This retrospective study investigated the correlation between the patients’ perception on LLD and different variables measured in the full-length standing anteroposterior radiographs after primary THA. In multiple logistic regression and The ROC curves, perceived LLD was associated with the difference in TD-TP and the difference in FL + TL. The sensitivity and specificity of the difference in TD-TP and the difference in FL + TL were 57.7%, 79.4% and 61.5%, 79.4%, respectively when the calculated thresholds of the difference in TD-TP and the difference in FL + TL were set at 9.0 mm.
The pelvic radiograph is widely used for LLD detection in clinical practice because of its simplicity and its low radiation exposure. In patients with unilateral DDH may present with LLD derived from both the femur and the tibial[14, 15]. Using the pelvic radiograph to predict LLD is not reliable. Zhang et al.  recommended that the use of full-length standing anteroposterior radiographs for LLD detection is advisable for patients with DDH because of its good accuracy and reliability.
The center of the femoral head is used as landmark for LLD measurement in the full-length standing anteroposterior radiographs[7-9]. However, the center of the femoral head in the operative side may be affected by the position of the acetabular shell. For the patients with Crowe type IV DDH in our institute, the acetabular shell was located in the posterior and inferior position of the true acetabulum. The height of the center of the acetabular shell in operative side was lower than that in the contralateral side, which may cause the difference of CH-TP to be inconsistent with the patients’ perception on LLD. Many studies also confirmed the patients’ perception on the difference of CH-TP had poor correlation and reliability in primary THA[8, 9].
In our study, the patients’ perception on the difference of GT-TP had also poor correlation and reliability. GT-TP was influenced by many factors, such as the SSTO and the femorotibial angle. In the setting of Crowe type IV DDH, SSTO may be necessary to safely reduce the hip to the true acetabulum, mitigate hip soft-tissue contractions and protect the neurovascular structures[11, 16]. The GT-TP in operative side after SSTO was significantly shorter than that contralateral side. In patients with Crowe type IV DDH, the femoral head was dislocated outwards and upward. In order to keep the leg alignment perpendicular to the ground, the valgus knee deformity was very common in the operative side. Therefore, the difference of GT-TP had poor accuracy and reliability on the true LLD and patients’ perception on LLD.
As described in most literature，LLD is defined as the difference of the distance between a femoral and a pelvic landmark on both sides[4, 17]. As a pelvic reference, the teardrop is used in many study[13, 17]. Because the teardrop is less affected by the position of the pelvis, it is more reliable than other pelvic landmarks. In our study, we found the difference of the TD-TP and the difference of FL + TL had good correlation and reliability on patients’ perception. Sectional measurement of the sum of FL and TL had a fair performance (AUC: 0.704) in the patients’ perception on LLD. However, the AUC of the difference of the TD-TP was only 0.679. The difference may be caused by valgus knee deformity in the operative side.
There is a broad consensus that less than 10 mm of LLD on radiographs is clinical acceptable[18, 19]. It also had been demonstrated previously that 10 mm inequality results in activation of compensatory mechanisms, such as a functional scoliosis or contraction of gluteus medius, which continues to increase in proportion to the imposed LLD[18, 20]. Lawrence et al.  proposed that no alterations in body posture or mechanics were activated at LLD of 6 mm and below. The study results in pelvic radiograph of Fujita et al.  show that 7 mm may be a reasonable threshold for reducing the residual discomfort. In our study, LLD of 9.0 mm may be a cutoff value to assess whether the patients’ perception on LLD, and the sensitivity and specificity of the difference in FL + TL and the difference in TD-TP were 57.7%, 79.4% and 61.5%, 79.4%, respectively.
The limitations of the study were (1) no data of preoperative LLD that might influence postoperative patient’s perception, and (2) no data of pelvic obliquity that was important for the postoperative patient’s perception on LLD. Zhang et al found the pelvic obliquity changes significantly in the first year after THA in patients with DDH. Therefore, we selected the patients with at least 12 months after THA, in order to reduce the influence of pelvic obliquity on LLD.