Background
The proportion of elderly people is increasing as the world’s population ages, resulting in an expected rise in the incidence of osteoporotic hip fractures. It is estimated that about 1.6 million hip fractures occurred in the year 2000 [
1], and the incidence of hip fractures is expected to increase to over six million worldwide by the year 2050 [
2]. About half of the hip fracture population has displaced femoral neck fracture (FNF, Garden type III or IV) of the subcapital region [
3]; displaced FNFs can result in non-union or avascular necrosis [
4]. Moreover, these fractures are associated with impaired mobility, loss of function, and personal dependence as well as with global economic health costs, and are significant causes of mortality and morbidity in the elderly [
5,
6]. The optimal treatment of displaced FNF in the elderly is an ongoing scientific and clinical debate [
7].
Surgical treatment options for displaced FNF include internal fixation, which is not recommended in elderly patients [
8‐
11], and arthroplasty. Both hemiarthroplasty (HA) and total hip arthroplasty (THA) are widely accepted methods of hip replacement after displaced FNF. Some evidence has suggested that THA leads to better functional outcome than HA [
7]; however, there are some advantages of HA compared with THA such as reduced dislocation rate, less complex surgery, shorter operation time, less blood loss, and lower initial costs [
12].
The prosthesis and the acetabulum in HA can be articulated using a unipolar or bipolar prosthesis. Parker et al. reported no difference in outcome between bipolar and unipolar prostheses in adults [
7]. However, recent studies have reported that bipolar HA (BHA) provides good outcomes for elderly patients with displaced FNFs. BHA after FNF has predictable and good medium- and long-term results, even when compared with internal fixation or unipolar HA [
8,
13,
14], and BHA displays a later onset of acetabular erosion compared with unipolar HA [
14,
15]. A review of data from national registries supports the continued use of BHA for FNF in the elderly [
16], implying that BHA should be the preferred treatment for elderly patients with displaced FNF. Elderly patients who receive BHA may also have a more favorable survival outcome compared with those who receive unipolar HA [
17]. Additionally, in elderly patients with FNFs who were fit and physiologically young, uncemented BHA seemed to achieve better functional outcomes [
18]. The clinical results from different groups cannot agree on whether to recommend BHA or THA [
19‐
21].
Therefore, we performed a meta-analysis of randomized controlled trials (RCTs) published up to May 2015. The aim of this study was to evaluate the clinical outcomes of BHA compared with THA. The results will improve understanding of the treatment options for displaced FNFs in healthy elderly patients.
Methods
Literature search
A protocol was developed prior to commencement of this meta-analysis following the Cochrane Back Review Group guidelines [
22]. We searched the following databases from inception to May 2015 for relevant RCTs without language restrictions: PubMed, the Cochrane Central Register of Controlled Trials (CENTRAL), Ovid MEDLINE and EMBASE, CINAHL, the China Biological Medicine Database (CBM), International Clinical Trials Registry Platform (ICTRP), Current Controlled Trials, and ClinicalTrials.gov. Other search methods included screening references listed in relevant systematic reviews and identified RCTs, searching the abstracts of relevant meetings, and personal communication with content experts in the field and authors of identified RCTs. Key words used for searching were arthroplasty, hip prosthesis implantation, total hip replacement, bipolar hemiarthroplasty, hip fractures, femoral neck fracture, and randomized controlled trial.
Study eligibility criteria
All RCTs comparing BHA with THA for the treatment of FNF were considered for this review. Trials including participants with displaced FNFs (Garden type III or IV) in the elderly (older than 65 years) were eligible. Patients with the following conditions were excluded from the study: (1) undisplaced FNF; (2) pathological fracture secondary to malignant disease; (3) osteoarthritis or rheumatoid arthritis of the hip; (4) severe cognitive dysfunction; or (5) non-first time experience with artificial joint replacement.
The first phase of the trial selection process involved screening of titles and abstracts, followed by a second phase of eligibility evaluation from the full-text format. Both phases were performed independently by two reviewers and checked by the principal reviewer. The observed percentage agreement between the reviewers for the assessment of inclusion was calculated using the κ test [
23,
24]. Disagreements were resolved by discussion.
Risk of bias assessment and evaluation of validity
The risk of bias (RoB) and methodological quality was assessed in duplicate using the Cochrane Collaboration recommendations, and evaluated independently by two reviewers [
22,
25]. The criteria included six items as follows: (1) adequate sequence generation; (2) allocation concealment; (3) blinding; (4) incomplete outcome data; (5) selective reporting; and (6) other bias.
The data were extracted from included reports independently by two reviewers, and any disagreements were resolved through discussion. The data extracted included: participant characteristics, number of participants, and loss to follow-up; study characteristics; intervention details; primary and secondary outcomes. The primary outcomes included the Harris Hip Score (HHS) and reoperation rate. The secondary outcomes included: mobility, dislocation, acetabular erosion, infection, general complications, 1-year mortality, EuroQol (EQ)index-5D score, pain rate, operation time and blood loss, and length of hospital stay.
Assessment of heterogeneity
Heterogeneity was informally tested visually by the eyeball test, and formally tested by the chi-squared test and the I2 statistic; however, the decision regarding heterogeneity was dependent on I2. Substantial heterogeneity was defined as ≥50 %, and the effect of the interventions was described if the results were too heterogeneous.
Measures of treatment effect
Attempts were made to statistically pool the data of homogeneous studies in order to obtain the primary and secondary outcomes. The results were expressed in terms of risk ratio (RR) and 95 % confidence intervals (95 % CI) for dichotomous outcomes, and in terms of weighted mean difference and 95 % CI for continuous outcomes. When the same continuous outcomes were measured by different scales, the standardized mean difference and 95 % CI were calculated. The individual and pooled statistics were calculated using the random effect model. Funnel plots were used to explore potential publication bias. The Cochrane Review Manager software (RevMan version 5.1, the Nordic Cochrane Centre, København, Denmark) software was used for data analysis.
Discussion
The latest meta-analysis comparing THA with HA found that THA was preferable for healthy elderly patients with displaced FNF, because of better functional outcomes and lower reoperation rates [
32]. However, in the last several years some results have suggested that BHA results in a better clinical outcome than unipolar HA [
14,
15,
17], and that BHA should be the preferred treatment for elderly patients with displaced FNF [
16]. There is still controversy over whether THA or BHA should be recommended in healthy elderly patients with displaced FNF.
Eight RCTs were included in the present analysis (total 1,014 patients; 523 with BHA versus 491 with THA). The different follow-up durations in the included studies meant that the data could not be directly pooled; thus the data were initially divided into four different subgroups (within 1 year, within 2 years, within 4 years, and more than 4 years). The results showed that the comparison of HHS was not different between BHA and THA in all subgroups, although HHS tended to be higher after THA. Both reoperation rate and acetabular erosion rates were higher after BHA after more than 4 years, while there was a higher risk of dislocation after THA. Some of the outcomes such as EQindex-5D, mobility and pain rate were in favor of THA, while operating time was in favor of BHA. No significant differences were found in other outcomes, including infection rate, general complication, 1-year mortality, blood loss, and length of postoperative hospital stay.
The present meta-analysis showed that HHS after THA was not significantly different from that after BHA in each subgroup; however, it tended to be higher after THA. It was reported in the last several meta-analyses that the HHS was higher after THA compared with HA [
32‐
35]. The present meta-analysis included a newly reported high quality RCT in which the HHS after BHA was compared with that after polycarbonate–urethane (PCU)-THA within 3 years [
31]. The PCU-THA comprised a PCU acetabular component coupled with a large-diameter metal femoral head for the treatment of displaced FNFs in elderly patients [
31]. There were no significant differences between THA and BHA; however, the HHS tended to be higher in the BHA group than in the THA group within 3 years, and the authors suggested that the patients in the PCU-THA group experienced significantly more pain than the HA patients [
31], which is one of the important factors affecting HHS [
36]. These results might have influenced the pooled results in the present meta-analysis; therefore, we performed a sensitivity analysis by omitting this report. The subsequent results showed that the HHS in the THA group was better than that in the BHA group within 2 years, while the HHS was non-significantly higher in the THA group after 2 years. We suggest that, even though THA might lead to better clinical outcomes, proper implants were of great importance for the patients. Moreover, heterogeneity was obvious between studies, which might also have influenced the validity of the pooled results.
In the present meta-analysis, there were no significant differences between BHA and THA regarding reoperation rate within 4 years; however, the reoperation rate was higher after BHA than that after THA after 4 years. This result is similar to previous results [
32,
33,
35,
37,
38]; however, it was different from the results of Kannan et al. [
16]. Kannan et al. analyzed national registry data and reported a lower reoperation rate after BHA than after THA after a follow-up of 9 years in the Australian and Italian registries [
16]. They concluded that the difference in reoperation rate after both procedures in Australia was not significant for patients under 75 years, but was significantly lower after BHA for those over 75 years and hence BHA may be better for this age group, considering the low functional demand and the morbidity of reoperation surgery in patients over 75 years of age [
16].
There was a higher acetabular erosion risk after BHA than after THA, which often results in persistent pain [
39‐
41] and reoperation after HA [
42,
43]; this was even higher in unipolar HA [
14,
15]. HA may lead to conversion from HA to THA, and Coates and Armour advocated that THA should be used in cases of acetabular erosion with HA [
44], while Kannan et al. suggested that using THA to prevent acetabular erosion is not supported by registry data [
16], although a longer follow-up may change this recommendation. However, the reoperation rate for acetabular erosion after HA was very low [
16,
45], and accurate and careful preoperative planning for HA to equalize limb lengths and restore the patient’s own femoral offset can reduce postoperative acetabular erosion and pain [
46]. Sen et al. introduced a new technique that was able to salvage a painful HA due to acetabular erosion [
40].
One of the main reasons for reoperation after THA is dislocation [
12]. In the present meta-analysis there was a higher dislocation risk after THA than after BHA, which was similar to previous reports [
32‐
35,
37,
38]. Factors contributing to dislocation may include the surgical approach and the size of the prosthetic head. A multivariable analysis of dislocation after primary THA for all diagnoses found that a posterolateral approach and a smaller prosthetic head were associated with a higher rate of dislocation [
47]. This was similar in BHA [
16], hence choosing optimal hardware and improving the surgical access route may help to reduce complications associated with this common procedure. Enocson et al. analyzed the possible reasons related to dislocation in THA, and recommended the anterolateral approach for THA in patients with FNF to minimize the risk of dislocation [
48]. Byström et al. studied the risk factors for prosthesis luxation leading to reoperation, and found that larger head size had been associated with fewer dislocations [
49]. Therefore, an accurate and careful preoperative plan that considers the most appropriate approach and prosthesis may help lower the dislocation rate.
Our review has some limitations. First, the search was restricted to RCTs published in peer-reviewed journals, excluding other sources of biomedical literature that could have possibly provided more relevant studies. In such a case, studies with positive or statistically significant results would be expected to be over-represented in our review, as such studies were more likely to be published, particularly in the English language. To counter this, we used funnel plots to investigate the potential influence of publication bias on our results. Second, the validity of our results is limited by the low quality of the included studies; double-blinding was unattainable for most of the trials, which may decrease the strength of our conclusions. Third, there is the potential for bias because of high heterogeneity (≥50 %) in some comparisons, which may have affected the pooled results. Studies brought together in a meta-analysis will inevitability differ, and any kind of variability among studies may be termed heterogeneity. The included studies had clinical heterogeneity caused by variability in the participants (age, gender, comorbidities, preoperative ambulatory status), interventions (instrumentation from different manufacturers, different surgeons) and outcomes (selective reporting, data deficiency), and methodological heterogeneity caused by variability in study design and RoB.
However, there were several improvements in the present meta-analysis compared with previous reviews. This review is the most current report on the topic and includes the latest published trials. We adopted more strict inclusion criteria, with quasi-RCTs and non-RCTs strictly excluded in order to guarantee the reliability of results. We pooled most data of comparable parameters using subgroup analysis with different follow-up durations in an attempt to reduce the bias. We conducted this first analysis comparing BHA with THA to provide alternative guidelines for the clinical treatment of healthy elderly patients with FNF.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
FW, HFZ and ZYZ conceived and designed the study. FW, HFZ, ZYZ, CBM and XMF screened the literature search results and identified articles for retrieval, reviewed full papers for inclusion, and extracted data for meta-analysis. FW, HFZ and ZYZ analyzed data and carried out comparisons. FW and HFZ wrote the manuscript. All authors read and approved the final manuscript.