In the present study, we established 3D postoperative models for CL radiograph simulation and data analysis. The advantages are that each variable can be set independently within a wide range, and confounding factors can be effectively controlled. The analysis showed that the PT had a significant effect on the measurement of the VA on CL radiographs. The formula we derived can convert the measured values under different PTs into actual values under the natural pelvic position. After comparing the corrected values from the clinical measurement data with the CT values, we found that they were highly consistent, which confirmed the high fidelity and reliability of the 3D simulation method and inferred formula. Our correction approach can effectively improve the accuracy of CL radiography when measuring the VA.
This study has several limitations. First, the data in our study were obtained from 3D models based on cases of a non-deformed pelvis because we sought to avoid the interference of other confounding factors (such as osteophytes and deformities) when determining the effects of the PT. The findings may not be suitable for abnormal cases. Second, we used the outline of the 3D images to replace the actual radiographs, enabling the setting of variables to be more accurate and extensive; however, the magnification of the radiographic edge cannot be well simulated, although it will not have a great impact on the measurement of the central visual field.
Reports in the literature with regard to the accuracy of CL radiographic measurement compared with that of CT measurement are conflicting [6, 11, 18, 19]. Some investigators found that the CL radiographic value was smaller than the CT value [11, 19], while others revealed the opposite [6, 18]. However, no control or influencing factors, such as the PT, was examined in those studies. The current study bridged the gap and identified the significant implications of the PT on the accuracy of CL radiographic measurement. When the PT changed by 1°, the measurement error could reach 0.76°. If the PT deviates significantly from the norm, the error may be considerable. Unfortunately, during the course of CL film imaging, the flexion of the lower contralateral extremities to avoid occlusion would lead to variations in the PT, which would be more pronounced in patients with contralateral hip stiffness. Moreover, for CT measurement, owing to the factors of anatomical specificity, pathological changes, and poor posture, there will inevitably be potential differences in the PT among different patients. When the PT was balanced between the CL radiographic and CT measurements, we were pleasantly surprised to find that the two measurements were highly consistent (ICC = 0.997, p < 0.001); however, these were in sharp contrast to those without control (ICC = 0.454, p = 0.203). Accordingly, we believe that the PT is the main factor affecting the accuracy of CL radiographic measurement.
To reduce this effect, we derived a correction formula, which can convert the measured values under different PTs into the actual VA values under the natural pelvis, and it was verified to be reliable after comparison with the CT values. Several methods have been described for correcting the error of VA measurement owing to the PT in previous studies [15, 20, 21]. Some of them used formulae as we did [15, 20], while others modified methods using special techniques, such as normograms . However, these methods are relatively complex and cumbersome. In contrast, our method can achieve reliable correction only through a simple equation.
Nevertheless, both the definition of the safety zone and the correction method were evaluated on the basis of the supine position. It has been reported that there are different changes in acetabular cup anteversion in different positions. Lazennec et al.  performed a CT scan in the supine and standing positions in patients undergoing hip replacement, finding that the VA significantly increased in the standing position because of the change in the included angle between the patient functional plane and the APP, which exceeded the limits of the safe zone defined by Lewinnek et al. . Therefore, only evaluating the acetabular component anteversion in the supine position is not sufficient. Mathematical algorithm may be employed to solve this problem. For instance, a patient was selected to receive CL radiography and additional lateral radiography to measure the VA and PT in the supine position 1 day after THA. One more lateral radiography at the standing position was performed on this patient to measure the PT 1 month after surgery. Through our equation, the actual VA in the standing position was calculated. If we include a few more patients for analysis, a new formula between the PT and VA in the standing position would be developed, indicating that by obtaining only a lateral film to measure the PT in the standing position, we can determine the actual VA using the equation. Thus, our next study is aimed at determining whether potential mathematical correlation could be explored to calculate actual acetabular anteversion in the standing position, making it possible to more easily and precisely assess the position of the acetabular cup after THA.