Isokinetic Testing of the Shoulder Abductors and Adductors: Windowed vs Nonwindowed Data Collection

Kevin E. Wilk, PT' Christopher A. Arrigo, MS, PT, ATCZ lames R. Andrews, MD3

sokinetic testing of the shoulder is a common practice in the sports physical therapy clinic. Numerous articles have published investigations regarding the isokinetic strength of baseball pitchers a t various levels (1-3, 5, 10, 15), tennis players (6, 12, 13), and nonathletic individuals (4, 7-1 0, 14). However, only a few have reported isokinetic test data regarding abduction and adduction (1, 11) of the shoulder. Isokinetic testing of the shoulder abductors/adductors involves the use of a long and heavy lever arm attachment that is affixed t o the dynamometer (see Figure 1). T h e test consists of subjects moving this apparatus through their available range of motion, which is preset prior t o testing. As the subject moves the lever arm apparatus through the testing motion and reaches the end stop of the preset range of motion, a significant spike in the torque curve can be observed (see Figures 2-4 and Figures 5 and 6). This torque spike occurs as a result of combining a long lever arm with an abrupt end stop during testing. T h e isokinetic system's computer misinterprets this end-stop oscillation as the actual peak torque point during data analysis, which will result in an inflated
JOSPT Volume 15 Number 2 February 1992

The manner of acquiring strength-testing data may influence the results of an investigation. The purpose of this study was to determine if a significant difference exists between windowed and unwindowed data collection during isokinetic testing of the shoulder abductorsladductors. Fifty healthy professional baseball pitchers participated in this study. Testing was performed on a Biodex isokinetic dynamometer at 180 and 30O0/sec for both the throwing and nonthrowingshoulders. Testing procedures regarding testing protocol, repetitions, positioning, and stabilization followed establishedguidelines for each subject. Statistical analysis was performed using a paired t-test with a p < 0.01 level of significance. Statistically significant differences were demonstrated between windowed and unwindowed mean peak torque data for both shoulders at both test speeds. The results indicated an average nonthrowingarm difference of 20.2 ftllbs at 180/sec and 51.7 ftllbs at 300/ sec for the abductors. In each instance, the unwindowed mean peak torque values were higher than the windowed values, and significant end range torque spikes were elicited during unwidowed data collection. The nonthrowing adductors exhibited an average of 39.3 and 48.3 ftllb differences at 180 and 300"sec, respectively. The throwing shoulder demonstratedaverage abductor differences of 25.6 ftllbs at 180/sec and 47.7 ftllbs at 30O0/sec. The average throwing adductor difference was 24.4 ftllbs and 54.6 ftllbs, respectively, at both test speeds. This investigation offers clinical relevance for those using isokinetic testing of the shoulder abductorsladductors in demonstrating the significant differences between windowed and unwindoweddata, identifying torque spike data misinterpretation, and describing a clinical means of controlling aberrant torque production during testing.

Key Words: isokinetics, shoulder testing, torque spikes, windowed data

Presented at the Sports Physical Therapy Section Team Concept Meeting, December 1991, New Orleans, M. ' Director, Clinical Services, Healthsouth Sports Medicine and Rehabilitation, Birmingham, Al. director of Rehabilitation Research, American Sports Medicine Institute, Birmingham, A1 Healthsouth Sports Medicine and Rehabilitation, Birmingham, AL 'Alabama Sports Medicine and Orthopaedic Center, Birmingham, Al. medical director, American Sports Medicine Institute, Birmingham, A1

torque measurement. A similar torque spike has also been termed "impact artifact data" by Sapega et al and has been reported during hip abduction testing (16). This torque spike occurred due to overshoot torque when the subject initiated a forceful movement (hip abduction).

This resulted in a rapid torque spike as the subject engaged the system, which resulted in overspeeding the system. Several isokinetic systems, Biodex (Biodex, Biodex Corp, Shirley, NY), Lido (Loredan Biomedical, Davis, CA), Cybex (Lumex Corp,

FIGURE 1. The lever arm apparatus used for isokinetic testing oi shoulder abductors/adductors.

ing the last 15-20" of the test motion. This is unfavorable because the subject is not moving through the range of motion at the preset isokinetic angular velocity. The subject would, therefore, be rapidly decelerating through a large part of the motion being tested. A possible solution for this dilemma in isokinetic evaluations is to window the isokinetic test data. T h e term "isokinetic window" refers to filtering out all data that have not been obtained at the preset isokinetic speed or 95 percent of that speed. This filtering appears to elim10

150

Torque Magnify

INV-FAST (1801180)

Rep 3 of

HARD CUSHION

(UNWINDOWED)

Torque spike occurred due to overshoot torque when the subject initiated a forceful movement.
inate the end range spike oscillation or impact artifact in the torque curve produced during shoulder abduction and adduction. T h e purpose of this investigation was to compare windowed isokinetic data with unwindowed data from the same shoulder abduction/adduction isokinetic tests to determine if a significant difference exists between these two and if end stop spike oscillations are eliminated from the data reported. Gross et al (8) reported that windowing data appears to improve the intermachine reliability of peak knee extension and flexion torque at both 60 and 180/sec. T o the investigators' knowledge, no previous study has examined this question during shoulder abduction/adduction testing. In addition, previous studies (1, 1 1) reporting shoulder abduction/ adduction results were performed on a Cybex 11, which does not allow the
Volume 15 Number 2 February 1992 JOSPT

FIGURE 2. lsokinetic unwindowed torque curve illustrating torque spike due to end stop impact at 18O"sec.

Torque Magnify

IN"-FAST.

(1801180)

Rep 3 of

1 0

1 11

HARD CUSHION

(WINDOWED)

FIGURE 3. lsokinetic windowed torque curve illustrating the elimination of torque spike.

Ronkonkoma, NY), and Kin-Com (Chattanooga Corp, Chattanooga, TN) allow the tester to adjust the end stop in an attempt to cushion

this abrupt stop in motion. However, by utilizing a soft stop cushion, the isokinetic angular velocity is decelerated to a lower isokinetic speed dur-

RESEARCH STUDY

11

150 1 Torque Normal

INV-SLOW: ( 8 1 8 ) 1010

Rep 3 of 1 0

tester the option of windowing the data generated during testing.

METHODS
Subjects
Fifty informed professional baseball pitchers participated in this investigation. T h e subjects included were all professional pitchers ranging in age from 18 to 32 years of age, with a mean age of 24.5 3.3 yrs. T h e subjects' mean height was 74.2 f 2.6 in, and their mean weight was 200.9 18 Ibs, (see Table 1). Forty-one of the subjects were righthanded pitchers, and the remainder were left-handed.

FIGURE 4. Overlay of one repetition illustrating shaded area is due to end stop impact artifact.

loo 88

Torque Magnify

INV-FAST (3001300)

Rep 3 of 1 0

75635038

A paired t-test
demonstrated a statistically significant difference between the windowed and unwindowed mean peak torque values.
All subjects tested had no recent history (past 6 months) of shoulder pain, tenderness, or any known pathology. All subjects exhibited a normal clinical examination. In addition, all subjects were involved in unrestricted throwing activities at the time of the study.

3.98

42 .3

44 .8

47 .3

4.98

52 .3

54 .8

57 .3

59 .8

FIGURE 5. lsokinetic unwindowed toque curve for shoulder abduction/adduction performed at 30O0/sec. Note large torque spikes' presence.

Torque Magnify

INV-FAST (3001300)

Rep 3 of

1 0

Testing Procedure
Each subject's shoulder was evaluated using a Biodex Multi-joint System for shoulder abduction/adduction in the seated position (see Figure 7). T h e testing apparatus was set up and subjects were positioned and stabilized as described by Wilk and Arrigo in the standardized testing

FIGURE 6. lsokinetic windowed toque curve exhibiting the elimination of torque spike at 300e/sec.

JOSPT Volume 15 Number 2 February 1992

RESEARCH STUDY

Mean
Age (years) Heigh~ (inches) Weight (pounds) 24.5 74.2 200.9

SD
3.3 2.6 18.0

Range
16-32 67-80 150-238

TABLE 1. Demographics (N = 50).

5 nlin, perfornling 600 kg/m of work at 75-90 revolutions per minute (rpnl). T h e isokinetic testing protocol was then initiated beginning with 180/sec and progressing to 300/ sec for each test. Fhch subject per- . formed six gradual subn~axinlal repetitions, progressing from 50-80 percent effort at 180/sec as an initial warm-up. This was followed by three rnaxinlal contractions at the same speed prior to testing. Ten maximal test repetitions were then performed. Subjects then performed the same warm-up at 300/sec, followed

by 15 nlaxirnal test repetitions. This procedure was then repeated for the dominant arm in the identical test position. Subjects received standardized verbal commands during the testing sequence. Subjects were unaware of test results until the completion of the testing sequence. T h e test data were collected with an end stop, which the authors define as a semihard cushion. This cushion is attained by adjusting the end stop dial '/4 turn short of a completely hard end stop. The data were then analyzed first without an isokinetic window, then with an isokinetic window in all test cases.

RESULTS
Mean peak torque values comparing windowed to unwindowed isokinetic data of the dominant throwing shoulder abductors/adductors are presented in Table 2. A paired t-test demonstrated a statistically significant difference ( p < 0.01) between the windowed and unwindowed mean peak torque values produced at both test speeds. Kesults indicated significantly greater ( p < 0.0 1) unwindowed peak torque values. The nonthrowing arm also showed similar trends with statistically significant differences ( p < 0.01) found between the windowed and nonwindowed values at both test speeds (Table 3).
FIGURE 8. During isokinetic testing of the shoulder< abductors/adductors, the chain is In a reclined position 15-20"nd the dynamometer i c tilted ISn, axis o i rotation through the acromioclavicular joint.

FIGURE 7. Isokmetic testlng position ior ,boulder abductron/adduction in the seated position w t h upper body stabilization.

protocol for the throwers' shoulder (17). Subjects were tested in the seated position utilizing the Biodex accessory chair. T h e back support of the chair was reclined 15" for c o n fort. In addition, the dynamometer was also tilted 15" in the same plane to align the subject with the dynanlometer (See Figure 8). 'The axis of rotation of the dynanlon~eter was aligned with the subject at 1 crn lateral to the acronlioclavicular joint (15). Data was obtained at two testing velocities, 180 and SOOO/secconcentrically, and the nondonlinant (nonthrowing) shoulder was always tested first. A standardized ~ ~ a r n l - u p was performed on an Upper Body ergometer (U BE, Cybex, Division of Lumex. Inc. Ronkomkoma, NY) for
110

DISCUSSION
A significant difference exists between windowed and nonwindowed
N = 50

Test Speed

Abduction Window ft-lbs Nonwindow ft-lbs

81.2 f 32 91.1 f 37

Adduction Window ft-lbs

69.1 f 13 55.7 f 14

Nonwindow ft-lbs
93.5 f 34 110.3 f 44

18O0/sec 30O0/sec

55.6 f 12 43.4 f 14

TABLE 2. Comparison of mean peak torque f SD (it-lh)between windowed and nonwindowed shoulder abduction/adduction values for the throwing arm.
Volume I3 Sunher 2 Februar). 1992 JOSPT

RESEARCH STUDY

Test Speed

Abduction Window ft-lbs Nonwindow

Adduction

ft-lbs

Window ft-lbs

Nonwindow ft-lbs

Respective pairs show signrhcant stat~stical diiierences (p < 0.01).

TABLE 3. Comparison of mean peak torque f SD (ft-lb) between windowed and nonwindowed shoulder abduction/adduction values for the nonthrowing arm.

data during isokinetic testing of the shoulder's abductors/adductors because the long lever arm apparatus used produces a large torque spike as the subject contacts the end stop in the terminal range of motion. Due t o the computerization of the isokinetic test interpretation, this torque spike is misinterpreted as the point of peak torque. Utilization of an "isokinetic window" eliminates these torque spike oscillations from the reported data. By windowing the data, all measures which are not obtained a t the preset isokinetic angular velocity o r a t 9 5 percent of that speed will not be recorded. Thus, the end range torque spikes that result during shoulder abduction/adduction a r e eliminated by windowing the data because these aberrant oscillations d o not occur at the preset isokinetic velocities. During shoulder abduction a t 180/sec, there exists a significant difference between windowed and unwindowed data for both the dominant and nondominant extremities. This windowed data represents a p proximately 70-74 percent of the values generated using unwindowed data. T h e windowed data a t 300 "/ sec from shoulder abduction represents only 42-47 percent of the torque produced with nonwindowed data. Thus, it appears that a t faster test speeds (300/sec), there is a significantly greater end range torque spike. This may be produced by the greater angular velocities generated during shoulder abduction, which create a large rebound effect as the
JOSPT * Volume 15 * Number 2 February 1992

end point is reached and movement begins in the opposite direction. T h e r e also exists a significant difference between the windowed and nonwindowed data a t both test speeds during shoulder adduction. T h e windowed data a t 180/sec r e p resents 73 percent of the nonwindowed data on the dominant throwing shoulder and 6 2 percent on the nondominant shoulder. At the faster test speed (300/sec), the nonwindowed data was approximately 5 0 percent greater than the windowed data. Again, it appears that the faster isokinetic speed produces a larger isokinetic torque spike in the terminal range of motion. It appears shoulder adduction produces a greater end stop torque spike. This may be due t o the significantly greater peak torque generated by the shoulders' adductors compared t o the abductors. This paper attempts to make clinicians aware of the problem of peak torque overshoot inherent in should e r abduction/adduction testing. Due t o the long lever arm apparatus used during this test motion, the end range torque spikes will be significantly greater than the mid-range peak torques generated. Therefore, the authors advise the utilization of an isokinetic windowing technique when analyzing shoulder abduction/ adduction tests. If a windowin'g technique is not utilized, and care is not taken to exclude the end stop oscillations (i.e. rebound torque spikes produced), clinicians may misinterpret the data generated. This may result

in peak torque data twice as great as the actual torque generated and, thus, gross misinterpretation of testing results. Although this is one of the first investigations to document end range torque spikes during isokinetic evaluation, these may occur during any test motion that employs a long lever arm. These motions include shoulder flexion/extension and horizontal abduction/adduction. Thus, it would appear that a similar end range torque spike may occur during these testing motions, and windowing techniques of data reduction may be required for proper interpretation. Testing of other joints with various short lever arm motions, such as knee flexion/extension, also appear t o produce these end range torque spikes. T h e authors feel that the end stop torque spike described in this paper is different than the impact artifact described by Sapega e t al (1 6). In their paper, the torque described is due to the subject engaging the system, which resulted in overspeeding the system. This usually occurs within the first 10-1 2" of motion. In our study, this occurred in a larger point of range of motion and, in addition, occurred in both ends of the torque curve. Gross e t al(8) has documented that impact artifact occurs during knee testing. In addition, Gross (8) reported that by windowing his Cybex and Biodex data, reliability was significantly improved compared t o unwindowed data. Based on this study and the work of others, it appears that both clinical interpretation and data reliability are enhanced by windowing isokinetic test results for the knee flexors/extensors (8)and the shoulder abductors/adductors. It is the opinion of the authors that isokinetic windowing should be utilized to eliminate end range torque spikes during shoulder abduction/adduction testing. It is also the authors' belief that additional re-

RESEARCH S T U D Y

search is needed t o determine i f end range torque spikes exist d u r i n g any other isokinetic test movements.

2.

CONCLUSIONS
Based o n the results o f this study regarding isokinetic testing o f the shoulder abductors/adductors, t h e following conclusions can be made: 1) T h e r e exists a significant end range torque spike d u r i n g shoulder abduction/adduction isokinetic testi n g at both 180 and 300/sec., and 2) These end range torque spikes may be misinterpreted as the peak torque point by the computer, resulting i n significantly greater torque measurements than t h e actual peak torques generated. JOSPT 3.

4.

5.

ACKNOWLEDGMENTS
T h e authors thanks Donna Erber and David Hinger f o r their assistance i n data collection, Michael Keirns f o r his assistance i n the statistical analysis, and Regina Biddings f o r her assistance i n the preparation o f this manuscript.

6.

7.

8.

REFERENCES
I . Alderink GI, Kuck Dl: lsokinetic shoulder strength of high school and col-

lege-aged pitchers. I Orthop Sports Phys Ther 7(4):163- 1 72, 1986 Bartlett LR, Sotrey MD, Simons BD: Measurement of upper extremity torque production and its relationship to throwing speed in the competitive athlete. Am j Sports Med l7(1):89-91, 1989 Brown LP, Niehues SL, Harrah A, Yarorsky P, Hirschman HP: Upper extremity range of motion and isokinetic strength of the internal and external shoulder rotators in major league baseball players. Am I Sports Med 16(6):577-585, 1988 Clarke WA: Reliability of an alternative method for measurements of shoulder rotation strength (abstr). Phys Ther 57:741, 1987 Cooke EE, Gray VL, Savinor-Nogue E, Medeiros I: Shoulder antagonistic strength ratios: A comparison between college-level baseball pitchers. I Orthop Sports Phys Ther 8(9):451-46 1, 1987 Ellenbecker TS: A total arm strength isokinetic profile of highly skilled tennis players. lsokin Exerc Sci 1:9-21, 1991 Greenfield BH, Donatelli R, Wooden MI, Wilkes 1: lsokinetic evaluation of shoulder rotational strength between plane of the scapula and functional plane. Am I Sports Med 18(2):124127, 1990 Cross MT, Huffman GM, Phillips CN, Wray ]A: lntramachine and intermachine reliability of the Biodex and Cybex I1 for knee flexion and extension peak torque and angular work. I Orthop Sports P 13(6):329-335, 1991 T

9. Hageman PA, Mason DK, Rydlund KW, Humpal SA: Effects of position and speed on eccentric and concentric isokinetic testing of the shoulder rotators. I Orthop Sports Phys Ther 1 1(2):64-69, 1989 10. Hinton RY: lsokinetic evaluation of shoulder rotational strength in high school baseball pitchers. Am I Sports Med 16(3):274-279, 1988 1 I . lvey FM, Calhoun jH, Rusche K, Bierschenk I: Normal values for isokinetic testing of shoulder strength (abstr). Med Sci Sports Exerc 16(2):274-279, 1988 12. Maddox REC, Kibler WB, Uhl T: Isokinetic peak torque and work values for the shoulder. I Orthop Sports Phys Ther 1 1:264, 1989 13. Ng LR, Kramer IS: Shoulder rotator torques in female tennis and nontennis players. I Orthop Sports Phys Ther 13:40, 1991 14. Otis IC, Warren RF, Backus 51, Santner TI, Mabrey ID: Torque production in the shoulder of the young male adult. Am Sports Med 18:1 19- 123, 1990 15. Pedegana LR, Elsner R, Roberts D, Lang I, Farewell V: The relationship of upper extremity strength to throwing speed. Am I Sports Med 10(6):352-354, 1982 16. Sapega AA, Nicholas )A, Sokolow D, Saraniti A: The nature of torque "overshoot" in Cybex isokinetic dynamometry. Med Sci Sports Exerc 14:368-375, 1982 17. Wilk KE, Arrigo CA, Andrews lR: A standardized isokinetic testing protocol for the throwers' shoulder: The throwers series. lsokin Exerc Sci 1(2):63-71, 1991

Volume 15 Number 2 February 1992 JOSPT