Nobody has run a trial to find out, and the one matched cohort that exists points favorably rather than conclusively. Across 1,391 pairs with both sickle cell disease and type 2 diabetes, followed for a year, painful vaso-occlusive crises occurred 36% less often among GLP-1 users, at a hazard ratio of 0.64 [1]. People with this disease are largely absent from the trials that built the evidence base for these drugs, and 84.5% of the cohort were Black. That is a group underrepresented in almost every study covered here, including the access picture in uptake after a heart attack or stroke.
The cardiovascular findings are roughly what this drug class produces elsewhere. Major adverse cardiovascular events came in at a hazard ratio of 0.61 (95% CI 0.49–0.76), and stroke at 0.47 (95% CI 0.23–0.97).
The novel result is the one specific to the disease. Vaso-occlusive crises — the painful episodes that define sickle cell disease and drive most of its hospital admissions — occurred 36% less often among GLP-1 users (HR 0.64, 95% CI 0.45–0.91). That is not a cardiovascular benefit restated; it is a claim about the mechanism of the underlying disease, and there is no obvious precedent for it in this drug class. Whether an anti-inflammatory route could plausibly carry such an effect is the open question. The SELECT inflammation substudy is where it is usually raised, because semaglutide cut high-sensitivity CRP by 37.8% at 104 weeks against placebo [2]. That is a marker rather than an outcome, read here in the inflammation substudy.
Two limits shape how far any of this travels. Everyone in the study had type 2 diabetes as well as sickle cell disease, which is an unusual combination and not the typical patient with either condition. The stroke interval is the other limit, because 0.23 to 0.97 excludes 1 by a margin thin enough that a modest amount of unmeasured confounding would erase it. The same caution applies with equal force to the glaucoma comparison, where the upper bound was 0.96.
What makes this worth reporting despite those limits is that the alternative is nothing. Rare-disease populations rarely accumulate enough participants for a trial, and a matched cohort of 1,391 pairs is a substantial body of evidence by the standards of the field it sits in. A result that would be preliminary in type 2 diabetes is, here, among the better evidence available — which is also true of the disease-specific findings in the Crohn's disease cohort.