Inter- and intraobserver reliability for diagnosing levator ani changes on magnetic resonance imaging

OBJECTIVES: To assess the inter- and intraobserver reliability of the diagnosis of pubovisceral muscle avulsions and measurements of the levator hiatus on magnetic resonance imaging (MRI). METHODS: Women with recurrent pelvic organ prolapse or in whom there was a discrepancy between clinical signs and symptoms of pelvic floor dysfunction underwent MRI and were eligible for inclusion. MRI datasets of the pelvic floor of 262 women were obtained and evaluated by two observers, who scored the presence and extent of pubovisceral muscle avulsions on each side using a scale from 0 to 3 and obtained measurements of the anteroposterior and transverse diameters and area of the levator hiatus. A random sample of 100 patients was reviewed a second time by one of the observers. Intraclass correlation coefficients (ICCs) with their 95% CI were calculated for all measurements. Mean differences with accompanying limits of agreement were calculated to estimate agreement between pairs of measurements and to detect possible systematic bias. RESULTS: Good interobserver reliability was found for the assessment of pubovisceral muscle avulsions (ICC = 0.76-0.79) and excellent agreement for measurements of the levator hiatus (ICC = 0.85-0.89). The intraobserver reliability for pubovisceral muscle avulsions and other levator hiatus measurements was also excellent (ICC = 0.80-0.97). A significant interobserver systematic bias was observed in the measurement of levator hiatus transverse diameter; however, narrow limits of agreement were observed. CONCLUSIONS: Pubovisceral muscle avulsions and levator hiatus measurements can be assessed with good to excellent reliability on MRI. Copyright (c) 2013 ISUOG. Published by John Wiley & Sons Ltd