This study evaluated both the relative and absolute reliability of ultrasound-based measurement of changes in medial knee joint space width during a standardized valgus stress test in healthy individuals. The aim was to determine measurement consistency (ICC) and the magnitude and type of measurement error (Bland-Altman, SEM, MDC 95 ), providing thresholds to distinguish true changes from measurement noise.
Seventeen healthy adult males (34 knees) were prospectively recruited. Exclusion criteria included prior knee surgery or trauma, subjective knee pain, significant leg malalignment, inflammatory joint disease, and BMI ≥ 30. Female participants were excluded to avoid confounding from hormonal influences on laxity. The mean participant age, height, and weight were reported in the study.
A diagnostic ultrasound system with an 8–15 MHz linear transducer was used to acquire medial knee images. Participants were supine with the knee fixed at 20° flexion and the ankle in neutral dorsiflexion using a custom fixation device. A single examiner applied a valgus moment of 10 Nm by delivering a lateral force to the distal tibia while stabilizing the lateral knee; force was individualized by calculating applied force from moment and leg length and was continuously monitored.
Two licensed physical therapists performed image acquisition: Examiner 1 (early-career) and Examiner 2 (>10 years’ experience). A third experienced therapist applied the valgus load. Examiners trained until intra-rater ICC ≥ 0.80 before formal measurements. Images were captured repeatedly until three successful trials were obtained under both unloaded and loaded conditions.
All ultrasound images were transferred to a PC and analyzed in ImageJ. The medial knee joint space width was defined as the distance between the distal end of the medial femoral edge (extension of the medial femoral margin) and the most proximal medial surface of the tibial plateau along that extension. Measurements were scaled and made with the ImageJ distance tool. Changes were computed as loaded minus unloaded values. Examiner 1 performed all image analyses blinded to subject identity, acquiring examiner, and trial number.
Imaging occurred on two days one week apart. Day 1 imaging was performed by Examiner 1; Day 2 imaging was performed by Examiners 1 and 2 to permit intra- and inter-rater comparisons. For cross-day assessments, skin markings were removed and re-identified; for same-day inter-rater assessment, markings were retained.
Relative reliability was quantified using intraclass correlation coefficients (ICC(3,1), two-way mixed-effects, absolute agreement, single-measure). Interpretation thresholds used in the study were: >0.90 excellent, 0.75–0.90 good, 0.50–0.75 moderate, <0.50 poor. Absolute reliability was assessed with Bland-Altman analysis to detect fixed or proportional bias and to calculate limits of agreement. Where no systematic bias was present, the standard error of measurement (SEM) and minimal detectable change at the 95% confidence level (MDC 95) were calculated. Paired t tests compared unloaded and loaded joint space widths.
Medial joint space width increased significantly under valgus loading (mean increase reported between 1.33 and 1.41 mm; p < 0.001). Within-day intra-rater ICC(3,1) values for changes ranged from 0.85 to 0.93. Across two separate measurement days, intra-rater ICC(3,1) for change was 0.88 (95% CI 0.77–0.94). Inter-rater ICC(3,1) for change across examiners was 0.84 (95% CI 0.70–0.92); the lower bound of the inter-rater CI extended into the moderate range per the study’s interpretation thresholds. These point estimates indicate generally good relative reliability both within and between examiners using the described protocol.
Bland-Altman analysis identified only random measurement error for both intra- and inter-rater comparisons; neither fixed nor proportional bias was detected.
For intra-rater measurements across days, the 95% limits of agreement ranged approximately from −0.52 to 0.51 mm. The SEM was 0.19 mm, yielding an MDC 95 of 0.51 mm. For inter-rater measurements, the 95% limits of agreement ranged approximately from −0.75 to 0.55 mm. The SEM was 0.23 mm, yielding an MDC 95 of 0.65 mm.
These absolute reliability metrics indicate that typical random measurement error under the present conditions was small (<0.25 mm SEM) and that observed changes in medial joint space width greater than the MDC thresholds (0.51 mm for same-examiner repeat measures; 0.65 mm for different examiners) are likely to exceed measurement error with 95% confidence.
Under the specific protocol used in this study, ultrasound provided acceptable relative reliability (ICC in the good range) and precise absolute reliability (SEM ~0.19–0.23 mm; MDC 95 < 1.0 mm) for quantifying changes in medial knee joint space width during a 10 Nm valgus stress in healthy young males. The authors recommend averaging multiple trials to reduce random error and improve precision.
Limitations noted by the investigators include the small, homogeneous sample (healthy young males only), performance by trained examiners with a defined protocol, and image analyses by a single blinded rater. The authors explicitly state that further research with larger samples and inclusion of patients with ligamentous injuries is required to confirm clinical applicability and to validate thresholds in patient populations.
Clinical application: within the constraints of this protocol, clinicians and researchers using ultrasound to quantify medial compartment opening during valgus stress can use the reported SEM and MDC 95 values as thresholds to interpret whether observed changes likely reflect true differences rather than measurement error. Caution is advised before generalizing these thresholds to other populations, different loading protocols, or operators without similar training.