Lower uterine segment thickness measurement in pregnant women with previous Cesarean section: reliability analysis using two- and three-dimensional transabdominal and transvaginal ultrasound

Authors

  • W. P. Martins,

    Corresponding author
    1. Departamento de Ginecologia e Obstetrícia da Faculdade de Medicina de Ribeirão Preto, Universidade de São Paulo and Escola de Ultra-sonografia de Ribeirão Preto, Ribeirão Preto, São Paulo, Brazil
    • Faculdade de Medicina de Ribeirão Preto da Universidade de São Paulo, Departamento de Ginecologia e Obstetrícia, Avenida dos Bandeirantes, 3900, 8° Andar, Ribeirão Preto, São Paulo 14049-900, Brazil
    Search for more papers by this author
  • D. A. Barra,

    1. Departamento de Ginecologia e Obstetrícia da Faculdade de Medicina de Ribeirão Preto, Universidade de São Paulo and Escola de Ultra-sonografia de Ribeirão Preto, Ribeirão Preto, São Paulo, Brazil
    Search for more papers by this author
  • F. M. P. Gallarreta,

    1. Departamento de Ginecologia e Obstetrícia da Faculdade de Medicina de Ribeirão Preto, Universidade de São Paulo and Escola de Ultra-sonografia de Ribeirão Preto, Ribeirão Preto, São Paulo, Brazil
    Search for more papers by this author
  • C. O. Nastri,

    1. Departamento de Ginecologia e Obstetrícia da Faculdade de Medicina de Ribeirão Preto, Universidade de São Paulo and Escola de Ultra-sonografia de Ribeirão Preto, Ribeirão Preto, São Paulo, Brazil
    Search for more papers by this author
  • F. M. Filho

    1. Departamento de Ginecologia e Obstetrícia da Faculdade de Medicina de Ribeirão Preto, Universidade de São Paulo and Escola de Ultra-sonografia de Ribeirão Preto, Ribeirão Preto, São Paulo, Brazil
    Search for more papers by this author

Abstract

Objective

To evaluate the reliability of two- and three-dimensional ultrasonographic measurement of the thickness of the lower uterine segment (LUS) in pregnant women by transvaginal and transabdominal approaches.

Methods

This was a study of 30 pregnant women who had had at least one previous Cesarean section and were between 36 and 39 weeks' gestation, with singleton pregnancies in cephalic presentation. Sonographic examinations were performed by two observers using both 4–7-MHz transabdominal and 5–8-MHz transvaginal volumetric probes. LUS measurements were performed using two- and three-dimensional ultrasound, evaluating the entire LUS thickness transabdominally and the LUS muscular thickness transvaginally. Each observer measured the LUS four times by each method. Reliability was analyzed by comparing the mean of the absolute differences, the intraclass correlation coefficients, the 95% limits of agreement and the proportion of differences < 1 mm.

Results

Transvaginal ultrasound provided greater reliability in LUS measurements than did transabdominal ultrasound. The use of three-dimensional ultrasound improved significantly the reliability of the LUS muscular thickness measurement obtained transvaginally.

Conclusions

Ultrasonographic measurement of the LUS muscular thickness transvaginally appears more reliable than does that of the entire LUS thickness transabdominally. The use of three-dimensional ultrasound should be considered to improve measurement reliability. Copyright © 2009 ISUOG. Published by John Wiley & Sons, Ltd.

Introduction

Three layers of the lower uterine segment (LUS) can be identified on ultrasound: the chorioamniotic membrane with decidualized endometrium; the middle muscular layer; and the uterovesical peritoneal reflection juxtaposed with muscularis and mucosa of the bladder1. Rozenberg et al.2 reported a significant relationship between the transabdominal sonographic measurement of the entire LUS thickness in women near term who had had a previous Cesarean section and the risk of uterine rupture or dehiscence, the separation of the muscular layer with an intact serosa3. No uterine rupture or dehiscence was noted when the entire LUS thickness was > 4.5 mm. Furthermore, the risk of uterine rupture or dehiscence was 0.66% when this measurement was ≥ 3.5 mm, compared with 11.7% in women who had a LUS thickness < 3.5 mm.

Several studies using various methods have been conducted to evaluate the correlation of LUS measurement with the risk of uterine rupture or dehiscence2–8, with relative success. In some studies, the sonographers measured the entire LUS by transabdominal ultrasound2, 5, while in others, only the middle muscle layer was assessed using transvaginal ultrasound3, 4, 6 and some studies used both approaches7, 8. Yet, despite this heterogeneity, only one study addressing the intra- and interobserver reliability of LUS measurement by transvaginal and transabdominal ultrasound has been published9. This study evaluated the reliability of transabdominal ultrasound in patients with full or half-full bladder and of transvaginal ultrasound, analyzing only the measurement of the entire LUS thickness. The authors concluded that reliability was not improved by a full patient bladder and that transvaginal ultrasound was more reliable than were the two transabdominal methods. However, this study did not examine the measurement of the LUS muscular layer, which is the most commonly used method for evaluation by the transvaginal approach3, 4, 6, 10.

It is possible that three-dimensional (3D) ultrasound could improve the reliability of LUS measurement since its reliability and validity for the measurement of in-vitro models11–13 have proved to be excellent. In addition, the reliability of endometrial thickness measurement in postmenopausal women, which is similar in thickness to the LUS, was shown to be greater when using 3D ultrasound14. This might be due to the use of the multiplanar display of 3D ultrasound, which permits simultaneous longitudinal, transverse and coronal views, thus avoiding measurements in oblique longitudinal views.

In this study, we evaluated the intra- and interobserver reliability of the two most commonly used means of measuring LUS thickness: the entire LUS thickness measured by transabdominal ultrasound and the LUS muscular thickness measured by transvaginal ultrasound. We also evaluated whether the use of 3D ultrasound was able to improve LUS measurement reliability.

Methods

Patients

The research protocol was conducted in accordance with the guidelines of the Declaration of Helsinki15 and was approved by the ethics and research committee of the Hospital das Clínicas da Faculdade de Medicina de Ribeirão Preto da Universidade de São Paulo. Informed written consent was obtained from all women after the nature of the procedures had been fully explained.

Between October and December 2006, 30 pregnant women who had had at least one previous Cesarean section and were between 36 and 39 weeks' gestation were enrolled into the study at Escola de Ultra-sonografia e Reciclagem Médica de Ribeirão Preto. All women had singleton pregnancies in cephalic presentation. The women ranged in age from 21.3 to 35.2 (mean ± SD, 27.4 ± 3.5) years. Twelve (40%) women had previously had only one Cesarean section, 10 (33.3%) women had previously had two, six (20%) women had previously had one Cesarean section and one vaginal delivery and two (6.7%) women had previously had one Cesarean section and two vaginal deliveries. By the time of the LUS evaluation, the gestational age ranged from 36 + 4 weeks to 38 + 3 weeks.

Ultrasound

The sonographic examinations were performed with an Accuvix (Medison Co., Ltd. Seoul, Korea) machine equipped with a 4–7-MHz transabdominal (3D4-7EK) and a 5–8-MHz transvaginal (3D5-8EK) probe. Measurements were performed by two sonographers, Observer 1 (D.A.B.) and Observer 2 (F.M.P.G.). The ultrasound scans followed this sequence: primary, transabdominal evaluation (Figure 1a and c) was performed with the woman in a supine position with a comfortably full bladder. One observer stored two images of the entire LUS thickness measurement (one cursor positioned at the urine–bladder interface and the other at the decidua–amniotic fluid interface5) and one 3D dataset, and then repeated this process. When this observer had completed the evaluation he left the examination room and the other observer entered to perform the same steps. After the transabdominal examinations were complete, the transvaginal examination (Figure 1b and d) was performed with the woman in the lithotomy position, within 20 min after voiding her bladder3. The same schedule of examinations was performed by the two observers, this time measuring the LUS muscular layer (one cursor positioned at the bladder–muscular interface and the other at the muscular–decidua interface6). The time taken between first placement of the probe on the abdomen or inside the vagina until storage of the first image (time to identify and measure LUS) was noted. The order of observers was swapped for each new patient.

Figure 1.

Images demonstrating measurement of the entire thickness of the lower uterine segment (LUS) by transabdominal two-dimensional (a) and three-dimensional (c) ultrasound, and of the muscular layer of the LUS by tranvaginal two-dimensional (b) and three-dimensional (d) ultrasound.

After all examinations had been completed, the two observers manipulated their acquired volumes on the multiplanar display, trying to avoid obliquity and searching for the thinnest portion of the LUS. They were blinded to the measurement by a label placed on the numeric display. For each observer, two images of the LUS measurement for each method were stored from each of the two previously recorded 3D datasets. The time spent between opening the 3D dataset and storage of the first LUS measurement image was also noted.

After all 30 patients had been evaluated, the 2D and 3D ultrasound images were reviewed without the label, and the 32 LUS thickness measurements from every pregnant woman were recorded (16 for each observer).

Statistical analysis

Statistical analysis was performed using SPSS 15.0 for Windows (SPSS Inc., Chicago, IL, USA) and GraphPad 4.0 for Windows (GraphPad Software, San Diego, CA, USA). Intraobserver reliability was examined by calculating the six absolute differences among the four measurements per patient per method performed by each operator, giving 180 absolute differences (30 patients) overall for each method and 720 intraobserver differences overall for each observer. To evaluate the interobserver reliability, we quantified the absolute differences between the measurements of Observer 1 and Observer 2: 16 absolute differences per patient per method performed by each operator, giving 480 absolute differences overall for each method and 1920 interobserver differences overall. The means were compared using ANOVA. Additionally, the proportions of the differences that were < 1 mm (1 mm being described as the maximum reasonable difference9)—were analyzed, and the differences among these ratios were compared using Chi-square test. In order to improve the reliability analysis, the intraclass correlation coefficients (ICCs) and their 95% CIs were obtained. Intraobserver reliability was assessed by one-way absolute agreement ICCs, taking into account all measurements of each observer. Interobserver reliability was examined by two-way mixed ICCs, considering only the smallest measurement of each observer9, 12, 13, 16. For each method, the agreement between the thinnest measurement of each of the two observers was checked using the Bland–Altman limits of agreement method17. The means obtained for the time spent with each method were compared by one-way ANOVA.

Results

When measured by 2D ultrasound, the entire LUS thickness range was 3.6–19.2 mm by the transabdominal approach and the LUS muscular thickness range was 1.0–9.7 mm by the transvaginal approach. When measured by 3D ultrasound, the entire LUS thickness range was 3.3–16.0 mm by the transabdominal approach and the LUS muscular thickness range was 0.6–8.5 mm by the transvaginal approach.

Transvaginal vs. transabdominal approach

The transvaginal approach was more reliable than was the transabdominal one (Table 1). The means of the absolute differences were significantly lower for the transvaginal LUS muscular thickness measurement for all comparisons. The proportion of differences < 1 mm was significantly higher for the transvaginal approach on all intra- and interobserver evaluations. The ICCs were higher for the transvaginal approach on all comparisons. The 95% limits of agreement were also narrower for the transvaginal approach both for 2D and 3D ultrasound (Figure 2).

Figure 2.

Interobserver differences in lower uterine segment thickness measurement in pregnant women with previous Cesarean section. Bland–Altman plot and 95% limits of agreement17 are shown for each of the four methods: (a) total lower uterine segment (LUS) by two-dimensional (2D) transabdominal ultrasound; (b) muscular LUS by 2D transvaginal ultrasound; (c) total LUS by three-dimensional (3D) transabdominal ultrasound; and (d) muscular LUS by 3D transvaginal ultrasound.

Table 1. Intra- and interobserver reliability in lower uterine segment thickness measurement in pregnant women with previous Cesarean section
 Transabdominal ultrasoundTransvaginal ultrasoundP
2D3D2D3D
  • *

    One-way ANOVA.

  • Chi-square test. 2D, two-dimensional; 3D, three-dimensional; ICC, intraclass correlation coefficient.

Intraobserver analysis: Observer 1
 Absolute differences (mm, mean ± SD)0.98 ± 0.540.93 ± 0.800.42 ± 0.230.37 ± 0.31< 0.01*
 Proportion of differences < 1 mm (%)60.0071.1191.6796.67< 0.01
 ICC (95% CI)0.89 (0.78–0.94)0.87 (0.80–0.93)0.93 (0.89–0.96)0.92 (0.86–0.95) 
Intraobserver analysis: Observer 2
 Absolute differences (mm, mean ± SD)0.88 ± 0.700.89 ± 0.710.42 ± 0.270.29 ± 0.21< 0.01*
 Proportion of differences < 1 mm (%)72.7872.2291.6797.78< 0.01
 ICC (95% CI)0.85 (0.76–0.92)0.83 (0.73–0.90)0.91 (0.85–0.95)0.95 (0.91–0.97) 
Interobserver analysis
 Absolute differences (mm, mean ± SD)1.55 ± 1.541.49 ± 1.870.57 ± 0.610.42 ± 0.46< 0.01*
 Proportion of differences < 1 mm (%)48.7557.7185.8391.25< 0.01
 ICC (95% CI)0.78 (0.68–0.87)0.69 (0.56–0.81)0.87 (0.80–0.93)0.90 (0.85–0.95) 

Two- vs. three-dimensional ultrasound

The use of 3D ultrasound significantly improved the interobserver reliability (Table 1). The mean of the absolute differences was significantly lower on 3D ultrasound when considering the interobserver differences on both transvaginal and transabdominal approaches, and the proportion of differences < 1 mm was significantly higher on 3D ultrasound. The 95% limits of agreement were narrower on 3D ultrasound using the transvaginal approach (Figure 2). The ICCs were very similar between 2D and 3D ultrasound (Table 1).

Time spent on examinations

The mean ± SD time spent by Observers 1 and 2 to identify and measure the entire LUS by 2D transabdominal ultrasound was 34.8 ± 12.3 s and 35.5 ± 13.2 s, respectively. The time taken to identify and measure the LUS muscular thickness by transvaginal ultrasound was 29.4 ± 11.1 s and 30.1 ± 10.7 s for Observers 1 and 2, respectively. No significant difference was detected by one-way ANOVA.

The identification and measurement of the entire LUS using the 3D ultrasound dataset obtained on transabdominal ultrasound took 35.9 ± 14.8 s and 39.4 ± 14.3 s for Observers 1 and 2, respectively. When the 3D ultrasound dataset was obtained by transvaginal ultrasound, the time taken to identify and measure the LUS muscular thickness was 31.7 ± 14.2 s and 34.2 ± 12.8 s for Observers 1 and 2, respectively. No significant difference was detected by one-way ANOVA.

Discussion

Uterine rupture is an uncommon though potentially fatal complication of vaginal birth following a Cesarean section18. The rate in women undergoing labor after a previous Cesarean section is about 0.6–1.5%19–21. The rate is similar in women who have had multiple previous Cesarean sections (0.9%)22. Uterine rupture can lead to many complications such as intrapartum stillbirth, hypoxic–ischemic encephalopathy, neonatal death, low 5-min Apgar score, hysterectomy, transfusions and even maternal death19, 21. Moreover, this event cannot be predicted by clinical factors23.

Several trials have evaluated the use of ultrasonography in the prediction of uterine rupture, but only one has evaluated the method's reliability9. Lack of reliability in a test is responsible for different readings of the same measurement when it is made by the same observer a second time or by a second observer16. Testing and screening are critical parts of the clinical process, but unsuitable tests may put patients at risk and entail a waste of resources24, 25.

We evaluated the reliability of the ultrasonographic LUS measurement, which is indicative of the risk of uterine rupture or dehiscence2, using both ICC, which measures variation between both observers and subjects26, and 95% limits of agreement, which is the range that encompasses 95% of the differences between measurements obtained by two observers17. Our decision was based on the fact that the use of both is encouraged when evaluating reliability16. We also evaluated the proportion of the differences that were < 1 mm, since the previous study evaluating LUS reliability also used this method9.

The greater the ICC, the more reliable is the test and values > 0.75 indicate that a test has good agreement27. The interobserver ICCs obtained by the transabdominal approach were around this threshold (0.69–0.78), while the values obtained by the transvaginal approach were higher (0.87–0.90). 3D ultrasound measurement of the LUS muscular thickness by the transvaginal approach resulted in the highest interobserver ICC (0.90).

Results were similar for the 95% limits of agreement. Measurement of the entire LUS thickness by 2D transabdominal ultrasound resulted in the widest 95% limits of agreement (−3.63 to 4.06 mm), while the 3D ultrasound measurement of LUS muscular thickness by the transvaginal approach resulted in the narrowest 95% limits of agreement (−0.97 to 0.65 mm). This was the only method in which the interobserver differences were all below or close to 1 mm, the value described as being the maximum reasonable difference9. The 95% limits of agreement for 3D transvaginal ultrasound were narrower than those obtained by Jastrow et al.9 when evaluating the entire LUS by 2D ultrasound transvaginally (−2.1 to 1.6 mm), while our 95% limits of agreement for 2D transvaginal ultrasound measurements of the LUS muscular thickness were very similar.

The results for the proportion of differences < 1 mm between the two observers were also very similar to the ICC and 95% limits of agreement analysis. The measurement of LUS by transvaginal compared with transabdominal ultrasound resulted in a higher proportion of differences < 1 mm (85.83–91.25% vs. 48.75–57.71%). The 3D ultrasound measurement of the LUS muscular thickness by the transvaginal approach resulted in the highest proportion of differences < 1 mm between the two observers (91.25%), concordant with the absolute differences and 95% limits of agreement analysis.

In conclusion, we found that ultrasonographic LUS muscular thickness assessed by transvaginal ultrasound was more reliable than entire LUS thickness measured by the transabdominal approach. Nonetheless, one should consider that the association between thin LUS muscular thickness measurement obtained by transvaginal ultrasound and the risk of uterine rupture has only been suggested: all patients evaluated by studies in which LUS was assessed by the transvaginal approach underwent Cesarean section and only uterine dehiscence was observed3, 6. The actual association between thin LUS measurement and uterine rupture (in women undergoing labor, with some undesirable uterine ruptures) has been assessed only using the transabdominal approach2, 8. The use of 3D ultrasound requires specific training of sonographers, longer examination time and ultrasound machines and probes with 3D capability. Nevertheless, the use of three-dimensional ultrasound should be considered, since it resulted in a significant improvement in reliability.

Ancillary