Introduction

Femoral neck fractures are common injuries among the elderly and represent a major source of morbidity and mortality [1, 2]. While some non-displaced fractures can be managed non-surgically, displaced femoral neck fractures generally require surgical treatment [3]. For elderly patients with multiple comorbidities, physiologic compromise, and decreased functional demands, hemiarthroplasty is often performed. The benefits of hemiarthroplasty include early ambulation, minimization of operative time, and improved stability [4,5,6].

Two important considerations when managing displaced femoral neck fractures are early mobilization and minimization of perioperative complications [1, 2, 7]. These parameters may be affected by choice of surgical approach. The direct anterior approach (DAA) is a popular technique for elective total hip arthroplasty (THA), with a growing body of literature supporting its association with reduced risk of dislocation, decreased postoperative pain, faster rehabilitation, and shorter length of hospital stay compared to other surgical approaches [5, 8,9,10]. Despite this trend in THA, most hemiarthroplasties for the treatment of femoral neck fractures are done through lateral or posterior approach variations [11, 12]. We performed a systematic review and meta-analysis to evaluate whether treatment of displaced femoral neck fractures with hemiarthroplasty using the DAA is associated with improved functional and clinical outcomes when compared with hemiarthroplasty performed using other surgical approaches.

Materials and methods

Prior to our literature search, we drafted a protocol outlining our search strategy, inclusion/exclusion criteria, and outcomes of interest. The study protocol was registered on PROSPERO, an international prospective register of systematic reviews and meta-analyses [13]. We conducted our review utilizing standard methodology as outlined in the Cochrane Handbook and reported results in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines [14, 15].

Eligible studies included randomized control trials and observational, comparative studies of hemiarthroplasty performed via direct anterior approach versus hemiarthroplasty performed through another surgical approach (anterolateral, lateral, posterolateral, posterior) for treatment of femoral neck fractures, in which at least one quantifiable pre-specified outcome measure was reported. Our primary outcome was postoperative functional mobility. Secondary outcomes included overall complication rate, dislocation rate, perioperative fracture rate, infection rate, re-operation rate, overall mortality, operative time, pain, perioperative blood loss, and length of stay.

We searched Medline [Ovid], EMBASE [Ovid], CINAHL [Ovid], and the Cochrane Library databases, from their date of inception until October 2016. No limits were applied to searches regarding time, publication status, or language. Google Scholar was searched using similar criteria, and the first 20 pages of search results were reviewed. To identify studies from unpublished sources, we searched for registered trials on ClinicalTrials.gov and reviewed publicly available abstract archives from the Orthopaedic Research Society (ORS), Orthopaedic Trauma Association (OTA), American Academy of Orthopaedic Surgeons (AAOS), and the American Association of Hip and Knee Surgeons (AAHKS). If no full manuscript could be located for identified abstracts, attempts were made via email to contact the authors and obtain full study datasets. Manuscripts written in a language other than English were translated by a medically trained individual fluent in that language.

Studies were selected for inclusion, in accordance with our pre-specified criteria, by two reviewers (SK, MS). Both reviewers performed data extraction independently, using a standardized data collection form. Discrepancies regarding study inclusion or data extraction were resolved through discussion and consensus by the reviewers. The surgical technique described in each manuscript was confirmed by blinded evaluation by a fellowship-trained arthroplasty surgeon (WM).

Risk of bias assessment was performed independently by two reviewers (SK, MS). An additional risk of bias assessment was performed by a third reviewer (RK) blinded to study title, authors, year, and journal of publication. Discrepancies were resolved through discussion and consensus by the reviewers. The Cochrane Collaboration risk-of-bias assessment tool was utilized for randomized trials, and the Newcastle–Ottawa Scale risk-of-bias assessment tool was utilized for non-randomized studies [16, 17].

Analysis

Qualitative analysis was used to evaluate outcomes for which substantial variability in method or timing of assessment among studies precluded meaningful quantitative analysis. Quantitative analysis was performed for comparably assessed outcomes if substantial heterogeneity of the data element was not present. Heterogeneity was assessed using the I 2 statistic, with the limit for substantial heterogeneity pre-defined as I 2 > 60% [14]. Quantitative analysis was performed using Review Manager 5.3 (The Nordic Cochrane Centre, The Cochrane Collaboration). When necessary for statistical comparison, conversion of median and range to mean and standard deviation was performed using the method of Hozo et al. [18]. Fixed effect odds ratios (OR) and 95% confidence intervals (CI) were used to analyze dichotomous variables; mean differences were used to analyze continuous variables [14]. Statistical significance was set at p < 0.05.

Subgroup analysis was performed to evaluate outcomes among approaches grouped by direction of intra-operative, surgical dislocation. Defined subgroups include DAA versus other approaches involving anterior intra-operative dislocation (anterolateral approach and lateral approach), and DAA versus posterior intra-operative dislocation (posterolateral approach and posterior approach).

Results

Eleven studies were identified that met inclusion criteria; however, two studies were excluded due to inability to obtain full text (Fig. 1). A summary of the 9 studies included in our analysis, including a total of 698 hips (330 direct anterior, 57 anterolateral, 89 lateral, 114 posterolateral, 108 posterior), is provided in Table 1. Among included studies, there was uniform distribution of age, sex, and comorbidity [as assessed by American Society of Anesthesiologists (ASA) score] of patients. Body mass index (BMI) did not vary substantially within studies, with mean BMI reported for DAA and other approaches at 22.6 (range 20.6–26.0) kg/m2 and 22.4 (range 19.8–26.0) kg/m2, respectively.

Fig. 1
figure 1

Study selection flow diagram: AAHKS American Association of Hip and Knee Surgeons, AAOS American Academy of Orthopaedic Surgeons, ORS Orthopaedic Research Society, OTA Orthopaedic Trauma Association, CDSR Cochrane Database of Systematic Reviews, CENTRAL Cochrane Central Register of Controlled Trials, CINAHL Cumulative Index to Nursing and Allied Health Literature, EMBASE Excerpta Medica dataBASE

Full size image
Table 1 Characteristics of included studies
Full size table

Methodological quality of included studies

Methodological quality assessment is presented in Fig. 2. Randomized studies demonstrated potential for systematic error due to lack of treatment allocation concealment, blinding of personnel/participants, and blinding of outcome assessment. Non-randomized studies demonstrated variable potential for systematic error with regard to comparability of cohorts and outcomes assessment. Publication bias was not assessed due to the limited number of papers included in our study.

Fig. 2
figure 2

Assessment of methodological quality

Full size image

Functional mobility

Eight studies reported functional outcomes assessment. The functional outcome measure utilized, as well as timing of assessments varied considerably among studies, precluding quantitative comparative analysis [4,5,6, 19,20,21,22,23]. Results of individual studies are reported in Table 2. Timing of postoperative functional mobility assessment ranged from 1 day to 1 year after surgery, with each study reporting at least one assessment that occurred prior to or around 1 month after surgery.

Table 2 Qualitative Summary of Functional Mobility Assessment
Full size table

Four studies reported significant functional mobility differences at various postoperative time points, all favoring the DAA [20,21,22,23]. Four studies reported no significant differences in functional mobility for the DAA compared to other approaches [4,5,6, 19]. Of the four studies that reported no statistically significant differences, two reported non-significant findings favoring the DAA, and two reported non-significant findings favoring other approaches [4,5,6, 19]. No studies reported statistically significant superior functional results with another surgical approach compared to the DAA.

Overall complication rate

All studies reported overall complication rates. No significant differences were found for overall complication rates in either overall (Fig. 3a) or subgroup analysis (Fig. 3b). The overall complication rate for the DAA was 20.3% and for other approaches was 25.5% [OR 0.92 (95% CI 0.62–1.35); p = 0.66].

Fig. 3
figure 3

a Overall complication rate for DAA versus control approaches and b subgroup analysis of overall complication rate for DAA versus anterolateral and direct lateral approach, and for DAA versus posterolateral and posterior approach [the solid squares denote the mean difference, the horizontal lines represent the 95% confidence intervals (CI), and the diamond denotes the cumulative odds ratio]

Full size image

Dislocation rate

Dislocation rate was reported in all studies. Meta-analysis demonstrated significantly fewer dislocations with the DAA (0.6%) than with other approaches (5.4%) [OR 0.28 (95% CI 0.06–0.54); p = 0.003] (Fig. 4a). Subgroup analysis demonstrated significantly fewer dislocations among DAA (1.1%) versus posterolateral and posterior approaches (7.8%) [OR 0.18 (95% CI 0.05–0.63); p = 0.007]; however, no significant difference was shown when DAA (0%) was compared with anterolateral and direct lateral approaches (1.9%) [OR 0.19 (95% CI 0.01–4.03); p = 0.29] (Fig. 4b).

Fig. 4
figure 4

a Overall dislocation rate for DAA versus control approaches and b subgroup analysis of dislocation rate for DAA versus anterolateral and direct lateral approach, and for DAA versus posterolateral and posterior approach [the solid squares denote the mean difference, the horizontal lines represent the 95% confidence intervals (CI), and the diamond denotes the cumulative odds ratio]

Full size image

Perioperative fracture rate

Fractures that occurred during surgery or the immediate postoperative recovery period were evaluated. Trinh et al. [4] reported a peri-prosthetic fracture in their control group which occurred several months after the index surgery. This was excluded from our analysis. No significant differences were found for perioperative fracture rate in either overall (Fig. 5a) or subgroup analysis (Fig. 5b). The overall perioperative fracture rate for DAA was 1.8% and for other approaches was 1.1% [OR 1.4 (95% CI 0.48–4.08); p = 0.67].

Fig. 5
figure 5

a Overall perioperative fracture rate for DAA versus control approaches and b subgroup analysis of perioperative fracture rate for DAA versus anterolateral and direct lateral approach, and for DAA versus posterolateral and posterior approach [the solid squares denote the mean difference, the horizontal lines represent the 95% confidence intervals (CI), and the diamond denotes the cumulative odds ratio]

Full size image

Infection rate

Infection rates were reported in all studies. No significant differences were found for infection rate in either overall (Fig. 6a) or subgroup analysis (Fig. 6b). The overall infection rate for DAA was 0.6%, and for other approaches was 1.6% [OR 0.58 (95% CI 0.18–1.88); p = 0.36].

Fig. 6
figure 6

a Overall infection rate for DAA versus control approaches and b subgroup analysis of infection rate for DAA versus anterolateral and direct lateral approach, and for DAA versus posterolateral and posterior approach [the solid squares denote the mean difference, the horizontal lines represent the 95% confidence intervals (CI), and the diamond denotes the cumulative odds ratio]

Full size image

Re-operation rate

Re-operation rates were reported in all studies. No significant differences in re-operation rate were found in either overall (Fig. 7a) or subgroup analysis (Fig. 7b). The overall re-operation rate for DAA was 1.2%, and for other approaches was 2.2% [OR 0.73 (95% CI 0.24–2.20); p = 0.58].

Fig. 7
figure 7

a Overall re-operation rate for DAA versus control approaches and b subgroup analysis of re-operation rate for DAA versus anterolateral and direct lateral approach, and for DAA versus posterolateral and posterior approach [the solid squares denote the mean difference, the horizontal lines represent the 95% confidence intervals (CI), and the diamond denotes the cumulative odds ratio]

Full size image

Overall mortality

Mortality rates ranged widely among included studies (from 0 to 25%), consistent with substantial variability in the length of time patients were followed for this outcome. No significant differences were found for mortality in either overall (Fig. 8a) or subgroup analysis (Fig. 8b). The mortality rate for DAA was 7.6%, and for other approaches was 6.8% [OR 1.17 (95% CI 0.65–2.11); p = 0.61].

Fig. 8
figure 8

a Overall mortality by final follow-up for DAA versus control approaches and b subgroup analysis of mortality by final follow-up for DAA versus anterolateral and direct lateral approach, and for DAA versus posterolateral and posterior approach [the solid squares denote the mean difference, the horizontal lines represent the 95% confidence intervals (CI), and the diamond denotes the cumulative odds ratio]

Full size image

Operative time

High heterogeneity (I 2 = 90%) precluded quantitative analysis of operative time. Three studies reported significantly longer operative times for the DAA [5, 6, 19]. One study reported significantly shorter operative times for the DAA group [24]. Mean operative duration ranged considerably from 47 to 85.6 min for the DAA, and 54 to 79.3 min for the other approaches. Mean differences within individual studies between DAA and other approaches ranged from 2 to 23.8 min.

Pain

Variability among pain assessment metrics utilized and timing of assessment precluded quantitative analysis. Five studies reported pain outcomes, with 3 favoring DAA and 2 favoring other approaches [5, 6, 21, 22, 24]. Langlois et al. [5] assessed visual analog scale (VAS) scores at postoperative day 5 and found no difference. However, at their final follow-up (mean 22 ± 5.1 months for DAA, 21 ± 5.1 months for controls) patients in the DAA group reported more pain (5.6 ± 1.1 vs. 4.9 ± 1.4, p = 0.03) on the Postel Merle d’Aubigné (PMA) metric. Pala et al. [24] reported numeric rating scale scores and found significantly less pain in the DAA group (1.5) versus controls (2.1), p = 0.001 (timing of assessment was not specified). Renken et al. [22] reported VAS scores at postoperative days 1, 5, 16, and 40. They reported lower median scores in the DAA group at day 16 (1 ± 1.33 vs. 2 ± 1.53, p = 0.035) and day 40 (0 ± 0.31 vs. 1 ± 0.82, p = 0.0004). Conversely, Auffarth et al. [6] assessed VAS scores at 12, 24, 48, 72, and 96 h postoperatively and reported higher pain scores at every time point in the DAA group, with a mean difference between the DAA and control groups at each time point of 0.58 (p = 0.024). Preininger et al. [21] reported number of days patients required PRN IV analgesics (in addition to baseline 550 mg naproxen BID) and found significantly longer requirements in the DAA group (4.4 ± 4 days) versus controls (3.5 ± 7 days) p = 0.04.

Perioperative blood loss

Eight studies reported perioperative blood loss. Variation in method of assessment precluded quantitative analysis. Of the 5 studies that evaluated postoperative hemoglobin drop, 4 showed no significant difference between the DAA and other approaches [4,5,6, 22]. Pala et al. [24] reported a significantly lower decrease in hemoglobin levels in the DAA group [1.5 (0–5.4) g/dL] versus controls [1.9 (0–4) g/dL], p = 0.02. Of the 3 studies that reported volumetric blood loss, 2 found no significant difference [4, 19]. Tsukada and Wakui [23] found significantly higher volumetric intra-operative blood loss in the DAA group (370.1 mL ± 192.1) compared to controls (230 mL ± 114.9) p = 0.0002. Three studies reported postoperative transfusion requirement and found no significant difference between approaches [4, 5, 21].

Length of stay

Substantial differences in length of stay (5.3–36.1 days), likely reflective of health systems and cultural differences, were present among the studies and precluded meaningful quantitative comparison. No significant differences in length of stay were reported within any of the individual studies.

Discussion

This is the first systematic review and meta-analysis to evaluate the direct anterior approach for hemiarthroplasty in the setting of femoral neck fracture. With regard to early functional mobility, 4 of the 8 included studies that reported functional outcome metrics favored DAA, and no study favored other approaches over the DAA, suggesting superior early functional outcome with the DAA. Conversely, qualitative analysis of operative time, postoperative pain, perioperative blood loss, and length of stay did not clearly favor either the DAA or other approaches. Moreover, with the exception of dislocation rate, there were no significant differences seen for any of the evaluated outcomes in our overall or subgroup quantitative analysis.

The overall risk of dislocation among included studies was 3.2% (0.6% for the DAA, 5.4% for other approaches). Subgroup analysis further demonstrated a significant difference between dislocation rates for the DAA and posterior capsular surgical approaches (posterolateral and posterior). Subgroup analysis of dislocation rates for the DAA versus other anterior capsular approaches (anterolateral and direct lateral), also favored the DAA, but failed to demonstrate statistical significance. These findings support prior literature reporting lower dislocation rates with anterior capsular approaches than posterior capsular approaches for hemiarthroplasty [25,26,27]. Moreover, the dislocation rate for DAA seen in our meta-analysis (0.6%, ranging from 0 to 2.6% among individual studies) is well below previously published dislocation rates for hemiarthroplasty overall (1.6–16%) and lower than those published for other anterior capsular approaches, anterolateral (0–6%) or direct lateral (0.9–3.3%) [7, 27,28,29,30,31,32,33].

Prior literature has demonstrated that dislocation is a serious complication which may be detrimental in terms of quality of life and mortality [29, 34, 35]. Mortality within 6 months of hemiarthroplasty dislocation has been reported to be as high as 65–73% [30, 34]. We believe the low risk of dislocation associated with the DAA merits consideration. Additionally, use of the DAA may allow for avoidance of restrictive postoperative hip precautions. Although postoperative hip precaution protocols were not assessed in this review, for elderly patients with high rates of comorbidities adversely affecting motor skills and cognition, this may also be a salient consideration.

There are several limitations of this systematic review and meta-analysis. The paucity of studies relevant to this topic did not allow for subgroup analysis of each individual surgical approach. Similarly, too few studies met our inclusion criteria for quantitative assessment of potential publication bias to be performed [14]. Furthermore, the limited number of randomized control trials available necessitated inclusion of non-randomized studies. Moreover, rigorous analysis of methodological quality identified some areas of potential bias in the included studies. This was largely related to lack of allocation concealment and participant/personnel blinding. Both of these limitations are common among surgical intervention studies, particularly in the setting of trauma. Also, analysis of functional outcomes was limited by variation among individual studies in type of metrics utilized and timing of assessment. This issue was previously noted by Hutchings et al. [36] in their review of outcomes assessment for proximal femoral fractures and remains a limitation for any literature-based analysis of functional outcomes in this population.

Our analysis was unable to fully account for some technical aspects of the surgical intervention, including use of cemented versus non-cemented prosthetic devices, type of prosthesis (bipolar vs. monopolar), and whether capsular repair was performed. Cement fixation of implants has been shown to confer benefits with regard to functional mobility and implant-related complications [11, 37, 38]. Among the included studies, 2 did not use cemented implants, and 3 failed to specify whether cement was used [4, 5, 19, 20, 23]. However, there did not appear to be substantial differences within individual studies regarding use of cemented implants for the DAA versus other approaches. Variability was found regarding the use of bipolar prostheses (8 studies reported their use in some or all patients, 1 study failed to specify prosthesis type). However, because current literature does not clearly support the superiority of either option, we believe it unlikely that this would substantially impact our measured outcomes [11, 38].

Several of the included studies did not provide information on whether capsular repair was performed, which may be a pertinent factor affecting dislocation rate among posterior capsular approach patients [39]. However, the three studies which compared the DAA to posterior or posterolateral approaches, and specified that capsular repair was performed, demonstrated a higher combined rate of dislocation (10/99 (10.1%)) among posterior capsular approach patients than the two studies of DAA versus posterior or posterolateral approach which did not explicitly specify whether capsular repair had been performed [5/94 (5.3%)] [5, 19, 20, 23, 24].

Of note, the experience level of surgeons in the included studies was varied (Table 1). This may be an important consideration as the DAA has been reported to have a relatively steep learning curve [4, 24, 40, 41]. However, among the 8 studies that specified surgeon experience level, there did not appear to be substantial intra-study variation in surgeon experience between approaches, with the exception of Trinh et al. [4], who compared the DAA to multiple other surgical approaches and whose data were not included in our subgroup analyses. Also, because these injuries are commonly treated by surgeons with a diversity of training and experience, we believe that variation in among surgeon experience level likely adds to the generalizability of our findings.

Conclusion

Despite the stated limitations, our analysis suggests that the DAA provides superior early functional mobility compared to other surgical approaches. Our analysis did not demonstrate a significant difference between the DAA and other approaches with regard to overall complications, perioperative fracture, infection, re-operation rate, mortality, operative time, postoperative pain, perioperative blood loss, or length of stay. The data do, however, demonstrate that the DAA is associated with a significantly lower rate of dislocation when compared with posterior capsular approaches. What proportion of this is attributable to the muscle-sparing aspects unique to the DAA, versus the anterior capsular dissection and direction of intra-operative dislocation shared with other anterior capsular approaches, remains unclear. Further high-quality randomized trials are needed to provide definitive insight into this question and evaluate effect of surgical approach on clinical and functional outcomes after hemiarthroplasty.