To evaluate the performance of external electronic fetal heart rate and uterine contraction monitoring according to maternal body mass index.
To evaluate the performance of external electronic fetal heart rate and uterine contraction monitoring according to maternal body mass index.
Secondary analysis of prospective equivalence study.
Three US urban teaching hospitals.
Seventy-four parturients with a normal term pregnancy.
The parent study assessed performance of two methods of external fetal heart rate monitoring (abdominal fetal electrocardiogram and Doppler ultrasound) and of uterine contraction monitoring (electrohystero-graphy and tocodynamometry) compared with internal monitoring with fetal scalp electrode and intrauterine pressure transducer. Reliability of external techniques was assessed by the success rate and positive percent agreement with internal methods. Bland–Altman analysis determined accuracy. We analyzed data from that study according to maternal body mass index.
We assessed the relationship between body mass index and monitor performance with linear regression, using body mass index as the independent variable and measures of reliability and accuracy as dependent variables.
There was no significant association between maternal body mass index and any measure of reliability or accuracy for abdominal fetal electrocardiogram. By contrast, the overall positive percent agreement for Doppler ultrasound declined (p = 0.042), and the root mean square error from the Bland–Altman analysis increased in the first stage (p = 0.029) with increasing body mass index. Uterine contraction recordings from electrohysterography and tocodynamometry showed no significant deterioration related to maternal body mass index.
Accuracy and reliability of fetal heart rate monitoring using abdominal fetal electrocardiogram was unaffected by maternal obesity, whereas performance of ultrasound degraded directly with maternal size. Both electrohysterography and tocodynamometry were unperturbed by obesity.
abdominal fetal electrocardiogram
body mass index
fetal heart rate
intrauterine pressure transducer
positive percent agreement
root mean square error
Performance of two external fetal heart rate monitoring techniques was measured in relation to maternal body mass index (BMI). Cardiotachometry using maternal abdominal electrodes for fetal electrocardiogram detection was unaffected by BMI. Performance of ultrasound detection of fetal heart rate, however, degraded as BMI increased.
The burgeoning epidemic of obesity in much of the world has not spared pregnant women [1-6]. The prevalence of obesity during pregnancy in some regions has been reported to be as high as 45% [2, 4]. There are many medical and obstetric complications associated with excess maternal body weight, and these have received considerable attention in the literature [4-8]. Sometimes overlooked is the fact that monitoring the fetal heart rate (FHR) or uterine contractions (UC) during labor can be difficult when the parturient is corpulent, potentially placing fetus and mother at risk .
Most intrapartum electronic FHR and UC monitoring is done externally. A Doppler ultrasound (US) transducer and tocodynamometer (TOCO), each held against the parturient's abdominal wall by a belt, are used most commonly. Another method, which determines the FHR and identifies UC from electrophysiological signals detected by maternal abdominal surface electrodes, has been recently introduced [10, 11].
Obesity creates potential problems for external monitoring techniques because the expanse of adipose tissue interposed between the abdominal surface instruments and the uterus could degrade the quality of the US, TOCO, abdominal fetal electrocardiogram (afECG), or electrohysterogram (EHG) signals used to identify fetal cardiac activity or uterine contractility. Also, maternal movement may more readily alter the spatial relation between the abdominal instrument and the fetal heart or uterus. The inferiority of external Doppler US and TOCO to internal techniques is well known [12, 13]. The severity of this problem is magnified by obesity, in which FHR monitoring is often suboptimal and may require frequent transducer adjustment by the staff. Practitioners often, therefore, resort to a more invasive method of FHR monitoring in obese patients, employing a fetal scalp electrode and intrauterine pressure transducer (IUPT) . This method cannot be used throughout labor. It is only feasible when the cervix is dilated and membranes are ruptured; moreover, it carries a risk of fetal and maternal infection .
In a previous study we evaluated the performance of both available external FHR and UC monitoring methods in comparison with simultaneous internal monitoring with a scalp electrode and IUPT [10, 11]. We found that the accuracy and reliability of afECG and EHG monitoring was at least equivalent to that of US and TOCO; in some respects afECG was superior to US. The aim of the present study was to analyze those data according to maternal body mass index (BMI) so as to determine whether monitoring performance was affected by the parturient's obesity.
We performed a secondary analysis of data from a trial in which women with a singleton term pregnancy underwent intrapartum FHR (n = 75) and UC (n = 74) monitoring simultaneously with three different methods [10, 11]. External US and afECG were compared with the scalp electrode, and external TOCO and EHG with the IUPT. The parent study was carried out at three hospitals in the USA with approval of each site's Institutional Review Board, and conformed to the guidelines of the World Medical Association Declaration of Helsinki. The specific aim of the parent study was to assess the accuracy and reliability of afECG-based and Doppler-based EFM in comparison to the optimal standard of acquiring the fetal ECG from a direct fetal scalp electrode. Further, the study assessed the performance of TOCO and EHG in comparison to that of an IUPT in the same manner.
Each study site used the Model 50XM monitoring system (Philips Healthcare, Andover, MA, USA) for internal scalp electrode and external US FHR monitoring, and for TOCO. The IUPT systems were either IPC5000 (Clinical Innovations, Salt Lake City, UT, USA) or Accutrace (Coviden, Mansfield, MA, USA). The afECG and EHG monitoring was done with the Model AN24 fetal–maternal monitor (Monica Healthcare, Ltd., Nottingham, UK).
Patients were monitored initially with US and TOCO. Once it was determined that these were working properly, the abdominal wall electrodes for detection of the fetal ECG and EHG were applied. Any patient who sub-sequently required internal monitoring for clinical reasons was entered into the study. Once the IUPT and fetal scalp electrode were placed, they were used for clinical decision-making. The output from the external techniques was still recorded continuously, but it was not available to the obstetric team. The protocol required the US and TOCO recordings to be checked for optimum transducer placement by a member of the research team every 20–30 min. The FHR and UC data from all devices were generated at 0.25-s intervals, and synchronized to within 0.25 s.
As measures of reliability we determined the success rate for afECG and TOCO and the positive percent agreement (PPA) for both external devices. Success rate was the percentage of time that the device reported a non-zero FHR value. The PPA indicated the percentage of time that the device generated an FHR value within 10% of an FHR value obtained from the scalp electrode or, for UC, provided a valid contraction signal. In this context, validity was defined as a deflection of at least 10% above baseline lasting 40–120 s.
To assess the accuracy of the external devices in the parent study we performed Bland–Altman analysis [15, 16]. The root mean square error (RMSE) was determined for the FHR differences between the internal and each external device and the regression line of the Bland–Altman plot. Those RMSE results were used in the current study as a quantitative reflection of accuracy. For UC we used a count of contractions in consecutive 10-min windows in the Bland–Altman analysis. We also calculated the sensitivity of UC detection (the frequency with which the external device displayed a valid contraction when the IUPT did), the false positive rate (percentage of time when the external device showed a contraction but the IUPT did not) and the positive predictive value (the number of times the EHG or TOCO displayed a valid contraction as a percentage of all the contractions that the device displayed).
The confusion rate (CR) was calculated as a reflection of the frequency with which the external devices likely reported the maternal heart rate rather than the FHR. The CR was the percentage of FHR determinations for which the afECG or US calculated an FHR that was both more than 5% different from that of the scalp electrode and within 5% of the maternal heart rate. The CR was calculated for 47 patients, the number of subjects in whom pulse oximetry was used for continuous detection of maternal heart rate during FHR monitoring.
Of the 75 cases who had FHR monitoring in the parent study, a BMI was not recorded for one patient. We divided the remaining 74 FHR and UC monitoring cases into three groups according to the maternal BMI computed on admission in labor (<30, 30–35 and >35 kg/m2) in order to ascertain whether characteristics of the patients differed according to their body size. We sought these differences using one-way analysis of variance for continuous variables and the chi-squared or Fisher's exact test for categorical variables.
To assess the relationship between maternal body mass and performance of the external monitors, we used linear regression, with BMI as the independent variable and measures of reliability and accuracy as dependent variables. For each regression the Pearson product-moment correlation coefficient R and the p-value were calculated, the latter by analysis of variance; p < 0.05 was considered significant.
The characteristics of the patients were similar across the BMI groups (Table 1), which were comparable in maternal age, gestational age and the use of epidural anesthesia. The range of BMI values was 19–54 (mean 32.6 ±7.5) kg/m2.
|BMI < 30 kg/m2 (n = 28)||BMI 30–35 kg/m2 (n = 26)||BMI > 35 kg/m2 (n = 20)||p-value|
|BMI (kg/m2)||25.8 ± 2.9||32.6 ± 2.0||42.3 ± 5.7||<0.001|
|Gestational age (weeks)||39.8 ± 1.0||37.8 ± 7.8||39.8 ± 1.1||0.233|
|Maternal age (years)||25.4 ± 5.2||25.2 ± 5.1||25.1 ± 5.8||0.977|
|Epidural analgesia (%)||77.3||84.2||85.0||0.777|
|FHR monitoring, total duration (min)||7190||6908||4830|
|FHR monitoring, duration/patient||(257 ± 180)||(266 ± 210)||(242 ± 157)||0.902|
|UC monitoring, total (min)||7611||6688||4854|
|UC monitoring, duration/patient||254 ± 174||291 ± 205||243 ± 159||0.632|
The results of the FHR monitoring assessment are summarized in Table 2. Regression analysis revealed no significant association between maternal BMI and any measure of reliability or accuracy for the afECG technique. R values were very small, and the regression lines were nearly horizontal. This was true both overall (data shown) and with results stratified according to stage of labor. In fact, the success rate in the subgroup of nine severely obese women with a BMI > 40 kg/m2 was essentially the same as in the leanest group with a BMI < 30 kg/m2 (86 ± 23% vs. 84 ± 15%; p = 0.432). The same applied to the PPA (81.4 ± 23.8% and 81.6 ± 15% in the heaviest and lightest groups, respectively; p = 0.398). The RMSE results similarly showed no significant change in accuracy of the afECG as BMI increased.
|Variable||Mean ± SD||R||Slope||p > F|
|Success rate (%)||83.4 ± 20.2||0.06||0.001||0.633|
|PPA (%)||81.5 ± 20.6||0.03||0.08||0.802|
|RMSE (bpm)||5.29 ± 2.4||0.09||−0.03||0.442|
|CR (%)||0.4 ± 0.7||0.15||0.001||0.302|
|Success rate (%)||82.5 ± 21.2||0.21||−0.006||0.076|
|PPA (%)||73.0 ± 24.8||0.24||−0.78||0.042|
|RMSE (bpm)||10.9 ± 5.8||0.08||0.07||0.461|
|CR (%)||9.0 ± 15.0||0.17||0.003||0.266|
Performance of the US technique was not as stable across the range of patient body mass as that of the afECG. The overall success rates for US FHR monitoring showed a downward trend as BMI increased (p = 0.076). The PPA for US declined significantly with increasing BMI (p = 0.042), and the RMSE from the Bland–Altman analysis in the first stage rose (p = 0.029). Although the overall US success rate of 82.5 ± 21.2% was similar to that of the afECG (83.4 ± 20.2%), the success rate of US in the nine women with a BMI > 40 kg/m2 was only 66 ± 30%, considerably lower than the 86 ± 23% in the afECG in those patients (p = 0.132).
The maternal–FHR CRs were very low for afECG. The overall mean was 0.5 ± 0.7% (median 0.1, 95% CI 0.2–0.6%), with a range of 0–2.5%. The CR was the same in the first and second stage of labor and did not change significantly as a function of maternal BMI. The CR for US, 9.0 ± 15% (median 2.1, 95% CI 4.5–13.6%), also did not show a trend related to maternal size, but the overall rates ranged from 0 to 80.9%, substantially higher than for afECG at every BMI level (p = 0.022).
The UC recordings (Table 3) from the EHG showed no significant trend related to maternal BMI for the success rate, positive predictive value, sensitivity or false positive rate. The exception was that there was a significant improvement in sensitivity in the first stage of labor as BMI increased (p = 0.0253). The TOCO functioned similarly, with no significant trend discernible according to BMI either overall, or in individual stages of labor.
|Variable||Mean ± SD||R||Slope||p > F|
|PPV (%)||78.6 ± 16.1||0.01||0.0002||0.950|
|PPA (%)||97.1 ± 8.5||0.09||0.103||0.454|
|RMSE (count/10 min)||0.89 ± 0.36||0.30||0.0005||0.933|
|Sensitivity (%)||86 ± 15||0.20||−0.004||0.090|
|False positive rate (UC/h)||5.3 ± 4.5||0.07||−0.046||0.531|
|PPV (%)||84.2 ± 17.3||0.09||0.002||0.468|
|PPA (%)||61.1 ± 27.6||0.09||−0.011||0.980|
|RMSE (count/10 min)||1.20 ± 0.46||0.12||−0.008||0.311|
|Sensitivity (%)||74 ± 18||0.02||0.001||0.842|
|False positive rate (UC/h)||3.1 ± 3.6||0.10||0.051||0.377|
As the prevalence of obesity in the obstetric population continues to increase  the need for noninvasive FHR and UC monitoring techniques that are reliable, accurate and easy to apply has become pressing. This is especially so because, apart from the technical barriers to adequate monitoring presented by obesity, excess body mass is associated with a number of comorbidities, especially diabetes mellitus and hypertensive disorders, that often require intensive fetal and uterine monitoring during labor. In addition, obesity per se is now recognized as a risk factor for adverse pregnancy outcome [6, 9].
Our analysis showed no detrimental effect of increasing maternal body mass on the performance of afECG and EHG monitoring during labor. In fact, the output of the monitoring device remained quite stable in all our measures of accuracy and reliability across the studied BMI spectrum. As importantly, the frequency with which the machine was likely reporting the mother's heart rate rather than that of the fetus was remarkably low. This is important because confusion between the maternal and FHRs can be a substantial source of erroneous clinical judgment .
By contrast, the Doppler US method of monitoring was negatively affected by high body mass. The PPA decreased and, in the first stage, the RMSE increased significantly as BMI increased. These findings suggest that US monitoring was less reliable in obtaining a valid FHR signal, and that those signals reported were less likely to be accurate as maternal corpulence increased. The CR for US heart rate detection did not deteriorate significantly in obese women. The mean CR was, however 18-fold higher than with afECG. Further, in subjects with a BMI > 35 kg/m2 the CR was 25 times higher for US than for afECG (p < 0.001).
Both external methods of UC monitoring were mostly unaffected by maternal size. There was no observed significant deterioration in any of the performance measures according to maternal BMI. The TOCO also showed no significant trend related to BMI. False-positive contractions (those noted when the IUPT did not register a contraction) were somewhat more common with the EHG method than with TOCO; in neither case did they become more frequent with advancing maternal obesity.
There is substantial evidence from several investigators to support the value of external FHR and UC monitoring using maternal abdominal electrodes. These techniques are at least equivalent, and in some respects superior, to external monitoring with US and TOCO [10, 11, 18-25].
Our findings are consistent with those of other investigators. In the only other study that analyzed UC monitoring using three methods concurrently (TOCO, EHG, IUPT), EHG correlated better with IUPT contraction identification than did TOCO in a sample of 25 obese women . In 2010 Graatsma et al. determined the recording quality of the FHR patterns obtained by afECG in a sample of women across a spectrum of BMIs (16–51 kg/m2) They found no diminution in recording quality as BMI increased .
Our study is the first to analyze the function of all three intrapartum monitoring techniques for FHR and UC obtained simultaneously. This design allowed assessment of the external modes in comparison with internal (fetal scalp electrode and IUPT) monitoring, the latter techniques considered to provide the best attainable accuracy and reliability. While the data from the parent study provided important insights, that study was not powered to support this secondary analysis. Therefore, the possibility of having made a type II error in interpreting some of our negative results exists. Moreover, while our study included a broad spectrum of maternal BMIs, there were only nine women in the severely obese group (BMI > 40 kg/m2). Therefore our conclusions regarding extreme obesity should be interpreted cautiously.
Optimal function of the Doppler US and the TOCO methods requires that the healthcare team adjust transducer positions when necessary. The parent study did not record the number of transducer repositionings made in each patient. Therefore, although the protocol required checking of the transducer function at least every 20–30 min, we cannot be certain that the TOCO and US were used optimally. Nor can we quantify the extra effort required to maintain adequate US/TOCO function. One advantage of the afECG/EHG method is that once the electrodes are placed on the abdomen they do not generally need to be moved, or attended to in any way, thus eliminating the need for frequent readjustment of the surface transducers, which is distracting for both patient and provider.
We conclude from our analysis that intrapartum external FHR monitoring using the afECG method was not affected by maternal obesity, whereas the performance of US degraded directly with increasing maternal corpulence. Both EHG and TOCO were unperturbed by obesity. The CR, although not affected by body mass with either technique, was significantly higher for US than for afECG at all BMI levels. In as much as the same electrode array can be used for FHR and UC detection, the afECG/EHG approach should be considered as the preferred technique for women with a high body mass.
We are grateful to Drs Raymond Brown, Sarmina Hassan, John Himsworth, Fadi Mirza, Sophia Ommani, Barry S. Schifrin and Molham Solomon, for their work on the parent study.
This and the parent study were supported by a grant from Monica Healthcare, Ltd., Nottingham, UK.