Comparing total hip arthroplasty and hemiarthroplasty for the treatment of displaced femoral neck fracture in the active elderly over 75 years old: a systematic review and meta-analysis of randomised control trials

Background: Displaced femoral neck fractures (DFNF) are increasingly common in elderly patients. Repair of DFNF can be completed using two methods, hemiarthroplasty (HA) or total hip Arthroplasty (THA). However, there is much controversy regarding whether HA or THA is superior in active elderly patients over 75 years old. Methods: We conducted the literature search by searching PubMed, Embase, the Cochrane Library, ClinicalTrials.gov, and Web of Science from the inception dates to June 1, 2019. Randomised controlled trials (RCT) were included according to the inclusion and exclusion criteria. Included studies were analysed according to Cochrane review methods. Results: Nine studies met the inclusion criteria totalling 631 participants (301 THA and 330 HA. Four of the studies conducted were identical to a previous study but look at different follow up measures. The analysis of previously collected data revealed that the THA group had a lower erosion rate, higher total pain and function HHS, and higher EQ-5Dindex. The re-operation rate was significantly lower in the THA group after five years of follow up after adding the patients with painful symptoms to the group of patients revised. However, THA had longer operative time and a higher dislocation rate in the first three years compared to the HA procedure. Moreover, there was an insignificant difference in mortality rate, general complications, wound infection, duration of hospital stay, and VTE prevalence among the two procedures. Conclusions: THA may be a preferred management option for active elderly patients over 75 years old, after careful evaluation of patients’ general condition and tolerance of surgery. Strict management and follow up protocols should be followed to prevent dislocation within the first three years following a

another recent study found that the frequency of HA procedures increase as age increases. However, in recent years, the utilisation of THA to treat DFNF has increased in the elderly population [18]. A systematic review and meta-analysis recommended THA in patients younger than 80 years old or those with a life expectancy of more than four years [10]. The findings of this systematic review were similar to other previously performed studies [19][20][21].
Moreover, Two RCTs with a 9-year and 12-year follow up found that there were no statistical differences in the functional outcomes, revision rates, and complication rates between HA and THA treatments for active elderly patients over 70 years old [22,23]. There is currently an ongoing RCT comparing THA and HA outcomes for DFNF in patients over 80 years old, but it is yet to be finished [12].
The option of arthroplasty in DFNF patients over 75 years old becomes more and more critical as the elderly continue to face greater demands. The previous metaanalyses did not include strict age restrictions and some of these studies also included patients who were unable to walk independently or received a polycarbonate-urethane (PCU) THA that would lead to high heterogeneity, therefore making it complicated to apply these findings to patients with older age [9,10,19,[24][25][26]. The purpose of this study is to evaluate the clinical outcomes of THA and HA in the treatment of DFNF in the active elderly over 75 years old by using the latest evidence from previously performed RCTs.

Methods
This systematic review and meta-analysis was performed according to the Systematic Reviews and Meta-Analyses (PRISMA) statement [27,28]. This study was registered at the International Prospective Register of Systematic Reviews

Outcome measures
We extracted the following information from each study: lead author, participant characteristics, follow up duration, details of the interventions, the primary outcomes, and secondary outcomes. The primary outcomes measured were intraoperative details (blood loss and duration of operation), mortality, functional outcomes (Harris Hip Score), re-operation rates, dislocation rates, erosion rates, infection rates, deep venous thrombosis prevalence, and general complication rates.
The secondary outcomes were prosthesis loosening, peri-prosthetic fractures, Europol (EQ) index-5D score and the duration of hospital stay.

Statistical analysis
We used both the Q2 test and I² test to assess the statistical heterogeneity among studies. If the P value resulting from the Q2 test was 0.1 or the value of I² test was >50%, it was indicative of statistical heterogeneity. We calculated the RR (relative risks) for dichotomous outcomes, and weight mean difference or standardised mean difference for continuous outcomes with corresponding 95% confidence intervals (CI). The results were pooled by a fixed effects model. If data was indicative of statistical heterogeneity, a random effects model was applied instead. If two studies were based on the same population, only the latter one was included in the forest plot. We performed the sensitivity analysis by excluding research to evaluate the stability of the results. We used the Cochrane Review Manager 5.3 software for data analysis and P<0.05 was considered to be statistically significant.

Study Characteristics
We identified 2186 studies, of which nine studies met the inclusion criteria[8, 20,  Table 1). The range of the duration of follow up was from 24.0 months to 194.0 months. One of these studies also included a third arm (internal fixation), but the data from these arms were not taken into account[32].

Risk of bias assessment
All nine Randomized Controlled Trials were assessed and found to have used adequate randomisation procedures, including the sealed-envelope technique,  Table 1). The range of the duration of follow up was from 24.0 months to 194.0 months. One of these studies also included a third arm (internal fixation), but the data from these arms were not taken into account[32].  Fig.7). There was low heterogeneity (I 2 =6%) among studies.

Erosion
Three studies reported the erosion rates in both groups (189 with THA and 202 with HA) during the follow up periods (mean 45.3m) [20,22,23]. The erosion rates were 0.48% in the THA group and 13.7% in the HA group with a significant difference (RR 0.05, 95% CI 0.01-0.20, Fig.9) and no significant between-study heterogeneity (I 2 =0%). The re-operation rate was 3.5% in the THA group and 5.6% in the HA group, however, there was no significant difference (RR 0.45, 95%CI 0.14-1.39, Fig.10) between the two techniques. If we added the patients with painful symptoms to the group of patients revised, the re-operation rate was significantly lower in the THA group after five years of follow up than in the HA group (RR 0.26, 95%CI 0.09-0.82,

Dislocation rate
Five studies reported the dislocation rate in both groups (301 with THA and 300 with HA). We divided the data into three subgroups according to follow up duration within three years and more than three years [20,22,23,30,32]. There was a significant statistical difference between dislocation rates in the THA group 4.0%) and the HA group 0.61% (RR 4.40, 95% CI 1.35-14.37, Fig.12). There were no dislocations reported in the subgroups that had more than three

Function HSS
Two studies assessed the HHS function in both groups (171 with THA and 192 with HA)[20, 23]. We divided the data into two subgroups according to the follow up duration (within one year and more than one year). Two studies assessed the HHS function within one year I 2 =56%) and the data tended to favour the THA group (MD 1.37, 95% CI -1.54-4.27, Fig,14)[20, 23]. Two studies assessed the HHS function after more than one year of follow up (I 2 =0%), and the data favoured the THA group (MD 3.48, 95% CI -0.7-6.26, Fig,14)[20, 23].

Hospital stay
Two studies assessed the length of hospital stay in both groups (132 with THA and 160 with HA). The data trended to favour the HA group (MD 1.30, 95% CI -0.43-4.57, revealed that the THA group had lower erosion rates, increased pain HHS, total HHS, function HHS and EQ-5Dindex. There was no significant difference between two groups in the re-operation rate. If we added the patients with painful symptoms to the re-operation group, the re-operation rate would be significantly lower in the THA group after five years of follow up. The function HHS was better in the THA group after one year of follow up. However, the THA group had a longer average operative time. The dislocation rate was higher within three years in the THA group. The general complication rates, mortality rates from one to five years and duration hospital stays favoured the HA group but without significant difference. Moreover, there was no significant difference in postoperative infection, peri-prosthetic fractures or VTE prevalence between the two groups. Compared to other systematic reviews and meta-analyses[9, 10, 19, 24-26], our study limited patients' age to greater than 75 years old, which was more specific than the previous studies. We excluded two RCTs because they included patients with compromised mental states[38, 39], and two RCTs because they used a THA procedure utilising a polycarbonate-urethane (PCU) acetabulum. The PCU components utilised in these studies were pliable, and therefore may have Re-operation criteria included different procedure, such as debridement, internal fixation, and revision. Some studies suggested that re-operation rates were higher in the HA groups compared to the THA groups [9, 10, 19, 22, 25, 26]. Avery et al.
[22] performed an RCT with nine years follow up and found that 75% of the revisions of HA were due to acetabular wear and erosion. However, studies contained in the Australian and Italian national registries identified HA to have lower re-operation rates than THA in patients exclusively over 75 years old[17]. Our trial found no significant difference in re-operation rate between the two groups in patients over 75 years old. The failure to find a difference may be the result of the fact that patients included in this study were older and had lower levels of baseline activity as well as lower tolerance to re-operation. After adding in the patients with painful symptoms, the re-operation rate was significantly lower in the THA group after five years follow up, which indicated that patients in THA group had a better quality of life five years post operation.
Many studies [9, 10, 22-26] found that dislocation rates were higher among the THA group, which correlated to the surgical approach and the femoral head size.
However, some studies [9, 10, 22, 23] suggested that there might be no significant differences in dislocation rates several years after the operation. Avery et al. Our trial found that the THA group had better HHS outcomes. These results were consistent with previously completed meta-analyses. However, these analyses did not have similar inclusion criteria that restricted the studies to patients over 75 years old and had different inclusion criteria [19,24] The heterogeneity among our studies might be due to other factors such as the differences in surgical technique, recovery programs and so on. In similar metaanalyses[9, 10], heterogeneity among studies in HHS was also high. Our trial found that THA had better pain HHS within one year follow up. However, the functional HHS was higher in the THA group only after one year. It indicated that patients could have the advantage of less pain in THA soon after the operation, but that it took a longer time to achieve the improvement in walking and living ability. The EQ-    Forest plot of comparison of re-operation, including patients with painful symptoms but witho