Visual rating scales for age-related white matter changes (leukoaraiosis): can the heterogeneity be reduced?

Stroke. 2002 Dec;33(12):2827-33. doi: 10.1161/01.str.0000038424.70926.5e.

Abstract

Background and purpose: It has been hypothesized that the use of different visual rating scales partly explains the discordant results of studies investigating risk factors and clinical correlates of age-related cerebral white matter changes (leukoaraiosis). We aimed to compare 6 widely used rating scales for leukoaraiosis and to calculate conversion coefficients of the score of 1 scale in the score of a second scale.

Methods: Two trained raters evaluated 80 pairs of CT and MRI scans using 2 CT and 4 MRI rating scales for white matter changes. Correlations among the scales were evaluated and regression lines were constructed with each of the CT and MRI scale scores as variables.

Results: A high correlation was observed in all the paired comparisons of the 6 scales (Spearman's rho ranging from 0.85 to 0.96, P<0.0001). Using regression analysis, we determined numeric parameters to transform the score of 1 scale to the corresponding score for each of the remaining scales and relative confidence intervals. The predictive values of these conversions expressed as R(2) ranged from 0.75 to 0.92.

Conclusions: The present findings support the view that a good correlation exists among the considered visual rating scales for white matter changes. With the limitation that conversion parameters are calculated by applying a linear regression to partly nonlinear scales, their use allows comparison of the results of previous studies that used different scales and to pool data from past and ongoing clinical trials.

Publication types

  • Comparative Study
  • Evaluation Study

MeSH terms

  • Aging*
  • Brain / diagnostic imaging*
  • Brain Diseases / complications
  • Brain Diseases / diagnosis*
  • Cognition Disorders / complications
  • Feasibility Studies
  • Gait Disorders, Neurologic / complications
  • Humans
  • Hypertension / complications
  • Linear Models
  • Magnetic Resonance Imaging*
  • Middle Aged
  • Predictive Value of Tests
  • Risk Factors
  • Sensitivity and Specificity
  • Statistics, Nonparametric
  • Tomography, X-Ray Computed*