Temporal distribution of recorded indications for first revision total knee arthroplasty: a multicenter Chilean cohort of 150 patients

Abstract

INTRODUCTION: /Objectives: Recorded indications for first revision total knee arthroplasty (TKA) may vary with time after the primary procedure. This study compared their distribution among early and late first revisions in a multicentre Chilean cohort. METHODS: This retrospective non-consecutive cohort included 150 unique patients who underwent first revision TKA at five Chilean centres between 2018 and 2026. Revisions were classified as early (<24 months) or late (≥24 months). Indication distributions were compared using categorical tests, including a fixed-margin Monte Carlo procedure for sparse cells. Logistic regression adjusted for age and sex. Sensitivity analyses incorporated revision year and centre and modelled time continuously, including assessment of non-linearity. RESULTS: Mean age was 70.1±9.5 years, and 97 patients (64.7%) were women. Fifty-three revisions (35.3%) were early and 97 (64.7%) were late. The indication distribution differed by timing (Monte Carlo p=0.0003; Cramér's V=0.41). Infection was the most frequent early indication (18/53, 34.0%), whereas aseptic loosening accounted for more than half of late revisions (52/97, 53.6%). After adjustment for age and sex, late revision was associated with higher odds that aseptic loosening was the recorded indication (adjusted odds ratio [OR], 6.70; 95% confidence interval [CI], 2.80-16.00; p<0.001). Early revision was associated with higher odds that infection was the recorded indication (adjusted OR, 3.21; 95% CI, 1.40-7.34; p=0.006). Estimates were similar after accounting for revision year and centre and when time was analysed continuously. CONCLUSION: Among the included first revisions, infection was the most frequent recorded indication before 24 months, whereas aseptic loosening was the most frequent indication thereafter. These within-cohort associations remain subject to selection bias, diagnostic misclassification and residual confounding, and do not estimate cause-specific revision risk after primary TKA. LEVEL OF EVIDENCE: III.