Professional Documents
Culture Documents
Obesity and The Mediterranean Diet: A Systematic Review of Observational and Intervention Studies
Obesity and The Mediterranean Diet: A Systematic Review of Observational and Intervention Studies
1
Unit of Nutrition, Environment and Cancer. Summary
Epidemiological Research Programme. World Health Organization projections estimate that worldwide approximately
Catalan Institute of Oncology-IDIBELL, one-third of adults are overweight and one-tenth are obese. There is accumulating
Barcelona, Spain; 2Mediterranean Diet research into the Mediterranean diet and whether it could prevent or treat obesity.
Foundation, University of Barcelona Science Therefore, the purpose of this paper was to systematically review and analyse the
Park, Barcelona, Spain; 3Department of Public epidemiological evidence on the Mediterranean diet and overweight/obesity. We
Health, University of Barcelona, Barcelona, identified 21 epidemiological studies that explored the relationship between the
Spain; 4Department of Clinical Sciences, Mediterranean diet and weight. These included seven cross-sectional, three cohort
University of Las Palmas de Gran Canaria, and 11 intervention studies. Of these, 13 studies reported that Mediterranean diet
Las Palmas de Gran Canaria, Spain adherence was significantly related to less overweight/obesity or more weight loss.
Eight studies found no evidence of this association. Exploring the relationship
Received 23 November 2007; revised 5 May between the Mediterranean diet and overweight/obesity is complex, and there are
2008; accepted 7 May 2008 important methodological differences and limitations in the studies that make it
difficult to compare results. Although the results are inconsistent, the evidence
Address for correspondence: G Buckland, points towards a possible role of the Mediterranean diet in preventing overweight/
Unit of Nutrition, Environment and Cancer, obesity, and physiological mechanisms can explain this protective effect. Despite
Epidemiological Research Programme, this, more research is needed to substantiate this association. Epidemiological
Catalan Institute of Oncology (ICO), Gran Via studies should use a consistent universal definition of the Mediterranean diet, and
s/n Km 2,7 Hospitalet. 08907, Barcelona, address common methodological limitations to strengthen the quality of research
Spain. E-mail: gbuckland@iconcologia.net in this area.
Although there have been reviews of specific types of cohort studies (21–23) and 11 intervention studies, of
epidemiological studies researching dietary patterns and which three were without a control group (24–26), and
weight (7,8), there has not been a systematic review of eight included a control group (27–34). Information
epidemiological studies, including both clinical trials and regarding the methodology and weight-related results for
observational studies, which have examined the association these studies is summarized in Tables 1 and 2.
between the MD and overweight/obesity specifically.
Obesity contributes towards approximately 30–40% of
Characteristics of study sample
cardiovascular diseases, a large proportion of type 2 dia-
betes and metabolic syndrome and, with a lack of exercise, The studies were carried out between 2000 and 2007
up to 30% of some cancers (9). World Health Organization in eight different countries. Although the majority of the
projections estimated that worldwide, in 2005, approxi- studies (13 out of 21) were from Mediterranean countries
mately 1.6 billion adults were overweight and 400 million (Italy, Spain, Cyprus and Greece), eight studies were also
adults were obese. The increasing trends in obesity in the from non-Mediterranean countries (Germany, Canada, the
USA and Europe are particularly striking, and Mediterra- USA and Hong Kong). The health status of the subjects
nean countries are also affected although to a lesser extent, varied between studies, and could be grouped into three
with adult obesity in Spain increasing from 15% to 21% main categories: (i) apparently healthy individuals without
between 1994–1995 and 1999–2000 (10). chronic diseases; (ii) overweight/obese individuals and (iii)
There has been a decrease in adherence to the MD in individuals with chronic diseases, such as cardiovascular
Southern European countries throughout a similar period disease or its risk factors.
(11,12), as well as a tendency to lead more sedentary lif-
estyles. However, it is not clear whether and to what extent
Association between the MD and obesity
the changing dietary patterns account for the increases in
obesity. As there is an increasing prevalence of overweight/ Out of the 21 studies identified, 13 reported that adherence
obesity, with its life-threatening co-morbidities worldwide, to an MD significantly reduced the probability of
it is important to question whether the MD could be used overweight/obesity, promoted weight loss, or resulted in
to prevent or treat obesity, or conversely whether the MD more weight loss than a control diet (14–16,20,21,24–
could contribute to obesity, because of its moderately high 26,29,30,32–34). In contrast, eight studies found that there
fat content (13). Therefore, the aim of this study was to was no significant association between MD adherence and
systematically review and analyse epidemiological studies overweight/obesity (17–19,22,23,27,28,31).
on the MD and overweight/obesity.
Cohort studies
Out of the three cohort studies, only the Spanish study
Methodology
from the European Prospective Investigation into Cancer
A MEDLINE search was carried out up to July 2007 and Nutrition (21) provided evidence of a significant pro-
to identify epidemiological studies on the MD and tective effect of the MD against obesity incidence. After
overweight/obesity, using the term ‘MD’ along with other excluding obese participants at baseline and taking into
key word(s): ‘obesity’, ‘overweight’, ‘body mass index account participant’s different energy intakes, as well as
(BMI)’, ‘weight’, ‘body fat’ or ‘weight loss/gain’. All human possible dietary under-reporting, overweight individuals
epidemiological studies with full text were considered. The with a high MD adherence were 27% (in women) and 29%
search was narrowed to include only articles examining the (in men) less likely to become obese. Although results from
effect on weight of an MD as a whole (combined effect of the Seguimiento Universidad de Navarra–Follow-up Uni-
key components). Studies were included if the authors versity of Navarra study did not find any association
described the use of an MD, without restricting its defini- between weight and the MD (22), there was a significant
tion to include specific foods or nutrients. In addition, reduction in waist circumference as MD adherence
studies were included if the main outcome was either increased (35).
weight related (overweight/obesity or weight change), a
chronic disease or metabolic alterations, but with weight Cross-sectional studies
as a secondary outcome. Out of the seven cross-sectional studies, four (14–16,20)
found that a higher adherence to an MD had a significantly
negative association with overweight/obesity. The strongest
Results
association was reported in the ATTICA study by Panagio-
A total of 21 studies were identified that met all the search takos (15), where individuals with a higher adherence to an
criteria. The studies were classified according to study type, MD were 51% less likely to be overweight/obese. Other
resulting in seven cross-sectional studies (14–20), three studies found that individuals with a high MD adherence
Cross-sectional studies n = 7
Rossi et al. (17), Italy Patients MDS (+): cereals, vegetables, fruit, legumes, Self-reported cs • ↑ Adh MD (+1p):
2007 n = 6 619 MUFA/SFA, (+m): alcohol (-): meat, milk and 씹 ↑BMI (b coef 0.05 [CI: -0.05, 0.14]), NS
(3 090 씹, 3 529 씸) dairy products 씸 ↓BMI (b coef -0.04 [CI: -0.15, 0.07]), NS
Obesity and the Mediterranean diet
58 years†
Panagiotakos Cyprus General population MDS: wholegrain cereals, vegetables, fruit, Measured cs • ↑ Adh MD (+10p): ↓Ob (OR = 0.88), S, (P = 0.001)
et al. (16), 2007 n = 150 (senior), without legumes, fish, olive oil, dairy products, chicken,
(53 씹, 97 씸) CVD nuts and seeds, olives, potatoes, eggs, sweets,
65–100 years red meat and meat products
Panagiotakos Greece General population, MDS: wholegrain cereals, vegetables, fruit, Measured cs • ↑ Adh MD (+5p): ↓Ow/Ob (OR = 0.49
et al. (15), 2006 n = 3 042 without CVD legumes, nuts and seeds, fish, olive oil, olives, [CI: 0.42, 0.56]), S
1 514 씹 (18–87 years) dairy products, chicken, potatoes, eggs, sweets, • ↑ Adh MD: ↓BMI (-4 kg m-2)‡, S, (P = 0.001)
G. Buckland et al.
1 528 씸 (18–89 years) wine, red meat and meat products, MUFA/SFA
Shubair et al. (14), Canada General population MDP (principal component analysis) (+): fruit and Self-reported cs • ↑ Adh MD: ↓BMI (b coef -0.186)
2005 n = 759 vegetables, olive oil and garlic, fish and shellfish, All ages, S, (P = 0.027)
(265 씹, 494 씸) Non MDP (-): meats and poultry, high SFA and Pop 40–49 years, S, (P = 0.011)
18–65 years TFA, foods high in added sugar and low in Pop 30–39 years, NS, (P = 0.056)
nutrients Pop 18–29 years, NS, (P = 0.999)
Trichopoulou et al. Greece General population, MDS (+): cereals, vegetables, fruit and nuts, Measured cs • ↑ Adh MD (+2p energy adjusted):
18, 2005 n = 23 597 healthy legumes, fish, MUFA/SFA, (+m): ethanol, (-): 씹 ↑BMI (b coef 0.08 [CI: -0.03, 0.20]), NS
(9 612 씹, 13 985 씸) meat and meat products, dairy products 씸 ↓BMI (b coef -0.06 [CI: -0.16, 0.04]), NS
20–86 years • ↑ Adh MD (+2p, not energy adjusted):
씹 ↑BMI (b coef 0.21 [CI: 0.10, 0.32]), S
씸 ↑BMI (b coef 0.05 [CI: -0.04, 0.15]), NS
Fung et al. (19), USA Nurses, healthy MDS (+): wholegrain cereals, vegetables (without Self-reported cs • ↑ Adh MD: ↓BMI (26.5 ⫾ 6.1 kg m-2)†
2005 n = 660 (씸) potatoes), fruit, nuts, legumes, fish, MUFA/SFA vs.
43–69 years (+m): alcohol, (-): red and processed meat • ↓ Adh MD: ↑BMI (27.1 ⫾ 6.8 kg m-2)†, NS,
(P = 0.43)
Schroder et al. (20), Spain General population MDS (+): cereals, vegetables, fruit, legumes, Measured cs • ↑ Adh MD: ↓Ob (OR = 0.61), S, (P = 0.01)
2004 n = 2 871 nuts, fish, (+m): red wine, (-): meat, high fat • ↑ Adh MD (+5p):
(1 403 씹, 1 468 씸) dairy products 씹 ↓BMI (-0.43 kg m-2)‡, S (P = 0.030)
25–74 years 씸 ↓BMI (-0.68 kg m-2)‡, S, (P = 0.007)
Journal compilation © 2008 International Association for the Study of Obesity. obesity reviews 9, 582–593
1467789x, 2008, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1467-789X.2008.00503.x by Universidad Nacional Autonoma De Mexico, Wiley Online Library on [09/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table 1 Continued
Cohort studies n = 3
Mendez et al. (21), Spain General population, MDS (+): cereals, vegetables, fruit, legumes, 0 years: 3.3 years • ↑ Adh MD:
2006 n = 27 827 normal weight fish, MUFA/SFA, (+m): ethanol, (-): meat measured 씸 ↓Ob (OR = 0.73 [CI: 0.57, 0.93]), S
(10 589 씹, 17 238 씸) or overweight 3.3 years: 씹 ↓Ob (OR = 0.71 [CI: 0.55, 0.93]), S
29–65 years self-reported 씸 ↓Ow (OR = 0.99 [CI: 0.78, 1.25]), NS
씹 ↑Ow (OR = 1.11 [CI: 0.81, 1.52]), NS
Sanchez-Villegas Spain University MDP (+): cereals, vegetables, fruit, legumes, Self-reported 2.4 years • ↑ Adh MD: ↓Ow/Ob (OR = 0.90 [CI: 0.59, 1.38]),
et al. (22), n = 6 319 (씸 씹) graduates, nuts, fish, olive oil, (+m): red wine, (-): meat and NS
2005 healthy meat products, whole fat dairy products • ↑ Adh MD: ↑BMI (+0.23 kg m-2)‡
vs.
↓Adh MD: ↑BMI (+0.26 kg m-2)‡, NS, (P = 0.279)
Woo et al. (23), Hong Kong General population MDS (+): cereals, vegetables, fruit and nuts, Measured 5–9 • ↑ Adh MD (+1SD): ↑Ow (OR = 1.35 [CI: 0.94,
2000 n = 1 010 (씸 씹) (Chinese), normal legumes, fish, MUFS/SFA, (+m): ethanol, (-): years 1.93]), NS
25–74 years BMI meat, poultry, dairy products
Journal compilation © 2008 International Association for the Study of Obesity. obesity reviews 9, 582–593
(230 씹, 788 씸) obese fish, olive oil, wine, dairy products, chicken, ↓BMI (30.1 ⫾ 4.4 kg m-2)†
14–76 years eggs, sweets, red meat vs.
MD + PA Programme • Adh MD (non-completers):
↓weight (-6.57 ⫾ 4.9 kg)‡
↓BMI (32.4 ⫾ 5.4 kg m-2)†
Difference between groups S (P < 0.001)
Goulet et al. (24), Canada General population MD: cereals, fruit, vegetables, legumes, nuts and Measured 3m • Adh MD: ↓BMI (m0: 25.8 ⫾ 3.9 kg m-2 to m3:
Obesity and the Mediterranean diet
2003 n = 73 (씸) (French Canadians) seeds, fish, olive oil, wine, dairy products, 25.6 ⫾ 3.8 kg m-2)†, S, (P < 0.01)
30–65 years without metabolic chicken, eggs, sweets, red meat • Adh MD: ↓weight (m0: 67.7 ⫾ 11.9 kg to m3:
diseases and with MD + Mediterranean style cooking classes 67.3 ⫾ 11 kg)†, S, (P < 0.01)
stable weight
G. Buckland et al.
585
1467789x, 2008, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1467-789X.2008.00503.x by Universidad Nacional Autonoma De Mexico, Wiley Online Library on [09/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
586
Table 1 Continued
vs. vs.
MD + 30 g d-1 of nuts and seeds (free) (n = 258) • Diet-C:
vs. ↓weight (-0.24 kg [CI: -0.48, 0.01])‡, NS
Diet-C: low fat (n = 257) ↓BMI (-0.21 kg m-2 [CI: -0.38, -0.05])‡, S
Difference between all groups, NS
Michalsen et al. Germany Patients with a MD (+): wholegrain cereals, fruit, vegetables, fish Measured 1 year • Adh DM vs. Diet-C: ↓BMI (-0.10 kg m-2)‡, NS,
(28), 2006 n = 101 (78 씹, 23 씸) history of CHD (fatty), poultry, olive oil, nuts and flaxseeds, (P = 0.969)
59 ⫾ 8.6 years† (+m): red wine, (-): meat (including
processed)
MD + healthy lifestyle programme (100 h)
vs.
Diet-C: written information about a healthy diet
Vincent-Baudry France Patients with one MD (35–38% energy from fat), (+): wholegrain Measured 3m • Adh DM: ↓BMI (-1.5 kg m-2)‡, S, (P = 0.010)
et al. (29), 2005 n = 212 (씸 씹) CVD risk factor cereals, vegetables, fruit, legumes, nuts, fish, vs.
18–70 years olive oil, poultry and sheep, (+m): wine, dairy • Diet-C: ↓BMI (-1.2 kg m-2)‡, S, (P = 0.010)
products (sheep and goat), (-): red meat Difference between groups NS
MD vs. Diet-C (30% of total energy from fat)
Journal compilation © 2008 International Association for the Study of Obesity. obesity reviews 9, 582–593
1467789x, 2008, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1467-789X.2008.00503.x by Universidad Nacional Autonoma De Mexico, Wiley Online Library on [09/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table 1 Continued
Journal compilation © 2008 International Association for the Study of Obesity. obesity reviews 9, 582–593
advise (n = 60) (P < 0.05)
• Adh DM vs. Diet-C: ↓weight (-11 kg)‡, ↓BMI
(-4.2 kg m-2)‡, S, (P < 0.001)
McManus et al. USA General population, MD: hypo-caloric (35% energy from fat) + healthy Measured 2.5 years • Adh DM: ↓weight (-4.1 ⫾ 6.5 kg)‡, ↓BMI
(34), 2001 n = 101 overweight and lifestyle programme (n = 50) (-1.6 ⫾ 2.5 kg m-2)‡, S, (P < 0.001)
(10 씹, 91 씸) obese, without vs. vs.
18–70 years chronic diseases Diet-C: low fat + healthy lifestyle programme • Diet-C: ↑weight (+2.9 ⫾ 7.7 kg)‡, ↑BMI
Obesity and the Mediterranean diet
Healthy, without chronic diseases or metabolic disorders (cholesterol, hypertriglyceridemia); CVD, cardiovascular diseases; CHD, coronary heart diseases; MD, Mediterranean diet; MDS, Mediterranean
diet score; MDP, Mediterranean diet pattern; MLP, Mediterranean lifestyle programme; (+), positive components; (-), negative components; (+m), components positive in moderation; Diet-C, diet-control;
SFA, saturated fatty acid; MUFA, monounsaturated fatty acid; TFA, trans fatty acid; CHO, carbohydrates; PA, physical activity; cs, cross-sectional; Adh, adherence; Ob, obesity; Ow, overweight; BMI,
587
body mass index; ↓, decrease; ↑, increase; S, significant; NS, not significant; OR, odds ratio; CI, confidence interval (95%); b coef, b coefficient; SD, standard deviation; m, month.
1467789x, 2008, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1467-789X.2008.00503.x by Universidad Nacional Autonoma De Mexico, Wiley Online Library on [09/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1467789x, 2008, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/j.1467-789X.2008.00503.x by Universidad Nacional Autonoma De Mexico, Wiley Online Library on [09/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
588 Obesity and the Mediterranean diet G. Buckland et al. obesity reviews
Table 2 Summary of results of epidemiological studies on the Mediterranean diet and overweight/obesity
Cross-sectional studies n = 7
Rossi et al. (17) 6 619 cs b coef, BMI (CI) 씹 0.05 (-0.05, 0.14)
씸 -0.04 (-0.15, 0.07)
Trichopoulou et al. (18) 23 597 cs 씹 0.08 (-0.03, 0.20)
씸 -0.06 (-0.16, 0.04)
Shubair et al. (14) 759 cs -0.186 (0.027)
Panagiotakos et al. (16) 150 cs OR (CI or P value) 0.88 (0.001)
Panagiotakos et al. (15) 3 042 cs 0.49 (0.42, 0.56)
Schroder et al. (20) 2 871 cs 0.61 (0.01)
Fung et al. (19) 660 cs Difference in BMI (kg m-2) 0.6 (0.43)
Cohort studies n = 3
Mendez et al. (21) 27 827 3.3 years OR (CI) 씸 0.73 (0.57, 0.93)
씹 0.71 (0.55, 0.93)
Sanchez-Villegas et al. (22) 6 319 2.4 years 씹 0.90 (0.59, 1.38)
Woo et al. (23) 1 010 5–9 years 1.35 (0.94, 1.93)
Intervention studies n = 11
Andreoli et al. (26) 47 4 months Difference in BMI (kg m-2) (CI or P value) -2 (<0.001)†
Bautista-Castaño et al.* (25) 1 018 5.7 months -2.3 (<0.001)‡
Goulet et al. (24) 73 3 months -0.2 (<0.01)†
Estruch et al. (27) 772 3 months -0.12 (-0.24, 0.06)†
-0.09 (-0.24, 0.05)†
Michalsen et al. (28) 101 1 years -0.10 (0.969)§
Vincent-Baudry et al. (29) 212 3 months -1.5 (0.010)†
Esposito et al. (30) 180 2 years -1.2 (0.001)†
†
Fernández de la Puebla et al. (31) 34 28 days NS
Toobert et al. (32) 279 6 months -0.57 (0.015)§
Esposito et al. (33) 120 2 years -5.2 (0.001)†
-4.2 (0.001)§
McManus et al. (34) 101 2.5 years -1.6 (0.001)†
were 39% less likely to be obese (20) or 12% less likely be compared with a control diet, and the significance of the
obese (16). All of the three studies that used a subjects weight loss within the MD group was not specified. Half of
self-reported weight (14,17,19) did not find any association the studies (four out of eight), which compared the weight
between MD adherence and weight, apart from a study by loss after following an MD and a control diet, found that
Shubair (14). This study reported that there was a decrease the MD group lost significantly more weight than the
in BMI with MD adherence, but only significant within the control group (30,32–34).
older age groups. Of the four studies whose study sample consisted of
overweight/obese individuals, (25,26,33,34), all found the
Intervention studies MD effective for losing weight, and also reported the most
There were two types of intervention studies: (i) all the pronounced reductions in weight. The greatest weight loss
subjects followed an MD (n = 3) or (ii) the subjects were was a group average of 14 kg, in a small group of Italian
divided into groups, and instructed to either follow an MD obese patients following a 2-year hypo-caloric MD as well
or a control diet (n = 8) with or without additional non- as a healthy lifestyle programme (33). The two studies,
dietary interventions. Over two-thirds of all of these inter- which specifically recruited individuals with type 2 diabetes
vention studies (eight out of 11) found that adherence to an mellitus and metabolic syndrome (30,32), both closely
MD significantly decreased weight/BMI (24–26,29,30,32– related to obesity, also found a significant reduction in
34), although in two of these studies (25,32), this was obesity after following an MD.
Discussion
Intervention studies
The intervention studies recommended different variations Epidemiological evidence assessing the association between
of the MD, partly because of the use of different definition MD adherence and overweight/obesity is limited and con-
sources as a reference. Four of the MD interventions were flicting. Only 21 epidemiological studies have assessed this
hypo-caloric (25,26,33,34), and all reported significant relationship, and the majority of these were intervention
weight losses. The MD intervention varied in terms of level and cross-sectional studies, with few cohort studies. No
studies reported that an MD significantly increased obesity. libitum energy intake between four trial meals rich in either
Just over half of the studies provided evidence that the fat, carbohydrate, protein or alcohol (41).
adherence to an MD was associated with less overweight/ Thirdly, the habitual use of olive oil in salads and veg-
obesity or promoted weight loss. The extent of the protec- etable and legume dishes enhances palatability of these
tive effect of the MD was reasonably strong in some foods. This increases consumption of foods high in dietary
studies. For example, in a cohort study (21), men with a fibre and low in energy density, resulting in greater satia-
high MD adherence were up to 29% less likely to become tion and satiety.
obese. An even stronger protective effect was seen in cross- Fourthly, diets rich in monounsaturated fat have been
sectional studies, with up to 51% less probability of being found to improve glucose metabolism (42), and increase
overweight or obese (15). In addition, some intervention postprandial fat oxidation, diet-induced thermogenesis and
studies reported important weight losses, of up to 14 kg in overall daily energy expenditure (43,44), compared with
one study (33). diets higher in saturated fats. This may provide physiologi-
There are several physiological explanations that could cal explanations of why olive oil consumption is less prone
explain why key components of MD might protect against to cause weight gain (22,45).
weight gain. The MD is rich in plant-based foods that Finally, the MD is highly palatable and, therefore, well
provide a large quantity of dietary fibre. This has been liked and tolerated among dieters, and compliance with the
shown to increase satiety and satiation through mecha- MD has been found to be reasonably high (27,29,34).
nisms, such as prolonged mastication, increased gastric As well as the favourable fatty acid profile of the MD, the
detention and enhanced release of cholecystokinin (38). variety in vegetable products and higher consumption of
Energy density has an important role in weight gain, as plant-based foods, compared with that of animal products,
palatable energy-dense food leads to poor appetite control provide a diet rich in both non-nutritional factors and
and consequently to over-consumption. The MD has a low micronutrients (especially antioxidants). These all give
energy density (38) and a low glycaemic load (39) com- additional health benefits, such as reducing risk of cardio-
pared with many other dietary patterns. These characteris- vascular disease and type 2 diabetes (6,38).
tics, together with its high water content, lead to increased Despite the physiological mechanistic evidence that can
satiation and a lower calorie intake, and thus help to explain how the MD can protect against weight gain, the
prevent weight gain. epidemiological evidence for this relationship is inconsis-
Some studies reported the percentage of total energy tent. Exploring the relationship between the diet and
derived from fat within the MD, and in most instances, it obesity is complex, and methodological differences may
was consistent with the moderately high fat content char- partly explain the incoherency between studies.
acteristic of the traditional MD (30–40% of energy from In terms of cohort and cross-sectional studies, a key
fat). However, some intervention studies (26,30,31,33) did issue is the use of inconsistent definitions of the MD.
use an MD with 25–30% of energy from fat. The conse- Several studies, in their definition, do not include key
quences of diets with a relatively high total fat content have characteristic components of a traditional MD, such as
been a topic of concern partly because of the possible olive oil or ‘wholegrain’ cereals. The methodology used to
effects on weight gain (13). However, there is some evi- construct MD indices could be a further issue because the
dence that high fat diets are not the major cause of obesity use of cut-offs, such as medians to indicate a high adher-
(39), as important cohort studies (1,40) and long-term ence, may not reflect a traditional MD. In non-
trials (33,34) have found no significant relation between Mediterranean countries, such as the USA or Hong Kong,
higher fat diets and obesity. the intakes of certain components of the MD, such as
The MD and a low-fat diet could both possess similarly olive oil or legumes, may be considerably lower. This
favourable effects on prevention of weight gain, because of would result in low median intakes of these components,
the high fibre content and the low energy density of the and therefore, show uncharacteristically low cut-offs to
components. However, the MD has several advantageous define MD adherence.
characteristics that also protect against obesity. In Mediterranean countries, recent research has shown
Firstly, the quality of fat is a key factor, as it is low in that the traditional MD is disappearing (2), and therefore,
cholesterol-rising fats (saturated and trans fats) and high in using intake medians to define a high adherence to an MD
monounsaturated fats (approximately 67% of fat energy) would again be influenced by current dietary patterns,
as found in oleic acid in olive oil. This fatty acid profile has and may not reflect a traditional MD. The comparison of
a range of important health benefits (6,13,38). weight outcomes of studies carried out in different coun-
Secondly, although fat is believed to be the least satiating tries using different MD definitions, not surprisingly, gen-
of the macronutrients, study findings have not always been erates contradictory results.
consistent. For instance, a recent intervention study found Another important methodological issue is the assess-
no differences in hunger or satiety sensations or in ad ment of diet in obese participants, as they are more prone
to under-report dietary intakes in comparison with lean intervention and the more time spent explaining the
participants (37). In addition, they may have changed intervention to the participants, the greater the weight
their previous dietary habits because of treatment of loss. The use of non-dietary approaches, especially mul-
obesity-related co-morbidities. These factors could dilute tidisciplinary healthy lifestyle programmes, promoted
the association between MD and obesity. The use of greater weight losses than studies that did not use these
studies that prospectively explore the effect of an MD on additional interventions. For example, the study by
weight can overcome these issues by excluding at baseline Esposito (33), which combined an MD with physical
obese participants and individuals with a poor concor- activity programmes, reported a mean weight loss of
dance of reported energy intakes to expenditures. This 14 kg. However, in studies such as these, it is difficult to
methodological approach was used in two of the three establish whether the weight loss is a consequence of the
cohort studies (21). An additional advantage of cohort MD or the non-dietary interventions. A further compli-
studies is that they are more capable of exploring cation is that in the majority of intervention studies the
obesity causality, which is a limitation of cross-sectional additional non-dietary interventions (when used) were
studies. only given to the MD group and not the control group
The use of different anthropometric measures, with (except in one study). Therefore, in these studies, com-
nearly half the cross-sectional studies using participants’ paring the effectiveness of the MD with a control diet is
self-reported weight as opposed to measured weight, confounded by these differences. However, the study by
creates similar issues to those mentioned above. Obese McManus (34), which did give similar non-dietary treat-
individuals are more likely to under-report weight com- ment to both the MD and control group, found that the
pared with lean individuals, again this could dilute the MD group lost more weight, providing some evidence for
relationship between the MD and obesity. Although self- the effectiveness of the MD on weight loss compared with
reported weight was validated in most studies, only one of other diets.
the studies using participants’ self-reported weight reported Finally, differences in statistical methods between inter-
that the MD was inversely associated with obesity. Dif- vention studies can have important effects on the results.
ferences in the statistical analyses in cohort and cross- Differences include the analysis of weight changes of all
sectional studies, such as number of confounders taken into participants, using an intention to treat method or analysis
account in regression models, measuring intake of MD of only participants who completed the intervention. The
components as a percentage of total energy intake or using sample sizes varied considerably within the different
validated dietary questionnaires, could also influence the studies, which could alter the statistical power to detect
reported MD–obesity relationship. significant associations, and contribute to the inconsisten-
In terms of intervention studies, it is well known that it is cies in results. However, the effects of these differences are
very difficult to change dietary habits. Therefore, the many not clear-cut.
methodological factors that affect the level of compliance Although the epidemiological evidence regarding the
and effectiveness of an MD can also contribute to the relationship between the MD and overweight/obesity is
differences in weight loss. Several of these factors are: inconsistent, it reveals that the MD is not related to any
health status of participants, type of recommended MD, increased risk of overweight/obesity. It actually points
length of MD intervention and additional healthy lifestyle towards a possible role of the MD in preventing
interventions. Again, this makes it difficult to compare the overweight/obesity, and physiological mechanisms could
results between studies. explain this protective effect. However, further research is
The countries where the intervention studies took place needed to substantiate these findings because of the incon-
could affect the availability, affordability or acceptability sistency of the results. Cohort or intervention studies that
of MD food used in the intervention. Consequently, greater are able to provide better evidence of causality, together
compliance would be expected in Mediterranean compared with the use of a consistent universal definition of the MD,
with non-Mediterranean countries, such as Germany or are necessary. MD interventions should provide the same
Hong Kong. non-dietary interventions to comparison groups, if the aim
Regarding the health status of participants, all MD inter- is to assess which diet is more effective for weight loss.
ventions carried out on overweight/obese participants or Longer interventions, which also assess compliancy, are
participants with closely related disorders, reported signifi- required to evaluate the long-term efficacy of the MD for
cant weight losses. The MD appears to be more successful promoting and preventing overweight/obesity.
at promoting weight loss when it is targeted at these
groups, possibly because it varies more from their usual
diet, or because they are more motivated for health reasons.
Conflict of Interest Statement
The way the MD intervention was delivered varied
considerably between studies. In general, the longer the No conflict of interest was declared.
a Mediterranean-style diet on endothelial dysfunction and markers food group intakes in Jamaican adults. Public Health Nutr 2004;
of vascular inflammation in the metabolic syndrome: a randomized 7: 9–19.
trial. JAMA 2004; 292: 1440–1446. 38. Schroder H. Protective mechanisms of the Mediterranean diet
31. Fernandez de la Puebla RA, Fuentes F, Perez-Martinez P, in obesity and type 2 diabetes. J Nutr Biochem 2007; 18: 149–160.
Sanchez E, Paniagua JA, Lopez-Miranda J, Perez-Jimenez F. A 39. Willett WC, Leibel RL. Dietary fat is not a major determinant
reduction in dietary saturated fat decreases body fat content in of body fat. Am J Med 2002; 113(Suppl. 9B): 47S–59S.
overweight, hypercholesterolemic males. Nutr Metab Cardiovasc 40. Seccareccia F, Lanti M, Menotti A, Scanga M. Role of body
Dis 2003; 13: 273–277. mass index in the prediction of all cause mortality in over 62 000
32. Toobert DJ, Glasgow RE, Strycker LA, Barrera M Jr, Radcliffe men and women. The Italian RIFLE Pooling Project. Risk factor and
JL, Wander RC, Bagdade JD. Biologic and quality-of-life outcomes life expectancy. J Epidemiol Community Health 1998; 52: 20–26.
from the Mediterranean Lifestyle Program: a randomized clinical 41. Raben A, Agerholm-Larsen L, Flint A, Holst JJ, Astrup A.
trial. Diabetes Care 2003; 26: 2288–2293. Meals with similar energy densities but rich in protein, fat, carbo-
33. Esposito K, Pontillo A, Di Palo C, Giugliano G, Masella M, hydrate, or alcohol have different effects on energy expenditure
Marfella R, Giugliano D. Effect of weight loss and lifestyle changes and substrate metabolism but not on appetite and energy intake.
on vascular inflammatory markers in obese women: a randomized Am J Clin Nutr 2003; 77: 91–100.
trial. JAMA 2003; 289: 1799–1804. 42. Due A, Larsen TM, Hermansen K, Stender S, Holst JJ, Toubro
34. McManus K, Antinoro L, Sacks F. A randomized controlled S, Martinussen T, Astrup A. Comparison of the effects on insulin
trial of a moderate-fat, low-energy diet compared with a low fat, resistance and glucose tolerance of 6-mo high-monounsaturated-
low-energy diet for weight loss in overweight adults. Int J Obes fat, low-fat, and control diets. Am J Clin Nutr 2008; 87: 855–862.
Relat Metab Disord 2001; 25: 1503–1511. 43. Soares MJ, Cummings SJ, Mamo JC, Kenrick M, Piers LS. The
35. Tortosa A, Bes-Rastrollo M, Sanchez-Villegas A, Basterra- acute effects of olive oil v. cream on postprandial thermogenesis
Gortari FJ, Nunez-Cordoba JM, Martinez-Gonzalez MA. Medi- and substrate oxidation in postmenopausal women. Br J Nutr
terranean diet inversely associated with the incidence of metabolic 2004; 91: 245–252.
syndrome: the SUN prospective cohort. Diabetes Care 2007; 11: 44. Piers LS, Walker KZ, Stoney RM, Soares MJ, O’Dea K. Sub-
2957–2959. stitution of saturated with monounsaturated fat in a 4-week diet
36. Trichopoulou A, Kouris-Blazos A, Wahlqvist ML, Gnardellis affects body weight and composition of overweight and obese
C, Lagiou P, Polychronopoulos E, Vassilakou T, Lipworth L, men. Br J Nutr 2003; 90: 717–727.
Trichopoulos D. Diet and overall survival in elderly people. BMJ 45. Bes-Rastrollo M, Sanchez-Villegas A, de la Fuente C, de Irala
1995; 311: 1457–1460. J, Martinez JA, Martinez-Gonzalez MA. Olive oil consumption
37. Mendez MA, Wynter S, Wilks R, Forrester T. Under- and and weight change: the SUN prospective cohort study. Lipids
over-reporting of energy is related to obesity, lifestyle factors and 2006; 41: 249–256.