P23.01: Doppler colour scoring system using offline video analysis in women with an incomplete miscarriage: inter‐ and intraobserver reproducibility study

Abstract

To assess inter- and intraobserver reproducibility of the Doppler colour scoring (DCS) system in women with 1st trimester incomplete miscarriage noted on transvaginal ultrasound (TVS). 32 digital videosets were recorded of women with incomplete miscarriage (presence of measurable focus of hyperechoeic material) on TVS. Vascularisation of retained products of conception (RPC) was recorded using DCS system of IOTA group: DCS 1=absence of vascularity, DCS 2=minimal vascularity, DCS 3=rather strong, DCS=4 very strong. CDS has been previously used to predict successful expectant management of incomplete miscarriage (Casikar et al. Hum Reprod 2013). Each videoset was then reviewed offline by 5 observers who apportioned a DCS to RPC. Same videos reanalysed, albeit in a different order, at least 7 days later, to assess intraobserver agreement. Inter- and intraobserver correlation was performed to determine agreement among the 5 observers. Interobserver agreement was also measured between each observer and reference standard (GC). Cohen's κ coefficient <0 suggests poor agreement, 0.01-0.20 slight, 0.21-0.40 fair, 0.41-0.60 moderate, 0.61-0.80 substantial, 0.81-0.99 almost perfect. Videoset sample size: assuming participants agree 60% of the time, a sample size of 31 is required with a relative error of 30%(Gwet, 2010). 61% and 89% of cases had agreement between >3 and >2 observers, respectively. Interobserver agreement ranged from moderate to substantial (0.48-0.751); agreement with GC ranged from moderate to substantial (0.495-0.678). 2 observers had almost perfect intraobserver agreement (0.905-0.912), 3 observers had substantial intraobserver agreement (0.653-0.691). Overall interobserver agreement for the CDS was substantial (0.626). CDS interobserver reproducibility between all observers and the reference standard ranged from moderate to substantial. CDS intraobserver reproducibility was substantial to almost perfect.