Patient-Specific Predictive Model Improves Accuracy in Hip Implant Sizes

A prediction model presented better estimates of the actual size of implants used in total hip arthroplasty and, in the long term, may help reduce the cost of health care by minimizing implant inventory costs, according to study results.

Researchers at the University of Texas Health Science Center at Houston – Medical School retrospectively reviewed preoperative radiographs of 468 patients who underwent total hip arthroplasty (THA). Preoperative template sizes were compared with the actual implants used during surgery. All preoperative radiographs were taken in the standing position.

The pelvis was centered over pubic symphysis and the hip was internally rotated 10 degrees and 15 degrees for THA templating in the anteroposterior view. Using the TraumaCad software system (Voyant Health), the researchers analyzed the digital radiographs. A multiple regression model was then used to create the model for predicting actual size of implants used, as well as to investigate each factor’s contribution to the model in predicting the implant’s actual size based on preoperative measurements. Variables studied included preoperative acetabular size, preoperative femoral size, BMI, age, gender, height and weight. The researchers used a backward stepwise algorithm to identify the variables in each model, which resulted in significant model demonstration.
According to the researchers, the backward stepwise model showed four significant predictors were achieved for acetabular component size, including template acetabulum size, height, BMI and template femoral size. Estimated acetabular size, according to the model, was 54.96 millimeters.

The model also demonstrated femoral component size can only be determined significantly with preoperative femoral component measurements. The estimated femoral size, according to the model, was 4.82 millimeters.

Using the model, two cases were predicted within ± 3 size and two cases were predicted within ± 4 size for acetabular component. The model predicted the femoral component with less accuracy, however. In 11 cases the model predicted implant size correctly in ± 3 size, whereas two cases were predicted within ± 4 size, according to the data.

The results showed that utilizing the model improved overall templating accuracy, with significant improvement in accuracy in the acetabular component within ±1 size and a 99 percent accuracy within ± 2 of the templated size.