Estimation of mechanical axis from tibiofemoral alignment measurements on short radiographs for total knee arthroplasty

Abstract

Background: The purpose of this study was to develop predictive models to estimate post-operative mechanical axis as measured on long radiographs using anatomic and component alignment measured on short radiographs. Methods: We retrospectively studied 653 participants (788 knees) who elected to undergo total knee arthroplasty (TKA) between September 2005 and December 2010. Multiple linear regression was conducted to predict post-operative mechanical axis (POSTMA) from pre-operative tibiofemoral angle (PRETFA), post-operative tibiofemoral angle (POSTTFA), and component alignment angles. Multinomial logistic regression was used to predict the risk of lower extremity alignment deformity from POSTTFA and component alignment angles. Results: The predictive accuracy of the linear regression models to estimate POSTMA from PRETFA and POSTTFA was 4% and 22%, respectively. Nearly 100% of the knees with actual neutral POSTMA of 0º ± 3º were accurately predicted to have a POSTMA of 0º ± 3º. There were zero knees with actual varus or valgus alignment over 3º that were accurately predicted to have a POSTMA over 3º of varus or valgus. Logistic regression analyses revealed the risk of having POSTMA deformity was significantly effected (p < 0.0001) by POSTTFA. The model indicated that for every 1º increase in valgus POSTTFA, the odds of predicting any degree of varus POSTMA decreased by 0.29 times, and the odds of predicting any degree of valgus POSTMA increased by 1.36 times. The model indicated that for every 1º increase in valgus POSTTFA, the odds of predicting a varus POSTMA > 3º decreased by 0.41 times and the odds of predicting a valgus POSTMA > 3º increased by 1.63 times. The percentages of knees correctly predicted into deformity category was high for neutral POSTMA but not for predicting the risk of varus or valgus deformity. Conclusions: The linear and logistic regression models were not adequate to estimate POSTMA using either PRETFA or POSTTFA. We do not recommend the use of short radiographs as a substitute for long radiographs when evaluating lower extremity alignment to determine component position before TKA and for assessment of postoperative alignment outcomes following TKA.