Three-Dimensional CT-Based Limb Length Evaluation Is Highly Dependent on Anatomical Landmark Selection and Pelvic Asymmetry

Background: Pelvic skeletal asymmetry can result in rotational differences and morphologic bony prominence variance between the left and right hemipelvis. When selecting bony reference points for modern computed tomography-based robotic total hip arthroplasty planning, it is unclear which bony landmarks are the most reliable and accurate, especially in the presence of significant pelvic asymmetry.

Methods: A retrospective study was conducted utilizing a database of computed tomography scans. Multiple bony landmarks in the pelvis and femur were selected for comparison, with the aim of measuring pelvic asymmetry. Specifically, the study measured the average difference in lateral offset between the left and right hemipelvis caused by pelvic asymmetry. Landmarks were also compared to determine the impact of pelvic asymmetry on hip length, femur length, and limb length discrepancies. Furthermore, a scenario was simulated in the software whereby a total hip replacement was inserted, potentially changing the hip length. The impact of pelvic reference point selection on the measurement of this simulated change in hip length was examined.

Results: This study population showed widespread pelvic asymmetry. The anatomical landmarks of the opposite side cannot be relied upon for predicting the anatomy of the affected side. The center of rotation axis is more reliable than the inferior obturator foramen axis for hip length discrepancy due to pelvic asymmetry (P < .05).

Conclusions: Current computer-assisted surgery THR software reports measurements of global offset and hip length that do not consider pelvic asymmetry. Surgeons are not given confidence ranges to represent the potential impact of asymmetry on the global offset and hip length values. Surgeons following these numbers to guide implant position may incur implant placement error should significant pelvic asymmetry be present in a given patient.