REWIND: Dulaglutide in a Primary-Prevention Population
Last verified May 2026 · Phase 3 · Finished; published July 13, 2019 · NCT01394952 ↗
Trials of this kind usually recruit people who have already survived a heart attack or a stroke, because their events pile up quickly and the study can then be smaller and finish sooner. That design answers a narrower question. REWIND enrolled a majority who had not, accepted the longer follow-up that requires, and still detected a difference. Reading the 12% next to SELECT's 20% without noting the difference in populations is the kind of comparison that makes a weaker trial look like a weaker drug.
- Enrollment
- 9,901
- Duration
- A median 5.4 years — the longest follow-up of any cardiovascular trial in this drug class
- Drug
- Dulaglutide 1.5 mg (Trulicity)
- Population
- Adults 50 and over with type 2 diabetes and an HbA1c of 9.5% or below, entering by one of two routes: having already had cardiovascular disease (31%), or carrying risk factors without an event (69%). ★ That 69% is what makes this trial distinctive — nearly every other trial in this class studied people who had already been damaged. Average age 66.2, 46.3% women, average HbA1c 7.3%, average BMI 32.3, and about 10.5 years since diagnosis.
Primary endpoint
Cardiac death, a nonfatal heart attack, or a nonfatal stroke — whichever came first
Treatment arm
594 of 4,949 (12.0%); 2.4 per 100 person-years
Comparator
663 of 4,952 (13.4%); 2.7 per 100 person-years
Treatment difference: Hazard ratio 0.88 (95% CI 0.79-0.99); P=0.026 for superiority
★ A 12% reduction, which is the smallest of the positive trials in this class and should not be read as the weakest drug. It comes from a population where most people had never had an event and where events are therefore rarer and harder to prevent. In absolute terms, about 72 people treated for five years to avoid one event. ⚠ The curves did not separate for roughly two years — this is not a fast-acting benefit.
Secondary endpoints
| Endpoint | Treatment | Comparator | Difference |
|---|---|---|---|
| Heart attacks that were not fatal ⚠ Essentially unchanged, which is a striking contrast: in the liraglutide and semaglutide obesity trials, heart attacks carried most of the benefit. | 205 of 4,949 (4.1%); 0.8 per 100 person-years | 212 of 4,952 (4.3%); 0.8 per 100 person-years | Hazard ratio 0.96 (95% CI 0.79-1.16); P=0.65 |
| Strokes that were not fatal ★ Down 24%, and the main driver of the headline. Between this trial, LEADER and SELECT, the component doing the work differs each time — a reminder that 'GLP-1s protect the heart' covers several different effects. | 135 of 4,949 (2.7%); 0.5 per 100 person-years | 175 of 4,952 (3.5%); 0.7 per 100 person-years | Hazard ratio 0.76 (95% CI 0.61-0.95); P=0.017 |
| Death from a cardiac cause Fewer, but the confidence interval crossed 1 — a trend, not a finding. | 317 of 4,949 (6.4%); 1.3 per 100 person-years | 346 of 4,952 (7.0%); 1.4 per 100 person-years | Hazard ratio 0.91 (95% CI 0.78-1.06); P=0.21 |
| Death from any cause at all ⚠ About 10% fewer, and not statistically established. Over 5.4 years in nearly 10,000 people, that is a real limit on what can be claimed. | 536 of 4,949 (10.8%); 2.1 per 100 person-years | 592 of 4,952 (12.0%); 2.4 per 100 person-years | Hazard ratio 0.90 (95% CI 0.80-1.01); P=0.067 |
| Admitted to hospital for heart failure ⛔ No reduction at all. This class does not move heart-failure admissions the way SGLT2 inhibitors do, and this is one of the clearest demonstrations of that gap. | 213 of 4,949 (4.3%); 0.9 per 100 person-years | 226 of 4,952 (4.6%); 0.9 per 100 person-years | Hazard ratio 0.93 (95% CI 0.77-1.12); P=0.46 |
| Kidney decline: new protein in the urine, a sustained 30% loss of function, or needing dialysis ⚠ A 15% reduction, driven mainly by new protein in the urine — the softest of the three components and the furthest from what patients care about. | 848 of 4,949 (17.1%); 3.5 per 100 person-years | 970 of 4,952 (19.6%); 4.1 per 100 person-years | Hazard ratio 0.85 (95% CI 0.77-0.93); P=0.0004 |
| Three-month average blood sugar over the whole trial A modest improvement that held for more than five years, in a trial where both arms could have other diabetes drugs added when needed. | 7.3% baseline; sustained reduction throughout follow-up | 7.3% baseline; less sustained reduction | Least-squares mean difference -0.61 percentage points (95% CI -0.65 to -0.58) favoring dulaglutide |
| Weight change over the trial ⛔ Around 1.5 kg against placebo. This drug is dosed for blood sugar, not weight, and nothing here should be read as evidence about it as a weight treatment. | 32.3 kg/m² baseline BMI | 32.3 kg/m² baseline BMI | Least-squares mean difference -1.46 kg (95% CI -1.61 to -1.31) favoring dulaglutide |
| Diabetic eye disease getting worse ★ No excess — unlike the semaglutide cardiovascular trial, which found a clear one. Worth knowing that the eye signal is not uniform across this class. | Numerically similar to placebo | Reference | No significant difference between arms |
| Stopped the drug over a side effect 9.5% against 6.0% on placebo, over more than five years of exposure. | 470 of 4,949 (9.5%) | 295 of 4,952 (6.0%) | Higher on dulaglutide, driven mainly by gastrointestinal events |
Adverse events
| Event | Treatment rate | Comparator rate |
|---|---|---|
| Any stomach or bowel problem ⚠ Read both columns: 47.4% on the drug and 34.1% on placebo. Over five years a third of people report a stomach complaint regardless — the gap is the drug's contribution, not the raw number. | 47.4% (2,347/4,949) | 34.1% (1,687/4,952) |
| Nausea Twice the placebo rate. | 16.9% (839/4,949) | 8.4% (415/4,952) |
| Diarrhea About 16% against 10%. | 15.7% (779/4,949) | 9.7% (482/4,952) |
| Vomiting Roughly double the placebo rate. | 9.0% (446/4,949) | 4.3% (213/4,952) |
| Pancreatitis No meaningful difference, and in line with what every other cardiovascular trial in this class has reported. | 0.5% (27/4,949) | 0.4% (21/4,952) |
| Pancreatic cancer ⚠ A small numerical excess — 21 cases against 16 — that does not reach significance. Similar imbalances have appeared and disappeared across trials in this class without ever being confirmed. | 0.4% (21/4,949) | 0.3% (16/4,952) |
| Medullary thyroid cancer ★ Not a single case in either arm across a median 5.4 years. This is the cancer behind the class's boxed warning, which comes from rodent studies. | 0 events | 0 events |
| Severe low blood sugar needing help from someone else No excess, and numerically lower than placebo. This class only pushes insulin out when glucose is already high. | 1.3% (64/4,949) | 1.5% (75/4,952) |
| Stopped the drug for good over a side effect 9.5% against 6.0%, almost all of it stomach and bowel intolerance. | 9.5% (470/4,949) | 6.0% (295/4,952) |
Subgroup analyses
- People with no prior cardiovascular disease — 6,838 participants, 69% of the trial: 13% fewer cardiac events, with a confidence interval just touching 1
★ The most-cited subgroup in this literature, because no other major trial in the class had enough of these people to look at.
- People who had already had cardiovascular disease — 3,114 participants, 31%: The same 13% reduction, with a near-identical confidence interval
★ Two groups with very different starting risk and effectively the same relative benefit. That is the strongest evidence in this class that the protection is not confined to people already damaged — and it is the reason this drug's label reaches further than liraglutide's.
- Women — 4,589 participants, 46.3%: 19% fewer events, a larger benefit than in men
★ This trial enrolled a higher proportion of women than any other cardiovascular trial in the class, most of which run around a quarter to a third. That makes this one of the few places the question can be asked at all.
- People whose blood sugar was already reasonably controlled: The same benefit as everyone else
More evidence that the cardiovascular effect does not run through lowering glucose.
Clinical significance
REWIND is the best evidence in this class that benefit extends to people at risk rather than only to people already damaged, which is a large population and a hard one to study. For readers, it is mostly context — dulaglutide is not a weight-management drug and is not what anyone here is choosing between. ⚠ Its relevance is to the argument that GLP-1 cardiovascular benefit is a class property, which the neutral EXSCEL result complicates.
Frequently Asked Questions
References
- 1.Gerstein HC, Colhoun HM, Dagenais GR, et al.; REWIND Investigators. Dulaglutide and cardiovascular outcomes in type 2 diabetes (REWIND): a double-blind, randomised placebo-controlled trial. Lancet. 2019. PMID: 31189511.
- 2.Gerstein HC, Colhoun HM, Dagenais GR, et al.; REWIND Investigators. Dulaglutide and renal outcomes in type 2 diabetes: an exploratory analysis of the REWIND randomised, placebo-controlled trial. Lancet. 2019. PMID: 31189509.
- 3.Eli Lilly and Company. Researching Cardiovascular Events With a Weekly INcretin in Diabetes (REWIND, NCT01394952) — Study Record. ClinicalTrials.gov. 2019. https://clinicaltrials.gov/study/NCT01394952
- 4.U.S. Food and Drug Administration. FDA Approves Trulicity (dulaglutide) for the Reduction of Major Adverse Cardiovascular Events in Adults With Type 2 Diabetes. FDA Drug Approval Announcement, February 21, 2020. 2020. https://www.fda.gov/drugs/drug-safety-and-availability/fda-approves-trulicity-cardiovascular-risk-reduction
- 5.Riddle MC, Gerstein HC, Xavier D, et al. HbA1c Reduction in Dulaglutide-Treated Patients Irrespective of Duration of Diabetes, Microvascular Disease, and BMI: A Post Hoc Analysis From the REWIND Trial. Diabetes Care. 2022. PMID: 35043140.