SELECT: did semaglutide prevent diabetes in people with heart disease?

Why this study matters

People living with obesity and heart disease are at high risk of developing type 2 diabetes. The SELECT trial enrolled only people without diabetes, and the researchers planned in advance to track blood sugar throughout. This gives a rare long-term, randomized view of whether semaglutide (Wegovy, Ozempic) changes how blood sugar moves over time. For Canadian readers with prediabetes and heart disease, it addresses a common question: does the medicine hold off diabetes, and does the effect last?

Who was in the study

The analysis used all 17,604 SELECT participants: 8,803 assigned to semaglutide and 8,801 to placebo. Everyone was aged 45 or older, had a BMI of 27 or higher, and had pre-existing cardiovascular disease. BMI (body mass index) compares weight with height. No one had diabetes, defined as an HbA1c below 6.5% at the start. HbA1c is a blood test that reflects average blood sugar over about three months.

People took the study medicine for a mean of 152 weeks and were followed for a mean of 176 weeks. Glycemic status was assessed at week 156.

What the researchers did

SELECT was a multicentre, randomized, double-blind trial. Randomized means chance decided the treatment. Double-blind means neither participants nor staff knew who was on the medicine. People injected semaglutide 2.4 mg or placebo once a week.

This paper is a prespecified secondary analysis, meaning the blood sugar questions were planned before the trial started but were not its main purpose. The main outcomes were HbA1c over time, the share of people reaching biochemical normoglycemia (HbA1c below 5.7%), and the share progressing to biochemical diabetes (HbA1c of 6.5% or higher). "Biochemical" means based on the blood test alone, not a clinical diagnosis. Missing values were filled in using multiple imputation, a statistical method, and the confidence intervals were not adjusted for multiple comparisons.

What they found

In both groups, HbA1c reached its lowest point at week 20. After that it rose at a similar rate in both groups (R32-C32). The gap between groups stayed steady: a mean difference of -0.32 percentage points favouring semaglutide (95% CI -0.33 to -0.30). A confidence interval is the range where the true value most likely lies.

Both differences were very unlikely to be chance (P less than 0.0001) (R32-C31). The authors calculated that 18.5 people would need to be treated to prevent one case of biochemical diabetes.

Body weight levelled off at 65 weeks and was 8.9% lower with semaglutide than placebo. The authors estimated that weight reduction explained 24.5% of the effect on progression to diabetes and 27.1% of the effect on return to normal, so weight loss accounted for only part of the effect. Both regression and progression depended on where blood sugar started.

Side effects and people who stopped

This paper reports blood sugar outcomes only. The summary we reviewed does not report adverse events or how many people stopped treatment. For side effects, see the main SELECT page and the SELECT safety analysis page.

What this study does not tell you

• It shows blood-test thresholds during treatment. It does not show whether diabetes is prevented for good after the medicine stops.
• HbA1c rose at a similar rate in both groups after the early drop, so the medicine did not slow the underlying drift upward; it shifted the starting point.
• Everyone had cardiovascular disease and was aged 45 or older. Results in younger people or people without heart disease are not covered.
• It was a secondary analysis without adjustment for multiple comparisons.
• The number of people who received a clinical diabetes diagnosis, rather than a blood-test result, was not the outcome.
• The summary we reviewed does not state the funding source, although SELECT itself was funded by the maker of semaglutide.

What it means in Canada

This analysis is often quoted as showing semaglutide prevents diabetes. The more precise reading is that, while on the medicine, fewer people crossed the blood-test threshold for diabetes, but blood sugar still climbed over time in both groups. For how the medicine is used and covered here, see semaglutide in Canada and public drug coverage by province.

Related reading

Other studies in this library that bear on the same question.

Source