A custom-designed instrumented knee prosthesis26
was implanted in the right knee of an 81-year-old male (170 cm, 64.5 kg) using a midvastus approach 1.5 years prior to testing, and the patient was well functioning at the time of testing. In vivo
tibial forces were measured after informed consent was obtained. The femur was cut at 6° anatomic valgus and 3° external rotation with the posterior condyles as references using intramedullary alignment. A standard cruciate-retaining Sigma PFC femoral component (Depuy, Warsaw, IN) was cemented in place. The tibial cut was made at 0° anatomic valgus without any posterior slope using intramedullary alignment. The tibial canal was reamed using custom instrumentation developed for the stem and keel. The instrumented tibial prosthesis was cemented. A 10-mm polyethylene insert (Sigma PLI) was used. Postoperative deep venous thrombosis prophylaxis and rehabilitation were the same as for routine primary knee arthroplasty. Intraoperative passive flexion showed reasonable balance between the medial and lateral soft tissues, defined as <10% difference between the medial and lateral forces over the range of from 0 to 90° of flexion. On postoperative full-length standing AP radiographs, femoral component alignment was in 6° valgus to the anatomic femoral shaft axis, defined as a line joining the mid-point of two transverse lines at the upper and lower region of the middle third of the femur; tibial tray alignment was 90.1° relative to the anatomic tibial shaft, defined by a line drawn from the mid-point of the tibial plateau to the center of the talus.
The instrumented prosthesis consisted of a titanium alloy tray instrumented with four uniaxial load cells, a microtransmitter, and an antenna.26
The uniaxial load cells were located 20.4 mm medial and lateral, and 9.8 mm anterior and posterior, of the center of the tibial tray, respectively (). The instrumented knee transmitted tibial force data from the four sensors at 70 Hz. Custom PC-based software was developed to read, display, and store data. Medial compartment contact loading was calculated as the sum of the medial anterior and medial posterior compressive loads.27
Similar calculations were performed for the lateral compartment. Total compressive contact force was calculated as the sum of the medial and lateral compressive loads (the sum of all four load cells). Loads were normalized to body weight (BW) for analyses.
The four load cells were located 20.4 mm medial and lateral, and 9.8 mm anterior and posterior, of the center of the instrumented prosthesis, respectively.
Gait analysis was performed simultaneously with tibial force measurements. The subject performed 3 walking trials at each of 3 speeds in random order: self-selected slow (1.00 ± 0.07 m/s), normal (1.23 ± 0.08 m/s), and fast (1.38 ± 0.06 m/s), and in 2 shoe conditions, his own personal walking shoes and variable-stiffness intervention shoes. The variable-stiffness shoe was a normally-appearing athletic shoe () with a sole made of compression molded ethylene vinyl acetate that has been custom-designed so that the lateral sole (Asker C durometer 55 ± 2) is 1.3-1.5 times stiffer than the medial sole (Asker C durometer 70-76 ± 2). The personal shoe had a sole of injection-molded EVA ethylene vinyl acetate (New Balance, Men's model 625).
Variable-stiffness shoe with greater lateral sole stiffness (1.3-1.5×) versus medial sole stiffness.
Reflective markers were placed on the leg along the anterior superior iliac spine, greater trochanter, lateral tibial plateau, lateral malleolus, lateral aspect of the calcaneus, and lateral head of the fifth metatarsal. An 8-camera optoelectronic system for 3D motion analysis (Qualisys Medical AB; Gothenburg, Sweden) was used to collect marker data for 5 secs for each trial. Ground reaction force data were collected using a multi-component force plate placed in the center of the walkway (Bertec Corporation; Columbus, OH). Kinematic and force data were collected at a frequency of 120 Hz. The kinematic, ground reaction force, and tibial force data were resampled at a common frequency during postprocessing. To calculate external moments at each joint center, each limb segment (foot, shank, thigh) was idealized to be a rigid body. Inertial properties of the segments were taken from the literature.28
The positions of the joint centers at the hip, knee, and ankle were located relative to the positions of the skin markers at the greater trochanter, the lateral joint line of the knee, and the lateral malleolus, respectively. The flexion-extension axis was assumed to remain perpendicular to the plane of progression; the abduction-adduction and internal-external rotation axes of the hip, knee, and ankle moved with the thigh, shank, and foot segments, respectively.7
The external knee adduction moment for each trial was calculated from marker, force plate, and inertial segment data using an inverse dynamics approach,29
and normalized to bodyweight and height (Bw*Ht).
The 1st and 2nd peak external knee adduction moments, medial contact forces, and total contact forces were calculated as the maximum moments or forces during the 1st or 2nd half of stance phase, respectively. No clear peaks existed for lateral force data; therefore, average lateral contact force over the stance phase was investigated. Average medial and total contact forces over stance phase (i.e., the mean value from 0 to 100% stance phase) were also examined.
For hypotheses 1 and 2, to account for the source of variability of the speeds, analysis of covariance (ANCOVA) was used to compare the knee adduction moment and contact forces between shoe conditions, using walking speed as a covariate (α=0.05). Changes in stride length,7
and average vertical ground reaction force were also analyzed using ANCOVA with speed as a covariate to investigate if changes in medial contact force between shoe conditions could be attributed to these sources. For hypothesis 3, linear regression analysis (α=0.05) was used to detect a relationship between change in first peak knee adduction moment and change in first peak medial contact force with the variable-stiffness intervention shoes versus the personal shoes. All statistical tests were performed in SPSS version 17.0 (SPSS Inc.; Chicago, IL).