Comparative diagnostic accuracy of knee adduction moments in knee osteoarthritis: a case for not normalizing to body size

Previous authors have questioned the practice of normalizing the external knee adduction moment during gait to body size when investigating dynamic joint loading in knee osteoarthritis (OA). The purpose of this study was to compare the abilities of non-normalized and normalized external knee adduction moments during gait in discriminating between patients with least and greatest severity of radiographic medial compartment knee OA. Subjects with mild (n=118) and severe (n=115) medial compartment knee OA underwent three-dimensional gait analysis. The peak external knee adduction moment was calculated and kept in its original units (Nm), normalized to body mass (Nm/kg) and normalized to body weight and height (%BW × Ht). Receiver Operating Characteristic (ROC) curve analysis indicated that non-normalized values better discriminated between patients with mild and severe knee OA. The area under the ROC curve for non-normalized peak knee adduction moments (0.63) was significantly (p<0.05) greater than when normalized to body mass (0.58), or to body weight times height (0.57). Post-hoc analysis of covariance indicated the mean difference in peak knee adduction moment between OA severity groups (7.23 Nm, p=0.003) was reduced by approximately 50% (3.60 Nm, p=0.09) when adjusted for mass. These findings are consistent with the suggestion that non-normalized values are more sensitive to radiographic disease progression. We suggest including knee adduction moment values that are not normalized to body size when investigating knee OA.