Metformin and Diabetes Prevention: The Long View

The Diabetes Prevention Program followed people long enough to ask several questions. Could metformin delay a diagnosis of type 2 diabetes? Would that difference remain years later? Would fewer diagnoses be followed by fewer complications or longer lives?

The answers came in separate papers. Reading them together gives a clearer account of prevention than any one headline.

The original trial, reported in 2002, enrolled 3,234 adults who had elevated blood glucose and were at high risk of developing diabetes. Their average age was about 51. They were assigned to metformin, a placebo, or an intensive program focused on diet, activity and weight loss.

The program involved individual sessions and continued support. The medication groups received standard lifestyle information. These were three specific approaches delivered within a monitored study.

After an average of 2.8 years, both active approaches had reduced new diabetes diagnoses, with the larger reduction in the lifestyle group. The estimated proportion diagnosed by three years was about 29 in every 100 people assigned placebo, 22 in every 100 assigned metformin, and 14 in every 100 assigned lifestyle support.

Those figures describe diagnoses during a defined period. They leave room for a later question: how many people would develop diabetes with more time?

The 2015 follow-up included 2,776 participants from the original trial. By year 15, the estimated proportions who had developed diabetes were 62% in the original placebo group, 56% in the metformin group, and 55% in the lifestyle group.

Differences remained, although they were smaller than in the early trial. Most of the separation had developed during the first few years. Many people in every group eventually received a diagnosis.

Care also changed. After the initial trial, all groups were offered lifestyle sessions. Metformin continued openly in its original group, and some people in the other groups later received it through their own clinicians. The follow-up compared people according to their original assignment as their care evolved.

The researchers also assessed a combined measure of kidney, eye and nerve complications. Across the whole cohort, there was no statistically clear difference between the three assigned groups. The report found a benefit of lifestyle support for this measure in women, in a planned analysis. That finding belongs with its specified group and outcome.

A further report published in 2025 examined diabetes incidence through February 2020, covering about 21 years after the trial began. The early separation between groups was still visible. It estimated an average of 1.2 additional diabetes-free years in the metformin group and 2.0 in the lifestyle group compared with placebo, within the follow-up window.

This gives the result another useful unit: time before diagnosis. It also keeps the publication date separate from the period when participants were assessed.

A 2021 analysis examined deaths over a median of 21 years. It found no statistically clear reduction in death from any cause with either metformin or lifestyle assignment compared with placebo.

For metformin, the hazard ratio was 0.99, with a 95% confidence interval from 0.79 to 1.25. The estimate was close to equal rates, while the interval still allowed a benefit or harm. This was a secondary analysis of a study originally designed around diabetes prevention.

The participants had also met entry criteria that excluded several serious illnesses. Their results describe this selected population and its subsequent care. They cannot settle every question about metformin in people who already have diabetes or other medical conditions.

A delayed diagnosis, a measured complication and survival each describe a different part of health. The DPP papers give each its own follow-up period and result.

The early reduction in diabetes diagnoses remained visible in the longer record. The later reports added detail about how that record unfolded. Keeping the outcome, the time frame and the care people received together makes the evidence easier to understand.