Risk Factors for Re-Revision After Aseptic Revision Total Knee Arthroplasty: Knees Revised for Instability Re-Fail Early in a Single-Center Cohort With a Mean 14-Year Follow-Up.

BACKGROUND: Aseptic revision total knee arthroplasty (TKA) is increasingly common, yet its long-term (≥ 10-year) durability and predictors of re-revision are less well characterized than for septic revision. We defined survivorship, risk factors, and failure modes.

METHODS: We retrospectively reviewed 168 consecutive aseptic revision TKAs at a single institution (2007 to 2016). The mean potential follow-up was 14.1 years (range, 9.6 to 19.1). Kaplan-Meier, competing-risk, and Cox analyses were used. Knee Society (KSS) and Oxford Knee (OKS) scores were compared preoperatively and at the latest follow-up. There were 28 patients (16.7%) who underwent re-revision, and 22 (13.1%) died.

RESULTS: Survivorship free of any-cause re-revision was 84.3% at 10 and 82.1% at 15 years (competing-risk cumulative incidence, 15.7 and 17.7%). There were three independent risk factors: prior revision (a knee revised before the index procedure) (hazard ratio [HR] 3.69; P = 0.001), severe bone loss (Anderson Orthopaedic Research Institute [AORI] Type 2B/3; HR 2.99; P = 0.007), and revision for instability (HR 2.70; P = 0.018). Severe bone loss and instability remained independent after adjustment and stratification for prior-revision status; however, in the single-revision subgroup these estimates were imprecise and unconfirmed, and the 43 multiply revised knees accounted for half of all re-revisions. Instability risk was front-loaded: seven of nine such re-revisions occurred within 3.2 years. Only two of these nine re-revisions were for recurrent instability. Re-revision was most often for aseptic loosening (39.3%) and instability (25.0%). Functional scores improved (KSS 49.3 to 79.9; OKS 23.6 to 32.6; both P < 0.001).

CONCLUSIONS: Aseptic revision TKA is durable, with 15-year survivorship exceeding 80%. Risk is highest with prior revision; knees revised for instability fail early, warranting the least constraint necessary and heightened early surveillance. In this mixed cohort, severe bone loss and instability predicted re-revision, but were not confirmed in first-time revisions.