Ascertaining maximal voluntary effort production during isokinetic knee strength testing of anterior cruciate ligament-reconstructed patients.

Authors: Sivan Almosnino, PhD Zeevi Dvir, PhD, LLB Davide D. Bardana, MD Elena D. Diaconescu, MD Joan M. Stevenson, PhD

Affiliations:

Strength

ORIGINAL RESEARCH ARTICLE

From the School of Kinesiology and Health Studies, Queen’s University, Kingston, Ontario, Canada (SA, ZD, JMS); Department of Physical Therapy, The Stanley Steyer School of Health Professions, Sackler Faculty of Medicine, Tel Aviv University, Ramat-Aviv, Israel (ZD); Human Mobility Research Centre, Syl & Molly Apps Medical Research Centre, Kingston General Hospital, Kingston, Ontario, Canada (SA, DDB, EDD, JMS); and Division of Orthopaedic Surgery, School of Medicine, Queen’s University & Kingston General Hospital, Kingston, Ontario, Canada (DDB).

Correspondence: All correspondence and requests for reprints should be addressed to: Sivan Almosnino, PhD, Biomechanics and Ergonomics Laboratory, School of Kinesiology and Health Studies, Queen’s University, 28 Division St, Room 300, Kingston, Ontario, Canada, K7L 3N6.

Disclosures: Supported by awards and stipends from the Natural Sciences and Engineering Research Council of CanadaYCollaborative Research and Training Experience (CREATE) Program in Bone and Joint Health Technologies (to S.A., E.D.D.). Presented in part at the 17th biannual Canadian Society of Biomechanics conference, June 6Y9, 2012, Vancouver, British Columbia, Canada. Financial disclosure statements have been obtained, and no conflicts of interest have been reported by the authors or by any individuals in control of the content of this article.

0894-9115/14/9302-0169 American Journal of Physical Medicine & Rehabilitation Copyright * 2014 by Lippincott Williams & Wilkins

Ascertaining Maximal Voluntary Effort Production During Isokinetic Knee Strength Testing of Anterior Cruciate LigamentYReconstructed Patients ABSTRACT Almosnino S, Dvir Z, Bardana DD, Diaconescu ED, Stevenson JM: Ascertaining maximal voluntary effort production during isokinetic knee strength testing of anterior cruciate ligamentYreconstructed patients. Am J Phys Med Rehabil 2014;93:169Y181.

Objective: The aim of this study was to assess the performance of prediction rules meant for declaration of efforts as being maximal or not during isokinetic strength testing in a cohort that underwent anterior cruciate ligament reconstruction. Design: Thirty-six individuals performed four sets of six reciprocal concentric knee extension/flexion repetitions at a testing speed of 60 degrees per second through a 60-degree range of motion. The sets consisted of a maximal voluntary effort, two nonmaximal sincere efforts at 50% and 75% of self-perceived maximum, and a set attempting to feign or exaggerate thigh muscle strength deficiencies. Strength curve derived set internal consistency measures, namely, cross-correlation and percent root mean square difference scores, were inputted into the prediction rules, whose performance is reported as specificity and sensitivity percentages.

Results: Dependent on the prediction rule used and when expressed on an individual participant basis, the corresponding specificity and sensitivity values ranged from 66.6% to 97.2% and 97.2% to 94.4%, respectively.

Conclusions: Using the prediction rules presented in this investigation, clinicians may be able to ascertain maximal effort production during isokinetic testing in those who have undergone surgical reconstruction of their anterior cruciate ligament. Key Words:

Strength, Knee, Dynamometry, Maximal Voluntary Contraction

DOI: 10.1097/PHM.0000000000000043

www.ajpmr.com

Isokinetic Knee Strength Testing Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

169

T

he assessment of musculature strength capabilities after injury is thought to provide information relevant for patient treatment. Irrespective of the joint musculature assessed, the validity of the obtained results is reliant on the notion that the participant exerted his/her maximal voluntary effort during testing.1 Given this, there is a definite clinical need to substantiate the level of effort (i.e., maximal or nonmaximal) exerted by the patient during testing. In addition, it may also be useful in particular situations to corroborate the type of effort produced. That is, it may be of value to ascertain whether the participant exerted a sincere effort or whether the effort was produced with an underlying intent to deceive the examiner into concluding deficiencies in musculature strength capabilities.2,3 In a previous investigation,4 predictive rules for ascertaining the level or type of effort produced were established for isokinetic-based, concentric knee extension and flexion strength measurements using a cohort of healthy participants. This particular investigation relied on two predictor variables, namely, cross-correlation (Rxy) and percent root mean square difference (%RMSD) scores, calculated between pairs of strength curves to quantify the degree of strength production consistency within a given set.4 The premise for use of these measures was that a higher degree of set internal consistency would be observed when exerting maximal efforts because of adoption of a neuromuscular strategy that attempts, during performance of consecutive repetitions, to recruit all possible motor units of agonist muscle at the highest possible frequency while minimizing antagonist muscle group activity.5,6 In effect, this would be expected to be manifested in pairs of strength curves within a given set being more similar in shape and magnitude, resulting in higher Rxy and lower %RMSD scores, respectively. On the other hand, during performance of nonmaximal efforts, it is assumed that only a portion of the available motor units are initially recruited, and activation frequency varies as a result of the central nervous system continuously modifying efferent signaling in attempt to produce the intended muscular force level, which, in turn, is gauged through afferent circuitry feedback.5,6 Practically, the adoption of this neuromuscular strategy is expected to result in a compromise in the consistency of strength production expected during performance of nonmaximal effort exertions.6,7 The aforementioned previous investigation,4 however, used only scores of healthy participants, which is a practice common to investigations concerned with differentiating between maximal and nonmaximal efforts, including those pertaining

170

Almosnino et al.

to the knee musculature (Table 1).7Y17 Nonetheless, determining whether prediction rules developed using healthy participants may also be applied in case of participants with or recovering from injury is pertinent for generalization purposes.3 As such, the purpose of this investigation was to assess whether the prediction rules for ascertaining maximal effort production during isokinetic-based concentric knee strength testing developed using healthy participants may be used for decision purposes in a patient population. For this purpose, the authors have chosen to test those who have undergone surgical reconstruction of the anterior cruciate ligament (ACL). The choice of this patient population was motivated by (1) the relatively large numbers of this particular surgical procedure performed at the authors’ hospital; (2) the prevalent use of isokinetic dynamometry for assessment of knee musculature strength in this patient population18; and, in relation, (3) findings suggestive that some of those who underwent ACL reconstructive surgery may be experiencing psychologic and physiologic changes that may limit exertion maximal voluntary muscular efforts of either the knee extensor or flexor muscle groups throughout the entire set (e.g., variable antagonistic muscular activity,19 quadriceps reflex inhibition,20 and fear of movement21). Given these, it should be noted that the issue of differentiating between maximal and nonmaximal efforts has been a focus primarily in research pertaining to the medicolegal realm1Y3 and that isokinetic assessment of ACL-reconstructed patients has received much attention with regard to return to participation in sports.18,22 However, because the validity of strength test results is reliant on maximal voluntary effort exertions in any setting (i.e., rehabilitative, athletic), it is perceived that the ability to detect the production of sincere nonmaximal effort exertions in this patient population is of value because, and as alluded previously, this would possibly alert clinicians of underlying psychologic and physiologic issues that may need to be addressed. In addition, and although arguably not a prevalent phenomenon, it has been pointed out that in certain cases, those injured in the athletic setting may, for a variety of reasons and incentives, produce insincere efforts such that examiners conclude of deficiencies in physical abilities.23,24 As such, the ability to identify insincere nonmaximal effort production is of some interest in these settings as well.

METHODOLOGY Participants Participants were recruited from a list of patients who have undergone surgical reconstruction of their

Am. J. Phys. Med. Rehabil. & Vol. 93, No. 2, February 2014

Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

ACL within a time frame of 2Y36 mos. The study exclusion criteria were identified using a self-report medical questionnaire and physician consultation and included being diagnosed with or being at risk for developing high blood pressure levels and carotid and coronary artery disease, current use of medication that may elevate blood pressure, and recurrent episodes of dizziness or fainting. In addition, the participants were screened by their orthopedic surgeon using standard clinical tests for any contraindications in performance of maximal knee musculature efforts using isokinetic dynamometry. The participants were informed verbally and by way of an introduction letter about the aims of the study, the procedures to be used, and any potential risks arising from them. Thereafter, written informed consent was obtained from all participants. The study procedures were reviewed for ethical compliance and received approval by the Queen’s University Health Sciences and Affiliated Teaching Hospitals Research Ethics Board. In total, 36 participants undertook the study (15 men, mean T SD age, 28 T 11 yrs; height, 178 T 6 cm; weight, 84 T 14 kg; 21 women, age, 29 T 10 yrs; height, 167 T 4 cm; weight, 64 T 11 kg). In 26 of these participants, a hamstring autograft was used for ACL reconstruction, whereas a patellar tendon autograft was used for ACL reconstruction in the remaining 10 patients. The mean T SD time of testing after surgery was 12 T 7 mos (range, 2.5Y28 mos). It should be noted that some of the participants had the injury in the work setting (n = 9); however, none were involved in a medicolegal situation. In addition, the participants were involved in a mean T SD of 2.2 T 1.1 hrs a week of recreational sports, which primarily consisted of weight lifting; running; walking; and team sports such as volleyball, ultimate Frisbee, and soccer. Four of the participants had previous experience with isokinetic testing of the knee joint musculature.

Isokinetic Strength Testing Procedures Testing procedures emulate to a large extent those reported previously.4 Upon arrival to the laboratory, the participants performed a general warm-up consisting of 5 mins of cycling on a stationary bicycle followed by examiner-guided stretching of the hamstring, quadriceps, and triceps surae muscle groups. Measurements of the concentric isokinetic strength of the knee extensor and flexor muscles of the ACL-reconstructed knee were performed using a commercial isokinetic dynamometer (Biodex Multi-Joint system 3; Biodex Medical Systems Inc, Shirley, NY). Testing was performed in a seated position, with the chair’s back www.ajpmr.com

rest set at 85 degrees, and the participant secured to the chair using two straps across the chest and single straps at the pelvis and the distal thigh of the tested limb. The dynamometers’ axis of rotation was aligned with the lateral femoral condyle, and the lever arm pad was secured to the participant’s distal shank at the level of the medial malleolus. Efforts were performed through a 60-degree range of motion (ROM; starting with the knee flexed at 90 degrees) and at a preset angular velocity of 60 degrees per second. The choice of ROM was largely dictated by the authors’ reluctance to test through full knee extension. In particular, because a minority of the participants (n = 5) were tested within a span of 2Y3 mos after the surgical procedure, the authors contemplated that testing in the last 30 degrees of knee extension may result in increasing magnitude and time-sustained ACL forces that may be contradictive for graft healing.25 In addition, because several participants underwent ACL reconstruction using a hamstring autograft, it was considered that performance through full knee extension may compromise these muscles because of being in an elongated position toward end ROM. Another factor relates to the hamstring’s protective beneficiary role in limiting anterior tibial translation, a function that is hampered when approaching full knee extension.26 Concurrently, the choice of the test angular velocity was based on previous reports indicating that a high level of set internal consistency may be expected at this particular velocity during production of maximal voluntary efforts by healthy participants17 and also, to some extent, by those who have gone ACL reconstruction.27 Following instructions related to the operation and general use of the dynamometer in rehabilitative settings, the participants performed 15 continuous knee flexion-extension repetitions at a self-perceived low effort level. This was followed by three repetitions at a self-perceived medium effort level and two to three practice sets consisting of two to three maximal repetitions. Throughout the pretesting isokinetic familiarization session and warm-up, the participants were instructed and reminded during effort exertions to grip the side handle bars, to maintain constant contact of their head with the chair’s headrest, and not to decelerate the lever arm at end ROMs. Testing involved performance of four sets, each consisting of six repetitions. The first set consisted of maximal, continuous knee extension/flexion repetitions. Before the second set, the participants were read a standardized vignette in which they were asked to feign or exaggerate deficiencies in knee muscular strengthYproducing capabilities for secondary gain purposes while trying to convince the examiner that Isokinetic Knee Strength Testing


171

172

Almosnino et al.



KinCom 500 Biodex System 3

NA

20 (17 m), healthy, 33.3 T 9.5 yrs old

Knee extension10,b Knee extension11,c

KinCom 500

KinCom 500

31 m, healthy, 25.3 T 4.5 yrs. old

31 m, healthy, 25.3 T 4.5 yrs old

Knee extension9

Knee extension8,a

40 (20 m, 31.2 T Cybex II 10.1; 20 w, 26.4 T 5.5), healthy

Knee extension7

Instrument

Participants (n, sex, age)

Joint/ Movement

4 randomized sets of max and 50% nonmax at 30 degrees and 180 degrees per second through 80-degree ROM, no mention of feedback

2 random sets (maximal and self-selected nonmaximal best replicated) of 4 reps at 60 degrees per second, ROM not specified 2 test days, 2 sets of isolated knee extension reps through 90-degree ROM at 60 degrees per second (max or feign), verbal feedback provided 2 test days, isometric and isolated knee extension reps through 90-degree ROM at 60 degrees per second (max or feign), verbal feedback provided NA

Protocol

Mean T SD Maximal

14% 10.5% 14.04% 8.76%

È17% È15% È25% È27%

At 30 degrees per second: È7% At 180 degrees per second: È8% At 30 degrees per second: È5% At 180 degrees per second: È7%

CV of peak moments

Average CV per degree

8%

9.1 T 4.4%

3.8 T 2.2%

Not proposed

8%

CV

15.0 T 9.2%

Not proposed, concludes based on score distribution that effort types are indistinguishable

Proposed Cutoff Score

13 T 30 N I m

3.9 T 2.5%

At 30 degree: 12.0% At 45 degrees: 14.6%

Mean T SD Nonmaximal

33 T 29 N I m

Difference between isokinetic and isometric peak moment

CV of peak moments

CV of moment at At 30 degrees: 5.8% 30 degrees and At 45 degrees: 45 degrees of 6.1% knee extension

Measure Used

95%

70%

100%

75%

95%

NC

95%

NC

Sp (%)

TABLE 1 Summary of previous investigations concerned with establishment of cutoff scores for differentiating between maximal and nonmaximal efforts of the knee musculature

75%

100%

90%

90%

72%

NC

75%

NC

Sn (%)

www.ajpmr.com

Isokinetic Knee Strength Testing


173

32 (21 w), healthy, 25.2 T 4.7 yrs old



30 (13 m), healthy, 18Y37 yrs old

Knee extension12,d

Knee extension13,e

Knee extension14,f

Knee extension15,g

Biodex (model 2 testing days, unknown) dominant leg, isometric at 60-degree knee extension, 2 sets of 5 reps Isokinetic: 90-degree ROM, 1 set of 5 reps at 60 degrees per second, and 15 at 180 degrees per second. Randomized max and 50% nonmax Biodex (model 2 sessions, nondominant unknown) leg, 2 sets of 6 reps at 180 degrees per second through 90 degrees, max and 50% nonmax BTE Dynatrak Isometric at 60-degree knee extension (performed first); isotonic through 90-degree ROM, 2 sets of 3 reps (max and 50% nonmax), no visual feedback KinCom II 4 intermittent Con-Ecc reps through 70 degrees at 30 degrees per second and 180 degrees per second, mixed order, low velocity first, visual feedback provided 10.1% T 7.8% 11.3% T 4.4%

1.83 T 1.33

4.2% T 2.8% 3.2% T 2.3%

0.33 T 0.27

CV of isometric strength CV of isotonic: velocity

DEC

19.25% across both days

7.1 across both days

0.75

10%

12%

15% 10%

Isometric Isometric 11% 4.4% T 2.2% 10.0% T 5.5% Isokin 60 degrees Isokin 60 degrees 19% per second per second 19.3% T 9.2% 8.7% T 4.3% Isokin 180 degrees Isokin 180 degrees 10% 16.4% T 8.5% per second 5.6% T 2.1%

CV of average moment of middle 3 reps

CV of peak or average moment

75%

100%

93.3

90.0

73%

20%

(Continued next page)

100%

100%

55% 90%

53%

100%

100% 80%

38%

100%

174

Almosnino et al.



16 (m), healthy, 21Y30 yrs old

Knee extension16,h

Biodex System 2

KinCom II

Instrument 80 degrees/60 degrees (20Y80 degrees)/4 intermittent, reciprocal, Con-Ecc, max first, low velocity first, no feedback for nonmaximal efforts. 30:180 (1:6) 5 consecutive reps, 90-degree ROM at 60 degrees per second, max and nonmax (feign pain), bilaterally, random order

Protocol 50% effort 0.81 T 0.43 BREP 0.78 T 0.41 50% effort 0.83 T 0.40 BREP 0.90 T 0.83 NC

Right 0.3 T 0.23

DEC

Visual observation NC of strength curve consistency

Left 0.42 T 0.29

Mean T SD Nonmaximal

Mean T SD Maximal

Measure Used

NC

1.26

1.02

Proposed Cutoff Score

NC

NC

Sn (%)

87.5%Y92.5% 82.5%Y95.0%

NC

NC

Sp (%)

Sp = true negatives/(true negatives + false positives) 100; Sn = true positives/(true positives + false negatives) 10; ROC = receiver operating characteristic. a CV averaged across testing days. ROC curve provided for evaluation of different cutoffs. b Mean values derived from graphs in article. In addition, scores across speeds were combined and a fatigue index was used. c Full article not available to the authors’ disposal, hence the missing methodological details. d CV also reported for both peak and average moment. Best classification for isometric was based on peak moment CV, whereas best classification for isokinetic efforts was obtained using CV of average moment. Use of the CV of the moment-angular position slope was also reported. e Only CV of average moment reported in the current table. The investigation also reports CV of peak moment used across three and five repetition scores. Sensitivity and specificity values in the current table correspond to the so-called classification rate in the article. f Does not report what strength or velocity parameter is used in calculations. g Tested crossover effect of effort order (max vs. nonmax and vice versa). DEC calculation used windowed average force because use of peak force resulted in inferior results. Cutoff score is the one that optimized Sp and Sn within the sample. h Used three testing conditions of maximal, 50%, and best reproducible effort (BREP condition). i Specificity and sensitivity values correspond to the range of correctly classifying curves as being from maximal or feigned effort attempts across four examiners with a varying degree of experience with isokinetic testing. Con, concentric; DEC, difference between the eccentric and concentric ratios; Ecc, eccentric; Isokin, isokinetic; m, men; NA, not available to the authors’ disposal; NC, not explicitly calculated; nonmax, nonmaximal effort; reps, number of repetitions; Sn, sensitivity; Sp, specificity; w, women.

Knee extension 20, healthy, and flexion17,i physical education students

Participants (n, sex, age)

Joint/ Movement

TABLE 1 (Continued)

they are exerting a maximal effort.4 The next two sets involved performance of efforts at a self-perceived 50% and 75% maximal effort. Before the performance of each of these two sets, the participants performed a self-selected number of practice repetitions to estimate as best as they could the level of effort required. The order of sets was such that the maximal effort set was always performed first, the feigned effort set was always performed second, and the two sincere nonmaximal sets were performed in a random order. The set order was dictated by the authors’ wish that the participants were fully aware of their maximal capabilities before performance of feigned and sincere nonmaximal efforts. In addition, the sincere nonmaximal sets were performed after the feigned set because the authors did not wish to consciously evoke the participants into considering performance of a feigned effort resembling a sincere nonmaximal one (although the participants did in fact perform such efforts as part of warm-up). A rest period of 2Y3 mins was provided in between sets. During testing, both visual feedback of the concurrent moment-time curve and audible encouragement on performance were provided. Such feedback was not given in the initial study because of the possibility of influencing outcomes.15,16 However, because of the possible use of test scores to determine readiness to return to regular activities, feedback was provided because it has been found to have a favorable effect on strength production.28,29 The same two examiners provided all instructions and performed all measurements. Data were sampled at 100 Hz using the manufacturer’s proprietary software and stored for offline analysis. To blind the examiner performing the data analysis (S. Almosnino) of the actual effort performed, all header data containing set information were removed, and all files were labeled as maximal effort attempts.4 A second examiner (E.D. Diaconescu) kept a separate list of the actual efforts corresponding to each data file.

Data Analysis Moment data obtained between 5 degrees and 55 degrees were segmented into individual extension and flexion repetitions. Data within the initial and final 5 degrees were discarded to avoid possible moment artifacts caused by acceleration and deceleration of the lever arm at end ranges.30 Moment data from the first and last repetitions were also discarded because it has been previously suggested (1) that the first repetition may assume an uncharacteristic shape in comparison with subsequent repetitions because of warm-up effects27 and possibly because of initiation of the first www.ajpmr.com

extension contraction from a static starting position and (2) that the strength level of the last repetition may be relatively low in comparison with previous repetitions because of fatigue.22 Thereafter, Rxy and %RMSD scores were computed between pairs of curves. Rxy and %RMSD scores were calculated for repetitions 2Y5 in the same direction (i.e., extension and flexion), each by each (i.e., between curves 2Y3, 2Y4, 2Y5, 3Y4, 3Y5, and 4Y5). These were then averaged across directions, achieved for Rxy values, using a Z transformation routine,4 to yield representative consistency scores for each set.

Statistical Analysis For descriptive purposes, the authors extracted for each participant the mean peak moment attained during performance of knee extension or flexion efforts in each set and expressed these as a percentage of maximal efforts. For classification purposes, the individual Rxy and %RMSD score were inputted into the previously established prediction rules.4 In brief, these prediction rules were based on Rxy and %RMSD scores obtained from a mixed-sex sample of 37 healthy participants who were tested on two occasions and performed the same effort sets as in the current investigation (i.e., maximal, feigned, and sincere nonmaximal efforts at 50% and 75% of selfperceived maximum). The initial prediction rule was constructed using binary logistic regression, which has several advantages in comparison with other statistical classification methods (e.g., discriminant analysis), including being well suited for the binary classification problem posed (i.e., are efforts maximal or not?) and the ability to accommodate variables that are not normally distributed or have unequal variance in each group.31 The prediction rule was originally set such that no maximal effort would be classified as nonmaximal, thus maintaining 100% specificity. The authors were then interested in the performance of the prediction rule in the population represented by the sample of healthy participants. To achieve this, Monte Carlo simulations methods were used to estimate the cutoff scores that would be required to maintain 99.0% and 99.9% specificity in future samples with distributions similar to this study’s sample. In specifics, 1 million pairs of Rxy and %RMSD were simulated on the basis of their assumed distributions and with the same means, standard deviations, and correlations as those attained by the healthy sample during maximal effort production attempts. The P value was then Isokinetic Knee Strength Testing


175

manually adjusted in the initial logistic regressionbased prediction rule such that 99.0% and 99.9% specificity were reached on the simulated data. In effect, the P value adjustment in the initial investigation allowed to maintain 100% specificity while lowering the sensitivity percentage to a more realistic level.4 In the current investigation, the aforementioned P value adjustment ostensibly results in correctly predicting a larger number of maximal efforts (i.e., greater specificity percentage) at the expense of misclassifying a larger number of nonmaximal efforts as maximal (lower sensitivity percentage). It should be noted that the initial prediction rule was constructed using Rxy and %RMSD obtained from both healthy men and women because, on average, no significant differences in scores were found (not originally reported). In the current sample, separate independent, two-tailed t test was conducted to discern any differences in parameter scores between men and women (> G 0.05). The results indicated that no statistical differences are apparent between sexes for Rxy scores (men, 0.94 T 0.03, vs. women, 0.93 T 0.03; t[34] = 0.98; P = 0.33) and %RMSD scores (men, 7.52 T 2.05, vs. women, 8.16 T 2.51; t[34] = 0.81, P = 0.42). The results give support for use of the prediction models in the assessment of both men and women. Scores were also inputted into a model meant to classify effort as sincere or feigned. This particular model was deemed to be of little clinical utility because of exhibiting an unacceptably low sensitivity level.4 However, this model was used in the current investigation for exploratory purposes. Each individual outcome was assessed by comparison with the actual effort performed, as it appeared in the list kept by the second examiner. Performance of the prediction rules is reported in terms of number of misclassifications per effort condition and as sensitivity and specificity values expressed as percentages of the total number of tested participants.

Lastly, for knowledge attainment, the authors assessed group differences in set internal consistency performance between the ACL-reconstructed patients and those in the healthy sample used for construction of the prediction rules.4 This latter group consisted of 37 participants4 (21 men, age, 23 T 4 yrs; mass, 81 T 15 kg; height, 1.78 T 0.07 m; 15 women, age, 23 T 3 yrs; mass, 65 T 10 kg; height, 1.66 T 0.07 m). This was achieved by comparing maximal effort Rxy and %RMSD using separate independent, two-tailed t tests (> G 0.05) and complimenting with effect size estimates (Cohen d).

RESULTS Table 2 presents a summary of the Rxy and %RMSD scores obtained for each effort set, as well as the mean peak moment level expressed as the percentage of maximal effort values. During performance of the sincere nonmaximal efforts, the participants tended to underestimate the level of effort requested to be exerted. In addition, the level of strength exerted during attempts to feign deficiencies in muscular strength was, on average, lower than that exerted during performance of the two sincere nonmaximal conditions. However, a wide dispersion in relative strength is noted for the feigned condition, suggesting that although some participants attributed a substantial reduction of strength to be of benefit when attempting to feign deficiencies in muscular strength capabilities, others did not. Comparatively, Rxy scores of those who have undergone ACL reconstructive surgery (0.93 T 0.03; range, 0.86Y0.98) were significantly lower than those recorded for the healthy controls (0.96 T 0.02; range, 0.93Y0.99; t[71] = 5.634; P G 0.01; effect size = 0.13). In addition, %RMSD values of the ACLreconstructed participants were significantly higher (7.89 T 2.32; range, 4.11Y13.81) than those obtained by the healthy participants (6.63 T 1.77; range,

TABLE 2 Average peak moment scores (newton meter) for each effort level and percentage of actual effort performed at each submaximal and feigned sets, whereas accompanying data in parentheses are efforts expressed in relation to maximal efforts. Also shown are average cross-correlation (Rxy) and %RMSD for each effort Effort

Extension APM

Maximal 165 T 57 (100%) 75% 106 T 56 (67% T 18%) 50% 83 T 58 (48% T 20%) Feigned 78 T 66 (41% T 26%)

Rxy

Flexion APM 81 T 30 (100%) 51 T 29 (64% T 18%) 39 T 29 (44% T 18%) 40 T 37 (41% T 23%)

0.93 0.78 0.76 0.60

T 0.03 (0.86Y0.98) T 0.15 (0.34Y0.97) T 0.15 (0.45Y0.97) T 0.18 (0.26Y0.89)

%RMSD 7.89 T 2.32 (4.11Y13.81) 20.56 T 7.98 (6.63Y36.71) 26.50 T 16.35 (9.94Y76.97) 51.67 T 36.07 (12.12Y207.22)

Values are presented as group mean T standard deviation, with score range in parentheses. Data are pooled for the male and female participants. APM, average peak moment of repetitions 2Y5 (expressed as the percentage of maximal effort attempts).

176

Almosnino et al.



3.70Y11.50) (t[71] = j2.619, P = 0.011, ES = 0.62). Taken together, the results indicate that during performance of maximal isokinetic knee musculature efforts, moment curve shapes are less similar and relative curve magnitude differences are larger in those who have undergone ACL reconstructive surgery in comparison with those who have not had injury to the knee. In terms of prediction rule performance, these are presented in Table 3 and illustrated in Figures 1AYB. Alluding to the model that exhibited the best results for declaring a particular effort as maximal or not (model 3), of the total of 36 maximal effort attempts, 35 were classified correctly, whereas 1 effort attempt was misclassified. Of the total of 108 sincere nonmaximal and feigned efforts, 6 were misclassified as being maximal (four trials at the 50% effort condition and two trials at the 75% effort condition). No

feigned effort attempt was misclassified as being maximal. Of the six total nonmaximal misclassification cases, two belonged to the same individual. As such, when expressed on an individual participant case-by-case basis, the prediction rule’s corresponding specificity and sensitivity values are 97.2% and 94.4%, respectively. The two other prediction models, on the other hand, yield lower specificity values and higher sensitivity values (Table 3). In addition, the performance of the prediction rule meant for differentiating between sincere (either maximal or sincere nonmaximal efforts) and feigned efforts showed that 17 of 36 latter effort attempts were misclassified as being sincere efforts. In addition, of the 108 total sincere efforts, 5 (all performed at a self-perceived 50% of maximal effort) were misclassified as feigning attempts. The corresponding specificity and sensitivity values for the

TABLE 3 Performance of decision rules for differentiating between types and levels of isokinetic knee extension/flexion efforts exerted during testing of ACL-reconstructed participants No. Misclassifications by Effort Type or Level No.a 1

Description Cutoff set to maximize the sensitivity while maintaining 100% specificity in development sample of healthy participants. Cutoff set to maximize the sensitivity while maintaining 99.0% specificity in the Monte Carlo simulation study based on healthy samples outcome score distribution. Cutoff set to maximize the sensitivity while maintaining 99.9% specificity in the Monte Carlo simulation study based on healthy samples outcome score distribution. Cutoff set to maximize the sensitivity while maintaining 100% specificity in development sample of healthy participants.

2b

3b

4

Decision Rule

Sn

Sp

Overall

Max

50% 75% Feigned

Declare nonmaximal 97.2 66.6 effort if: 67.9(Rxy) j1.02 (%RMSD) G53.5

15/144

12/36 1/36 2/36

0/36


12/144

7/36

3/36 2/36

0/36


6/144

1/36

4/36 2/36

0/36

Declare feigned effort if: 0.16(Rxy) j2.1 (%RMSD) Q5.87

22/144

0/36

5/36 0/36

17/36

52.7 95.5

a

Models 1, 2, and 3 are meant at predicting the level of effort exerted (maximal or not). Model 4 is meant at predicting effort type (sincere or not). b A 1 million iteration Monte Carlo simulation was done to generalize results to new or larger samples of healthy participants. The difference in the decision rule between model 2 and model 3 relates to the specificity level attained (see description). Max, maximal effort; Sn, sensitivity (%); Sp, specificity.

www.ajpmr.com


177

FIGURE 1 A, Scatterplot of the average %RMSD scores vs. the average of the cross-correlation scores (Rxy) illustrating the performance of model 4 meant for predicting feigned efforts (ordinate values were truncated for improving figure resolution). B, Scatterplot focusing on the performance of models 1, 2, and 3, meant for predicting maximal efforts.

prediction rule meant at classifying effort type are 95.3% and 52.7%, respectively.

DISCUSSION This investigation was primarily concerned with validating previously established probabilitybased prediction rules meant for declaring efforts as being maximal or not during isokinetic testing of the thigh musculature. Because, to the authors’ knowledge, there are no clear guidelines as to what constitutes acceptable sensitivity and specificity levels, the authors evaluate the performance of the prediction rules presented in the current investigation by comparison with diagnostic tests that are currently used on a routine basis as well as compare the results of this study with those investigations concerned with differentiating between maximal and nonmaximal knee musculature efforts.

178

Almosnino et al.

First, the model that exhibited the highest sensitivity and specificity values outperforms clinical tests routinely used in those who have had injury to the ACL. For example, the authors report sensitivity and specificity values that are superior to those reported for the three most common clinical tests for diagnosing ACL rupture (i.e., the Lachman test, the Anterior Drawer test, and the Pivot-shift test).32,33 In addition, the test sensitivity and specificity values were attained even though this study’s sample of participants may be considered heterogeneous in terms of age, sex, type of surgical procedure performed, and time after surgical intervention. Hence, the prediction rule seems to exhibit some robustness to possible variations in scores because of such factors. However, it should be noted that several other aspects need to be further assessed. In specifics, it is unclear whether the prediction rule, which was in fact constructed with consideration of between-day



variations in scores of healthy participants,4 takes into account possible between-day variations in patient scores. That is, the precision of measurements of the outcomes used in this investigation needs to be established for those who have undergone ACL reconstruction. In addition, the prediction rule needs to be assessed in other samples of similar characteristics by different laboratories to assess the influence of examiner- and machine-induced variations in scores. On a related point, because the best-performing prediction rule was also the most lenient one in terms of accommodating less consistent performance during maximal effort attempts in those who underwent ACL reconstructive surgery, a natural question arises as to whether some of these efforts were in fact performed at a nonmaximal level. That is, if one of the more stringent prediction rules was used, then some of the efforts would have been classified as nonmaximal. The dilemma that arises is which prediction rule is most suited for clinical use. This query is admittedly difficult to answer and perhaps is a function of the importance of the test for decision-making purposes as well as the characteristics of the individual patient. In any case, if a particular effort is classified as nonmaximal, performance of a second set within the same test session is recommended to ascertain the classification achieved. The performance of a second set in such cases may be of value because it would perhaps alleviate a possible learning effect or psychologic factor that may have hindered maximal effort production in the initial set. When attempting to compare the sensitivity and specificity values attained in the current investigation with previous ones concerned with differentiating between maximal and nonmaximal efforts of the knee musculature (Table 1), several difficulties arise. First, with the exception of the study of David et al.15 and Dvir and David,16 all other investigations used performance of maximal efforts and only one level or type of nonmaximal effort (i.e., participants were asked to either feign muscular strength capabilities or perform at a predetermined percentage of self-perceived maximum). Second, with the exception of Ayalon et al.,17 all previous investigations used analyzed only knee extension moment output. Third, those investigations relying on a within-set variation parameter for differentiating between maximal and nonmaximal knee musculature efforts have primarily used the coefficient of variation (CV). Severe criticism has been raised toward the use of the CV for such purposes,1,34,35 criticisms that also extend toward its general use for assessment of variability.36 One limiwww.ajpmr.com

tation of the CV relates to the fact that, when examining its computational formula (i.e., the standard deviation divided by the mean value of a certain parameter, such as the peak moment), it is apparent that if setting the standard deviation at a constant value (i.e., absolute variability is unchanged), then small values of the mean will result in higher CV values.7,34,35 This in itself may introduce a bias against those that are weaker because of seemingly larger score variations.34 In this context, it should be noted that the outcome measures used in the current investigation, namely, the Rxy and %RMSD, are insensitive to the absolute level of strength exerted and thus eliminate the possibility of a potential calculation bias toward those that are inherently weaker. Lastly, it should be noted that all previous studies have used scores of only healthy participants for cutoff score development, and, to the authors’ knowledge, none have formally reported the performance of these cutoffs in those with or recovering from injury. Addressing practical testing considerations, the authors emphasize the need to provide participants with practice of maximal effort repetitions. This need arises when considering that in general, performance of maximal efforts within a prescribed ROM at a constant angular velocity is likely to be a novel movement experience for most participants. In addition, it has been documented that some of those recovering from ACL reconstructive surgery may be fearful of movement.21 As such, the authors contemplate that performance of maximal practice repetitions is necessary to minimize any potential learning effects as well as possibly alleviate psychologic hindrances. This latter point in itself is worthy of future investigations because it is unclear whether or to what extent routines used by clinicians to motivate participants to exert maximal efforts affect such psychologic constructs. Another practical consideration relates to the authors’ choices of testing ROM and angular velocity. With regard to the ROM, their decision may be considered conservative because others have reported isokinetic testing of those who have undergone ACL reconstruction through full knee extension within a span of 3Y32 mos after surgery.27 In addition, it has been suggested that performance at higher angular velocities may be beneficial during early rehabilitation.37 As such, establishment of prediction rules such as those presented in the current investigation for different ROMs and angular velocities is warranted. Although not the primary aim of this investigation, the results of this study continue to highlight deficiencies in the ability to declare muscular efforts Isokinetic Knee Strength Testing


179

of being a certain type (i.e., whether these were performed sincerely). The exploration of this issue is of importance because both sincere nonmaximal efforts and attempts to feign or exaggerate deficiencies of muscular capabilities may be encountered in the clinical realm.1Y3,23,24 One limitation of this study relates to testing of only the involved knee musculature. This was done because it was contemplated that possible neuromuscular control and strength deficiencies would be more pronounced on the involved side and thus would be particularly of interest for validation of the prediction rules previously established. However, bilateral strength comparisons are routine in the rehabilitative assessment of those who have undergone ACL reconstruction,18 and because the validity of strength scores attained from the uninjured limb is also dependent on exertions of maximal voluntary efforts, it would be of interest to assess the performance of the prediction rules in ascertaining such efforts.

CONCLUSIONS Using the prediction rules presented in this investigation, clinicians may be able to ascertain maximal effort production during isokinetic testing in those who have undergone surgical reconstruction of their ACL. Future investigation is necessary for establishment of the precision of measurements as well as development of prediction rules for testing performed at different ranges of knee motion and different angular velocities. Lastly, validation of the current prediction rules in other pathologic populations is warranted. REFERENCES 1. Dvir Z: Isokinetics: Muscle Testing, Interpretation and Clinical Applications, ed 2. Edinburgh, Scotland, Churchill Livingstone, 2004 2. Lechner DE, Bradbury SF, Bradley LA: Detecting sincerity of effort: A summary of methods and approaches. Phys Ther 1998;78:867Y88 3. Robinson ME, Dannecker EA: Critical issues in the use of muscle testing for the determination of sincerity of effort. Clin J Pain 2004;20:392Y8 4. Almosnino S, Stevenson JM, Day AG, et al: Differentiating between types and levels of knee musculature efforts using isokinetic dynamometry. J Electromyogr Kinesiol 2011;21:974Y81

180

7. Bohannon RW, Smith MB: Differentiation of maximal and submaximal knee extension efforts by isokinetic testing. Clin Biomech 1988;3:219Y21 8. Birmingham TB, Kramer JF, Speechley M, et al: Measurement variability and sincerity of effort: Clinical utility of isokinetic strength coefficient of variation scores. Ergonomics 1998;41:853Y63 9. Birmingham TB, Kramer JF: Identifying submaximal muscular effort: Reliability of difference scores calculated from isometric and isokinetic measurements. Percept Mot Skills 1998;87:1183Y91 10. Birmingham TB, Kramer JF, Speechley M, et al: Testretest reliability of the coefficient of variation as a measure of sincerity of effort during isometric testing. Physiother Can 1997;49:184Y90 11. Colombo R, Demaiti G, Sartorio F, et al: Sincerita` dello sforzo: Valutazione isocinetica nell’estensione del ginocchio. G Ital Med Lav Ergon 2008;30:370Y6 12. Lin PC, Robinson ME, Carlos J Jr, et al: Detection of submaximal effort in isometric and isokinetic knee extension tests. J Orthop Sports Phys Ther 1996;24:19Y24 13. Robinson ME, Sadler IJ, O’Connor PD, et al: Detection of submaximal effort and assessment of stability of the coefficient of variation. J Occup Rehabil 1997;7:207Y15 14. Robinson ME, O’Connor PD, Riley JL, et al: Variability of isometric and isotonic leg exercise: Utility for detection of submaximal effort. J Occup Rehabil 1994;4:163Y9 15. David G, Dvir Z, Mackintosh S, et al: Validity study of a novel test protocol for the identification of submaximal muscular effort. Isokinet Exerc Sci 1998;6:139Y44 16. Dvir Z, David G: Suboptimal muscular performance: Measuring isokinetic strength of knee extensors with new testing protocol. Arch Phys Med Rehabil 1996;77:578Y81 17. Ayalon M, Rubinstein M, Barak Y, et al: Identification of feigned strength test of the knee extensors and flexors based on the shape of the isokinetic torque curve. Isokinet Exerc Sci 2001;9:45Y50 18. Pua YH, Bryant AL, Steele JR, et al: Isokinetic dynamometry in anterior cruciate ligament injury and reconstruction. Ann Acad Med Singapore 2008;37: 330Y40 19. Bryant AL, Clark RA, Pua YH: Morphology of hamstring torque-time curves following ACL injury and reconstruction: Mechanisms and implications. J Orthop Res 2011;29:907Y14 20. Snyder-Mackler L, De Luca PF, Williams PR, et al: Reflex inhibition of the quadriceps femoris muscle after injury or reconstruction of the anterior cruciate ligament. J Bone Joint Surg Am 1994;76:555Y60

5. Kroemer KHE, Marras WS: Evaluation of maximal and submaximal static muscle exertions. Hum Factors 1981;23:643Y53

21. Kvist J, Ek A, Sporrstedt K, et al: Fear of re-injury: A hindrance for returning to sports after anterior cruciate ligament reconstruction. Knee Surg Sports Traumatol Arthrosc 2005;13:393Y7

6. Kroemer KHE, Marras WS: Towards an objective assessment of the Bmaximal voluntary contraction[ component in routine muscle strength measurements. Eur J Appl Physiol Occup Physiol 1980;45:1Y9

22. Eitzen I, Eitzen TJ, Holm I, et al: ACL deficient potential copers and non-copers reveal different isokinetic quadriceps strength profiles in the early stage after injury. Am J Sports Med 2010;38:586Y93

Almosnino et al.



23. Heil J: Specialized treatment approaches: Problems in rehabilitation, in Heil J (ed): Psychology of Sport Injury. Champaign, IL, Human Kinetics, 1993, pp. 195Y218 24. Rotella RJ, Ogilvie BC, Perrin DH: The malingering athlete: Psychological considerations, in Pargman D (ed): Psychological Bases of Sport Injuries, ed 2. Morgantown, WV, Fitness Information Technology Inc, 1999, pp. 111Y24 25. Serpas F, Yanagawa T, Pandy M: Forward-dynamics simulation of anterior cruciate ligament forces developed during isokinetic dynamometry. Comput Methods Biomech Biomed Engin 2002;5:33Y43

30. Schwartz FP, Bottaro M, Celes RS, et al: The influence of velocity overshoot movement artifact on isokinetic knee extension tests. J Sports Sci Med 2010;9:140Y6 31. Hosmer DW Jr, Lemeshow S, Sturdivant RX: Applied Logistic Regression. New York, NY, Wiley, 2013 32. Benjaminse A, Gokeler A, van der Schans CP: Clinical diagnosis of an anterior cruciate ligament rupture: A meta-analysis. J Orthop Sports Phys Ther 2006;36: 267Y88 33. Peeler J, Leiter J, MacDonald P: Accuracy and reliability of anterior cruciate ligament clinical examination in a multidisciplinary sports medicine setting. Clin J Sports Med 2010;20:80Y5

26. Yanagawa T, Shelburne K, Serpas F, et al: Effects of hamstrings muscle action on stability of the ACLdeficient knee in isokinetic extension exercise. Clin Biomech 2002;17:705Y12

34. Shechtman O: The coefficient of variation as a measure of sincerity of effort of grip strength, part I: The statistical principle. J Hand Ther 2001;14:180Y7

27. Ayalon M, Barak Y, Rubinstein M: Qualitative analysis of the isokinetic moment curve of the knee extensors. Isokinet Exerc Sci 2002;10:145Y51

35. Shechtman O: Using the coefficient of variation to detect sincerity of effort of grip strength: A literature review. J Hand Ther 2000;13:25Y32

28. Kim HJ, Kramer JF: Effectiveness of visual feedback during isokinetic exercise. J Orthop Sports Phys Ther 1997;26:318Y23

36. O’Dwyer N, Smith R, Halaki M, et al: Independent assessment of pattern and offset variability of time series waveforms. Gait Posture 2009;29:285Y9

29. O’Sullivan A, O’Sullivan K: The effect of combined visual feedback and verbal encouragement on isokinetic concentric performance in healthy females. Isokinet Exerc Sci 2008;16:47Y53

37. Yu¨ksel HY, Erkan S, Uzun M: Factors affecting isokinetic muscle strength before and after anterior cruciate ligament reconstruction. Acta Orthop Belg 2011;77:339Y48

www.ajpmr.com


181

RELATIONSHIP BETWEEN ISOKINETIC KNEE STRENGTH AND JUMP CHARACTERISTICS FOLLOWING ANTERIOR CRUCIATE LIGAMENT RECONSTRUCTION.

Comparison of knee laxity and isokinetic muscle strength in patients with a posterior cruciate ligament injury.

Reduced Maximal Force during Acute Anterior Knee Pain Is Associated with Deficits in Voluntary Muscle Activation.

A study of isokinetic strength and laxity with and without anterior cruciate ligament injury.

Isokinetic Identification of Knee Joint Torques before and after Anterior Cruciate Ligament Reconstruction.

Differences in Isokinetic Strength of the Knee Extensors and Flexors in Men With Isolated and Combined Cruciate-Ligament Knee Injury.

Core stability, knee muscle strength, and anterior translation are correlated with postural stability in anterior cruciate ligament-reconstructed patients.

Isokinetic training effect of ankle positions on knee extensor strength.

Evaluation of isokinetic and isometric strength measures for monitoring muscle function recovery after anterior cruciate ligament reconstruction.

Effect of knee angle on quadriceps strength and activation after anterior cruciate ligament reconstruction.

Obtaining reliable measurements of knee extensor torque produced during maximal voluntary contractions: an experimental investigation.

Association of Anterior Cruciate Ligament Width With Anterior Knee Laxity.

Pacing strategies during repeated maximal voluntary contractions.

Dynamic joint forces during knee isokinetic exercise.

Drop-Landing Performance and Knee-Extension Strength After Anterior Cruciate Ligament Reconstruction.

Anterior cruciate ligament injury after more than 20 years. II. Concentric and eccentric knee muscle strength.

Anterior cruciate ligament reconstruction and knee osteoarthritis.

Voluntary activation of human quadriceps during and after isokinetic exercise.

Strength deficits of the shoulder complex during isokinetic testing in people with chronic stroke.

Serial changes in knee muscle strength after anterior cruciate ligament reconstruction using hamstring tendon autografts.

Isokinetic angle-specific moments and ratios characterizing hamstring and quadriceps strength in anterior cruciate ligament deficient knees.

Neuromuscular function of the quadriceps muscle during isometric maximal, submaximal and submaximal fatiguing voluntary contractions in knee osteoarthrosis patients.

Effects of between-set interventions on neuromuscular function during isokinetic maximal concentric contractions of the knee extensors.

Psychological factors affecting isokinetic trunk strength testing in patients with work-related chronic low back pain.