| Abstract | The effects of different rejection limits on the results of analysis of neonatal heart rate variability from electrocardiograms by a microprocessor-based system were studied on fourteen infants after normal labor and on ten infants after elective cesarean section. In addition to the infant's movements and crying, two further main sources of error in the calculation of variability indices were detected: premature beats and problems in the shape of QRS complexes in a neonatal electrocardiogram. No noteworthy problem was observed in the calculation of the interval index (II), which describes the long-term variability of heart rate. In the calculation of the differential indices (DI), which describe the short-term component of heart rate, distorted QRS complexes created very high DI's when the rejection logic was 10 beats per minute. When stricter rejection limits were used (5 bpm), the DI values even in these cases fell within the normal range. The DI values calculated using the rejection limit of 5 bpm were always lower than those calculated using a rejection limit of 10 bpm, but the reference values of neonatal II and DI after normal labor were similar to those presented previously using a rejection limit of 10 bpm. In conclusion, the neonatal electrocardiogram appears to be a noisy signal, comparable with the abdominal fetal electrocardiogram, and strict rejection limits are useful in processing it for indices of variability. |