A comparison between visual and computer analysis of antepartum fetal heart rate tracings

This study was designed to determine the intraobserver and interobserver variability in the visual assessment of fetal heart rate tracings and to evaluate the accuracy of the visual detection of accelerations and decelerations when compared with computerized fetal heart rate analysis.

One hundred antepartum fetal heart rate tracings, of good quality and of 30 minutes' duration, were visually assessed by five expert observers on three occasions during a 12-month period. There were a total of seven questions related to either judgment or accuracy of each recording. Visual detection of fetal heart rate acceleration, deceleration, and estimated baseline was compared with computerized analysis. Statistical significance was determined by kappa coefficient and contingency table chi 2 analysis.

Analysis with kappa coefficient reflecting intraobserver and interobserver agreement of judgment-related questions indicated poor agreement between observers when assessing short-term fetal heart rate variability (kappa 0.18), when making a decision to stop or continue recording (kappa 0.39), and when judging whether there is concern regarding fetal heart rate tracing (kappa 0.26). When compared with computerized analysis, clinicians had a tendency to recognize tracings as normal. In particular, they failed to identify 35% of tracings with 0 or 1 acceleration and failed to detect 92% of fetal heart rate decelerations. Variable decelerations associated with fetal movements were the major source of disagreement between observers and the computer. Only the estimation of baseline fetal heart rate had a high level of accuracy.

We concluded that the poor level of accuracy in several components of the nonstress test was responsible for the low interobserver agreement seen in simple judgment-related questions such as decision to continue or stop fetal heart rate recordings.