UOG Journal Club: August 2013

1. UOG Journal Club: August 2013 Sonographic measurement of lower uterine segment thickness to predict uterine rupture during a trial of labor in women with previous Cesarean section: a meta-analysis N. Kok, I. C. Wiersma, B. C. Opmeer, I. M. De Graaf, B. W. Mol, E. Pajkrt Volume 42, Issue 2, Date: August 2013, pages 132–139 Journal Club slides prepared by Dr Tommaso Bignardi (UOG Editor for Trainees)

2. Background • The risk of uterine rupture in laboring women with a previous Cesarean section (CS) varies between 0.2% and 1.5% after induction of labor, compared to 0.5% in women with spontaneous labor after a previous CS. • Several studies have proposed that thinning of the lower uterine segment (LUS) measured by ultrasonography is a predictor of uterine rupture. • Accurate prediction of uterine rupture would allow women at low risk to proceed with a trial of labor (TOL), whereas women at high risk for uterine rupture could undergo a planned CS.

3. Sonographic measurement of lower uterine segment thickness to predict uterine rupture during a trial of labor in women with previous Cesarean section: a meta-analysis Kok et al., UOG 2013 Objective To evaluate the accuracy of antenatal sonographic measurement of lower uterine segment (LUS) thickness in the prediction of risk of uterine rupture during a trial of labor (TOL) in women with a previous Cesarean section (CS).

4. Sonographic measurement of lower uterine segment thickness to predict uterine rupture during a trial of labor in women with previous Cesarean section: a meta-analysis Kok et al., UOG 2013 Inclusion criteria • Studies on pregnant women with at least one previous CS • Studies that reported on sonographic appearance of LUS during pregnancy in relation to uterine defects observed during or immediately after delivery • Studies that allowed construction of 2×2 tables comparing LUS thickness measurement and the occurrence of uterine scar defects (uterine scar dehiscence or uterine scar rupture) • 1980 – December 2011

5. Sonographic measurement of lower uterine segment thickness to predict uterine rupture during a trial of labor in women with previous Cesarean section: a meta-analysis Kok et al., UOG 2013 Definitions • Uterine scar dehiscence: loss of continuity of myometrial layer without complete rupture of LUS • Uterine rupture: complete separation of the uterine scar resulting in communication between the uterine and peritoneal cavities • Full LUS thickness: distance between bladder wall and amniotic cavity • Myometrial thickness: minimum thickness overlying amniotic cavity at the level of uterine scar (only myometrium is measured)

6. Total citations (n = 297) screened for relevance: PubMed (n = 143); EMBASE (n = 150); Reference lists (n = 4) References excluded because of duplication (n = 150) References excluded after screening title (n = 84) References excluded after screening abstract (n = 29) Citations retrieved for more detailed evaluation of full manuscripts (n = 34) Studies excluded because of inappropriate reporting of outcome (n = 10) or language restrictions (n = 3) Studies included in systematic review (n = 21)

7. Sonographic measurement of lower uterine segment thickness to predict uterine rupture during a trial of labor in women with previous Cesarean section: a meta-analysis Kok et al., UOG 2013 Methods • Methodological quality assessment: Quality Assessment of Diagnostic Accuracy Studies (QUADAS) tool • Data independently extracted by four reviewers • Data extracted: sonographic LUS thickness during pregnancy (index test), definition of uterine scar defect, full or myometrial LUS measurement, transabdominal or transvaginal measurement, level of experience of ultrasound examiners, number of examiners, number of measurements, gestational age at measurement, a priori determined threshold for LUS thickness, blinding, setting, study population, study design, data collection, number of participants, adverse neonatal or maternal outcome, VBAC success rate and prevalence of a uterine defect

8. Sonographic measurement of lower uterine segment thickness to predict uterine rupture during a trial of labor in women with previous Cesarean section: a meta-analysis Kok et al., UOG 2013 Methods • For each study a two-by-two table was constructed, cross-classifying LUS thickness measured by ultrasound and the presence of LUS defect after delivery • A bivariate meta-regression model was used to calculate pooled estimates of sensitivity and specificity and to calculate the corresponding summary ROC (sROC) curve • Separate sROC curves were calculated for full and myometrial LUS measurements

9. Results: characteristics of studies included Results NR, not reported; Prosp., prospective cohort; Retro., retrospective cohort; TAS, transabdominal sonography; TVS, transvaginal sonography

10. Results Summary of the Quality Assessment of Diagnostic Accuracy Studies (QUADAS) tool for articles included

11. Results: sROC curves (−) and pooled sensitivity and specificity ( ●) for prediction of uterine defects* sensitivity sensitivity 1-specificity 1-specificity Myometriallower uterine segment (LUS) thickness Fulllower uterine segment (LUS) thickness *Rectangles show the observed accuracy for each cut-off point in each study

12. Results sensitivity sensitivity 1-specificity 1-specificity Myometrial LUS thickness cut-off ranges 0.6–2.0 mm: ○ observed accuracy; ● pooled sens/spec; ─ sROC curve 2.1–4.0 mm: □ observed accuracy; ■ pooled sens/spec; - - sROC curve Full LUS thickness cut-off ranges 2.0–3.0 mm: ○ observed accuracy; ● pooled sens/spec; ─ sROC curve 3.1–5.1 mm: □ observed accuracy; ■ pooled sens/spec; - - sROC curve

13. Sonographic measurement of lower uterine segment thickness to predict uterine rupture during a trial of labor in women with previous Cesarean section: a meta-analysis Kok et al., UOG 2013 Results • Full LUS thickness measurement between 2.0 and 3.0 mm reached a specificity of 0.91 (95% CI, 0.80–0.96) at a sensitivity of 0.61 (95% CI, 0.42–0.77). • Full LUS thickness measurement between 3.1 and 5.1 mm reached a specificity of 0.63 (95% CI, 0.30–0.87) at a sensitivity of 0.96 (95% CI, 0.89–0.98) • The accuracy of TVS and TAS could not be compared statistically

14. Sonographic measurement of lower uterine segment thickness to predict uterine rupture during a trial of labor in women with previous Cesarean section: a meta-analysis Kok et al., UOG 2013 Discussion • Strong negative correlation between LUS thickness and risk of • uterine defect • Similar ROC curves for myometrial and full LUS thickness, • indicating no significant difference in any of the three parameters: • accuracy, shape and position

15. Sonographic measurement of lower uterine segment thickness to predict uterine rupture during a trial of labor in women with previous Cesarean section: a meta-analysis Kok et al., UOG 2013 Strengths of the study • Thorough search without language restrictions Limitations • Considerable amount of heterogeneity among studies with use of • different cut-offs and variable definition of uterine defect • Large number of small studies (inclined to overestimate the predictive • capacity of LUS thickness) • More than 75% of studies not blinded

16. Sonographic measurement of lower uterine segment thickness to predict uterine rupture during a trial of labor in women with previous Cesarean section: a meta-analysis Kok et al., UOG 2013 • Conclusions • This meta-analysis provides support for the use of antenatal LUS • measurements in the prediction of a uterine defect during TOL. • Clinical applicability should be assessed in prospective observational • studies using a standardized method of measurement.

17. Sonographic measurement of lower uterine segment thickness to predict uterine rupture during a trial of labor in women with previous Cesarean section: a meta-analysis Kok et al., UOG 2013 • Discussion points • Do we need a consensus on terms and definitions regarding ‘uterine defects’? • What is the clinical significance of a ‘silent’ uterine scar dehiscence? • Do we need a consensus regarding which layer(s) of the LUS should be measured and by which route? • What is the agreement on such measurements between different observers? • Can clinical factors influence the accuracy of this tool?