# An Oral Pill in a Diabetes Population

By Nora Bissett, Pricing Editor (https://glpwatchdog.org/authors/nora-bissett) | Published: 2026-08-20 | Updated: 2026-10-30

Canonical page: https://glpwatchdog.org/research/an-oral-pill-in-a-diabetes-population

> Over 72 weeks a daily tablet took 9.6% off the scale for people carrying obesity alongside type 2 diabetes. Why we will not set that figure beside the injectable trials.

## Key points

- −9.6% at 72 weeks on 36 mg against −2.5% on placebo, where everyone also had type 2 diabetes.
- Clean dose-response: 5.1% / 7.0% / 9.6% across 6, 12 and 36 mg.
- We do not compare it across trials with the injectables, and the 'fairer' comparison fails too.
- 6.1–9.9% stopped for adverse events vs 4.1% on placebo.
- The abstract's account of one death contains two clauses that contradict each other. We print both.

ATTAIN-2 tested orforglipron, a daily tablet with no food or water restrictions, for weight in 1,613 adults who had both obesity and [type 2 diabetes](https://glpwatchdog.org/conditions/type-2-diabetes). The top dose reached **−9.6%** at 72 weeks against −2.5% on placebo.[1] The number most readers will want next is how that compares to the injections, and this page explains why we will not give it to you.

## What the trial did

Across 136 sites in ten countries, people whose BMI was at least 27 and whose HbA1c sat between 7% and 10% took a once-daily tablet at one of three strengths (6, 12 or 36 mg) or a matching placebo, alongside lifestyle modification, for 72 weeks. The allocation ran 1:1:1:2, deliberately doubling the placebo arm, which is why 630 people took placebo against roughly 330 in each active group. Nine in ten of the 1,613 enrolled saw the trial through, 1,444 in all. The trial used an investigational formulation; the Foundayo label reports these same three arms as its 5.5, 9 and 17.2 mg tablets.[2]

Participants started at an average weight of 101.4 kg, a BMI of 35.6, and an HbA1c of 8.05%. What the trial set out to measure was the percentage of starting weight lost, analyzed on the treatment-regimen estimand. That means everyone randomized is counted regardless of whether they stopped the drug, which is the more conservative of the two ways to report a weight trial.

*ATTAIN-2: percent change in body weight at 72 weeks, treatment-regimen estimand. ETD is the difference against placebo.[1]*

| Dose | Weight change | 95% CI | vs placebo (ETD) |
| --- | --- | --- | --- |
| Orforglipron 6 mg | −5.1% | −6.0 to −4.2 | −2.7 (−3.7 to −1.6) |
| Orforglipron 12 mg | −7.0% | −7.8 to −6.2 | −4.5 (−5.5 to −3.6) |
| Orforglipron 36 mg | −9.6% | −10.5 to −8.7 | −7.1 (−8.2 to −6.1) |
| Placebo | −2.5% | −3.0 to −1.9 | — |

Every dose separated from placebo at p<0.0001, and the dose-response is orderly: each step up produced more weight loss, which is what you want to see and do not always get. Every prespecified weight and cardiometabolic measure, HbA1c included, improved significantly.

## Why we will not put this beside SURMOUNT-1

> **Setting −9.6% next to a number from a different trial is not a comparison.** SURMOUNT-1 enrolled people *without* diabetes, over 72 weeks, with its own placebo arm and its own population. ATTAIN-2 enrolled people *with* type 2 diabetes. Those are different questions, and reading across two trials as though they were arms of one is the move this register criticizes when a seller does it.

The obvious rejoinder is that there is a fairer comparator: the tirzepatide trial that also enrolled people with diabetes. That is a better instinct, and it is still a cross-trial comparison, with different investigators, sites, eligibility rules, placebo responses and years. The gap between two trials’ placebo arms alone can be larger than the difference anyone is trying to detect.

What we can say without inventing a comparison: **type 2 diabetes reliably lowers weight loss on every drug in this class**, in every trial that has looked. That pattern is robust across the literature, which means a reader with diabetes should calibrate against trials run in people with diabetes, whatever the drug. Our [tirzepatide timeline](https://glpwatchdog.org/research/tirzepatide-results-timeline) and [semaglutide timeline](https://glpwatchdog.org/research/semaglutide-results-timeline) each report their own trials on their own terms for this reason.

> The comparison you want requires a trial that has not been run. Saying so is more useful than a number that looks like an answer.

One formal method gets closer than reading across headlines by anchoring two trials through their placebo arms. We take apart what it can and cannot settle in [comparing trials that never met](https://glpwatchdog.org/research/comparing-trials-that-never-met).

## Safety, and one sentence we cannot resolve

Discontinuations for adverse events ran from 6.1% to 9.9% across the orforglipron doses against 4.1% on placebo, driven mainly by gut effects. For an oral GLP-1 that is the number to watch; [danuglipron worked on weight, but most people in its trial did not finish it](https://glpwatchdog.org/research/danuglipron-the-pill-most-people-stopped). Those were rarely severe, and they bunched into the weeks when the dose was climbing. That is the same shape the injectables show, and the reason a titration schedule exists at all.

Ten deaths occurred: six on orforglipron, four on placebo. Here the published abstract says something we cannot fully reconcile. It reports that investigators deemed all deaths unrelated to study treatment *except* one case in the placebo group and one in the 12 mg orforglipron group. Then, in the next sentence, it says that for the orforglipron case *no treatment-related association was reported*.

> Those two clauses point in opposite directions, and the abstract does not say which governs. We are printing both rather than picking the reading that suits a conclusion. The full paper may resolve it; the abstract does not, and a reader weighing this drug should know the ambiguity exists rather than inherit whichever half a summary happened to quote.

In absolute terms, ten deaths across 1,613 people over 72 weeks (six on drug against four on a placebo arm about two-thirds the size of the three drug arms combined) is a small number in a population with obesity and type 2 diabetes, and a trial this size is not designed to detect a mortality difference in either direction.

## What it is useful for

- **It is double-blind and placebo-controlled**, which the pill-versus-pill head-to-head is not. [The first pill-versus-pill trial](https://glpwatchdog.org/research/the-first-pill-versus-pill-trial) explains why that matters with these particular drugs.
- **It reports the conservative estimand.** The treatment-regimen figure counts people who stopped; the efficacy estimand, which flatters every weight trial, was supportive only.
- **The dose-response is clean**, which is evidence the effect is the drug rather than the trial.
- **It says nothing about cardiovascular or kidney outcomes.** A separate trial, ACHIEVE-4, has since tested heart safety against insulin glargine, and orforglipron’s rate of serious cardiovascular events stayed inside the prespecified margin (hazard ratio 0.84, 95% CI 0.59 to 1.20).[3] That is a safety result. Semaglutide has shown outcome benefit; orforglipron has not.

For what orforglipron is and how it is labeled, see [Foundayo](https://glpwatchdog.org/research/foundayo-orforglipron-oral-glp1).

## Frequently asked questions

### How much weight does orforglipron cause people to lose?

ATTAIN-2 enrolled people carrying both obesity and type 2 diabetes. Over 72 weeks the strongest tablet averaged 9.6% against 2.5% on placebo; the middle strength reached 7.0% and the lowest 5.1%.

### Is that less than the injections?

That comparison is not available from this trial. ATTAIN-2 enrolled people with type 2 diabetes, and the injectable trials most often quoted did not. Reading across separate trials with different populations and different placebo arms is not a comparison, however tempting the arithmetic looks.

### Why does having diabetes matter?

Type 2 diabetes lowers weight loss on every drug in this class, in every trial that has examined it. Someone with diabetes should calibrate expectations against trials run in people with diabetes, whichever drug they are considering.

### How many people stopped because of side effects?

Between 6.1% and 9.9% across the orforglipron doses, against 4.1% on placebo. Most stopped for gut effects, which were seldom severe and were concentrated in the weeks the dose was being raised.

### Were there deaths in the trial?

Ten: six on orforglipron and four on placebo, in a trial whose placebo arm was deliberately doubled in size. The published abstract's account of whether any was treatment-related contains two clauses that point in opposite directions, and we cannot resolve it from the abstract alone.

## References

1. Horn DB, Ryan DH, Kis SG, et al. Orforglipron, an oral small-molecule GLP-1 receptor agonist, for the treatment of obesity in people with type 2 diabetes (ATTAIN-2): a phase 3, double-blind, randomised, multicentre, placebo-controlled trial. The Lancet. 2026. PMID: 41275875. https://pubmed.ncbi.nlm.nih.gov/41275875/
2. Eli Lilly and Company FOUNDAYO (orforglipron) tablets — prescribing information: clinical studies (Trials 1 and 2). DailyMed. 2026. https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=8ac446c5-feba-474f-a103-23facb9b5c62
3. Klein KR, Wysham C, Tuttle KR, et al. Cardiovascular safety of orforglipron versus insulin glargine in adults with type 2 diabetes at increased cardiovascular risk (ACHIEVE-4): a phase 3, event-driven, randomised, open-label, non-inferiority, active comparator trial. The Lancet. 2026. PMID: 42815506. https://pubmed.ncbi.nlm.nih.gov/42815506/

---

Cite as: GLP Watchdog, Nora Bissett, "An Oral Pill in a Diabetes Population", published 2026-08-20, updated 2026-10-30. https://glpwatchdog.org/research/an-oral-pill-in-a-diabetes-population

Canonical HTML page: https://glpwatchdog.org/research/an-oral-pill-in-a-diabetes-population

Medical disclaimer: Nothing here is medical advice. We report on this subject; we do not treat anyone. Whether to begin a medication, supplement, or treatment, whether to adjust it, and whether to stop are questions for a clinician licensed to answer them, who knows your history.
