Which method of hip joint centre localisation should be used in gait analysis?

•Accuracy of predictive and functional hip joint centre estimates against EOS.•Functional geometrical sphere fitting technique; 1.1 cm from EOS.•Harrington et al. predictive equations; 1.7 cm from EOS.•Predictive equations best suited for population with hip/motor control problems.

Accurate localisation of the hip joint centre is required to obtain accurate kinematics, kinetics and musculoskeletal modelling results. Literature data showed that conclusions drawn from synthetic data, adult normal subjects and cerebral palsy children may vary markedly. This study investigated the localisation accuracy of the hip joint centre against EOS. The EOS system allowed us to register the hip joint centres with respect to the skin markers on standing subjects. A comprehensive set of predictive and functional calibration techniques were tested. For the functional calibration techniques, our results showed that algorithm, range of motion and self-performance of the movement were factors significantly affecting the results. Best results were obtained for comfortable range and self-performance of the movement. The best method in this scenario was the functional geometrical sphere fitting method which localised the hips 1.1 cm from the EOS reference in average and 100% of the time within 3 cm. Worst results for functional calibration methods occurred when the movement was assisted with a reduced range of movement. The best method in this scenario was the Harrington et al. regression equations since it does not rely on a functional calibration movement. Harrington et al. equations put the hips 1.7 cm from the EOS reference in average and 97% of the time within 3 cm. We conclude that accurate localisation of the hip joint centre is possible in gait analysis providing that method to localise the hip joint centres are adapted to the population studied: functional geometrical sphere fitting when hip calibration movements are not a problem and Harrington et al. predictive equations otherwise.