The effectiveness and sustainability of supervised balance training in chronic ankle instability with grade III ligament injury: a one-year prospective study

From September 2018 to April 2019, 20 CAI patients who were diagnosed as grade III ligament injury and ready for rehabilitation were included in the study. A priori power analysis was completed using data from a previous study in which the researchers examined the effects of a similar balance-training program [12]. The study was approved by the IRB Medical Committee of our hospital (IRB00006761-M2019164) and the written consent was obtained from all patients. The study was registered in Chinese Clinical Trial Registry (ChiCTR), and the number was ChiCTR1900023999​.

The inclusion criteria were (i) aged 18 to 40 years, (ii) unilateral injury with a history of at least one significant lateral ankle sprain (at least 12 months prior to study enrolment) that caused inflammatory symptoms and disrupted activity, (iii) the most recent ankle sprain occurred > 3 months prior to study participation, (iv) reports of the previously injured joint “giving way” and/or recurrent sprain and/or “feelings of instability” (v) scoring < 24 on the Cumberland Ankle Instability Tool (CAIT) [13], (vi) the first time sprain was diagnosed as grade III [14, 15] injury of anterior talofibular ligament (ATFL) and/or calcaneofibular ligament (CFL) confirmed by both MRI and positive anterior drawer test (increased translation of 3 mm compared to the uninjured side or an absolute value of 10 mm of displacement) [16] and talar tilt test (10° of absolute talar tilt or 5° difference compared to the contralateral side) [17] by TELOS SD 900 Stress Device (Austin & Associates, inc. USA). All patients presented without a history of neurological or orthopedic impairment. Patients with combined intra-articular lesions (OCLs, osteophyte, impingement, loose body, etc.), any history of surgery, fracture requiring realignment and/or acute injury to the musculoskeletal structures (bone, joint structure and/or nerve) in either lower limb were excluded.

Upon enrollment, all the patients’ basic information was collected and evaluated, including gender, age, height, weight, involved side, pre-duration, sprain time and the Beighton score (Table 1). The Beighton score ≥ 4 was defined as the generalized joint hypermobility (gJHM). The postintervention data-collection session occurred within 48 h after the intervention ended. A follow-up session was performed 3 months, 6 months and 1 year since the pre-intervention data-collection session. Participants were instructed to cease all interventions during the follow-up session. During each data-collection session, we administered the patient-oriented outcomes (Foot and Ankle Ability Measure (FAAM), Cumberland Ankle Instability Tool (CAIT)) before evaluating the disease-oriented outcomes (isometric ankle strength, foot pressure and static and dynamic postural control).

The patients underwent three trials of single-limb stance on each leg with eyes closed on a force plate (AMTI; Watertown, MA, USA) for 10s [10, 18]. COP data were calculated from the three-dimensional force and moment signals and sampled at a rate of 50 Hz¹⁸. Subjects were instructed to stand as still as possible during testing with arms folded across their chests, holding the opposite limb at approximately 45° of knee flexion and 30° of hip flexion in accordance. If a subject touched down onto the ground, contacted the stance limb, or was unable to maintain standing posture during the 10-s trial, the trial was terminated and repeated. TTB measures and COP measures were computed separately in the ML and the AP directions using previously described methods and smaller TTB and larger COP indicated shorter time to reach the balance boundaries and more risk to fall [10, 18].

Following the standard calibration trial, the subjects were asked to walk barefoot at a self-chosen, comfortable, moderate velocity along the walkway. All subjects were allowed to familiarize themselves with the procedures before data collection. Subjects walked six times over the pressure plate (Footscan, RSscan International, Olen, Belgium) with 120 Hz sampling rate [19]. Three valid left and three valid right stance phases were measured. A trial was considered to be valid when the following criteria were met: stance phase registered as a heel strike pattern and no visual adjustments in step length or step frequency to aim on the pressure plate. Each print consisted of a time peak-force curve for eight regions of interest on the foot. The regions of interest, which were analyzed automatically by the system software, were medial heel (HM), lateral heel (HL), 1st to 5th metatarsal heads (M1 ~ M5), and toes (T1). The time variables were calculated as the ratio of time from the start of the stance to peak force under the region of interest and the total stance time (%). The peak force variables were calculated as the ratio of the peak force under the region of interest and were normalized by the body weight (N/kg) [20].

As described in TW Kaminski’s research [21], isokinetic strength was assessed with a Biodex isokinetic dynamometer (Biodex Medical Systems Inc., Shirley, NY). Each subject’s foot was securely fastened on the Biodex chair, with the hip angle 80° flexion (0° neutral position) and 20°to 30° of knee flexion. Each subject was allowed three submaximal (50% capacity) warm-up repetitions at each velocity to become familiar with the isokinetic test procedure, then performed three maximal concentric test repetitions at 60 and 120°/s on both ankles. The resting interval was approximately one minute between tests for each motion, velocity, and side. At the end of testing, peak torque data were extracted from the torque curves.

As shown in Table 2, the detailed balance training protocol was designed based on the widely used protocol from previously published papers [22, 23]. The protocol includes single-legged stance, wobble board, resistant band and hop exercises. All participants underwent the 3-month balance training intervention. The progressive balance-training program consisted of 24 supervised training sessions (two 60-min sessions per week).

Based on the ankle sprain recurrence after the balance training program through one year, the patients were divided into the sprain recurrence group and control group. The pre-training variables for each patient was used to analyze the risk factor for the sprain recurrence after the balance training.

The self-reported function (FAAM and CAIT), postural control measures (TTB measures and COP-based measures, plantar pressure measures) were analyzed separately at pre-training, post-training, 3 months, 6 months and 1 year. Paired sample t test was used to assess changes in the dependent measures before and after balance training. Shapiro-Wilk test was used to assess normality of data. Group comparison was only conducted after the formation of both groups (1 year). Univariate analysis and multivariable logistic regression models were used to explore the risk factors of sprain recurrence after the balance training, alpha level was set a priori at p < 0.05. An a priori power analysis was completed using data from a previous study [23] in which the researchers examined the effects of a similar balance-training program. Based on an α level of .05, a power of 0.95, and an effect size of 0.97 determined by the FAAM-Sport, 16 participants were needed. Therefore, we enrolled 20 participants to account for up to 20% attrition.

The changes of self-reported function questionnaires are presented in the Fig. 1. Compared with the pre-training values there was a significant increase for the FAAM ADL (p < 0.05) and FAAM Sport (p < 0.05) and CAIT (p < 0.05) at all follow-up timepoints. Numbers of recurrent sprains decreased from 3.29 ± 2.87 to 1.63 ± 2.02 at one-year follow-up. However, there were four patients (20%) that still reported a sense of instability with at least once sprain recurrence while other 16 patients (80%) patients went back to pre-injury sport and general life.

During walking, a significant increase in the peak force under the M2 (2.4 ± 1.2 vs 3.2 ± 1.9 N/kg, T = 3.79, p = 0.011) and HM (3.2 ± 0.9 vs 4.0 ± 1.2 N/kg, T = 3.99, p = 0.034) was observed at 3 months follow-up, and the changes of peak force in HM (3.6 ± 0.9 vs 5.0 ± 1.7, T = 4.03, p = 0.032) lasted until 6 months (Fig. 2. a, b). For the time to peak force (TPF), there was a significant decrease in M2 (79.5 ± 8.5 vs 72.3 ± 7.9%, T = 2.98, p = 0.031) and increase in HM (12.6 ± 4.6 vs 19.1 ± 4.9%, T = -2.65, p = 0.047) only at 3 months follow-up. No other differences were found at 6 months follow-up compared with the pre-training value. The foot pressure distribution changes in a three-dimension model of two individuals from respective control group and recurrence group are shown in the Fig. 3.

During single leg standing, TTB measurements were found to be significantly different with the pre-training level after 3-month training (Fig. 2. c, d). After 3 months’ balance training program, TTBAPa (T = 3.58, p = 0.02) and TTBMLa (T = 3.36, p = 0.02) temporarily increased and then decreased to the pre-training level while TTBAPm and TTBMLm showed continued improvement at 6 months and one-year (p < 0.05). There were no significant differences for the COP measures at all time points (p > 0.05).

As shown in Fig. 4, all the isokinetic contraction strength was significantly improved after 3 months of balance training (p < 0.05). Continuously significant increase in the 120°/s, 60°/s PF and 120°/s IV strength was observed at the one year follow up (p < 0.05), but the dorsiflexion and eversion muscle strength decreased to the pre-training level (p > 0.05) after six months until one year. Besides, 60°/s inversion strength decreased slightly after one year.

To preliminarily explore the potential risk factors of recurrence, the univariate analysis and multivariable logistic regression model were used. Based on whether the scores < 24 CAIT or there was an obvious ankle sprain during the follow up, the patients were divided into the sprain recurrence group (n = 4) and the control group (n = 16). The baseline values in those two type individuals were analyzed with univariate analysis. Significant differences were found in the 60°/s inversion strength, Beighton scores, peak force under first metatarsal, medial hindfoot and lateral hindfoot (Appendix file 1, all p < 0.05). Using univariate analysis results (variables p < 0.1) and forward selection to create a multivariable logistic regression model, we found that the lower 60°/s isokinetic inversion muscle strength (OR 1.46; 95% CI, 1.23–1.98; p = 0.012) and a higher Beighton score (OR 1.27; 95% CI, 1.03–1.32; p = 0.032) could be the risk factors (Table 3).