Inter-observer variability in the assessment of ultrasound features of polycystic ovaries

Abstract Objective To evaluate inter-observer variability in the assessment of ultrasound features of polycystic ovaries. Design Prospective cohort study. Population 60 females with known polycystic ovarian syndrome (PCOS). Setting Two tertiary care hospitals in Amman, Jordan; Prince Hamza Hospital and Albashir Hospital. Methods A total of 120 transvaginal scans were performed on 60 study participants with known PCOS by two gynecologists. The ovaries were evaluated for the presence or absence of PCOS criteria: The number of follicles and the volume of each ovary. The correlation coefficient was calculated. Bland-Altman plots were used to analyze any discrepancies in the measurement of ovaries between the observers. Main outcome measures Identification of ultrasound features of PCOS will be reproducible by different gynecologists. Results The mean follicle count and ovarian volume in PCOS patients reported in this study were 21 cm 3 follicles and 11.5 cm 3 respectively. The Pearson correlation coefficient was calculated and it can be seen that there was low inter-observer correlation in follicular count (0.560) and a moderate correlation in measuring ovarian volume (0.770). Bland-Altman plots show low inter-observer agreement in follicular count and high inter-observer agreement in the measurement of ovarian volume. Conclusion The inter-observer agreement in the assessment of ultrasound features of polycystic ovaries is not acceptable. The diagnostic criteria of polycystic ovarian morphology need to be revisited.