A weight-bearing knee test found a marked difference in the operated legs of adults who had undergone anterior cruciate ligament reconstruction, or ACLR. Their knee-valgus proprioception, the body's sense of joint position, was poorer than in both their intact legs and the limbs of healthy controls. The ACLR group also performed worse on a measure of dynamic balance. Because the study was cross-sectional, with measurements taken at one point in time, it shows a pattern associated with ACLR status rather than proving that surgery caused it.
A test built around weight bearing
The researchers set out to test whether Knee-VEDA could reliably measure weight-bearing knee-valgus proprioception and distinguish people after ACLR from healthy adults. During the formal test, participants bore weight with the knee flexed to 15 degrees while four valgus positions were presented in random order. Each position was shown five times, making 20 trials. The score summarized the mean area under each participant's discrimination curve, which reflects how accurately the person detected the different positions.
The angles were estimated from the sideways movement of a disk and the person's leg length, rather than measured directly. For leg lengths between 35 and 45 centimeters, the reported angles ranged from 5.2 to 10.1 degrees. The result should therefore be read as performance on this standardized task, not as a direct measurement of every knee movement.
The measurement held up, with a weaker control estimate
The laboratory case-control study included 63 people: 32 ACLR participants and 31 age-matched healthy controls. The ACLR participants were 18 to 50 years old, at least four months beyond unilateral reconstruction, able to perform daily activities and above a Tegner activity score of 4. They could not have a major additional lower-limb injury or surgery. Controls had no lower-limb injury or symptoms. Thirty-six participants returned for a second assessment at least seven days later.
Knee-VEDA scores showed good overall test-retest reliability, meaning they were reasonably consistent when the test was repeated. The overall intraclass correlation coefficient, a common consistency measure, was 0.85. It was 0.84 among ACLR participants and 0.64 among controls. Only 15 controls took part in the retest, making the control estimate less certain than the overall result.
On a receiver operating characteristic analysis, which tests how well a measure separates two groups, the area under the curve was 0.835. At the reported cutoff of 0.768, the test had 71.9% sensitivity for ACLR cases and 87.9% specificity for controls. The figures suggest useful separation within this sample, but they do not establish that the same threshold would work in other groups.
The clearest signal came from the operated limb
The difference in proprioception was large. The operated limb scored worse than the healthy-control limb, with a standardized difference of 1.47, and worse than the participant's intact limb, with a standardized difference of 1.52. Both comparisons had p-values below 0.001. By contrast, the intact ACLR limb and the control limb were almost indistinguishable, with a standardized difference of 0.05 and p = 0.91.
The same ACLR group also had lower composite Y-Balance performance, a measure of dynamic balance. The difference was moderate, with a standardized effect size of 0.58, a 95% confidence interval from 0.09 to 1.07, and p = 0.02.
Better test performance tracked balance and symptoms
Among ACLR participants, better Knee-VEDA performance was associated with better Y-Balance performance and better scores on the KOOS Symptoms patient questionnaire. The correlation with Y-Balance was 0.50, with p below 0.001; the correlation with KOOS Symptoms was 0.55, with p = 0.001. No significant associations were found with the other outcome measures. These relationships were measured at the same assessment, so they do not show that one caused the other.
Muscle recordings told a different story
The researchers also recorded surface electromyography, or sEMG, from seven lower-limb muscles to examine muscle activity during the task. Yet no between-group or between-limb sEMG difference remained statistically significant after correction for multiple comparisons. The study therefore found a clear proprioception difference without a matching persistent group-level difference in the muscle recordings.
A separate correlation appeared only among healthy controls: proprioceptive acuity was linked with mean activity in the vastus medialis muscle, with r = 0.53 and an adjusted p-value of 0.002. The ACLR group showed no significant sEMG associations, with p-values from 0.14 to 0.96. Time since surgery was also not correlated with sEMG outcomes, with correlations from -0.04 to 0.07 and p-values from 0.66 to 0.82.
A useful signal, not a finished clinical tool
The evidence remains bounded. This was a cross-sectional laboratory study in adults selected by specific eligibility rules, not a prospective study of recovery. The ACLR group had varied postoperative duration and incomplete documentation of surgery, rehabilitation and return-to-sport status. The reliability retest included only 15 healthy controls, and the control estimate was imprecise. Knee-valgus angles were estimated from disk displacement and leg length rather than directly measured.
Taken together, the findings support Knee-VEDA as a potentially useful way to assess weight-bearing proprioception in this sample, and they justify further testing of the measure. Larger reliability studies in healthy participants, prospective validation and direct movement measurements are needed before routine clinical use. The study did not test whether rehabilitation guided by the test improves outcomes.
Paper data and sources
Original title: Knee valgus proprioception, muscle activity, patient-reported outcomes and dynamic balance control after anterior cruciate ligament reconstruction: a cross-sectional case-control study.
Authors: Mengde Lyu, Adrian Pranata, Joshua Farragher et al.
Journal/Repository: Knee surgery & related research
Status: Peer-reviewed
First online: 2026-08-20
DOI: 10.1186/s43019-026-00341-5
Original paper · Full text