ABSTRACT: Accurate measurement of muscle strength and voluntarymuscle activation is important in the assessment of disorders that affect themotor pathways or muscle. We designed a multipurpose system to assessthe variability and reproducibility of isometric torque measurements obtainedduring maximal voluntary efforts of the knee flexor, knee extensor, ankledorsiflexor, and ankle plantarflexor muscles on each side. It used twoisometric myographs mounted on an adjustable frame. Measurements ofmaximal voluntary torque (range, 25–188 Nm) displayed low variabilitywithin a testing session and over five testing sessions (coefficient of variationrange, 5–11%). We used the same equipment to measure voluntary activa-tion of the triceps surae muscles. Voluntary activation, measured with asensitive twitch interpolation method, increased with increasing voluntarycontraction torque (P � 0.001) and was very high during maximal efforts(mean, 97.8 � 2.1%; median, 98.5%). Furthermore, measurements of vol-untary activation during maximal efforts were reproducible across testingsessions with very little variability (coefficient of variation, �2%). The myo-graph system and the testing procedures should allow accurate measure-ment of strength and voluntary drive in longitudinal patient studies.

Muscle Nerve 29: 834–842, 2004

MEASUREMENT AND REPRODUCIBILITY OFSTRENGTH AND VOLUNTARY ACTIVATIONOF LOWER-LIMB MUSCLES

GABRIELLE TODD, BSc,1,2 ROBERT B. GORMAN, BE,1,2 and SIMON C. GANDEVIA, MD, DSc1,2

1 Prince of Wales Medical Research Institute, Barker St., Randwick, New South Wales 2031, Australia2 University of New South Wales, Sydney, New South Wales, Australia

Accepted 9 January 2004

Weakness and excessive fatigue are important indi-cators of disease progression in disorders that affectthe motor pathways or muscle.25 Therefore, it isimportant that measurements of maximal voluntarycontraction (MVC) in these patients are accurateand reliable over time. Strength may be assessedunder controlled conditions with a myograph or bymeasurement of the largest weight that can be lifted.Strength is also widely assessed subjectively by man-ual muscle testing against resistance (e.g., the Med-ical Research Council grading scale). Although thisis the simplest method, requiring no equipment, it isnot desirable for long-term assessment, because itdoes not detect small changes in strength and relieson the clinician’s memory.

Hand-held dynamometers are easily and widelyused clinically for measurement of maximal force forthe knee flexor and extensor and ankle dorsiflexormuscles. Such measurements of maximal force forthese muscle groups are apparently reproduciblewithin a testing session, with 7% variation betweenmeasurements.37 However, reproducibility of maxi-mal force measurements made on separate days islower,31 perhaps due to difficulty with standardiza-tion of the testing position and thus of muscle lengthand with the examiner’s ability to hold the dyna-mometer stationary. Others reported a high test–retest reproducibility of maximal force measure-ments of the knee extensor and flexor and ankledorsiflexor muscles with the use of a fixed springgauge.24 However, fewer data are available on thevariability of maximal isometric force measurementsfor the plantarflexor muscles, because of difficultiesassociated with stabilization of the leg.

Measurement of force around the knee and an-kle joint normally requires two separate pieces ofequipment. To provide a single system for longitu-dinal measurement of lower-limb strength, we devel-oped a system whereby force generated by the flexor

Abbreviations: ANOVA, analysis of variance; CV, coefficient of variation;ICC2,1, intraclass correlation coefficient; MVC, maximal voluntary contractionKey words: maximal voluntary torque; reproducibility; triceps surae; variabil-ity; voluntary activationCorrespondence to: S. C. Gandevia at the Prince of Wales Medical Re-search Institute; e-mail: S.Gandevia@unsw.edu.au

© 2004 Wiley Periodicals, Inc.Published online 24 March 2004 in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/mus.20027

834 Strength and Voluntary Activation MUSCLE & NERVE June 2004

and extensor muscles of the knee and ankle jointscan be assessed by two isometric myographsmounted in the same piece of equipment. Theequipment is highly adjustable, compact, and easy touse with patients. We assessed the reproducibilityand variability of maximal strength measures ob-tained with the new system from the four musclegroups tested. The system was also designed to allowmeasurement of voluntary activation during contrac-tions of the plantarflexor muscles using a sensitiveform of twitch interpolation.19 This is technicallydifficult, because of instability of the leg at highforces.

Voluntary activation refers to the level of neuraldrive to muscle during exercise18 and is commonlyestimated by interpolation of a single supramaximalelectrical stimulus to the motor nerve during an iso-metric voluntary contraction (twitch interpolation27).The interpolated stimulus evokes a superimposedtwitch, the amplitude of which decreases with increas-ing voluntary force.4,8,27 During maximal efforts, thepresence of a superimposed twitch implies that thestimulated axons were not all recruited voluntarily orwere discharging at subtetanic rates and hence thatvoluntary activation is incomplete.8,17,20,27,34 Impair-ments of voluntary activation occur in patients who areunable to drive motoneurons sufficiently. This may

occur with impairment of corticospinal drive, as instroke30; lower motoneuron disorders, such as priorpoliomyelitis3; and with muscle pain and joint pathol-ogy.32

Longitudinal assessment of voluntary activa-tion is uncommon. Similar voluntary activationscores between testing sessions are reported forhealthy subjects1,7,22 and for patients with priorpoliomyelitis when tested up to 3 years apart.2 Inthis study, we measured voluntary activation oftriceps surae and assessed the reproducibility andvariability of voluntary activation scores for thismuscle.

MATERIALS AND METHODS

Maximal isometric torque was assessed, for the leftand right leg separately, with a compact new myo-graph in four muscle groups of the lower limb: theknee flexors and extensors, and the ankle dorsiflex-ors and plantarflexors. For the ankle plantarflexormuscles, voluntary activation was also assessed dur-ing submaximal and maximal contractions usingtwitch interpolation. Healthy subjects sat uprightwith the hip, knee and ankle flexed at 90° (Fig. 1A).All experimental procedures were undertaken with

FIGURE 1. Experimental apparatus used for measurement of flexion and extension torque around the right knee (A) and ankle joint (B).To test muscles on the other side, the platform was slid to the left. Voluntary muscle activation was calculated for the triceps surae muscleby measurement of the torque responses to electrical stimulation over the muscle. Double-ended arrows indicate components of theapparatus that can be repositioned. (Abbreviations: DF, dorsiflexion; PF, plantarflexion.)

Strength and Voluntary Activation MUSCLE & NERVE June 2004 835

approval of the institutional ethics committee, andinformed consent was obtained.

Torque Recordings. Maximal isometric force (kg)was measured for the knee flexors and extensorsusing a linear strain gauge (Xtran, Melbourne, Aus-tralia; linear to 1 kN). In each subject, force wasconverted to torque (Nm) by measurement of thedistance between the axis of joint rotation and thepoint of application of the strain gauge. A secondlinear strain gauge was used to measure maximalisometric torque about the ankle joint. The axis ofrotation of the ankle was colinear with the axis ofrotation of the platform. Torque signals were sam-pled at 200 Hz for later analysis using a data acqui-sition system (CED 1401 interface with Spike 2 soft-ware; Cambridge Electronic Design, Cambridge,UK). For knee flexion and extension, force was ap-plied to the strain gauge, 20 cm above the sole of thefoot, via a strap and rigid bracket. Subjects pushed orpulled against it for knee extension or flexion, re-spectively. To minimize friction, a movable perspexplate was placed between the sole of the foot and theunderlying platform of the myograph.

For ankle dorsiflexion and plantarflexion, theplate was secured to the platform and a strain gaugewas positioned under it, 16 cm horizontally from theaxis of joint rotation (Fig. 1B). For measurement ofdorsiflexion torque, subjects pulled against a wideadjustable strap that secured the foot to the under-lying platform. To minimize elevation of the heelduring ankle plantarflexion, a wide strap ran verti-cally over the top of the knee and was secured to theunderlying platform. Tension in the strap was ad-justed with the use of a block-and-tackle pulley sys-tem with a 5-to-1 mechanical advantage. The subjectattempted to plantarflex the ankle and tighten thestrap over the knee. (A detailed drawing of the ap-paratus is available from the authors.)

The apparatus containing the knee and anklestrain gauges was rapidly adjustable along aluminumstructural extrusions to enable correct positioning ofthe lower limb and measurement of torque fromeither leg. For knee flexion and extension, the lowerlimb was positioned with the lateral malleolus andhead of the fibula vertically aligned and at 90° to theunderlying platform. For ankle dorsiflexion andplantarflexion, the lateral malleolus was verticallyaligned with the axis of rotation and head of thefibula.

Stimulation. For twitch interpolation, a computertriggered a single electrical stimulus of 1-ms dura-tion (constant current, Digitimer DS7AH; Digitimer

Ltd., Welwyn Garden City, UK; up to 1 A) to intra-muscular nerves of the triceps surae. Pilot experi-ments were conducted to locate the best electrodeposition to evoke a resting twitch from the tricepssurae. As a result, the cathode and anode (Ag–AgCl,10 mm in diameter) were positioned along a verticalline between the horizontal midpoint of the poste-rior crease of the knee and the Achilles tendon. Thecathode was positioned at the point where the cir-cumference of the calf was greatest, approximately atthe lower third of the gastrocnemius muscle. Theanode was positioned 8 cm inferiorly. We chose notto stimulate over the tibial nerve because of spreadof the stimulus to the common peroneal nerve,which innervates antagonist dorsiflexor muscles.The stimulus intensity was set 10% above the levelrequired to evoke a resting muscle twitch of maximalamplitude (132–385 mA).

Protocol. Subjects participated in one or more stud-ies. The aim of the first study was to assess reproduc-ibility and variability of maximal torque measure-ments made within one testing session and betweentesting sessions. Subjects participated in five identi-cal testing sessions performed approximately 1 weekapart (three women and three men; age, 34 � 9years [mean � SD]). In each testing session, subjectsperformed five brief MVCs (2–3 s in duration) of theright and left knee flexor, knee extensor, ankle dor-siflexor, and ankle plantarflexor muscles, separately.The order of testing of the four muscle groups wasthe same for each of the five testing sessions.

In the second study, the variability and reproduc-ibility of estimates of voluntary activation of the tri-ceps surae muscle were assessed during maximalcontractions performed within a testing session andbetween sessions. The triceps surae muscle was se-lected because of its critical role in stance and loco-motion and because it is approximately three timesstronger than the antagonist ankle dorsiflexor mus-cles,35 so any current spread would have only a smalleffect on the evoked torque. Subjects participated infive identical testing sessions, which involved 12 briefMVCs (2–3 s) of the ankle plantarflexor muscles(three women and one man; age, 31 � 10 years).The first two MVCs were not included in the analysis.Each session was performed on a separate day. Dur-ing each contraction, a single supramaximal electri-cal stimulus was delivered over the triceps suraemuscle. This was followed by another stimulus 5 slater while at rest.

In the third study we assessed voluntary activationof the ankle plantarflexor muscles at different levelsof voluntary torque. Subjects (three women and two

836 Strength and Voluntary Activation MUSCLE & NERVE June 2004

men; age, 36 � 9 years) performed between 11 and23 pairs of contractions in the same session. Eachpair involved a brief maximal effort followed �8 slater by a brief submaximal contraction of varyingstrength. A single supramaximal electrical stimuluswas delivered to the triceps surae muscle duringeach contraction and while at rest between the twocontractions.4

In all three studies, maximal contractions wereperformed at least 1 min apart, to avoid fatigue, andcare was taken to ensure that the knee and anklejoints remained correctly positioned in the appara-tus.

For measurement of torque during each MVC,we followed the criteria suggested by Gandevia.17 Allmaximal efforts were accompanied by a consistentset of instructions. Subjects were instructed to “braceand prepare to push” or “pull back” against thebracket positioned on the lower leg, for knee exten-sion or flexion, respectively. In some knee extensioncontractions, recruitment of the plantarflexor mus-cles occurred, and this may move the tibia forwardagainst the myograph, thereby increasing the mea-sured knee extension torque. To ensure that thecontribution of the plantarflexor muscles to kneeextension torque was minimal, we observed the strat-egy used during each knee extension contraction. Insome cases, subjects were instructed to lift their toesslightly, which limited the contribution of the plan-tarflexor muscles. For ankle dorsiflexion, subjectswere instructed to “brace and prepare to pull up”against a strap positioned on the dorsum of the foot,and for ankle plantarflexion, to “brace and prepareto push down” on the platform and against the strappositioned on the top of the knee. Verbal encour-agement and real-time visual feedback of torqueproduction were provided during every contraction.Subjects were allowed to reject any effort they didnot regard as maximal. This occurred in fewer than2% of contractions. Subjects received no feedbackabout the variability in their performance betweensessions.

Data Analysis. Estimates of voluntary activationwere quantified by measurement of the torque re-sponses to stimulation over the triceps surae muscleusing twitch interpolation. Any evoked increment inplantarflexion torque during a contraction (super-imposed twitch; Fig. 2) was expressed as a fraction ofthe amplitude of the maximal response evoked bythe same stimulus in the potentiated relaxed muscle(resting twitch; Fig. 2).1,8,12 The level of voluntarydrive was then quantified as a percentage: voluntaryactivation (%) � (1 � a/b) � 100.

The experimental apparatus was sensitive, andwe were able to detect a superimposed twitch of only0.2% of the resting twitch (0.03 Nm). This corre-sponded to a voluntary activation of 99.8%. Becausevalues for voluntary activation are not normally dis-tributed (i.e., cannot exceed 100%), we have in-cluded median scores.

For some analyses, peak torques were normalizedto the largest MVC recorded during a testing session.The term “average torque” refers to the average peaktorque for contractions performed within a testingsession. In the text, group data are presented as themean � standard deviation (SD), whereas in fig-ures, the mean � standard error of the mean (SEM)is shown. Statistical analysis involved repeated-measures, one-way analysis of variance (ANOVA) forcomparison within a testing session and betweentesting sessions. A repeated-measures, one-wayANOVA was also used for between-leg comparisons.Post hoc discriminations were made with the Stu-dent Newman–Keuls procedure. To assess the vari-ability of measurements of torque and voluntary ac-tivation within and across sessions, coefficients ofvariation (SD/mean � 100; coefficient of variation,CV) were calculated for each subject. The reproduc-ibility of torque and voluntary activation measure-ments across sessions was quantified with the intra-class correlation coefficient (ICC2,1) based on arepeated-measures ANOVA with testing session as

FIGURE 2. Typical plantarflexion torque following a supramaxi-mal electrical stimulus over the triceps surae muscle while at rest(resting twitch, indicated by “b”) and during a maximal voluntarycontraction (superimposed twitch, indicated by “a”). Arrow indi-cates the timing of the stimulus.

Strength and Voluntary Activation MUSCLE & NERVE June 2004 837

the independent variable.33 The model takes intoaccount differences between subjects, differences be-tween testing sessions, and error variance. In addi-tion, the 95% confidence interval was calculated formeasurements of voluntary activation across testingsessions. The association between voluntary activa-tion and voluntary torque was assessed using a linearregression analysis. Statistical significance was set atthe 5% level.

RESULTS

Reproducibility and Variability of Lower-Limb MVCs.

In the first study, an adjustable multipurpose isomet-ric myograph was used to assess the reproducibilityand variability of measurements of maximal torquein five maximal efforts performed in a testing sessionand on multiple days for the knee flexor, knee ex-tensor, ankle dorsiflexor, and ankle plantarflexormuscles.

Table 1 shows the reproducibility and variabilityof the torque measurements. For each musclegroup, CV was less than 7% for five maximal effortsperformed in a testing session and was less than 11%for the average MVC or for the largest MVC mea-sured on multiple days. The variability was signifi-cantly different between muscle groups, with theknee extensor muscles exhibiting the greatest vari-ability (P � 0.03). Across multiple days, maximalefforts of the knee flexor, ankle dorsiflexor, andankle plantarflexor muscles were very reproducible,with intraclass correlation coefficients exceeding0.90.

Across subjects, maximal torques ranged from 25to 188 Nm (Fig. 3A). The average torque for theknee extensor muscles was 43 � 36% larger than theknee flexor muscles, and the ankle plantarflexormuscles was 258 � 74% larger than the ankle dorsi-flexor muscles. Although the proportional differ-ence in strength between the flexor and extensormuscles varied between subjects, the average abso-

lute maximal voluntary torque for the right and leftleg was similar for the four muscle groups (P �0.05). For the muscle groups tested, the averagedifference between the right and left leg was only2.6 � 1.7%. Within a testing session, 77% of thelargest maximal efforts occurred between the thirdand fifth contractions for each muscle group.

Reproducibility and Variability of Voluntary Activation

during Maximal Efforts. In the second study, the re-producibility and variability of measurements ofmaximal torque and voluntary activation were calcu-lated for 10 maximal efforts of the plantarflexormuscles. The average torque within a testing sessionwas consistent across multiple days for each subject(Fig. 3B). Measures of reproducibility and variabilityfor the average torque were similar to those in thefirst study (CV within a testing session, 5.1 � 1.8%;CV across sessions, 4.7 � 1.8%; ICC2,1, 0.98).

The average amplitude of the unpotentiated rest-ing twitch evoked by stimulation over the tricepssurae muscle was 12.4 � 2.4% MVC (14.7 � 4.5 Nm)and was reproducible across testing sessions (CV,7.6 � 2.2%; ICC2,1, 0.91). The average time to peakamplitude and half-relaxation time of the unpoten-tiated resting twitch were 115.0 � 5.8 ms and 91.3 �9.0 ms, respectively. Once potentiated by a briefMVC, the average amplitude increased to 14.3 �3.0% MVC (16.6 � 3.1 Nm) and was also reproduc-ible across testing sessions (CV, 8.0 � 1.1%; ICC2,1,0.87). With potentiation, there was a small but sig-nificant decrease in the average time to peak ampli-tude (91.9 � 5.3 ms; P � 0.006) and half-relaxationtime (84.4 � 9.2 ms; P � 0.007).

In 95% of brief maximal efforts, stimulation overthe triceps surae muscle evoked a superimposedtwitch. Thus, voluntary activation of the musclegroup was rarely complete. Across all maximal ef-forts, the average amplitude of the superimposedtwitch was 0.3 � 0.4% MVC with a time to peak

Table 1. Measures of variability (CV) and reproducibility (intraclass correlation coefficient) of maximal torque measurements.

Statistic Testing sessionKnee

flexorsKnee

extensorsAnkle

dorsiflexorsAnkle

plantarflexors

CV Within a testing session 4.6 � 2.4% 6.9 � 3.0% 3.7 � 1.6% 6.6 � 4.1%Across sessions: average MVC* 5.2 � 2.3% 9.7 � 5.0% 7.7 � 4.5% 7.9 � 5.7%Across sessions: largest MVC† 5.2 � 2.6% 10.0 � 5.8% 7.5 � 4.6% 7.1 � 4.3%

ICC Across sessions: average MVC* 0.94 0.81 0.92 0.90Across ssessions: largest MVC† 0.93 0.79 0.92 0.91

*Average maximal torque of 5 MVCs performed within a testing session.†Largest maximal voluntary contraction torque (MVC) obtained within a testing session.

838 Strength and Voluntary Activation MUSCLE & NERVE June 2004

amplitude of 13.1 � 10.8 ms. This translated to anaverage voluntary activation of 97.8 � 2.1% (me-dian, 98.5%; Fig. 2; see Methods). The lower 95%confidence interval for estimates of voluntary activa-tion across testing sessions was 93.7%. The variabilityin voluntary activation was small when measuredduring 10 maximal efforts performed in a session oracross multiple days (Fig. 4). The average CV was

1.5 � 1.0% and 0.9 � 0.4%, respectively. Reproduc-ibility, measured with the intraclass correlation coef-ficient, was poor for the average voluntary activationacross multiple testing sessions (ICC2,1, 0.60), eventhough voluntary activation varied only slightly be-tween testing sessions. This may be because of thesmall variability among subjects’ voluntary activationscores (see Discussion).

FIGURE 3. Single-subject data showing the largest maximal torque (circles) and the average maximal torque (bars, �SEM) in fivesessions. (A) Upper panel: right knee flexor (gray bars) and extensor muscles (white bars). Lower panel: right ankle dorsiflexor (gray bars)and plantarflexor muscles (white bars). (B) Right ankle plantarflexor muscles (white bars) and average amplitude of the potentiatedresting twitch evoked by stimulation over triceps surae (gray bars).

Strength and Voluntary Activation MUSCLE & NERVE June 2004 839

Voluntary Activation during Submaximal Contractions.

In the third study, voluntary activation of the tricepssurae was assessed at different levels of voluntarycontraction. The amplitude and time to peak ampli-tude of the superimposed twitch decreased with in-creasing contraction strength (P � 0.001). The re-sult was a near-linear correlation between voluntaryactivation and torque for each subject (mean r2 �0.91 � 0.04; Fig. 5). However, at high torque levels,the increase in voluntary activation was smaller for agiven change in voluntary torque. The r2 was mar-ginally improved by a polynomial fit (mean r2 �0.96 � 0.01).

DISCUSSION

Torque measurements obtained with the new systemdescribed earlier have low variability and high repro-ducibility both within a testing session and acrossmultiple sessions. Furthermore, the system allowedreproducible measurement of maximal voluntary ac-tivation of triceps surae.

Variability and Reproducibility of Maximal Torque Mea-

surements. The maximal isometric torque mea-surements obtained in this study were similar tothose previously published for the flexor and exten-sor muscles of the knee and ankle joint, with com-parable joint positioning.9,11,13,23,24,28,29,36 In some in-stances, authors reported the maximal isometricforce or torque after only two contractions.29,37 Foraccurate assessment of maximal strength, our resultsindicate that patients should perform five MVCs toensure that the largest maximal torque is recorded.

Furthermore, if the fifth MVC is the largest, an ad-ditional MVC should be performed.

Clinically, the assessment of strength is oftenused to determine disease progression and, conse-quently, the need for or result of therapeutic inter-ventions. Therefore, a high degree of reproducibilityis essential for measurements across testing sessions.In our study, measurements of maximal torque werevery reproducible for each muscle group duringmaximal efforts and were consistent with those re-ported by others.24,26 Our measurement techniqueswere also suitable for small plantarflexion torques,because the resting twitch responses were reproduc-ible. It is helpful to know the variability of maximaltorque measurements within and between testingsessions even though such information is infre-quently reported. Thus, when a patient presents witha reduction in strength between testing sessions, it ispossible to gauge whether the strength reduction issignificant. We found, for each muscle group, thatthe variability of measurements of maximal torquewas low (mean CV, 5.5 � 1.6%) for five MVCs per-formed within a testing session. Furthermore, thevariability between testing sessions was also low (av-erage torque, mean CV � 7.7 � 4.1%, largest MVC,mean CV � 7.7 � 4.2%) and compares favorablywith data reported by others.22,26,35 Maximal torquemeasurements for knee extension had the largestvariability, possibly because of inadvertent recruit-ment of the plantarflexor muscles, which could in-

FIGURE 5. Data from the individual subjects showing voluntaryactivation measured during brief maximal and submaximal con-tractions of the ankle plantarflexor muscles. Results from eachsubject are shown with different symbols.

FIGURE 4. Data from the individual subjects showing voluntaryactivation measured during brief maximal contractions of theankle plantarflexor muscles in five testing sessions.

840 Strength and Voluntary Activation MUSCLE & NERVE June 2004

fluence knee extension torque in our system unlessprecautions are taken (see Methods). Nonetheless,measurements of knee extension torque were stillreproducible between testing sessions.

Three factors probably underlie this low variabil-ity and high reproducibility of measurements ofmaximal torque. First, subjects were provided with astandard set of instructions prior to each contrac-tion. Second, care was taken to ensure accurate po-sitioning of the joints within the myograph. Third,during each contraction, subjects were provided witha real-time visual display of torque production andverbal encouragement was also given.17 A corollaryof these results is that voluntary activation also wasconsistent. This was formally confirmed for torquesproduced by ankle plantarflexion, a torque that iscritically important for stance and locomotion.

Variability and Reproducibility of Voluntary Activation.

There is limited information available on voluntaryactivation of the triceps surae muscle, primarily be-cause of problems in stabilization of the leg due tohigh torques during maximal contractions.7,9 Withour technique, in 95% of brief maximal contrac-tions, a single stimulus over triceps surae evoked asmall superimposed twitch with a twitch-like timecourse. The torque measurements were sufficientlysensitive to detect small evoked increments intorque. Voluntary activation was high but rarely max-imal (mean, 97.8 � 2.1%). Thus, the sensitivity ofour system appears optimal because complete volun-tary activation of 100% was rarely achieved but sub-jects could consistently reach high levels. Some au-thors reported that complete voluntary activation iscommonly achieved for this muscle group but thesmallest resolvable twitch amplitude and the degreeof stimulus spread to antagonist muscles is unclear inthe reports.6,10 In our study, a superimposed twitchamplitude of less than 0.2% of the resting twitchcould be resolved.

In the triceps surae muscle, we observed very lowvariability of measurements of voluntary activationrecorded within and between testing sessions. Theintraclass correlation coefficient was low (0.60),which would normally suggest poor reproducibility.However, when there is little variability among sub-jects’ voluntary activation scores, as occurred here, itis difficult to obtain a high intraclass correlationcoefficient, even though voluntary activation scoresdid not vary much between trials.26 To overcomethis, the lower 95% confidence interval was calcu-lated. Across testing sessions, 95% of voluntary acti-vation scores exceeded 93.7%. The amplitude of theunpotentiated and potentiated resting twitch was

also reproducible between testing sessions (ICC2,1 �0.87) with minimal variability (CV � 9.0%). Thissuggests that the stimulus was similar for each testingsession and that low torques were accurately re-solved.

Voluntary Activation during Submaximal Contractions.

For triceps surae, the relationship between voluntaryactivation and voluntary torque is linear until higheffort, where increases in voluntary torque occurwith less change in voluntary activation. This wasreported previously for triceps surae8 and other mus-cle groups.4,8,15,16 Such nonlinearity may be due inpart to differential voluntary activation of the syner-gistic muscles, to excessive currents used to stimulatenerves to the triceps surae inadvertently excitingantagonist dorsiflexor muscles, thus reducing thesuperimposed twitch, and other biomechanical fac-tors.4,5,14 In a seated position, triceps surae contrib-utes approximately 80% of the total plantarflexiontorque.21 The peroneus longus and brevis are un-likely to be affected by the stimulus and may make agreater contribution during progressively strongercontractions.8 The influence of any stimulus spreadto antagonist muscles is less for triceps surae than forother muscle groups, because the plantarflexor mus-cles are 2.5 times as strong as the antagonist dorsi-flexor muscles.35

In summary, our study shows that maximaltorque measurements for lower-limb muscles, ob-tained with a new isometric myograph system, arereproducible between testing sessions and exhibitlow variability in healthy subjects. Furthermore, mea-surements of voluntary activation of the triceps suraeare high during maximal efforts and show low vari-ability between testing sessions.

This work was supported by the National Health and MedicalResearch Council of Australia (3206). The authors are grateful toMr. Hilary Carter for technical assistance in construction of theapparatus and to Dr. Hylton Menz and Dr. Janet Taylor forcomments on the manuscript.

REFERENCES

1. Allen GM, Gandevia SC, McKenzie DK. Reliability of measure-ments of muscle strength and voluntary activation usingtwitch interpolation. Muscle Nerve 1995;18:593–600.

2. Allen GM, Gandevia SC, Middleton J. Quantitative assess-ments of elbow flexor muscle performance using twitch in-terpolation in post-polio patients: no evidence for deteriora-tion. Brain 1997;120:663–672.

3. Allen GM, Gandevia SC, Neering IR, Hickie I, Jones R,Middleton J. Muscle performance, voluntary activation andperceived effort in normal subjects and patients with priorpoliomyelitis. Brain 1994;117:661–670.

4. Allen GM, McKenzie DK, Gandevia SC. Twitch interpolationof the elbow flexor muscles at high forces. Muscle Nerve1998;21:318–328.

Strength and Voluntary Activation MUSCLE & NERVE June 2004 841

5. Awiszus F, Wahl B, Meinecke I. Influence of stimulus crosstalk on results of the twitch-interpolation technique at thebiceps brachii muscle. Muscle Nerve 1997;20:1187–1190.

6. Behm DG, St-Pierre DM. Effects of fatigue duration and mus-cle type on voluntary and evoked contractile properties.J Appl Physiol 1997;82:1654–1661.

7. Behm DG, Whittle J, Button D, Power K. Intermuscle differ-ences in activation. Muscle Nerve 2002;25:236–243.

8. Belanger AY, McComas AJ. Extent of motor unit activationduring effort. J Appl Physiol 1981;51:1131–1135.

9. Belanger AY, McComas AJ, Elder GB. Physiological propertiesof two antagonist human muscle groups. Eur J Appl PhysiolOccup Physiol 1983;51:381–393.

10. Bellemare F, Woods JJ, Johansson R, Bigland-Ritchie B. Mo-tor-unit discharge rates in maximal voluntary contractions ofthree human muscles. J Neurophysiol 1983;50:1380–1392.

11. Bemben MG, Massey BH, Bemben DA, Misner JE, Boileau RA.Isometric muscle force production as a function of age inhealthy 20- to 74-yr-old men. Med Sci Sport Exer 1991;23:1302–1310.

12. Bigland-Ritchie B, Furbush F, Woods JJ. Fatigue of intermit-tent submaximal voluntary contractions: central and periph-eral factors. J Appl Physiol 1986;61:421–429.

13. Clarkson PM, Kroll W, Melchionda AM. Age, isometricstrength, rate of tension development and fiber type compo-sition. J Gerontol 1981;36:648–653.

14. Cresswell AG, Loscher WN, Thorstensson A. Influence ofgastrocnemius muscle length on triceps surae torque devel-opment and electromyographic activity in man. Exp BrainRes 1995;105:283–290.

15. De Serres SJ, Enoka RM. Older adults can maximally activatethe biceps brachii muscle by voluntary command. J ApplPhysiol 1998;84:284–291.

16. Dowling JJ, Konert E, Ljucovic P, Andrews DM. Are humansable to voluntarily elicit maximum muscle force? NeurosciLett 1994;179:25–28.

17. Gandevia SC. Spinal and supraspinal factors in human musclefatigue. Physiol Rev 2001;81:1725–1789.

18. Gandevia SC, Allen GM, McKenzie DK. Central fatigue. Crit-ical issues, quantification and practical implications. Adv ExpMed Biol 1995;384:281–294.

19. Hales JP, Gandevia SC. Assessment of maximal voluntary con-traction with twitch interpolation: an instrument to measuretwitch responses. J Neurosci Methods 1988;25:97–102.

20. Herbert RD, Gandevia SC. Twitch interpolation in humanmuscles: mechanisms and implications for measurement ofvoluntary activation. J Neurophysiol 1999;82:2271–2283.

21. Hoy MG, Zajac FE, Gordon ME. A musculoskeletal model ofthe human lower extremity: the effect of muscle, tendon, andmoment arm on the moment–angle relationship of musculo-tendon actuators at the hip, knee, and ankle. J Biomech1990;23:157–169.

22. Jacobsen S, Wildschiodtz G, Danneskiold-Samsoe B. Isoki-netic and isometric muscle strength combined with transcu-taneous electrical muscle stimulation in primary fibromyalgiasyndrome. J Rheumatol 1991;18:1390–1393.

23. Kawakami Y, Sale DG, MacDougall JD, Moroz JS. Bilateraldeficit in plantar flexion: relation to knee joint position,muscle activation, and reflex excitability. Eur J Appl PhysiolOccup Physiol 1998;77:212–216.

24. Lord SR, Menz HB, Tiedemann A. A physiological profileapproach to falls risk assessment and prevention. Phys Ther2003;83:237–252.

25. McComas AJ, Miller RG, Gandevia SC. Fatigue brought on bymalfunction of the central and peripheral nervous systems.Adv Exp Med Biol 1995;384:495–512.

26. Menz HB, Tiedemann A, Kwan MM, Latt MD, Sherrington C,Lord SR. Reliability of clinical tests of foot and ankle charac-teristics in older people. J Am Podiatr Med Assoc 2003;93:380–387.

27. Merton PA. Voluntary strength and fatigue. J Physiol (Lond)1954;123:553–564.

28. Miaki H, Someya F, Tachino K. A comparison of electricalactivity in the triceps surae at maximum isometric contractionwith the knee and ankle at various angles. Eur J Appl PhysiolOccup Physiol 1999;80:185–191.

29. National Isometric Muscle Strength Database Consortium.Muscular weakness assessment: use of normal isometricstrength data. Arch Phys Med Rehabil 1996;77:1251–1255.

30. Newham DJ, Davies JM, Mayston MJ. Voluntary force gener-ation and activation in the knee muscles of stroke patientswith mild spastic hemiparesis. J Physiol (Lond) 1995;483:128P.

31. Phillips BA, Lo SK, Mastaglia FL. Muscle force measuredusing “break” testing with a hand-held myometer in normalsubjects aged 20 to 69 years. Arch Phys Med Rehabil 2000;81:653–661.

32. Rutherford OM, Jones DA, Newham DJ. Clinical and experi-mental application of the percutaneous twitch superimposi-tion technique for the study of human muscle activation.J Neurol Neurosur Psychiatry 1986;49:1288–1291.

33. Shrout PE, Fleiss JL. Intraclass correlations: uses in assessingrater reliability. Psychol Bull 1979;86:420–428.

34. Thomas CK, Woods JJ, Bigland-Ritchie B. Impulse propaga-tion and muscle activation in long maximal voluntary contrac-tions. J Appl Physiol 1989;67:1835–1842.

35. Tornvall G. Assessment of physical capabilities. Acta PhysiolScand 1963;58(Suppl 201):1–102.

36. Vollestad NK, Sejersted I, Saugen E. Mechanical behaviour ofskeletal muscle during intermittent voluntary isometric con-tractions in humans. J Appl Physiol 1997;83:1557–1565.

37. Wang C-Y, Olson SL, Protas EJ. Test–retest strength reliability:hand-held dynamometry in community-dwelling elderly fall-ers. Arch Phys Med Rehabil 2002;83:811–815.

842 Strength and Voluntary Activation MUSCLE & NERVE June 2004