Background/Purpose: Varus thrust, bowing-out of the knee during gait, i.e., appearance (or worsening) of varus during stance improving in late stance or swing, was associated with medial tibiofemoral OA progression in a single-site study. To determine replicability in a multi-center study and if thrust is associated with incident OA, we undertook an OAI ancillary study. We hypothesized that: 1) in knees without OA [KL<2 at enrollment (0m) and 12m visits], thrust at 12m (our study baseline) is associated with subsequent a) incident OA (KL≥2) and b) medial joint space narrowing (JSN); and 2) in knees with OA (KL≥2 at 0m or 12m), thrust at 12m is associated with subsequent medial JSN. We further considered the hypothesized associations adjusted for static alignment, anticipating some but not complete attenuation.

Methods: Gait was observed for thrust at 4 sites by 2-3 trained examiners/site at 12m. In knees KL<2 at 0m and 12m, incident OA was analyzed as subsequent incident KL≥2, medial JSN by whole grade and by partial grade (all dichotomous), and annualized JSN in mm (continuous) after 12m. In knees KL≥2 at 0m or 12m, progression was analyzed as subsequent medial JSN by whole grade and by partial grade, and annualized JSN. Outcomes were assessed up to 96m. Alignment was measured in 3 other sub-studies: mechanical axis (hip-knee-ankle angle, HKA) using 2 full-limb x-ray measurement approaches (HKA-DC and HKA-JD), and anatomical axis (femorotibial angle) on knee x-ray (FTA). All analyses were knee-level; we used multivariable logistic and linear regression methods with GEE to account for between-limb correlation.  

Results: The incident OA sample included 4187 knees from 2610 persons [age 60.5 (SD 9.2), BMI 27.8 (4.5), 1455 (56%) women]; the progression sample included 3421 knees/2284 persons [age 62.6 (9.0), BMI 29.6 (4.8), 1339 (59%) women]. Thrust was not associated with incident KL≥2 or medial JSN in knees without OA (Table 1, upper half), but was associated with all progression outcomes (Table 1, lower half). The thrust/progression association was attenuated (Table 2) but an independent association persisted in partial grade and annualized JSN models including HKA-JD and FTA.  After adjusting for HKA-DC, thrust was no longer associated with progression.

Conclusion: Over up to 7 years of follow-up observation, varus thrust was associated with medial knee OA progression but not incident OA. Gait observation for thrust may offer a simpler (vs. radiographic methods for alignment) approach to predicting knee OA progression, which is translatable to larger scale studies with multiple examiners.