Evaluation of the accuracy of three popular regression equations for hip joint centre estimation using computerised tomography measurements for metal-on-metal hip resurfacing arthroplasty patients

We investigated the accuracy of the regression equations by Bell et al., Davis III et al. and Harrington et al. for hip joint centre (HJC) estimation against the gold standard of computerised tomography (CT) measurements of HJC for 18 patients with metal-on-metal hip resurfacing arthroplasty (MoMHRA). The HJCs were estimated based on the position of the left and right Anterior Superior Iliac Spine (ASIS) and the left and right Posterior Superior Iliac Spine (PSIS) identified from a CT scan. Of the three tested regression equations, only those of Harrington et al. produced results that were not significantly different from the patient's 'true' HJCs as measured from the CT scan in all three directions when analysing left and right hips together for both resurfaced and native hips. When native and resurfaced hips were pooled and analysed for left and right, separately, the Harrington et al. regression equations showed significantly different results in the ML direction. Similar estimation errors were observed for native and resurfaced hips. Since none of the methods tested performed particularly well, we suggest using medical imaging if accurate estimates of HJCs are required.

► The HJC regression equations of Bell et al., Davis III et al. and Harrington et al. were evaluated for MoMHRA patients. ► Similar estimation errors between native and resurfaced hips were found. ► The regression equation of Harrington et al. generally produced the best results.