Can we predict early recurrence in acute stroke?

Published

Journal Article

BACKGROUND: The prevention of early recurrent stroke, which worsens outcomes after a cerebral infarction, is a major objective for acute stroke therapy. The ability to predict which patients are at risk for early recurrence would be useful for both the management and design of clinical trials. METHODS: Using the prospective database with the 1,266 stroke patients admitted in the TOAST study, we analyzed all the patients who had suffered either a transient ischemic attack (TIA) or a recurrent stroke within 3 months after stroke, and their possible association with 20 selected clinical variables. Both univariate and stepwise regression analyses were performed. RESULTS: Sixty-two patients (4.9%) had a second stroke, and 47 patients (3.7%) had at least one TIA. No particular high-risk period was observed. Early recurrent stroke was associated with the large artery atherosclerosis subtype. A prior history of TIA increased the odds for recurrent stroke (OR = 2.52; 1.16-5.46) or poststroke TIA (OR = 3.46; 1.59-7.48). In addition, patients who had a TIA after the stroke had a 17% chance of having an early recurrent stroke, as compared with 4.4% among those that did not (p = 0.001). CONCLUSION: Our present ability to identify patients at risk for early recurrence based on baseline clinical features remains limited. While the presence of TIA before or after the stroke denotes a subgroup of acute stroke patients at higher risk for early recurrence in the first 3 months, no other factors reliably identify high-risk patients.

Full Text

Cited Authors

  • Leira, EC; Chang, K-C; Davis, PH; Clarke, WR; Woolson, RF; Hansen, MD; Adams, HP

Published Date

  • January 2004

Published In

Volume / Issue

  • 18 / 2

Start / End Page

  • 139 - 144

PubMed ID

  • 15218280

Pubmed Central ID

  • 15218280

Electronic International Standard Serial Number (EISSN)

  • 1421-9786

International Standard Serial Number (ISSN)

  • 1015-9770

Digital Object Identifier (DOI)

  • 10.1159/000079267

Language

  • eng