Picture a 51-year-old walking into a clinic in 1998. Prediabetes. Slightly overweight. The kind of patient who sits across from you every single day. She gets randomized into a study, spends about three years in an intensive lifestyle program (individual counseling sessions at first, then monthly follow-ups) loses some weight, picks up a walking routine. But by year six, her weight and activity levels look a lot like everyone else’s in the trial. By her mid-70s, you’d assume whatever happened in those early years has long since washed out.
The data say otherwise, though the explanation may be more complicated than simple behavior change.
That’s the quietly striking takeaway from the newest analysis of the Diabetes Prevention Program, published in JAMA this June. Salive and colleagues followed 1,173 of the original DPP participants for up to 21 years, linking their Medicare claims all the way through 2021, and asked a bigger question than the trial was originally built to answer. Not “did we prevent diabetes?” but “did we prevent the pileup, the two, three, five chronic conditions that define what aging actually looks like for most Americans?”
The Headline You’ll Want to Sit With
Adults randomized to the intensive lifestyle intervention had a 21% lower risk of developing multimorbidity, two or more chronic conditions, compared with placebo. Push the threshold to three or more conditions and the gap widens: a 25% lower risk. And for the costliest, most burdensome disease combinations, the dyads that drive hospitalizations and drain healthcare budgets, lifestyle was tied to a 43% lower risk.
Here’s the part worth reading twice: the intensive lifestyle intervention lasted about three years, with 16 individual sessions in the first six months followed by monthly maintenance visits. After that, only limited group sessions were offered, and attendance was low. Weight-loss and physical activity differences between the lifestyle and other groups had largely converged by year six. Yet two decades later, the statistical fingerprint of that early randomization was still visible in the data.
What drove that durability is an open question. It may reflect lasting behavioral shifts in some participants, early metabolic or physiological changes that altered disease trajectories, or some combination of both. The study was not designed to untangle the mechanism, but the association itself, persisting across multiple sensitivity analyses, is notable.
And Then There’s Metformin
This is where it gets interesting for those of us who think about pharmacotherapy. Metformin, the medication that did prevent and delay diabetes in the original DPP, showed no statistically significant reduction in multimorbidity. Not for two conditions. Not for three. Not for the high-cost dyads.
That contrast deserves careful interpretation rather than a simple verdict. The study may have been underpowered to detect a modest metformin effect, only about a third of the original cohort was included, and metformin adherence declined over time.
Still, the difference between the two interventions is suggestive. Lifestyle modification acts on multiple risk factors simultaneously, weight, fitness, blood pressure, lipids, insulin sensitivity, while metformin’s primary mechanism targets hepatic glucose production. When the outcome broadens from a single disease to the full accumulation of chronic conditions, an intervention that touches many pathways may have an inherent advantage. The study’s authors note this pleiotropic quality, though they appropriately stop short of claiming it as proven mechanism.
The lifestyle benefit even held up when diabetes was removed entirely from the multimorbidity definition, meaning the intervention wasn’t just delaying diabetes and calling it a day. It was associated with a lower trajectory of chronic disease through pathways that reach beyond glucose.
Before You Frame This on the Wall
Credit where it’s due: this is the longest follow-up of any lifestyle intervention studied for multimorbidity, outdistancing FINGER (2 years) and Look AHEAD (8 years) by a wide margin. But it earns its caveats, and honest reading demands we name them.
This was not a trial designed with multimorbidity as its primary endpoint. It was an observational follow-up conducted long after the original randomized diabetes prevention trial ended. The original randomization provides stronger causal inference than a typical observational study, but it’s not the same as a prospective multimorbidity prevention trial. Only about a third of the original 3,234 participants made it into this analysis, the people who survived, stayed enrolled, reached Medicare age, and consented to data linkage. That group skewed slightly older, leaner, more educated, and higher-income than those left out, which is the textbook setup for survivorship and selection bias.
The claims-based method of counting conditions has its own well-known quirks, including potential upcoding in Medicare Advantage encounter data (though the authors took steps to mitigate this). And the crossover inherent in the study design, all groups eventually received some lifestyle counseling, placebo was discontinued, metformin adherence waned, means the true contrast between groups was diluted over time. If anything, that dilution makes the persistent lifestyle signal more remarkable, but it also makes precise effect estimation harder.
None of this erases the finding, but it means we hold it as strong, directional evidence rather than definitive proof.
What This Means in Practice
Because the counseling that shaped these trajectories, dietary modification, 150 minutes of weekly movement, a 7% weight-loss target, is structured and repeatable, it fits well into team-based care models where pharmacists, nurses, dietitians, and health coaches reinforce the same message. This study hands clinicians a genuinely compelling data point for the patient who wants medication alone: metformin is worth considering for diabetes prevention, but the early investment in behavior change appears to carry benefits that extend far beyond glucose, and far beyond the years in which the behaviors were most actively supported.
That’s not a lecture. That’s a 21-year dataset suggesting that the structured, early work, the coaching, the follow-up, the behavior support we sometimes undervalue because it doesn’t come in a bottle, may set in motion a health trajectory that a pill alone does not replicate. The mechanism isn’t fully clear. But the association, as the study’s authors put it, suggests that intensive lifestyle modification “may prevent or delay multimorbidity in middle and older age among adults with high risk of diabetes.”
Sources & Further Reading: The primary study — Salive ME, Tjaden AH, Ames JR, et al. “Lifestyle and Metformin Interventions and Risk of Multimorbidity in Adults With Prediabetes.” JAMA. Published online June 15, 2026 (doi:10.1001/jama.2026.8492) — anchors this piece. Supplementary context and effect-size confirmation drawn from the NIH/NIA press release, the American College of Cardiology Journal Scan summary, and Pennington Biomedical Research Center’s coverage. For screening background, see the USPSTF evidence report on prediabetes and type 2 diabetes (JAMA. 2021;326(8):744-760).
