P17.07: Cervical length assessment: How accurate are sonographers?

Abstract

Objectives: To determine how local inter-operator and intraoperator variability of sonographers performing trans-vaginal cervical length measurements compares with previous published studies. Methods: 27 women attending for TV US were recruited over a 6 month period for cervical length assessment. 3 images of the cervix were taken using a standardised technique and image criteria. Each image was measured blind twice during the examination and re-measured blind by 3 sonographers at a later date. Each sonographer selected the measurement they considered to be their best measurement for each participant. Intra-operator variability was assessed by comparing the 2 measurements sonographers took of their best image (method 1) and the largest difference between measurements for each woman (method 2). In total 162 pairs of measurements were compared for in the intra-operator analysis. 162 pairs of measurements were also compared in the inter-operator analysis; 81 pairs compared the sonographer’s best measurements (method 3) and a further 81 pairs compared the measurements each sonographer took of the image the sonographer who scanned the patient selected as the best image (method 4). Results: Intra-operator variability calculated by method 1 produced repeatability coefficient, 1.3 to 3.5 mm and method 2 repeatability coefficient, 8.3 to 9.4 mm. Inter-operator variability calculated by Method 3 produced variability of 7.2 to 7.9 mm and method 4 variability of 7.1 to 10.0 mm. Conclusions: The intra-operator variability was worse than that reported by previous studies. The inter-operator variability was comparable to previous similar sized studies. The relatively large intra-operator and inter-operator variability between different images is a concern as it has the potential to influence patient management, particular when the cervix measures between 15 mm–25 mm, a stage when intervention may be considered. This suggests that the standardised method for image acquisition has not been successful at limiting variability in measurements.