Correspondence to: Prof. H. P. Dietz, Sydney Medical School Nepean, University of Sydney, Nepean Hospital, Penrith NSW 2750, Australia (e-mail: firstname.lastname@example.org)
Levator avulsion is common after vaginal delivery and is strongly associated with prolapse and prolapse recurrence. The aim of this study was to compare assessment by digital palpation and two ultrasound methods, one using rendered volumes and the other multislice imaging, for the diagnosis of levator avulsion.
We retrospectively analyzed randomly identified datasets of 266 women seen at a tertiary urogynecology unit. Each patient had undergone an interview, vaginal examination and 3D/4D translabial ultrasound examination. Analysis of the retrieved ultrasound volumes was performed offline, with the operator blinded to all clinical data, using two techniques: assessment of rendered volumes and evaluation on multislice imaging. We tested agreement between the three methods and the association of each method's results with symptoms and signs of pelvic organ prolapse.
Agreement between the findings on palpation and the two ultrasound methods with regard to diagnosis of levator avulsion ranged from 80% to 87% (Cohen's kappa, 0.35−0.56). The findings for all methods were significantly associated with symptoms, signs and ultrasound findings of pelvic organ prolapse (P = 0.007 to < 0.001), with no single method appearing superior to the others.
Depending on the availability of local expertise and equipment, any of the three methods tested in this study may be used to document avulsion of the puborectalis muscle.
Levator avulsion injury is a form of maternal birth trauma that is known to occur in 10−35% of women after a first vaginal delivery, with the use of forceps being the main risk factor[1-6]. This trauma has substantial implications for pelvic organ support. It is strongly associated with female pelvic organ prolapse, especially cystocele and uterine prolapse, and with recurrence after pelvic reconstructive surgery[8-12]. The latter implies that examination for levator trauma is very likely to become a standard part of the diagnostic work-up of patients presenting with female pelvic organ prolapse.
Several different diagnostic methods have been developed to diagnose levator avulsion. Vaginal palpation[13-15] is simple, with moderate validity and reproducibility, but requires substantial training. 3D/4D ultrasound imaging is probably more valid and reproducible but requires expensive equipment. Magnetic resonance imaging (MRI) was the first modality proposed for the diagnosis of levator trauma but involves by far the highest cost of the three methods, along with difficulty of access and difficulty in obtaining truly dynamic imaging. Ultrasound is much simpler and more readily accessed, and has the advantage of true real-time dynamic imaging capability. For 3D/4D pelvic floor ultrasound this applies to any user-definable plane, including oblique axial planes which define the plane of minimal hiatal dimensions. Two-dimensional ultrasound imaging requires either side-firing endoprobes for access to the axial plane (which are rarely available) or oblique sagittal planes obtained using abdominal curved arrays (which suffer from the absence of a well-defined point of reference). Most authors in this field seem to employ 3D/4D translabial ultrasound for the diagnosis of levator trauma, using rendered volumes placed at the level of the plane of minimal hiatal dimensions or else tomographic or multislice imaging which is now available on virtually all modern 3D/4D ultrasound systems used for perinatal imaging. While rendered volume imaging, to our knowledge, has never been compared to other methods, there is some evidence that tomographic or multislice imaging compares favorably with MRI in the diagnosis of levator trauma.
In order to support the clinical establishment of simple, inexpensive and readily tolerated diagnostic procedures for levator trauma we compared the three methods that currently seem to offer the greatest potential clinical utility: vaginal palpation and sonographic diagnosis using either rendered volume imaging or else tomographic or multislice imaging.
PATIENTS AND METHODS
We retrospectively analyzed randomly identified datasets of 266 women who had presented for urodynamic testing in a tertiary urogynecology unit between March 2006 and November 2008. Each patient had undergone an interview, a vaginal examination, multichannel urodynamic testing (Neomedix Acquidata system, Neomedix, Ryde, Australia) and translabial ultrasound (Voluson 730 Expert, GE Kretz Ultrasound, Zipf, Austria). The interview (in-house, non-validated) queried stress and urge incontinence, frequency, nocturia (all according to definitions of the International Continence Society (ICS)), symptoms of voiding dysfunction (hesitancy, poor stream, stop-start voiding) and symptoms of prolapse (vaginal lump or bulge, or dragging sensation). During vaginal examination, prolapse was graded using the Pelvic Organ Prolapse Quantification system (POP-Q) approved by the ICS; levator strength using modified Oxford grading and integrity were also assessed by palpation at the original examination (Figure 1). Ultrasound imaging analysis was performed over 1 year later by a junior trainee (F.M.) with fewer than 3 months of experience in pelvic floor ultrasound, using proprietary software (Kretz 4D View V 5.0, GE Kretz Ultrasound) on a PC, with the operator blinded to all clinical data. In order to exclude any influence of previous pelvic floor surgery, we also analyzed a subset of women without previous surgery for incontinence and prolapse (n = 201).
Diagnosis by palpation was based on a method developed by the senior author and validated in our unit[15, 24]. In short, the index finger of the dominant hand is placed parallel to the urethra, with the fingertip at the bladder neck. The fingertip is then turned and retracted towards the inferior pubic ramus, and the patient is asked to contract the pelvic floor muscles. One should be able to palpate contractile tissue on the inferior pubic ramus, and the gap between urethra and muscle should be about the breadth of one finger. If there is no contractile tissue palpated on the pubic ramus there will be room for two or more fingers between urethra and lateral pelvic sidewall, and a diagnosis of avulsion is made. Partial trauma may be recognized by palpation as thinning of the entire muscle, slit-like defects or defects of the most inferior or superior aspects of the puborectalis muscle (Figure 1). It is understood that mapping of defects as shown in Figure 1 is less repeatable, and the repeatability of digital findings will depend crucially on training. In a previous study we obtained a Cohen's kappa of 0.41 for agreement in the diagnosis of levator defects between the senior author and a subspecialty trainee with 3 months of training, signifying moderate agreement, and a Cohen's kappa of 0.495 for agreement with the results of tomographic ultrasound imaging TUI. The current series comprises the datasets of patients seen subsequently by the senior author.
Rendered volumes obtained on maximal pelvic floor contraction were set at a thickness between 1.5 and 2.5 cm, with the plane of minimal dimensions included in the region of interest which delineates the rendered volume, with the rendering direction set from caudal to cranial (Figure 2). Thickness was adjusted to optimize visualization of insertion of the puborectalis muscle, which depends on a number of factors including the patient's muscle mass, image quality and quality of a contraction. The standard rendering setting was surface/minimum 80/20, with transparency set at 50. An avulsion was diagnosed on analysis of rendered volumes if insertion of the puborectalis muscle on the inferior pubic ramus was clearly abnormal. This did not require complete discontinuity between sidewall and muscle. Figure 2 shows typical findings in a patient with a unilateral right-sided avulsion. As there were no reproducibility data available for this method, we performed an assessment of interobserver reproducibility between senior author and trainee (H.P.D. and F.M.) in a series of 43 patients. We used archived volume datasets, with at least 12 individual volumes per patient for each of the examiners to choose from.
TUI was performed as previously described using volume datasets obtained at maximum pelvic floor contraction, to produce a set of eight slices in the axial plane at intervals of 2.5 mm, from 5 mm caudad to 12.5 mm cephalad of the plane of minimal hiatal dimensions (Figure 3). This method is highly repeatable and correlates well with the diagnosis of avulsion on MRI. An avulsion was rated as present if the plane of minimal dimensions, as well as the two slices cephalad of that plane, showed abnormal insertions.
To validate the three methods used to diagnose avulsion we determined maximum organ descent in volume datasets obtained on Valsalva, using a previously published methodology.
This project was approved as an extension of a previously approved parent project (SWAHS HREC 05–029). We used Cohen's kappa to determine agreement between methods as well as χ2 tests and t-tests to assess any associations between the diagnosis of avulsion by the three tested methods and clinical or sonographic signs of female pelvic organ prolapse. All continuous data were normally distributed as ascertained by histograms and Kolmogorov–Smirnov testing.
Of 266 datasets identified for this study, seven were irretrievable due to clerical error, leaving 259. Mean age of patients was 56 (range, 22–88) years. Patients presented with stress incontinence (n = 211; 79%), urge incontinence (n = 188; 71%), frequency (n = 106; 40%), nocturia (n = 137; 52%), symptoms of voiding dysfunction (n = 70; 28%) and symptoms of prolapse (n = 119; 45%). Most patients (n = 247; 93%) were vaginally parous; 101 (38%) had undergone a hysterectomy and 58 (22%) an incontinence or prolapse procedure. A total of 156 (59%) patients had a significant prolapse on examination. On urodynamic testing, 186 (71%) patients were diagnosed with urodynamic stress incontinence, and 66 (25%) and 77 (29%) with detrusor overactivity and voiding dysfunction, respectively. Mean bladder neck descent was 30 (range, 0–67) mm. Mean hiatal area on Valsalva was 29 (12–63) cm2.
On palpation, 54 (20%) patients had a complete avulsion on at least one side, 47 (18%) on the right and 25 (9%) on the left. On rendered volume imaging, 65 (25%) patients were diagnosed with avulsion, 43 (17%) on the right and 31 (12%) on the left. On TUI, 79 (30%) women were found to have an avulsion, with 70 right-sided and 50 left-sided defects. The interobserver reproducibility series for the evaluation of rendered images for levator avulsion yielded a Cohen's kappa of 0.57 (95% CI, 0.32–0.76). Table 1 shows agreement between methods which was moderate to fair. Table 2 shows results of validating the diagnosis of avulsion against symptoms and signs of prolapse as well as against sonographic findings. The results of all three methods were strongly associated with symptoms and signs of prolapse. Almost identical results were obtained on analyzing a subset of women without previous surgery for incontinence and prolapse (n = 201).
Table 1. Agreement between palpation at the time of examination and retrospective analysis of stored ultrasound volume datasets using either rendered images or tomographic ultrasound imaging (TUI) in the diagnosis of levator avulsion
Cohen's kappa (95% CI)
Palpation vs rendered volume
Rendered volume vs TUI
Palpation vs TUI
Table 2. Validation of diagnosis of avulsion using each method against symptoms and signs of prolapse as well as against sonographic findings associated with prolapse
Symptoms of prolapse
Significant prolapse (POP-Q stage, 2+)
Maximum bladder descent on US
Maximum hiatal area on Valsalva
The results shown are the chi-square statistic for categorical data and t-statistic for continuous data (both with associated P-values) for the association between the diagnosis of avulsion using each method and differences observed in the variables considered. Unless stated otherwise, operators were blinded to each other's findings (n = 266). *Not blinded. †n = 259. ‡n = 252. US, ultrasound.
χ2 = 39.8, P < 0.001*
χ2 = 91.1, P < 0.001*
t = 4.22, P < 0.001
t = −6.92, P < 0.001†
χ2 = 25.8, P < 0.001†
χ2 = 64.3, P < 0.001†
t = 2.73, P = 0.007†
t = −3.46, P < 0.001‡
χ2 = 13.8, P < 0.001†
χ2 = 58.3, P < 0.001†
t = 3.78, P < 0.001†
t = −7.04, P < 0.001†
Examination for levator avulsion is likely to become a standard component of the diagnostic work-up of patients presenting with symptoms and/or signs of female pelvic organ prolapse. The strongest motivation for such a change is the fact that avulsion seems to be a strong risk factor for prolapse recurrence8−12, with relative risks/odds ratios of between 2 and 4. This implies that avulsion should be diagnosed clinically prior to prolapse surgery, especially if the use of mesh is contemplated.
Of the different methods proposed for the diagnosis of levator avulsion in the literature, MRI seems to be the most expensive and the least accessible and practical. Translabial ultrasound is non-invasive and seems to perform as well as or better than MRI. In addition, it has multiple other advantages such as the imaging of prolapse, paraurethral abnormalities, slings and meshes and anorectal abnormalities including sphincter trauma[27, 28]. Originally, the sonographic diagnosis of avulsion was described in ‘rendered volumes’, i.e. representations of ‘boxes’ of grayscale pixels, processed by user-definable algorithms in order to obtain semi-transparent representations of anatomical structures. The method was originally developed to allow imaging of the surface of fetal structures, especially limbs and faces. Fortuitously, the settings used for imaging of the fetal face, and the transducers employed for that purpose, are highly suitable for imaging of the inferior aspects of the levator ani muscle (Figure 2) and avulsion of the puborectalis muscle is readily apparent. The first reported use dates to 2004, and the first publications describing the research use of ultrasound for the diagnosis of avulsion employed this technique[1, 21]. More recently, multislice or tomographic imaging was developed and is in the process of becoming the standard for levator assessment. The method is sufficiently repeatable for clinical use and compares favorably with MRI. However, TUI often requires optional software and may involve additional cost. It is certainly premature to expect widespread availability of this modality.
Palpation, on the other hand, is universally available, at least in principle, and comes at no additional cost. However, it requires substantial training13−15, and such training is rarely available in clinical practice. The first studies documented in the literature utilized palpation to diagnose levator trauma[29, 30]. However, Gainey's work was never followed up by other authors; it is quite likely that this is due to a lack of training opportunities, and palpation of the levator ani never became part of our diagnostic repertoire. Laycock introduced the palpatory assessment of levator strength in the late 1980s, and physiotherapists have since made this a routine part of their assessment of women with pelvic floor dysfunction, but the diagnosis of morphological abnormalities has played no role in physiotherapeutic investigations until very recently. In our unit we have developed a method for examining the integrity of the bony insertion of the puborectalis muscle that seems moderately repeatable after an acceptable period of training. It involves placement of the palpating finger parallel to the urethra to assess the inferior pubic ramus on pelvic floor muscle contraction. An intact muscle will be evident immediately lateral to the operator's fingertip.
In this study we compared the three methods currently in clinical use, i.e. sonographic diagnosis by rendered volume, sonographic diagnosis by TUI and palpation, in a retrospective study on a population of women seen in a tertiary urogynecology unit. The three methods all seem to be moderately repeatable; they correlate moderately well with each other and findings with all three methods are significantly associated with symptoms, signs and ultrasound findings of female pelvic organ prolapse (P = 0.007 to < 0.001). TUI seems to reveal the highest number of avulsions, and it is not clear whether this represents overdiagnosis or higher sensitivity. From our own experience it seems likely that palpation may identify small remnants of puborectalis muscle that are not visible on tomographic imaging. This may also explain why palpation performed surprisingly well in the validation phase of the study; however, this may be due partly to the fact that the examiner performing the palpation could not be consistently blinded against symptoms and clinical examination, and also due to the greater experience of the palpation operator.
There are some weaknesses of this study that have to be acknowledged. Our conclusions are necessarily limited to the population tested, i.e. largely Caucasian women complaining of symptoms of pelvic floor dysfunction. The retrospective nature of the study is another disadvantage, but selection bias seems very unlikely since the entry criteria were assessment during the time period stipulated above and the availability of ultrasound volume datasets. We did not exclude women with previous pelvic surgery, although a subanalysis of women without such a history gave nearly identical results.
The main weakness of this comparison is that, while all imaging analysis was undertaken by a junior operator (F.M.) blinded to all clinical data, palpation data had been obtained at the time of the original appointment by the senior author who was therefore not consistently blinded to history and other clinical examination findings. It is possible that this may have resulted in a stronger relationship between the diagnosis of avulsion on palpation and symptoms and signs of prolapse. To exclude such an effect would require a prospective study design and standardized training. However, since the diagnostic performance of palpation depends on operator experience to a much greater degree than does imaging, any relative rating of diagnostic performance of palpation would pertain to individuals only and hence be of very limited utility.
Despite these disadvantages we feel that this study supports the validity of all three methods examined here. The authors feel that in women with pelvic floor dysfunction, especially in those for whom surgical management of prolapse is considered, assessment of the integrity of the levator plate should form part of routine investigations. Depending on the availability of local expertise and equipment, any of the three methods tested in this study could serve this purpose.
Within the last 2 years, Dr Dietz has received an educational grant from GE Medical (value USD 9000) and has received equipment support from GE Medical and Siemens for workshops. He also acted as consultant for Materna Medical (San Francisco, USA). The other authors have no potential conflict of interest to declare.