Analysis of antepartum fetal heart rate tracing by physician and computer

Abstract

Objective To compare the interpretation of fetal heart rate (FHR) tracings by three obstetricians with that of a computer analysis program. Methods Our study population consisted of high-risk pregnant women referred as outpatients for antepartum FHR monitoring. A total of 121 FHR tracings, from a series of 54 consecutive women, were interpreted by three physicians and a computer program (Oxford Sonicaid System 8000, Oxford Sonicaid Ltd., Chichester, UK). The physicians used a modified FHR scoring system to interpret the tracings. Total scores were categorized as 0–4: abnormal, 5–7: questionable, and 8–10: normal. The computer program used overall variation, categorized as normal: longer than 30 ms, abnormal: shorter than 20 ms, and questionable: 20–30 ms. Results Significant differences were found among the physicians and between the physicians and the computer analysis for the individual elements of FHR tracings. There was very good agreement between two physicians and the computer in the assessment of the FHR baseline. When physicians used a FHR scoring system to classify the tracings as normal, questionable, or abnormal, the agreement was poor (kappa values ranged from -0.037 to 0.28). The computerized analysis identified two FHR tracings as questionable but both were classified as normal by all three physicians. Conclusions The level of agreement in the interpretation of FHR tracings was poor among physicians and between physicians and the computer analysis. A FHR scoring system did not improve the level of agreement between physicians. J. Matern.-Fetal Med. 9:181–185, 2000. © 2000 Wiley-Liss, Inc.