Diabetes Prevention Lasts 22 Years: Lifestyle Beats Metformin Long-Term
The landmark DPP trial shows lifestyle intervention delays diabetes onset by 3.5 years, metformin by 2.5 years, over two decades of follow-up.
Summary
The Diabetes Prevention Program tracked 3,195 adults with prediabetes for 22 years, comparing intensive lifestyle intervention (ILS), metformin, and placebo. Intensive lifestyle reduced diabetes incidence by 24% and metformin by 17% over the full period, translating to median delays in diabetes onset of 3.5 and 2.5 years respectively. The bulk of benefit stemmed from large early reductions during the original 3-year trial. Crucially, effects were heterogeneous: lifestyle worked best in those at highest baseline risk, while metformin showed no benefit—and possibly harm—in older participants. These findings support precision-targeting of preventive interventions toward highest-risk individuals.
Detailed Summary
Type 2 diabetes affects hundreds of millions globally, yet most prevention trials only follow participants for a few years. The Diabetes Prevention Program and its follow-up study (DPPOS) now offer one of the longest randomized intervention datasets in medical history, spanning 22 years and offering rare insight into whether early prevention benefits endure.
The original DPP enrolled 3,234 adults with prediabetes (elevated fasting and post-load glucose, BMI ≥24 kg/m²) at 27 US centers, randomly assigning them to placebo, metformin (850 mg twice daily), or intensive lifestyle intervention (ILS) targeting ≥7% weight loss and ≥150 min/week physical activity. After the 3-year trial ended in 2001, the DPPOS continued with placebo discontinued, open-label metformin maintained, and group lifestyle classes offered to all participants. The current analysis includes 3,195 participants followed through February 2020.
Over 22 years, cumulative diabetes incidence reached approximately 70%, 64%, and 66% in the placebo, metformin, and ILS groups respectively. Compared to placebo, ILS reduced diabetes incidence by 24% (HR=0.76; rate difference −1.59 cases/100 person-years), and metformin by 17% (HR=0.83; RD −1.17). These translate to median delays in diabetes-free survival of 3.5 years for ILS and 2.5 years for metformin. Importantly, almost all of this benefit was generated during the initial 3-year DPP phase; after unmasking and protocol changes, rates in the three groups largely converged, though cumulative separation was preserved.
A key finding is the heterogeneity of treatment effects. Participants with higher baseline fasting glucose, HbA1c, or composite multivariable risk indices derived greater absolute benefit from ILS. Metformin showed a clear age interaction: younger participants benefited substantially, while those aged ≥60 at baseline showed no benefit and slightly worse outcomes with metformin. Older participants did respond to ILS with an 8-year median delay in diabetes. Sex, BMI, and race/ethnicity subgroups showed varying but generally consistent patterns.
The findings carry significant implications for precision prevention. Rather than applying uniform interventions to all individuals with prediabetes, clinicians may achieve greater population-level impact by directing intensive lifestyle programs toward those at highest metabolic risk, and reserving metformin primarily for younger high-risk individuals. The diminishing returns over time also suggest that maintaining intervention intensity is critical for sustaining long-term benefit.
Key Findings
- Intensive lifestyle intervention delayed median diabetes onset by 3.5 years over 22 years of follow-up.
- Metformin delayed median diabetes onset by 2.5 years, driven largely by the initial 3-year trial phase.
- Nearly all long-term benefit originated from large early effects during the original DPP randomized phase.
- ILS benefit was greatest in participants with highest baseline fasting glucose, HbA1c, and composite risk scores.
- Metformin showed no benefit—and a slight negative trend—in participants aged 60 or older at baseline.
Methodology
This is a 22-year intention-to-treat analysis of the DPP randomized clinical trial (n=3,195) comparing placebo, metformin, and intensive lifestyle intervention. The DPPOS continuation phase maintained metformin and offered group lifestyle classes to all arms after unmasking in 2001. Statistical methods included Kaplan-Meier survival analysis, Cox proportional hazards models, restricted mean survival time estimation, and subgroup interaction testing using Wald tests and treatment-covariate product terms.
Study Limitations
After the DPP trial ended, placebo was discontinued and all groups received group lifestyle education, limiting long-term treatment contrast and making post-DPP effects difficult to attribute to original assignments. HbA1c was not used as a primary eligibility or outcome criterion, limiting direct clinical translation to current ADA diagnostic standards. Subgroup analyses, while pre-specified, involved multiple comparisons and cross-sectional risk indices that may not fully capture dynamic changes in individual risk over time.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.