Characteristics of 4 stroke scales for the detection of changes in clinical signs in the acute phase of stroke

Abstract

Objectives: Detection of minor changes in clinical signs of stroke may be of interest when evaluating treatment interventions. This study analyzes the internal structure of four frequently used stroke scales and compares them for their sensitivity to detect changes in neurologic signs in the first week after acute stroke. Methods: A cohort of 77 hospitalized acute stroke patients was scored by the Mathew, the National Institutes of Health (NIH), the Scandinavian, and the Orgogozo scales within 48 hours of hospital admission and again 7 days later. Results: Scores on different scales correlated well with each other (range of absolute value of Spearman R,.82-.91; P <.001 in all comparisons). Scales reflected significant changes from entry to reexamination: P =.0013 for the Scandinavian scale (P =.004 for prognostic and P =.009 for long-term items, respectively); P =.00009 for the Orgogozo scale; P =.000007 for the Mathew scale; and P <.000001 for the NIH scale. This difference in sensitivity coincided with the number of factors extracted by principal component analysis: higher sensitivity of a scale was associated with a larger number of factors. Initial scores differed significantly among patients who were discharged, patients who died, and patients who remained hospitalized 7 days after the first examination (Kruskal-Wallis ANOVA, P <.01 for all scales). Conclusions: There are considerable differences in the internal structure of the different scales as reflected by the different number of factors extracted from the scale items. The application of the NIH scale is recommended for the most sensitive detection of changes in stroke signs. Copyright © 2001 by National Stroke Association