Component-specific prognostic signals within a multidimensional classification for periprosthetic joint infection and the impact of repeated surgical episodes: a retrospective cohort study
Yohei Kumabe, Kenichi Sawauchi, Yutaka Matsumiya, Tomoaki Fukui, Shinya Hayashi, Tomoyuki Matsumoto, Ryosuke Kuroda, Keisuke OeBackground
The PJI-TNM classification provides a multidimensional description of periprosthetic joint infection, but the prognostic contribution of its individual components remains uncertain.
Objective
To evaluate component-specific associations with infection-related reoperation and determine how repeated surgical episodes within patients influence risk estimates.
Methods
We retrospectively analysed 116 hip or knee periprosthetic joint infection surgical episodes in 84 patients treated from April 2016 to February 2026. The outcome was the first unplanned infection-related reoperation after each index operation. Kaplan-Meier and Cox analyses were performed, using patient-clustered robust variance estimates to account for repeated episodes. A patient-level sensitivity analysis used the first observed episode per patient.
Results
Infection-related reoperation occurred after 42 episodes. Reoperation-free survival was 70.8% at 180 days, 63.1% at 1 year, and 55.1% at 2 years. In the primary clustered model, T2 showed an elevated but imprecise hazard estimate (hazard ratio 2.61, 95% confidence interval 0.89-7.72), whereas T1 did not. The N component showed no clear association with outcome. M2 produced a high estimate but comprised only three episodes from two patients. In the patient-level sensitivity analysis, the T2 hazard ratio was 4.78 (95% confidence interval 1.05-21.81).
Conclusion
Prognostic information within PJI-TNM was unevenly distributed across components. Severe soft-tissue compromise represented by T2 showed the most consistent adverse signal, while repeated-episode analyses materially affected precision. Multicentre validation is required.