Abstract
We have developed a 10-item scoring system for estimation of initial severity of neurologic deficit and its subsequent course in stroke patients. The present study evaluated correlations between 10 clinical tests and assessed inter- and intraobserver variation when the scoring system was used by physicians without extensive neurological training. The intraobserver variation for the total score of each patient ranged from 0 to 9; mean, 1. Coefficient of variation was 6.6%. Interobserver variation varied from 0 to 4.5; mean, 2. Coefficient of variation was 8.5%. The pairwise correlations between individual items ranged from insignificant to 0.94. Based on a very high correlation between items 7 and 8 and between items 9 and 10, it was believed that items 8 and 10 could be deleted. We propose a revised scale in which all items have been designed with 4 steps. There is reason to expect that improvement may be obtained by different weighting of items. Assignment of weighing factors must, however, await a longitudinal cohort study.