British Biomedical Bulletin

Objective To assess the possibility to diagnose placental insufficiency and metabolic acidosis by analysing the fetal heart rate (FHR) variability by using non-linear methods; symbolic dynamics or traditional linear stochastic theory. Study design The recorded FHR signal sets from the last hour of delivery in 21 live born fetuses were analysed. The complexity of the FHR variability was described by an entropy parameter using symbolic dynamics as well as by standard deviation of each FHR. Cases with placental insufficiency and metabolic acidosis were identified from cord artery blood gas analyses. Results No significant correlation was found between the entropy parameter and the arterial pH or between the standard deviation and the arterial pH. Results are improving when combining the two (entropy parameter and standard deviation) in a multiple regression analysis. Only one of the analysed cases was identified with metabolic acidosis (cord artery pH < 7,0 and Base Deficit > 12,0 mmol/L). Conclusion To know if it is possible to diagnose placental insufficiency and metabolic acidosis by analysing the FHR variability by using nonlinear methods, symbolic dynamics or traditional linear stochastic theory, a study with FHR not influenced by uterine contractions during labour is needed to conclude if the lack of correlation is caused by uterine contractions or insufficient mathematical methodology. © 2015 British Biomedical Bulletin. All rights reserved Storm et al______________________________________________________ ISSN-2347-5447 BBB[3][2][2015] 190-198 Introduction Placental insufficiency is a condition that can be fatal to the fetus. The decrease in oxygen may lead to fetal hypoxia and metabolic acidosis. Placental insufficiency and fetal hypoxia may be suspected by typical changes in the CTG in terms of baseline fetal heart rate, variability, accelerations and decelerations. However, CTG is an unreliable method and a Doppler ultrasound examination of the umbilical arteries and the vessels in the fetal brain is needed to further confirm fetal hypoxia. These procedures are both time consuming and requires experienced midwifes to recognise the CTG patterns and a special trained obstetrician to perform the ultrasound examination. Several studies show that the use of non-linear dynamics (chaos theory) is better than ordinary linear stochastic theory, when trying to detect pathology in the human organs. One study reported that chaos theory measures a decrease in heart rate variability in those patients who are in high risk of developing ventricular fibrillation (p<0.001, sensitivity 91%, specificity 85%). The patients’ heart frequencies standard deviation (SD) could not tell whom of the patients that would experience ventricular fibrillation. Using non-linear dynamics has shown good correlation between reduced heart beat complexity and death rate in pigs with myocardial ischema inflicted on them. Moreover, one study have examined sheep fetuses and found that fetuses exposed to long-term hypoxia are associated with a reduction in fetal heart rate (FHR) variability compared to normal fetuses. This is possibly due to a delay in the normal maturational changes of the autonomic control of FHR. The acute reaction (until 20 hours past the first hypoxic event) was a transitory increased FHR variability. The FHR variability is normally increasing with advancing gestational age and hypoxemia seems to delay the normal maturation of the autonomic control. Another study showed no decrease in FHR variation when healthy women at term pregnancy were exposed to short-term hypoxia. However, in a group of women with pregnancy-related complications (severe pre-eclampsia and/or severe growth retardation), a reduced FHR variation was found in relation to increasing resistance in the placenta. This change in FHR variability will possibly vary depending on what caused the hypoxia and its duration, e.g. a sudden event during labour or placental insufficiency. Chronic hypoxia in the fetus will lead to a decrease in the natural variations in the fetus heart rate. Labour is a potential threat to fetal wellbeing. Most fetuses will have sufficient metabolic reserve to withstand the effect of reduced oxygen supply during uterus contractions. A distressed fetus could be due to limited oxygen reserves caused by placental insufficiency. We are searching a way to predict placental insufficiency in advance of delivery by use of symbolic dynamics. The purpose of this study is to apply the symbolic dynamic model on the fetus heart rates and test the hypothesis that symbolic dynamics can be used to indicate which fetuses suffer from hypoxia due to placental insufficiency. We also want compare the use of symbolic dynamics with the standard deviation as a predictor of hypoxia when analyzing the FHR variability. Materials and Methods Materials We analysed FHR signal sets from the last hour of delivery in 35 live born fetuses. Storm et al____________________________________________________________________ BBB[3][2][2015] 190-198 The recordings were collected from three Norwegian and Swedish delivery units from June 1998 to January 1999 as a part of a Nordic observational multi-centre study. Those eligible for the study were women in active labour at more than 36 completed gestational weeks, and for whom a clinical decision had been made to apply a fetal scalp electrode for continuous internal CTG recording. The multi-centre study originally recorded FHR from 573 cases. The acid-base status of the cord artery and vein was recorded to assess the condition of the child at birth. Thirty-three non-hypoxic fetuses were randomly picked from the Nordic database to form our material in this study. None of the fetuses had major cardiac anomalies. The ethics committees of the participating hospitals approved the study and all mothers gave informed consent. In our material we have information about umbilical cord blood gas in 35 fetuses. However only 21 of those were from the cord artery limiting our population to 21. Umbilical cord blood gas Umbilical cord blood gas analyses have been used to evaluate fetal oxygenation. Both arterial pH and Pco2/Base Deficit are essential values. The Pco2 /Base Deficit are used to differentiate between respiratory or metabolic acidosis. The umbilical cord PO2 or O2 saturation is not useful, as many normal newborns are initially hypoxemic until normal extrauterine respiration is established. Respiratory acidosis is not predictive of newborn injury or long-term injury. According to the Nordic multi-centre study we have used a cut-off for metabolic acidosis: cord artery pH<7.00 and Base Deficit >12.0 mmol/l or in case of cord vein sample only pH<7.10 and Base Deficit >12.0 mmol/l. These limits should identify if a fetus was clinically affected from intrapartum hypoxia. Data acquisition and signal processing The fetal ECG was recorded during delivery with an intrauterine scalp electrode using a STANS 21 monitor (Neoventa Medical, Gothenburg, Sweden). The fetal unipolar ECG lead configuration consisted of a single helix scalp electrode and a maternal skin electrode. The R-peaks were detected and R-R intervals were measured and digitised (sampling rate 500 Hz). The R-R interval data sets were stored on a PC hard disc. Calculation of a symbolic dynamics estimate of the heart rate variability We have calculated a symbolic dynamics estimate for the fetal heart rate variability. We used the procedure below to analyse the ECG time series (fig.1a): The control of the autonomic regulation of the heart rate is done by the sinus node. The sinus rhythm should be derived from the onsets of the P-waves. However, the P-wave signal cannot always be extracted and the intervals between the Rpeaks are chosen for further analysis. Step 1. Find the sequence of the R-R interval (length between two R-peaks) This is shown in fig.1b with the interval number along the x-axis and the interval duration in milliseconds along the yaxis. At this stage it is often possible to recognize the classic differences between healthy fetus and patient with a normal and a reduced variability. Step 2. Calculate the first derivative, i y (fig.1c) of the R-R interval   i i x 1  Storm et al____________________________________________________________________ BBB[3][2][2015] 190-198   i i n