging for detection 26. Ros C, Martinez-Franco E, Wozniak MM, (BSUG) Annual Meeting, London, UK, November 9, 2018,
of anatomic defects in the anal sphincter com- et al. Postpartum two- and three- dimensional and at the International Urogynecology Association (IUGA)
plex muscles. Clin Gastroenterol Hepatol ultrasound evaluation of anal sphincter com- Meeting 2016 41st Annual Meeting, Cape Town, South
2009;7:205–11. plex in women with obstetric anal sphincter Africa, August 2e6, 2016.
16. Roos AM, Abdool Z, Sultan AH, Thakar R. injury. Ultrasound Obstet Gynecol 2017;49: Corresponding author: Ranee Thakar, FRCOG. ranee.
The diagnostic accuracy of endovaginal and 508–14. [email protected]

MONTH 2019 American Journal of Obstetrics & Gynecology 1.e9