Changes in the force-frequency relationship of the human quadriceps femoris muscle following electrically and voluntarily induced fatigue.

Neuromuscular electrical stimulatioin (NMES) is used for a variety of purposes, including the transcutaneous stimulation of skeletal muscle to produce muscle strengthening (ie, increases in peak torque output). [1-7] The force output of the targeted muscle during training has been shown to be positively correlated with the observed strength gains. [5] Therefore, stimulation protocols that allow skeletal muscle to produce the maximum force during training should be used.

A number of factors determine the force output of skeletal muscle during training with NMES. In addition to stimulation intensity, another important factor is the stimulation frequency. The stimulation frequency helps to determine the peak force produced (ie, the force-frequency relationship [FFR]) and the rate of force fatigue. [8-10] The FFR is a plot of the force that is produced (y-axis) when a muscle is stimulated with a range of frequencies (x-axis). At the lowest stimulation frequencies (eg, <5 pulses per second [pps] for the human quadriceps femoris muscle), individual twitches can be seen, with complete, or nearly complete, relaxation between pulses (Fig. 1). At slightly higher stimulation frequencies, complete relaxation of the twiches does not occur and the twitches begin to summate to produce greater force outputs. This summation produces the rising portion of the FFR (Fig. 2). The longer the contraction time of the twitch, the lower the frequency needed to produce summation and the further to the left the rising portion of the FFR lies. As the frequency continues to be increased, greater summation is observed and greater forces are observed. Eventual, maximum summation is observed, and the force output plateaus. The disadvantage of using high stimulation frequencies to produce maximum force is the rapid rate of fatigue that is observed when using high stimulation frequencies. [9] Therefore, when using NMES to strengthen muscle, to minimize fatigue, and to maximize the force output during training, it would appear reasonable to use the minimum frequency that produces the maximum force output. [11]

All studies that have used NMES to strengthen muscle have used a consistent stimulation frequency throughout each training session. Although the use of a consistent stimulation frequency presumes that the FFR remains constant during each training session, this does not appear to be the case. The contractile rate of skeletal muscle has been shown to slow during voluntarily induced fatigue in humans [12] and during electrically induced fatigue of cat single motor units. [13] It has been suggested that this contractile showing should shift the FFR toward lower frequencies (ie, the FFR should shift to the left) by producing greater summation at lower frequencies and allowing lower frequencies to produce the maximum available force. [14]

In addition to contractile slowing repetitive activation of skeletal muscle has been shown to produce a selective loss of force at low stimulation frequencies (ie, <20 pps). [15-18] That is, there is a greater attenuation of force when a muscle is stimulated with subtetanic frequencies that when the muscle is tetanically stimulated. This type of fatigue is termed "low frequency fatigue" (LFF). Low frequency fatigue, by producing a greater attenuation of force at the lower stimulation frequencies, would produce an apparent shift in the FFR to the right (see "Discussion" section for explanation). The shift in the FFR produced by LFF is in the opposite direction to that suggested to be produced by contractile slowing. Therefore, depending on the extent of contractile slowing and LFF, the actual shift in the curve may vary.

The purpose of this study was to identify the changes in the FFR following voluntarily and electrically elicited contractions using fatiguing protocols similar to those used to strengthen skeletal muscle. The NMES protocol used to elicit fatigue utilized a stimulation frequency (60 pps) similar to that used clinically to strengthen muscle. [1-7] Both the NMES and voluntary protocols used a contraction duration (8 seconds) similar to that used to strengthen muscle. To produce significant amounts of fatigue, however, the rest times between contractions (12 seconds) was markedly shortened compared with those of clinical protocols used to strengthen skeletal muscle. [1-7] Knowing how the FFR of the human quadriceps femoris mucle changes with intermittent fatiguing contraction may aid clinicians in selecting the stimulation frequency needed to maximally stimulate muscle with minimum fatigue

Method

Subjects

Twenty nondisabled subjects (10 female, 10 male) ranging in age from 20 to 37 years ([unkeyable] = 23.1, SD = 4.2) voluntarily participated in this study. All subjects signed informed consent forms and completed the study.

Procedure

All subjects participated in one training session and one experimental session to test the efforts of electrically induced fatigue. At least 72 hours separated these two sessions. Because of scheduling and other practical difficulties, only 10 of these subjects also participated in a second experimental session at least 1 week later, which tested the effects of voluntarily induced fatigue. Subjects were asked to refrain from any strenuous activity 24 hours prior to testing.

Subjects were seated on a computer-controlled isokinetic dynamometer, (*) which recorded the force generated during all contractions. Their left hips and knees were flexed to 70 and 90 degrees, respectively. The dynamometer axis was aligned with the knee joint axis (ie, lateral epicondyle of the femur). Each subject's leg, thigh, and pelvis were held securely by the force transducer pad, a thigh strap, and a belt across the pelvis. The bottom of the force transducer pad was placed 4 cm proximal to the lateral malleolus. Subjects were asked to keep their backs against the backrest during teting with their arms folded across the chest. The position of the seat, thigh support, and force transducer was recorded for each subject to ensure consistency across sessions.

A velocity setting of 0 [degrees]/s was used to allow only isometric contractions during all testing procedures. The force data were digitized by the dynamometer's computer with a sampling frequency of 100 samples per second.

A Grass S8800 stimulator, (+) SIU5 stimulus isolation unit, (+) and CCU1A constant-current unit (+) were used to deliver 47-mA, monophasic, square-wave pulses of 350 microseconds' duration to all subjects. Forty-seven mA was the maximum output the stimulator was capable of delivering and was well tolerated by all subjects. This intensity was used for all electrically elicited contractions durin the experimental session (ie, for FFR testing and to induce fatigue electrically). Each subject's left quadriceps femoris muscle was stimulated using two 10-x 10-cm carbon-rubber electrodes placed over sponges that were saturated with tap water. One electrode was placed distally over the vastus medialis muscle, and the other electrode was placed over the proximal third of the rectus femoris muscle (anterior proximal thigh). Exact placement was determined by shifting the electrodes to find the motor points in these areas. The electrode positions that produced the maximum force output during a brief train of pulses (frequency=60 pps, train duration=3 seconds) were identified as the motor points. Electrode placement was kept constant from session to session by outlining the electrodes on the subject's thigh with a marker. The electrodes were secured to the thigh by 10-cm-wide Velcro[R] (++) straps that were long enough to be wrapped twice around each subject's thigh.

Training session. During the training session, each subject was first trained to perform a maximal voluntary contraction (MVC). The twitch superimposition technique, as demonstrated by Rutherford et al, [19] was used to determine whether a true MVC was being performed. With this procedure, a single electrical pulse is delivered to the subject while the subject attempts to perform an MVC. If a true MVC is being produced, the pulse produces no measurable increase in force output. If less than a maximum force is being produced, a twitch is superimposed on the voluntary contraction. Subjects were also trained to relax during 1.5-second trains of NMES delivered to the quadriceps femoris muscle. Subjects were assumed to be relaxed if a flat baseline force was observed prior to stimulation and if there was a smooth and complete return to the baseline force immediately following the response to the last pulse of the train. The frequency of each successive train was randomly selected from the frequencies to be used during testing. Training was continued until each subject could perform an MVC and relax during stimulation. Adequate rest time (ie, at least 60 seconds) was allowed between contractions to minimize fatigue. All subjects were able to produce an MVC and relax within one training session.

Experimental sessions. During the experimental sessions, each subject first performed a 10-second MVC of the quadriceps femoris muscle to maximally potentiate the muscle. [20] Next, within 60 seconds following the MVC and with the subject relaxed, a single twitch was elicited. This twitch was later used to measure the contractile speed of the muscle. Ten seconds later, to determine the prefatigue FFR, delivery of a series of 20 1.5-second trains was begun. Each train was separated by approximately 10 seconds. The frequencies used were 1, 3, 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 90, and 100 pps. They were presented in a random order. After the prefatigue FFR was determined, the subjects were exercised to produce fatigue. Finally, within 60 seconds following the induction of fatigue, another twitch was elicited and the postfatigue FFR was determined using the same protocol (including the same random order of train frequencies) as that used prior to fatigue.

To produce fatigue by using electrically elicited contractions, 8-second trains at 60 pps were used. A 12-second rest separated each train. The specific train duration and frequency were chosen because, as previously noted, they approximated those used clinically to strengthen muscle and the rest time was brief enough to allow significant fatigue (>50% reduction) for all subjects. [21] Stimulation was continued until the peak force declined to just less than 50% of the initial electrically elicited contraction.

To test for voluntarily induced fatigue, an identical protocol was used, except that fatigue was included using 8-second MVCs. The subjects were given visual feedback of their force output in addition to verbal encouragement. To determine whether subjects were applying maximal effort, twitches were randomly superimposed on the contractions. If the twitches were observed to produce an increase in force, the tester (LRM) would conclude that a submaximal effort was being applied and encourage the subjects to push harder. Voluntary contractions were repeated until the peak force declined to just less than 50% of MVC of the subject being tested. The forces produced during the contractions performed to elicit fatigue (ie, both the 8-second electrically and voluntarily elicited contractions) were monitored but not recorded.

Data Management

and Analysis

The peak force produced at each stimulation frequency (ie, the FFR) and the twitch characteristics were determined using the dynamometer's software. All statistical tests were performed using a personal computer and a statistical software package. [*1] To compare the relative prefatigue and postfatigue FFRs, all data were normalized. Each subject's prefatigue FFR data were normalized by dividing the peak force produced at each frequency by the maximum peak force produced during the prefatigue FFR testing (this force was generally, but not always, produced during the 100-pps train). Similarly, each subject's postfatigue force data were normalized by dividing the peak force produced at each frequency by the maximum peak force produced during the postfatigue FFR testing. Normalized peak forces were plotted against frequency of stimulation to determine the FFR before and after fatigue. Normalized data were pooled across subjects, and groups means were determined.

The twitch contraction times, one-half relaxation times, ratio of the peak forces produced at 20 and 50 pps, and frequency needed to produce 50% of maximum normalized force were all calculated before and after fatigue. The twitch contraction time, defined as the time from the onset of the force to the peak of the twitch, and the one-half relaxation time, defined as the time to return to half the twitch force from the peak twitch force, are commonly used measures of contractile rates. [12,13] The ratio of the peak forces produced at 20 and 50 pps is a measure of the degree of LFF. [16,18] The frequency needed to produce 50% of maximum normalized force is often used as a simple measure to describe the position of the FFR. [21,22]

Two-way repeated-measures analyses of variance (ANOVAs) were used to compare the normalized peak forces produced at each frequency prior to and following fatigue (ie, separate ANOVAs were performed for the electrically and voluntarily induced fatigue). Similarly, a two-way repeated-measures ANOVA was used to compare the subjects' postfatigue data in response to electrically and voluntarily induced fatigue. Following the ANOVAs, if a significant frequency X treatment interaction was observed, Student's paired t tests were used as the method of post hoc testing to determine the specific frequencies at which differences existed between the treatments (eg, prefatigue versus postfatigue data). Paired t tests were also used to compare the prefatigue and postfatigue twitch characteristics, ratios of the peak forces produced at 20 and 50 pps, and frequencies needed to produce 50% of maximum normalized force in response to electrically and voluntarily induced fatigue and to compare the postfatigue electrically and voluntarily induced fatigue for the same variables. Significance was accepted at the .05 level.

Reliability of this procedure was determimed by calculating the intraclass correlation coefficient (ICC) [24] of the peak forces produced at each stimulation frequency for the two sets of prefatigue data collected from those subjects who participated in both the electrically induced fatigue and voluntarily induced fatigue sessions. In addition, paired t tests were used to compare these subjects' prefatigue data from each session for each of the variables described. The procedure used was shown to be highly reliable (ICC[2,1] = .8748), and none of the variables tested showed significant prefatigue differences.

Results

The observed FFR for the subjects' quadriceps femoris muscle prior to fatigue was nearly flat between 1 and 5 pps, increased steeply from 10 to 30 pps, and then slowly increased at higher frequencies (Fig. 2). The lowest frequency needed to produce near-maximum force (>90% of maximum) was approximately 55 pps before fatigue. The average maximum tetanic force (ie, the force produced at 100 pps [approximately 109 N]) was approximately 20% of the average MVC for all subjects, with no subject producing a tetanic force that was more than 50% of his or her MVC.

The fatiguing protocols (ie, the 8-second electrically and voluntarily induced contractions) were both able to produce the targeted level of fatigue (50% reduction in force) in all subjects tested. On average, 19 electrically induced contractions and 27 MVCs were required to produce the targeted level of fatigue. Comparison of the prefatigue and postfatigue FFRs for both electrically and voluntarily induced fatigue revealed that the force output decreased at all frequencies tested (Fig. 2). Figure 1 shows the force output over time for a single subject before and after fatigue for six of the frequencies tested.

The relative change in the force output at each frequency of stimulation was determined by comparing the normalized forces before and after fatigue. Figures 3 and 4 show that, consistent with the occurrence of LFE, both electrically and voluntarily induced fatigue produced greater attenuation of the normalized force at the lower frequencies than at the higher frequencies. Comparison of the normalized forces produced at 20 and 50 pps before and after fatigue revealed significant LFF with both electrically and voluntarily induced fatigue (Tabs. 1 and 2). The results of the two-way repeated-measures ANOVAs for the electrically and voluntarily induced fatigue data are shown in Tables 3 and 4, respectively. Significant (P<.001) treatmentXfrequency interactions were found for both treatments (ie, electrically and voluntarily induced fatigue). Table 5 shows that electrically induced fatigue produced significant differences in the

[TABULAR DATA OMITTED]

normalized force output for all frequencies tested between 1 and 55 pps (excluding 50 pps) and no significant differences for any frequency above 55 pps. The largest differences between the prefatigue and postfatigue data were between 10 and 35 pps. Voluntarily induced fatigue produced significant differences between 1 and 25 pps and produced the largest differences in frequency between 10 and 25 pps (Tab. 6).

Changes were also seen in the twitch characteristics with fatigue (Tabs. 1 and 2). Twitch contraction times decreased significantly, whereas one-half relaxation times increased significantly, with electrically induced fatigue (Tab. 1). With voluntarily induced fatigue, there were no significant changes in either twitch contraction times or one-half relaxation times (Tab. 2).

Figures 3 and 4, as well as the frequencies needed to produce 50% of maximum normalized force for electrically and voluntarily induced fatigue (Tabs. 1 and 2), show that the FFR did not shift to the left as suggested by Bigland-Ritchie et al, [14] but rather appeared to shift to the right. This shift was due to the observed LFF, which produced greater attenuation of force at the lower frequencies of stimulation. No significant differences were seen in the postfatigue data between electrically induced and voluntarily induced fatigue for any of the characteristics analyzed.

Discussion

The results of this research support and extend the findings of previous studies reporting LFF in human muscle [15,17] to now include training frequencies (in the case of the electrically induced fatigue at 60 pps) and contraction duration (for both fatiguing

[TABULAR DATA OMITTED]

protocols) often used to strengthen skeletal muscled. Cooper et al, [15] using intermittent trains of "programmed" stimulation at 1, 10, 20, 50, and 100 Hz (1 second each) to the human adductor pollicis muscle, and

Table 3. Two-Way Repeated-Measures Analysis-of-Variance Results for Electrically Induced Prefatigue Versus Postfatigue Force-Frequency Relationship

Source                 df    SS     MS     F (a)
Treatment               1    1.01   1.01    34.08
Error                  18    0.54   0.03
Frequency              19   56.34   2.96   468.23
Error                 342    2.17   0.01
Treatmentxfrequency    19    1.34   0.07    21.06
Error                 342    1.15   0.00
  (a) P<.001.

Table 4. Two-Way Repeated-Measures Analysis-of-Variance Results for Voluntarily Induced Prefatigue Versus Postfatigue Force-Frequency Relationship

Source                 df   SS      MS     F
Treatment              1     0.18   0.18     3.57 (a)
Error                   9    0.47   0.05
Frequency              19   30.49   1.60   219.21 (b)
Error                 171    1.25   0.01
Treatmentxfrequency    19    0.29   0.02     3.25 (b)
Error                 171    0.79   0.00
  (a) Not significant, P=.091.
  (b) P<.001.

Jones et al, [17] using brief MVCs of the human elbow flexors, found an LFF that produced an apparent shift in the FFR toward higher frequencies (ie, to the right). Neither of the fatiguing protocols used during these two previous studies was similar to the activation patterns used clinically to strengthen muscle with electrical stimulation. Though the rest period used in out study was selected to allow the production of fatigue with a reasonable number of contractions, it was much shorter that is commonly used for the muscle strengthening. [1-7] Nevertheless, this study demonstrated LFF using a clinically relevant training frequency and contraction duration.

Fatigue has been defined as a decrease in the force-generating ability of a muscle. [24] Several causes of fatigue have been suggested. Electrical stimulation of skeletal muscle has been shown to produce two distinct types of fatigue: high frequency fatigue (HFF) and LFF. [25] Edwards [25] defines HFF as a selective loss of force when a high stimulation frequency is used to produce the fatigue. Possible mechanisms for this phenomenon include impairment of neuromuscular transmission and propagation of the muscle action potential. [25] Low frequency fatigue is a selective loss of force at low stimulation frequencies that is long lasting (ie, hours to days). 15-18] It appears to result from an impairment of excitation-contraction

[TABULAR DATA OMITTED]

coupling. [15,18] This decrease in excitation-contraction coupling is not significant at high frequencies because the decrease in activation per stimulus is negligible as a result of a greater number of total action potentials. Thus, the implications of LFF on the FFR would be a decrease in force output at the lower frequencies and no change in force production at the higher frequencies. This study showed LFF to occur concurrently, with an overall attentuation of actual force recorded at all frequencies tested.

We propose that the apparent shift to the right in the FFR was produced by the LFF. Although a decrease in actual force output was observed with fatigue at all frequencies (Fig. 2), the postfatique data were normalized using the maximum postfatique tetanic force to determine the relative shape and position of the FFR. If the percentage of reduction in force was uniform across all frequencies, no shift in the FFR should have been observed; however, this was not the case. Both electrically and voluntarily induced fatique caused a greater depression of force at the frequencies than at the higher frequencies. This nonuniform attentuation of roce at lower stimulation frequencies caused the curve to appear to shift to the right with fatique.

In addition to LFF, the muscle's contractile properties (eg, twitch contraction time and one-half relaxation time) affect the FFR. The total duration of a muscle twitch should determine the minimum frequency (maximum interpulse interval) that allows successive twitches to summate and the frequency required for tetanic fusion. [23,26] An increase in the duration of the twitch, which can be evaluated by summing the twitch contraction time and the one-half relaxation time, should cause the FFR to shift to the left.

The findings of this study revealed a decrease in twitch contraction time and an increase in one-half relaxation time with electrically induced fatigue and no significant change with voluntarily induced fatigue. Previous studies have also investigated the changes in twitch characteristics with fatigue. Bigland-Ritchie et al [12] showed a 37% increase in one-half relaxation time and a 5% decline in twitch contraction time with voluntarily induced fatigue of the human adductor pollicis muscle produced by 60-second, continuous MVCs. Dubose et al, [13] however, found an increase in both twitch contraction time and one-half relaxation timein motor units of the cat medial gastrocnemius muscle using brief (325-millisecond), repetitive, electrically elicited contractions. The apparent inconsistencies among these findings could be due to the variation in the fatiguing protocols or the fatigue characteristics of the specific muscles used. Obviously, the contractile slowing (ie, increase in one-half relaxation time) observed in our study was not sufficient to compensate for the decreased twitch contraction time and LFF; therefore, fatigue produced a downward shift in the normalized forces at the lower frequencies of stimulation and an overall shift in the normalized FFR to the right.

The clinical significance of the results of this study are difficult to assess. The absence of differences in the stimulation frequencies needed to produce the greatest percentages of available force from the muscle (ie, >0.90 normalized force) after fatigue suggests that relatively high frequencies of stimulation (eg, [is greater than or equal to] 60 pps) are always most appropriate if maximum force generation is desired (eg, for muscle strengthening). Previous studies, [9,21] however, have shown that, during both sustained and intermittent contractions, decreasing the frequency of stimulation as a muscle begins to fatique actually reduces the amount of force decline observed. This finding argues for a progressive decline in stimulation frequency to minimize fatigue. Finally, if the clinician is attempting to produce less than maximum tetanic force, the results of our study suggest that marked increases in frequency (or intensity of current) may be necessary to produce a consistent force output from a muscle as it fatigues. For example, if we are attempting to produce 50% of miximum tetanic force, Figure 2 shows that, on average, approximately 14 pps would be needed prior to fatigue and 38 pps during fatigue. This represents an increase of nearly 175% in the frequency needed to produce the same tetanic force. Future studies are needed to determine the exact effect frequency may have on the ability of electrical stimulation to strength skeletal muscle and on the changes in the FFR of other human muscles with fatigue.

In addition to providing information regarding the changes in FFR with fatigue, this study also provides the clinician with data on the FFR in young, nondisabled subjects' quadriceps femoris muscles. To our knowledge, such information has not been published previously. Current work in our laboratory is attempting to identify the FFR in a sample of patients with marked atrophy.

Conclusion

This study has shown that both electrically and voluntarily induced fatigue affect the FFR in healthy, human quadricepts femoris muscle in a nonuniform manner. Both fatiguing protocols produced LFF, with the greatest attentuation of normalized forces at the lower frequencies of stimulation. These

[TABULAR DATA OMITTED]

results suggest that, depending on the specific therapeutic goal and level of tetanic force desired, the clinician may want to consider alteration of the stimulation frequency as a muscle begins to fatigue.

SA Binder-Macleod, PhD, PT, is Assistant Professor, Program in Physical Therapy, School of Life and Health Sciences, University of Delaware, 315 McKinly Laboratory, Newark, DE 19716 (USA). Address all correspondence to Dr Blinder-Macleod.

LR McDermond, BS, Pt, is Staff Physical Therapist, Rehabilitation Medicine, Johns Hopkins Hospital, Osler 1, 600 N Wolfe St, Baltimore, MD 21205. Ms McDermond was a student in the Program in Physical Therapy, School of Life and Health Sciences, University of Delaware, when this study was completed in partial fulfillment of the requirements for her Bachelor of Science in Physical Therapy degree.

This work was supported by the University of Delaware Honors Program, a Peter White Fellowship to Ms McDermond, and grants from the University of Delaware Research Foundation and the Foundation for Physical Therapy Inc to Dr Binder-Macleod.

This study was approved by the University of Delaware Human Subjects Review Board.

This article was submitted November 26, 1990, and was accepted September 9, 1991.

(++) Velcro USA Inc, 406 Brown AVe, Manchester, NH 03103.

(*1) SYSTAT Inc, 1800 Sherman Ave, Evanston, IL 60201-3973.

(*) Kin-Com(TM) II, Chattecx Corp, 101 Memorial Dr, PO Box 4287, Chattanooga, TN 34705.

(+) Grass Instrument Co, 101 Old Colony Ave, Quincy, MA 02169.

References

[1] Delitto A, Rose SJ, McKowen JM, et al. electrical stimulation versus voluntary exercise in strengthening thight musculature after anterior cruciate ligament surgery. Phys Ther. 1988;68:660-663.

[2] Milner-Brown HS, Miller RG. Muscle strengthening through electrical stimulation combined with low-resistance weights in patients with neuromuscular disorders. Arch Phys Med Rehabil. 1988;69:20-24.

[3] Mohr T, Carlson B, Sulentic C, Landry R. Comparison of isometric exercise and high volt galvanic stimulation on quadriceps femoris muscle strength. Phys Ther. 1985;65:606-612.

[4] Morrissey MC, Brewster CE, Shields CL, Brown M. The effects of electrical stimulation on the quadriceps during post-operative knee immobilization. Am J Sports Med. 1985;13:40-45.

[5] Selkowitz DM. Improvement in isometric strength of the quadriceps femoris muscle after training with electrical stimulation. Phys Ther. 1985;65:186-196.

[6] Soo C-L, Currier DP, Threlkeld AJ. Augmenting voluntary torque of healthy muscle by optimization of electrical stimulation. Phys Ther. 1988;68:333-337.

[7] Wigerstad-Lossing I, Grimby G, Honsson T, et al. Effects of electrical muscle stimulation combined with voluntary contractions after knee ligament surgery. Med Sci Sports Exerc. 1988;20:93-98.

[8] Binder-Macleo SA, Barker CB. Use of a catchlike property of human skeletal muscle to reduce fatigue. Muscle Nerve. 1991;14:850-857.

[9] Jones DA, Bigland-Ritchei B, Edwards RHT. Excitation frequency and muscle fatigue: mechanical responses during voluntary and stimulated contractions. Exp Neurol. 1979;64:401-413.

[10] Marsden CD, Meadows JC, Merton PA. "Muscular wisdom" that minimizes fatigue during prolonged effort in man: peak rates of motoneuron discharge and slowing of discharge during fatigue. Adv Neurol. 1983;39:169-211.

[11] Delitto A, Synder-Mackler L. Two theories of muscle strength augmentation using percutaneous electrical stimulation. Phys Ther. 1990;70:158-164.

[12] Bigland-Ritchie B, Johansson R, Lippold OCJ, Woods JJ. Contractile speed and EMG changes during fatigue of sustained maximal voluntary contractions. J Neurophysiol. 1983;50:313-324.

[13] Dubose L, Schelhorn TB, Clamann HP. Changes in contractile speed of cat motor units during activity. Muscle Nerve. 1987;10:744-752.

[14] Bigland-Ritchie B, Johansson R, Lippold OCJ, et al. Changes in motonenuron firing rates during sustained maximal voluntary contractions. J Physiol. (Lond). 1983;340:335-346.

[15] Cooper RG, Edwards RJT, Gibson H, Stokes, MJ. Human muscle fatigue: frequency dependence of excitation and force generation. J Physiol (Lond). 1988;397:585-599.

[16] Edwards RHT, Hill DK, Jones DA, Merton PA. Fatigue of long duration in human skeletal muscle after exercise. J Physiol (Lond). 1977;272:769-778.

[17] Jones DA, Newham DJ, Torgan C. Mechanical influences on long-lasting human muscle fatigue and delayed-onset plain. J Physiol (Lond). 1989;412:415-427.

[18] Stokes MJ, Edwards RHT, Cooper RG. Effect of low frequency fatigue on human muscle strength and fatigability during subsequent stimulated activity. Eur J Appl Physiol. 1989;59:278-283.

[19] Rutherford OM, Jones DA, Newham DJ. Clinical and experimental application of the twitch superimposition technique for the study of human muscle activation. J Neurol Neurosurg Psychiatry. 1986;49:1288-1291.

[20] Vandervoort AA, Quinlan J, McComas AJ. Twitch potentiation after voluntary contraction. Exp Neurol. 1983;81:141-152.

[21] Binder-Macleod SA, Guerin T. Preservation of force output through progressive reduction of stimulation frequency in human quadriceps femoris muscle. Phys Ther. 1990; 70:619-625.

[22] Botterman BR, Iwamoto GA, Gonyea WJ. Gradation of isometric tension by different activation rates in motor units of cat flexor carpi radialis muscle. J Neurophysiol. 1986; 56:494-506.

[23] Kernell D, Eerbeek O, Verhey BA. Relation between isometric force and stimulation rate in cat's hindlimb motor units of different twitch contraction times. Exp Brain Res. 1983;86:420-428.

[24] Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability. Psychol Bull. 1979;86:420-428.

[25] Edwards RHT. Human muscle function and fatigue. Ciba Found Symp. 1981; 82:1-18.

[26] Bigland-Ritchie B, Woods JJ. Changes in muscle contractile properties and neural control during muscular fatigue. Muscle Nerver. 1984;7:691-699.