Patients and Methods

A retrospective review of 91 patients who underwent revision THA using 96 CFACs was undertaken, comparing features between radiologically failed and successful cases. Patient characteristics (demographic, clinical, and radiological) and implant features (design characteristics and intraoperative features) were collected. There were 74 women and 22 men; their mean age was 62 years (31 to 85). The mean follow-up was 24.9 months (sd 27.6; 0 to 116). Two sets of statistical analyses were performed: 1) univariate analyses (Pearson’s chi-squared and independent-samples Student’s t-tests) for each feature; and 2) bivariable logistic regressions using features identified from a random forest analysis.

Methods

We reviewed 96 CFAC cases performed in 91 patients between 2004 and 2017, from which 36 variables were collected spanning patient demographics, pre-operative clinical and radiographic features, intraoperative information, and implant design characteristics. Patient demographics and relevant clinical features were collected from individual medical records. Radiographic review included analysis of pre-operative radiographs, computer tomographic (CT) scans, and serial post-operative radiographs. Radiographic failure was defined as loosening or gross migration as determined by a board-certified orthopedic surgeon. CFAC implant design characteristics and intra-operative features were collected from the design record, surgical record and post-operative radiograph for each case respectively.

Two sets of statistical analyses were performed with this dataset. First, univariate analyses were performed for each variable, comprising of a Pearson chi-square test for categorical variables and an independent t-test for continuous variables. Second, a random forest supervised machine learning method was applied to identify the most influential variables within the dataset, which were then used to perform a bivariable logistic regression to generate odds ratios. Statistical significance for this study was set at p < 0.05.

Materials and Methods

A total of 60 lumbar vertebrae were divided into six groups: PLS submitted to instant pull-out and fatigue-resistance testing (groups A1 and A2, respectively), TPS submitted to instant pull-out and fatigue-resistance testing (groups B1 and B2, respectively) and PLS augmented with 2 ml polymethylmethacrylate, submitted to instant pull-out and fatigue-resistance testing (groups C1 and C2, respectively). The PPF and normalized PPF (PPFn) for bone mineral density (BMD) were compared within and between all groups.

Methods

First, we validated the accuracy and repeatability of a custom-made mechanical loading equipment set-up. Eight lumbar specimens underwent CT scanning in flexion, neural, and extension positions by using the equipment set-up. The changes in the disc height and distance between adjacent two pedicle screw entry points (DASEP) of the posterior approach at different lumbar levels (L3/4, L4/5 and L5/S1) were examined on three-dimensional lumbar models, which were reconstructed from the CT images.