Congenital Uterine Malformation by Experts (CUME): diagnostic criteria for T‐shaped uterus

ABSTRACT Objectives To identify uterine measurements that are reliable and accurate to distinguish between T‐shaped and normal/arcuate uterus, and define T‐shaped uterus, using Congenital Uterine Malformation by Experts (CUME) methodology, which uses as reference standard the decision made most often by several independent experts. Methods This was a prospectively planned multirater reliability/agreement and diagnostic accuracy study, performed between November 2017 and December 2018, using a sample of 100 three‐dimensional (3D) datasets of different uteri with lateral uterine cavity indentations, acquired from consecutive women between 2014 and 2016. Fifteen representative experts (five clinicians, five surgeons and five sonologists), blinded to each others' opinions, examined anonymized images of the coronal plane of each uterus and provided their independent opinion as to whether it was T‐shaped or normal/arcuate; this formed the basis of the CUME reference standard, with the decision made most often (i.e. that chosen by eight or more of the 15 experts) for each uterus being considered the correct diagnosis for that uterus. Two other experienced observers, also blinded to the opinions of the other experts, then performed independently 15 sonographic measurements, using the original 3D datasets of each uterus. Agreement between the diagnoses made by the 15 experts was assessed using kappa and percent agreement. The interobserver reliability of measurements was assessed using the concordance correlation coefficient (CCC). The diagnostic test accuracy was assessed using the area under the receiver‐operating‐characteristics curve (AUC) and the best cut‐off value was assessed by calculating Youden's index, according to the CUME reference standard. Sensitivity, specificity, negative and positive likelihood ratios (LR– and LR+) and post‐test probability were calculated. Results According to the CUME reference standard, there were 20 T‐shaped and 80 normal/arcuate uteri. Individual experts recognized between 5 and 35 (median, 19) T‐shaped uteri on subjective judgment. The agreement among experts was 82% (kappa = 0.43). Three of the 15 sonographic measurements were identified as having good diagnostic test accuracy, according to the CUME reference standard: lateral indentation angle (AUC = 0.95), lateral internal indentation depth (AUC = 0.92) and T‐angle (AUC = 0.87). Of these, T‐angle had the best interobserver reproducibility (CCC = 0.87 vs 0.82 vs 0.62 for T