IPM Take
GLP-1 medicines are no longer simply diabetes drugs that happen to cause weight loss.
They are increasingly becoming obesity therapies, cardiovascular medicines and potentially tools for preventing diabetes before it starts. That expansion changes the policy argument around access because the value of treatment can no longer be judged solely by kilograms lost or HbA1c reduced.
New preliminary data from more than a quarter of a million adults with obesity and prediabetes push that argument further. Compared with matched patients receiving metformin, people prescribed GLP-1-based medicines were more likely to return to normal glycaemia and less likely to develop type 2 diabetes. They also experienced lower rates of myocardial infarction, stroke, heart failure, atrial fibrillation and all-cause death.
Those numbers are attention-grabbing.
They are not the result of a randomized head-to-head trial.
The study used retrospective electronic health records, and the GLP-1 group included semaglutide, liraglutide, dulaglutide and tirzepatide, the latter being a dual GIP and GLP-1 receptor agonist. Prescriptions do not prove adherence, residual differences between patient groups may remain after matching and the analysis cannot determine whether weight loss, direct pharmacological effects or other differences explain the associations.
That distinction is essential before this becomes another argument that metformin has been “beaten”.
The more consequential signal is that cardiometabolic treatment is moving earlier in the disease trajectory.
Current diabetes guidance already recommends looking beyond glucose alone in people with obesity and prediabetes, with weight loss, prevention of hyperglycaemic progression and cardiovascular risk all part of person-centred prevention goals.
The access question is beginning to change accordingly.
Instead of asking whether an expensive obesity medicine is justified before diabetes appears, payers may increasingly have to ask what waiting for diabetes to develop will cost.
Executive Summary
Preliminary research released for the American Heart Association Scientific Sessions 2026 analysed electronic health records from 256,492 adults with obesity and prediabetes treated with either metformin or GLP-1-based medicines.
Participants had an average age of approximately 52 years, and 76% were women. Data came from 104 institutions in the TriNetX electronic health-record network between 2010 and 2023, with approximately three years of average follow-up.
Researchers matched patients according to factors including age, hypertension, sleep apnoea, nicotine dependence and medication use.
Compared with metformin, GLP-1-based treatment was associated with approximately a 1.5-fold greater likelihood of returning from prediabetes to normal glycaemia and a 31% lower risk of progressing to type 2 diabetes.
The analysis also reported a 41% lower risk of all-cause mortality, alongside 15% fewer myocardial infarctions, 20% fewer strokes, 25% fewer heart-failure events and 8% fewer cases of atrial fibrillation.
These figures describe associations and should not be interpreted as randomized estimates of treatment effects.
The analysis combined semaglutide, liraglutide, dulaglutide and tirzepatide into a single treatment category despite differences in pharmacology, indications, effectiveness and prescribing patterns.
The study relied on diagnosis and prescription codes, did not capture actual medication adherence or dose reliably and remains available only as a conference abstract rather than a peer-reviewed full manuscript.
The findings nevertheless align with existing randomized evidence showing that substantial pharmacological weight loss can delay progression from prediabetes to diabetes. In the three-year SURMOUNT-1 analysis, tirzepatide reduced progression to type 2 diabetes compared with placebo among people with obesity and prediabetes.
The new study asks a different and clinically important question by comparing GLP-1-based treatment with metformin rather than placebo.
It does not yet answer it definitively.
Why it matters
- HTA bodies: Assessment of obesity pharmacotherapy is increasingly moving beyond percentage weight loss. Diabetes prevention, cardiovascular outcomes, kidney outcomes, quality of life and durability of benefit are becoming part of the value proposition. However, comparative effectiveness claims against inexpensive metformin should require stronger evidence than retrospective electronic-record analyses, particularly when several pharmacologically distinct medicines are pooled together.
- Payers: Prediabetes creates a difficult reimbursement problem because the intervention occurs before the costly disease endpoint appears. Metformin is inexpensive and supported by decades of prevention evidence, while modern GLP-1-based treatments carry substantially greater acquisition costs. The relevant economic question is therefore not whether GLP-1 medicines lower glucose more effectively, but whether preventing diabetes, cardiovascular events and obesity-related complications can offset enough downstream cost to justify earlier treatment in clearly defined high-risk patients.
- Industry / innovation partners: The next competitive frontier is likely to be prevention rather than treatment of established diabetes. Randomized trials designed around progression to diabetes, cardiovascular outcomes and comparison with active preventive strategies such as metformin could materially expand the evidence base for earlier incretin therapy. Real-world analyses can identify the opportunity, but they cannot substitute for those trials.
Prediabetes occupies an uncomfortable place in modern healthcare.
It identifies people whose glucose regulation is already abnormal but who have not yet crossed the diagnostic threshold for type 2 diabetes. Clinicians know the condition carries substantial future metabolic risk, yet the treatment pathway often remains comparatively conservative until disease progresses.
Lifestyle intervention remains foundational. Metformin has decades of evidence behind it and is inexpensive, familiar and generally well tolerated. For many patients, that combination makes clinical and economic sense.
The arrival of highly effective obesity medicines is beginning to destabilise that model.
GLP-1 receptor agonists and related incretin therapies can produce weight loss on a scale that older pharmacological interventions could not reliably achieve. Several agents also have cardiovascular or kidney evidence in defined populations, meaning their value increasingly extends beyond glycaemic control.
The question is therefore shifting.
If obesity and prediabetes are part of the pathway toward diabetes and cardiovascular disease, should health systems wait until more advanced disease appears before using the most effective available treatment?
A large electronic-health-record analysis released ahead of the American Heart Association Scientific Sessions 2026 provides an important new signal.
Researchers examined records from 256,492 adults with both obesity and prediabetes. About half had received a GLP-1-based medicine and half metformin.
The GLP-1 category included semaglutide, liraglutide and dulaglutide, as well as tirzepatide, which combines GIP and GLP-1 receptor agonism.
Patients in the two groups were matched on characteristics including age, existing health conditions and medication use in an effort to create more comparable populations.
The resulting differences were substantial.
People prescribed GLP-1-based therapy had around a 1.5-fold greater likelihood of returning to normal blood glucose than those receiving metformin.
Progression to type 2 diabetes was 31% lower.
And the associations extended beyond glycaemia.
The analysis reported 15% fewer myocardial infarctions, 20% fewer strokes, 25% fewer heart-failure events and 8% fewer cases of atrial fibrillation among people receiving GLP-1-based treatment.
All-cause mortality was 41% lower.
At first glance, those numbers look like a straightforward therapeutic victory.
They should not be read that way.
This was not a randomized trial in which adults with identical clinical profiles were assigned to GLP-1 therapy or metformin and followed prospectively.
It was an observational study built from electronic health records.
That means the reasons one person received semaglutide or tirzepatide while another received metformin may themselves be associated with subsequent health outcomes.
Insurance coverage may differ. Socioeconomic circumstances may differ. Healthcare engagement may differ. Severity of obesity may differ. The ability to tolerate and continue expensive injectable therapy may identify a patient population different from those treated with metformin.
Researchers can match the variables they can see.
They cannot guarantee balance in everything they cannot measure.
This becomes particularly important when the reported difference in mortality reaches 41%.
A large mortality association should invite additional scrutiny rather than automatic confidence because treatment selection, access to healthcare and unmeasured patient characteristics can strongly influence mortality in real-world datasets.
The conference abstract also relies on prescription records rather than direct measurement of medication use.
Being prescribed a drug does not establish that the patient filled every prescription, administered it correctly or remained on treatment at the intended dose.
The researchers additionally grouped several drugs together.
Semaglutide, liraglutide and dulaglutide are GLP-1 receptor agonists, while tirzepatide acts through both GIP and GLP-1 receptors. Their weight-loss efficacy differs considerably, as do approved indications, dose regimens and existing cardiovascular evidence.
Treating them as a single class may be useful for generating a broad clinical signal.
It makes the findings less useful for determining what an individual medicine actually does.
There is another terminology problem worth avoiding.
Secondary coverage has described the cardiovascular findings as a reduction in major adverse cardiovascular events, or MACE.
The AHA release instead reports myocardial infarction, stroke, heart failure and atrial fibrillation as separate outcomes. Heart failure and atrial fibrillation are not conventionally components of the standard three-point MACE composite of cardiovascular death, myocardial infarction and stroke.
Without a peer-reviewed manuscript defining a prespecified composite outcome, the safer interpretation is that several cardiovascular events were individually less frequent in the GLP-1-treated group.
That methodological restraint does not make the results unimportant.
It tells policymakers what evidence is still missing.
There is already randomized evidence that major pharmacological weight loss can dramatically change diabetes progression.
In the long-term SURMOUNT-1 trial, people with obesity and prediabetes were randomly assigned to tirzepatide or placebo and followed for 176 weeks.
Type 2 diabetes developed in 1.3% of tirzepatide-treated participants compared with 13.3% receiving placebo during treatment, alongside substantial sustained weight reduction.
Seventeen weeks after treatment ended, the difference narrowed somewhat but remained pronounced.
That trial demonstrated something the observational analysis cannot: in a randomized setting, treating obesity pharmacologically can substantially delay progression to type 2 diabetes in a high-risk population.
But placebo is not metformin.
That is why the new AHA analysis matters.
Healthcare systems rarely decide between a modern medicine and doing nothing.
They decide between interventions with different prices, evidence bases, treatment burdens and expected benefits.
Metformin remains especially relevant because its diabetes-prevention evidence is unusually mature.
The Diabetes Prevention Program established that intensive lifestyle intervention was highly effective in people at increased diabetes risk and that metformin also reduced progression, with particularly strong benefit in selected higher-risk groups.
Those data have been followed for years rather than months.
The 2026 American Diabetes Association Standards of Care therefore continue to recommend considering metformin for adults at particularly high risk of type 2 diabetes, especially younger adults with severe obesity, higher fasting glucose or HbA1c and those with a history of gestational diabetes.
But the same guidance now reflects how much the prevention landscape has changed.
For adults with overweight or obesity who are at high risk of diabetes, treatment goals should include not only preventing worsening glucose but also achieving and maintaining weight loss and addressing cardiovascular risk.
Pharmacotherapy can be considered to support those goals.
That is a materially broader framework than simply asking which drug lowers glucose most cheaply.
It also exposes why comparing GLP-1 therapy and metformin as if they were interchangeable diabetes drugs can be misleading.
Metformin is primarily being used here as a metabolic prevention medicine.
A modern incretin therapy may simultaneously treat obesity, reduce glycaemic progression and, depending on the molecule and patient population, potentially address other cardiovascular or obesity-related risks.
The interventions may therefore produce overlapping but not identical value.
That creates a reimbursement problem.
Metformin is generic and inexpensive.
GLP-1-based therapies are not.
If health systems expand access to millions of people with obesity and prediabetes, the budgetary consequences could be enormous even if treatment is clinically effective.
Payers will therefore demand something stronger than an association between prescription records and fewer events.
They will need to know which people benefit enough to justify treatment.
Prediabetes is not a uniform risk state.
Some people return to normoglycaemia without medication. Others remain stable for years. Some progress rapidly to type 2 diabetes.
Treating every person with prediabetes identically would therefore produce substantial overtreatment alongside prevention.
Risk stratification becomes crucial.
The ADA already identifies features such as BMI of at least 35 kg/m², higher fasting glucose, HbA1c of at least 6.0% and previous gestational diabetes as markers of particularly high progression risk.
Future incretin-prevention strategies may need to become even more precise.
The most compelling candidate may not simply be “a person with prediabetes”.
It may be a person with obesity, worsening glycaemia, high lifetime cardiovascular risk and additional obesity-related complications for whom substantial sustained weight loss offers several simultaneous benefits.
That would change the economic calculation.
The price of treatment could be weighed against prevention of multiple downstream conditions rather than diabetes alone.
It would also change how outcomes should be measured in trials.
A future active-comparator trial between metformin and GLP-1-based treatment should not stop at HbA1c.
It should examine progression to type 2 diabetes, magnitude and durability of weight loss, cardiovascular and kidney outcomes, adverse effects, treatment discontinuation, quality of life and cost-effectiveness.
Duration will matter too.
Obesity pharmacotherapy is increasingly understood as chronic treatment because stopping effective therapy frequently leads to weight regain and recurrence of metabolic risk.
Diabetes prevention therefore cannot be evaluated solely by asking what happens while a person is receiving a medicine.
Policymakers need to know what happens over years, what occurs after discontinuation and whether preventing diabetes delays disease or permanently changes its trajectory.
The SURMOUNT-1 data already hint at this issue because some glycaemic benefit was lost after treatment stopped.
That is not evidence of treatment failure.
It reflects the chronic biology of obesity.
But it creates a different reimbursement proposition from a short course of preventive treatment.
Access is therefore likely to become the central political question if evidence for earlier GLP-1 use strengthens.
Restricting treatment until type 2 diabetes develops could become clinically difficult to defend for some high-risk patients if an earlier intervention is proven to prevent that progression.
At the same time, paying for costly long-term therapy across enormous prediabetes populations could threaten healthcare budgets and widen inequalities if access is determined primarily by private coverage or ability to pay.
That tension is already visible in obesity medicine.
Prediabetes could make it sharper.
There is also a danger in turning this into an artificial contest between lifestyle, metformin and modern pharmacotherapy.
The interventions do not need to be ideological rivals.
Lifestyle intervention remains a central component of diabetes prevention. Metformin remains inexpensive, extensively studied and appropriate for many high-risk patients. GLP-1-based medicines offer a different level of weight-loss efficacy and may provide additional benefits in selected patients.
The useful question is not which intervention should defeat the others.
It is how to match intensity of treatment to intensity of risk.
The new observational analysis pushes that conversation forward, but it does not settle it.
Its scale makes the findings difficult to ignore. Its design makes them impossible to treat as definitive.
What it provides is a strong reason to test the comparison properly.
Because if randomized evidence eventually shows that modern obesity pharmacotherapy prevents diabetes and cardiovascular disease more effectively than established prevention at an acceptable cost, the argument for waiting until type 2 diabetes develops will become increasingly difficult.
And that would represent a deeper change than simply adding another drug to a guideline.
It would mean treating metabolic disease before the diagnostic threshold becomes the trigger for action.
GLP-1 therapies are moving upstream. The next policy fight will be over how far prevention should move with them, and who gets access before diabetes arrives.

