SELECT’s Undercovered Result: Fewer Hospitalizations on Semaglutide

The SELECT result that matters to hospitals

Semaglutide's famous SELECT headline was a 20% relative reduction in major adverse cardiovascular events. That deserved the coverage. But the more practical result may be this: people on semaglutide spent less time in the hospital, and they were admitted less often, across more than one disease category.

In a 2026 prespecified exploratory analysis of SELECT by Nicholls et al., weekly semaglutide 2.4 mg cut total hospital admissions from 20.4 to 18.3 per 100 patient-years, with a mean ratio of 0.90 (95% CI 0.85 to 0.95; p<0.001). Hospital days fell from 176.2 to 157.2 per 100 patient-years, rate ratio 0.89 (95% CI 0.82 to 0.98; p=0.01). Serious-adverse-event-related admissions also dropped, from 17.1 to 15.2 per 100 patient-years, mean ratio 0.89 (95% CI 0.84 to 0.94; p<0.001).

Scaled to 1,000 patients over four years, that works out to about 84 avoided admissions and 760 avoided hospital days. That is harder to turn into a splashy TV segment than "weight-loss drug," but it is the sort of number health systems actually care about.

In SELECT, semaglutide did not just lower one composite endpoint. It reduced how often patients were hospitalized and how long they stayed there.

What SELECT actually tested

The main SELECT trial, published by Lincoff et al. in 2023, randomized 17,604 adults with BMI of at least 27 kg/m², established atherosclerotic cardiovascular disease, and no diabetes at baseline to semaglutide 2.4 mg weekly or placebo. This was a double-blind, placebo-controlled trial. Median follow-up in the hospitalization analysis was 41.8 months.

This population matters. These were not generally healthy people trying to fit back into old jeans. They had obesity or overweight plus established cardiovascular disease, and they did not have diabetes. That makes SELECT the key trial for asking whether semaglutide's cardiovascular benefits extend beyond glucose lowering. The answer appears to be yes. In a 2024 prespecified analysis by Lingvay et al., the cardiovascular benefit did not depend on baseline HbA1c or on how much HbA1c changed during treatment.

The original primary endpoint was MACE: cardiovascular death, nonfatal myocardial infarction, or nonfatal stroke. Semaglutide reduced that composite by 20%, with a hazard ratio of 0.80 (95% CI 0.72 to 0.90; p<0.001). The hospitalization paper did not replace that result. It extended it.

Why the hospitalization signal looks real

Secondary analyses can mislead. That is always the risk. This one has a few features working in its favor.

First, the hospitalization analysis was prespecified, not fished out after the fact. That does not give it the same evidentiary weight as a primary endpoint, but it is better than a late statistical treasure hunt. Second, three related outcomes moved in the same direction: total admissions, hospital days, and serious-adverse-event-related admissions. Third, the effect did not meaningfully differ across BMI, age, or sex subgroups in the Nicholls analysis.

The safety data line up with the hospitalization data in a way that makes clinical sense. In a 2025 SELECT safety analysis by Kushner et al., serious adverse events were less common with semaglutide than placebo: 33.4% versus 36.4% (p<0.001). The reductions were not confined to cardiac disorders. Cardiac serious adverse events occurred in 11.5% versus 13.5% of participants. Infections and infestations occurred in 7.1% versus 8.4%. Surgical and medical procedures occurred in 4.9% versus 6.2%. Nervous system disorders trended lower, 5.0% versus 5.6%, but that difference was not statistically significant (p=0.08). Neoplasm rates were the same at 4.6% in both groups.

That pattern matters because it argues against a narrow story in which semaglutide only changes plaque biology and nothing else. Fewer infections and fewer procedure-related serious events suggest a broader reduction in illness burden. That is consistent with obesity as a systemic inflammatory condition, although consistency is not proof of mechanism. Human trials can tell you what happened. They are less good at settling exactly why.

Durable weight loss is part of the story

If semaglutide's benefits depend partly on sustained loss of adiposity, durability matters. Here SELECT looks better than older obesity drugs, which often faded over time.

In a 2024 Nature Medicine analysis by Ryan et al., participants assigned to semaglutide had an intention-to-treat weight change of -10.2% at 208 weeks, with a treatment difference of -8.7 percentage points versus placebo (95% CI -9.42 to -7.88; p<0.0001). Weight loss reached its low point around week 65 and then stayed there for roughly three more years of follow-up. In the on-treatment analysis, weight loss at 208 weeks was -11.7%.

That does not prove weight loss explains every downstream benefit, but it makes the hospitalization result easier to believe. The intervention kept changing the risk factor it was designed to change, and it kept changing it for years.

Semaglutide also appears usable in a heart-failure-heavy population that usually punishes wishful thinking. In a 2024 prespecified SELECT heart failure analysis by Deanfield et al., semaglutide reduced ischemic cardiovascular events in participants with prevalent heart failure without evidence of harm in that subgroup. A broader 2024 pooled Lancet analysis by Kosiborod et al. found improvement in heart-failure symptoms and exercise capacity in patients with mildly reduced or preserved ejection fraction across related semaglutide trials.

What the evidence does not settle

This is good evidence. It is not perfect evidence, and the gaps matter.

The hospitalization result comes from one prespecified exploratory analysis within one Novo Nordisk-funded trial. That is stronger than a retrospective registry exercise and weaker than independent replication. No second randomized cardiovascular outcomes trial has yet confirmed the same hospitalization endpoint.

Tolerability is the other major caveat. In the main SELECT report, 16.6% of participants assigned semaglutide discontinued treatment, versus 8.2% on placebo, largely because of gastrointestinal adverse events. Intention-to-treat analyses keep those patients in their original groups, which can dilute the measured benefit among people who actually stay on treatment. But the same fact creates a real-world problem: patients who stop the drug do not collect its long-term benefit.

Safety was broadly reassuring, but not spotless. The Kushner analysis found no excess serious psychiatric events, pancreatitis, or malignancy. It did find more gallbladder events, driven by cholelithiasis. The main SELECT publication reported gallbladder event rates of about 2.8% with semaglutide versus 2.3% with placebo over roughly 40 months, an absolute excess of about 0.5%. That is not large, but it is real. If you prevent 84 admissions per 1,000 patients over four years, then add a few gallstone-related problems back into the ledger, the net still favors treatment. Medicine, however, is rarely neat enough to fit on a mug.

There is also a treatment-dependency problem. SELECT did not study what happens after stopping semaglutide. Outside SELECT, the STEP 1 extension reported by Wilding et al. found that about two-thirds of lost weight returned within 12 months of discontinuation. It is reasonable to expect at least some loss of hospitalization benefit after stopping, but SELECT does not quantify that.

Generalizability is narrower than the marketing blur. SELECT participants were 72.3% male, 84.0% White, and had a mean age of 61.6 years. All had established atherosclerotic cardiovascular disease and none had diabetes at baseline. That supports conclusions for a high-risk cardiovascular population. It does not justify a free-for-all extrapolation to younger adults, lower-risk groups, or every patient seeking weight loss.

How this fits with the broader GLP-1 literature

The hospitalization result does not sit in isolation. It sits on top of a large cardiovascular literature for GLP-1 receptor agonists. A 2025 meta-analysis by Galli et al. covering 99,599 patients found class-wide cardiovascular benefits and acceptable tolerability. In obesity specifically, a 2025 Cochrane review of semaglutide and a 2024 meta-analysis by Singh et al. support cardiovascular and weight-loss benefits in overweight or obese adults without diabetes.

That said, the comparative question is still open. A 2025 Cochrane review of tirzepatide suggests comparable or greater weight loss than semaglutide in obesity trials. What it does not provide is a SELECT-equivalent cardiovascular outcomes trial showing the same reduction in MACE, hospitalizations, or hospital days in a non-diabetic population with established cardiovascular disease. Better weight loss does not automatically mean better hard outcomes. Sometimes it does. Sometimes biology declines to read our slide decks.

The practical verdict

For high-risk adults with obesity and established cardiovascular disease but no diabetes, SELECT makes a stronger case for semaglutide than the usual public conversation does. The key point is not just fewer heart attacks and strokes. It is fewer total admissions, fewer hospital days, and fewer serious adverse events across several organ-system categories during nearly four years of follow-up.

The result still comes with conditions: the evidence is concentrated in one trial program, discontinuation was common, gallbladder events increased, post-cessation outcomes are unknown, and cost-effectiveness remains unsettled at current U.S. prices. But if you want the part of the semaglutide story that feels most connected to real clinical burden, it is not the before-and-after photo. It is the empty hospital bed.