Journal of Electromyography and Kinesiology 9 (1999) 245–252

www.elsevier.com/locate/jelekin

Electromyographic patterns of individuals suffering from lateral

tennis elbow
Jeffrey A. Bauer *, Robert D. Murray
Department of Exercise and Sport Sciences, University of Florida, Gainesville, FL, USA

Received 30 January 1998; received in revised form 12 November 1998; accepted 20 November 1998

Abstract

This study investigated the applicability of using surface electromyography (EMG) as a tool for differentiating between persons
suffering from lateral tennis elbow and the healthy age-matched adults. Temporal muscle activation patterns of the tennis elbow
group were evaluated to determine if they varied between subject groups and if noted variations might be interpreted as arresting
or exacerbating the injury. Sixteen subjects (Healthy Controls, n ⫽ 6; Tennis Elbow, n ⫽ 10) were tested under simulated tennis
playing conditions. All subjects were males (Healthy group (CON) 38.8 ⫾ 13.1, Injured group (INJ) 40.8 ⫾ 10.8 yrs). EMG
response data, temporal and spatial muscle activities, of the forearm extensors (Ext), the forearm flexors (Flex) and the triceps (Tri)
were recorded for each subject during a single test session using all combinations of three different velocities on three different
racket head impact locations. Data were collected at a frequency of 1000 Hz. Statistical analysis was performed using a 2 ⫻ 3 ⫻
3 (Health status ⫻ Impact velocity ⫻ Impact location) ANOVA with repeated measures. Results indicated statistically significant
differences (p ⬍ 0.05) between the CON and INJ subject groups for the response variables associated with forearm extensor muscle
activation. During simulated play, the INJ group employed an earlier, longer, and greater activation of Ext than the CON group,
such changes may be considered detrimental to the healing process. These results support the use of surface EMG to quantify
differences in muscle activation strategies employed by individuals suffering from soft tissue muscle microtrauma injuries and
healthy controls.  1999 Elsevier Science Ltd. All rights reserved.

Keywords: Tennis elbow; Healthy vs. injured; Biomechanics; Tennis impacts; Forearm extensors

1. Introduction the activity that caused the condition to appear. Nirschl

Diagnosis and treatment of soft tissue damage are
often complicated by reliance on self-report to determine
the extent of injury. Recently, the clinician’s ability to
more accurately view the damage through quantitative
means, such as MRIs, has been greatly improved. How-
ever, such evaluation is expensive and may be deemed
unnecessary for many non-life-threatening conditions.
This study was conducted to determine if simple surface
electromyographic analysis of muscles associated with
lateral tennis elbow could be used to assist in accurately
diagnosing this common soft tissue injury.
While rest is the most frequently prescribed treatment
for most soft tissue injury, for athletes and industrial
workers, it may not be possible to cease or severely limit
* Corresponding author. Tel.: ⫹ 1-352-392-0584, ext: 263; fax: ⫹
1-352-392-0316; e-mail: jbauer@hhp.ufl.edu.
[12] has proposed using the term tendonosis due to the
lack of inflammation accompanying common lateral ten-
nis elbow. With an injury such as lateral epicondyle ten-
donosis, because the absence of tissue inflammation
accompanying common tennis elbow (TE), the pain
associated with the injury may actually encourage coun-
ter productive muscle activation [12]. It is therefore
important to determine if: (1) the patient is, in fact, suf-
fering from the condition, i.e. exhibiting signs different
from the normal population, and (2) possible detrimental
muscle activation patterns have been developed in
response to the injury. Electromyographic (EMG)
responses showing increases in muscular activation, fre-
quency, duration, and earlier onset related to specific
events would contradict Nirschl’s [12] assertion that rest
is the most effective initial form of treatment.
Assuming that a person sought medical attention for
his/her tennis elbow, a simple elbow exam may be per-
formed to determine the location of the pain. Self-report

1050-6411/99/$ - see front matter  1999 Elsevier Science Ltd. All rights reserved.
PII: S 1 0 5 0 - 6 4 1 1 ( 9 8 ) 0 0 0 5 1 - 0
246 J.A. Bauer, R.D. Murray / Journal of Electromyography and Kinesiology 9 (1999) 245–252
of pain is usually sufficient for a clinician to make a
recommendation of treatment, which typically includes
rest, strengthening and re-evaluation of activities
believed to have caused the injury [11,21,14,12]. How-
ever, absent from these evaluations is the ability to deter-
mine if muscle usage patterns have been altered by the
injury. EMG data from the extensor carpi radialis brevis
(ECRB) showing increased activity during tennis spe-
cific movements would indicate that muscular strain on
the damaged soft tissue, usually the hyaline region of
the ECRB [20], would also increase. Such data, as seen
in Fig. 3, would support our theory that individuals suf-
fering from TE employ muscle activation strategies that
actually increase their likelihood of further soft tissue
damage or at the very least slow the healing process.
1.1. History of tennis elbow
Although much has changed since the early days of
tennis, unfortunately, one aspect has remained remark-
ably consistent: those individuals who play the game
often will be physically injured by their participation in
the sport [16]. The most common of the upper body
injuries is “tennis elbow” or more correctly, lateral
epicondyle tendonosis [16]. Morris [10] documented this
injury just 10 years after Major Wingfield first intro-
duced the game. Morris [10] named the condition “lawn
tennis arm”. While the medical community has known
about the injury for over one hundred years, there still
seems to be little that physicians or physical therapists
can do to prevent the injury. Recent studies have also
provided a more sobering evaluation; although pain tem-
porarily disappears the tendon injury is irreversible [18].
When a player strikes a tennis ball with a racket, the
whole arm complex i.e., hand, wrist, forearm, upper arm,
and shoulder is mechanically stressed. Renstrom and
Johnson [19] calculated the magnitude of force stressing
the arm for a ball–racket impact at 13.9 m/s, the equival-
ent of jerking up a 25-kg weight with one arm. The lower
arm, consisting of the hand, wrist and forearm, sustains
most of the impact-generated stress. The soft tissue of
the lower arm is poorly suited to withstand this repeated
stress of hitting a ball.
While the exact frequency and level of mechanical
stress necessary to cause tennis elbow have not been
documented, it is now generally agreed that tennis elbow
is an overuse injury [13,6,21] whose origin can often be
traced to the backhand stroke in tennis. Even Morris [10]
stated as much over one hundred years ago in his first
reference to the injury when he noted that during the
execution of hitting a backhand the extensor muscles are
most stressed and most likely to be injured. Since the
back of the racket hand is parallel to the racket face and
on the leading side of the racket grip during the impact
portion of a typical one-handed backhand stroke, only
the fingers can provide bony support for the racket. The
extensor muscles are activated eccentrically to stabilize
the hand–wrist during the momentum transfer that
occurs during impact. It is well known that eccentric
muscle actions generate forces much greater than those
seen during normal concentric muscle actions [3]. These
large, eccentric forces put a great deal of stress on the
tendon ligament complex of the extensor muscles and
usually can be safely absorbed if the tendon insertion is
made over a relatively large area of bone. However, the
tendon most often identified as the site of tennis elbow
pain is the Extensor Carpi Radialis Brevis (ECRB), a
small muscle of the forearm, which has a distinctly poor
biomechanical design for withstanding high loads [20].
The ECRB inserts into the lateral epicondyle of the
humerus, an extremely small insertion area that Groppel
[5] hypothesized as the major reason for injury occurring
regularly at this point. None of the other extensor
muscles active during the backhand stroke have such a
biomechanical disadvantage.
Chop [2] has estimated that only five percent of tennis
elbow sufferers are tennis players. The injury may affect
carpenters, dental technicians, or computer operators.
Furthermore, it is believed that between 40 and 50% of
people who play tennis will develop tennis elbow during
their playing career [1,6,15,16]. For those who seek
medical advice about the injury, the most recommended
treatment is rest. Even if an athlete agrees to stop playing
tennis, there is still no assurance this action will prevent
further aggravation of the injury. People suffering from
tennis elbow want to continue the activity that is causing
the pain because it is part of their occupation or a sport
which they enjoy playing. During a tennis stroke they
attempt pain alleviation by trying different techniques to
limit the amount of wrist flexion during a swing. To
decrease pain temporarily and to stabilize the wrist, a
player flexes and rotates the wrist to grip the handle
tighter. This grip may decrease their pain but increases
the tension on the muscle and increases the likelihood
of injury at the lateral epicondlye of the humerus.
Although these techniques may temporarily reduce pain
they lead to greater structural damage.
While the medical and tennis playing communities
have known about tennis elbow for many years, a
definitive etiology has yet to be determined. Further-
more, the criteria used in assessing treatment success are
(1) the patient’s diminished perceived level of pain and
(2) his/her ability to participate in activities previously
prohibited by the pain. Currently there are no quantitat-
ive methods used to determine if treatments allow the
injured tendon to fully heal or to assess if the injury
results from muscle activity modification, possibly
inhibiting the healing of the damaged tissue.
The purpose of this study was to determine under
simulated tennis conditions if (1) surface EMG could be
used to accurately delineate between healthy subjects
and subjects suffering from tennis elbow and, (2) if acti-
 J.A. Bauer, R.D. Murray / Journal of Electromyography and Kinesiology 9 (1999) 245–252
vation patterns are different, to assess whether the differ-
ences should be considered helpful or detrimental to the
healing process.
2. Methods
This investigation attempted to determine the ability
of surface EMG to accurately distinguish the health
status of individuals suffering from a soft tissue injury
(tennis elbow) from that of a healthy control group based
on observed responses to forced wrist flexion similar to
that experienced during tennis play. Analysis was perfor-
med to determine if the INJ group employed muscle acti-
vation patterns that may be counterproductive to speedy
recovery from tennis elbow. Additional main effects of
ball velocity at impact, as well as impact location, were
investigated. Since impact velocity and location vary
during the course of normal play, these two main factors
were evaluated to determine if statistically significant
differences in the responses could be linked to such
effects.
2.1. Selection of subjects
Subjects were selected based on their physical con-
dition at the time of the study. CON subjects were volun-
teers who had no history of arm injuries and were not
suffering from tennis elbow at the time of their partici-
pation in this study. Those assigned to the INJ group
were clinically diagnosed as suffering from tennis elbow
at the time of their participation in this study. None of
the INJ subjects had altered their physical activities sig-
nificantly due to their injured state and, at the initiation
of the study, were not suffering pain sufficient to alter
activities of daily living.
Various researchers [8,13,16] have reported that those
most likely to suffer from tennis elbow are 35 to 45 years
of age. While the 35–45 age group may be at the highest
risk, Priest [16], Priest et al. [17], Goldie [4], Ilfeld and
Field [7], have reported verified cases of tennis elbow
in persons ranging in age from the teens to the seventies.
The mean age for all subjects in this study was 39.8
years. The injury shows no gender preference [8] but,
for reasons of convenience, all individuals tested were
males. Table 1 provides a summary of subject ages, hand
and forearm mass information.
Table 1
Subject summary data
Health status Age (yrs) Hand mass (g) Forearm mass (g)
Healthy (mean) 38.8 107.45 280.61
(Std. dev.) 13.1 6.38 17.67
Injured (mean) 40.8 104.49 270.92
(Std. dev.) 10.8 3.95 12.95
247
Fig. 1. Illustration of the physical dimensions, impact locations and
sensor placement of the tennis racket used during all trials.
2.2. Apparatus and subject preparation
The racket arm of each subject was instrumented with
surface silver–silver chloride disposable electromyo-
graphic electrodes. The electrodes were placed over the
muscle bellies of the Extensor Carpi Radialis Brevis,
Flexor Carpi Ulnaris and the long head of the Tricep
muscle with an interelectrode distance of 2.5 cm. It was
understood that the surface electrodes provide an aggre-
gate of the EMG activity in the various locations. The
locations were selected to provide as much input from
the specified muscles as possible since they are con-
sidered central to the injury pathology.
Pilot work was performed during which a wide range
of ball velocities and impact locations were tried to
determine the ability of the ball machine to deliver a ball
at a consistent velocity and to the desired impact
location. Once ball velocities and impact locations were
determined, data collection began.
Nine separate impact conditions were presented in
random order to each subject during the test session. The
nine conditions were comprised of each combination of
three impact velocities and three impact locations of ten-
nis balls on the racket face. All participants used the
same racket. Fig. 1 shows information about the dimen-
sions of the racket and location of the impacts. Table 2
summarizes the nine conditions of velocity and impact
location.
Each of the nine experimental trials lasted 30 seconds
with an average mean time between impact of 5 seconds.
The trial duration assured a minimum of five ball
impacts, of which at least three had to be in the pre-
scribed string face location for the trial to be considered
Table 2
Impact and location abbreviations
Impact Centre Long axis Torsional
Location speed
Low (11.94 m/s) CL LL TL
Medium (17.13 m/s) CM LM TM
High (22.95 m/s) CH LH TH
248 J.A. Bauer, R.D. Murray / Journal of Electromyography and Kinesiology 9 (1999) 245–252
successful, this was determined after testing completion
using video analysis. To accomplish this both frontal and
overview views of each impact were recorded using 30
Hz video. It was previously estimated through on-court
study that no more than 3 seconds were necessary for
racket preparation during normal match play. Following
each impact the subjects reoriented the racket to the
required starting position with the help of a free-standing
guide mechanism and remained stationary while waiting
for the next ball. Subjects were instructed to complete a
“block” volley, which is a shot with no backswing and
minimal forward follow through. All racket motion was
forward toward the ball machine.
It was determined that wait intervals of greater than
5 seconds proved to be too long a period between shots
and often resulted in subjects “false starting” the volley.
With rest–wait intervals of longer than 5 seconds, the
subjects became anxious and reported feeling fatigued
by the end of data collection. To avoid fatigue but pre-
vent boredom, 90-second rests were taken between trials.
This interval period proved adequate to change the velo-
city setting on the ball machine, orient the subject to the
new starting location of the racket for the next trial, and
to remove and replace a memory chip card in the data
logger unit (Paromed DLS-8C-System, Paromed Med-
zintechnik, Germany).
The data logger unit worn about the subject’s waist
was used to collect and store all sensor data at 1000 Hz.
Wire leads from 3 sets of surface EMG electrodes and
the cable from an accelerometer mounted on the racket
were connected to the data logger. The electrodes were
site-amplified with a gain of 50 and amplified by a factor
of 8 at the controller unit. A ground electrode was
affixed to the acromion process of the shoulder. A 0.5-
gram uniaxle accelerometer placed on the test racket pro-
vided a signal that was used as an event marker indicat-
ing the instant of ball–racket impact. Fig. 2 illustrates
a sample EMG output for a single experimental ball–
racket impact.
2.3. Calculation of ball impact force and ball velocity
To determine the peak impact force exerted on the
string surface resulting from ball impacts having differ-
ent velocities, a ball machine shot balls at a Bertec Cor-
poration model 4060 force plate fixed in a vertical pos-
ition 2 meters in front of it. A calibration factor was
calculated after placing known weights on the plate
immediately prior to and following the vertical mount-
ing. The instrumented tennis racket was fixed such that
the frame of the racket was supported 1.25 cm above
the force plate on a wooden frame. The force plate was
connected to a MetraByte Dash-16 data acquisition
board housed in a Toshiba 3200SX-laptop computer.
Data were collected at 1000 Hz for all impacts. The data
acquisition board’s configuration was as follows: 16 bit
resolution, 0 to 5-volt range, unipolar. Voltage resolution
was accurate to 0.01 volts. The conversion factor calcu-
lated for this experiment was 7.4 N/0.01 volts. Three
good trial values for each impact velocity, 11.95, 17.16
and 22.95 m/s were collected and averaged to provide
the force value per impact. The impact force values were
determined to be: 236.8, 495.8 and 732.6 N for the low,
medium and high velocity impacts, respectively.
Initially, these impact forces appear to be rather high
for a tennis ball hitting a racket at the indicated velo-
cities. However, a quick calculation shows that they are
indeed close to what should occur under normal game
conditions. The mass of a tennis ball is 52 grams. The
impact time to peak force for the impacts was calculated
to be 1 to 2 milliseconds. Using the equation F ⫽ ma
and substituting in the incoming velocity of 22.95 m/s
for the high velocity impact with an assumed time of
1.5 milliseconds for the time to peak force, the force is
calculated to be 795.6 N. This is accomplished as fol-
lows:
F ⫽ ma ⫽ > mv/time
⫽ > (0.052 kg)(22.95 m/s)/0.0015 sec ⫽ 795.6 N
Moment calculations for the three impact locations
follow: For a Central Impact the ball would strike the
racket 51 cm from the butt end of the racket. It was
assumed that the wrist would act as the axis of rotation
for this calculation and that it was located at the butt
end of the racket. If the moment arm was 51 cm then the
flexor moment experienced at the wrist for the impacts
conducted would equal those found in Table 3. Simi-
larly, the Long Axis impact, which had a moment arm
of 61 cm, was calculated. The flexor moment generated
by the Torsional impact would be the same as that for
the Central impacts since the moment arm would still
be 51 cm, however an additional twisting moment would
be generated and these values are found in Table 3.
3. Data analysis
3.1. Data collection and processing
The analog outputs of the EMG and accelerometer
sensors were collected at a sampling frequency of 1000
Hz for all test conditions. The data logger system col-
lected and stored the data with 8 bit resolution. All chan-
nels were polled within 20 micro-seconds during each
0.001 second sampling interval, providing complete
synchronization of the data logger sensor data. Two
software packages provided with the data logger system
(DLM.exe and ADGRAF.exe), assisted in initializing the
chip cards, downloading and quantitatively analyzing the
captured data. Custom software (10S-Filt.exe) was used
 J.A. Bauer, R.D. Murray / Journal of Electromyography and Kinesiology 9 (1999) 245–252 249

Fig. 2. Sample output showing Extensor, Flexor and Triceps activation relative to an impact. Accelerometer data used to determine the time of
impact is shown in channel 4.

Table 3 aging the values recorded during the most active 100-
Calculated flexion and rotational moments experienced for each impact msec interval during each MVC.
location and impact velocity
Once the impact time point was established the muscle
Impact location Central Long axis Torsional (In activation curves were analyzed to determine start and
addition to end times of muscle activation using the 20% threshold
central moment) values already established. This provided the muscle
activation duration (MAD) response value.
Impact velocity
Low (11.94 m/s) 120.7 Nm 144.4 Nm 11.8 Nm Calculating the area under the rectified EMG curve
Medium (17.13 m/s) 252.9 Nm 302.4 Nm 24.8 Nm for each muscle group during its total time of activation
High (22.95 m/s) 373.6 Nm 446.9 Nm 36.6 Nm for each impact generated an integrated EMG value
(IEMG).

to low pass filter EMG channel data as described by
Winter [22]. The data were downloaded to a computer
for analysis.
Data reduction for selected impacts were conducted
as follows: EMG data channels were full-wave rectified
and passed through a 6 Hz zero lag, 4th order But-
terworth filter generating a linear envelope. “On” thres-
hold values for each muscle group were calculated as
values of 20% of the peak signal, collected prior to the
start of each experimental trial block from maximum iso-
metric voluntary contractions (MVC) of the extensor,
flexor and triceps muscles. 20% was selected based on
pilot work that showed active preparation activity to be
greater than 20% of peak signal and passive limb orien-
tation activity below 20% of the peak signal. MVCs
were performed for each of the three muscle groups at
the beginning of each experimental block. A mean
maximum activation level was then determined by aver-
3.2. Electromyographic measurements
There are no reliable methods of directly comparing
the raw EMG readings from one individual to another.
However, since comparison of EMG responses was
necessary for this study, the condition that provided the
least stress on the arm during ball–racket impact was
selected and utilized to normalize several EMG vari-
ables. The central medium speed impact condition (CM)
was selected based on verbal responses provided by sub-
jects during pilot testing. They indicated that the impacts
experienced during CM trials were the least mechan-
ically stressful of the nine experimental test conditions.
Values of maximum EMG amplitude during MVC, mus-
cle activation pre-impact (NMAI) and integrated EMG
(IEMG) generated during the CM condition for each
individual were determined. The CM responses were
assigned a normalized value of 1.0. The total muscle
250 J.A. Bauer, R.D. Murray / Journal of Electromyography and Kinesiology 9 (1999) 245–252

activation duration (MAD) was not normalized and the

actual time that the EMG level was above 20% of the
MVC value collected during the CM trial was evaluated.

3.3. Statistical analysis

EMG magnitude values were normalized versus the

calibration values recorded during the maximum volun-
tary contract recordings. All response variable data were
analyzed using analysis of variance designs to determine
factor significance. The following model was used for
the ANOVAs which were performed for each dependent
response variable (i.e. MAD and IEMG). The model
used was as follows: Fig. 3. Mean muscle activation duration (MAD) of all muscle groups
for all subjects and all impact conditions. Only Extensor group data
indicated a significant difference between activation times based on
Yijkl ⫽ ␮... ⫹ ␣i ⫹ ␳j(i) ⫹ ␤k ⫹ ␥l ⫹ (␣␤)ik ⫹ (␣␥)il
health status (p ⬍ 0.05).
⫹ (␤␥)kl ⫹ (␣␤␥)ikl ⫹ ⑀(ijkl)

where ␣ ⫽ health status, ␳ ⫽ subject, ␤ ⫽ velocity, ␥
⫽ impact location, i ⫽ 1, 2; j ⫽ 1, . . ., n; k ⫽ 1, 2, 3;
l ⫽ 1, 2, 3.
The statistical calculations were performed using
MINITAB Release 9.2 for Windows (MINITAB Inc.,
State College, PA.). The statistical treatment was
designed to determine response variable significance
versus the test condition matrix for both inter subject
and between group pairings. Mean values (n ⫽ 10 for
INJ and n ⫽ 6 for CON) of the various measured para-
meters were compared with repeated measures analysis
of variance (ANOVA) designs. A series of 2 ⫻ 3 ⫻ 3
ANOVAs were performed to determine the significance
of the main effects health status, velocity and impact
location. Table 4 shows the ANOVA design matrix for
the repeated measure dependent response variable of
muscle activation duration (MAD).
4. Results
ditions used during this study. Fig. 1 illustrates the
impact locations studied.
Figs. 3–5 illustrate the responses evaluated during this
study. Fig. 3 shows the aggregate activation duration of
the all impact conditions relative to muscle group. Only
the Extensor group revealed a statistically significant dif-
ference in activation time (p ⬍ 0.05). Fig. 4 illustrates
the mean activation times of the forearm extensor
muscles for each impact conditions comparing the CON
vs. INJ responses. Fig. 6 shows the results of the inte-
grated EMG for all impact conditions of CON vs. INJ
for the extensor muscle group.
Statistical comparisons to evaluate interactions
between Health Status (HS), impact Location (LOC) and
impact Velocity (VEL) were performed. A significant
three way (HS-LOC-VEL) interaction (p ⬍ 0.05) was
found for the MAD response variable for the extensor
group activity.
The post hoc analysis of the three-way interaction for

The variable health status proved to be statistically

significant for all muscle activity responses for the exten-
sor group (p ⬍ 0.05). Activity levels of the flexors and
triceps were not statistically different between CON and
INJ participants (p > 0.05). Table 2 lists the abbrevi-
ations for the nine impact and ball velocity impact con-

Table 4
Example of the coding for the condition central impact, low velocity,
for healthy subject number one

Health Subject Velocity Location Muscle activation duration

status (MAD)
Fig. 4. Activation duration of Extensor muscle group for each impact
␣ ␳ ␤ ␥ Trial 1 Trial 2 Trial 3 location (Center, Long Axis, Torsional) response data are displayed.
1 1 1 1 xx xx xx In all instances there was a statistical difference relative to health status
(p ⬍ 0.05).
 J.A. Bauer, R.D. Murray / Journal of Electromyography and Kinesiology 9 (1999) 245–252
Fig. 5. Activation duration of Extensor muscle group for each impact
velocity (Low, Medium, High) response data are displayed. In all
instances there was a statistical difference relative to health status (p
⬍ 0.05).
Fig. 6. Integrated EMG signals for each of the nine impact con-
ditions. Response variables indicated significant differences for each
condition relative to health status (p ⬍ 0.05).
MAD revealed that there were thirteen pairwise signifi-
cant differences. Of the differences, twelve occurred
with interactions involving differing HS. It appears that
for MAD the condition of torsional impact location at
high speed (TH) most clearly delineates the subject
group responses. The differences created by TH can be
seen in Figs. 4 and 5. IEMG significance values were
not associated with any interaction terms and simply
diverged on the basis of HS.
5. Discussion
The results of this study indicate that surface EMG
provides useful data in evaluating biomechanical
changes that take place in the case of tennis elbow. This
result is significant since it is often very difficult to quan-
tify the magnitude of changes that result from soft
tissue injury.
The present data show that muscle activation patterns
are statistically different under identical conditions
251
between tennis elbow sufferers and players not suffering
from such an injury. Our findings that there is a differ-
ence in muscle activation for the extensor group with
respect to health status, are consistent with those
observed by Kelley, Lombardo, Pink, Perry and Gian-
garra [9] who reported significantly increased muscle
activity in the ECRB of injured tennis players while hit-
ting groundstrokes. It is also apparent that the changes
made by the tennis elbow sufferers are contrary to what
is prescribed by health care professionals to effectively
treat this condition, namely to rest the affected region.
Although, increases in EMG activity do not relate
directly to increases in force production by the muscles,
there is a strong correlation in non-pathologic groups
when muscles are not fatigued. It appears that the sub-
jects with tennis elbow were attempting to limit the
amount of forced flexion during backhand shots
(resulting from the ball impact) by increasing the activity
of the muscle group mainly affected by tennis elbow,
that is, the extensor group. The strategy of earlier exten-
sor activation and longer post impact activity displayed
in Fig. 7, which the INJ group adopted, was ineffective
in reducing pain due to the magnitude of the impact
forces.
It is likely that the earlier and longer activation of the
extensors relative to impact would be counter-productive
to minimizing the damaging effect of ball–racket
impacts. The subject’s early muscle activation assures
muscle stiffness at the time of impact, causing currently
damaged tendons to be pre-loaded prior to impact. More-
over, this activation forces forearm muscle–tendon com-
Fig. 7. Impact duration for each muscle group. Impact occurs at time
0 with pre-impact activation indicated as positive time and post-impact
activation indicated as negative time.
252 J.A. Bauer, R.D. Murray / Journal of Electromyography and Kinesiology 9 (1999) 245–252
plex into flexion and decreases eccentric load absorption
without directly stressing the injured hyaline region of
the ECRB tendon. However, the subsequent impact load
would then proceed to increase the likelihood of generat-
ing additional microtrauma in the already affected region
because there is little elastic deformation left to help dis-
sipate the forced lengthening of the muscle–tendon com-
plex.
By observing EMG responses of the major movers
and stabilizers of the racket arm, we were able to deter-
mine that the INJ group activated their forearm extensor
muscles earlier, longer, and attained greater levels of
activation than did their CON counterparts during simu-
lated tennis shot making conditions. This increased mus-
cle activation strategy was an attempt by the INJ injured
group to reduce the painful effects of the forced wrist
flexion associated with hitting a backhand volley. Kine-
matic video analysis of the motion revealed that the strat-
egy failed and that the forced flexion occurred to the
same magnitude for both INJ and CON participants for
all test conditions. Therefore, the increased activation
did not produce the desired outcome for the INJ subjects
and, in fact, would contribute to increased strain on the
injury soft tissue and promote muscle fatigue. Both the
increased strain and fatiguing characteristics of the acti-
vation strategy would be counter-productive to the heal-
ing process.
While this study dealt specifically with tennis-induced
lateral tennis elbow, regardless of its causes and out-
comes, the implication of our results are potentially sig-
nificant for evaluation and treatment of the 95% of tennis
elbow sufferers who develop the condition from non-
tennis related causes. The protocols established in this
study can assist the clinician and sport scientist in
improving diagnosis and treatment for soft tissue injury
sufferers like those who experience the pain of tennis
elbow. Further, the methodology and analysis presented
here may have value in evaluation and treatment of other
over-use micro-trauma type soft tissue injuries.
References
[1] Allman FL. Tennis elbow: who’s likely to get it, and how. Phys-
ician and Sports Medicine 1975;3(7):43–58.
[2] Chop WM. Tennis elbow. Postgraduate Medicine 1989;88:301–8.
´ ¨ ¨
[3] Friden J, Sjostrom M, Ekblom B. Myofibrillar damage following
intense eccentric exercise in man. International Journal of Sports
Medicine 1983;4:170–6.
[4] Goldie I. Epicondylitis lateralis. Acta Chirurgica Scandanavica,
Suppl. 1964;339:104–9.
[5] Groppel JL. The biomechanics of tennis: an overview. Inter-
national Journal of Sport Biomechanics 1986;2:141–55.
[6] Gruchow HW, Pelletier D. An epidemiologic study of tennis
elbow. American Journal of Sports Medicine 1979;14:195–200.
[7] Ilfeld FW, Field SM. Treatment of tennis elbow. Journal of
American Medical Association 1966;195:67–70.
[8] Kamien M. A rational management of tennis elbow. Sports Medi-
cine 1990;9(3):173–91.
[9] Kelley JD, Lombardo SJ, Pink M, Perry J, Giangarra CE. Electro-
myographic and cinematographic analysis of elbow function in
tennis players with lateral epicondylitis. The American Journal
of Sports Medicine 1994;22(3):359–63.
[10] Morris H. Lawn tennis elbow. British Medical Journal
1883;2:557.
[11] Nirschl RP. The etiology and treatment of tennis elbow. Journal
of Sports Medicine 1974;2(6):308–19.
[12] Nirschl RP. Elbow tendinosis/tennis elbow. Clinics in Sports
Medicine 1992;11(4):852–5.
[13] Nirschl RP, Pettrone FA. Tennis elbow. Journal of Bone and Joint
Surgery 1979;61A:832–9.
[14] Nirschl RP, Solilck J. Conservative treatment of tennis elbow.
Physician and Sports Medicine 1981;6:43–54.
[15] Nirschl R. Soft tissue injuries about the elbow. Clinics in Sports
Medicine 1986;5:637–52.
[16] Priest JD. Tennis elbow: the syndrome and a study of average
players. Minnesota Medicine 1976;59:367–71.
[17] Priest JD, Braden V, Gerverich SG. The elbow and tennis, part
2: a study of players with pain. Physician and Sports Medicine
1980;8(5):77–85.
[18] Regan W, Wold LE, Coonrad R, Morrey BF. Microscopic histo-
pathology of chronic refractory lateral epicondylitis. American
Journal of Sports Medicine 1992;20(6):746–9.
[19] Renstrom P, Johnson RJ. Overuse injuries of sports. Sports Medi-
cine 1985;2:316–33.
[20] Stoeckart R, Vleeming A, Snijders CJ. Anantomy of the extensor
carpi radialis brevis muscle related to tennis elbow. Clinical
Biomechanics 1989;4(4):210–2.
[21] Wadsworth TG. Tennis elbow: conservative, surgical and
manipulative treatment. British Medical Journal 1987;294:621–4.
[22] Winter DA. Biomechanics and motor control of human move-
ment. New York: John Wiley and Sons Inc., 1990:139–46.
Dr Jeffrey A. Bauer has been an Assistant Pro-
fessor and Director of the Biomechanics Labora-
tory at the University of Florida since 1994. He
completed his doctoral work in the “water
tower” at Penn State and was selected to start
an emphasis of study in biomechanics at the gra-
tuate level UF’s department of Exercise and
Sport Sciences. His primary research interests
are in determining effects of both repetitive and
acurate dynamic loading to the human body. His
teaching responsibilities include an undergrad-
uate Kinesiology course and Graduate courses
in Biomechanics and Biomechanical Instrumentation.
Robert Murray is a graduate of the College of
Health and Human Performance with a major in
Exercise and Sport Science emphasizing Exer-
cise Physiology. He minored in Chemistry,
Anthropology, and French. His plans to go to
medical school and someday practice orthop-
edic surgery.