Development of a brief ataxia rating scale (BARS) based on a modified form of the ICARS

To develop a brief ataxia rating scale (BARS) for use by movement disorder specialists and general neurologists. Current ataxia rating scales are cumbersome and not designed for clinical practice. We first modified the International Cooperative Ataxia Rating Scale (ICARS) by adding seven ataxia tests (modified ICARS, or MICARS), and observed only minimally increased scores. We then used the statistics package R to find a five-test subset in MICARS that would correlate best with the total MICARS score. This was accomplished first without constraints and then with the clinical constraint requiring one test each of Gait, Kinetic Function-Arm, Kinetic Function-Leg, Speech, and Eye Movements. We validated these clinical constraints by factor analysis. We then validated the results in a second cohort of patients; evaluated inter-rater reliability in a third cohort; and used the same data set to compare BARS with the Scale for the Assessment and Rating of Ataxia (SARA). Correlation of ICARS with the seven additional tests that when added to ICARS form MICARS was 0.88. There were 31,481 five-test subtests (48% of possible combinations) that had a correlation with total MICARS score of > or =0.90. The strongest correlation of an unconstrained five-test subset was 0.963. The clinically constrained subtest validated by factor analysis, BARS, had a correlation with MICARS-minus-BARS of 0.952. Cronbach alpha for BARS and SARA was 0.90 and 0.92 respectively; and inter-rater reliability (intraclass correlation coefficient) was 0.91 and 0.93 respectively. BARS is valid, reliable, and sufficiently fast and accurate for clinical purposes.