Predicting Major Bleeding in Patients With Stroke on Antiplatelet Agents

There are a limited number of prediction models that identify patients at risk of major bleeding after stroke or TIA.

Researchers from The Netherlands have developed the S2TOP-BLEED score which estimates the 3-year risk of major bleeding in patients with a history of ischemic stroke or transient ischemic attack (TIA) treated with antiplatelet agents. Results from their validation study were published in Neurology.

Antiplatelet therapy is the cornerstone of secondary prevention in patients with a history of stroke of arterial origin and is tied to a 25% decrease in serious vascular events. However, there is an estimated 2-fold increase in bleeding with aspirin compared with placebo, and this risk further increases with dual antiplatelet therapy regimens.

There are limited numbers of prediction models to identify patients at increased risk for bleeding with antiplatelets for secondary prevention. Risk stratification and bleeding prediction in this group of patients could be helpful for balancing the risk-benefit ratio of antiplatelet therapy.  

Nina A. Hilkens, MD, of the University Medical Center Utrecht in The Netherlands, and colleagues, sought to develop and validate a prediction model for major bleeding risk in patients with a history of TIA or ischemic stroke on antiplatelet therapy.

The investigators analyzed individual patient data from 6 randomized trials that studied the efficacy of long-term antiplatelet therapy for secondary prevention in patients with an ischemic stroke or TIA. The trials included were CAPRIE, ESPS-2, MATCH, CHARISMA, ESPRIT, and PRoFESS.

Based on the data, a risk prediction model was developed and then validated on the PERFORM trial, a randomized clinical trial of more than 18,000 patients with recent ischemic stroke or TIA who were randomly assigned to receive aspirin or terutroban.

Analysis was conducted of 43,112 participants after excluding for possible cardioembolic origin and patients treated with dipyridamole alone. The investigators identified 1530 participants with major bleeding, with death occurring in 10% and nonfatal intracranial bleeding in 18%. The mean risk of bleeding over 1 and 3 years was 1.9% and 4.6%, respectively. The predictors of major bleeding included male sex, smoking, modified Rankin Scale score of ≥3, type of antiplatelet agent, lower body mass index, hypertension, prior stroke, Asian ethnicity, being elderly, and diabetes.

Related Articles

The S2TOP-BLEED model had a c-statistic of 0.63 (95% CI, 0.61-0.64) and the “calibration plot showed good correspondence between predicted and observed risks,” the investigators noted. The risk for major bleeding ranged from 2% in younger patients (45 to 54) without risk factors to 10% in older patients (75 to 84) with multiple risk factors.

The validation population had a slightly higher observed 3-year risk of major bleeding (5.5% vs 4.6%) and the S2TOP-BLEED model had a c-statistic of 0.61 with slight underestimation of major bleeding risk in this population.

The development of this prediction model was limited by only using data from previous trials, which may have excluded patients at the highest risk of bleeding leading to a potential underestimation of bleeding risk.

“We developed and externally validated a practical score that can generate individualized risk predictions of major bleeding after a TIA or ischemic stroke,” the investigators wrote. “Whether this model can guide treatment decisions needs to be investigated in a decision analytical study in which the risk of major bleeding is balanced against the risk of recurrent ischemic events,” the investigators noted.

Disclosures: The authors report multiple disclosures. Please see the study for a full list of disclosures.


Hilkens NA, Algra A, Diener HC, et al. Predicting major bleeding in patients with noncardioembolic stroke on antiplatelets: S2TOP-BLEED. Neurology. 2017;89:1-8.