Approximate entropy, a statistic of regularity, applied to fetal heart rate data before and during labor

To determine whether approximate entropy (ApEn), a new statistic of regularity, when applied to fetal heart rate (FHR) data antepartum or in labor, would offer an advantage over standard statistics of variation in predicting outcome.

A large data base of antepartum FHR records collected in clinical practice over 10 years was available. Two data sets in labor were stored on disk in small computers interfaced to fetal monitors on-line. Outcomes were assessed using blood gas values on delivery and Apgar scores.

Antepartum, when the most favorable form of ApEn was used on 769 good-quality FHR records, the correlation with measurement of short-term variation was high. This was especially true when the fetal pulse interval variation fell below the normal range (less than 6 milliseconds short-term; r = 0.93) and in 20 other records with sinusoidal variation (r = 0.96). Approximate entropy varied with fetal sleep cycles and took longer to calculate than FHR variation. During the last hour of labor, in 319 records, there was no significant correlation between umbilical artery base deficit values on delivery and ApEn measurement. In 871 additional good-quality records of fetuses with normal outcome, the mean (+/- standard error [SE]) ApEn (0.95 +/- 0.005) was significantly greater than in 22 records (0.88 +/- 0.028) from fetuses with abnormal outcome (umbilical artery base deficit more than 12 mmol/L and Apgar score of 3 or less at 1 minute). However, consideration of the frequency distributions of these measurements showed that ApEn did not discriminate between normal and abnormal outcomes. The SD of fetal pulse intervals rose in labor whereas ApEn values fell, confirming that this new statistic of regularity differs from standard statistics of variation.

Approximate entropy offered no advantage over measurement of short-term FHR variation antepartum, and neither measurement predicted outcome in labor.