Abstract

The aims of this study were to determine the important prognostic factors at presentation which identify patients with transient ischaemic attacks (TIA) who are at high risk (and low risk) of serious vascular events and to derive a prediction model (equation) for each of the major vascular outcome events. A cohort of 469 TIA patients referred to a University hospital, without prior stroke, were evaluated prospectively and followed up over a mean period of 4.1 years (range 1-10 years). The major outcome events of interest were 1) stroke 2) coronary event and 3) stroke, myocardial infarction or vascular death (whichever occurred first). Prognostic factors and their hazard ratios were identified by means of the Cox proportional hazards multiple regression analysis. The significant adverse prognostic factors (in order of strength of association) for stroke were an increasing number of TIAs in the three months before presentation, increasing age, peripheral vascular disease, left ventricular hypertrophy and TIAs of the brain (compared with the eye); the prognostic factors for coronary event were increasing age, ischaemic heart disease, male sex, and a combination of carotid and vertebrobasilar TIAs at presentation; and for stroke, myocardial infarction or vascular death they were increasing age, peripheral vascular disease, increasing number of TIAs in the three months before presentation, male sex, a combination of carotid and vertebrobasilar TIAs at presentation, TIAs of the brain (compared with the eye), left ventricular hypertrophy and the eye), left ventricular hypertrophy and the eye), left ventricular hypertrophy and the presence of residual neurological signs after the TIA. Prediction models (equations) of both the relative risk and absolute risk of each of the major outcome events were produced, based on the presence or level of the significant prognostic factors and their hazard. Before it can be concluded that our equations accurately predict prognosis and can be generalised to other populations, their predictive power needs to be validated in other, independent samples of TIA patients (which we are currently doing).