Selection of acute stroke patients for treatment of visual neglect.

Although visual neglect is a predictor of poor outcome after stroke, some patients regain independence, whilst others take up considerable rehabilitation resources. Intensive treatment of visual neglect is available and a knowledge of the predictive features in the recovery of these patients would be helpful in the early selection of patients for treatment. A study was therefore carried out to determine the prognosis of patients presenting with visual neglect at two to three days after stroke. Linear logistic regression showed that the initial degree of paralysis (measured by the Motricity Index), the severity of neglect (measured by the Visual Neglect Recovery Index) and the patient's age were the significant predictors of independence (Barthel score 20), mild dependence (Barthel 15-19), and moderate/severe dependence (Barthel 0-14) in surviving patients at three months and at six months. Regression equations correctly predicted 78% of outcomes, and had a sensitivity and specificity for "independence" of 84% and 90% respectively, and a sensitivity and specificity for "moderate/severe dependence" of 89% and 80%. It is suggested that these equations may be useful in selecting comparable groups of patients for randomised controlled trials of treatment of visual neglect.