MR pelvimetry measurements, analysis of inter- and intra-observer variation

To set reference standards and then to evaluate the measurement variation in magnetic resonance (MR) pelvimetry between observations and observers.

The study was carried out retrospectively using 100 MR pelvimetry examinations performed in North Carelian Central Hospital between September 2006 and January 2008. Pelvimetric parameters of pelvic inlet and outlet were measured four times to determine the standard reference for each measurement and then intra- and inter-observer variations were compared.

The accuracy of MR pelvimetry, defined as a deviation of <5mm from the reference standard, was better in pelvic inlet measurements than in the corresponding outlet measurements (95-99% vs. 86-89%). Intra-observer variation was acceptable in all of the measurements with intraclass correlation coefficient (ICC) in the range 0.956-0.981 in all parameters. Inter-observer variation was higher than intra-observer variation. The largest variation of measurements was in pelvic outlet parameters between observers with the ICC in the range of 0.710-0.813.

MR pelvimetry measurement should be conducted in a centralized location to decrease observer-related variation. Clinicians should be aware that millimeter differences are not reliable in MR pelvimetry and therefore the use of millimeter accurate limits are not recommended in obstetric decision making.