Fetal heart rate patterns in labor and the risk of neonatal encephalopathy: A case control study.

OBJECTIVE: To describe the accuracy of intrapartum fetal heart rate abnormalities as defined by National Institute of Health and Care Excellence guidelines to predict moderate-severe neonatal encephalopathy of apparent hypoxic-ischemic etiology. STUDY DESIGN: A case-control study of HIE risk factors was conducted. Eligible babies were born in a single maternity hospital in Dublin, Ireland between September 2006, and November 2017 at ≥35 + 0 weeks' gestational age. Cases were eligible babies with moderate-severe neonatal encephalopathy of definite or apparent hypoxic-ischemic etiology. Controls were eligible babies born before and after each case with normal Apgar scores. The included subjects who had intrapartum fetal heart rate recordings were identified. Pattern features (baseline rate, variability, accelerations, decelerations [early, late, variable, prolonged], bradycardia, sinusoidal pattern) were manually identified blind to all clinical details by one of the authors. Each 15-minute segment was then algorithmically categorized (uninterpretable, normal, suspicious, pathological). RESULTS: Of 88 cases and 176 controls, 71 cases (81%) and 146 controls (83%) were admitted to the delivery suite in labor. From that group, intrapartum FHR traces longer than 15 min were available for 52 (73%) cases and 118 (83%) controls. The FHR pattern feature with the largest area under the receiver operating characteristic curve was the maximum number of consecutive segments in which the baseline was >160 bpm (0.71 [95% confidence interval: 0.62-0.80]). The category variable with the highest area under the curve was the number of suspicious segments (0.76 [95% confidence interval: 0.67-0.84]). A tri-variate logistic regression model incorporating the total number of segments, the number of "suspicious" segments classed, and the number of "pathological" segments achieved an area under the curve of 0.78 (95% confidence interval: 0.70-0.86). With 95% specificity, this model correctly identified 17 cases (33%) at a median time before delivery of 2 h and 18 min (interquartile range: 01:19-04:40). CONCLUSIONS: The power of fetal heart rate analysis to predict neonatal encephalopathy is hampered by poor specificity given the rarity of the outcome. When analyzing a suspicious trace, it is beneficial to consider the overall duration of the suspicious pattern.

Uterine activity in labour and the risk of neonatal encephalopathy: a case control study.

OBJECTIVE: To determine the relationship between intrapartum contraction frequency, rest interval duration, and cervical dilation speed and the risk of neonatal hypoxic-ischemic encephalopathy (HIE). STUDY DESIGN: This was a retrospective case-control study conducted in a maternity hospital in Dublin, Ireland. Babies born without major congenital anomalies between September 2006 and November 2017 at ≥ 35 + 0 weeks' gestational age were eligible. Cases were diagnosed with moderate-severe HIE. The controls were the first eligible baby born before and after each case with normal Apgar scores and not admitted to the neonatal unit. Intrapartum uterine activity was assessed by automated analysis of external tocography recordings. Cervical dilation was assessed by linear interpolation between vaginal examination measurements. The speed of cervical dilation was expressed as the times from 4 to 6 cm, >6 cm to the start of pushing, and from pushing to delivery. RESULTS: Intrapartum tocographs results were available in 49 of 88 cases and 121 of 176 controls. The median contraction rate in cases was 7.7 (Interquartile range [IQR]: 6.6-9.0) compared to 7.0 in controls (IQR: 6.2-7.9) (p = 0.021). The median rest interval duration was 56 s (IQR: 38-76) in cases and 62 s (IQR: 50-79) in controls (p = 0.058). Cases took longer to progress from > 6 cm to the start of pushing (cases: 02:58 [01:14-04:49], controls: 01:48 [00:51-03:34], p = 0.020) and from pushing to delivery (cases: 00:34 [00:24-01:10], controls: 00:27 [00:13-00:56], p = 0.036). CONCLUSIONS: Higher contraction frequencies and slower progress towards the end of labour are both independently associated with the risk of moderate-severe HIE. Inter-contraction rest interval duration as measured by external tocography does not provide additional accuracy.

A wavelet-based algorithm for automated analysis of external tocography: How does it compare to human interpretation?

BACKGROUND: Studies which use external tocography to explore the relationship between increased intrapartum uterine activity and foetal outcomes are feasible because the technology is safe and ubiquitous. However, periods of poor signal quality are common. We developed an algorithm which aims to calculate tocograph summary variables based on well-recorded contractions only, ignoring artefact and excluding sections deemed uninterpretable. The aim of this study was to test that algorithm's reliability. METHODS: Whole recordings from labours at ≥35 weeks of gestation were randomly selected without regard to quality. Contractions and rest intervals were measured by two humans independently, and by the algorithm using two sets of models; one based on a series of pre-defined thresholds, and another trained to imitate one of the human interpreters. The absolute agreement intraclass correlation coefficient (ICC) was calculated using a two-way random effects model. RESULTS: The training dataset included data from 106 tocographs. Of the tested algorithms, AdaBoost showed the highest initial cross-validated accuracy and proceeded to optimization. Forty tocographs were included in the validation set. The ICCs for the per tocograph mean contraction rates were; human B to human A: 0.940 (0.890-0.968), human A to initial models: 0.944 (0.898-0.970), human A to trained models 0.962 (0.927-0.980), human B to initial models: 0.930 (0.872-0.962), human B to trained models: 0.948 (0.903-0.972). CONCLUSIONS: The algorithm described approximates interpretation of external tocography performed by trained humans. The performance of the AdaBoost trained models was marginally superior compared to the initial models.