Clinical Care/Education/Nutrition/Psychosocial Research

O R I G I N A L A R T I C L E

Reduction in the Incidence of Type

2 Diabetes With the Mediterranean Diet
Results of the PREDIMED-Reus nutrition intervention randomized trial
JORDI SALAS-SALVADÓ, MD, PHD1,2 MARIA ISABEL COVAS, DPHARM, PHD2,6

T
he increasing incidence of type 2 di-
MONICA BULLÓ, BSC, PHD1,2 DOLORES CORELLA, DPHARM, PHD2,7 abetes throughout the world, closely
NANCY BABIO, BSC, PHD1,2 FERNANDO ARÓS, MD, PHD2,8 linked to westernized dietary pat-
MIGUEL ÁNGEL MARTÍNEZ-GONZÁLEZ, MD,
2,3
VALENTINA RUIZ-GUTIÉRREZ, DPHARM, PHD9 terns, physical inactivity, and raising rates
PHD EMILIO ROS, MD, PHD2,10 of obesity, is a challenging health prob-
NÚRIA IBARROLA-JURADO, RD1,2 FOR THE PREDIMED STUDY
JOSEP BASORA, MD1,2,4 lem. Lifestyle changes are effective mea-
INVESTIGATORS sures to prevent diabetes, and weight loss
RAMON ESTRUCH, MD, PHD2,5
is the main predictor of success (1). Five

Downloaded from http://diabetesjournals.org/care/article-pdf/34/1/14/606990/zdc00111000014.pdf by guest on 30 October 2024

clinical trials that examined the effects of
OBJECTIVE — To test the effects of two Mediterranean diet (MedDiet) interventions versus
reputedly healthy, energy-restricted diets
a low-fat diet on incidence of diabetes. together with increased physical activity
in individuals with impaired glucose tol-
RESEARCH DESIGN AND METHODS — This was a three-arm randomized trial in erance, a prediabetic stage, showed risk
418 nondiabetic subjects aged 55– 80 years recruited in one center (PREDIMED-Reus, north- reductions between 30 and 70% (2– 6).
eastern Spain) of the Prevención con Dieta Mediterránea [PREDIMED] study, a large nutrition The results of these studies provide con-
intervention trial for primary cardiovascular prevention in individuals at high cardiovascular vincing evidence that lifestyle modifica-
risk. Participants were randomly assigned to education on a low-fat diet (control group) or to one tion reduces the incidence of diabetes
of two MedDiets, supplemented with either free virgin olive oil (1 liter/week) or nuts (30 g/day).
among high-risk individuals. In four of
Diets were ad libitum, and no advice on physical activity was given. The main outcome was
diabetes incidence diagnosed by the 2009 American Diabetes Association criteria. these studies (2–5), diabetes rates de-
creased in relation to substantial reduc-
RESULTS — After a median follow-up of 4.0 years, diabetes incidence was 10.1% (95% CI tions in body weight, whereas in the
5.1–15.1), 11.0% (5.9 –16.1), and 17.9% (11.4 –24.4) in the MedDiet with olive oil group, the Indian trial (6) lifestyle intervention was
MedDiet with nuts group, and the control group, respectively. Multivariable adjusted hazard successful despite no weight loss. Obser-
ratios of diabetes were 0.49 (0.25– 0.97) and 0.48 (0.24 – 0.96) in the MedDiet supplemented vational studies have also shown that di-
with olive oil and nuts groups, respectively, compared with the control group. When the two ets rich in vegetables and low in red meat
MedDiet groups were pooled and compared with the control group, diabetes incidence was and whole-fat dairy products are associ-
reduced by 52% (27– 86). In all study arms, increased adherence to the MedDiet was inversely
associated with diabetes incidence. Diabetes risk reduction occurred in the absence of significant
ated with a decreased risk of diabetes,
changes in body weight or physical activity. whereas dietary patterns rich in red
meats, processed foods, refined grains,
CONCLUSIONS — MedDiets without calorie restriction seem to be effective in the preven- and sweets increase diabetes risk (7).
tion of diabetes in subjects at high cardiovascular risk. The traditional Mediterranean diet
(MedDiet), characterized by high con-
Diabetes Care 34:14–19, 2011 sumption of vegetables, legumes, grains,
fruits, nuts, and olive oil, moderate con-
sumption of fish and wine, and low con-
sumption of red and processed meat and
● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●
whole-fat dairy products, is widely recog-
nized as a healthy dietary pattern (8). Two
From the 1Human Nutrition Unit, Hospital Universitari de Sant Joan, Departament de Bioquímica i Biotec-
nologia, Institut d’Investigació Sanitària Pere Virgili, Universitat Rovira i Virgili, Reus, Spain; the 2CIBER
prospective studies from Southern Eu-
Fisiopatologia de la Obesidad y Nutrición, Instituto de Salud Carlos III, Madrid, Spain; the 3Department rope suggested a lower incidence of dia-
of Preventive Medicine and Public Health, University of Navarra, Pamplona, Spain; the 4SAP Reus- betes with increasing adherence to the
Altebrat, Institut Català de la Salut, Reus, Spain; the 5Department of Internal Medicine, Institut d⬘Inves- MedDiet in previously healthy individu-
tigacions Biomèdiques August Pi Sunyer, Hospital Clínic, Barcelona, Spain; the 6Cardiovascular Risk and
Nutrition Research Group, Institut Municipal d’Investigació Mèdica, Barcelona, Spain; the 7Department
als (9) or myocardial infarction survivors
of Preventive Medicine and Public Health, University of Valencia, Valencia, Spain; the 8Department of (10). Recently, a clinical trial showed that,
Cardiology, University Hospital Txagorritxu, Vitoria, Spain; the 9Instituto de la Grasa, Consejo Superior compared with a low-fat diet, a MedDiet
de Investigaciones Científicas, Seville, Spain; and the 10Lipid Clinic, Service of Endocrinology and Nu- allowed better glycemic control and de-
trition, Institut d⬘Investigacions Biomèdiques August Pi Sunyer, Hospital Clínic, Barcelona, Spain. layed the need for antidiabetes drug treat-
Corresponding author: Jordi Salas-Salvadó, [email protected].
Received 6 July 2010 and accepted 28 September 2010. Published ahead of print at http://care.diabetesjournals. ment in patients with newly diagnosed
org on 7 October 2010. DOI: 10.2337/dc10-1288. Clinical trial reg. no. ISRCTN35739639, ISRCTN.org. diabetes (11). However, the role of the
© 2011 by the American Diabetes Association. Readers may use this article as long as the work is properly MedDiet in the prevention of diabetes has
cited, the use is educational and not for profit, and the work is not altered. See http://creativecommons. not been tested in a clinical trial.
org/licenses/by-nc-nd/3.0/ for details.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby We conducted a randomized con-
marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. trolled trial to compare the effect on dia-

14 DIABETES CARE, VOLUME 34, NUMBER 1, JANUARY 2011 care.diabetesjournals.org

 Salas-Salvadó and Associates

betes incidence of three non– calorie- of olive oil for cooking and dressing, 2) unsaturated fatty acid (MUFA)-to-
restricted nutritional interventions: a increased consumption of fruit, vegeta- saturated fatty acid (SFA) ratio, 3) high
low-fat diet (control diet), a MedDiet en- bles, legumes, and fish, 3) reduction in olive oil consumption (ⱖ20 g/1,000 kcal/
riched with virgin olive oil, and a MedDiet total meat consumption, recommending day), 4) high nut consumption (ⱖ10
enriched with mixed nuts. white meat instead of red or processed g/1,000 kcal/day), 5) high dietary fiber in-
meat, 4) preparation of homemade sauce take (ⱖ14 g/1,000 kcal/day), 6) substan-
RESEARCH DESIGN AND with tomato, garlic, onion, and spices tial weight loss (ⱖ5% of initial body
METHODS — The Prevención con with olive oil to dress vegetables, pasta, weight), and 7) high physical activity
Dieta Mediterránea (PREDIMED) study is rice, and other dishes, 5) avoidance of (ⱖ395 kcal/day, the top tertile). Changes
a multicenter, randomized, parallel group butter, cream, fast food, sweets, pastries, in weight and physical activity were not
primary prevention trial conducted in and sugar-sweetened beverages, and 6) in intervention goals but were assessed be-
Spain to assess the effects of two Med- alcohol drinkers, moderate consumption cause of their well-known association
Diets, supplemented with either extra vir- of red wine. with diabetes.
gin olive oil or mixed nuts, versus a low- At inclusion and quarterly thereafter, The primary outcome was new-onset
fat control diet on cardiovascular and dietitians administered both individual diabetes, diagnosed according to Ameri-
other chronic disease outcomes in indi- interviews and group sessions, separately can Diabetes Association criteria (15),

Downloaded from http://diabetesjournals.org/care/article-pdf/34/1/14/606990/zdc00111000014.pdf by guest on 30 October 2024

viduals at high cardiovascular risk. Full for each group. Sessions consisted of in- namely fasting plasma glucose ⱖ7.0
details of the PREDIMED protocol have formative talks and delivery of written mmol/l or 2-h plasma glucose ⱖ11.1
been published elsewhere (12) and are material with elaborate descriptions of mmol/l after a 75-g oral glucose load,
available at www.predimed.org and www. typical foods for each dietary pattern, sea- measured yearly. A second test using the
predimed.es. Recruitment took place be- sonal shopping lists, meal plans, and rec- same criteria was required for confirma-
tween October 2003 and June 2008, ipes. Participants assigned to MedDiet tion. Case ascertainment was done by the
including 7,232 participants randomly as- groups were given free allotments of ei- PREDIMED Clinical Event Committee,
signed to the three interventions. ther virgin olive oil (1 liter/week) or whose members were blinded to inter-
In only one of the PREDIMED centers mixed nuts (30 g/day). Participants as- vention group. When diabetes was diag-
(PREDIMED-Reus) was a yearly oral glu- signed to the low-fat diet received recom- nosed, participants and their primary care
cose tolerance test (OGTT) in nondiabetic mendations to reduce all types of fat, from physicians were informed and no further
participants part of the protocol. The both animal and vegetable sources, but no OGTTs were scheduled. Every effort was
present report represents a nested sub- free foods. Instead, to encourage adher- made to retain participants and to ascer-
study with the aim of assessing the effects ence, at quarterly visits they were given tain vital status, including telephone calls
of the three interventions on the inci- small gifts, such as oil dispensers, aprons, and home visits by PREDIMED investiga-
dence of diabetes using a yearly OGTT as shopping bags, or cookbooks. Energy re- tors if necessary.
a diagnostic tool. The local institutional striction was not advised, nor was physi-
review board approved the study proto- cal activity promoted. Statistical analysis
col, and all participants provided written At baseline and at each annual visit Comparisons among groups for qualita-
informed consent. we administered 1) a short-questionnaire tive variables were done with the ␹2 test.
Candidates for the study were com- about lifestyle variables, medical condi- We fitted Cox regression models to assess
munity-dwelling men aged 55– 80 years tions, and medication use, 2) a 14-item the relative risk of diabetes by allocation
and women aged 60 – 80 years without questionnaire of adherence to the Med- group, estimating hazard ratios and 95%
prior cardiovascular disease but having at Diet (12), 3) a 137-item validated food CIs. The time variable was the interval be-
least three cardiovascular risk factors, frequency questionnaire (13), and 4) the tween randomization and the date of last
namely smoking, hypertension, dyslipi- validated Spanish version of the Minne- follow-up, death, or diabetes diagnosis,
demia, overweight (BMI ⱖ25 kg/m2), and sota Leisure-Time Physical Activity ques- whichever occurred first. Participants
family history of premature cardiovascu- tionnaire (14). Staff involved in collecting who were free of diabetes or who were lost
lar disease (ⱕ55 years in men and ⱕ60 questionnaires and physical measures during follow-up were censored at the
years in women). Participants with prev- were unblinded to intervention group. date of the last visit. The assumption of
alent diabetes were excluded from the Energy and nutrient intakes were calcu- proportional hazards was tested using
present analysis. Other exclusion crite- lated from Spanish food composition ta- time-dependent covariates. In all analy-
ria were any severe chronic illness, al- bles as described previously (12). At ses, we fitted a Cox regression model ad-
cohol or drug abuse, BMI ⱖ40 kg/m2, yearly visits, weight was recorded, sam- justed for age and sex. In a subsequent
and history of allergy or intolerance to ples of fasting blood were taken, and an model, we adjusted additionally for base-
olive oil or nuts (12). OGTT was scheduled. Plasma glucose line energy intake, BMI, waist circumfer-
A behavioral intervention promoting concentrations were centrally analyzed by ence, physical activity, smoking status,
the MedDiet was implemented, as de- the glucose-oxidase method. Laboratory fasting serum glucose, use of lipid-
scribed previously (12). In brief, on the technicians were blinded to intervention lowering drugs, MedDiet score, and
basis of the initial assessment of individ- group. weight change during the study. The last
ual scores of adherence using a 14-item We assessed the proportion of partic- model was repeated after merging the two
questionnaire, dietitians gave personal- ipants in each group attaining prespecific MedDiet groups into a single category.
ized dietary advice to participants ran- lifestyle goals on at least 50% of the fol- Multiplicative interaction (effect modifi-
domly assigned to both MedDiets, with low-up visits. Goals included 1) improved cation) between the intervention (“Medi-
instructions directed to scale up the score, adherence to the MedDiet (ⱖ10 points in terranean diets,” i.e., the two groups
including, among others, 1) abundant use the 14-point score), 2) a high (ⱖ2) mono- merged into one category) and age, sex,

care.diabetesjournals.org DIABETES CARE, VOLUME 34, NUMBER 1, JANUARY 2011 15

MedDiet and diabetes incidence

BMI, and baseline fasting glucose were as- Table 1—Characteristics of the study population at baseline
sessed using the likelihood ratio test for
multiplicative product terms introduced MedDiet with MedDiet with Control diet
in fully adjusted Cox models. Kaplan- VOO group nuts group group
Meier survival curves were plotted to es-
timate the probability of remaining free of n 139 145 134
diabetes during follow-up. Analyses were Age (years) 67.4 ⫾ 6.1 66.6 ⫾ 5.8 67.8 ⫾ 6.1
based on the intention-to-treat principle. Male sex (%) 40 47 38
All P values are two-tailed at the ⬍0.05 Current smoker (%) 11 15 15
level. Statistical analysis were performed Weight (kg) 75.3 ⫾ 10.3 76.1 ⫾ 10.5 76.2 ⫾ 11.3
with SPSS (version 17.0; SPSS, Chicago, BMI (kg/m2) 29.7 ⫾ 3.3 29.6 ⫾ 3.1 30.0 ⫾ 3.3
IL) software. Waist circumference(cm) 101.1 ⫾ 8.6 100.3 ⫾ 8.5 102.2 ⫾ 9.4
Leisure-time physical activity (kcal/day) 372 ⫾ 280 389 ⫾ 267 338 ⫾ 209
RESULTS — Of 1,125 eligible candi- Plasma biomarkers
dates, 870 fulfilled the inclusion criteria LDL cholesterol (mmol/l) 3.7 ⫾ 0.9 3.5 ⫾ 0.8 3.7 ⫾ 0.9
and entered the trial. Of these, 452 were HDL cholesterol (mmol/l) 1.5 ⫾ 0.3 1.5 ⫾ 0.4 1.5 ⫾ 0.4

Downloaded from http://diabetesjournals.org/care/article-pdf/34/1/14/606990/zdc00111000014.pdf by guest on 30 October 2024

excluded because of a prior diagnosis of Triglycerides (mmol/l) 1.5 ⫾ 0.6 1.6 ⫾ 0.8 1.6 ⫾ 0.8
diabetes. A total of 418 nondiabetic vol- Non–HDL cholesterol (mmol/l) 4.3 ⫾ 0.9 4.2 ⫾ 0.9 4.4 ⫾ 1.0
unteers were randomly assigned into the Fasting glucose (mmol/l) 5.5 ⫾ 0.8 5.5 ⫾ 0.9 5.5 ⫾ 0.9
three groups (supplementary Fig. 1, avail- 2-h postload glucose (mmol/l)* 7.1 ⫾ 2.6 6.9 ⫾ 2.4 7.4 ⫾ 2.9
able in an online appendix at http://care. Fasting insulin (␮U/ml)† 5.8 ⫾ 3.6 5.4 ⫾ 3.2 6.2 ⫾ 4.4
diabetesjournals.org/cgi/content/full/ HOMA-IR† 1.41 ⫾ 0.87 1.34 ⫾ 0.87 1.60 ⫾ 1.17
dc10-1288/DC1). The first participant Medication use (%)
entered the study in October 2003 and Lipid-lowering drugs 46.8 47.6 40.3
the last one in June 2008. Most partici- Antihypertensive medication 82.0 82.1 78.4
pants (98.8%) were enrolled for at least 1 Estrogen replacement therapy 2.4 0.0 2.4
year, 88.3% were enrolled for ⱖ3 years, Energy, food, and nutrient intake
and 19.9% were enrolled for ⱖ6 years. Total energy (kcal/day) 2,320 ⫾ 579 2,365 ⫾ 570 2,314 ⫾ 580
The median follow-up was 4.0 years (in- Carbohydrate (% energy) 41 ⫾ 6 40 ⫾ 6 41 ⫾ 7
terquartile range, 3.0 –5.0). Attrition rates Protein (% energy) 16 ⫾ 3 16 ⫾ 2 16 ⫾ 2
were low and were almost exclusively due Fat (% energy) 41 ⫾ 6 41 ⫾ 6 40 ⫾ 7
to major disease events or death. MUFA-to-SFA ratio 1.9 ⫾ 0.4 2.0 ⫾ 0.4 1.9 ⫾ 0.5
Table 1 shows baseline characteristics Total fiber (g/day) 23.7 ⫾ 7.6 23.6 ⫾ 8.0 23.0 ⫾ 7.7
of participants according to intervention Olive oil (g/day) 41.2 ⫾ 17.7 42.0 ⫾ 16.5 40.1 ⫾ 20.4
arm. The mean age was 67.3 years, and Nuts (g/day) 13.1 ⫾ 15.0 14.4 ⫾ 15.4 9.3 ⫾ 12.1
58.4% of participants were women. The Vegetables (g/day) 309 ⫾ 129 310 ⫾ 141 286 ⫾ 117
groups were well balanced with respect to Fruits (g/day) 298 ⫾ 185 315 ⫾ 164 286 ⫾ 168
most relevant variables. However, partic- Legumes (g/day) 18 ⫾ 7.9 19 ⫾ 9.0 18 ⫾ 8.4
ipants in the control group used fewer lip- Cereals (g/day) 248 ⫾ 98 245 ⫾ 99 251 ⫾ 105
id-lowering drugs and had a lower Red meat and meat products (g/day) 80 ⫾ 44 86 ⫾ 46 84 ⫾ 47
MedDiet score than the other two groups. Milk and dairy products (g/day) 355 ⫾ 201 348 ⫾ 183 346 ⫾ 207
Participants refused the OGTT on 17% of Seafood (g/day) 107 ⫾ 42 104 ⫾ 43 99 ⫾ 41
the scheduled occasions, and 41 (9.8%) Alcohol (g/day) 8 ⫾ 11 11 ⫾ 14 9 ⫾ 12
of them had none performed. Red wine (ml/day) 57 ⫾ 95 75 ⫾ 101 64 ⫾ 89
The diets were well tolerated. Up to Score of adherence to the MedDiet 8.4 ⫾ 1.9 8.4 ⫾ 1.9 7.9 ⫾ 1.9
3% of participants in each treatment arm Data are means ⫾ SD or %. VOO, virgin olive oil. *Data were available for 263 participants. †Data were
reported difficulties in following the pre- available for 307 participants.
scribed diets, which were solved in all
cases by the dietitians through individual
counsel, negotiation, and small diet ad- gumes, and fish, and a MedDiet score the nut diet group, and ⫺0.6 ⫾ 4.3 kg for
justments. The number of participants in ⱖ10 were achieved more frequently by the low-fat diet group (P ⫽ 0.74 for the
each group who reported improved participants allocated to the two MedDi- comparison between groups). Likewise,
bowel motions (5.2– 8.1%) was approxi- ets than by those in the control group, physical activity changes were similar in
mately twice the number of those who whereas a small proportion of partici- the three groups: ⫺17.4 ⫾ 336, ⫺58.8 ⫾
complained of newly developed constipa- pants in each group reached substantial 297, and ⫺35.8 ⫾ 257 kcal/day, respec-
tion (3.3–3.7%). Supplementary Table 1 dietary fiber intakes. Only 21% of partic- tively (P ⫽ 0.50). As shown in supplemen-
(available in an online appendix) shows ipants in the control group achieved the tary Table 1, at the end of the study,
the proportion of participants in each goal specific for this group of total fat in- participants sustained weight loss ⬎5% to a
group who attained prespecified goals take ⬍35% of energy. Weight changes similar extent in the three groups, and a
during the trial. The goals of higher among the 418 participants at the end of lower proportion of those in the control
MUFA-to-SFA ratios, higher intakes of to- follow-up were ⫺0.2 ⫾ 4.6 kg for the group were in the top tertile of physical ac-
tal olive oil, nuts, fruit and vegetables, le- olive oil diet group, ⫺0.6 ⫾ 4.2 kg for tivity. There were few changes in medica-

16 DIABETES CARE, VOLUME 34, NUMBER 1, JANUARY 2011 care.diabetesjournals.org

 Salas-Salvadó and Associates

tion during the trial. Hypolipidemic and

antihypertensive therapy was initiated by
15.6 and 6.1% of participants and discon-
tinued by 5.4 and 3.2%, respectively, with a
similar distribution among groups.
During the study, 54 individuals de-
veloped new-onset diabetes. Supplemen-
tary Table 2 (available in an online
appendix) shows that the rate per 1,000
person-years of diabetes incidence was
24.6 (95% CI, 13.5– 40.8) for the Med-
Diet with virgin olive oil group, 26.8
(15.3– 43.0) for the MedDiet with nuts
group, and 46.6 (30.1– 68.5) for the con-
trol group. Cumulative incidence of dia-
betes was 10.1 (5.1–15.1) in the MedDiet

Downloaded from http://diabetesjournals.org/care/article-pdf/34/1/14/606990/zdc00111000014.pdf by guest on 30 October 2024

with olive oil group and 11.0 (5.9 –16.1)
in the MedDiet with nuts group, whereas
it was 17.9 (11.4 –24.4) in the control
group. Figure 1 shows that cumulative di-
abetes-free survival was lower in the con-
trol group compared with both MedDiet
groups. After adjustment for various con-
founders, incident diabetes was reduced
by 51% in the MedDiet with olive oil
group and by 52% in the MedDiet with Figure 1—Cumulative diabetes free-survival by group of intervention. Cox regression models
nuts group in comparison with the con- with outcome of diabetes onset and exposure to MedDiet intervention group vs. control diet group,
trol group (Table 2). Thus, when the two adjusted by sex, age, baseline energy intake, BMI, waist circumference, physical activity, smoking
MedDiet groups were merged into a sin- status, fasting serum glucose, use of lipid-lowering drugs, Mediterranean diet score, and weight
gle category in a similarly adjusted model, change during the study. a, MedDiet and virgin olive oil group; b, MedDiet and nuts group; c,
diabetes incidence was reduced by 52%. control diet group.
In multivariable regression models, sex,
age, baseline obesity, and baseline fasting riched with high-fat foods of vegetable after a median follow-up of 4.0 years. Di-
glucose were unrelated to outcomes. origin decreased the incidence of diabetes abetes rates were reduced by 51 and 52%
Diabetes incidence was lower in par- in individuals at high cardiovascular risk by the consumption of MedDiets supple-
ticipants assigned to the two MedDiets
(considered together), who attained ⱖ4
of the 7 prespecified goals or achieved a Table 2—Hazard ratios (95% CI) of diabetes by intervention group
MedDiet score ⱖ10 (supplementary Fig.
2, available in an online appendix). Thus, MedDiet with VOO MedDiet with nuts vs. Both MedDiets vs.
6.3% of participants in the MedDiet vs. control diet control diet control diet
groups developed diabetes if they attained
ⱖ4 goals compared with 15.0% of those Crude model 0.53 (0.27–1.09) 0.58 (0.31–1.10) 0.55 (0.32–0.95)
who reached ⬍4 goals (P ⫽ 0.02). Rates Age- and sex-adjusted model 0.52 (0.27–1.00) 0.55 (0.29–1.00) 0.53 (0.31–0.92)
depending on attainment or not of a high Multivariate adjusted model* 0.49 (0.25–0.97) 0.48 (0.24–0.96) 0.48 (0.27–0.86)
score (ⱖ10) of adherence to the MedDiet Sex†
were 9.4 vs. 20.6% (P ⫽ 0.07), respec- Male 0.48 (0.16–1.46) 0.65 (0.21–2.00) 0.55 (0.21–1.43)
tively, in the control group. Changes in Female 0.47 (0.19–1.17) 0.32 (0.11–0.93) 0.40 (0.18–0.90)
weight or physical activity did not differ Age†
among participants in each intervention ⱕ67 years 0.50 (0.18–1.39) 0.65 (0.26–1.61) 0.58 (0.26–1.31)
arm developing or not developing diabe- ⬎67 years 0.26 (0.08–0.83) 0.27 (0.07–0.98) 0.26 (0.09–0.76)
tes at the end of the study. In the control BMI†
group, however, subjects developing dia- ⱕ30 kg/m2 0.56 (0.21–1.49) 0.52 (0.19–1.41) 0.54 (0.24–1.22)
betes had sustained a mean weight gain of ⬎30 kg/m2 0.50 (0.18–1.42) 0.62 (0.22–1.76) 0.56 (0.23–1.43)
1.8 ⫾ 3.3 kg, whereas those remaining Fasting glucose†
diabetes-free had an average weight loss ⱕ6.1 mmol/l 0.44 (0.16–1.25) 0.60 (0.24–1.50) 0.53 (0.23–1.20)
of 1.1 ⫾ 4.4 kg, a nearly significant dif- ⬎6.1 mmol/l 0.29 (0.09–0.95) 0.39 (0.11–1.37) 0.32 (0.11–0.98)
ference (P ⫽ 0.10). Cox regression models to assess the relative risk of diabetes by allocation group, estimating the hazard ratios
(95% CI), were performed. Pinteraction (MedDiet ⫻ sex) ⫽ 0.496; Pinteraction (MedDiet ⫻ age) ⫽ 0.195;
Pinteraction (MedDiet ⫻ BMI) ⫽ 0.592; Pinteraction (MedDiet ⫻ fasting glucose) ⫽ 0.932. VOO, virgin olive oil.
CONCLUSIONS — In this nutrition *Adjusted for sex, age, baseline energy intake, BMI, waist circumference, physical activity, smoking status,
intervention study we found that a non– fasting serum glucose, use of lipid-lowering drugs, Mediterranean diet score, and weight changes during the
calorie-restricted traditional MedDiet en- study. †Adjusted for the same variables as in footnote *, except for the variable of interest.

care.diabetesjournals.org DIABETES CARE, VOLUME 34, NUMBER 1, JANUARY 2011 17

MedDiet and diabetes incidence

mented with virgin olive oil or mixed of weight loss. In the same study (12), a ful tool for preventing diabetes. Because
nuts, respectively, compared with a con- reduction in circulating inflammatory bi- other studies have shown that the benefit
trol diet consisting of advice on a low-fat omarkers was observed in the two Med- of lifestyle modification in reducing dia-
diet. When the results of the two MedDiet Diet groups. Because chronic low-grade betes risk extends beyond the termination
groups were merged, risk reduction was inflammation is a pathogenetic factor in of active intervention (23–25), education
52%. These results extend those of prior diabetes, synergy among the anti- of the population on the MedDiet might
studies showing that lifestyle interven- inflammatory properties of the MedDiet be a safe public health approach to delay
tions can substantially reduce the inci- and those specific to virgin olive oil (18) or prevent development of diabetes as
dence of diabetes in individuals at high and nuts (19) might also be relevant to well as that of other prevalent chronic dis-
risk (2– 6). However, in these studies, the diabetes risk reduction. Regarding nuts, eases (22). Further research is needed to
interventions consisted of advice on a cal- reports from large prospective studies elucidate the mechanisms leading to dia-
orie-restricted diet plus physical activity suggest that usual intake relates inversely betes risk reduction independently of
and, except for one study (6), weight loss to future diabetes risk in women (20) but weight loss.
was a major driving force in reducing the not in men (21). No such data are avail-
incidence of diabetes. Of note, in our able for olive oil consumption and risk of
study, diabetes risk reduction occurred in diabetes. However, a former report of the Acknowledgments — This study was funded,

Downloaded from http://diabetesjournals.org/care/article-pdf/34/1/14/606990/zdc00111000014.pdf by guest on 30 October 2024

in part, by the Spanish Ministry of Health
the absence of significant changes in body PREDIMED trial (16) showed that, com- (Instituto de Salud Carlos III) (projects
weight or physical activity. pared with the control diet, both Med- PI051839, PI070240, PI1001407, G03/140,
Our estimate of the magnitude of the Diets, particularly the nut-enriched diet, and RD06/0045), Fondo Europeo de Desar-
effect of the MedDiet can be viewed as had a favorable effect on metabolic syn- rollo Regional, and the Public Health Division
conservative because the data were ana- drome status after intervention for 1 year. of the Department of Health of the Autono-
lyzed by intention to treat, even though The fact that in our study participants in mous Government of Catalonia in collabora-
some participants in all treatment arms any treatment arm who were more com- tion with Merck Sharp & Dohme. The
might not have been fully compliant with pliant with the MedDiet had two to three Fundación Patrimonio Comunal Olivarero
the intended dietary modifications. As de- times lower incidence of diabetes than and Hojiblanca SA (Málaga, Spain), California
scribed in other reports of the PREDIMED those with lesser scores supports a bene- Walnut Commission (Sacramento, CA),
Borges SA (Reus, Spain), and Morella Nuts SA
study (12,16), a sizable proportion of par- ficial effect of the whole MedDiet pattern. (Reus, Spain) donated the olive oil, walnuts,
ticipants in the control group, despite be- There are some limitations to our almonds, and hazelnuts, respectively, used in
ing advised to follow the low-fat diet, did study. First, the Mediterranean cohort the study.
not substantially reduce total fat intake studied was older in age and at high risk J.S.-S. has received research funding from
(supplementary Table 1), because of a for cardiovascular disease. The generali- the International Nut Council (Reus, Spain)
long-lasting preference for using olive oil zation of our findings to younger and/or and is a nonpaid member of the Scientific Ad-
in the kitchen and at the table in Mediter- healthier individuals from other geo- visory Board of the International Nut Council.
ranean cultures. In fact, diabetes risk was graphical locations is uncertain. Never- E.R. has received research funding from the
reduced to a similar extent in participants theless, it is plausible that the beneficial California Walnut Commission (Sacramento,
of all treatment arms who reported higher effect of the MedDiet on diabetes risk may CA) and is a nonpaid member of its Scientific
Advisory Committee.
scores of adherence to the MedDiet (sup- be reproduced in other populations, as it No other potential conflicts of interest rele-
plementary Fig. 2). has been shown for all-cause mortality, vant to this article were reported.
Our results are consistent with prior cardiovascular disease incidence, and None of the funding sources played a role in
evidence suggesting a protective effect of cancer mortality in U.S. populations (22). the design, collection, analysis, or interpreta-
the MedDiet against diabetes (7,9 –11). Second, the lifestyle score used in our tion of the data or in the decision to submit the
Characteristically, the MedDiet is a high- study to determine whether changes in manuscript for publication. CIBER Fisiopato-
fat, high-unsaturated fat dietary pattern, a dietary goals related to diabetes incidence logia de la Obesidad y Nutrición is an initiative
feature that was maximized in our study could not reflect the totality of dietary of Instituto de Salud Carlos III, Spain.
by the free provision of virgin olive oil changes, thus making difficult to show J.S-S. conceived the study concept and de-
(rich in MUFAs) and mixed nuts (rich in significant differences among interven- sign, obtained funding, acquired data, ana-
lyzed and interpreted data, wrote the
MUFAs and polyunsaturated fatty acids) tions. Third, some participants did not manuscript, and reviewed/edited the manu-
to participants in the MedDiet groups. As undergo an OGTT, thus limiting an even- script. M.B. conceived the study concept and
suggested by the known associations tual diagnosis of diabetes to a fasting design, acquired data, analyzed and inter-
among subtypes of dietary fat and diabe- blood glucose ⱖ7.0 mmol/l confirmed by preted data, and reviewed/edited the manu-
tes risk (17), the increased unsaturated fat a second test, which might have falsely script. N.B. acquired data, analyzed and
load of our MedDiets was probably in- lowered overall incident rates. Finally, interpreted data, performed statistical analy-
strumental in achieving diabetes risk our sample size was relatively small, be- sis, wrote the manuscript, and reviewed/
reduction. cause the results are based on fewer than edited the manuscript. M.A.M.-G. conceived
The results of a prior PREDIMED 55 incident cases, and the CIs for our es- the study concept and design, obtained fund-
study report (12) support the protective timates are wide. Longer follow-up of the ing, analyzed and interpreted data, performed
statistical analysis, wrote the manuscript, and
role of olive oil and nuts against diabetes PREDIMED cohort may eventually pro- reviewed/edited the manuscript. N.I.-J. ac-
risk, as both MedDiets were associated vide stronger evidence of diabetes pre- quired data, analyzed and interpreted data,
with improved fasting glucose in diabetic vention by the MedDiet. and reviewed/edited the manuscript. J.B. con-
participants and decreased insulin resis- In summary, the results show that a ceived the study concept and design, acquired
tance in those without diabetes after a non– energy-restricted traditional Med- data, and reviewed/edited the manuscript.
3-month follow-up, again in the absence Diet high in unsaturated fat can be a use- R.E. conceived the study concept and design,

18 DIABETES CARE, VOLUME 34, NUMBER 1, JANUARY 2011 care.diabetesjournals.org

 Salas-Salvadó and Associates

obtained funding, analyzed and interpreted and metformin prevent type 2 diabetes in abetes Care 2008;31(Suppl. 1):S55–S60
data, wrote the manuscript, and reviewed/ Asian Indian subjects with impaired glu- 16. Salas-Salvadó J, Fernández-Ballart J, Ros
edited the manuscript. M.I.C. conceived the cose tolerance (IDPP-1). Diabetologia E, Martínez-González MA, Fitó M, Es-
study concept and design and reviewed/edited 2006;49:289 –297 truch R, Corella D, Fiol M, Gómez-Gracia
the manuscript. D.C. conceived the study con- 7. Kastorini CM, Panagiotakos DB. Dietary E, Arós F, Flores G, Lapetra J, Lamuela-
cept and design and reviewed/edited the patterns and prevention of type 2 diabe- Raventós R, Ruiz-Gutiérrez V, Bulló M,
manuscript. F.A. conceived the study concept tes; from research to clinical practice; a Basora J, Covas MI, PREDIMED Study In-
and design and reviewed/edited the manu- systematic review. Curr Diabetes Rev vestigators. Effect of a Mediterranean diet
script. V.R.-G. conceived the study concept 2009;5:221–227 supplemented with nuts on metabolic
and design, obtained funding, and reviewed/ 8. Martínez-González MA, Bes-Rastrollo M, syndrome status: one-year results of the
edited the manuscript. E.R. conceived the Serra-Majem L, Lairon D, Estruch R, PREDIMED randomized trial. Arch Intern
study concept and design, obtained funding, Trichopoulou A. Mediterranean food pat- Med 2008;168:2449 –2458
and reviewed/edited the manuscript. tern and the primary prevention of 17. Riserus U, Willett WC, Hu FB. Dietary
We thank the participants for their enthusi- chronic disease: recent developments. fats and prevention of type 2 diabetes.
astic collaboration, the PREDIMED personnel Nutr Rev 2009;67(Suppl. 1):S111–S116 Progr Lipid Res 2009;48:44 –51
for excellent assistance, and the personnel of 9. Martínez-González MA, de la Fuente-Ar- 18. Covas MI, Konstantinidou V, Fitó M. Ol-
all affiliated primary care centers. rillaga C, Nunez-Cordoba JM, Basterra- ive oil and cardiovascular health. J Car-
Gortari FJ, Beunza JJ, Vazquez Z, Benito S, diovasc Pharmacol 2009;54:477– 482

Downloaded from http://diabetesjournals.org/care/article-pdf/34/1/14/606990/zdc00111000014.pdf by guest on 30 October 2024

Tortosa A, Bes-Rastrollo M. Adherence to 19. Ros E. Nuts and novel biomarkers of car-
References Mediterranean diet and risk of developing diovascular disease. Am J Clin Nutr 2009;
1. American Diabetes Association, Bantle JP, diabetes: prospective cohort study. BMJ 89:1649S–1656S
Wylie-Rosett J, Albright AL, Apovian CM, 2008;336:1348 –1351 20. Jiang R, Manson JE, Stampfer MJ, Liu S,
Clark NG, Franz MJ, Hoogwerf BJ, Lich- 10. Mozaffarian D, Marfisi R, Levantesi G, Sil- Willett WC, Hu FB. Nut and peanut
tenstein AH, Mayer-Davis E, Mooradian letta MG, Tavazzi L, Tognoni G, Valagussa butter consumption and risk of type 2
AD, Wheeler ML. Nutrition recommen- F, Marchioli R. Incidence of new-onset di- diabetes in women. JAMA 2002;288:
dations and interventions for diabetes: a abetes and impaired fasting glucose in pa- 2554 –2560
position statement of the American Dia- tients with recent myocardial infarction 21. Kochar J, Gaziano JM, Djoussé L. Nut
betes Association. Diabetes Care 2008;31: and the effect of clinical and lifestyle risk consumption and risk of type II diabetes
S61–S78 factors. Lancet 2007;370:667– 675 in the Physicians’ Health Study. Eur J Clin
2. Pan XR, Li GW, Hu YH, Wang JX, Yang 11. Esposito K, Maiorino MI, Ciotola M, Di Nutr 2010;64:75–79
WY, An ZX, Hu ZX, Lin J, Xiao JZ, Cao Palo C, Scognamiglio P, Gicchino M, Pe- 22. Sofi F, Cesari F, Abbate R, Gensini GF,
HB, Liu PA, Jiang XG, Jiang YY, Wang JP, trizzo M, Saccomanno F, Beneduce F, Ce- Casini A. Adherence to Mediterranean
Zheng H, Zhang H, Bennett PH, Howard riello A, Giugliano D. Effects of a diet and health status: meta-analysis. BMJ
BV. Effects of diet and exercise in prevent- Mediterranean-style diet on the need for 2008;11:337:a1344
ing NIDDM in people with impaired glu- antihyperglycemic drug therapy in pa- 23. Lindström J, Ilanne-Parikka P, Peltonen
cose tolerance. The Da Qing IGT and tients with newly diagnosed type 2 diabe- M, Aunola S, Eriksson JG, Hemiö K, Hä-
Diabetes Study. Diabetes Care 1997;20: tes: a randomized trial. Ann Intern Med mäläinen H, Härkönen P, Keinänen-Kiu-
537–544 2009;151:306 –314 kaanniemi S, Laakso M, Louheranta A,
3. Tuomilehto J, Lindström J, Eriksson JG, 12. Estruch R, Martínez-González MA, Mannelin M, Paturi M, Sundvall J, Valle
Valle TT, Hämäläinen H, Ilanne-Parikka P, Corella D, Salas-Salvadó J, Ruiz-Gutiérrez TT, Uusitupa M, Tuomilehto J, Finnish
Keinänen-Kiukaanniemi S, Laakso M, Lou- V, Covas MI, Fiol M, Gómez-Gracia E, Diabetes Prevention Study Group. Sus-
heranta A, Rastas M, Salminen V, Uusitupa López-Sabater MC, Vinyoles E, Arós F, tained reduction in the incidence of type 2
M, Finnish Diabetes Prevention Study Conde M, Lahoz C, Lapetra J, Sáez G, Ros diabetes by lifestyle intervention: fol-
Group. Prevention of type 2 diabetes melli- E, PREDIMED Study Investigators. Ef- low-up of the Finnish Diabetes Preven-
tus by changes in lifestyle among subjects fects of a Mediterranean-style diet on car- tion Study. Lancet 2006;368:1673–1679
with impaired glucose tolerance. N Engl diovascular risk factors: a randomized 24. Li G, Zhang P, Wang J, Gregg EW, Yang
J Med 2001;344:1343–1350 trial. Ann Intern Med 2006;145:1–11 W, Gong Q, Li H, Li H, Jiang Y, An Y,
4. Kosaka K, Noda M, Kuzuya T. Prevention 13. Fernández-Ballart JD, Piñol JL, Zazpe I, Shuai Y, Zhang B, Zhang J, Thompson TJ,
of type 2 diabetes by lifestyle intervention: Corella D, Carrasco P, Toledo E, Perez- Gerzoff RB, Roglic G, Hu Y, Bennett PH.
a Japanese trial in IGT males. Diabetes Res Bauer M, Martínez-González MA, Salas- The long-term effect of lifestyle interven-
Clin Pract 2005;67:152–162 Salvadó J, Martín-Moreno JM. Relative tions to prevent diabetes in the China Da
5. Knowler WC, Barrett-Connor E, Fowler validity of a semi-quantitative food-fre- Qing Diabetes Prevention Study: a 20-
SE, Hamman RF, Lachin JM, Walker EA, quency questionnaire in an elderly Medi- year follow-up study. Lancet 2008;371:
Nathan DM, Diabetes Prevention Pro- terranean population of Spain. Br J Nutr 1783–1789
gram Research Group. Reduction in the 2010;103:1808 –1816 25. Diabetes Prevention Program Research
incidence of type 2 diabetes with lifestyle 14. Elosua R, Marrugat J, Molina L, Pons S, Group, Knowler WC, Fowler SE, Ham-
intervention or metformin. N Engl J Med Pujol E. Validation of the Minnesota Lei- man RF, Christophi CA, Hoffman HJ,
2002;346:393– 403 sure Time Physical Activity Questionnaire Brenneman AT, Brown-Friday JO, Gold-
6. Ramachandran A, Snehalatha C, Mary S, in Spanish men. The MARATHOM Inves- berg R, Venditti E, Nathan DM. 10-year
Mukesh B, Bhaskar AD, Vijay V, Indian tigators. Am J Epidemiol 1994;139: follow-up of diabetes incidence and
Diabetes Prevention Programme (IDPP). 1197–1209 weight loss in the Diabetes Prevention
The Indian Diabetes Prevention Pro- 15. American Diabetes Association. Diagnosis Program Outcomes Study. Lancet 2009;
gramme shows that lifestyle modification and classification of diabetes mellitus. Di- 374:1677–1686

care.diabetesjournals.org DIABETES CARE, VOLUME 34, NUMBER 1, JANUARY 2011 19