Diagnostic Evaluation of Mechanical Ankle Instability by Comparing Injured and Uninjured Contralateral Ankles Using Arthrometry

A cross-sectional study was conducted between February 2022 and August 2022 to quantitatively investigate the bilateral difference in ankle joint laxity between ankles in the MAI and control groups.

We recruited participants with MAI from our institution, where they were hospitalized and scheduled to undergo ankle arthroscopy after the experiment. All patients provided written informed consent for arthroscopic procedures. A total of 38 participants (age = 31.24 ± 7.90 years, height = 168.93 ± 7.69 cm, mass = 65.72 ± 10.47 kg) with unilateral MAI were enrolled (MAI group). All participants were confirmed to have impaired lateral ankle ligaments via arthroscopic examination (Figure 1). A total of 38 individuals without MAI (age = 32.10 ± 7.10 years, height = 166.59 ± 7.89 cm, mass = 62.93 ± 10.72 kg) were recruited from the local community through advertisements and were matched by age, sex, height, and mass with participants who had MAI (control group). All participants provided written informed consent, and the study was approved by the Huashan Institutional Review Board.

Inclusion and exclusion criteria were based on the criteria of the International Ankle Consortium¹⁰  and arthroscopic findings. For the MAI group, the following inclusion criteria were used: (1) age 18 to 50 years, (2) ≥1 episode of substantial unilateral ankle sprain sustained ≥12 months before recruitment, (3) ≥1 interrupted day of desired physical activity due to the ankle injury, (4) a history of ≥2 episodes of unilateral sprains or perceived instability or giving way in the 6 months before the study, (5) a Cumberland Ankle Instability Tool (CAIT) score of <24, and (6) direct visualization of lateral ligament injuries via ankle arthroscopy after arthrometric testing. For the control group, the following inclusion criteria were used: (1) no history of ankle injury, instability, or surgery; (2) normal ankle range of motion and muscle strength; and (3) a CAIT score of 29 or 30. The following exclusion criteria were used for both groups: (1) age not between 18 and 50 years, (2) a history of surgery or fracture in either lower extremity, or (3) a history of acute injury to the lower extremity within 3 months before enrollment.

We calculated sample size using the following formula: sample size = (Z_1−α/2)² (SD)²/d²,¹¹  where Z_1−α/2 is the standard normal variate, which equals 1.96 at 5% type I error; SD is the standard deviation of the variable taken from previous work,⁹  which was 1.566; and d is the absolute error or precision decided in the earlier study,⁹  which was set at 0.5 for this investigation. Thereafter, the sample size was calculated to be at least 37 ankles in each group.

Two researchers (Y.C. and S.C.) who were experienced in use of the arthrometer and were blinded to the conditions of participants, including group, injured side, and CAIT score, performed instrumented ADT using the Ligs Digital Arthrometer (Innomotion Software Technologies). All participants in the MAI and control groups completed the arthrometric testing in a treatment room. All data were transferred to a laptop computer and were processed by a third researcher (C.W.). In the MAI group, this test was performed 1 day before the arthroscopic procedure. All arthroscopic procedures were to be performed by a senior surgeon (X.W.), who was blinded to the results of the arthrometric testing. The arthrometer consisted of a motor unit and a sensor unit. A researcher (C.Z.) demonstrated the use of the arthrometer to participants and assisted them in maintaining the position of the tested lower limb. The knee was flexed to 90°, and the ankle was placed in a neutral position (0° of flexion angle) using a goniometer. The neutral position was measured from the plantar surface of the foot relative to the anterior tibia. With the participant’s calf and heel fixed, the motor unit pushed the anterior tibia posteriorly at a constant speed of 3 N/s until the maximum loading force of 75 N was reached. Loading force and joint-displacement data were recorded continuously at a frequency of 30 Hz by the sensor unit. To reduce the influence of the calf musculature, the recording started the moment after the load exceeded 10 N. The load is accurate to 1 N, and displacement is accurate to 0.1 mm.⁹ 

All participants underwent testing of both ankles. For the MAI group, the differences between the injured and uninjured ankles were calculated. For the control group, the differences between the left and right ankles were calculated and recorded as absolute values. Ankle arthrometer calculations were based on the average values of 3 consecutive measurements. The arthrometer has been shown to have excellent test-retest and intertester reliability.⁹ 

Traditionally, static measures are used to assess the laxity of the ankle joint by measuring the anterior translation of the talus with a maintained loading force applied by an arthrometer.^7,12  However, more recently, researchers^8,9  have suggested that dynamic measures using load-displacement curves might better reflect the difference in laxity between injured and uninjured ankles. Chen et al⁹  reported that the arthrometer could best distinguish an injured from an uninjured ankle using displacement at a loading force of 75 N (D75) as a static reference standard. When using dynamic measurement by depicting load-displacement curves, Chen et al showed that the difference in the load-displacement ratio (LDR) at 10 to 40 N (LDR 10–40) had the largest effect size (ES) between mechanically stable and unstable ankles (Figure 2).⁹  The LDR 10–40 was the slope of the linear regression between 10 and 40 N on the load-displacement curve. Therefore, D75 and LDR 10–40 were used for further analysis.

Characteristics of the 2 groups were compared using 2-tailed paired t tests. Differences in D75 and LDR 10–40 between bilateral ankles of participants in the 2 groups were calculated using a 2-tailed paired t test. We used subtraction to establish absolute differences between ankles in each participant and a 2-tailed between-subjects t test to examine between-groups absolute limb differences in D75 and LDR 10–40. The Cohen d was used to compute ESs, which were interpreted as small (0.20), medium (0.50), or large (0.80).¹³  A receiver operating characteristic (ROC) curve analysis was generated to identify the cutoff points that discriminated between the control and MAI groups for bilateral absolute differences in D75 and LDR 10–40. The upper left coordinate of each ROC curve was the cutoff point. The area under the ROC curve (AUC) value was calculated to reflect the diagnostic accuracy of D75 and LDR 10–40. Sensitivity and specificity were determined. The α level was set a priori at .05. Statistical analyses were performed using SPSS (version 24.0; IBM Corp).

Characteristics of the MAI and control groups are shown in Table 1. We observed differences in CAIT scores between the groups (t₃₇ = 10.96, P < .001).

In the MAI group, the mean D75 of injured ankles was 11.32 ± 1.93 mm, and the mean D75 of uninjured ankles was 8.03 ± 0.76 mm (t₃₇ = 9.78, P < .001, ES = 2.24; Table 2). The mean LDR 10–40 of injured ankles was higher than that of uninjured ankles (0.222 ± 0.046 mm/N versus 0.138 ± 0.026 mm/N, respectively; t₃₇ = 9.80, P < .001, ES = 2.25). In the control group, no differences were found between the left and right ankles in either D75 (t₃₇ = 0.92, P = .14) or LDR 10–40 (t₃₇ = 1.16, P = .07).

Absolute differences between each participant’s ankles were also calculated using subtraction. The difference in D75 was larger in the MAI group than in the control group (3.28 ± 1.72 mm versus 1.02 ± 0.81 mm, respectively; t₃₇ = 7.33, P < .001, ES = 1.68; Figure 3). The difference in LDR 10–40 was also larger in the MAI group than in the control group (0.084 ± 0.048 mm/N versus 0.018 ± 0.013 mm/N, respectively; t₃₇ = 8.18, P < .001, ES = 1.88; Figure 4).

The results of the ROC curve analysis of bilateral absolute differences in ankle D75 and LDR 10–40 are presented in Table 3. When the cutoff value of D75 was set at 1.85 mm, the highest AUC value was derived (0.914 [95% CI = 0.844, 0.984]), and the sensitivity and specificity were 0.800 and 0.933, respectively. For LDR 10–40, the highest AUC value was 0.959 (95% CI = 0.907, 1.000) when the cutoff value was set at 0.0351 mm/N, and the sensitivity and specificity were 0.900 and 0.933, respectively. The ROC curves are displayed in Figure 5.

Our most important finding was that ankle arthrometry had excellent diagnostic accuracy for MAI when comparing injured and uninjured contralateral ankles. Compared with static measurement using D75, dynamic measurement using LDR 10–40 provided superior diagnostic accuracy based on the higher sensitivity and specificity.

We observed that arthrometric stress testing had sensitivity and specificity of 0.900 and 0.933, respectively, when using LDR 10–40 for bilateral comparisons. A dynamic reference standard using LDR 10–40 was more sensitive than a static reference standard using D75 for diagnosing MAI (Table 3). The diagnostic accuracy of arthrometric stress testing is superior to that of manual ADT.⁶  The clinical practice of manual ADT is highly dependent on the tester’s subjectivity and experience. In a recent meta-analysis of 885 cases, the researchers¹⁴  reported that manual ADT had a sensitivity of only 54%, despite a relatively high specificity of 87%. Li et al¹⁵  noted a difference in the diagnostic performance of manual ADT between a junior and a senior examiner: sensitivity was only 5.3% for the junior examiner but 39.5% for the senior examiner. By contrast, arthrometric ADT has the advantage of objectively quantifying the test results, and sensitivity has been described as ranging from 74% to 91.7%.^7–9,16 

The load-displacement curves reflected the laxity of the ankle joints under arthrometric loading (Figure 2). Under a lower loading force (10–40 N), patients with MAI had a steeper slope than the control individuals, indicating a greater extent of anterior talar translation caused by lateral ligament failure.⁸  However, when the loading force was increased (>40 N), the curves of patients with MAI and control individuals tended to be parallel and have decreased slopes, which suggested that the talus was being forced to the end position and soft tissues around the ankle caused further displacement.

In previous studies related to the diagnostic accuracy of ankle arthrometry, the authors compared patients with CAI and a healthy control group.^7,17,18  Yet factors such as sex, foot size, local anatomy of the ankle, and soft tissue elasticity vary when performing the ADT.¹⁹  Eliminating all of these potentially confounding factors that may affect the results of ankle stress testing is not possible, but a bilateral ankle comparison is not influenced by these factors. Chen et al⁹  compared 153 ankles with CAI and 160 uninjured ankles and found sensitivity of 0.804 and specificity of 0.863 using LDR 10–40 as a dynamic reference standard. Nonetheless, Chen et al compared only CAI and uninjured ankles and did not assess bilateral differences between the CAI and control groups. Using the same experimental procedure and device, we found sensitivity and specificity of 0.900 and 0.933, respectively, using LDR 10–40 for bilateral comparisons. Thus, the ankle arthrometer displayed superior diagnostic accuracy for bilateral comparisons when only 1 ankle had MAI.

Differences between injured and uninjured contralateral ankles in the ADT results have varied among studies. Liu et al²⁰  demonstrated mean increased laxity of 12.8% in injured ankles compared with uninjured contralateral ankles. Tourné et al²¹  reported that the average bilateral difference in patients with MAI was 2 mm. Hubbard and Cordova²²  observed an average difference of 5.5 mm between the injured and uninjured contralateral ankles of patients with CAI. These investigators used different methods to measure the change in anterior talofibular ligament (ATFL) length, including displacement of the arthrometer transducers,^20,22  stress radiography,²¹  stress ultrasonography,¹⁹  and stress magnetic resonance imaging.²³  Different standards for representing ATFL lengthening led to various results. Transducer displacement essentially reflected translation of the ankle complex rather than the ATFL. Cho et al⁶  used stress radiography to define displacement as the distance between the posterior malleolus and the talus, whereas ultrasonography and stress magnetic resonance imaging were used to directly measure the distance between the origin and insertion of the ATFL. Cho et al described the sensitivity of stress radiography as 86%, whereas that of stress ultrasonography was as high as 100%. Moreover, ATFL examination depends on the examiner’s experience and positioning of the participant’s foot,²⁴  which were markedly diverse in different studies. For example, Iwata et al²⁵  found that the variability of stress ultrasonography between the experienced and beginner groups was too high to be clinically acceptable. Regarding foot positioning, the anterolateral drawer test allowed more internal rotation of the ankle than the ADT and detected more subtle degrees of ankle instability.¹⁵  However, the instrumented anterolateral drawer test was only performed in a cadaveric study, and its clinical use requires further development of ankle arthrometers.²⁶  Some authors have classified MAI as a difference of >3 mm of anterior displacement between the involved and uninvolved ankles,²⁷  whereas other researchers have used a difference of 2 mm^28–31 ; these thresholds were set arbitrarily. Croy et al⁷  demonstrated that different thresholds yielded different sensitivity and specificity, even if the arthrometric testing procedures were identical. For the proper diagnosis of MAI via an arthrometer, the best approach might be bilateral comparisons with a standardized testing procedure and a reasonable diagnostic threshold, which should be derived from calculation. In our work, the average bilateral difference was 3.28 mm. Furthermore, we determined that the 1.85-mm threshold for the bilateral difference in anterior D75 offered the highest sensitivity and specificity.

In this study, we calculated the AUC of 2 reference standards, D75 and LDR 10–40. The LDR had a higher AUC, indicating superior diagnostic accuracy. The LDR 10–40 is an emerging reference standard that may have an advantage in reflecting ATFL laxity. Among the researchers who used arthrometer-measured LDR, Lohrer et al⁸  measured the LDR between 40 and 60 N to represent the laxity of the ankle joint, and the resulting sensitivity and specificity were 0.81 and 0.93, respectively. However, Chen et al⁹  demonstrated that LDR 10–40 had a larger ES and AUC according to the load-displacement curves of the arthrometers. This might explain why we observed higher sensitivity and specificity. Nauck et al¹⁶  also used LDR to analyze ankle stiffness and reported sensitivity of 0.917 and specificity of 0.625. Nauck et al tested cadaveric ankles; some with intact lateral ligaments might have had structural changes resulting in increased laxity, which would have been misdiagnosed by the arthrometer (false positive). This could explain the relatively low specificity they identified. Chen et al⁹  showed that arthrometer-measured LDR 10–40 had sensitivity and specificity of 0.804 and 0.863, respectively; however, they did not further subdivide participants with CAI into MAI and FAI groups. Patients with FAI have intact ankle ligaments, which might have influenced the arthrometric measurements and decreased the sensitivity compared with other studies.

We recognize that many factors affect the diagnostic accuracy of ankle arthrometers. Differences in outcome measures yield different sensitivity and specificity values. Reviewing the literature on the sensitivity and specificity of ankle arthrometry, we found that dynamic measures, such as LDR, generally had better sensitivity and specificity than static measures using anterior displacement.^7–9,17  Dynamic measures reflected the entire process when the lateral ligaments of the ankle joints were being stretched during the ADT, whereas static measures represented only the ending status. Moreover, the setup and participants also played roles in the outcomes of different investigations. In a cadaveric study, the researchers³²  easily controlled the damage to ankle ligaments by severing ≥1 lateral ligament. However, participants with MAI from in vivo studies sustained ligament failure due to recurrent sprains that stretched their ligaments, which was different from cutting of the structures.³³  Another factor was that the loads differed greatly among the different arthrometers, and the resulting anterior displacement ranged widely from 3.2 ± 2.8 mm to 21 ± 2.5 mm.³⁴  In addition, the methodologic approaches to recruiting patients with MAI were diverse, including questionnaires, sonography, stress radiographs, magnetic resonance imaging, and clinical examination, all of which had different validities in assessing MAI.³⁴ 

We acknowledge several limitations of our work. First, we did not include patients with FAI. Differentiating the diagnosis between FAI and MAI using arthrometers may be useful because the interventions for these conditions differ. Second, the arthrometric talar tilt test was not performed. Future authors should assess diagnostic accuracy in bilateral comparisons.

The ankle arthrometer quantitatively measured the difference in bilateral ankle laxity in patients with MAI. An absolute difference of >1.85 mm in D75 or >0.0351 mm/N in LDR 10–40 for bilateral ankles suggests the diagnosis of MAI. Using LDR 10–40, the arthrometer had high diagnostic accuracy with sensitivity and specificity of 0.900 and 0.933, respectively. The arthrometer could be a useful tool for diagnosing patients with MAI in clinical settings.