PREDICTION OF PROBABILITY OF THE LONG TERM FATAL OUTCOME OF STROKE BY THE DATA OF TERRITORY AND POPULATIONAL REGISTRY

Abstract

Aim. To develop a prediction model of individual probability of long term (within first 28 days from the onset) outcome of stroke.Material and methods. By the method of territory-populational registry, in 2009-2016 in 16 regions of Russia, an analysis performed, of significant predictors of fatal stroke outcome. Overall, 50902 strokes registered in persons older 25 y. In 1553 there were no data on long term mortality. By the results of revealed significant predictors of the fatal outcome of stroke by LOTUS method, a tree-branching was done for development of probability of long term fatal outcome during the first 28 days from disease presentation.Results. The significant predictors were acquired, of the fatal outcome of stroke, and graded by odds ratio. Based on the definition of significant predictors of fatal outcome, first time a prediction model developed of probability of long term stroke fatal outcome taken heterogeneity and multifactorial nature of the disease. Clinical guidelines proposed.Conclusion. The developed prediction model of individual probability of long term stroke fatal outcome shows high level of sensitivity and specificity. Application of such model and of proposed clinical guidelines at various stages of patients management will facilitate diagnostical search, management strategy selection and improvement of prevention.

Keywords

• stroke
• prediction of long term fatal outcome
• solution tree
• lotus method