Total Hip Arthroplasty: Direct Anterior Approach Versus Posterior Approach in the First Year of Practice

Abstract

Background

The direct anterior approach (DAA) for total hip arthroplasty (THA) has been popularized as a less invasive technique, however outcomes within the first year of practice after fellowship have not been investigated. The primary aim was to determine differences in complications and outcomes between DAA and posterior approach (PA) in the first year of practice. The secondary aim was to determine if there was a learning curve factor in DAA and PA after fellowship training.

Methods

THA cases performed by two surgeons during their first year of practice were reviewed. Overall, 181 THAs (91 DAA, 90 PA) in 168 patients, were performed. Intraoperative differences (blood loss, operative time), hospital stay, complications, reoperations, and revisions were compared.

Results

Overall surgical complications were similar between DAA and PA (11% vs. 9%, p=0.64), but complication profiles were different: dislocation (1% vs. 4%, p=0.17), intraoperative femoral fracture (2% vs. 1%, p=0.32), postoperative periprosthetic fractures (2% vs. 3%, p=0.64). neuropraxia (3% vs. 0%, p=0.08). There was no difference in rate of reoperation (1% vs. 3%, p=0.31). There was a difference in rate of revision at final follow-up (0% vs. 6%, p=0.02). DAA consisted of longer operative time (111 vs. 99 minutes; p<0.001), however was only significant in the first 50 cases (p<0.001), while the subsequent cases were similar (p=0.31). There was no difference in the first 50 cases compared to the subsequent cases for either approach regarding blood loss, complications, reoperations, or revisions.

Conclusion

DAA and PA for THA performed within the first year of practice exhibit similarly low complication rates, but complication profiles are different. In our series, PA did demonstrate a higher risk of revision at final follow-up. A learning curve is not unique to the DAA. Both DAA and PA THA exhibited a learning curve in the first 50 cases performed at the start of a surgeon’s practice.

Level of Evidence: III

Introduction

The direct anterior approach (DAA) for total hip arthroplasty (THA) is promoted as a less invasive, muscle-sparing approach to the hip with a true internervous plane. Previous studies report its benefits including soft tissue preservation, less postoperative pain, shorter hospital stay, improved gait mechanics, and faster recovery time.^1-7 Early studies proposed that patients may benefit from improved early functional outcomes including a lower risk of dislocation;^3,8 however, longer-term studies have suggested that patient reported outcomes (PROs) and rates of instability are equivalent between DAA and other approaches.^3,9-14 The caveat to potential early perioperative benefits is the reported “learning curve” associated with the DAA.¹⁵ Recent literature has reported significantly longer surgical times, more intraoperative blood loss, increased risk of intraoperative femur fracture, and increased wound complication rates using the DAA during a surgeon’s early caseload, including the first 50 cases.^15-21

Most residency graduates currently undergo specialized fellowship training, and the impact of formalized fellowship training on the learning curve associated with complicated arthroplasty procedures has not been previously studied.^22-25 The purpose of this study was to compare complications, reoperations, and PROs between the DAA and the posterior approach (PA) in the first year of practice. The secondary aim was to determine if there was a learning curve factor in DAA and PA after fellowship training. We hypothesized that there would be no differences between the DAA and PA THA in the first year of practice following arthroplasty fellowship regarding learning curve, operative data, complications, reoperations, and revisions.

Methods

Following Institutional Review Board approval, a retrospective review was performed on patients who underwent primary THA by two joint arthroplasty fellowship-trained orthopedic surgeons at one academic institution between September 1, 2017, and August 30, 2018. This time interval was chosen due to this being the first year of orthopedic arthroplasty practice after fellowship training for both surgeons. Both surgeons were trained in the DAA and PA as fellows at academic institutions. Exclusion criteria included age less than 18 years, direct lateral, or anterolateral approaches and THA performed due to malignancy.

Two-hundred and fourteen primary DAA or PA THA procedures (91 DAA;122 PA) in 196 total patients were performed during the study period. 32 THAs were excluded from analysis selection of PA due to significant comorbidities, leaving 91 DAA and 90 PA THAs in the final analysis (Figure 1). Fifteen patients underwent bilateral procedures (6 DAA; 9 PA) during the study period with 1 DAA and 2 PA under the same anesthetic. All other bilateral surgeries were performed during separate hospital admissions. Indications for THA included 139 primary osteoarthritis (67 DAA, 72 PA), 19 aseptic osteonecrosis (9 DAA, 10 PA), 12 fractures (9 DAA, 3 PA), 6 secondary to osteoarthritis due to previous fracture or rheumatologic disease (3 DAA, 3 PA), 5 secondary osteoarthritis due to dysplasia (3 DAA, 2 PA).

Figure 1.

Flowchart of Inclusion/Exclusion Criteria.

There was no difference in demographics or medical comorbidities between the two groups except for different in body mass index (BMI) . The mean body mass index BMI for DAA was 27 kg/m² and for PA was 33 kg/m² (p<0.001). The number of patients at or above BMI 40 kg/m² (0% vs. 26%) were different for the two approaches (p<0.001, Table 1, Table 2). The average time for final follow-up was 11 months (range: 0 to 29 months), with no difference between DAA and PA for final follow-up (10 vs. 11 months; p=0.17).

Table 1.

Direct Anterior Approach and Posterior Approach Demographics

	Total	DAA	PA	p value
Total Hips	181	91	90
Age (range)	61.0 (26.3-93.5)	62.7 (26.3-87.8)	59.27 (29.8-93.5)	p = 0.08
Sex				p = 0.94
Male	89	45	44
Female	92	46	46
BMI (range)	30.0 (14-50)	27 (14-39)	33 (18-50)	p < 0.001*
BMI 30-39	58	22	36	p = 0.02
BMI 40-49	20	0	20	p < 0.001*
BMI ≥50	2	0	2	p = 0.15
Laterality				p = 0.83
Right	99	50	49
Left	76	39	37
Bilateral	3	1	2

DAA - Direct Anterior Approach; PA - Posterior Approach; BMI: Body Mass Index; ^* indicates significance.

Table 2.

Direct Anterior Approach and Posterior Approach Comorbidities

	DAA	PA	p value
ASA			p = 0.45
1	7	5
2	46	41
3	32	41
4	6	3
Medical Comorbidities			p = 0.89
DM	13 (14.3%)	12 (13.3%)	p = 0.85
Tobacco within 1 year	18 (19.8%)	16 (17.8%)	p = 0.73
Dyspnea	7 (7.7%)	9 (10.0%)	p = 0.58
COPD	6 (6.6%)	8 (8.9%)	p = 0.56
CHF	1 (1.1%)	3 (3.3%)	p = 0.31
HTN	41 (45.1%)	41 (45.6%)	p = 0.95
ARF/Dialysis	0 (0%)	0 (0%)	NA
Immunosuppressant (including corticosteroids)	8 (8.8%)	7 (7.8%)	p = 0.81

DAA - Direct Anterior Approach; PA- Posterior Approach; DM: Diabetes Mellitus; COPD: Chronic Obrstructive Lung Disease; CHF: Chronic Heart Failure; HTN: Hypertension; ARF: Acute Renal Failure; ^* indicates significance.

Procedure

All DAA procedures were performed supine on a Hana® fracture table (Mizuho OSI, Union City, CA). All PA procedures were performed lateral decubitus on a Capello board. The implants used by the senior authors varied according to preference and availability, however all acetabular components were press-fit porous-coated, with or without screw fixation. There was a mix of press-fit and cemented stems depending on the patient’s age and bone quality. All patients received the same standardized postoperative care. Patients in the PA group received standard posterior hip precautions. The DAA patients were given no restrictions.

Data Collection

Demographic data including age, sex, BMI, American Society of Anesthesia (ASA) classification, Charlsteon Comorbidity Index, smoking status, and medical comorbidities were analyzed. Patient charts were reviewed for postoperative complications including: surgical site infection (SSI), periprosthetic joint infection (PJI), venous thromboemoblism (VTE), length of hospital stay (LOS), discharge destination, emergency department (ED) visit within 30 days, and hospital readmission within 30 days. SSI was defined as infections that did not extend past the fascia and were managed with oral antibiotics, wound cares, or superficial irrigation and debridement. PJI was defined using the Musculoskeletal Infection Society (MSIS) criteria.²⁶ Operative reports were examined for surgical approach, anesthetic type, operative time, and intraoperative complications. Blood loss was determined with computed total blood loss per the OSTHEO study calculations outlined by Rosencher and colleagues.²⁷ Patients undergoing bilateral THA under the same anesthetic (n=3) were excluded from blood loss analysis. Additionally, two more patients were excluded from this analysis due to lack of post-operative hematocrit as they discharged the same day.

Reoperation was considered a return to the operating room without removing/altering the bony prosthetic components.. Revision was defined as removal of any prosthetic component at the bone interface.

Patient reported outcomes (PROs) were collected for all patients at least preop within 3 months (90 days) prior to surgery and follow-up at least 3 months postop (90 days). The Patient Reported Outcome Measurement Information System (PROMIS) global health scores for mental health (MH) and physical function (PF) were analyzed.

Learning Curve

The first 50 cases were compared to the subsequent procedures, as well as the first six months (9/1/17–2/28/18) to the last six months (3/1/18–8/30/18) within the specific approach.^15,17 Additionally, the first 6 months (42 DAA and 39 PA cases) were compared to the subsequent cases (49 DAA and 51 PA) cases during the first year of practice.

Statistical Methods

An unpaired, unequal variance, two-tailed t-test was used to compare continuous variables. Fisher exact and Pearson Chi-Square tests were used to evaluate significant differences in categorical variables including complications between the groups. Stata (version 15.1; StataCorp, College Station, TX, USA) was used for all statistical analysis, with significance determined by a P value <0.05.

Results

Operative details and hospital stay

Mean procedure time was significantly longer for DAA (111 vs. 99 minutes; p<0.001). There was no difference in the mean LOS (2.2 vs. 2.0 days, p=0.53) or patient disposition location; however, more DAA patients were discharged to a nursing facility (13% vs. 7%; p=0.15; Table 3) The average computed blood loss was 1183 mL (range: 126-3136 mL) with no difference between the approaches (p=0.06; Table 4).

Table 3.

Hospital Stay Characteristics and Complications

	All Approaches	DAA	PA	p value
Cases	181	91	90
Anesthesia
Regional	119	65	54
General	62	26	36	p = 0.11
Procedure
Duration (minutes)	105	111	99	p = 0.001*
Post-operative
PRBC Transfusion (total units)	4	3	1	p = 0.32
In Hospital Complications	3	0	3	p = 0.37
Pneumonia	1	0	1	p = 0.32
UTI	2	0	2	p = 0.15
DVT/PE	0	0	0	NA
Hospital Stay (days)	2.0	2.2	2.0	p = 0.53
Disposition				p = 0.15
Home	156	76	80
Home, which is a Facility	3	1	2
Skilled Nursing Facility	18	12	6
Acute Rehab Facility	1	1	0
Follow-up (months)	10.7	10.0	11.5	p = 0.17

DAA - Direct Anterior Approach; PA - Posterior Approach; PRBC: Packed Red Blood Cells; UTI: Urinary Track Infection; DVT: Deep Vein Thrombosis; PE: Pulmonary Embolism; ^* indicates significance.

Table 4.

Direct Anterior Approach and Posterior Approach Total Blood Loss

	All Approaches	DAA	PA	p value
Computed Total Blood Loss (mL)	1183	1109	1260	p = 0.06

DAA - Direct Anterior Approach; PA- Posterior Approach; ^* indicates significance.

Patient Reported Outcomes

There were a total of 44 (48.4%) DAA and 39 (43.3%) PA PRO were completed. The mean preoperative and postoperative PROMIS-PH scores were 40.3 (±7.7) and 46.3 (±8.2) for DAA and 38.6 (±7.3) and 45.0 (±10.3) for PA. Mean PROMIS-MH scores were 48.2 (±8.1) and 49.1 (±8.3) for DAA and 49.5 (±9.7) and 49.2 (±8.6) for PA. There was no difference between the groups when comparing the change of preoperative to postoperative scores (p=0.31, p=0.71).

Dislocations

Overall, 2.8% (n=5) primary THAs had a dislocation event. There was no significant difference in dislocation rate between the two approaches (1% DAA vs. 4% PA, p=0.17). One case in the DAA group (1.1%) dislocated, that was managed by closed reduction. Four dislocations occurred in the PA group (4.4%), of which two were successfully closed reduced and two eventually underwent revision due to recurrent instability. (Table 5)

Table 5.

Complications, Reoperations, Revisions

	DAA	PA
Total Hips	91	90
Overall Reasons for Reoperations and Revisions	1	8
Superficial Surgical Site Infection	0	1
Recurrent Instability	0	2
PJI	0	3
Periprosthetic Femur Fracture	2	3
Aseptic Loosening	0	0
Other Complications
Intraoperative Fracture	2	1
Superficial Infection	3	3
Dislocation	1	4
Nerve Injury	3	0

DAA - Direct Anterior Approach; PA - Posterior Approach; PJI - Periprosthetic Joint Infection; ^* indicates significance

Periprosthetic Fractures

Overall, an intraoperative femoral fracture occurred in 1.7% of THA procedures (2 DAA vs. 1 PA, p=0.48). All fractures were successfully managed with intraoperative cables and/or postoperative protected weight bearing. There were five (2.8%) postoperative periprosthetic fractures (2 DAA vs. 3 PA, p=0.64). Postoperative periprosthetic fracture events are detailed in Table 6.

Table 6.

Periprosthetic Femur Fracture

Age	Sex	BMI	Comorbidities	Time (post op)	MOI	Vancouver Classification	Management
Direct Anterior Approach
65	Female	13	Myotonic dystrophy, Osteopenia	2 weeks	Felt a “pop” and severe pain during abduction while getting out of bed	Vancouver A (Greater trochanter with 3 mm displacement)	Non-operative: TTWB, No active abduction
65	Female	22	DDH, Osteoarthritis of Lumbar Spine with fusion to Pelvis	4 weeeks	TTWB due to intraoperative fracture treated by removing the stem, placing 2 proximal and one distal cerclage wires around the fracture to reduce it, and passing the stem through the reduced fracture. Placed NWB 2 weeks post op due to subsidance of stem. Sustained a ground level fall resulting in Vancouver C periprosthetic fracture.	Vancouver C	Reoperation: Retention of stem. Proximal femur lateral locking plate and cerclage wiring
Posterior Approach
89	Female	40	COPD, Right MCA stroke with contralateral weakness, Lumbar fusion	5 weeks	Ground level fall while getting out of bed	Vancouver A (Greater trochanter)	Non-operative: TTWB, No active abduction
52	Female	33	Crohn’s disease (long term corticosteroid use), chronic pain, chronic alcohol use disorder, anxiety, migraine, borderline personality	4 weeks	Relapse of her depression as well as her alcohol abuse, has had multiple falls. Admitted to psychiatric unit for suicidal attempt. Ground level fall directly on the ipsilateral hip in the shower. Radiographs demonstrated ubsidence of her femoral component and medial fracture confirmed on CT scan	Vancouver B	Revision: New femoral stem with cerclage wiring
81	Female	27	Cervical Myelopathy, ataxia lumbar spinal stenosis, scoliosis and symptoms of neurogenic claudication, CKD	3 weeks	Missed hand grip of walker resulting in a ground level fall	Vancouver A (Medial Cortex above Lesser Trochanter)	Revision: New femoral stem with cerclage wiring

DAA: Direct Anterior Approach; PA: Posterior Approach; MOI: Mechanism of Injury; BMI: Body Mass Index; ^* indicates significance.

Infection

There was no difference in the rate of SSI between approaches (3 DAA vs. 3 PA; p=0.64). Two PA hips were diagnosed with acute PJI with MSIS major criteria (>2 positive intraoperative cultures (staphylococcus lugdunensis, morganella morganii). One PA hip was diagnosed with chronic PJI using MSIS major criteria (>2 positive tissue cultures of staphylococcus aureus), 18 months from primary procedure.

Nerve Related Complications

Three (3%) neurologic injuries occurred in the DAA group (2 lateral femoral cutaneous nerve dysesthesias, 1 partial femoral nerve neurapraxia; p=0.4) and none in the PA group. The three nerve injuries in the DAA approach remained present at final follow-up (Table 5).

Medical Complications

There was no difference between the two approaches in regard to readmission rates within 30 days of primary procedure (1.1% vs. 3.3%, p=0.30). A total of 3 (3.4%) PA patients had medical complications. None of the DAA cohort (0%) had medical complications (Table 3).

Reoperations

Overall, 4 hips (2.4%) required reoperation (Table 5). One (1.1%) DAA underwent reoperation due to Vancouver C periprosthetic fracture, requiring cable fixation and lateral locking plate (Table 6). As previously reported, three PA hips underwent reoperation for infection, with one case requiring superficial irrigation due to SSI, with no evidence of arthrotomy. Two cases underwent irrigation and debridement with polyethylene exchange due to acute PJI. There was no difference in rate of reoperation between the two groups at final follow-up (p=0.31) (Table 5).

Revisions

There were no revision procedures performed in the DAA cohort, compared to 5 (6%) in the PA cohort. Two revision cases were due to recurrent instability. Two revision cases were due to postoperative periprosthetic fractures. Both cases underwent successful revision of the femoral component to a fluted, tapered design. One revision was due to the previously mentioned chronic PJI that occurred 18 months after primary procedure. There was a difference in rate of revisions between the two groups at final follow-up (p=0.02) (Table 5). In subgroup analysis with exclusion of patients with a BMI ≥40 kg/m² (n=22), there continued to be a difference present (p=0.04).

Learning Curve

The BMI was lower in the first 50 DAA cases when compared to the subsequent cases (p=0.02). Although the DAA consisted of longer operative time when compared to the PA, this was only significant in the first 50 cases (p<0.001), while the subsequent cases were similar (p=0.31).

There was a longer LOS when comparing the first 50 DAA cases to the subsequent cases (p=0.03). Age, blood loss, rate of complications, reoperations, and revisions did not change between the early cases to the subsequent cases. Additionally, there was no difference in complications, reoperations, and revisions when comparing the first months of practice to the subsequent cases in the first year. (Table 7)

Table 7.

Direct Anterior Approach and Posterior Approach Learning Curve

	First 50 Cases	Subsequent Cases		Months 1-6	Months 7-12
DAA
Age (years)	63	62	p = 0.63	64	61	p = 0.23
BMI	26	28	p = 0.02*	26	28	p = 0.04*
Computed Total Blood Loss (mL)	1111	1108	p = 0.98	1041	1135	p = 0.57
Duration (minutes)	114	108	p = 0.15	114	109	p = 0.29
Length of Stay (days)	2.7	1.6	p = 0.03*	2.6	1.8	p = 0.13
Reoperations and Revisions	1	0	p = 0.36	1	0	p = 0.29
PA
Age (years)	59	60	p = 0.64	59	59	p = 0.95
BMI (kg/m2)	32	34	p = 0.43	32	33	p = 0.52
Computed Total Blood Loss (mL)	1263	1256	p = 0.96	1114	1296	p = 0.14
Duration (minutes)	96	104	p = 0.09	96	101	p = 0.30
Length of Stay (days)	2.1	1.9	p = 0.66	2.1	2.0	p = 0.86
Reoperations and Revisions	6	2	p = 0.25	5	3	p = 0.20

DAA - Direct Anterior Approach; PA- Posterior Approach; BMI - Body Mass Index (kg/m2) ^* indicates significance.

Discussion

This study took advantage of the unique opportunity to follow the practices of two new, fellowship trained surgeons. We investigated differences between the DAA and the PA during the first year of practice after completing a joint arthroplasty fellowship. We found no difference in blood loss, transfusion rate, LOS, complications, readmissions, reoperations, or PRO between the two approaches. The DAA did consist of longer procedure duration, while the PA had a higher incidence of revisions at final follow-up. This study also assessed the impact of learning curve by comparing the first 50 cases with subsequent cases and cases performed in the first 6 months versus the second 6 months. We found minimal evidence for DAA or PA learning curve except in operative time.

There continues to be conflicting evidence on the differences in blood loss, operative time, LOS, and institutional disposition between the DAA and conventional approaches,^14,28,29 with several studies demonstrating higher surgical blood loss, longer procedure duration, and shorter LOS associated with the DAA approach.^14,16,30 We found no difference in blood loss or transfusion rate, LOS and non-institutional discharge between the approaches.

Intra-operative fractures of the proximal femur are often discussed in the DAA, with a reported incidence of 1.4-2.3%.^31-33 Femoral fracture risk appears to be relatively higher in the DAA.³⁴ While the current study demonstrated no statistical significant difference between the two approaches, the incidence of an intraoperative fracture was more common in the DAA compared to the PA, consistent with previous research.

Due to its intermuscular nature and preservation of the posterior soft tissue envelope, the DAA has been historically known to have a lower dislocation risk compared to the PA.⁸ More recent studies have reported similar dislocation rates in the DAA and PA without significant difference between the two approaches, citing the positive effects of modern techniques of posterior soft tissue repair, larger femoral head sizes, and increased offset stems in reducing dislocation rates.^30,35-39 While our results demonstrated no statistical difference in dislocation rates between the approaches, instability was more common in the PA patients. It is important to note the patient selection bias involved in the current study, as the DAA surgeon did reserve PA for heavier patients early in the learning curve, who may have been at risk for dislocation due to elevated BMI.

Transient neuropraxia of the LFCN has been reported to be as high as 81% of patients undergoing DAA THA,^40-43 and represents the most common complication using this approach.³³

However, most studies report significant improvement or resolution of sensory disturbance over time.^40,41 In PA THA, the sciatic nerve is most at risk, with a reported 0.3-1.3% incidence of injury.^44-47 In this study’s cohort, no PA cases sustained nerve injuries, while two DAA cases sustained neuropraxia to the LFCN and one case with reported partial femoral neuropathy.

In our study, there was no difference in reoperation rates between the two groups; however, five revisions were performed in the PA group while no DAA cases required revision

during the study period. Revisions in the PA group were performed specifically for two periprosthetic fractures, two cases of recurrent instability, and one patient with PJI that occurred 18 months after primary procedure. Consistent with our findings, Gwam et al. reviewed 258,461 revision cases using the National Inpatient Sample database and demonstrated that persistent dislocation events are the most common etiology for revision THA.⁴⁸ Sheth et al. examined a decade’s worth of data and compared DA, PA, direct lateral, and anterolateral approaches in regard to aseptic revision, revision for infection, and dislocation.⁴⁹ They demonstrated no difference in adjusted risk for revision in any of the groups, however the anterolateral and DAA groups had a significantly lower risk of dislocation when compared to the PA. In contrast to our findings and the previously mentioned studies, Pincus et al. performed a large propensity-score matched analysis of 5,986 cases and suggested that the DAA is associated with both a higher risk of revision arthroplasty, deep infection requiring reoperation, and dislocation required closed or open reduction.⁵⁰ However, this study matched the DAA cases (n=2,993) to a heterogeneous group that included only 22% PA cases (n=667) and 78% direct lateral cases (n=2326), providing confounding conclusions when comparing DAA to PA.

Learning Curve

The learning curve has been a significant interest when discussing advanced surgical approaches. Previous studies have investigated complication rates in surgeons changing approaches, resulting in a steep learning curve with reports of 20 to 300 cases.^18,21,51-56 This in turn has many senior surgeons questioning the benefits of transition and incorporation of the DAA into their practice.^57,58 However, there are no known studies investigating the learning curve and associated outcomes within the first year of practice after a dedicated arthroplasty fellowship.

In a comparison of DAA and PA used by surgeons early in their training, Spaans et al. evaluated two equal numbered groups of age-and comorbidity-matched patients across a year-long time period.¹⁸ Patients undergoing DAA THA experienced nearly twice the operative time and blood loss, in addition to a higher complication rate. The study did trend a decrease in operative time as surgeons gained experience, but this was not found to be statistically significant. We demonstrated similar findings, with a difference between DAA and PA operative time only being present when comparing the first 50 cases.

Limitations

Cases were reviewed from a single academic center, which limits the generalizability of the findings. Additionally, the specific approaches were performed by separate surgeons. This study’s cohort had mean BMI of 30, including 22 patients ≥40 kg/m² in the PA cohort and 0 in the DAA patient cohort. While there continued to be a difference with exclusion of this subgroup, the significant proportion of obese patients in the PA group may have impacted the results of this study as it is known that morbid obesity is clearly associated with intraoperative differences (blood loss, operative time), hospital stay, complications, reoperations, and revisions.^59-63 Additionally, the amount of PRO forms that patients completed were minimal, providing a lack of conclusions on PRO scores.

Conclusion

Both DAA and PA for THA have similarly low complication rates within the first year of practice following fellowship training. A learning curve is not unique to one approach. Femoral complications and nerve injuries are more common in DAA, while dislocation is more common in the PA.