OC07.01: The International Endometrial Tumor Analysis (IETA) study: interim analysis of measurement differences between centres

In multicentre studies, measurements will vary between centres for various reasons (e.g. subjectivity of measurements, differences in equipment, deviations from the protocol and differences in patient populations). The aim of this interim analysis by the International Endometrial Tumor Analysis group is to identify the variables with the strongest intercentre differences. 1864 consecutive women presenting with abnormal uterine bleeding underwent ultrasound examination by one of 20 international investigators. Women with an available histological diagnosis (n = 1504) were included in the analysis. We studied the variability in measurements of the endometrial thickness (unenhanced ultrasound examination (UUE) and fluid instillation sonography (FIS)), endometrial junction (UUE), presence of fibroids, patient age, BMI, use of anticoagulant therapy and use of hormonal therapy. The statistic used was the residual intraclass correlation (RICC) after adjustment for the type of lesion. An RICC of 0% indicates that the variability in the measurements is not due to differences between investigators. Normal endometria (n = 593) were distinguished from endometria with malignant (n = 100), focal (n = 609), diffuse (n = 116) and other (n = 86) lesions. The strongest differences between investigators were observed for the assessment of the endomyometrial junction as irregular (RICC = 56.9%), undefined (48.4%), or interrupted (36.9%) vs. regular. The use of anticoagulant therapy also varied across investigators (RICC = 41.7%), with 11 of 20 investigators not reporting anticoagulant therapy for any patient. The reported presence of fibroids, BMI and thickness of the endometrium (UUE) exhibited the smallest differences between investigators (RICC < 6%). The results suggest that the assessment of the endomyometrial junction is difficult and subjective and that anticoagulant therapy use is underreported. This should be considered when analysing the data.