Retrograde Intramedullary Nailing Versus Locked Plating for Periprosthetic Distal Femur Fractures: A Propensity-matched Cohort from TriNetX.

Introduction: Periprosthetic distal femur fractures (PDFF) after total knee arthroplasty (TKA) are increasing with the growth of arthroplasty volume and longevity of implants; reported incidences for primary TKA range from ~0.3% to 2.5%, underscoring a clinically meaningful burden. Comparative data suggest both retrograde intramedullary nailing (RIMN) and locked plating (LP) achieve acceptable union with broadly similar complication profiles, but uncertainty persists regarding construct-specific trade-offs. Materials and Methods: We conducted a retrospective propensity-matched cohort study within the TriNetX U.S. Collaborative Network, a federated platform of de-identified electronic health records from dozens of U.S. health systems. Adult patients (≥50 years) with PDFF around a stable knee prosthesis were identified using International Classification of Disease-10 (ICD-10) diagnosis and current procedural terminology/ICD-10-procedure coding system procedure codes. The index event was the first qualifying fixation - RIMN or LP - with outcomes observed for 365 days post-index. Propensity score matching (1:1, greedy nearest neighbor without replacement) balanced age, sex, and comorbidities (e.g., diabetes, hypertension, cardiovascular, renal, pulmonary, hepatic disease, osteoporosis, obesity, tobacco use, depression, rheumatoid arthritis). Outcomes included revision/re-operation, hardware removal, mechanical implant complications, deep infection, non-union/malunion, re-fracture coding events, and venous thromboembolism (deep vein thrombosis/pulmonary embolism [DVT/PE]). We report risk difference (RD), risk ratio (RR, 95% confidence interval [CI]; P-matching RD P-values), and time-to-event hazard ratios (HR, 95% CI) from Kaplan-Meier/log-rank analyses. TriNetX governance ensures de-identification and standardized analytics. Results: After matching, 1,152 RIMN patients were compared with 1,152 LP patients with similar baseline characteristics and follow-up (median ~11-12 months). Revision/re-operation within 1 year was uncommon and comparable (1.0% RIMN vs. 1.5% LP; RD-0.4%, P = 0.35; RR 0.71, 95% CI 0.34-1.47; HR 0.77, log-rank P = 0.49). Hardware removal occurred more often after RIMN (6.0% vs. 4.2%; RD + 1.8%, 95% CI 0.0-3.6%; P = 0.046; RR 1.44, 95% CI 1.00-2.06; HR 1.60, log-rank P = 0.012). Mechanical implant complications were less frequent with RIMN (3.6% vs. 5.6%; RD -2.0%, 95% CI -3.7--0.3%; P = 0.023; RR 0.65, 95% CI 0.44-0.94; HR 0.68, log-rank P = 0.047). Deep infection rates were similar (6.3% vs. 7.4%; RD -1.0%, P = 0.32; RR 0.86, 95% CI 0.64-1.16; HR 0.89, P = 0.48), as were non-union/malunion (5.0% vs. 5.6%; RD -0.5%, P = 0.58; RR 0.91, 95% CI 0.64-1.28; HR 0.94, P = 0.75). DVT (4.6% vs. 6.2%; RD -1.6%, P = 0.097; RR 0.75, 95% CI 0.53-1.06; HR 0.77, P = 0.14) and PE (3.7% vs. 4.4%; RD -0.7%, P = 0.40; RR 0.84, 95% CI 0.57-1.26; HR 0.86, P = 0.46) did not differ significantly. These findings align with prior comparative literature showing broadly similar healing and complication rates between constructs, with nuanced differences by endpoint. Conclusion: In a large real-world, propensity-matched cohort, RIMN and LP produced comparable 1-year union, infection, thromboembolism, and revision rates for PDFF after TKA. RIMN was associated with fewer mechanical implant failures but a higher frequency of elective hardware removal, reflecting a clinically relevant trade-off. Construct selection should be individualized to prosthesis design, fracture geometry, distal bone stock, and patient priorities. Findings complement prior meta-analyses and multicenter series and support shared decision-making rather than a one-size-fits-all approach.