HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Maner, W. L. Articles by Saade, G. Search for Related Content PubMed PubMed Citation Articles by Maner, W. L. Articles by Saade, G.

Predicting Term and Preterm Delivery With Transabdominal Uterine Electromyography

From the Division of Reproductive Sciences, Department of Obstetrics and Gynecology, University of Texas Medical Branch, Galveston, Texas.

Address reprint requests to: Robert E. Garfield, PhD, University of Texas Medical Branch, Department of Obstetrics and Gynecology, 301 University Boulevard, Galveston, TX 77555-1062; E-mail: rgarfiel{at}utmb.edu.

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
OBJECTIVE: To determine whether delivery can be predicted using transabdominal uterine electromyography.

METHODS: A total of 99 patients were grouped as either term (37 weeks or more) or preterm (less than 37 weeks). Uterine electrical activity was recorded for 30 minutes in clinic. Electromyographic "bursts" were evaluated to determine the power density spectrum. Measurement-to-delivery time was compared with the average power density spectrum’s peak frequency. Receiver operating characteristic curve analysis was performed for 48, 24, 12, and 8 hours from term delivery, and 6, 4, 2, and 1 day(s) from preterm delivery.

RESULTS: The power density spectrum peak frequency increased as the measurement-to-delivery interval decreased. Receiver operating characteristic curve analysis gave high positive and negative predictive values for both term and preterm delivery. At term, the average power density spectrum peak frequency was significantly higher for the 24-or-fewer-hours-to-delivery group than for the more-than-24-hours-to-delivery group, whereas at preterm, the average power density spectrum peak frequency was significantly higher in the 4-or-fewer-days-to-delivery group than in the more-than-4-days-to-delivery group (P < .05).

CONCLUSION: Transabdominal uterine electromyography predicts delivery within 24 hours at term and within 4 days preterm. This methodology offers many advantages and benefits that are not available with present uterine monitoring systems.

It is widely accepted that uterine contractions are generated by the electrical activity originating from the depolarization and repolarization of billions of smooth-muscle myometrium cells.¹ When such polarization alternations involve many myometrial cells and happen in immediate succession, "bursts" of activity are generated. This electrical activity is low and uncoordinated early in gestation² but becomes intense and synchronized later in pregnancy, eventually building to a peak at term.³ Early in pregnancy, poor electrical coupling between myometrial cells is partly responsible for the relatively inactive uterus.⁴ The development of vast numbers of gap junctions undoubtedly also plays a role in the electrical evolution of the myometrium.^5–7 Other factors may also play a role, but it is generally thought that the uterus undergoes a critical transition to become electrically prepared for labor and delivery. Estimating changes in the electrical signal characteristics when this transition occurs (for term and preterm patients) and establishing the associated predictive parameters are the subjects of this study.

Measurement of electrical activity has previously been accomplished by placing electrodes directly on the uterus⁸ and, more recently, on the abdominal surface.^9,10 Even more recent studies have demonstrated that it is possible to accurately record myometrial activity from the abdominal surface¹¹ and that power spectrum analysis can be used to effectively quantify the data.¹² However, until now, electrical characterization was incomplete and required more research. We intend to show that not only can the electrical progression of the uterus be described quantitatively, but also that the period during which the transition to electrical preparedness occurs can be estimated for both term and preterm patients, leading to the possibility of the prediction of parturition.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
The institutional review board of the University of Texas Medical Branch gave approval for this study, and all patients were required to sign written consent forms. Of 128 pregnant patients who consented to participate in the study, 99 patients of varying gestational ages (27–42 weeks) were ultimately included. Patients were recruited over a 22-month period. All patients presented to the clinic with contractions to rule out the possibility of labor. Each was placed into one of two groups: term (37 or more weeks’ gestation, n = 57) or preterm (less than 37 weeks’ gestation, n = 42). Maternal ages ranged from 17 to 36 years. In the term group, approximately 53% were Hispanic, 33% were white, and 14% were black, whereas the preterm group composed 41% Hispanic, 26% white, and 33% black patients. Inclusion criteria for this study were singleton gestation, ultimate spontaneous vaginal delivery (29 patients were excluded owing to cesarean delivery), more than 24 weeks’ gestation, intact membranes, cervical dilatation 2 cm or less and effacement less than 80%, and no signs of infection. Patients with unusual distress were excluded. Furthermore, to ensure optimal electrical traces from the abdominal surface, those patients weighing more than 230 lb were excluded. Average gestational ages at recording were 39.26 ± 1.29 weeks and 31.76 ± 5.11 weeks for term and preterm patients, respectively, whereas average gestational ages at the time of delivery were 39.52 ± 1.28 weeks and 36.36 ± 3.87 weeks, respectively. All patients included were admitted to the labor and delivery area of the University of Texas Medical Branch at Galveston.

Electrode attachment sites were prepared by first cleaning away excess oil with alcohol pads, and then using a mild abrasive and impedance-reducing gel to gently rub off the outer layers of the skin, improving electrical conduction to the electrode. The electrodes were self-adhesive Ag₂Cl models, each approximately 2 cm² in area (Quinton, Bothell, WA). Two sets of these bipolar electrodes were attached to the abdomen near the navel. Each electrode was separated from its respective partner by approximately 3 cm. Grounding was accomplished by placing another lead laterally on the patient’s hip. Sampling was done at 100 Hz. The differential signal was analog band-pass filtered from 0.05 Hz to 4 Hz to remove unwanted signal components and to prevent aliasing. The information was then amplified and stored in a personal computer. Analysis was performed using Chart 4.0 software (AD Instruments, Castle Hill, Australia). Patients were monitored with this system continuously for 30 minutes. Respiration rate and heart rate were checked periodically during the recording and noted.

Power spectrum methods (Fourier transform) were used for analysis. Only the uterine electrical bursts observed in the recordings were used. No information was quantified for those periods of the record during which activity was quiescent. Using the Chart 4.0 software (AD Instruments) for the Fourier analysis, data from all 99 patients were examined. Fourier analysis is predicated on the notion that virtually any signal can be constructed from a sum of sinusoidal components.¹³ The Fourier transform deconstructs a signal into its components. Because the recording electrodes are located on the abdomen in the immediate vicinity of the myometrium, and because the myometrium becomes such a relatively large muscle, the contributions to the electrical signal from the uterus should be predominant in the recordings over other biologic events in the prone, relaxed patient. Therefore, aside from artifacts, such as possible patient movement, respiration, or skin potentials,¹⁴ the frequency at which the highest power occurs should correspond primarily to the contributions from the myometrium. The power density spectrum peak frequency was therefore chosen as the parameter of interest, as in our previous studies.¹² Furthermore, to ensure that only myometrial signals were being analyzed, only the bursts of uterine activity were selected for power spectrum analysis. Moreover, within the power spectra generated, only the activity from 0.34 Hz to 1.0 Hz was searched for peaks (see below). It was thought that if changes in this parameter occurred as the time to delivery approached zero, it could be indicative of how prepared the uterus was for labor.

Reviewing the spectral content (with Signal Processing Toolbox, Matlab, Mathworks Inc., Natick, MA) in the power density spectrum for uterine bursts revealed that about 98% of the uterine power spectral components reside below 1 Hz. Furthermore, assessing the data on patient respiration showed that about 95% of patients maintained respiration rates at or below 20 events per minute, or about 0.33 Hz, during the entire recording. All the patients involved in this study showed cardiac rates higher than 60 beats per minute, or 1 Hz, during recording. Therefore, to additionally eliminate respiratory and cardiac signal contributions, while also reducing the effects of low-frequency patient motion artifact and any other nonuterine signals from the analysis, the highest-magnitude peaks in the power density spectrum were selected only from within the range of 0.34 Hz to 1 Hz. Very-low-frequency components of the uterine electromyography signals, those below 0.34 Hz, are often hidden or convoluted with motion and respiration artifact and can be difficult to isolate and analyze. Therefore, low-frequency components are intrinsically more erroneous than the higher uterine frequencies around which this study revolved.

Power spectrum peak frequency values from subsequent bursts within a recording were averaged for each patient. Then True Epistat software (Epistat Services, Round Rock, TX) was used for receiver operating characteristic (ROC) curve analysis. This was performed on the averaged power density spectrum peak frequency values to determine how well the data predicted delivery. A ROC curve displays sensitivity against 1 – specificity for each patient. Depending on whether a test is more sensitive or more specific, the best cutoff can be different. However, because overall prediction of labor and delivery was the main point of interest for this study, the cutoff at which the sum of the positive and negative predictive values was highest was used. From the number of true and false positive results and from the number of true and false negative results, positive predictive values and negative predictive values for 48-, 24-, 12-, and 8-hour gold standards, or end points, were generated for term patients. For the preterm group, 6-, 4-, 2-, and 1-day end points were used to calculate the positive and negative predictive values.

Sigma-Stat software (SPSS Inc., Chicago, IL) was implemented for statistical comparison of groups. Mann–Whitney rank-sum was used for comparing any two given groups, and one-way analysis of variance (Dunn test) was used for comparing more than two groups. A P value of less than .05 was considered statistically significant.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Uterine electromyography consisted of bursts of activity separated by periods of quiescence. The uterine electrical signals, as measured by transabdominal electromyography, corresponded to uterine mechanical activity, as recorded by tocodynamometer (Figure 1). Power spectrum analysis of the bursts of uterine electrical activity indicated prominent peaks, within the 0.34–1-Hz range, that corresponded to uterine mechanical activity (Figure 2), and that were not present during the quiescent electrical periods. In general, the frequency (of the largest-amplitude peak, as determined by the power spectrum) for electrical activity within bursts increased as the measurement-to-delivery interval decreased in term patients (Figure 3A) and in preterm patients (Figure 3B). This phenomenon is associated with the myometrial cells’ ability to more rapidly undergo depolarization and repolarization as the uterus becomes electrochemically prepared for labor. The division of term patients into two subgroups (Figure 4A), 24 or fewer hours from recording to delivery, and more than 24 hours from recording to delivery, resulted in a significant difference between the two sets (0.4768 ± 0.0144 Hz versus 0.4042 ± 0.0185 Hz). A similar result was seen (Figure 4B) when comparing patients in the 4-or-fewer-days group to those who delivered more than 4 days from measurement in the preterm labor group (0.5113 ± 0.0279 Hz versus 0.3899 ± 0.0061 Hz). Average interobserver variability in the power spectrum peak frequency measurements was ±0.0046 Hz, whereas average intraobserver variability was ±0.0019 Hz.

View larger version (28K): [in this window] [in a new window]   Figure 1. Uterine electrical activity is responsible for uterine contractions. The two phenomena are seen corresponding in time. Maner. Predicting Delivery. Obstet Gynecol 2003.

View larger version (15K): [in this window] [in a new window]   Figure 2. Power spectrum of a typical burst from a patient in labor showing uterine spectral components in the 0.34–1-Hz range. The power density spectrum is derived by dividing the power by the frequency at which the power occurs. Note the power density spectrum peak frequency (frequency at which the largest-amplitude component occurs) is at approximately 0.5 Hz. Maner. Predicting Delivery. Obstet Gynecol 2003.

View larger version (18K): [in this window] [in a new window]   Figure 3. A) Plot showing the change in power density spectrum (PDS) peak frequency vs the measurement-to-delivery time for term patients. The increase is most dramatic within 24 hours of delivery. The number of patients at each time point is also indicated. B) Plot showing the change in PDS peak frequency vs the measurement-to-delivery time for preterm patients. The increase is most dramatic within 4 days of delivery. The number of patients at each time point is also indicated. Maner. Predicting Delivery. Obstet Gynecol 2003.

View larger version (17K): [in this window] [in a new window]   Figure 4. A) Comparison of average power density spectrum (PDS) peak frequency values for term patients delivering within 24 hours of measurement with those delivering more than 24 hours from measurement. The 24-or-fewer-hours group is statistically higher. *P < .01. Standard errors shown. B) Comparison of average PDS peak frequency values for preterm patients delivering within 4 days of measurement with those delivering more than 4 days from measurement. The 4-or-fewer-days group is statistically higher. *P < .05). Standard errors shown. Maner. Predicting Delivery. Obstet Gynecol 2003.

View larger version (15K): [in this window] [in a new window]   Figure 5. Receiver operating characteristics curve for term patients (n = 57) showing Z value, area under the curve (AUC), and positive and negative predictive values (PPV and NPV, respectively). An end point of 24 or fewer hours to delivery was used for the curve. Maner. Predicting Delivery. Obstet Gynecol 2003.

View larger version (15K): [in this window] [in a new window]   Figure 6. Receiver operating characteristics curve for preterm patients (n = 42) showing Z value, area under the curve (AUC), and positive and negative predictive values (PPV and NPV, respectively). An end point of 4 or fewer days to delivery was used for the curve. Maner. Predicting Delivery. Obstet Gynecol 2003.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
The present work has shown that noninvasive, transabdominal recording of uterine activity can be used to predict labor and delivery. Frequency parameters are especially well suited for quantifying uterine electromyography data over and above intensity-dependent variables. Term and preterm patients can be successfully monitored and evaluated for preparedness for labor based on these variables, and delivery time forecasts can be made using positive and negative predictive values.

Neither simple amplitude analyses nor raw signal integration seem reliable in predicting delivery, because attenuation of myometrial signals occurs more for some patients and less for others, depending on a variance in subcutaneous tissues and a variance in conductivity at the skin–electrode interface. A thicker layer of tissue between the myometrium and the pickup electrodes will result in a smaller signal. In addition, poor conductivity due to high impedance, possibly related to salinity levels or other chemical factors at the skin surface, serves to reduce the measured uterine signal strength and to increase the background noise in the recordings of some patients.

Work has previously been done to characterize the electrical activity of the uterus noninvasively by using the frequency parameter.¹⁵ The present work also shows that the uterine transition to labor involves an increase in the average power density spectrum peak frequency from lower to higher frequencies. At the cellular level, there exist myometrial action potentials¹⁶ in a frequency range of about 0.05–1.5 Hz. Long before delivery, during the nonlaboring phase of the uterus, the electrochemical processes (specifically the physical properties of ion channels or the established levels of the resting potential and threshold potential) governing these action potentials limit the myometrium to slower depolarizations and repolarizations.^16,17 When the uterus becomes more excitable and signal propagation distance and strength increase, the electrochemical conditions are favorable to higher frequency cycles within the bursts of activity. This is evidenced by the fact that a larger percentage of spectral components of uterine electrical bursts are found at higher frequencies in term patients that are within 24 hours of delivery at the time of measurement, as compared with those who go on to deliver at more than 24 hours from recording. A similar change seems to occur earlier in gestation for some patients, and in these preterm patients the transition occurs approximately 4 days before delivery. However, there seems to be no significant difference between the average power density spectrum peak frequency of preterm patients who are within 24 hours of delivery and term patients who are within 24 hours of delivery. This suggests that the two groups at least share a common mechanism of uterine development, partially independent of gestational age.

The longer measurement-to-delivery times after a shift in uterine spectral components for the preterm group may be the result of insufficiently ripened cervices, which require longer labors for delivery. Another possible reason for a longer measurement-to-delivery time for the preterm group is that a preterm laboring patient generally has an underdeveloped and smaller uterus that is inadequate for quick or effective expulsion of the fetus. However, the higher Z values and greater area under the ROC curves associated with the preterm group suggest that preterm delivery may be even more favorable to predict than term delivery. At any rate, the general increase in the power density spectrum peak frequency observed as a patient enters the final phase of uterine readiness indicates that a greater fraction of the power of uterine activity resides at higher frequencies in patients just before labor than in those far removed from delivery.

The incidence and extent to which shifts in uterine electrical spectral components occurred for patients as measurement-to-delivery time decreased implies that these changes may be used reasonably successfully in predicting delivery, at least within days of delivering. Also, the ability to make a successful negative prediction for labor when no spectral shift is seen is of great use to obstetricians, patients, and hospitals, where subjective analysis might often fail.

In addition to the power density spectrum peak frequency analysis presented here, other parameters for evaluation should be considered. The integral under the power spectrum, median power, total energy (sum of power times burst duration), and power and frequency ratios, along with many other possible variables, should be taken into consideration. If the signal quality is kept high and if signals could be standardized for conduction and propagation differences between patients, one or more such variables (possibly used in conjunction) may lead to even more effective characterization and prediction of delivery in both preterm and term patients.

All devices and methods currently in use, such as tocodynamometer, intrauterine pressure catheters, fetal fibronectin, and ultrasound, have little capability of predicting labor and delivery.¹⁸ Noninvasive uterine electromyogram monitoring will change that. Because positive predictive results using power density spectrum peak frequency were best at around 2 days from delivery in term patients, and the overall predictive capability is best at about 1 day from delivery, it may be useful to evaluate the uterine electromyography of term patients, for example, on consecutive days beginning several days to 1 week before their expected due date to assess their condition and react appropriately in managing the patient. As a precautionary measure, a similar procedure could be established much earlier in pregnancy for patients at high risk for preterm labor.¹⁹

Many other potential uses and benefits of transabdominal uterine electromyography recording have been previously described.^13,18 These briefly include the following: objective rather than subjective measurements for obstetricians to use in evaluating preparedness for labor, prevention of unnecessary admissions in term patients, improving perinatal outcome, including prevention of preterm labor, and better-defined treatments, such as the use of tocolytics or oxytocin, as the case may be.

	Footnotes

 
Supported by a grant from the National Institutes of Health (HD37480-01). Studies were conducted in the General Clinical Research Center at the University of Texas Medical Branch at Galveston, funded by grant M01RR0073 from the National Center for Research Resources, National Institutes of Health, United States Public Health Service.

doi:10.1016/S0029-7844(03)00341-7

Received October 9, 2002. Received in revised form December 9, 2002. Accepted December 26, 2002.

	REFERENCES

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
1. Marshall JM. Regulation of the activity in uterine muscle. Physiol Rev 1962;42:213–27.

2. Kuriyama H, Csapo A. A study of the parturient uterus with the microelectrode technique. Endocrinology 1967; 80:748–53.[Medline]

3. Harding R, Poore ER, Bailey A, Thorburn GD, Jansen CAN, Nathanielsz PW. Electromyographic activity of the nonpregnant and pregnant sheep uterus. Am J Obstet Gynecol 1982;142:448–57.[Medline]

4. Demianczuk N, Towell ME, Garfield RE. Myometrial electrophysiological activity and gap junctions in the pregnant rabbit. Am J Obstet Gynecol 1984;149:485–91.[Medline]

5. Verhoeff A, Garfield RE, Ramondt J, Wallenburg HCS. Electrical and mechanical uterine activity and gap junctions in peripartal sheep. Am J Obstet Gynecol 1985;153: 447–54.[Medline]

6. Verhoeff A, Garfield RE, Ramondt J, Wallenburg HC. Myometrial activity related to gap junction area in periparturient and in ovariectomized estrogen treated sheep. Acta Physiol Hung 1986;67:117–29.[Medline]

7. Garfield RE. Role of cell-to-cell coupling in control of myometrial contractility and labor. In: Garfield RE, Tabb TN, eds. Control of uterine contractility. Boca Raton, Florida: CRC Press, 1994:40–81.

8. Wolfs GM, Van Leeuwen M. Electromyographic observations on the human uterus during labor. Acta Obstet Gynecol Scand Suppl 1979;90:1–61.[Medline]

9. Devedeux D, Marque C, Mansour S, Germain G, Duchene J. Uterine electromyography: A critical review. Am J Obstet Gynecol 1993;169:1636–53.[Medline]

10. Figueroa JP, Honnebier MB, Jenkins S, Nathanielsz PW. Alteration of 24-hour rhythms in the myometrial activity in the chronically catheterized pregnant rhesus monkey after 6-hour shift in the light-dark cycle. Am J Obstet Gynecol 1990;163:648–54.[Medline]

11. Garfield RE, Saade G, Buhimschi C, Buhimschi I, Shi L, Shi SQ, et al. Control and assessment of the uterus and cervix during pregnancy and labour. Hum Reprod Update 1998;4:673–95.[Abstract/Free Full Text]

12. Buhimschi C, Boyle MB, Saade GR, Garfield RE. Uterine activity during pregnancy and labor assessed by simultaneous recordings from the myometrium and abdominal surface in the rat. Am J Obstet Gynecol 1998;178:811–22.[Medline]

13. Garfield RE, Chwalisz K, Shi L, Olson G, Saade GR. Instrumentation for the diagnosis of term and preterm labour. J Perinat Med 1998;26:413–36.[Medline]

14. Marque C, Duchêne J, Lectercq S, Panczer G, Chaumont J. Uterine EMG processing for obstetrical monitoring. IEEE Trans Biomed Eng 1986;33:1182–7.[Medline]

15. Buhimschi C, Boyle M, Garfield RE. Electrical activity of the human uterus during pregnancy as recorded from the abdominal surface. Obstet Gynecol 1997;90:102–11.[Abstract]

16. Garfield RE, Yallampalli C. Structure and function of uterine muscle. In: Chard T, Grudzinskas JG, eds. The uterus. Cambridge reviews in human reproduction. Cambridge, UK: 1994;54–93.

17. Wynn R, Jollie W. Biology of the uterus, 2nd ed. London: Plenum Publishing, 1989.

18. Garfield RE, Maul H, Shi L, Maner W, Fittkow C, Olsen G, et al. Methods and devices for the management of term and preterm labor. Ann N Y Acad Sci 2001;943:203–24.[Abstract/Free Full Text]

19. Garfield RE, Yallampalli C. Control of myometrial contractility and labor. In: Chwalisz K, Garfield RE, eds. Basic mechanisms controlling term and preterm birth. New York: Springer-Verlag, 1993:1–28.

This article has been cited by other articles:

				W. J. E. P. Lammers, H. Mirghani, B. Stephen, S. Dhanasekaran, A. Wahab, M. A. H. Al Sultan, and F. Abazer Patterns of electrical propagation in the intact pregnant guinea pig uterus Am J Physiol Regulatory Integrative Comp Physiol, March 1, 2008; 294(3): R919 - R928. [Abstract] [Full Text] [PDF]

				R. E. Garfield Is knowledge of the pattern of electrical activity in the pregnant uterus helpful to our understanding of uterine function? Focus on "Patterns of electrical propagation in the intact pregnant guinea pig uterus" by Lammers et al. Am J Physiol Regulatory Integrative Comp Physiol, March 1, 2008; 294(3): R917 - R918. [Full Text] [PDF]

				J. D. Roizen and L. J. Muglia Understanding the Timing of Birth: The Continuing Challenge to Prevent Prematurity NeoReviews, March 1, 2006; 7(3): e151 - e159. [Full Text] [PDF]

				M. Doret, R. Bukowski, M. Longo, H. Maul, W. L. Maner, R. E. Garfield, and G. R. Saade Uterine Electromyography Characteristics for Early Diagnosis of Mifepristone-Induced Preterm Labor Obstet. Gynecol., April 1, 2005; 105(4): 822 - 830. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Maner, W. L. Articles by Saade, G. Search for Related Content PubMed PubMed Citation Articles by Maner, W. L. Articles by Saade, G.