Introduction

After anterior cruciate ligament (ACL) injury, many individuals undergo reconstructive surgery to restore knee stability and function [30, 38], yet despite ACL reconstruction (ACLR), movement asymmetries exist at least 6 months to 1 year after ACLR [12, 16, 19, 20, 36, 41, 43] and may persist longer [36]. Movement asymmetries during gait are associated with the development of early osteoarthritis [4, 41]; thus, developing strategies to mitigate these asymmetries is an important area of research.

To improve movement asymmetries in individuals after ACL injuries, neuromuscular training programs have been suggested. One type of neuromuscular training is perturbation training [8, 13, 15, 25, 42], which consists of external perturbations applied by a therapist while the participant stands on an unstable surface (such as a roller board or rocker board) [8, 15, 25, 42]. Perturbation training, when applied preoperatively, improves gait asymmetries [8, 13, 20]. Moreover, an extended preoperative physical therapy program including perturbation training results in higher success rates and longer maintained functional status compared with control subjects with extended physical therapy but no perturbation training [15]. Unfortunately, improvements in gait asymmetry that occur from preoperative perturbation training are not retained postoperatively [16, 36]. Postoperative interventions beyond traditional physical therapy are likely needed to restore gait symmetry after ACLR given the prevalence of movement abnormalities that persist [6, 16, 27, 28, 36, 37, 40, 41].

Although previous studies show promise for the efficacy of perturbation training, the effect of postoperative perturbation training on movement patterns is unknown. To address this gap, we developed the Anterior Cruciate Ligament Specialized Post-Operative Return-to-Sports training protocol (ACL-SPORTS) [42] for athletes after ACLR and traditional physical therapy. The ACL-SPORTS training protocol consists of 10 progressive sessions of comprehensive strengthening, agility, and secondary prevention exercises (SAP) or this SAP protocol with the addition of perturbation training (SAP+PERT).

The purpose of this study was to evaluate the effect of two versions of the ACL-SPORTS training protocol on hip and knee gait mechanics in men 1 and 2 years after ACLR. Specifically, among men undergoing ACLR, we sought to compare SAP treatment with SAP+PERT with respect to (1) gait mechanics; and (2) elimination of gait asymmetries 1 and 2 years after ACLR.

Patients and Methods

A detailed description of the patients and methodology for this study may be found in Cummer et al. [1]. Briefly, we enrolled and randomized 40 male athletes (mean age ± SD at surgery: 23 ± 7 years) after unilateral ACLR (autograft = 27, allograft = 13) when they met the following criteria for enrollment: ≥ 12 weeks after ACLR, ≥ 80% quadriceps femoris muscle strength symmetry, minimal knee effusion, full ROM, no reports of pain, and completion of a running progression [3]. Participants were randomized to two treatment groups: SAP group (n = 20) and SAP+PERT group (n = 20) (Fig. 1). The SAP group received 10 training sessions (twice per week) of ACL injury prevention exercises, agility drills, and plyometric exercises, whereas the SAP+PERT group received 10 training sessions (twice per week) consisting of all these exercises plus perturbation training (ie, neuromuscular training requiring selective muscle activation in response to surface perturbations applied by a physical therapist) [15]. After completing training and achieving objective return-to-sports criteria [3, 18, 42] (≥ 90% quadriceps strength index, ≥ 90% limb symmetry on four single-leg hop tests [34], and ≥ 90% score on the Knee Outcome Survey-Activities of Daily Living Scale [26]), participants were cleared to begin a gradual, patient-specific return-to-sport progression; rehabilitation after this time was not standardized.

Fig. 1
figure 1

The CONSORT flow diagram outlines enrollment, allocation and intervention, testing time points, and followup.

Full size image

We analyzed participants’ walking patterns 1 and 2 years postoperatively using an eight-camera motion capture system (VICON, Oxford, UK) and embedded force platform (Bertec Corporation, Columbus, OH, USA). Thirty-nine retroreflective markers were placed on the bilateral lower extremities and pelvis before motion analysis testing. Participants then walked overground at a self-selected gait speed maintained to ± 5% across trials and time points. Kinematic data were captured at 120 Hz, whereas kinetic data were captured at 1080 Hz. Data were processed using commercial software (Visual3D; C-Motion, Germantown, MD, USA) and normalized to 100% of stance phase. Joint moments were calculated through inverse dynamics and normalized to body weight and height (Nm/Kg*m) to allow comparisons between participants [32].

Variables of interest included (1) sagittal and frontal plane hip and knee angles and moments at peak knee flexion angle (pKFA); (2) sagittal plane hip and knee angles and moments at peak knee extension angle (pKExtA); (3) sagittal plane hip and knee excursion during weight acceptance (ie, pKFA – initial contact); and (4) sagittal plane hip and knee excursion during midstance (ie, pKExtA – pKFA). Sagittal plane joint excursions during weight acceptance and midstance are of particular interest after ACLR given the role of the quadriceps muscle in eccentrically and concentrically controlling knee motion during these phases of the gait cycle.

Statistical Analyses

To compare the demographic characteristics of participants in each group (SAP versus SAP+PERT), we used Student’s t-tests and Pearson chi-square tests of proportions and odds ratios (Table 1). To compare gait biomechanical variables, we used 2 × 2 × 2 analysis of variance with three factors: (1) time (1 versus 2 years postoperatively); (2) group (SAP versus SAP+PERT); and (3) limb (uninvolved [UN] versus involved [INV]). Alpha was set at 0.05 a priori for all comparisons. We conducted post hoc t-tests and calculated mean differences (with 95% confidence intervals).

Table 1 Demographics and anthropometrics were similar for subjects in the SAP and SAP+PERT groups
Full size table

We also compared interlimb differences (UN – INV) with previously established minimal clinically important difference (MCID) [13] values to assess for meaningful interlimb asymmetries. MCID values are 3° for sagittal plane hip and knee kinematics during weight acceptance and 0.06 Nm/Kg*m and 0.04 Nm/Kg*m for sagittal plane hip and knee kinetics, respectively, at peak knee flexion angle [13]. Clinically meaningful interlimb asymmetry was defined as an absolute difference that met or exceeded the MCID [13] for hip angles, moments, and excursions and knee moments. Only smaller [13] involved (versus uninvolved) limb knee angles and excursions were deemed clinically meaningful given the prevalence of reduced knee flexion angles and excursions after ACLR. We computed the proportion of SAP and SAP+PERT athletes who walked with clinically meaningful interlimb asymmetry in sagittal plane hip and knee variables and compared these proportions using odds ratios.

Post Hoc Power Analysis

We conducted a post hoc power analysis on our primary outcome (ie, knee flexion angle) using the observed SDs (for peak knee flexion angle) and number of subjects per group. With 80% power and an α of 0.05, we could have detected a group difference of 3.5° for knee flexion angle. The previously established MCID for peak knee is 3° [13], which is only marginally smaller than what we were powered to detect. Moreover, none of our group differences for knee flexion angles or excursions at 1 or 2 years postoperatively exceeded 2.0°: the largest measured group difference for knee flexion angles/excursions in the involved limb was 1.8° (95% confidence interval [CI], −2.2° to 5.7°; p = 0.37) for pKFA 2 years postoperatively; the largest measured group difference for knee flexion angles/excursions in the uninvolved limb was −2.0° (95% CI, −5.0° to 1.0°; p = 0.18) for knee excursion during midstance 1 year postoperatively.

Results

Gait Mechanics

There were no differences between the SAP or SAP+PERT groups for the biomechanical gait variables. Athletes walked with similar sagittal plane hip and knee angles at both 1 and 2 years postoperatively in both their involved and uninvolved limbs (Table 2). Likewise, hip and knee excursions were similar across groups at both time points. Notably, the involved limb’s knee excursion during midstance for the SAP (mean ± SD: 1 year: 15° ± 5°; 2 years: 16° ± 5°) and SAP+PERT (1 year: 16° ± 5°; 2 years: 15° ± 4°) groups was similar between groups at both 1 year (mean difference: −1°; 95% CI, −5° to 2°; p = 0.49) and 2 years (mean difference: 1°; 95% CI, −2° to 4°; p = 0.54) postoperatively. Hip extension moments at peak knee flexion angle decreased from 1 year to 2 years postoperatively but did not differ between groups.

Table 2 Sagittal plane hip and knee angles at peak knee flexion angle did not differ between SAP and SAP+PERT groups for the involved or uninvolved limbs at 1 or 2 years after ACLR
Full size table

Gait Asymmetries

There were no differences between SAP and SAP+PERT training regarding the elimination of gait asymmetries 1 or 2 years after ACLR, and gait asymmetries persisted to a large degree in both groups at both time points. The majority of both SAP and SAP+PERT athletes walked with meaningful interlimb asymmetries for sagittal plane hip and knee moments at pKFA, knee excursions during weight acceptance and midstance, and hip excursion during midstance at 1 year postoperatively, but these proportions did not differ between groups (Table 3). Likewise, the proportion of athletes who walked with clinically meaningful hip and knee angles, moments, and excursions did not differ at 2 years after ACLR (Table 3).

Table 3 This table displays the proportions of SAP and SAP+PERT athletes who walked with clinically meaningful interlimb asymmetries for sagittal plane hip and knee angles and moments at 1- 2-years followup
Full size table

When comparing the interlimb difference in mean UN and INV limbs within each group, gait asymmetries existed in both SAP and SAP+PERT groups, but were more prevalent at 1 year versus 2 years postoperatively regardless of group (Table 4). Both groups walked with similar hip (Fig. 2A–B) but smaller knee excursions (Fig. 2C–D) during weight acceptance and clinically smaller hip (Fig. 3A–B) and smaller knee (Fig. 3C–D) excursions during midstance in the involved versus uninvolved limb. Post hoc t-tests revealed meaningful [13] interlimb mean differences at 1 year postoperatively for knee excursions during weight acceptance (SAP: UN 18° ± 3° versus INV 14° ± 4°, mean interlimb difference [95% CI]: 4° [1° to 7°], p = 0.004; SAP+PERT: UN 18° ± 3° versus INV 15° ± 4°, mean interlimb difference [95% CI]: 3° [1° to 5°], p = 0.007) and midstance (SAP: UN 19° ± 5° versus INV 15° ± 5°, mean interlimb difference [95% CI]: 4° [1° to 8°], p = 0.019; SAP+PERT: UN 21° ± 4° versus INV 16° ± 5°, mean interlimb difference [95% CI]: 5° [2° to 8°], p = 0.002). At 1 year postoperatively, SAP athletes walked with smaller hip excursion during midstance (SAP: UN 34° ± 6° versus INV 30° ± 6°, mean interlimb difference [95% CI]: 5° [0° to 9°], p = 0.031), whereas SAP+PERT athletes tended to walk with clinically smaller hip excursions during midstance (SAP+PERT: UN 34° ± 5° versus INV 31° ± 5°, mean interlimb difference [95% CI]: 3° [0° to 6°], p = 0.065). At 2 years followup, hip and knee excursions were clinically asymmetrical for only the SAP group’s hip (UN 33° ± 6° versus INV 29° ± 5°; mean interlimb difference: 3°; 95% CI, 0° to 7°; p = 0.083) and the SAP+PERT group’s knee (UN 18° ± 4° versus INV 15° ± 4°; mean interlimb difference: 3°; 95% CI, 0° to 6°; p = 0.024) during midstance. Pooling across groups, athletes walked with smaller peak knee flexion angles in their involved versus uninvolved limbs; they also walked with smaller sagittal plane hip and knee moments in their involved versus uninvolved limbs at peak knee extension angle.

Table 4 Sagittal plane interlimb asymmetries (determined by the difference in the mean values of the uninvolved and involved limbs for each group) exceeding MCID values [31] existed across both groups, as indicated by the check marks (√)*
Full size table
Fig. 2A–D
figure 2

These figures display the mean sagittal plane hip (A–B) and knee (C–D) excursions during weight acceptance at 1 (A, C) and 2 (B, D) years postoperatively (whiskers are SDs). Both SAP and SAP+PERT athletes walked with meaningfully smaller knee excursions during weight acceptance at 1 year postoperatively.

Full size image
Fig. 3A–D
figure 3

These figures display the mean sagittal plane hip (A–B) and knee (C–D) excursions during midstance at 1 (A, C) and 2 (B, D) years postoperatively (whiskers are SDs). Both SAP and SAP+PERT athletes walked with meaningfully smaller hip and knee excursions during midstance at 1 year postoperatively; however, at 2 years, only SAP athletes walked with meaningfully smaller hip excursions and only SAP+PERT athletes walked with meaningfully smaller knee excursions when comparing differences in mean values.

Full size image

Discussion

Movement asymmetries during walking are prevalent after ACL injury and reconstruction and may increase the risk of posttraumatic osteoarthritis (OA) early after ACLR [6, 41]. Previous research has found that a specialized type of neuromuscular training (ie, perturbation training) improves gait asymmetries and functional outcomes when applied to ACL-deficient patients [8, 10, 13, 15, 20]. However, movement asymmetries exist after ACLR even in individuals who had preoperative perturbation training and postoperative physical therapy [10, 20, 36]. Thus, additional training is likely needed to restore movement symmetry during gait. However, the effect of postoperative perturbation training on gait mechanics is previously unknown. Therefore, we conducted a randomized controlled trial to compare SAP with SAP+PERT training. We found no difference between SAP and SAP+PERT training on the gait mechanics or resolution of gait asymmetries in men after ACLR, and both groups walked with meaningful gait asymmetries 1 and 2 years postoperatively.

There are several limitations to consider when interpreting the results of our study. The primary limitation of this study is that we only included men in this analysis; thus, the effect of SAP and SAP+PERT training on the gait mechanics of women is unknown. We are currently recruiting our final female participants for this study and plan to analyze and report these outcomes once enrollment and testing are complete. We also did not include the pretraining and posttraining data from our male participants; thus, we do not know what their gait mechanics were before and immediately after SAP or SAP+PERT training, and the present study is unable to quantify the pre- to post-changes or immediate effects of SAP versus SAP+PERT training on gait mechanics. We will analyze the pretraining and posttraining time points for both men and women once all women have completed training. Additionally, all participants were Level I/II (that is, jumping, cutting, and pivoting) [9, 22] athletes; thus, it is unknown whether our findings are applicable to less athletic populations. However, the majority of individuals who tear their ACL do so in Level I/II [9, 22] sports. Our findings also may not apply to individuals with severe concomitant injuries (eg, large osteochondral defects, multiple ligament injury, previous ACL injury) given our exclusion criteria; by excluding these patients, however, we created a more homogenous cohort for randomization and analysis. We did not standardize surgical method (performed by 21 different surgeons); thus, we do not know what effect surgical technique may have had on our findings. However, because we randomized our subjects to treatment group, the findings should be more generalizable across surgical intervention. Finally, we did not analyze electromyographic data or use musculoskeletal modeling [5, 29] to estimate joint contact forces, which have implications for the development of OA [27, 41]. Further analysis is warranted.

Our findings suggest that SAP+PERT training does not alter gait mechanics in men 1 or 2 years after ACLR compared with SAP training alone. Although the authors are unaware of any prior study investigating the effect of postoperative SAP versus SAP+PERT training on gait mechanics after ACLR, previous studies have investigated preoperative perturbation training on gait mechanics [8, 13, 15, 20, 25]. In these prior studies, changes in gait from pre- to post-intervention occurred; however, these changes were more prevalent among women, who, as compared with men and healthy control subjects, demonstrated a more impaired gait strategy (including smaller knee flexion angles [8] and muscle activation imbalances [25]) before training [8, 13, 20, 25]. In contrast to the changes seen among ACL-deficient women, ACL-deficient men walked with similar gait patterns before and after preoperative perturbation training [13]. Moreover, men who received preoperative strength training and physical therapy with or without perturbation training walked with stable and persistent truncated knee excursions pre- and post-intervention (before surgery) and 6 months after ACLR [10]. Similarly, in another study by Risberg et al. [35], lower extremity biomechanics in ACL-injured patients were largely unchanged even after 20 rehabilitation sessions including neuromuscular and strength training: smaller knee excursions persisted in both walking and hopping despite improvements in functional outcomes. Current rehabilitation programs may not be effective at changing gait, which is an automatic activity that may be resistant to change [10, 14, 35]. Future work should investigate the pre- to post-intervention effects of SAP versus SAP+PERT training, these programs on the gait mechanics of women, and novel paradigms to improve gait mechanics after ACLR.

There were no differences between SAP and SAP+PERT groups regarding the elimination of gait asymmetries at 1 or 2 years postoperatively, and clinically meaningful asymmetries existed to a large degree in both groups at both time points. The presence of gait asymmetries among individuals after ACLR is consistent with previous work [12, 16, 19, 20, 36, 41, 43]. Previous work has shown that limb asymmetries are prevalent in the short and medium term after ACLR [12, 16, 19, 20, 36, 41, 43], persist in both ACL-injured and ACL-reconstructed athletes despite rehabilitation programs including preoperative perturbation training [10, 13] and postoperative strengthening and neuromuscular rehabilitation [35], and may continue up to 2 years postoperatively [36]. The present study corroborates these findings and underscores that limb asymmetries remain present 2 years postoperatively even among individuals who are well rehabilitated after ACLR and have returned to sports [1]. The present study adds to a growing body of rehabilitation paradigms that have not succeeded in restoring gait symmetry [10, 21, 35]. Notably, gait asymmetries (including smaller INV versus UN limb knee flexion angles) persisted among patients after ACLR even when towing a sled or wearing a weighted vest [21]. Interestingly, previous studies have typically evaluated gait mechanics in the short and medium term by comparing group means rather than the proportion of individuals who walk with clinically meaningful asymmetries, as was done in the present study. By comparing the proportions of individual athletes who walked asymmetrically (Table 3) with the presence of meaningful asymmetries in group means (Table 4), it is apparent that many athletes in both the SAP and SAP+PERT groups walked asymmetrically despite generally symmetric means when pooling limb data across groups. These findings suggest that meaningful asymmetries may be even more present than once thought, even among participants who were functioning at a high level. Further development of rehabilitation paradigms is likely necessary to restore gait symmetry after ACLR, but further investigation to identify an effective program is needed.

Our findings suggest that a postoperative strength, agility, and secondary prevention training program with or without perturbation training is not effective at ameliorating gait asymmetries in men 1 or 2 years after ACLR. Moreover, regardless of treatment group, meaningful interlimb asymmetries persisted during gait at both 1 and 2 years after ACLR. Interestingly, by comparing the proportion of athletes who walked with interlimb asymmetries as well as each group’s mean, our findings suggest that interlimb asymmetries may be even more ubiquitous when comparing individuals versus group means. Although impaired gait patterns may be more prevalent among those with poorer functional and clinical performance [2, 12, 17, 41, 43], gait impairments may be present even in the absence of functional or clinical deficits [24, 43]. Therefore, although neither of the rehabilitation paradigms tested in the present study (SAP or SAP+PERT) restored gait symmetry or altered gait mechanics 1 or 2 years after ACLR, there may be other benefits to these programs. Previous work suggests neuromuscular training programs may be efficacious in improving functional performance and patient-reported outcomes [7, 31, 33, 35, 39] and facilitating return to sport while lowering second injury risk [11, 23, 44]. Future work should not only further investigate the functional, clinical, and biomechanical outcomes of SAP versus SAP+PERT training and compare them with outcomes of other programs and no additional training, but also explore new interventions to improve gait mechanics and ameliorate gait asymmetry in athletes after ACLR.