Can someone help me understand the CI and significance issue? The overall MACE result was significant (p < 0.001) but CV death individually was not (CI crossed 1.0). Does that mean semaglutide doesn't actually reduce cardiovascular death?

I'm trying to wrap my head around how the composite can be significant but individual components might not be.

Great question. This comes down to statistical power. SELECT was powered for the composite MACE endpoint, not for individual components. There were 569 primary endpoint events but only 223 CV deaths. With fewer events, you have wider confidence intervals and less power to detect a statistically significant difference for each individual component.

The CV death HR of 0.85 (95% CI: 0.71-1.01) actually trends toward benefit — the point estimate is consistent with the overall result — but the upper bound of the CI barely crosses 1.0. This is what we call "directionally consistent but not individually significant."

The reason we use composite endpoints in CV outcome trials is precisely because individual events are too infrequent to study in isolation without enormous sample sizes. The composite approach has been standard since the TIMI group formalized it in the early 2000s.[4]

Bottom line: the CV death result is not evidence of no effect. It's insufficient evidence to conclude there is an effect. Those are very different statements.

Jumping in with an important nuance about the all-cause mortality finding. The HR of 0.81 (95% CI: 0.71-0.93) is striking, but remember this was not a pre-specified hierarchical endpoint. The trial used a gate-keeping strategy — if the primary MACE composite was significant, they would then test CV death, then non-fatal MI, etc. All-cause mortality was essentially a secondary/exploratory endpoint.

That said, the totality of evidence is compelling. The 20% MACE reduction is robust, the safety profile in 17,000 patients is reassuring (no pancreatitis signal, no thyroid cancer signal with 40 months of follow-up), and the consistency across subgroups is excellent.

For those asking whether this changes guidelines: the ACC already added GLP-1 RAs to the secondary prevention algorithm in their 2023 decision pathway update. The 2024 AHA/ACC chronic coronary disease guidelines will almost certainly incorporate SELECT. We're watching a paradigm shift in real time.

One more analytical point. I compared SELECT to other landmark secondary prevention trials using the per-year ARR to standardize across different follow-up periods:

• 4S (simvastatin): per-year ARR for MACE ~1.0%[5]
• HOPE (ramipril): per-year ARR ~0.75%[6]
• SELECT (semaglutide): per-year ARR ~0.46%
• FOURIER (evolocumab): per-year ARR ~0.55%[7]

SELECT's per-year ARR is somewhat lower than the statin and ACE-I trials, but those were conducted in an era with less background therapy. When you account for the fact that SELECT participants were already well-treated, the incremental benefit is impressive.

The absolute risk reduction matters most for cost-effectiveness. At current semaglutide pricing (~$1,350/month), the cost per MACE event averted is roughly $350,000. That's above the typical willingness-to-pay threshold of $100-150K/QALY. Price reductions or generic entry will be needed to make this cost-effective at population scale.

Excellent cross-trial comparison. The cost-effectiveness question is the elephant in the room, and frankly, the pharmacoeconomics will determine adoption more than the clinical evidence.

But let's not lose sight of what SELECT represents scientifically: the first definitive evidence that pharmacologic weight management reduces hard cardiovascular outcomes in non-diabetic patients. That's a breakthrough. It validates the hypothesis that obesity is a causal, treatable risk factor for atherosclerotic CVD — not just a correlated comorbidity.

Whatever happens with pricing and access, SELECT has permanently changed how we think about the relationship between adiposity and cardiovascular risk.