Design: prespecified cognitive secondary outcomes of two parallel randomised clinical trials, blinded central assessment
Population: 2,165 patients aged 35+ with ≥70% asymptomatic carotid stenosis, mean age 70.1, across 146 sites
Primary outcome: fully adjusted composite cognitive z score at 4 years
Effect: stenting trial difference 0.03 (95% CI −0.09 to 0.15); endarterectomy trial 0.02 (−0.09 to 0.14); threshold for importance 0.50 SD

Asymptomatic carotid stenosis has long been associated with worse cognitive test performance, and that association has been used to argue for intervening in patients who have no neurological symptoms. CREST-2 was designed to test it. Two parallel randomised trials compared carotid stenting plus intensive medical management, and carotid endarterectomy plus intensive medical management, each against intensive medical management alone, in patients aged 35 and over with at least 70% asymptomatic stenosis. Cognitive testing was prespecified.

Of 2,366 US patients, 2,165 (91.5%) took part, mean age 70.1 years, 37.8% women. Five tests covering learning, memory, executive function, attention and processing speed were administered before treatment and annually for up to four years, scored centrally by telephone and blinded to allocation. At four years the fully adjusted composite z score differed by 0.03 (95% CI −0.09 to 0.15) in the stenting trial and 0.02 (−0.09 to 0.14) in the endarterectomy trial. The prespecified threshold for a clinically important difference was 0.50 standard deviations. No individual test differed either.

This matters beyond the vascular clinic because the argument it settles — that fixing a narrowed vessel should restore the function downstream of it — recurs across medicine. Here the intervention was done, the outcome was measured properly, and nothing moved. Cognitive concern is no longer a reason to offer revascularisation to a patient with an asymptomatic stenosis; the decision rests on stroke prevention alone.

Do not cite cognitive benefit when discussing revascularisation for asymptomatic carotid stenosis
Base the conversation on stroke risk and procedural risk, which is where the evidence is
Investigate cognitive complaints in these patients on their own merits rather than attributing them to the stenosis
Ensure intensive medical management is actually in place — it was the comparator, not nothing

Why it matters

It removes an argument for operating on patients who have no symptoms.

The statistics, in plain English

A between-group difference of 0.03 standard deviations against a threshold of 0.50 is not a near miss — it is effectively zero, and the confidence intervals rule out anything approaching a clinically meaningful effect. This is a genuinely negative result rather than an underpowered one: 2,165 patients and 4,954 test batteries. Note that both groups improved slightly, which is what repeated cognitive testing usually produces through practice effects.