Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Apr 28.
Published in final edited form as: Lancet Diabetes Endocrinol. 2025 Apr 28;13(6):469–481. doi: 10.1016/S2213-8587(25)00022-1

Long-term effects and effect heterogeneity of lifestyle and metformin interventions on type 2 diabetes incidence over 21 years in the US Diabetes Prevention Program randomised clinical trial

William C Knowler 1, Lindsay Doherty 1, Sharon L Edelstein 1, Peter H Bennett 1, Dana Dabelea 1, Mary Hoskin 1, Steven E Kahn 1, Rita R Kalyani 1, Catherine Kim 1, F Xavier Pi-Sunyer 1, Sridharan Raghavan 1, Vallabh O Shah 1, Marinella Temprosa 1, Elizabeth M Venditti 1, David M Nathan 1
PMCID: PMC12414453  NIHMSID: NIHMS2081085  PMID: 40311647

SUMMARY

Background

Lifestyle and metformin interventions reduce diabetes incidence among many adults with prediabetes. We now assess heterogeneity of their long-term effects.

Methods

In the Diabetes Prevention Program (DPP), a three-year randomized clinical trial in 3,234 adults with prediabetes, diabetes incidence was reduced by 58% (hazard ratio [HR]=0∙42, rate difference [RD]= −6∙2 cases/100 person-years) with intensive lifestyle (ILS) intervention and by 31% (HR=0∙69, RD= −3∙2) with metformin compared with placebo. Subsequently, the trial was continued with protocol modifications as the DPP Outcomes Study (DDPOS). Placebo was discontinued, metformin continued, and all participants offered group-based ILS. We now assess long-term persistence and heterogeneity of intervention effects on diabetes incidence. Follow-up is reported for the combined study from 31 July 1996 to 23 February 2020. The trial is registered: NCT00004992 (DPP), NCT00038727 (DPPOS).

Findings

During 22 years of follow-up, compared with placebo, diabetes incidence was reduced in the ILS group by 24% (HR=0∙76, 95%CI=0.68, 0.85, RD=−1·59 [−2·25, −0·93] cases/100 person-years), and in the metformin group by 17% (HR=0∙83 [0∙74, 0∙93], RD= −1·17 [−1·85, −0·49]), with corresponding increases in median diabetes-free survival of 3∙5 and 2∙5 years/person, and mean diabetes-free survival of 2∙0 [1∙2, 2∙8] and 1∙2 [0∙4, 2∙0] years/person, respectively. The overall treatment effects resulted from the large early effects during DPP. Absolute intervention effects were greater with ILS in subgroups with higher baseline fasting glucose, HbA1c, and multivariable risk indices, and with metformin in younger participants.

Interpretation

The large initial intervention effects resulted in sustained reductions in cumulative diabetes incidence for 22 years. Intervention effects were heterogeneous, with greater benefit in subgroups at highest risk, and with little or no effect in those with low risk of diabetes. These findings should guide precision interventions to stem the current diabetes epidemic.

INTRODUCTION

In randomized clinical trials of adults at high risk of developing type 2 diabetes, behavioral and medication interventions have delayed or prevented diabetes onset for a few years, but evidence for longer-term effects is sparse.1–5 In the three-year Diabetes Prevention Program (DPP), compared with placebo, diabetes incidence was reduced by 58% (HR=0∙42) in the group randomly assigned to intensive lifestyle (ILS) weight-loss intervention with an absolute rate difference (RD) of −6∙2 cases/100 person-years and by 31% (HR=0∙69) in the metformin group with RD of −3∙2.1

After the DPP ended in 2001, outcome assessments and modified interventions continued in the DPP Outcomes Study (DPPOS) to determine if the original effects on diabetes incidence persisted and to identify heterogeneity in responses to interventions. Medication treatment assignment was unmasked, open-label metformin was continued in the original metformin group, in the ILS group the intervention decreased in intensity, and participants in all three intervention groups were offered a less intensive lifestyle program.6 We report results of approximately 22 years of follow-up from randomization until February 2020.

METHODS

Study design

Detailed methods, including recruitment, eligibility, randomization methods, outcome assessments, and adverse events have been published for DPP1,7 and DPPOS.6 The closing date for this report is 23 February 2020; subsequent outcome assessments were interrupted by the COVID-19 pandemic during which assessment procedures were modified. The institutional review board at each center approved the protocols that are available at https://dppos.bsc.gwu.edu/web/dppos/dpp and https://dppos.bsc.gwu.edu/web/dppos/dppos. All participants gave written informed consent.

Participants

Major DPP eligibility criteria included fasting glucose 95–125 mg/dl (5∙3–6∙9 mmol/l; or ≤125 mg/dl in the American Indian centers), two-hour post-load glucose 140–199 mg/dl (7∙8–11∙0 mmol/l), body mass index (BMI) ≥24 kg/m2 (≥22 kg/m2 among Asian-American participants), and age ≥25 years.1,7 Sex and race/ethnicity were ascertained by interviewer-administered questionnaire, with male or female as the possible responses for sex and white, African American, Hispanic, American Indian, and Asian as the possible responses for race/ethnicity. Random assignment resulted in similar baseline characteristics across the three treatment groups.1 Of the 3150 eligible surviving DPP participants, 88% enrolled in DPPOS with similar baseline characteristics in each treatment group (88% of each sex and 87 to 88% of each treatment group enrolled).6

Interventions

DPP participants were randomly assigned to one of three intervention groups: placebo, metformin, or ILS.1 Metformin, 850 mg twice per day as tolerated, and matching placebo tablets were masked to participants and investigators. The unmasked ILS intervention was individually administered and designed to produce ≥7% body weight loss through diet and moderately intensive physical activity ≥150 minutes per week. During DPPOS, placebo was discontinued, unmasked metformin, unless contraindicated, was provided to the original metformin group while the participant remained without diabetes.6 Until 2014, twice-annual booster-intervention classes were offered to the original ILS group and group-implemented quarterly lifestyle classes were offered to all participants.

Outcomes and assessments

The primary outcome during DPP and DPPOS was diabetes incidence defined by American Diabetes Association criteria with a semi-annual fasting plasma glucose ≥126 mg/dl (7∙0 mmol/l) or 2-hour glucose ≥200 mg/dl (11∙1 mmol/l) in an annual 75-gm oral glucose tolerance test, confirmed within six weeks.1,7 HbA1c was not used as an eligibility or outcome criterion, except in secondary analyses.8 Exposure to metformin, administered through the study and prescribed outside the study, was calculated from semi-annual questionnaires.

Statistical methods

We report intention-to-treat analyses for the entire period of observations. The treatment groups are referred to by their original DPP assignments as the placebo, metformin, and ILS groups, regardless of subsequent treatment and protocol changes. The metformin and ILS groups are compared with the placebo group, in all participants and in subsets defined by baseline characteristics.

Time to diabetes compared metformin and ILS with placebo on a modified product-limit life table distribution with a log-rank test statistic. Follow-up was censored at the participant’s last diabetes assessment if diabetes had not developed. Incidence rates, as new cases per 100 person-years at risk, were computed over the entire study period from randomization and in 2-year periods, i.e., from randomization to the second annual examination, the second to the fourth annual examination, etc. Rate differences (RD) on an absolute scale, between the metformin and ILS groups and the placebo group, were expressed in cases per 100 person-years based on treatment-specific crude rates calculated as the number of diabetes events divided by the total number of person-years of follow-up using a Poisson model under the assumption that the incidence is constant over time and applies to all participants in the study population. Proportional hazards regression models were used to estimate hazard ratios and assess heterogeneity. Since hazard ratios were not always constant (“proportional”) over time, they represent averages over the entire follow-up time. The statistical significance of treatment group (vs. placebo) by subgroup interactions were computed for rate differences using a composite Wald test and for hazard ratios by including treatment by subgroup product terms in the proportional hazards models.

Cumulative incidence of diabetes was computed from Kaplan-Meier curves with half-year increments. The point at which each curve crossed 50% defined the median diabetes-free time, i.e., the time at which half the participants had developed diabetes. In the few subgroups for which the curve did not reach 50%, the median time was noted as >22 years. Treatment-group differences in restricted mean diabetes-free survival times were also estimated.9 Based on the cumulative incidence curve, this statistic estimates the mean diabetes-free time per person from randomization until diagnosis of diabetes or censoring. The “restricted” time frame of this analysis was 23 years, the maximum follow-up time to diabetes or censoring. Mean survival times were also computed at each 6-month follow-up period, for the diabetes-free survival time from randomization to that time point. They were expressed as absolute values in each intervention group, as a percentage of total follow-up time to that point, and as differences and ratios of the metformin and ILS groups compared with the placebo group.

To examine heterogeneity, we stratified by a priori defined subgroups based on baseline diabetes risk factors previously identified in the DPP: age, sex, race/ethnicity, BMI, waist circumference, fasting and 2-hour post-load plasma glucose, HbA1c, estimated insulin sensitivity and response to oral glucose, self-reported histories of anti-depression medication and of gestational diabetes mellitus.1,8,10–12 We assessed treatment interactions (effect modification) by subgroups of individual baseline risk factors. Assessing interactions with groups of two or more risk factors was impractical because cross-classification resulted in cells with too few participants for precise estimation. Therefore, we computed two multivariable indices that predict diabetes using data from all treatment groups13. A “clinical” multivariable risk index was derived from routinely available demographic and clinical variables. A “physiologic” risk index included the variables in the clinical model plus insulin sensitivity index and the 30-minute delta insulin to delta glucose ratio.

The funding agencies were represented in the steering and executive committees and contributed to the study design and implementation. They had no role in data analysis, data interpretation, or writing this report.

RESULTS

Randomization occurred from 31 July 1996 to 28 May 1999. Follow-up is now reported to 23 February 2020. The study period averaged 22 years: 3.2 years in DPP6 plus 18.6 years from the end of DPP (31 July 2001 to 23 February 2020). The number of participants randomized and the 3,195 included in the present analyses are shown in the study flowchart (Figure 1). At baseline mean age was 51 years, 32% were men, and 45% were self-identified members of U.S. minority groups.1 Mean BMI was 34∙0 kg/m2. All had prediabetes by American Diabetes Association criteria.7 Other characteristics are shown in Appendix p. 2.

Figure 1.

Figure 1.

Open in a new tab

Study flowchart. The 39 participants “with no diabetes follow-up” had no examinations after randomization or had an examination at which glucose concentrations were not measured. All 3,195 participants with any amount of follow-up for diabetes were included in the analyses.

Mean weight loss by treatment group and over time among participants who had not yet developed diabetes are shown in Appendix p. 3A. During the DPP, weight loss was greatest in the ILS group, but the ILS and metformin groups had similar mean weights until 17 years after randomization, after which weights were lower in the metformin group. Mean weight was highest in the placebo group at all times after randomization. Metformin was rarely used in the placebo or ILS groups prior to diabetes diagnosis. The fraction of participants taking metformin declined to <60% over time in the metformin group and remained very low in the other groups before diabetes diagnosis (Appendix p. 3B). From randomization until diagnosis of diabetes or censoring, the average fractions of participants taking metformin (administered by the study or prescribed outside) were 7∙7, 3·1, and 68·4% in the placebo, ILS, and metformin groups, respectively.

Cumulative diabetes incidence rates are shown by randomized treatment group (Figure 2A); details are shown in Appendix p. 4–5. At 21 years after randomization, the cumulative incidence rates (with 95% confidence intervals) were 70% (67, 73%), 64% (61, 68%), and 66% (62, 69%) in the placebo, metformin, and lifestyle groups, respectively. Figure 2B shows incidence rates in cases/100 person-years in two-year intervals after randomization. During the first two years, incidence rates were markedly reduced by metformin and ILS; subsequently, however, rates in the placebo and metformin groups fell to approximately equal those in the ILS group. The incidence rates continued falling during follow-up, especially in the metformin group. Over the entire 22 years, compared with placebo, ILS reduced diabetes incidence by 24% (HR=0∙76 [95%CI=0∙68, 0∙85], RD= −1∙59 [−2∙25, −0∙93] cases/100 person-years) (Table 1). Metformin reduced it by 17% (HR=0∙83 [0∙74, 0∙93], RD= −1∙17 [−1∙85. −0∙49]). Appendix p. 6 shows the results as rate differences and ratios for the metformin and ILS intervention groups compared with placebo over the entire study’s time course.

Figure 2.

Figure 2.

Open in a new tab

A. Cumulative incidence of diabetes throughout DPP and DPPOS by treatment group in all participants. The times from randomization at which the cumulative incidence rates reach 50% are shown with arrows for each treatment group (P=placebo, M=metformin, L=intensive lifestyle). Numbers below the horizontal axis are the numbers of participants at risk at 0, 4, 8, 12, 16, and 20 years from randomization. Further details are in the Appendix p. 4–5. Potential follow-up time in the study averaged 22 years. Individuals’ follow-up times were shorter for this analysis because they stopped at diabetes diagnosis or when censored for other causes. Individual follow-up times ranged from 0∙2 to 23∙2 years; the mean was 10∙0 years, median 8∙0 years, and interquartile range 3∙0 to 18∙0 years. Over the entire follow-up, for metformin vs. placebo, HR=0∙83, 95% confidence interval=0∙74, 0∙93. For Lifestyle vs. placebo, HR=0∙76 (0∙68, 0∙85).

B. Incidence rates of diabetes in cases/100 person-years by treatment group in 2-year time periods.

Table 1.

Long-term treatment results on diabetes incidence and diabetes-free survival time.

PLAC (n=1066) MET (n=1061) ILS (n=1068)
Incidence rate (cases/100 p-yr) 6∙69 5∙52 5∙10
Rate difference (cases/100 p-yr) *  -1∙17 (−1∙85, −0∙49)  -1∙59 (−2∙25, −0∙93)
Hazard ratio * 0.83 (0.74, 0.93) 0.76 (0.68, 0.85)
Rate reduction (%) * 17 (7, 26) 24 (15, 32)
Median diabetes-free years 9∙5 12∙0 13∙0
Mean diabetes-free yearsǂ 11∙6 (11·0, 12·1) 12∙8 (11·5, 14·2) 13∙6 (12·3, 14·9)
Median delay (years) * 2∙5 3∙5
Mean delay (years)ǂ * 1∙2 (0∙4, 2∙0) 2∙0 (1∙2, 2∙8)
Open in a new tab

Note: the incidence rates are averages over the 22-year follow-up time. They were not uniform over time as shown in Figure 2.

PLAC = placebo group, MET = metformin group, ILS = intensive lifestyle group.

p-yr = person-years at risk.

Statistics in parentheses are 95% confidence intervals.

*

reference group for the metformin and ILS comparisons.

ǂ

from the restricted mean survival time (RMST).

The cumulative incidence curves separated early, especially in the first three years, with lower rates in the metformin and ILS groups (Figure 2A). The metformin and ILS curves progressively converged with longer follow-up. Median diabetes-free times were 9∙5, 12∙0, and 13∙0 years in the placebo, metformin, and ILS groups, respectively, and compared with placebo, metformin and ILS led to delays of 2∙5 and 3∙5 years. The treatment-group differences in mean diabetes-free survival times were somewhat less than the medians: 1∙2 and 2∙0 years for metformin and ILS (Table 1). The evolution of treatment benefits can also be quantified from the standpoint of extra diabetes-free survival time attributable to each intervention. The absolute diabetes-free time increased with increasing follow-up time but declined as a percentage of follow-up time (Appendix p. 7, panels A and C). As a percentage of time in the study, the maximal intervention benefit of ILS occurred at about six years after randomization, and the benefit continued to increase gradually for metformin after six years (Appendix p. 7, panel D).

Treatment effects differed by some baseline variables, as shown for age in Figure 3. Compared with younger participants, participants ≥60 years old at baseline had lower incidence rates reflected in longer median and mean diabetes-free survival times in all treatment groups. The older group had a greater median diabetes-free survival time (8 years) with ILS but no benefit from metformin (minus 3 years). Treatment heterogeneity was also examined by other baseline risk factors, including the multivariable risk indicators described in Appendix p. 9–10. Subgroups with higher fasting glucose, BMI, or multivariable clinical risk indices had higher diabetes incidence (Appendix p. 11), with the cumulative incidence curves shifted to the left and shorter median and mean diabetes-free times. Among high-risk groups, the curves separated by treatment group earlier than in lower-risk groups. In the highest tertile of the physiological risk index, mean diabetes-free survival was increased by 1∙0 years with metformin and 2∙8 years with ILS compared with 0∙9 and 1∙3 years, respectively, in the lowest tertile (Appendix p. 12–14). ILS had a greater RD in the highest tertile of the physiological risk index (−5∙77 cases/100 person-years) than in the lowest tertile (RD= −0∙56) with interaction p<0∙001.

Figure 3.

Figure 3.

Open in a new tab

Cumulative incidence of diabetes throughout DPP and DPPOS by treatment group and baseline age. The median diabetes-free survival times, i.e., times at which the cumulative incidence curves reach 50%, are shown with arrows for each treatment group (P=placebo, M=metformin, L=intensive lifestyle). Numbers below the horizontal axis are the numbers of participants at risk at 0, 4, 8, 12, 16, and 20 years from randomization.

Treatment effects in subgroups also varied with follow-up time (Appendix p. 15). In the first two years, diabetes incidence rates were reduced by both ILS and metformin in younger participants (<45 years at baseline), and by ILS, but not metformin, in those ≥60 years old. Although neither intervention had substantial effects within any age group in later follow-up, the reduction in diabetes incidence was much greater with ILS over the full period in participants ≥60 years old at baseline (Figure 3). Incidence rates were higher in those with baseline fasting glucose 110 to 125 mg/dl than in those with lower glucose concentrations, but the convergences of incidence rates with time were similar regardless of baseline fasting glucose (Appendix p. 15). The clinical and physiological risk indices (Appendix p. 9–10) were derived within the study and accordingly were more strongly associated with diabetes incidence than were individual risk factors (Figure 4). Again, their strong early effects diminished during follow-up (Appendix p. 15).

Figure 4.

Figure 4.

Open in a new tab

Diabetes incidence throughout DPP and DPPOS in the Metformin and ILS groups relative to the Placebo group. Panel A: incidence rates in the Placebo group in all participants and in subgroups defined by baseline variables. Panel B: incidence rate differences (RD) for the lifestyle and metformin groups relative to the Placebo group. Panel C: hazard ratios (HR) for the lifestyle and metformin groups relative to the Placebo group. Filled squares are the point estimates and the lines are 95% confidence intervals. Boxes indicate statistically significant subgroup by treatment interactions with p-interaction < 0∙05.

Incidence rates, RDs, and HRs among subgroups with metformin and ILS relative to placebo are shown in Figure 4 and Appendix p. 12–14. There were no significant treatment by subgroup interactions throughout follow-up among subgroups defined by sex, race/ethnicity, or baseline BMI despite these variables significantly predicting diabetes incidence (Appendix p. 10). ILS effects were uniform (no significant heterogeneity) on the relative (HR) scale for all individual variables except baseline HbA1c, where the effect was greater at higher values (Figure 4). By contrast, there was marked treatment heterogeneity on the absolute (RD) scale for ILS with baseline fasting glucose, HbA1c, and the clinical and physiological risk indices, with much greater RDs in the subgroups at highest risk. Metformin effects were uniform on the HR scale for all the individual variables but heterogeneous by age on the absolute (RD) scale with diabetes incidence reduced in the younger (−2∙21 cases/100 person-years) subgroup but not (+0∙41 cases/100 person-years) in older participants (p-heterogeneity = 0∙04).

DISCUSSION

The primary aim of diabetes prevention interventions is to delay or prevent the onset of diabetes, thereby increasing the time that individuals remain free of the disease. Although interventions can clearly reduce diabetes incidence over a period of a few years,1–5 it is not well known how long their effects persist, especially when interventions cease or become less intensive. In the present study, compared with placebo, ILS and metformin substantially reduced annual diabetes incidence rates during the initial few years but not subsequently (Figure 2B). Thus, the continued separation of the ILS and metformin cumulative incidence curves from the placebo curve during DPPOS (Figure 2A) depended on the large initial intervention effects and the absence of a rebound during long-term follow-up. Similarly, the Da Qing clinical trial in Chinese adults with impaired glucose tolerance had a large separation in diabetes cumulative incidence between the lifestyle intervention and comparison groups during the six-year active intervention.2 During the next 24 years, annual diabetes incidence rates were lower than in the first six years, but similar between the treatment groups, again resulting in maintained separation of the cumulative incidence rates.14

Once the treatment effects in DPP were established, ethical considerations required protocol changes including unmasking the medicines, stopping the placebo, and offering lifestyle coaching to all participants.6 Administration of metformin was continued in the original metformin group. The large persistent separation in metformin exposure between the metformin and other groups provided an excellent test of the long-term effects of metformin administration on diabetes prevention. Interpretation of the long-term effects of ILS is less clear, given the less intensive group lifestyle program offered to all DPPOS participants. The intensive intervention followed by less intensive maintenance was consistent with common practice in weight-loss interventions. The US Preventive Services Task Force recommended behavioral weight-loss interventions for preventing obesity-related morbidity and mortality based on a review of interventions, most of which lasted 1 to 2 years.15

Among all participants and in subgroups with the highest diabetes incidence, incidence rates declined during follow-up. As previously shown, the overall decline was not explained by the lifestyle coaching offered to all participants but was likely due, in part, to “exhaustion” of the population most susceptible to diabetes who developed it earlier in the study.16 Weight loss was a major contributor to the reduction in diabetes incidence in the ILS and metformin groups.17, 18 Appendix p. 3 shows continued weight loss with longer follow-up. Some of the apparent weight loss in each group may be due to removal of persons who lost less weight (because they would have had greater diabetes incidence). Regardless of the causes, the lower weights of participants who remained at risk of diabetes might account for some of the decrease in incidence rates compared with those seen earlier in the trial.

The heterogeneity of intervention effects according to some baseline variables should inform and help refine future prevention efforts. Metformin was more effective on the absolute scale in those who were younger, had higher fasting glucose, had a history of gestational diabetes, or were at greater risk according to the multivariable risk indices, although the test for treatment heterogeneity with these subgroups was significant only for age (p<0∙05). By contrast, on the relative (HR) scale, the ILS effects were homogeneous across all subgroups except for baseline HbA1c. The ILS effects were heterogeneous on the absolute scale for baseline fasting glucose, HbA1c, and the two risk indices, with greater effects in the higher risk subgroups. By design, the multivariable risk indices had the strongest associations with diabetes incidence, and participants in the highest risk groups derived the greatest benefit by either absolute or relative diabetes reduction. These multivariable risk indices were derived to enable evaluating treatment interactions with more than one risk factor simultaneously. They were not intended or validated for clinical use.

Despite several baseline factors modifying the absolute treatment effects (RDs), the relative treatment effects (HRs) were mostly homogeneous. The HR may be more relevant to causal inferences, but the RDs and diabetes-free survival times describe the overall impact of the intervention. Interpretation of the HR is straightforward when it is constant over time. In DPP/DPPOS this is not the case and the HR can be interpreted as an average over time. Intervention effects on mean diabetes-free survival time also vary with time since randomization, as illustrated on Appendix p. 7. As a percentage of follow-up time, the extra diabetes-free survival time peaked at about six years for ILS and plateaued after six years in the metformin group. Treatment effects on median diabetes-free times are generally greater than effects on mean times (Table 1, Appendix p. 12–14). The medians are based only on events in a group until the cumulative incidence reaches 50% but are unaffected by subsequent events. By contrast, the mean time is computed from the entire cumulative incidence curve. Therefore, the treatment-group differences in mean diabetes-free time represent average treatment effects of each intervention throughout follow-up, whereas the differences in median diabetes-free time represent only earlier treatment effects. This is consistent with differences in treatment effects being strong early in DPP but essentially disappearing later (Figure 2, Appendix p. 6).

How do these results guide the prevention of diabetes in persons with prediabetes? Although they may apply only to persons meeting the DPP eligibility criteria of BMI and fasting and 2-hour post-load glucose, DPP participants represent a substantial fraction of the US adult population. Based on US data from 2007–2016, an estimated 10∙7% of non-diabetic adults aged 20–64 years and 30∙2% of those aged ≥65 years would meet the DPP eligibility criteria.19 Whether the interventions would be similarly effective in persons with prediabetes defined more broadly is unknown.20 In Japan, a lifestyle intervention for diabetes prevention was effective in persons with impaired fasting glucose only if they had an additional risk factor such as elevated HbA1c or impaired glucose tolerance.21 In two more recent diabetes prevention clinical trials, interventions were less effective22 or equally effective23 in persons with isolated impaired fasting glucose compared with impaired glucose tolerance with or without impaired fasting glucose. A meta-analysis of lifestyle interventions in prediabetes suggested they were effective in persons with impaired glucose tolerance with or without impaired fasting glucose, but not in those with isolated impaired fasting glucose.24 Another limitation to generalizability is that diabetes was assessed systematically and often with an oral glucose tolerance test in the DPP/DPPOS and other clinical trials, whereas outside of research settings, diabetes assessment may be less frequent or accurate.

During the DPP, ILS was much more effective than metformin, but in the long term, there was little difference between the two in cumulative diabetes incidence. However, metformin was provided for many years to the original metformin group while the ILS intervention was reduced in intensity at the end of DPP and provided to all participants, regardless of randomized intervention group. Metformin was ineffective in those ≥60 years old at baseline, but effective at younger ages. By contrast, ILS prevented or delayed diabetes in all age groups studied.

Current evidence for effective and safe means of preventing type 2 diabetes long-term supports lifestyle intervention, emphasizing weight loss and moderate physical activity at any adult age and metformin in younger adults. The relative benefits of ILS and drug interventions may change with future research on newer medicines including the GLP-1 receptor agonists25 that reduce hyperglycemia and body weight and with improvements in long-term maintenance of behavioral weight loss.

Supplementary Material

1
2

RESEARCH IN CONTEXT.

Evidence before this study

The possibility of preventing or delaying the onset of type 2 diabetes has been shown in several randomized clinical trials conducted in adults with prediabetes. These trials have tested lifestyle interventions, drugs, or both. The development of diabetes has been prevented or delayed by most interventions tested, but follow-up of most trials was limited to only a few years. To our knowledge, only two such trials had follow-up of 20 years or more, and only one of these, the US Diabetes Prevention Program (DPP) and its long-term extension, the DPP Outcomes Study (DPPOS), included a drug intervention. Therefore there is little information about the long-term effects of these interventions - a critical question given the difficulty of sustaining behavioral change, including medication adherence, beyond a few months or years. In the lifestyle modification trial in Da Qing, China with 577 participants, the cumulative incidence of diabetes was significantly lower after the six-year intervention and remained lower in the intervention groups than in a control group after 30 years of follow-up. The DPP randomly assigned 3234 participants to metformin, matching placebo, or an intensive lifestyle (ILS) intervention aimed at weight loss and moderate physical activity. Participants have now been followed for approximately 22 years in the combined DPP and DPPOS.

Added value of this study

During 22 years of follow-up, the cumulative incidence rates of diabetes remained lower in the metformin and ILS groups than in the placebo group. These overall treatment effects resulted from large early effects of the interventions but little continuing effects on annual incidence rates during long-term follow-up. ILS effects on relative rate reduction were largely homogeneous, but absolute rate reductions were greatest in those with the greatest baseline risk factors. Relative and absolute effects of metformin were heterogeneous by age, with greater treatment effects in younger participants.

Implications of all the available evidence

Clinicians and public health programs should aim for maximal effects within the first few years of diabetes prevention interventions. Even when intervention intensity cannot be maintained for many years, the initial intervention effects result in a legacy of lower cumulative diabetes incidence for at least 22 years.

Acknowledgements

The DPP Research Group gratefully acknowledges the commitment and dedication of the participants of the DPP and DPPOS. Research reported in this publication was supported by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) of the National Institutes of Health (NIH) under award numbers U01 DK048489, U01 DK048339, U01 DK048377, U01 DK048349, U01 DK048381, U01 DK048468, U01 DK048434, U01 DK048485, U01 DK048375, U01 DK048514, U01 DK048437, U01 DK048413, U01 DK048411, U01 DK048406, U01 DK048380, U01 DK048397, U01 DK048412, U01 DK048404, U01 DK048387, U01 DK048407, U01 DK048443, and U01 DK048400, by providing funding during DPP and DPPOS to the clinical centers and the Coordinating Center for the design and conduct of the study, and collection, management, analysis, and interpretation of the data. Funding was also provided by the National Institute of Child Health and Human Development, the National Institute on Aging, the National Eye Institute, the National Heart Lung and Blood Institute, the National Cancer Institute, the Office of Research on Women’s Health, the National Institute on Minority Health and Health Disparities, the Centers for Disease Control and Prevention, and the American Diabetes Association. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The Southwestern American Indian Centers were supported directly by the NIDDK, including its Intramural Research Program, and the Indian Health Service. The General Clinical Research Center Program, National Center for Research Resources, and the Department of Veterans Affairs supported data collection at many of the clinical centers. Merck KGaA provided medication for DPPOS. DPP/DPPOS have also received donated materials, equipment, or medicines for concomitant conditions from Bristol-Myers Squibb, Parke-Davis, and LifeScan Inc∙, Health O Meter, Hoechst Marion Roussel, Inc∙, Merck-Medco Managed Care, Inc∙, Merck and Co∙, Nike Sports Marketing, Slim Fast Foods Co∙, and Quaker Oats Co. McKesson BioServices Corp∙, Matthews Media Group, Inc∙, and the Henry M. Jackson Foundation provided support services under subcontract with the Coordinating Center. The sponsor of this study was represented on the Steering Committee and played a part in study design, how the study was done, and publication. All authors in the writing group had access to all data. The opinions expressed are those of the study group and do not necessarily reflect the views of the funding agencies. A complete list of Centers, investigators, and staff can be found in the Appendix.

Funding:

US National Institute of Diabetes and Digestive Diseases and other agencies.

Footnotes

Conflict of Interest

All authors have no relevant conflicts to disclose.

Data Sharing Statement

In accordance with the NIH Public Access Policy, we continue to provide all manuscripts to PubMed Central including this manuscript. DPP/DPPOS has provided the protocols and lifestyle and medication intervention manuals to the public through its public website (https://www.dppos.org). The DPPOS abides by the NIDDK data sharing policy and implementation guidance as required by the NIH/NIDDK (https://www.niddkrepository.org/studies/dppos/).

References

  • 1.Knowler WC, Barrett-Connor E, Fowler SE, Hamman RF, Lachin JM, Walker EA, Nathan DM; Diabetes Prevention Program Research Group. Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin. N Engl J Med. 2002. Feb 7;346(6):393–403. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Pan XR, Li GW, Hu YH, Wang JX, Yang WY, An ZX, Hu ZX, Lin J, Xiao JZ, Cao HB, Liu PA, Jiang XG, Jiang YY, Wang JP, Zheng H, Zhang H, Bennett PH, Howard BV. Effects of diet and exercise in preventing NIDDM in people with impaired glucose tolerance. The Da Qing IGT and Diabetes Study. Diabetes Care. 1997. Apr;20(4):537–44. [DOI] [PubMed] [Google Scholar]
  • 3.Tuomilehto J, Lindström J, Eriksson JG, Valle TT, Hämäläinen H, Ilanne-Parikka P, Keinänen-Kiukaanniemi S, Laakso M, Louheranta A, Rastas M, Salminen V, Uusitupa M; Finnish Diabetes Prevention Study Group. Prevention of type 2 diabetes mellitus by changes in lifestyle among subjects with impaired glucose tolerance. N Engl J Med. 2001. May 3;344(18):1343–50. [DOI] [PubMed] [Google Scholar]
  • 4.Ramachandran A, Snehalatha C, Mary S, Mukesh B, Bhaskar AD, Vijay V; Indian Diabetes Prevention Programme (IDPP). The Indian Diabetes Prevention Programme shows that lifestyle modification and metformin prevent type 2 diabetes in Asian Indian subjects with impaired glucose tolerance (IDPP-1). Diabetologia. 2006. Feb;49(2):289–97. [DOI] [PubMed] [Google Scholar]
  • 5.Knowler WC, Crandall JP, Chiasson JL, Nathan DM. Prevention of Type 2 Diabetes. In: Cowie CC, et al. , editors. Diabetes in America. 3rd ed. Bethesda (MD): National Institute of Diabetes and Digestive and Kidney Diseases (US); 2018. Aug. chapter 38. [Google Scholar]
  • 6.Diabetes Prevention Program Research Group; Knowler WC, Fowler SE, Hamman RF, Christophi CA, Hoffman HJ, Brenneman AT, Brown-Friday JO, Goldberg R, Venditti E, Nathan DM. 10-year follow-up of diabetes incidence and weight loss in the Diabetes Prevention Program Outcomes Study. Lancet. 2009. Nov 14;374(9702):1677–86. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.The Diabetes Prevention Program Research Group. The Diabetes Prevention Program. Design and methods for a clinical trial in the prevention of type 2 diabetes. Diabetes Care. 1999. Apr;22(4):623–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Diabetes Prevention Program Research Group. HbA1c as a predictor of diabetes and as an outcome in the diabetes prevention program: a randomized clinical trial. Diabetes Care. 2015. Jan;38(1):51–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Royston P, Parmar MK. Restricted mean survival time: an alternative to the hazard ratio for the design and analysis of randomized trials with a time-to-event outcome. BMC Med Res Methodol. 2013. Dec 7;13:152. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Kitabchi AE, Temprosa M, Knowler WC, Kahn SE, Fowler SE, Haffner SM, Andres R, Saudek C, Edelstein SL, Arakaki R, Murphy MB, Shamoon H; Diabetes Prevention Program Research Group. Role of insulin secretion and sensitivity in the evolution of type 2 diabetes in the diabetes prevention program: effects of lifestyle intervention and metformin. Diabetes. 2005. Aug;54(8):2404–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Ratner RE, Christophi CA, Metzger BE, Dabelea D, Bennett PH, Pi-Sunyer X, Fowler S, Kahn SE; Diabetes Prevention Program Research Group. Prevention of diabetes in women with a history of gestational diabetes: effects of metformin and lifestyle interventions. J Clin Endocrinol Metab. 2008. Dec;93(12):4774–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Rubin RR, Ma Y, Peyrot M, Marrero DG, Price DW, Barrett-Connor E, Knowler WC; Diabetes Prevention Program Research Group. Antidepressant medicine use and risk of developing diabetes during the diabetes prevention program and diabetes prevention program outcomes study. Diabetes Care. 2010. Dec;33(12):2549–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Kent DM, Paulus JK, van Klaveren D, D’Agostino R, Goodman S, Hayward R, Ioannidis JPA, Patrick-Lake B, Morton S, Pencina M, Raman G, Ross JS, Selker HP, Varadhan R, Vickers A, Wong JB, Steyerberg EW. The Predictive Approaches to Treatment effect Heterogeneity (PATH) Statement. Ann Intern Med. 2020. Jan 7;172(1):35–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Gong Q, Zhang P, Wang J, Ma J, An Y, Chen Y, Zhang B, Feng X, Li H, Chen X, Cheng YJ, Gregg EW, Hu Y, Bennett PH, Li G; Da Qing Diabetes Prevention Study Group. Morbidity and mortality after lifestyle intervention for people with impaired glucose tolerance: 30-year results of the Da Qing Diabetes Prevention Outcome Study. Lancet Diabetes Endocrinol. 2019. Jun;7(6):452–461. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.US Preventive Services Task Force; Curry SJ, Krist AH, Owens DK, Barry MJ, Caughey AB, Davidson KW, Doubeni CA, Epling JW Jr, Grossman DC, Kemper AR, Kubik M, Landefeld CS, Mangione CM, Phipps MG, Silverstein M, Simon MA, Tseng CW, Wong JB. Behavioral Weight Loss Interventions to Prevent Obesity-Related Morbidity and Mortality in Adults: US Preventive Services Task Force Recommendation Statement. JAMA. 2018. Sep 18;320(11):1163–1171. [DOI] [PubMed] [Google Scholar]
  • 16.Hamman RF, Horton E, Barrett-Connor E, Bray GA, Christophi CA, Crandall J, Florez JC, Fowler S, Goldberg R, Kahn SE, Knowler WC, Lachin JM, Murphy MB, Venditti E, Diabetes Prevention Program Research Group. Factors affecting the decline in incidence of diabetes in the Diabetes Prevention Program Outcomes Study (DPPOS). Diabetes. 2015. 64:989–98. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Hamman RF, Wing RR, Edelstein SL, Lachin JM, Bray GA, Delahanty L, Hoskin M, Kriska AM, Mayer-Davis EJ, Pi-Sunyer X, Regensteiner J, Venditti B, Wylie-Rosett J. Effect of weight loss with lifestyle intervention on risk of diabetes. Diabetes Care. 2006. 29:2102–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Lachin JM, Christophi CA, Edelstein SL, Ehrmann DA, Hamman RF, Kahn SE, Knowler WC, Nathan DM; DDK Research Group. Factors associated with diabetes onset during metformin versus placebo therapy in the diabetes prevention program. Diabetes. 2007. 56:1153–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Lee AK, Warren B, Liu C, Foti K, Selvin E. Number and Characteristics of US Adults Meeting Prediabetes Criteria for Diabetes Prevention Programs: NHANES 2007–2016. J Gen Intern Med. 2019. Aug;34(8):1400–1402. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Knowler WC. Prevention of type 2 diabetes: comment on “Lifestyle modification and prevention of type 2 diabetes in overweight Japanese with impaired fasting glucose levels”. Arch Intern Med. 2011. Aug 8;171(15):1361–2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Saito T, Watanabe M, Nishida J, Izumi T, Omura M, Takagi T, Fukunaga R, Bandai Y, Tajima N, Nakamura Y, Ito M, Zensharen Study for Prevention of Lifestyle Diseases Group. Lifestyle modification and prevention of type 2 diabetes in overweight Japanese with impaired fasting glucose levels: a randomized controlled trial. Arch Intern Med. 2011. Aug 8;171(15):1352–60. [DOI] [PubMed] [Google Scholar]
  • 22.Weber MB, Ranjani H, Staimez LR, Anjana RM, Ali MK, Narayan KM, Mohan V. The Stepwise Approach to Diabetes Prevention: Results From the D-CLIP Randomized Controlled Trial. Diabetes Care. 2016. Oct;39(10):1760–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Zhang L, Zhang Y, Shen S, Wang X, Dong L, Li Q, Ren W, Li Y, Bai J, Gong Q, Kuang H, Qi L, Lu Q, Cheng W, Liu Y, Yan S, Wu D, Fang H, Hou F, Wang Y, Yang Z, Lian X, Du J, Sun N, Ji L, Li G; China Diabetes Prevention Program Study Group. Safety and effectiveness of metformin plus lifestyle intervention compared with lifestyle intervention alone in preventing progression to diabetes in a Chinese population with impaired glucose regulation: a multicentre, open-label, randomised controlled trial. Lancet Diabetes Endocrinol. 2023. Aug;11(8):567–577. [DOI] [PubMed] [Google Scholar]
  • 24.Sathish T, Khunti K, Narayan KMV, Mohan V, Davies MJ, Yates T, Oldenburg B, Thankappan KR, Tapp RJ, Bajpai R, Anjana RM, Weber MB, Ali MK, Shaw JE. Effect of conventional lifestyle interventions on type 2 diabetes incidence by glucose-defined prediabetes phenotype: an individual participant data meta-analysis of randomized controlled trials. Diabetes Care. 2023. Nov 1;46(11):1903–1907. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Kahn SE, Deanfield JE, Jeppesen OK, Emerson SS, Boesgaard TW, Colhoun HM, Kushner RF, Lingvay I, Burguera B, Gajos G, Horn DB, Hramiak IM, Jastreboff AM, Kokkinos A, Maeng M, Matos ALSA, Tinahones FJ, Lincoff AM, Ryan DH; SELECT Trial Investigators. Effect of Semaglutide on Regression and Progression of Glycemia in People With Overweight or Obesity but Without Diabetes in the SELECT Trial. Diabetes Care. 2024. Aug 1;47(8):1350–1359. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS2081085-supplement-1.docx (58.5KB, docx)
2
NIHMS2081085-supplement-2.docx (1.5MB, docx)

Data Availability Statement

In accordance with the NIH Public Access Policy, we continue to provide all manuscripts to PubMed Central including this manuscript. DPP/DPPOS has provided the protocols and lifestyle and medication intervention manuals to the public through its public website (https://www.dppos.org). The DPPOS abides by the NIDDK data sharing policy and implementation guidance as required by the NIH/NIDDK (https://www.niddkrepository.org/studies/dppos/).

RESOURCES