Total hip arthroplasty compared to bipolar and unipolar hemiarthroplasty for displaced hip fractures in the elderly: a Bayesian network meta-analysis

PURPOSE: Displaced femoral neck fractures (FNF) usually require surgical treatment with either a total hip arthroplasty (THA), unipolar hemiarthroplasty (U-HHA), or bipolar hemiarthroplasty (B-HHA). However, there is still controversy regarding the optimal implant. This network meta-analysis compared the outcomes and complication rates of THA versus B-HHA and versus U-HHA in elderly patients with FNF. MATERIAL AND METHODS: This study was conducted according to the PRISMA extension statement for reporting of systematic reviews, and incorporated network meta-analyses of health care interventions. The literature search was performed in September 2020. All randomized clinical trials comparing two or more of the index surgical interventions for displaced FNF in the elderly were eligible for inclusion. For the Bayesian network meta-analysis, the standardized mean difference (SMD) and Log Odd Ratio (LOR) were used. RESULTS: Data from 24 RCTs (2808 procedures) were analysed. The mean follow-up was 33.8 months. The THA group had the longest surgical time (SMD 85.74) and the greatest Harris Hip Score (SMD − 17.31). THA scored similarly in terms of mortality (LOR 3.89), but had lower rates of revision surgeries (LOR 2.24), higher rates of dislocations (LOR 2.60), and lower rates of acetabular erosion (LOR − 0.02). Cementless implants required a shorter surgical duration (− 18.05 min; P = 0.03). Mortality was positively associated with acetabular erosion (P = 0.006), female gender (P = 0.007), revision (P < 0.0001). CONCLUSION: THA led to the highest Harris Hip scores and lowest rate of revision surgery compared to B-HHA and U-HHA. However, B-HHA had the lowest dislocation rate when compared with U-HHA and THA. No significant differences in functional outcomes and complication rates were found between cemented and uncemented implants; however, a tendency for lower mortality, revision and dislocation rates in cemented implants was evidenced. LEVEL OF EVIDENCE: I, Bayesian network meta-analysis of RCTs.