The accuracy of auscultatory detection of fetal cardiac decelerations: a computer simulation

To evaluate current practices of auscultation for the detection of decelerations, we used a computer to generate contractions and late decelerations and perform the counting. The baseline rate ranged from 110 to 180 beats/min. The duration of the deceleration ranged from 1 to 2 minutes, and the amplitude of the deceleration ranged from 10 to 90 beats/min. The onset of the decelerations ranged from 0.4 to 0.7 of the length of the contraction. Counting was begun at 80%, 100%, and 120% of the contraction length. The duration of counting varied between 15 and 60 seconds. A multicount algorithm obtained three 10-second counts separated by 5 seconds. Results were classified by the ability to detect rates below 120, 100, or 80 beats/min (threshold determination) or 20 and 25 beats/min below the baseline rate (subtraction determination). The baseline rate and deceleration amplitude had the greatest effect on accuracy. The higher the baseline rate and the smaller the deceleration amplitude, the less accurate was detection. Multiple counts were more accurate than the single-count strategy, and subtraction detection was more accurate than threshold detection. The effects of counting error are briefly described. This model, which requires clinical confirmation, nevertheless emphasizes the potential inaccuracies of many popular schemes of auscultatory surveillance, even for the detection of prolonged or sustained decelerations. Certain modifications of auscultatory practice may improve the accuracy of this technique.