Objective and subjective assessment of abnormal uterine artery Doppler flow velocity waveforms

Objective Abnormal uterine artery Doppler waveforms in the second trimester of pregnancy can predict adverse outcome. The presence of a diastolic notch is a better predictor than measurement of impedance indices. A notch has generally been assessed subjectively, which has raised questions as to the reproducibility of the method. The aim of this study was to compare subjective and objective assessments of 50 abnormal waveforms. Design This was a retrospective study of 50 abnormal uterine artery waveforms collected during a previous screening study. Subjective and objective assessments were compared and related to adverse outcomes. Five impedance indices incorporating a measurement of the diastolic notch were compared. Three indices predictive of adverse outcome were compared to subjective assessment. Results There was good agreement between experienced operators in their subjective assessment of uterine artery waveforms. Adverse outcome was predicted by only moderately or severely abnormal waveforms. Three impedance indices predictive of adverse outcome were the pulsatility index (PI), AIC and (D-C)IB (where A is the peak systolic, D is the maximum diastolic and B is the end-diastolic frequency, and C is the nadir of the notch). The cut-off values combining the highest sensitivity for the lowest false-positive rate were 1.5 for PI; 3.0 for AIC and 0.15 for (D - C)IB. A PI of 1.5 had the highest sensitivity (100%) and positive predictive value (55%) for adverse outcome in this preselected group. These three indices were compared to subjective assessment using the kappa index and the (D-C)IB ratio had the closest agreement, although all three indices showed at least moderate agreement. The cut-off values of 1.5, 3.0 and 0.15 for PI, AIC and (D - C)IB, respectively appear to differentiate between mildly abnormal waveforms and moderately/severely abnormal waveforms. Conclusion Objective assessment of uterine artery waveforms using cut-off values of PI, AIC and (D - C)IB provides an estimation similar to that provided by subjective assessment by an experienced operator. These ratios will now need to be used prospectively in an unselected population to compare their screening properties.