I want to discuss what I believe is the most consequential finding from the SELECT trial, and one that received insufficient attention relative to the primary MACE endpoint: the signal for all-cause mortality reduction.

In SELECT, all-cause mortality was a pre-specified secondary endpoint. The results:^[1]

"All-cause mortality: semaglutide 4.3% vs. placebo 4.7% (HR 0.81; 95% CI, 0.71–0.93)."

This represents a 19% relative reduction in death from any cause. While the trial was not powered for all-cause mortality as a primary endpoint, the point estimate and confidence interval are compelling. The upper bound of the CI (0.93) does not cross 1.0, indicating statistical significance.

Why does this matter more than the MACE endpoint?

1. All-cause mortality is immune to adjudication bias. Death is an unambiguous endpoint. You can debate whether a troponin elevation was a Type 1 or Type 2 MI, but you can't debate whether someone died.
2. It captures safety signals. If semaglutide reduced MI but increased cancer, suicide, or pancreatitis deaths, the all-cause mortality endpoint would be neutral or harmful. A favorable all-cause mortality signal means the drug is not "borrowing" from one cause of death to pay for another.
3. It is the endpoint patients actually care about. Patients don't want to avoid MI only to die of something else. They want to live longer.

Let that sink in: in a 40-month trial, semaglutide reduced the probability of death by approximately 1 in 5 among the events observed. For a drug initiated for "weight management," this is extraordinary.

^[1] Lincoff AM, Brown-Frandsen K, Colhoun HM, et al. Semaglutide and Cardiovascular Outcomes in Obesity without Diabetes. N Engl J Med. 2023;389(24):2221-2232.

This is the discussion I've been waiting for on this forum. The mortality signal from SELECT is, in my view, potentially the most important cardiovascular finding since the original statin mega-trials (4S, HPS).

Consider the historical context of mortality reductions from major cardiovascular interventions:

SELECT's mortality HR of 0.81 is in the same league as PARADIGM-HF and is achieved in a broader population. The only recent trial with a more dramatic mortality reduction was EMPA-REG OUTCOME (HR 0.68), but that was in a more specific population (T2DM with established CVD).

We are looking at a drug that prevents death. Period.

Let me break down the mortality data more granularly. In SELECT, the causes of death were categorized:

The all-cause mortality benefit is driven primarily by the cardiovascular death reduction. Non-cardiovascular deaths were essentially equal between groups, which is reassuring — it means semaglutide is not increasing cancer mortality, infection mortality, or deaths from other causes.

The absolute risk reduction for all-cause mortality was 0.4 percentage points over ~40 months, yielding a NNT of approximately 250 over 3.3 years. While this sounds high, extrapolated over longer treatment periods (as would occur in real-world practice), the NNT improves substantially.

As a patient who had an MI at 54 and is now on semaglutide as part of my secondary prevention regimen, this data is incredibly reassuring. My cardiologist explicitly cited the mortality data when he added semaglutide to my regimen (alongside atorvastatin, metoprolol, aspirin, and lisinopril).

I asked him: "If I stay on this medication for 10 years, how much does it improve my odds?" His answer: based on extrapolation of the SELECT data and the consistency with SUSTAIN-6 and FLOW, the cumulative mortality benefit over 10 years could be substantial, potentially a 5-7 percentage point absolute reduction in all-cause mortality in high-risk populations.

That's not a marginal benefit. That's tens of thousands of lives at a population level.

Your cardiologist's extrapolation is reasonable but I want to offer a nuance: the mortality curves in SELECT were diverging at the end of the trial (they had not reached a plateau). This means the mortality benefit may actually be greater with longer follow-up, as the separation between curves tends to widen over time for treatments that modify disease progression rather than merely treating events.

This is in contrast to drugs that only treat acute events (e.g., anticoagulants for VTE prevention), where the benefit tends to plateau. GLP-1 agonists appear to be modifying the underlying atherosclerotic process — reducing plaque inflammation, improving plaque stability, slowing new lesion formation — which suggests that the benefit should compound over time.

We won't know for certain until longer-term data are available, but the mechanistic argument for increasing benefit with longer treatment duration is strong.