Sonographic Lower Uterine Segment Thickness and Risk of Uterine Scar Defect: A Systematic Review

Objective: To study the diagnostic accuracy of sonographic measurements of the lower uterine segment (LUS) thickness near term in predicting uterine scar defects in women with prior Caesarean section (CS). Data Sources: PubMed, Embase, and Cochrane Library (1965-2009). Methods of Study Selection: Studies of populations of women with previous low transverse CS who underwent third-trimester evaluation of LUS thickness were selected. We retrieved articles in which number of patients, sensitivity, and specificity to predict a uterine scar defect were available. Data Synthesis: Twelve eligible studies including 1834 women were identified. Uterine scar defect was reported in a total of 121 cases (6.6%). Seven studies examined the full LUS thickness only, four examined the myometrial layer specifically, and one examined both measurements. Weighted mean differences in LUS thickness and associated 95% confidence intervals between women with and without uterine scar defect were calculated. Summary receiver operating characteristic (SROC) analysis and summary diagnostic odds ratios (DOR) were used to evaluate and compare the area under the curve (AUC) and the association between LUS thickness and uterine scar defect. Women with a uterine scar defect had thinner full LUS and thinner myometrial layer (weighted mean difference of 0.98 mm; 95% CI 0.37 to 1.59, P = 0.002; and 1.13 mm; 95% CI 0.32 to 1.94 mm, P = 0.006, respectively). SROC analysis showed a stronger association between full LUS thickness and uterine scar defect (AUC: 0.84 +/- 0.03, P < 0.001) than between myometrial layer and scar defect (AUC: 0.75 +/- 0.05, P < 0.01). The optimal cut-off value varied from 2.0 to 3.5 mm for full LUS thickness and from 1.4 to 2.0 for myometrial layer. Conclusion: Sonographic LUS thickness is a strong predictor for uterine scar defect in women with prior Caesarean section. However, because of the heterogeneity of the studies we analyzed, no ideal cut-off value can yet be recommended, which underlines the need for more standardized measurement techniques in future studies.