A

urTfsbbcam©

K

1

atTds(aoaor

(

0d

Journal of Neuroscience Methods 162 (2007) 1–7

Standardization of H-reflex analyses

R.S.A. Brinkworth a, M. Tuncer a,b, K.J. Tucker a,S. Jaberzadeh a,c, K.S. Turker a,∗

a Research Centre for Human Movement Control, School of Molecular and Biomedical Sciences,Discipline of Physiology, University of Adelaide, SA 5005, Australia

b Hacettepe University, Faculty of Medicine, Department of Physiology, Sihhiye-Ankara 06100, Turkeyc Physiotherapy, School of Primary Health Care, Monash University, VA 3199, Australia

Received 20 June 2006; received in revised form 23 November 2006; accepted 30 November 2006

bstract

Variability in the H-reflex can make it difficult to identify significant changes using traditional pooled analysis techniques. This study wasndertaken to introduce a normalisation approach to calculate both the relative size and the relative stimulus intensity required to elicit the H-reflexesponse so that comparisons can be made not only with results obtained during different experimental session but also between different subjects.his normalisation process fits the size of the measured M-responses and H-reflexes over the entire stimulus range with model curves to better

acilitate the calculation of important parameters. This approach allows normalisation of not only the size of the response but also the relativetimulus intensity required to elicit the response. This eases the comparison of the reflex responses under various situations, and is capable ofringing out any genuine differences in the reflex in a reliable manner not previously possible. This study illustrates that comparison of the reflexetween days is problematic, even in the same subject, as both the reflex size and the relative stimulus intensity required to obtain this reflex

hanged in all subjects. We suggest that H-reflex studies need to use normalisation not only for size of the reflex but also for the stimulus intensity,nd also that all experiments for a single subject should be performed in the same session or during the same day using some level of backgrounduscle activity in the muscle concerned as the variability of the muscle at rest was found to be larger. 2006 Elsevier B.V. All rights reserved.

imTh1miatc

eywords: Methodology; Human; Hoffmann reflex; Muscle

. Introduction

The H-reflex has been described as a monosynaptic reflex,nd as an excellent tool in determining strength and distribu-ion of spindle input to a motoneuronal pool (Hugon, 1973;aborıkova, 1973). However it cannot be described as a trueirect measure of �-motoneuron excitability due to the pre-ynaptic and postsynaptic effects on motoneuron excitabilityfor reviews see Capaday, 1997; Zehr, 2002). The H-reflex shape,nd especially the latter part of the reflex, is known to haveligosynaptic contributions from the spindle afferents as well

s the involvement of inhibitory Ib effects from Golgi tendonrgans supporting further the non-monosynaptic nature of thiseflex (Pierrot-Deseilligny et al., 1981; Burke et al., 1984).

∗ Corresponding author. Tel.: +61 8 8303 5311; fax: +61 8 8303 3356.E-mail address: kemal.turker@adelaide.edu.au (K.S. Turker).URL: http://www.physiology.adelaide.edu.au/people/Turker.html

K.S. Turker).

h(

1ussd

165-0270/$ – see front matter © 2006 Elsevier B.V. All rights reserved.oi:10.1016/j.jneumeth.2006.11.020

The H-reflex is relatively easy to elicit, and has been usedn many neurophysiological investigations in human to esti-

ate a number of parameters that cannot be measured directly.hese include: the effects of active or passive movements on theuman soleus H-reflex (Brooke et al., 1995; Capaday and Stein,986), estimating synaptic efficacy of various sensory inputs tootoneuron pool (Crone et al., 1990; Meinck, 1980), estimat-

ng the threshold depolarisation of resting motoneurons (Turkernd Miles, 1991), estimating the membrane potential trajec-ory (Turker, 1995), and estimating the strength of the Renshawell inhibition (Katz and Pierrot-Deseilligny, 1999). The reflexas also found a wide range of use in clinical neurophysiologyDownes et al., 1995; Misiaszek, 2003).

However, despite extensive reviews on this reflex (Capaday,997; Zehr, 2002) one particular aspect, the statistical analysis

sing standardized measurements remains ambiguous. It is pos-ible that the variations in the H-reflex described in previoustudies may have been due to a small number of stimuli beingelivered (Zehr and Stein, 1999; Tucker and Turker, 2004), or

2 Neur

apimftrmsa

2

swa1tiawtc

2

ii(rd5btsv(S1

2

tbaowaaettoc

2

e(wiiScuaurMMsMsMattstarTsDaAc

2

MbatTbtea

M

wr

R.S.A. Brinkworth et al. / Journal of

shift in the H-reflex recruitment curve due to a change in theosition of the stimulating electrode/mixed nerve, and/or record-ng electrode/muscle relationship. In this paper a novel analysis

ethod for the H-reflex is presented which uses normalisationor both the stimulus amplitude and the reflex size. We suggesthat this new approach can elucidate genuine changes in theeflex circuitry and compensate for shifts in the H-reflex recruit-ent curve independent of the method used to determine the

ize of the reflex (area or peak-to-peak measurements; Tuckernd Turker, 2005).

. Methods

In order to illustrate the use of this new method three con-enting adult volunteers took part in the study. The subjectsere comfortably seated in a dental chair. The angle of the knee

nd foot joint, on both sides, was kept constant at ∼120о and00о, respectively, by means of an immobile footplate. Becausehe soleus H-reflex can be influenced by posture (reviewedn Schiepatti, 1987) the head was supported on a headrest,nd the forearms and hands were positioned at 0о pronationith the fingers slightly flexed. These positions ensured that

he muscles of the arm and leg were relaxed during the restondition.

.1. Electromyographic recordings

After careful preparation of the skin (shaving, abrad-ng, and cleaning with alcohol) in order to obtain lowmpedance (<10 k�), a bipolar surface electromyographySEMG) recording electrode (Duo-Trode® silver/silver chlo-ide, with a 12.5 mm active surface and 19 mm inter electrodeistance) was placed on the belly of the left soleus, aboutcm below the gastrocnemii–Achilles tendon junction. Theipolar surface EMG recordings were amplified in a cus-om made EMG amplifier with built in stimulus artefactuppressor and high pass (20 Hz) filter, and were recordedia a custom designed computer program (Brinkworth, 2004)LabVIEW®: National Instruments) at a sampling rate of 2 kHz.ubjects were grounded by lip-clip electrodes (Turker et al.,988).

.2. Feedback of voluntary contraction level

Soleus muscle activity was used to provide visual feedback ofhe subject’s level of muscle contraction. The SEMG activity wasand-pass filtered (20–500 Hz; Turker, 1993), full wave rectifiednd low passed filtered (0.3 Hz). The image was then displayedn a feedback monitor. Maximum voluntary contraction (MVC)as obtained over a series of five trials, whereby the subject was

sked to contract their calf muscles as hard as they could, forperiod of 5 s. A minimum of 2 min rest was given between

ach MVC. Trials were repeated without visual feedback, and

he strongest contraction over this period was considered to beheir MVC. A preset target line of 50% MVC was displayedn the feedback monitor, as required during the contractionondition.

H

wn

oscience Methods 162 (2007) 1–7

.3. Stimulus

The motor response (M-wave) and H-reflex were elicited vialectrical stimulation of the tibial nerve in the left leg. A large∼100 cm2) copper anode was placed just above the patella,hile the silver ball cathode, with a 1 cm diameter, was placed

n the popliteal fossa, at the point where the weakest stimulusntensity was required to elicit a H-reflex (Miles et al., 1989).quare pulse stimuli of 0.5 ms duration were triggered from aomputer and delivered by a Digitimer® constant current stim-lator (model DS7A). The stimulus artefact was removed usingcustom made artefact-suppressing amplifier. The lowest stim-lus intensity used during the experiment was 2 mA below thatequired to elicit an H-reflex in the soleus during rest and at 50%

VC. The stimulus intensity required to obtain the maximal-wave was determined by increasing the stimulus intensity in

mall increments, and assessing with online feedback when the-wave reached its maximum size. The upper stimulus inten-

ity used was 3 mA higher than required to obtain a maximal-wave at rest. During the two experimental conditions (rest

nd 50% MVC) 15 stimulus intensities were selected (includinghe pre-determined minimum and maximum intensities), andhen randomly ordered. Fourteen of the stimuli were equallyeparated on a logarithmic scale with the 15th placed betweenhe upper two intensities. This was done to give better resolutiont the lower end of the scale and to ensure that there was enoughesolution at the upper end to observe the M-wave levelling off.en blocks of these 15 stimulus intensities were given with eachtimulus delivered at random intervals between 5 and 10 s apart.uring the 50% MVC trials, the stimuli were given immedi-

tely after the subject reached the required contraction level.t least 2 min rest was provided between each experimental

ondition.

.4. Curve fitting

This method does not rely on the way that the size of the-wave and H-reflex are calculated and works equally well on

oth the peak-to-peak amplitude (of the averaged traces) andrea (of the rectified averaged traces), although for clarity onlyhe results based on area measurements are included in the text.his is because the results from each method have been found toe almost identical (Tucker and Turker, 2005). Once the size ofhe M-wave and H-reflex were calculated they were then mod-lled using a hyperbolic function for the M-wave (Eq. (1)) andGaussian for the H-reflex (Eq. (2)).

(x) = acx

b + cx(1)

here a is the maximum M-wave value, b the point of 50%esponse, c the slope parameter and x is the stimulus intensity.(

(xc − x/b)2)

(x) = a exp −2

(2)

here a is the maximum H-reflex value, exp is e the base of theatural logarithm b the standard deviation of the curve, x the

of Ne

sut

rotowt1aecp

2

etrtacdmrrt

Mitewd

3

eautsmmldsnma

ts

Fvdbs

R.S.A. Brinkworth et al. / Journal

timulus intensity that produces the largest response, x the stim-lus intensity and c is the a slope parameter to compensate forhe non-symmetrical shape of the H-reflex about the maximum.

H-reflex curves were calculated not only for the mean H-eflex response at each stimulus intensity but also the mean plusne standard deviation and the mean minus one standard devia-ion. In order to estimate the effect of using a different numberf repeats for each stimulus intensity, 95% confidence intervalsere calculated by taking the difference between the mean and

he upper and lower curve fitted H-reflex values, multiplying by.96 (conversion factor for standard error to confidence interval)nd dividing by the square root of 3, 5 and 10 to simulate theffect of using a smaller number of repeats. Non-overlappingonfidence intervals indicated significance at the 5% level (i.e.,< 0.05).

.5. Stimulus normalisation

In order to compare the results between subjects it was nec-ssary to normalise the stimulus intensities used. To do this,he stimulus intensity required to give 95% of the maximalesponse in the curve fitted M-wave was defined as 95 m whilehe value required to give 5% of maximal response was defineds 5 m. Using these two points, a linear scale was extrapolated toover the entire stimulus range used. It should also be noted thatue to the symmetrical nature of the curve fitted M-wave (Sig-

oid) the stimulus intensity required to give 50% of maximal

esponse was 50 m, while the elongated tails of the responseesulted in stimulus intensities of less than 0 m and greaterhan 100 m. The scale was given the unit of m for normalised

reeH

ig. 1. Measured variability in the H-reflex and M-response. The variability of the Eoluntary contraction. (A) H-reflex and M-response traces superimposed to show theeviation for the area of the H-reflex and M-response over the full stimulus paradigoth responses. Only raw results are shown, i.e. before curve fitting and normalisationimilar curves to the curves obtained using the area calculations (not shown).

uroscience Methods 162 (2007) 1–7 3

-wave. This normalisation procedure was used for each exper-ment in each subject. Stimulus normalisation was necessaryo account for any shifts in the stimulus intensity threshold tolicit M-wave and H-reflex, given that the threshold is differentithin and across subjects tested on the same day or in differentays.

. Results

The trial-to-trial variability (within the 10 blocks of reflexeslicited at a given stimulus intensity) was noted in all subjectst all stimulus intensities. The variability was strongest at stim-lus intensities that induced both H-reflexes and M-waves. Aypical finding is illustrated in Fig. 1. Note that the stimuli wereo arranged to include the lowest possible values to induce ainimum H-reflex and the highest possible values to generate aaximum M-wave. The number and distribution of the stimu-

us steps was selected to allow the examination of the H-reflexevelopment and to build a precise stimulus/response relation-hip without using too many stimulus steps. In this figure alsoote that neither the response nor the stimulus values were nor-alised and that these data points indicate typical EMG values

nd the stimulus current required to elicit them.Average values at each stimulus were curve fitted (see Sec-

ion 2) so that normalisation could be achieved both for theize of the reflex and for the location of the H-reflex curve in

elation to the M-wave curve, i.e., stimulus normalisation, anxample of which is shown in Fig. 2. Curve fitting was nec-ssary to accurately determine the maximum H-reflex and the-reflex size at various points along the recruitment curve. Stim-

MG events as seen in one subject at a muscle activation level of 50% maximalvariability at five different stimulus levels (n = 10). (B) Mean ± one standard

m. The variability in the waves is largest at stimulus intensities which induce. Peak-to-peak calculations of the H-reflex and M-response amplitudes induced

4 R.S.A. Brinkworth et al. / Journal of Neur

Fig. 2. Curve fitting the H-reflex and M-response. Modelled (solid lines) andraw (dashed lines and circles) H-reflex (black) and M-response (grey) curvesover the full stimulus range. The x-axis has been normalised as described in themethods such that 5, 50 and 95 m correspond to 5, 50 and 95% of the curvefitted M-response value, while the y-axis has been normalised to the modelledmaximal M value. Raw data are based on the average area of 10 responsesfsf

uttactteiec

b

vfim0

dcrmarwoiil

4

Fistrwpw

FoctTdt

rom one subject during a 50% maximal voluntary contraction. Note the non-ymmetrical nature of the H-reflex curve, which has a sharper rising phase thanalling phase.

lus normalisation was necessary to observe real changes inhe excitability of the H-reflex circuit. Since the position ofhe M-wave was normalised to the stimuli and hence ‘fixed’,ny significant shift in the H-reflex curve indicated a genuinehange in excitability of the reflex circuit and were independento changes in the electrode/nerve relationship. Fig. 3 illustrateshis point clearly where M-wave curves obtained during differentxperiments on the same day or on different days were super-mposed via stimulus normalisation. This superimposition then

xposed any shift in the position or the magnitude of the H-reflexurve.

The accuracy of the curve-fitting procedure was determinedy measuring the correlation co-efficient, r2. The average r2

amps

ig. 3. Repeatability of the H-reflex and M-response. Curve fitted results from all threver the entire stimulus range at both 0 and 50% maximal voluntary contraction (orrespond to 5, 50 and 95 m on the stimulus scale. By normalising the M-response ihat the true variability in the H-reflex can be exposed. The dashed lines show one shis figure illustrates that the H-reflex is repeatable only when the experiments are dramatically in the same subject between days. Hence it is suggested that between suhe same subject between different experimental days.

oscience Methods 162 (2007) 1–7

alue for the M-wave fits was 0.995 ± 0.002 (mean ± 95% con-dence interval, n = 15) and 0.962 ± 0.009 for the H-reflex. Theinimum r2 value for any of the M-wave fits was 0.984 and

.909 for the H-reflex.When 50% MVC was used, the trials conducted in the same

ay produced results that were not significantly different (95%onfidence intervals overlapped), however the trials that usedelaxed muscles (0% MVC) generated H-reflex curves that wereuch more variable and significantly different from each other,

s can be seen in Table 1. When the same experiments wereeplicated on the same subjects on different days the resultsere variable and produced large shifts in the relative amplitudef the stimulus required to elicit maximal H-reflex. This findings illustrated in Fig. 4, which also shows the effect of averag-ng different numbers of the same stimuli on the confidenceimits.

. Discussion

There were a number of important findings in this study.irstly, trial-to-trial variability was observed at all stimulus

ntensity levels, however the greatest variability existed at inten-ities that induced both the H-reflex and M-responses. Secondly,he importance of curve fitting and normalisation of the stimulusesponse curve to the M-response curve were illustrated. Thirdly,hile H-reflex curves determined from experiments that wereerformed on the same day were similar, especially when thereas some background activity in the muscle, day-to-day vari-

bility in these curves were observed in all subjects and at bothuscle conditions (rest and 50% MVC). Fourthly, while it was

ossible to determine confidence limits from using only threetimuli for each stimulus intensity, a relatively large level of

e subjects showing the average area (solid lines) of the H-reflex and M-responseMVC). M-responses (green) have been normalised such that 5, 50 and 95%t is possible to remove a number of sources of variability in the experiment sotandard error of the mean from 10 repeats of 15 different stimulus intensities.one in the same subject during the same session. The H-reflex values change

bject comparisons would therefore be totally unreliable given the variability in

R.S.A. Brinkworth et al. / Journal of Neuroscience Methods 162 (2007) 1–7 5

Table 1H-reflex recruitment threshold and maximal response

H-threshold H-max

S Muscle activity D1 × T1 D1 × T2 D2 × T1 D2 × T2 Mean S.D. D1 × T1 D1 × T2 D2 × T1 D2 × T2 Mean S.D.

Raw stimulus intensity (mA)1 0% MVC 5.46 4.15 5.09 4.09 4.70 0.687 8.31 6.39 8.46 7.00 7.54 1.010

50% MVC 5.49 5.16 4.59 4.55 4.95 0.456 9.80 9.84 9.95 10.09 9.92 0.1322 0% MVC 4.51 4.41 6.52 6.65 5.52 1.227 8.22 7.83 11.01 11.21 9.57 1.788

50% MVC 4.09 3.97 5.63 5.24 4.73 0.829 7.44 7.36 10.24 9.45 8.62 1.4503 0% MVC 6.55 6.33 6.41 5.86 6.29 0.300 11.72 11.83 9.88 9.54 10.74 1.202

50% MVC 6.76 6.87 6.79 6.55 6.74 0.139 12.58 12.45 12.22 12.18 12.36 0.190

Normalised stimulus intensity (m)1 0% MVC −39.92 −18.33 −50.93 −40.26 −37.36 13.67 8.66 20.00 3.99 9.46 10.53 6.76

50% MVC −26.08 −23.65 −50.92 −37.35 −34.50 12.47 16.14 17.60 5.01 11.47 12.56 5.672 0% MVC −11.16 −16.10 −23.40 −32.86 −20.88 9.44 25.16 23.71 16.18 13.28 19.58 5.76

50% MVC −68.61 −58.69 −65.32 −56.90 −62.38 5.51 −11.44 −5.00 −9.10 −5.68 −7.80 3.013 0% MVC −109.45 −114.54 −28.01 −36.90 −72.22 46.11 −27.55 −29.76 16.13 12.85 −7.08 24.96

50% MVC −94.59 −93.64 −40.32 −65.74 −73.57 25.9 −18.95 −16.91 11.19 1.39 −5.82 14.57

The minimum stimulus intensity required to elicit a H-reflex (mean from curve fitted response) and the stimulus intensity required to elicit the maximal H-reflex(mean of curve fitted response) for each subject (S 1; 2; 3) at both levels of maximal voluntary contraction (MVC). Data are separated by both day (D) and trialw elow d( (indica

uifntsra

4r

pafrariesv1rln2dtfpHw

otl11(tirti1mthgseTsiwliltist

ithin each day (T). In the top table data are shown in raw units (mA) while bS.D.) indicates the spread of the results over all days and trials and was lowerll cases.

ncertainty in the measurement of the maximum H-reflex sizendicated that at least five, but preferably 10, stimuli be usedor each intensity as this would make the determination of sig-ificant changes much more likely. Finally it was determinedhat at least 15 stimulus steps are required within the wholetimulus response curve to determine the stimulus intensityequired to produce the maximum H-reflex size with sufficientccuracy.

.1. Trial-to-trial variation of the magnitude of the H-reflexesponse

The size of the H-reflex is a function of three factors: therecision of stimulus delivery, excitability of the entire H-reflexrch, and accuracy of the recording. Since none of these threeactors can be predicted and fixed, it is self-evident that H-eflex will vary not only between subjects and between days, butlso from trial-to-trial in the same subject. We have previouslyeviewed the methods for the stimulus delivery and EMG record-ng, and suggested normalisation for these procedures (Tuckert al., 2005). However, even when one follows these suggestedtimulation and recording procedures closely, the trial-to-trialariability of the reflex continues to occur (Funase and Miles,999; Tucker and Turker, 2005). This variability occurs as aesult of many factors, including changes in the effective stimu-us intensity (due to changes in the stimulating electrode/mixederve, Tucker et al., 2005), axonal excitability (Burke et al.,001), responsiveness of the Ia synapse (via primary afferentepolarisation; Pierrot-Deseilligny, 1997), as well as changes inhe motoneuron excitability (Funase and Miles, 1999). There-

ore, the size of the H-reflex response, which represents therecision of the stimulus delivery, the excitability of the entire-reflex arch and the precision of the recording from the muscle,ill vary from trial-to-trial.

hssr

ata are shown in units normalised to the M-wave (m). The standard deviationating less spread in the measurements) during a 50% MVC than during rest in

To minimise variability, most researchers average a numberf H-reflex responses around the time of stimulation. However,here is no agreement on the number of times the same stimu-us is applied. It has been suggested that the reflex response to0 stimuli be averaged for each experimental condition (Hugon,973), however, in more recent investigations, as few as eightJaberzadeh et al., 2004), five (Christie et al., 2004) or evenhree stimuli (Tucker and Turker, 2004) have been used. Whilencreasing the number of trials increases the signal to noiseatio, it can also generate additional problems since the dura-ion of the study for a given variable will increase, resultingn further complications such as fatigue (Bigland Ritchie et al.,983; Miles et al., 1989). Therefore, it is best to establish ainimum number of reflexes elicited in a given experimen-

al condition that give a reliable estimate of the H-reflex size,ence the excitability of the entire H-reflex circuitry during aiven condition. Furthermore, many studies have used only onetimulus intensity to study changes in H-reflex size rather thanntire reflex recruitment curve (reviewed in Tucker et al., 2005).he results presented in this paper support previous work thatuggests H-reflex excitability can be modified by different exper-mental conditions. It goes further however to suggest that thehole H-reflex curve moves relative to the normalised stimu-

us intensity, and that therefore using only one level of stimulusntensity to determine H-reflex excitability would generate mis-eading results. Even when 15 stimulus steps are used, deliveringhree stimuli for each of the steps will generate large variabilityn the reflex sizes seen. This variability may therefore obscureignificant changes in the size of the H-reflex when the reflex isested under differing experimental conditions. This experiment

as found that at least five stimuli should be delivered at eachtimulus intensity in order to produce confidence limits that aremall enough to be useful for studying changes in the H-reflexesponse under different experimental conditions.

6 R.S.A. Brinkworth et al. / Journal of Neur

Fig. 4. Variability of the H-reflex. Average size of the curve fitted maximumH-reflex and the stimulus required to elicit it. Error bars show the 95% confi-dence intervals. The y and x co-ordinates represent the average of 10 normalisedmaximal H-reflex sizes and 15 normalised stimulus intensities, respectively (seeSection 2). Data are shown separately for each subject and have been separatedby the level of muscle activity (MVC%; maximum voluntary contraction) andday, with 0% MVC in black triangles and 50% MVC in grey squares, day 1 datapoints are filled while day 2 data points are hollow. Trials on the same day withthe same level of muscle activity are denoted by the same symbol. Uncertaintyin the size of the maximum response (y-axis) can be reduced by increasing thenumber of the same stimuli averaged together. This has been done and is evidentiib

4

parmst2acm

rcc

H12icobtb

4

pBfsHtdrfmtowfcoutot(i

4

rmrm2aa(tr

n the three different size error bars which represent n = 10, 5, and 3. Uncertaintyn the size of the stimulus required to elicit the maximum H-reflex (x-axis) cane reduced by increasing the number of different stimulus intensities used.

.2. H-reflex variability between conditions and day-to-day

The current investigation supports the suggestion that a com-arison of the entire recruitment curve is more useful thanttempting to compare a single measurement of the H-reflexesponse. It is well understood that the entire H-reflex curve canove to the left (hence requires less current to reach the same

ize), increase in size relative to the M-response during volun-ary activation of muscles (Loscher et al., 1996; Fumoto et al.,

002; Tucker and Turker, 2004). The size of the M-response canlso change during different muscle lengths and experimentalonditions (Tucker et al., 2005). Therefore reporting H-reflexagnitude relative to a fixed maximal M-response magnitude

ie‘W

oscience Methods 162 (2007) 1–7

epeated during different experimental conditions may not fullyapture the changes that occur in the effectiveness of this reflexircuitry.

In contrast to the findings of this methods paper, the soleus-reflex magnitude has been claimed to be highly reliable fromday to another (Hopkins and Wagie, 2003; Christie et al., 2004,005). However, studies considering the between day reliabil-ty of the H-reflex only consider one point on the recruitmenturve, and therefore cannot provide a complete understandingf the reliability of the H-reflex and M-wave recruitment curvesetween days. The results from the current investigation suggesthat the entire recruitment curve should be investigated to ensureoth between condition and between day reliability.

.3. Importance of curve fitting

Curve fitting makes it possible to determine the size and theosition of the H-reflex relative to the M-response with ease.y using a number of stimulus intensities, and exploiting the

act that the H-reflex recruitment curve closely matches a Gaus-ian curve, it was possible to find not only the point of maximal-reflex excitability (mean of the curve) but to also estimate

he confidence interval of this measurement by using the stan-ard deviation of the underlying function. By using a number ofepeats at each stimulus intensity, it was possible to fit a Gaussianunction not only to the mean response but also to calculate theean ± one standard deviation at each stimulus, and to fit curves

o this as well, in order to provide an estimate of the variabilityf the size of the response at any stimulus intensity. In this way itas possible to get not only means, but also confidence intervals

or each subject. The use of 15 different stimulus steps was suffi-ient to accurately curve fit both the H-reflex and the M-responsever the entire stimulus recruitment curve, however the resultingncertainty in the measurement of the stimulus required to elicithe maximum H-reflex was considerable, due to the large rangef the H-reflex recruitment curve. It is therefore recommendedhat a larger number of stimulus steps (20–25) with a numbers5–10) of repeats at each step be used to obtain reliable resultsn H-reflex investigations.

.4. The importance of normalisation

There are a number of papers in the literature that use the H/Mecruitment curves as evidence for stimulating low thresholduscle spindle primary afferents and subsequent blocking of the

eflex by the antidromically conducting action potentials in theotor axons (Hugon, 1973; reviewed in Capaday, 1997; Zehr,

002). These curves, and values obtained from these curves,re also used widely to normalise the H-reflex response withinnd between experimental sessions as well as between subjectsChristie et al., 2004). Both the existence and importance ofhese H/M curves in recognising and normalising the H-reflexesponse is well recognised (Koceja et al., 1995). However when

t comes to comparing the H-reflex responses under varyingxperimental conditions most investigators use only one selectedrepresentative’ point or slope from these curves (Hopkins and

agie, 2003; Christie et al., 2005).

of Ne

rmiaajcse

aatfittI

innnpminMti

R

B

B

B

B

B

C

C

C

C

C

D

F

F

F

H

H

J

K

K

L

M

M

M

P

P

S

T

T

T

T

T

T

T

Ttrode for use in human electrophysiology. Brain Res Bull 1988;21:139–41.

R.S.A. Brinkworth et al. / Journal

Given the large number of variables that can affect the H-eflex curve, between subject comparisons can become evenore problematic, even when normalisation factors for stimulus

ntensity and maximum M-wave size have been used. Despitell the uncontrollable parameters that may influence the H-reflexrch from one experimental session to another, or in one sub-ect to the next, the magnitude of the reflex has routinely beenompared on different days and/or with different experimentalettings (Funase et al., 1994; Hopkins and Wagie, 2003; Christiet al., 2005).

By performing normalisation of both the size of the responsend the position of the reflex curve, changes in the positionnd the size of the H-reflex can be clearly illustrated. Sincehe position of the H-reflex curve was normalised to the curve-tted and stimulus-normalised M-wave, significant changes in

he relative position of the H-reflex curve indicate that eitherhe excitability of the motoneurons, or the effectiveness of thea/�-motoneuron synapses, has changed.

Since normalising the stimulus intensity to the M-wave curvendicates the effective stimulus current arriving in the mixederve and exposes real changes in the H-reflex circuit that areot related to the alterations in the stimulating electrode/mixederve relationship, without such normalisation it is impossible torecisely determine the changes that influence the H-reflex. Theethod used in this paper can clearly indicate genuine changes

n the H-reflex circuitry that occur between experiments since itormalises the H-reflex to the full M-wave curve and keeps the-wave recruitment curve constant between trials, thus ensuring

hat the stimuli are equally effective in activating the motor fibresn the mixed nerve, and that the results obtained are truly reliable.

eferences

igland Ritchie B, Johansson R, Lippold OC, Smith S, Woods JJ. Changes inmotoneurone firing rates during sustained maximal voluntary contractions.J Physiol 1983;340:335–46.

rinkworth RSA. Response of the human jaw to mechanical stimulationof teeth. PhD thesis. University of Adelaide, Adelaide, Australia. 2004http://thesis.library.adelaide.edu.au/public/adt-SUA20050110.160608.

rooke JD, Cheng J, Misiaszek JE, Lafferty K. Amplitude modulation of thesoleus H reflex in the human during active and passive stepping movements.J Neurophysiol 1995;73:102–11.

urke D, Gandevia SC, McKeon B. Monosynaptic and oligosynaptic contribu-tions to human ankle jerk and H-reflex. J Neurophysiol 1984;52:435–48.

urke D, Kiernan MC, Bostock H. Excitability of human axons. Clin Neuro-physiol 2001;112:1575–85.

apaday C, Stein RB. Amplitude modulation of the soleus H-reflex in the humanduring walking and standing. J Neurosci 1986;6:1308–13.

apaday C. Neurophysiological methods for studies of the motor system infreely moving human subjects. J Neurosci Methods 1997;74:201–18.

hristie A, Lester S, LaPierre D, Gabriel DA. Reliability of a new measure ofH-reflex excitability. J Clin Neurophysiol 2004;115:116–23.

hristie AD, Inglis JG, Boucher JP, Gabriel DA. Reliability of the FCR H-reflex.

J Clin Neurophysiol 2005;22:204–9.

rone C, Hultborn H, Mazieres L, Morin C, Nielsen J, Pierrot-Deseilligny E.Sensitivity of monosynaptic test reflexes to facilitation and inhibition as afunction of the test reflex size: a study in man and the cat. Exp Brain Res1990;81:35–45.

Z

Z

uroscience Methods 162 (2007) 1–7 7

ownes L, Ashby P, Bugaresti J. Reflex effects from Golgi tendon organ (Ib)afferents are unchanged after spinal cord lesions in humans. Neurology1995;45:1720–4.

umoto M, Komiyama T, Nishihira Y. Soleus H-reflex dynamics during fastplantarflexion in humans. J Electromyogr Kinesiol 2002;12:367–74.

unase K, Imanaka K, Nishihira Y, Araki H. Threshold of the soleus muscle H-reflex is less sensitive to the change in excitability of the motoneuron poolduring plantarflexion or dorsiflexion in humans. Eur J Appl Physiol OccupPhysiol 1994;69:21–5.

unase K, Miles TS. Observations on the variability of the H reflex in humansoleus. Muscle Nerve 1999;22:341–6.

opkins JT, Wagie NC. Intrasession and intersession reliability of the quadricepsHoffmann reflex. Electromyogr Clin Neurophysiol 2003;43:85–9.

ugon M. Methodology of the Hoffman reflex in man. In: Desmedt JE, editor.New Developments in Electromyography and Clinical Neurophysiology,vol. 3. Basel: Karger; 1973. p. 277–93.

aberzadeh S, Scutter S, Warden-Flood A, Nazeran H. Between-days reliabil-ity of H-reflexes in human flexor carpi radialis. Arch Phys Med Rehabil2004;85:1168–73.

atz R, Pierrot-Deseilligny E. Recurrent inhibition in humans. Prog Neurobiol1999;57:325–55.

oceja DM, Markus CA, Trimble MH. Postural modulation of the soleus Hreflex in young and old subjects. Electroencephalogr Clin Neurophysiol1995;97:387–93.

oscher WN, Cresswell AG, Thorstensson A. Central fatigue during a long-lasting submaximal contraction of the triceps surae. Exp Brain Res1996;108:305–14.

einck HM. Facilitation and inhibition of the human H reflex as a function ofthe amplitude of the control reflex. Electroencephalogr Clin Neurophysiol1980;48:203–11.

iles TS, Turker KS, Le TH. Ia reflexes and EPSPs in human soleus motoneu-rones. Exp Brain Res 1989;77:628–36.

isiaszek JE. The H-reflex as a tool in neurophysiology: its limitations anduses in understanding nervous system function. Muscle Nerve 2003;28:144–60.

ierrot-Deseilligny E. Assessing changes in presynaptic inhibition of Ia afferentsduring movement in humans. J Neurosci Methods 1997;74:189–99.

ierrot-Deseilligny E, Bergego C, Katz R, Morin C. Cutaneous depression ofIb reflex pathways to motoneurones in man. Exp Brain Res 1981;42:351–61.

chiepatti M. The Hoffmann reflex: a means of assessing spinal reflex excitabil-ity and its descending control in man. Prog Neurobiol 1987;23:345–76.

aborıkova H. Supraspinal influences on H-reflexes. In: Desmedt JE, editor. NewDevelopments in Electromyography and Clinical Neurophysiology. Basel:Karger; 1973. p. 328–35.

ucker KJ, Turker KS. Muscle spindle feedback differs between the soleus andgastrocnemius in humans. Somatosens Mot Res 2004;21:189–97.

ucker KJ, Turker KS. A new method to estimate signal cancellationin the human maximal M-wave. J Neurosci Methods 2005;149:31–41.

ucker KJ, Tuncer M, Turker KS. A review of the H-reflex and M-wave in thehuman triceps surae. Hum Mov Sci 2005;24:667–88.

urker KS. Electromyography: some methodological problems and issues. PhysTher 1993;73:698–710.

urker KS. The shape of the membrane potential trajectory in tonically-activehuman motoneurons. J Electromyogr Kinesiol 1995;5:3–14.

urker KS, Miles TS. Threshold depolarization measurements in resting humanmotoneurones. J Neurosci Methods 1991;39:103–7.

urker KS, Miles TS, Le HT. The lip-clip: a simple, low-impedance ground elec-

ehr PE. Considerations for use of the Hoffmann reflex in exercise studies. EurJ Appl Physiol 2002;86:455–68.

ehr EP, Stein RB. Interaction of the jendrassik maneuver with segmental presy-naptic inhibition. Exp Brain Res 1999;124:474–80.