2026 American Academy of Neurology Abstract Website

Objective:

To evaluate whether outcomes in the “standard-of-care” control arm of secondary stroke prevention randomized controlled trials (RCTs) outperformed historical standard of care clinical outcomes.

Background:

The “clinical trial effect” proposes that clinical trial participants experience improved outcomes compared to similarly treated patients not enrolled in a clinical trial. Evidence for the clinical trial effect remains mixed. We assessed whether control-arm outcomes in secondary stroke prevention trials outperform expectations based on real-world data.

Design/Methods:

We identified secondary stroke prevention RCTs published between 2000-2025 in NEJM, Lancet, JAMA, Neurology, and Stroke. Trials were included if their primary otucome was recurrant stroke and if their estimated control-arm event rates were based on observational, registry, or natural-history data. Expected event rates were obtained from trial protocols or statistical analysis plans; observed rates were extracted from the primary publications. For each study, expected-to-observed (E/O) event-rate ratios were calculated, with E/O>1.0 indicating fewer observed events than predicted.

Results:

The search yielded 407 results, of which 80 studies were retained after title/abstract screening. An additional 41 studies were excluded because the primary outcome did not include stroke recurrence. A further 30 studies were ineligible because their expected outcome rates were not derived from real-world clinical data. Nine eligible RCTs (>11,000 control patients) were identified. Expected control arm event rates ranged from 1.8-24.7%; observed rates ranged from 0.9-14%. In 8 of 9 trials, observed outcomes were lower than expected (E/O: 1.05-2.93), with NAVIGATE-ESUS as the only exception (E/O: 0.81). The largest trial effects (E/O>2.0) were identified in SPS3, VIST, RESPECT PFO, and POINT.

Conclusions:

In this exploratory analysis, secondary stroke prevention trial participants consistently had less recurrent stroke than expected based upon real-world clinical data. Our findings may support the presence of a clinical trial effect, but should be interpreted cautiously due to non-contemporaneous control bias.