Infant Behavior & Development 42 (2016) 142–151
Contents lists available at ScienceDirect
Infant Behavior and Development
Comparison of fetal and maternal heart rate measures using electrocardiographic and cardiotocographic methods Barbara S. Kisilevsky a,b,∗ , C. Ann Brown a a b
School of Nursing, Queen’s University, Kingston, ON, Canada Department of Obstetrics & Gynaecology, Kingston General Hospital, Kingston, ON, Canada
a r t i c l e
i n f o
Article history: Received 9 November 2015 Received in revised form 23 December 2015 Accepted 27 December 2015 Keywords: Fetus Pregnant women Heart rate ECG Cardiotocograph Voice Reliability
a b s t r a c t Purpose: To determine the reliability at term of: (1) two methods of measuring fetal heart rate (HR), electrocardiographic (ECG, the ‘gold standard’) and cardiotocographic (CTG) and (2) two ECG methods of measuring maternal HR variability over relatively brief periods of time (s–min). Methods: During 20 min of rest (N = 39) and during 2 min of auditory stimulation (mother’s recorded voice, n = 19), fetal HR data were collected using an ECG (Monica AN24) and a Hewlett-Packard Model 1351A CTG. Simultaneously, maternal HR data (n = 37) were collected using the same ECG device (Monica AN24) and a second stand-alone cardiac monitor (Spacelab 514T cardiac monitor with a QRS detector). Results: During 20 min of maternal rest, correlations of individual fetal CTG with ECG measures of HR at each second were moderate to high (r = .57–.97) for 77% of fetuses. Correlations of HR averaged over fetuses and over each of the 20 min were high (r = .93–.97); fetal HR averaged over 20 min varied between devices from 0.0 to 0.8 bpm. During 2 min of maternal voice presentation, correlations of fetal HR over each second were moderate to high (r = .54–.99) for 95% of fetuses and high (all rs = .99) when averaged across fetuses in 30 s or 2 min epochs. Average fetal HR between devices over the 2 min voice varied from 0.0 to 0.6 bpm. Correlations and/or % agreement between the two ECG methods of measuring maternal HR were high. Average maternal HR over 10 min showed 81% of pairs with a difference of ≤1 bpm; correlations for HR variability measures varied from r = .89 to .97. Conclusions: Good reliability was demonstrated between individual spontaneous and auditory induced fetal CTG and ECG with high correlations when HR data were averaged over fetuses or in 30–120 s epochs. High reliability of maternal HR measures was obtained using two ECG devices. © 2016 Elsevier Inc. All rights reserved.
1. Introduction Heart rate measures are one of the few readily available, non-invasive measures for the study of human fetal neurological and behavioral development. For research with infants, children, and adults, cardiac measures including heart rate [HR; in beats per minute (bpm)] and HR variability (i.e., measures of the balance between the sympathetic and parasympathetic branches of the autonomic nervous system influencing heart rate) are extracted from the R–R interval (the time, in ms,
∗ Corresponding author at: School of Nursing, Queen’s University, Kingston, ON K7L 3N6, Canada. E-mail addresses: [email protected] (B.S. Kisilevsky), [email protected] (C.A. Brown). http://dx.doi.org/10.1016/j.infbeh.2015.12.004 0163-6383/© 2016 Elsevier Inc. All rights reserved.
B.S. Kisilevsky, C.A. Brown / Infant Behavior & Development 42 (2016) 142–151
143
between two successive ventricular contractions) of an electrocardiogram (ECG) using an electrocardiographic instrument. Although an ECG is considered the ‘gold standard’ for heart rate measures, for fetal research during gestation when the membranes are intact, often cardiac measures are obtained using a cardiotocograph (CTG) which measures muscular contraction of the heart and employs autocorrelation techniques to compare successive Doppler ultrasound heart beat signals and provide a measure corresponding to the R–R interval of an ECG (Cesarelli, Romano, & Bifulco, 2009). ECG technology has not been readily available for fetal studies because of difficulties inherent in reliably separating the maternal and fetal electrocardiograms (for a technical discussion of fetal HR signal processing issues such as relatively low signal-to-noise ratio and limited knowledge of cardiac function and development see, e.g., Andreotti et al., 2014; Clifford, Silva, Behar, & Moody, 2014; Sameni & Clifford, 2010). As well, from about 24 to 34 weeks gestational age (GA), the vernix caseosa on the fetal skin impedes conduction of fetal electric signals to the maternal surface (e.g., Oostendorp, Oosterom, & Jongsma, 1989). Although the use of in-house ECG technology (e.g., Groome et al., 2000) and commercial equipment (e.g., FEMO, David, Hirsch, Karin, Toledo, & Akselrod, 2007) occasionally have been reported in fetal psychobiological research, little is known regarding the relationship between fetal HR measures collected using CTG vs. ECG (the gold standard) in non-laboring women. See Footnote #1. In uneventful pregnancies during the third trimester of pregnancy, only one report (Ibrahimy, Mohd, Zahedi, & Tsuruoka, 2002) could be found of a direct comparison between ECG and CTG measures. Ibrahimy et al. simultaneously collected FHR data for 10 min during maternal rest from 5 fetuses between 35 and 40 weeks gestation using three ECG electrodes (above and below the maternal umbilicus, right wrist) and a commercial Doppler fetal monitor (IFM-500, BioSys Co., Ltd). The authors reported that the two fetal HR measurements varied ±5 beats per minute (bpm) for 84% of recording. The purpose of the present study was to replicate and extend this earlier work using a larger sample size (n = 39) at term and comparing heart rate measures collected using two commercial fetal HR machines (Monica AN24 ECG vs. Hewlett-Packard CTG) during maternal rest using a longer rest period (20 min) as well as during fetal auditory stimulation which could increase fetal movement activity and interfere with signal detection. In addition, given that the Monica AN24 simultaneously captures and separates both the maternal and fetal ECG, a second maternal ECG was obtained using a stand-alone standard Spacelab 514T cardiac monitor (with a QRS detector) to compare two maternal ECG measures (Monica AN24 ECG vs. Spacelab ECG). Knowledge and understanding of the reliability between fetal ECG and CTG measures collected for relatively brief periods of time (seconds or minutes) are especially relevant for the interpretation of research findings in studies of fetal neurobiological development because the equipment was designed and is most often used and tested for clinical monitoring of heart rate over extended periods of time (hours).1 Studies employing the two different technologies, CTG before birth with ECG measures after birth, have shown a relationship between spontaneous fetal HR and HR variability from 24 weeks GA through the second year of life (DiPietro, Costigan, Pressman, & Doussard-Roosevelt, 2000; DiPietro, Bornstein, Hahn, Costigan, & Achy-Brou, 2007). Moreover, a continuity between term fetal and newborn elicited cardiac responses to similar kinds of stimulus materials has been demonstrated (Kisilevsky & Muir, 1991). Such findings suggest that the two fetal measures yield similar results and that such comparisons are reliable. See example, Footnote #1. 2. Material and methods 2.1. Participants A total of 39 maternal–fetal pairs at term provided the data for this study. Participants were recruited from antenatal clinics at a community teaching hospital in southern Canada. Inclusion criteria were a maternal age of at least 18 years and a low-risk, uneventful, singleton pregnancy at term with delivery of a healthy newborn. Exclusion criteria were any pharmacologically treated co-morbid conditions (e.g., diabetes, hypertension, thyroid disease, depression). Gestational age was calculated from the first day of the last menstrual period if periods were reliable (accuracy rate 75–85%) or from early ultrasound (SD ± 1 week). Data from an additional 15 fetuses were not included because the fetal HR could not be reliably captured by the ECG (n = 14) or CTG (n = 1) technology. Testing of fetuses and newborns was conducted in the Maternal-Fetal-Newborn Studies Laboratory located adjacent to the obstetrical outpatient and inpatient services of the hospital. Sex of the infant was determined at birth. Information on race and socio-economic status are not routinely collected in the Canadian health care system and were not collected for this study. The study was carried out according to ethics approval from the University and Affiliated Teaching Hospitals Research Ethics Board. Women provided informed, voluntary, written consent prior to participation. 2.2. Equipment/stimuli Maternal blood pressure (BP) was obtained using a Lifesource One Step Auto-Inflation Blood Pressure Monitor Model UA-767. The User Manual reports a SD = 3 mmHg.
1 There is a substantive literature examining fetal CTG and/or ECG [obtained from the maternal abdomen (e.g., Reinhard et al., 2013) or fetal scalp electrode (e.g., Clifford, Sameni, Ward, Robinson, & Wolfberg, 2011)] during labor. The studies are conducted for clinical purposes (e.g., see reviews by Amer-Wahlin & Kwee, 2015; Neilson, 2013; commentary by Sholapurkar, 2014) rather than to elucidate normal fetal psychobiological development. They are carried out during a time of substantial physiological stress for the mother and fetus rather than under resting conditions and are not reviewed here.
144
B.S. Kisilevsky, C.A. Brown / Infant Behavior & Development 42 (2016) 142–151
Maternal and fetal ECG signals were captured using a Monica AN24 (Monica Healthcare, Nottingham, UK) wireless fetal monitor. The Monica AN24 is a relatively small (fits in the palm of the hand), wearable, battery-powered device, attached via a lead assembly and five ECG electrodes (4 electrodes, 1 ground) placed on the maternal abdomen. It records electrical signals at the electrodes which are converted into digital format, processed in real-time, and can be downloaded to a computer. The device is a commercially available instrument that provides maternal and fetal HR in beats per minute (bpm) and R–R intervals in ms in singleton pregnancies from 20 weeks gestational age (GA) until the onset of labor for recording sessions lasting up to 16 h. Manufacturer’s software (Monica DK version 1.6) for data acquisition (Bluetooth) and computer storage was used to extract maternal and fetal beat-to-beat heart rate, providing HR in bpm in 0.25 s or 2 s epochs as well as beat-by-beat R–R intervals (duration between consecutive ventricular depolarizations measured by the R waves of the QRS complexes) in 0.25 s epochs or ms. A second continuous recording of fetal HR was obtained using a Hewlett-Packard (HP) cardiotocograph (Model 1351A). The HP cardiotocograph is a ‘desk-top’, electrically powered device which employs Doppler ultrasound, emitted and captured through a circular paddle placed on the maternal abdomen to detect contractions of the fetal heart muscle while the mother is seated or supine. The signals are converted to fetal HR in bpm using autocorrelation techniques with HR in bpm displayed on the instrument screen and a hard copy paper graph of HR over seconds printed. The HP cardiotocograph can be used for single or twin (two paddles) fetuses throughout pregnancy and labor. It was available in North America at the time of this study through the independent measurement company, Agilent Technologies (San Jose, CA). HR in bpm based on cardiac muscular contraction was obtained directly from the device using custom software which captured and averaged 4 readings per second; the average at each second was stored in a computer text file. Both the Monica AN24 and the HP cardiotocograph were designed and are used primarily for clinical assessment of fetal well-being. A second maternal ECG was obtained using a Spacelab 514T cardiac monitor (Squibb Vitatek Inc., Hillsboro) with a QRS detector and three (2 electrodes, 1 ground) latex-free standard surface ECG electrodes in bipolar Lead II configuration on the maternal chest; beat-by-beat R–R intervals were recorded continuously. An analog-digital converter (DAS-16, Metrabyte Corp., Multitest Electronics, Scarborough, ON, Canada) digitized the analog R–R interval output, providing an R–R interval accuracy of 1 ms through a sampling rate of 1000 Hz (Yamamoto & Hughson, 1991). The digital R–R interval output was analyzed using a custom computer software program for spectral analysis of R–R interval variability (Yamamoto & Hughson, 1991). Simultaneous onset and offset of observational (mother at rest) and experimental (auditory stimulus) recording sessions among the three devices were indicated manually using the built-in event markers of the Monica AN24 and HP instruments. For the surface electrode ECG, the maternal observational recording session was indicated using a computer key. Recording and play back of auditory stimuli were carried out using a computer running Pre-recorded Auditory Acoustic Testing (PRAAT version 4.4.27) freeware. For fetal testing, a 2 min voice stimulus was generated by each mother reading the story of Bambi (Salten & Chambers, 1928). The stimuli were augmented by a Yamaha Natural Sound Stereo Amplifier (Model AX-497) and delivered to the fetus through a loud speaker (Auratone 5C Super-Sound-Cube). Sound intensity was measured in air at the maternal abdomen above the site of the fetal head, at a distance of 10 cm from the speaker, using the A scale of a Bruel and Kjaer Sound Pressure Level meter (Model 2235). A Sony portable CD player and Panasonic noise cancellation headphones were used to play masking music to the mother. A Siemens Sonoline ultrasound scanner (Model SI-450) with a built-in video-recording system (Panasonic AG-6200) was used for obtaining an estimate of amniotic fluid. 2.3. Procedure Data were obtained from each mother-fetal pair in one laboratory visit lasting from 1 to 2 h. Upon arrival, the mother provided consent, demographic/obstetrical information (e.g., age, education, weight, height, gravida), had her blood pressure taken three times in the right upper arm while seated, and read the story of Bambi for 2 min while being audio recorded. Next, with the mother in a semi-recumbent position on a hospital bed, the abdomen was first washed with soap and water and then swabbed with alcohol. The five leads of the Monica AN24 were applied in accordance with the manufacturer’s instructions; one electrode was placed near the umbilicus, a second was placed above the symphysis pubis in a vertical line from the umbilicus, a third and a fourth were placed on opposite sides of the abdomen horizontal to the umbilicus; the ground was placed on the upper right side. The electrical signals were captured via Bluetooth technology on a computer running Monica customized software. For each lead, the computer program indicated that it was capturing the maternal and fetal √ cardiac signals by changing from a red X to a green check mark ( ). The program automatically began recording maternal and fetal ECG signals only when all leads were functional. Placement of leads was manipulated for a maximum of 1 h to try and obtain ECG signals. If signals could not be obtained within an hour, the study was abandoned. Subsequently, three adult electrodes were placed on the maternal chest in a Lead II configuration to obtain a second measure of maternal ECG and R–R interval variability, using custom software (Yamamoto & Hughson, 1991); the wand of a Hewlett-Packard cardiotocograph was placed on the maternal abdomen to obtain a second measure of fetal HR. When the three devices were in place and operational, the built-in event markers of the Monica AN24 and HP devices and a computer key for the surface electrode ECG were pressed simultaneously to synchronize the signals. The maternal and fetal HR were recorded for a minimum of 20 consecutive min while the mother was at rest in a semi-recumbent position wedged left, with no talking and the room
B.S. Kisilevsky, C.A. Brown / Infant Behavior & Development 42 (2016) 142–151
145
dimly lit. At the end of recording, the Spacelab maternal ECG was stopped. For 19 fetuses, the maternal–fetal ECG signals and fetal CTG heart rate continued to be recorded continuously during a 6 min fetal voice processing procedure including: a 2 min pre-voice, no-sound baseline period, a 2 min mother’s voice period, and a 2 min post-voice, no-sound period. The voice recordings were delivered at an average of 95 dB A (estimated uterine attenuation 35–40 dB) through a loudspeaker placed 10 cm above the maternal abdomen. The ambient noise level with all equipment running averaged 50 dB A. During the auditory procedure, mothers wore headphones and listened to soft jazz or classical guitar music to mask the sounds being played to the fetus. Finally, an estimate of amniotic fluid volume was obtained by measuring the largest visible pocket of fluid in two dimensions in each of the four maternal abdominal quadrants. An amniotic fluid index (AFI) was calculated by summing the four vertical dimensions. 2.4. Data analyses IBM SPSS 22 was used for data analyses. For all repeated measures analyses, Greenhouse–Geisser p values are reported. To compare average maternal HR and HR variability measures collected using two ECG instruments, the onset of recording while the mother was at rest was matched, the 1st 10 min served as stabilization, and the 2nd 10 min of recording was used in analyses To compare the Monica AN24 ECG and Hewlett-Packard CTG collected measurements of spontaneous fetal HR, onset of recordings was matched and 20 consecutive, simultaneous min of data from the two instruments was analyzed. For the comparison of voice elicited fetal HR, the recordings were matched at voice onset. The Monica AN24 ECG beat-to-beat data were extracted in bpm in 0.25 s intervals and averaged over each 4 consecutive 0.25 s, resulting in a HR for each second of the 20 min recording and for each second during the 2 min of the playing of the voice, similar to that extracted from the Hewlett-Packard CTG. 3. Results Demographic and obstetrical characteristics of the sample are displayed in Table 1. The good health of the fetuses can be seen in the delivery outcome variables (e.g., mean birth weight = 3638 g; mean 5 min Apgar score = 8.8). 3.1. Comparison of ECG and CTG measures of spontaneous fetal HR changes The association of ECG and CTG HR in bpm at each second for the 1200 s of recording while the mother was at rest was determined for the 39 fetuses. Analyses revealed 34 positive, 4 negative, and 1 no correlation. The correlations and p values for each individual fetus are displayed in Table 2. Of the positive correlations, 26 (67%) were high ranging from .73 to .97; 4 (10%) were moderate ranging from .57 to .65; and 4 (10%) were low ranging from .08 to .26. The negative correlations were all low, ranging from −.13 to −.40. Comparing the average spontaneous fetal HR obtained using the ECG with the CTG at 1 min intervals for the 20 continuous min aggregated over fetuses revealed high correlations between the two measurements, ranging from .93 to .99 (all p values = .000). The average fetal HR and correlations at each minute are shown in Table 3. Paired t-tests used to determine differences between instruments in the average fetal HR at each min showed no differences between any of the minutes (all p values >.05). Mean differences at each minute for the aggregate data varied from 0.0 to 0.8 bpm. Table 1 Means (±SD) and frequencies (%) for maternal and fetal demographic and obstetrical characteristics of the sample (n = 39). Variables
Mean (±SD)
Maternal age (years) Gestational age at testing (weeks) BMI prepregnancy BMI at testing Systolic blood pressure (mmHg) Diastolic blood pressure (mmHg) Amniotic fluid index Gestational age at delivery Newborn Apgar at 5 min Newborn birth weight (g)
28 (6) 37.8 (0.7) 24.5 (4.7) 30.8 (6.1) 120 (10) 75 (8) 118 (22) 39.6 (1.3) 8.8 (0.6) 3638 (453) Frequencies (%)
Maternal educational level Less than secondary graduation Secondary graduation Post-secondary Primiparous Caesarian section delivery Infant sex – male
1 (2.6%) 8 (20.5%) 30 (76.9%) 17 (44%) 7 (17.9%) 21 (53.8%)
146
B.S. Kisilevsky, C.A. Brown / Infant Behavior & Development 42 (2016) 142–151
Table 2 Correlations of fetal heart rate at each second over 20 consecutive minutes while the mother was at rest for measurements obtained using a Monica AN24 electrocardiograph with a Hewlett-Packard cardiotocograph for each participant separately. Fetus #
r
p
Fetus #
r
p
Fetus #
r
p
01 02 03 04 05 06 07 08 09 11 17 18 20
.80 .84 .85 .86 .87 −.40 .83 .85 .92 .57 .65 .77 −.37
.000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000
22 23 25 26 28 29 30 31 33 37 38 39 40
.86 .65 .93 .97 .94 .87 .91 .08 −.25 .93 .84 .59 .94
.000 .000 .000 .000 .000 .000 .000 .01 .000 .000 .000 .000 .000
41 42 43 46 47 48 49 50 52 53 54 55 56
.80 .78 .96 .02 .08 .93 .87 .27 .95 .73 −.13 .26 .93
.000 .000 .000 .52 .01 .000 .000 .000 .000 .000 .000 .000 .000
Repeated measures ANOVAs with 2 within factors (machine – 2; time – 1200 s) used to examine differences in average spontaneous fetal HR between the two machines over 20 min of consecutive recording revealed no effect of machine, F(1, 30) = 0.438, p = .51, 2 = .01, and no machine × time interaction, F(1199, 35,970) = 0.926, p = .52, 2 = .03. Repeating the analyses for each 5 consecutive minutes (1–5, 6–10, 11–15, 16–20 min) had the same results, no effects of machine or time × machine interactions. For clarity, Fig. 1 displays the average HR over seconds limited to the first consecutive 5 min period.
3.2. Comparison ECG and CTG measures of fetal HR during auditory stimulation During the 120 s of the playing of the mother’s voice, correlations between ECG and CTG measurements of fetal HR in bpm were moderate to high, with r values ranging from .54 to .99, for 18/19 of the fetuses and negative (r = −.45) for one fetus. The correlations and average maternal voice elicited fetal HR over the 2 min voice period are displayed in Table 4. When the associations were determined by averaging fetal HR over each 30 s epoch and the entire 120 s, all were high, r = .99 (all p values = .000). Fig. 2 displays the average HR over 120 s for the 19 fetuses during the playing of the mother’s voice for the ECG and CTG measurements shown separately. Over the 2 min voice period, the average fetal HR between devices varied from 0.0 to 0.6 bpm (paired t-tests, all p values >.05).
Table 3 Correlations of the average spontaneous fetal heart rate obtained over 60 s intervals for 20 continuous minutes while the mother was at rest for measurements obtained by a Monica AN24 electrocardiograph with the Hewlett-Packard cardiotocograph aggregated over participants. Average spontaneous fetal heart rate for each minute over 20 consecutive minutes Minute
ECG Mean (±SD)
CTG Mean (±SD)
r
p
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
139.0 (10.6) 139.5 (9.3) 140.6 (10.5) 140.3 (11.6) 140.1 (10.2) 141.9 (11.9) 142.1 (9.5) 139.5 (10.3) 140.9 (11.9) 140.4 (12.4) 140.1 (9.0) 140.9 (9.8) 141.3 (10.8) 143.1 (9.5) 140.8 (10.4) 140.8 (10.2) 141.3 (13.2) 141.4 (11.0) 140.8 (12.2) 138.6 (11.5)
139.0 (10.4) 139.6 (9.5) 140.2 (10.2) 140.9 (11.5) 139.8 (9.8) 141.1 (11.5) 142.0 (9.9) 138.9 (11.2) 141.4 (11.6) 140.6 (11.5) 140.4 (9.5) 141.5 (9.3) 141.4 (10.3) 143.4 (10.2) 140.9 (10.5) 141.3 (10.2) 142.0 (12.1) 141.6 (10.8) 140.6 (12.5) 138.3 (11.9)
.99 .99 .98 .98 .99 .96 .98 .97 .99 .95 .96 .97 .96 .93 .99 .99 .97 .96 .95 .99
.000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000
B.S. Kisilevsky, C.A. Brown / Infant Behavior & Development 42 (2016) 142–151
147
Fig. 1. Mean spontaneous fetal heart rate at each second over the first consecutive 5 min of recording while the mother was at rest using a Monica AN24 electrocardiograph and a Hewlett-Packard cardiotocograph shown separately.
Table 4 Correlation and average maternal voice elicited fetal heart rate obtained over 2 min during the playing of an audio-recording of the mother reading a story for a Monica AN24 electrocardiograph and a Hewlett-Packard cardiotocograph for each participant shown separately. Average maternal voice elicited fetal heart rate (bpm) over 2 min voice period Fetus #
ECG HR mean (±SD)
CTG HR mean (±SD)
r
p
01 02 06 07 09 11 17 18 20 22 23 25 26 28 29 30 37 38 39
126.6 (2.9) 137.6 (5.4) 149.2 (3.7) 135.8 (4.6) 135.5 (6.7) 143.6 (11.3) 130.3 (14.8) 155.6 (6.5) 142.5 (10.1) 140.6 (8.5) 144.4 (4.9) 125.3 (13.6) 139.3 (6.8) 135.6 (11.6) 157.2 (10.4) 143.3 (7.0) 149.1 (13.0) 151.8 (7.9) 164.6 (4.6)
126.0 (2.6) 137.5 (5.0) 149.8 (3.3) 135.6 (4.5) 135.6 (6.4) 143.2 (11.0) 130.8 (14.0) 155.6 (6.4) 144.9 (8.8) 140.5 (8.3) 144.8 (4.5) 125.2 (13.4) 140.1 (7.2) 135.7 (11.3) 156.8 (11.9) 140.6 (3.5) 148.9 (12.8) 152.1 (7.6) 165.0 (4.5)
.66 .65 .79 .62 .90 .92 .90 .91 .76 .86 .54 .96 .97 .90 .90 −.45 .99 .88 .74
.000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000
148
B.S. Kisilevsky, C.A. Brown / Infant Behavior & Development 42 (2016) 142–151
Fig. 2. Average fetal heart rate over 120 s during the playing of the mother’s voice for Monica AN24 electrocardiograph and Hewlett-Packard cardiotocograph measurements shown separately.
3.3. Comparison of Monica ECG and Spacelab cardiac monitor ECG measures of average maternal HR and HR variability Data were available from both ECG devices for 37 women. The correlation between the Spacelab and Monica AN24 ECG measures for the maternal HR averaged over the 2nd 10 min of recording was high, r = .997, p = 000. Percent of agreement also was high. As can be seen in Table 5, 81% of the pair averages differed by ≤1 bpm with a maximum difference score of 3 bpm.
Table 5 Individual average maternal heart rates (bpm) and the difference in heart rates over 10 min of rest for data collected by a surface electrocardiograph and a Monica AN24. Mother #
ECG (bpm)
Monica (bpm)
Difference
Mother #
ECG (bpm)
Monica (bpm)
Difference
02 03 04 05 06 07 08 09 11 17 18 20 22 23 25 26 27 28 30
62.5 78.0 82.2 71.9 82.8 89.6 81.6 72.6 77.2 82.0 75.6 97.5 99.3 78.3 86.4 79.3 88.0 77.8 87.6
59.5 77.0 81.4 72.0 80.8 90.4 81.1 71.6 76.4 81.3 74.5 97.2 98.5 78.0 86.1 79.1 87.5 77.0 87.3
3.0 1.0 0.8 −0.1 2.0 −0.8 0.5 1.0 0.8 0.7 1.1 0.3 0.8 0.3 0.3 0.2 0.5 0.8 0.3
31 33 37 38 39 40 41 42 43 47 48 49 50 52 53 54 55 56
102.4 92.5 76.8 80.1 79.8 80.8 76.5 85.2 80.8 81.5 70.9 94.8 91.1 59.2 84.9 96.9 82.2 107.7
102.2 92.0 76.5 78.6 79.1 80.0 76.3 84.5 79.0 80.9 70.6 94.2 92.7 59.1 86.7 96.4 81.8 106.8
0.2 0.5 0.3 1.5 0.7 0.8 0.2 0.7 1.8 0.6 0.3 0.6 −1.6 0.1 −1.8 0.5 0.4 0.9
B.S. Kisilevsky, C.A. Brown / Infant Behavior & Development 42 (2016) 142–151
149
Table 6 Correlations between heart rate variability measures using R–R intervals collected using a surface electrocardiograph and a Monica AN24 for 37 pregnant women during rest. Heart rate variability measure
Correlation r
p
Low frequency power High frequency power Total power PNS indicator SNS indicator
.891 .959 .920 .946 .968
.000 .000 .000 .000 .000
PNS, parasympathetic nervous system; SNS, sympathetic nervous system.
HR variability measures [low frequency (LF: 0.04–0.15 ms2 /Hz), high frequency (HF: 0.15–0.5 ms2 /Hz), and total power (ms2 /Hz), parasympathetic nervous system indicator (PNS = HF/total power), and sympathetic nervous system indicator (SNS = LF/HF)] were calculated based on the Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology (1996). Maternal R–R intervals extracted from the Monica ECG were used with custom software to calculate the measures. Maternal R–R interval data from the Spacelab cardiac monitor were analyzed with the Yamamoto and Hughson (1991) program. Correlations for each pair from the two ECG machines are displayed in Table 6; all correlations are high and significant, ranging from r = .89 to .97 (all p values = .000). 4. Discussion The purpose of the study was to compare at term: (1) two methods of measuring fetal HR (CTG vs. ECG) and (2) two ECG methods of measuring maternal HR variability. The results showed good reliability of CTG with ECG measures of spontaneous and auditory elicited fetal HR when comparisons were made between individual fetal data over each second and very high reliability when data were aggregated over longer periods of time (30–1200 s epochs) and/or averaged over fetuses. During maternal rest, comparisons of individual fetal spontaneous HR changes at each second over 20 min ranged from r = .57–.97 (Table 2) for 77% of fetuses; when HR for each min of the 20 min recording was averaged over participants, correlations were all high, r = .93–.97 (Table 3). When HR change was elicited during 2 min of playing the mother’s voice, individual 1 s measurements again showed moderate to high (r = .54–.99) associations for 95% of fetuses; when the data for all fetuses were averaged either over the 2 min of the voice recording or in four, 30 s epochs, correlations were all high (r = .99). Figs. 1 and 2 which display the average spontaneous fetal HR over 5 min of rest and during the playing of the mother’s voice, respectively, demonstrate that the two methods yield essentially similar patterns of results over time. These findings indicate that CTG used in fetal neurobiological studies which typically aggregate data to characterize development are a good approximation to an ECG, the ‘gold’ standard for cardiac measurements. They extend the findings of earlier reports of the concordance of fetal ECG vs. CTG measures (Ibrahimy et al., 2002) using spontaneous HR changes over longer recording times as well as auditory induced HR changes. They support the effectiveness of earlier studies using CTG before birth and ECG after birth which showed a relationship between spontaneous HR (DiPietro et al., 2000, 2007) and a continuity between sensory elicited cardiac responses to similar kinds of stimulus materials (Kisilevsky & Muir, 1991). The greater variability observed in correlations computed for 1 s intervals for individual fetuses is most likely the result of differences in the methods of heart rate data collection (ECG using electrical signals, CTG using muscular contraction) and the algorithms employed by the two technologies (for mathematical descriptions of algorithms see for example, Ibrahimy et al., 2002 and Goncalves, Bernardes, Rocha, & Ayers-de-Campos, 2007). In general, the Monica AN24 (Reference Operating Manual, 2008) provides HR data based upon the general noise in the recorded data and characteristic shape of the ECG complex. If there is a question of accuracy of either the fetal or maternal HR, it is not made available (i.e., data are not ‘filled in’). In contrast, the HP cardiotocograph which uses autocorrelation techniques and rule-based methods to calculate the fetal heart rate will ‘fill in’ data. It could be that rule-based methods account for the negative correlation (r = −.45) observed here when the average maternal voice elicited fetal heart rate over the 2 min voice period was compared between the two instruments. As shown in Fig. 3, the elicited fetal HR captured using an ECG vs. CTG showed a much greater magnitude (>20 bpm vs.
Contents lists available at ScienceDirect
Infant Behavior and Development
Comparison of fetal and maternal heart rate measures using electrocardiographic and cardiotocographic methods Barbara S. Kisilevsky a,b,∗ , C. Ann Brown a a b
School of Nursing, Queen’s University, Kingston, ON, Canada Department of Obstetrics & Gynaecology, Kingston General Hospital, Kingston, ON, Canada
a r t i c l e
i n f o
Article history: Received 9 November 2015 Received in revised form 23 December 2015 Accepted 27 December 2015 Keywords: Fetus Pregnant women Heart rate ECG Cardiotocograph Voice Reliability
a b s t r a c t Purpose: To determine the reliability at term of: (1) two methods of measuring fetal heart rate (HR), electrocardiographic (ECG, the ‘gold standard’) and cardiotocographic (CTG) and (2) two ECG methods of measuring maternal HR variability over relatively brief periods of time (s–min). Methods: During 20 min of rest (N = 39) and during 2 min of auditory stimulation (mother’s recorded voice, n = 19), fetal HR data were collected using an ECG (Monica AN24) and a Hewlett-Packard Model 1351A CTG. Simultaneously, maternal HR data (n = 37) were collected using the same ECG device (Monica AN24) and a second stand-alone cardiac monitor (Spacelab 514T cardiac monitor with a QRS detector). Results: During 20 min of maternal rest, correlations of individual fetal CTG with ECG measures of HR at each second were moderate to high (r = .57–.97) for 77% of fetuses. Correlations of HR averaged over fetuses and over each of the 20 min were high (r = .93–.97); fetal HR averaged over 20 min varied between devices from 0.0 to 0.8 bpm. During 2 min of maternal voice presentation, correlations of fetal HR over each second were moderate to high (r = .54–.99) for 95% of fetuses and high (all rs = .99) when averaged across fetuses in 30 s or 2 min epochs. Average fetal HR between devices over the 2 min voice varied from 0.0 to 0.6 bpm. Correlations and/or % agreement between the two ECG methods of measuring maternal HR were high. Average maternal HR over 10 min showed 81% of pairs with a difference of ≤1 bpm; correlations for HR variability measures varied from r = .89 to .97. Conclusions: Good reliability was demonstrated between individual spontaneous and auditory induced fetal CTG and ECG with high correlations when HR data were averaged over fetuses or in 30–120 s epochs. High reliability of maternal HR measures was obtained using two ECG devices. © 2016 Elsevier Inc. All rights reserved.
1. Introduction Heart rate measures are one of the few readily available, non-invasive measures for the study of human fetal neurological and behavioral development. For research with infants, children, and adults, cardiac measures including heart rate [HR; in beats per minute (bpm)] and HR variability (i.e., measures of the balance between the sympathetic and parasympathetic branches of the autonomic nervous system influencing heart rate) are extracted from the R–R interval (the time, in ms,
∗ Corresponding author at: School of Nursing, Queen’s University, Kingston, ON K7L 3N6, Canada. E-mail addresses: [email protected] (B.S. Kisilevsky), [email protected] (C.A. Brown). http://dx.doi.org/10.1016/j.infbeh.2015.12.004 0163-6383/© 2016 Elsevier Inc. All rights reserved.
B.S. Kisilevsky, C.A. Brown / Infant Behavior & Development 42 (2016) 142–151
143
between two successive ventricular contractions) of an electrocardiogram (ECG) using an electrocardiographic instrument. Although an ECG is considered the ‘gold standard’ for heart rate measures, for fetal research during gestation when the membranes are intact, often cardiac measures are obtained using a cardiotocograph (CTG) which measures muscular contraction of the heart and employs autocorrelation techniques to compare successive Doppler ultrasound heart beat signals and provide a measure corresponding to the R–R interval of an ECG (Cesarelli, Romano, & Bifulco, 2009). ECG technology has not been readily available for fetal studies because of difficulties inherent in reliably separating the maternal and fetal electrocardiograms (for a technical discussion of fetal HR signal processing issues such as relatively low signal-to-noise ratio and limited knowledge of cardiac function and development see, e.g., Andreotti et al., 2014; Clifford, Silva, Behar, & Moody, 2014; Sameni & Clifford, 2010). As well, from about 24 to 34 weeks gestational age (GA), the vernix caseosa on the fetal skin impedes conduction of fetal electric signals to the maternal surface (e.g., Oostendorp, Oosterom, & Jongsma, 1989). Although the use of in-house ECG technology (e.g., Groome et al., 2000) and commercial equipment (e.g., FEMO, David, Hirsch, Karin, Toledo, & Akselrod, 2007) occasionally have been reported in fetal psychobiological research, little is known regarding the relationship between fetal HR measures collected using CTG vs. ECG (the gold standard) in non-laboring women. See Footnote #1. In uneventful pregnancies during the third trimester of pregnancy, only one report (Ibrahimy, Mohd, Zahedi, & Tsuruoka, 2002) could be found of a direct comparison between ECG and CTG measures. Ibrahimy et al. simultaneously collected FHR data for 10 min during maternal rest from 5 fetuses between 35 and 40 weeks gestation using three ECG electrodes (above and below the maternal umbilicus, right wrist) and a commercial Doppler fetal monitor (IFM-500, BioSys Co., Ltd). The authors reported that the two fetal HR measurements varied ±5 beats per minute (bpm) for 84% of recording. The purpose of the present study was to replicate and extend this earlier work using a larger sample size (n = 39) at term and comparing heart rate measures collected using two commercial fetal HR machines (Monica AN24 ECG vs. Hewlett-Packard CTG) during maternal rest using a longer rest period (20 min) as well as during fetal auditory stimulation which could increase fetal movement activity and interfere with signal detection. In addition, given that the Monica AN24 simultaneously captures and separates both the maternal and fetal ECG, a second maternal ECG was obtained using a stand-alone standard Spacelab 514T cardiac monitor (with a QRS detector) to compare two maternal ECG measures (Monica AN24 ECG vs. Spacelab ECG). Knowledge and understanding of the reliability between fetal ECG and CTG measures collected for relatively brief periods of time (seconds or minutes) are especially relevant for the interpretation of research findings in studies of fetal neurobiological development because the equipment was designed and is most often used and tested for clinical monitoring of heart rate over extended periods of time (hours).1 Studies employing the two different technologies, CTG before birth with ECG measures after birth, have shown a relationship between spontaneous fetal HR and HR variability from 24 weeks GA through the second year of life (DiPietro, Costigan, Pressman, & Doussard-Roosevelt, 2000; DiPietro, Bornstein, Hahn, Costigan, & Achy-Brou, 2007). Moreover, a continuity between term fetal and newborn elicited cardiac responses to similar kinds of stimulus materials has been demonstrated (Kisilevsky & Muir, 1991). Such findings suggest that the two fetal measures yield similar results and that such comparisons are reliable. See example, Footnote #1. 2. Material and methods 2.1. Participants A total of 39 maternal–fetal pairs at term provided the data for this study. Participants were recruited from antenatal clinics at a community teaching hospital in southern Canada. Inclusion criteria were a maternal age of at least 18 years and a low-risk, uneventful, singleton pregnancy at term with delivery of a healthy newborn. Exclusion criteria were any pharmacologically treated co-morbid conditions (e.g., diabetes, hypertension, thyroid disease, depression). Gestational age was calculated from the first day of the last menstrual period if periods were reliable (accuracy rate 75–85%) or from early ultrasound (SD ± 1 week). Data from an additional 15 fetuses were not included because the fetal HR could not be reliably captured by the ECG (n = 14) or CTG (n = 1) technology. Testing of fetuses and newborns was conducted in the Maternal-Fetal-Newborn Studies Laboratory located adjacent to the obstetrical outpatient and inpatient services of the hospital. Sex of the infant was determined at birth. Information on race and socio-economic status are not routinely collected in the Canadian health care system and were not collected for this study. The study was carried out according to ethics approval from the University and Affiliated Teaching Hospitals Research Ethics Board. Women provided informed, voluntary, written consent prior to participation. 2.2. Equipment/stimuli Maternal blood pressure (BP) was obtained using a Lifesource One Step Auto-Inflation Blood Pressure Monitor Model UA-767. The User Manual reports a SD = 3 mmHg.
1 There is a substantive literature examining fetal CTG and/or ECG [obtained from the maternal abdomen (e.g., Reinhard et al., 2013) or fetal scalp electrode (e.g., Clifford, Sameni, Ward, Robinson, & Wolfberg, 2011)] during labor. The studies are conducted for clinical purposes (e.g., see reviews by Amer-Wahlin & Kwee, 2015; Neilson, 2013; commentary by Sholapurkar, 2014) rather than to elucidate normal fetal psychobiological development. They are carried out during a time of substantial physiological stress for the mother and fetus rather than under resting conditions and are not reviewed here.
144
B.S. Kisilevsky, C.A. Brown / Infant Behavior & Development 42 (2016) 142–151
Maternal and fetal ECG signals were captured using a Monica AN24 (Monica Healthcare, Nottingham, UK) wireless fetal monitor. The Monica AN24 is a relatively small (fits in the palm of the hand), wearable, battery-powered device, attached via a lead assembly and five ECG electrodes (4 electrodes, 1 ground) placed on the maternal abdomen. It records electrical signals at the electrodes which are converted into digital format, processed in real-time, and can be downloaded to a computer. The device is a commercially available instrument that provides maternal and fetal HR in beats per minute (bpm) and R–R intervals in ms in singleton pregnancies from 20 weeks gestational age (GA) until the onset of labor for recording sessions lasting up to 16 h. Manufacturer’s software (Monica DK version 1.6) for data acquisition (Bluetooth) and computer storage was used to extract maternal and fetal beat-to-beat heart rate, providing HR in bpm in 0.25 s or 2 s epochs as well as beat-by-beat R–R intervals (duration between consecutive ventricular depolarizations measured by the R waves of the QRS complexes) in 0.25 s epochs or ms. A second continuous recording of fetal HR was obtained using a Hewlett-Packard (HP) cardiotocograph (Model 1351A). The HP cardiotocograph is a ‘desk-top’, electrically powered device which employs Doppler ultrasound, emitted and captured through a circular paddle placed on the maternal abdomen to detect contractions of the fetal heart muscle while the mother is seated or supine. The signals are converted to fetal HR in bpm using autocorrelation techniques with HR in bpm displayed on the instrument screen and a hard copy paper graph of HR over seconds printed. The HP cardiotocograph can be used for single or twin (two paddles) fetuses throughout pregnancy and labor. It was available in North America at the time of this study through the independent measurement company, Agilent Technologies (San Jose, CA). HR in bpm based on cardiac muscular contraction was obtained directly from the device using custom software which captured and averaged 4 readings per second; the average at each second was stored in a computer text file. Both the Monica AN24 and the HP cardiotocograph were designed and are used primarily for clinical assessment of fetal well-being. A second maternal ECG was obtained using a Spacelab 514T cardiac monitor (Squibb Vitatek Inc., Hillsboro) with a QRS detector and three (2 electrodes, 1 ground) latex-free standard surface ECG electrodes in bipolar Lead II configuration on the maternal chest; beat-by-beat R–R intervals were recorded continuously. An analog-digital converter (DAS-16, Metrabyte Corp., Multitest Electronics, Scarborough, ON, Canada) digitized the analog R–R interval output, providing an R–R interval accuracy of 1 ms through a sampling rate of 1000 Hz (Yamamoto & Hughson, 1991). The digital R–R interval output was analyzed using a custom computer software program for spectral analysis of R–R interval variability (Yamamoto & Hughson, 1991). Simultaneous onset and offset of observational (mother at rest) and experimental (auditory stimulus) recording sessions among the three devices were indicated manually using the built-in event markers of the Monica AN24 and HP instruments. For the surface electrode ECG, the maternal observational recording session was indicated using a computer key. Recording and play back of auditory stimuli were carried out using a computer running Pre-recorded Auditory Acoustic Testing (PRAAT version 4.4.27) freeware. For fetal testing, a 2 min voice stimulus was generated by each mother reading the story of Bambi (Salten & Chambers, 1928). The stimuli were augmented by a Yamaha Natural Sound Stereo Amplifier (Model AX-497) and delivered to the fetus through a loud speaker (Auratone 5C Super-Sound-Cube). Sound intensity was measured in air at the maternal abdomen above the site of the fetal head, at a distance of 10 cm from the speaker, using the A scale of a Bruel and Kjaer Sound Pressure Level meter (Model 2235). A Sony portable CD player and Panasonic noise cancellation headphones were used to play masking music to the mother. A Siemens Sonoline ultrasound scanner (Model SI-450) with a built-in video-recording system (Panasonic AG-6200) was used for obtaining an estimate of amniotic fluid. 2.3. Procedure Data were obtained from each mother-fetal pair in one laboratory visit lasting from 1 to 2 h. Upon arrival, the mother provided consent, demographic/obstetrical information (e.g., age, education, weight, height, gravida), had her blood pressure taken three times in the right upper arm while seated, and read the story of Bambi for 2 min while being audio recorded. Next, with the mother in a semi-recumbent position on a hospital bed, the abdomen was first washed with soap and water and then swabbed with alcohol. The five leads of the Monica AN24 were applied in accordance with the manufacturer’s instructions; one electrode was placed near the umbilicus, a second was placed above the symphysis pubis in a vertical line from the umbilicus, a third and a fourth were placed on opposite sides of the abdomen horizontal to the umbilicus; the ground was placed on the upper right side. The electrical signals were captured via Bluetooth technology on a computer running Monica customized software. For each lead, the computer program indicated that it was capturing the maternal and fetal √ cardiac signals by changing from a red X to a green check mark ( ). The program automatically began recording maternal and fetal ECG signals only when all leads were functional. Placement of leads was manipulated for a maximum of 1 h to try and obtain ECG signals. If signals could not be obtained within an hour, the study was abandoned. Subsequently, three adult electrodes were placed on the maternal chest in a Lead II configuration to obtain a second measure of maternal ECG and R–R interval variability, using custom software (Yamamoto & Hughson, 1991); the wand of a Hewlett-Packard cardiotocograph was placed on the maternal abdomen to obtain a second measure of fetal HR. When the three devices were in place and operational, the built-in event markers of the Monica AN24 and HP devices and a computer key for the surface electrode ECG were pressed simultaneously to synchronize the signals. The maternal and fetal HR were recorded for a minimum of 20 consecutive min while the mother was at rest in a semi-recumbent position wedged left, with no talking and the room
B.S. Kisilevsky, C.A. Brown / Infant Behavior & Development 42 (2016) 142–151
145
dimly lit. At the end of recording, the Spacelab maternal ECG was stopped. For 19 fetuses, the maternal–fetal ECG signals and fetal CTG heart rate continued to be recorded continuously during a 6 min fetal voice processing procedure including: a 2 min pre-voice, no-sound baseline period, a 2 min mother’s voice period, and a 2 min post-voice, no-sound period. The voice recordings were delivered at an average of 95 dB A (estimated uterine attenuation 35–40 dB) through a loudspeaker placed 10 cm above the maternal abdomen. The ambient noise level with all equipment running averaged 50 dB A. During the auditory procedure, mothers wore headphones and listened to soft jazz or classical guitar music to mask the sounds being played to the fetus. Finally, an estimate of amniotic fluid volume was obtained by measuring the largest visible pocket of fluid in two dimensions in each of the four maternal abdominal quadrants. An amniotic fluid index (AFI) was calculated by summing the four vertical dimensions. 2.4. Data analyses IBM SPSS 22 was used for data analyses. For all repeated measures analyses, Greenhouse–Geisser p values are reported. To compare average maternal HR and HR variability measures collected using two ECG instruments, the onset of recording while the mother was at rest was matched, the 1st 10 min served as stabilization, and the 2nd 10 min of recording was used in analyses To compare the Monica AN24 ECG and Hewlett-Packard CTG collected measurements of spontaneous fetal HR, onset of recordings was matched and 20 consecutive, simultaneous min of data from the two instruments was analyzed. For the comparison of voice elicited fetal HR, the recordings were matched at voice onset. The Monica AN24 ECG beat-to-beat data were extracted in bpm in 0.25 s intervals and averaged over each 4 consecutive 0.25 s, resulting in a HR for each second of the 20 min recording and for each second during the 2 min of the playing of the voice, similar to that extracted from the Hewlett-Packard CTG. 3. Results Demographic and obstetrical characteristics of the sample are displayed in Table 1. The good health of the fetuses can be seen in the delivery outcome variables (e.g., mean birth weight = 3638 g; mean 5 min Apgar score = 8.8). 3.1. Comparison of ECG and CTG measures of spontaneous fetal HR changes The association of ECG and CTG HR in bpm at each second for the 1200 s of recording while the mother was at rest was determined for the 39 fetuses. Analyses revealed 34 positive, 4 negative, and 1 no correlation. The correlations and p values for each individual fetus are displayed in Table 2. Of the positive correlations, 26 (67%) were high ranging from .73 to .97; 4 (10%) were moderate ranging from .57 to .65; and 4 (10%) were low ranging from .08 to .26. The negative correlations were all low, ranging from −.13 to −.40. Comparing the average spontaneous fetal HR obtained using the ECG with the CTG at 1 min intervals for the 20 continuous min aggregated over fetuses revealed high correlations between the two measurements, ranging from .93 to .99 (all p values = .000). The average fetal HR and correlations at each minute are shown in Table 3. Paired t-tests used to determine differences between instruments in the average fetal HR at each min showed no differences between any of the minutes (all p values >.05). Mean differences at each minute for the aggregate data varied from 0.0 to 0.8 bpm. Table 1 Means (±SD) and frequencies (%) for maternal and fetal demographic and obstetrical characteristics of the sample (n = 39). Variables
Mean (±SD)
Maternal age (years) Gestational age at testing (weeks) BMI prepregnancy BMI at testing Systolic blood pressure (mmHg) Diastolic blood pressure (mmHg) Amniotic fluid index Gestational age at delivery Newborn Apgar at 5 min Newborn birth weight (g)
28 (6) 37.8 (0.7) 24.5 (4.7) 30.8 (6.1) 120 (10) 75 (8) 118 (22) 39.6 (1.3) 8.8 (0.6) 3638 (453) Frequencies (%)
Maternal educational level Less than secondary graduation Secondary graduation Post-secondary Primiparous Caesarian section delivery Infant sex – male
1 (2.6%) 8 (20.5%) 30 (76.9%) 17 (44%) 7 (17.9%) 21 (53.8%)
146
B.S. Kisilevsky, C.A. Brown / Infant Behavior & Development 42 (2016) 142–151
Table 2 Correlations of fetal heart rate at each second over 20 consecutive minutes while the mother was at rest for measurements obtained using a Monica AN24 electrocardiograph with a Hewlett-Packard cardiotocograph for each participant separately. Fetus #
r
p
Fetus #
r
p
Fetus #
r
p
01 02 03 04 05 06 07 08 09 11 17 18 20
.80 .84 .85 .86 .87 −.40 .83 .85 .92 .57 .65 .77 −.37
.000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000
22 23 25 26 28 29 30 31 33 37 38 39 40
.86 .65 .93 .97 .94 .87 .91 .08 −.25 .93 .84 .59 .94
.000 .000 .000 .000 .000 .000 .000 .01 .000 .000 .000 .000 .000
41 42 43 46 47 48 49 50 52 53 54 55 56
.80 .78 .96 .02 .08 .93 .87 .27 .95 .73 −.13 .26 .93
.000 .000 .000 .52 .01 .000 .000 .000 .000 .000 .000 .000 .000
Repeated measures ANOVAs with 2 within factors (machine – 2; time – 1200 s) used to examine differences in average spontaneous fetal HR between the two machines over 20 min of consecutive recording revealed no effect of machine, F(1, 30) = 0.438, p = .51, 2 = .01, and no machine × time interaction, F(1199, 35,970) = 0.926, p = .52, 2 = .03. Repeating the analyses for each 5 consecutive minutes (1–5, 6–10, 11–15, 16–20 min) had the same results, no effects of machine or time × machine interactions. For clarity, Fig. 1 displays the average HR over seconds limited to the first consecutive 5 min period.
3.2. Comparison ECG and CTG measures of fetal HR during auditory stimulation During the 120 s of the playing of the mother’s voice, correlations between ECG and CTG measurements of fetal HR in bpm were moderate to high, with r values ranging from .54 to .99, for 18/19 of the fetuses and negative (r = −.45) for one fetus. The correlations and average maternal voice elicited fetal HR over the 2 min voice period are displayed in Table 4. When the associations were determined by averaging fetal HR over each 30 s epoch and the entire 120 s, all were high, r = .99 (all p values = .000). Fig. 2 displays the average HR over 120 s for the 19 fetuses during the playing of the mother’s voice for the ECG and CTG measurements shown separately. Over the 2 min voice period, the average fetal HR between devices varied from 0.0 to 0.6 bpm (paired t-tests, all p values >.05).
Table 3 Correlations of the average spontaneous fetal heart rate obtained over 60 s intervals for 20 continuous minutes while the mother was at rest for measurements obtained by a Monica AN24 electrocardiograph with the Hewlett-Packard cardiotocograph aggregated over participants. Average spontaneous fetal heart rate for each minute over 20 consecutive minutes Minute
ECG Mean (±SD)
CTG Mean (±SD)
r
p
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
139.0 (10.6) 139.5 (9.3) 140.6 (10.5) 140.3 (11.6) 140.1 (10.2) 141.9 (11.9) 142.1 (9.5) 139.5 (10.3) 140.9 (11.9) 140.4 (12.4) 140.1 (9.0) 140.9 (9.8) 141.3 (10.8) 143.1 (9.5) 140.8 (10.4) 140.8 (10.2) 141.3 (13.2) 141.4 (11.0) 140.8 (12.2) 138.6 (11.5)
139.0 (10.4) 139.6 (9.5) 140.2 (10.2) 140.9 (11.5) 139.8 (9.8) 141.1 (11.5) 142.0 (9.9) 138.9 (11.2) 141.4 (11.6) 140.6 (11.5) 140.4 (9.5) 141.5 (9.3) 141.4 (10.3) 143.4 (10.2) 140.9 (10.5) 141.3 (10.2) 142.0 (12.1) 141.6 (10.8) 140.6 (12.5) 138.3 (11.9)
.99 .99 .98 .98 .99 .96 .98 .97 .99 .95 .96 .97 .96 .93 .99 .99 .97 .96 .95 .99
.000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000
B.S. Kisilevsky, C.A. Brown / Infant Behavior & Development 42 (2016) 142–151
147
Fig. 1. Mean spontaneous fetal heart rate at each second over the first consecutive 5 min of recording while the mother was at rest using a Monica AN24 electrocardiograph and a Hewlett-Packard cardiotocograph shown separately.
Table 4 Correlation and average maternal voice elicited fetal heart rate obtained over 2 min during the playing of an audio-recording of the mother reading a story for a Monica AN24 electrocardiograph and a Hewlett-Packard cardiotocograph for each participant shown separately. Average maternal voice elicited fetal heart rate (bpm) over 2 min voice period Fetus #
ECG HR mean (±SD)
CTG HR mean (±SD)
r
p
01 02 06 07 09 11 17 18 20 22 23 25 26 28 29 30 37 38 39
126.6 (2.9) 137.6 (5.4) 149.2 (3.7) 135.8 (4.6) 135.5 (6.7) 143.6 (11.3) 130.3 (14.8) 155.6 (6.5) 142.5 (10.1) 140.6 (8.5) 144.4 (4.9) 125.3 (13.6) 139.3 (6.8) 135.6 (11.6) 157.2 (10.4) 143.3 (7.0) 149.1 (13.0) 151.8 (7.9) 164.6 (4.6)
126.0 (2.6) 137.5 (5.0) 149.8 (3.3) 135.6 (4.5) 135.6 (6.4) 143.2 (11.0) 130.8 (14.0) 155.6 (6.4) 144.9 (8.8) 140.5 (8.3) 144.8 (4.5) 125.2 (13.4) 140.1 (7.2) 135.7 (11.3) 156.8 (11.9) 140.6 (3.5) 148.9 (12.8) 152.1 (7.6) 165.0 (4.5)
.66 .65 .79 .62 .90 .92 .90 .91 .76 .86 .54 .96 .97 .90 .90 −.45 .99 .88 .74
.000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000
148
B.S. Kisilevsky, C.A. Brown / Infant Behavior & Development 42 (2016) 142–151
Fig. 2. Average fetal heart rate over 120 s during the playing of the mother’s voice for Monica AN24 electrocardiograph and Hewlett-Packard cardiotocograph measurements shown separately.
3.3. Comparison of Monica ECG and Spacelab cardiac monitor ECG measures of average maternal HR and HR variability Data were available from both ECG devices for 37 women. The correlation between the Spacelab and Monica AN24 ECG measures for the maternal HR averaged over the 2nd 10 min of recording was high, r = .997, p = 000. Percent of agreement also was high. As can be seen in Table 5, 81% of the pair averages differed by ≤1 bpm with a maximum difference score of 3 bpm.
Table 5 Individual average maternal heart rates (bpm) and the difference in heart rates over 10 min of rest for data collected by a surface electrocardiograph and a Monica AN24. Mother #
ECG (bpm)
Monica (bpm)
Difference
Mother #
ECG (bpm)
Monica (bpm)
Difference
02 03 04 05 06 07 08 09 11 17 18 20 22 23 25 26 27 28 30
62.5 78.0 82.2 71.9 82.8 89.6 81.6 72.6 77.2 82.0 75.6 97.5 99.3 78.3 86.4 79.3 88.0 77.8 87.6
59.5 77.0 81.4 72.0 80.8 90.4 81.1 71.6 76.4 81.3 74.5 97.2 98.5 78.0 86.1 79.1 87.5 77.0 87.3
3.0 1.0 0.8 −0.1 2.0 −0.8 0.5 1.0 0.8 0.7 1.1 0.3 0.8 0.3 0.3 0.2 0.5 0.8 0.3
31 33 37 38 39 40 41 42 43 47 48 49 50 52 53 54 55 56
102.4 92.5 76.8 80.1 79.8 80.8 76.5 85.2 80.8 81.5 70.9 94.8 91.1 59.2 84.9 96.9 82.2 107.7
102.2 92.0 76.5 78.6 79.1 80.0 76.3 84.5 79.0 80.9 70.6 94.2 92.7 59.1 86.7 96.4 81.8 106.8
0.2 0.5 0.3 1.5 0.7 0.8 0.2 0.7 1.8 0.6 0.3 0.6 −1.6 0.1 −1.8 0.5 0.4 0.9
B.S. Kisilevsky, C.A. Brown / Infant Behavior & Development 42 (2016) 142–151
149
Table 6 Correlations between heart rate variability measures using R–R intervals collected using a surface electrocardiograph and a Monica AN24 for 37 pregnant women during rest. Heart rate variability measure
Correlation r
p
Low frequency power High frequency power Total power PNS indicator SNS indicator
.891 .959 .920 .946 .968
.000 .000 .000 .000 .000
PNS, parasympathetic nervous system; SNS, sympathetic nervous system.
HR variability measures [low frequency (LF: 0.04–0.15 ms2 /Hz), high frequency (HF: 0.15–0.5 ms2 /Hz), and total power (ms2 /Hz), parasympathetic nervous system indicator (PNS = HF/total power), and sympathetic nervous system indicator (SNS = LF/HF)] were calculated based on the Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology (1996). Maternal R–R intervals extracted from the Monica ECG were used with custom software to calculate the measures. Maternal R–R interval data from the Spacelab cardiac monitor were analyzed with the Yamamoto and Hughson (1991) program. Correlations for each pair from the two ECG machines are displayed in Table 6; all correlations are high and significant, ranging from r = .89 to .97 (all p values = .000). 4. Discussion The purpose of the study was to compare at term: (1) two methods of measuring fetal HR (CTG vs. ECG) and (2) two ECG methods of measuring maternal HR variability. The results showed good reliability of CTG with ECG measures of spontaneous and auditory elicited fetal HR when comparisons were made between individual fetal data over each second and very high reliability when data were aggregated over longer periods of time (30–1200 s epochs) and/or averaged over fetuses. During maternal rest, comparisons of individual fetal spontaneous HR changes at each second over 20 min ranged from r = .57–.97 (Table 2) for 77% of fetuses; when HR for each min of the 20 min recording was averaged over participants, correlations were all high, r = .93–.97 (Table 3). When HR change was elicited during 2 min of playing the mother’s voice, individual 1 s measurements again showed moderate to high (r = .54–.99) associations for 95% of fetuses; when the data for all fetuses were averaged either over the 2 min of the voice recording or in four, 30 s epochs, correlations were all high (r = .99). Figs. 1 and 2 which display the average spontaneous fetal HR over 5 min of rest and during the playing of the mother’s voice, respectively, demonstrate that the two methods yield essentially similar patterns of results over time. These findings indicate that CTG used in fetal neurobiological studies which typically aggregate data to characterize development are a good approximation to an ECG, the ‘gold’ standard for cardiac measurements. They extend the findings of earlier reports of the concordance of fetal ECG vs. CTG measures (Ibrahimy et al., 2002) using spontaneous HR changes over longer recording times as well as auditory induced HR changes. They support the effectiveness of earlier studies using CTG before birth and ECG after birth which showed a relationship between spontaneous HR (DiPietro et al., 2000, 2007) and a continuity between sensory elicited cardiac responses to similar kinds of stimulus materials (Kisilevsky & Muir, 1991). The greater variability observed in correlations computed for 1 s intervals for individual fetuses is most likely the result of differences in the methods of heart rate data collection (ECG using electrical signals, CTG using muscular contraction) and the algorithms employed by the two technologies (for mathematical descriptions of algorithms see for example, Ibrahimy et al., 2002 and Goncalves, Bernardes, Rocha, & Ayers-de-Campos, 2007). In general, the Monica AN24 (Reference Operating Manual, 2008) provides HR data based upon the general noise in the recorded data and characteristic shape of the ECG complex. If there is a question of accuracy of either the fetal or maternal HR, it is not made available (i.e., data are not ‘filled in’). In contrast, the HP cardiotocograph which uses autocorrelation techniques and rule-based methods to calculate the fetal heart rate will ‘fill in’ data. It could be that rule-based methods account for the negative correlation (r = −.45) observed here when the average maternal voice elicited fetal heart rate over the 2 min voice period was compared between the two instruments. As shown in Fig. 3, the elicited fetal HR captured using an ECG vs. CTG showed a much greater magnitude (>20 bpm vs.
