The contributions of this paper are twofold. One is measurement of loads and corresponding strains at failure in both the anterior cruciate (ACL) and medial collateral (MCL) ligaments of the knee, the two ligaments most commonly injured. To make these measurements, a five degree-of-freedom automated system was used to apply either a single load or a combination of two loads at flexion angles ranging from 0° to 45°. Approximately half of the specimens were loaded to fail the ACL and half the MCL. An anterior force consistently failed the ACL while a valgus moment and/or an external axial moment failed the MCL. Despite this consistency, there was wide variation in failure loads between specimens with more than a twofold difference in load levels. Also, there was a large variation in ligament strain at failure with a mean value of 6.7% and a coefficient of variation of 0.64.
The second contribution of this paper is the development and testing of empirical models to predict ligament failure. To provide data for these models, prior to the load-to-failure tests of the knee specimens, the load application system was used to apply sub-injury level loads corresponding to varus/valgus moments, axial moments, medial/lateral forces, anterior/posterior forces, and compressive force at flexion angles of 0°, 15°, and 30° for single loads and 0° and 30° for combination loads. Regression analysis was performed to determine empirical equations relating strain to both load and flexion angle. The ability to extrapolate these models to predict strain measured in the load-to-failure tests was evaluated. In general, the empirical models did not accurately predict strain at injurious load levels. The inability to make an accurate prediction was traced to the conservative sub-injury load levels used to develop the empirical models.