The long-standing radiograph (LSR) is the reference tool for assessing knee alignment after total knee arthroplasty (TKA). However, its value is debatable, as many factors can influence measurement accuracy. Computer-assisted surgery (CAS) provides accurate measurements. Few studies have compared LSR and computer-assisted measurements of knee alignment. The objective of this study was to compare hip-knee-ankle (HKA) angle values obtained before and after TKA on LSRs to those obtained during CAS.

The HKA angle values measured on LSRs before and after surgery are identical to those measured during CAS.

The HKA angles of 126 knees were measured on bipedal full-weight-bearing LSRs obtained before and 3 months after TKA. The results were compared to the values obtained during CAS.

Before surgery, the standard deviation was 2.672, with limits of agreement of {−5.391; + 5.082}. The intra-class coefficients were good for the overall measurements (0.9), good for detecting>10° varus (0.89), fair for<10° varus and valgus (0.66 and 0.71, respectively), poor for>10° valgus (0.43) and poor for normal alignment (0). Post-operatively, the standard deviation was 3.113, with limits of agreement of {−6.426; +5.776}. The intra-class coefficient was poor for the overall measurements (0.20), negative for normal alignment (−0.05) and<10° valgus (−0.05), and positive for<10° varus (0.017) and for>10° varus and valgus (0.33).

Before TKA, the LSR overestimates the deformity compared to CAS. After surgery, the two methods are not comparable. These findings underline the relevance of routinely obtaining LSRs and for using LSR results to estimate costs for healthcare insurance reimbursement purposes.

IV, retrospective observational cohort study.