Learning process for performing and analyzing 3D/4D transperineal ultrasound imaging and interobserver reliability study

Transcription

1 Ultrasound Obstet Gynecol 2013; 41: Published online in Wiley Online Library (wileyonlinelibrary.com). DOI: /uog Learning process for performing and analyzing 3D/4D transperineal ultrasound imaging and interobserver reliability study F. SIAFARIKAS*, J. STÆR-JNSN*, I. H. BRÆKKN, K. BØ* and M. LLSTRÖM NGH* *Department of Obstetrics and Gynaecology, Akershus University Hospital, Lørenskog, Norway; Faculty Division Akershus University Hospital, University of Oslo, Oslo, Norway; Department of Sports Medicine, Norwegian School of Sport Sciences, Oslo, Norway KYWORDS: interobserver reliability; learning process; levator hiatus dimension; pelvic floor; transperineal ultrasound ABSTRACT Objectives To evaluate the learning process for acquiring three- and four-dimensional (3D/4D) transperineal ultrasound volumes of the levator hiatus (LH) dimensions at rest, during pelvic floor muscle (PFM) contraction and on Valsalva maneuver, and for analyzing the ultrasound volumes, as well as to perform an interobserver reliability study between two independent ultrasound examiners. Methods This was a prospective study including 22 women. We monitored the learning process of an inexperienced examiner (I) performing 3D/4D transperineal ultrasonography and analyzing the volumes. The examination included acquiring volumes during three PFM contractions and three Valsalva maneuvers. LH dimensions were determined in the axial plane. The learning process was documented by estimating agreement between the I and an experienced examiner () using the intraclass correlation coefficient. Agreement was calculated in blocks of 10 ultrasound examinations and analyzed volumes. After the learning process was complete the interobserver reliability for the technique was calculated between these two independent examiners. Results For offline analysis of the first 10 ultrasound volumes obtained by, good to very good agreement between and I was achieved for all LH measurements except for the left and right levator urethra gap and pubic arc. For the next 10 analyzed volumes, agreement improved for all LH measurements. Volumes that had been obtained by I and were then re-evaluated by I, and good to very good agreement was found for all LH measurements indicating consistency in volume acquisition. The interobserver reliability study showed excellent ICC values (ICC, ) for all LH measurements except the pubic arc (ICC = 0.67). Conclusion 3D/4D transperineal ultrasound is a reliable technique that can be learned in a short period of time. Copyright 2012 ISUOG. Published by John Wiley & Sons, Ltd. INTRODUCTION Pelvic floor dysfunction, such as pelvic organ prolapse and urinary and fecal incontinence, impairs quality of life for a large number of women 1. The pelvic floor muscles (PFM) play a critical role in pelvic organ support 2,3.The pubovisceral muscle (PVM) is the most medial part of the levator ani complex bordering on the levator hiatus (LH) 4. In recent years, three- and four-dimensional (3D/4D) transperineal ultrasound has become a useful diagnostic tool for visualizing the PVM, and has been shown to be reliable and valid However, while it has been suggested that it is easy to learn how to perform a 3D/4D transperineal ultrasound examination 11, a PubMed search did not reveal any studies investigating the learning process of both volume acquisition and analysis of the recorded volumes. Previous intra- and interobserver studies of 3D/4D transperineal ultrasound of the pelvic floor have shown acceptable reliability 5,7,8,10, but to our knowledge these studies only tested reliability of the offline analysis. As far as we have been able to ascertain, there have been no studies investigating the reliability between two independent examiners for complete transperineal ultrasound examination including instructing the patient, and recording and analyzing the ultrasound volumes. The primary aim of this study was to monitor the learning process for acquiring 3D/4D transperineal ultrasound volumes and for their offline analysis. A second aim was to perform an interobserver reliability study for the entire ultrasound procedure, including both volume acquisition and offline analysis. Correspondence to: Dr F. Siafarikas, Department of Obstetrics and Gynaecology, Akershus University Hospital, Lørenskog, Norway ( franziska.richter@medisin.uio.no) Accepted: 7 May 2012 Copyright 2012 ISUOG. Published by John Wiley & Sons, Ltd. ORIGINAL PAPR

2 Learning process for 3D/4D transperineal ultrasound 313 MTHODS Subjects This prospective study was conducted at the Department of Obstetrics and Gynaecology of Akershus University Hospital from June to September 2010 and included 22 women (two nulliparous pregnant women at 22 weeks of gestation and 10 at 37 weeks of gestation, and 10 primiparous women 6 weeks after delivery). The women were participating in an ongoing cohort study designed to investigate changes in PFM morphology and function during pregnancy and after childbirth. thics committee approval was granted by the Regional Medical thics Committee (2009/170) and the Norwegian Social Science Data Service ( ). An inexperienced examiner (I) was introduced into an existing research team, as part of which an experienced examiner () had performed more than 500 ultrasound examinations and analyses. I was a physician undergoing her 4 th year of specialist training in gynecology and obstetrics. She was comfortable with performing transvaginal and abdominal ultrasound examinations but had no previous experience of examining the pelvic floor using 3D/4D transperineal ultrasound. After an initial introduction procedure, for which four volunteers from the staff were recruited, I s learning process for volume acquisition and offline analysis was monitored by evaluating the agreement of her findings in the 22 women with those of. When I s findings achieved acceptable agreement with s, an interobserver study was performed between these two examiners. They were both blinded to previously collected data and to each other s results. Ultrasound Ultrasound examinations were performed using a G Voluson 8 system (G Medical Systems, Zipf, Austria), equipped with a 4 8-MHz curved array 3D/4D ultrasound transducer (RAB4-8l/obstetric). The field of view angle was set to its maximum of 70 in the sagittal plane with the depth at 6.5 or 8.3 cm and the focus at 3 cm. The acquisition angle was set to 85 in the coronal plane. Before the examination, all women were instructed by a physiotherapist as to how to perform a pelvic floor contraction correctly. 3D/4D volumes were acquired with the women in the lithotomy position and with an empty bladder, at rest, during contraction and on Valsalva 7,9. ach maneuver was recorded three times. The ultrasound volumes were stored on the hard disc of the ultrasound machine and were transferred to a laptop. Analysis was carried out using 4D View version 10 (G Medical Systems) software. The volume with the best contraction, defined as the one with the most cranial displacement of the levator plate, and the volume with the best Valsalva maneuver, defined as the one with the most caudal displacement of the levator plate, were chosen for analysis 6,7. The rest position was defined as the most caudal position of the levator plate before the best PFM contraction. All measurements of the LH were determined in the axial plane at the level of minimal hiatal dimensions, as described previously 6 8. Minimal hiatal dimension was defined as the minimum distance between the hyperechogenic posterior aspect of the symphysis pubis and the hyperechogenic anterior border of the PVM 8,9. LH area (LHarea) was measured as the area bordered by PVM, symphysis pubis and inferior pubic ramus. The transverse diameter of the LH from right to left (LHrl) was defined as the widest part, perpendicular to the anteroposterior diameter (LHap) 9. The levator urethra gap (LUG) was measured from the insertion of the PVM on the left (LUG-l) and right (LUG-r) sides to the midurethra 12. The pubic arc (the bony part of the hiatal circumference) was measured along the inner margin of the pubic ramus between the insertions of the PVM. Cocontraction of PVM was diagnosed when LHap was less on Valsalva compared with in the resting position 13. Learning procedure For the introduction procedure, using four volunteers, I learned how to instruct the patient and to record an ultrasound examination consisting of three PFM contractions and three Valsalva maneuvers. Thereafter, I was instructed on how to find the image with the best contraction and the one with the most effective Valsalva maneuver and how to analyze the chosen image in the axial plane. To evaluate the introduction procedure, I and both performed an ultrasound examination on two of the 22 study recruits, who were 37 weeks gestation, and analyzed the recorded volumes separately. It was stated a priori that the coefficient of variation between examiners measurements should not exceed 10% for I to be considered to having finished the introduction procedure and be allowed to move on to the learning procedure. The learning procedure involved 3D/4D ultrasonography of the remaining 20 recruited women. ach woman was examined twice, once by I and once by, consecutively and in alternating order. The ultrasound volumes were then analyzed by I and to evaluate the learning process and for the interobserver study. To evaluate the learning procedure for the offline analysis, the 20 volumes recorded by were used to compare measurements of LH dimensions between the two examiners (Figure 1a). Agreement was estimated using the intraclass correlation coefficient (ICC) in blocks of 10 volumes. When good to very good agreement was achieved for measurements of LHap, LHrl and LHarea in the analysis procedure, I analyzed the 20 ultrasound volumes recorded by herself. This allowed assessment of the learning procedure for performing the ultrasound volume acquisition (Figure 1b), by calculating the agreement between volumes recorded by I and by, as analyzed by I. Interobserver reliability Finally, when I s findings achieved acceptable agreement with those of, the interobserver reliability of the

3 314 Siafarikas et al. (a) Learning process: offline analysis 20 patients RSULTS Introduction procedure Volume acquisition Offline analysis (b) Learning process: volume acquisition Volume acquisition I 20 patients I I During three 1-hour learning sessions I learned how to record an ultrasound examination, and during four 5-hour sessions she learned how to analyze the recorded volumes. The calculated coefficient of variation between I and for performing the ultrasound examination and offline analysis of two women was good (< 10% for all LH measurements), so the introduction procedure was considered complete and I continued the learning procedure. Offline analysis I I Learning procedure Learning how to analyze recorded volumes (c) Interobserver study Volume acquisition Offline analysis 20 patients Figure 1 Diagram illustrating learning procedure of an inexperienced examiner (I) in performance of three-dimensional transperineal ultrasound volume acquisition and offline analysis for measurement of levator hiatus dimensions, and interobserver study between two independent examiners. (a) Learning process of offline analysis: I and an experienced examiner () analyzed volumes recorded by. (b) Learning process of ultrasound volume acquisition: I analyzed volumes recorded by and I. (c) Interobserver study: each investigator analyzed her own recorded volumes. entire procedure was tested by having the two examiners each analyze the results of their own 20 examinations separately (Figure 1c). Statistical analysis Statistical analysis was performed using SPSS version 18.0 (SPSS Inc., Chicago, IL, USA). The difference between the measurements made by I and by in the introduction procedure was calculated using the coefficient of variation. To evaluate agreement between measurements, the intraclass correlation coefficient (ICC) was calculated using a general linear univariate model with 95% CI to identify different variance components 9. ICC values < 0.2 were considered poor, was considered fair, moderate, good and excellent 14. The Bland Altman method calculates the mean difference between two methods of measurement ( bias ) and 95% limits of agreement as the mean difference (1.96 SD) 15. To test for systematic bias, a one-sample t-test was used to verify the hypothesis that the difference between the two examiners did not deviate from zero. P < 0.05 was considered statistically significant. I I For analysis of the first 10 ultrasound volumes obtained by, good to excellent agreement between the examiners was achieved for all LH measurements except for LUG and pubic arc during contraction, LHrl at rest and LHrl on Valsalva (Table 1 and Figure 2a). ICC was poor only for the pubic arc. In the next 10 ultrasound volumes, one Valsalva examination was excluded from further analysis owing to poor image quality. For analysis of these ultrasound volumes, agreement between I and was improved in all parameters and only ICC values for the pubic arc remained below the level of good agreement (Table 1 and Figure 2a). Learning how to perform volume acquisitions For the first 10 women who underwent ultrasound examinations, the agreement between volumes recorded by I and by (as analyzed by I), was excellent in all LH dimensions, at rest and during contraction, except for the pubic arc (Table 1 and Figure 2b). The ICC of LH dimensions on Valsalva was good for LHrl and excellent for LHap and LHarea. Over the next 10 ultrasound examinations, agreement improved for pubic arc and for all LH dimensions on Valsalva, until ICC values were > 0.95, indicating consistency in volume acquisition. Three women (two nulliparous and one examined 6 weeks after delivery) performed a co-contraction of the PFM on Valsalva when examined by I. One of these women (a nullipara) performed a correct maneuver when examined by. Interobserver reliability One Valsalva examination was excluded from the analysis of interobserver reliability for the entire procedure, owing to poor image quality. All measurements of the LH dimensions except the pubic arc showed excellent reliability during contraction, at rest and on Valsalva (Table 2). The measurements of the pubic arc showed good reliability. There was no systematic bias found between the two examiners.

4 Learning process for 3D/4D transperineal ultrasound 315 Table 1 Learning procedure: interobserver differences between an inexperienced and an experienced examiner in performance of three-dimensional transperineal ultrasound offline analysis and volume acquisition, for measurement of levator hiatus (LH) during contraction (-C), at rest (-R) and on Valsalva (-V) Intraclass correlation coefficient (95% CI) Parameter Analyzed volume 1 10 Analyzed volume Volume acquisition 1 10 Volume acquisition LHap-C 0.94 (0.79, 0.98) 0.96 (0.84, 0.99) 0.97 (0.88, 0.99) 0.96 (0.84, 0.99) LHrl-C 0.73 (0.27, 0.93) 0.77 (0.34, 0.94) 0.86 (0.57, 0.96) 0.89 (0.63, 0.97) LHarea-C 0.90 (0.67, 0.97) 0.93 (0.77, 0.98) 0.93 (0.76, 0.98) 0.96 (0.85, 0.99) Pubic arc 0.13 ( 0.49, 0.67) 0.39 ( 0.25, 0.80) 0.69 (0.18, 0.91) 0.81 (0.42, 0.95) LUG-right 0.47 ( 0.15, 0.83) 0.70 (0.21, 0.92) 0.82 (0.47, 0.95) 0.82 (0.46, 0.95) LUG-left 0.49 ( 0.12, 0.84) 0.69 (0.19, 0.91) 0.88 (0.60, 0.97) 0.90 (0.67, 0.97) LHap-R 0.87 (0.58, 0.97) 0.94 (0.79, 0.99) 0.96 (0.86, 0.99) 0.84 (0.50, 0.96) LHrl-R 0.39 ( 0.28, 0.82) 0.81 (0.41, 0.95) 0.90 (0.68, 0.97) 0.94 (0.78, 0.98) LHarea-R 0.84 ( 0.51, 0.96) 0.90 (0.65, 0.98) 0.97 (0.89, 0.99) 0.96 (0.91, 0.99) LHap-V 0.85 (0.51, 0.96) 0.95 (0.83, 0.99) 0.82 (0.46, 0.95) 0.96 (0.87, 0.99) LHrl-V 0.56 ( 0.74, 0.88) 0.98 (0.92, 0.99) 0.79 (0.39, 0.94) 0.96 (0.87, 0.99) LHarea-V 0.84 (0.46, 0.96) 0.99 (0.98, 1.00) 0.87 (0.57, 0.96) 0.97 (0.88, 0.99) Twenty women were examined, and ultrasound examinations were considered in blocks of 10 for the learning process. ap; anteroposterior diameter; LUG, levator urethra gap; rl, transverse diameter. (a) ICC (b) ICC LHap-C LHap-C LHrl-C LHrl-C LHarea-C LHarea-C Pubic arc Pubic arc LUGr LUGl LUGr LUGl LHap-R LHrl-R LHarea-R LHap-R LHrl-R LHap-V LHrl-V LHarea-V LHarea-R LHap-V LHrl-V LHarea-V Figure 2 Plots illustrating agreement between an inexperienced and an experienced examiner in three-dimensional transperineal ultrasound offline analysis (a) and volume acquisition (b) for measurement of the levator hiatus (LH) during contraction (-C), at rest (-R) and on Valsalva (-V). Agreement based on numbers of ultrasound volume acquisitions or offline analyses: 1 10 ( )or ( ). LH dimensions measured included: anteroposterior diameter (ap), transverse diameter (rl), levator urethra right (LUGr) and left (LUGl) gaps and pubic arc. In the second set of volumes (volumes 11 20) analyzed, there was one Valsalva examination with insufficient imaging quality, which was therefore excluded from the evaluation of offline analysis (a). ICC, intraclass correlation coefficient. DISCUSSION To our knowledge, this is the first study of the learning process for using 3D/4D transperineal ultrasound. We found that an examiner who had never performed a transperineal ultrasound examination was able to accomplish a volume acquisition with good to very good agreement with an experienced investigator after only 3 hours of learning. After the first 10 volume acquisitions, except for the pubic arc, the ICC values for LH dimensions at rest and during PFM contraction were excellent. On Valsalva, the ICCs for LH dimensions ranged between 0.79 and 0.87 and improved over the next 10 volume acquisitions to ICC In our study no women were excluded because of incomplete scanning of the LH on Valsalva, which has been reported as a problem in previous studies 7,16. It was apparently easier for I to instruct women to perform efficient PFM contractions than to perform proper Valsalva maneuvers, perhaps because all participants had already been taught how to perform a correct PFM contraction by a physiotherapist. In this study, Valsalva maneuver appeared more instructor-dependent. Valsalva can be confounded by co-contraction of the PFM, which results in lower LH diameter and area and is reported to be common in nulliparous women 13. Approximately 50% of our study population were nulliparous. It is imperative that the investigator is able to recognize potential co-contractions of the PFM as insufficient increase in anteroposterior diameter in the sagittal plane while recording in real time. Biofeedback, verbal instruction and repetition may help women to avoid co-activation during assessment of the Valsalva maneuver. Learning the analysis procedure was more timeconsuming: 20 hours were required to learn the offline analysis. After analysis of the first 10 ultrasound volumes, ICCs at rest, during PFM contraction and on Valsalva ranged widely, between 0.13 and Agreement

5 316 Siafarikas et al. Table 2 Interobserver reliability of three-dimensional transperineal ultrasound for measurement of levator hiatus (LH) dimensions during contraction (-C), at rest (-R) and on Valsalva (-V): interobserver differences between two independent examiners Limits of agreement Parameter Mean (95% CI) n ICC Bias SD Lower Upper LHap-C (cm) xaminer ( ) xaminer ( ) 20 LHrl-C (cm) xaminer ( ) xaminer ( ) 20 LHarea-C (cm 2 ) xaminer ( ) xaminer ( ) 20 Pubic arc (cm) xaminer ( ) xaminer ( ) 20 LUG-right (cm) xaminer ( ) xaminer ( ) 20 LUG-left (cm) xaminer ( ) xaminer ( ) 20 LHap-R (cm) xaminer ( ) xaminer ( ) 20 LHrl-R (cm) xaminer ( ) xaminer ( ) 20 LHarea-R (cm 2 ) xaminer ( ) xaminer ( ) 20 LHap-V (cm) xaminer ( ) xaminer ( ) 19 LHrl-V (cm) xaminer ( ) xaminer ( ) 19 LHarea-V (cm 2 ) xaminer ( ) xaminer ( ) 19 ap, anteroposterior diameter; ICC, intraclass correlation coefficient; LUG, levator urethra gap; rl, transverse diameter. improved within the next 10 ultrasound volumes analyzed, with good to excellent reliability achieved for LHap, LHrl and LHarea during PFM contraction and at rest. These results are in accordance with findings in previous inter- and intraobserver studies testing the analysis process 5 8,10,17. We also found excellent reliability for all LH dimensions on Valsalva after analysis of 20 volumes. Reported reliability for LHap and LHarea on Valsalva vary from excellent to moderate 7,8,12. However, while others have reported good or excellent reliability for the pubic arc and LUG 12,17,incontrast, we found the lowest reliability was achieved for these measurements, with ICCs of 0.40 and 0.70, respectively. Most investigators measure the pubic arc and LUG during PFM contraction when the insertion angle of the PVM into the pubic rami is most acute 17. Defining the insertion of the PVM can be difficult, especially postpartum. In the 10 women examined 6 weeks postpartum, the PVM displayed low contrast, and demarcation was challenging. As far as we have been able to ascertain, this interobserver reliability study is the first study using two independent examiners performing the entire transperineal ultrasound procedure, both ultrasound volume acquisition and offline analysis. We found excellent reliability between the two observers for all LH measurements except for the pubic arc. Previous test retest, intra- and interobserver studies of 3D/4D transperineal ultrasound have found acceptable reliability for measurements of LH dimensions 5 8,10. In all these studies, ultrasonography was performed by the same examiner. We did not find any studies evaluating reliability for volume acquisition with transperineal ultrasound between two different examiners. Weinstein et al. 18 performed an intra- and interobserver reliability study of PFM morphology, in which two independent examiners performed volume acquisition and offline analyses using a transvaginal probe, and found good to very good reliability for LHap and LHarea during PFM contraction and at rest. The limits of agreement in our study are

6 Learning process for 3D/4D transperineal ultrasound 317 comparable to those of the interobserver study of Majida et al. 7. A limitation of our study is that only one inexperienced examiner was included. Our main aim was to introduce a new co-worker into an already established team as quickly as possible, ascertaining good-quality ultrasound data with minimal variation. The generalizability of our findings might be questioned, not only because of the limited number of inexperienced examiners, but also because of the patient characteristics. For example, in a group of patients with pelvic organ prolapse, transperineal ultrasonography might be more difficult to accomplish owing to the larger LH area on Valsalva and to the limited acquisition angle of the transducer 8. Yet, 50% of our study women were examined 6 weeks postpartum, when anatomical structures display low contrast. Regarding sample size, the number of women included was in line with other published reliability and validation studies in this area 5,6,9. Although better ICC values for LUG and pubic arc might have been achieved by including more women, I had already accomplished the ultrasound examination and offline analysis with good and very good agreement with for the other measurements after 20 examinations, indicating that measurements of LUG and pubic arc were more difficult to assess. To include 3D/4D transperineal ultrasound in a routine gynecological examination of women with pelvic floor disorders, the technique has to fulfill at least two criteria. First and most importantly, information from a transperineal ultrasound examination should give clinically useful information. To date, information about the PVM has been found to be helpful in selecting patients with high risk of prolapse recurrence 19. Second, the examination should be easy to perform and the imaging reliable and easy to interpret. In this study we have confirmed that 3D/4D transperineal ultrasound can be learned with an acceptable level of effort. The excellent ICC values of the interobserver study show that it is a reliable tool for examining the PVM. In conclusion, 3D/4D transperineal ultrasound is a reliable technique that can be learned in a short period of time and could be incorporated easily into examinations of the pelvic floor, in the same way in which ultrasound of the uterus and adnexa is included in routine gynecological examinations today. ACKNOWLDGMNTS This study was supported by grants from the South- astern Regional Health Authority in Norway. RFRNCS 1. MacLennan AH, Taylor AW, Wilson DW, Wilson D. The prevalence of pelvic floor disorders and their relationship to gender, age, parity and mode of delivery. BJOG 2000; 107: Olsen AL, Smith VJ, Bergstrom JO, Colling JC, Clark AL. pidemiology of surgically managed pelvic organ prolapse and urinary incontinence. Obstet Gynecol 1997; 89: DeLancey JO. The hidden epidemic of pelvic floor dysfunction: achievable goals for improved prevention and treatment. Am J Obstet Gynecol 2005; 192: DeLancey JO. The anatomy of the pelvic floor. Curr Opin Obstet Gynecol 1994; 6: Braekken IH, Majida M, ngh M, Bo K. Test-retest reliability of pelvic floor muscle contraction measured by 4D ultrasound. Neurourol Urodyn 2009; 28: Braekken IH, Majida M, llstrom-ngh M, Dietz HP, Umek W, Bo K. Test-retest and intra-observer repeatability of two-, three- and four-dimensional perineal ultrasound of pelvic floor muscle anatomy and function. Int Urogynecol J Pelvic Floor Dysfunct 2008; 19: Majida M, Braekken IH, Umek W, Bo K, Saltyte Benth J, llstrom ngh M. Interobserver repeatability of three- and four-dimensional transperineal ultrasound assessment of pelvic floor muscle anatomy and function. Ultrasound Obstet Gynecol 2009; 33: Dietz HP, Shek C, Clarke B. Biometry of the pubovisceral muscle and levator hiatus by three-dimensional pelvic floor ultrasound. Ultrasound Obstet Gynecol 2005; 25: Majida M, Braekken IH, Bo K, Benth JC, ngh M. Validation of three-dimensional perineal ultrasound and magnetic resonance imaging measurements of the pubovisceral muscle at rest. Ultrasound Obstet Gynecol 2010; 35: Yang JM, Yang SH, Huang WC. Biometry of the pubovisceral muscle and levator hiatus in nulliparous Chinese women. Ultrasound Obstet Gynecol 2006; 28: Dietz HP. Ultrasound imaging of the pelvic floor. Part II: threedimensional or volume imaging. Ultrasound Obstet Gynecol 2004; 23: Dietz HP, Abbu A, Shek LK. The levator urethra gap measurement: a more objective means of determining levator avulsion? Ultrasound Obstet Gynecol 2008; 32: Orno AK, Dietz HP. Levator co-activation is a significant confounder of pelvic organ descent on Valsalva maneuver. Ultrasound Obstet Gynecol 2007; 30: Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1: Myles PS, Cui J. Using the Bland Altman method to measure agreement with repeated measures. Br J Anaesth 2007; 99: Dietz HP, Steensma AB. Posterior compartment prolapse on two-dimensional and three-dimensional pelvic floor ultrasound: the distinction between true rectocele, perineal hypermobility and enterocele. Ultrasound Obstet Gynecol 2005; 26: Thyer I, Shek C, Dietz HP. New imaging method for assessing pelvic floor biomechanics. Ultrasound Obstet Gynecol 2008; 31: Weinstein MM, Jung SA, Pretorius DH, Nager CW, den Boer DJ, Mittal RK. The reliability of puborectalis muscle measurements with 3-dimensional ultrasound imaging. Am J Obstet Gynecol 2007; 197: 68 e Dietz HP. Clinical consequences of levator trauma. Ultrasound Obstet Gynecol 2012; 39: