Cahill, Alison G. MD, MSCI; Roehl, Kimberly A. MPH; Odibo, Anthony O. MD, MSCE; Macones, George A. MD, MSCE
The 2008 Consensus Conference on electronic fetal monitoring defined four types of decelerations: early, late, variable, and prolonged. However, in everyday practice, additional features associated with decelerations such as shoulders or overshoots or variability at the nadir of the deceleration are seen often. Unfortunately, the meaning of these features is uncertain, because there is scant prior research on the association between these atypical deceleration features and acid-base status at birth. The goal of this study was to estimate the incidence of atypical deceleration characteristics in term labor and to estimate their association with acidemia at birth.
MATERIALS AND METHODS
We performed a retrospective cohort study of all consecutive births between 2004 and 2008 at Washington University in St Louis Medical Center after approval from the Washington University School of Medicine Human Research Protection Office. Women were included if they carried a singleton gestation in vertex presentation, labored, reached complete dilation, had intrapartum electronic fetal heart monitoring, and an umbilical cord arterial pH. Women were excluded if they carried a fetus with known anomalies, did not reach complete dilation (10 cm), and did not have sufficient electronic fetal heart monitoring recording in the 30 minutes before delivery. Institutional policy is one of universal electronic fetal heart monitoring during labor and arterial umbilical cord gas pH at delivery.
The 30 minutes of electronic fetal heart monitoring before delivery was extracted by two formally trained obstetrics nurses certified in electronic fetal heart monitoring interpretation and blind to clinical data and outcomes. Women were excluded if they did not have at least 10 minutes of electronic fetal heart monitoring in the final 30 minutes before delivery. The nurses used the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) three-tiered category system and definitions for decelerations to extract the electronic fetal heart monitoring as well as extracting additional characteristics of each deceleration. Specific to this study, the presence of the following deceleration characteristics were extracted: overshoot (defined as a secondary heart rate acceleration, continuous with the deceleration, beyond return to the fetal baseline of 15 beats per minute or greater for at least 20 seconds),1 shoulder (defined as an increase in baseline of at least 15 beats per minute for at least 15 seconds, immediately preceding or after the deceleration),1 slow return (defined as any recovery from nadir to baseline that was 30 seconds or greater longer than duration of onset to nadir, Fig. 1), or nonmoderate variability within the deceleration (absent, minimal, marked, Fig. 2). Any deceleration with one or more of these characteristics was considered “atypical.” Each patient was given an “atypical score,” defined as the number of atypical decelerations present in the final 30 minutes before delivery. Because “slow return” decelerations were the most commonly occurring, we explored whether the number of “slow returns” alone was associated with acidemia. Lastly, given that the definition of “slow return” is somewhat arbitrary, we secondarily analyzed the incidence and association of “slow return” decelerations defined as nadir to recovery 60 seconds or greater longer than onset to nadir.1,2 The variability at the nadir of each deceleration also was recorded.1 Decelerations with absent, minimal, or marked variability were compared with those with moderate variability.
Detailed maternal and pregnancy data were also extracted, including medical and obstetric history, pregnancy course and complications, medication exposure, labor course, and delivery and neonatal outcomes. Use of internal monitors for fetal heart rate monitoring and contractions, maternal examination and vital signs, and use and type of anesthesia were also recorded. Umbilical cord gas arterial pH as well as CO2 and base excess were also recorded. The primary outcome was acidemia, defined as arterial umbilical cord gas pH 7.10 or less. The secondary outcome was metabolic acidemia, defined as arterial umbilical cord gas pH 7.10 or less and base excess less than −8. We defined a neonatal composite adverse outcome as at least one of the following: arterial umbilical cord gas pH less than 7.20, Apgar score at 5 minutes less than 7, or neonatal intensive care or special care unit admission. To estimate the incidence of atypical deceleration features in neonates born with no evidence of neonatal depression, we described the incidence and distributions of atypical deceleration features in neonates born without the composite adverse outcome (apparently normal).
The cohort was described and baseline characteristics were compared between women who delivered a neonate with acidemia and those who did not. Student’s t test and Mann-Whitney U tests were used for continuous variables and χ2 and Fisher’s exact tests were used for dichotomous variables as appropriate. Continuous variables were tested for normality visually and with the Shapiro-Francia test.3 Relative risk of acidemia and 95% confidence intervals were calculated for each of the electronic fetal heart monitoring deceleration types as specified by the NICHD as well as atypical features. Atypical scores and types of variability within the deceleration nadirs were also compared between women delivering neonates with acidemia and those without. Stratified analyses were performed to identify potentially confounding factors, which were considered in multivariable analyses. Multivariable logistic regression was performed in a backward, stepwise fashion to refine estimates of association between electronic fetal heart monitoring characteristics and acidemia by eliminating nonsignificant factors. The final model, adjusting for nulliparity, regional anesthesia, and obesity (body mass index [calculated as weight (kg)/[height (m)]2] 30 or greater) was tested with the Hosmer-Lemeshow goodness-of-fit test. Multivariable analyses were not performed for any characteristics with rare incidence, occurring fewer than five times in either group. A subanalysis was performed among only women who contributed all 30 minutes of electronic fetal heart monitoring. Lastly, descriptive analyses were performed among only those “apparently normal” neonates by excluding any with the neonatal composite adverse outcomes. All analyses were performed using STATA 10 special edition and SAS 9.2.
Of 5,388 women meeting inclusion criteria, 57 (1.1%) delivered a neonate with acidemia (pH 7.10 or less). Of those, 50 delivered neonates with metabolic acidemia (pH 7.10 or less and base excess −8.0 or less). There was no difference in the amount of electronic fetal heart monitoring available for interpretation between groups (P=.22): all 30 minutes, acidemia n=51 (89.5%) compared with no acidemia n=5,063 (95.0%); 20–30 minutes, acidemia n=4 (7.0%) compared with no acidemia n=208 (3.0%); and 10–20 minutes, acidemia n=2 (3.5%) compared with no acidemia n=59 (1.1%). Comparing those with acidemia (pH 7.10 or less) with those without, the groups were similar with respect to maternal age, gestational week at delivery, maternal race, incidence of diabetes and preeclampsia, and neonate birth weight on average (Table 1). Women delivering a neonate with acidemia were more likely to be nulliparous, have regional anesthesia, and deliver by operative vaginal delivery or cesarean delivery. Women delivering a neonate without acidemia tended to be less likely to have a maternal fever or to be attempting a vaginal birth after cesarean delivery, but neither reached statistical significance.
Of the atypical deceleration features analyzed, “shoulders” occurred most frequently (n=2,914 [54.1%]) followed by “slow returns” (n=2,618 [48.6%]), and nonmoderate (absent, minimal, marked) variability in the deceleration (n=2,014 [37.4%]), the majority of which was minimal deceleration variability (n=430 [8.0%]), and absent deceleration variability (n=4 [0.07%]). There was no difference in incidence between those with acidemia and those without (Table 2). “Overshoots” occurred rarely (n=1). Neither the presence of “shoulders” (adjusted odds ratio [OR] 1.06, 95% confidence interval [CI] 0.63–1.81) nor the presence of “slow returns” (adjusted OR 0.91, 95% CI 0.54–1.53) was associated with an increased risk of acidemia, even after adjusting for nulliparity, obesity (body mass index 30 or greater), and regional anesthesia. Similarly, compared with patients with moderate variability within deceleration nadirs, neither minimal (adjusted OR 0.82, 95% CI 0.43–1.55) nor marked (adjusted OR 0.65, 95% CI 0.27–1.55) variability was significantly associated with acidemia. Even when the presence of atypical deceleration characteristics were combined in aggregate as an atypical score, there was no association with acidemia (4.6 compared with 3.6 average atypical decelerations, P=.12); an atypical score of greater than five was not significantly associated with acidemia at birth (adjusted OR 1.24, 95% CI 0.70–2.20). In subanalysis including only women with 30 minutes of electronic fetal heart monitoring, results were similar.
When the incidence and association of atypical decelerations with metabolic acidemia was examined (Table 3), we found similar results. There was no association among “shoulders,” “overshoots,” “slow-returns,” nonmoderate variability in decelerations, or the overall atypical score and metabolic acidemia. Testing the definition of “slow returns,” we further compared the incidence of “slow return” decelerations defined as a difference in recovery time from the onset of 60 seconds or more and found no difference (Table 4). In fact, the women with the most “slow return” decelerations (90th percentile or greater, or greater than six) defined as recovery that took 30 seconds or longer than the onset to nadir all gave birth to neonates without acidemia.
Finally, we excluded all women who delivered neonates with any evidence of neonatal morbidity (composite neonatal morbidity). After excluding women who gave birth to a neonate with any of the following: arterial umbilical cord gas pH less than 7.20, 5-minute Apgar score less than 7, or neonatal intensive care unit or special care admission, the frequency and distribution of atypical decelerations were described (Table 5). Even in apparently normal term neonates, decelerations with atypical characteristics occurred commonly; 54% had one or more “shoulders” and 48% had one or more “slow returns.” Twenty-two percent women had more than five atypical decelerations in the 30 minutes before delivery but still delivered a neonate without evidence of neonatal morbidity.
We found no association between atypical deceleration characteristics in the final 30 minutes before delivery, alone or in combination, and risk of acidemia at birth. Specifically, we found that “shoulders,” “slow returns” with varied definition, and minimal or marked variability within the deceleration demonstrated no association with acidemia or metabolic acidemia at birth. Given the high frequency of these atypical deceleration characteristics in neonates born with a normal arterial umbilical cord gas pH, normal 5-minute Apgar score, and admission to the low-risk nursery, they are likely a normal variant, at least in the final 30 minutes before term birth.
The Practice Bulletin from the American College of Obstetricians and Gynecologists addressing electronic fetal heart monitoring nomenclature in 2009,4 after the 2008 workshop on electronic fetal heart monitoring guidelines sponsored in part by American College of Obstetricians and Gynecologists, the Society for Maternal-Fetal Medicine, and the NICHD5 specified atypical deceleration features including “slow returns,” “shoulders,” and “overshoots” as periodic or episodic decelerations within category II (also termed indeterminate category).4 However, no published evidence was available to guide this portion of the classification system with respect to neonatal acid base status or outcome.6 This is particularly important because the association data between category II tracings and outcomes is void,7 leaving health care practitioners with no guidance when these atypical decelerations are noted in practice.
In 1983, Krebs et al1 published their findings of a retrospective cohort study of electronic fetal heart monitoring tracings during labor after 34 weeks of gestation in 1,996 women between 1975 and 1977. Similar to the present study, they examined the presence of atypical decelerations in the 30 minutes before delivery. They found 988 women with decelerations before delivery and described their association to fetal pH from scalp sampling and 1- and 5-minute Apgar scores. The authors reported that atypical decelerations (having any one of the atypical features) were present in 19% of electronic fetal heart monitoring tracings but predicted 38% of 1- and 5-minute Apgar scores less than 7. Among other limitations, because women without fetal scalp samples were not included, it is not possible to calculate the test performance of these features and the reduced sample size further restricted the information that could be gleaned from these data. However, this was one of the largest studies to examine these atypical deceleration characteristics in human electronic fetal heart monitoring tracings and remains so through the present time.
More recently, Hamilton et al8 examined the last 4 hours on electronic fetal heart monitoring patterns before birth after 37 weeks of gestation in 3,695 women. The authors compared electronic fetal heart monitoring patterns of women with normal neonates with two groups: those with metabolic acidosis but phenotypically normal and those with acidosis and evidence of encephalopathy using computer recognition software. The authors stated that as a result of the rare occurrence of the third group, cases were collected from other sources including other hospital series and medicolegal files. They reported loss of (moderate) variability within the deceleration, but none of the other atypical features, to be discriminatory between neonates with metabolic acidosis (defined by base excess) and those who were normal. However, the area under the receiver operator curve was only slightly better than chance (0.56), which others might argue is nondiscriminatory. The definition of acidemia used by these authors and the processes to acquire their sample identified a greater number of cases of acidemia, offering greater power compared with the present study. However, despite this, they similarly found a lack of discriminatory ability of all features in common with our study excess variability within the deceleration. The significant possibility of selection bias as well as the inability to adjust for relevant confounding factors known to be associated with acidemia resulting from the lack of clinical data makes the interpretation of these findings limited. In our unselected birth cohort, we were able to adjust for relevant confounding factors to best estimate the association between atypical deceleration features and acidemia.
A unique strength of this cohort study was the interpretation of the fetal heart rate tracing deceleration characteristics by to dedicated obstetric research nurses formally trained in electronic fetal heart monitoring interpretation who were blind to all clinical factors and outcome data, including arterial umbilical cord gas pH. Although some may argue that human interpretation can be unreliable, we would offer that these nurses underwent formal inter- and intraobserver reliability testing and ongoing retraining and further that human interpretation is how electronic fetal heart monitoring is used in everyday practice and thus offers generalizability. We were also able to use a composite marker of neonatal morbidity to exclude any neonates with short-term markers of at-risk neonates to enable the description of the frequency with which atypical decelerations occur in apparently normal term neonates. Lastly, we assigned a score to objectively estimate any evidence of dose–effect in the setting of multiple atypical decelerations and acidemia.
In contrast, there are some potential limitations that are important to consider with respect to our study. Choosing the 30 minutes before delivery may limit generalizability of our findings, specifically in the interpretation of atypical deceleration characteristics at other times during labor. However, we chose this time period because it is most proximal to the measurement of acidemia in modern practice (at birth). Although we chose atypical deceleration features that are most commonly referenced in historical texts and have been previously described, our study was certainly not exhaustive and there are many other possible features of electronic fetal heart monitoring patterns, and decelerations specifically, that might be associated with acidemia but have yet to be tested. Lastly, despite our large sample size, acidemia at term occurs rarely, and thus the lack of association found in our study could be the result of type II error. Although we had 80% power to detect a 2.1-fold increased risk of acidemia in the setting of atypical decelerations, which we would offer is a reasonable threshold for clinical significance, it is possible that associations exist of lesser magnitude that could not be detected among this term-born sample.
Despite these potential limitations, we feel that our findings contribute to the existing literature associating electronic fetal heart monitoring patterns with acidemia and birth outcomes. We found no evidence that “shoulders,” “slow returns,” or nonmoderate deceleration variability, alone or in combination, are associated with acidemia. It remains to be studied whether the persistence of these deceleration features for longer than 30 minutes before delivery might have an association with acidemia. However, given the high incidence of atypical features in apparently normal term-born neonates, it appears more likely that these are normal variants within the definitions used in this study. These data support the absence of these specific atypical deceleration characteristics from the recognized definitions of decelerations stipulated by the NICHD in 20085 given their lack of association with acidemia or neonatal depression.