Effects of low maternal heart rate on fetal growth and birthweight

Abstract

Objective:

To evaluate the association between birthweight and maternal heart rate (MHR) or heart rate variability (HRV) under resting conditions at 20–24 gestational weeks and 34 weeks or later (34+ weeks).

Methods:

Data were retrospectively reviewed from the Safe Passage Study, a prospective longitudinal cohort study of alcohol use in pregnancy and birth outcomes in Cape Town, South Africa, between August 2007 and January 2015. Using custom-designed software, MHR and indicators of HRV were obtained from the recorded maternal electrocardiograms and compared with birthweight and z-scores of birthweight adjusted for gestation and gender.

Results:

Data from 5655 women were included. MHR increased from 84.6 bpm at 20–24 weeks to 88.3 bpm at 34+ weeks. Increasing MHR from 70–80 to 80–90 and 90–100 bpm at 20–24–weeks was associated with increasing birthweight from 2940 to 2998 and 3032 g, respectively (P<0.05). MHR and HRV contributed 29% to the variability associated with birthweight, whereas maternal body mass index at recruitment contributed 44%. Similar associations were observed for MHR at 34+ weeks.

Conclusion:

The observed association of low maternal heart rate with birthweight might help to identify pregnancies at risk of poor fetal growth.

1. INTRODUCTION

Very few studies have examined the relationship between maternal heart rate (MHR) or its variability and fetal growth [1–3]. Heart rate variability (HRV), resulting from temporal variation in R-to-R intervals [4], has been employed as biomarker for autonomic nervous system function [5]. Reduced HRV has been linked to autonomic neuropathy in individuals with diabetes before clinical symptoms are apparent, anxiety, depression, asthma, and sudden infant death [4, 6–8]. Most studies, however, do not include measurements in pregnancy.

Little is known about effects of periods of low or high MHR in pregnancy [10, 11]. Reports on periodic changes in MHR before labor are even more limited. One study observed periodic changes in MHR in 10.9% of term pregnancies before elective cesarean section. However, most of the changes (84.6%) were associated with uterine activity [12].

Because MHR is potentially related to fetal growth [1–3], it is essential to know more about this parameter in pregnancy. In addition, the electrocardiogram (ECG) recordings collected for the Safe Passage Study (SPS), a prospective longitudinal cohort study of alcohol use in pregnancy and birth outcomes [13], revealed unusual MHR patterns for some women (Figure 1a, b). To determine whether MHR and HRV are linked to fetal development and growth, the aim of the present study was to evaluate the association between birthweight and MHR or HRV under resting conditions at 20–24 gestational weeks and 34 weeks or later (34+ weeks).

Figure 1.

Maternal heart rate patterns. a) The top recording reflects large maternal heart rate accelerations. The bottom recording shows uterine activity. b) The top recording reflects decelerations of the maternal heart rate. The lower recording shows uterine activity.

2. MATERIALS AND METHODS

In a retrospective analysis, data were reviewed from the SPS, a prospective longitudinal cohort study of pregnant women from a well-defined low-income residential area of Cape Town, South Africa, between August 1, 2007, and January 31, 2015. Approval for the study was obtained from the Health Research Ethics Committee of Stellenbosch University (N06/10/210), and all participants provided informed written consent.

In the SPS, pregnant women were recruited at the local prenatal clinic during their first prenatal visit. The inclusion criteria were at least 16 years of age, able to communicate in Afrikaans or English, gestational age of at least 6 weeks but not in labor, desire to continue at the same clinic with pregnancy, and a single fetus. Depending on the timing of enrolment, participants completed up to three prenatal assessments at 20–24 weeks, 28–32 weeks (for a subgroup of women), and 34 weeks or later (34+ weeks), in accordance with the aims of the SPS study. Gestational age at enrolment was determined by ultrasound performed at the first prenatal visit, because this has been found to be most appropriate for the study community [14].

An essential part of the SPS assessments was the non-invasive transabdominal collection of fetal and maternal electrocardiograms (ECGs) for at least 30 minutes under resting conditions in a quiet room. Participants were fetched from home by a driver from the SPS. After having a light snack, they were positioned in a 15° right or left lateral position, with a wedge placed under one of the buttocks to prevent supine hypotension. Detail on the placement of the electrodes on the abdominal wall has been described previously [12, 15]. The five electrodes were then connected to the fetal ECG monitor (AN24, Monica Health Care, Nottingham, United Kingdom).

Measurements, such as weight and length of the mother and fetus, respectively, were taken twice, and separately transcribed on a case report form; if the two values differed by more than 10%, a third measurement was taken. If the fetal position was inappropriate at the assessment, the biometry was delayed until a suitable position was observed. Data from the case report form were entered by trained data capturers on a web-based data capture sheet.

The present analysis of MHR was based on the digitized (300 samples/s) data available from SPS participants. The raw data from the ECG monitor were stored as proprietary formatted raw files. These files were extracted by using Monica DK version 2.2 software (Monica Health Care, Nottingham, United Kingdom), stored as beat-to-beat R-to-R intervals in a text file, and imported into MATLAB (MATLAB Release 2015a, The MathWorks, Natick, MA, USA). A feature extraction framework was written to characterize patterns. Many of the algorithms that were implemented in the feature extraction framework were adapted from the Dawes–Redman guidelines [16], one of the most widely used frameworks for quantifying heart rate features.

Pre-processing of the raw data was done in two steps [17]. First, the 4-Hz signals were processed to remove unwanted artefacts and unreliable sections; second, sparsity-based epoch rejection was applied to the resulting signals. The same general approach was followed for the beat-to-beat heart rate data.

The 4-Hz data were used to calculate mean and basal heart rates, lost and gained beats, and the mean minute range. The beat-to-beat data were used to calculate the mean R-to-R interval (MRR), standard deviation of normal-to-normal interval (SDNN), and root mean square of successive R-to-R interval differences (RMSSD).

To estimate increases or decreases in MHR, the baseline signal was first determined as described by Pardey et al. [16]. Gained or lost beats were computed as indices and quantified as accelerations in MHR or decelerations in MHR, respectively. The area below the periodic increases in MHR area, and above the periodic decreases in MHR area were used to calculate respective gained and lost beats to quantify the cumulative effect of periodic deviations above and below the baseline (Figure 2). To determine the number of gained and lost beats due to periodic increases and decreases in MHR, the integral of the difference between the MHR and baseline signal were calculated for all potential increases and decreases (Figure 2).

Figure 2.

Graph showing the calculation of gained and lost beats. Abbreviation: MHR, maternal heart rate.

A potential increase/decrease was defined as the wavelet between two adjacent crossing points between the MHR and baseline signal. The maximum difference and duration of all potential increases/decreases were then considered to classify the potential increases/decreases as either an increase/decrease or not. The duration of the recording was used to adjust the total gained and lost beats to a period of 60 minutes. Mean and basal MHR, MRR, SDNN, and RMSSD [18] were calculated from the cleaned MHR signal.

The different MHR features were examined with respect to fetal growth, which was assessed by birthweight and by z-scores of birthweight based on the Intergrowth–21st study [19]. Further analyses also quantified the relationships between MHR variables and maternal body mass index (BMI, calculated as weight in kilograms divided by the square of height in meters) because obesity might be associated with maternal tachycardia [20]. For these analyses, women with late spontaneous abortion, a fetus with congenital abnormalities, termination of pregnancy, ethnicity other than colored, more than one enrolment in the SPS, and unknown z-scores of the birthweight were excluded.

Statistical analyses were done with Statistica version 13 (Dell, Aliso Viejo, CA, USA). Continuous variables were compared by either analysis of variance (ANOVA) or repeated-measures ANOVA with the compound symmetry assumption on correlations between repetitions. Bonferroni corrections for multiple comparisons were used to identify significant differences among the means. Spearman correlations were used to measure associations among variables. Intra-class correlations for agreement and consistency were also reported for some variables. Maximum likelihood χ² tests were used to analyze associations between nominal variables. To explore the contribution of different variables to birthweight, principal component analysis of all variables was performed. A P value of less than 0.05 was taken to be statistically significant.

3. RESULTS

The SPS recruited 7060 pregnant women. Among these women, digitized data were available for 5655 participants for an investigation of MHR and birthweight. Prior history of heart disease and hyperthyroidism was found in 0.68% (30/4448) and 0.18% (8/4424) of participants, respectively. Most of the pregnancies were uncomplicated, but hypertension was recorded in 682 (12.1%), pre-eclampsia in 212 (3.9%), placental abruption in 49 (0.9%), and diabetes in 32 (0.6%) cases. There were 14 stillbirths.

The raw MHR data of 4421 participants were analyzed and compared with maternal age, gravidity, BMI, gestational age at delivery, birthweight, and z-score of birthweight (Table 1). As determined by the study protocol, more recordings were done at 20–24 and 34+ weeks than at 28–32 weeks. The mean duration of the recording for the three windows ranged from 0.60 to 0.92 hours (Table 1). The mean birthweight was 2983 g and the mean z-score was −0.4. The missed beat and arrhythmia rejection algorithm removed, on average, approximately 0 ± 10.9% of the points in the beat-to-beat data, and the sparsity based rejection removed 4.6 ± 13.1%.

Table 1.

Maternal and fetal characteristics, and gestational age at recordings.

Characteristics	No. of women	Mean ± SD	Median (range)
Maternal age, y	5655	24.5 ± 6.0	23 (16–45)
Gravidity	5639	2.1 ± 1.3	2 (1–10)
BMI	5505	25.5 ± 5.7	24.2 (14.4–55.9)
20–24-wk recording
GA, d	2991	157.4 ± 6.3	159 (140–167)
Duration of recording, h	2931	0.67 ± 0.15	0.62 (0.12–1.68)
28–32-wk recording
GA, d	743	204.6 ± 6.7	203 (196–223)
Duration of recording, h	696	0.60 ± 0.11	0.57 (0.31–1.17)
34–38 wk recording
GA, d	3792	244.15 ± 6.7	242 (238–276)
Duration of recording, h	3719	0.92 ± 0.10	0.90 (0.09–1.77)
GA at delivery, d	5655	271.7 ± 17.4	275 (146–313)
Birthweight, g	5588	2983.5 ± 598.6	3010 (190–5140)

Abbreviations: BMI, body mass index (calculated as weight in kilograms divided by the square of height in meters); GA, gestational age.

Table 2 summarizes the MHR and HRV characteristics for each of the recording windows. The mean basal MHR was 84.1 bpm at 20–24 weeks, but increased to 90.0 and 87.6 bpm at 28–32 and 34+ weeks, respectively. Indicators of maternal HRV changed little during the course of pregnancy.

Table 2.

Maternal heart rate variables at different gestational ages^a.

HR variable	20–24 wk (n=2931)	28–32 wk (n=696)	≥34 wk (n=3719)
Mean HR, bpm	84.6 ± 9.9	90.4 ± 10.3	88.3 ± 11.2
Basal HR, bpm	84.1 ± 10.1	90.0 ± 10.4	87.6 ± 11.3
SDNN, ms	47.6 ± 16.8	40.7 ± 15.2	48.9 ± 19.0
RMSSD, ms	31.4 ± 17.6	23.3 ± 15.0	26.8 ± 17.5
MRR, ms	716.9 ± 84.6	671.0 ± 78.5	688.3 ± 90.6
No. of gained beats	2120 ± 2644	1920 ± 3510	3374 ± 3896
No. of lost beats	307 ± 1073	369 ± 975	619 ± 3160

Abbreviations: HR, heart rate; MRR, mean R-to-R interval; RMSSD, root mean square of successive R-to-R interval differences; SDNN, standard deviation of normal to normal interval.

^aValues are given as mean ± SD.

Table 3 summarizes the correlations between MHR measures at 20–24 gestational weeks and birthweight, BMI, and z-score of birthweight. Birthweight correlated positively with MHR, but negatively with SDNN and RMSSD. BMI correlated positively with MHR, but negatively with MRR, SDNN, RMSSD, and maternal gained beats. Birthweight z-score correlated positively with MHR, and negatively with MRR, SDNN, and RMSSD. There was also a significant negative correlation between SDNN and basal MHR at 20–24 weeks (Spearman correlation, r=–0,68; P<0.001).

Table 3.

Correlation of MHR variables at 20−24 wk with birthweight, maternal BMI, and z-score of birthweight.

MHR variable	Birthweight		BMI		z-Score of birthweight
	r	P value	r	P value	r	P value
Mean heart rate	0.0330	>0.05	0.0769	<0.001	0.0697	<0.001
Basal heart rate	0.0355	>0.05	0.0834	<0.001	0.0717	<0.001
MRR	−0.0364	>0.05	−0.0846	<0.001	−0.0692	<0.001
SDNN	−0.0437	<0.02	−0.1496	<0.001	−0.0751	<0.001
RMSSD	−0.0627	<0.01	−0.1221	<0.001	−0.772	<0.001
Gained beats	−0.0382	<0.05	−0.1192	<0.001	−0.0346	>0.05
Lost beats	0.0020	>0.05	−0.0345	>0.05	−0.0058	>0.05

Abbreviations: BMI, body mass index (calculated as weight in kilograms divided by the square of height in meters); MHR, maternal heart rate; MRR, mean R-to-R interval; RMSSD, root mean square of successive R-to-R interval differences; SDNN, standard deviation of normal to normal interval.

Table 4 summarizes correlations between MHR measures at 34+ weeks and birthweight, BMI, and z-score of birthweight. Birthweight correlated positively with MHR, but negatively with MRR, SDNN, RMSSD, and maternal gained beats. BMI correlated positively with MHR, but negatively with indicators of HRV and maternal gained or lost beats. Birthweight z-score correlated positively with MHR, but negatively with indicators of HRV and gained beats.

Table 4.

Correlation of MHR variables at 34+ wk with birthweight, maternal BMI, and z-score of birthweight.

MHR variable	Birthweight		BMI		z-Score of birthweight
	r	P value	r	P value	r	P value
Mean heart rate	0.2526	<0.01	0.0457	<0.01	0.2138	<0.01
Basal heart rate	0.2540	<0.01	0.0564	<0.01	0.2159	<0.01
MRR	−0.2633	<0.01	−0.0573	<0.01	−0.2196	<0.01
SDNN	−0.1977	<0.01	−0.1977	<0.01	−0.1697	<0.01
RMSSD	−0.2213	<0.01	−0.0774	<0.001	−0.1870	<0.01
Gained beats	−0.0364	<0.05	−0.1701	<0.01	−0.0359	<0.05
Lost beats	−0.0122	>0.05	−0.0529	<0.01	−0.0027	>0.05

Abbreviations: BMI, body mass index (calculated as weight in kilograms divided by the square of height in meters); MHR, maternal heart rate; MRR, mean R-to-R interval; RMSSD, root mean square of successive R-to-R interval differences; SDNN, standard deviation of normal to normal interval.

Principal component analysis was used to assess the contributions of the variables measured at 20–24 weeks to birthweight (Figure S1). The horizontal axis or principal component 1 (PC1) explained 44% of the total variation, whereas PC2 explained 29% of the total variation. Thus, PC1 and PC2 explained 73% of the total variation of all variables. Notably, the z-score and BMI were almost parallel with the vertical axis, showing that they were highly correlated with PC2, whereas MHR and SDNN were almost parallel with the horizontal axis, showing that they were highly correlated with PC1. Both SDNN and MHR contributed very little to birthweight.

Increasing MHR at 20–24 weeks from 70–80 bpm to 80–90 and 90–100 bpm was associated with a significant increase in birthweight (Figure S2). Increasing MHR from 70–80 bpm to 80–90 and 90–100 bpm at 34+ weeks was also associated with a significant increase in birthweight (Figure S3).

4. DISCUSSION

The present study found that MHR at 20–24 and 34+ gestational weeks are significantly associated with neonatal birthweight, confirming the findings of Everett et al. [3]. This convergence of results was apparent even though the methods of the two studies differed. Everett et al. [3] studied singleton pregnancies at risk of pre-eclampsia and determined MHR at 26–27 weeks, whereas the present study included all singleton pregnancies from a population with poor socioeconomic conditions and a high prevalence of alcohol consumption and cigarette smoking during pregnancy [21]. In addition, Everett et al. [3] used a z-score developed for their maternity unit, whereas the present study used the z-score recommended by the Intergrowth-21st project for comparison in multi-ethnic populations [19]. The present findings also correspond with those of another study that found that depressed women who gave birth at term had significantly higher heart rates and delivered heavier newborns as compared with pregnant women who were not depressed [22].

A possible reason for lower neonatal birthweight in pregnancies with lower MHR is that low MHR is a marker of lower cardiac output resulting from poor maternal hemodynamic adaptation, as seen in cases of impaired fetal growth [1,23,24]. The primary cause of low cardiac output is probably poor placental implantation or inadequate trophoblast invasion, leading to impaired vasodilation and reduced plasma volume expansion, which is usually responsible for the physiologic increase in MHR and increased cardiac output. Poor placentation is therefore associated with low MHR and lower birthweight. The present study also observed an association between MHR and BMI, consistent with the findings of Carson et al. [20], who reported tachycardia in 29% of obese pregnant women.

Heart rate variability is regulated by a complex interplay between sympathetic and parasympathetic branches of the autonomic nervous system and has been assessed by many methodologies [4]. Although lower HRV seems to be associated with poor outcome in many medical conditions, little is known about it in pregnancy. In the present study, there was an inverse correlation between parameters of HRV and birthweight, but also significant negative correlations between heart rate and the three indicators of variability. The negative correlation between birthweight and HRV might therefore be explained by the increase in MHR because MHR and HRV are negatively correlated. Beyond the contribution of MHR to birth weight, HRV contributed very little to birthweight.

Other conditions that might be associated with changes in MHR are the position during recording and uterine activity. In a prospective observational study of the effect of position and uterine activity before elective cesarean section, Ibrahim et al. [12] observed periodic changes in MHR for 13 (10.9%) women. Most of these changes (84.6%) were associated with uterine activity and not maternal position. They concluded that, for a subgroup of pregnant women at term, uterine activity was associated with periodic changes in MHR. They postulated that displacement of blood from the uterus and choriodecidual space into the venous circulation increased the preload to the heart. This in turn increased myocardial contractility, which is associated with a short increase in MHR [25].

The mechanism for periodic MHR decelerations is probably different because aortocaval compression might occur in the supine position. Ibrahim et al. [12] speculated that the reduced venous return to the heart might still occur in a lateral position because they also observed a decline in MHR in this position. A possible explanation is that aortacaval compression probably occurs more often when the uterus contracts. When the uterus is relaxed, it will mold around the great vessels, causing little compression. During a contraction, however, compression might happen more readily.

Periodic increases or decreases in MHR, as reflected by total gained and lost beats, are small and infrequent in pregnancy. Although the mechanisms of these changes remain uncertain, there seems to be an association with uterine activity in some pregnant women [12]. The present finding that periodic increases in MHR above the baseline, as reflected by gained beats, were negatively associated with birthweight and its z-score at 34+ gestational weeks indicates that periodic increases in MHR are associated with fetal growth; however, the underlying mechanisms are not yet fully understood. The present finding that the association between lost beats and birthweight was stronger at 34+ weeks than at 20–24 weeks may reflect the effects of a larger uterus.

Because the MHR was obtained with automatic identification of R-peaks, and the heart rate parameters were extracted with an algorithm, there should not have been any subjectivity in the data acquisition. Another strength of the study is monitoring of FHR patterns in uncomplicated pregnancies as non-stress tests are usually reserved for complicated pregnancies. A limitation is that the fetuses or recordings were not selected to represent similar fetal sleep states. However, it would have been extremely difficult select the fetal state by a computer program.

In conclusion, the correlation of low MHR in early pregnancy with birthweight or low z-score of birthweight for gestational age might be helpful in identifying pregnancies at risk of poor fetal growth. In addition, it might aid additional assessments to identify fetuses at risk of stillbirth, especially among primigravidas who have no previous pregnancy to highlight potential complications during pregnancy.

Supplementary Material

Supp FigS1

Figure S1 Principal component analysis of the contributions of the different variables at 20–24 gestational weeks to birthweight. Principal component 1 (PC1) on the horizontal axis explains 44% of the total variation, whereas principal component 2 (PC2) on the vertical axis explains 29% of the total variation. Therefore, these two components explain 73% of the total variation in birthweight. Note that z-score and BMI are almost parallel to the vertical axis factor 2, indicating that they are highly correlated with each other and with PC2. The variable mean heart rate and SDNN were closer to the horizontal axis and thus more correlated with PC1.

Supp FigS2

Figure S2 Analysis of variance of birthweight versus maternal heart rate at 20–24 gestational weeks. F_{6, 2924}=2.1942, P<0.05; Kruskal–Wallis test, P<0.05. Different letters above the vertical bars indicate a significant difference.

Supp FigS3

Figure S3 Analysis of variance of birthweight versus maternal heart rate at 34+ gestational weeks. F_{6, 3712}=43.286, P<0.001; Kruskal–Wallis test, P<0.01. Different letters above the vertical bars indicate a significant difference.

Synopsis:

The association observed between low maternal heart rate during pregnancy and low birthweight might contribute to earlier identification of high-risk pregnancies.