Optimal surgical treatment for periprosthetic distal femoral fractures after total knee arthroplasty: a Bayesian-based network analysis

BACKGROUND: The surgical methods for periprosthetic distal femoral fractures (PDFFs) after total knee arthroplasty included locking compression plate (LCP), retrograde intramedullary nailing (RIMN), and distal femoral replacement (DFR). However, the optimal treatment remains controversial. We performed a network meta-analysis (NMA) to provide the optimal surgical method for PDFFs. MATERIALS AND METHODS: Electronic databases, including Embase, Web of Science, Cochrane Library, and PubMed, were searched for studies that compared LCP, RIMN, and DFR for PDFFs. The quality of the included studies was assessed according to the Newcastle–Ottawa scale. Pairwise meta-analysis was performed by Review Manager version 5.4. The NMA was conducted in Aggregate Data Drug Information System software version 1.16.5. We calculated odds ratios (ORs) and 95% confidence intervals (CIs) for postoperative complications and reoperations. RESULTS: A total of 19 studies and 1198 patients were included, of whom 733 for LCP, 282 for RIMN, and 183 for DFR. Pairwise meta-analysis comparing LCP to RIMN and LCP to DFR showed no significant difference in complications and reoperations except that RIMN had a higher risk of malunion comparing to LCP (OR 3.05; 95% CI 1.46–6.34; P = 0.003). No statistically significant effects were found in the NMA of overall complications, infection, and reoperation. However, results of rank probabilities showed that DFR ranked best in overall complications and reoperation, RIMN ranked best in infection but worst in reoperation, and LCP ranked worst in infection and middle in reoperation. DISCUSSION: We found similar complication rate and reoperation rate between LCP, RIMN, and DFR. The results of rank probabilities favored DFR, and further studies with high-level evidence are expected to verify the optimal surgical method for PDFFs. LEVEL OF EVIDENCE: Level II; network meta-analysis.