Consistency and reproducibility of sonographic diagnosis of adenomyosis: an interobserver variability study

To study whether there is a marked interobserver discrepancy in the diagnosis of adenomyosis based on reviewing stored sonographic images of infertile patients?

Three reproductive endocrinology and infertility specialists reviewed stored ultrasound images of 2134 patients attending tertiary IVF-centers. The physicians were blinded to each other as well as to the original diagnosis. The revised Morphological Uterus Sonographic Assessment (MUSA) criteria was used.

The overall three-rater agreement for adenomyosis (Yes/No) was 62.3%, with Fleiss' κ = 0.317 (z = 25.4; p < 0.001), indicating fair interobserver reliability. Within cases where all three raters diagnosed adenomyosis (n = 179), interobserver reliability for revised MUSA signs ranged from slight to fair. Overall, even among adenomyosis-positive cases, reproducibility of individual signs varied considerably, with myometrial cysts being the most commonly agreed upon direct feature (K 0.315, 95% CI 0.213-0.412) and asymmetrical myometrial thickening for the indirect features (K 0.406, 95% CI 0.321-0.491).

Our findings demonstrate only fair interobserver reproducibility for the sonographic assessment of adenomyosis based on the assessment of stored static ultrasound images using the revised MUSA criteria. This degree of variability highlights the challenges associated with consistent interpretation of sonographic features and suggests a need for further refinement and improved standardization of diagnostic approaches.