Hartmann 2009

Clinical Section
Gerontology 2009;55:259–268 Received: November 9, 2007

Accepted after revision: May 29, 2008
DOI: 10.1159/000172832
Published online: November 10, 2008
Reproducibility of an Isokinetic
Strength-Testing Protocol of the Knee
and Ankle in Older Adults
Antonia Hartmann a, c Ruud Knols b Kurt Murer a Eling D. de Bruin a, b
a
Institute of Human Movement Sciences and Sport, ETH, b Department of Rheumatology and Institute of
Physical Medicine, University Hospital Zurich, Zurich, Switzerland; c Care and Public Health Research Institute,
Maastricht University, Maastricht, The Netherlands
Key Words tar flexion with SEMs that varied from 14 to 17% and RLOAs
Isokinetic test protocols ⴢ Isokinetic dynamometers ⴢ from 39 to 48%. Conclusion: The results of this study dem-
Protocol reproducibility ⴢ Knee/ankle contractions onstrate that the Biodex System 3 is a reliable device when
used for elderly living independently. The ability of the de-
vice to determine a real change in isokinetic ankle and knee
Abstract contractions is better on a group level than on an individual
Background: Muscle power assessed by isokinetic dyna- level. The Biodex System 3 can be employed with confidence
mometers has the potential for playing an important role in in studies to determine the effect of exercise intervention
investigating functional status in older subjects. Researchers programs on physical activity.
and clinicians are interested in the reliability of isokinetic test Copyright © 2008 S. Karger AG, Basel
protocols for the confidential assessment of status, as this
affects the interpretation of the results of an intervention
program. Objective: The current study investigated the in- Introduction
ter- and intrarater reliability of an isokinetic strength-testing
protocol of the knee and ankle preceded by a familiarization The loss of muscle strength and muscle power that
session. Methods: Twenty-four independently living elderly leads to a decline in functional status and an increased
subjects (6 males, 18 females, mean age 71.2 8 5.5 years) risk of falls is a common consequence of aging. Several
were assessed 3 times in two test sessions. The main out- basic tasks of daily life (climbing stairs, walking, chair
comes were the intraclass correlation coefficient, standard rise, etc.) are, above all, related to the ability to generate
error of measurements (SEM) and ratio of limits of agree- power around the ankle and knee joint [1–3], and more
ment (RLOA) for isokinetic knee and ankle contractions, as precisely to the ability to generate power at low velocities
measured with the Biodex System 3. Results: The intraclass [4]. Some studies showed that muscle power decreases at
correlation coefficients of the isokinetic variables varied a much greater rate [3, 5–7] – and may be more directly
from 0.81 to 0.99 representing ‘good’ to ‘very good’ reliabil- related to impaired physical performance in the elderly
ity. Most SEM and RLOA indexes represented acceptable [3, 6] – than isometric strength, making it a potentially
agreement which varied from 6 to 13 and 18 to 37%, respec- major etiological factor in age-related functional decline.
tively. Nonacceptable agreement was found for ankle plan- Moreover, Skelton et al. [2] showed, in a cross-sectional
Siriraj Medical Library, Mahidol University
202.28.191.34 - 3/17/2015 7:48:20 AM
© 2008 S. Karger AG, Basel Antonia Hartmann

0304–324X/09/0553–0259$26.00/0 Institute of Human Movement Sciences and Sport, ETH Hönggerberg
Fax +41 61 306 12 34 Wolfgang-Pauli-Str. 27, CH–8093 Zurich (Switzerland)
E-Mail karger@karger.ch Accessible online at: Tel. +41 44 632 3152, Fax +41 44 632 1383
Downloaded by:
www.karger.com www.karger.com/ger E-Mail antonia.hartmann@move.biol.ethz.ch

study, significant differences in leg power in fallers and Therefore, the aim of this study was to determine in-
nonfallers and no differences in leg strength. For such tra- and interrater reliability of a newly developed iso-
studies, precise, sensitive power testing methods are kinetic strength-testing protocol around the knee and
needed to investigate muscle function in older subjects. ankle joint in an older adult population.
One such measurement device is the isokinetic dyna-
mometer. However, data on isokinetic measurements in
aged populations are, at present, scarce. Methods
Little is known about the reliable use of isokinetic tests
for the lower limbs in older adults. Studies have focused Subjects
on the intrarater reliability of knee strength tests [8–12], Twenty-four independently living older subjects participated
only one report described ankle strength tests [8], and in the study. Community-dwelling subjects were recruited from
the elderly residence Diakoniewerk Neumunster-Zurich (n = 3)
one study investigated the interrater reliability on ankle and from the local community at Zollikerberg, Zurich (n = 21).
strength [13]. These few studies on isokinetic tests in old- The inclusion criteria were a minimum age of 65 years. Exclusion
er adults reveal variable intraclass correlation coefficients criteria were cardiovascular diseases (a myocardial infarction in
(ICCs). The ICCs for highest peak torque, average peak the previous 6 months, untreated ischemic heart disease and hy-
torque, power and work found by Symons et al. [9, 12], pertension), neuromuscular and musculoskeletal limitations in
the lower back and extremities (e.g. knee and foot osteoarthritis)
ranged from 0.84 to 0.93. Capranica et al. [11] reported which might impair the ability to maximally contract the muscles
values that ranged from 0.29 to 0.80. The results of abso- of the leg. There were no selection criteria for habitual physical
lute reliabilities, as expressed by the coefficient of varia- activity. All elderly participating in the study provided informed
tion (CV) and the ratio of limits of agreement (RLOA), consent. The local ethics committee approved of the study.
reveal suboptimal values. CVs ranged from 7 to 21% as
Instruments
reported by Symons et al. [9, 12]. Ordway et al. [8] and A Biodex System 3 isokinetic dynamometer with a sampling
Symons et al. [9, 12] described RLOAs that ranged from rate of 100 Hz (Biodex Medical Systems, Shirley, N.Y., USA) was
21 to 43% and 21 to 57%, respectively. used for all isokinetic strength testing. The mechanical reliability
It can be hypothesized that the isokinetic test proto- of this dynamometer has been shown to be excellent [18]. For all
cols used are, at present, maladapted to the motor perfor- testing trials, the Biodex System 3 Advantage software (version
3.2) calculated the highest peak torque, the average peak torque
mance abilities of the elderly. Connelly et al. [14] showed and average power for each movement and speed combination.
that for older people effects of learning the (isokinetic) The Biodex was calibrated in accordance with the manufacturer’s
motor task should be accounted for in test protocols. specifications before the start of each test session.
From this viewpoint, it seems clear that the use of a sin-
gle-session test protocol, as often used for younger sub- Design of the Total Assessment Protocol
jects, should not be recommended for aged persons [12] To minimize the learning effect, the total assessment protocol
started with a familiarization session for isokinetic muscle test-
and the learning process of the elderly should be consid- ing, followed by the two test sessions (fig. 1). The two test sessions
ered in test protocols for elderly. were performed at approximately the same time of day, and were
Researchers and clinicians are interested in the relative separated by 5–10 days, under the assumption that the isokinetic
and absolute reliability of a measurement instrument, as strength of the subjects does not change over this time. Each sub-
this affects the interpretation of results of an intervention ject was assessed three times (t1, t2, t3) in the two test sessions,
twice in the first (t1, t2) and once in the second test session (t3).
program. Furthermore, an important feature of clinical The interrater reliability was established in the first test session
measurements is that they are performed in individuals (t1, t2) with 1-hour rest between the assessments by two different
and not only in groups. Therefore, clinicians are con- raters. The intrarater reliability was calculated from one of two
cerned about the reliability of individual measurement re- measurements of the first test session (randomly selected by flip-
sults [15]. It has been suggested that the CV or the stan- ping a coin: t1 or t2) and from the single measurement in the sec-
ond test session (t3), where assessment was performed by one rat-
dard error of measurement (SEM), depending on whether er only.
heteroscedasticity is present or not, can be used to indi- For reasons of time, isokinetic strength testing was performed
cate the limit for the smallest change that indicates a real on the dominant lower leg only. In general, leg dominance is de-
improvement for groups of subjects, whereas for a single fined as the leg used to manipulate an object [19] and, therefore,
person the RLOA or the limits of agreements (LOA) indi- leg dominance was determined in this study according to leg pref-
erence while kicking a ball. We maintained the order of isokinet-
cate a real change [16, 17]. If the changes are less than the ic strength testing for all participants as follows: concentric knee
SEM or CV and RLOA or LOA, this reflects measurement contractions at 60°/s were performed first, followed by concentric
noise and is most likely meaningless. knee contractions at 120°/s, and finally concentric ankle contrac-
202.28.191.34 - 3/17/2015 7:48:20 AM
260 Gerontology 2009;55:259–268 Hartmann /Knols /Murer /de Bruin

Downloaded by:
Familiarization
session Test session 1 Test session 2
3–7 days 1h 5–10 days
break break break
Fig. 1. Timetable of the total assessment t1 t2 t3
protocol – one familiarization session, R1 or R2 R1 R2 R1 or R2
three tests (t1, t2, t3) and two raters (R1,
R2).
Color version available online

tions at 60°/s. These velocities were chosen because most daily
activities are, above all, related to the ability to generate power at
low velocities [4]. The positioning was recorded in the familiar-
ization session to ensure consistency of participant setup for the
two test sessions.
Each participant was seated in the Biodex chair in an upright
position with the back of the seat tilted at an angle of 85°. Stabili-
zation was provided by two shoulder straps that crossed the par-
ticipant’s chest, a waist strap, and a thigh strap. The lateral femo-
ral epicondyle was aligned with the axis of rotation of the dyna-
mometer. The length of the attachment was adjusted to ensure
that the ankle pad rested comfortably above the lateral and me-
dial malleoli. Range of motion (ROM) was determined individu-
ally by each subject. Prior to the knee joint measurements, the
dynamometer performs gravity compensation at an approximate
angle of 30°, while the effect of gravity on the torques is measured
with the subject completely relaxed.
On the testing days, the subjects completed a standardized 5-
min period of warm-up. All subjects performed 6–8 passive rep-
etitions followed by 3 submaximal contractions at each speed. All
tests started with knee flexion followed by knee extension. The
subjects performed 4 continuous maximal voluntary flexion-ex-
tension contractions at each speed with 1 min of rest between the
2 speeds. The first knee flexion and the last knee extension were
not recorded. This procedure was chosen to ensure that the test
conditions of all recorded contractions would be equal. The sub-
jects were instructed to push and pull ‘as fast and hard as possible’
through the full available ROM at every trial. All raters gave stan-
dardized instructions and verbal encouragement.
The subjects rested on the same dominant side while the Bio- Fig. 2. Subject performing an isokinetic ankle test with the knee
dex was configured for the ankle protocol. The subjects were po- in 30° of flexion and with the back of the seat tilted at an angle of
sitioned semi-reclined for ankle testing with the knee in 30° of 40°.
flexion and with the back of the seat tilted at an angle of 40°
(fig. 2). A pad was fixed under the thigh for stabilization. The
flexed-knee protocol allows full ROM at the ankle and prevents
the hip and knee muscles from contributing. Stabilization was Statistical Analyses
provided by a waist strap and a thigh strap. The subject’s foot was The highest peak torque, the average peak torque and the aver-
positioned so that the axis of rotation of the ankle was aligned age power were recorded for analysis for each movement and
with the axis of rotation of the dynamometer. The foot was se- speed combination. Means and standard deviations (SDs) were
cured to the footplate with two straps, one strap just distal to the calculated for each test by the raters (at t1, t2, t3). Bland-Altman
ankle and the second over the metatarsal bones. The test proce- plots were generated to provide a visual representation of het-
dure of the ankle was identical to the procedure of the knee, how- eroscedasticity, systematic bias, random error and LOA by plot-
ever, without gravity compensation. The subjects performed 4 ting the individual subject differences between the two tests
maximal voluntary contractions after the warm-up procedure against the individual mean of the two tests [17]. The normal dis-
starting with dorsiflexion; the first dorsiflexion and the last plan- tribution of the data was tested by the Kolmogorov-Smirnov test
tar flexion were not recorded. [20]. Statistically significant differences between the two raters
and the two test sessions (same rater) were calculated by means of
202.28.191.34 - 3/17/2015 7:48:20 AM
Reproducibility of an Isokinetic Gerontology 2009;55:259–268 261

Strength-Testing Protocol
Downloaded by:
Table 1. Subjects’ demographics (mean 8 SD) Intrarater Reliability
Table 2 presents means, SDs and p values with regard
Gender Subjects Age, years Body mass index to intrarater reliability. There were no statistically sig-
Women 18 69.985.0 25.0 8 4.4 nificant differences between the values of day 1 and day
Men 6 75.085.6 25.2 8 3.1 2, except for highest peak torque (p = 0.001), average peak
Total 24 71.285.5 25.0 8 4.0 torque (p = 0.019) and average power (p = 0.026) for knee
flexion measurements at 120°/s. The relative and absolute
intrarater reliabilities are shown in table 3. ICCs varied
from 0.85 to 0.98, SEMs from 8.0 to 14.9%, systematic bi-
a paired t test. Results were considered significant at p ! 0.05. The ases from –0.1 to 3.2 Nm for peak torques and from –0.2
levels of inter- and intrarater reliability were expressed as ICC, to 3.7 W for power, random errors from 2.3 to 19.7 Nm
SEM, associated 95% confidence intervals (CIs) of ICC and SEM, for peak torques and from 1.6 to 21.3 W for power and
systematic bias, random error, LOA and RLOA [21]. The smallest
detectable difference (SDD), which also indicates the limits for the
RLOAs from 22.1 to 41.4%. All average peak torque and
real change for a single person, is equivalent to the LOA statistic average power measures had an ICC greater than 0.90,
[17]. For this reason, we do not use the expression SDD. Two-way representing ‘very good’ reliability. The ICCs for highest
random effects models (ICC2,1 and ICC2,3) and two-way mixed peak torque ranged from 0.85 to 0.96.
models (ICC3,1 and ICC3,3) were used for relative inter- and intra-
rater reliabilities, respectively. Single-measure reliabilities (ICC2,1
and ICC3,1) were used for highest peak torque, whereas average-
Interrater Reliability
measure reliabilities (ICC2,3 and ICC3,3) were used for average Table 4 presents means, SDs, and p values from the
peak torque and average power [21]. The SEM was calculated as first test session (t1, t2). Significantly higher values in
the square root of the mean square of error (residual mean square) the second test were recorded for the average power in
between the measurements from the ANOVA results: SEM = 冪(re- knee extension measurements at 60°/s (p = 0.035) and for
sidual mean square) [17, 21]. The 95% CIs of SEM were detected
as 95% CI = 81.96 ! SEM [17]. LOA were detected as LOA = (av- all values on ankle plantar flexion at 60°/s (p values
erage difference between the two tests) 81.96 ! SD [20, 21]. The ranged from 0.005 to 0.042). The relative and absolute
systematic bias represents the average difference between the two interrater reliabilities are shown in table 5. ICCs varied
tests, and random error represents the difference between the from 0.81 to 0.99, SEMs from 6.3 to 17.2%, systematic
LOA and the systematic bias. All statistical analyses were per- biases from –0.8 to 5.3 Nm for peak torques and from
formed using the SPSS 15.0 and Microsoft Excel.
0.1 to 4.7 W for power, random errors from 1.7 to 19.8
Nm for peak torques and from 1.4 to 21.4 W for power,
and RLOAs from 17.6 to 47.7%. The relative and absolute
Results reliability for ankle plantar flexion tended to be less
good (0.81–0.88) than that of the other measurements
Subjects (0.89–0.99).
All subjects completed the study and no subject re-
ported any discomfort throughout the total assessment.
A summary description of the demographic variables of Discussion
the subjects is presented in table 1. The activity levels of
the subjects ranged from sedentary to regular exercisers. This study was performed to establish the relative and
None of the subjects was familiar with isokinetic dyna- absolute reliabilities of an isokinetic strength-testing pro-
mometry. tocol around the knee and ankle joint preceded by a fa-
miliarization session in an elderly population. The pres-
Reliability ent study showed ‘very good’ relative reliabilities for knee
All power measurements with regard to intrarater reli- extension and flexion at 60°/s and 120°/s and for ankle
ability are illustrated in figure 3 by way of Bland-Altman dorsiflexion at 60°/s, and ‘good’ reliability for ankle plan-
plots. There were similar findings for plots of highest tar flexion at 60°/s. The results of absolute reliabilities
peak torque, average peak torque and with regard to in- were partly satisfactory. The SEMs were moderate for an-
terrater reliability (results not shown). All data were nor- kle plantar flexion and low for all other contractions, and
mally distributed and showed no heteroscedasticity. The the LOAs were moderate to high.
systematic bias, random error and the LOA are presented Only few studies have collected data in older adults us-
on the plots. ing an isokinetic dynamometer with the same contrac-
202.28.191.34 - 3/17/2015 7:48:20 AM

Downloaded by:
Knee flexion at 60º/s Knee flexion at 120º/s
30
Difference between t1 and t2 (W)

15
10 20
5 10
0 0
−5 −10
−10 −20
−15
−30
10 15 20 25 30 35 40 10 20 30 40 50 60
Mean between t1 and t2 (W) Mean between t1 and t2 (W)
Knee extension at 60º/s Knee extension at 120º/s

40

20
10 20
0 0
−10 −20
−20
−40
30 40 50 60 70 80 90 40 60 80 100 120
Ankle dorsiflexion at 60º/s Ankle plantar flexion at 60º/s

3
15
2
10
1 5
0 0
−1 −5
−10
−2
−15
−3
2 4 6 8 10 10 20 30 40
Fig. 3. Individual subject differences between day 2 and day 1 are fies systematic bias; dashed lines signify LOAs (81.96 SD); ran-
plotted against each individual’s mean for the two tests for the dom error represents difference between limits of agreements and
average power with regard to intrarater reliability. Solid line signi- systematic bias.
202.28.191.34 - 3/17/2015 7:48:20 AM

Downloaded by:
Table 2. Means 8 SD and p values with regard to intrarater reliability
Day 1 Day 2 p value
Knee flexion at 60°/s

Highesta Peak torque, Nm 40.4812.5 42.3814.6 0.217
Averageb Peak torque, Nm 38.0811.9 40.4814.0 0.116
Power, W 25.688.7 27.289.4 0.075
Highesta Peak torque, Nm 33.7811.0 36.6812.7 0.010*
Averageb Peak torque, Nm 32.1810.7 35.0812.6 0.019*
Power, W 36.8814.6 40.4816.8 0.026*
Knee extension at 60°/s
Highesta Peak Torque, Nm 88.9822.0 89.7822.4 0.690
Averageb Peak Torque, Nm 84.7822.6 85.6822.3 0.657
Power, W 55.5814.3 58.5814.1 0.067
Power, W 81.0822.7 84.7824.8 0.116
Dorsiflexion at 60°/s
Power, W 5.382.7 5.382.5 0.736
Plantar flexion at 60°/s
Power, W 21.588.9 23.188.9 0.129
* p < 0.05. a Highest value of 3 repetitions. b Average of 3 repetitions.
tion velocities as in the present study. The magnitudes of [17] and data about absolute reliabilities of a test are im-
the highest peak torque and average power values from portant for clinical use.
this study are comparable with what has been previously The ICCs (0.81–0.99) that we found were similar to
published [8, 10, 13, 22]. Due to the small number of sub- values (0.84–0.92) found by Symons et al. [9, 12] for con-
jects included in our study, it is uncertain whether the centric knee extension in older, healthy subjects (mean
subjects in the present study are truly representative of age: 72 8 5 years). Capranica et al. [11] found clearly low-
the older adult population as a whole. However, we strived er ICCs (0.23–0.85) for knee extension and flexion in old-
to obtain a sample that is as representative as possible by er, independent women (mean age: 68 8 5 years). Fur-
formulating as few exclusion criteria as possible. The ac- thermore, the present values were similar to those (0.93–
tivity level of the elderly participants in our study varied 0.95) for knee extension and flexion for a young adult
from sedentary to regular exerciser behavior that repre- population (mean age: 20 8 1 years) found by Sole et al.
sented a heterogeneous group and had a wide range of [23]. There were no data concerning SEMs of isokinetic
isokinetic strength values. The relative reliability is the tests in a comparable population. Sole et al. [23] found
degree to which individuals maintain their test results in only slightly lower SEMs (6–10%) in the young adult pop-
a sample with repeated measurements and is affected by ulation, and Flansbjer et al. [16] found similar SEMs (9–
sample heterogeneity, which means: the more heteroge- 12%) for concentric knee extension in a group of stroke
neous a sample is, the higher the relative reliability be- subjects (mean age: 58 8 6 years). The other studies in-
comes. Therefore, a high correlation may still mean un- volving reliability of isokinetic dynamometers in older
acceptable measurement error for some analytical goals adults calculated CVs, and therefore they are not directly
202.28.191.34 - 3/17/2015 7:48:20 AM

Downloaded by:
Table 3. Intrarater reliability of isokinetic knee and ankle measurements (ICC; 95% CI; SEM; SEM as percentage of group average;
systematic bias; random error; LOA lower boundary, LOALB; LOA upper boundary, LOAUB; ROLA)
ICCa 95% CI SEM SEM 95% CI System- Random LOALB LOAUB RLOA
for ICC % for SEM atic bias error %

Highestb Peak torque, Nm 0.85 0.68–0.93 5.3 12.8 810.4 2.0 14.6 –12.7 16.6 35.4
Averagec Peak torque, Nm 0.92 0.80–0.97 5.0 12.7 89.7 2.4 13.8 –11.4 16.2 35.2
Power, W 0.94 0.85–0.97 2.9 11.1 85.8 1.6 8.1 –6.5 9.7 30.8
Power, W 0.93 0.82–0.97 5.2 13.5 810.2 3.7 14.4 –10.8 18.1 37.3
Power, W 0.92 0.81–0.97 5.2 9.1 810.1 2.9 14.3 –11.4 17.2 25.1
Power, W 0.94 0.86–0.98 7.7 9.3 815.1 3.7 21.3 –17.6 25.0 25.7
Highestb Peak torque, Nm 0.96 0.91–0.98 0.9 8.4 81.7 –0.2 2.4 –2.6 2.2 23.4
Averagec Peak torque, Nm 0.98 0.96–0.99 0.8 8.7 81.6 –0.1 2.3 –2.3 2.2 24.0
Power, W 0.98 0.95–0.99 0.6 10.6 81.1 –0.1 1.6 –1.6 1.5 29.3
Power, W 0.92 0.81–0.97 3.3 14.9 86.5 1.5 9.2 –7.7 10.8 41.4
a ICC3,1 for highest peak torque; ICC3,3 for average peak torque and average power.
b
Highest value of 3 repetitions.
c Average of 3 repetitions.
comparable to the SEMs. The CVs ranged from 7 to 20% al. [23] and Flansbjer et al. [16]. The results of our study
[9] and 8–13% [12] found by Symons et al. [9, 12]. In the are higher compared to the younger adult populations
present study, the RLOAs ranged from 18 to 35%, except (12–23%) reported by Sole et al. [23] and slightly lower
for the ankle plantar flexion (39–48%). Ordway et al. [8] compared to the group of stroke subjects (26–55%) in the
found similar systematic biases (1–4 Nm), random errors study by Flansbjer et al. [16].
(7–13 Nm) and RLOAs (21–43%) for average peak torque In this study, the SEMs were moderate for ankle plan-
for knee extension and flexion and for ankle plantar and tar flexion and low for the other contractions. Compared
dorsiflexion in 33 older adults (mean age: 72 8 6 years). with the other studies, it can be concluded that our iso-
Symons et al. [9, 12] found similar RLOAs for highest and kinetic strength-testing protocol can be used to detect
average peak torque (21–33%) in two studies for concen- real changes in an older adult group after, e.g., a training
tric knee flexion, but higher results for average power intervention. However, training studies with elderly
(42–52%). Because these RLOAs were larger than expect- where this protocol is used should substantiate this as-
ed in these two studies, the authors do not recommend sumption. Furthermore, the testing protocol used showed
the use of a single-session test protocol in elderly indi- moderate to high RLOAs, and therefore it can be expect-
viduals. Because SDDs and LOAs are equivalent [17], it is ed that it is not sufficiently sensitive to detect real chang-
possible to compare our results with the results of Sole et es in older, single subjects. However, there are only lim-
202.28.191.34 - 3/17/2015 7:48:20 AM

Downloaded by:
Table 4. Means 8 SD and p values with regard to interrater reliability
Test 1 Test 2 p value

Power, W 27.088.4 27.5838.8 0.727
Power, W 38.0814.9 40.4816.1 0.050
Power, W 57.2814.7 60.2814.2 0.035*
Power, W 83.2825.1 87.9824.0 0.050
Power, W 5.282.6 5.382.5 0.456
Highesta Peak torque, Nm 42.0816.3 47.2818.0 0.016*
Averageb Peak torque, Nm 39.1815.2 43.7816.9 0.042*
Power, W 22.189.0 25.8810.4 0.005*
* p < 0.05. a Highest value of 3 repetitions. b Average of 3 repetitions.
ited amounts of training studies with elderly where pow- ficult to perform than against gravity. These significant-
er has been used as an outcome measure; therefore, we do ly increased means may indicate a learning effect that still
not know how big a change might be expected based on takes place in spite of the familiarization session. Ploutz-
muscle training interventions for the elderly. Snyder and Giamis [24] compared the number of testing
It has been speculated that normalizing the highest sessions required to achieve consistent 1 repetition max-
peak torque with body weight may decrease the absolute imum strength measurements in untrained older and
reliability. However, the results showed negligible de- younger women. The older subjects required significant-
creased values of SEMs and RLOAs (data not shown; ly more familiarization testing sessions (8–9 sessions)
available upon request). compared with the younger subjects (3–4 sessions) to
The ICCs of the two average variables (ICCs range achieve the same absolute consistency of measurements.
from 0.88 to 0.99) were slightly higher than the ICCs for However, multiple sessions may not be practical for older
the maximal variable (ICCs range from 0.81 to 0.98). This adults. It may be speculated that the period of warm-up
pattern was not reflected in the absolute reliability. Be- in our protocol should be modified to minimize the sys-
cause of the small differences in reliability, we recom- tematic bias, while including more submaximal contrac-
mend to rather use an average value from a set of 3 rep- tions and one or two maximal contractions.
etitions instead of the highest score. The LOAs that include the systematic biases and the
With few exceptions, paired t tests revealed small sys- random errors together were moderate to high and there-
tematic biases between the two tests (table 2 and 4). Sub- fore not always satisfactory. Because the systematic biases
jects often found the movement toward gravity more dif- were small in the majority of cases, we conclude that the
202.28.191.34 - 3/17/2015 7:48:20 AM

Downloaded by:
Table 5. Interrater reliability of isokinetic knee and ankle measurements (ICC; 95% CI; SEM; SEM as percentage of group average;
systematic bias; random error; LOA lower boundary, LOALB; LOA upper boundary, LOAUB; RLOA)
ICCa 95% CI SEM SEM 95% CI System- Random LOALB LOAUB RLOA
for ICC % for SEM atic bias error %

Highestb Peak torque, Nm 0.92 0.83–0.97 3.6 8.5 87.1 –0.8 10.1 –10.8 9.3 23.5
Power, W 0.95 0.89–0.98 2.7 9.8 85.2 0.3 7.4 –7.1 7.7 27.3
Power, W 0.96 0.90–0.98 4.0 10.1 87.8 2.4 11.0 –8.6 13.4 28.1
Power, W 0.94 0.85–0.98 4.5 7.6 88.8 3.0 12.4 –9.4 15.4 21.2
Power, W 0.94 0.85–0.98 7.7 9.0 815.1 4.7 21.4 –16.6 26.0 24.9
Power, W 0.98 0.96–0.99 0.5 9.2 81.0 0.1 1.4 –1.2 1.5 25.6
Power, W 0.88 0.61–0.95 4.0 16.8 87.9 3.7 11.2 –7.5 14.9 46.6
a ICC2,1 for highest peak torque; ICC2,3 for average peak torque and average power.
b
Highest value of 3 repetitions.
c
Average of 3 repetitions.
random errors were large. The random error can be af- ject-linked variability is higher than rater-linked vari-
fected by instrument, data processing, rater and subject- ability. Particularly the fluctuations in daily condition
linked variability, test procedure, and protocol errors and/or motivation may strongly affect the results of
[25]. The mechanical reliability of the Biodex System 3 strength measurements in older adults. An average of
isokinetic dynamometer has been shown to be excellent, multiple test sessions may minimize these factors, where-
with ICCs of 0.99 for torque, position and velocity [18]. as in clinical practice the number of test sessions is nor-
To minimize test procedure and protocol-linked errors, mally restricted.
the test procedure was standardized, the subject position- A limitation of this study is the questionable relation
ing was recorded and the two raters gave standardized between test specificity of strength and power measure-
instructions and verbal encouragement. The subject- ments together with the transferability on everyday func-
linked variability is affected by different factors, such as tional requirements. Although we know from a cross-
an effect of learning or fatigue, fluctuations in daily con- sectional study design applied on very old people [6] that
dition and/or motivation. We tried to reduce the learning there is a close correlation between leg extensor power
effects by including a familiarization session. Because the and chair-rising, stair climbing and walking speeds, and
interrater reliability showed slightly better results com- stair-climbing power, we cannot state that this is also a
pared to the intrarater reliability, we conclude that sub- causal relation. Until some prospective studies are per-
202.28.191.34 - 3/17/2015 7:48:20 AM

Downloaded by:
formed that indicate such a causality between isokinetic on a group level than on an individual level. The Biodex
measures of power with functional everyday activities, System 3 may be employed in studies to determine the
we will be mainly restricted to assessing the effects of re- effect of exercise intervention programs on physical ac-
sistance training on isokinetic strength testing. In these tivity.
cases, the form of resistance training should resemble the
isokinetic test procedure.
Acknowledgements
Conclusions The authors wish to thank all the subjects who participated in
the study and the student apprentices who collected the data.
The results of this study demonstrate that the Biodex They also wish to thank the Stiftung Diakoniewerk Neumün-
ster – Schweizerische Pflegerinnenschule for their financial sup-
System 3 is a reliable device when used on independently port, and we gratefully acknowledge the support of PROXOMED쏐
living elderly. The ability of the device to determine a real for providing the test material for this study. We also thank Le-
change in isokinetic ankle and knee contractions is better anne Pobjoy for help in preparing the manuscript.
References
1 Skelton DA, Young A, Greig CA, Malbut KE: 9 Symons TB, Vandervoort AA, Rice CL, 17 Atkinson G, Nevill AM: Statistical methods
Effects of resistance training on strength, Overend TJ, Marsh GD: Reliability of iso- for assessing measurement error (reliability)
power, and selected functional abilities of kinetic and isometric knee-extensor force in in variables relevant to sports medicine.
women aged 75 and older. J Am Geriatr Soc older women. J Aging Phys Act 2004;12:525– Sports Med 1998;26:217–238.
1995;43:1081–1087. 537. 18 Drouin JM, Valovich-McLeod TC, Shultz SJ,
2 Skelton DA, Kennedy J, Rutherford OM: Ex- 10 Frontera WR, Hughes VA, Dallal GE, Evans Gansneder BM, Perrin DH: Reliability and
plosive power and asymmetry in leg muscle WJ: Reliability of isokinetic muscle strength validity of the biodex system 3 proisokinetic
function in frequent fallers and non-fallers testing in 45- to 78-year-old men and wom- dynamometer velocity, torque and position
aged over 65. Age Ageing 2002; 31:119–125. en. Arch Phys Med Rehabil 1993; 74: 1181– measurements. Eur J Appl Physiol 2004; 91:
3 Skelton DA, Greig CA, Davies JM, Young A: 1185. 22–29.
Strength, power and related functional abil- 11 Capranica L, Battenti M, Demarie S, Figura 19 Sadeghi H, Allard P, Prince F, Labelle H:
ity of healthy people aged 65–89 years. Age F: Reliability of isokinetic knee extension Symmetry and limb dominance in able-bod-
Ageing 1994;23:371–377. and flexion strength testing in elderly wom- ied gait: a review. Gait Posture 2000; 12: 34–
4 Ferri A, Scaglioni G, Pousson M, Capodaglio en. J Sports Med Phys Fitness 1998; 38: 169– 45.
P, Van Hoecke J, Narici MV: Strength and 176. 20 Bland JM, Altman DG: Measuring agree-
power changes of the human plantar flexors 12 Symons TB, Vandervoort AA, Rice CL, ment in method comparison studies. Stat
and knee extensors in response to resistance Overend TJ, Marsh GD: Reliability of a sin- Methods Med Res 1999;8:135–160.
training in old age. Acta Physiol Scand 2003; gle-session isokinetic and isometric strength 21 Portney LG, Watkins MP: Foundations of
177:69–78. measurement protocol in older men. J Geron- Clinical Research: Applications to Practice.
5 Metter EJ, Talbot LA, Schrager M, Conwit tol A Biol Sci Med Sci 2005; 60:114–119. New Jersey, Prentice Hall, 2000.
RA: Arm-cranking muscle power and arm 13 Morris-Chatta R, Buchner DM, de Lateur BJ, 22 Signorile JF, Carmel MP, Czaja SJ, Asfour SS,
isometric muscle strength are independent Cress ME, Wagner EH: Isokinetic testing of Morgan RO, Khalil TM, Ma F, Roos BA: Dif-
predictors of all-cause mortality in men. J ankle strength in older adults: assessment of ferential increases in average isokinetic pow-
Appl Physiol 2004;96:814–821. inter-rater reliability and stability of strength er by specific muscle groups of older women
6 Bassey EJ, Fiatarone MA, O’Neill EF, Kelly over six months. Arch Phys Med Rehabil due to variations in training and testing. J
M, Evans WJ, Lipsitz LA: Leg extensor pow- 1994;75:1213–1216. Gerontol A Biol Sci Med Sci 2002; 57:M683–
er and functional performance in very old 14 Connelly DM, Carnahan H, Vandervoort M690.
men and women. Clin Sci (Lond) 1992; 82: AA: Motor skill learning of concentric and 23 Sole G, Hamren J, Milosavljevic S, Nicholson
321–327. eccentric isokinetic movements in older H, Sullivan SJ: Test-retest reliability of iso-
7 Lexell J, Taylor CC, Sjostrom M: What is the adults. Exp Aging Res 2000;26:209–228. kinetic knee extension and flexion. Arch
cause of the ageing atrophy? Total number, 15 Roebroeck ME, Harlaar J, Lankhorst GJ: The Phys Med Rehabil 2007;88:626–631.
size and proportion of different fiber types application of generalizability theory to reli- 24 Ploutz-Snyder LL, Giamis EL: Orientation
studied in whole vastus lateralis muscle from ability assessment: an illustration using iso- and familiarization to arm strength testing
15- to 83-year-old men. J Neurol Sci 1988;84: metric force measurements. Phys Ther 1993; in old and young women. J Strength Cond
275–294. 73:386–395; discussion 396–401. Res 2001;15:519–523.
8 Ordway NR, Hand N, Briggs G, Ploutz- 16 Flansbjer UB, Holmback AM, Downham D, 25 Dvir Z: Isokinetics: Muscle Testing, Inter-
Snyder LL: Reliability of knee and ankle Lexell J: What change in isokinetic knee pretation and Clinical Applications. Edin-
strength measures in an older adult popula- muscle strength can be detected in men and burgh, Churchill Livingstone, 2004.
tion. J Strength Cond Res 2006;20:82–87. women with hemiparesis after stroke? Clin
Rehabil 2005;19:514–522.
202.28.191.34 - 3/17/2015 7:48:20 AM

Downloaded by:

Hartmann 2009

Uploaded by

Copyright:

Available Formats

Hartmann 2009

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Hartmann 2009

Uploaded by

Copyright:

Available Formats

Clinical Section

Gerontology 2009;55:259–268 Received: November 9, 2007

© 2008 S. Karger AG, Basel Antonia Hartmann

www.karger.com www.karger.com/ger E-Mail antonia.hartmann@move.biol.ethz.ch

260 Gerontology 2009;55:259–268 Hartmann /Knols /Murer /de Bruin

Color version available online

Reproducibility of an Isokinetic Gerontology 2009;55:259–268 261

262 Gerontology 2009;55:259–268 Hartmann /Knols /Murer /de Bruin

Difference between t1 and t2 (W)

Knee extension at 60º/s Knee extension at 120º/s

Difference between t1 and t2 (W)

Ankle dorsiflexion at 60º/s Ankle plantar flexion at 60º/s

Reproducibility of an Isokinetic Gerontology 2009;55:259–268 263

Day 1 Day 2 p value

Knee flexion at 60°/s

* p < 0.05. a Highest value of 3 repetitions. b Average of 3 repetitions.

264 Gerontology 2009;55:259–268 Hartmann /Knols /Murer /de Bruin

Knee flexion at 60°/s

Reproducibility of an Isokinetic Gerontology 2009;55:259–268 265

Test 1 Test 2 p value

Knee flexion at 60°/s

* p < 0.05. a Highest value of 3 repetitions. b Average of 3 repetitions.

266 Gerontology 2009;55:259–268 Hartmann /Knols /Murer /de Bruin

Knee flexion at 60°/s

Reproducibility of an Isokinetic Gerontology 2009;55:259–268 267

268 Gerontology 2009;55:259–268 Hartmann /Knols /Murer /de Bruin

You might also like