Skip to content
This GLP
← Research
Evidence

The authors do not believe their own hazard ratio

A study found 39% lower all-cause mortality on GLP-1 drugs, then said in its conclusion that the number is probably confounding rather than benefit.

Glenn Torres7 min read
Hazard ratios, BMI under 27 with type 2 diabetesvs DPP-4ideath, any cause0.61vs SGLT2ideath, any cause0.89vs DPP-4icoded heart failure0.71vs SGLT2icoded heart failure0.56The mortality benefit disappears against the stronger comparator.

A 39% reduction in death from any cause is the kind of number that ends up in a headline, and the people who calculated it wrote a sentence asking you not to treat it that way. It is worth understanding why they did, because the reasoning is visible inside their own results — the same kind of internal check as a comparison run where a difference is expected.

Who was studied

Adults with type 2 diabetes and a body mass index below 27, starting a GLP-1 drug, compared separately against new users of DPP-4 inhibitors, SGLT2 inhibitors and usual care. [1] Matching left 7,389, 8,969 and 8,610 GLP-1 users in the three comparisons, followed up to 24 months.

That population is the reason the study exists. The cardiovascular evidence for these drugs comes almost entirely from trials in people with overweight or obesity, and a person with diabetes and a body mass index of 25 has been largely absent from it — the same gap between a trial population and a treated one that eligibility figures keep exposing.

What the three comparisons gave

Against DPP-4 inhibitors: all-cause mortality 0.61, 95% CI 0.52 to 0.71. Major adverse cardiovascular events 0.83, 95% CI 0.71 to 0.98. Coded heart failure 0.71, 95% CI 0.60 to 0.85.

Against SGLT2 inhibitors: cardiovascular events 0.81, 95% CI 0.69 to 0.95, and coded heart failure 0.56, 95% CI 0.47 to 0.65 — but all-cause mortality 0.89, 95% CI 0.76 to 1.05, which crosses one.

What they do stand behind

Coded heart failure, which was lower across all three comparisons. That consistency is the thing a confounding explanation has more trouble with, and it is what the authors present as their most reliable result.

It is still a coded event — a diagnosis appearing in a record rather than an adjudicated one — so it carries the same caveat as any outcome assembled from what somebody wrote down.

And a harm in the same study

Gastroparesis was more frequent on GLP-1 drugs than on DPP-4 inhibitors.

Slowed stomach emptying is the mechanism these drugs work partly through, so a rise in its clinical extreme is coherent rather than surprising, and it belongs in the same summary as the benefits — alongside what is already known about that.

What a reader takes from it

Not a result about weight-loss buyers. Everybody here had type 2 diabetes and a body mass index under 27, which is nobody this site’s sellers are writing prescriptions for.

What travels is the method: when an observational result is huge against a weak comparator and gone against a strong one, the comparator is doing the work. Authors who say so about their own headline are rarer than the finding.

Frequently asked

Do GLP-1 drugs reduce death in people who are not overweight?
This study found a hazard ratio of 0.61 against DPP-4 inhibitors, and the authors state that result likely reflects residual confounding rather than a benefit of that size.
Why does the SGLT2 comparison matter?
Because mortality did not differ against it, 0.89, 95% CI 0.76 to 1.05. A real drug effect would be expected to shrink rather than disappear; confounding produces exactly this pattern.
What result do the authors stand behind?
Lower coded heart failure, which was consistent across all three comparisons. Consistency is harder for a confounding explanation to account for.
Was anything worse?
Gastroparesis was more frequent on GLP-1 drugs than on DPP-4 inhibitors in the same analysis.

Sources

  1. [1] Chen SC, et al. (2026). Cardiorenal Mortality and Safety Outcomes of GLP-1 Receptor Agonists in Type 2 Diabetes With BMI Below 27 kg/m(2): A Target Trial Emulation Diabetes, Obesity and Metabolism. PMID 42642357

Where to get it

Best GLP-1 injections

Every injectable seller we can verify, with the price each one publishes and an honest read of what the trials measured.

Compare providers →

More in Evidence