Variability of concentric angle-specific isokinetic torque and impulse assessments of the knee extensors

Isokinetic assessments are frequently utilized in a variety of settings, including sports performance (Delextrat et al 2019), rehabilitation (Kannus and Yasuda 1992, Timmins et al 2015) and musculoskeletal and neuromuscular research (Bowers et al 2004, Yeung and Yeung 2008, Philippou et al 2009, Noorkoiv et al 2015, El-Ashker et al 2019). For example, determining the optimal angle (i.e. the angle of peak torque) via isokinetic assessments has been proposed as a proxy for changes in muscle fascicle length and injury prediction (Timmins et al 2015), as many injuries occur at, or near, full knee extension (Escamilla et al 2012, Lui et al 2017), and end-range of motion (ROM) strength is a strong indicator of recovery following anterior cruciate ligament reconstruction (Cavanaugh and Powers 2017). Additionally, the concentric optimal angle has been widely used as a non-invasive means of estimating exercise-induced muscle damage (Bowers et al 2004, Yeung and Yeung 2008, Philippou et al 2009). For instance, Philippou et al (2009) reported that 50 maximal eccentric knee extensions caused significant shifts towards longer quadriceps muscle lengths, lasting up to 16 d, using concentric optimal angle assessment. Similarly, Yeung and Yeung (2008) reported greater increases in quadriceps optimal angle and soreness, and reductions in peak torque, in the eccentric versus concentric limb following a bout of step-ups.

While the peak torque and the optimal angle have traditionally been the focus of isokinetic assessments, the optimal angle may have high variability and is a poor predictor of performance or injury risk (Timmins et al 2015). Alternatively, multiple-angle maximal voluntary isometric contractions allow for reliable and robust length-tension assessments and the quantification of the rate of force development and impulse (Oranchuk et al 2019c). Additionally, isometric assessments are commonly utilized to examine muscular activation and the variability of maximal and submaximal motor outputs (Sosnoff et al 2010, Ofori et al 2018, Lanza et al 2019). However, obtaining a full isometric length-tension profile requires an extensive number of contractions (Noorkoiv et al 2014, Oranchuk et al 2019b). For example, Noorkoiv et al (2014) evaluated knee-extension force at eight angles, totaling 16 maximal isometric contractions per limb, with one and two minutes of rest between contractions and joint-angles, respectively. While robust, the aforementioned evaluation may be excessively time-consuming (~23 minutes per limb after warmup), thereby limiting practicality, and potentially inducing a training stimulus (Green et al 2014).

A potential solution to the limitations with isometric length-tension assessments would be to implement angle-specific isokinetic (torque-angle) evaluations throughout the ROM). Additionally, the angular impulse may provide better diagnostic information to the practitioner, and therefore deliver more informative programming information for athletic and clinical populations. However, while concentric peak torque is highly reliable (Timmins et al 2015), little is known regarding the between-session variability of angle-specific torque outputs. For example, recent studies examining angle-specific torque reported no reliability data (Delextrat et al 2019, El-Ashker et al 2019), while others report the coefficient of variation (CV) (Kannus and Yasuda 1992) or intraclass correlation coefficient (ICC) (Arnold et al 1993) in isolation, which fail to provide a robust assessment of variability as the ICC is overly reliant on between-subject variability, whereas typical error of measure (TEM) and CV are minimally affected by this (Hopkins 2000, Prescott 2019). Finally, to the authors' knowledge, there are currently no reports outlining the variability of angle-specific impulse measures. Therefore, the primary purpose of this note was to determine isokinetic derived angle-specific concentric torque and angular impulse intersession variability, through the entire knee-joint ROM, and determine the need for familiarization.

Experimental design

Subjects

Testing procedures

Statistical analysis

Isokinetic peak torque, optimal angle, and angle-specific torque data are presented in table 1. Peak torque was found to have small intersession variability (ICC  =  0.94–0.95, CV  =  6.2–7.4%, TEM  =  0.22–0.26), whereas optimal angle was found to have moderate to large variability (ICC  =  0.58–0.64, CV  =  7.3%–8%, TEM  =  0.76–0.86). Angle-specific torque was observed to have small to moderate variability (ICC  =  0.79–0.94, CV  =  6.4%–14.1%, TEM  =  0.25–0.51) except for at the most extreme angles of 10° and 100° (ICC  =  0.57–0.83, CV  =  12.9%–42.9%, TEM  =  0.40–0.89).

The isokinetic angular impulse data can be observed in table 2. The variability of total impulse through the entire ROM was found to be small to moderate (ICC  =  0.93–0.97, CV  =  6.4%–8.1%, TEM  =  0.20–0.31). Like angle-specific torque, the variability of all angular impulse bands was found to be small to moderate (ICC  =  0.78–0.96, CV  =  6.6%–13.6%, TEM  =  0.27–0.53), except for the most extreme angles (100–90° and 20–10°) (ICC  =  0.75–0.85, CV  =  11%–16.5%, TEM  =  0.36–0.60).

A comprehensive understanding of the variability associated with isokinetic torque and angular impulse across multiple angles, in a homogenous resistance-trained population, was previously lacking. As such, the main purpose of the study was to examine the intersession variability of traditional isokinetic assessments (i.e. peak torque, optimal angle), and the potentially more consistent diagnostic metrics of angle specific torque and angular impulse. Our primary findings were as follows: (1) variability was minimal for isokinetic peak torque and total impulse throughout the quadriceps ROM; (2) optimal angle demonstrates moderate to large variability; and (3) angle-specific isokinetic torque and angular impulse were found to have small to moderate variability, except for the ends of ROM.

Minimal variability was associated with concentric peak torque (ICC  =  0.94–0.95, CV  =  6.3%–7.4%, TEM  =  0.22–0.26), enabling the use of this metric without necessarily reporting optimal angle (Noorkoiv et al 2015, Oranchuk et al 2019b). Interestingly, while the mean optimal angle was nearly identical between sessions, moderate to large viabilities were found (ICC  =  0.58–0.64, CV  =  7.3%–8.0%, TEM  =  0.76–0.86), likely due to between-subject changes in rank-order. The aforementioned result question the common practice of analyzing the optimal angle, especially as it is likely that clinical populations would have greater biological variability than the population of resistance-trained men used in this study. The much less studied angle-specific isokinetic torque was found to be small to moderate in variability at 90–20° of knee flexion (ICC  =  0.79–0.94, CV  =  6.4%–14.1%, TEM  =  0.25–0.51) while 100° and 10° were more variable (ICC  =  0.57–0.83, CV  =  12.9%–42.9%, TEM  =  0.40–0.89). These results suggest that even healthy, resistance-trained subjects may have difficulty in consistently initiating and completing isokinetic contractions. These findings are important to consider as peak anterior cruciate ligament loading typically occurs between 10–30° of knee flexion (Escamilla et al 2012). While a direct investigation is required, it would be expected that rehabilitative populations would have greater difficulty consistently initiating and completing contractions (Cavanaugh and Powers 2017), thus resulting in more variable torque and impulse outputs near end ROM. Therefore, isometric contractions, with lower movement variability, may prove a more reliable measurement of injured populations.

Variabilities reported here are similar to those of Arnold et al (1993) (ICC  =  0.83–0.87) and are more consistent when compared to Kannus and Yasuda (1992) (CV  =  35%–47%); although these authors report across two sessions at only 30°, 60° and 75°, and 15° and 75° of knee-flexion, respectfully. Furthermore, to our knowledge, this is the first paper reporting angle-specific angular impulse, the reliability of which was more consistent (mean ICC  =  0.84–0.88, CV  =  9.9%–11.8%, TEM  =  0.34–0.44) as compared to angle-specific torque (mean ICC  =  0.82–0.86, CV  =  11.7%–14.5%, TEM  =  0.36–0.45). Additionally, all angular impulse brackets held small to moderate variabilities between sessions two and three, suggesting that only a single familiarization session is required to confidently evaluate the entire length-tension relationship. Alternatively, while the variability of angle-specific torque was more consistent for sessions two-three (mean ICC  =  0.88, CV  =  9.9%, TEM  =  0.38) versus one-two (mean ICC  =  0.84, CV  =  11.8%, TEM  =  0.44), variation at 100° remained moderate to large, suggesting that more than one familiarization session may be required to adequately detect small changes at long muscle lengths. Thus, we recommend utilizing bracketed ranges over torque at a single point in the ROM when practitioners must be sure that a change has occurred, especially when evaluating the knee extensors near full extension or flexion.

While the primary aim of this paper was accomplished, it is important to note limitations and directions for future research. We did not account for limb dominance, which may have altered intersession variability. However, it is common practice to evaluate both limbs in research (e.g. within-participant parallel conditions) and clinical (e.g. inter-limb asymmetry) settings. A complex relationship between fatigue, adaptations, and performance exists. As such, it is possible that each testing session could have led to small adaptations and/or resulted in residual fatigue, potentially affecting our results. However, while familiarization effects can be conferred in fewer than 10 maximal contractions (Green et al 2014), it is unlikely that noticeable levels of fatigue remained by the second and third testing sessions as even high responders have been shown to recover upwards of 80% of maximal voluntary isometric contraction just four days following 50 maximal eccentric contractions in untrained participants (Hubal et al 2007).

This study only examined the knee extensors; therefore, muscle groups such as the knee flexors, or humeral rotators must be directly examined before researchers and practitioners can understand the utility of angle-specific torque or impulse measurements at other muscle groups and joints. Furthermore, the utility of assessing agonist/antagonist strength ratios via angle-specific torque or impulse analyses remains unknown. Similarly, studies examining the variability of submaximal and maximal motor-outputs and muscle activation during non-isometric contractions should be undertaken. Additionally, the reader needs to be cognizant that only males with substantial resistance training experience were recruited for this investigation. Thus, the response of untrained, youth, elderly, female and rehabilitative populations could be different. Future researchers will also need to determine the variability of eccentric contractions and contractions at different velocities to gain a full understanding of the utility of these measures in a comprehensive torque-angle profile. Most importantly, future researchers should determine the validity of angle specific torque and impulse to predict injury and monitor recovery processes. Finally, while precedence exists for the specific inference cut-offs in this article (Oranchuk et al 2019a, 2019c, Smith and Hopkins, 2011), it is important to note that consensus to such thresholds is not universal (Hopkins 2000, Prescott 2019). Consequently, readers may wish to apply their own inferences based on their specific contexts.

Isokinetic concentric torque and angular impulse have low to moderate variability at all but extreme joint angles. Additionally, angle-specific angular impulse intersession variability at long muscle length is relatively low, and therefore could be used instead of angle-specific torque when multiple familiarization sessions are not practical and/or different diagnostic information is sought. Depending on the methodological purpose, researchers and practitioners can utilize angle-specific concentric torque or angular impulse as a time-efficient evaluation of contractile and neuromuscular qualities when the quantification of the rate of force development is not required.