Development of a novel, weighted, quantifiable stroke scale - Japan strokescale

Background and Purpose-Several stroke scales are available for estimation o f the severity of stroke, but none of them provides information regarding t he relative weights of the observed variables. To define an integrated seve rity of stroke, we developed a quantifiable stroke scale with weighted vari ables that apply conjoint analysis to calculate the relative weight of each item. Methods-We selected 10 variables (consciousness, language, neglect, hemiano psia, gaze, pupillary abnormality, facial palsy. plantar reflex, sensation, and weakness) based on the multivariate analysis of the Keio Stroke Patien t Database Battery. The variables were categorized and evaluated for their distribution and sensitivity. The categorizations were then modified and re checked. The procedure was repeated until the appropriate categorization wa s obtained from 198 patients. A temporary stroke scale without weight was t hen formulated, and the reliability of the scale was examined and revised w ith 80 new stroke patients. As a next step, 150 neurologists were asked to rank a set of 27 virtual patients, each with a different combination of var iables, according to severity. From these rankings, conjoint analysis was u sed to derive utility scores (weights) for each factor level. Results-The relative weights of each of the factors were as follows: consci ousness 49.8%, language 9.9%, weakness of lower extremity 7.3%, pupillary a bnormality 6.8%, gaze palsy 5.6%, weakness of arm 4.3%, weakness of hand 3. 7%, neglect 3.7%, facial palsy 2.4%, plantar reflex 2.2%, hemianopsia 2.2%, and sensory impairment 2.1%. The total score for a patient could be calcul ated from the sum of the scores for each of the variables ranging from -0.3 8 to 27.86. Scoring of 100 patients with acute stroke was carried out, and the changes in scores were followed for validation. Longitudinal clinical m onitoring of the patients correlated well with the scores in each patient. The inter-rater and intrarater reliabilities of the scale were excellent (w eighted kappa 0.83; Cronbach's alpha 0.998). Conclusions-The Japan Stroke Scale is a parametric stroke scale that provid es a quantitative measure of the severity of stroke. Each of the variables of the scale has a relative weight according to the severity of stroke. Rel iability and responsiveness were proved to be excellent. The present data r evealed a potentiality for the Japan Stroke Scale to be a universally accep ted and reliable standardized system from the clinimetrical point of view.