Janet A. DiPietro Department of Population and Health Sciences Johns Hopkins University Baltimore, MD 21205 Kathleen A. Costigan Eva K. Pressman Division of Maternal±Fetal Medicine Johns Hopkins School of Medicine Baltimore, MD 21205 Antenatal Origins of Individual Differences in Heart Rate Jane A. Doussard-Roosevelt Institute of Child Study University of Maryland at College Park, College Park, MD 20742 Received 11 May 1999; accepted 25 July 2000 ABSTRACT: This study examines prenatal-to-postnatal stability in heart rate and variability from mid-gestation through the ®rst year of life. Fetal heart rate data were collected from 52 healthy fetuses at 24, 30, and 36 weeks gestation, and again at 2 weeks and 12 months of age. Fetal heart rate measures were stable during gestation and positively associated with neonatal and infant measures. Maternal pulse rate and oxygen saturation were moderately associated with fetal heart rate. Together, fetal cardiac (heart rate and variability) and maternal physiologic measures (blood pressure and oxygen saturation) explained 40 and 48% of the variance in heart rate and variability, respectively, at 1 year of age. These common measures of individual differences in autonomic function are enduring characteristics that originate during fetal development. ß 2000 John Wiley & Sons, Inc. Dev Psychobiol 37: 221±228, 2000 Keywords: fetal development; fetal heart rate; infant heart rate; infant development ``It will be seen that the infant heart rate is suggestively lower than the fetal heart rate, and is signi®cantly more variable.'' (Sontag & Richards, 1938, p. 25) Documentation of the developmental trajectory of heart rate from the prenatal to postnatal periods was one of the many goals of the 1930s Fels studies of fetal Correspondence to: J. A. DiPietro ß 2000 John Wiley & Sons, Inc. behavior, but methodological limitations obscured discovery of within-individual consistency between the fetus and the infant at that time. Recently, models of antenatal programming of autonomic function and disease in adulthood (Barker, 1995; Phillips & Barker, 1997) have contributed to a resurgence in interest in the fetal period. Although it is clear that parturition does not represent a signi®cant damarcation in neural development (Als, 1982; Prechtl, 1984), relatively few studies have attempted to document intraindividual, prenatal-to-postnatal stability in the same domain. These have been limited to investigations of motor activity (DiPietro, Hodgson, Costigan, & Johnson, 1996; Groome et al., 1999; Shadmi Homburg, & 222 DiPietro et al. Insler, 1986) and behavioral state (Groome, Swiber, Atterbury, Bentz, & Holland, 1997). We are aware of only two published studies which have examined stability in heart rate. In the ®rst, correlations between late third trimester fetal heart rate and infant heart rate measured at multiple ages through the ®rst year were consistently positive (r ranges from .25 to .73), although sample size did not exceed 17 at any age (Lewis, Wilson, Ban, & Baumel, 1970). The second study (Thomas, Haslum, MacGillivray, & Golding, 1989), based on a nationally representative dataset of 11,000 10-year-old children, determined that normal fetal heart rate during labor (i.e., 120±160 bpm) was associated with signi®cantly higher heart rate during childhood when compared to cases with heart rate outside this range, although the magnitude of the mean difference was small (1.4 bpm). The focus of the current study is on the development of mean heart rate and variability from the fetus through the ®rst year of life. Cardiac measures are commonly used indicators of autonomic function in developmental research (e.g., Fox, 1989; Huffman et al., 1998; Porges, 1992; Richards, 1985). Measures include both tonic and reactive heart rate as well as a variety of metrics for variability, ranging from global, time-based descriptives (e.g., standard deviation, mean squares of successive differences) to more complex methods designed to isolate speci®c frequency components (e.g., cardiac vagal tone, RSA) (Bernston et al., 1997). Moderate stability in measures of heart rate and variability have been demonstrated in preterm infants prior to term (DiPietro, Caughy, Cusson, & Fox, 1994) and in full-term infants during infancy (Fox, 1989; Fracasso, Porges, Lamb, & Rosenberg, 1994; Izard et al., 1991; Snidman, Kagan, Riordan, & Shannon, 1995), although there are inconsistencies across studies in the patterning of signi®cant relations. Establishing stability is requisite to the orientation that patterns of heart rate represent underlying constitutional attributes of neural function and autonomic regulation. Stability in fetal heart rate and variability within individuals has been demonstrated in fetuses beginning at 20 weeks gestation (DiPietro et al., 1996; Nijhuis et al., 1998). Factors which contribute to baseline fetal heart rate and variability are not well documented, although tonic maternal heart rate (Patrick, Campbell, Carmichael, & Probert, 1982) and sympathetic arousal induced by physical exercise (Artal et al., 1986) exert some in¯uence. The aim of the current study is to document prenatal-topostnatal stability in heart rate, and to evaluate maternal contributions to these measures before and after birth. METHOD Participants Participants were 52 nonsmoking women with singleton pregnancies and their offspring. Inclusion criteria included low-risk, uncomplicated pregnancies with gestational age dating based on a pregnancy test within 2 weeks of missed menstrual period and/or ®rst trimester obstetric/ultrasound examination. Demographic and medical data were collected by interview and medical chart review. The sample consisted of primarily healthy, well-educated, and employed women (M maternal age 29.9 yrs (SD 3.5); M education 16 (SD 2.6)). All infants included in this analysis were delivered at term (M GA 39.6, SD 1.1, range 37±41 weeks), of normal birthweight (M 3502 g, SD 470) and discharged from the regular newborn nursery according to routine schedules (5 min Apgar M 8.9, range 7±10). Sixty percent (n 31) of offspring were boys. Design and Procedures Fetal Data Collection and Quanti®cation. Data were collected at 24, 30, and 36 weeks gestational age. Women were monitored in a left lateral recumbent position while resting quietly. Maternal radial pulse rate, blood pressure, and oxygen saturation (SpO2) were measured at the beginning of each recording, at 20 min into the recording and again after 40 min.1 These three values were averaged. All 52 subjects provided data at each gestational age. Fetal heart rate was collected from a fetal cardiotocograph (Toitu, MT320) using a single wide array Doppler transducer positioned on the maternal abdomen with an elastic belt. The current generation of fetal heart rate monitors detect fetal heart motions and quantify heart rate by processing small segments of serial Doppler-generated waveforms using autocorrelation techniques. Data were collected during 50 min of undisturbed monitoring and digitized at 5 Hz using an A-D converter board and commercial data collection package (LabVIEW NB, National Instruments Corp., Austin, TX). A series of error rejection procedures, developed in our laboratory and based on moving averages of acceptable values were 1 Pulse oxygen saturation was measured using a digit probe (Nellcor Pulse Oximeter, Hayward, CA) and re¯ects the proportion of hemoglobin-bound oxygen in pulsatile arterial blood. Fetal-Infant Heart Rate applied to remove movement artifact.2 Data were interpolated to preserve temporal integrity but interpolated data were not used in data analyses. Fetal heart rate and variability, computed as standard deviation,3 were analyzed in 30 s epochs; means were calculated for the entire 50 min recording. The mean amount of rejected data was 6.5, 5.2, and 5.4% at 24, 30, and 36 weeks, respectively. Postnatal Data Collection and Quanti®cation. Forty-one neonates (79%) returned for testing at 2 weeks postnatal age (M 14.2 days postpartum, SD 1.9); the average length of ECG recording was 8 min (SD 2.4). ECG was recorded using three disposable electrodes triangulated on the infant chest. The signal was ampli®ed (PhysioControl, Model Lifepak 5, Plainview, NY) and recorded on an instrumentation tape recorder (Vetter FM, Model C4). Because state exerts an in¯uence on heart rate, effort was made to collect data during a consistent period of active (i.e., REM) sleep state. Five infants could not be recorded due to prolonged fussiness or high motor activity, four were successfully recorded during a period of either drowsiness or quiet wakefulness, and the remainder during predominantly active sleep. Most of the subjects were breast-fed (n 35), but because feeding method in infancy affects cardiac patterns (Butte, Smith, & Garza, 1991; DiPietro, Larson, & Porges, 1987; Zeskind, Marshall, & Goff, 1992), analyses were conducted on both the full sample and without the six exclusively formulafed infants. At 1 year, 35 (67%) subjects returned for follow-up (M 53.4 weeks; SD 1.0); the average length of ECG recording was 7 min (SD 1.6). ECG data were ampli®ed and recorded in the same manner, while the child was seated quietly on the parent's lap and looking at a picture book. Data for three subjects were unusable as a result of excessive movement artifact. The main reasons for the relatively low 2 Distinguishing artifactual from real data is a dif®cult but critical component in quantifying FHR early in the third trimester because motor activity can result in poor quality signal if the fetal heart moves beyond the Doppler ®eld. The digital data underwent a series of error rejection procedures based on 5-point moving medians of acceptable values. In brief, values were rejected if they were beyond the criterion limits of 1.5±.75 of the previous median; the speci®c values against which each prior data point was compared were adjusted by .03 per rejected s, until attaining the criterion limits. These algorithms were developed after comparing the polygraphic output of the monitor of the computerized output of several hundred records and ultimately validated against visual inspection of 7,500 min of collected polygraphic data. Details of the error rejection program are available upon request. 3 Root mean square (RMS) deviation values were also computed as a measure of variability but not used in this analysis in order to maintain consistency with neonatal metrics. 223 follow-up rate involved participants moving from the area and non availability of the testing site during evenings and weekends. However, there were no signi®cant differences in neonatal or demographic characteristics between tested and untested subjects at either age. The ECG data were quanti®ed of¯ine by digitizing the tape recorded data using a Vagal Tone Monitor (Delta-Biometrics, Inc., Bethesda, MD), which detects R-waves and times sequential heart periods to the nearest millisecond. Data were time-sampled to the nearest millisecond (1000 Hz). Heart periods were output to a personal computer and Mxedit software (Delta-Biometrics) was used to display the data, edit outliers and quantify heart rate measures. Heart rate, variability (standard deviation of heart rate values), and cardiac vagal tone4 using the Porges method (Porges, 1985) were computed in 30-second epochs. Means were calculated for the entire recording. RESULTS Development of Heart Rate and Variability Descriptive measures of fetal and infant heart rate and variability for the full sample are presented in Table 1. Preliminary analyses revealed no effects based on fetal sex during either the fetal or postpartum period. T-tests were used to examine whether there were signi®cant, or near-signi®cant, differences in the fetal measures between the tested and untested samples at each age. None were detected. Stability over Time Pearson correlation matrices for HR and HRV from 24 weeks gestation through the end of the ®rst year of life are presented in Table 2. There is evidence of stability in both fetal HR and fetal HRV during the last half of gestation (r ranges from .30 to .72). Fetal and neonatal HR and HRV measures are also positively associated; 4 Cardiac vagal tone was quanti®ed through a series of procedures developed by Porges as follows: (1) the duration between successive heartbeats was time sampled every 200 ms for neonates and 250 ms for infants; (2) linear and complex trends were removed using a 21point cubic moving polynomial stepped through the heart period data. The smooth template from this procedure was subtracted from the dataset, providing a trend-free residual heart period series; (3) time-series analyses were conducted to extract the component of heart period variance in the frequency band associated with spontaneous breathing in neonates (.30±1.30 Hz) and older infants (.24±1.04 Hz). The natural logarithm of this variance measure produced the vagal tone index, the statistic used to estimate cardiac vagal tone from the amplitude of respiratory sinus arrhythmia. 224 DiPietro et al. Table 1. Fetal and Infant Heart Rate and Variability Values Heart rate Heart rate variability n Mean SD Mean SD Fetus 24 weeks 30 weeks 36 weeks 52 52 52 146.0 139.9 138.4 5.5 6.1 7.6 3.9 5.1 5.7 .9 1.4 1.7 Infant 2 weeks 1 year 36 32 142.9 129.1 9.2 8.7 6.0 5.9 .5 .7 Note. Fetal data have been presented previously (DiPietro et al., 1998). Table 2. Intercorrelations of Fetal, Neonatal, and Infant Heart Rate Measures Fetal (n 52) 24 weeks 30 weeks 36 weeks Neonatal (n 36) Infant (n 32) Heart rate 24 weeks 30 weeks 36 weeks Neonatal Ð Ð Ð Ð .73*** Ð Ð Ð .56*** .59*** Ð Ð .22 / .27a .26 / .32* .31*/.35* Ð .42** .32* .40** .36*b Heart rate variability 24 weeks 30 weeks 36 weeks Neonatal Ð Ð Ð Ð .68*** Ð Ð Ð .30* .61*** Ð Ð .02 / .03a .18/ .28 .26/.33* Ð .04 .29* .47** .25b a Values in second neonatal column are correlations with bottle-fed infants (n 6) excluded. n 26. *p < :05. **p < :01. ***p < :001. Based on one-tailed test. b for these pairs, correlations >.27 are signi®cant based on one-tailed criteria. Analyses conducted excluding the exclusively formula-fed infants indicated that removing this source of variance tends to increase the magnitude of the coef®cients, also presented in Table 2. For both HR and HRV, associations from fetal to 1 year recordings are somewhat larger in magnitude than those to the neonatal period. Associations between the two cardiac measures were also investigated. During gestation, fetal HRV and HR were not correlated at any gestational age. Postnatally, HR and HRV were negatively related (r (36) ÿ .24 in the neonate and r (32) ÿ .73 at 1 year). Table 3 presents cross-correlations between fetal HRV and infant HR, and fetal HR and infant HRV. Higher fetal HR is negatively associated with infant, but not neonatal, HRV (r ranges from ÿ .28 to ÿ .35). Greater fetal HRV is associated with faster HR in neonates, but slower HR at 1 year. Doppler-based fetal HR data are not appropriate for calculating fetal vagal tone,5 but because vagal tone is a speci®c measure of variability, associations between antenatal heart rate measures and later vagal tone are of interest. Correlations between fetal HR and vagal tone during the neonatal and infant periods were similar to those 5 In the current study we have relied on standard cardiotocography to measure heart rate. In contrast to ECG-based detection of R-waves using high sampling rates, Doppler-generated fetal heart rate data do not quantify interbeat intervals with the degree of precision necessary for detection of RSA. In addition, the fetus displays fetal breathing movements episodically, making continuous quanti®cation of RSA inappropriate. Fetal ECG data are most often collected during the intrapartum period from scalp electrodes following rupture of membranes. More recent techniques to collect fetal ECG data noninvasively through transabdominal monitoring are in development, but this technology is cumbersome, not widely available, and has not been well implemented in mid-gestation. We did not consider it to be a useful candidate for this longitudinal study. Fetal-Infant Heart Rate Table 3. Cross correlations among fetal heart rate and infant measures Heart rate variability Heart rate 24 weeks 30 weeks 36 weeks Heart rate variability 24 weeks 30 weeks 36 weeks Neonate Infant .11/.03a .09/.08 ÿ .11/ ÿ .20 ÿ .29 ÿ .28 ÿ .35* Heart rate .35*/.36*a .29*/.31* .26/.28 ÿ .02 ÿ .29* ÿ .28 a Values in second neonatal column are correlations with bottlefed infants (n 6) excluded. *p < :05, based on one-tailed test. for HRV, although less consistent and of lower magnitude. Associations with fetal HR at 24, 30, and 36 weeks are r .28, .26, and ÿ .08, respectively, in the neonate; r ÿ .15, ÿ .15, and ÿ .30 in infants. Correlations between vagal tone and fetal HRV at each fetal period are r ÿ .12, .20, and .17, respectively, in the neonate; r ÿ .07, .22, and .37 at 1 year. Associations between Maternal and Fetal Measures Mean values for the maternal measures for the full sample, and results of repeated measures analysis of variance for changes over gestation are presented in Table 4. Blood pressure was analyzed as mean arterial pressure (MAP), a composite measure of diastolic and systolic components computed as: (2*diastolic) systolic) 3. Maternal pulse rate and MAP were relatively stable during gestation (r .59 to .74 for pulse rate; r .41 to .69 for MAP); pulse oxygenation intracorrelations were lower (r .20 to .32). The contemporaneous in¯uence of maternal pulse rate, oxygenation (SpO2), and blood pressure on fetal HR and HRV during gestation were evaluated using mixed effects models (SAS PROC MIXED). Developmental trends during gestation are robust for heart rate and variability (p <.0001); similar results regarding the change in HR and HRV over gestation, generated by repeated measures analysis of variance but exclusive of maternal factors, have been presented previously (DiPietro, Costigan, Shupe, Pressman, & Johnson, 1998). Results of the mixed effects analysis for the association between maternal physiologic and fetal heart rate measures revealed no signi®cant relations between fetal HRV and maternal measures (all ts<1). Howver, fetal HR was moderately associated with maternal pulse (t ÿ 2.09, p<.04) and oxygen saturation (t 1.93, p<.06). The association with maternal MAP approached a trend level of signi®cance (t ÿ 1.63, p .11). Prediction of 1 Year Fetal Heart Rate Measures Multiple regression was used to model the prediction of 1-year heart rate measures from antenatal variables. Because of the small sample size at 1 year, preliminary bivariate analyses were conducted to exclude weak associations. Measures considered were of maternal heart rate, oxygen saturation, and blood pressure, averaged over the three time points, as well as demographic variables including maternal age and education level. Selection of fetal measures was gui- Table 4. Mean maternal physiologic values by gestational age Mean Pulse rate (bpm) 24 weeks 30 weeks 36 weeks Oxygen saturation (SpO2; %) 24 weeks 30 weeks 36 weeks Mean arterial pressure (MAP; mm Hg) 24 weeks 30 weeks 36 weeks @ p < :10; ***p < :001. 225 SD 84.8 87.4 86.2 9.1 9.9 11.0 97.9 96.8 97.1 1.0 1.8 1.0 70.8 72.0 76.7 7.1 6.9 9.9 F, Time 2.54@ 11.53*** 15.44*** 226 DiPietro et al. Table 5. Results of Multiple Regressions in Prediction of Infant Heart Rate from Fetal Heart Rate (HR), Fetal Heart Rate Variability (FHRV), Maternal Mean Arterial Blood Pressure (MAP), and Oxygen Saturation (SpO2) B I. Dependent measure: Infant heart rate 24 weeks FHR .58 30 weeks FHRV ÿ 1.81 Multiple R .48; R2 .23; F(2,29) 4.38* Maternal MAP ÿ .28 Maternal SpO2 ÿ 3.24 Multiple R .63; R2 .40; F(4,27) 4.45** II. Dependent measure; Infant heart rate variability 36 weeks FHRV .19 36 weeks FHR ÿ .03 Multiple R .60; R2 .36; F(2,29) 8.27*** Maternal MAP ÿ .01 Maternal SpO2 ÿ .28 Multiple R .69; R2 .48; F(4,27) 6.14*** t pr 2.34* ÿ 1.95@ .41* ÿ .35 ÿ 1.35 ÿ 1.89@ ÿ .25 ÿ .34 3.67** ÿ 2.05* .58 ÿ .37 .75 2.07* .14 .37 Note. @p < :10; *p < :05; **p < :01; ***p < :001. ded by the pattern of correlations observed in Tables 3 and 4; the earliest associations to either attain a correlation of .30, or the highest correlation across gestation were used. For infant HR, 24-week fetal HR (r .42) and 30-week fetal HRV (r ÿ .29) values were used; for infant HRV, 36-week fetal HR (r ÿ .35) and HRV (r .47) were used. Maternal pulse rate, age, or education did not near signi®cance. Both fetal measures were entered in the ®rst block, followed by both maternal variables. The predictive models for infant HR and HRV, including unstandardized regression coef®cients (B), t-values, and partial correlations for each variable are presented in Table 5. In each model, fetal HR and HRV account for signi®cant variance in prediction of infant measures (r2 .23 and .36 for HR and HRV, respectively). Addition of maternal blood pressure and oxygen saturation signi®cantly increase the explained variance in infant HR and HRV r2 change .16, (F(4,27) 3.69, p<.05) and r2 change .11 (F(4,27) 2.92, p <.05) respectively. DISCUSSION The results of this longitudinal investigation demonstrate that individual differences in heart rate and variability originate before birth. We have replicated earlier ®ndings from a smaller sample of intraindividual stability in both fetal HR and HRV during the second half of gestation (DiPietro et al., 1996) and documented consistency in cardiac measures from the fetus to 1-year-old infant. Together, these provide strong support that heart rate and variability are enduring characteristics which index aspects of autonomic function. Our detection of signi®cant associations between 36-week fetal HR and HRV con®rm those reported in an earlier study (Lewis et al., 1970), although the magnitude of the correlations vary. A recent report of preliminary data in which subjects were strati®ed into low and high HR and HRV groups based on 36-week fetal data did not detect stability between fetal and neonatal HR within 3 days postpartum, but a signi®cant association was found at 2 months of age across sleep state (Fifer et al., 1998). The time-intensive nature of fetal data collection typically results in small sample sizes, but convergent ®ndings across studies using different methodologies provides con®dence in these data. Our data further reveal that predictive relations between fetal and postnatal HR begin by 24 weeks gestation. Positive relations with fetal HRV do not emerge until 30 weeks gestation. This observation is likely due to the later maturation of neuroregulatory in¯uences on variability than those that control rate (Martin, 1978). Longitudinal relations were observed not only within cardiac measures, but also across them. Fetal HR was not associated with neonatal variability but higher fetal HR was associated with lower HRV at 1 year. Higher fetal HRV was associated with higher heart rate in the neonatal period, but also with lower HR at 1 year. Since both the current data as well as that reported by others (DiPietro et al., 1994; Fracasso et al., 1994; Izard et al., 1991) reveal negative postnatal within age relations between rate and Fetal-Infant Heart Rate variability,6 the 1 year ®ndings are expected but the positive neonatal associations are not. Given the small sample size, we are hesitant to propose hypotheses for this unexpected result. However, the lack of any relation between HR and HRV during gestation indicates that the contemporaneous associations develop after birth. The positive relation fetal HRV and neonatal HR may re¯ect this transition and the in¯uences of cardioacceleratory processes during these periods. Fetal HR and HRV explained 23% of the variance in 1 year HR and 36% of the variance in 1 year HRV, despite many sources of error that might have limited our ability to detect these relations. Efforts to control known sources of variance (i.e., method of feeding) tended to increase the magnitude of the relations between fetal and neonatal HR and HRV. Prominent among these is the different methods of heart rate detection used during prenatal and postnatal recordings. That is, fetal heart rate was quanti®ed through Doppler detection of fetal heart motions in contrast to detection of R-waves in the infant. Finally, collecting ECG data in infancy is challenging and subject to uncontrollable in¯uences of behavioral state and motor activity during the recording, which limits the duration of data collection feasible during infancy. The higher individual stability in HR and HRV observed during the fetal period may be related to regulation provided from maternal sources, but may also be related to the longer recording period (50 min) as compared to the necessarily brief period of data collection after birth (< 10 min at either age). During gestation, higher maternal oxygen saturation was associated with lower fetal heart rate and there was a near-signi®cant relation between maternal and fetal heart rate. There is relatively little understanding of how maternal physiologic functioning contributes to fetal functioning, although a robust relation between maternal and fetal heart rate has been documented near term using averaged values over 24hour continuous recordings (Patrick et al., 1982). Because our measure of maternal heart rate was a crude one based on palpated radial pulse, we would expect more precise ascertainment of maternal HR through continuously recorded ECG to provide more robust relations with fetal HR; these efforts are currently underway in our laboratory. Similarly, our failure to ®nd relations between maternal pulse and fetal HRV at this time should not be considered as suf®cient evidence for lack of a relation. In contrast to the few modest contemporaneous relations between maternal variables and fetal func6 Typically, positive relations between heart period and variability are reported. 227 tioning, maternal oxygen saturation and blood pressure added signi®cant unique variance to the prediction of infant HR and HRV, increasing the explained variance by 17 and 12%, respectively. In particular, higher maternal SpO2 was predictive of both lower infant heart rate and higher variability. A relationship between increased SpO2 and decreased fetal heart rate has been described in healthy term pregnancies following induced hyperoxia (Polvi, Pirhonen, & Erkkola, 1995). Higher normotensive maternal blood pressure during pregnancy has been correlated with greater neonatal irritability (Chisholm, Woodson, & DaCosta-Woodson, 1978). We believe the current results to be the ®rst to document maternal physiologic effects on infant autonomic functioning. Lower HR and higher variability are considered representative of higher vagal tone; thus higher maternal perfusion or oxygen saturation may enhance uteroplacental function and accelerate development. In summary, these results provide the most comprehensive report of the stability in development of HR and HRV from the fetus to infant. They provide support for the use of cardiac measures in research of individual differences, which is often predicated on assumptions of constitutionality. Support is also provided for the utility of both global descriptives of heart rate variability as well as the suf®ciency of cardiotocography in measurement of fetal heart rate. Further elucidation of the role of intrinsic and extrinsic antenatal in¯uences on subsequent autonomic function is key to understanding the extent and nature of the programming of subsequent function that may occur prior to birth. NOTES This research was supported by grant R01 HD27592, National Institute of Child Health and Human Development, awarded to the ®rst author, and by R01 HD22628 awarded to S. W. Porges and the fourth author. The investigators wish to thank the diligent and generous participation of our study families, without which this research would not have been possible. REFERENCES Als, H. (1982). Toward a synactive theory of development: Promise for the assessment and support of infant individuality. Infant Mental Health Journal, 3, 229±243. Artal, R., Rutherford, S., Romem, Y., Kammula, R., Dorey, F., & Wiswell, R. (1986). Fetal heart rate responses to maternal exercise. American Journal of Obstetrics and Gynecology, 155, 729±733. 228 DiPietro et al. Barker, D. J. (1995). The fetal and infant origins of disease. European Journal of Clinical Investigation, 25, 457±463. Bernston, G., Bigger, J., Eckberg, D., Grossman, P., Kaufmann, P., Malik, M., Nagaraja, H., Porges, S., Saul, J., Stone, P., & van-der-Molen, M. (1997). Heart rate variability: Origins, methods, and interpretative caveats. Psychophysiology, 34, 623±648. Butte, N., Smith, E., & Garza, C. (1991). Heart rates of breast and formula-fed infants. Journal of Pediatric Gastroenterology and Nutrition, 13, 391±396. Chisholm, J., Woodson, R., & DaCosta-Woodson, E. (1978). Maternal blood pressure in pregnancy and newborn irritability. Early Human Development, 2, 171±178. DiPietro, J. A., Caughy, M. O. B., Cusson, R., & Fox, N. A. (1994). Cardiorespiratory functioning of preterm infants: Stability and risk associations for measures of heart rate variability and oxygen saturation. Developmental Psychobiology, 27(3), 137±152. DiPietro, J. A., Costigan, K. A., Shupe, A. K., Pressman, E. K., & Johnson, T. R. B. (1998). Fetal neurobehavioral development: Associations with socioeconomic class and fetal sex. Developmental Psychobiology, 33, 79±91. DiPietro, J. A., Hodgson, D. M., Costigan, K. A., & Johnson, T. R. B. (1996). Fetal antecedents of infant temperament. Child Development, 67, 2568±2583. DiPietro, J. A., Larson, S. K., & Porges, S. W. (1987). Behavioral and heart rate pattern differences between breast-fed and bottle-fed neonates. Developmental Psychology, 23, 467±474. Fifer, W. P., Hurtado, A., Garcia, W., & Myers, M. M. (1998). Fetal to newborn continuities in cardiac control. Developmental Psychobiology, 33, 371. Fox, N. A. (1989). Psychophysiological correlates of emotional reactivity during the ®rst year of life. Developmental Psychology, 25, 364±372. Fracasso, M., Porges, S., Lamb, M., & Rosenberg, A. (1994). Cardiac activity in infancy: Reliability and stability of individual differences. Infant Behavior and Development, 17, 277±284. Groome, L., Swiber, M., Holland, S., Bentz, L., Atterbury, J., & Trimm, R. (1999). Spontaneous motor activity in the perinatal infant before and after birth: Stability in individual differences. Developmental Psychobiology, 35, 15±24. Groome, L. J., Swiber, M. J., Atterbury, J. L., Bentz, L. S., & Holland, S. B. (1997). Similarities and differences in behavioral state organization during sleep periods in the perinatal infant before and after birth. Child Development, 68(1), 1±11. Huffman, L., Bryan, Y., delCarmen, R., Pedersen, F., Doussard-Roosevelt, J. A., & Porges, S. (1998). Infant temperament and cardiac vagal tone: Assessments at twelve weeks of age. Child Development, 69, 624±635. Izard, C., Porges, S. W., Simons, R., Haynes, O., Hyde, C., Parisi, M., & Cohen, B. (1991). Infant cardiac activity: Developmental change and relations with attachment. Developmental Psychology, 27, 432±439. Lewis, M., Wilson, C., Ban, P., & Baumel, M. (1970). An exploratory study of resting cardiac rate and variability from the last trimester of prenatal life through the ®rst year of postnatal life. Child Development, 41, 799±811. Martin, C. (1978). Regulation of the fetal heart rate and genesis of FHR patterns. Seminars in Perinatology, 2, 131±146. Nijhuis, I., tenHof, J., Mulder, E., Nijhuis, J., Narayan, H., Taylor, D., Westers, P., & Visser, G. (1998). Numerical fetal heart rate analysis: Nomograms, minimal duration of recording, and intrafetal consistency. Prenatal and Neonatal Medicine, 3, 314±322. Patrick, J., Campbell, K., Carmichael, L., & Probert, C. (1982). In¯uence of maternal heart rate and gross fetal body movements on the daily pattern of fetal heart rate near term. American Journal of Obstetrics and Gynecology, 144, 533±538. Phillips, D., & Barker, D. (1997). Association between low birthweight and high resting pulse in adult life: Is the sympathetic nervous system involved in programming the insulin resistance syndrome? Diabetic Medicine, 14, 673±677. Polvi, H., Pirhonen, J., & Erkkola, R. (1995). The hemodynamic effects of maternal hypo- and hyperoxygenation in healthy term pregnancies. Obstetrics and Gynecology, 86, 795±799. Porges, S. (1985). Method and apparatus for evaluating rhythmic oscillations in aperiodic physiological response systems, Washington DC: US Patent Of®ce. Porges, S. (1992). Vagal tone: A marker of stress vulnerability. Pediatrics, 90, 498±504. Prechtl, H. F. R. (1984). Continuity and change in early neural development. In H. Prechtl (Ed.), Continuity in neural functions from prenatal to postnatal life (pp. 1±15). Richards, J. E. (1985). Respiratory sinus arrhythmia predicts heart rate and visual responses during visual attention in 14 and 20 week old infants. Psychophysiology, 22, 101± 108. Shadmi, A., Homburg, R., & Insler, V. (1986). An examination of the relationship between fetal movements and infant motor activity. Acta Obstetrica Gynecologica Scandinavia, 65, 335±339. Snidman, N., Kagan, J., Riordan, L., & Shannon, D. (1995). Cardiac function and behavioral reactivity during infancy. Psychophysiology, 32, 199±207. Sontag, L. W., & Richards, T. W. (1938). Studies in fetal behavior: I. Fetal heart rate as a behavioral indicator. Monographs of the Society for Research in Child Development, 3(4 (Serial No. 17)), 1±67. Thomas, P. W., Haslum, M. N., MacGillivray, I., & Golding, M. J. (1989). Does fetal heart rate predict subsequent heart rate in childhood? Early Human Development, 19, 147±152. Zeskind, P., Marshall, T., & Goff, D. (1992). Rhythmic organization of heart rate in breast-fed and bottle-fed newborn infants. Early Development and Parenting, 1, 79±87.