The effects of selected stimulus waveforms on pulse and phase characteristics at sensory and motor thresholds.

Clinical use of transcutaneous electrical stimulation is frequently associated with the excitation of sensory and motor nerves. Commercial stimulators provide many different waveforms and a variety of pulse settings. These stimulators are used for treatment of numerous clinical conditions that require stimulation of various body sites.[1,2] Researchers have attempted to identify a preferred waveform during neuromuscular excitation.[3-9] Most of these researchers tested the comfort perception to different waveforms either at the highest tolerable intensity[3,5] or at fixed levels of muscle torque production.[4,6-9] Wong stimulated the forearm of healthy subjects and reported that a monophasic waveform was more comfortable and generated more torque than an asymmetric biphasic waveform. Bowman and Baker[6] found that most subjects preferred a symmetric biphasic waveform over a monophasic or asymmetric biphasic waveform for stimulation of the quadriceps femoris muscle. The upper and lower extremities, therefore, may respond differently and may require different waveforms to optimize comfort.[7-9]

Determining whether waveforms differently affect nerve excitation in the upper and lower limbs, or whether waveforms affect sensory and motor nerves differently, can be studied through testing of the minimum, threshold excitation values of the various waveforms. Use of invasive electrodes with animal models [10,11] or mathematical modeling[12,13] suggests that the depolarization and action potential propagation patterns are independent of waveform. Stimulus characteristics, however, vary appreciably with waveforms when human subjects are stimulated via surface electrodes at threshold levels. Kantor and colleagues[14,15] collected preliminary data from healthy subjects and reported that peak voltage, peak current, and total pulse charge all vary significantly between different waveforms. In contrast, they found that phase charge values were the same at threshold excitation irrespective of waveform.

Minimum levels of sensory and motor nerve stimulation are common in clinical practice. Thus, establishing minimum values of stimulus characteristics, at thresholds that excite peripheral nerves of both the upper and lower extremities, may help clinicians in setting these waveform levels. The data may also help to identify potential advantages and disadvantages of each waveform and may indicate whether there is a preferred waveform for excitation. In addition, the safety of stimulation may depend in part on stimulus waveform.[2]

The purpose of our investigation was to document the effect of five waveforms on four variables: peak current, peak voltage, phase charge, and total pulse charge during threshold excitation of peripheral nerves in the forearm arm and leg. The data were used to establish minimum values for nerve excitation and to compare (1) sensory and motor nerves and (2) upper- and lower-extremity nerves.

Method

Subjects

Sixteen female and two male subjects, all healthy volunteers with a mean age of 26.9% years (SD=7.5, range=19-45), participated in the study. All subjects stated they had no neuromuscular abnormality and that they had normal tactile and pressure perception.

Instrumentation

A stimulation and recording system was developed at the Center for Devices and Radiological Health (Rockville, Md) to generate, control, and record various waveforms (four pulsed and one sinusoidal) currently used in the clinic.[1,2] The system consisted of two constant-voltage signal generators (Wavetek model 175* and HP model 3314A [dagger])controlled by an HP model 5000 microcomputer [dagger] and an HP model 5180A waveform recorder [dagger] (Fig. 1). The accuracy of the waveform recorder was reported by the manufacturer to be [+ or -] 3% over the range of the voltages of the signal. Using a software program written by one of us (HSH), the system generated five different waveforms: monophasic (MP); symmetric biphasic (SBP) bursts of 10 or 25 symmetrical pulses (10 SP and 25 SP); and amplitude modulated (AM), also known as premodulated interferential current.[2] Figures 2 through 5 show typical plots for four of the five voltage waveforms and associated current waveforms. The phase duration and pulse repetition rate of all five waveforms were 200 microseconds and 50 pulses per second (pps), respectively. The 25-SP waveform is not shown as it was al@ most identical to the 10-SP waveform. During stimulation, the stimulus voltage and current waveforms were recorded and peak voltage, peak current, phase charge, and total pulse charge were determined by use of the computer. Phase charge (current integrated over phase duration, namely between two zero crossings) was calculated by the computer through digital integration using the formula [Q.sub.P] = It, where [Q.sub.P] represents the phase charge, I is the peak current, and t is the phase duration. The algorithm to calculate phase charge included (1) threshold detection of peak current at 0.3% of the peak, (2) 20 consecutive counts of peak current above threshold, and (3) summation of peak current over time until the signal dropped below threshold for 20 consecutive counts of peak current. Sampling of the signal was done every 2 microseconds. Total pulse charge was calculated as the sum of all phase charges contained in the pulse, irrespective of polarity (absolute values).

Procedure

Each subject signed a consent form, after which he or she assumed a sitting position, resting the right lower extremity on a chair with the knee in full extension and the foot supported in O degrees of dorsiflexion. The right upper extremity was positioned on an armrest with the shoulder in the anatomical position, the elbow at 90 degrees of flexion, the forearm in full pronation, and the hand and fingers fully supported yet relaxed on the armrest.

Two conductive polymer, self-adhesive surface electrodes (model 651-1842 [double dagger]), each 8.8 cm long and 3.8 cm wide, were used in the study. The right forearm of each subject was cleansed with water, and the proximal electrode was placed perpendicular to the segment on the dorsal surface over the extensor muscle group, 4 cm distal to the lateral epicondyle. The second electrode was placed 8 cm distal to the first electrode. During stimulation of the lower extremity, the same electrode application procedure was repeated over the dorsiflexor muscle group, but the placement of the proximal electrode was 10 cm distal to the tibial plateau. Determination of sensory nerve threshold excitation was achieved by gradual increase of stimulus amplitude at a rate of 0.2 V/s. At the instant the subject began to perceive tingling, the amplitude was held constant for a 30-second "on" period, then switched "off" for 15 seconds and then switched on again. If the subject perceived the stimulation throughout the two on periods, the peak voltage, peak current, phase charge, and total pulse charge were recorded. Peak current and phase charge were measured for the highest phase component of each waveform. The same procedure was repeated for motor threshold excitation by observing and/or palpating the initiation of minimal muscle contraction beneath the electrodes. Motor threshold was determined by the investigator, not the subject.

The sequence of testing the five waveforms was ordered as follows. Waveform presentation to subject I was selected at random and followed the order AM, MP, SBP, 10 SP, and 25 SP. Subject 2 started with MP followed by SBP, 10 SP, 25 SP, and AM, and so on. This sequential ordering ensured that each waveform was introduced as first, second, third, forth, or fifth the same number of tests and minimized the effect of novelty on threshold determination. The time between presentation of each waveform was 2 minutes.

Data Analysis

Three main factors were considered in this study: the five waveforms, the sensory and motor excitation thresholds, and the forearm and leg segments. The four measured variables were peak voltage, peak current, phase charge, and total pulse charge. Each of the four variables was subjected to a separate analysis of variance (ANOVA) for repeated measures on afl three factors. Significant F ratios ([alpha] =.05) were further analyzed by Newman-Keuls post hoc tests ([alpha] =.05) to identify significant differences among means. We examined all three factors to permit comparison of pulse characteristics at different excitation thresholds, and between the upper and lower extremities.

Results

All four ANOVAs indicated significant interaction among the three factors in each of the four measured variables (Table). The post hoc analysis of the interaction of 20 means of peak voltage age showed that the peak voltage of the MP and AM waveforms was higher than that of the SBP, 10-SP, and 25-SP waveforms during motor excitation of either the forearm or the leg. In contrast, the peak current of the AM waveform was lower than that of the other four waveforms in all test conditions. The peak currents of the SBP, MP, 10-SP, and 25-SP waveforms were not different.

The five different waveforms had little effect on the amount of phase charge required to excite the sensory nerves of the forearm. Likewise, the wave@ form did not affect the phase charge during sensory excitation of the leg or motor nerve excitation in the forearm or leg. Unlike the absence of a discernible change in phase charge, there was a marked increase in total pulse charge associated with the AM and 25-SP waveforms compared with the MP, SBP, and 10-SP waveforms. The differences were particularly noticeable during forearm and leg motor nerve excitation. The total pulse charge of the 10-SP waveform was higher than that of the SBP and MP waveforms, which did not differ from each other. The means and standard deviations of the four measured stimulus characteristics are presented in Figures 6 through

Discussion

In general and irrespective of waveforms, sensory nerve excitation required less stimulus magnitude of peak voltage, peak current, phase charge, and total pulse charge than did motor stimulation. Likewise, forearm stimulation required less magnitude of all four stimulus characteristics than did leg stimulation. In contrast, the sensory threshold in the forearm was elicited with an average of 18.4% (SD=4.0%) less stimulus peak voltage, peak current, phase charge, and total pulse charge compared with the sensory threshold in the leg. The mean difference for motor threshold was more pronounced, averaging 41.4% (SD=2.3%) more output for the measured variables in the leg than in the forearm.

Our data verified the clinical observation that many different waveforms can be used to excite peripheral nerves.[1,16-19] The results are in agreement with those of Johnston and Kasper,[10] who stimulated a frog nerve-muscle preparation and reported that all five studied waveforms induced very similar compound action potentials. Our data, however, demonstrate that the five waveforms had diverse effects on stimulus peak voltage, peak current, phase charge, and total pulse charge.

Peak Voltage and Current

Peak voltage and peak current were less variable than total pulse charge but more variable than phase charge. We could find neither published references nor reasonable explanations for peak voltages of the MP and AM waveforms being significantly higher than those of the SBP, 10-SP, and 25-SP waveforms during motor excitation. Conceivably, the MP and AM waveforms require higher voltages to generate the appropriate phase charges when a regulated voltage stimulator is used.

The peak current of the AM stimulus was significantly lower than the peaks of the other waveforms because it was the only one in which the phase duration of the current waveform was the same (200 microseconds) as that of the voltage waveform. This finding was due to the sinusoidal shape of the AM waveform. In the other four waveforms, the phase duration of the current waveform was much shorter (Figs. 2-4). The human body exhibits nonlinear, resistive-capacitive, frequency-dependent conduction characteristics.[12,20] The four square waveforms contained multiple frequencies, whereas the sinusoidal waveform had only one frequency (2,500 Hz). Consequently, the square waveforms exhibited a very fast discharge of the current, which resulted in a much shorter effective phase duration.[1] In accordance with the excitatory rule, which states that a shorter phase duration requires a higher peak current, our results are consistent with the expected relationship between phase duration and peak current amplitude.[12,21]

Comparing peak current between the MP and SBP waveforms showed that they are not different. Gorman and Mortimer[11] observed that the peak current of the SBP waveform was higher than that of a monophasic pulse, and they concluded that the biphasic waveform is less desired for nerve excitation. Our results did not support this conclusion. Gorman and Mortimer studied anesthetized cats, using surgically implanted electrodes, and they did not test the statistical significance of the reported differences. Their results should therefore be limited to the studied model.

Further examination of the differences in peak current indicated that the 10-SP and 25-SP waveforms required somewhat lower peak current than the SBP waveform. This result was also observed by Reilly and associates,[12] who reported a decreased threshold of excitation with increased number of sinusoidal cycles. Theoretically, the addition of more phases to the SBP waveform may facilitate the depolarization of the nerve membrane through temporal summation. But how many additional phases are needed to minimize the peak current? Mathematical modeling by Buetikoffer and Lawrence[13] suggested that three symmetrical pulses would be the optimal number in what they termed "quantal stimulation." Clinical verification of this modeling is currently not available.

Phase Charge

When evoking sensory and motor responses, phase charge was less variable than the other stimulus characteristics. Whereas there were differences between sensory and motor thresholds and between forearm and leg stimulation, there were no differences in phase charge between the five waveforms. These results are not in accord with the findings of Bowman and Baker,[6] who found dissimilar values of phase charge during motor stimulation using SBP and asymmetric biphasic waveforms. Unfortunately, Bowman and Baker did not test the statistical significance of those differences. Furthermore, we tested threshold excitation levels of the wrist and ankle extensors, whereas Bowman and Baker studied the quadriceps femoris muscles at moderately strong contraction levels. Whether our results are specific to the test conditions and muscles under investigation, or whether they can be extended to higher levels of excitation and different body regions, cannot be determined at present.

There are indications that phase charge is the least affected variable of the waveform during excitation of peripheral nerves at various levels of intensity. This conclusion was implied by Laquicque and Weiss, according to a review by Geddes.[20] We believe this is because phase charge is represented by the area under the curve of the phase.[2,4,6,7] The phase charge, therefore, reflects the attributes of phase duration, current amplitude, and stimulus shape. We suggest monitoring and reporting phase charge values as part of any research. Doing so will permit comparisons among studies that used different waveforms to stimulate peripheral nerves.

We believe that considering and reporting only peak current (or peak voltage) and phase duration (as is usually done) may not allow comparison of data concerning excitation of peripheral nerves. Not knowing the shape of the waveform and whether the stimulus is generated as constant voltage or constant current may provide for high variability of peak current and peak voltage. Because phase charge reflects not only phase duration and current amplitude but also stimulus shape, it is likely to be the most consistent and thus reproducible stimulus characteristic of the different waveforms. Furthermore, the repeatability of phase charge values has been demonstrated to be independent of three electrodes of significantly different conductivity,[22] and this repeatability held true whether constant-current or constant-voltage stimulators were used.[23] Hayden et al[24] added the dimension of time to phase charge reproducibility. They concluded that the phase charge quantities recorded during maximally tolerated stimulation of the lumbar erector spinae muscle were very consistent over 4 weeks of weekly stimulation.

There are limitations to our hypothesis on phase charge reproducibility. Our data show that the amount of phase charge varies between sensory and motor excitation, averaging 44.3% and 60.6% higher values during motor stimulation of the forearm and leg, respectively. Likewise, stimulation of the sensory and motor nerves in the forearm required a 34% lower average phase charge than for the leg nerves.

Alon et al[21] presented data on healthy subjects where the amount of phase charge necessary to excite sensory and motor nerves declined as phase duration became shorter. The amount of charge necessary to generate 1 N.m of elbow extension torque was fairly uniform between 20 and 200 microseconds and sharply increased with phase durations longer than 300 microseconds. We infer from these findings that the stability and reproducibility of phase charge can be expected as long as phase duration is approximately in the range of 20 to 200 microseconds.

Total Pulse Charge

The most dramatic differences were observed in total pulse charge. The threshold excitation with the SBP waveform was statistically no different than with the MP waveform. Further@ more, the mean differences of 18% to 21% were smaller than what could be extrapolated from the mathematical modeling of excitation thresholds published by Reilly et al.[12] The fact that total pulse charge did not differ statistically between the MP and SBP waveforms suggests that current mathematical models may not adequately address and thus may not accurately predict a multifactorial, in vivo human response. An improved model may require the consideration of a number of variables including electrode size, electrode location relative to the nerves, and method of threshold determination, all of which are inherently much less controlled when testing human subjects. Furthermore, the difference in total pulse charge between the MP and SBP waveforms might have been even smaller if the SBP waveform had an interphase interval.[2] Adding an interval of 75 to 100 microseconds was advocated by several investigators[6,11,12] but was not included in our study due to technical difficulty.

The 10-SP waveform's total pulse charge was only 8.6 to 9.5 times greater than that of the SBP waveform, whereas theoretically a tenfold increase would have been anticipated. The 25-SP waveform's total pulse charge was approximately 21 to 24 times greater and the AM waveform's total pulse charge was 25 to 31 times greater than that associated with the SBP waveform. These differences may be related in part to inaccuracy in determining the exact instant of sensory or motor threshold excitation. Our method is probably less sensitive and accurate than those that used direct measurements of nerve excitation.[10,12] Our approach, however, is practical and applicable to clinical settings, in which therapists usually resort to palpation and patient report rather than invasive techniques of recording directly from the nerves. Furthermore, our results are in agreement with those of Snyder-Mackler et al,[16] who noted, without measuring, a similar contractile response comparing the SBP waveform with 25-SP and AM waveforms. The authors reported, that the SBP waveform required much less total pulse charge to induce such contraction.

We believe our data support the hypothesis that greater total pulse charge does not necessarily add to a greater perception of stimulation or stronger muscle contraction. Further support for this view can be extrapolated from the study by Tracy et al,[25] in which the SBP and 25-SP waveforms equally augmented peripheral blood flow in the lower extremity. Once a nerve membrane is depolarized, most likely by the first few phases of the pulse,[11,13] additional charges that fall within the absolute or relative refractory period will not affect the excitation. The 10-SP, 25-SP, and AM waveforms, therefore, may deposit electrical energy that does not add to the excitation. Because total pulse charge multiplied by pulse frequency equals total current (in coulombs per second), the potential exists for the 25-SP and AM waveforms to reach total current levels, which may result in an adverse skin reaction, particularly when used with very small electrodes.[2]

Effects on Sensory and

Motor Nerves

We demonstrated that sensory and motor nerves respond in a seemingly identical way to transcutaneous stimulation and that varying the waveform has no meaningful influence on the responses. Sensory excitation always preceded motor excitation irrespective of waveform or stimulated sites, a finding that is in accord with the well-documented fact that motor nerve thresholds are higher.[1,21] We believe the reason for these higher thresholds is not necessarily related to a unique excitatory characteristic of the motor nerves. Rather, it seems to be associated with the fact that motor nerves are often situated deeper in the tissue than sensory nerves and therefore more stimulus intensity (phase charge) is required to cause their discharge.[12]

One practical implication of our findings is a need to reexamine the classification of stimulators into transcutaneous electrical nerve stimulators and neuromuscular electrical stimulators. Both types of devices excite the peripheral nerves in the same way; the only difference is that a higher phase charge is required for neuromuscular electrical stimulation than for transcutaneous electrical nerve stimulation.

Clinical implications

From a clinical perspective, the selection of a waveform should be based in part on criteria that include the ability to excite the target nerves with minimum electrical energy. All five waveforms were effective in exciting both sensory and motor nerves. The SBP waveform's peak current, peak voltage, and phase charge, with one exception, were equal to or lower than those of the MP, 10-SP, 25-SP, and AM waveforms. Yet, the total pulse charge of the SBP waveform was much smaller than that of the 10-SP, 25-SP, and AM waveforms. Based on these findings, we believe that the SBP waveform may be the preferred waveform for peripheral nerve stimulation. Not only does the SBP waveform minimize the total pulse charge, and thus the electrical energy involved in the stimulation, but it eliminates the potential skin irritation associated with monophasic pulses[1,2] and minimizes the discomfort of stimulation.[6,16]

The current Association for Advancement of Medical Instrumentation (AAMI) standard for transcutaneous electrical nerve stimulation has a safety requirement that maximum charge per pulse should not exceed 70 [micro] C.[26] In our threshold study, the 25-SP and AM waveforms reached respective means of 151.2 and 198.5 [micro] C of total pulse charge during motor stimulation. Because we used only very low levels of stimulation, we are unable to explain the limits advocated in the AAMI standard. We may speculate that the discrepancy is in terminology and the standard refers to phase rather than total pulse charge. Alternatively, the AAMI standard values are restricted to MP or biphasic waveforms and do not include pulses with multiple phases. As more and more waveforms are available in the clinic today, it seems that reevaluation of the AAMI standard is warranted.

Our results may also indicate that there is no need for multiple-waveform stimulators in the clinic. It seems that if the physiological objective of stimulation is to excite peripheral nerves, then one waveform should be enough to achieve the desired effect. The presence of redundant waveforms only complicates the decision-making process of the clinician and probably adds unnecessarily to the cost of the stimulator. Redundant waveforms also do not appear to have an advantage in the elicitation of strong muscle contractions.[12,19]

Conclusion

Within the limits of this study, we conclude that all five studied waveforms excited both sensory and motor nerve fibers in the forearm and leg. Phase charge was the most repeatable of the studied variables and should be reported if the objective of stimulation is to excite peripheral nerves. Adding phases to the SBP pulse by creating 10-SP, 25-SP, or AM wave@ forms does not enhance the excitation, although it increases the total pulse charge dramatically. We thus conclude that the SBP waveform may be the preferred waveform for excitation of peripheral nerves. We likewise conclude that irrespective of waveform, the motor threshold requires higher stimulus characteristics than the sensory threshold and the leg thresholds are higher than those of the forearm. The sources of these differences are not clear at the present time and deserve further study.

References

[1] Alon G, DeDomenico G. High Voltage Stimulation: An integrated Approach to Clinical Electrotherapy. Chattanooga, Tenn: Chattanooga Corp; 1987: chaps 7-11. [2] Alon G. Principles of electrical stimulation. In: Nelson RM, Currier DP, eds. Clinical Electrotherapy. East Norwalk, Conn: Appleton and Lange; 1987:29-80. [3] Wong RA. High voltage versus low voltage electrical stimulation: force of induced muscle contraction and perceived discomfort in healthy subjects. Phys Ther. 1986;66:1209-1214. [4] Gracanin F, Trinkcozy A. Optimal stimulus parameters for minimum pain in the chronic stimulation of innervated muscle. Arch Phys Med Rehabil. 1975;56:243-249. [5] Vodovnik L, Long C, Regenos E, et al. Pain response to different tetanizing currents. Arch Phys Med Rehabil. [6] Bowman BR, Baker LL. Effects of waveform parameters on comfort during transcutaneous neuromuscular electrical stimulation. Ann Biomed Eng. 1985;13:59-74. [7] Baker LL, Bowman BR, McNeal DR. Effects of waveform on comfort during neuromuscular electrical stimulation. Clin Orthop. 1988;233: 75-85. [8] Baker LL, Borup C, Mann M. Comparison of three waveforms for comfort of electrical stimulation in the upper extremity. Phys Ther. 1989;69:395. Abstract. [9] Baker LL, McNeal DR, Dewart KP, et al. Effects of carrier frequency on comfort with medium frequency electrical stimulation. Phys Ther. 1989;69:395. Abstract. [10] Johnston RM, Kasper S. Compound nerve action potentials produced by signals from clinical stimulators. Phys Ther. 1986;66:85. Abstract. [11] Gorman PH, Mortimer JT. The effect of stimulus parameters on the recruitment characteristics of direct nerve stimulation. IEEE Trans Biomed Eng. 1983;30:407-414. [12] Reilly JP, Freeman VT, Larkin WD. Sensory effects of transient electrical stimulation: evaluation with a neuroelectric model. IEEE Trans Bio Med Eng. 1985;32:1001-1011. [13] Buetikoffer R, Lawrence PD. Electrocutaneous nerve stimulation, II: stimulus waveform selection. IEEE Trans Biomed Eng. 1979;26: 69-75. [14] Kantor G, Alon G, Ho HS. Threshold excitation of peripheral nerves with human subjects. In: Proceedings of the 12th Annual International IEEE-EMBS Conference. 1989:1660-1661. [15] Kantor G, Alon G, Ho HS. Charges associated with threshold excitation of peripheral nerves using various waveforms. In: Proceedings of the 11th Annual international IEEE-EMBS Conference. [16] Snyder-Mackler L, Garrett M, Roberts M. A comparison of torque-generating capabilities of three different electrical stimulating currents. J Orthop Sports Phys Ther. 1987;8:297-301. [17] Barr JO, Nielsen DH, Soderberg GL. Transcutaneous electrical nerve stimulation characteristics for altering pain perception. Phys Ther. 1986;66:1515-1521. [18] DeDomenico G, Strauss GR. Maximum torque production in the quadriceps femoris muscle group using a variety of electrical stimulators. Australian Journal of Physiotherapy. 1986;32:51-56. [19] Reisman MA. A comparison of electric stimulators in electrical muscle contraction. Phys Ther. 1984;64:751. Abstract. [20] Geddes LA. A short history of the electrical stimulation of excitable tissue. Physiology. [21] Alon G, Allin J, Inbar GE. Optimization of pulse duration and pulse charge during transcutaneous electrical stimulation. Australian Journal of Physiotherapy. 1983;29:195-201. [22] Kantor G, Alon G, Ho HS. Intra- and inter-electrode charge distributions and their effects on threshold excitation of human peripheral motor nerves. In@ Proceedings of the 13th Annual International IEEE-EMBS Conference. [23] Kantor G, Alon G, Ho HS. Phase charge significance in peripheral nerve excitation with constant voltage and constant current stimulation. In: Proceedings of the 15th Annual International IEEE-EMBS Conference. 1993:1255-1256. [24] Hayden VL, Bathurst MC, Clark LM, et al. Pulse rate variation and its effects on maximal volitional contraction of the erector spinae muscle group. In: Proceedings of the North American Regional Meeting of the international Society for Electrophysiology and Kinesiology, Baltimore, MD. 1987:10. [25] Tracy JE, Currier DP, Threlkeld AJ. Comparison of selected pulse frequencies from two different electrical stimulators on blood flow in healthy subjects. Phys Ther. 1988;68: 1526-1532. [26] American National Standard for TENS (ANSI/AAMI NS4). American National Standards Institute/Association for Advancement of Medical Instrumentation. 1985.