LIRN PAPER

Reliability of Ankle Isometric, Isotonic, and Isokinetic Strength and Power Testing in Older Women Sandra C. Webber, Michelle M. Porter

Background. Ankle strength (force-generating capacity) and power (work pro- duced per unit of time or product of strength and speed) capabilities influence physical function (eg, walking, balance) in older adults. Although strength and power parameters frequently are measured with dynamometers, few studies have examined the reliability of measurements of different types of contractions.

Objective. The purpose of this study was to examine relative and absolute intrarater reliability of isometric, isotonic, and isokinetic ankle measures in older women.

Design. This was a prospective, descriptive methodological study.

Methods. The following dorsiflexion (DF) and plantar-flexion (PF) measures were assessed twice (7 days apart) by the same examiner in 30 older women (mean age�73.3 years, SD�4.7): isometric peak torque and rate of torque development (RTD), isotonic peak velocity, average acceleration and peak power, and isokinetic peak torque and peak power (30°/s and 90°/s). Several statistical methods were used to examine relative and absolute reliability.

Results. Intraclass correlation coefficients (ICCs) for the DF tests (ICC�.76 –.97) were generally higher than ICCs for matched PF tests (ICC�.58 –.93). Measures of absolute reliability (eg, coefficient of variation of the typical error [CVTE]) also demonstrated more reliable values for DF tests (5%–18%) compared with PF tests (7%–37%). Isotonic peak velocity tests at minimal loads were associated with the lowest CVTE and ratio limits of agreement values for both DF (5% and 14%, respec- tively) and PF (7% and 18%, respectively). Isometric RTD variables were the least reliable (CVTE�16%–37%).

Limitations. This study was limited to a relatively homogeneous sample of older women.

Conclusions. Test-retest reliability was adequate for determining changes at the group level for all strength and power variables except isometric RTD. Minimal detectable change scores were determined to assist clinicians in assessing meaningful change over time in ankle strength and power measurements within individuals.

S.C. Webber, MSc, BMR(PT), is a PhD candidate, Department of Physiology, Faculty of Medicine, University of Manitoba, Winnipeg, Manitoba, Canada.

M.M. Porter, BPHE, MSc, PhD, is Professor, Faculty of Kinesiology and Recreation Management and the Department of Physiology, University of Manitoba, 207 Max Bell Centre, Winnipeg, Manitoba, Canada R3T 2N2. Address all cor- respondence to Dr Porter at: [email protected].

[Webber SC, Porter MM. Reliabil- ity of ankle isometric, isotonic, and isokinetic strength and power testing in older women. Phys Ther. 2010;90:1165–1175.]

© 2010 American Physical Therapy Association

Research Report

Post a Rapid Response to this article at: ptjournal.apta.org

August 2010 Volume 90 Number 8 Physical Therapy f 1165

D ow

nloaded from https://academ

ic.oup.com /ptj/article/90/8/1165/2738033 by guest on 14 S

eptem ber 2020

Loss of neuromuscular mass,strength (force-generating capac-ity), and power (work produced per unit of time or product of strength and speed) are closely asso- ciated with functional decline, loss of independence, and mortality in older adults.1 Some studies suggest that the rate of loss of neuromuscu- lar power exceeds the rate of loss of strength with age.2,3 Cross-sectional studies also have demonstrated that functional capabilities such as the ability to get up and down from a chair, climb stairs, and walk quickly may be more closely associated with power than strength4 – 6 and that loss of power may be more related to the etiology of falls.7 In order to deter- mine the true nature of the relation- ships among changes in strength and power with age and changes in func- tion, reliable testing techniques are required.

Isokinetic dynamometers frequently are used to assess neuromuscular function because they provide de- tailed torque, velocity, and position data with high mechanical reliabili- ty.8 Although researchers have in- vestigated the reliability of dyna- mometer strength assessment in older adults, the focus has largely been on the knee.9 –14 However, the distal leg muscles exhibit reductions in strength and power with ag- ing15,16 and are important for walk- ing,17,18 maintaining balance, avoid- ing falls,7,19 and braking a vehicle. Greater limitations in ankle strength (as opposed to knee strength) have been associated with falls in nurs-

ing home residents.20,21 Dorsiflexion (DF) power has been found to be closely associated with function in community-dwelling older women in terms of their ability to get up and down from a chair and climb stairs.4

Plantar-flexion (PF) strength has been shown to be positively related to both habitual gait speed and fast gait speed.4,18 Despite the important role that ankle function plays in mobility, there have been only a few evalua- tions of the reliability of ankle strength protocols11,12,22 and only one assessment concerned with reli- ability associated with measures of ankle power in this population.11

Because power is defined as work (force � distance) divided by time, it is influenced by both strength and speed. Peak or average power (watts�newton-meters � radians/s) can be measured using either the iso- kinetic or isotonic mode on a dyna- mometer. Other indirect measures that may be associated with the abil- ity of the neuromuscular system to generate force or torque rapidly can be evaluated using either the isomet- ric mode (rate of torque develop- ment [RTD]) or the isotonic mode (velocity, acceleration).2,16,23,24 Hart- mann et al11 reported reliability scores for ankle average isokinetic power tests, but reliability of iso- tonic and isometric measures related to power has not been investigated previously in older adults. Prelimi- nary findings suggest that isokinetic evaluations of power may not be as reliable as strength measures9,11 and that RTD averaged over a specified range (eg, from 30% to 60% of peak torque) may yield more consistent results compared with peak RTD.24

Further research is needed to com- pare reliability of strength and power measures and to determine which measures are associated with lower levels of measurement error for use in research and clinical situations.

Physical therapists and other health care providers need to be able to properly interpret measurement change to determine the relative ef- fectiveness of different interventions. Test-retest studies provide informa- tion about relative reliability, that is, the degree to which repeated mea- surements reveal consistent ranking of individuals’ scores within a group. Measures of absolute reliability de- scribe individual variability and mea- surement error and, therefore, are important in determining levels for clinically significant change. Although the literature suggests that power may be more important than strength in terms of function in older adults, very little is known about the reliability of different measures of power or power-related variables (eg, velocity during isotonic move- ments). The objective of this study was to determine the relative and absolute intrarater reliability for an- kle strength and power measure- ments obtained using isometric, iso- tonic, and isokinetic tests on a dynamometer in older women.

Method Participants Thirty older women (mean age� 73.3 years, SD�4.7) were recruited to take part in this study. This sam- ple size was consistent with those traditionally chosen for studying dynamometer measures in older adults.9 –12,14,25 A convenience sam- ple consisting of women who had expressed an interest or participated in previous research in our labora- tory was used. However, none of the women had been tested previously on an isokinetic dynamometer or participated in an exercise-related study. Exclusion criteria included acute or unstable chronic disease and neurological or musculoskeletal impairment that would interfere with testing. In addition to the 30 women who participated, another 22 women were approached to be involved in the study but either did not meet the in-

Available With This Article at ptjournal.apta.org

• Audio Abstracts Podcast

This article was published ahead of print on May 20, 2010, at ptjournal.apta.org.

Ankle Measures in Older Women

1166 f Physical Therapy Volume 90 Number 8 August 2010

D ow

nloaded from https://academ

ic.oup.com /ptj/article/90/8/1165/2738033 by guest on 14 S

eptem ber 2020

clusion and exclusion criteria or were not interested in participating. Participant characteristics are pre- sented in Table 1. A physician’s signed Physical Activity Readiness Medical Examination (PARmed-X) form was re- quired when potential contraindica- tions to exercise (eg, history of cardiac disease, uncontrolled hypertension, hernia, detached retina) were identi- fied. All testing took place in a univer- sity research laboratory over the sum- mer of 2008. At the initial evaluation session, participants provided their written informed consent.

The median number of comorbidi- ties reported by participants was 2 (range�0 –3). Arthritis (18), hyper- tension (14), previous cancer diag- nosis (5), and diabetes (4) were most commonly reported. Seven partici- pants reported that they had fallen in the past year, and 1 participant reg- ularly used a cane. The median num- ber of medications prescribed was 1 (range�0 –3). None of the partici- pants changed their medications dur- ing their testing week.

Procedure Upon admission to the study, partic- ipants completed a health/demo- graphic questionnaire. All other tests were conducted twice, by the same physical therapist, exactly 1 week apart, at the same time of day. Infor- mation on the isokinetic dynamome- ter setup from session 1 was used for session 2, but the examiner was blinded to the results from session 1 until after session 2 was conducted. Resting blood pressure, heart rate, body mass, and height were mea- sured using standard procedures. Ac- tive range of motion was measured for both DF and PF (right ankle) with the participant in a seated position with the knee supported in exten- sion. Two measurements were taken in each direction and then averaged to determine active range of motion. A universal goniometer was used for the measurements, with the axis of

the goniometer aligned with the lat- eral malleolus of the ankle, the prox- imal arm aligned with the head of the fibula, and the distal arm parallel to the lateral border of the fifth meta- tarsal. Ankle range of motion was measured to ensure that participants had adequate range of motion to tolerate the starting positions used for dynamometer testing. No partic- ipants were excluded because they lacked sufficient range of motion.

Dynamometer Tests Dorsiflexion and PF torque, position, and velocity were measured using a Biodex System 3 Pro dynamometer.* The mechanical reliability of this dy- namometer has been shown to be excellent.8 Calibration of the dyna- mometer was verified each day prior to testing. Participants warmed up by walking for 4 minutes on a tread- mill before being seated on the Bio- dex dynamometer (right lateral mal- leolus aligned with the axis of rotation, right knee flexed 45°–55°, trunk reclined 5° from vertical). Only the dominant leg (defined as the leg that would be used to kick a ball) was tested. All participants re- ported right leg dominance. Each participant kept her arms folded across her chest during testing, and belts provided stabilization around the waist and over the right thigh. The end limits of range of motion were set at 10 degrees of DF and 30 degrees of PF for all tests. Partici- pants performed isometric tests, concentric isotonic tests, and con- centric isokinetic tests, always in that order, with DF contractions pre- ceding PF contractions. Standard- ized, consistent verbal encourage- ment was provided for all tests using a script.

Isometric Tests Following 3 practice trials, 3 maxi- mal voluntary isometric contractions

were performed for DF (at 25° of PF) and then for PF (at 0°). Test angles were chosen based on previous liter- ature2,12,16,26 to correspond approxi- mately to the angles at which maxi- mum isometric torques can be produced. Participants were strongly encouraged to contract “fast” and to hold each contraction for 3 to 5 sec- onds. They were given 90 seconds of rest between trials.

Isotonic Tests The dynamometer then was switched to the isotonic mode. The DF and PF movements were each performed against 2 set resistance levels: (1) a minimal resistance level (DF�1 N�m, PF�15 N�m) and (2) a load equal to 50% of isometric peak torque. These resistance levels rep- resent the boundaries of those previ- ously published for the dorsiflexors.2

Isotonic DF trials were initiated from 30 degrees of PF, and PF trials were initiated from 10 degrees of DF. Again, participants were strongly en- couraged to move “fast.” Two prac- tice trials preceded 5 test trials for each of the 4 conditions (DF and PF, 2 loads each). Thirty seconds of rest was provided between all repetitions.

Isokinetic Tests Maximal-effort isokinetic concentric DF and PF tests were performed at 30°/s and 90°/s (in that order). These velocities were chosen be- cause they are within the range of velocities typically studied for mea-

* Biodex Medical Systems Inc, 20 Ramsey Rd, Shirley, NY 11967.

Table 1. Participant Characteristics

Characteristic Mean (SD)

Age (y) 73.3 (4.7)

Body mass (kg) 73.8 (11.9)

Height (cm) 159.9 (4.8)

Body mass index (kg/m2) 28.8 (4.1)

Active dorsiflexion range of motion (°)

11 (5)

Active plantar-flexion range of motion (°)

53 (6)

Ankle Measures in Older Women

August 2010 Volume 90 Number 8 Physical Therapy f 1167

D ow

nloaded from https://academ

ic.oup.com /ptj/article/90/8/1165/2738033 by guest on 14 S

eptem ber 2020

surements about the ankle in older adults.4,25,27,28 In addition, restricting the highest velocity to 90°/s allowed for constant velocity to be main- tained over approximately one third (12°) of the total 40-degree range of motion excursion (once acceleration and deceleration were accounted for). The passive mode on the dyna- mometer was used for these con- stant velocity tests because it pro- vided a passive return to the start position after each concentric con- traction; therefore, all concentric DF contractions were completed before PF testing began (ie, DF and PF con- tractions were not performed imme- diately back-to-back). Furthermore, many participants would not be able to generate enough DF torque to overcome the torque related to the combined mass of the foot and foot- plate to initiate DF movement. Matching concentric contractions with the onset of passive movement avoided this difficulty. Participants were given 3 to 5 submaximal prac-

tice trials for familiarization before 5 test trials were conducted for each movement, at each velocity. A 2-minute rest period was provided between velocities. One participant was unable to generate torque in the DF direction at the higher velocity; therefore, DF peak torque and peak power values were recorded at 90°/s for 29 participants.

Data Analysis Biodex data were collected at a fre- quency of 100 Hz and exported for analyses in SigmaPlot (version 11.0).† All variables used in the anal- yses were means of the repetitions performed (isometric measures� mean of 3 repetitions, isotonic and isokinetic measures�mean of 5 rep- etitions). Because all scores inher- ently include some random error (which either adds to or subtracts from the true score), using mean

scores may reduce the magnitude of the error component contributing to the total score.29

For each isometric contraction, peak torque (in newton-meters) was iden- tified and RTD (in newton-meters per second) was calculated by 2 methods. Change in newton-meters/ time was first determined from 0% to 50% of peak torque and then from 40% to 80% of peak torque (Fig. 1). Calculating RTD over a specified range has been shown to be more reliable than determining peak RTD.24 These specific ranges were chosen to allow comparison of RTD reliability between relatively steep sections of the isometric torque curve (0%–50% of peak torque) and less steep sections (40%– 80% of peak torque).

For each isotonic contraction, peak velocity (in degrees per second), average acceleration (peak velocity/ time to reach peak velocity, mea- sured in degrees per second squared), and peak power (watts� newton-meters � radians/s) were determined (Fig. 2). Although the dy- namometer was set to the isotonic mode for these tests, torque is not held absolutely constant throughout the range of motion on this setting (Fig. 2). As has been noted previous- ly,30,31 the sampling rate (100 Hz) of the dynamometer does not permit adjustments in speed to occur fast enough to result in a continuous torque level. For each isokinetic con- traction, peak torque (in newton- meters) and peak power (watts�N�m � radians/s) were analyzed.

Statistical analyses were conducted using SPSS (version 15.0)‡ and Sig- maPlot. Means and standard devia- tions were calculated for each vari- able tested at time 1 and time 2. Paired t tests were conducted to look

† Systat Software Inc, 1735 Technology Dr, Ste 430, San Jose, CA 95110.

‡ SPSS Inc, 233 S Wacker Dr, Chicago, IL 60606.

Figure 1. Torque recorded during one repetition of isometric dorsiflexion for one representative participant. Rate of torque development was calculated as the change in newton- meters/change in time from 0% to 50% of peak torque and from 40% to 80% of peak torque. Torque is designated as negative in the dorsiflexion direction.

Ankle Measures in Older Women

1168 f Physical Therapy Volume 90 Number 8 August 2010

D ow

nloaded from https://academ

ic.oup.com /ptj/article/90/8/1165/2738033 by guest on 14 S

eptem ber 2020

for significant bias between test ses- sions (P�.05). The intraclass corre- lation coefficient (ICC [2,k]) was used to evaluate both systematic and random errors that may affect rela- tive test-retest reliability.32 Specifi- cally, ICC [2,3] was used for all iso- metric measures because they were based on the mean of 3 repetitions, and ICC [2,5] was used for all iso- kinetic and isotonic measures that were scored as the mean of 5 repe- titions. Normality of the difference scores was assessed using the Shapiro-Wilk test. Data were checked visually with Bland-Altman plots for the presence of heterosce- dasticity, and Pearson correlation co- efficients were calculated between absolute differences and the means of the 2 tests.

Measures of absolute reliability were expressed using standard error of the measurement (SEM), coefficients of variation of the typical error (CVTE), limits of agreement (LOA), ratio limits of agreement (RLOA), and the minimal detectable change (MDC). Absolute reliability describes within-subject variation and the de- gree to which observed scores will vary with repeated measurements.33

Generally, CVTE and RLOA are used to describe heteroscedastic data, and SEM and LOA are used to describe homoscedastic data.34 The majority of the strength and power variables studied did not demonstrate het- eroscedasticity (greater measure- ment error when measured values were larger) and, therefore, could be adequately described using SEM and LOA.34 However, because a few vari- ables demonstrated a positive rela- tionship between the degree of mea- surement error and the magnitude of the measured value, CVTE and RLOA statistics also are included.

The SEM was determined as the square root of the residual mean square error term from the analysis of variance table.35 The SEM de-

scribes (in units of the actual mea- sure) the limits for change required to indicate a real increase or de- crease for a group of individuals fol- lowing some sort of intervention.25

Whereas SEM values express typical error in original units, CVTE ex- presses typical error as a percentage, making it useful for comparing reli- ability among different measures and across different studies. Typical error was calculated as the standard devi- ation of the differences scores be- tween sessions, divided by �2.33 Co- efficient of variation of the typical error is defined as typical error di- vided by the mean of all trials from both sessions, multiplied by 100.36

The LOA was calculated as the sys- tematic bias: (mean difference be- tween 2 test sessions) � the random error component (1.96 � standard deviation of the difference between the 2 test sessions), which is identi- cal to systematic bias � MDC95

34

(MDC95�1.96 � �2 � SEM 37,38).

The MDC95 values provide informa-

tion about the confidence limits as- sociated with measurement error so that, for example, it can be stated with 95% confidence that an individ- ual’s change score that exceeds the LOA represents a true change. The MDC95 values also were expressed as a percentage in order to allow for comparisons among measures and across studies (RLOA�MDC95/mean of all observations � 100).

Role of the Funding Source Ms Webber was supported by a Ca- nadian Institutes of Health Research, Institute of Aging fellowship.

Results Means and standard deviations for the isometric, isotonic, and isoki- netic strength and power variables are presented in Table 2. There were no significant differences between session 1 and session 2 for almost all of the variables; however, PF iso- metric torque and RTD increased (P�.05). In addition, changes in DF

Figure 2. Torque, angle, velocity, and power measurements during one repetition of isotonic plantar flexion (against 50% of isometric peak torque) for one representative participant.

Ankle Measures in Older Women

August 2010 Volume 90 Number 8 Physical Therapy f 1169

D ow

nloaded from https://academ

ic.oup.com /ptj/article/90/8/1165/2738033 by guest on 14 S

eptem ber 2020

Table 2. Means and Standard Deviations for Isometric, Isotonic, and Isokinetic Tests

Measure Test 1 Test 2 P a

Dorsiflexion

Isometric results

Peak torque (N�m) 21.6 (5.1) 21.2 (5.5) .28

RTDb (to 50% of peak torque, N�m/s) 97.8 (28.6) 95.3 (34.9) .55

RTD (40%–80% of peak torque, N�m/s) 62.3 (18.6) 59.3 (23.1) .30

Isotonic results

Peak velocity (1-N�m load, °/s) 160.9 (31.0) 158.5 (28.9) .24

Average acceleration (1-N�m load, °/s2) 685.5 (183.0) 662.0 (176.7) .05

Peak power (1-N�m load, W) 14.7 (6.1) 13.8 (5.6) .18

Peak velocity (50% of maximum isometric load, °/s) 78.2 (18.5) 80.0 (15.0) .53

Average acceleration (50% of maximum isometric load, °/s2) 345.2 (95.3) 350.9 (81.9) .57

Peak power (50% of maximum isometric load, W) 15.7 (6.3) 15.5 (6.2) .64

Isokinetic results

Peak torque (30°/s, N�m) 14.0 (4.6) 13.9 (4.8) .68

Peak torque (90°/s, N�m) 10.5 (4.2) 10.6 (4.1) .68

Peak power (30°/s, N�m) 7.2 (2.3) 7.1 (2.5) .76

Peak power (90°/s, N�m) 11.2 (4.5) 10.9 (4.4) .35

Plantar flexion

Isometric results

Peak torque (N�m) 71.0 (21.5) 77.5 (24.0) .03

RTD (to 50% of peak torque, N�m/s) 113.5 (60.1) 142.0 (65.3) .02

RTD (40%–80% of peak torque, N�m/s) 68.9 (30.1) 90.3 (48.8) .02

Isotonic results

Peak velocity (15-N�m load, °/s) 275.2 (47.8) 274.8 (50.1) .93

Average acceleration (15-N�m load, °/s2) 1,686.4 (477.2) 1,698.3 (460.5) .79

Peak power (15-N�m load, W) 171.3 (73.0) 180.0 (71.2) .24

Peak velocity (50% of maximum isometric load, °/s) 224.6 (44.2) 217.9 (43.6) .34

Average acceleration (50% of maximum isometric load, °/s2) 1,304.7 (334.8) 1,235.6 (318.5) .17

Peak power (50% of maximum isometric load, W) 158.9 (59.1) 162.7 (57.0) .51

Isokinetic results

Peak torque (30°/s, N�m) 66.7 (20.0) 69.7 (20.2) .11

Peak torque (90°/s, N�m) 61.4 (15.8) 62.0 (18.5) .82

Peak power (30°/s, N�m) 35.0 (10.3) 37.1 (10.6) .06

Peak power (90°/s, N�m) 76.2 (18.0) 77.8 (21.2) .54

a P values from paired t tests, except dorsiflexion isotonic peak power (1-N�m load) and plantar-flexion isometric peak torque, which were analyzed with the signed rank test. b RTD�rate of torque development.

Ankle Measures in Older Women

1170 f Physical Therapy Volume 90 Number 8 August 2010

D ow

nloaded from https://academ

ic.oup.com /ptj/article/90/8/1165/2738033 by guest on 14 S

eptem ber 2020

Table 3. Relative and Absolute Reliability Scores for Isometric, Isotonic, and Isokinetic Testsa

Measure ICCb 95% CI for ICC

SEM (Units)

95% CI for SEM (Units)

CVTE (%)

Systematic Biasc � MDC95 (Units)

Ratio LOA (% of Mean)

Dorsiflexion

Isometric results

Peak torque (N�m) .97 0.94–0.99 1.3 �2.4 6 0.4�3.5 16

RTD (0%–50%, N�m/s) .86 0.71–0.94 15.8 �31.0 16 2.5�43.8 45

RTD (40%–80%, N�m/s) .84 0.67–0.92 11.0 �21.5 18 3.0�30.4 50

Isotonic results

Peak velocity (1-N�m load, °/s) .96 0.92–0.98 8.0 �15.8 5 2.5�22.3 14

Average acceleration (1-N�m load, °/s2) .97 0.93–0.98 43.7 �85.6 6 23.5�121.1 18

Peak power (1-N�m load, W) .90 0.79–0.95 2.5 �4.9 17 0.9�6.9 48

Peak velocity (50%,°/s) .76 0.50–0.89 10.6 �20.7 13 �1.7�29.2 37

Average acceleration (50%, °/s2) .90 0.79–0.95 37.9 �74.3 11 �5.7�105.1 30

Peak power (50%, W) .95 0.90–0.98 1.9 �3.7 12 0.2�5.3 34

Isokinetic results

Peak torque (30°/s, N�m) .95 0.89–0.98 1.5 �3.0 11 0.2�4.2 30

Peak torque (90°/s, N�m) .96 0.92–0.98 1.2 �2.3 11 �0.1�3.2 31

Peak power (30°/s, N�m) .94 0.88–0.97 0.8 �1.6 11 0.1�2.3 32

Peak power (90°/s, N�m) .97 0.94–0.99 1.0 �2.0 9 0.3�2.9 26

Plantar flexion

Isometric results

Peak torque (N�m) .90 0.74–0.95 9.2 �18.0 12 �6.5�25.4 34

RTD (0–50%, N�m/s) .63 0.20–0.82 44.7 �87.6 35 �28.4�123.8 97

RTD (40–80%, N�m/s) .58 0.12–0.80 29.9 �58.6 37 �19.5�82.9 104

Isotonic results

Peak velocity (15-N�m load, °/s) .93 0.85–0.97 18.2 �35.6 7 0.4�50.4 18

Average acceleration (15-N�m load, °/s2) .93 0.86–0.97 168.2 �329.6 10 �11.9�466.1 28

Peak power (15-N�m load, W) .92 0.83–0.96 27.6 �54.1 16 �8.6�76.5 44

Peak velocity (50%, °/s) .77 0.51–0.89 27.0 �53.0 12 6.7�74.9 34

Average acceleration (50%, °/s2) .79 0.56–0.90 192.0 �376.2 15 69.1�532.1 42

Peak power (50%, W) .92 0.68–0.93 22.2 �43.5 14 �13.8�61.5 40

Isokinetic results

Peak torque (30°/s, N�m) .89 0.77–0.95 8.7 �17.1 13 �3.8�24.2 36

Peak torque (90°/s, N�m) .85 0.68–0.93 8.9 �17.5 14 �0.5�24.7 40

Peak power (30°/s, N�m) .88 0.75–0.95 4.6 �9.0 13 �2.4�12.8 35

Peak power (90°/s, N�m) .86 0.71–0.93 9.8 �19.2 13 �1.6�27.1 35

a ICC�intraclass correlation coefficient, CI�confidence interval, SEM�standard error of measurement, CVTE�coefficient of variation of typical error, MDC95�minimal detectable change with 95% confidence. b ICC (2,3) for isometric results and ICC (2,5) for isotonic and isokinetic results. c Systematic bias�average difference between the 2 tests (time1�time2).

Ankle Measures in Older Women

August 2010 Volume 90 Number 8 Physical Therapy f 1171

D ow

nloaded from https://academ

ic.oup.com /ptj/article/90/8/1165/2738033 by guest on 14 S

eptem ber 2020

isotonic average acceleration (1-N�m load) and PF isokinetic peak power at 30°/s were very close to being statistically significant (P�.05 and P�.06, respectively).

Table 3 reports the reliability data for all DF and PF tests. The ICC values for DF tests (ICC�.76 –.97) were higher (signed rank test, P�.001) than ICC values for matched PF tests (ICC�.58 –.93), with the exception of 2 isotonic values (peak power against minimal load and peak veloc- ity against 50% of maximum isomet- ric load). Measures of absolute reli- ability (CVTE) also demonstrated more reliable values for all DF tests (5%–18%) compared with PF tests (7%–37%), except for the same 2 iso- tonic measures (signed rank test, P�.001). Isotonic peak velocity tests at minimal loads were associated with the lowest CVTE and RLOA val- ues for both DF (5% and 14%, respec- tively) and PF (7% and 18%, respec- tively). Isometric RTD0%–50% and RTD40%– 80% demonstrated the high- est levels of variability between test sessions for both DF (CVTE�16% and 18%, respectively, and RLOA�45% and 50%, respectively) and PF (CVTE�35% and 37%, respectively, and RLOA�97% and 104%, respec- tively). The MDC values, considered to be the minimal amount of change in an outcome measure that can be measured for an individual that is not due to systematic or chance variation in measurement,37,39 are included for all variables. Specifically, MDC95 val- ues indicate that a person can be 95% confident in the true nature of changes that exceed these levels. The LOA were equal to the system- atic bias � the MDC95 value.

Bland-Altman plots (individual par- ticipant differences plotted against the mean for both test sessions) were created for all outcome vari- ables to look for systematic bias, outliers, and the presence of het- eroscedasticity. Pearson correlation

coefficients were not significant for heteroscedasticity for 23 of the 26 tests, but isotonic PF peak velocity (against 50% of isometric peak torque), PF RTD0%–50%, and PF RTD40%– 80% did demonstrate signifi- cant positive correlations (r�.45– .65, P�.01). Reliability statistics as- sociated with the RTD and peak isotonic velocity (against 50% of iso- metric peak torque) variables were relatively poor; therefore, other strength or power variables should be chosen in test-retest situations. No data transformations were conducted.

Discussion This study was conducted to estab- lish relative and absolute reliability scores for isometric, isotonic, and isokinetic strength- and power- related measures about the ankle in older women. Although the reliabil- ity of some of these measures (eg, isokinetic tests) has been investi- gated previously, other parameters (eg, isotonic values) have been re- ported infrequently in the literature with no associated reliability infor- mation provided. Results demon- strated that isometric, isotonic, and isokinetic measures of strength and power were associated with good relative reliability (all ICCs�.75, with the exception of PF RTD)29 and measures of absolute reliability were similar to previously published re- sults involving both younger and older individuals.11,24,25,36

Virtually all isometric, isotonic, and isokinetic DF and PF measures dem- onstrated good relative reliability, in- dicating that these measures gener- ally exhibited consistency for repeated measurements at the group level. With the exception of PF RTD0%–50% (ICC [2,k]�.63) and PF RTD40%– 80% (ICC [2,k]�.58), all ICC values ex- ceeded .75, and more than half of the values reached .90 or greater. The 95% confidence intervals associ- ated with the ICCs (Tab. 3) provide a more thorough understanding of the

reliability of these measurements. In the majority of cases (18/26), the lower confidence interval did not fall below 0.70; however, at the worst extreme, PF RTD measurements demonstrated lower confidence lim- its of 0.20 and 0.12, indicating very poor test-retest reliability.

In terms of strength, women in this study obtained DF and PF isokinetic peak torque values similar to those previously reported.11,12,16,40 The ICC values associated with isokinetic DF and PF peak torque and peak power (ICC�.85–.97) also were very similar to those reported in a previ- ous study (ICC�.92–.98) of older women and men tested at 60°/s.11

The current study is the first to re- port reliability statistics associated with isometric and isotonic tests about the ankle in older women; therefore, no comparisons of these variables could be made.

Clinically, SEM values (expressed in absolute units) and CVTE values (ex- pressed as a percentage) can be used to determine whether signifi- cant change has occurred in a group over time. The SEM and CVTE results were similar to those reported in other ankle strength and power stud- ies for isokinetic parameters11,25,36

and isometric RTD.24 The CVTE val- ues for DF and PF isometric peak torque were relatively low in this study (6% and 12%, respectively), whereas CVTE results were slightly higher for isokinetic peak torque and peak power results, ranging from 9% to 14%.

Clinicians can use MDC95 and RLOA values to determine whether true change has occurred over time in individual patients. Based on our results, changes in isokinetic peak torques in individual patients would need to exceed the following thresholds to exceed measurement error: 4.2 N�m (DF�30°/s), 3.2 N�m (DF�90°/s), 24.2 N�m (PF�30°/s),

Ankle Measures in Older Women

1172 f Physical Therapy Volume 90 Number 8 August 2010

D ow

nloaded from https://academ

ic.oup.com /ptj/article/90/8/1165/2738033 by guest on 14 S

eptem ber 2020

and 24.7 N�m (PF�90°/s). Isometric peak torques would need to exceed 3.5 N�m (DF) and 25.4 N�m (PF). The MDC95, LOA (systematic bias � MDC95), and RLOA results in this study were similar to those previ- ously reported.11,12,25 Small differ- ences among study results may be attributed to differences in the par- ticipants (eg, sex, age), the raters, or the test protocol itself (eg, test velocities, measurement of peak ver- sus average power, and participant positioning).

The older women in our study reached peak isotonic velocities of 275°/s for PF and 161°/s for DF when testing was conducted against minimal loads. Only one previous study has reported peak isotonic ve- locities about the ankle in older adults.2 The current study adds to the literature by providing detailed information about relative and abso- lute reliability associated with differ- ent isotonic parameters that have been measured infrequently to date. In the present study, isotonic peak velocity and average acceleration were associated with low CVTE val- ues when the load was minimal (CVTE for DF�5% and 6% for peak velocity and average acceleration, re- spectively, CVTE for PF�7% and 10%, respectively) and slightly higher CVTE values (DF�13% and 11%, re- spectively, and PF�12% and 15%, re- spectively) when the load was equal to 50% of isometric peak torque. Iso- tonic peak velocity measured against low loads was associated with less variation compared with other iso- tonic and isokinetic variables. Fur- ther research involving other joint movements and different populations is needed to determine whether peak velocity is consistently more re- liable than other more traditionally measured parameters. This informa- tion may be important clinically, as the isotonic setting allows for evalu- ation of contractions in which veloc- ity is not constrained, and results,

therefore, may be more functionally relevant compared with isokinetic tests.

In all but 2 instances (isotonic peak power against minimal load and iso- tonic peak velocity against 50% of maximum isometric load), DF scores demonstrated better reliability com- pared with PF scores. This result is in agreement with the findings re- ported by Hartmann et al.11 In both studies, participants were positioned with the knee flexed for PF tests. Although the upper body and thigh were well stabilized with straps, it is conceivable that attempts to extend the knee or hip may have occurred during PF movements, adding vari- ability to these PF measurements that did not occur with DF move- ments. Reliability of ankle PF mea- sures may be improved with differ- ent positioning during testing (eg, hip in neutral and knee extended with the individual in a prone position).

It has been suggested that from a functional perspective, increases in RTD may represent one of the most important adaptations that occurs in response to resistance training in older adults.41 That is, the ability to generate moderate forces quickly may be more important than being able to generate high forces, espe- cially when quick action is required (eg, to regain balance and avoid a fall). Improvements in RTD are likely associated with a greater capacity to generate power. Although RTD may be a functionally important variable, our study demonstrated that it had the lowest absolute reliability of all the power-related variables studied about the ankle. The PF results were especially variable. Positioning used in this study (sitting with the hip and knee partially flexed) and the longer duration associated with iso- metric testing (3 seconds) likely con- tributed to some of this variability (greater potential contributions of

hip or knee extension accompanying isometric PF attempts). It should be noted that isokinetic dynamometers may not be as reliable for isometric tests as other devices that are inher- ently more stable (eg, custom-made isometric rigs). It is recommended that the reliability of measurements of RTD about the ankle be examined in future studies using different joint and body positions and possibly us- ing different types of strength testing equipment.

In this study, a familiarization session on the dynamometer was not pro- vided before the 2 test sessions. This lack of a familiarization session may represent a limitation of the study if learning had an effect on the scores during the second testing session. However, familiarization sessions are rarely provided in clinical situations and may not always be feasible in research circumstances because of time constraints, associated costs, and availability of equipment. For these reasons, we elected to omit a familiarization session. Measured lev- els of systematic bias were minimal for most variables, indicating that there was no substantial learning effect.

This study involved a relatively ho- mogeneous sample of community- dwelling older women. Future stud- ies should continue to investigate the reliability of strength and power measurements attained using differ- ent modes on the dynamometer about different joints and in other segments of the older population (eg, older men, individuals who are more frail).

Conclusions Interpreting and setting threshold levels for acceptable reliability re- sults depends on the particular test- ing circumstance.29 In this study, many variables demonstrated good ICC results and CVTE values in the range of 6% to 13%, which are com-

Ankle Measures in Older Women

August 2010 Volume 90 Number 8 Physical Therapy f 1173

D ow

nloaded from https://academ

ic.oup.com /ptj/article/90/8/1165/2738033 by guest on 14 S

eptem ber 2020

parable to previous strength and power assessments in younger and older people.9,10,25,36 These levels are likely adequate to determine gross changes in strength- and power-related parameters among groups over the course of a training period; but ideally, more reliable measures would provide greater confidence in interpreting clinically meaningful change within individu- als. Further research is needed to examine the reliability of isotonic variables that have been studied in- frequently using dynamometers. These measures may prove to be more reliable and relevant to func- tion in older adults than the more commonly reported isometric and isokinetic strength and power pa- rameters. In the meantime, MDC95 scores have been presented for all DF and PF isometric, isotonic, and isokinetic variables to provide mean- ingful thresholds for clinicians and researchers to identify changes in in- dividuals beyond those expected by measurement error.

Both authors provided concept/idea/re- search design, data collection, and project management. Ms Webber provided writing and data analysis. Dr Porter provided fund procurement, facilities/equipment, and con- sultation (including review of manuscript be- fore submission).

Ethical approval for this study was granted by the Education/Nursing Research Ethics Board of the University of Manitoba.

Some of the results specific to isotonic tests were presented orally at the Canadian Soci- ety for Exercise Physiology meeting; Novem- ber 11–14, 2009; Vancouver, British Colum- bia, Canada.

Ms Webber was supported by a Canadian Institutes of Health Research, Institute of Ag- ing fellowship.

This article was submitted November 30, 2009, and was accepted April 10, 2010.

DOI: 10.2522/ptj.20090394

References 1 Roubenoff R. Sarcopenia: effects on body

composition and function. J Gerontol A Biol Sci Med Sci. 2003;58:1012–1017.

2 McNeil CJ, Vandervoort AA, Rice CL. Pe- ripheral impairments cause a progressive age-related loss of strength and velocity- dependent power in the dorsiflexors. J Appl Physiol. 2007;102:1962–1968.

3 Hakkinen K, Hakkinen A. Muscle cross- sectional area, force production and relax- ation characteristics in women at different ages. Eur J Appl Physiol Occup Physiol. 1991;62:410 – 414.

4 Suzuki T, Bean JF, Fielding RA. Muscle power of the ankle flexors predicts func- tional performance in community- dwelling older women. J Am Geriatr Soc. 2001;49:1161–1167.

5 Bean JF, Leveille SG, Kiely DK, et al. A comparison of leg power and leg strength within the InCHIANTI study: which influ- ences mobility more? J Gerontol A Biol Sci Med Sci. 2003;58:728 –733.

6 Bean JF, Kiely DK, Herman S, et al. The relationship between leg power and physical performance in mobility-limited older people. J Am Geriatr Soc. 2002;50: 461– 467.

7 Skelton DA, Kennedy J, Rutherford OM. Explosive power and asymmetry in leg muscle function in frequent fallers and non-fallers aged over 65. Age Ageing. 2002;31:119 –125.

8 Drouin JM, Valovich-mcLeod TC, Shultz SJ, et al. Reliability and validity of the Biodex system 3 pro isokinetic dynamometer ve- locity, torque and position measurements. Eur J Appl Physiol. 2004;91:22–29.

9 Symons TB, Vandervoort AA, Rice CL, et al. Reliability of isokinetic and isometric knee-extensor force in older women. J Ag- ing Phys Act. 2004;12:525–537.

10 Symons TB, Vandervoort AA, Rice CL, et al. Reliability of a single-session isoki- netic and isometric strength measurement protocol in older men. J Gerontol A Biol Sci Med Sci. 2005;60:114 –119.

11 Hartmann A, Knols R, Murer K, de Bruin ED. Reproducibility of an isokinetic strength-testing protocol of the knee and ankle in older adults. Gerontology. 2009; 55:259 –268.

12 Ordway NR, Hand N, Briggs G, Ploutz- Snyder LL. Reliability of knee and ankle strength measures in an older adult popu- lation. J Strength Cond Res. 2006;20:82– 87.

13 Frontera WR, Hughes VA, Dallal GE, Evans WJ. Reliability of isokinetic muscle strength testing in 45- to 78-year-old men and women. Arch Phys Med Rehabil. 1993;74:1181–1185.

14 Capranica L, Battenti M, Demarie S, Figura F. Reliability of isokinetic knee extension and flexion strength testing in elderly women. J Sports Med Phys Fitness. 1998; 38:169 –176.

15 Lanza IR, Towse TF, Caldwell GE, et al. Effects of age on human muscle torque, velocity, and power in two muscle groups. J Appl Physiol. 2003;95:2361–2369.

16 Thelen DG, Schultz AB, Alexander NB, Ashton-Miller JA. Effects of age on rapid ankle torque development. J Gerontol A Biol Sci Med Sci. 1996;51:M226 –M232.

17 McGibbon CA, Krebs DE. Effects of age and functional limitation on leg joint power and work during stance phase of gait. J Rehabil Res Dev. 1999;36:173–182.

18 Bendall MJ, Bassey EJ, Pearson MB. Factors affecting walking speed of elderly people. Age Ageing. 1989;18:327–332.

19 Mackey DC, Robinovitch SN. Mechanisms underlying age-related differences in abil- ity to recover balance with the ankle strat- egy. Gait Posture. 2006;23:59 – 68.

20 Wolfson L, Judge J, Whipple R, King M. Strength is a major factor in balance, gait, and the occurrence of falls. J Gerontol A Biol Sci Med Sci. 1995;50 Spec No:64 – 67.

21 Whipple RH, Wolfson LI, Amerman PM. The relationship of knee and ankle weak- ness to falls in nursing home residents: an isokinetic study. J Am Geriatr Soc. 1987; 35:13–20.

22 Morris-Chatta R, Buchner DM, de Lateur BJ, et al. Isokinetic testing of ankle strength in older adults: assessment of inter-rater reliability and stability of strength over six months. Arch Phys Med Rehabil. 1994;75:1213–1216.

23 Stauber WT, Barill ER, Stauber RE, Miller GR. Isotonic dynamometry for the assess- ment of power and fatigue in the knee extensor muscles of females. Clin Physiol. 2000;20:225–233.

24 Sleivert GG, Wenger HA. Reliability of measuring isometric and isokinetic peak torque, rate of torque development, inte- grated electromyography, and tibial nerve conduction velocity. Arch Phys Med Re- habil. 1994;75:1315–1321.

25 Holmback AM, Lexell J. Reproducibility of isokinetic ankle dorsiflexor strength and fatigue measurements in healthy older sub- jects. Isokinet Exerc Sci. 2007;15:263–270.

26 McNeil CJ, Rice CL. Fatigability is in- creased with age during velocity- dependent contractions of the dorsiflex- ors. J Gerontol A Biol Sci Med Sci. 2007; 62:624 – 629.

27 Porter MM, Vandervoort AA, Kramer JF. A method of measuring standing isokinetic plantar and dorsiflexion peak torques. Med Sci Sports Exerc. 1996;28:516 –522.

28 Connelly DM, Vandervoort AA. Effects of isokinetic strength training on concen- tric and eccentric torque development in the ankle dorsiflexors of older adults. J Gerontol A Biol Sci Med Sci. 2000;55: B465–B472.

29 Portney LG, Watkins MP. Foundations of Clinical Research: Applications to Prac- tice. 3rd ed. Upper Saddle River, NJ: Pear- son Education Inc; 2009.

30 Macaluso A, De Vito G. Comparison be- tween young and older women in explo- sive power output and its determinants during a single leg-press action after op- timisation of load. Eur J Appl Physiol. 2003;90:458 – 463.

Ankle Measures in Older Women

1174 f Physical Therapy Volume 90 Number 8 August 2010

D ow

nloaded from https://academ

ic.oup.com /ptj/article/90/8/1165/2738033 by guest on 14 S

eptem ber 2020

31 Nadeau S, Gravel D, Arsenault AB. Rela- tionships between torque, velocity and power output during plantarflexion in healthy subjects. Scand J Rehabil Med. 1997;29:49 –55.

32 Stratford PW. Letter to the editor on “Test- retest reliability and minimal detectable change on balance and ambulation tests, the 36-Item Short-Form Health Survey, and the United Parkinson Disease Rating scale in people with parkinsonism. Phys Ther. 2008;88:888 – 890.

33 Hopkins WG. Measures of reliability in sports medicine and science. Sports Med. 2000;30:1–15.

34 Atkinson G, Nevill AM. Statistical methods for assessing measurement error (reliabil- ity) in variables relevant to sports medi- cine. Sports Med. 1998;26:217–238.

35 Stratford PW, Goldsmith CH. Use of the standard error as a reliability index of in- terest: an applied example using elbow flexor strength data. Phys Ther. 1997;77: 745–750.

36 Holmback AM, Porter MM, Downham D, Lexell J. Reliability of isokinetic ankle dor- siflexor strength measurements in healthy young men and women. Scand J Rehabil Med. 1999;31:229 –239.

37 Beckerman H, Roebroeck ME, Lankhorst GJ, et al. Smallest real difference, a link between reproducibility and responsive- ness. Qual Life Res. 2001;10:571–578.

38 Schmitt JS, Di Fabio RP. Reliable change and minimum important difference (MID) proportions facilitated group respon- siveness comparisons using individual threshold criteria. J Clin Epidemiol. 2004; 57:1008 –1018.

39 Haley SM, Fragala-Pinkham MA. Interpret- ing change scores of tests and measures used in physical therapy. Phys Ther. 2006; 86:735–743.

40 Ferri A, Scaglioni G, Pousson M, et al. Strength and power changes of the human plantar flexors and knee extensors in re- sponse to resistance training in old age. Acta Physiol Scand. 2003;177:69 –78.

41 Aagaard P, Simonsen EB, Andersen JL, et al. Increased rate of force development and neural drive of human skeletal muscle following resistance training. J Appl Physiol. 2002;93:1318 –1326.

Ankle Measures in Older Women

August 2010 Volume 90 Number 8 Physical Therapy f 1175

D ow

nloaded from https://academ

ic.oup.com /ptj/article/90/8/1165/2738033 by guest on 14 S

eptem ber 2020