Variables associated with hospital arrival time after stroke: effect of delay on the clinical efficiency of early treatment.

Abstract

A limiting criterion for the eligibility of patients in clinical trials investigating acute stroke therapies is that time between onset of symptoms and arrival in the hospital should fall within the "therapeutic window." The aims of this study were to estimate hospital arrival time in an unselected sample of stroke patients, to assess the association with some clinical and demographic variables, and to evaluate the effects of the delay on the clinical efficiency of an effective treatment. We evaluated the delay in hospital arrival time in 189 patients (84 men, 105 women; mean age, 76.5 years) prospectively collected in the S Orsola-Malpighi Community Teaching Hospital in Bologna, Italy. Cutoffs of 2 and 5 hours were chosen to allow for hypothetical treatment within 3 and 6 hours, respectively. Exact multiple logistic regression was used to predict the delay as a function of dichotomized age, sex, symptoms on awakening, day of the week, hour of the day, area of residence, level of consciousness, and level of motor power defect. We then projected the effectiveness of tissue plasminogen activator (TPA) on disability as estimated with the aid of the odds ratio from the National Institute of Neurological Disorders and Stroke (NINDS) rt-PA Stroke Trial onto our unselected sample to evaluate clinical efficiency of treatment as a function of arrival time and of hypothetical effects of educational efforts to reduce it. The mean interval between onset of symptoms and hospital arrival was 680 minutes; 59 patients (31%) arrived within 2 hours and 100 (53%) within 5 hours. Onset of symptoms when awake, drowsiness or coma, and paralysis of at least one limb were the only independent predictors of hospital arrival within 2 and 5 hours in both the total sample and the subgroup of patients who were awake at stroke onset. The effectiveness of 17%, extrapolated with the aid of the odds ratio of 1.6 of having a favorable outcome (Barthel Index > or = 95 at 3 months) in treated versus untreated patients in the NINDS rt-PA Stroke Trial, corresponded to a projected clinical efficiency of 5%. This could be doubled by hypothesizing a 100% effect of educational efforts in reducing the delay in hospital arrival time. Patients with milder symptoms, for whom treatment might be more effective, were less likely to arrive in time for therapy. The proposed model of the relationship between the delay in hospital presentation after a stroke and the clinical efficiency of a given treatment might be useful for planning future clinical trials on early stroke treatment and predicting the impact of an educational program aimed at shortening arrival time.