© 2015, The Author(s). Purpose: The purpose of this study was to determine the test–retest reliability and the repeatability over multiple days of a robotic testing device when used to measure laxity of the lower leg during a simulated dial test. Methods: Ten healthy subjects were evaluated using an instrumented robotic lower leg testing system over 4 days. Three testing cycles were performed each day. Each leg was rotated into external and then internal rotation by servomotors until a torque threshold of 5.65 N m was reached. Load–deformation curves were generated from torque and rotation data. Both average-measure and single-measure intraclass correlation coefficients (ICC) were compared across the curves. ICC scores were also compared for features of the curves including: maximum external rotation at −5.65 N m of torque, maximum internal rotation at 5.65 N m of torque, rotation at torque 0, compliance (slope of load–deformation curve) at torque 0, endpoint compliance in external rotation, endpoint compliance in internal rotation, and play at torque 0. Play at torque 0 was defined as the width of the hysteresis curve at torque 0. Results: Average-measure ICC scores and test–retest scores were >0.95 along the entire load–deformation curve except around zero torque. ICC scores at maximum internal and external rotation ranged from 0.87 to 0.99 across the left and right knees. ICC scores for the other features of the curves ranged from 0.61 to 0.98. The standard error of the mean ranged from 0.0497 to 1.1712. Conclusions: The robotic testing device in this study proved to be reliable for testing a subject multiple times both within the same day and over multiple days. These findings suggest that the device can provide a level of reliability in rotational testing that allows for clinical use of test results. Objective laxity data can improve consistency and accuracy in diagnosing knee injuries and may enable more effective treatment.

Purpose: The purpose of this study was to collect knee laxity data using a robotic testing device. The data collected were then compared to the results obtained from manual clinical examination. Methods: Two human cadavers were studied. A medial collateral ligament (MCL) tear was simulated in the left knee of cadaver 1, and a posterolateral corner (PLC) injury was simulated in the right knee of cadaver 2. Contralateral knees were left intact. Five blinded examiners carried out manual clinical examination on the knees. Laxity grades and a diagnosis were recorded. Using a robotic knee device which can measure knee laxity in three planes of motion: anterior–posterior, internal–external tibia rotation, and varus–valgus, quantitative data were obtained to document tibial motion relative to the femur. Results: One of the five examiners correctly diagnosed the MCL injury. Robotic testing showed a 1.7° larger valgus angle, 3° greater tibial internal rotation, and lower endpoint stiffness (11.1 vs. 24.6 Nm/°) in the MCL-injured knee during varus–valgus testing when compared to the intact knee and 4.9 mm greater medial tibial translation during rotational testing. Two of the five examiners correctly diagnosed the PLC injury, while the other examiners diagnosed an MCL tear. The PLC-injured knee demonstrated 4.1 mm more lateral tibial translation and 2.2 mm more posterior tibial translation during varus–valgus testing when compared to the intact knee. Conclusions: The robotic testing device was able to provide objective numerical data that reflected differences between the injured knees and the uninjured knees in both cadavers. The examiners that performed the manual clinical examination on the cadaver knees proved to be poor at diagnosing the injuries. Robotic testing could act as an adjunct to the manual clinical examination by supplying numbers that could improve diagnosis of knee injury. Level of evidence: Level II.