For years, the standard response to prediabetes has been relatively straightforward: improve diet, increase physical activity, lose excess weight and, for some high-risk patients, consider metformin.
A new analysis released ahead of the American Heart Association’s Scientific Sessions 2026 raises a provocative question: could GLP-1-based medicines produce better outcomes for some patients before type 2 diabetes develops?
In a comparison involving health records from 256,492 adults with prediabetes and obesity, people prescribed GLP-1-based medicines were approximately 1.5 times more likely to return to normal-range blood glucose than those prescribed metformin. They also had a reported 31% lower risk of progressing to type 2 diabetes.
The other numbers were even more striking. The GLP-1 group had a 41% lower risk of death from any cause, alongside lower reported risks of heart attack, stroke, heart failure and atrial fibrillation.
Those figures come from a preliminary observational analysis announced by the American Heart Association on November 2, 2026. They do not establish that GLP-1 treatment caused the differences.
That distinction is central to the metformin vs GLP-1 debate. A large health-records analysis can identify patterns across thousands of patients, but it cannot automatically separate the effects of a medicine from the characteristics of the people who receive it.
The results are interesting enough to influence research priorities. They are not sufficient to rewrite diabetes-prevention guidelines on their own.
What the Metformin vs GLP-1 Study Actually Measured
The research was led by Vinicius de Sousa Barzon Serra, a cardiovascular medicine research fellow at Mayo Clinic in Scottsdale, Arizona.
His team analyzed electronic health records from the TriNetX global multicenter network, drawing on information from 104 healthcare institutions between 2010 and 2023.
The analysis included 256,492 adults diagnosed with both prediabetes and obesity. Their average age was approximately 52, and 76% were women.
Approximately half had received prescriptions for GLP-1-based treatment, while the other half had been prescribed metformin.
The GLP-1 category included semaglutide, liraglutide, dulaglutide and tirzepatide. The last of these is technically a dual GIP and GLP-1 receptor agonist rather than a GLP-1-only medicine.
Those medicines differ in their molecular design, clinical indications, dosing schedules and average effects on body weight. Combining them into one treatment category makes the analysis easier to conduct at scale, but prevents reliable conclusions about the performance of any one drug.
The researchers used matching methods to pair patients with similar recorded characteristics, including age, other diagnoses, high blood pressure, sleep apnea, nicotine dependence and certain medication histories.
Patients were followed for an average of approximately three years. The American Heart Association’s announcement also described mortality comparisons over a five-year observation period, although it did not provide enough methodological detail to explain the distinction fully.
This was an observational study. No investigator randomly assigned patients to semaglutide, tirzepatide, another GLP-1-based medicine or metformin.
Instead, researchers examined what happened in routine care among people who had already received different treatments.
That approach can reveal potentially important associations, particularly in populations larger and more diverse than those enrolled in many randomized trials.
But it cannot eliminate every difference between treatment groups. Matching on recorded characteristics is not equivalent to randomization, particularly when medication choice is strongly affected by insurance coverage, treatment access and personal health circumstances.
The research was identified as abstract MDP1524 for the American Heart Association’s Scientific Sessions 2026, held November 6–9 in Chicago.
The AHA explicitly characterized the findings as preliminary conference research that had not undergone the full peer-review process required for publication as a journal manuscript.
The Numbers Behind the Headlines
The diabetes-related results were the central finding.
Compared with patients prescribed metformin, those prescribed GLP-1-based medicines had an approximately 1.5-times higher chance of returning from prediabetes to normal-range glucose measurements.
They also had a reported 31% lower relative risk of developing type 2 diabetes during follow-up.
The distinction between relative and absolute risk matters. A 31% relative reduction does not mean 31 out of every 100 patients avoided diabetes because of treatment.
To calculate that kind of absolute benefit, readers would need to know the actual event rates in each group and the time period over which they were measured.
Those details were not provided in the AHA news release.
The cardiovascular and mortality findings require even more care.
The analysis reported a 41% lower relative risk of death from any cause among GLP-1 users than among metformin users. Heart attacks were 15% less frequent, strokes 20% less frequent, heart failure events 25% less frequent and atrial fibrillation events 8% less frequent.
All of these figures describe associations observed in the analyzed records.
The public summary did not include the underlying absolute event counts, confidence intervals or detailed time-to-event estimates necessary for readers to evaluate the size and statistical precision of each outcome.
A relative reduction can appear dramatic even when the absolute difference is small. Conversely, a modest relative reduction may be clinically important when the underlying outcome is common.
Without those details, the percentages are incomplete as measures of clinical benefit.
The mortality result deserves particular scrutiny. A 41% difference in death risk between two medication groups is large, especially in an observational comparison where treatment selection may reflect numerous unmeasured factors.
It is a finding that warrants investigation, not proof that switching from metformin to a GLP-1 medicine will reduce an individual’s risk of death by that amount.
Similarly, a return to normal-range blood glucose is an encouraging metabolic outcome, but it does not necessarily mean that the underlying susceptibility to diabetes has permanently disappeared.
The value of these results lies in the consistency of the reported direction across several outcomes. Their limitation is that the public evidence cannot establish precisely how much of the difference was caused by treatment.
Big relative risk reductions from observational records are a reason to investigate a treatment. They are not a substitute for randomized evidence.
Why the Comparison Matters for Prediabetes Treatment
Prediabetes affects an enormous population. According to the American Heart Association’s 2026 statistical report, nearly 40% of U.S. adults have blood-glucose measurements within the prediabetes range.
Prediabetes means glucose levels are elevated but have not reached the diagnostic threshold for type 2 diabetes. The condition increases future diabetes risk, but progression is not inevitable.
The probability of developing diabetes varies substantially. Some high-risk populations may experience annual progression rates approaching 10%, while others have much lower rates.
Current prevention guidance emphasizes lifestyle intervention. The American Diabetes Association’s 2026 Standards of Care recommend structured programs that support sustained weight reduction through dietary changes and regular physical activity.
Metformin is an established, inexpensive generic medicine with substantial long-term evidence for delaying diabetes in selected high-risk patients.
However, it is not automatically recommended for every person with prediabetes.
The ADA recommends considering metformin particularly for adults aged 25–59 with high-risk characteristics, including a BMI of at least 35, higher fasting glucose or HbA1c measurements, or a history of gestational diabetes.
Its use specifically for diabetes prevention is off-label in the United States. The FDA has approved metformin for type 2 diabetes treatment, but not as a medicine specifically indicated to prevent the disease.
GLP-1-based medicines occupy a different position. Several are approved for treating obesity, which itself is a major risk factor for developing type 2 diabetes. Some are approved for treating diabetes or other specified conditions.
But the FDA has not approved any pharmacological product solely for preventing type 2 diabetes in people with prediabetes.
The distinction becomes especially relevant when discussing Ozempic for prediabetes. Ozempic contains semaglutide and is approved for specified type 2 diabetes indications. Wegovy contains the same active ingredient but has a separate indication for chronic weight management and certain additional approved uses.
A patient with prediabetes and obesity may qualify for an obesity medicine based on the applicable label and clinical assessment, but that does not mean the medicine has a separate FDA-approved diabetes-prevention indication.
There is randomized evidence supporting improvement in glucose measurements with semaglutide.
The Phase 3 STEP 10 trial, published in The Lancet Diabetes & Endocrinology in 2024, compared once-weekly semaglutide 2.4 mg with placebo in adults with obesity and prediabetes.
After 52 weeks, 81% of participants receiving semaglutide had returned to normoglycemia, compared with 14% receiving placebo.
Average body weight declined by 13.9% in the semaglutide group versus 2.7% in the placebo group.
That trial established a substantial difference between semaglutide and placebo in the studied population. It did not directly compare semaglutide with metformin.
It therefore supports the idea that obesity treatment can improve glycemic status, without independently validating the new observational comparison.
Grey Pepper has also covered a randomized investigation of type 2 diabetes remission among young adults, illustrating why lifestyle intervention and weight management remain central to metabolic treatment.
The practical issue is not whether patients must choose lifestyle changes or medication. It is how different approaches can be combined appropriately for individual risk, health circumstances and treatment goals.
The Confounding Problem Behind the Headline
The biggest weakness in the new comparison is not its sample size. It is the reason different people received different medicines.
In routine clinical practice, treatment assignment is rarely random.
Metformin is inexpensive, widely available and often prescribed in primary care. GLP-1-based treatments may be considerably more expensive, require different insurance authorization and involve specialist consultations or obesity-management programs.
Those differences can create systematic distinctions between the people receiving each treatment.
Patients who obtain GLP-1 medicines may have different insurance coverage, income, access to clinicians, treatment goals, weight-loss histories or engagement with healthcare services.
Some of those characteristics may be recorded in an electronic medical record. Others may be missing, inconsistently documented or impossible to measure reliably.
The researchers attempted to address this by matching patients on recorded characteristics such as age, coexisting diseases and other medications.
That is an important statistical step, but it does not eliminate all residual confounding.
For example, two people of the same age with the same diagnoses may have very different patterns of physical activity, dietary intake, healthcare access or treatment adherence.
Those differences could contribute to future diabetes diagnoses and cardiovascular events regardless of the medicine prescribed.
Prescription records introduce another limitation. They may identify a treatment recommendation or prescription, but not whether the patient filled it, took it regularly, remained on the treatment or reached the intended dose.
The AHA release specifically notes that medication adherence and dose were not adequately reflected in the available records.
Diagnosis codes may also fail to capture some outcomes or classify patients consistently. The reliability of a coded diagnosis depends on the quality and completeness of the underlying healthcare record.
The analysis also combined four different GLP-1-based drugs into one category, despite substantial differences in their average effects on body weight and other outcomes.
It therefore cannot establish whether semaglutide, liraglutide, dulaglutide or tirzepatide was responsible for a particular observed association.
The large mortality difference is especially vulnerable to questions about residual confounding and selection effects.
It is entirely possible that some of the observed benefit reflects medication effects. It is also possible that other differences between the groups explain part of the association.
The study design cannot reliably determine the contribution of each explanation.
As the researchers themselves acknowledge, the findings do not establish cause and effect.
More than 256,000 records can make an association statistically compelling without making it causally certain.
Is Weight Loss Doing the Heavy Lifting?
One explanation for the findings is familiar to obesity researchers: patients who lose substantial weight often improve their glucose regulation and several cardiovascular risk factors.
Robert Eckel, professor of medicine emeritus at the University of Colorado Anschutz Medical Campus and a former American Heart Association president, was not involved in the new study.
In comments included with the AHA announcement, he said the findings were not surprising and suggested that weight loss probably accounted for an important part of the observed benefit.
That interpretation is biologically plausible.
Excess adiposity contributes to insulin resistance and metabolic dysfunction. Sustained weight loss can improve insulin sensitivity, reduce glucose levels and lessen the likelihood that prediabetes progresses to type 2 diabetes.
GLP-1-based obesity medicines generally produce larger average weight reductions than metformin, particularly at doses approved for chronic weight management.
That difference could account for some of the observed separation between the treatment groups.
But the new AHA release does not provide the detailed weight-change data or mediation analyses needed to measure how much of the lower diabetes or cardiovascular risk was attributable to weight reduction.
The possibility of additional benefits beyond weight loss also remains relevant.
GLP-1 receptor agonists have effects on glucose-dependent insulin secretion, appetite and other physiological processes, while several agents have shown cardiovascular benefits in randomized trials conducted in specific populations.
Those established benefits should not be confused with proof that the new prediabetes comparison is causal.
A randomized cardiovascular outcome trial involving one medicine and one patient population cannot automatically validate a retrospective comparison involving four different medicines and patients selected through routine healthcare records.
Grey Pepper has previously examined this problem in its analysis of observational cardiovascular comparisons involving Ozempic and Mounjaro.
The same caution applies here. Real-world studies can strengthen the case for further investigation, but large differences in outcomes should be interpreted in the context of study design, missing variables and the available randomized evidence.
There is also a practical question that statistics alone cannot answer.
Metformin is a low-cost generic medicine with extensive clinical experience. GLP-1-based treatments are generally much more expensive and may require ongoing injections, dose escalation and monitoring for adverse effects.
Some GLP-1-based medicines are available orally, but the class is not uniform in price, tolerability or convenience.
For patients with obesity and prediabetes, the relevant decision involves more than which therapy produces the largest average weight reduction.
It also involves cardiovascular risk, potential benefits, side effects, affordability, insurance coverage and the likelihood of maintaining treatment over time.
Eckel suggested that GLP-1 medicines may deserve consideration as an additional treatment option rather than simply as a replacement for metformin.
That is a more defensible interpretation of the new evidence than declaring the older medicine obsolete.
The study raises the possibility of benefits beyond conventional prevention strategies. It does not establish that replacing metformin with a GLP-1 medicine is the best choice for every patient.
What Would Settle the Metformin vs GLP-1 Question?
The clearest answer would come from a randomized trial specifically designed to compare treatment strategies for adults with obesity and prediabetes.
Such a study could assign eligible participants to an appropriate GLP-1-based medicine, metformin or a combination of the two, alongside comparable lifestyle support.
Researchers would then follow participants long enough to measure progression to type 2 diabetes, sustained return to normoglycemia, cardiovascular outcomes, body-weight changes and adverse events.
Absolute event rates would be particularly important.
A relative risk reduction is useful for comparing groups, but patients and healthcare systems also need to know the number of people who experience an outcome with and without treatment.
Those figures help establish the potential clinical benefit relative to medication costs, treatment burden and side effects.
A rigorous trial would also need to address what happens when treatment is interrupted or discontinued.
Weight regain following cessation of GLP-1 obesity therapy is well documented in adult research. For a long-term prevention strategy, maintaining benefit may require maintaining treatment or other effective weight-management interventions.
Comparing medications over a short interval would therefore answer only part of the question.
A health system considering expanded preventive use must also evaluate cost-effectiveness and affordability over years, not merely the clinical effects observed during initial treatment.
The new AHA findings do not settle those questions.
They establish that in a very large observational dataset, people receiving GLP-1-based medicines experienced more favorable recorded diabetes and cardiovascular outcomes than people receiving metformin.
That is a meaningful research signal. But its strength is limited by the absence of random treatment assignment, incomplete medication-exposure information, missing absolute outcome data and the grouping of several different drugs into a single category.
For patients, the practical conclusion remains more modest than the headlines.
Prediabetes does not inevitably progress to type 2 diabetes. Lifestyle intervention remains a foundation of prevention, metformin remains an evidence-based option for selected high-risk patients, and GLP-1-based obesity medicines provide additional treatment possibilities when clinically appropriate.
Serra emphasized in the AHA announcement that the point was not to abandon metformin, but to support better discussions between clinicians and patients about the options available.
That is a reasonable message from preliminary research.
The harder question is whether the greater expense and treatment burden of newer medicines produce enough additional long-term benefit to justify their broader use for diabetes prevention.
Until randomized evidence directly addresses that comparison, the 256,492 health records should be treated as a reason to investigate more carefully, not as a verdict that one treatment strategy has definitively defeated another.
The study found better outcomes among GLP-1 users. It did not prove that the medicines caused those outcomes, or that the most expensive prevention strategy should become the first one offered.
Sources and Further Reading
- American Heart Association: GLP-1 Meds Linked to Lower Diabetes, Heart Risks Than Standard Diabetes Prevention Medication. Scientific Sessions 2026, Abstract MDP1524, November 2, 2026.
- The American Journal of Managed Care: GLP-1 Use Tied to Lower Diabetes, Cardiovascular Risk vs Metformin in Prediabetes, November 2, 2026.
- Reuters: GLP-1 Drugs Help Control Prediabetes Better Than Current Standard Medicine, November 4, 2026.
- American Diabetes Association: Prevention or Delay of Diabetes and Associated Comorbidities — Standards of Care in Diabetes 2026.
- McGowan BM, et al. Efficacy and Safety of Once-Weekly Semaglutide 2.4 mg Versus Placebo in People With Obesity and Prediabetes (STEP 10). The Lancet Diabetes & Endocrinology, 2024.
- Indian Health Service: Diabetes Prevention — Clinical Guidance, Lifestyle Intervention and Metformin’s Regulatory Status.
- Diabetes Prevention Program Research Group: Reduction in the Incidence of Type 2 Diabetes With Lifestyle Intervention or Metformin. New England Journal of Medicine, 2002.
