Effects of Muscle Croup and Placement Site on
Reliability of and-~eld Dynamometry
Strength Measurements
Lori Mitchell M c M a h o n , MSE,
PT'
C. Burdett, PhD, PTZ
Susan L. Whitney, MS, PT, ATC3
Ray
T
he use of manual muscle testing as a clinical
indicator of strength
has been subject to
criticism for various
reasons, including subjectivity at
higher muscle grades, an ordinal
measurement scale, and relatively
large increments between muscle
grades (2, 15). Hand-held dynamometers (HHD) have been proposed as
an alternative to manual muscle testing for obtaining strength measures
in the clinic (5,7,8,lO). During a
measurement session, the tester stabilizes the HHD against the extremity being tested while the patient exerts a maximal force against the dynamometer. T h e HHD is objective,
uses a measurement scale with equal
increments, and provides small increments between measured values.
Muscle strength is defined as the
amount of torque that a muscle
group can exert (14). This torque is
generally measured as the amount of
external force that can be exerted at
a given lever arm distance (torque =
force x lever arm), where the lever
arm is the perpendicular distance
from the joint center to the force
action line. HHD outputs are force
readings, and, therefore, are not direct measures of muscle strength.
When the HHD is applied at a con-
Studies of measuring muscle strength with hand-held dynamometers have produced a variety
of results. The purpose of this research was to further investigate the effect of muscle group and
placement site on reliability. The purpose of Part I of this study was to examine reliabilities of force
measurements generated by four specific muscle groups using a hand-held dynamometer (HHD).
Part 11's purpose was to determine the effects of HHD placement site on the variability of HHD force
measurements. In Part I, two testers obtained measurements of right shoulder abductor, wrist
extensor, hip flexor, and ankle dorsiflexor forces in 20 subjects. Two-way analysis of variance
indicated a main effect due to tester, but no tester by session interaction and no main effect due to
session (p < 0.5). lntraclass correlation coefficients ranged from .76-.93 for within-session, intratester reliabilities, .67-.84, for between-session intratester reliabilities, and .30-.83 for within-session,
intertester reliabilities. Reliability tended to be higher when HHD placement sites were iarther from
joint centers. Part 11 explored the hypothesis that HHD forces would be less variable if measured
distally. One tester measured shoulder abductor forces for 30 subjects at three sites on the upper
extremity. Bartlett's Test for homogeneity of variance indicated a lower variability at the distal
placement site (p < 0.05).
Key Words: muscle strength, reliability, hand-held dynamometer
' Research engineer, Department of Otolaryngology, The Eye & Ear, Institute of Pittsburgh, Pittsburgh PA;
formerly instructor, Departrnent of Physical Therapy, University of Pittsburgh
2Associateprofessor, Department of Physical Therapy, University of Pittsburgh, Pittsburgh, PA
'Assistant professor, Department of Physical Therapy, University of Pittsburgh, Pittsburgh, PA
sistent position, however, lever arm
length remains constant and, therefore, the force reading can be used
as an indicator of muscle strength.
According to Blesh (3). reliability
may be given a subjective rating according to four ranges. Values
greater than 0.9 are considered to
be high; those ranging from 0.80 to
0.89 are termed good; those between 0.70 and 0.79 are fair; and
reliabilities less than 0.70 are considered poor. Intratester reliability of
HHD measurements has been shown
to be high in both healthy and patient populations within a session
(4,ll). Riddle et al also reported
high within-session intratester reliability (38-.98) for paretic and nonparetic sides of braindamaged patients. For sessions 2 days apart,
however, they reported high intratester reliability for the paretic
side (.9-.98) and varied intratester
reliabilities for some muscle groups
of the nonparetic side (.31-.93) (1 2).
Volume 15 Number 5 May 1992 JOSPT
Studies of intertester reliability
have generated a variety of results.
Bohannon and Andrews (6), using a
varied patient sample, found relatively high intertester reliabilities
(.88-.94). Byl ( I I) used healthy subjects and reported lower intertester
reliabilities (.52-.84) than Bohannon
and Andrews. Agre et al (1) found
good t o high intertester reliabilities
for the upper extremities (.79-.99)
but poor to fair intertester reliabilities for the lower extremities (.49.8 1). Agre suggested several factors
that may contribute t o low intertester reliability, including strength
of the tested muscle group, ability of
the tester t o stabilize the instrument
when obtaining H H D measurements, and distance from joint centers t o H H D placement sites. Riddle
et al (1 2) examined the relationship
between muscle strength and reliability and reported a Pearson product-moment correlation of 0.1. This
result does not support the speculation by Agre e t al (1) that strength of
the muscle being tested is a factor in
H H D reliability. Recently, however,
Wikholm and Bohannon (16) reported that tester strength influences
the magnitude and reliability of
H H D measurements for muscle
forces greater than 120 N. No studies have investigated the effects of
H H D distance from the joint center
on reliability of H H D force measurements.
This study was conducted in two
parts. T h e purpose of the first part
was to add t o the current knowledge
base regarding intertester and intratester reliability of H H D measurements. Specifically, within-session intertester and intratester reliabilities
as well as between-session intratester
reliabilities were determined for four
muscle groups (shoulder abductors,
wrist extensors, hip flexors, and ankle dorsiflexors). In the second part,
the relationship between H H D
placement sites and reliability of
shoulder abductor force measurements was investigated.
JOSPT Volume 15 Number 5 May 1992
PART I
METHOD
Subjects
Twenty young adult females
(x= 22.8 years, range 19-30 years)
with n o known muscle strength limitations o r orthopaedic pathology
served as subjects rbr this part of the
study. All subjects signed an informed consent form approved by
the Biomedical Internal Review
Board of the University of Pittsburgh.
dynamometers have
been proposed as an
alternative to manual
muscle testing for
obtaining strength
measures in the clinic
--
Data Collection
A spring-loaded H H D with a
measurement range of 0-267 N in
4.45 N increments was used
throughout the study (SPARK Instruments 8c Academics. Inc, P O
Box 5 123. Coralville IA 5224 1).
T h e dynamometer was calibrated
prior t o data collection and at the
end of the study by applying 133.5
N of free weights t o the dynamometer in 44.5-N increments. Over the
course of 2 weeks of data collection,
the calibration of the dynamometer
remained linear and varied less than
4.45 N with the 133.5 N of loading.
During the study, some subjects exerted greater than 133.5 N of force,
but the dynamometer's linearity
above this value was not determined.
Measurements were recorded t o the
nearest 2.23 N.
T w o physical therapists served as
testers for this study. A I-hour practice session was held t o familiarize
them with the use of the H H D and
the procedures for measuring the
four muscle groups in Part I of this
study. Neither tester had been using
H H D with regularity in the clinic.
Isometric "make" tests were performed in standard, against-gravity
positions for four muscle groups:
right shoulder abductors, wrist extensors, hip flexors, and ankle dorsiflexors. Positions and stabilization
strategies are listed in Table 1. During an isometric "make" test, the
tester stabilizes the H H D while the
subject exerts a maximum isometric
force against it. Fach subject's force
outputs were measured during session one by each tester and remeasured 1 week later (session two).
again by both testers. During each of
the sessions, three dynamometer
measurements were obtained by
both testers for each of the four
muscle groups. Thirty seconds of
rest were allowed between each trial.
T h e tester placed the dynamometer
at the appropriate position on the
upper extremity, stabilized the dynamometer perpendicularly to the
body segment, and asked the subject
to exert a maximal force for 3 seconds until told to relax. T h e other
tester read the dynamometer and recorded the value. After tester one
completed data collection from all
four muscle groups, tester two completed the same procedure with the
subject. Subjects were assigned numbers, with tester one performing initial data collection on the even-numbered subjects and tester two performing the initial data collection on
the odd-numbered subjects. Testing
was performed in the same order
during both sessions.
Data Analysis
In comparing force measurements between testers and between
R E S E- A R .C-H - S-T-U- D Y------.
----
--.-- -----
-- -- ----- ------
Muscle Group
Position
Stabilization Strategy
Shoulder abductors
.Sitting: shoulder at 90" abduction; elbow fully extended
Force applied 3 cm proxima1 to tip of olecranon
process
.Sitting: forearm pronated
and resting on treatment
table surface; wrist and
hand over edge of table;
wrist at neutral; fingers
flexed
Force applied just proximal
to MCP joint of third digit
.Sitting: hips and knees
flexed to 90'; thigh and
foot unsuppofied
.Force applied 5 cm proximal to superior pole of patella
.Sitting: hips and knees
flexed to 90"; ankle in 0"
DF
Force applied on MTP joints
.Subject's contralateral UE
holding chair seat
Wrist extensors
Hip flexors
Ankle dorsiflexors
-- -
.Subject's feet flat on floor
.Subject uses contralateral
hand to stabilize tested
forearm
.Subject's hands gripping
edge of plinth
.Subject's heel resting on
floor
-the testers in the two sessions. T h e
results of the four two-way
ANOVAs are shown in Table 3. For
the shoulder abductors, hip flexors,
and wrist extensors, there was n o
tester X session interaction and no
main effect due t o session. T h e r e
was, however, a main effect due to
tester for these three muscle groups.
This indicates that measurements
obtained by each tester were consistent from session to session, but that
there was a significant difference between the measurement obtained by
the two testers for these muscle
groups. Tester two recorded higher
force measurements for all of these
muscle groups. For the ankle dorsiflexors, there was a significant inter-
TABLE 1. Positioningfor HHD testing.
sessions, the highest of the three
force readings obtained by each
tester was used. T h e highest reading
was used rather than the mean value
to limit the effects of fatigue and
learning. T h e reliability of the HHD
force measurements was examined
by two methods: 1 ) two-way analysis
of variance (ANOVA) with repeated
measures and 2) intraclass correlation coefficients (ICC) (1 3).
T h e two-way repeated measures
ANOVA was used with tester as one
factor and session as the other factor. T h e main effects of tester and
session, as well as any interaction effects between tester and session,
were determined ( p < 0.05). In the
absence of interaction, a nonsignificant tester effect would tend to indicate a high intertester reliability. A
nonsignificant session effect would
tend to indicate high intratester reliability.
Within-session intratester reliability was determined for each therapist and each muscle group by calculating ICCs, comparing the three
measurements of each muscle group
for that particular session. As a mea-
Tester l
Tester 2
Session Session
Muscle Group Session Session
2
1
2
,
----
X S K S S ~ S K S
Hip flexors
Shoulder
abductors
Ankle
dorsiflexors
Wrist
extensors
190 37 200 36 208 33 212 35
127 20 123 25 135 25 133 25
167 36 176 33 186 40 179 34
The highest reliability
was for measurement
of the shoulder
abductors and the hip
flexors, while the
lowest reliability was
for the wrist extensors.
99 12 99 11 113 19 113 19
TABLE 2. Muscle force measurements (N).
sure of between-session intratester
reliability, ICCs were calculated for
each tester using the forces measured at the first and second sessions.
As a measure of within-session intertester reliability, intraclass correlation coefficients were calculated for
each session using the maximum
forces measured by the two testers.
RESULTS
Table 2 shows the means and
standard deviations of the force
measures obtained for the four muscle groups, as measured by each of
action between testers and sessions
(Figure 1). There was a large difference between force measurements
recorded by the two testers during
session one and little difference in
session two. There was no main effect, however, due to session o r
tester.
T h e reliabilities as indicated by
the ICC's are shown in Table 4. T h e
within-session intratester reliabilities
ranged from 0.76-0.93, with measurements of the shoulder abductors
and hip flexors showing the highest
reliabilities and measurement of the
wrist extensors showing the lowest
reliability. Between-session intratester reliabilities ranged from 0.670.84, with the shoulder abductors
Volume 15 Number 5 May 1992
JOSPT
RESEARCH S T U D Y
Muscle Group
Source
Hip flexors
df
SS
MS
F
Tester
Session
Tester x Session
Tester
Session
Tester x Session
Tester
Session
Tester x Session
Tester
Session
Tester x Session
Shoulder abductors
Wrist extensors
Ankle dorsiflexors
'Significant at p < 0.05.
TABLE 3. Results of four two-way repeated measures analysis of variance.
Shoulder
Abductors
1 3 8 ~
Hip Flaws
Team 1
Session 1
Session 1
Session 2
Session 2
Wrist Extensors
I2OT
st
I02+
I
I
Session 1
I
Tester 2
Tester 1
Session 1
Session 2
FIGURE 1. Mean forces measured by both testers in both sessions. Shoulder abductors, hip flexors, and wrist
extensors showed a main effect due to differences between testers, while the ankle dorsiflexors showed a tester
X session interaction.
having the highest reliability and hip
flexors the lowest. T h e intertester
reliabilities ranged from 0.3 1-0.83.
Measurement of the shoulder abductors again showed the highest reliability, and measurement of the wrist
extensors generated the lowest.
JOSPT Volume 15 Number 5 May 1992
The muscle groups
with the shortest
moment arms (wrist
extensors and ankle
dorsiflexors) had
consistently lower
reliabilities than the
hip flexor and
shoulder abductor
muscle groups.
.
I
Session 2
of Bohannon (4), who found high
test-retest reliabilities for ankle dorsiflexion, shoulder abduction, wrist
extension, and hip flexion as well as
other muscles during a single testing
session. Riddle et al (12) reported
good to high within-session intratester reliabilities for ankle dorsiflexors,
wrist extensors, and hip flexors in
the paretic and nonparetic limbs of
brain-damaged patients. Riddle et
al's testing included three of the four
muscle groups tested in the present
study.
Between-session intratester reliabilities were good for the shoulder
abductor muscle group, but fair for
DISCUSSION
Using the guidelines suggested
by Blesh (3). results from the present
study indicate good within-session intratester reliability for all four muscle groups. T h e results support those
the hip flexor, wrist extensor, and
ankle dorsiflexor measurements.
These results are similar to those reported by Riddle et al (1 2). Their
research generated high reliabilities
for paretic limb force outputs, but
fair to poor ICCs for nonparetic
limb forces. T h e ICC for nonparetic
wrist extensors was reported to be
0.3 1.
Wikholm and Bohannon (16) reported that the strongest tester
measured the highest HHD force
readings for shoulder external rotator, elbow flexor, and knee extensor
---- -
R .E-S-E-A-R-C-H S T U D Y
Muscle Group
lntratester Reliability within a Session
Tester 2
Tester 1
Day 1
Day 2
Day 1
Day 2
Mean ICC per
Group
.79
.89
.76
.83
.91
.93
.77
.89
.93
.80
.80
.90
.91
.89
.88
.88
.88
.80
.86
Shoulder abductors
Hip flexors
Wrist extensors
Ankle dorsiflexors
Muscle Group
.84
Intratester Reliability between
Sessions
Tester 1
Tester 2
Mean
Intertester Reliability within Sessions
Session 1
Session 2
Mean
-
Shoulder abductors
Hip flexors
Wrist extensors
Ankle dorsiflexors
.84
.67
.76
.84
.81
.75
.74
.70
.83
.71
.75
.77
.83
.76
.31
.55
.79
.68
.33
.48
.81
.72
.32
.52
TABLE 4. Reliabilities based on K C .
measurements. In the present study,
however, the stronger tester (tester
one) measured consistently lower
H H D force readings for all four
muscle groups. Wikholm and Bohannon also reported a decrease in reliabilities of H H D measurements with
increased tested muscle group force
production. T h e present authors did
not find such a relationship between
reliabilities and muscle group force
production, but this may be the result of testing four muscle groups
with a narrow range of force outputs: Wikholm and Bohannon's
force outputs across muscle groups
ranged from 1 14-429 N, while
the present study's ranged from
99-212 N.
T h e intertester reliabilities in the
current study were poor for the
wrist extensors and ankle dorsiflexors, fair for the hip flexors, and
good for the shoulder abductors.
Note that the muscle groups with
the shortest measurement moment
arms (wrist extensors and ankle dorsiflexors) had consistently lower reliabilities than the hip flexor and
shoulder abductor muscle groups.
O n the basis of this observation, the
authors decided t o test the hypothesis that a more distal placement site
would yield more reliable H H D
measurements for a particular muscle group.
PART II
METHOD
Subjects
Eighteen female and 12 male
healthy adult subjects (y = 24 years,
range 20-35 years) participated in
the study. Subjects excluded from
the study included those with histories of upper extremity fracture.
shoulder subluxation/dislocation,
shoulder instability, rotator cuff injury, and axillary nerve pathology.
Informed, written consent was obtained from each subject in compliance with the Human Subjects Committee of the Biomedical Internal
Review Board a t the University of
Pittsburgh.
Data Collection
A design was employed in which
three force measurements were obtained at each of three sites on the
dominant arms of all subjects, using
the same type of H H D as in Part I of
this study. Upper extremity dominance was determined by handedness preference. T h e positioning of
the dynamometer a t each of the
three sites was standardized by marking the subjects' skin with a felt-tip
marker. Marks were made both 10
cm and 2.5 cm proximal t o the tip of
the olecranon process (sites A and B,
respectively) as well as the proximal
aspect of the ulnar styloid process
(site C). Markings of all sites for all
subjects were performed by tester
one. In order to account for subject
fatigue, order of the testing sites was
randomized, and the subjects were
permitted to rest for 2 minutes between each measurement.
Body position during measurement of shoulder abduction was the
same as in Part I (see Table 1). Subjects were instructed to allow their
nondominant hands t o rest on their
laps. Tester two stood on a 6-in stool
behind the subject, positioned the
subject's upper extremity to 90" of
abduction, and placed the instrument on the appropriate site. Subjects were asked to perform a maximal isometric "make" contraction of
their dominant shoulder abductor
muscles for each of the nine measurements. T h e dynamometer was removed from the subject's arm after
3 seconds. Tester two then gave the
H H D to tester one for reading and
recording of the measurement. Dynamometer measurements were read
t o the nearest 2.23 N. Tester two
performed the measurements for all
subjects.
DATA ANALYSIS
T o determine the reliabilities of
the force measurements, an ICC (1 3)
was calculated for each of the three
sites. Also, Bartlett's Test for homogeneity of variance (9) was used t o
test the null hypothesis that the variances were equal at all three sites.
T h e alternate hypothesis stated that
at least one of the sites' variances differed significantly from the others.
RESULTS
Table 5 summarizes the ICC values as well as the within subject variance for each of the three sites. T h e
Volume 15 Number 5 May 1992 JOSPT
R E S E A R C H S .T- U- D .Y.
ICC values were >0.90 for each site,
indicating that all three sites yielded
highly reliable intratester dynamometer measurements. Bartlett's Test
for homogeneity of variance resulted
in rejection of the null hypothesis a t
the .05 level of significance. In order
t o determine which of the sites' variance differed significantly from the
others, the confidence intervals
(level = 0.9) were calculated and
plotted (Figure 2). Only site C's variance (most distal site) was determined to be significantly different,
since its confidence interval plot did
not intersect with the other two sites'
plots.
DISCUSSION
Measurement of shoulder abduction forces proved to be highly reliable for each of the dynamometer
placement sites. These results agreed
with those of Part I of this study in
which shoulder abduction was found
t o be highly reliable when using site
B. O n the basis of the ICC values
-
Placement Site
X (N)
A: 10 cm proximal to 212.33
olecranon process
B: 2.5 cm proximal to 176.71
olecranon process
96.92
C: just proximal to
ulnar styloid
process
ICC Variance
.91
1485.2
.94
2058.0
.94
824.6
TABLE 5. Mean, K C , and variance values lor force
measurements at three sites on the upper extremity
during shoulder abduction testing.
alone, it was not possible t o state
with confidence that one of the sites
was more reliable than any of the
other sites. However, Bartlett's Test
for homogeneity of variance showed
that site C, the most distal site, generated significantly smaller measurement variance than the other two
sites. T h e results of this test indicated that the most distal site was the
most reliable for measuring force
output of the shoulder abductors using HHD.
A possible reason for reliability
differences at the three H H D measurement sites may be due t o differences in lever arm measurement
lengths. An error in lever arm distance will change the H H D force
reading. For example, if a patient's
shoulder abductor muscle strength is
4 0 Nm, placement of the dynamometer 0.2 m from the center of the
glenohumeral joint would result in
an H H D reading of 200 N (neglecting the effect of gravity). If during a
retest of the same patient, a 2-cm
error occurred so that the dynamometer was placed 0.18 m from
the joint center, the H H D reading
would be (4O)/(O. 18) o r 222.2 N, resulting in a difference of 22 N in the
two measurements with n o actual
change in strength. If a more distal
testing site is used, for example, 0.4
m from the joint center, the H H D
reading would be 100 N. If a 2-cm
error in placement again occurred
during a retest (HHD a t 0.38 m
from the joint center), the H H D
Area of
Overlap
n
Variance (N2,
FIGURE 2. Ninety percent confidence intervals for three sites on the upper extremity. Site C's variance (most
distal site)was significantly smaller than the proximal sites' variances.
JOSPT * Volume 15 * Number 5 * May 1992
reading would be 105.3 N, resulting
in a force measurement difference of
only 5.3 N. T h e same 2-cm error in
dynamometer placement results in a
larger error in the H H D reading a t
the more proximal site. These examples show that, theoretically, the
more distal placement site would be
less sensitive to placement errors.
This may help to explain the significantly smaller variance obtained at
the distal measurement site (site C).
Due t o the short lever arm distance, proximal H H D placements
yield higher force readings. It is
more difficult, therefore, for a tester
to stabilize an H H D at proximal
rather than distal placement sites.
T h e tester's own strength may affect
his o r her ability to stabilize the
HHD, and this may also contribute
to the increase in proximal site variance.
O n the basis of these study results, the authors recommend a
choice of distal H H D placement sites
in order to minimize variance in
force measurements. T h e r e may be
other factors, however, that discourage distal placement sites. Such factors may include pathology of a joint
falling between the muscle group
being tested and the H H D placeJOSPT
ment site.
REFERENCES
Agre IC, Magness lL, Hull SZ, Wright
KC, Baxter TL, Patterson R, Stradel L:
Strength testing with a portable dynamometer: Reliability for upper and
lower extremities. Arch Phys Med Rehabil68:454-457, 1987
Beasley WC: Influence of method on
estimates of normal knee extensor
force among normal and post polio
children. Phys Ther Rev 36:2 1-4 1,
1956
Blesh TE: Measurement in Physical Education (2nd Ed), New York: The Ronald Press Co, 1 974
Bohannon RW: Test-retest reliability of
hand-held dynamometry during a single session of strength assessment.
Phys Ther 66:206-209, 1986
Bohannon RW: The clinical measure-
RESEARCH S T U D Y
6.
7.
8.
9.
- -
ment of strength. Clin Rehabil 1:5-16,
1987
Bohannon RW, Andrews AW: Accuracy of spring and strain gauge handheld dynamometers. I Orthop Sports
Phys Ther 10:323-325, 1989
Borden R, Colachis SC: Quantitative
measurement of the good and normal
ranges in muscle testing. Phys Ther
48:839-843, I968
Darcus HD: Strain-gauge dynamometer for measuring the strength of muscle contraction and for re-educating
muscles. Ann Phys Med l:l63-177,
19-52
Class CV, Hopkins KD: Statistical
Methods in Education and Psychology,
.
-
.
(2nd Ed), Englewood Cliffs, NI: Prentice-Hall, Inc, 1984
10. Newman LB: A new device for measuring muscle strength: The myometer.
Arch Phys Med Rehabil l:l63-177,
1952
11. Nies By1 N: lntrarater and interrater
reliability of strength measurements of
the biceps and deltoid using a handheld dynamometer. I Orthop Sports
Phys Ther 9:399-405, 1988
12. Riddle DL, Finucane SD, Rothstein lM,
Walker ML: lntrasession and intersession reliability of hand-held dynamometer measurements taken on
brain-damaged patients. Phys Ther
69: 182- 194, 1989
" .
13. Shrout PE, Fleiss /L: lntraclass correlations: Uses in assessing rater reliability.
Psych Bull 86:420-428, 1979
14. Smidt CL, Rogers MW: Factors contributing to the regulation and clinical
assessment of muscular strength. Phys
Ther 62: 1283- 1290, 1982
15. Wadsworth CT, Krishan R: lntrarater
reliability of manual muscle testing
and hand-held dynametric muscle
testing. Phys Ther 67: 1342-1347,
1987
16. Wikholm )B, Bohannon RW: Handheld dynamometer measurements:
Tester strength makes a difference )
Orthop Sports Phys Ther 13:191- 198,
1991
Volume 15 Number 5 May 1992
JOSPT