Visual-Motor Control of Drop Landing After Anterior Cruciate Ligament Reconstruction

We recruited participants from the university community and used an online survey to determine whether volunteers met the inclusion criteria: a minimum score of 5 on the Tegner Activity Scale and engagement at least once a week in a running or cutting/change-of-direction activity on the Marx Activity Scale. The participants with ACLR were individually matched to participants serving as healthy controls by age; sex; height; mass; upper and lower extremity dominance; history of and current physical activity level, including sport participation; and education level. Of the 502 individuals screened, 30 (15 with ACLR, 15 controls) fit the criteria and satisfied the matching requirements. We excluded volunteers with a history of other lower extremity injuries. The participants with ACLR (7 men, 8 women; age = 21.41 ± 2.60 years, height = 1.72 ± 0.09 m, mass = 69.24 ± 15.24 kg, Tegner Activity Scale score = 7.30 ± 1.30, time since surgery = 36.18 ± 26.50 months, hamstrings grafts = 13, patellar tendon grafts = 2) had undergone reconstruction 6 months to 5 years before the study, had been cleared for full return to activity by their physicians, and were engaged in regular physical activity. Control participants (7 men, 8 women; age = 23.15 ± 3.48 years, height = 1.73 ± 0.09 m, mass = 69.98 ± 14.83 kg, Tegner Activity Scale score = 6.77 ± 1.48) had no history of lower extremity injury. We observed no demographic differences between groups. All participants provided written informed consent, and the study was approved by The Ohio State University Institutional Review Board.

A 3-dimensional passive motion-capture system (model MX-F40; Vicon, Los Angeles, CA) with 10 cameras and the point-cluster marker technique^33–35  and two 40- × 60-cm force plates (Bertec Corp, Columbus, OH) were used to capture the kinematics and kinetics of all participants. For the DVJ assessment, participants fell forward from a 30-cm box, immediately performed a vertical jump, raised both upper extremities, and hit a target (Vertec Power Systems, Knoxville, TN) set at 90% of their maximal jump height.³⁶  The primary biomechanical outcomes collected were sagittal- and frontal-plane knee excursions and peak moments from initial contact to peak stance (peak knee flexion) during landing. These outcomes are highly reliable and constitute commonly used standard analyses for knee control and predictors for primary and secondary ACL injury risks.^36–41  We selected joint-angle excursion rather than focusing on initial contact or peak angle because researchers^31,42,43  have reported that knee mechanics at initial contact were stable but the landing movement pattern from contact to peak stance was altered when visual feedback was modified.

The DVJ was completed under 3 conditions: full vision, low SVFD, and high SVFD. Three trials under each condition were completed and averaged for each participant. Before the SVFD conditions, individuals completed an accommodation protocol consisting of a 5-minute ball toss during which the rate of SVFD increased after each set of 5 successful catches to allow accommodation to the SVFD and to limit effects due to novelty.^44,45  A minimum of 2 practice DVJs were completed before each condition.

The SPARQ Vapor Strobe goggles (Nike, Inc, Beaverton, OR) imposed the SVFD condition. These goggles, which are similar to sunglasses, have a wraparound strap for a secure fit and lenses constructed with battery-powered liquid crystal displays. They do not block vision continuously but strobe to block vision for milliseconds at a time. The length and frequency of these periods of “lost” and “intact” vision can be customized to 8 levels, with constant 100-millisecond transparent and 50- to 900-millisecond opaque periods. We used 2 levels of disruption: low (100-millisecond opaque, 100-millisecond transparent) and high (250-millisecond opaque, 100-millisecond transparent). In our pilot testing, visual-disruption levels greater than 250 milliseconds resulted in complete loss of vision during the DVJ and great difficulty in hitting the in-air target. These glasses have been used to improve reaction time, visual processing, and anticipatory ability, but they have not been assessed for their influence on lower extremity neuromuscular control.^44,45 

Data analysis was performed in Visual3D (version 5.0; C-Motion Inc, Germantown, MD) and MATLAB (version R2013B; The MathWorks Inc, Natick, MA). Initial contact of each limb was defined as the point when the vertical ground reaction force first exceeded 20 N. The landing phase was defined as the period from initial contact to peak knee flexion. The mean of 3 trials for each condition was used for statistical analysis.

We used a low-pass Butterworth filter with a cutoff frequency of 15 Hz to filter marker trajectories. Using a previously described method,⁴⁶  we estimated hip-joint centers to improve joint-center approximation. The midpoint between the medial and lateral knee- and ankle-joint markers defined the joint centers for the knee and ankle, respectively. Knee flexion and adduction were described as positive values. The changes in adduction and flexion angle from initial contact to peak knee flexion were extracted for analysis. We used a low-pass Butterworth filter at a matched cutoff frequency of 15 Hz to filter force data. External (force acting on the body) knee-flexion and knee-abduction moments were described as positive values. Given that we observed no difference between SVFD conditions, an SVFD difference score was calculated by subtracting the peak mean of the 2 SVFD conditions from the mean of the full-vision trials. This SVFD effect was used for the following statistical analysis, Tables 1 and 2, and Figures 1 through 5.

Power analysis with pilot data from our laboratory and previous reports^47,48  in which similar methods were used to detect movement differences showed that a sample size of 14 per group was required to detect differences in our primary outcome variable of interest: knee excursion (2.4 ± 3.7).³⁸  To determine whether the SVFD condition altered knee mechanics beyond previously established within-session error, a 1-sample t test was used for the absolute SVFD effect (full-vision condition minus SVFD condition, as described) versus the previously established measurement error.^41,49  This 1-sample t test for joint excursion and moment was calculated relative to the previously established error threshold to determine whether the SVFD effects were beyond the error of the measure and typical variations in human performance.

A paired-samples t test between groups (ACLR, matched control) was conducted on the SVFD effect (full-vision condition minus SVFD condition, as described). A paired analysis was selected because the groups varied by sex, height, mass, sport, activity level, and limb dominance but were pairwise matched; therefore, for example, a collegiate-level soccer player was compared with a collegiate soccer player of the same level and position to limit between-subjects variability beyond the history of ACLR. In this way, we ensured that the comparison was between each matched pair (ACLR versus matched control), whereas the typical group analysis would compare the ACLR and control group means, which would be a major limitation for our design (heterogeneous groups but homogeneous pairs). This participant-level paired design was planned a priori because this investigation was completed in parallel with a neuroimaging study²⁰ ; given the large variations in participants' brain activity due to various experiential factors (eg, sport, activity, education), we completed a participant-level pairwise matched design and carried this design strength forward to this analysis. The P values are reported without correction because we were analyzing a novel intervention. A Pearson product moment correlation was completed on each variable of interest and time from surgery for the ACLR cohort to determine whether time since surgery influenced the data (Table 2). We used the guidelines of Cohen⁵⁰  to determine the strength of effect sizes: weak (<0.02), small (0.21–0.5), moderate (0.51–0.8), or large (>0.8). All statistical analyses were conducted using SPSS (version 24.0; IBM Corp, Armonk, NY). The α level was set a priori at P ≤ .05.

We observed no differences in any dependent variable between sides; therefore, we focused our analysis on the ACLR (always left in this case) and matched control knees. The mean and standard deviation for each variable of interest under each visual condition are presented in Table 3. The SVFD condition induced effects beyond within-session error for knee-flexion and knee-adduction excursion and ground reaction force but not for knee-flexion or knee-abduction moment (Table 1). Knee-flexion excursion increased more under the SVFD condition in participants with ACLR than in controls, but we observed no other kinematic or kinetic differences between groups (Table 2). Figures 1 through 5 display the SVFD effect for each participant with ACLR next to the matched control for each variable of interest.

The purpose of our study was to evaluate the effects of SVFD on landing mechanics. Our results suggest that this disruption to visual-motor processing can influence landing motor control.

The primary effect of SVFD was in the sagittal plane, where it induced altered knee excursion during landing. Our finding of increased knee flexion with visual disruption was contrary to reports^42,43  of decreased knee flexion when vision was fully removed in healthy participants. This is likely due to methodologic differences, including no vertical-jump component after landing⁴²  (our study involved a jump to an in-air target to better simulate sport), increased drop height (our study was set at 30.5 cm, whereas other researchers have used 50 cm^42,43  or variable heights, ranging from 20 to 80 cm⁵¹ ), and comparison of only full- and blind-vision conditions with no component of decreased visual input via stroboscopic glasses.^42,43  The limited effect of SVFD, as opposed to fully obscuring vision via a blindfold, may have allowed some of our participants to alter motor planning to accommodate the depressed visual feedback and not execute a stiffening strategy as previously reported.

The ACLR pairwise increase in knee-flexion excursion may indicate a specific visual-motor adaption associated with the injury or recovery process. However, some individuals in the control group demonstrated an ACLR-like adaptation in motor control by also increasing knee flexion. Eight participants with ACLR increased knee flexion above the typical error of 3.2°, whereas only 4 control participants experienced a similar change (Figure 1). Although such a small change in knee flexion (absolute mean change of 4°) may seem clinically unimportant, it is on a level similar to an in-flight perturbation, with a 3.1° change in knee flexion due to a 15% body mass lateral perturbation during the drop phase before ground contact.⁵²  It is also in line with the effects of an in-season neuromuscular injury-prevention training program, with changes in knee flexion of 3.1°,⁵³  a plyometric–jump-training program effect with a change of 3.0°,⁵⁴  and the immediate effects of a landing feedback intervention of 3.5°.⁵⁵  Altering knee flexion with SVFD on levels similar to the given examples and a large effect size (Table 2) indicate that visual-motor control may influence knee neuromuscular control on a clinically important level.

Knee frontal-plane excursion changed with the visual condition but was not influenced by injury history. The SVFD effect on knee-adduction angle was more consistent regardless of group, with most participants going into more abduction (knee-valgus) excursion during landing due to SVFD. Seven participants with ACLR increased knee-abduction excursion beyond the minimal detectable threshold of 0.9°, whereas only 4 matched control participants experienced a similar increase (Figure 2). The 1.98° absolute change in knee frontal-plane excursion due to disrupted visual feedback is also consistent with previous reports on the effect of a multiweek neuromuscular-training program (1.7° change),⁵³  a self-feedback intervention (0.96° change),⁵⁵  and an in-flight perturbation (2.31° change)⁵²  and, combined with a moderate effect size, indicates that SVFD may influence frontal-plane knee control generally, with no particular effect for ACLR (Table 2).

The absolute SVFD effect for peak vertical ground reaction force was 38.72% of body mass (large effect size) with no effect of ACLR history (Table 1). This is similar in magnitude to the effects of ACL injury-prevention programs that decrease vertical ground reaction force by 18% to 38%.⁵⁶  However, the responses to SVFD were more varied, with 18 individuals decreasing and 12 increasing peak vertical ground reaction forces (Figure 3). The SVFD effect on knee sagittal- and frontal-plane peak moments was not greater than the previously reported measurement error, and we observed no difference between the ACLR and control cohorts. We found it interesting that whereas overall, we observed no change in knee-flexion moment in the 1-sample t test for the absolute SVFD effect, 16 individuals experienced a change in knee-flexion moment greater than the typical error, with an almost even split between those with ACLR (n = 8) and controls (n = 7; Figure 4). This may warrant further investigation to discover the neuromuscular factors that caused these individuals to respond to the SVFD. The SVFD altered the knee-abduction moment beyond the typical error in only 1 participant with ACLR and no control participants, indicating that SVFD had little effect on knee-abduction moment (Figure 5).

Physical activity and athletic participation place high demands on the visual-motor system to maintain environmental interaction as well as neuromuscular control.^57,58  The alterations in knee mechanics under SVFD indicate that lower extremity landing movement control may be influenced by the amount of visual feedback. Our data further supported the role of visual-motor function on landing motor control; specifically, visual feedback may provide a larger input to sustain sagittal-plane neuromuscular control after ACL injury. It interested us that some individuals demonstrated a movement pattern more dependent on visual feedback for motor control, with larger changes during the SVFD condition, and others experienced little effect. Using SVFD may allow clinicians to replicate the visual-motor demands of the complex athletic environment before return to play.^59–61 

The ACLR cohort was heterogeneous relative to sport and time from surgery. However, despite the cohort's demographic variability, the response to visual disruption was still unique in some ways in the ACLR group. The group heterogeneity was mitigated by the matched-pair design and analysis, and we controlled for any sex, sport, activity-level, or limb-dominance differences between groups in the paired analysis because participants were matched on these key variables that may influence landing mechanics. We selected the study design a priori to scale the visual disruption from none to low and then to high to limit any effects of task novelty from the SVFD. By having participants always perform practice and full-vision test trials first, we ensured they had as much familiarity with the task as possible before their vision was disrupted to limit the effect of task novelty on performance. Simply wearing the stroboscopic glasses can be a sufficient challenge to the neuromuscular control system, and we wanted to avoid the combined challenge of a new task and visual disruption. Conducting the combined practice trials and full-vision trials first ensured a motor pattern that was as stable as possible before inducing visual disruption. In addition, we selected the progressive difficulty to mimic how this technology may be used clinically.

The lack of longitudinal data limited our ability to determine whether any alteration in visual-motor control was present before injury or partially induced by the trauma, surgery, and rehabilitation. Researchers^62–66  have noted altered knee neuromuscular control during visual-motor interactions in healthy active participants. Swanik et al⁶⁷  prospectively reported that decreased visual-processing ability increased the risk of experiencing a noncontact ACL injury. The role of visual-motor processing to facilitate preparation of the neuromuscular control system is imperative to sport function, whereby visual feedback must be handled with minimal preparation time.^58,68,69  The ACLR likely induces increased use of visual feedback to program motion, but such a motor-control strategy may also be present to a degree in healthy individuals and may be accentuated after injury.

The variability in the individual responses indicated that some participants may have had a relative visual-motor processing bias for landing motor control. Capturing other metrics of function, such as proprioception, strength, and psychological factors, may help answer why some participants responded to a greater degree to SVFD. Given that the DVJ is a bilateral task and brain visual processing is not lateralized, we did not detect an SVFD effect for side. More complex or unilateral tasks may be better able to detect variations in visual-motor control after ACL injury. Intervention studies with traditional neuromuscular-control training and visual-motor–focused neuromuscular training may provide a mechanistic understanding of how altering visual-motor processing ability influences landing mechanics.

Visual-feedback disruption via stroboscopic glasses altered bilateral landing kinematics. Recognizing the visual-motor implications of maintaining neuromuscular control may help clinicians mitigate injury risk beyond traditional measures.^68,69  Using a visual-disruption technology, such as stroboscopic glasses, provides an opportunity to supplement traditional interventions and may more closely mimic the cognitive stress of sport in the clinic.^59–61 

This study was funded by doctoral grants from the National Athletic Trainers' Association Foundation and the National Strength and Conditioning Association Foundation and by the Ohio State University School of Medicine (Dr Grooms and Dr Onate). We thank the members of the Ohio State University Movement Analysis and Performance Laboratory, specifically Michael McNally, for their technical expertise during data collection.