Because it can reduce complications and result in early recovery, minimally invasive surgery (MIS) has recently gained popularity as an approach to performance of hip hemiarthroplasty (HA) for treatment of femoral neck fractures (FNFs)^1-4). In addition, the Barthel index (BI) at discharge has been reported as a prognosis predictor^5,6). A higher BI score at discharge showed an association with a good prognosis.

The superior approach (SA) to performance of HA is a minimally invasive posterior approach (PA)^7-9). The PA is a conventional approach. A surgical technique for performance of the PA was described by Moore in 1957¹⁰⁾. The short external rotator muscles (piriformis muscle, conjoined tendon, and external obturator muscle) and quadratus femoris muscle were resected. The hip joint was flexed at 90° and internally rotated to 90°. However, use of these methods resulted in a higher incidence of postoperative dislocations¹¹⁾. Use of the SA enables preservation of short external rotator muscles, reaching the hip joints between the piriformis muscles and the gluteus medius and minimus muscles, and does not involve an over-ranged hip position⁸⁾. No study comparing early recovery and prognosis for SA-HAs and PA-HAs for treatment of FNFs has been reported.

The objective of this study was to clarify the features of SA-HAs, with comparison of the outcomes of SA-HAs and PA-HAs in terms of early recovery and mid-term outcomes.

In total, 85 SA-HAs and 45 PA-HAs were performed at our hospital for treatment of FNFs between June 2013 and May 2018. Ten patients who had already undergone surgery on the contralateral hip joint were excluded. A total of 120 HAs performed for treatment of a primary FNF with a healthy opposite hip joint were included in this retrospective survey.

The results of a previous survey showed that 23% of patients who underwent HAs did not reach the gait required for evaluation of walking ability. Injury at home can influence walking ability¹²⁾. Thus, propensity score matching in regard to age, sex, body mass index (BMI), walking ability before injury, injury at home, time to surgery, and American Society of Anesthesiologists-Physical Status (ASA-PS) between 79 SA-HA patients and 41 PA-HA patients was performed. The final analysis of the time to surgery, operation duration, intraoperative bleeding, postoperative walking ability, hospital stay, BI (at the start of postoperative rehabilitation and discharge), follow-up period, and complications (postoperative dislocations, periprosthetic fractures, sciatic nerve injury, acetabular erosions, superficial surgical site infection, wound complications, contralateral fractures, and death), including cerebrovascular disorders and other fractures included 34 patients who underwent SA-HAs and 34 patients who underwent PA-HAs. Other fractures included any other fractures, except periprosthetic and contralateral fractures.

Assessment of the complication-free survival rate and the periprosthetic fracture-free survival rate was performed for 120 patients before matching and 68 patients before and after matching. The endpoints were the complications described above and the periprosthetic fracture, respectively. This study received approval from the Institutional Review Board of the Yamaguchi Red Cross Hospital (approval No. H27-11). The written informed consent was obtained from all patients.

Walking ability was 8-10 (walking independence), according to the Hospital for Special Surgery-Hip Rating System (HSS)¹³⁾. The scores were as follows: 10=no support or an appreciable limp and unrestricted travel distance; 8=no support with a limp or one support with no limp; 6=no support and less than one block (50 m) traveled distance, one support and up to five blocks (250 m) traveled distance, or bilateral support and unrestricted travel distance; 4=no support and housebound, one support with less than one block (50 m) traveled distance, or bilateral support and less than three blocks (150 m) traveled distance; 2=wheelchair or transfer activities with a walker; and 0=bedridden. This system uses a block as the walking distance index. One block was defined as 50 m.

1. Surgeons

PA-HAs were performed by six surgeons, four of whom were young (≤6 years), and two who were experienced. SA-HAs were performed by seven surgeons. Among seven surgeons, six surgeons were young (≤6 years) and one surgeon was experienced. In this survey, both the PA and SA were used by two of the young surgeons.

2. Surgery

When performing SA, the patient was placed in the lateral decubitus position. The hip joint position was 60°-70° flexion, maximum abduction, and 10°-20° internal rotation (sleep position)¹⁴⁾. A skin incision was made from the tip of the greater trochanter 8-12 cm proximal to the extended femoral axis, and the gluteus maximus was bluntly divided. The gluteus medius muscles were retracted anteriorly and the piriformis muscles were retracted posteriorly. The gluteus minimus was then retracted anteriorly with the gluteus medius to expose the capsules. After cutting the capsule, the femur was lifted over the femoral head (overhead lift) to allow for easy rasping. A neck cut was made in the rasp placed on the femur to prevent intraoperative femoral fracture, followed by removal of the femoral head. A trial reduction was performed, followed by implantation. Overrange of the hip joint did not occur during surgery. Three instruments were used: an outer head for attachment of the inner head within the acetabular fossa, a trial outer head with a groove to facilitate trial reduction, and flat broaches to facilitate performance of a neck osteotomy, leaving the broach in place^7,8). Use of these devices can facilitate performance of SA-HA procedures (Fig. 1).

Figure 1. Hip hemiarthroplasty surgery using a superior approach. (A) A flat broach to facilitate neck osteotomy, leaving the broach in place. (B) A trial outer head with a groove to facilitate trial reduction. (C) Trial reduction. (D) After placement. P: piriformis muscle, M: gluteus medius.

When using the PA, the piriformis, conjoined tendon, and external obturators were resected along with repair of the piriformis muscle and conjoined tendon post-implantation. The hip was positioned at 90° of flexion, maximum abduction, and 90° of internal rotation.

3. Statistical Analysis

The t-test was used for continuous variables and Fisher’s exact test was used for categorical variables using Easy R software (EZR, ver. 1.52; Saitama Medical Center, Jichi Medical University)¹⁵⁾. In propensity score matching, multivariate analysis was used for analysis of patient characteristics (age, sex, BMI, walking ability before injury, injury at home, time to surgery, and ASA-PS). Complications was the endpoint for determining the survival rate. Survival analyses were performed using postoperative dislocations, periprosthetic fractures, contralateral fractures, death, and other fractures as factors described in the Kaplan–Meier plot, and an evaluation of statistical differences between the two survival curves was performed using the log-rank test. Multiple regression analysis was performed to determine the factors that influenced BI at discharge, which included age, approach, injury at home, walking ability before injury, and sex.

The time to surgery was longer (P=0.001) and the complication rate was lower (P=0.037) for SA-HAs compared with PA-HAs (Table 1). No conversion of SA to PA was observed. The 4-year complication-free survival rates were 71.7% and 48.7% for SA-HAs and PA-HAs, respectively (P=0.054; Fig. 2). The 4-year periprosthetic fracture-free survival rates were 90.8% and 76.1% for SA-HAs and PA-HAs, respectively (P=0.446; Fig. 3).

Table 1 . Patients Characteristics and Complications before the Propensity Score Matching.

	PA (n=41)	SA (n=79)	P-value
Age (yr)	83.4±6.7	81.4±9.7	0.224
Sex, female	34 (82.9)	55 (69.6)	0.129
BMI (kg/m2)	20.8±3.3	20.1±3.9	0.344
Injury at home	25 (61.0)	53 (67.1)	0.548
Walking ability before injury	8.2±2.6	7.5±2.6	0.121
Time to surgery days (day)	1.8±1.5	3.1±3.0	0.001
Time to surgery days <2 days	33 (80.5)	49 (62.0)	0.042
ASA-PS, I/II/III/IV	0/4/36/1	1/13/63/2	0.826
Surgeons, young/experienced	5/2	7/1	0.569
Young surgeons operations	32 (78.0)	62 (78.5)	>0.999
Follow-up (day)	872±827	902±768	0.843
Total complications	17 (41.5)	18 (22.8)	0.037
IOF	2 (4.9)	1 (1.3)	0.269
Sciatic nerve injury	0	0	-
Postoperative dislocation	0	0	-
Periprosthetic fracture	3 (7.3)	6 (7.6)	>0.999
Contralateral fracture	4 (9.8)	3 (3.8)	0.229
Death	3 (7.3)	4 (5.1)	0.689
DVT	0	1 (1.3)	>0.999
Superficial surgical site infection	0	0	-
Wound complication	0	0	-
Other fractures	4 (9.8)	2 (2.5)	0.178
AE	0	1 (1.3)	>0.999
CVD	1 (2.4)	0	0.342

Values are presented as mean±standard deviation, number (%), or number only..

PA: posterior approach, SA: superior approach, BMI: body mass index, ASA-PS: American Society of Anesthesiologists-Physical Status, IOF: intraoperative fracture, DVT: deep venous thrombosis, AE: acetabular erosion, CVD: cerebrovascular disease..

Figure 2. Complication free survival rate before and after matching, compared with PA and SA. PA: posterior approach, SA: superior approach.

Figure 3. Periprosthetic fracture free survival rate before and after matching, compared with PA and SA. PA: posterior approach, SA: superior approach.

1. Outcomes of Propensity Score Matching

The duration of surgery was 57.1 minutes and 72.1 minutes (P=0.001) for SA-HAs and PA-HAs, respectively. The scores for walking ability at postoperative one week were 4.9±1.4 and 4.2±1.0 (P=0.021) for SA-HAs and PA-HAs, respectively. The BIs at the start of rehabilitation were 26.2±18.7 and 17.4±16.3 (P=0.042) for SA-HAs and PA-HAs, respectively (Table 2). The 4-year complication-free survival rates were 74.2% and 56.3% for SA-HAs and PA-HAs, respectively (P=0.310, Fig. 2). The 4-year periprosthetic fracture-free survival rates were 91.6% and 80.7% for SA-HAs and PA-HAs, respectively (P=0.579; Fig. 3).

Table 2 . Patients Characteristics and Outcomes after the Propensity Score Matching.

	PA (n=34)	SA (n=34)	P-value	SMD
Age (yr)	84.1±5.8	80.7±9.5	0.077	0.266
Sex, female	27 (79.4)	28 (82.4)	>0.999	0.155
BMI (kg/m2)	20.3±3.2	20.5±3.8	0.862	0.034
Injury at home	22 (64.7)	22 (64.7)	>0.999	<0.001
Walking ability before injury	8.1±2.6	8.5±2.2	0.486	0.024
Time to surgery days (day)	2.1±1.4	1.7±1.1	0.253	0.041
ASA-PS, I/II/III/IV	0/4/30/0	0/9/25/0	0.217	0.372
Time (min)	72.1±21.8	57.1±14.2	0.001
Bleeding (g)	113.2±125.4	134.4±74.2	0.400
Surgeons, young/experienced	4/2	5/1	>0.999
Young surgeons operations	26 (76.5)	22 (64.7)	0.425
Walking ability score at
PO 1 week	4.2±1.0	4.9±1.4	0.021
PO 2 weeks	4.7±1.5	6.1±2.1	0.004
Hospital stay (day)	34.3±20.0	31.5±14.6	0.507
Discharge to home	12 (35.3)	17 (50.0)	0.327
Barthel index
At the start	17.4±16.3	26.2±18.7	0.042
At the discharge	58.7±33.2	74.9±25.7	0.028
Follow-up (day)	855.3±813.1	1,102.2±770.2	0.203
Total complications	13 (38.2)	11 (32.4)	0.800
IOF	2 (5.9)	1 (2.9)	>0.999
Peri-prosthetic fractures	3 (8.8)	3 (8.8)	>0.999
Contralateral fractures	2 (5.9)	2 (5.9)	>0.999
Death	3 (8.8)	2 (5.9)	>0.999
DVT	0	1 (2.9)	>0.999
Other fractures	2 (5.9)	2 (5.9)	>0.999
CVD	1 (2.9)	0	>0.999

Values are presented as mean±standard deviation or number (%)..

PA: posterior approach, SA: superior approach, SMD: standardized mean difference, BMI: body mass index, ASA-PS: American Society of Anesthesiologists-Physical Status, PO: postoperative, IOF: intraoperative fracture, DVT: deep venous thrombosis, CVD: cerebrovascular disease..

2. Outcomes of Multiple Regression Analysis

The BI at discharge was influenced by female sex (P=0.012), SA (P=0.028), injury at home (P=0.001), and walking ability before injury (P<0.01) (Table 3).

Table 3 . Factors to Influence Barthel Index at Discharge.

	Estimate	Standard error	t value	P-value
Age	–0.253	–1.098 to 0.590	–0.599	0.551
SA	13.0	1.444 to 24.500	2.249	0.028
Injury at home	21.8	8.796 to 34.900	3.348	0.001
Walking ability before injury	4.3	1.533 to 7.129	3.094	0.003
Sex, male	–18.4	–32.800 to –4.140	–2.576	0.012

SA: superior approach..

Patients who underwent SA-HAs showed significantly earlier recovery in terms of walking ability and BI at the start of postoperative rehabilitation and discharge. The 4-year complication-free survival rates were 74.2% and 56.3% for the SA-HAs and PA-HAs groups, respectively (P=0.310). Multivariate analysis showed that SA contributed to higher BI at discharge.

However, few studies on recovery after HAs have been reported. Tsukada and Wakui³⁾ reported that use of HA with a direct anterior approach (DAA) resulted in earlier recovery of walking compared with PA-HA, HSS walking ability 5.3±1.5 and 4.2±2.3 at postoperative one month. However, no differences in walking ability were observed at postoperative one year. DAA was performed using a minimally invasive anterior supine approach. Soft tissue sparing can lead to early recovery. When using the PA, testing of spare short external rotators was performed (Table 4)^7,8,10,16). The short external rotator muscles were preserved in the SA-HA group. Compression of the surface of the hip joint by external rotational torque can lead to hip stability. In particular, the external rotator muscles play an important role during performance of pelvic and trunk rotational activities while standing on one leg¹⁷⁾. Postoperative unrestricted motion and early recovery can be achieved by preserving surrounding tissues.

Table 4 . The Feature of the Posterior Approach and the Muscle Sparing Posterior Approaches.

Approach	Resection of short rotator muscles				ISFL resection	Risk of sciatic nerve injury	Dislocated hip position
	Piriformis	Conjoined tendon	External obturator	Quadratus
Posterior (1957)10)	Yes	Yes	Yes	Yes Proximal 1/3	Superior Inferior	High	Yes
External rotator preserved (2008)16)	No	IG	Yes	No	Inferior	High	Yes
Superior (2017)7,8)	No	No	No	No	Superior	Low	No

ISFL: ischiofemoral ligaments, IG: inferior gemellus..

The BI at discharge is an important index for predicting prognosis^5,6). Renken et al.¹⁾ reported that the modified BI score was higher for DAA-HA compared with HA using the anterolaretal supine approach, 20±13.6 and 10±10.2 at postoperative five days. The findings of our study showed that BI at discharge was influenced by younger age, female sex, injury at home, walking ability before injury, and SA. Among these factors, SA, a surgical approach, was the only factor that affected the device after injury. The findings of this study indicate that there are differences in the processes of recovery for SA and PA. Depending on the surgical approach, the goals of rehabilitation should include achievement of early recovery of walking and BI.

Before propensity score matching, the complication rate was lower and the survival rate was higher for SA-HAs than PA-HAs. This result suggests that use of a muscle-sparing approach to HA may be a predictor of prognosis. Albumin level, male sex, age, and cognitive impairment were also predictors of prognosis^18-21). In this study, cerebrovascular disease and fractures were a cause of bedridden conditions and influenced healthy life expectancy and were therefore included as complications. Thus, preventing patients from becoming bedridden after a fracture is important. Use of a multidisciplinary approach and MIS-HA has led to a decrease in complications, resulting in early recovery^22,23).

This study had some limitations. The first issue was the choice of approach. Selection of PA or SA by surgeons was based on their surgical skill maturity. They first mastered the PA and then employed the SA. Thus, most were accustomed to using the PA but were unfamiliar with the SA. The learning curve for SA-HAs plateaued at approximately 10 cases²⁴⁾. In this study, of the 10 cases, SA-HAs were not performed by any of the young surgeons. Another limitation was that the propensity score analysis did not control for unmeasured confounders. Data on comorbidities were not collected and prescribed drugs were excluded. Alternatively, we believe that the scores for injury at home, walking ability before injury, and ASA-PS and time to surgery days (TTS) reflected the patients’ general condition. Before propensity score matching, TTS was longer for SA-HAs. The reasons for TTS over a week were fever (pneumonia and urinary tract infection), except in two cases, cerebrovascular bleeding and refusal to undergo surgery. Under these poor general conditions, the surgeons might have chosen the SA due to lack of a postoperative contraindicated limb position. Third, total hip arthroplasties (THAs) were excluded. In our hospital, THAs for treatment of FNFs is performed in patients aged <60 years or in those with coxarthrosis, such as rheumatoid arthritis. Therefore, these outcomes are limited to SA-HAs and generalizing to THAs may not be possible. When performing SA-THA, preparation of the acetabulum is difficult, and the short external rotator muscles are not preserved in all cases. Otherwise, the findings of this study demonstrated that preserving the short external rotator muscles was important for achievement of early recovery after HAs because the short external rotator muscles were preserved in all SA-HAs.

SA-HA is performed without torsion of the muscles or ligaments surrounding the hip joint. A significant feature of SA-HAs was the early recovery of walking ability and BI at the start of postoperative rehabilitation. Use of SA-HAs may lead to improved prognosis for patients with FNFs.

No potential conflict of interest relevant to this article was reported.