Comparing the Effects of Differential and Visuo-Motor Training on Functional Performance, Biomechanical, and Psychological Factors in Athletes after ACL-Reconstruction: A Randomized Controlled Trial

1. Introduction

Despite significant progressions in surgical procedures and suggestions to optimize rehabilitation, the short- and long-term outcomes after ACL reconstruction remain disappointing [1]. The risk of sustaining a second ACL injury is nearly 1 in 4 in athletes younger than 25 years returning to high-risk sports activity [2]. In addition, an ACL injury and reconstruction are unequivocally associated with the `development of knee osteoarthritis [3]. Moreover, an ACL injury can have detrimental psychological effects on the athlete too (self-efficacy, fear of movement/re-injury) [4].

Asymmetries in leg coordination are commonly observed during daily and sport activities after an ACL injury and/or following ACL-rupture (ACLR) [5,6,7,8,9]. Unfortunately, contemporary rehabilitation plans do not effectively target dysfunctional movement patterns and motor control [10]. Dysfunctional movement control has been linked with an augmented risk for ipsilateral or contralateral secondary injury and the evolution of premature beginning of knee osteoarthritis [11,12]. Based on the continued neuromuscular control deficits, it is apparent that traditional rehabilitation does not restore normal motor function in all patients after ACLR [13].

At the same time, most traditional rehabilitation approaches still rely heavily on so-called master teachings which are characterized by person-independent movement prototypes that must be performed correctly and repeatedly with error correction, albeit at a slowly increasing rate of movement as the recovery process progresses.

More recent rehabilitation approaches that include general principles of motor learning have been proposed [13] with more emphasis on the individuality and self-organization of movements [14]. These approaches pursue a more holistic strategy that is based on principles of system dynamics and neurophysiology, relies on the offer of increased noise in terms of more variations [15], and thereby fosters implicit learning [16], such as the differential learning (DL) [17,18]. Both general motor learning approaches have in common that they try to reduce the dominance of the working memory [19,20] and thus the activation of the frontal lobe [21]. While the implicit learning approach relies on the modification of instructions and staying below the capacity threshold of the working memory, the DL intends to overload the working memory by adding stochastic perturbations [22] or increased noise to the to-be-learned movement to trigger a qualitative change of frontal lobe activation [23,24]. A more specific approach, so far mainly related to rehabilitation after ACLR, that focuses more on visual perturbations is suggested with the visuo-motor training approach [25].

In more detail, the DL approach models the patient or learner as a complex adaptive system that uses the versatile stimulations of the action-apperception system through increased fluctuations in the surrounding of the movement to-be-learned to make the system instable. In its most extreme form, no repetition and no augmented feedback is given to the learner. By not giving the learner explicit information about a possible solution, a true self-organization process is initiated [25]. The increased noise, transmitted mainly by means of mechano-sensors (e.g., Vater-Pacini-, Merkel cells, Golgi-, Ruffini organs, muscle spindles), vestibular organs, and the visual system, leads to a broader spectrum of input signals to the neural networks of the different brain areas, allowing easier discovery of new and more effective activation and movement patterns [18,26]. Preference is given to the proprioceptive systems during the learning of a movement technique due to the highly parallel processing outside the visual and conscious control. This course of holistic action is in analogy with the noisy training of artificial neural networks for higher learning rates, whose principal working mechanisms were originally derived from the behavior of neurons [27]. As an alternative to the pedagogically enticing [28] but epistemologically problematic [29] constraints model, the measures that trigger the increased noise of interventions in DL are differentiated into internal and external. Due to the different time scales of adaptation the internal measures e.g. are further distinguished into rather metabolism- and emotion-related measures (e.g. varying fatigue, mood, …) on one side and a rather cognition oriented measure (e.g. problems to be solved) on the other side. The external measures are associated with varying e.g. equipment, clothes, landscape, obstacles, etc. This structure goes along with the differentiation of objective (external) and subjective (internal) information proposed from cybernetic pedagogy [25,30] and emphasizes the closer interconnection of the two internal subsystems. In addition to numerous studies in the field of sports [18], there are increasingly positive findings in the field of rehabilitation [31]; e.g. stroke patients [32], focal dystonia [33], and hip replacement [34].

Alternatively, the visuo-motor training (VMT) assumes a “more accurate feedforward motor control” by stronger emphasis on visuo-motor information during practice [35,36,37]. In a more specific context, the VMT was most recently proposed as a potential avenue to augment ACL rehabilitation to facilitate sensory reweighting (nervous system adjustment of relative sensory input/processing for motor control) by shifting the post-injury reliance stronger exclusion of the visual part during motor control to remaining proprioceptive inputs the joint capsule, other ligaments, muscle spindles [38]. In particular, the use of visual obstruction training aimed at sensitizing the esthetic and visual input during standard rehabilitation exercises [39]. VMT intends to reduce the dependence on vision by shifting neural processing towards proprioception and increasing the efficiency of visual processing [38]. This is mainly derived theoretically from studies suggesting neurophysiological changes in athletes after ACLR that include: (a) modifying visual input combined with altered sensorimotor processing, which may induce (b) increased visual and somatosensory processing to plan movement and maintain neuromuscular control and (c) increased cortical top-down motor control strategies [38].

To date, there is a paucity of studies comparing DL and VMT in patients after ACLR. Accordingly, this study aimed to compare the effect of the DL and VMT compared to a control group on functional performance, biomechanical, and psychological factors in athletes with a history of ACL reconstruction. We hypothesized that: (a) both intervention groups have superiority over the control group, and (b) that DL and VMT should achieve comparable improvements.

2. Materials and Methods

Study design

This study was a single monocentric assessor blind randomized controlled trial (RCT) prepared and reported following CONsolidated Standards of Reporting Trials (CONSORT) guideline [40], and all intervention conditions conformed to the Declaration of Helsinki [41]. This RCT was approved by the Ethics Committee of (blinded for submission) and retrospectively registered at UMIN-CTR (ID number: UMIN000047952). All patients provided written informed consent prior to enrollment.

Participants recruitment and eligibility criteria

Forty-five competitive male handball, volleyball, and basketball players with a primary ACL reconstruction, who completed conventional post-operative rehabilitation, voluntarily enrolled in a group training program. After assessment of the study’s criteria, they were randomly assigned into one of three groups: DL group (n = 15), VMT (n = 15), and control group (n = 15). The study was conducted in the biomechanics lab of the Kharazmi University, Iran, from March of 2020 to July of 2020. Athletes were enrolled through one of the authors, while a blinded investigator was responsible for the randomization.

The following eligibility criteria were applied:

Inclusion criteria were: Having a unilateral hamstring tendon autograft ACL reconstruction performed by the same surgeon 6–12 months prior to participation, athletes were required to exhibit no pain, no effusion, report pain-free knee active range of motion (via electro goniometer), achieved 80% or greater quadriceps strength index, [42] limb symmetry via a handle-held dynamometer, and complete all hop tests without pain and at an equivalent distance/rate of at least 80% of the contralateral limb. Moreover, they were cleared by their medical teams to resume sports participation [43,44].

Exclusion criteria were having a concomitant injury to another knee structure (e.g., medial collateral ligament, meniscus) or experiencing a post-operative re-injury [43].

Demographic and health data

At baseline, there were no significant differences (p > 0.05) in demographic data among the three intervention groups (Table 1) (age (p = 0.67), mass (p = 0.56), height (p = 0.71), Body Mass Index (P = 0.64), time since surgery (P = 0.39)). The same was valid between the groups at baseline for any of the dependent variables of interest (P ≥0.100), indicating that the groups were comparable in terms of initial anthropometry and injury history.

Randomization, allocation, and implementation

Concealed allocation was performed using a computer-generated http://randomizer.org/ (Social Psychology Network, Connecticut, USA) block randomized table of numbers (1, for the control group; 2, for the DL group; and 3, for the VMT group). The random numerical sequence was placed in sealed opaque envelopes in a box. According to the group assignment, another researcher (blinded to the baseline assessment) opened envelopes and proceeded with training.

Sample size calculation

A sample size estimate indicated that 15 athletes per group (15 total athletes) would provide adequate statistical power to detect a group-by-time interaction for moderate effect size (partial eta squared = 0.06) [45]. This determination was made based on biomechanical and joint position sense data. Using these data, an alpha of 0.05, a beta of 0.20, the effect size of ηp2 = 0.06, and assuming a correlation among repeated measures of 0.85 for our sample size estimate, we arrived at the total of 45 athletes needed. The value used for the correlation among repeated measures was based on the test-retest reliability reported for isokinetic testing. G*Power software was used for sample size estimation [17].

All athletes in the experimental groups participated in each scheduled training session (100% compliance). In addition, all athletes who completed baseline testing also returned for follow-up testing (Figure 1).

Flow Chart 1. Flow Diagram of the study. Abbreviations: DL: differential learning; VMT: visuo-motor training

Flow Chart 1. Flow Diagram of the study. Abbreviations: DL: differential learning; VMT: visuo-motor training

Procedures

Before participating, a licensed athletic trainer with five years of experience conducted a preliminary assessment (based on inclusion and exclusion criteria) to ensure that participation was safe for all included athletes.

Baseline and post-intervention were completed using two blinded assessors before and after the interventions. Athletes in the control group also completed baseline and post-intervention and performed their typical training regimen, like improving technique and sport-related skills over a similar period. The primary outcome was functional performance (triple hop test), while secondary outcomes were dynamic balance during Star Excursion Balance Test (SERT), biomechanics during single leg drop jump (knee flexion (KF), ankle dorsiflexion (AD), knee valgus (KV), and vertical ground reaction force (VGRF)), and kinesiophobia.

Athletes executed a standardized warm-up protocol, including double-leg squats (2 × 8 repetitions) and double-leg maximum jumps (2 × 5 repetitions), followed by calf-stretching with a straight and bent knee [46]. In addition, all athletes were asked to refrain from training, maintain a regular diet, and avoid smoking, caffeine, and alcohol for 24 h prior to testing sessions [46].

Each test was performed on the non-involved side and repeated on the involved side, except for descent to fatigue. For this test, the initial limb test to control the potential fatigue of the hip and core muscles may affect the function of the second limb.

Primary outcome measure

Functional performance (triple hop test)

The athletes completed three trials of a single-leg, horizontal, triple hop test. Athletes performed practice trials to become familiar with the task prior to testing. A standard tape positioned perpendicular to the starting line was used to measure the distance hopped. The single-leg triple hop task is a common and reliable (ICC= 0.93-0.98) functional performance task used for athletes with ACL reconstruction [47].

Secondary outcome measures

Dynamic balance (Star Excursion Balance Test)

The Star Excursion Balance Test (SEBT) as a reliable tool (ICC = 0.90) was used for measuring dynamic balance [48]. Athletes were asked to stand in the center of eight lines forming an eight-pointed star, reach the farthest possible direction with their non-stance limb while maintaining balance on one leg, and return to the center of the grid. Athletes were also told to keep their hands on their hips to standardize torso and limb movements. If the athletes removed their hands from their hips, lifted the stance foot from the floor, lost balance during the trial, or the toe-touch was heavy or prolonged, the trial was considered an error and repeated. The reach distance was normalized and expressed as a percentage of the maximum reaching leg’s length. Each direction was tested for three trials with 30 s rest between each trial and an one minute break between each direction. The represented SEBT for each direction was an average of the three trials [48,49].

Biomechanics during single-leg drop-landing task

For the single-legged drop-landing task, athletes were asked to have a single- injured legged standing position on a platform of 30.5 cm height placed next to the force plate. Athletes landed on a force plate with the same limb and then jumped upward as high as possible. Each athlete was allowed to try the landing task 3 times. Three trials were collected for each participant [50]. The mean of these three single-legged drop-landings was adopted for statistical analysis. No augmented feedback was given during data collection.

Data collection

Kinematic data were recorded at 250 Hz using a 6-camera Motion Analysis system (raptor E with associated Cortex software). Kinetic data were collected at 1500 Hz using an AMTI force plate (AMTI, Watertown, Massachusetts) synchronized with the motion capture system. Retroreflective markers were placed on anatomic landmarks of the pelvis and lower extremities following the Plug-in-Gait lower body marker set. A static calibration trial was conducted while athletes stood in the anatomical position. Following the static calibration trial, the athletes completed a standardized warm-up (running and jumping tasks).

Kinematic and kinetic data from the single-leg landing trials were low pass filtered using a 4th order, zero-lag, recursive Butterworth filter. A cutoff frequency of 15 Hz was used for the marker data, and a cutoff frequency of 50 Hz was adopted for the force data. Three-dimensional joint angles were calculated for the trunk, hip and knee using a XYZ Cardan sequence, which resulted in joint angles corresponding with flexion/extension, adduction/abduction, and internal/external rotation ((knee flexion (KF), ankle dorsiflexion (AD), knee valgus (KV)). Joint angles reflected the orientation of the local coordinate system of the distal segment relative to the local coordinate system of the proximal segment. All kinetic variables were identified during the first 100 ms following initial contact with the force plate. Loading rates were calculated by dividing the peak z-component of GRF by the time to peak force. All kinetic variables were normalized to body weight (BW) as appropriate. All data processing was performed using custom MATLAB scripts (The MathWorks, Inc., Natick, MA, USA) to extract kinetic and kinematics data during the initial landing phase of the single-legged landing task. The three trial mean was calculated for each of these aforementioned dependent variables.

Kinesiophobia (TSK)

Kinesiophobia was assessed using the Iranian version of the Tampa Scale of Kinesiophobia (TSK). The TSK has 11 items, with scores ranging from 11 to 44 points, and higher scores indicate greater pain-related fear of movement/re-injury. Therefore, the TSK-11 is a valid measure (ICC= 0.81) of fear of movement/re-injury in the later stages of rehabilitation after ACL reconstruction [45,51].

Interventions

Differential learning (DL)

Participants of the DL group executed their exercises for 8-weeks under the supervision of an experienced athletic trainer. Each week comprised three sessions of 25-30 minutes (odd days). The exercises were performed under different conditions (Table 2), including exercises in the dark, on the sand with shoes and without shoes, and with loud music [13]. Also, variations of the double-legged jump were included, e.g., double legged jump on a Bosu-ball, double-legged jump over an obstacle, and double-legged jump with an air target.

Visual-motor training (VMT)

Athletes of the VMT group also attended an 8-week training period, consisting of three sessions per week (even days), with every training session lasting approximately 25 and 30 minutes. VMT protocols were developed for stroboscopic intervention and integrated into regular training (Table 3) [52]. The VMT protocols consisted of tasks including a Tap-test, agility ladder drills, single-leg stance (on foam), vertical jumps, and squat jumps. In addition, an error scoring system with detailed criteria to assess behavioral performance while wearing stereoscopic glass (SG) is described in Table 4.

Clinicians should first verify that their athlete can perform all exercises successfully before incorporating Stereoscopic glass. Then clinicians may expose their athlete to Stereoscopic glass by beginning at the easiest difficulty level (highest frequency of fluctuation between transparent and opaque states). As their athlete improves performance behaviorally, clinicians may increase Stereoscopic glass difficulty to increase the visual-cognitive demand.

For VMT, athletes performed the protocol under stroboscopic conditions. Each week, the shutter glasses' settings (frequency [Hz] and duty cycle [%]) were adjusted to compensate for expected improvements in visuomotor performance and to avoid adaptation to a specific setting [52]. Moreover, athletes should refrain from looking at their feet and focusing on how to exercise while performing the movements.

Control

Athletes in the control group received no specific treatment and were encouraged to continue their typical training regimen, like improving technique and sport-related skills.

Statistical analysis

The Shapiro–Wilk and Levene's tests were conducted to evaluate the normality and homogeneity assumptions, and all the tested variables showed p > 0.05. Group demographics were compared using a one-way analysis of variance (ANOVA). To determine differences between the three groups and time (pretest and posttest) 3 × 2 repeated measures ANOVA was directed followed by post hoc comparison with Bonferroni correction [53]. Within group factor (pretest to posttest) as a main effect of time, and between group factor as a main effect of group were considered. Also, 95% confidence intervals (CI95%) were computed based on the adjusted group mean differences, and Cohen’s d effect size (ES) of 0.8, 0.5 and 0.2 were considered “large”, “moderate” and “small” [54]. Statistical significance was set at p ≤ 0.05. All data analyses were calculated by means of SPSS software version 22.

Results

Functional performance (triple hop test)

The triple hop test revealed a statistically significant group × time interaction (F2,42 =3.861; p= 0.029) and main effect of time (F2,42 =16.226; p= 0.001). There was only trend for a significant main effect of group (F2,42= 2.609; p=0,085). Statistically significant differences in triple hop test distance were found between DL and control (p=0.017, ES=1.18) groups. Post hoc test showed that DL (p=0.001, ES=1.15) and VMT groups (p=0.003, ES=0.90) have significant larger improvement than control group (Table 5).

Dynamic balance (SEBT)

Significant group × time interaction effects, main effect of time and main effect of group were found for the eight directions (p< 0.05). There was no significant between group differences in medial direction. At anterior (p=0.001, ES=3.74), antero-medial (p=0.001, ES=1.90), postero-medial (p=0.001, ES=2.70), posterior (p=0.016, ES=2.13), posterolateral (p=0.001, ES=2.75), lateral (p=0.001, ES=2.92) and antero-lateral (p=0.001, ES=2.64) directions significant differences were found between DL and control groups. Also, significant differences in anterior (p=0.001, ES=3.71), antero-medial (p=0.001, ES=1.87), postero-medial (p=0.001, ES=2.88), posterior (p=0.032, ES=2.05), postero-lateral (p=0.002, ES=2.45), lateral (p=0.008, ES=2.64) and antero-lateral (p=0.001, ES=2.35) directions were found between VMT and control groups. Differences between intervention groups were not statistically significant. Post hoc test showed that DL and VMT groups have significant larger improvements in anterior (p=0.001, ES=2.90; p=0.001, ES=3.40), anteromedial (p=0.001, ES=1.94; p=0.001, ES=1.86), medial (p=0.001, ES=1.94; p=0.001, ES=1.86), posteromedial (p=0.001, ES=2.70; p=0.001, ES=2.88), posterior (p=0.001, ES=2.39; p=0.001, ES=2.22), posterolateral (p=0.001, ES=2.92; p=0.001, ES=2.53), lateral (p=0.001, ES=2.78; p=0.001, ES=2.68), and anterolateral (p=0.001, ES=2.90; p=0.001, ES=2.37) directions (Table 6).

Biomechanics (hip and knee flexion, ankle dorsiflexion, knee valgus, and VGRF)

Significant group × time interaction effects, main effect of time and main effect of group were found for the hip and knee flexion, ankle dorsiflexion, knee valgus, and VGRF (p< 0.05). The main effect of group was not significant at hip flexion angle. Significant differences in hip (p=0.001, ES=1.13) and knee (p=0.001, ES=2.70) flexion, knee valgus (p=0.001, ES=4.74) and VGRF (p=0.001, ES=-2.05) were found between DL and control groups. Also, at knee flexion (p=0.001, ES=3.74), knee valgus (p=0.001, ES=3.67) and VGRF (p=0.001, ES=-2.40) significant differences were found between VMT and control groups. Differences between intervention groups (p > 0.05) were not statistically significant. Post hoc test showed that DL and VMT groups have significant increase in hip (p=0.001, ES=1.05; p=0.001, ES=0.78) and knee (p=0.001, ES=1.86; p=0.001, ES=2.95) flexion and ankle dorsiflexion (p=0.001, ES=1.68; p=0.001, ES=1.90) angles and significant decrease in knee valgus (p=0.001, ES=6.5; p=0.001, ES=4.24) and VGRF (p=0.001, ES=-3.16; p=0.001, ES=-1.02) (Table 7).

Kinesiophobia (TSK)

A statistically significant group × time interaction (F2,42 =6.154; p= 0.001), main effect of time (F2,42 =50.047; p= 0.001) and main effect of group (F2,42 =3.438; p= 0.029) was reported for the TSK test (p <0.001) (Table 5). Significant decrease in TSK test was found between DL and control (p=0.001, ES=-2.59) groups and between VMT and control groups (p=0.001, ES=-2.59). Differences between the DL and the VMT groups were not statistically significant. Post hoc test showed that DL (p=0.001, ES=-2.42) and VMT groups (p=0.003, ES=-2.64) have significant decrease in TSK test (Table 8).