Reliability and validity of two methods of three-dimensional cervical volume measurement

Objectives To determine if cervical length obtained with three-dimensional ultrasound correlated with the 'true cervical volume' and to evaluate the reliability and validity of transabdominal and transvaginal three-dimensional cervical volume measurement. Methods This was a prospective observational study. Three-dimensional cervical volume measurements were made prior to hysterectomy in 28 women. Following hysterectomy the amputated cervical volume was calculated using water displacement. For the assessment of intra- and interobserver reliability, the intraclass correlation coefficient (ICC) was used. The index of concordance between the sonographic cervical volumes and those obtained by the reference standard (true cervical volume) was assessed with the limits of agreement method and the ICC. Results Transabdominal cervical length and transvaginal cervical length correlated moderately with actual cervical volume; correlation coefficients were 0.64 and 0.57 (P < 0.05), respectively. Intraobserver reliability for both transabdominal and transvaginal cervical volume assessment was good (> 0.75). Interobserver reliability for transvaginal cervical volumes was similarly good (ICC = 0.90). However, for transabdominal measurements the interobserver reliability was poor (ICC = 0.51). The validity of both methods of three-dimensional volume assessment was poor (ICC < 0.75). This was reflected in the wide limits of agreement, which ranged from approximately - 25 mL to + 30 mL. Conclusion The reliability and validity of three-dimensional cervical volume measurement are poor. Clinical introduction of cervical volume measurement should be avoided at this time. Copyright (C) 2003 ISUOG. Published by John Wiley & Sons, Ltd.