Intraobserver and interobserver reproducibility of third trimester uterine artery pulsatility index

The primary aim of the study was to investigate intraobserver and interobserver reproducibility of uterine artery (UtA) pulsatility index (PI) in the third trimester of pregnancy. The secondary aim of the study was to examine whether high maternal body mass index (BMI) or gestational age (GA) influence the reliability of this measurement.

Singleton pregnancies in women with known BMI were recruited between 30⁺⁰ and 40⁺⁰  weeks. UtA PI Doppler measurements were performed prospectively, independently, and blindly by 2 Fetal Medicine Foundation-accredited operators. Intraobserver and interobserver reproducibility was assessed by concordance correlation coefficient (CCC) and intraclass correlation coefficient (ICC); Bland-Altman plots were built and limits of agreement (LoA) were calculated. The analysis was performed for both raw numbers and percentiles. To estimate the intraoperator and interoperator agreement in defining normal and pathological measurements, the assessments were divided in 2 categories-UtA PI <95th percentile/ ≥95th percentile-and Cohen's kappa coefficients were calculated. Results were interpreted according to the cutoffs reported by the True Reproducibility of Ultrasound Techniques review. Correlation between maternal BMI and GA and accuracy of UtA measurements was studied with Spearman's correlation coefficient.

Measurements were available in 101 women. For intraobserver reproducibility, ICCs and CCCs were calculated for raw values and percentiles and were 0.912 and 0.835, and 0.837 and 0.716, respectively. For interobserver reproducibility, ICCs and CCCs were 0.809 and 0.732, and 0.677 and 0.576, respectively. This indicates a poor-moderate reproducibility of third trimester UtA PI. LoA were also wide (from a minimum of -0.30-0.35 to a maximum of -0.53-0.62). Cohen's kappa coefficients were 0.478 and 0.418, showing a moderate intraoperator and interoperator agreement in distinguishing between normal and pathological values. No correlation was found between maternal BMI and GA and reproducibility of the measurements.

Intraobserver and interobserver reproducibility of third trimester UtA PI as assessed by ICC, CCC, and LoA is only moderate-poor. The agreement between operators in defining pathological and normal measurements is moderate.