Consenting patients recruited from April 2002 to March 2014 and surviving for 3 months after their first stroke in the study period (index stroke) were included in the analysis, to focus on the period beyond 90‐days poststroke that is not conventionally captured in stroke trials. Analyses were censored at 31 July 2015.
The proportions of patients who were (1) disabled, (2) dead/disabled, (3) dead from all causes, and (4) dead from index‐stroke‐related causes at 1, 2, 3, 4, and 5 years were calculated and categorized according to the 3‐month mRS (or the mRS between 1 and 3 months if the 3‐month mRS was missing). Logistic regression was used to adjust the associations of 3‐month mRS and long‐term outcomes for age and sex. Survival to 5 years after index stroke was assessed using Kaplan‐Meier techniques categorized according to 3‐month mRS. Differences in survival in relation to 3‐month mRS scores were assessed using age‐ and sex‐adjusted Cox proportional hazards models.23The above analyses were repeated for the subgroups defined below:

Treatable major strokes or nonhyperacute/minor strokes: The former were defined as patients seeking medical attention within 6 hours of symptom onset, presenting either to hospital or emergency services with National Institutes of Health Stroke Scale ≥5. These were “minimal criteria” to capture the subset that have been the focus of hyperacute stroke trials and would potentially be eligible for thrombolysis/thrombectomy.24, 25 The remaining patients were classified as nonhyperacute/minor strokes, to facilitate direct comparison.

Atrial fibrillation (AF)‐related or non‐AF‐related strokes: The former were defined as patients with a prestroke diagnosis of AF or AF at presentation, meeting criteria for cardioembolic etiology per the TOAST trial (Trial of Org 10172 in Acute Stroke Treatment) classification system.26

Lacunar or nonlacunar strokes: Classified by TOAST criteria for small‐vessel occlusion.26

We were specifically interested in examining the long‐term death and/or disability outcomes of AF‐related and lacunar strokes because they have been the focus of several recent stroke trials, particularly of secondary prevention strategies.27, 28We also plotted changes in mRS between 3 months and 1 year and between 1 and 5 years after the index stroke for all patients, categorized according to their 3‐month mRS.
Statistical analyses were performed using STATA 13.1 (Statacorp, College Station, TX).
Free full text: Click here