DEV (WILEJ) LEFT INTERACTIVE

Shawn F. Reed University of Maryland College Park, MD 20742 A Neural Explanation of Fetal Gonen Ohel Rahav David Patterns: A Test of Poriya Hospital Tiberias, Israel the Polyvagal Theory Stephen W. Porges University of Maryland College Park, MD 20742

Received 11 March 1998; accepted 5 February 1999

ABSTRACT: The current study applies a neurophysiological model based on the Polyvagal Theory (Porges, 1995) to interpret fetal heart rate patterns. Beat-to-beat heart rate data from 7 fetuses monitored during the ®rst and second stages of labor were analyzed. Transitory heart rate accelerations and reduced beat-to-beat variability reliably preceded heart rate decelera- tions. The data are interpreted within the context of the Polyvagal Theory, which provides a plausible explanation of the neurophysiological mechanisms that mediate fetal heart rate de- celerations. Speci®cally, it is proposed that both the transitory heart rate accelerations and the depression of the respiratory rhythm in the beat-to-beat heart rate pattern re¯ect a withdrawal of the determined by myelinated vagal pathways originating in the nucleus ambig- uus. Functionally, withdrawal of vagal tone originating in the results in the cardiac pacemaker becoming vulnerable to sympathetic in¯uences and to the more-primitive unmyelinated vagal pathways originating in the dorsal motor nucleus of the vagus, which may contribute to clinically relevant . ᭧ 1999 John Wiley & Sons, Inc. Dev Psychobiol 35: 108±118, 1999

Keywords: electronic fetal heart rate monitoring; ; respiratory sinus ar- rhythmia; vagus

Based upon clinical observations and limited experi- mechanisms mediating the speci®c features of the mental manipulations, speci®c features of the fetal heart rate patterns used in the clinical assessment of heart rate pattern have been accepted by the clinical the fetus. community as having diagnostic utility. Despite the The primary goal of fetal heart rate monitoring is the widespread acceptance of fetal heart rate monitoring, detection of fetal hypoxia at its earliest stage (Kruse, especially in high-risk pregnancies, there remains a 1982). Speci®c characteristics of the heart rate pattern limited understanding of the neurophysiological (i.e., early decelerations, variable decelerations, late de- celerations, beat-to-beat variability, , brady- Correspondence to: Stephen W. Porges, Department of Human cardia, and sinusoidal patterns) have been identi®ed and Development/Institute for Child Study, 3304 Benjamin Bldg., Uni- proposed as re¯ections of the neurophysiological reg- versity of Maryland, College Park, MD 20742±1131. ulation of the heart and thus, as sensitive indices of fetal Contract grant sponsor: National Institute of Child Health and Human Development condition. However, because the features of the fetal Contract grant number: HD 22628 heart rate pattern are often unreliable diagnostic indi- short ᭧ 1999 John Wiley & Sons, Inc. CCC 1012-1630/99/020108-11 standard long DEV (WILEJ) RIGHT INTERACTIVE

Fetal Heart Rate Patterns 109 cators, controversy still exists over the use and inter- 1990). This central respiratory rhythm is an emergent pretation of fetal heart rate monitoring. property of the brainstem with signi®cant contribu- Dif®culties in the diagnostic application of fetal tions from the interneuronal communication between heart rate monitoring may be due, in part, to uncertain the nucleus ambiguus and the nucleus of the solitary and contradictory neurophysiological explanations tract. The functional impact of the nucleus ambiguus (Goodlin & Haesslein, 1977; Rosen & Dickinson, vagal pathways on the is to produce a 1993). For example, beat-to-beat variability is as- respiratory rhythm in the heart rate pattern. The am- sumed to be mediated by the vagus with high levels plitude of these rhythmic increases and decreases in of variability indicating fetal well-being (Kruse, 1982; heart rate (i.e., respiratory sinus arrhythmia, RSA) is, Paul, Suidan, Yeh, Schifrin, & Hon, 1975). However, thus, a valid indicator of vagal out¯ow from the nu- heart rate decelerations, indicators of fetal compro- cleus ambiguus. Even in the absence of actual mise, are assumed also to be mediated by the vagus. breathing, there are reports of beat-to-beat changes in The current article attempts to reconcile this para- fetal heart rate that track a rhythm similar to the spon- dox by interpreting fetal heart rate patterns within the taneous breathing of a neonate (Donchin, Caton, & context of the Polyvagal Theory (Porges, 1995, 1997). Porges, 1984; Groome, Mooney, Bentz, & Wilson, The Polyvagal Theory focuses on the phylogenetic 1994). changes in the functional neuroanatomy of the vagus and how these modi®cations are represented in the neural regulation of heart rate patterns. BRADYCARDIA

Across vertebrate species, the control of cardiac func- THE NEUROPHYSIOLOGY OF tion exhibits a phylogenetic trend towards greater neu- MAMMALIAN HEART RATE PATTERNS ral involvement. The Polyvagal Theory (Porges, 1995) focuses on the speci®c phylogenetic changes in the The vagus, the 10th cranial nerve, conveys neural in- vagal efferents and the medullary nuclei from which ¯uences to the cardiac pacemaker (i.e., sinoatrial these pathways originate. Vagal efferent pathways node). The release of acetylcholine on the sinoatrial originate in two medullary nuclei, the nucleus ambig- node of the heart, triggered by vagal activity, not only uus and the dorsal motor nucleus of the vagus. Ac- inhibits cardiac pacemaker activity but also attenuates cording to the theory, because these two vagal systems sympathetic in¯uences on the heart (Levy, 1984; Van- have different phylogenetic and embryologic origins, houtte & Levy, 1979). Therefore, heart rate slows with they have different and potentially contradictory re- increases in vagal tone and heart rate speeds with de- sponse strategies when challenged. The unmyelinated creases in vagal tone. Vagal output to the heart origi- vagal ®bers that originate in the dorsal motor nucleus nates in two brainstem nuclei, the dorsal motor nucleus of the vagus evolved before the myelinated vagal ®- of the vagus and the nucleus ambiguus. Vagal pathways bers of the nucleus ambiguus. The dorsal motor vagal originating in both nuclei are capable of producing heart system evolved as a passive avoidance system and re- rate decelerations (Jones, Wang, & Jordan, 1995). The sponds to environmental challenges by sending inhib- vagal ®bers originating in the dorsal motor nucleus of itory impulses to the heart, which reduces cardiac out- the vagus are unmyelinated. During most states, the un- put in an effort to conserve metabolic resources. In myelinated vagal ®bers have little in¯uence on heart reptiles, the dorsal motor vagal system is the predom- rate levels (Ford, Bennett, Kidd, & McWilliam, 1990; inant vagal system in the regulation of cardiopulmo- Jones et al., 1995) and exhibit little or no spontaneous nary processes. This system promotes apnea and activity or variability (Ford et al., 1990). In contrast, the bradycardia by constricting the bronchi and by inhib- vagal ®bers originating in the nucleus ambiguus are my- iting the cardiac pacemaker. These physiological re- elinated, in¯uence both heart rate level and beat-to-beat sponses are neurophysiologically consistent with the variability (McAllen & Spyer, 1978), and maintain commonly observed reptilian avoidance strategies of tonic in¯uence over the heart. submerging or immobilizing. In mammals, the nucleus ambiguus vagal system is the predominant vagal sys- tem in the regulation of cardiopulmonary processes. RESPIRATORY SINUS ARRHYTHMIA IN This phylogenetically more-recent system has myeli- THE FETUS nated vagal ®bers that enable a rapid and dynamic reg- ulation of cardiac output via transitory increases and The vagal efferent ®bers originating in the nucleus am- decreases in tonic vagal output to the heart. Although short biguus have a respiratory rhythm (Richter & Spyer, both nuclei are capable of slowing heart rate in mam- standard long DEV (WILEJ) LEFT INTERACTIVE

110 Reed et al. mals, stimulation of vagal ®bers originating in the dor- may produce clinically signi®cant bradycardia. Two sal motor nucleus of the vagus has been proven inef- neurophysiological factors contribute to this possibil- fective in producing large bradycardia (Jones et al., ity. First, nucleus ambiguus vagal tone may protect the 1995). pacemaker from the in¯uence of ®bers originating in Research examining the relative in¯uence of dorsal the dorsal motor nucleus. Thus, removal of nucleus motor nucleus vagal ®bers in promoting the clinically ambiguus vagal tone would create a vulnerability of relevant bradycardia observed in fetal distress is not the pacemaker to the cholinergic in¯uences of the va- conclusive at this time (Hopkins, Bieger, de Vente, & gal ®bers originating in the dorsal motor nucleus. Sec- Steinbusch, 1996). However, it is possible that hy- ond, because hypoxia potentiates cholinergic action on poxia itself potentiates the cardioinhibitory actions of the pacemaker, stimulation of only a few vagal ®bers vagal ®bers on the sinoatrial node. Potter and Mc- during hypoxic states might be suf®cient to produce Closkey (1986) report a feedback system between du- massive bradycardia. Potter and McCloskey (1986) ration of hypoxia and vagal efferent discharge, which demonstrated that during progressive asphyxic hy- results in a potentiation of the vagal output on the poxia in dogs, not only was cardiac vagal activity elic- heart. Although there is a massive decline in vagal ited, but also the sensitivity of the sinoatrial node to ®ring during hypoxia, this system functions to main- vagal efferent in¯uences was potentiated. Thus, during tain bradycardia by potentiating the in¯uence of vagal hypoxic conditions, the withdrawal of nucleus ambig- ®ring on the sinoatrial node. Under these conditions, uus in¯uence on the heart (i.e., indexed by low-am- the magnitude of the bradycardia is determined, in plitude RSA) and the increased sensitivity of the si- part, by peripheral mechanisms. Therefore, it may be noatrial node to vagal efferent in¯uences may create possible that bradycardia could be mediated by a rel- a situation in which vagal output from the dorsal motor atively small number of vagal ®bers originating in the nucleus of the vagus might be suf®cient to produce dorsal motor nucleus of the vagus potentiated in the clinically relevant bradycardia. periphery by hypoxia.

APPLICATION OF THE POLYVAGAL TACHYCARDIA THEORY

In humans and other mammals, under most conditions, The current article applies the Polyvagal Theory as a the primary vagal control of the heart is mediated via model to interpret fetal heart rate patterns. Based on nucleus ambiguus pathways. During these states, the the theory, several predictions are made. First, the ini- nucleus ambiguus vagal tone potentially may protect tial transitory response to withdrawal of nucleus am- the mammalian heart from both the inhibitory surges biguus vagal tone would be a transitory heart rate ac- from the dorsal motor nucleus of the vagus and the celeration. Second, withdrawal of nucleus ambiguus excitatory surges from the sympathetic nervous sys- vagal tone would leave the pacemaker vulnerable to tem. However, when metabolic demands are high, the activity from vagal pathways originating in the dorsal adaptive response of the mammal is to increase heart motor nucleus of the vagus. If the above were ob- rate by reducing the nucleus ambiguus vagal tone to served, heart rate decelerations would be anticipated the pacemaker. Because, at the level of the sinoatrial by a transitory heart rate acceleration in a background node, the cholinergic vagal system is inhibitory on the of low-amplitude RSA. Third, recovery from the heart pacemaker and the sympathetic adrenergic system rate decelerations would be characterized by a return (e.g., Levy, 1984; Vanhoutte & Levy, 1979), the re- to, and in many cases, an increase in RSA re¯ecting moval of cholinergic in¯uences (i.e., the withdrawal the dynamic adjustment of nucleus ambiguus vagal of nucleus ambiguus vagal tone) results in a potentia- tone. Fourth, chronic fetal distress would be charac- tion of sympathetic activity. Thus, the withdrawal of terized by low beat-to-beat heart rate variability and nucleus ambiguus vagal in¯uence results in an instan- the massive and frequent heart rate decelerations that taneous tachycardia. characterize the clinical bradycardia indicative of fetal distress. Hypothetically, this response pro®le might be due to low nucleus ambiguus vagal tone and a pace- FETAL DISTRESS maker that is vulnerable to surges in the dorsal motor nucleus of the vagus. According to the Polyvagal Theory, prolonged with- The current study was conducted to test the Poly- drawal of nucleus ambiguus vagal tone creates a phys- vagal Theory by statistically examining heart rate pat- short iological vulnerability to other vagal mechanisms that terns in archived fetal heart rate data. Based on the standard long DEV (WILEJ) RIGHT INTERACTIVE

Fetal Heart Rate Patterns 111 above four predictions, three hypotheses were tested: erationalized as the point in time when the heart period (a) Heart rate decelerations will be preceded by tran- stopped decreasing after reaching baseline levels. sitory accelerations re¯ecting the withdrawal of nu- Heart period levels immediately before the decel- cleus ambiguus vagal tone, (b) recovery from decel- erations were evaluated to determine if a signi®cant eration will be characterized by an increase in RSA, heart rate acceleration (i.e., decrease in heart period) and (c) chronic low-amplitude RSA will be associated occurred. Signi®cant accelerations were statistically with more-frequent and severe heart rate decelerations. de®ned according to the following steps. First, for each deceleration, the range of heart period values was calculated during the preceding 60 s. Second, the MATERIAL AND METHODS lower 5th percentile of the range was calculated. Third, 95% con®dence intervals were constructed around the Heart period data (i.e., R±R intervals) were collected heart period value representing the lower 5th percen- at the Poriya Government Hospital in Israel. To insure tile. The con®dence intervals were estimated using a accurate detection of beat-to-beat measures of heart bootstrapping technique similar to the method de- period, data were collected during the ®rst and second scribed by Lunneborg (1986). This bootstrapping tech- stages of labor using scalp electrodes. Electrodes were nique produces con®dence intervals that are not nec- attached to the fetal scalp through the opening of the essarily symmetrical about the 5th percentile. Fourth, cervix during the initial stages of labor. Data from 11 the heart period value immediately prior to the onset fetuses were obtained. The number of subjects was of deceleration was contrasted with the lower limit of governed by the clinician's decision to monitor the the con®dence intervals. If this heart period was out- fetus with scalp electrodes. Ten cases were normal in side the con®dence interval, it was de®ned as a sig- terms of pregnancy, labor, clinical fetal heart rate pat- ni®cant acceleration (See Figure 1). terns during labor, APGAR scores, umbilical cord To evaluate the hypothesized recovery of RSA fol- blood pH, and maternal oxygen saturation. In addition, lowing decelerations, a within subjects analysis of RSA 1 case of intrauterine fetal death was analyzed. The contrasted the amplitude of RSA measured during the intrauterine fetal death case was diagnosed prior to 60 s prior to and following the decelerations. Due to delivery and had pronounced fetal hydrocephalus and the temporal proximity of several decelerations, 9 de- an abnormal heart rate pattern. celerations did not have the required 60 s of heart period The ECG signal was digitized on-line in the deliv- data following the deceleration; thus, the analysis of ery room using a Vagal Tone Monitor (Delta-Biomet- RSA was conducted on 19 of the 28 decelerations. rics, Bethesda, MD). The Vagal Tone Monitor digit- RSA was quanti®ed using the Porges (1985) izes the ECG analog output of the fetal heart rate method, the steps of which are illustrated in Figure 2. monitor at 1 kHz and identi®es the peak of the R-wave Approximately 200 s of fetal heart period time sam- and the R±R intervals with 1-ms accuracy. This de- pled every 200 ms during a stable state are plotted in gree of accuracy, which is uncommon in standard hos- Figure 2a. To identify and quantify RSA, the data in pital monitors, is necessary for the accurate quanti®- Figure 2a were ®rst smoothed with a 3rd order 21- cation of the low-amplitude RSA observed in the fetus point moving polynomial to create a template time se- (Donchin et al., 1984; Groome et al., 1994). ries (Figure 2b) and the template time series is then Of the 10 normal cases, 6 were used in the ®nal subtracted from the data in Figure 2a to generate a analysis. Two cases contained excessive ECG artifact residual time series that contains only the heart period and could not be analyzed, and of the 8 remaining cases, variability in a frequency band above .3 Hz (Figure only 6 contained at least one heart rate deceleration. 2c). Note the high-frequency oscillations with ampli- Decelerations were operationally de®ned as a transitory tudes between Ϯ5 ms. These oscillations represent fe- decrease in heart rate slower than 85 beats per min (i.e., tal RSA. To quantify the amplitude of these oscilla- a heart period longer than 700 ms). From these six re- tions, the template time series is bandpassed to output cordings, 28 heart rate decelerations were identi®ed. a time series consisting only of variances associated A computer graphics display of the beat-to-beat with the frequency band from .3 to 1.3 Hz, the spon- heart periods enabled a visual identi®cation of the be- taneous breathing frequencies of a human newborn. ginning and end of each heart rate deceleration. The The variance of the bandpassed time series is used as beginning of the deceleration was operationalized as a measure of the amplitude of RSA. To stabilize the the point at which the heart period recording began to statistical estimates of the amplitude of RSA, the nat- increase without a subsequent return to predecelera- ural logarithms of these variance measures were used tion (i.e., baseline) levels until after the 700-ms crite- in the subsequent analyses. short rion was reached. The end of the deceleration was op- In cases characterized by frequent decelerations, standard long DEV (WILEJ) LEFT INTERACTIVE

112 Reed et al.

FIGURE 1 Example of a signi®cant acceleration prior to the onset of a deceleration. such as during fetal distress, it may be impossible to spectral representation of the higher-frequency oscil- accurately quantify the amplitude of RSA. When the lation. Thus, a high-frequency peak in the spectrum fetal heart rate pattern is characterized by frequent de- might represent the sum of both the amplitude of the celerations, the slopes of the decelerations are similar true physiological oscillation (e.g., RSA) and a har- to the slope of RSA, thus creating a situation in which monic component from a slow oscillation. The moving it is statistically impossible to accurately quantify and polynomial approach, by removing the variance as- interpret RSA. To deal with the limitations in method, sociated with low-frequency activity, optimizes the the analyses required periods in which frequent de- quanti®cation of the heart rate oscillation at the res- celerations did not occur. piratory frequency. Two statistical factors contribute To demonstrate the effectiveness of the methodol- to this enhanced ability to detect the frequency and to ogy relative to other applications of spectral analyses quantify the amplitude of RSA. First, the variance as- (e.g., Davidson, Rankin, Martin, & Reid, 1992), a rel- sociated with the harmonics of slow-frequency activity atively stable segment of fetal heart period was ana- is removed, thus allowing the quanti®cation of the var- lyzed. In Figure 3a, the time-sampled fetal heart period iance at the respiratory rhythm to be statistically in- is plotted. Note in the ®gure that there are several ob- dependent of the harmonic in¯uences from high-am- servable oscillations. In Figure 4a, the power spectrum plitude, slow-frequency components. Second, if the is illustrated for the data in Figure 3a. The spectrum low- frequency components are removed prior to spec- re¯ects a dominant periodicity at approximately 0.04 tral analysis, statistical windowing procedures in the Hz with no clearly identi®able periodicity at any frequency domain do not include the low-frequency higher frequency within the normal frequencies of activity, which if included would distort the spectral spontaneous breathing. Because physiological perio- representation by increasing the amplitude and shifting dicities are not perfect sine waves, all slow heart rate the peak to slower frequencies. oscillations have higher-frequency components that Our approach starts by attempting to describe the appear in the Fourier transform at integer harmonics amplitude of oscillations in a frequency band that of the peak frequency. When the higher-frequency would re¯ect the spontaneous breathing rhythm in a peak is of low amplitude relative to the amplitude of human newborn infant. In human newborns, this fre- the very-slow oscillation, there is the possibility that quency would be in the range of 20 to 80 breaths per short the slow periodicity is contributing variance to the min or approximately 0.3 Hz to 1.3 Hz. We approach standard long DEV (WILEJ) RIGHT INTERACTIVE

Fetal Heart Rate Patterns 113

FIGURE 3 (a) Time-sampled fetal heart period data; (b) FIGURE 2 Sequential steps in calculating RSA: (a) time- data illustrated in Figure 3a smoothed with moving poly- sampled fetal heart period (i.e., 200-ms intervals); (b) data nomial procedure. illustrated in Figure 2a smoothed with a 3rd order 21-point moving polynomial; (c) residual time series generated by subtracting data in Figure 2b from data in Figure 2a.

eration being outside the 5th percentile was greater than the statistical expectation (i.e., 5%). All subjects exhibited accelerations outside the range for at least this problem by applying the moving polynomial pro- one deceleration and in 4 of the 6 fetuses, virtually cedure (Porges & Bohrer, 1990) to statistically remove every deceleration (10/11) was preceded by a transi- virtually all of the variance below .3 Hz from the fetal tory acceleration. heart period time series. In Figure 3b, the fetal heart rate data illustrated in Figure 3a are processed by the moving polynomial procedure described above. Fig- An Increase in RSA Is a Component of the ure 4b illustrates the power spectrum of the residual Recovery From a Deceleration time series illustrated in Figure 3b. The power spec- Postdeceleration levels of RSA (M ϭ 2.49, SD ϭ .45) trum has a prominent peak at approximately .40 Hz, a were consistently higher than predeceleration levels frequency similar to the spontaneous breathing fre- (M ϭ 1.34, SD ϭ 1.05). All 6 subjects displayed an quency of healthy newborn infants. average increase in RSA. These differences are statis- tically signi®cant, binomial test, p Ͻ .05. Figure 5 il- lustrates this characteristic pattern. Note the obvious RESULTS increase in beat-to-beat variability and the quanti®ed index of the amplitude of RSA. Convergent with the Transitory Acceleration Precedes increase in RSA, the stabilized postdeceleration base- Decelerations line was characterized by longer heart periods (i.e., Table 1 contains the number and percent of decelera- slower heart rate). All 6 subjects displayed an average tions which have been preceded by signi®cant accel- increase in heart period from predeceleration levels erations for each of the 6 subjects. The data illustrate (M ϭ 432 ms, SD ϭ 25) to postdeceleration levels that for all subjects the probability of the heart period (M ϭ 452 ms, SD ϭ 26). These differences are statis- short value immediately preceding the onset of the decel- tically signi®cant, binomial test, p Ͻ .05. Although the standard long DEV (WILEJ) LEFT INTERACTIVE

114 Reed et al.

number of decelerations provided by each of the fe- tuses varied, of the 19 decelerations observed across the 6 fetuses, 18 (95%) of the decelerations were as- sociated with increased RSA and 12 (63%) episodes with increased heart period (or heart rate slowing) dur- ing the postdeceleration recovery phase.

Fetuses With Chronic Low-Amplitude RSA Have More-Frequent and Severe Bradycardia The heart period pattern observed in the intrauterine fetal death case provides the only illustration in our restricted database of frequent and clinically relevant bradycardia occurring in a heart rate pattern devoid of RSA. As illustrated in Figure 6a, the heart period pat- tern is characterized by frequent bradycardia. In Figure 6b, one bradycardic event denoted by the two dotted lines in 6a is plotted. Note the ¯at beat-to-beat pattern that precedes and follows the bradycardia. Figure 6c further ampli®es the pattern that precedes the brady- cardia. Additionally, transitory tachycardia did not precede any of the bradycardia, suggesting an inability to further withdraw nucleus ambiguus vagal tone, which was already chronically depressed. However, as with the other fetuses, no signi®cant correlation was found between RSA and the magnitude of the brady- cardia or tachycardia.

DISCUSSION

FIGURE 4 (a) Power spectrum of fetal heart period data The Polyvagal Theory provides a useful model for ex- plotted in Figure 3a; (b) Power spectrum of fetal heart period plaining fetal heart rate patterns and for predicting the data plotted in Figure 3b. likelihood of decelerations during normal delivery. According to this theory, fetal heart rate decelerations may be mediated via a primitive unmyelinated vagal system (i.e., vagal pathways originating in the dorsal motor nucleus of the vagus) in the absence of in¯u- ences from a more phylogenetically advanced vagal

Table 1. Number and Percentage of Decelerations Showing Transitory Accelerations Total Number of Number with Signi®cant Percent with Signi®cant Subject Decelerations Accelerations Accelerations 4478 6 1 17 5972 3 3 100 6851 11 6 55 6919 3 2 67 7374 4 4 100 8372 1 1 100 short Totals 28 17 standard long DEV (WILEJ) RIGHT INTERACTIVE

Fetal Heart Rate Patterns 115

FIGURE 5 (a) Heart period pattern demonstrating the in- FIGURE 6 Segment of heart period data, shown at in- crease in RSA following a deceleration; (b) The 30-s seg- creasing magni®cation from (a) to (c), collected from a hy- ment prior to the deceleration shows low-amplitude RSA; drocephalic fetus demonstrating the frequent and severe (c) The 30-s segment following a deceleration shows in- bradycardic activity that can occur in the absence of RSA. creased-amplitude RSA. (Goodlin & Haesslein, 1977). In addition to the heart, system (i.e., vagal pathways originating in the nucleus efferent ®bers of the dorsal motor nucleus of the vagus ambiguus). Tonic depression of nucleus ambiguus va- innervate the muscles of the sphincter. Thus, it is theo- gal tone may be due to either physiological compro- retically consistent that if hypoxia stimulated vagally mise or developmental immaturity. The current study mediated relaxation of the sphincter muscles and the provides preliminary support for three hypotheses de- release of meconium, this same potent efferent system rived from the Polyvagal Theory. First, the onset of a might produce bradycardia. heart rate deceleration is reliably preceded by a tran- According to the Polyvagal Theory, transitory ac- sitory tachycardia (i.e., decrease in heart period) con- celerations in fetal heart rate suggest that the nucleus sistent with the predicted transitory effect on the pace- ambiguus vagal pathways are being used as a ªvagal maker due to the withdrawal of nucleus ambiguus brakeº to regulate heart rate. A transitory removal of vagal tone. Second, recovery from a deceleration is the vagal brake would produce a short latency accel- regularly characterized by an increase in RSA ampli- eration and the reinstatement of the vagal brake would tude demonstrating a reestablishment of nucleus am- produce an immediate slowing of heart rate to a more- biguus vagal tone. And ®nally, the absence of RSA normal or optimal level. Thus, transitory accelerations has been shown to be associated with frequent and would provide an indication that the nucleus ambiguus massive bradycardia in at least 1 fetal distress subject, vagal system was controlling the fetal heart rate. From consistent with the proposed vulnerability of the si- this perspective, the Polyvagal Theory accounts well noatrial node to vagal in¯uence from the dorsal motor for research demonstrating fetal heart rate accelera- nucleus of the vagus during physiological states char- tions to be indicators of fetal well-being (DiPietro, acterized by low nucleus ambiguus tone. Hodgson, Costigan, Hilton, & Johnson, 1996; Emory Additional evidence of the role of the dorsal motor & Noonan, 1984; Krebs, Petres, Dunn, & Smith, nucleus of the vagus in bradycardic activity is pro- 1982). It should also be noted that the Polyvagal The- vided by research demonstrating an association be- ory offers an alternative explanation to that given by short tween bradycardic activity and meconium staining James et al. (1976) for the transient tachycardia ob- standard long DEV (WILEJ) LEFT INTERACTIVE

116 Reed et al. served prior to bradycardia during hypoxic conditions. nation, fetal bradycardia would be due to the potentia- While James and colleagues attribute the tachycardia tion of unmyelinated vagal pathways originating in the to sympathetic activation, the Polyvagal Theory pres- dorsal motor nucleus of vagus by hypoxia, a state in ents a plausible alternative explanation in which these which vagal tone from the myelinated vagal pathways patterns could be the result of transient withdrawal of originating in the nucleus ambiguus are depressed. parasympathetic tone originating in the nucleus am- The second explanation assumes that the ªnewº biguus and functional disinhibition of the pacemaker. myelinated vagal system originating in the nucleus The current study offers only one example to sup- ambiguus might have a phylogenetically ordered re- port the hypothesis that reduced RSA is associated sponse hierarchy. According to the second explana- with increased likelihood of bradycardic activity. Al- tion, under normal conditions, the vagal ®bers from though the fetus had neurological anomalies, the heart the nucleus ambiguus convey a respiratory rhythm to period pattern was consistent with the previously re- the heart and produce RSA. However, once blood gas ported relationship between increased risk of fetal status is compromised, the respiratory rhythm would bradycardia and low beat-to-beat variability (Dawes et be depressed (i.e., similar to neurophysiological pro- al., 1993). Therefore, the protective role of the nucleus cesses that would produce apnea postpartum) and the ambiguus proposed by the Polyvagal Theory, and the more-tonic in¯uences to the bronchi and heart would high level of RSA associated with nucleus ambiguus be expressed through the myelinated vagal ®bers orig- tone, is also consistent with the view of heart rate vari- inating in the nucleus ambiguus. Thus, the bradycardia ability as a positive indicator of fetal well-being in the term fetus might be dependent upon the mye- (Kruse, 1982). linated vagal ®bers to the heart in mammals, which Because it has been questioned whether the un- during conditions of compromise would function like myelinated vagal ®bers originating in the dorsal motor the reptilian unmyelinated vagal ®bers that originate nucleus of the vagus are capable of producing the mas- in the dorsal motor nucleus of the vagus. This latter sive bradycardia observed in the fetus (e.g., Hopkins model might explain both the lack of evidence dem- et al., 1996; Jones et al., 1995), an alternative expla- onstrating potent bradycardia in mammals via the un- nation might be considered. For example, it is possible myelinated pathways and the occurrence of bradycar- that the myelinated vagal ®bers originating in the dia during states characterized by depressed RSA. nucleus ambiguus might contribute to clinical brady- Consistent with the points addressed by Paul et al. cardia. This explanation would suggest that the mye- (1975), there is a relation between the ability to inter- linated vagal pathways might, under different condi- pret data and the technology available to collect, store, tions, respond to two regulatory systems. During and display fetal heart rate data. This point is critical conditions when blood gases are within normal range in the clinical interpretation of fetal heart rate patterns, the vagal ®bers of the nucleus ambiguus, as a part of where the neurally mediated oscillations (i.e., RSA) in a ªrhythmicº vagal system, may produce a coordinated fetal heart rate are of low amplitude and relatively high respiratory rhythm in heart rate and bronchial activity frequency. Therefore, the accuracy of R-wave detec- to facilitate oxygen diffusion. However, during con- tion, the quality of the ECG signal, the digitizing rate, ditions of compromise when blood gas status threatens the resolution of fetal heart rate monitors (e.g., often survival, not only would RSA be depressed but the fetal heart rate monitors provide moving averages of respiratory drive involving the nucleus ambiguus also heart rate rather than true beat-to-beat values), and would be suppressed. In the absence of the rhythmic even the limitations of the statistical models used to vagal system, a ªtonicº vagal system may drop heart analyze data may change the characteristics of the data rate level to reduce metabolic demands. Perhaps dur- and limit or bias the interpretation. Because the im- ing fetal distress, a tonic increase in the in¯uence of portant information regarding neural tone to the heart the myelinated vagal ®bers to the heart (i.e., producing via the nucleus ambiguus is conveyed in slight varia- bradycardia) is paralleled by the increase in the tonic tions measured in changes of only a few milliseconds, in¯uence of the unmyelinated vagal ®bers to the gut it has been recommended that heart period data be (i.e., producing meconium). quanti®ed with an accuracy of 1 ms (Berntson et al., Consistent with Jacksonian principles (Jackson, 1997; Riniolo & Porges, 1997). 1958) and the Polyvagal Theory (Porges, 1995, 1997), The current study has limitations. First, uterine con- two alternative neurophysiological mechanisms might traction activity and clinical judgment of decelerations explain the clinical bradycardia observed in the human (i.e., early, late, or variable decelerations) were not fetus. The ®rst explanation is consistent with the phy- temporally indexed with the archived heart rate data. logenetic distinctions between the vagal systems of Thus, it is not possible to relate the current data to the short mammals and reptiles. According to the ®rst expla- more-traditional clinical indices of heart rate deceler- standard long DEV (WILEJ) RIGHT INTERACTIVE

Fetal Heart Rate Patterns 117 ation. Most likely, because 6 fetuses were normally 10th International Conference on Infant Studies, Providence, delivered, the decelerations quanti®ed were of the RI, 1996. variable type. Second, interpretations of the current The authors thank Todd Riniolo for technical assistance results, as with any such research, are dependent on and Jane Doussard-Roosevelt for critical reading of the the particular de®nitions of heart rate decelerations manuscript. and accelerations described herein. Third, the small sample of subjects may have precluded an ability to evaluate the relation between amplitude of RSA and REFERENCES magnitude of deceleration. However, the amplitude of RSA and the magnitude of deceleration might be re- Berntson, G. G., Bigger, J. T., Eckberg, D. L., Grossman, lated, but not in a manner that would be detected with P., Kaufmann, P. G., Malik, M., Nagaraja, H. N., Porges, linear correlation. Perhaps hypoxia or an amplitude of S. W., Saul, J. P., Stone, P. H., & van der Molen, M. W. RSA below some absolute threshold is required to po- (1997). Heart rate variability: Origins, methods, and in- tentiate the magnitude of the deceleration. If these terpretive caveats. Psychophysiology, 34, 623±648. points are supported in future research, then the hy- Davidson, S. R., Rankin, J. H., Martin, C. B., Jr., & Reid, D. L. (1992). Fetal heart rate variability and behavioral pothesis could be supported by group differences be- state: Analysis by power spectrum. American Journal of tween distressed and normal fetuses, even in the situ- Obstetrics and Gynecology, 167, 717±722. ation in which no signi®cant correlations are observed Dawes, G. S., Lobb, M. O., Mandruzzato, G., Moulden, M., within either the normal group, as reported in this Redman, C. W., & Wheeler, T. (1993). Large fetal heart study, or a distressed group. The data set did not pro- rate decelerations at term associated with changes in fetal vide an adequate test of the hypothesis within a dis- heart rate variation. American Journal of Obstetrics and tressed group because there was only 1 compromised Gynecology, 168, 105±111. fetus. Fourth, because the observations are limited by DiPietro, J. A., Hodgson, D. M., Costigan, K. A., Hilton, S. the sample (i.e., primarily healthy fetuses) and by the C., & Johnson, T. R. (1996). Development of fetal move- temporal constraints of the analyses (at least 60 s be- ment Ðfetal heart rate coupling from 20 weeks through tween decelerations), the study may have been biased term. Early Human Development, 44, 139±151. Donchin, Y., Caton, D., & Porges, S. W. (1984). Spectral to study only variable decelerations. Future research analysis of fetal heart rate in sheep: The occurrence of is needed to ascertain whether the Polyvagal Theory respiratory sinus arrhythmia. American Journal of Ob- provides a plausible explanation for the physiological stetrics and Gynecology, 148, 1130±1135. mechanisms determining early and late decelerations. Emory, E. K., & Noonan, J. R. (1984). Fetal cardiac re- In summary, the Polyvagal Theory provides a new sponding: A correlate of birth weight and neonatal behav- and unique perspective on the interpretation of fetal ior. Child Development, 55, 1651±1657. heart rate patterns. The theory provides an explanation Ford, T. W., Bennett, J. A., Kidd, C., & McWilliam, P. N. for the paradoxical observation of vagal bradycardia (1990). Neurons in the dorsal motor vagal nucleus of the in the absence of tonic vagal tone manifested in beat- cat with nonmyelinated axons projecting to the heart and to-beat variability. By increasing our understanding of lungs. Experimental Physiology, 75, 459±473. the mechanisms responsible for the onset of and re- Goodlin, R. C., & Haesslein, H. C. (1977). Fetal reacting bradycardia. American Journal of Obstetrics and Gyne- covery from bradycardia, methods might be imple- cology, 129, 845±855. mented to alert clinicians to states in which potentially Groome, L. J., Mooney, D. M., Bentz, L. S., & Wilson, J. life-threatening bradycardia are most likely to occur. D. (1994). Vagal tone during quiet sleep in normal human With this knowledge, the probability of timely inter- term fetuses. Developmental Psychobiology, 27, 453± vention, and thus, the survival prospects during deliv- 466. ery of the fetus at risk, might be improved. Further Hopkins, D. A., Bieger, D., de Vente, J., & Steinbusch, research evaluating RSA level as a predictor of the W. M. (1996). Vagal efferent projections: Viscerotopy, frequency of bradycardia and the time course of vagal neurochemistry, and effects of vagotomy. Progress in responses relative to changing environmental varia- Brain Research, 107, 79±96. bles may bring us closer to this goal. Jackson, J. H. (1958). Evolution and dissolution of the ner- vous system. In J. Taylor (Ed.), Selected writings of John Hughlings Jackson (pp. 45±118). London: Stapes Press. James, L. S., Yeh, M. N., Morishima, H. O., Daniel, S. S., NOTES Caritis, S. N., Niemann, W. H., & Indyk, L. (1976). Um- bilical vein occlusion and transient acceleration of the fe- Portions of this manuscript have been presented at the tal heart rate. Experimental observations in subhuman pri- Thirty-Fifth Annual Meeting of the Society for Psycho- mates. American Journal of Obstetrics and Gynecology, short physiological Research, Toronto, Canada, 1995, and the 126, 276±283. standard long DEV (WILEJ) LEFT INTERACTIVE

118 Reed et al.

Jones, J. F. X., Wang, Y., & Jordan, D. (1995). Heart rate of the neural regulation of the . responses to selective stimulation of cardiac vagal C ®bers In C. S. Carter, B. Kirkpatrick, & I. I. Lederhendler (Eds.), in anesthetized cats, rats, and rabbits. Journal of Physi- The integrative neurobiology of af®liation (pp. 62±77). ology (London), 489, 203±214. Annals of the New York Academy of Science, Vol. 807. Krebs, H. B., Petres, R. E., Dunn, L. J., & Smith, P. J. Porges, S. W., & Bohrer, R. E. (1990). Analyses of periodic (1982). Intrapartum fetal heart rate monitoring. VI. Prog- processes in psychophysiological research. In J. T. Ca- nostic signi®cance of accelerations. American Journal of cioppo & I. G. Tassinary (Eds.), Principles of psycho- Obstetrics and Gynecology, 142, 297±305. physiology: Physical, social, and inferential elements (pp. Kruse, J. (1982). Electronic fetal monitoring during labor. 708±753). New York: Cambridge University Press. Journal of Family Practice, 15, 35±42. Potter, E. K., & McCloskey, D. I. (1986). Effects of hypoxia Levy, M. N. (1984). Cardiac sympathetic±parasympathetic on cardiac vagal efferent activity and on the action of the interactions. Federal Proceedings, 43, 2598±2602. at the heart in the dog. Journal of the Auto- Lunneborg, C. E. (1986). Con®dence intervals for a quartile nomic Nervous System, 17, 325±329. contrast: Applications of the bootstrap. Journal of Applied Richter, D. W., & Spyer, K. M. (1990). Cardiorespiratory control. In A. D. Loewy & K. M. Spyer (Eds.), Central Psychology, 71, 451±456. regulation of autonomic function (pp. 189±207). New McAllen, R. M., & Spyer, K. M. (1978). Two types of vagal York: Oxford University Press. preganglionic motoneurons projecting to the heart and Riniolo, T. C., & Porges, S. W. (1997). Inferential and de- lungs. Journal of Physiology (London), 282, 353±364. scriptive in¯uences on measures of respiratory sinus ar- Paul, R. H., Suidan, A. K., Yeh, S. Y., Schifrin, B. S., & rhythmia: Sampling rate, r-wave trigger accuracy, and Hon, E. H. (1975). Clinical fetal monitoring. VII. The variance estimates. Psychophysiology, 34, 1±9. evaluation and signi®cance of intrapartum baseline FHR Rosen, M. G., & Dickinson, J. C. (1993). The paradox of variability. American Journal of Obstetrics and Gynecol- electronic fetal monitoring: More data may not enable us ogy, 123, 206±210. to predict or prevent infant neurologic morbidity. Amer- Porges, S. W. (1985). Method and apparatus for evaluating ican Journal of Obstetrics and Gynecology, 168, 745± rhythmic oscillations in aperiodic physiological response 751. systems. U.S. patent 4,510,944. Vanhoutte, P. M., & Levy, M. N. (1979). Cholinergic inhi- Porges, S. W. (1995). Orienting in a defensive world: Mam- bition of adrenergic neurotransmission in the cardiovas- malian modi®cations of our evolutionary heritage. A po- cular system. In C. M. Brooks, K. Koizumi, & A. Sato lyvagal theory. Psychophysiology, 32, 301±318. (Eds.), Integrative functions of the autonomic nervous Porges, S. W. (1997). : An evolutionary by-product system (pp. 156±176). Tokyo: University of Tokyo Press.

short standard long