Reliability of the Tardieu Scale for assessing spasticity in children with cerebral palsy.

OBJECTIVE To measure the Tardieu Scale's reliability in children with cerebral palsy (CP) when used by raters with and without experience in using the scale, before and after training. DESIGN Single-center, intrarater and interrater reliability study. SETTING Institutional ambulatory care. PARTICIPANTS Referred children with CP in the pretraining phase (n=5), during training (n=3), and in the posttraining phase (n=15). INTERVENTIONS The Tardieu Scale involves performing passive muscle stretch at 2 velocities, slow and fast. The rater derives 2 parameters; the Spasticity Angle X is the difference between the angles of arrest at slow speed and of catch-and-release or clonus at fast speed; the Spasticity Grade Y is an ordinal variable that grades the intensity (gain) of the muscle reaction to fast stretch. In phase 1, experienced raters without formalized training in the scale graded elbow, knee, and ankle plantar flexors bilaterally, without and with a goniometer. In phase 2, after training, the experienced and nonexperienced raters graded the same muscles unilaterally. MAIN OUTCOME MEASURES Intrarater and interrater reliability of the Tardieu Scale. RESULTS After training, nonexperienced raters had mean +/- SD intrarater and interrater agreement rates across all joints and parameters of 80%+/-14% and 74%+/-16%, respectively. For experienced raters, intrarater and interrater agreement rates before training were 77%+/-13% and 66%+/-15%, respectively, versus 90%+/-8% and 81%+/-13%, respectively, after training (P<.001 for both). Specific angle measurements at the knee were less reliable for the angles of catch measured at fast speed. Across all joints, agreement rates were similar using visual or goniometric measurements. CONCLUSIONS Both parameters of the Tardieu Scale have excellent intrarater and interrater reliability when assessed at the elbow and ankle joints of children with CP, with no difference noted between visual and goniometric measurements. Angle measurements were less reliable at the knee joints. Training was associated with a highly significant improvement in reliability.