Secondary Logo

Journal Logo

Original Article

Clinical validation of electromyography and acceleromyography as sensors for muscle relaxation

Hänzi, P.*; Leibundgut, D.*; Wessendorf, R.; Lauber, R.*; Zbinden, A. M.*

Author Information
European Journal of Anaesthesiology: October 2007 - Volume 24 - Issue 10 - p 882-888
doi: 10.1017/S0265021506002353



Mechanomyography (MMG) is generally used as gold standard to measure neuromuscular function [1]. However, its clinical use in daily practice is limited [2]. Electromyography (EMG) gives comparable results to MMG and can be used as well [1]. Subjective estimation of the extent of the train-of-four (TOF, T4/T1) fade is poor, especially when the TOF ratio exceeds 0.40-0.50 [3,4]. Manually measured counts cannot be used as a continuous input for a feedback-controlled system. However, an advantage of the TOF is the fact that no calibration is needed before the measurement [5]. This reduces induction time and avoids the problem of baseline shift. The disadvantage is the fact that it cannot be determined once there are only three counts left. For this reason, we used T1% (percentage of first twitch to initial twitch before relaxation) as reference value, which had to be calibrated before starting the measurements. A waiting period of 10 min was introduced to minimize the effect of the baseline drift.

Despite potential adverse effects of using neuromuscular blocking agents, there is no quantitative routine monitoring of neuromuscular function during anaesthesia [6] using MMG or EMG as these measurement techniques tend to be more time consuming. Clinical estimation of neuromuscular blockade may be biased [13]. Acceleromyography (AMG) was introduced in 1988 by Viby-Mogensen as a new method for monitoring neuromuscular function, fulfilling the basic requirements for a simple and reliable clinical monitoring tool [7]. It became popular because of its straightforward and easy application [8]. As a research instrument, it may be used after careful calibration [8,9]. However, AMG cannot be directly compared to MMG [10,11] or EMG with respect to the estimation of neuromuscular blockade and stimulating patterns [12]. Furthermore, as AMG works dynamically and MMG works isometrically, it is not possible to use the two methods simultaneously on the same arm. AMG has been investigated and comparisons have been made with EMG and MMG [1,11,13-18]; however, sensitivity and robustness to artefacts have not been investigated in AMG and EMG so far.

For delivering short-acting newer neuromuscular blocking agents, a continuous, accurate and reliable signal would be useful, specially if these agents should be given using automatic feedback control systems [19]. Input signals for control loops should be discrete and of a higher resolution, compared to a categorical signal such as a TOF value. They must give a rapid and true measurement of what should be controlled and should not be subjected to artefacts or to drifts as these are difficult to detect by an automatic control system.

The objective of this study was to answer the question if the AMG response to the ulnar nerve stimulation using a common clinically established device gives reliable, discrete input signal, which could be routinely used. We tested EMG against AMG using two manipulations: (a) the change of infusion rate (lowering the dose) and (b) the position change of the sensor hand, answering the fact that the two measurement methods quantify different effects but answer the same clinical question.

Materials and methods

The study was approved by the Ethics Committee of the District of Bern. Written informed consent was obtained from each patient. Fourteen patients (nine females and five males, mean age 41.7 yr (26-55), mean body mass index (BMI) 24.85 kg m−2 (17.3-32.5)), ASA Class I or II, who were scheduled for elective surgery, were included in the study. The patients were free of any known neuromuscular diseases, or renal and hepatic diseases, and did not take any drugs known to interfere with neuromuscular transmission.

Anaesthesia and the measurement of neuromuscular blockade was performed in accordance with the Good Clinical Research Practice (GCRP) in pharmacodynamic studies of neuromuscular blocking agents [1] to keep bias as low as possible. All patients were premedicated orally with 7.5 or 15 mg of midazolam or 1 mg of lorazepam 30-60 min before induction of anaesthesia. On arrival in the operating theatre, an intravenous (i.v.) catheter was inserted to administer fluids and drugs. Pulse oximetry, electrocardiogram and non-invasive blood pressure were monitored on the same arm. The other arm was used for neuromuscular monitoring. Bispectral index was routinely monitored. After preoxygenation, anaesthesia was introduced with bolus doses of propofol 2% 2-2.5 mg kg−1, fentanyl (0.1-0.2 mg) and remifentanil (1-2 μg kg−1). The trachea of the patients was intubated without the use of neuromuscular blocking agents. In case of problems or danger for the patient, succinylcholine 1 mg kg−1 was used for relaxation. Anaesthesia was maintained with an infusion of remifentanil (100-800 μg h−1), a target controlled infusion of propofol 2% 1.5-8 μg mL−1 and bolus doses of fentanyl (0.05-0.1 mg). Core temperature was monitored and maintained using forced air warming blankets (Bair Hugger™; Augustine Medical Inc., Eden Prairie, MN, USA).

The ulnar nerve was stimulated on the forearm with the TOF (four supramaximal square wave pulses of 0.2 ms duration and a frequency of 2 Hz). The recording electrodes for EMG were set up following the manufacturer guidelines at the adductor pollicis muscle. We compared EMG (Datex-Ohmeda AS/3; Helsinki, Finland) with AMG (TOF-Watch SX, Organon Teknika; Boxtel, The Netherlands), both devices being installed on the same stimulus-electrodes of the arm of the patient using an electrical switch. The electrodes (Ag/AgCl-ECG electrodes for children, recording diameter of 10 mm; REF 1008, Nessler Medizinaltechnik, Innsbruck, Austria) were stuck to the skin, which had been cleaned beforehand. Temperature of the stimulated hand was measured with a surface electrode (TOF-Watch SX) and kept constantly above 32°C [1]. The acceleration transducer was attached to the flexor-side of the thumb over the distal interphalangeal joint. The thumb was able to move freely while hand and arm were fixed with the splint from Organon-Teknika on a rigid board [10,20]. The counts (T1-T4), T1% (T1/Tref*100) and the TOF were continuously measured by computer, separately for each device (TOF-Watch SX-Monitor, software version 1.2 by Organon-Teknika), for TOF-Watch SX and Labview (National Instruments Corporation, Austin, TX, USA) and for the EMG-Monitor (Datex-Ohmenda, Helsinki, Finland). After induction, the AMG was calibrated according to the instructions of the manufacturer using its automatic start-up procedure. A period of at least 10 min was allowed for baseline drift of the nerval responses by using the 0.1 Hz mode (supramaximal stimuli with square wave pulses of 0.2 ms duration every 10 s). A second calibration of the AMG was performed in a similar fashion. EMG was set up and also calibrated according to the instructions of the manufacturer. Supramaximal current was determined separately for each device. The T1% of EMG was used as reference value for all measurements. After the stabilization phase and second calibration, both devices were started with a time difference of 30 s and then the TOF was measured every minute for each device.

Patients were normoventilated and end-tidal CO2 was kept constant. A syringe pump (Asena GH; Alaris Medical Systems, Basingstoke, Hampshire, UK) was used to apply a continuous infusion of mivacurium chloride (mivacron®; Glaxo Wellcome Gmbh & Co, Zeneca Gmbh) with an initial rate of 0.2-0.3 mg kg−1 h−1. Once T1% of the EMG was stable (criteria for stabilization of T1%: amplitude within 15%, during 15 min), the infusion rate of mivacurium was lowered to 0.05 or 0.1 mg kg−1 h−1. After this change, we waited again for signal stabilization using the above-mentioned criteria, then turned the sensor-hand by 90° from a vertical position with the thumb up to a palmar-side-down position (hand turn) (Fig. 1; first trace). Mean T1% over 5 min before lowering the mivacurium dose (=HD), mean T1% over 5 min before hand turn (=LDb) and mean T1% over 5 min after hand turn (=LDa) were used for comparing these events.

Figure 1.
Figure 1.:
Sample trial data traces. The vertical lines mark the events for lowering the mivacurium dose and the hand turn. The first trace shows the mivacurium infusion rate. The second, the T1% values for both EMG and AMG. The last two traces show the TOF ratio and the count values for EMG and the bottom one for AMG.

Statistical analysis

To measure the reaction to lowering the mivacurium dose, HD was compared with the LDb of each device. The effect of turning the hand was shown by comparing LDb with LDa. To compare AMG with EMG for both effects, lowering the dose or changing the hand position, the differences of the mean T1% before the event to the mean T1% after the event were compared using t-test (normality test passed) or U-test (normality test failed). To compare EMG and AMG for their potential to be used in situations with high neuromuscular blockade, no-twitch response measurements of one method were compared to twitch responses of the other, and vice versa. To compare the overall deviation of EMG-counts to AMG-counts, the L1-Norm value was used:

Δt = sampling time.

T = total time of observation of one period.

n = total number of measurements during observation period.

The differences were considered as statistically significant when P < 0.05 (F0.95).


Seventeen patients were selected to participate, of which 14 are presented in the results. Three patients dropped out because of surgical complications during operation, the duration of the operation or because of inadequate depth of relaxation with respect to the surgical demands. The average temperature of the sensor-hands was 33.7°C (31.5-35.6°C). The mean BMI was 24.85 kg m−2 (range: 17.3-32.5 kg m−2); three of the patients were obese (BMIs of 30.7, 31.2 and 32.5 kg m−2). Succinylcholine was used due to expected difficult intubation in three patients.

The stimulation power of both devices after searching for the supramaximal current for each sensor was significantly different (P < 0.001, using a U-test): the median current of AMG was 60 mA (range 55-60 mA) while the median current of EMG was 37.5 mA (range 23-64 mA), despite the fact that the same stimulating electrodes were used. The mivacurium dose was not lowered according to a scheme but based on the experience of the anaesthetist and depending on the phase of surgery. Figure 1 shows a sample recording of one experiment, with inaccurate high T1% values at the beginning of mivacurium infusion.

On lowering the dose (mean of 0.086 mg kg−1 h−1), both devices reacted comparably in the expected direction. Both EMG and AMG show a significant change of mean T1% between the phase before lowering the dose to the phase afterwards (P = 0.015 and P = 0.004, respectively, Table 1).

Table 1
Table 1:
Influence of lowering the mivacurium-infusion-rate (HD vs. LDb) and of hand turn (LDb vs. LDa) on EMG and AMG.

The hand turn disturbed the EMG signal (P = 0.863) much less than the AMG signal (P = 0.007) (Table 1). For control, we compared two phases under stable conditions that showed no significant differences (AMG: P = 0.953 and EMG: P = 0.972) (Fig. 2). The comparison of EMG and AMG showed no significant difference with respect to lowering the dose (P = 0.306) compared to turning the hand (P = 0.008) (Table 2). The difference of mean T1% before and after hand turn was significantly (P = 0.008) smaller for EMG (−0.26%) than for AMG (−10.01%) (Fig. 3).

Figure 2.
Figure 2.:
Comparison of two phases (each 5 min) under stable conditions for AMG (left) and EMG (right). The box plots show median (notched), 25% and 75% quartiles, range of data and outliers (for values beyond 1.5 times the inter quartile range of upper or lower quartile).
Table 2
Table 2:
Differences of T1% values of EMG and AMG at lowering mivacurium-infusion rate and at hand turn.
Figure 3.
Figure 3.:
Comparison of the two devices for hand turn. Mean T1% before the hand turn minus mean T1% after the turn. The box plots show median (notched), 25% and 75% quartiles, range of data and outliers (for values beyond 1.5 times the inter quartile range of upper or lower quartile).

In situations with high neuromuscular blockade, a comparison of the twitch response records between EMG and AMG is as follows: in measurements where EMG recorded no answers, AMG detected one or more count in 30% of the non-twitch response periods; whereas where AMG recorded no answers, EMG detected one or more counts in 51% of the non-twitch response periods (Fig. 4). The overall deviation of EMG counts to AMG counts calculated as mean L1 was 0.56 (0.47).

Figure 4.
Figure 4.:
No twitch response of EMG compared to the twitch response of AMG (shaded) and reverse (black). Number of twitch responses outside the valid range of 0 to 4 counts are summarized under ‘Invalid’.


This study compared the AMG and EMG sensors with respect to accuracy and artefact tolerance, with the objective of evaluating which sensor is more suitable to integrate in a feedback-controlled system for the application of muscle relaxants. AMG, in contrast to EMG, was more affected by external disturbances such as movement and was less sensitive at high degrees of neuromuscular blockade. The difference between the measurements of AMG and EMG may be caused by the fact that these devices do not measure the same physiological phenomenon [21]. AMG and EMG are preferred in the clinical routine to MMG because they are easier to use.

We followed the guidelines for GCRP in pharmacodynamic studies of neuromuscular blocking agents [1]. We allowed a period of 10 min for stabilization of the baseline, as according to earlier studies the most significant drift occurs during this time [22]. TOF ratios may go up to 1.3 without any neuromuscular blocking agent on board as also reported in other studies [23], even though the thumb was able to move freely with the hand totally fixed to a board. Non-relaxed thumbs might not return to the starting point and might not move in one direction only. Previous studies using AMG and EMG showed that the measurement of neuromuscular blockade cannot be compared between both arms of one individual [10]. For that reason, the patient's response to neuromuscular stimulation was determined on the same arm, stimulating with the same electrodes for both devices. To take into consideration that EMG and AMG measure different physical effects, supramaximal stimulation current is different as well. So far, no further data on supramaximal stimulation current while comparing different methods have been published. AMG and EMG recorded on the same limb are published by Kopman and colleagues [24] and showed an overestimation of AMG TOF values compared to EMG TOF.

In operations where full relaxation is required, such as brain or ophthalmic surgery, any sudden movement by the patients may result in complications. It is therefore essential to use a system that provides reliable and highly sensitive measurements. The L1-norm showed a difference between the two devices during the observed periods of half a count: EMG sensors showed higher sensitivity and seemed to be more reliable.

In conclusion, these findings suggest that in situations where reliable monitoring is essential such as in cases where short acting drugs are applied, feedback control is used, or in especially critical operations, EMG should be given preference over AMG despite the fact that its installation is more time-consuming. EMG, as a more accurate method, is preferable for research, while AMG, as a simpler and stable method, is preferable for routine practice.


Stiftung zur Foerderung der wissenschaftlichen Forschung an der Universitaet Bern.


1. Viby-Mogensen J, Engbaek J, Eriksson LI et al. Good clinical research practice (GCRP) in pharmacodynamic studies of neuromuscular blocking agents. Acta Anaesthesiol Scand 1996; 40: 59-74.
2. Dahaba AA, von Klobucar F, Rehak PH. The neuromuscular transmission module vs. the relaxometer mechanomyograph for neuromuscular block monitoring. Anesth Analg 2002; 94: 591-596.
3. Viby-Mogensen J, Jensen NH, Engbaek J, Ording H, Skovgaard LT, Chraemmer-Jorgensen B. Tactile and visual evaluation of the response to train-of-four nerve stimulation. Anesthesiology 1985; 63: 440-443.
4. Kopman AF, Mallhi MU, Justo MD, Rodricks P, Neuman GG. Antagonism of mivacurium-induced neuromuscular blockade in humans. Edrophonium dose requirements at threshold train-of-four count of 4. Anesthesiology 1994; 81: 1394-1400.
5. Kopman AF, Klewicka MM, Neuman GG. The relationship between train-of-four fade and single twitch depression. Anesthesiology 2002; 96: 583-587.
6. Hayes AH, Mirakhur RK, Breslin DS, Reid JE, McCourt KC. Postoperative residual block after intermediate-acting neuromuscular blocking drugs. Anaesthesia 2001; 56: 312-318.
7. Viby-Mogensen J, Jensen E, Werner M, Nielsen HK. Measurement of acceleration: a new method of monitoring neuromuscular function. Acta Anaesthesiol Scand 1988; 32: 45-48.
8. Kopman AF. Measurement and monitoring of neuromuscular blockade. Curr opin Anesthesiol 2002; 15: 415-420.
9. Viegas O, Kopman AF, Klevicka MM. An open label, parallel group, comparative randomized multicenter trial to compare the time course of the neuromuscular effects and safety of Raplon (rapacuronium bromide) for injection and mivacurium in adults (abstract). Anesth Analg 2001; 92: 211.
10. Kirkegaard-Nielsen H, Helbo-Hansen HS, Lindholm P, Pedersen HS, Severinsen IK, Schmidt MB. New equipment for neuromuscular transmission monitoring: a comparison of the TOF-Guard with the myograph 2000. J Clin Monit Comput 1998; 14: 19-27.
11. Dahaba AA, Rehak PH, List WF. Assessment of accelerography with the TOF-GUARD: a comparison with electromyography. Eur J Anaesth 1997; 14: 623-629.
12. Nakata Y, Goto T, Saito H et al. Comparison of acceleromyography and electromyography in vecuronium-induced neuromuscular blockade with xenon or sevoflurane anaesthesia. J Clin Anesth 1998; 10: 200-203.
13. Engbaek J, Mortensen CR. Monitoring of neuromuscular transmission. Ann Acad Med Singapore 1994; 23: 558-565.
14. May OP, Kirkegaard-Nielsen H, Werner MU. The acceleration transducer - an assessment of its precision in comparison with a force displacement transducer. Acta Anesthesiol Scand 1988; 32: 239-243.
15. Werner MU, Kirkegaard-Nielsen H, May O, Djernes M. Assessment of neuromuscular transmission by the evoked acceleration response (an evaluation of the accuracy of the acceleration transducer in comparison with a force displacement transducer). Acta Anesthesiol Scand 1988; 32: 395-400.
16. Harper NJN, Martlew R, Strang T, Wallace M. Monitoring neuromuscular block by acceleromyography: comparison of the Mini-Accelerograph with the Myograph 2000. Brit J Anaesth 1994; 72: 411-414.
17. Lepage JY, Malinovski JM, Lechevalier T, Cozian A, Pinaud M. Neuromuscular junction: neuromuscular transmission analyser: mechanomyography vs. acceleromyography. Anesthesiology 1995; 83: A891.
18. Loan PB, Paxton LD, Mirakhur RK et al. The TOF-Guard neuromuscular transmission monitor. A comparison with the Myograph 2000. Anaesthesia 1995; 50: 699-702.
19. Lendl M, Schwarz UH, Romeiser HJ, Unbehauen R, Georgieff M, Geldner GF. Nonlinear modul-based predictive control of non-depolarizing muscle relaxants using neural networks. J Clin Monit Comput 1999; 15: 271-278.
20. Dubois PE, Broka SM, Jamart J, Joucken KL. TOF-tube, a new protection for acceleromyography, compared with the TOF-Guard/TOF-watch arm board. Acta Anaesth Belg 2002; 53(1): 33-38.
21. Elorbany M, Wafai Y. Electromyography and acceleromyography do not measure the same physiological event. Brit J Anaesth 2001; 86: 737-738.
22. Meretoja OA, Brown TCK. Drift of the evoked thenar EMG-signal. Anesthesiology 1989; 71: A825.
23. Kopman AF, Klewicka MM, Neuman GG. The relationship between train-of-four fade and single twitch depression. Anesthesiology 2002; 96: 583-587.
24. Kopman AF, Chin W, Cyriac J. Acceleromyography vs. electromyography: an ipsilateral comparison of the indirectly evoked neuromuscular response to train-of-four stimulation. Acta Anesthesiol Scand 2005; 49: 316-322.


© 2007 European Society of Anaesthesiology