Clinical Neurophysiology 115 (2004) 1748–1759

www.elsevier.com/locate/clinph

EMG for assessing the recovery of voluntary movement

after acute spinal cord injury in man
Blair Calanciea,b,*, Maria R. Molanoa, James G. Brotona
a
The Miami Project to Cure Paralysis and Department of Neurological Surgery, University of Miami School of Medicine, Miami, FL, USA
b
Department of Neurosurgery, SUNY’s Upstate Medical University, Syracuse, 750 E. Adams St, IHP #1213, Syracuse, NY, USA
Accepted 3 March 2004
Available online 15 April 2004

Abstract
Objective: Multi-channel electromyogram (EMG) was used to examine the pattern and time-course of voluntary contraction recovery in
subjects with acute traumatic spinal cord injury (SCI), concentrating on the latest time after injury at which a given muscle would begin to
show voluntary recruitment.
Methods: We conducted repeated measures of voluntary contractions of 12 lower limb muscles (for all subjects) and 12 upper-limb
muscles (for subjects with cervical injury), beginning within days of the injury and extending for 1 or more years post-injury. The EMG
interference pattern was scored in a blinded fashion from tape records.
Results: We recruited 229 subjects, including 152 from whom repeated measures were made. Several different patterns of recovery were
identified. For persons with motor-incomplete injury to the cervical or thoracic spine, EMG recruitment had not yet occurred by 5 weeks
post-injury in roughly 1/2 of all lower limb muscles, and prolonged delays between injury and recruitment onset were sometimes seen. Injury
to the thoracolumbar spine was frequently associated with very long delays (i.e. . 1 year) between injury and resumption of volitional
contraction of distal lower limb muscles.
Discussion: The incidence of neurologically incomplete SCI is rising. In such subjects, delays of 1 or more months between injury and the
onset of voluntary contraction are common for muscles of the distal upper limbs (for cervical injury) and lower limbs. Given the abbreviated
period of in-patient rehabilitation now routine in the United States, these subjects in particular will benefit from frequent follow-up
evaluations to assess spontaneous recovery and design appropriate rehabilitation strategies to maximize functional independence. Moreover,
the potential for delayed recovery must be considered when designing and implementing novel clinical interventions for treating SCI, to
better differentiate between spontaneous and treatment-related improvements in neurologic function.
q 2004 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved.
Keywords: Electromyogram; Spinal cord injury; Human; Recovery; Voluntary contraction

1. Introduction the MMT as a primary measure, often using that form

The ability to voluntarily contract and relax a limb
muscle following spinal cord injury (SCI) provides nearly
unambiguous evidence of the retention (or recovery) of
cortical control over that muscle’s motoneuron pool. In
humans, this property is appreciated through widespread use
of the ‘manual muscle test’ (MMT), in which subjects are
asked to attempt a voluntary contraction of specific muscle
groups, and the strength of contraction is assessed by the
tester (Daniels and Worthingham, 1980). Virtually all major
clinical studies of outcomes following SCI have included
* Corresponding author. Tel.: þ1-315-464-9935; fax: þ 1-315-464-9848.
E-mail address: calancib@upstate.edu (B. Calancie).
1388-2457/$30.00 q 2004 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.clinph.2004.03.002
endorsed by the American Spinal Injury Association (ASIA)
(Marino et al., 2003).
An earlier publication from this laboratory using a subset
of the present study’s subject population showed that a 6
point scale of electromyogram (EMG) activity can be
substituted for MMT measures, and was accurate for
predicting contractile force within multiple muscle groups
at both acute and chronic stages following SCI (Calancie
et al., 2001). In the present study, we report on our findings
of recruitment recovery based on EMG measures from a
large population of subjects with acute SCI in who repeated
follow-up evaluations were made.
Other studies of human SCI emphasize recovery of
overall muscle strength, expressing outcomes as
 B. Calancie et al. / Clinical Neurophysiology 115 (2004) 1748–1759
a composite score for all 4 limbs (Burns and Ditunno, 2001;
Waters et al., 1998). Rather than adopt this approach, we
concentrated on the early stages of recovery as indicated by
the presence and magnitude of EMG activity on a muscle-
by-muscle basis. We paid particular attention to the period
of time following injury when subjects were going to, and
being discharged from, the rehabilitation service, as the
overall length of stay in the rehabilitation setting following
traumatic SCI has declined precipitously in the United
States over the past 2 decades (Kirshblum et al., 2002;
National Spinal Cord Injury Statistical Center, Birmingham
AB, 2002).
Two key goals of the present study, then, were to:
(1) identify common patterns of EMG recovery across
populations of subjects in relation to the severity of injury;
and (2) for multiple upper- and lower-limb muscles,
establish the range of times post-injury during which
muscles initially paralyzed by the SCI could begin to
show motor unit recruitment (i.e. EMG) during voluntary
contraction attempts. In other words, how long after injury
might one expect to regain some degree of voluntary
contraction in various upper- and lower-limb muscles
affected by SCI. Our findings are relevant to the manner
and timing of follow-up assessments and rehabilitation
intervention, as well as to questions of subject selection and
result interpretation when implementing novel interventions
for treating SCI (Burns et al., 2003; Fawcett, 1998).
2. Methods
2.1. Recruitment
We enrolled subjects with acute, traumatic SCI who were
admitted to Miami’s Jackson Memorial Hospital (JMH).
More than 90% of subjects were recruited in a 3 year span
from 1997 to 2000. All subjects gave their informed consent
(for adults) and assent (for children) to participate in this
study, which was approved by the University of Miami’s
Institutional Review Board.
2.2. Instrumentation
EMG was recorded with pairs of surface electrodes
positioned on the skin overlying target muscles. (In fewer
than 5% of these studies, EMG was recorded with pairs of
1/200 EEG-type needle electrodes placed subcutaneously
over the target muscles (Hugon, 1973; Nuwer et al., 1993).
EMG was amplified, filtered (100 – 2.5 kHz) and stored on
magnetic tape (PCM 4000 (A.R. Vetter, Co.; initial studies)
or DT-1600 (Micro-Data Instruments, Co.)) for off-line
analysis. Individual EMG records were displayed visually
on a computer screen (Computerscope; R.C. Electronics)
and audibly, through mixers (Rane RM 26; Rane Corp., or
Roland M120; Roland Corp.) and a loudspeaker (MS101;
Yamaha Corp.).
1749
2.3. Protocol
The primary measure of muscle contraction was the
EMG interference pattern. Early studies included indepen-
dent measures of muscle strength via the MMT. This
traditional form of strength testing was conducted by a
physical therapist or physician experienced in this form of
testing and who did not have knowledge of the subject’s
EMG findings. We discontinued MMT testing after
establishing that there was a strong positive correlation
between MMT and EMG values for multiple upper- and
lower-limb muscles at both acute and chronic stages after
SCI (Calancie et al., 2001).
For persons with cervical injuries, EMG was recorded
from the following muscles: biceps brachii (biceps), triceps
brachii (triceps), wrist extensors (including extensor carpi
radialis (ECR)), wrist flexors (including flexor carpi radialis
(FCR)), thenar group (including abductor pollicis brevis
(APB)), hypothenar group (including abductor digiti minimi
(ADM)), hip flexors (including psoas major (psoas)),
quadriceps (quads), hamstring (hams), tibialis anterior
(TA), soleus, and abductor hallucis (AbH; an intrinsic
muscle of the foot). For persons with injury at or caudal to
the 3rd thoracic vertebrae, EMG for voluntary contractions
was recorded from only the lower-limb muscles described
above.
Subjects were asked to make an isolated contraction of
each muscle or muscle group in a particular direction (or
axis), and were given auditory feedback of that muscle’s
EMG. When necessary, resistance to contraction was
provided by one of the investigators, such that all
contractions were isometric (or nearly so, in the case of
normal contractile strength). The instructions were to
attempt to move a joint in the direction indicated, smoothly
and quickly increasing to maximal effort upon the word
‘contract’, and to maintain that contraction until asked to
‘relax’ (subjects were cautioned against making an abrupt
increase in force (i.e. a ‘ballistic’ contraction)).
In persons with cervical injury, we began with voluntary
contraction attempts of left-side muscles. As this study was
part of a larger, comprehensive evaluation of spinal cord
properties after acute SCI, we followed the EMG recruit-
ment with measures of spinal cord reflex excitability, and
used transcranial magnetic stimulation to examine central
motor conduction in some subjects. Next, recording leads
were connected to comparable electrodes on right-side
muscles, and these measures were repeated.
The entire assessment protocol, including TMS,
required approximately 1.5 h in a person with cervical
SCI. Most subjects were tested acutely while supine or in
side-lying postures, and while sitting or reclining in later
evaluations. Follow-up evaluations were scheduled at 1, 2,
4, 8, 12, 26, and 52 weeks relative to the initial
examination. Many subjects were receiving some form
of injury-related medications at the time of initial
1750 B. Calancie et al. / Clinical Neurophysiology 115 (2004) 1748–1759

evaluation (e.g. acetaminophen; narcotics) or at follow-up two-thirds of the population was neurologically ‘incom-
(e.g. baclofen, gabapentin). plete’ at the time of initial evaluation, based on ASIA
criteria (i.e. ASIA ‘B’ through ‘E’).
2.4. Analysis Fractures of the cervical spine were the most common
cause of SCI, occurring in 62.5% of the total subject
Analysis was done off-line, from taped records. EMG population (Fig. 1A). Forty-seven subjects (20.5%) sus-
scores were assigned on a 6 point scale, comparable to that tained a fracture of at least one thoracic vertebrae, and
used for manual muscle testing. Scoring criteria were lumbar fractures accounted for 13.5% ðn ¼ 31Þ of the
established to approximate a MMT score, as follows: 0, no sample population. In the cervical spine, fractures of C2 and
volitional recruitment or rate-modulation of spontaneously C5 were common, as were fractures of the T12 and L1
active motor units; 1, recruitment of 1 or 2 motor units, vertebrae in the thoracolumbar region. Eight (3.5%) of the
unsustained discharge; 2, recruitment of 1 – 3 motor units, 229 subjects in our study were admitted to the hospital with
sustained discharge, slow onset and/or offset; 3, recruitment neurologic evidence of SCI, yet no radiologic evidence of
of . 3 motor units, but not enough to become indistinct
audibly, sustained discharge, rapid onset and offset; 4,
recruitment of multiple units, indistinguishable visually and
audibly, rapid onset and offset, moderate amplitude at peak;
and 5, recruitment of multiple units, indistinguishable
visually and audibly, rapid onset and offset, high amplitude
(. 1 mV) at peak (Calancie et al., 2001). The same author
(JGB) assigned EMG scores for muscle contractions. He
was not present during data collection, nor did he have
specific knowledge of each patient’s clinical status. Mean
EMG scores were calculated for each limb. We also
expressed EMG scores individually, using a ‘pie-chart’
format to allow rapid visual assessment of a large number of
individual data values across time (see Fig. 3).
Contraction in muscles distant to EMG electrodes may
be recorded when the muscle underlying the electrodes is
electrically silent. Such ‘cross-talk’ (De Luca and Merletti,
1988; Farina et al., 2002; van Vugt and van Dijk, 2001) can
introduce error into the EMG scoring, for which a low EMG
score might be assigned to a muscle that is not, in fact, being
recruited. In the recording configuration used in the current
study, the occurrence of cross-talk appeared to be highest
from triceps brachii (during strong contractions of biceps
brachii; and see (Calancie et al., 2001)) and the psoas
muscle group (during contraction of quadriceps). Such
cross-talk was typically seen as a dense interference pattern
(i.e. motor units indistinct, rapid firing rates) of low
amplitude (, 50 mV peak to peak), evident for contractions
of both the target muscle and of the muscle from which
cross-talk was likely originating. A score of ‘0’ was entered
Fig. 1. Breakdown of the study population incidence, based on cranial-most
when cross-talk was suspected. This was most common in level of spine fracture (A), time (in days) between injury and the first
the ‘motor-complete’ group for triceps brachii contractions evaluation (B), and the number of evaluations performed on each subject
(, 5% of trials). In cases in which cross-talk was suspected (C). Fracture data shown in (A) exclude sites of more caudally located
but one or more clearly differentiated single motor unit fractures in those subjects with multiple fractures. Cases of spinal cord
injury without radiologic abnormality (SCIWORA) are represented to the
potentials were superimposed on this activity, a score of 1 or
far right of the figure. In (B), subjects first studied within 24 h of injury lie
2 was assigned. within the first bin, subjects studied between 24 and 48 h post-injury are in
the second bin, etc. The most common interval was 3 days from date of
injury to date of initial evaluation. Mean and SD for the total population
3. Results (‘all’) is shown. Dashed line represents an arbitrary cut-off excluding 15
subjects whose recruitment into the study was especially delayed.
Excluding these subjects, a revised mean (,19 days) is also shown. In
A total of 229 subjects were included in this study. The (C), the largest group by far had only one evaluation performed. In contrast,
average subject age was 40.0 ^ 17.5 years (mean ^ SD). as many as 12 evaluations were carried out on 2 subjects. A total of 802
Males constituted 76.4% of the population. Approximately evaluations were carried out over the course of the study.
 B. Calancie et al. / Clinical Neurophysiology 115 (2004) 1748–1759 1751

fracture could be found (i.e. SCI without radiologic
abnormality—SCIWORA).
Fig. 1B summarizes the time at which subjects were first
evaluated following their injury. This value across all
subjects was 6.8 ^ 8.7 days (mean ^ SD). This includes
data from 12 subjects who were transferred to JMH
after initially being hospitalized elsewhere, and another 3
subjects whose life-threatening injuries prevented us from
recruiting them at an earlier time-point. The average period
from injury to date of first evaluation for these 15 subjects
(to the right of the dashed vertical line in Fig. 1B) was
32.9 ^ 15.5 days. Excluding these 15 subjects, the average
delay from time of injury to initial evaluation was 5.0 ^ 3.8
days for the remaining 214 subjects (i.e. . 93% of the total).
Twenty-six subjects were first tested within 24 h of their
injury.
We completed a total of 802 evaluations of spinal cord
neurophysiology in the 229 subjects enrolled in this study.
Fig. 1C illustrates the distribution of the total numbers of
evaluations per subject across the population. The sampling
frequency was dominated by subjects ðn ¼ 77Þ in whom we
completed only a single evaluation following acute injury.
In the majority of these cases, subjects were discharged or
transferred shortly after our initial evaluation, and were lost
to follow-up evaluations. As this manuscript concentrates
on the pattern of recovery, those subjects tested only a single
time are not considered further in this report. Thus we
averaged just under 5 evaluations per subject on those from
whom we had multiple evaluations.
Fig. 2 illustrates EMG records following two unin-
terrupted series of contractions in left- (A) and right-side
muscles (B) of a 17 year old girl who sustained a C4 fracture
after falling from a bicycle approximately 7 weeks prior to
this recording session. The short horizontal lines above and
below the upper- and lower-most data traces in each panel
indicate the time when the subject was asked to contract
each of the biceps (‘1’) through AbH (‘12’) muscles. This
subject was able to contract and relax each of her left-side
muscles without hesitation (Fig. 2A). Co-contraction of
upper-limb muscles was modest. Contractions of leg
muscles were accomplished with little or no upper-limb
muscle activity. The only left-side lower extremity muscles
showing considerable co-contraction were the TA and AbH
muscles, active along with soleus during ankle plantar-
flexion contractions.
Contraction of her right-side muscles (Fig. 2B) was much
more difficult for this subject. She had no volitional APB,
TA, or soleus recruitment during this evaluation, and only
trace recruitment in triceps brachii, ADM, and hamstring
groups when attempting to contract each of these in
isolation. However, several of these muscles were clearly
recruited when she attempted contractions of other muscles.
Note also that there were multiple upper-limb muscles
recruited during right-side lower-limb contraction attempts.
Co-contraction was in fact very common amongst persons
with SCI, many of whom would contract all muscles
receiving voluntary innervation in an effort to recruit those
muscles left paretic or paralyzed by the injury.
Fig. 3A summarizes the average EMG scores by side (left
and right) and limb (upper and lower) for all of the recording
sessions carried out in the subject whose data are shown in
Fig. 2. There was a clear asymmetry between left- and right-
side strength (left-side stronger), evident from the initial
evaluation onwards. EMG scores from this subject’s left
side showed substantial and steady improvement from the
time of injury through a 6 week span, while improvement in
right-side muscles was delayed in onset and more gradual.
Three of the 4 limb averages reached their highest value at
the final evaluation conducted (6 months post-injury).
Right-side limb scores in particular would likely have

Fig. 2. EMG recruitment in left-side (A) and right-side (B) muscles, during attempted voluntary contraction of each muscle in isolation. Each panel represents a
continuous record of 12 channel activity recorded simultaneously. The horizontal lines at the top and bottom of each panel (with accompanying number at the
top) indicate the time during which the subject was asked to contract the muscle indicated (position of each line estimated from the voice record on tape giving
the cue to ‘contract’ and ‘relax’). Contractions always started with biceps brachii (‘1’) and continued in a proximal-to-distal manner. There were several trials
of right-side (Fig. 2B) muscle contractions which were repeated by the subject (e.g. ECR (‘3’), hams (‘9’), TA (‘10’), soleus (‘11’), and AbH (‘12’)).
Neurologic status at admission: ASIA ‘C’.
1752 B. Calancie et al. / Clinical Neurophysiology 115 (2004) 1748–1759

Fig. 3. Summary of average EMG scores for each limb at different times post-injury (A) and individual EMG scores for each muscle separated by left-
and right-sides (B). This serves as an example of a ‘slow, subtotal’ recovery pattern. In (B), each circle is divided into 5 ‘slices’. An EMG score of ‘0’
has no fill, an EMG score of ‘5’ leads to a completely filled pie, and scores of 1 through 4 are indicated by progressively more ‘slices’ being filled. Note
that this subject was much weaker in her right-side muscles than in muscles on her left. In fact she had no detectable EMG in any right-side lower-limb
muscle during the first 3 evaluations (EMG average scores of ‘0’ and represented by ‘diamonds’ in (A), and open ‘pies’ for all right-side lower limb
muscles in the first 3 evaluations in (B)).

increased further at later times following injury, but longer
term evaluations were not possible with this subject.
Fig. 3B illustrates individual muscle scores for each of
the 6 recording sessions summarized in Fig. 3A. Depiction
of data in this manner reveals a number of aspects of
recovery following SCI that are not apparent when
comparing overall muscle scores, or even when comparing
mean limb scores. It was not unusual to see an individual
muscle group with minimal recruitment early after an injury
then fall silent upon the next follow-up evaluation, as
occurred in the right-side FCR findings between week 1
(score of ‘1’) and 2 (score of ‘0’). A substantial increase in
proximal (psoas and quads) right lower-limb muscle
strength was largely responsible for the big improvement
in average scores for this limb between the 3rd (week 4) and
4th (week 7.5) recording sessions (as seen in Fig. 3A),
whereas right upper-limb EMG was almost unchanged
between these sessions. While not contributing much to the
overall limb averages, 3 different muscle groups (right-side
APB, TA, and soleus) showed new recruitment at the week
14 session compared to evaluations at and prior to the week
7.5 examination (see Fig. 2B). Finally, the modest decline in
average EMG scores of the left lower limb from week 7.5 to
week 26 (squares in Fig. 3A) reflects a drop of ‘1’ in scores
across each of 4 different muscles (psoas, quads, hamstring,
and soleus). It is not clear in this individual’s case whether
this modest decline across multiple muscles reflected a
genuine loss of function (unlikely, given that distal
musculature on the right side was continuing to improve
in EMG score), or whether she had simply not achieved her
maximal contraction effort prior to relaxation.
The pattern depicted in Fig. 3 is consistent with a
relatively slow, subtotal improvement in function spanning
months. Fig. 4 (up-pointed triangles) graphically depicts this
as one of 4 recovery patterns into which we were able to
group the majority of subjects in whom repeated measures
were carried out. Sixty-two percent of subjects with cervical
and thoracic incomplete SCI fit this ‘slow, subtotal’ recovery
pattern. Other subjects showed rapid and substantial motor
recovery within days of their injury, leading to essentially
normal motor function (down-pointed triangles in Fig. 4).
Fig. 4. Hypothetical single EMG score averaged for all muscles examined,
reflecting a synthesis of 4 different recovery patterns encountered in
subjects from this study. Subjects with a ‘rapid, near-total’ pattern showed
the best clinical improvement, in which function had recovered to
essentially normal within a few weeks of injury. This profile accounted
for 25% of the cervical, motor-incomplete cohort. Persons with a ‘slow,
subtotal’ recovery pattern experienced a more prolonged period of clinical
improvement, continuing over a period of months following injury. This
was typically due to a combination of recruitment of previously silent
muscles, and greater levels of motor unit recruitment/rate-coding in
muscles that had already shown recovery. More severe SCI led to a ‘slow,
minimal’ recovery pattern, which is typical for subjects with neurologi-
cally-complete (i.e. ASIA ‘A’) SCI. It is characterized by gradual
strengthening of muscles innervated from levels at or rostral to the SCI
epicenter, and no recruitment in muscles innervated from more caudal
levels of the spinal cord. Finally, some subjects with fracture to the caudal
thoracic or lumbar spine experienced a ‘slow, root-sparing’ recovery
profile. These subjects typically showed preservation of recruitment in
proximal lower-limb musculature evident almost immediately following
injury, and gradual recovery of recruitment capability in more distal
muscles.
 B. Calancie et al. / Clinical Neurophysiology 115 (2004) 1748–1759
This pattern described the recovery profile in 25% of the
cervical incomplete population, and 8% of subjects with
thoracolumbar injury.
A third pattern of EMG recovery encountered in this
study is referred to as a ‘slow, minimal’ recovery pattern
(diamonds in Fig. 4). This is seen in subjects with
neurologically complete SCI to the cervical or thoracic
levels. Five subjects with cervical incomplete SCI recovered
only minimal upper- and lower-limb function, yet were not
included in this ‘complete’ recovery pattern because of clear
evidence for voluntary recruitment of 1 lower-limb muscle
(AbH in all 5 cases). By definition, all subjects with cervical
motor-complete SCI were included in this pattern, in whom
recovery was restricted to proximal upper-limb musculature
(e.g. biceps brachii, wrist extensors). Similarly, persons
with thoracic, motor-complete SCI failed to show any motor
recovery in lower limbs.
Finally, some subjects with thoracolumbar injury showed
what we refer to as a ‘slow, root-sparing’ recovery pattern,
in which modest improvement in lower-limb EMG scores
were seen spanning a very prolonged time-frame (Fig. 4,
circles). Recovery of EMG recruitment, when seen at all,
was in a proximal-to-distal manner, with continued
improvement in EMG scores within proximal muscles at
the same time that more distal muscles began to show
recruitment (i.e. individual muscle scores . 0).
Not all subjects improved in neurologic function from
the most acute stage and beyond. A small number of our
subjects showed distinct deterioration of EMG recruitment
scores in the weeks following their initial evaluations (i.e.
‘acute deterioration’). In each case these subjects showed a
widespread deterioration of clinical function, and ultimately
died of injuries secondary to—or sustained at the same time
as—their SCI.
Finally, longer-term complications and neurologic
deterioration have been reported in persons with chronic
SCI. In our subject population, modest levels of deterio-
ration in mean EMG scores (a decline in mean EMG score
of ‘1’ or more for any limb studied) at time points in excess
of 1 year post-injury were seen in 21% of subjects with
neurologically incomplete injury to the cervical or thoracic
spine. Such deterioration was seen in 1 of the 7 subjects with
thoracolumbar injury followed beyond 1 year post-injury.
We did not see any cases of ‘delayed deterioration’ in
upper-limb muscles of persons with cervical, motor-
complete SCI.
We now shift attention to the time after injury at which
EMG recruitment first becomes evident in a particular
muscle, without consideration of the magnitude of the
recruitment. In these cases, a given muscle score had to be
‘0’ at the initial evaluation, and greater than ‘0’ at some
subsequent evaluation. We were especially interested in
learning how long after injury one might look for, and not
see, signs of EMG recruitment during a voluntary contrac-
tion attempt in a muscle or muscle group that ultimately did
show signs of recovery.
1753
Fig. 5 illustrates data from upper-limb muscles of
subjects with neurologically incomplete injury to the
cervical spine. Circles represent the latest time (in weeks)
after injury when the muscle indicated was tested and found
to be silent during an attempted voluntary contraction (i.e.
EMG score of ‘0’), while the overlying triangle represents
the first time after injury that recruitment was seen in that
muscle during an attempted contraction (i.e. EMG
score . 0). Using the top-most panel (‘biceps’) in Fig. 5
as an example, the right-most data set shows that the biceps
muscle in this particular subject was not recruited during
repeated measures up to and including a trial at approxi-
mately 28 weeks post-injury, but then showed recruitment at
approximately 46 weeks post-injury (the next time after the
28 week evaluation that this particular subject was tested).
Also shown in the heading of Fig. 5 panels is the mean
number of weeks for which EMG recruitment was absent
(circles) for each muscle, along with the median time of
EMG absence for that muscle population. Again using the
biceps brachii as an example, the mean time at which EMG
was looked for and found to be absent in this population was
4.7 weeks. However, this value was disproportionately
affected by the prolonged time without recovery of biceps
function (silent at , 28 weeks) of the right-most data set.
Overall, more than one-half of the subjects had recovered
biceps recruitment at 2– 3 weeks post-injury, with a median
time of 1.4 weeks post-injury at which EMG was looked for
and found to be absent.
The mean and median times of EMG absence for distal
muscles of the upper limb (APB and ADM) were
considerably higher than those for more proximal muscles
of the arm in motor-incomplete subjects with cervical injury
(Fig. 5). Approximately 1/3 of all hand muscles did not
show recruitment by 5 weeks post-injury, a time chosen to
reflect a stage during which many of these subjects were
now on the rehabilitation service, or had been discharged
from the hospital altogether. Finally, there were a few cases
for hand muscles (n ¼ 5 and 8 for APB and ADM,
respectively) in which recruitment recovery was delayed
for even longer periods, extending beyond 20 weeks post-
injury. This prolonged delay between injury and recruitment
recovery was seen only once for the biceps, and was not
seen in any of the triceps, ECR or FCR muscles for this
population.
Fig. 6 illustrates the distribution of recruitment delay
within lower-limb muscles in subjects with cervical and
motor-incomplete SCI. Both the mean and median times of
the last examination date without recruitment were
considerably longer in lower-limb muscles than those
values from upper-limb muscles in this subject population.
Roughly 1/2 of all cases across the 6 lower-limb muscles in
this cohort had not shown recruitment by 4– 5 weeks post-
injury. The incidence of especially late recruitment (. 20
weeks for the last examination without recruitment) varied
between a low of 10% (for AbH) to a high of 27% (for
hamstring). Note that the filled circles in the ‘Hams’ and
1754 B. Calancie et al. / Clinical Neurophysiology 115 (2004) 1748–1759
Fig. 5. Data from subjects with motor-incomplete cervical SCI. Each panel
corresponds to a different upper-limb muscle, showing the greatest number
of weeks post-injury at which we confirmed an EMG score of ‘0’ (circles)
for that muscle, and the time of the next evaluation when a non-zero EMG
score was seen (triangles). There were 2 cases (1 each for APB and ADM
muscles) in which the time of non-zero confirmation was not illustrated, in
order to avoid excess ‘compression’ of the ordinate. Recruitment was seen
at 101 and 83 weeks for these two cases, respectively. Since most subjects
had non-zero EMG scores in biceps at the very first evaluation, the sample
size for this group was much smaller than that for other upper-limb muscles.
‘AbH’ data records of Fig. 6 represent data from a single
subject who failed to show recruitment in these two muscles
at 127 weeks post-injury, but they were recruited during his
next evaluation (264 weeks post-injury).
Given the more serious nature of their spinal cord
injuries, there were far fewer cases of delayed muscle
recovery in persons with cervical and motor-complete SCI.
By definition, these subjects did not recover recruitment in
any lower-limb muscle. Within upper-limb muscles, the few
cases encountered of distal recovery usually became evident
within the first 5 weeks post-injury. There were only two
cases of prolonged recovery within this cohort, both
occurring in the triceps muscle (absent recruitment at 28
and 55 weeks post-injury, with recovery at some later time
still).
A small number of subjects with thoracic-level SCI went
on to show recovery of recruitment within one or more
lower-limb muscles. The delays in recruitment recovery in
this population were similar to those for lower-limb muscles
in persons with cervical, motor-incomplete injury. Recovery
had not yet become evident in approximately 1/2 of all
lower-limb muscles by 5 weeks post-injury in this cohort.
Across all categories of injury, there were a few cases of
what we term ‘very prolonged’ recruitment delays (i.e.
confirmed absence of recruitment at a time-point exceeding
1 year, with known recruitment at a still later point in time).
Based on absolute numbers, the incidence of such very
prolonged recruitment was identical for the ‘cervical
incomplete’ and ‘thoracolumbar’ groups. However, as a
percentage of the total number of cases in each group, very
prolonged recruitment delays made up only 3.5% of the
‘cervical incomplete’ cases, 4% of the ‘cervical complete’
cases, and a much higher incidence (57%) of the
‘thoracolumbar’ cases. Based on Chi-square analysis, the
difference between the ‘cervical incomplete’ and ‘thoraco-
lumbar’ groups in the incidence of cases very prolonged
recruitment delay was significant ðP , 0:001Þ:
4. Discussion
In this study, we found the incidence of neurologically-
incomplete SCI to be considerably higher than has been
reported in other large-scale studies of SCI (Bracken et al.,
1984, 1990, 1997). This trend is in keeping with recent
reports (National Spinal Cord Injury Statistical Center,
Birmingham AB, 2002; Sekhon and Fehlings, 2001), and
Data were arranged in order of increasing time from injury to date of
confirmed absent EMG recruitment. Following each muscle name is the
mean time (weeks ^ SD) of confirmed recruitment absence for each
muscle, followed by the median for this muscle’s sample. Large differences
between these two values, most evident in the APB and ADM records,
reflect the contribution to the mean value of especially long intervals
between injury and EMG absence/recruitment (to the far right of these
panels).
 B. Calancie et al. / Clinical Neurophysiology 115 (2004) 1748–1759
Fig. 6. As in Fig. 5, but now showing recruitment delays for lower-limb
muscles in subjects with motor-incomplete, cervical SCI. Filled circles in the
‘Hams’ and ‘AbH’ panels represent cases of very prolonged delay between
injury and recovery of recruitment capability. These values are 131 and 127
weeks, respectively. Recruitment was evident in both of these muscles at the
time of the next evaluation (189 and 264 weeks post-injury, respectively).
may reflect increased usage of seatbelts as well as improved
head/neck immobilization and patient transport techniques
following injury (Sekhon and Fehlings, 2001). We also
found that there was usually a marked asymmetry between
1755
left- and right-side muscles in the initial deficit of persons
with incomplete injury, as well as in the pace and extent of
their neurologic recovery (Waters et al., 1993). Such
asymmetry was also seen in persons with complete SCI at
the acute stage, but these differences between sides tended
to diminish over time in this latter population.
There was a high incidence of delayed recovery of EMG
function in motor-incomplete subjects. In persons with
cervical injury, approximately one-third of all distal hand
muscles did not show recruitment at 5 weeks post-injury,
but recovered recruitment some time afterwards. In leg
muscles of these subjects, and for persons with motor-
incomplete injury to the thoracic spine, the proportion of
delayed (. 5 week) recruitment recovery was closer to one-
half of all the lower-limb muscles sampled.
This time-frame of 5 or more weeks is beyond the most
acute stage during which drug studies to improve outcome
have been initiated (Bracken et al., 1990; Geisler et al.,
2001), all of which used the MMT as an outcome measure.
Therefore assessments of muscle strength in this population
at the acute stage would fail to detect force in a relatively
large number of muscle groups that will eventually show
recovery. Assuming that persons with incomplete injury are
candidates for novel intervention studies now being
considered for treating SCI (Fawcett, 1998; Jones et al.,
2001; Schwartz, 2003), this high probability for delayed,
spontaneous improvement in neurologic function must be
considered when interpreting findings emerging from new
intervention studies.
In addition to this concern (which is largely theoretical at
the present time), a practical consequence of our findings
relates to the timing of hospital discharge and a patient’s
access to rehabilitation. A significant number of subjects in
our cervical, motor-incomplete group had already been
discharged from the hospital before this 5 week period had
been reached: the median length of stay for this cohort was
only 16 days, while the average length of stay for this group
was 38.4 days. (This latter value is somewhat less than a
mean length of stay of 59 days for acutely injured patients
treated within a Model System center, a value derived from
all treated patients (complete and incomplete) and dating
back to 1998 (National Spinal Cord Injury Statistical
Center, Birmingham AB, 2002). Thus in the present study,
many subjects had already been discharged at a time when
some of their lower limb muscles were still completely
paralyzed. Access to a rehabilitation facility for timely
evaluations and implementation of appropriate therapy as
new muscle groups regained voluntary innervation would
therefore be especially crucial to this population, who might
have the most to gain in terms of functional independence
from structured out-patient therapy (Kirshblum et al., 2002).
We attempted to recruit subjects at an acute, but not
‘hyper-acute’ stage following injury (as would be the case
with assessments in the emergency room setting, for
example). Hyper-acute assessments are necessary for
examining the role of neuroprotective agents – such as
1756 B. Calancie et al. / Clinical Neurophysiology 115 (2004) 1748–1759
methylprednisolone –but it can be difficult to conduct a valid
neurologic exam at such an early time after injury, due to
pain, mild brain trauma, sedative use (or alcohol effects),
and psychological unrest. In fact a delay of several days
from SCI to research assessment is associated with an
improved accuracy of prognosis (Brown et al., 1991).
4.1. EMG as an outcome measure
In an earlier study of subjects with SCI (Calancie et al.,
2001), we showed that there was a strong, positive
relationship between the MMT score and scores from the
EMG system used in the present study. This has allowed us
to substitute EMG for MMT testing in this study. Multiple
channel EMG recording and analysis requires relatively
expensive instrumentation, and adds time (electrode place-
ment) to the evaluation compared to a typical MMT
protocol. However, EMG monitoring does confer certain
advantages over traditional MMT testing for assessing
motor recovery after SCI.
First, EMG recording is, without question, more sensitive
than the MMT, since only one single motor unit need be
recruited to score a ‘1’ with EMG testing. Such limited
recruitment would be unlikely to cause palpable contraction
in a subject with acute or subacute injury (in whom
peripheral sprouting had not yet occurred (Thomas et al.,
1997)), particularly if the motor unit’s rate of discharge was
relatively low. Paralyzed muscles tend to develop faster
twitch contraction times—hence require higher stimulus
rates to cause fusion—than those under supra-spinal control
(Cameron and Calancie, 1995; Stein et al., 1992; Thomas
et al., 1997; Yang et al., 1990).
Another advantage of EMG recording is that one can
examine multiple muscle groups simultaneously for indi-
cations of co-contraction, excessive amounts of which are
associated with central motor disorders (Boorman et al.,
1996). In a related vein, the use of manual muscle testing in
conditions involving altered (i.e. increased) muscle ‘tone’
has been discouraged, due to the difficulty in attributing
contractile force to one muscle alone when it more likely
reflects the net effect of multiple muscles acting simul-
taneously across a joint (Daniels and Worthingham, 1980).
Monitoring multiple EMG channels can help confirm
that a subject is genuinely attempting to contract a paretic or
paralyzed muscle, since such attempts typically lead to
motor unit recruitment in muscle(s) innervated at a level
rostral to the cord injury (Sherwood et al., 1996). An
example can be seen in Fig. 2B, in which the right biceps
was recruited during attempted contractions in most of the
other right-sided muscles of this subject.
Despite all these advantages, the time and expense of
EMG monitoring for assessing persons with acute SCI
likely renders it unsuitable for routine clinical use in most
centers. However, novel interventions which may call for
expanded outcome measures for assessing the intervention’s
efficacy would likely benefit from inclusion of multi-
channel EMG recording, particularly since this approach
allows for quantitative electrophysiologic studies of spinal
cord properties (e.g. reflexes and central motor conduction)
not accessible to traditional clinical examination.
4.2. Recovery patterns
The ‘slow, limited’ EMG recovery pattern illustrated in
Fig. 4 is characteristic of a severe, neurologically complete
SCI at the cervical level (i.e. ASIA ‘A’ category). There is
typically some degree of improved function in upper-limb
muscles initially paretic or paralyzed but receiving inner-
vation from levels at or rostral to the injury epicenter, but
this improvement does not extend to muscle groups that are
more caudally innervated. Muscle contraction capabilities
tend towards greater symmetry between left- and right-sides
in such cases than is usually the case for motor-incomplete
injuries (Waters et al., 1993). It is this type of injury that is
traditionally associated with SCI, and a large proportion of
animal models continue to use this severe-injury profile (e.g.
transection, or high-momentum weight-drop) for studies of
SCI interventions (Bunge, 1994; Geller and Fawcett, 2002;
Schwab and Bartholdi, 1996).
The majority of subjects in this study showed higher
degrees of EMG recovery than that depicted by the ‘slow,
limited’ pattern in Fig. 4, based on both the numbers of
muscles showing recruitment, and the magnitude of
recruitment. In some cases substantial recovery was seen
very quickly (i.e. within days) of the injury (e.g. Fig. 4,
‘rapid, near-total’), reflecting a ‘neuropraxia’-like state of
axonal dysfunction and recovery. More prolonged recovery
(weeks to months) was also common. Possible mechanisms
underlying this later recovery include nerve regrowth
(centrally and/or at the neuromuscular level (Hill et al.,
2001)), axonal remyelination (Honmou et al., 1996), or a
redistribution of excitable channels (Waxman et al., 1994).
There were 5 subjects in the present study who had clear
recruitment in multiple muscles during our initial evalu-
ation, and in whom we noted a substantial decline in motor
function upon follow-up evaluation (i.e. acute deterio-
ration). In one case the patient’s deterioration followed
surgical spinal cord decompression and stabilization of her
spine fracture. The other 4 cases of deterioration were
associated with multiple co-morbidities secondary to motor
vehicle accident. The decline in contraction capabilities in
each of these subjects appeared to reflect deteriorating
central motor function. Significant cognitive dysfunction
was ruled out in these cases, as each of these subjects
indicated their ongoing willingness to participate in the
study protocol at subsequent evaluations. Outside of these 5
cases, there was no evidence for deterioration of neurologic
function to values below those encountered upon the initial
evaluation, in contrast to a previous report of transient
deterioration after acute SCI (Brown et al., 1991). Other
reports of deterioration after acute SCI are thought to be
related to thrombosis of a key vessel or to inflammatory
 B. Calancie et al. / Clinical Neurophysiology 115 (2004) 1748–1759
and/or immune mechanisms (Belanger et al., 2000),
possibly due to episodes of hypotension (Aito et al.,
1999). Whatever the causes, it is evident from our own
findings and the limited numbers of published reports that
deterioration of neurologic function after acute SCI in
humans is rare (Aito et al., 1999). Put another way, subjects
are typically in their worst neurologic status at the time they
first reach the hospital following SCI, and thereafter
experience varying degrees of neurologic improvement.
Cases of ‘delayed deterioration’ were relatively common
in subjects with cervical and thoracic incomplete SCI. Several
studies have reported on the incidence of post-traumatic
myelopathy, frequently associated with cystic cavitation and
tethering between spinal cord and dura mater (Asano et al.,
1996; Falci et al., 1999; Lee et al., 2001; Smith and Rekate,
1994). Such a diagnosis of spinal cord tethering was made in
one of our subjects, who experienced marked loss of
neurologic function approximately 1.5 years post-injury. He
was treated surgically (cervical laminoplasty, dural untether-
ing and duraplasty) and had good resolution of symptoms. For
our other cases of ‘delayed deterioration’, lowered EMG
values may have simply been due to a submaximal
contraction effort, but one cannot rule out the possibility of
spinal cord tethering (Bursell et al., 1999; Little et al., 1992),
reflecting the early stages of this known complication
following spine trauma. Alternatively, loss of spinal moto-
neurons secondary to trans-synaptic degeneration has been
reported following central lesions (Aisen et al., 1992; Chang,
1998; Eidelberg et al., 1989), and may have contributed to
diminished muscle EMG noted in these subjects.
4.3. Time-course of recovery onset
This manuscript concentrates more on when neural
signals are arriving at the muscle, and less on the absolute
amount of functional recovery one can expect to get after
acute SCI. Thus while we are comfortable presenting and
comparing findings about the onset of muscle contraction
after injury, our study design does not address the maximal
degree of recovery one can expect, since our more limited
data and that of others indicates this process of strengthen-
ing can continue for 2 or more years following SCI (Ditunno
and Formal, 1994; Piepmeier and Jenkins, 1988).
In Figs. 5 and 6, we quantified the latest time after injury
at which EMG was looked for and not seen, rather than the
first time after injury during which recruitment was seen.
This decision reflects the wide disparity of timing for
follow-up evaluations across different subjects. Had we
been able to carry out follow-up evaluations across all
subjects with an invariant schedule between time of injury
and time of evaluation, we might have presented these data
differently. In each record of these figures, a given muscle
fell under the volitional control of the subject between the
time indicated by the circle (EMG not present with
contraction attempt) and the triangle (EMG now present
upon contraction attempt). A brief interval between these
1757
symbols (e.g. a week or two) gave a very accurate indication
of when that recovery took place relative to the time of
injury. If, conversely, the interval between these two pivotal
evaluations was very prolonged (e.g. months to 1 year or
more), then considerable error could have been introduced
by equating the date of EMG recovery to the date of
evaluation. In contrast, no such error can be associated with
quantifying the latest date (relative to the date of injury)
upon which EMG was looked for, and not seen. As such,
presenting the mean and median for these latter values was
felt to be more conservative, thereby avoiding overestimates
of the anticipated time of recovered motor function on a
muscle-by-muscle basis after traumatic SCI.
As previously discussed, absence of recruitment at
5 weeks in muscles which ultimately recovered some
function was very common in persons with cervical (or
thoracic) motor-incomplete injury, but less common at 20
weeks. There were very few cases in these subjects where
we could unequivocally show an absence of recruitment at
one year post-injury in a muscle group that ultimately went
on to show recruitment. It is important to note, however, that
our outcome measure does not directly measure force. A
muscle group atrophied from disuse (likely in a person with
chronic SCI) may therefore be limited in force production
capability and/or fatigue resistance, despite the presence of
a measurable EMG signal during contraction efforts.
Training of these muscles via circuit weight or related
protocols (Duran et al., 2001; Jacobs et al., 2001) might then
lead to improvement in force production and fatigue
resistance, translating into higher functional muscle scores
at these delayed times following injury.
Cases of thoracic injury leading to a ‘motor-incomplete’
status were infrequent, and the time-course of recovery in
these few cases was roughly comparable to that of lower-
limb recovery in persons with cervical and motor-incom-
plete SCI. In contrast, injury to the thoracolumbar spine was
associated with a relatively high probability for very
prolonged periods of recovery. In several cases for which
data were unambiguous, recovery of muscle recruitment
first became evident well over a year after injury. Given the
high probability of nerve root (i.e. lumbosacral plexus)
involvement for injury to the thoracolumbar spine, one
cannot rule out long-distance regeneration of motor axons as
a possible contributor to this very delayed recovery (Little
and DeLisa, 1986). In fact there is evidence—albeit
limited—for such late recovery following surgical repair
or transplantation of damaged ventral roots in humans
(LeBlanc et al., 1969; Petrov, 1971).
There were 13 subjects in our study who had no lower-
limb EMG recruitment upon the initial evaluation, but went
on to recover contraction in one or more muscles of both their
lower limbs at a time period exceeding 2 weeks post-injury
(i.e. all were motor-complete for at least 2 weeks following
injury). Six of these subjects went on to ambulate
independently. Within this subgroup, all but one subject
had recovered recruitment in at least one muscle of both
1758 B. Calancie et al. / Clinical Neurophysiology 115 (2004) 1748–1759
lower limbs by 11 weeks post-injury. The one exception
(who recovered ambulation) showed bilateral lower limb
muscle recruitment sometime between evaluations at 10
weeks (no recruitment) and 21 weeks (positive recruitment)
post-injury.
For subjects in this cohort who did not recover the ability
to walk, recovery delay typically exceeded 10 weeks for all
lower limb muscles in both extremities. In other words, a
complete absence of lower-limb EMG at the acute stage
after injury did not preclude walking recovery, but if this
absence extended beyond 10 weeks post-injury the prospect
for the spontaneous recovery of independent walking was
very low. These findings are reasonably consistent with
those of Waters and colleagues who reported on individual
recovery after SCI (Waters et al., 1993), while extending the
period of time within which functional recovery can be seen
by approximately 4 weeks relative to the findings of Waters
and colleagues (2224).
4.4. Corticospinal influence over upper- versus
lower-limb recovery
It is highly likely that the density of corticospinal
innervation to distal upper-limb muscles exceeds that to
most, if not all lower-limb muscles in man (Jankowska et al.,
1975; Maertens de Noordhout et al., 1999). Nathan’s studies
provide strong evidence that recovery of voluntary function
after SCI reflects sparing of fibers in the corticospinal tract
exclusively (Nathan, 1994; Nathan et al., 1990). If recovery
of motoneuron recruitment after cervical or thoracic SCI
reflects the ability of an intact, but previously-dysfunctional
supra-spinal motor axon(s) to now conduct action poten-
tials, the longer recovery time for lower-limb muscles may
simply be an issue of probability. That is, with a lower
innervation density, the probability that the ‘recovered’
axon would innervate a lower-limb motoneuron pool would
be expected to be lower than that for innervation of a
motoneuron pool to a distal upper-limb muscle. The same
argument should hold true for recovery of distal lower-limb
muscles compared to more proximal muscles of the lower-
limb. That is, AbH should recover voluntary function before
more proximal lower-limb muscles. As reported elsewhere
in preliminary form, we found that this was in fact the case
in the majority of our motor-incomplete subjects (Calancie
et al., 2000).
Acknowledgements
We thank Marcello DiStasio for writing the Excel macro
to display EMG scores. This work was supported in part by
grants from the National Institutes of Health (NS28059,
HD31240, NS36542) and by The Miami Project to Cure
Paralysis.
References
Aisen ML, Brown W, Rubin M. Electrophysiologic changes in lumbar
spinal cord after cervical cord injury. Neurology 1992;42:623–6.
Aito S, El Masry WS, Gerner HJ, Di Lorenzo N, Pellican G, D’Andrea M,
Fromm B, Freund M. Ascending myelopathy in the early stage of spinal
cord injury. Spinal Cord 1999;37:617–23.
Asano M, Fujiwara K, Yonenobu K, Hiroshima K. Post-traumatic
syringomyelia. Spine 1996;21:1446–53.
Belanger E, Picard C, Lacerte D, Lavallee P, Levi ADO. Subacute
posttraumatic ascending myelopathy after spinal cord injury. Report of
three cases. J Neurosurg 2000;93(Spine 2):294–9.
Boorman GI, Lee RG, Becker WJ, Windhorst UR. Impaired natural
reciprocal inhibition in patients with spasticity due to incomplete
spinal cord injury. Electroencephalogr. Clin. Neurophysiol. 1996;101:
84 –92.
Bracken MB, Collins WF, Freeman DF, Shepard MJ, Wagner FW, Silten
RM, Hellenbrand KG, Ransohoff J, Hunt WE, Perot PLJ, Grossman
RG, Green BA, Eisenberg HM, Rifkinson N, Goodman JH, Meagher
JN, Fisher B, Clifton GL, Flamm ES, Rawe SE. Efficacy of
methylprednisolone in acute spinal cord injury. J Am Med Assoc
1984;251:45–52.
Bracken MB, Shepard MJ, Collins WF, Holford TR, Young W, Baskin DS,
Eisenberg HM, Flamm E, Leo-Summers L, Maroon J, Marshall LF,
Perot PL, Piepmeier J, Sonntag VKH, Wagner FC, Wilberger JE, Winn
HR. A randomized, controlled trial of methylprednisolone or naloxone
in the treatment of acute spinal-cord injury. N Engl J Med 1990;322:
1405–11.
Bracken MB, Shepard MJ, Holford TR, Leo-Summers L, Aldrich EF, Fazl
M, Fehlings M, Herr DL, Hitchon PW, Marshall LF, Nockels RP,
Pascale V, Perot PLJ, Piepmeier J, Sonntag VK, Wagner F, Wilberger
JE, Winn HR, Young W. Administration of methylprednisolone for 24
or 48 hours or tirilazad mesylate for 48 hours in the treatment of acute
spinal cord injury. Results of the Third National Acute Spinal Cord
Injury Randomized Controlled Trial. National Acute Spinal Cord Injury
Study. J Am Med Assoc 1997;277:1597–604.
Brown PJ, Marino RJ, Herbison GJ, Ditunno Jr. JF. The 72-hour
examination as a predictor of recovery in motor complete quadriplegia.
Arch Phys Med Rehabil 1991;72:546–8.
Bunge RP. The role of the Schwann cell in trophic support and
regeneration. J Neurol 1994;242:S19–S21.
Burns AS, Ditunno JF. Establishing prognosis and maximizing functional
outcomes after spinal cord injury: a review of current and future
directions in rehabilitation management. Spine 2001;26:S137– 45.
Burns AS, Lee BS, Ditunno Jr. JF, Tessler A. Patient selection for clinical
trials: the reliability of the early spinal cord injury examination.
J Neurotrauma 2003;20:477–82.
Bursell JP, Little JW, Stiens SA. Electrodiagnosis in spinal cord injured
persons with new weakness or sensory loss: central and peripheral
etiologies. Arch Phys Med Rehabil 1999;80:904 –9.
Calancie B, Molano MDR, Broton JG. Neural plasticity as revealed
by the natural progression of movement expression—both voluntary
and involuntary—in humans after spinal cord injury. In: Seil FJ,
editor. Progress in brain research. Amsterdam: Elsevier; 2000.
p. 71–88.
Calancie B, Molano MDR, Broton JG, Bean JA. Relationship between
EMG and muscle force after spinal cord injury. J Spinal Cord Med
2001;24:19–25.
Cameron T, Calancie B. Mechanical and fatigue properties of wrist flexor
muscles during repetitive contractions after cervical spinal cord injury.
Arch Phys Med Rehabil 1995;76:929–33.
Chang CW. Evident trans-synaptic degeneration of motor neurons after
stroke: a study of neuromuscular jitter by axonal microstimulation.
Electroencephalogr. Clin. Neurophysiol. 1998;109:199–202.
Daniels L, Worthingham C. Muscle testing: techniques of manual
examination. Philadelphia, PA: Saunders; 1980.
 B. Calancie et al. / Clinical Neurophysiology 115 (2004) 1748–1759
De Luca CJ, Merletti R. Surface myoelectric signal cross-talk among
muscles of the leg. Electroencephalogr Clin Neurophysiol 1988;69:
568–75.
Ditunno JF, Formal CS. Chronic spinal cord injury. N Engl J Med 1994;
330:550–6.
Duran FS, Lugo L, Ramirez L, Eusse E. Effects of an exercise program on
the rehabilitation of patients with spinal cord injury. Arch Phys Med
Rehabil 2001;82:1349–54.
Eidelberg E, Nguyen LH, Polich R, Walden JG. Transsynaptic degener-
ation of motoneurones caudal to spinal cord lesions. Brain Res Bull
1989;22:39 –45.
Falci SP, Lammertse DP, Best L, Starnes CA, Prenger EC, Stavros AT,
Mellick D. Surgical treatment of posttraumatic cystic and tethered
spinal cords. J Spinal Cord Med 1999;22:173–81.
Farina D, Merletti R, Indino B, Nazzaro M, Pozzo M. Surface EMG
crosstalk between knee extensor muscles: experimental and model
results. Muscle Nerve 2002;26:681–95.
Fawcett JW. Spinal cord repair: from experimental models to human
application. Spinal Cord 1998;36:811–7.
Geisler FH, Coleman WP, Grieco G, Poonian D. The Sygen multicenter
acute spinal cord injury study. Spine 2001;26:S87–S98.
Geller HM, Fawcett JW. Building a bridge: engineering spinal cord repair.
Exp Neurol 2002;174:125–36.
Hill CE, Beattie MS, Bresnahan JC. Degeneration and sprouting of
identified descending supraspinal axons after contusive spinal cord
injury in the rat. Exp Neurol 2001;171:153 –69.
Honmou O, Felts PA, Waxman SG, Kocsis JD. Restoration of normal
conduction properties in demyelinated spinal cord axons in the adult rat
by transplantation of exogenous Schwann cells. J Neurosci 1996;16:
3199–208.
Hugon M. A discussion of the methodology of the triceps surae T- and H-
reflexes. In: Desmedt JE, editor. New developments in EMG and
clinical neurophysiology. Basel: Karger; 1973. p. 773 –80.
Jacobs PL, Nash MS, Rusinowski JW. Circuit training provides
cardiorespiratory and strength benefits in persons with paraplegia.
Med Sci Sports Exerc 2001;33:711–7.
Jankowska E, Padel Y, Tanaka R. Projections of pyramidal tract cells to a-
motoneurones innervating hind-limb muscles in the monkey. J Physiol
1975;249:637 –67.
Jones LL, Oudega M, Bunge MB, Tuszynski MH. Neurotrophic factors,
cellular bridges and gene therapy for spinal cord injury. J Physiol 2001;
533:83–9.
Kirshblum S, Ho CH, House JC, Druin E, Nead C, Drastal S. Rehabilitation
of spinal cord injury. In: Kirshblum S, Campagnolo DI, DeLisa JA,
editors. Spinal cord medicine. Philadelphia, PA: Lippincott Williams
and Wilkins; 2002. p. 275–98.
LeBlanc HJ, Gray LW, Kline DG. Stab wounds of the cauda equina. Case
report. J Neurosurg 1969;31:683– 5.
Lee TT, Alameda GJ, Camilo E, Green BA. Surgical treatment of post-
traumatic myelopathy associated with syringomyelia. Spine 2001;26:
S119–27.
Little JW, DeLisa JA. Cauda equina injury: late motor recovery. Arch Phys
Med Rehabil 1986;67:45–7.
Little JW, Robinson LR, Goldstein B, Stewart D, Micklesen P.
Electrophysiologic findings in post-traumatic syringomyelia: impli-
cations for clinical management. J Am Paraplegia Soc 1992;15:44–52.
1759
Maertens de Noordhout A, Rapisarda G, Bogacz D, Gerard P, De Pasqua V,
Pennisi G, Delwaide PJ. Corticomotoneuronal synaptic connections in
normal man. An electrophysiological study. Brain 1999;122:1327– 40.
Marino RJ, Barros T, Biering-Sorensen F, Burns SP, Donovan WH, Graves
DE, Haak M, Hudson LM, Priebe MM. International standards for
neurological classification of spinal cord injury. J Spinal Cord Med
2003;26(1):S50 –6.
Nathan PW. Effects on movement of surgical incisions into the human
spinal cord. Brain 1994;117:337 –46.
Nathan PW, Smith MC, Deacon P. The corticospinal tracts in man. Course
and location of fibres at different segmental levels. Brain 1990;113:
303– 24.
National Spinal Cord Injury Statistical Center, Birmingham AB. Spinal
cord injury: Facts and Figures at a glance. J Spinal Cord Med 2002;25:
139– 40.
Nuwer MR, Daube J, Fisher C, Schramm J, Yingling CD. Neuromonitoring
during surgery. Report of an IFCN committee. Electroencephalogr Clin
Neurophysiol 1993;87:263 –76.
Petrov MA. Transplantations of nerves and roots of the spinal cord.
Reconstr Surg Traumatol 1971;12:250–62.
Piepmeier JM, Jenkins NR. Late neurological changes following traumatic
spinal cord injury. J Neurosurg 1988;69:399 –402.
Schwab ME, Bartholdi D. Degeneration and regeneration of axons in the
lesioned spinal cord. Physiol Rev 1996;76:319–70.
Schwartz M. Helping the body to cure itself: Immune modulation by
therapeutic vaccination for spinal cord injury. J Spinal Cord Med 2003;
26(1):S6–S10.
Sekhon LH, Fehlings MG. Epidemiology, demographics, and pathophy-
siology of acute spinal cord injury. Spine 2001;26:S2– 12.
Sherwood AM, McKay WB, Dimitrijevic MR. Motor control after spinal
cord injury: assessment using surface EMG. Muscle Nerve 1996;19:
966– 79.
Smith KA, Rekate HL. Delayed postoperative tethering of the cervical
spinal cord. J Neurosurg 1994;81:196 –201.
Stein RB, Gordon T, Jefferson J, Sharfenberger A, Yang JF, Totosy de
Zepetnek J, Belanger M. Optimal stimulation of paralyzed muscle after
human spinal cord injury. J Appl Physiol 1992;72:1393–400.
Thomas CK, Broton JG, Calancie B. Motor unit forces and recruitment
patterns after cervical spinal cord injury. Muscle Nerve 1997;20:
212– 20.
van Vugt JPP, van Dijk JG. A convenient method to reduce crosstalk in
surface EMG. Clin Neurophysiol 2001;112:583 –92.
Waters RL, Adkins RH, Yakura JS, Sie I. Motor and sensory recovery
following complete tetraplegia. Arch Phys Med Rehabil 1993;74:
242– 7.
Waters RL, Adkins R, Yakura J, Sie I. Functional and neurologic recovery
following acute SCI. J Spinal Cord Med 1998;21:195–9.
Waxman SG, Utzschneider DA, Kocsis JD. Enhancement of action
potential conduction following demyelination: experimental
approaches to restoration of function in multiple sclerosis and spinal
cord injury. In: Bloom F, editor. Progress in brain research.
Neuroscience: from the molecular to the cognitive. London: Elsevier;
1994. p. 233– 43.
Yang JF, Stein RB, Jhamandas J, Gordon T. Motor unit numbers and
contractile properties after spinal cord injury. Ann Neurol 1990;28:
496– 502.