Relationship of femoral offset and clinical outcomes of femoral neck fracture patients following bipolar hip arthroplasty

Purpose This study aimed to investigate the relationship between femoral offset(FO) and clinical outcomes of patients with femoral neck fracture following bipolar hip arthroplasty(BHA). Materials and Methods This study recruited 520 patients who underwent BHA for femoral neck fracture from December 2003 to September 2018 with a minimum follow-up of one year. Excluding those with unclear medical records, a previous surgical history on the affected or contralateral hip, a congenital deformity and preoperative trauma history of the hip, high-energy multiple trauma, a history of neurodegenerative disease or cerebrovascular disease, a total of 77 patients were included in the analysis. The subjects were 54 females and 23 males. The mean age of patients was 74.6(65-95) years old and the mean follow-up period was 30.5 (12-136) months. For clinical assessment, postoperative pain visual analogue scale(VAS) and Harris hip score (HHS) were analyzed at each follow-up period. For radiological assessment, the difference in leg length discrepancy (LLD) and FO was measured. FO nearest to the real value was calculated using the values corrected for magnification and rotation errors on the anteroposterior view of plain radiographs taken at the first year of follow-up. In statistical analyses, the statistical significance between FO and clinical outcomes was analyzed using the Pearson correlation tests with SPSS v. 25.0. A value of p < 0.05 was considered statistically significant.


Introduction
The incidence of hip fracture is steadily increasing with a rising life expectancy 1) . Intertrochanteric fracture and femoral neck fracture account a large percentage of all hip fractures in older patients, and hip joint arthroplasty is commonly performed. Undisplaced femoral neck fractures that are not severe are typically treated with internal fixation, but displaced fractures at the time of admission to the hospital are more common. Hip arthroplasty is immediately performed to patients at high risk for fixation failure due to old age 2) . In recent years, total hip arthroplasty (THA) is increasingly performed in patients with femoral neck fracture, and favorable clinical outcomes have been reported [3][4] .
However, since THA is more extensive surgery and has the higher risk of dislocation compared to bipolar hip arthroplasty(BHA), BHA remains the preferred treatment in Korea 5) . In hip arthroplasty, biomechanical restoration is important and femoral offset (FO) is the most critical factor towards a successful hip replacement [6][7][8][9][10][11] . A large number of studies have explored FO in THA, but only a few studies have focused on BHA. Therefore, we aimed to investigate the correlation between Those with a history of neurodegenerative disease or cerebrovascular disease A total of 77 patients were included in the analysis. The subjects were 54 females and 23 males. The mean age of patients was 74.6 (65-95) years old and the mean follow-up period was 30.5 (±23.8) months. (Table 1).

Surgical methods and rehabilitation methods
All operations were performed in the lateral position using a posterolateral approach. Preoperative templating was performed for predicting the size of implant used in surgery. Before implant insertion during surgery, intraoperative portable X-ray was taken to confirm the size and location of the implant. The shuck test and stability test were done to assess stability including dislocation and impingement, and the actual implant was placed. A cementless femoral stem was used in all operations. Considering patient's systemic condition, surgeons allowed quadriceps exercise started from the second postoperative day. Standing and walker ambulation were begun in patients who actively flexed the hip. Walker ambulation was maintained and then shifted to cane ambulation in patients at high risk for slip down. In outpatient follow-up visits, full weight bearing was permitted in patients with good muscle strength and at low risk for slip down.

Clinical results
To identify the impact of the change in FO on clinical results of patients with femoral neck fracture after BHA, pain VAS at 1 st postoperative year and HHS at 1 st , 3 rd , 6 th and 12 th months postoperatively were assessed. HHS was categorized into pain domain with a maximum of 44 points and functional domain with a maximum of 47 points.

Radiological assessment
The difference of FO was defined as the change in the FO of the contralateral normal hip on preoperative 3-dimensional computed tomography (3D CT) compared to that of the operated side on plain radiographs obtained 1 year after BHA. To enhance the accuracy of FO of the normal side, 3D CT (Aquilion ONE ® , Canon Medical Systems, Otawara, Japan) was used and the perpendicular distance from the center of rotation of the femoral head to the femoral shaft axis was measured in the 3D axial, sagittal and coronal planes using the PACS system (Marosis M-view TM 5.4, Marotech, Seoul, Korea). (Fig. 2) FO of the operated hip was measured in the anterior-posterior pelvic views obtained 1 year after the surgery using the PACS system. The projected femoral offset (FOP) was defined as the perpendicular distance from the hip center of rotation to the femoral shaft axis on plain X-ray images. Since measurement errors occur on radiographs due to magnification rates by projection distance from the X-ray source to the anatomical structure, magnification correction factor (MCF) was calculated and used to reduce errors in the real FO value 12) . Since the shape of the acetabular cup is a hemisphere, the real cup diameter (CDR) can be obtained via correction for error rates according to projection distance because the maximum diameter of the cup represents the true hemispherical diameter on a plane in orthographic projection due to the geometrical nature. The real cup diameter (CDR) is the projected cup diameter (CDP) measured on plain radiographs using the PACS multiplied by MCF.
Therefore, MCF can be calculated using the following equation. (Fig. 3)

MCF = CDR / CDP
Using this formula, the real leg length discrepancy (LLDR) was calculated by multiplying MCF by the projected leg length discrepancy (LLDP), the measurement of the perpendicular distance between a horizontal line drawn through the inferior aspect of the teardrops and the prominent point of the lesser trochanter on plain radiographs taken 1 year after the surgery.

LLDR = LLDP x MCF
A caution is to be careful when assessing FO on plain radiographs because the more internally rotated the hip is, the smaller the real offset will appear [13][14][15][16] . To minimize measurement errors resulted from the degree of rotation, rotation correction factor (RCF) according to hip rotation was obtained

Discussion
The number of elderly patients with hip fracture is continuously increasing, and hip arthroplasty is commonly performed in older patients with femoral head fracture 1,2) . BHA was first introduced by Bateman and Giliberity in 1974. Unlike unipolar hemiarthroplasty, BHA is designed to provide the inner bearing between the femoral head and the polyethylene surface and the outer bearing between the acetabular cup and the acetabulum [2][3][4] . Compared with THA, BHA is more frequently performed type of arthroplasty because this procedure is a less complicated surgery without acetabular replacement and has the advantage of an increased jumping distance that lowers the risk for dislocation 5) . Like in THA, the FO is important in BHA. FO is the perpendicular distance between the center of rotation of the femoral head and a line drawn down the centre of the femoral shaft, and this is a critical in biomechanical studies on artificial hip joints [6][7][8][9][10][11] . The hip joint is the lever arm of body weight and the abductor mechanism and hip abductor muscles act as the lever arm extending from the lateral aspect of the greater trochanter and the femoral head, and a greater FO reduces forces Several studies have reported that errors can occur due to internal rotation and projection distance in measurement of hip offsets after THA, and an underestimation of the offset value on plain radiographs due to internal rotation has been commonly accepted [12][13][14][15][16] .
In this study, magnification error by projection distance was adjusted with MCF. Bae et al. 12) suggested that magnification rate was smaller as projection distance shortened, and magnification rate of plain X-ray of the femoral head was approximately 11.4%. In the present study, MCF was 0.88 (±0.04) and we verified that the size needed to be reduced by about 12% on plain X-ray to be nearest to the actual size.
Merle et al. 13) suggested that FO was frequently underestimated by approximately 13% on plain radiographs due to internal rotation of the hip, and recommended careful templating using CT because measurement errors in preoperative templating in THA may influence surgical outcomes.
Weber et al. 14) proposed that the measurement of FO with CT was more accurate than with plain radiographs, by revealing a mean error of 1.0 (±2.0).
Pasquier et al. 15) asserted and proved in their study that CT was a more accurate way for measuring FO and LLD. Therefore FO on the preoperative contralateral normal side was assessed using 3D CT in our study. This was possible because most patients with femoral neck fracture are admitted to hospital via emergency room and radiographic images are obtained by performing routine CT scanning before admission to emergency room. However, FO is assessed using plain radiography at postoperative follow-ups. The reasons for insufficient radiographic data are avoidance of high radiation doses from CT scans, high cost, patient compliance and others. To calculate postoperative FO nearest to the real value in plain radiographs, RCF was used. Philipp et al. 16) calculated RCF using centrum column diaphyseal angle(CCD) of inserted implants including the femoral stem and lag screw in 222 patients with proximal femoral nailing. In the Friedman test, there was no significant difference in the mean value between the mean FO measured using CT and the mean FO obtained using RCF on plain radiographs. In our study, the exclusion criteria were reinforced to find out definite differences in clinical outcomes during the follow-up period. Although this may reflect selection bias, it was appropriate for our study in which we aimed to identify even a small difference in clinical results based on patients' clear responses to postoperative inquiries. For example, there would be insignificant difference in clinical scores even 1 year after surgery in patients with inability to walk due to hemiparesis caused by a preoperative history of cerebral infarction. In addition, the evaluation of clinical symptoms and improvement would be less accurate in patients having difficulty communicating due to dementia.  18) . In the present study, HHS and VAS were used in the assessment of clinical outcomes, because HHS is the most commonly used tool to measure hip function in domestic and international literature and the evaluation criteria are easier for patients to understand. In a study on 50 patients who received hip arthroplasty and surgeons, Mahomed et al. 19) reported that items recorded by patients were almost the same as items recorded by surgeons in the HHS survey questionnaire.
The VAS has been used extensively in many studies. Data collection rate was good because VAS was one of the requirements in EMR in our institution. The mean VAS score was 1.5 (±1.1) in the first follow-up year, revealing no significant difference with the mean HHS pain score of 40.8 (±3.1).
In the current study, the Pearson correlation coefficient between FO and HHS within 1-year follow-up was -0.38, and a significant negative correlation was observed (p = 0.001). The Pearson coefficient for HHS function domain in 1-year follow-up was -0.42, revealing a higher correlation (p = 0.0001).
However, the Pearson coefficient for HHS pain domain in 1-year follow-up was -0.21, showing no significant correlation (p = 0.06). Through these results, we have inferred that FO after the first postoperative year was not correlated with pain severity, but was correlated with HHS function in femoral neck fracture patients with BHA. No significant difference in HHS functional domain at 1 st , 3 rd , and 6 th postoperative months seems to be resulted from the fact that functional domain was measured when functional rehabilitation was not fully completed and functional outcome is believed to be affected by limitations in daily activities such as pain.
Ji et al. 20) measured rotation-free FO using preoperative CT scans of the contralateral hip joint, the FO was changed by less than 20% in 77% of patients and more than 20% in 23% of patients, indicating clinically unfavorable outcomes. reported that anatomical restoration of an artificial hip joint had an insignificant effect on short-term functional outcome by showing in-hospital mortality of 5.7% and one-year mortality of 30% among elderly patients who received BHA for femoral neck fractures. Differences with our study is, the average age of the patient group was 84(78-94) years, relatively higher than in our study. And the exclusion criteria were only those with preoperative proximal femur fracture and a surgical history on the contralateral side. For this reason, it is believed patients having no difference between pre-and post-operative clinical results appear to account for a large percentage. Furthermore, since FO was not corrected for rotation, this likely that the measured FO was underestimated.
The strength of this study is that FO was obtained nearest to the real value by correcting for rotation and magnification errors in FO measurements. Moreover, this study strictly specified preoperative exclusion criteria to clarify clinical outcomes in compliance with the study purpose and to exclude obscure cases in determining any difference between pre-and post-operative results. We also aimed to identify the changes in pre-and post-operative results by investigating clinical outcomes according to the follow-up period. However, there are some limitations to note. First, this study has relatively small sample size. Second, the study was not conducted by a single surgeon and retrospectively. Availability of data and materials All data generated or analyzed during this study are included in this published article.

Competing interests
The authors declare that they have no competing interests.

Funding
This study was supported by the Dong-A University research fund.
Authors' contributions HJK designed the study and analyzed the data. SYL performed statistical analysis and language editing. KSS was a major contributor in writing the manuscript. All authors read and approved the final manuscript.    The scatter graph between the Harris hip score function domain and femoral offset difference in femoral neck fracture patients following bipolar hip arthroplasty