Reliability of isokinetic dynamometry of the plantarflexors in knee flexion and extension

1.1 Position of testing

By utilising knee flexion on isokinetic dynamometers it is possible to reduce the torque capacity of Gastrocnemius and bias testing to Soleus force production [11]. Some of the Achilles tendinopathy studies have completed testing in varying degrees of knee flexion but with little explanation for why this was done [12, 13]. The rational for testing Soleus in relative isolation from Gastrocnemius relates to the muscles independent role in locomotion [14, 15]. During ground contact Soleus controls knee flexion whist Gastrocnemius opposes this action and acts to extend the knee. In later stance Soleus decelerates the leg and propels the trunk forwards [16], producing vertical forces approaching 8 times body weight [17]. Gastrocnemius in comparison acts to accelerate the leg and decelerate the trunk producing [16] forces of 3 times body weight [17]. Recent studies suggest that the Soleus weakness may be most associated with Achilles tendinopathy [18, 19]. Due to these new findings it is important to determine the reliability of testing in positions where both Gastrocnemius and Soleus are responsible for force output (extended knee) and positions where Gastrocnemius is inhibited, therefore testing the force capacity of Soleus (flexed knee).

1.2 Consideration of angular velocity

Runners are the highest risk group for developing Achilles tendinopathy with incidence rates varying from 7% up to 50% [20–23], as such they are commonly recruited into research projects. The typical angular velocity used in isokinetic research is normally the equivalent of moving the joint through its available range within one second, for the ankle the suggestion is often 45°/sec, however this has little application to running or walking where the typical angular velocity of the ankle is around 90°/sec [24–26] and the peak is closer to 200°/sec [26, 27].

1.3 Contraction mode

Much of the previous reliability studies only report on concentric muscle contraction modes [13, 28–32] Only 4 reliability studies have included eccentric function of the plantarflexors, none of these utilised angular velocities relevant to running (i.e. 90°/sec) [13, 33–35].

1.4 Parameters of measurement

Currently the most common measure reported appears to be peak torque (Nm) presented as best repetition data. Whilst this parameter may be of interest it fails to take into account the weight of the individual and as such may lack clinical relevance [12], especially for anti-gravity muscles like the plantarflexors. Whilst this parameter appears logically important no published studies have reported either the reliability or relevance of this factor.

The current isokinetic literature has failed to utilise muscle contraction modes (concentric/eccentric) that are completed at functionally relevant speeds (angular velocities) for runners, targeting muscles most likely involved in diseases affecting the Achilles tendon. Of the studies that have assessed the plantarflexors all use units that do not account for the weight of the individual [10, 36–39], this may significantly influence the relevance of these studies [12].

1.5 Aims

Primary: Determine the test- re-test relative reliability of the Cybex NORM^® isokinetic dynamometer when biasing plantarflexor output to Soleus (knee flexion) or Gastrocnemius and Soleus (knee extension) at angular velocities and muscle contraction modes relevant to human locomotion.

Secondary: Evaluate the reliability of force as a percentage of body weight (Nm/Kg*100).

2.1 Subjects

A sample of 37 healthy volunteers (mean±SD, age: 22.9±3.2, height: 175.7±8.6 cm, weight: 73.2±11.1 kg) was recruited [40]. 28 males and 9 females, with 34 right-foot dominant and 4 left-foot dominant, defined as the foot used to kick a ball [41]. The study included both males and females because this mimics the clinical studies previously completed.

Inclusion criteria:

Over 18 years old

Ethical approval was granted from the research ethics committee of the University of Leicester. Written informed consent was obtained from all participants prior to the commencement of the study

2.2 Experimental design

Subjects were tested twice, within a 3–7 day interval [42, 43]. This test interval is similar to several other studies assessing plantarflexor reliability [13, 34]. Each subject’s Plantarflexor muscle strength was assessed using a Cybex NORM® (CSMI, CA, USA) isokinetic dynamometer. The dynamometer was calibrated prior to starting each day of data collection and set up in accordance with the standard manufacture guidelines for positioning.

2.3 Test positions

Two test positions were used to measure Plantarflexor strength in participants. The first position maintained the knee in extension. This ensured that both the Gastrocnemius and Soleus were able to function [11, 44]. The second position involved knee flexion to 90 degrees. This position has been shown to inhibit the force generated by Gastrocnemius thereby testing Soleus force production [11, 45]. Each test position adhered to the manufacturer’s guidance on subject positioning. A handheld goniometer was used to measure and confirm joint angles. The lateral malleolus of the ankle joint was aligned with the rotational axis of the dynamometer lever arm. A neutral foot position of 0° (Plantargrade) was used as the starting point. The range of motion was defined as 20° of dorsi-flexion to 30° of Plantarflexion. Whilst this ROM appears larger than other studies it has already been shown to be suitable in healthy controls and studies using subjects with Achilles tendinopathy [2, 46–48]. Participants were advised to wear flat-soled shoes to limit movement of the foot during testing.

2.4 Test protocol

The test protocol involved bilateral testing in both knee positions. Subjects were positioned and checked by two investigators (MA and CK). The protocols were all standardised with a submaximal warmup exercise in each test position and at each speed, this was completed to ensure familiarisation with the test procedure for each subject. Prior to starting the test all participants were instructed to push and pull as hard and as fast as possible. No further verbal encouragement was given during the test; this was done to reduce the effect verbal encouragement may have on force output [49] and to standardise the test procedure.

Participants performed the test in the flexed-knee position initially, then the extended-knee position. This was determined after extensive pilot work and in order to establish a protocol for examination in a clinical population. The pilot data assessed for an order effect and found there to be none.

The first phase was the concentric test at 90°/sec, with 7 familiarisation manoeuvres prior to the 5 repetitions at maximal output. The second test consisted of 10 concentric repetitions at 225°/sec with 7 familiarisation manoeuvres. 10 repetitions were used as our preliminary testing showed peak torque output often occurred later than the 5th repetition [2] during the faster angular velocities. The third phase was an eccentric test at 90°/sec, this used 7 familiarisation exercises prior to the 5 repetition test. A 30 second rest time was used between each phase. Our pilot testing showed this to be sufficient for recovery of maximum force output.

Data was collected bilaterally for Plantarflexor: peak torque (Nm), work per repetition (Nm), and average power per repetition (watts). These parameters are expressed as the highest value across all the repetitions (best repetition). Best repetition data was used, the convention in isokinetic research is to use this parameter, as such it is the most commonly reported parameter. Data was also expressed as a percentage of the individual’s body weight. This equates to a total of 77 parameters to report. These measures were identified prior to testing as potential important parameters for the Plantarflexors. Peak torque, using best repetition data, is the most frequently reported and has been shown to be an important factor previously [2, 37].

2.5 Statistical analysis

Data from the Cybex NORM^® isokinetic dynamometer was exported from the computer program to Microsoft Excel and statistical analysis was performed using SPSS (v.20, SPSS Inc., Chicago, IL, USA). The test re-test reliability analysed within-subject results from the first and second visits. All the parameters measured are expressed as a mean and standard deviation (SD). An intra-class correlation coefficient (ICC) with a two-way random effects model of single measures was used to evaluate the relative reliability. Relative reliability concerns the degree to which individuals maintain their position in a repeated samples measure. Good reliability was evaluated as an ICC value greater than 0.7. This threshold is commonly reported in clinical research as a good reliability threshold [50] To test for random or systematic alterations in the results between the two testing points standard error of the mean differences were calculated with 95% confidence intervals. Bland Altman plots, with 95% limits of agreement, were produced for all tests. The Bland Altman graphs allow visual assessment for any systematic biases and variability to be assessed. In total there were 72 separate test comparisons to view for the Bland Altman graphs, they are therefore not included in this paper due to the over extensive nature of the graphical representation. The Minimal Detectable Change was also calculated. The MDC represents the amplitude of change required to exceed the measurement error between the two measures. Knowledge of the MDC is essential to determine if any change in performance could be attributed to an intervention or injury rather than within the normal error of the measurement. MDC was calculated as MDC 95 = SEM × 1 . 96 × 2 .(1)

The 95% confidence interval as 1.96 being the two-sided tabled Z value for 95% CI and √ 2 is the variance between the two measures

3.1 ICC values

ICC values can be observed in Tables 1–6 with all above the 0.7 threshold (0.70–0.87). Best repetition data expressed as a % BW mostly fell above the 0.7 threshold except when testing the left leg in knee flexion. Testing in knee flexion using 225°/s on the left leg (non-dominant) produced 3 results that fell below the 0.7 threshold (Peak torque/% BW 0.61, Work per repetition/% BW 0.62, average power per repetition/% BW 0.59). Eccentric 90°/sec completed on the left leg (non-dominant) fell below the 0.7 threshold twice, peak torque/% BW (ICC 0.68) and Average power per repetition/% BW (ICC 0.69). Importantly all of the tests completed in an extended knee position produced reliable results, ICC levels ranged from 0.71–0.87.

3.2 MDC

The overall MDC varied between 2.1–35.9. The maximum MDC relates to eccentric contraction modes represented as a % BW whilst the minimum was found in the concentric 225°/sec flexed knee test.

4.1 Test velocity

ICC values were lower for the concentric 225°/sec muscle contractions, most notably when performed on the left limb (non-dominant) with more ICC values below the 0.7 threshold. This suggests that speeds of 225°/sec may not be suitable for testing due to their reliability. However it is important to consider the parameter of force output reported, peak torque, work per rep and average power per rep were all >0.7 with good SEM, 95% CI and MDC. It was only 225°/sec expressed as a % BW that led to the lower ICC values, Peak torque presented as a % BW had less reliability compared to peak torque in Nm. It is unclear why this was the case but may be explained by the ratio of torque to bodyweight e.g. A small increase in torque for a person with low bodyweight will give a substantial change in the % BW recorded.

4.2 Flexed versus extended knee testing

A flexed-knee position was used to investigate the Soleus and an extended-knee position was used to investigate the combined force production of both Plantarflexors. This paper is the first to analyse both positions in an attempt to determine reliability. ICC values for the all best rep parameters were >0.7 when using the slower 90°/sec concentric or eccentric assessment except on two occasions the left leg fell below the 0.7 cut off when tested eccentrically in knee flexion and presented as a % BW.

The eccentric ICC values are similar to previous work, although there are fewer extremes in the range. The ICC for Peak Torque ranged from 0.68–0.87 for the eccentric tests compared to a previous study that ranged from 0.37–0.98 [34]. Our investigation involved a larger sample population and would appear to be more robust. The reliability of Work per rep (ICC = 0.62–85) and Average Power per rep (ICC = 0.59–0.85) were also good. The lower ICC values represent the flexed-knee phases of testing on the left leg and involve the parameter being reported as a % BW. Testing in knee extension was more reliable than testing in knee flexion especially at the 225°/sec and on the non-dominant limb (left).

4.3 Learning effect

There is a trend in the data towards a slightly higher force output on the second visit. The data shows a trend towards a higher result during the second visit. A number of factors may have influenced this such as motivation, familiarity with the procedure, or a process of learning. Given the timeframe between visits one can presume that a degree of learning may have taken place rather than physical adaptations [56]. Previous research has identified that when conducting a complex repetitive static movement the improvement measured is attributed to a learning process [56]. Conversely, other studies investigated the learning effect using isokinetic dynamometry and found that learning does not interfere with the results of testing protocols performed over more than two visits [34]. It is important to note these studies did not show increased force on their second visit [13], unlike ours. This may be due to the different sample size (n = 14–2 versus 37 in ours), the population group, or the muscle groups investigated. The relative reliability as measured by the ICC gives an indication of the position of each participants result relative to their second visit, this still maintains a good level as shown by our ICC values.

4.4 Minimal detectable change

This is the first study to determine the MDC for the plantarflexors. The results show relatively moderate MDC values. Changes with interventions need to be greater than the MDC to support an actual change in strength rather than a measurement error. The MDC found within this study is less than the observed strength change typically observed with clinical interventions [10, 46]. The MDC is also greater than the level typical observed when comparing limbs [19, 46]. If only female or male subjects were tested, as opposed to the mixed sex subjects we utilised it may reduce the MDC. However this study represents typical clinical studies that utilise mixed genders and different levels of symptoms often leading to higher standard deviations and therefore higher MDC values. It is critical that further studies consider the MDC when reporting differences between groups or adaptation to an intervention.