Secondary Logo

Journal Logo

Original Articles

Electrogastrographic Norms in Children: Toward the Development of Standard Methods, Reproducible Results, and Reliable Normative Data

Levy, Joseph*; Harris, Jennifer*; Chen, Jiande; Sapoznikov, Dan; Riley, Benita*; De La Nuez, Wendy*; Khaskelberg, Anna*

Author Information
Journal of Pediatric Gastroenterology and Nutrition: October 2001 - Volume 33 - Issue 4 - p 455-461
  • Free


Electrogastrography (EGG) is a measurement of gastric myoelectrical activity collected via cutaneous electrodes. Its noninvasive nature and the relative simplicity of the testing and analysis process make it an attractive technique, especially for use in pediatrics. The EGG was introduced in 1921 by Alvarez et al. (1). Validation of the gastric source of the cutaneous signal came years later, when cutaneous recordings were shown by several investigators to correspond with direct intraluminal and serosal recordings of the gastric slow wave generated by the stomach's pacemaker, the interstitial cells of Cajal of the greater curvature (2,3).

The gastric slow wave, sometimes called electrical control activity, is always present. Although it determines the spatial and temporal organization of gastric contractions, it is not associated directly with contractile activity. Furthermore, although gastric contractions, or electrical response activity, may contribute to the amplitude and breadth of peaks in the recorded tracing, they are not always present, not always perceptible even when present, and not always synchronous with the gastric slow wave (3–5). Thus, the EGG provides insight into the regulation of peristaltic activity, but is not a direct reflection of motility.

Interest in the clinical use of the EGG has surged in the past decade as a result of technical advances in recording equipment and more reliable programs for signal analysis. Past investigators have demonstrated a correlation between abnormal EGG results and the presence of gastrointestinal symptoms and conditions, particularly nausea, gastroparesis, intestinal pseudoobstruction, and various dyspeptic symptoms. Although much work has been carried out with adult patients, fewer data are available on the EGG in children (6–13).

As with any novel technique, definition of standards of recording and analysis and the establishment of normal values are prerequisites for reliable application across populations. Studies performed in adults have shown the importance of a standard meal, emotional factors, sleeping state, and menstrual phase, but have found no major differences among age groups (14–19).

A study of EGG results in a large group of children and adolescents was conducted to define normative data for healthy children and adolescents from 6 to 18 years of age. Given the developmental and maturational considerations pertinent in conducting studies in the pediatric age group, the study was designed to investigate the influence of age on gastric electrical activity within this group. Body mass index (BMI) and gender were also considered as possible factors affecting EGG results in healthy pediatric participants.


Participants and test conditions

Eighty-nine volunteers from 6 to 18 years of age were enrolled in the study. The Presbyterian Hospital Institutional Review Board approved the study protocol, and written informed consent and assent were obtained from all participants' parents.

A detailed questionnaire covering a broad range of gastrointestinal, nutritional, and general health issues was completed for every participant. Using the questionnaire, potential participants' parents rated their children's lack of appetite, overeating, nausea, vomiting, abdominal pain, bloating, sleep disturbance, and diarrhea on scales of frequency and severity. (We developed the questionnaire for this purpose, and it has not been validated.) Any respondent noting a symptom beyond “mild” in severity or beyond “sometimes” in frequency of occurrence was excluded from the study. Unformed stool, presence of blood in the stool, or lack of stool control also disqualified potential participants from enrolling in the study.

Eight of the 89 volunteers were excluded from the study because of their gastrointestinal symptoms and two because of medication intake that could affect gastrointestinal function. Of the remaining 79, 24 participants' tracings were not used. Three of these tracings were flat lines caused by errors in signal collection, and two were lost because of errors in downloading or storing data. In two cases, a skin electrode was found to be loose at some point during the test. Four participants' questionnaires were not received or were not usable. Two participants who ate a nonstandard meal and two who ate less than half of the test meal were excluded. Four tests were too short or too frequently interrupted, four others were designated “unusable” by the attending technician for an unrecorded reason, and one test became the pilot for an induced artifact study.

Tracings from the remaining 55 participants were analyzed. Twenty-seven participants (49%) were female. Ages ranged from 6 to 18 years, with a mean of 11.7 years. No participant had a history of chronic gastrointestinal complaints, surgery, or medication intake. None was taking medication thought to influence gastrointestinal function. Each participant's BMI was calculated from stadiometer-measured heights and weights taken at the time of the procedure (20,21). Participants' characteristics are summarized in Table 1.


Participants fasted for at least 3 hours before the study. Preparation consisted of cleaning and gently rubbing participants' skin with Omniprep (D.O. Weaver & Co., Aurora, CO) to the point of erythema, applying a small amount of electrode gel (Signa Gel; Parker Laboratories, Orange, NJ), and placing three electrodes (Cleartrace; ConMed Corp., Utica, NY) over the gastric area. The umbilicus, xiphoid, and costal margins were used as landmarks for probe placement (22).

During the test, participants watched a movie in the recumbent position while avoiding movement as much as possible. An assistant monitored the procedure and noted any possible sources of motion artifact. In early tests, the assistant stayed at the bedside throughout to note every movement, but after familiarity with artifactual patterns in the signal developed, this was not necessary. In these cases, the assistant stayed in the room with the participant long enough to reinforce the instructions, to answer questions, and to verify that the instruments were functioning. Thereafter, he or she stepped back into the room occasionally to see that the participant was comfortable, to reinforce the importance of lying still, and to administer the test meal.

A continuous EGG recording was made, including 60 minutes of preprandial baseline, up to 15 minutes of meal consumption, and 60 minutes of postprandial conditions. The test meal consisted of a turkey sandwich on white bread, 8 potato chips, 1.5 cookies, and a glass of fruit juice. This 448-kcal meal is composed of 20.5 g protein, 17.5 g fat, and 2.0 g fiber. The test meal was received well by nearly all participants. Participants less than 10 years of age were permitted to eat only half of the meal if they reported being too full to continue.

Signal collection and analysis

The EGG signal was captured into a Synectics Digitrapper or Polygraph (Synectics, Shoreview, MN), with a sampling rate of 1 Hz and low and high bandwidth filters set at 0.025 and 0.15 Hz. The digitized signal was downloaded into a personal computer.

The raw signal, plotted as a tracing of amplitude versus time, was then exported to analysis software created by Jiande Chen, Ph.D. (CGG version 1.0, copyright 1999). The raw signal was edited to remove motion artifacts identified by visual inspection. The edited signal was subjected to running spectral analysis, which reveals the frequency of the gastric slow wave and shows variations in its frequency over time. The CGG software conducts running spectral analysis by an adaptive, parametric method based on the autoregressive moving average model (23,24). The product of this analysis is a pseudo–three-dimensional plot of time and power (sharing the vertical axis) versus frequency. This relationship is plotted as a series of spectral lines, each representing a 1-minute epoch of the test period. Thus, even brief shifts in frequency within the test period are visible. A second report, the smoothed power spectral analysis, is a two-dimensional plot of power versus frequency for the test period as a whole. A typical set of pre-and postprandial running spectral analysis reports and corresponding smoothed power spectral analysis report are presented as Figures 1, 2, and 3.

FIG. 1.
FIG. 1.:
Running spectral analysis, preprandial period. Normogastric activity 80.30% of test time, bradygastric 7.58%, tachygastric 10.61%, arrhythmic 1.53%. IC (F) = 0.4239.
FIG. 2.
FIG. 2.:
Running spectral analysis, postprandial period. Normogastric activity 89.58% of test time, bradygastric 8.33%, tachygastric 2.08%, arrhythmic 0%. IC (F) = 0.2041. The increased density of spectral line peaks between 2 and 4 cpm visible on the postprandial report indicates a greater portion of test time spent in the normogastric range after meal consumption.
FIG. 3.
FIG. 3.:
Smoothed power spectral analysis report for the same participant, with pre-and postprandial periods plotted on the same set of axes. The higher postprandial power peak indicates increased dominant power, and that peak's location to the right of the preprandial peak indicates higher dominant frequency—typical responses to meal consumption. Dominant power rose from 31.27 to 38.82 dB, whereas dominant frequency shifted from 2.70 to 3.05 cpm from the pre-to the postprandial period.

Key parameters

Percentage of time in the normogastric range

The frequency of the gastric slow wave in humans is normally approximately 3 cycles per minute (cpm), but gastric myoelectrical activity may be observed at frequencies of 0.5 to 9 cpm (25). Frequencies of 2 to 4 cpm are considered normogastric, with slower waves (1–2 cpm) defined as bradygastric, and faster ones (4–9 cpm) tachygastric. Arrhythmic waves have no discernible dominant frequency within this range (22).

Based on the running spectral analysis report, it is possible to calculate the percentage of total test time spent in each of the frequency ranges. We have given particular attention to the percentage of time spent in the normogastric range, a measure of the regularity of gastric slow waves recorded by the EGG (26).

Power ratio

The amplitude (A) of the raw signal represents the potential difference, in microvolts, between the two recording electrodes on the antral axis. Power (P, reported in decibels), is defined by the formula, P = 10 × log 10A2 . The absolute value of EGG power is influenced by several factors (e.g., electrode placement, thickness of abdominal wall, skin preparation, and method of spectral analysis), so that comparisons among patients are not meaningful and only relative changes in power can be interpreted clinically. Dominant power (peak power) provides an estimation of the amplitude and regularity of the gastric slow wave. We have examined the ratio of postprandial to preprandial dominant power (26).

Dominant frequency

Dominant frequency is the frequency at which most of the signal power of the EGG is concentrated. It is displayed graphically as the x-axis coordinate of the power peak on the smoothed power spectral analysis report (26).

Instability coefficient of dominant frequency

The variable IC (F) is defined as the standard deviation for dominant frequency readings over the course of the test period divided by the mean of these readings. IC (F) indicates the degree to which the frequency of the gastric slow wave fluctuates (26).

Data analysis

Three technicians independently edited each test for motion artifact. To determine interobserver variability, their results were compared by analysis of variance. All data presented are average values ± standard deviation based on the three readers' edited tracings.

Paired t test was used to determine the significance of changes from the pre- to postprandial state within each group and overall. Regression analysis was used to determine the degree to which each variable depended on age and BMI. Based on BMI percentile for age, participants were categorized as normal weight, low weight, or obese. Nonpaired t test was used to compare results for normal weight and obese participants and for male and female participants.


Effect of meal


A statistically significant shift in dominant frequency, from 2.9 ± 0.40 to 3.1 ± 0.35 cpm, was observed after ingestion of the test meal (Table 2). IC (F) decreased from 0.34 ± 0.13 to 0.26 ± 0.12, indicating stabilization of the dominant frequency with meal consumption.

Effect of meal


The absolute value of EGG power is influenced by several factors (e.g., electrode placement, thickness of abdominal wall, skin preparation, and method of spectral analysis), so that comparisons among patients are not meaningful and only relative changes in power can be interpreted clinically. In our participant group, dominant power rose significantly, from 41.1 ± 9.91 to 47.1 ± 9.68 dB, in response to the meal challenge, indicating an increase in the intensity and regularity of the gastric slow wave after feeding and suggesting increased and more coordinated impulses for recruitment of muscle cells for peristalsis. Contractile activity may also contribute to postprandial power, and the approach of the wall of the more distended full stomach to the recording electrodes may amplify this effect somewhat. We consider distention to be a relatively minor factor in the postprandial power increase, based on the discussion of postprandial power by Smout et al. (3), the results of efforts by Riezzo et al. (27) to quantify distention effects, and the power increases observed by Chen et al. (28) and Stern et al. (29) in response to sham feeding.

Frequency distribution over time

The percentage of normal slow waves increased significantly with meal consumption. This shift in frequency distribution toward the normogastric range is consistent with the stabilization of frequency shown by the postprandial drop in IC (F).

Effect of age, gender, and body mass index

The dependence of power ratio (postprandial divided by preprandial in linear scale or postprandial minus preprandial in decibels) on age did reach statistical significance. However, a weak and inverse relationship is demonstrated between the two variables, for which no physiologic explanation suggests itself. A weak relationship also appears between preprandial dominant frequency and BMI and between percentage of time spent in the normogastric range and BMI. Again, we do not attach clinical meaning to these results, although they are marginally statistically significant. Obese participants as a group did not differ significantly from those of normal weight in any parameter. Effects of age, gender, and BMI on key parameters are presented in Table 3.

Effect of age and BMI


Effect of meal consumption

Our results, which show a significant change in EGG power, dominant frequency, stability of frequency (IC (F)), and frequency distribution in response to a meal, depart in some respects from those of a pediatric study by Riezzo et al. (27). We demonstrated a significant decrease in IC (F), as has been shown in adults, whereas Riezzo et al. observed a postprandial increase in this measure. In addition, we observed a significant increase in dominant frequency in the postprandial period, which Riezzo et al. did not find. Our participants showed a significant increase in the percentage of time spent in the normogastric range, an effect not seen in the study group of Riezzo et al. In these aspects, our results resemble those of Chen et al. (22) in a very small study across a range of ages and are similar to results found in healthy adults (22).

Age and gender

The independence of EGG results from age in our group of healthy children and adolescents, aged 6 to 18 years, confirms the findings of Riezzo et al. (27) in children 6 to 12 years old, again with the exception of IC (F) results. Riezzo et al. documented a significantly different postprandial IC (F) in girls more than 9 years old compared with other age or gender groups. We did not find this difference when patients were grouped by age and gender and compared by analysis of variance, and regression analysis showed no dependence of IC (F) on age. No significant differences could be found between boys and girls in any parameter.

The stability of the EGG over the age range studied suggests that maturation of gastric slow waves is complete early in life. In recent studies, other investigators have disagreed on whether an immature myoelectrical feed response can be demonstrated in small groups of preterm infants (30–32).

Body mass index

It may be suspected that high BMI would obstruct a reliable EGG recording, because a thicker abdominal wall would distance electrodes from the signal source. Our results led us to conclude that good EGG tracings can be obtained from patients with a high BMI. Moreover, pediatric norms do not need to be adjusted to account for BMI.

It is also reasonable to ask whether obese participants may be considered healthy or normal in terms of gastric myoelectrical function. Obesity was common in our study population. Seventeen of our 55 participants (31%) were above the age-adjusted 90th percentile for BMI. No difference was found between obese and normal weight patients in any EGG parameter. These results confirm the findings of Riezzo et al. (19,27), who also investigated EGG results in obese and normal weight pediatric and adult participants in two studies. These results suggest that gastric dysrhythmias are neither an underlying factor nor an accompanying symptom in pediatric obesity.


Low signal-to-noise ratios have been a persistent problem for EGG recording, despite improvements in equipment. Although a bandwidth filter dampens electromagnetic waves at frequencies below 1.5 and above 9 cpm, other electrophysiologic signals of nongastric origin have frequencies that overlap with the amplified range. These may compete with the gastric slow wave and disrupt the recorded tracing (25).

Motion, a common source of noise, creates artifacts that must be removed from the tracing to avoid distortion of data. Often, these appear as sudden, high-amplitude deflections in the signal, which are easily identified by eye and removed by hand and mouse. In other cases, however, these disruptions are subtle and can be difficult to distinguish from authentic fluctuations in the signal's character. Although automated neural network systems for the deletion of noise from the EGG signal are in development, none have been introduced for general use (33). The process for identifying and removing motion artifacts remains manual and subjective, introducing error and bias (34).

Our data for key parameters exhibit high standard deviations. Given the fact that these standard deviations are much higher across participants than across observers within a participant, it seems that variability is a true characteristic of test results, and not merely the result of variability in editing technique, the subjective aspect of the testing and analysis process.

Comparisons of values across centers, even for normal, healthy participants, cannot be made without considering differences in hardware settings, test conditions, and analysis methods. Some of these differences are fairly easy to understand and correct, but differences among the various computer applications designed to extract numerical data from the complex EGG tracing are not easily accessible to the clinician or physician-investigator. It is important to realize that these analysis methods may differ substantially in their underlying mathematics, and that these differences may influence EGG results. For example, the consistently lower percentage of normal slow waves and higher percentage of tachygastric slow waves in the study by Riezzo et al. (27) could have a variety of sources, among them a different expression of the EGG tracing in spectral analysis or a different algorithm for identifying the dominant peaks of the spectral lines.


Our normative data for children and adolescents in the 6- to 18-year age group are quite comparable with those described in some adult studies, with no significant impact from age, gender, or BMI (18). The expected response to a meal, demonstrated in our results and shown by others in adults, is an increase in normogastric slow wave activity, a decrease in the instability coefficient of the dominant frequency (IC (F)), and a rise in signal amplitude. The underlying physiologic correlates of these changes are not fully understood and are only partially explained by the “entrainment” of more muscle units at the time of gastric contractions and by the gastric distension resulting from meal consumption. Substantial deviation from normal ranges and lack of normal meal response have been shown to accompany gastrointestinal dysfunction. However, consensus on what percentage of time in the brady- or tachygastric range should be considered extreme or pathologic has not been reached, and the use of the words bradygastria and tachygastria as diagnostic terms in themselves should be avoided. Based on the results of our study and others, healthy children and adults usually spend some portion of test time outside the normogastric frequency range (17–19,22,26,35).

With standards still in development, clinicians who want to use the EGG as a reliable measure of electrical control activity should adhere to a well-established and validated protocol using hardware and software that has been tested systematically for the task. Normal values used for comparison with observed results should be based on that same equipment and protocol or should be self-developed. The effort presented in this paper also illustrates the importance of using a reproducible method of artifact removal before subjecting the signal to quantitative spectral analysis. Unfortunately, identification of artifacts remains a subjective exercise, although new approaches using neural network algorithms may offer a better alternative in the future.

Use of the EGG as a diagnostic tool remains an attractive but unsubstantiated proposition. Validation studies, longitudinal measurements in specific clinical settings, correlative tests with other indicators of motility (e.g., manometry or gastric emptying) will be needed to confirm its usefulness. Finally, attention to the technical details involved in the performance of the test and the subsequent signal analysis will enhance the reproducibility of the test, allowing meaningful comparisons among clinicians.


The authors thank John Lazar and Zev Davidovics for their contributions to the study as research assistants.


1. Alvarez WC. The electrogastrogram and what it shows. JAMA 1922; 78: 1116–9.
2. Familioni BO. Validity of the cutaneous electrocardiogram. In: Chen JZ, McCallum RW, eds. Electrogastrography Principles and Applications. New York: Raven Press, 1994; 103–25.
3. Smout AJ, van der Schee EJ, Grashuis JL. What is measured in electrogastrography? Dig Dis Sci 1980; 25: 179–87.
4. Sun WM, Smout A, Malbert C, et al. Relationship between surface electrogastrography and antropyloric pressures. Am J Physiol 1995; 268: G424–30.
5. Mintchev MP, Kingma YJ, Bowes KL. Accuracy of cutaneous recordings of gastric electrical activity. Gastroenterology 1993; 104: 1273–80.
6. Koch KL, Stern RM. Nausea and vomiting and gastric dysrhythmias. In: Chen JZ, McCallum RW, eds. Electrogastrography Principles and Applications. New York: Raven Press, 1994; 309–30.
7. Stern RM, Koch KL. Using the electrogastrogram to study motion sickness. In: Chen JZ, McCallum RW, eds. Electrogastrography Principles and Applications. New York: Raven Press, 1994; 199–218.
8. Leahy A, Besherdas K, Clayman C, et al. Abnormalities of the electrogastrogram in functional gastrointestinal disorders. Am J Gastroenterol 1999; 94: 1023–8.
9. Soykan I, Lin Z, Jones S, et al. Gastric myoelectrical activity, gastric emptying and correlations with dyspepsia symptoms in patients with gastroesophageal reflux. J Invest Med 1997; 45: 483–7.
10. Cucchiara S, Salvia G, Borrelli O, et al. Gastric electrical dysrhythmias and delayed gastric emptying in gastroesophageal reflux disease. Am J Gastroenterol 1997; 92: 1103–8.
11. Cucchiara S, Franzese A, Salvia G, et al. Gastric emptying delay and gastric electrical derangement in IDDM. Diabetes Care 1998; 21: 438–43.
12. Devane SP, Ravelli AM, Bisset WM, et al. Gastric antral dysrhythmias in children with chronic idiopathic intestinal pseudoobstruction. Gut 1992; 33: 1477–81.
13. Riezzo G, Chiloiro M, Guerra V, et al. Comparison of gastric electrical activity and gastric emptying in healthy and dyspeptic children. Dig Dis Sci 2000; 45: 517–24.
14. Levanon D, Zhang M, Orr WC, et al. Effects of meal volume and composition on gastric myoelectrical activity. Am J Physiol 1998; 274: G430–4.
15. Hu S, Saladino PJ, Jaworski BC, et al. The effects of engaging in mental tasks on the severity of motion sickness and gastric myoelectric activity in humans. In: Chen JZ, McCallum RW, eds. Electrogastrography Principles and Applications. New York: Raven Press, 1994; 219–25.
16. Orr WC, Crowell MD, Lin B, et al. Sleep and gastric function in irritable bowel syndrome: derailing the brain-gut axis. Gut 1997; 41: 390–3.
17. Parkman HP, Harris AD, Miller MA, et al. Influence of age, gender, and menstrual cycle on the normal electrogastrogram. Am J Gastroenterol 1996; 91: 127–33.
18. Pfaffenbach B, Adamek RJ, Kuhn K, et al. Electrogastrography in healthy subjects. Evaluation of normal values, influence of age and gender. Dig Dis Sci 1995; 40: 1445–50.
19. Riezzo G, Pezzolla F, Giorgio I. Effects of age and obesity on fasting gastric electrical activity in man: a cutaneous electrogastrographic study. Digestion 1991; 50: 176–81.
20. National Center for Health Statistics and National Center for Chronic Disease Prevention and Health Promotion. Girls: 2 to 20 years CDC US growth charts body mass index. Ross Products Division Abbott Laboratories Inc., Sept. 2000.
21. National Center for Health Statistics and National Center for Chronic Disease Prevention and Health Promotion. Boys: 2 to 20 years CDC US growth charts body mass index. Ross Products Division Abbott Laboratories Inc., Sept. 2000.
22. Chen JD, Co E, Liang J, et al. Patterns of gastric myoelectrical activity in human subjects of different ages. Am J Physiol 1997; 272: G1022–7.
23. Lin Z, Chen JZ. Comparison of three running spectral analysis methods. In: Chen JZ, McCallum RW, eds. Electrogastrography Principles and Applications. New York: Raven Press, 1994; 77–99.
24. Chen J, Vandewalle J, Sansen W, et al. Adaptive spectral analysis of cutaneous electrogastric signals using autoregressive moving average modelling. Med Biol Eng Comput 1990; 28: 531–6.
25. Kaiho T, Shimoyama I, Nakajima Y, et al. Gastric and non-gastric signals in electrogastrography. J Auton Nerv Syst 2000; 79: 60–6.
26. Chen JZ, McCallum RW. Electrogastrographic parameters and their clinical significance. In: Chen JZ, McCallum RW, eds. Electrogastrography Principles and Applications. New York: Raven Press, 1994; 45–73.
27. Riezzo G, Chiloiro M, Guerra V. Electrogastrography in healthy children: evaluation of normal values, influence of age, gender, and obesity. Dig Dis Sci 1998; 43: 1646–51.
28. Chen JD, Pan J, Orr WC. Role of sham feeding in postprandial changes of gastric myoelectrical activity. Dig Dis Sci 1996; 41: 1706–12.
29. Stern RM, Crawford HE, Stewart WR, et al. Sham feeding. Cephalic-vagal influences on gastric myoelectric activity. Dig Dis Sci 1989; 34: 521–7.
30. Liang J, Co E, Zhang M, et al. Development of gastric slow waves in preterm infants measured by electrogastrography. Am J Physiol 1998; 274: G503–8.
31. Riezzo G, Indrio F, Montagna O, et al. Gastric electrical activity and gastric emptying in term and preterm newborns. Neurogastroenterol Motil 2000; 12: 223–9.
32. Cucchiara S, Salvia G, Scarcella A, et al. Gestational maturation of electrical activity of the stomach. Dig Dis Sci 1999; 44: 2008–13.
33. Liang J, Cheung JY, Chen JD. Detection and deletion of motion artifacts in electrogastrogram using feature analysis and neural networks. Ann Biomed Eng 1997; 25: 850–7.
34. Verhagen MA, Van Schelven LJ, Samsom M, et al. Pitfalls in the analysis of electrogastrographic recordings. Gastroenterology 1999; 117: 453–60.
35. Di Lorenzo C. EGG like EKG: Are we there yet? J Pediatr Gastroenterol Nutr 2000; 30: 134–6.

Electrogastrography; Pediatric studies; Normal values; Body mass index

© 2001 Lippincott Williams & Wilkins, Inc.