Twenty women involved in an existing study of healthy aging  were recruited through a women's health clinic and advertising in the local media. The study was approved by the ethics committees of Alfred Hospital (Prahran, Victoria, Australia), Caulfield Hospital (Caulfield, Victoria, Australia), and La Trobe University (Melbourne, Victoria, Australia). All participants gave informed consent.
Exclusion criteria were a history of knee OA, radiological OA or any history of symptoms requiring medical treatment, any knee pain for more than 1 day in the month prior to testing, previous or planned knee joint replacement, inflammatory arthritis, malignancy, fracture in the last 10 years, contraindication to MRI (for example, pacemaker, cerebral aneurysm clip, cochlear implant, presence of shrapnel in strategic locations, metal in the eye, and claustrophobia), inability to walk 50 feet without the use of assistive devices, hemiparesis, and any other musculoskeletal, cardiovascular, or neurological condition that would impair normal gait as previously described .
Weight was measured to the nearest 0.1 kg (shoes, socks, and bulky clothing removed) using a single pair of electronic scales. Height was measured to the nearest 0.1 cm (shoes and socks removed) using a stadiometer. Body mass index (BMI) (weight in kilograms divided by height squared in metres squared) was calculated. A history of knee trauma and knee surgery was obtained.
Magnetic resonance imaging
MRI was performed on the dominant knee (that is, the leg from which a subject stepped off from when initiating walking) as previously described . The following sequence and parameters were used: a T1-weighted fat-suppressed three-dimensional gradient recall acquisition in the steady state; flip angle 55°; repetition time 58 ms; echo time 12 ms; field of view 16 cm; 60 partitions; 512 (frequency direction, superior-inferior) × 512 (phase-encoding direction, anterior-posterior) matrix; one acquisition, time 11 minutes 56 seconds. Sagittal images were obtained at a partition thickness of 1.5 mm and an in-plane resolution of 0.31 × 0.31 mm (512 × 512 pixels).
Meniscal tears were assessed in the sagittal view and confirmed in coronal and axial views by experienced radiologists (André Pelletier and Josée Thériault) as previously described [3, 7, 16]. The presence of a tear was based on the presence of a signal, which was line-shaped, brighter than the dark meniscus, and reached the surface of the meniscus at both ends within six defined regions (anterior horn, body and posterior horn at both medial and lateral tibiofemoral compartments). A semi-quantitative lesion assessment of meniscal tears was also performed. Our scoring system for meniscal damage referred to the accepted MRI nomenclature for meniscal anatomy, which is in accordance with arthroscopic literature . The proportion of the menisci affected by tears was scored separately using the following semi-quantitative scale : 0 = no damage; 1 = one out of three meniscal areas involved (anterior, middle, posterior horns); 2 = two out of three areas involved; 3 = all three areas involved. The intra- and inter-reader correlation coefficients ranged from 0.86 to 0.96 for the meniscal tears .
Gait analyses were conducted in the gait laboratory in the Musculoskeletal Research Centre, La Trobe University. A six-camera Vicon motion analysis system (Oxford Metrics Ltd., Oxford, UK) was used to capture three-dimensional kinematic data during four walking trials on the dominant leg at the subjects' self-selected speed to capture normal gait patterns. Ground reaction forces were measured by a Kistler 9281 force-platform (Kistler Instruments, Winterthur, Switzerland). Inverse dynamic analyses were performed using 'PlugInGait' (Oxford Metrics Ltd.), which is based on a previously proposed model , to obtain joint moments calculated about an orthogonal axis system located in the distal segment of a joint as previously described [8, 12]. Inter-ASIS (anterior superior iliac spine) distance was measured using a calliper, allowing the medial-lateral and proximal-distal coordinates of the hip joint centre to be determined by the method previously described . The ASIS to greater-trochanter measurement provided the anterior-posterior coordinate of the hip joint. A knee alignment device (KAD) was used to calculate knee joint axes. The coronal plane of the thigh was defined as the plane containing the hip joint centre, knee marker, and lateral KAD marker. The coronal plane of the shank contained the knee joint centre and lateral malleolus marker. The angle formed by the knee and ankle joint axes measured tibial torsion.
Foot rotation was measured about an axis perpendicular to the foot vector and the ankle flexion axis. It is defined as the angle between the foot vector and the sagittal axis of the shank, projected into the foot transverse plane. This differs from the toe-out angle, which is measured from the long axis of the foot, relative to the line of progression of the body. The foot is defined by the single vector joining the ankle joint centre to the second toe. The relative alignment of this vector and the long axis of the foot is calculated from a static trial using an additional calibration marker from the heel. The foot vector is established by making two rotations about the orthogonal axis. This measure is equal to the angle between the line joining the heel marker and the toe marker, projected in the plane perpendicular to the ankle flexion axis (sagittal). The second rotation is about a foot rotation axis that is perpendicular to the foot vector and the ankle flexion axis. This measure is equal to the angle projected in the plane perpendicular to the foot rotation axis (transverse). The angle is measured between the line joining the heel and toe markers and the line joining the ankle centre and toe marker as previously described [12, 19] and according to the protocol stipulated by the Vicon technology in the gait laboratory . Positive values correspond with internal rotation (Vicon Clinical Manager's User Manual ). Subjects were instructed to walk barefoot at their normal pace over level ground, to capture their natural gait patterns.
Gait data were initially examined for normality and linearity. The peak external knee adduction moment and degree of foot rotation occurring when the adductor moment peaked during early and late stance were averaged over four walking trials. Peak external knee adduction moments were normalised to percentage body weight multiplied by height. Linear regression analyses were used to determine the relationship between meniscal tear presence (yes/no) and severity (grade) (independent variables) and peak external knee adduction moments and foot rotation during early and late stance (outcome variables). Age and gender are associated with meniscal tears and also with gait. Our study used restriction to reduce any confounding associated with gender and included age within our multivariate regression analysis. Moreover, since six participants reported a past knee injury, a history of knee injury (yes/no) was also included in the regression analyses. Furthermore, to see whether rotation effects on the menisci were independent of the adductor moment, this was included within the model. Results in which there were P values of less than 0.05 (two-tailed) were considered to be statistically significant. All analyses were performed using SPSS (version 11.0.1; SPSS Inc., Cary, NC, USA).