CPAK classification shows limited side‐to‐side agreement and limited reliability as a standalone surrogate for constitutional alignment in rheumatoid arthritis knees

Abstract Purpose The Coronal Plane Alignment of the Knee classification estimates constitutional coronal alignment. In rheumatoid arthritis knees, disease‐related structural changes may affect the bony parameters used for this classification. This study described its distribution and side‐to‐side agreement in rheumatoid arthritis knees and explored radiographic differences associated with discordance. Methods This single‐centre retrospective study included 57 patients and 93 knees treated with unilateral or bilateral total knee arthroplasty for rheumatoid arthritis between January 2014 and December 2024. Classification type was determined from preoperative standing full‐length anteroposterior radiographs. Side‐to‐side agreement was assessed in 36 bilaterally evaluable patients using complete agreement and unweighted Cohen's κ . Side‐to‐side differences in the medial proximal tibial angle (MPTA), lateral distal femoral angle (LDFA), joint line convergence angle (JLCA) and Larsen score were compared between agreement and discordance groups. Results In 93 knees, Type I was most frequent at 36.6% (34 of 93), followed by Type II at 25.8% (24 of 93) and Type III at 24.7% (23 of 93). Types VII–IX were not observed. Complete side‐to‐side agreement was 36.1% (13 of 36), and unweighted Cohen's κ was 0.152 (95% confidence interval, −0.050 to 0.356; p  = 0.093). Side‐to‐side differences in the MPTA, LDFA and JLCA did not differ significantly between groups. A Larsen score difference of at least 1 was more frequent in the discordance group than in the agreement group (78.3% [18 of 23] vs. 38.5% [5 of 13]; p  = 0.030), with an odds ratio of 5.760 (95% confidence interval, 1.290–25.640). Conclusions The classification showed limited side‐to‐side agreement in rheumatoid arthritis knees, and discordance was associated with asymmetric radiographic structural damage, indicating limited reliability as a standalone surrogate for constitutional alignment or contralateral knee morphology. Level of Evidence Level IV.