Reliability of intracortical and corticomotor excitability estimates
obtained from the upper extremities in chronic stroke§,§§
Lisa Koskia,b,*, Janice Chien-Ho Linc, Allan D. Wub,e, Carolee J. Winsteinc,d
aDivisions of Geriatrics and Clinical Epidemiology, Faculty of Medicine, Department of Neurology and Neurosurgery, McGill University, Canada
bTranscranial Magnetic Stimulation Laboratory, Ahmanson-Lovelace Brain Mapping Center, University of California, Los Angeles, United States
cMotor Behavior and Neurorehabilitation Laboratory, University of Southern California, United States
dDepartment of Neurology, Keck School of Medicine, University of Southern California, United States
eDepartment of Neurology, University of California, Los Angeles, United States
Received 24 August 2006; accepted 15 January 2007
Available online 20 January 2007
We estimated the trial-to-trial variability and the test–retest reliability of several intracortical and corticomotor excitability parameters for
the upper extremity in chronic stroke patients. Nine patients with hemiparesis of the upper extremity were enrolled 8–17 months after a
unilateral stroke. Transcranial magnetic stimulation was used to obtain repeated measures over a two week interval of motor evoked potential
(MEP) recruitment curves and cortical silent periods in the first dorsal interosseus muscle of each hand. Five trials would have provided
accurate estimates of the MEP amplitude and silent period duration for the unlesioned side in all patients, but 25% of the datasets from the
lesioned side provided poor estimates of MEP amplitude even with 10 trials. Intraclass correlations were >0.70 for all parameters obtained
from the lesioned side and for the MEP amplitude, slope of the recruitment curve, silent period, and silent period slope from the unlesioned
side. MEP amplitude varied across sessions within subject by 20% on both sides, whereas other parameters showed less variability on the
unlesioned side relative to the lesioned side. The Fugl-Meyer upper extremity motor score and the time to complete the 6 fine-motor items
from the Wolf Motor Function Test (WMFT) were also found to be highly reliable over this interval. We conclude that the functional and
most of the excitability parameters are reliable across time in patients with variable lesions due to stroke. Due to high intrasubject variability,
the use of some excitability parameters as indicators of functional neuroplasticity in response to treatment may be limited to interventions
with large effect sizes.
# 2007 Published by Elsevier Ireland Ltd and the Japan Neuroscience Society.
Keywords: TMS; Motor excitability; Stroke; Rehabilitation; Intrasubject reliability
Transcranial magnetic stimulation (TMS) is increasingly
used to obtain estimates of the excitability of intracortical
and corticomotor pathways after stroke. A considerable body
of evidence now supports the relationship between cortico-
motor excitability estimates and recovery of upper extremity
motor function after stroke (Turton et al., 1996; Cicinelli
et al., 1997a,b; Traversa et al., 1998; Liepert et al., 2000;
Traversa et al., 2000; Liepert et al., 2001; Delvaux et al.,
2003; Wittenberg et al., 2003; Platz et al., 2005). These
results support the validity of corticomotor excitability
parameters as markers of plastic change in studies of
therapeutic interventions after stroke. A few studies have
response to interventions aimed at improving the use of
the affected extremity (Liepert et al., 2000, 2001; Stinear and
Byblow, 2004; Platz et al., 2005).
Motor thresholds, motor evoked potentials (MEPs), and
silent periods are reliable estimates of corticomotor and
Neuroscience Research 58 (2007) 19–31
§This work was supported by General Clinical Research Centers Program
M01-RR00865 (UCLA), NIH NS45485 from NINDS to CW, and K23-
NS045764-01 from NINDS to AW.
§§Parts of this work were presented at the Organization for Human Brain
Mapping Annual Meeting, June 12–16 2005, Toronto, Canada.
* Corresponding author at: Royal Victoria Hospital, R4.74 687 Pine Avenue,
Montreal, Quebec, Canada H3A 1A1. Tel.: +1 514 934 1934x35652, voicemail
extension 34420; fax: +1 514 843 1734.
E-mail address: email@example.com (L. Koski).
0168-0102/$ – see front matter # 2007 Published by Elsevier Ireland Ltd and the Japan Neuroscience Society.
intracortical excitability in studies of the upper extremity
muscles in normal subjects (Mortifee et al., 1994; Cicinelli
et al., 1997a,b; Fritz et al., 1997; Miranda, 1997; Carroll et al.,
2001; Stewart et al., 2001; Maeda et al., 2002; Uy et al., 2002;
Wassermann, 2002; De Gennaro et al., 2003; Tergau et al.,
2005; Koski et al., 2005; Malcolm et al., 2006). Yet there is
almost no evidence for the test–retest reliability of these
measures in stroke patients. One reliability study showed that
the group mean of the motor thresholds, MEP amplitudes, map
sizes and recruitment curve slopes for the extensor digitorum
communalis muscle did not change when tested three times at
one-week intervals in 10 chronic stroke patients (Butler et al.,
2005). In treatment studies that included repeated baseline
evaluations there was no change in map size for the wrist flexor
and extensor muscles over a period of at least seven days
(Stinear and Byblow, 2004), or the abductor pollicis brevis over
a period of one (Liepert et al., 2001) or two weeks (Liepert
et al., 2000), before beginning treatment.
Little is known, however, about other important aspects of
test–retest reliability, such as the absolute agreement between
the values obtained at different assessment sessions (intraclass
correlation coefficients) and the amount of intrasubject
variation (known variously as measurement error, the standard
error of measurement, or the typical error of measurement).
Furthermore, the number of trials required to obtain reliable
estimates of motor evokedpotential amplitudes and latencies in
patients with post-stroke hemiparesis has not been studied. The
amount of time required to obtain reliable estimates of
excitability is a significant consideration for older persons with
a brain lesion who tend to fatigue easily. In normal control
subjects, as few as 5 trials may be sufficient to obtain mean
MEPamplitudes within 20% of the ‘‘true’’ MEP when studying
young healthy adults (Brasil-Neto et al., 1992a,b). However,
variability in MEPs increases with age (Pitcher et al., 2003) so
that this number of trials may not be appropriate in studies of
older persons with stroke.
The goals of this study were to determine: (1) the number of
trials per session required to obtain accurate estimates of
session (test–retest) reliability of motor excitability parameters
for the first dorsal interosseus (FDI) muscle in chronic stroke
patients. Patients were tested twice with an interval of two
weeks. Our measures of corticomotor excitability were MEP
amplitude and latency and the slope of the MEP stimulus–
response recruitment curve. Our measures of intracortical
excitability were the silent period duration and the slope of the
stimulus–response relation for silent period duration.
MEPs were measured during active contraction of the
muscle (Byrnes et al., 1999, 2001) to minimize the potential
impact of subthreshold muscle activation on the variability of
the MEP. To our knowledge, this is the first paper to describe
stimulus–response functions for silent period duration and the
variability of the silent period in patients with stroke.
Throughout this paper we use the term excitability when
referring to intracortical and corticomotor excitability
The nine participants in this study were recruited from a population of
patients who were participating in a within-subject clinical trial of constraint-
induced movement therapy that included a control period of two weeks without
intervention (Winstein et al., 2003). As such, they represent a relatively
homogenous group in terms of the degree of motor impairment. Inclusion
criteria were a first cerebrovascular accident at least 6 months previously and
deficient use of the affected upper extremity, but with minimum amounts of
voluntary control of wrist extension and abduction/extension of the thumb and
fingers and good passive range of movement. Exclusion criteria were: previous
cerebrovascular accident with neurological sequelae, frequent use of the
affected upper extremity at the time of enrollment (>2.5 on Motor Activity
Log Amount of Use scale(Kunkel et al., 1999; Uswatte et al., 2005; Wolf et al.,
2005)), inability to provide informed consent (<18 years old, or <24 on the
Folstein Mini-Mental Status Exam (Folstein et al., 1975)), major medical
incapacity, current participation in rehabilitation or use of antispasticity med-
ications, and contraindication for TMS as defined by international guidelines,
including presence of pacemaker, metal in head, pregnancy, other neurological
disorder, current use of stimulants or medications known to lower seizure
threshold, and personal or family history of seizure disorder (Wassermann,
1998). A structural magnetic resonance image of the brain was obtained to
verify the location and extent of the lesion in each patient.
This study was approved by UCLA’s Medical IRB, where the TMS
procedures were conducted, and by the USC Health Sciences IRB, where
the behavioural tests and treatment intervention were conducted.
This is a repeated-measures study in which measures of corticomotor and
intracortical excitability were obtained twice in each participant with a two-
week interval between sessions. No effort was made to control the amount of
motor activity performed between sessions although none of the patients was
engaged in a formal rehabilitation program at the time. Data were obtained at a
mean time of 12.8 months since stroke (range: 8–17 months). Two measures of
functional motor performance, the Wolf Motor Function Test (WMFT) and the
Fugl-Meyer upper extremity motor score, were included to assist in interpreta-
tion of the reliability data for excitability.
2.3. Procedures for EMG data collection
Electromyographic (EMG) signals were acquired using surface electrodes
digitized at 2000 Hz. The data were visually displayed and stored for later
analysis in 600 ms samples beginning 100 ms before TMS onset (Labview,
dorsal interosseus (FDI) muscle at 10% of the subject’s maximum voluntary
contraction for the more affected extremity as determined at baseline. Maximal
force was measured in millimeters of mercury using a pressure gauge pinched
Civardi et al., 2001). Both subject and experimenter were able to visually
monitor the level of muscle contraction and the TMS pulse was timed to occur
within seconds of onset of the contraction, when the appropriate level of
contraction had been achieved. The rationale for recording MEPs in the active
musclewas to reduce thevariability in the signal caused by random fluctuations
in the excitability of corticospinal and segmental motoneurons (Kiers et al.,
1993) and to standardize the level of alertness during testing by requiring
subjects to monitor their level of contraction during each trial. Since the motor
threshold is lower for MEPs in the active muscle, a greater number of stimulus
intensities could be tested within each subject, which improved our ability to
model the stimulus–response curve.
by the identical procedures for stimulation of the unlesioned hemisphere. A
L. Koski et al./Neuroscience Research 58 (2007) 19–3120
in one session and then last in another session would have introduced noise into
our estimates of test–retest reliability for that hemisphere. Stimulation was
performed using a Magstim Super Rapid biphasic stimulator and an air-cooled,
figure-8 coil (diameter: 9 cm/wing). The choice of a biphasic rather than
monophasic stimulator was based on availability of the appropriate stimulator
and coils. The point of intersection of the figure-8 coil was placed against the
skull and the coil was held at a 458 angle to the sagittal with the handle oriented
posterolaterally (Brasil-Neto et al., 1992a,b).
A tightly elastic cap marked with a 1 cm ? 1 cm grid relative to the vertex
was fitted for each subject at baseline and reapplied consistently for subsequent
sessions to facilitate systematic sampling of scalp locations during determina-
tion ofthe hotspot,to providevisual landmarks formaintaininga consistentcoil
angle within and across sessions, and to record the hotspot at each session. At
the beginning of each session, a relatively high intensity of stimulation
(typically 80% of maximum stimulator output) was used to locate the scalp
position from which the MEPs of highest amplitude could be obtained and the
location was marked on the cap to ensure consistent targeting of this ‘‘hotspot’’
throughout the session.
The stimulator intensity was then dropped to 30% maximum stimulator
output (MSO) and 10 pulses were delivered over the hotspot at a frequency of
one every 5–10 s. If no silent period or MEP was observed in those 10 trials, the
stimulator intensity was raised to 40% MSO and 10 more trials were sampled.
This procedure was repeated until the first evidence of an MEP or silent period
appeared in the EMG trace. At this point, the stimulation threshold was
determined by varying the stimulus intensity in 2% increments until the
minimum intensity was identified at which a silent period was obtained in
eye than were MEPs). The stimulus intensity was then raised in increments of
5%, with 10 trials obtained at each intensity, up to the maximum stimulator
intensity.The intensitieswere alwaysdelivered in ascendingorderand thesame
intensities used at the first session were applied in all subsequentsessions for an
tiring for the participants.
2.4. Excitability outcome measures
The following measures of excitability were obtained and analyzed: the
largest value obtained among the averages of the 10 peak-to-peak MEP
amplitudes calculated for each stimulation intensity, the average MEP latency
obtained at the same stimulation intensity selected for amplitude, the slope of
the intensity-amplitude function for MEP, the largest mean silent period
duration (after averaging the 10 durations calculated at each intensity), slope
of the intensity-duration function for silent period. All data were analyzed off-
line by the same investigator using a modular MATLAB-based (Mathworks,
MA) software tool developed by A. Wu for analysis of time-series data
(dataWizard, A.D.W., UCLA). The algorithms used to label relevant signal
events are described below.
The MEP was measured for each trial individually. Movement artifact was
present in <1% of trials and was treated by the application of a finite impulse
response(FIR)filterwith high-pass(low-end)cutoff at 50 Hz (Daskalakisetal.,
2003). MEP onset was defined as the first sustained crossing of the rectified
EMG trace prior to the first MEP peak that exceeded a threshold, where this
threshold was defined as three standard deviations above the averaged rectified
EMG signal computed in the 80 msec prior to TMS pulse. The MEP minimum
signal observed within a window between MEP onset and 50 ms after the TMS
pulse. These labels were used to calculate the MEP latency (MEP onset - TMS
onset, in ms) and peak-to-peak amplitude (maximum deflection–minimum
deflection, in uV). The largest MEP amplitude (MEPmax) was defined as
the largest value obtained when averaging together the amplitudes calculated
for the 10 trials, regardless of stimulus intensity. Typically this corresponded to
the average of the MEP amplitudes obtained at 100% of the maximum
stimulator output. To compute the silent period duration for each trial we first
implemented a high-pass finite impulse response (FIR) filter with cutoff
frequency of 50 Hz and a passband of 0.01 (1% of sample rate) as specified
time points (2.5 ms). Integrating the signal provided smoothing in order to
clearly identify sustained return of CSP activity. The silent period offset was
labeledasthe firstreturnofEMGactivityofatleast 50%ofbackgroundactivity
the silent period (ms) was calculated relative to TMS onset and then averaged
across trials for each stimulus intensity. The TMS onset was used instead of the
MEP onset to anchor the beginning of the silent period because the threshold of
thesilentperiodis lower thanthatoftheMEP(Wassermannetal., 1993) andwe
wanted to quantify silent periods at lower intensities than those capable of
eliciting a measurable MEP. The largest silent period duration (SPmax) was
defined as the largest value obtained when averaging together the durations
calculated for the 10 trials, regardless of stimulus intensity.
The MEP amplitudes and the silent period durations were each plotted as a
function of stimulus intensity. The slope of the linear relationship between
stimulus intensity and MEP amplitude was estimated in a manner similar to
Wolf and colleagues. (Wolf et al., 2004), from the best-fit line for all points
between the average MEP amplitude obtained at threshold and the MEPmax,
inclusive. The same approach was used to estimate the slope of the relationship
between stimulus intensity and silent period duration. Consistent with previous
Lang and Cohen, 2000; Carroll et al., 2001), we also fit the data using a sigmoid
curve and estimated a peak slope of the stimulus–response curve. In this
approach, the relationship between stimulus intensity and MEP amplitude
was fit using the following equation: y = d + a/(1 + exp(c(b ? x))). This is a
modified sigmoid curve with a vertical offset, where exp(n) = e raised to the
power of n; y = MEP amplitude; x = input intensity in % maximum stimulator
output;d = lowerlimitofthesigmoidfit(verticaloffsetfromzero);a = rangeof
sigmoid fit (i.e. upper limit–lower limit of the fit); b = input intensity where the
sigmoid fit reaches halfway between lower limit and upper limit; and c = para-
meter proportional to the peak slope achieved by the fit (where input intensi-
ty = b).
2.5. Functional outcome measures
The motor function section of the Fugl-Meyer Scale for Upper Extremity
Assessment (FM) (Fugl-Meyer et al., 1975; Berglund and Fugl-Meyer, 1986;
Filiatrault et al., 1991) and the Wolf Motor Function Test were used to
evaluateupperextremitymotor functionateachof thetwotimepointsfordata
collection. WMFT is a laboratory-based performance test designed to quan-
tify motor function in the affected upper extremity in stroke patients (Wolf
et al., 1989; Taub et al., 1993). The time to complete the tasks provides a
reliable and valid indicator of motor task performance by the affected hand
(Morris et al., 2001; Wolf et al., 2001; Wolf et al., 2005) and is responsive to a
variety of interventions aimed at improving motor function in subacute or
chronic stroke patients (Blanton and Wolf, 1999; Kunkel et al., 1999; Whitall
et al., 2000; Page et al., 2001; Schaechter et al., 2002; Sterr et al., 2002;
Duncan et al., 2003; Pierce et al., 2003; Lum et al., 2004; Page et al., 2004).
Only the 6 items requiring fine-motor control were included in the present
analysis because they are most likely to require corticomotor control of the
muscle from which we recorded in this study. The six items are: lifting a can,
liftinga pencil, liftinga paper clip, stacking checkers,flipping cards, turninga
key in a lock.
2.6. Data analysis
2.6.1. Trial-to-trial variability
For each subject we calculated the percent error in estimating with 90%
confidence the true amplitude of the MEPmax based on the mean of 10 trials,
and the minimum number of trials needed to be within 20% of the true MEP
amplitude, as described previously for normal control subjects (Brasil-Neto
et al., 1992a,b). The same approach was applied to the MEPmax latencies and
the CSPmax durations. We note that reduced MEP amplitudes on the lesioned
side will result in higher coefficients of variation even when the absolute
variation in signal (standard deviation) is comparable for the two hemispheres.
Therefore, we used the standard deviation as an aid to interpreting differences
between the two hemispheres in trial-to-trial variation.
L. Koski et al./Neuroscience Research 58 (2007) 19–3121
2.6.2. Goodness-of-fit for curve fitting methods
silent periods were compared using a 3-way repeated measures ANOVAs with
factors fit method (sigmoid, linear), side (lesioned, unlesioned) and session
2.6.3. Test–retest reliability
Threeindicators oftest–retest reliability were calculated for each excitability
estimate: reliability of the group mean (p > 0.05 for a two-tailed paired t-test of
the hypothesis that the group mean changes over time), the intraclass correlation
variation (in %, CV). An ICC of 0.70 or greater was interpreted as indicating
‘‘acceptable’’ reliability. The typical error of measurement (standard error of
the values observed at different points in time vary from the ‘‘true’’ value of that
reliable the estimate. This estimate was used to establish a cutoff value
(1.81 ? typical error) for the group for each excitability measure. When inter-
pretingobserved changes in excitability in an individual patient, this cutoff value
could be used to indicate the level beyond which the change can be considered
statistically significant. The value selected corresponds to a probability of 80%
thatadifference inthe estimatesobservedbeforeandaftertreatmentinanygiven
patient reflects a statistically reliable or ‘‘true’’ change (Hopkins, 2002).
2.6.4. Interside differences in excitability
Significant differences in excitability between the two cerebral hemispheres
were evaluatedin termsofa differenceinthe meansoutsidethe 95%confidence
interval (Hopkins, 2000).
2.6.5. Relationship between excitability and motor function
Pearson’s correlation coefficients were used to evaluate the relationship
between excitability and motor function at session one. We also calculated
reliability indicators for the functional scores derived from the 6-items of the
Demographic and lesion data for each participant are shown
in Table 1.
3.1. Trial-to-trial variation
Table 2 presents the number of trials required to estimate the
MEPmax latency and amplitude, as well as the CSPmax
duration, within 20% of the truevalue. Latencyof the MEPmax
could have been accurately estimated with only four trials for
both hemispheres in all patients. For the unlesioned hemi-
sphere, 5 trials would have been sufficient to estimate the
amplitude of the MEPmax in all but one time point for one
subject. For the lesioned hemisphere 10 trials were required on
average to obtain accurate amplitude estimates; however, >10
trials would be required for at least one time point in 4 of the 9
patients tested. In 7 of 9 patients the standard deviation of the
MEPmax amplitude on the lesioned side was less than or equal
Patient demographics, treatment group, lesion characteristics, and motor impairment
Fronto-temporo-parietal (MCA territory, cortical, subcortical)
Posterior limb, internal capsule
Periventricular, caudate nucleus tail
Frontal and parietal lesion, primary motor cortex
Medial occipital lobe (lingual gyrus and cuneus)
Posterior limb, internal capsule
M:male; F:female;L: left hemisphere; R:righthemisphere;maximalforce:pinchforce forthe affectedhand, measured in kiloPascals(kPa); FM: Fugl-Meyermotor
score prior to first session; MCA: middle cerebral artery. Lesion size is in cubic centimeters.
Number of stimuli required, for each participant, to produce mean excitability estimates within 20% of the true value for that parameter
MEP AmplitudeMEP Latency CSP Duration
Lesioned Unlesioned LesionedUnlesionedLesionedUnlesioned
T1T2 T1T2T1T2T1 T2T1T2 T1T2
T1: session 1; T2: session 2.
L. Koski et al./Neuroscience Research 58 (2007) 19–3122
to the standard deviation for the unlesioned side. This suggests
that the greater proportional error in estimating the mean on the
lesioned side was a function of the asymmetry in amplitudes
between the two sides. However, for two patients the standard
deviation was greater for the lesioned hemisphere, pointing to a
true increase in trial-to-trial variation for that side. These two
patients did not differ from the others on any demographic or
lesion characteristics measured.
On average, the silent period duration could have been
accurately estimated with a single trial on the unlesioned side
and with 6 trials on the lesioned side. Eight trials would have
yielded acceptable estimates for both sessions in all but two
patients. Both the coefficient of variation and the standard
deviation were higher for the lesioned side, pointing to greater
variability in the silent periods measured in the lesioned
hemisphere after stroke.
3.2. The relationship between stimulus intensity and
ShowninTable 3arethegoodness-of-fit statistics (R2values)
for the peak (sigmoid fit) and linear slopes of the MEP
recruitment curves and the silent period curves. Stimulus-
response curve-fitting results are shown for a patient with good
fits in Fig. 1 and for two patients with poor fits in Fig. 2. On
average the fits for the MEP recruitment curves were good
although there was a trend for worse fits to the data from the
lesioned side as compared with the unlesioned side (average
lesioned: R2= 0.78 ? 0.23; average unlesioned R2= 0.89 ?
0.08; F = 3.7, p = 0.09). For silent period curves, the effect of
side was significant, with the lesioned side fitting less well than
theunlesionedside(averagelesioned:R2= 0.77 ? 0.23;average
unlesioned R2= 0.92 ? 10.7; F = 10.1, p = 0.01). For the most
part the pattern of better fits for the unlesioned side held true for
individual patients as well, as illustrated by the data for a
MEPs and silent period data we observed a nonsignificant
interactionofslopecalculationmethodandsession(p = 0.08and
p = 0.07, respectively), with slightly higher R2values using the
sigmoid curve method but only for session 1. We conclude that
neither curve-fitting method was reliably superior to the other.
3.3. Reliability of excitability measures
The mean was stable across the two time points for all
excitability parameters tested, with the exception of the peak
slope of the MEP recruitment curve, which showed a small but
significant increase between the two sessions. Estimates of
test–retest reliability are presented in Table 4.
The ICC was in an acceptable range, i.e., >.70, for all
measures from the lesioned hemisphere, with the exception of
the peak slope of the silent period curve. The lower limits of the
confidence interval were positive for all measures and >0.70
for the amplitude of the MEPmax and the peak slope. This
suggests that the latter parameters would still be found reliable
in a different sample from the same population. Intrasubject
variability was low for MEP latency (7%); moderate for MEP
amplitude (20%), sigmoid-based (peak) recruitment curve
slope (34%), silent period duration (20%), and silent period
slope (23%); and high (poor) for the linear slope of the MEP
recruitment curve (87%) and the peak slope (sigmoid) of the
silent period curve (89%).
The ICCs for MEP-related parameters for stimulation of the
unlesioned hemisphere were lower than those observed for the
lesioned hemisphere. The ICCs for amplitude of the MEPmax,
linear slope of the MEP recruitment curve, silent period
duration, and linear slope of the silent period were all in the low
end of the acceptable range. However, the lower limit of the
confidence interval for these estimates ranged from 0.21–0.46,
suggesting that these results may require replication in a larger
study. Variability was low for MEP latency (7%), silent period
duration (10%) and linear silent period slope (10%); moderate
for MEP amplitude (20%), the linear slope of the MEP
recruitment curve (30%) and the peak (sigmoid) slope of the
Goodness-of-fit statistics (R2values in %) for MEP and silent period curve fitting
MEP recruitment curvesCSP recruitment curves
Sigmoid relation Linear relationSigmoid relationLinear relation
Lesioned UnlesionedLesionedUnlesionedLesioned Unlesioned LesionedUnlesioned
t1 t2t1 t2t1 t2t1t2 t1t2 t1t2t1 t2 t1t2
# of Fits7698669878988599
MEP: peak-to-peak amplitude for motor evoked potential, CSP: duration of the cortical silent period; t1: time 1 (baseline); t2: time 2 (two weeks after baseline); # of
Fits: number of patients whose stimulus–response relation was adequately described by the sigmoid curve or the linear relation, as defined by an R2value ?0.75.
L. Koski et al./Neuroscience Research 58 (2007) 19–3123
silent period (24%); and high (poor) for the peak slope of the
MEP recruitment curve (82%).
3.4. Interside differences in excitability
Table 5 shows the side-to-side difference in excitability
across the cerebral hemispheres for the data collected at
baseline. The MEP amplitude and linear slope of the MEP
unlesioned side. MEP latencies were greater on the lesioned
side. In contrast, the peak slope of the silent period recruitment
curve was greater on the lesioned side than on the unlesioned
When comparing corticomotor excitability in patients with
cortical versus subcortical lesions we observed trends that are
consistent with the results observed previously in acute stroke
patients (Liepert et al., 2005). Specifically, the latency of the
MEP was prolonged, the amplitude was suppressed, and the
silent periods (lower half). The raw data are represented as dots, the sigmoid curve that was fit to the data is represented by a dashed line, and the linear fit to the data
from the threshold to the maximum MEP amplitude (or silent period duration) is shown as a solid line. MEP = motor evoked potential amplitude in uV.
CSP = duration of the cortical silent period.Peak slope = the peak slope estimatedfrom the sigmoid curve fitting method.Linear slope = the b parameter of the linear
fit. MEP Data: The MEPs on the lesioned side approached a plateau at 100% maximum stimulator output (MSO) in session 1 and clearly pleateaued in session 2.
MEPs on the unlesioned side plateaued earlier. The difference between the two sessions mirrored that seen on the lesioned side, with a slightly earlier plateau in
session 2 (75% MSO) as compared with session 1 (80% MSO). The same pattern was reproduced in the maximum MEP amplitudes, peak slopes and linear slopes,
with slightly higher sensitivity to stimulation observed for both sides in session 2 as compared with session 1. CSP Data: Intracortical inhibition in the lesioned
hemisphereplateauedat aroundthesame stimulusintensityatbothsessions.Intheunlesionedhemispheretherewasno clearplateauforeithersessionat100%MSO,
such that a straight line better describes the relationship between stimulus intensity and silent period duration than a sigmoid curve. For both hemispheres the
maximum silent period duration was slightly higher in session 2 than in session 1. This follows the pattern observed for the MEP amplitudes.
L. Koski et al./Neuroscience Research 58 (2007) 19–3124
slope of the stimulus–response curve was reduced on the
lesioned side in patients with subcortical stroke, although the
interaction between lesion site and hemisphere did not reach
significance in our small sample (N = 3 cortical, 6 subcortical;
see Fig. 3). In contrast, silent period duration and the slope of
the stimulus-duration recruitment curve were longer in both
hemispheres for the cortical group compared with the
subcortical group, a result opposite to that found in acute
stroke (Liepert et al., 2005).
3.5. Reliability of functional measures
The average time to complete the 6 fine-motor items
correlated highly with the average time to complete all 15
WMFT scale items (0.98). Time to complete the 6-item WMFT
with the affected upper extremity did not change significantly
over the two-week period (average decrease: 0.6 s, S.D.: 9.3)
The standard error of measurement was 6.6 s (CV = 24%) and
the ICC was 0.96. A difference of greater than ?11.9 s in a
given subject would be considered to reflect a true change as
definedinthisstudy.Fortheunaffected extremity,thechange in
means was not significant (decrease 0.1 s, S.D.: 0.6), the
standard error was 0.4 s (CV = 16%) and the cutoff was ?0.8 s.
The ICC was only 0.51. The Fugl-Meyer Motor score for the
affected upper extremity showed an average change of 0.6
points (S.D.: 3.7), a standard error of measurement of 2.6
(CV = 5%), and an ICC of 0.86. The cutoff for this score was
Fig. 2. Data from two patients who showed noisy and/or unreliable stimulus–response relationships on the lesioned side for motor evoked potentials (upper half,
subject 12) and cortical silent periods (lower half, subject 06). We present the data for both the lesioned and unlesioned side for comparison, noting that the data from
the unlesioned side behaved in the predicted fashion. See Fig. 1 legend for labeling conventions. MEP Data: For subject 12, the MEPamplitudes on the lesioned side
did not show a clear monotonic increase with stimulus intensity, resulting in poor curve fits especially in session 2. CSP Data: For subject 06, the silent period
durations on the lesioned side did not show a clear monotonic increasewith stimulus intensity. Our approach for linear modeling of the data included only the values
between the threshold and the maximum silent period and thus yielded a reasonable fit, whereas the more inclusive sigmoid curve fitting method performed less well
for these data.
L. Koski et al./Neuroscience Research 58 (2007) 19–31 25
3.6. Relationship between excitability and motor function
We did not observe any correlations between parameters of
excitability and measures of motor function, with one
exception. There was a negative correlation between perfor-
mance on the Fugl-Meyer motor score at baseline and latency
of the MEP from the lesioned hemisphere at baseline, such that
subjects with normal latencies showed better motor function
than those with atypically long latencies (r = ?0.77, p = 0.01).
4.1. Trial-to-trial variability in stroke patients
Earlier work in normal control subjects suggested that five
trials were sufficient to estimate the true MEP amplitudewithin
20% (Brasil-Neto et al., 1992a,b), and the silent period duration
within 10% (Kimiskidis et al., 2005). In the present study, ten
trials were more than sufficient to estimate both the latency of
the MEPmax and the duration of the CSPmax with a high
degree of accuracy for every patient with stroke. It also allowed
estimation of the amplitude of the MEPmax within 20% for the
unlesioned side. Unfortunately, the amplitude of the MEPmax
on the lesioned side showed much greater variability. For most
patients this was probably an artifact of the lower signal
obtained on the lesioned side; however, two patients did show
higher absolute variation in MEPs on the lesioned side and this
was observed consistently in both testing sessions. The present
results suggest that for stimulation of the lesioned hemisphere
after stroke, a minimum of 10 trials is required on average to
obtain acceptable estimates of the true MEP. Pilot testing on
every patient could allow investigators to titrate the number of
trials to the individual; however, acquiring more than 10 trials
may not be feasible due to the extended time involved for the
patient. This error in estimating the true MEP amplitudes for
the lesioned side also contributederror to our estimates ofinter-
that rely on measurements of MEPamplitude in stroke patients.
4.2. Intraclass correlation coefficients for excitability
For all but one parameter the group mean was stable over
time, as reported previously for the finger extensor muscles in a
similar population of stroke patients (Butler et al., 2005). The
reliability coefficients suggest that parameters reflecting
corticomotor excitability (maximum MEP amplitude), the
capacity for increasing recruitment of corticospinal projections
(linear slope of the recruitment curve), intracortical inhibitory
circuits (silent period duration) and their recruitment as a
function of stimulus intensity (linear slope of the silent period)
are reliable enough for use as an outcome measure in studies of
patients with similar characteristics to those tested here.
The ICCs for amplitude of the MEP were within the range
reported in studies of normal control subjects (Carroll et al.,
2001; Kamen, 2004) and slightly better than those reported for
Test-retest reliability estimates
Mean (S.D.) T1 Mean (S.D.) T2Mean changeS.D. changeICCLLCISEMCut-off (?)
Amplitude of MEPmax (mV)
Latency of MEPmax
Linear slope to MEPmax
MEP peak slope (sigmoid)
Duration of SPmax
Linear slope to SPmax
SP peak slope (sigmoid)
Amplitude of MEPmax
Latency of MEPmax
Linear slope to MEPmax
MEP peak slope (sigmoid)
Duration of SPmax
Linear slope to SPmax
SP peak slope (sigmoid)
ICC: intraclass correlation coefficient; LLCI: lower limit of the 90% confidence interval for the intraclass correlation coefficient; SEM: standard error of
measurement; Cut-off: limits of normal variation for demonstrating statistically reliable change defined as 1.81 * SEM (80% probability of a ‘‘true’’ change); Mean:
mean of the change scores; S.D.: standard deviation of the change scores. The symbol - refers to correlation coefficients in the negative range, i.e., invalid.
*: Significant increase in mean from session 1 to session 2 with 95% confidence.
Side-to-side differences in motor excitability (Unlesioned-Lesioned)
MEP Peak slope
MEP Linear slope
Duration of SPmax
SP peak slope
SP linear slope
?5 ms (5)*
*: Difference between sides significant at 95% confidence level.
L. Koski et al./Neuroscience Research 58 (2007) 19–3126
MEPs obtained at a lower stimulation intensity (McDonnell
et al., 2004). On the lesioned side, the ICC for the peak slope of
the MEP recruitment curve was comparable to that seen in
normal subjects (Devanne et al., 1997; Carroll et al., 2001;
Malcolm et al., 2006). However, the average goodness-of-fit of
the sigmoid curve and the average peak slope for the lesioned
side were significantly different across the two testing sessions.
Moreover, the sigmoid model includes a larger number of
parameters than the simple linear model without improving the
fit to the data.We conclude that the recruitment of corticospinal
projections as indicated by the stimulus–response curve can be
more reliably represented by a simple linear slope than by a
The ICCs for the silent period duration in both hemispheres
were superior to those obtained in a previous reliability study
from ourlaboratory,wheresilent periods were measured during
stimulation just above the resting motor threshold (Koski et al.,
The reliability coefficients were comparable between
hemispheres for the parameters based on the silent period
but lower on the unlesioned side compared with the lesioned
side for parameters based on the MEP. However, in interpreting
this difference in reliability between the hemispheres we must
consider that intrasubject reliability coefficients are influenced
by intersubject variation. High ICCs can be obtained simply by
increasing the heterogeneity of the sample tested. For example,
MEP latencies on the lesioned side were normal in some
patients and abnormally lengthened in others. This intersubject
variation produced an excellent ICC on the lesioned side
compared with the poor reliability for the unlesioned side,
where there was little variation between subjects (see Table 4,
4.3. Stimulus–response curves for silent period duration
Studies on the nature of the relationship between stimulus
intensity and silent period duration in normal subjects yielded
conflicting results (Kimiskidis et al., 2005). Kimiskidis and
colleagues reported a plateau in the increase of silent period
duration with increasing stimulus intensities such that the
Fig. 3. Plots of cell means (with standard deviation error bars) demonstrating trends in the effects of lesion site (C = cortical; S = subcortical) and hemisphere (open
circles = Lesioned;filledsquares = Unlesioned).Dataare collapsedacrosssession.(A)LatencyofthemaximumMEPin ms.(B)AmplitudeofthemaximumMEPin
uV. (C) Peak slopeof the sigmoid MEPrecruitmentcurve. (D) Linear slope of the MEP recruitment curve. (E) Durationof the maximumsilent period in ms. (F) Peak
slope of the sigmoid silent period intensity-response curve. (G) Linear slope of the silent period intensity response curve. Three-way mixed analyses of variance
(lesion site ? hemisphere ? session) revealed longer MEP latencies (p ? 0.05), smaller amplitudes (p = 0.06) and lower recruitment slopes (p ? 0.05) in the
patients with subcortical lesions, and longer MEP latencies (p ? 0.05), smaller amplitudes (p = 0.06) and lower recruitment slopes (p ? 0.05) in the lesioned
hemisphere. The interaction suggested by the cell means for latency, amplitude and slope of the sigmoid recruitment curve did not reach significance in this small
sample(n = 3cortical,n = 6subcortical).Thesilentperioddurationwaslongerinthegroupwithcorticallesionsregardlessofside(p ? 0.05),andasimilarresultwas
seen for the linear slope of the silent period recruitment curve (p ? 0.05).
L. Koski et al./Neuroscience Research 58 (2007) 19–3127
relationship was best described by a sigmoid curve. They
concluded that there is a threshold for activating all inhibitory
interneurons participating in a particular network, although
they also acknowledged conflicting data from other investiga-
tors who have observed a linear stimulus–response relationship
with no plateau.
In the present study, the peak slope of the sigmoid curve
representing the stimulus–response relation for silent period
showed no correlation across time within subjects for the
lesioned side and correlated only moderately (0.54) on the
a curve to some data sets, which added noise to the data as a
whole. However, in examining the individual response curves
not for others. The absence of a plateau was observed more
often, but not consistently, for the unaffected side and was seen
for both sides in two patients. The presence or absence of a
plateau could not be explained on the basis of specific lesion
characteristics. Thus we conclude that the absence of a plateau
may simply reflect normal variation that places the threshold
for activating all inhibitory interneurons outside limits of the
4.4. Standard error of measurement for excitability
Measurement error was higher on the lesioned side than on
the unlesioned side for the MEP latency and for the parameters
show comparable or even greater reliability coefficients (ICCs)
for the lesioned side relative to the unlesioned side, an
intervention must produce a comparatively large change on the
lesioned side to exceed the normal session-to-session variation
in these parameters.
The opposite pattern was observed for the MEP amplitude
and peak slope of the MEP recruitment curve, where the
standard error was smaller on the lesioned side. The MEP
amplitude on the lesioned hemisphere also varied less between
subjects. We suppose that this difference may be related to a
ceiling effect in the MEP amplitudes for the lesioned
hemisphere, such that there is an upper limit on the ability
to recruit additional corticospinal projections by increasing the
intensity of stimulation on the lesioned hemisphere. This is
consistent with our observation of significant interside
differences in the slope of the recruitment curves at baseline.
The earlier study by Butler and colleagues (Butler et al., 2005)
also suggested a trend toward greater slopes on the unlesioned
relative to the lesioned hemisphere although this did not reach
their threshold for statistical significance.
In normal control subjects measurement error has been
of variation. Comparing the CVs for the unlesioned side in our
patients with those reported in the normal control literature, we
observe comparable values for the MEPmax latency (Nielsen,
13% (Nielsen, 1996) and 6% (Kamen, 2004; Orth and
Rothwell, 2004)). The coefficient of variation for the lesioned
side was even higher (46%) but a comparison of the standard
deviations for the two sides reveals that this is an artifact of the
small MEP amplitudes entering calculation in the denominator.
There is little published data on variation in MEP recruitment
curve slopes across repeated sessions, although substantial CVs
(?70%) were reported for the linear slope of the recruitment
curve in the finger extensor muscles (Wolf et al., 2004).
An early study of silent periods in normal subjects and in
stroke patients indicated that the variation across five sessions
did not exceed 15% (Kukowski and Haug, 1992). Orth and
Rothwell (Orth and Rothwell, 2004) reported that the median
test–retest CV in a group of 11 subjects was <5%, for silent
periods measured during stimulation at 150% of active motor
threshold and we have observed similar results when
stimulating just above the motor threshold in normal control
subjects (Koski et al., 2005). In the present study, the typical
error expressed as a CV was acceptable (10%) for the
unlesioned side but high (48%) for the lesioned side. The
duration of the silent period did not differ between the two
hemispheres, supporting the conclusion that the silent period is
less reliable in the lesioned hemisphere.
4.5. Reliability of functional measures
The ICC of 0.96 for time to complete the 6 fine-motor items
of the WMFT with the affected hand was identical to that
reported previously for the whole scale in a larger sample of
stroke patients (Wolf et al., 2005), suggesting that the 6-item
version can be used reliably when fine-motor coordination of
the affected hand is the focus of study. Reliability for the less-
affected hand was poor and far below that reported for the time
to complete derived from the full 15-item WMFT (0.51 versus
0.92). This difference in the ICCs for the affected and less-
affected extremities can be attributed to high intersubject
variation in the time to complete tasks with the affected
extremity, reflecting variability in the level of motor impair-
ment. For measuring changes in functional performance, the
ratio of affected hand to less-affected hand performance yields
astandarderrorof2.6(affected hand2.6timesslowerthan less-
affected hand), an ICC of 0.97, and a cutoff of ?4.7.
4.6. Limitations of this study
This study is limited in part by the small number of patients
tested. The lower limit of the confidence interval for our
estimates of ICC was at least 0.50 for the maximum MEP
amplitude, the recruitment curve slopes and the silent period
duration in the lesioned hemisphere, suggesting that our
estimates of these measures would still be found to be reliable
even if more subjects were added. For other excitability
measures, including those from the unlesioned side, our
conclusions must remain tentative.
Application of these results are limited to use in stroke
patients with similar characteristics to those tested in this study,
and where the methods for estimating excitability are similar.
Beyond these parameters, our estimates of reliability should be
used as guidelines only. Degree of motor impairment was
L. Koski et al./Neuroscience Research 58 (2007) 19–3128
relatively homogenous in the group tested here but lesions
differed greatly, with some lesions affecting the cortex and
others restricted to subcortical and brainstem regions. Future
studies of excitability will be needed to clarify (a) whether the
effect of lesion site interacts with time elapsed since stroke and
(b) whether the observed differences in excitability influence
test–retest reliability. However, the range of lesions included in
the present study is typical of the patient groups enrolled in
current clinical rehabilitation trials involving measurement of
excitability and thus increases the generalizability of our
We have provided estimates for multiple indicators of the
test–retest reliability of corticomotor and intracortical excit-
ability measures in the lesioned and unlesioned hemisphere of
individuals who have sustained a stroke. These data could be
used to estimate the required sample size in a study in which
excitability parameters are used as outcome measures.
Alternatively they might guide the selection of one excitability
indicator over another when practical constraints require
limiting the number of indicators that can be measured. We
have also provided cutoff values that can be used to identify
statistically reliable change in patients undergoing therapeutic
intervention. These data should be useful for interpreting
individual treatment effects in rehabilitation studies where
variability in the response to treatment is the norm. For
example,a recent paper by Liepert (Liepert, 2006) reported that
improvements in motor function following constraint-induced
movement therapy were accompanied in some patients by
increases in intracortical inhibition and in other patients by
decreases in intracortical inhibition within the lesioned
hemisphere.Hypotheses derivedfromthe analysisofindividual
treatment effects may be tested further with the aim of
identifyingsubgroupsofstroke patientswhorespond totherapy
through specific mechanisms. However, the contribution of
such studies will be strengthened if we are able to discriminate
between variability in therapeutic response and test–retest
reliability of the neurophysiological measure.
The selection of excitability measures obtained during
active muscle contraction has implications for rehabilitation
studies. MEPs obtained from a muscle at rest are sensitive to
peripheral interventions that increase corticomotor excitabil-
ity, such as ischemic anaesthesia or amputation of the distal
extremity, whereas MEPs from a contracted muscle are not
affected by these manipulations (Ridding and Rothwell, 1995,
1997). Ridding and Rothwell have argued that the increases in
motor map size that have been observed after therapeutic
interventions in stroke patients might reflect, not long-term
rather an increase in excitability. In the absence of any clinical
trial data it remains an open question whether the MEPs
obtained from a contracted muscle are sensitive to rehabili-
tative interventions and whether they might provide a more
valid indicator of long-term plastic reorganization within the
We gratefully acknowledge the contributions of Michelle
Prettyman, Chris Hahn, Steve Cen, Samantha Underwood, and
all others who assisted in the coordination of this study and in
the collection and processing of the clinical and functional
movement data. This work was supported by General Clinical
Research Centers Program M01-RR00865 (UCLA), NIH
NS45485 from NINDS to CW, and K23-NS045764 from
NINDS to AW.
Berglund, K., Fugl-Meyer, A.R.,1986.Upper extremityfunction in hemiplegia.
A cross-validation study of two assessment methods. Scand. J. Rehabil.
Med. 18 (4), 155–157.
Blanton, S., Wolf, S.L., 1999. An application of upper-extremity constraint-
induced movement therapy in a patient with subacute stroke. Phys. Ther. 79
Brasil-Neto, J.P., Cohen, L.G., Panizza, M., Nilsson, J., Roth, B.J., Hallet, M.,
1992a. Optimal focal transcranial magnetic brain stimulation of the human
motor cortex: effects of coil orientation, shape of the induced current pulse,
and stimulus intensity. J. Clin. Neurophysiol. 9, 132–136.
Brasil-Neto, J.P., McShane, L.M., Fuhr, P., Hallett, M., Cohen, L.G., 1992b.
Topographic mapping of the human motor cortex with magnetic stimula-
tion: factors affecting accuracy and reproducibility. Electroencephalogr.
Clin. Neurophysiol. 85 (1), 9–16.
Butler, A.J., Kahn, S., Wolf, S.L., Weiss, P., 2005. Finger extensor variability in
Byrnes, M.L., Thickbroom, G.W., Phillips, B.A., Mastaglia, F.L., 2001. Long-
term changes in motor cortical organisation after recovery from subcortical
stroke. Brain Res. 889 (1–2), 278–287.
Byrnes, M.L., Thickbroom, G.W., Phillips, B.A., Wilson, S.A., Mastaglia, F.L.,
1999. Physiological studies of the corticomotor projection to the hand after
subcortical stroke. Clin. Neurophysiol. 110 (3), 487–498.
Carroll, T.J., Riek, S., Carson, R.G., 2001. Reliability of the input-output
properties of the cortico-spinal pathway obtained from transcranial mag-
netic and electrical stimulation. J. Neurosci. Methods 112 (2), 193–202.
Cicinelli, P., Traversa, R., Bassi, A., Scivoletto, G., Rossini, P.M., 1997a.
Interhemispheric differencesof handmusclerepresentationin humanmotor
cortex. Muscle. Nerve. 20 (5), 535–542.
Cicinelli, P., Traversa, R., Rossini, P.M., 1997b. Post-stroke reorganization of
brain motor output to the hand: a 2–4 month follow-up with focal magnetic
transcranial stimulation. Electroencephalogr. Clin. Neurophysiol. 105 (6),
Cincotta, M., Borgheresi, A., Guidi, L., Macucci, M., Cosottini, M., Lam-
bruschini, P., Benvenuti, F., Zaccara, G., 2000. Remote effects of cortical
dysgenesis on the primary motor cortex: evidence from the silent period
following transcranial magnetic stimulation. Clin. Neurophysiol. 111 (8),
Civardi, C., Boccagni, C., Vicentini, R., Bolamperti, L., Tarletti, R., Varrasi, C.,
Corneal,S.F., Butler, A.J., Wolf, S.L., 2005. Intra-and intersubjectreliabilityof
abductor pollicis brevis muscle motor map characteristics with transcranial
magnetic stimulation. Arch. Phys. Med. Rehabil. 86 (8), 1670–1675.
Daskalakis, Z.J., Molnar, G.F., Christensen, B.K., Sailer, A., Fitzgerald, P.B.,
Chen, R., 2003. An automated method to determine the transcranial
magnetic stimulation-induced contralateral silent period. Clin. Neurophy-
siol. 114 (5), 938–944.
De Gennaro, L., Ferrara, M., Bertini, M., Pauri, F., Cristiani, R., Curcio, G.,
Romei, V., Fratello, F., Rossini, P.M., 2003. Reproducibility of callosal
effects of transcranial magnetic stimulation (TMS) with interhemispheric
paired pulses. Neurosci. Res. 46 (2), 219–227.
L. Koski et al./Neuroscience Research 58 (2007) 19–31 29
Delvaux, V., Alagona, G., Gerard, P., De Pasqua, V., Pennisi, G.A., de
Noordhout, A.M., 2003. Post-stroke reorganization of hand motor area: a
1-year prospective follow-up with focal transcranial magnetic stimulation.
Clin. Neurophysiol. 114 (7), 1217–1225.
Devanne, H., Lavoie, B.A., Capaday, C., 1997. Input-output properties and
gain changesinthe humancorticospinal pathway. Exp. Brain Res. 114(2),
Duncan, P., Studenski, S., Richards, L., Gollub, S., Lai, S.M., Reker, D.,
Perera, S., Yates, J., Koch, V., Rigler, S., Johnson, D., 2003. Randomized
clinical trial of therapeutic exercise in subacute stroke. Stroke 34 (9),
Filiatrault, J., Arsenault, A.B.,Dutil, E., Bourbonnais, D., 1991. Motor function
and activities of daily living assessments: a study of three tests for persons
with hemiplegia. Am. J. Occup. Ther. 45 (9), 806–810.
Folstein, M.F., Folstein, S.E., McHugh, P.R., 1975. Mini-mental state A
practical method for grading the cognitive state of patients for the clinician.
J. Psychiatr. Res. 12 (3), 189–198.
Fritz, C., Braune, H.J., Pylatiuk, C., Pohl, M., 1997. Silent period following
transcranial magnetic stimulation: a study of intra- and inter-examiner
reliability. Electroencephalogr. Clin. Neurophysiol. 105 (3), 235–240.
Fugl-Meyer, A.R., Jaasko, L., Leyman, I., Olsson, S., Steglind, S., 1975. The
post-stroke hemiplegic patient. 1. a method for evaluation of physical
performance. Scand. J. Rehabil. Med. 7 (1), 13–31.
Hopkins, W., 2000. Reliability from consecutive pairs of trials (Excel spread-
sheet). In: A new view of statistics, sportsci.org: Internet Society for Sport
Hopkins, W., 2002. Calculating likely (confidence) limits and likelihoods for
true values (Excel spreadsheet). In: A new view of statistics, sportsci.org:
Internet Society for Sport Science, 2005.
Kaelin-Lang, A., Cohen, L.G., 2000. Enhancing the quality of studies using
transcranial magnetic and electrical stimulation with a new computer-
controlled system. J. Neurosci. Methods 102 (1), 81–89.
Kamen, G., 2004. Reliability of motor-evoked potentials during resting and
active contraction conditions. Med. Sci. Sports Exerc. 36 (9), 1574–1579.
Kiers, L., Cros, D., Chiappa, K.H., Fang, J., 1993. Variability of motor
Clin. Neurophysiol. 89 (6), 415–423.
Kimiskidis, V.K., Papagiannopoulos, S., Sotirakoglou, K., Kazis, D.A., Kazis,
A., Mills, K.R., 2005. Silent period to transcranial magnetic stimulation:
construction and properties of stimulus–response curves in healthy volun-
teers. Exp. Brain Res. 163 (1), 21–31.
Koski, L., Schrader, L.M., Wu, A.D., Stern, J.M., 2005. Normative data on
changes in transcranial magnetic stimulation measures over a ten hour
period Original article. Clin. Neurophysiol. 116, 2099–2109.
Kukowski, B., Haug, B., 1992. Quantitative evaluation of the silent period,
evoked by transcranial magnetic stimulation during sustained muscle
contraction, in normal man and in patients with stroke. Electromyogr Clin.
Neurophysiol. 32 (7–8), 373–378.
Kunkel, A., Kopp, B., Muller, G., Villringer, K., Villringer, A., Taub, E., Flor,
H., 1999. Constraint-induced movement therapy for motor recovery in
chronic stroke patients. Arch. Phys. Med. Rehabil. 80 (6), 624–628.
Liepert, J., 2006. Motor cortex excitability in stroke before and after constraint-
induced movement therapy. Cogn. Behav. Neurol. 19 (1), 41–47.
Liepert, J., Bauder, H., Wolfgang, H.R., Miltner, W.H., Taub, E., Weiller, C.,
2000. Treatment-induced cortical reorganization after stroke in humans.
Stroke 31 (6), 1210–1216.
Liepert, J., Restemeyer, C., Kucinski, T., Zittel, S., Weiller, C., 2005. Motor
Liepert, J., Uhde, I., Graf, S., Leidner, O., Weiller, C., 2001. Motor cortex
plasticity during forced-use therapy in stroke patients: a preliminary study.
J. Neurol. 248 (4), 315–321.
Lum, P.S., Taub, E., Schwandt, D., Postman, M., Hardin, P., Uswatte, G., 2004.
Automated Constraint-Induced Therapy Extension (AutoCITE) for move-
ment deficits after stroke. J. Rehabil. Res. Dev. 41 (3A), 249–258.
Maeda, F., Gangitano, M., Thall, M., Pascual-Leone, A., 2002. Inter- and intra-
individual variability of paired-pulse curves with transcranial magnetic
stimulation (TMS). Clin. Neurophysiol. 113 (3), 376–382.
Malcolm, M.P., Triggs, W.J., Light, K.E., Shechtman, O., Khandekar, G.,
Gonzalez Rothi, L.J., 2006. Reliability of motor cortex transcranial mag-
netic stimulation in four muscle representations. Clin. Neurophysiol. 117
McDonnell, M.N., Ridding, M.C., Miles, T.S., 2004. Do alternate methods of
analyzing motor evoked potentials give comparable results? J. Neurosci.
Methods 136 (1), 63–67.
McGraw, K., Wong, S., 1996. Forming inferences about some intraclass
correlation coefficients. Psychol. Methods 1 (1), 30–46.
Miranda, P.C., de Carvalho, M., Conceicao, I., Luis, M.L., Ducla-Soares, E.,
1997. A new method for reproducible coil positioning in transcranial
magnetic stimulation mapping. Electroencephalogr. Clin. Neurophysiol.
105 (2), 116–123.
Morris, D.M., Uswatte, G., Crago, J.E., Cook, E.W., Taub 3rd., E., 2001. The
reliability of the wolf motor function test for assessing upper extremity
function after stroke. Arch. Phys. Med. Rehabil. 82 (6), 750–755.
Mortifee, P., Stewart, H., Schulzer, M., Eisen, A., 1994. Reliability of tran-
scranial magnetic stimulation for mapping the human motor cortex. Elec-
troencephalogr. Clin. Neurophysiol. 93 (2), 131–137.
action potentials evoked by transcranial magnetic stimulation. J. Clin.
Neurophysiol. 13 (5), 423–434.
Orth, M., Rothwell, J.C., 2004. The cortical silent period: intrinsic variability
and relation to thewaveform of the transcranial magnetic stimulation pulse.
Clin. Neurophysiol. 115 (5), 1076–1082.
Page, S.J., Sisto, S., Levine, P., McGrath, R.E., 2004. Efficacy of modified
constraint-induced movement therapy in chronic stroke: a single-blinded
randomized controlled trial. Arch. Phys. Med. Rehabil. 85 (1), 14–18.
Page, S.J., Sisto, S.A., Levine, P., Johnston, M.V., Hughes, M., 2001. Modified
constraint induced therapy: a randomized feasibility and efficacy study. J.
Rehabil. Res. Dev. 38 (5), 583–590.
Pierce, S.R., Gallagher, K.G., Schaumburg, S.W., Gershkoff, A.M., Gaughan,
J.P., Shutter, L., 2003. Home forced use in an outpatient rehabilitation
program for adults with hemiplegia: a pilot study. Neurorehabil. Neural.
Repair. 17 (4), 214–219.
motor cortex input-output characteristics. J. Physiol. 546, 605–613 (Pt 2).
Platz, T., van Kaick, S., Moller, L., Freund, S., Winter, T., Kim, I.H., 2005.
stroke: a fTMS study. J. Neurol..
Ridding, M.C., Rothwell, J.C., 1995. Reorganisation in human motor cortex.
Can. J. Physiol. Pharmacol. 73 (2), 218–222.
Ridding, M.C., Rothwell, J.C., 1997. Stimulus/response curves as a method of
measuring motor cortical excitability in man. Electroencephalogr. Clin.
Neurophysiol. 105 (5), 340–344.
Schaechter, J.D., Kraft, E., Hilliard, T.S., Dijkhuizen, R.M., Benner, T.,
Finklestein, S.P., Rosen, B.R., Cramer, S.C., 2002. Motor recovery and
cortical reorganization after constraint-induced movement therapy in
stroke patients: a preliminary study. Neurorehabil. Neural. Repair. 16
Sterr, A., Elbert, T., Berthold, I., Kolbel, S., Rockstroh, B., Taub, E., 2002.
Longer versus shorter daily constraint-induced movement therapy of
chronic hemiparesis: an exploratory study. Arch. Phys. Med. Rehabil. 83
Stewart, L.M., Walsh, V., Rothwell, J.C., 2001. Motor and phosphene thresh-
olds: a transcranial magnetic stimulation correlation study. Neuropsycho-
logia 39 (4), 415–419.
Stinear, J.W., Byblow, W.D., 2004. Rhythmic bilateral movement training
modulates corticomotor excitability and enhances upper limb motricity
poststroke: a pilot study. J. Clin. Neurophysiol. 21 (2), 124–131.
Taub, E., Miller, N.E., Novack, T.A., Cook 3rd., E.W., Fleming, W.C., Nepo-
muceno, C.S., Connell, J.S., Crago, J.E., 1993. Technique to improve
chronic motor deficit after stroke. Arch. Phys. Med. Rehabil. 74 (4),
Tergau, F., Wischer, S., Somal, H.S., Nitsche, M.A., Mercer, A.J., Paulus, W.,
Steinhoff, B.J., 2003. Relationship between lamotrigine oral dose, serum
level and its inhibitory effect on CNS: insights from transcranial magnetic
stimulation. Epilepsy Res. 56 (1), 67–77.
L. Koski et al./Neuroscience Research 58 (2007) 19–31 30
Traversa, R., Cicinelli, P., Oliveri, M., Giuseppina Palmieri, M., Filippi, M.M., Download full-text
Pasqualetti, P., Rossini, P.M., 2000. Neurophysiological follow-up of motor
cortical output in stroke patients. Clin. Neurophysiol. 111 (9), 1695–1703.
Traversa, R., Cicinelli, P., Pasqualetti, P., Filippi, M., Rossini, P.M., 1998.
Follow-up of interhemispheric differences of motor evoked potentials from
Turton, A., Wroe, S., Trepte, N., Fraser, C., Lemon, R.N., 1996. Contralateral
and ipsilateral EMG responses to transcranial magnetic stimulation during
recovery of arm and hand function after stroke. Electroencephalogr. Clin.
Neurophysiol. 101 (4), 316–328.
Uswatte, G., Taub, E., Morris, D., Vignolo, M., McCulloch, K., 2005. Relia-
bility and validity of the upper-extremity Motor Activity Log-14 for
measuring real-world arm use. Stroke 36 (11), 2493–2496.
Uy, J., Ridding, M.C., Miles, T.S., 2002. Stability of maps of human motor
cortex made with transcranial magnetic stimulation. Brain Topogr. 14 (4),
Wassermann, E. M., 1998. Risk and safety of repetitive transcranial magnetic
stimulation: report and suggested guidelines from the International Work-
7, 1996. Electroencephalogr. Clin. Neurophysiol. 108(1): 1–16.
Wassermann, E.M., 2002. Variation in the response to transcranial magnetic
brain stimulation in the general population. Clin. Neurophysiol. 113 (7),
Wassermann, E.M., Pascual-Leone, A., Valls-Sole, J., Toro, C., Cohen, L.G.,
Hallett, M., 1993. Topography of the inhibitory and excitatory responses to
transcranial magnetic stimulation in a hand muscle. Electroencephalogr.
Clin. Neurophysiol. 89 (6), 424–433.
Whitall, J., McCombe Waller, S., Silver, K.H., Macko, R.F., 2000. Repetitive
bilateral armtraining with rhythmic
motor function in chronic hemiparetic stroke. Stroke 31 (10), 2390–
Winstein, C.J., Miller, J.P., Blanton, S., Taub, E., Uswatte, G., Morris, D.,
Nichols, D., Wolf, S., 2003. Methods for a multisite randomized trial to
investigate the effect of constraint-induced movement therapy in improving
upper extremity function among adults recovering from a cerebrovascular
stroke. Neurorehabil. Neural. Repair. 17 (3), 137–152.
Wittenberg, G.F., Chen, R., Ishii, K., Bushara, K.O., Eckloff, S., Croarkin, E.,
Taub, E., Gerber, L.H., Hallett, M., Cohen, L.G., 2003. Constraint-induced
therapyin stroke: magnetic-stimulationmotormaps and cerebral activation.
Neurorehabil. Neural. Repair. 17 (1), 48–57.
Wolf, S.L., Butler, A.J., Campana, G.I., Parris, T.A., Struys, D.M., Weinstein,
S.R., Weiss, P., 2004. Intra-subject reliability of parameters contributing to
maps generated by transcranial magnetic stimulation in able-bodied adults.
Clin. Neurophysiol. 115 (8), 1740–1747.
Wolf, S.L., Catlin, P.A., Ellis, M., Archer, A.L., Morgan, B., Piacentino, A.,
2001. Assessing Wolf motor function test as outcome measure for research
in patients after stroke. Stroke 32 (7), 1635–1639.
Wolf, S.L., Lecraw, D.E., Barton, L.A., Jann, B.B., 1989. Forced use of
hemiplegic upper extremities to reverse the effect of learned nonuse
among chronic stroke and head-injured patients. Exp. Neurol. 104 (2),
Wolf, S.L.,Thompson,P.A., Morris,D.M., Rose,D.K.,Winstein,C.J.,Taub, E.,
Giuliani, C., Pearson, S.L., 2005. The EXCITE Trial: Attributes of the Wolf
Motor Function Test in Patients with Subacute Stroke. Neurorehabil.
Neural. Repair. 19, 194–205.
L. Koski et al./Neuroscience Research 58 (2007) 19–31 31