020702 reduction in the incidence of type 2 diabetes
C o py r ig ht 2 0 0 2 by t he Ma s s ac h u s e t t s Me d ic a l S o c ie t y
V O L U M E 3 4 6 REDUCTION IN THE INCIDENCE OF TYPE 2 DIABETES WITH LIFESTYLE INTERVENTION OR METFORMIN
DIABETES PREVENTION PROGRAM RESEARCH GROUP*
non-insulin-dependent diabetes mellitus, is
ly 8 percent of adults in the United States. Some risk
a serious, costly disease affecting approxi-
factors — elevated plasma glucose concentrations in
the fasting state and after an oral glucose load, over-
States.1 Treatment prevents some of its devastating
weight, and a sedentary lifestyle — are potentially
complications2,3 but does not usually restore normo-
reversible. We hypothesized that modifying these
glycemia or eliminate all the adverse consequences.
factors with a lifestyle-intervention program or the
The diagnosis is often delayed until complications are
administration of metformin would prevent or delay
present.4 Since current methods of treating diabetes
remain inadequate, prevention is preferable. The hy-
pothesis that type 2 diabetes is preventable5,6 is sup-
persons with elevated fasting and post-load plasmaglucose concentrations to placebo, metformin (850
ported by observational studies and two clinical tri-
mg twice daily), or a lifestyle-modification program
als of diet, exercise, or both in persons at high risk
with the goals of at least a 7 percent weight loss and
for the disease7,8 but not by studies of drugs used to
at least 150 minutes of physical activity per week.
The mean age of the participants was 51 years, and
The validity of generalizing the results of previous
the mean body-mass index (the weight in kilograms
prevention studies is uncertain.9 Interventions that
divided by the square of the height in meters) was
work in some societies may not work in others, be-
34.0; 68 percent were women, and 45 percent were
cause social, economic, and cultural forces influence
diet and exercise. This is a special concern in the
United States, where there is great regional and ethnic
incidence of diabetes was 11.0, 7.8, and 4.8 cases per
diversity in lifestyle patterns and where diabetes is es-
100 person-years in the placebo, metformin, and life-style groups, respectively. The lifestyle intervention
pecially frequent in certain racial and ethnic groups,
reduced the incidence by 58 percent (95 percent con-
including American Indians, Hispanics, African Amer-
fidence interval, 48 to 66 percent) and metformin by
31 percent (95 percent confidence interval, 17 to 43
The Diabetes Prevention Program Research Group
percent), as compared with placebo; the lifestyle in-
conducted a large, randomized clinical trial involv-
tervention was significantly more effective than met-
ing adults in the United States who were at high risk
formin. To prevent one case of diabetes during a
for the development of type 2 diabetes. The study
period of three years, 6.9 persons would have to par-
was designed to answer the following primary ques-
ticipate in the lifestyle-intervention program, and 13.9
tions: Does a lifestyle intervention or treatment with
metformin both reduced the incidence of diabetes inpersons at high risk. The lifestyle intervention was
The writing group (William C. Knowler, M.D., Dr.P.H., Elizabeth Bar-
rett-Connor, M.D., Sarah E. Fowler, Ph.D., Richard F. Hamman, M.D.,
more effective than metformin. (N Engl J Med 2002;
Dr.P.H., John M. Lachin, Sc.D., Elizabeth A. Walker, D.N.Sc., and David
M. Nathan, M.D.) takes responsibility for the content of this article.
Copyright 2002 Massachusetts Medical Society.
Address reprint requests to the Diabetes Prevention Program Coordinat-
ing Center, Biostatistics Center, George Washington University, 6110 Ex-ecutive Blvd., Suite 750, Rockville, MD 20852.
*The members of the Diabetes Prevention Program Research Group are
N Engl J Med, Vol. 346, No. 6 · February 7, 2002 · www.nejm.org · 393
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
metformin, a biguanide antihyperglycemic agent, pre-
during the first 24 weeks after enrollment, was flexible, culturally
vent or delay the onset of diabetes? Do these two
sensitive, and individualized. Subsequent individual sessions (usu-
interventions differ in effectiveness? Does their ef-
ally monthly) and group sessions with the case managers were de-signed to reinforce the behavioral changes.
fectiveness differ according to age, sex, or race orethnic group?
The primary outcome was diabetes, diagnosed on the basis of
an annual oral glucose-tolerance test or a semiannual fasting plasma
We conducted a clinical trial involving persons at 27 centers
glucose test, according to the 1997 criteria of the American Diabe-
who were at high risk for diabetes. The methods have been de-
tes Association: a value for plasma glucose of 126 mg per deciliter
scribed in detail elsewhere,6 and the protocol is available at http://
(7.0 mmol per liter) or higher in the fasting state or 200 mg per
www.bsc.gwu.edu/dpp. The institutional review board at each
deciliter (11.1 mmol per liter) or higher two hours after a 75-g
center approved the protocol, and all participants gave written in-
oral glucose load. In addition to the semiannual measurements,
fasting plasma glucose was measured if symptoms suggestive ofdiabetes developed. The diagnosis required confirmation by a sec-
ond test, usually within six weeks, according to the same criteria. If diabetes was diagnosed, the participants and their physicians
Eligibility criteria included an age of at least 25 years, a body-
were informed and glucose-tolerance tests were discontinued, but
mass index (the weight in kilograms divided by the square of the
fasting plasma glucose was measured every six months, with gly-
height in meters) of 24 or higher (22 or higher in Asians), and a
cosylated hemoglobin measured annually. As long as the fasting
plasma glucose concentration of 95 to 125 mg per deciliter (5.3
plasma glucose concentration was less than 140 mg per deciliter,
to 6.9 mmol per liter) in the fasting state («125 mg per deciliter
participants were asked to monitor their blood glucose and to con-
in the American Indian clinics) and 140 to 199 mg per deciliter
tinue their assigned study treatment. If the fasting plasma glucose
(7.8 to 11.0 mmol per liter) two hours after a 75-g oral glucose
concentration reached or exceeded 140 mg per deciliter, the study
load. These concentrations are elevated but are not diagnostic of
medication was discontinued and the participant was referred to his
diabetes according to the 1997 criteria of the American Diabetes
or her physician for treatment. Measurements of glucose and gly-
Association.11 Before June 1997, the criterion for plasma glucose in
cosylated hemoglobin (HbA ) were performed centrally. All tests
the fasting state was 100 to 139 mg per deciliter (5.6 to 7.7 mmol
were performed without interrupting the assigned treatment, ex-
per liter), or «139 mg per deciliter in the American Indian clinics.
cept that placebo or metformin was not taken on the morning of
Eligible persons were excluded if they were taking medicines known
the test. The investigators and the participants were unaware of the
to alter glucose tolerance or if they had illnesses that could seriously
results of these measurements and were informed only if the results
reduce their life expectancy or their ability to participate in the trial.
exceeded the specified threshold for a change in the treatment.
Recruitment was designed to enroll approximately half the partic-
Self-reported levels of leisure physical activity were assessed annu-
ipants from racial or ethnic minority groups. A four-step screen-
ally with the Modifiable Activity Questionnaire.16 The physical-
ing and recruitment process was developed to identify eligible par-
activity level was calculated as the product of the duration and fre-
quency of each activity (in hours per week), weighted by an estimateof the metabolic equivalent of that activity (MET) and summed
for all activities performed, with the result expressed as the aver-
Eligible participants were randomly assigned to one of three in-
age MET-hours per week for the previous year. Usual daily caloric
terventions: standard lifestyle recommendations plus metformin
intake during the previous year, including calories from fat, carbo-
(Glucophage) at a dose of 850 mg twice daily, standard lifestyle
hydrate, protein, and other nutrients, was assessed at base line and
recommendations plus placebo twice daily, or an intensive pro-
at one year with the use of a modified version of the Block food-
gram of lifestyle modification. The study initially included a fourth
intervention, troglitazone, which was discontinued in 1998 be-cause of the drug’s potential liver toxicity.6 The results in the tro-
Statistical Analysis and Early Closure
glitazone group are not reported here.
Random treatment assignments were stratified according to the
Treatment with metformin was initiated at a dose of 850 mg
clinical center. Assignments to metformin and placebo were dou-
taken orally once a day, with placebo tablets also given once a day
ble-blinded. The study design and analysis followed the inten-
initially. At one month, the dose of metformin was increased to
tion-to-treat principle. Nominal (unadjusted) P values and confi-
850 mg twice daily, unless gastrointestinal symptoms warranted
a longer titration period. The initiation of treatment with half a
The blinded treatment phase was terminated one year early, in
tablet was optional. Adherence to the treatment regimen was as-
May 2001, on the advice of the data monitoring board, on the
sessed quarterly on the basis of pill counts and structured inter-
basis of data obtained through March 31, 2001, the closing date
views. The standard lifestyle recommendations for the medication
for this report. By then, we had obtained evidence of efficacy on
groups were provided in the form of written information and in
the basis of 65 percent of the planned person-years of observation.
an annual 20-to-30-minute individual session that emphasized the
To maintain a type I error level of 0.05 for significance in pairwise
importance of a healthy lifestyle. Participants were encouraged to
comparisons of the risk of diabetes between groups, with adjust-
follow the Food Guide Pyramid14 and the equivalent of a National
ment for repeated interim analyses, the group-sequential log-rank
Cholesterol Education Program Step 1 diet,15 to reduce their
test18 required a P value of less than 0.0159. For pairwise compar-
weight, and to increase their physical activity.
isons of other outcomes, a Bonferroni-adjusted criterion of P<
The goals for the participants assigned to the intensive lifestyle
0.0167 was used. The study design provided 90 percent power to
intervention were to achieve and maintain a weight reduction of
detect a 33 percent reduction from an incidence of 6.5 cases of
at least 7 percent of initial body weight through a healthy low-
diabetes per 100 person-years, with a 10 percent rate of loss to
calorie, low-fat diet and to engage in physical activity of moderate
intensity, such as brisk walking, for at least 150 minutes per week.
The time to the outcome was assessed with the use of life-table
A 16-lesson curriculum covering diet, exercise, and behavior mod-
methods.19 Modified product-limit curves for the cumulative in-
ification was designed to help the participants achieve these goals.
cidence of diabetes were compared with the use of the log-rank
The curriculum, taught by case managers on a one-to-one basis
test. The estimated cumulative incidence at three years and the
394 · N Engl J Med, Vol. 346, No. 6 · February 7, 2002 · www.nejm.org R E D U C I N G T H E I N C I D E N C E O F T Y P E 2 D I A B ET E S W I T H L I F E ST Y L E I N T E RV E N T I O N O R M ET F O R M I N
Greenwood estimate of the standard error were used to calculate
erage of 2.8 years (range, 1.8 to 4.6). At the close of
the number of persons who would need to be treated in order to
the study, 99.6 percent of the participants were alive,
prevent one case of confirmed diabetes during a period of three
of whom 92.5 percent had attended a scheduled vis-
years and the associated 95 percent confidence interval. Risk re-duction, heterogeneity among strata, and interactions between
treatment assignments and covariates were assessed by proportion-al-hazards regression. Fixed-effects models with the assumption
Adherence to Interventions
of normally distributed errors20 were used to assess differences
Fifty percent of the participants in the lifestyle-
over time in body weight and plasma glucose and glycosylated
intervention group had achieved the goal of weight
hemoglobin values among the three groups.
loss of 7 percent or more by the end of the curricu-
lum (at 24 weeks), and 38 percent had a weight lossof at least 7 percent at the time of the most recent
Study Cohort and Follow-up
visit; the proportion of participants who met the
From 1996 to 1999, we randomly assigned 3234
goal of at least 150 minutes of physical activity per
study participants to one of the three interventions
week (assessed on the basis of logs kept by the par-
(1082 to placebo, 1073 to metformin, and 1079 to
ticipants) was 74 percent at 24 weeks and 58 percent
the intensive lifestyle intervention). Base-line charac-
at the most recent visit. Dietary change was assessed
teristics, including all measured risk factors for dia-
only at one year. Daily energy intake decreased by a
betes, were similar among the three study groups
mean (±SE) of 249±27 kcal in the placebo group,
(Table 1).12 The participants were followed for an av-
296±23 kcal in the metformin group, and 450±26
TABLE 1. BASE-LINE CHARACTERISTICS OF THE STUDY PARTICIPANTS.* METFORMIN LIFESTYLE CHARACTERISTIC (N=3234) (N=1082) (N=1073) (N=1079)
Leisure physical activity — MET-hr/wk¶
*Plus–minus values are means ±SD.
†Twenty Pacific Islanders were included in this category.
‡Information was not available for one participant.
§To convert the values for glucose to millimoles per liter, multiply by 0.05551.
¶Data are based on responses to the Modifiable Activity Questionnaire.16 MET denotes metabolic
equivalent. MET-hours represent the average amount of time engaged in specified physical activitiesmultiplied by the MET value of each activity.
N Engl J Med, Vol. 346, No. 6 · February 7, 2002 · www.nejm.org · 395
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
kcal in the lifestyle-intervention group (P<0.001).
ipants taking placebo and 84 percent of those taking
Average fat intake, which was 34.1 percent of total
metformin were given the full dose of one tablet
calories at base line, decreased by 0.8±0.2 percent in
(850 mg in the case of metformin) twice a day; the
the placebo and metformin groups and by 6.6±0.2
remainder were given one tablet a day to limit side
percent in the lifestyle-intervention group (P<
0.001). The proportion of participants who took at
Changes in weight and leisure physical activity in all
least 80 percent of the prescribed dose of the study
three groups and adherence to the medication regi-
medication was slightly higher in the placebo group
men in the metformin and placebo groups are shown
than in the metformin group (77 percent vs. 72 per-
in Figure 1. Participants assigned to the lifestyle in-
cent, P<0.001). Ninety-seven percent of the partic-
tervention had much greater weight loss and a great-
Figure 1. Changes in Body Weight (Panel A) and Leisure Physical Activity (Panel B) and Adherence to Medication Regimen (Panel C) According to Study Group.
Each data point represents the mean value for all participants examined at that time. The number ofparticipants decreased over time because of the variable length of time that persons were in the study. For example, data on weight were available for 3085 persons at 0.5 year, 3064 at 1 year, 2887 at 2 years,and 1510 at 3 years. Changes in weight and leisure physical activity over time differed significantlyamong the treatment groups (P<0.001 for each comparison). 396 · N Engl J Med, Vol. 346, No. 6 · February 7, 2002 · www.nejm.org R E D U C I N G T H E I N C I D E N C E O F T Y P E 2 D I A B ET E S W I T H L I F E ST Y L E I N T E RV E N T I O N O R M ET F O R M I N
er increase in leisure physical activity than did par-
ticipants assigned to receive metformin or placebo. The average weight loss was 0.1, 2.1, and 5.6 kg in theplacebo, metformin, and lifestyle-intervention groups,
Incidence of Diabetes
The cumulative incidence of diabetes was lower in
the metformin and lifestyle-intervention groups thanin the placebo group throughout the follow-up pe-
riod (Fig. 2). The crude incidence was 11.0, 7.8, and
4.8 cases per 100 person-years for the placebo, met-formin, and lifestyle-intervention groups, respective-
ly (Table 2). The incidence of diabetes was 58 per-cent lower (95 percent confidence interval, 48 to 66
percent) in the lifestyle-intervention group and 31
Figure 2. Cumulative Incidence of Diabetes According to Study
percent lower (95 percent confidence interval, 17 to
43 percent) in the metformin group than in the pla-
The diagnosis of diabetes was based on the criteria of the
cebo group. The incidence of diabetes was 39 per-
American Diabetes Association.11 The incidence of diabetes dif-
cent lower (95 percent confidence interval, 24 to 51
fered significantly among the three groups (P<0.001 for eachcomparison).
percent) in the lifestyle-intervention group than inthe metformin group. The results of all three pair-wise group comparisons were statistically significantby the group-sequential log-rank test. None of theseresults were materially affected by adjustment for
values for those variables. Neither interaction was
base-line characteristics. The estimated cumulative
explained by the other variable or by age. The ad-
incidence of diabetes at three years was 28.9 percent,
vantage of the lifestyle intervention over metformin
21.7 percent, and 14.4 percent in the placebo, met-
was greater in older persons and those with a lower
formin, and lifestyle-intervention groups, respective-
body-mass index than in younger persons and those
ly. On the basis of these rates, the estimated number
of persons who would need to be treated for threeyears to prevent one case of diabetes during this pe-
riod is 6.9 (95 percent confidence interval, 5.4 to
In the first year, there was a similar reduction in
9.5) for the lifestyle intervention and 13.9 (95 per-
the mean fasting plasma glucose values in the met-
cent confidence interval, 8.7 to 33.9) for metformin.
formin and lifestyle-intervention groups, whereas thevalues rose in the placebo group (Fig. 3). The values
Treatment Effects among Subgroups
rose in parallel in all three groups in subsequent
Incidence rates and risk reductions within sub-
years. There was a similar temporal pattern in the
groups of participants and the results of tests of the
values for glycosylated hemoglobin, except that the
homogeneity of risk reduction among subgroups are
values in the metformin group were in between
shown in Table 2; 95 percent confidence intervals for
those in the lifestyle-intervention and placebo groups.
the subgroup data indicate the precision of the risk-
Figure 4 shows the percentage of participants who
reduction estimate for each stratum. The study had
had normal glucose concentrations (fasting values,
inadequate power to assess the significance of effects
post-load values, and both) at each annual examina-
within the subgroups, nor were such tests planned.
tion. Metformin and the lifestyle intervention were
Significant heterogeneity indicates that treatment ef-
similarly effective in restoring normal fasting glucose
fects differed according to the values of the covariates.
values, but the lifestyle intervention was more effec-
Treatment effects did not differ significantly accord-
tive in restoring normal post-load glucose values.
ing either to sex or to race or ethnic group (Table 2). The lifestyle intervention was highly effective in all
subgroups. Its effect was significantly greater among
The rate of gastrointestinal symptoms was highest
persons with lower base-line glucose concentrations
in the metformin group, and the rate of musculo-
two hours after a glucose load than among those
skeletal symptoms was highest in the lifestyle-inter-
with higher base-line glucose values. The effect of met-
vention group (Table 3). Hospitalization and mor-
formin was less with a lower body-mass index or a low-
tality rates were unrelated to treatment. No deaths
er fasting glucose concentration than with higher
were attributed to the study intervention.
N Engl J Med, Vol. 346, No. 6 · February 7, 2002 · www.nejm.org · 397
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
TABLE 2. INCIDENCE OF DIABETES. VARIABLE PARTICIPANTS (%) INCIDENCE REDUCTION IN INCIDENCE (95% CI)*
†P<0.05 for the test of heterogeneity across strata. Age, body-mass index, and plasma glucose were analyzed as continuous variables.
‡This category includes 20 Pacific Islanders.
§The eligibility criterion was a body-mass index of at least 22 for Asians and at least 24 for all other persons.
¶To convert the values for glucose to millimoles per liter, multiply by 0.05551.
¿This category includes American Indian participants who had a fasting glucose concentration that was less than 95 mg per deciliter, ac-
**This category includes 54 participants with a fasting glucose concentration of 126 to 139 mg per deciliter who were enrolled before
June 1997,6 when the eligibility criteria were changed to conform to the diagnostic criteria of the American Diabetes Association, publishedthat year.11
turally diverse population of the United States. The
Our results support the hypothesis that type 2 di-
risk reduction associated with the lifestyle interven-
abetes can be prevented or delayed in persons at
tion in our study was the same as that in a study
high risk for the disease. The incidence of diabetes
conducted in Finland,8 and was higher than the re-
was reduced by 58 percent with the lifestyle inter-
ductions associated with diet (31 percent), exercise
vention and by 31 percent with metformin, as com-
(46 percent), and diet plus exercise (42 percent) in
pared with placebo. These effects were similar in
men and women and in all racial and ethnic groups.
Our lifestyle intervention was systematic and in-
The intensive lifestyle intervention was at least as ef-
tensive, with the study participants receiving de-
fective in older participants as it was in younger par-
tailed, individualized counseling. The study, however,
ticipants. The results of our study extend previous
was not designed to test the relative contributions of
data showing that lifestyle interventions can reduce
dietary changes, increased physical activity, and weight
the incidence of diabetes7,8 and demonstrate the ap-
loss to the reduction in the risk of diabetes, and the
plicability of this finding to the ethnically and cul-
effects of these components remain to be determined. 398 · N Engl J Med, Vol. 346, No. 6 · February 7, 2002 · www.nejm.org R E D U C I N G T H E I N C I D E N C E O F T Y P E 2 D I A B ET E S W I T H L I F E ST Y L E I N T E RV E N T I O N O R M ET F O R M I N Figure 3. Fasting Plasma Glucose Concentrations (Panel A) and Glycosylated Hemoglobin Values (Pan- el B) According to Study Group.
The analysis included all participants, whether or not diabetes had been diagnosed. Changes in fastingglucose values over time in the three groups differed significantly (P<0.001). Glycosylated hemoglobinvalues in the three groups differed significantly from 0.5 to 3 years (P<0.001). To convert the valuesfor glucose to millimoles per liter, multiply by 0.05551.
The incidence of diabetes in our placebo group
our study by the selection of persons who were over-
(11.0 cases per 100 person-years) was higher than
weight and had elevated fasting and post-load glu-
we had anticipated6 and was higher than the inci-
cose concentrations — three of the strongest risk
dence in observational studies,21 perhaps owing to
the greater frequency of glucose testing or to the se-
Previous studies have not demonstrated that drugs
lection of persons at higher risk in our study. The
used to treat diabetes are effective for its prevention,
incidence of diabetes in the placebo group was similar
perhaps because of small samples and the lack of data
among racial and ethnic groups despite differences in
on adherence to the prescribed regimens.5 In con-
these subgroups in observational, population-based
trast, metformin was effective in our study, although
studies.10 Racial and ethnic-group differences in the
less so than the lifestyle intervention. Metformin was
incidence of diabetes were presumably reduced in
less effective in persons with a lower base-line body-
N Engl J Med, Vol. 346, No. 6 · February 7, 2002 · www.nejm.org · 399
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
Participants with Normal Plasma Glucose (%)
Figure 4. Participants with Normal Plasma Glucose Values, According to Study Group.
Panel A shows the proportions of participants with normal glucose values in the fasting state (<110 mg per deciliter[6.1 mmol per liter]), Panel B the proportions with normal values two hours after an oral glucose load (<140 mg perdeciliter [7.8 mmol per liter]), and Panel C the proportions with normal values for both measurements. Persons in whoma diagnosis of diabetes had been made were considered to have abnormal values, regardless of the actual values at thetime. By design, no participants had normal post-load glucose values at base line, but base-line fasting glucose valueswere normal in 67 percent of persons in the placebo group, 67 percent of those in the metformin group, and 68 percentof those in the lifestyle-intervention group. Metformin and lifestyle intervention were similarly effective in restoringnormal fasting glucose concentrations, but lifestyle intervention was more effective in restoring normal post-load glu-cose concentrations. 400 · N Engl J Med, Vol. 346, No. 6 · February 7, 2002 · www.nejm.org R E D U C I N G T H E I N C I D E N C E O F T Y P E 2 D I A B ET E S W I T H L I F E ST Y L E I N T E RV E N T I O N O R M ET F O R M I N
analysis of the main secondary outcomes — reduc-
TABLE 3. ADVERSE EVENTS.
tions in risk factors for cardiovascular disease, in theproportion of participants with atherosclerosis, and
PLACEBO METFORMIN LIFESTYLE
in the proportion with cardiovascular disease, whichis the leading cause of death among patients with
Gastrointestinal symptoms (no. of events/
Optimal approaches to identifying candidates for
preventive measures remain to be determined. Al-
though elevation of either the fasting or the post-load
glucose concentration strongly predicts diabetes,26,27
both were required for eligibility in this study.
Whether the results would be similar in persons with
an isolated elevation of the fasting or post-load glu-
*Gastrointestinal symptoms included diarrhea, flatulence, nausea, and
cose concentration or other risk factors for diabetes
†P<0.0167 for the comparison with placebo.
In summary, our study showed that treatment
‡Most participants with musculoskeletal symptoms had myalgia, arthri-
with metformin and modification of lifestyle were
two highly effective means of delaying or preventingtype 2 diabetes. The lifestyle intervention was par-ticularly effective, with one case of diabetes prevent-ed per seven persons treated for three years. Thus, itshould also be possible to delay or prevent the de-
mass index or a lower fasting plasma glucose con-
velopment of complications, substantially reducing
centration than in those with higher values for these
the individual and public health burden of diabetes.
variables. The reduction in the average fasting plas-ma glucose concentration was similar in the lifestyle-
Supported by the National Institutes of Health through the National In-
intervention and metformin groups, but the lifestyle
stitute of Diabetes and Digestive and Kidney Diseases, the Office of Re-
intervention had a greater effect than metformin on
search on Minority Health, the National Institute of Child Health and Hu-man Development, and the National Institute on Aging; the Indian Health
glycosylated hemoglobin, and a larger proportion of
Service; the Centers for Disease Control and Prevention; the General Clin-
participants in the lifestyle-intervention group had
ical Research Center Program, National Center for Research Resources;the American Diabetes Association; Bristol-Myers Squibb; and Parke-
normal post-load glucose values at follow-up. These
findings are consistent with the observation that met-
Dr. Hamman owns stock in Bristol-Myers Squibb, which sells metformin
formin suppresses endogenous glucose production,
the main determinant of fasting plasma glucose con-
We are indebted to the participants in the study for their dedica-tion to the goal of preventing diabetes; to Lipha Pharmaceuticals for
Rates of adverse events, hospitalization, and mor-
the metformin and placebo; to LifeScan, Health-O-Meter, Hoechst
tality were similar in the three groups, except that
Marion Roussel, Merck-Medco Managed Care, Merck, Nike, Slim-
the rate of gastrointestinal symptoms was highest in
Fast Foods, and Quaker Oats for materials, equipment, and medi-cines for concomitant conditions; and to McKesson BioServices, Mathews
the metformin group and the rate of musculoskele-
Media Group, and the Henry M. Jackson Foundation for support
tal symptoms was highest in the lifestyle-interven-
services provided under subcontract with the Coordinating Center.
tion group. Thus, the interventions were safe in ad-dition to being effective. APPENDIX
An estimated 10 million persons in the United
The following investigators were members of the Diabetes Prevention
States resemble the participants in the Diabetes Pre-
Program Research Group (asterisks indicate principal investigators, and daggers program coordinators): Pennington Biomedical Research Center
vention Program in terms of age, body-mass index,
— G.A. Bray,* I.W. Culbert,† C.M. Champagne, M.D. Crow, L. Dawson,
and glucose concentrations, according to data from
B. Eberhardt, F.L. Greenway, F.G. Guillory, A.A. Hebert, M.L. Jeffirs, K.
the third National Health and Nutrition Examination
Joyce, B.M. Kennedy, J.C. Lovejoy, S. Mancuso, L.E. Melancon, L.H. Morris, L. Reed, J. Perault, K. Rau, D.H. Ryan, D.A. Sanford, K.G. Smith,
Survey.23 If the study’s interventions were implement-
L.L. Smith, S.R. Smith, J.A. St. Amant, M. Terry, E. Tucker, R.T. Tulley,
ed among these people, there would be a substantial
P.C. Vicknair, D. Williamson, and J.J. Zachwieja; University of Chicago — D.A. Ehrmann,* M.J. Matulik,† B. Clarke, D.A. Collins, K.B. Czech, C.
reduction in the incidence of diabetes. Ultimately, the
DeSandre, G. Geiger, S. Grief, B. Harding-Clay, R.M. Hilbrich, D. le
benefits would depend on whether glucose concen-
Grange, M.R. McCormick, W.L. McNabb, K.S. Polonsky, N.P. Sauter,
trations could be maintained at levels below those
A.R. Semenske, K.A. Stepp, and J.A. Tobian; Jefferson Medical College — P.G. Watson,* J.T. Mendoza,† K.A. Smith,† J. Caro, B. Goldstein, C. Lark,
that are diagnostic of diabetes and whether the main-
L. Menefee, L. Murphy, C. Pepe, and J.M. Spandorfer; University of Mi-
tenance of these lower levels improved the long-term
ami — R.B. Goldberg,* P. Rowe,† J. Calles, P. Casanova, R.P. Donahue,
outcome. These questions should be addressed by
H.J. Florez, A. Giannella, G. Larreal, V. McLymont, J. Mendez, P. O’Hara, J. Ojito, R. Prineas, and P.G. Saab; The University of Texas Health Science
continued follow-up of the study participants and by
Center — S.M. Haffner,* M.G. Montez,† C. Lorenzo, H. Miettinen,
N Engl J Med, Vol. 346, No. 6 · February 7, 2002 · www.nejm.org · 401
The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
C.M. Mobley, L.A. Mykkanen, and M.M. Rozek; University of Colorado
des, W. Grant, R.L. Hanson, E. Horse, G. Hughte, L.E. Ingraham, M.C.
— R.F. Hamman,* P.V. Nash,† L. Testaverde,† D.R. Anderson, L.B. Bal-
Jackson, P.A. Jay, R.S. Kaskalla, D. Kessler, K.M. Kobus, J. Krakoff, C.
lonoff, A. Bouffard, B.N. Calonge, M. Farago, W.J. Georgitis, J.O. Hill,
Manus, T. Morgan, Y. Nashboo (deceased), J. Nelson, G.L. Pauk, S. Poir-
S.R. Hoyer, B.T. Jortberg, J.A. Merenich, M. Miller, J.G. Regensteiner,
ier, E. Polczynski, M. Reidy, J. Roumain, D.H. Rowse, R.J. Roy, S. Sang-
H.M. Seagle, C.M. Smith, S.C. Steinke, and B. VanDorsten; Joslin Diabe-
ster, J. Sewemaenewa, D. Tonemah, C. Wilson, and M. Yazzie; George tes Center — E.S. Horton,* K.E. Lawton,† R.A. Arky, M. Bryant, J.P. Washington University Biostatistics Center — S. Fowler,* T. Brenne-
Burke, E. Caballero, K.M. Callaghan, D. Devlin, T. Franklin, O.P. Ganda,
man,† S. Abebe, R. Bain, J. Bamdad, J. Callaghan, S.L. Edelstein, Y. Gao,
A.E. Goebel-Fabbri, M. Harris, S.D. Jackson, A.M. Jacobsen, L.M. Kula,
K.L. Grimes, N. Grover, K. Hirst, S. Jones, T.L. Jones, R.J. Katz, J.M.
M. Kocal, S. Ledbury, M.A. Malloy, C. Mullooly, M. Nicosia, C.F. Old-
Lachin, R. Orlosky, C.E. Stimpson, C. Suiter, M.G. Temprosa, and F.E.M.
mixon, J. Pan, C. Pomposelli, M. Quitongan, S. Rubtchinsky, D. Schwei-
Walker-Murray; National Institute of Diabetes and Digestive and Kidney
zer, E.W. Seely, D. Simonson, F. Smith, C.G. Solomon, J. Tyson, and J. Diseases Program Office — S. Garfield,† R. Eastman, and J. Fradkin; Na-
Warram; University of Washington and Veterans Affairs Puget Sound tional Institute on Aging — R. Andres; Centers for Disease Control and Health Care System — S.E. Kahn,* B.K. Montgomery,† M. Alger, E. Prevention — M.M. Engelgau, K.M. Venkat Narayan, and D.F. William-
Allen, T. Barrett, D. Bhanji, J. Cowan, J. Cullen, W.Y. Fujimoto, B. Katz,
son; University of Michigan — W.H. Herman; Central Biochemistry Lab-
R.H. Knopp, E.W. Lipkin, M. Marr, B.S. McCann, J.P. Palmer, R.S. oratory (University of Washington) — S.M. Marcovina,* A. Aldrich, and
Schwartz, and D. Uyema; University of Tennessee — A.E. Kitabchi,* M.E.
W.L. Chandler; Central ECG Reading Center (Wake Forest University)
Murphy,† W.B. Applegate, M. Bryer-Ash, J.H. Coble, A. Crisler, G. Cun-
— P.M. Rautaharju,* N.T. Pemberton, R. Prineas, F.S.R. Rautaharju, and
ningham, A.W. Franklin, S.L. Frieson, D.L. Green, R. Imseis, C.L.
Z. Zhang; Nutrition Coding Center (University of South Carolina) —
Kennedy, H.C. Lambeth, K.A. Latif, L.C. Lichtermann, M.D. McIntyre,
E.J. Mayer-Davis,* T. Costacou, M. Martin, and K.L. Sparks; Central Ca-
J.D. Nault, H. Oktaei, M.L. O’Toole, H. Ricks, L.M.K. Rutledge, S.C. rotid Ultrasound Unit (New England Medical Center) — D.H.
Schussler, A.R. Sherman, C.M. Smith, J.E. Soberman, K.J. Stewart, D.L.
O’Leary,* L.R.C. Funk, K.A. O’Leary, and J.F. Polak; CT-Scan Reading
VanBrunt, and B.J. Williams-Cleaves; Northwestern University Medical Unit (University of Colorado) — E.R. Stamm* and A.L. Scherzinger; School — B.E. Metzger,* M.K. Johnson,† C. Behrends, M.L. Cook, M. Lifestyle Resource Core (University of Pittsburgh) — R.R. Wing,* B.P.
Fitzgibbon, M.M. Giles, D. Heard, C. Johnson, D. Larsen, A. Lowe, M.
Gillis, C. Huffmyer, A.M. Kriska, and E.M. Venditti; Medication Resource
Lyman, D. McPherson, M.E. Molitch, T. Pitts, R. Reinhart, S. Roston, and
Workgroup (Albert Einstein College of Medicine) — E.A. Walker* and T.
P.A. Schinleber; Massachusetts General Hospital — D.M. Nathan,* C.
Harroun; Quality of Well Being Reading Unit (University of California,
McKitrick,† K. Abbott, E. Anderson, L. Bissett, E. Cagliero, S. Crowell,
San Diego) — T.G. Ganiats,* E.J. Groessl, P.R. Beerman, K.M. David,
L. Delahanty, S. Fritz, K. Hayward, E. Levina, T. Michel, D. Norman, J.
R.M. Kaplan, and W.J. Sieber; Data Monitoring Board — S.M. Genuth
O’Keefe, A. Poulos, L. Ronan, M. Rosal, M. Salerno, M. Schneider, C.
(chair), G.F. Cahill, F.L. Ferris III, J.R. Gavin III, J.B. Halter, and J. Wittes;
Shagensky, B. Steiner, H. Turgeon, and A. Young; University of Califor- Ancillary Studies Subcommittee — R.R. Henry and S.M. Haffner nia, San Diego — J.M. Olefsky,* M.L. Carrion-Petersen,† E. Barrett-Con-
(chairs); Behavioral Scientists — R.R. Rubin (chair); Clinic Operations
nor, M. Beltran, K. Caenepeel-Mills, S.V. Edelman, R.O. Ford, J. Garcia,
— B.K. Montgomery (chair); Concomitant Conditions — R.E. Ratner
M. Hagerty, R.R. Henry, M. Hill, J. Horne, D. Leos, J. Matney, S. Mu-
(chair); Economic Evaluation — W.H. Herman (chair); Interventions —
daliar, G. Petersen, A. Pollard, W. Polonsky, S. Szerdi, J. Torio-Hurley, and
S.E. Kahn, J.V. Santiago (deceased), and J. Olefsky (chairs); Lifestyle Advisory
K. Vejvoda; St. Luke’s–Roosevelt Hospital — F.X. Pi-Sunyer,* J.E. Lee,† Group — R.R. Wing (chair); Outcomes — C. Saudek (chair); Outcomes Clas-
D.B. Allison, N. Agharanya, N.J. Aronoff, M. Baldo, S.T. Foo, S. Hagamen,
sification — R.E. Ratner (chair); Program CoordinatorsSubcommittee —
C. Pal, K. Parkes, M. Pena, and G.E.H. Van Wye; Indiana University —
M. Montez and K. Kramer (chairs); Protocol Oversight — R.F. Hamman
D.G. Marrero,* M.S. Kukman-Kelly,† Y.F. Dotson, S.E. Fineberg, J.C.
(chair); Publications and Presentations — W.C. Knowler (chair); Quality
Guare, A. Hadden, B. Hills, J.M. Ignaut, M.A. Jackson, M.S. Kirkman, K. Control — R.B. Goldberg (chair); Recruitment — W.Y. Fujimoto (chair);
Mather, G. McAtee, B.D. Porter, M.J. Prince, and M.L. Wheeler; Medstar Recruitment Coordinators Subcommittee — J. Charleston (chair); Reten- Research Institute — R.E. Ratner,* G. Youssef,† S. Shapiro,† A. Bonar, tion — R.R. Rubin (chair); Screening and Eligibility — R.F. Hamman
M. Bronsord, E. Brown, W.W. Cheatham, S. Cola, A. Comfort, G. Boggs,
(chair); Study Chair and Vice Chair — D.M. Nathan and R.F. Hamman.
C. Eagle, C. Evans, E. Gorman, R. Johnson, C. Levetan, T. Kellum, M. Lagarda, A.K. Nair, M.D. Passaro, and W. Phillips; UCLA Medical School REFERENCES
— M.F. Saad,* M. Budgett,† S. Fahmi,† S.D. Jinagouda,† B. Bernaba, S.L. Bodkin, V. Ciobanu, R. Commisso, C. Cosenza, T. Dinh, M. Gonzalez,
1. Harris MI, Flegal KM, Cowie CC, et al. Prevalence of diabetes, im-
A. Khan, D. Kumar, G. Lui, V. Mehta, A. Sharma, S. Soukiazian, K. Sza-
paired fasting glucose, and impaired glucose tolerance in U.S. adults: the
mos, A. Tramanian, A. Vargas, and N. Zambrana; Washington University,
Third National Health and Nutrition Examination Survey, 1988-1994. Di-
St. Louis — N.H. White,* A.S. Santiago,† S. Das,† A.L. Brown, S. Da-
gogo-Jack, E.B. Fisher, E. Hurt, T. Jones, M. Kerr, L. Ryder, J.V. Santiago
2. UK Prospective Diabetes Study (UKPDS) Group. Intensive blood-glu-
(deceased), and C. Wernimont; Johns Hopkins School of Medicine — C.D.
cose control with sulphonylureas or insulin compared with conventional
Saudek,* V.L. Bradley,† T. Fowlkes,† H. Joseph,† F.L. Brancati, J.B.
treatment and risk of complications in patients with type 2 diabetes
Charleston, J.M. Clark, K. Horak, D. Jiggetts, H. Mosley, R.R. Rubin, A.
(UKPDS 33). Lancet 1998;352:837-53. [Erratum, Lancet 1999;354:
Samuels, K.J. Stewart, L. Thomas, and P. Williamson; University of New Mexico School of Medicine — D.S. Schade,* K.S. Adams,† L.F. Atler, A. 3. Idem. Effect of intensive blood-glucose control with metformin on
Bland, D.A. Bowling, P.J. Boyle, M.R. Burge, L. Butler, J.L. Canady, L.
complications in overweight patients with type 2 diabetes (UKPDS 34).
Chai, K.M. Colleran, M. Guillen, Y. Gonzales, M. Gutierrez, D. Horn-
Lancet 1998;352:854-65. [Erratum, Lancet 1998;352:1557.]
beck, C. Johannes, P. Katz, C. King, E.N. Libby III, R. McCalman, D.A. 4. Harris MI, Eastman RC. Early detection of undiagnosed diabetes mel-
Montoya, A. Rassam, S. Rubinchik, and W. Senter; Albert Einstein Col-
litus: a US perspective. Diabetes Metab Res Rev 2001;16:230-6. lege of Medicine — H. Shamoon,* J.O. Brown,† J. Adames, E. Blanco, L. 5. Knowler WC, Narayan KMV, Hanson RL, et al. Preventing non-insu-
Cox, J.P. Crandall, H. Duffy, S. Engel, A. Friedler, T. Harroun, C.J.
lin-dependent diabetes. Diabetes 1995;44:483-8.
Howard-Century, S. Kloiber, N. Longchamp, D. Pompi, E. Violino, E.A. 6. The Diabetes Prevention Program Research Group. The Diabetes Pre-
Walker, J. Wylie-Rosett, E. Zimmerman, and J. Zonszein; University of
vention Program: design and methods for a clinical trial in the prevention
Pittsburgh — R.R. Wing,* M.K. Kramer,† S. Barr, M.A. Boraz, L. Clif-
of type 2 diabetes. Diabetes Care 1999;22:623-34.
ford, R. Culyba, M. Frazier, R. Gilligan, L. Harris, S. Harrier, W. Hender-
7. Pan XR , Li GW, Hu YH, et al. Effects of diet and exercise in preventing
son, S. Jeffries, G. Koenning, A.M. Kriska, K. Maholic, Q. Manjoo, M.
NIDDM in people with impaired glucose tolerance: the Da Qing IGT and
Mullen, A. Noel, T.J. Orchard, A. Otto, L.N. Semler, C. Smith, M. Smith,
Diabetes Study. Diabetes Care 1997;20:537-44.
V. Stapinski, J. Viteri, T. Wilson, K.V. Williams, and J. Zgibor; University 8. Tuomilehto J, Lindström J, Eriksson JG, et al. Prevention of type 2 di- of Hawaii — R.F. Arakaki,* R.W. Latimer,† N.K. Baker-Ladao, R.M.
abetes mellitus by changes in lifestyle among subjects with impaired glu-
Beddow, R. Braginsky, M. Calizar, L.M. Dias, N. Durham, D.A. Dupont,
cose tolerance. N Engl J Med 2001;344:1343-50.
L.L. Fukuhara, J. Inouye, M.K. Mau, K. Mikami, P. Mohideen, S.K. 9. Tataranni PA, Bogardus C. Changing habits to delay diabetes. N Engl
Odom, B. Sinkule-Kam, J.S. Tokunaga, R.U. Twiggs, C.Y. Wang, and J. Vi-
ta; Southwest American Indian Center for Diabetes Prevention — W.C. 10. Diabetes in America. 2nd ed. Bethesda, Md.: National Institute of Di-
Knowler,* N.J. Cooeyate,† M.A. Hoskin,† C.A. Percy,† K.J. Acton, V.L.
abetes and Digestive and Kidney Diseases, 1995. (NIH publication no. 95-
Andre, S. Antone, N.M. Baptisto, R. Barber, S. Begay, P.H. Bennett, M.B.
Benson, S. Beyale, E.C. Bird, B.A. Broussard, M. Chavez, T.S. Dacawyma,
11. Report of the Expert Committee on the Diagnosis and Classification
M.S. Doughty, R. Duncan, C. Edgerton, J.M. Ghahate, M. Glass, D. Goh-
of Diabetes Mellitus. Diabetes Care 1997;20:1183-97. 402 · N Engl J Med, Vol. 346, No. 6 · February 7, 2002 · www.nejm.org R E D U C I N G T H E I N C I D E N C E O F T Y P E 2 D I A B ET E S W I T H L I F E ST Y L E I N T E RV E N T I O N O R M ET F O R M I N 12. The Diabetes Prevention Program Research Group. The Diabetes Pre- 22. DeFronzo RA. Pharmacologic therapy for type 2 diabetes mellitus.
vention Program: baseline characteristics of the randomized cohort. Dia-
23. Trends in the prevalence and incidence of self-reported diabetes melli- 13. Idem. The Diabetes Prevention Program: recruitment methods and re-
tus — United States, 1980–1994. MMWR Morb Mortal Wkly Rep 1997;
sults. Control Clin Trials (in press). 14. The Food Guide Pyramid. Washington, D.C.: Department of Agricul- 24. Gillum RF, Mussolino ME, Madans JH. Diabetes mellitus, coronary
ture, Center for Nutrition Policy and Promotion, 1996. (Home and Gar-
heart disease incidence, and death from all causes in African American and
European American women: the NHANES I epidemiologic follow-up
15. Step by step: eating to lower your high blood cholesterol. Bethesda,
study. J Clin Epidemiol 2000;53:511-8.
Md.: National Heart, Lung, and Blood Institute Information Center, 1987. 25. Kuller LH, Velentgas P, Barzilay J, Beauchamp NJ, O’Leary DH, Sav- 16. Kriska AM, Caspersen CJ. Introduction to a collection of physical ac-
age PJ. Diabetes mellitus: subclinical cardiovascular disease and risk of in-
tivity questionnaires. Med Sci Sports Exerc 1997;29:Suppl:S5-S9.
cident cardiovascular disease and all-cause mortality. Arterioscler Thromb
17. Mayer-Davis EJ, Vitolins MZ, Carmichael SL, et al. Validity and repro-
ducibility of a food frequency interview in a multi-cultural epidemiology
26. Gabir MM, Hanson RL, Dabelea D, et al. The 1997 American Dia-
betes Association and 1999 World Health Organization criteria for hyper-
18. Lan KKG, Lachin JM. Implementation of group sequential logrank
glycemia in the diagnosis and prediction of diabetes mellitus. Diabetes Care
tests in a maximum duration trial. Biometrics 1990;46:759-70. 19. Lachin JM. Biostatistical methods: the assessment of relative risks. 27. de Vegt F, Dekker JM, Jager A, et al. Relation of impaired fasting and
postload glucose with incident type 2 diabetes in a Dutch population: the
20. Diggle PJ, Liang K-Y, Zeger SL. Analysis of longitudinal data. New
York: Oxford University Press, 1994. 21. Edelstein SL, Knowler WC, Bain RP, et al. Predictors of progression from impaired glucose tolerance to NIDDM: an analysis of six prospective studies. Diabetes 1997;46:701-10.
Copyright 2002 Massachusetts Medical Society.
POSTING PRESENTATIONS AT MEDICAL MEETINGS ON THE INTERNET
Posting an audio recording of an oral presentation at a medical meeting on theInternet, with selected slides from the presentation, will not be considered priorpublication. This will allow students and physicians who are unable to attend themeeting to hear the presentation and view the slides. If there are any questionsabout this policy, authors should feel free to call the Journal’s Editorial Offices.
N Engl J Med, Vol. 346, No. 6 · February 7, 2002 · www.nejm.org · 403
Facelifts As part of the aging process which happens to all of us sooner or late, our skin progressively loses its elasticity and our muscles tend to slacken. The stresses of daily life, effects of gravity and exposure to sun can be seen on our faces. The folds and smile lines deepen, the corners of the mouth droop, the jaw line sags and the skin of the neck becomes slack. Around the eyes,
QUALITY GRAND ROUNDS Series Editors: Robert M. Wachter, MD; Kaveh G. Shojania, MD; Sanjay Saint, MD, MPH; Amy J. Markowitz, JD; and Mark Smith, MD, MBA The Cognitive Psychology of Missed Diagnoses Donald A. Redelmeier, MD Cognitive psychology is the science that examines how people ence (in which people stop thinking when confronted with au- reason, formulate judgments, and make decis