Factors associated with dislocation after bipolar hemiarthroplasty through an (antero-)lateral approach in elderly patients with a femoral neck fracture: a retrospective cohort study with a nested case–control subanalysis of radiographic parameters

Hip fractures are common injuries in elderly patients implicating a major burden to healthcare systems worldwide. It is estimated that the incidence of hip fracture will rise from 1.66 million in 1990 to 6.26 million by 2050 [1]. For displaced femoral neck fractures in elderly patients, that are most commonly affected, arthroplasty is the treatment of choice [2]. However, there is still an ongoing debate whether the optimal treatment in these cases is hip total versus hip hemiarthroplasty. Functional results are reported to be in favor of total hip arthroplasty (THA), but the advantage seems to be below [3] or at [4] the threshold for a minimal clinically relevant difference. On the other hand, hip hemiarthroplasty (HHA) is associated with a lower blood loss [3‐5] and a shorter duration of the surgical procedure [3, 5, 6]. Regarding surgical complications and the risk of secondary procedures THA and HHA show similar results within the first 2 years after index operation [3, 5‐7]. However, the risk of hip dislocation is 2- to 3-times higher after THA compared to HA in these patients [3, 4, 8]. Biomechanically, higher hip stability after HHA is explained by its large head to neck ratio. Dislocation occurs beyond the jump distance which is a femoral head displacement out of the acetabulum greater than half of its diameter [2]. Hence, dislocation rates after HHA are reported to be low ranging from 0 to 2.4% by most studies [3, 5, 8‐10] while some report moderate to high rates of 6–11% [11‐13]. Irrespective of the incidence, hip dislocation represents a potentially devastating complication with a reduced quality of life [14], a 2.5-fold increased mortality at 3 months [15] and a total mortality of up to 65% after 6 months [9].

Multiple patient- and surgery-related risk factors are discussed for hip dislocation after HHA. A posterior surgical approach [16, 17] and a reduced postoperative femoral offset [12, 18] seem to be associated with higher dislocation rates. On the patients’ side, mental impairment [12, 19] and anatomical aspects, as a reduced center edge angel (CEA) of the acetabulum [12, 18], are reported to promote dislocation. With regard to hip anatomy measured by radiologic parameters on plain pelvic radiographs, as coverage of the femoral head or reconstruction of lateral femoral offset, only limited information is available. Therefore, the primary aim of this study was to analyze radiologic parameters that potentially identify patients at risk for dislocation after HHA. In addition, further clinical risk factors related to patients and surgery were analyzed.

In 527 performed bipolar HHA, 10 dislocations were identified (1.9%). Baseline characteristics (Table 1) displayed no significant differences between dislocation group (DG) and control group (CG). Apart from greater trochanteric fractures (GTF), perioperative and surgical parameters did not differ significantly between the groups (Table 2).

In both the groups, no significant differences in terms of radiographical acetabular parameters between fractured and unaffected side could be found. Although no dysplastic hips were identified, the acetabular CEA was significantly smaller in the dislocation group compared to the control group (p = 0.013). In accordance, the FHEI was substantially higher in the DG (p = 0.003). Correspondingly, the BHEI was also significantly higher in the DG (p = 0.006). On the femoral side, a decreased or more varus NSA of the non-operated side was found to be associated with an increased risk of dislocation (p = 0.011). The postoperative FO did not differ between the groups (p = 0.534) (Table 3). Comparing pre- and postoperative FO demonstrated a minimal reduction by approximately 2 mm in the DG (p = 0.246) as well as in the CG (p = 0.566). A standard stem with a 135° NSAP was used in all cases. Comparing the preoperative NSA with the NSAP, a significant increase was exhibited only in the DG (p = 0.003), but not in the CG (p = 0.741).

The primary aim of the study was to identify clinical aspects and radiographic parameters of the hip that put patients at risk for dislocation after HHA for femoral neck fractures.

In the recent literature, the reported rate of dislocation after HHA in patients treated for femoral neck fractures ranged from 0 to 11% [5, 13, 21]. This is in line with our study that found a rate of 1.9%. Risk factors for dislocation of a HHA can be divided in those related to patient characteristics or those related to surgical aspects. Radiographical parameters on plain pelvic radiographs address both and may help to identify patients who have an individual anatomical risk or support postoperative analysis of potential surgical failure.

In our study, the CEA was significantly smaller in the DG compared to the CG which is in line with other studies [12, 13, 18, 22]. Only one study clearly found no influence of the CEA on dislocation rates. Noteworthy, this study exhibited a considerable change (> 10%) of the postoperative FO in more than 50% of HHA. This relevant FO difference was also identified as a significant risk factor for dislocation in this study, and subsequently, might have masked a potential effect of the CEA [23].

Another radiographical risk factor for dislocation in our study was an increased FHEI as well as the BHEI corresponding with a reduced acetabular coverage. The BHEI has been confirmed as a risk factor for dislocation by Zhang et al. [22]. In opposition, Kizkapan et al. [24] did not identify FHEI or postoperative BHEI as a risk factor for dislocation. THA has the advantage of correcting a shallow acetabulum and adjusting acetabular inclination and version which is not possible with HHA. Thus, identifying morphological acetabular risk factors on preoperative X-rays may aid the surgeon in deciding when to use total over hemiarthroplasty and to reduce dislocation rates.

The NSA on the operated side was comparable between groups which was also described by others [12, 18, 24]. However, the NSA of the unaffected side differed significantly and displayed lower values in the dislocation group. Since a standard 135° stem was used in all HHA with a dislocation, it can be concluded that preoperative NSA was not addressed correctly in our study. Postoperative FO did not influence dislocation rate in our analysis. No difference of postoperative FO in dislocated and not dislocated HHA was also confirmed by Zhang et al. [22], whereas others demonstrated an association between a smaller postoperative FO and dislocation [12, 13, 18, 24, 25]. An explanation for these diverging findings may be that FO of the unaffected side did not differ between groups and postoperative FO was restored quite well with only marginal, not significant differences to the contralateral unaffected side in our study.

Most of the few studies that analyzed radiographical parameters on plain radiographs of the pelvis presented substantial higher dislocation rates ranging from 6 to 11% [12, 13, 18, 24, 25]. One reason might be that the majority of these studies applied a posterolateral surgical approach for HHA [13, 18, 24]. The posterolateral approach tends to higher dislocation rates compared to anterior approaches [15, 16, 23, 26]. One study with a posterolateral approach found a comparatively low dislocation rate of 3% [22]. In contrast to the other studies using a posterolateral approach [13, 18, 24], this study found no difference of postoperative FO between dislocated and not dislocated HHA. Only one study that analyzed radiographic parameters exclusively applied an anterolateral approach [25]. They reported a dislocation rate of 6%, comparable to those with a posterolateral approach [13, 18, 24]. Interestingly, this study demonstrated a significantly higher FO discrepancy [25]. It might be hypothesized that restoration of anatomic FO is more important than the chosen surgical approach with regard to dislocation rate after HHA.

A higher rate of HHA dislocation was reported in patients in which surgery was delayed beyond 24 h [18, 21]. A possible explanation might be that surgery within 24 h might not always allow an optimal arrangement of the surgical team. However, experience of the surgeon did not influence dislocation frequency in our study as in others [21, 23].

Intraoperative GTF in HHA for femoral neck fractures potentially adds to hip abductor weakness and hip instability. Consequently, limping, poor clinical outcome and hip dislocation may be associated with GTF [27, 28]. Following HHA for femoral neck fractures, GTF occurred in up to 6% [29, 30]. Effectively, our study ascertained a significant association between a GTF and dislocation.

Apart from anatomical issues on the patients’ side, cognitive impairment and neurologic disease with peripheral deficiencies may be associated with postoperative HHA dislocations due to incompliance and muscular insufficiency, imbalance, or contracture. In patients with HHA dislocation, the percentage of patients with dementia was 3-times higher compared to patients without a dislocation [12, 25]. Our study revealed a dislocation rate of approximately 2% in both cognitive impaired and healthy patients, which is in line with other study results that did not identify mental dysfunction as a risk factor for dislocation after HHA [18, 21]. We and others [31] also found no influence of neurologic diseases that affect muscular function at the hip as sequelae of stroke or Parkinson’s disease.

Mortality at 1 year might be increased after HHA dislocation due to potentially necessary additional surgery, prolonged immobility and complications associated with revision surgery in this vulnerable geriatric patient cohort. In fact, earlier studies reported a high mortality rate of up to 65% within the first 3–6 months after surgery [9, 15]. In contrary, other studies exhibited only a tendency towards a higher 1-year mortality [12, 13] or could not demonstrate any impact of HHA dislocation on 1-year mortality at all as shown by us and Salem et al. [21].

This study is limited by the retrospective design and a low number of dislocations inherently associated with an (antero-)lateral approach. A nested case–control subanalysis for radiographic parameters was performed which is advantageous over a full cross-sectional cohort design and provides equal accuracy when actual disease or complication prevalence in subjects is low [32]. Due to the low number of complications, subgroups of single and recurrent dislocations were not investigated. Thus, more detailed information about HHA dislocation may have been missed. Second, cognitive deficiencies were assessed by medical records on admission and not quantified using standardized questionnaires. Third, experience of the surgeons was graded by an educational level rather than the annually performed number of hip arthroplasties. Fourth, it is not possible to address version of the acetabulum and torsion of the implant stem accurately on two-dimensional radiographs which are relevant factors in HHA dislocation and might have influenced results. Finally, two different approaches, the lateral and anterolateral approach, were used which may add to some bias.

The results of this study suggest that a smaller CEA, a higher FHEI and a smaller or more varus FNA not addressed by correct choice of an implant stem with a corresponding NSA are radiographic risk factors for dislocation in bipolar HHA through an (antero-)lateral approach. Identifying these morphological parameters on preoperative radiographs which can be addressed by THA and correct choice of implant components may reduce the rate of HHA dislocations. GTF were also associated with an increased risk for HHA dislocation. This again underlines the need for a careful surgical approach to these procedures.