Predictive Model for AOS Response in Total Ankle Replacement: Doctor, Will My Symptoms Get Better After the Ankle Replacement?

Carlos Albarrán, Marianne Koolen, Tudor Trache, Sultan Alharbi, Andrea Veljkovic, Murray Penner, Alastair Younger, Kevin Wing

Introduction: Ankle osteoarthritis (AO) is a condition which generates a level of disability comparable to that of heart failure and chronic kidney failure. To date, no predictive model has been described to determine which patients will experience a more pronounced improvement in disability following total ankle replacement.

Method: A prospective registry (2003 – 2023) was screened for primary TARs; revisions and patients who declined participation were excluded. “Good response” was defined as ≥ 28-point improvement in the Ankle Osteoarthritis Scale (AOS) at 1–2 years. Stepwise multivariable logistic regression assessed demographics (age, smoking, diabetes, inflammatory disease, BMI), OA aetiology, implant type (Infinity, InBone, InVision, Agility, Hintegra, Mobility, Zimmer, STAR) and baseline AOS. Models were compared with Akaike’s Information Criterion; over-fitting was checked with 1000-bootstrap resampling and calibration with the Hosmer–Lemeshow test.

Results: Six-hundred-thirty-one primary TARs (mean age 64.9 ± 9.7 yr) met inclusion criteria; the most common implants were Hintegra (159), Infinity (151), Zimmer (127) and Agility (78). The final, parsimonious model retained two independent predictors: higher baseline AOS (OR 1.07 per point, 95 % CI 1.05–1.08; p < 0.01) and ankle instability aetiology (OR 4.69, 95 % CI 1.50–14.71; p < 0.01). Discrimination was acceptable (AUC 0.76); bootstrap validation showed no loss of performance, and the Hosmer–Lemeshow test confirmed good fit.

Conclusion: Greater pre-operative disability and an instability-related aetiology markedly increase the likelihood of achieving a clinically important improvement after TAR. This internally validated, two-variable model is easy to implement in clinic and can refine patient selection, set realistic expectations, and support shared decision-making. External validation in other centres is warranted.