Is it possible to reduce observer variability in skin score assessment of scleroderma? The ad hoc International Group on the Assessment of Disease Outcome in Scleroderma

J Rheumatol. 1995 Jul;22(7):1277-80.


Objective: A total skin thickness score based on summating the numerical severity grade at a specific number of skin sites is a widely accepted method of disease assessment in scleroderma. Our aim was to simplify the modified Rodnan skin thickness score (grade 0-3 at 17 sites) by either (1) reducing the number of grades or (2) reducing the number of sites, and to increase agreement without losing precision.

Methods: Eight patients were examined in random order by 16 experienced clinicians using the modified Rodnan method. The data were analyzed to determine (1) whether there was an improvement in agreement by reducing severity grades to a 0-2 scale, (2) whether the precision of the score could be maintained by using a reduced number of sites.

Results: Collapsing adjacent grades from the 0-3 scale to generate 3 new 0-2 scales performed similarly (intraclass correlation coefficients 0.66-0.72). A regression analysis showed that 95% of the variance in skin score can be predicted by inclusion of scores from just 5 sites: forearms, lower legs, anterior chest, upper arms, and fingers.

Conclusion: A simpler skin score may be useful in clinical studies depending on the precision required.

MeSH terms

  • Humans
  • Observer Variation
  • Random Allocation
  • Regression Analysis
  • Scleroderma, Systemic / pathology*
  • Skin / pathology*