To maximize the accuracy of intraoperative fluoroscopy, the influence of pelvic tilt and rotation needs to be eliminated during surgery. We generally confirmed the following two points: the pubic symphysis must be vertical and overlying the coccyx, and the obturator foramen has a similar appearance on the fluoroscopic image as on preoperative standing pelvic plain radiography. These results are consistent with the protocol reported in a previous study . However, some unavoidable factors of intraoperative fluoroscopy may still create errors . First, intraoperative fluoroscopy was performed with a posterior-anterior beam direction. Second, the film focus distance is small. Third, the central beam is usually centered on the femoral head. Previous studies using the intraclass correlation coefficient (ICC) showed that the measurements obtained on intraoperative fluoroscopy images were correlated with those obtained with postoperative radiography . However, correlation analysis cannot be used to assess the agreement between the two methods of clinical measurement [21, 22]. In this study, Bland–Altman plots were used to analyze the agreement between intraoperative fluoroscopy and postoperative radiographs in assessing the outcome of PAO. The results indicated that the biases between the imaging modalities could be neglected.
This study has some limitations. First, because we excluded some patients with previous hip surgeries, simultaneous PFO, a nonspherical femoral head, and a subluxated contralateral hip, our findings cannot be generalizable to patients with these conditions. In addition, all measurements were performed by a single observer (J.P.). Since multiple prior studies have reported on interrater and intrarater reliability in measuring the LCEA, AI, AWI, PWI, EI, and MO, we did not repeat such assessments. Third, we determined that the biases of AI, PWI, and MO as measured on fluoroscopy images was acceptable based on the reference normal values reported in previous studies. Further studies are needed to determine whether these biases affect hip function after PAO.
The LCEA of Wiberg, AI, and EI were used to assess the lateral coverage of the femoral head. Correction of the LCEA between 25° and 40°, AI between 0° and 10°, and EI ≤ 20% were defined as the target ranges after PAO based on previously published normative values[8, 23]. Charles’ study indicated that EI were less correlated, with ICCs of 0.66 (0.46–0.79) . Unlike the results from Charles, we found high agreement for EI between intraoperative fluoroscopic images and postoperative standing AP pelvis radiographs. The LCEA and AI have demonstrated a high correlation between intraoperative fluoroscopy and postoperative plain radiographs in previous studies [12, 13]. However, a high correlation does not imply good agreement between the two methods; correlation analysis only quantifies the degree to which two variables are related . Stefanie’s study indicated an acceptable agreement between the two images using kappa statistics ; however, intraoperatively, they inclined the C-arm by approximately 5° to imitate a pelvic-centered image, differing from conventional methods. In this study, LCEA also showed high agreement between intraoperative fluoroscopic images and postoperative standing AP pelvis radiographs. In contrast to previous studies, our study found that the AI acquired from the postoperative radiographs was larger than that measured on intraoperative fluoroscopy (p < 0.05). We suspect that this difference may be due to the difficulty in determining the medial margin of the acetabular sourcil on fluoroscopic images. Charles also considered fluoroscopic images to have poorer resolution than plain radiographs , potentially making it more difficult to find the necessary landmarks for measurement. Through Bland–Altman analysis, we considered this difference to be acceptable (mean bias: -0.97°).
Proper acetabular reorientation includes not only lateral but also anterior and poster coverage. Excessive anterior coverage is a detriment to posterior coverage and may cause impingement and adversely affect the long-term survival of PAO . An anterior center-edge angle of Lequesne (ACEA), created on the false-profile view, of < 20° can be indicative of structural instability . Most surgeons prefer to obtain an oblique view of the iliac crest during surgery to achieve a false-profile view. Previous studies have shown that the intraoperative ACEA was greatly correlated with that obtained on postoperative radiographs, with ICCs of 0.71 (95% CI: 0.54–0.82)  and 0.80 (95% CI: 0.71–0.86) . We chose not to measure the ACEA intraoperatively to assess the improvement in anterior coverage; although we can imitate the version of the standing pelvis by tilting the C-arm beam, we cannot simulate the version of the standing pelvis when obtaining an oblique image. Klaus recommended the AWI and PWI to quantify anterior and posterior coverage . According to their report, the mean AWI and PWI were 0.41 and 0.91, respectively, for normal hips. Because these parameters for judging anterior and posterior coverage are measured on images simulating the standing pelvis version, we prefer this method to using the ACEA. In this study, the AWI obtained on intraoperative fluoroscopy images highly agreed with that obtained on postoperative radiographs. Although the mean PWI obtained on intraoperative fluoroscopy images was smaller than that on postoperative radiographs, this difference was acceptable since the mean bias was only 0.11.
A lateralized hip center was considered to be a sign of structural instability. The hip center was considered lateralized if the medial aspect of the femoral head was greater than 10 mm from the ilioischial line . Medialization of the fragment could decrease the joint contact forces by decreasing the bodyweight lever arm. Troelsen found that an MO distance greater than 20 mm correlated with a poor 6.8-year survivorship of PAO . Charles recommended placing the medial aspect of the femoral head only 5 to 15 mm lateral to the ilioischial line . In their study, intraoperative fluoroscopic measurement of the MO distance had the lowest correlations (ICCs: 0.46) with those obtained on postoperative AP pelvis radiographs. Our data indicate that the femoral head was more medial on intraoperative fluoroscopic images than on postoperation AP pelvis radiography. This difference is partly due to the different central beam positions. On the other hand, the effect of the imaging magnification ratio on MO was more significant than that on the angle (LCEA, AI) measurements. To eliminate this effect, the MO was calibrated using the ratio of the femoral head diameter as measured on preoperative CT positioning images to those measured on intraoperative fluoroscopic images and pelvic radiographs. In this study, the mean bias was only − 1.11 mm, and the 95% LOA was − 5.55 mm-3.29 mm. Compared to the acceptable range of 15–20 mm [20, 26], this error is completely negligible.