Accruing evidence on benefits of adherence to the Mediterranean diet on health: an updated systematic review and meta-analysis1,2

 
CONTINUE READING
Accruing evidence on benefits of adherence to the Mediterranean diet
on health: an updated systematic review and meta-analysis1,2
Francesco Sofi, Rosanna Abbate, Gian Franco Gensini, and Alessandro Casini

ABSTRACT                                                                  great deal of attention has been given to tools that estimate the
Background: The Mediterranean diet has long been reported to              adherence of individuals to the Mediterranean diet because of the
be protective against the occurrence of several different health          usefulness of these tools to identify the whole dietary pattern
outcomes.                                                                 instead of single foods or nutrients (4). Hence, computational
Objective: We aimed to update our previous meta-analysis of pub-          scores have been created and used in several large epidemiologic
lished cohort prospective studies that investigated the effects of        studies to seek whether they could be useful to estimate the risk of
adherence to the Mediterranean diet on health status.                     disease in the general population (5). Some large epidemiologic

                                                                                                                                                       Downloaded from ajcn.nutrition.org by guest on February 10, 2015
Design: We conducted a comprehensive literature search through            studies conducted in different cohorts evidenced an association
electronic databases up to June 2010.                                     between a greater adherence to Mediterranean diet, a reduced risk
Results: The updated review process showed 7 prospective studies          of mortality, and the incidence of major chronic diseases (6–9). In
published in the past 2 y that were not included in the previous          2008, we performed a meta-analysis that included all cohort
meta-analysis (1 study for overall mortality, 3 studies for cardiovas-    prospective studies that investigated this issue, and we observed
cular incidence or mortality, 1 study for cancer incidence or mor-        that a 2-point increase of adherence to Mediterranean diet
tality, and 2 studies for neurodegenerative diseases). These recent       conferred a significant protection against mortality, the occur-
studies included 2 health outcomes not previously investigated (ie,       rence of cardiovascular diseases, and major chronic degenerative
mild cognitive impairment and stroke). The meta-analysis for all          diseases (3). However, the previous meta-analysis is 3 y old, and
studies with a random-effects model that was conducted after the          more prospective cohort studies have since been published.
inclusion of these recent studies showed that a 2-point increase in          Therefore, the aim of this study was to update our previous
adherence to the Mediterranean diet was associated with a signifi-
                                                                          systematic review and meta-analysis conducted in cohort pro-
cant reduction of overall mortality [relative risk (RR) = 0.92; 95%
                                                                          spective studies that investigated the effects of adherence to
CI: 0.90, 0.94], cardiovascular incidence or mortality (RR = 0.90;
                                                                          Mediterranean diet on several health outcomes and to use these
95% CI: 0.87, 0.93), cancer incidence or mortality (RR = 0.94;
                                                                          data to conduct a meta-regression analysis to explore a potential
95% CI: 0.92, 0.96), and neurodegenerative diseases (RR = 0.87;
                                                                          source of heterogeneity in studies.
95% CI: 0.81, 0.94). The meta-regression analysis showed that sam-
ple size was the most significant contributor to the model because it
significantly influenced the estimate of the association for overall      METHODS
mortality.
Conclusion: This updated meta-analysis confirms, in a larger num-         Literature search
ber of subjects and studies, the significant and consistent protection       According to the statement of the Preferred Reporting Items for
provided by adherence to the Mediterranean diet in relation to the        Systematic Reviews and Meta-Analyses (10), we systematically
occurrence of major chronic degenerative diseases.          Am J Clin     searched published cohort prospective studies that investigated
Nutr 2010;92:1189–96.                                                     the association between adherence to the Mediterranean diet and
                                                                          health outcomes in electronic databases: MEDLINE (source:
                                                                          PubMed, 1966 to June 2010; www.pubmed.com), EMBASE (1980
INTRODUCTION                                                              to June 2010; www.embase.com), Web of Science (isiweb
   The Mediterranean diet has long been reported to be the                ofknowledge.com), The Cochrane Library (source: The Cochrane
optimal diet for preventing noncommunicable diseases and
                                                                             1
preserving good health (1–3). The Mediterranean-style diet is not              From the Department of Medical and Surgical Critical Care, Thrombo-
a specific diet, but rather a collection of eating habits traditionally   sis Centre, University of Florence, Florence, Italy (FS, RA, and GFG); the
followed by people in the different countries bordering the               Agency of Nutrition, Azienda Ospedaliero-Universitaria Careggi, Florence,
Mediterranean Sea. The diet refers to a dietary profile commonly          Italy (FS and AC); and the Don Carlo Gnocchi Foundation, Centro Santa
                                                                          Maria agli Ulivi, Onlus IRCCS, Florence, Italy (FS and GFG).
available in the early 1960s in the Mediterranean regions and                2
                                                                               Address correspondence to F Sofi, Department of Medical and Surgical
characterized by a high consumption of fruit, vegetables, le-             Critical Care, Thrombosis Centre, University of Florence, Viale Morgagni
gumes, and complex carbohydrates, with a moderate consump-                85, 50134 Florence, Italy. E-mail: francescosofi@gmail.com.
tion of fish, and the consumption of olive oil as the main source of         Received April 14, 2010. Accepted for publication July 27, 2010.
fats and a low-to-moderate amount of red wine during meals. A                First published online September 1, 2010; doi: 10.3945/ajcn.2010.29673.

Am J Clin Nutr 2010;92:1189–96. Printed in USA. Ó 2010 American Society for Nutrition                                                        1189
1190                                                             SOFI ET AL

Central Register of Controlled Trials, 2010, issue 2; www.the            ment for confounders that include demographic, anthropometric,
cochranelibrary.com/), clinicaltrials.org, and Google Scholar            and traditional risk factors were considered of high quality.
(scholar.google.com). Relevant keywords relating to the Medi-
terranean diet in combination as MeSH terms and text words
(“Mediterranean diet,” or “diet” or “dietary pattern” “Mediterra-        Statistical analyses
nean,” or “adherence” or “score” and their variants) were used in           We used the Review Manager program (RevMan, version
combination with words relating to health status (“health,” or           5.0.18 for Macintosh, 2008; The Cochrane Collaboration,
“mortality” or “morbidity,” or “cardiovascular diseases,” or             Copenhagen, Denmark) and the Statistical Package for Social
“neoplastic diseases,” or “cancer,” or “neoplasm,” or “de-               Sciences software (SPSS, version 18.0 for Macintosh; SPSS,
generative diseases,” or “Alzheimer’s disease,” or “Parkinson’s          Chicago, IL) to pool and analyze results from the individual
disease,” or “cerebrovascular disease,” or “dementia,” or “cog-          studies. The methods and results of all recently identified cohort
nition disorder,” or “stroke,” or “outcome,” or “prospective,” or        prospective studies were added to the previous table, and data
“follow-up,” or “cohort” and their variants). The search strategy        were formally combined. Pooled results were reported as relative
had no language restrictions. References from the extracted articles     risks (RRs) and presented with 95% CIs with 2-sided P values by
and reviews were also consulted to complete the data bank. When          using a random-effects model (DerSimonian and Laird method)
multiple articles for a single study were present, we used the latest    and the general variance-based method. P , 0.05 was consid-
publication and supplemented it, if necessary, with data from the        ered statistically significant. When available, we used the results
most complete or updated publication.                                    of the original studies from multivariate models with the most
                                                                         complete adjustment for potential confounders; the confounding
                                                                         variables included in the analyses are shown in Table 1.

                                                                                                                                                 Downloaded from ajcn.nutrition.org by guest on February 10, 2015
Study selection                                                             Statistical heterogeneity was evaluated by using the I2 statistic,
   We identified studies that prospectively evaluated the asso-          which assessed the appropriateness of pooling the individual
ciation of an a priori score used for assessing adherence to             study results. The I2 value provided an estimate of the amount of
Mediterranean diet and adverse clinical outcomes. We excluded            variance across studies because of heterogeneity rather than
studies if they had a cross-sectional or case-control design, if they    chance. Where I2 was .50%, the heterogeneity was considered
analyzed adherence to a nonspecific dietary pattern or to a rec-         substantial. A small study bias or publication bias was appraised
ommended dietary guideline and not to Mediterranean diet, if             by visual inspection of the funnel plot of effect size against the
they evaluated cohort of patients who suffered from a prior              SE and, analytically, by the Egger’s test. Moreover, to in-
clinical event (ie, in a secondary prevention), if they did not adjust   vestigate possible sources of heterogeneity across studies, we
for potential confounders, and if they did not report an adequate        performed a meta-regression analysis to investigate the effects
statistical analysis.                                                    of various characteristics of studies on the study estimates of
                                                                         RRs.
Data extraction
   Two investigators (FS and AC) assessed potentially relevant           RESULTS
articles for eligibility. The decision to include or exclude studies        The updated search from recent years resulted in the identi-
was hierarchical and initially made on the basis of the study title,     fication of 7 additional prospective studies (11–17) published up
then of the study abstract, and finally of the complete study            to June 2010 that were not identified and included previously.
manuscript. Two researchers independently completed searches,            Characteristics of these recent studies are displayed in Table 1. Of
identification of studies, data abstraction, and tabulation, and         these, 1 study presented the overall mortality as a clinical out-
discordances were resolved by discussions. The following base-           come (14), 3 studies presented the incidence and/or mortality
line characteristics were extracted from the original articles by        from cardiovascular diseases (13, 15, 16), 1 study presented the
using a standardized data extraction form and included in the            incidence and/or mortality from neoplastic diseases (17), and 2
meta-analysis: lead author, year of publication, cohort name,            studies presented the incidence of neurodegenerative diseases
country of origin of the cohort, sample size of the cohort, number       (11, 12). Notably, 2 of the 7 studies reported clinical outcomes not
of outcomes, duration of follow-up, age at entry, sex, outcomes,         previously investigated, such as mild cognitive impairment and
components of the Mediterranean diet adherence score, and                stroke (12, 13). On the other hand, one study was an updated
variables that entered into the multivariable model as potential         analysis of a study already reported in the previous meta-analysis
confounders (Table 1).                                                   for the overall mortality outcome; therefore, only the most
   Outcomes of interests were overall mortality, mortality and/or        updated study was added to this updated final analysis (14). Taken
incidence from cardio- and cerebrovascular diseases, mortality           altogether, a total of 18 cohort prospective studies were included
and/or incidence from cancer, and incidence of neurodegenera-            and entered into the final analysis. As a whole, the updated
tive diseases.                                                           analysis determined a study population that reached a total of
                                                                         2,190,627 subjects analyzed. Included samples ranged in size
                                                                         from 161 to 485,044, with a follow-up time that ranged from 4 to
Quality assessment                                                       20 y.
   Study quality was assessed according to the following criteria:          Considerable variations among studies were observed in the
1) number of study participants, 2) duration of follow-up, and 3)        number of participants, sex and age of participants, length of
adjustment for potential confounders. Studies with a high                follow-up, and dietary components used to measure the adher-
number of participants, long duration of follow-up, and adjust-          ence score to the Mediterranean diet. Studies conducted in
TABLE 1
Study characteristics of the recent prospective studies investigating adherence to the Mediterranean diet and health outcomes1
Author, year (cohort)          Country          Sample size/total cohort         Outcome          Follow-up      Age      Sex     Components of the adherence score                  Adjustment

                                                       n/total n                                      y          y
Feàrt et al,           France                  50/1410                     Alzheimer                 5        68–952     M/F    1)   High legume intake                  Age, sex, BMI, hypertension,
  2009 (11)                                                                   disease                                            2)   High whole-grain products intake      hypercholesterolemia,
  (Three-City                                                                                                                    3)   High fruit intake                     smoking habit, education,
  Study)                                                                                                                         4)   High vegetable intake                 marital status, total energy,
                                                                                                                                 5)   High fish intake                      physical activity, medications,
                                                                                                                                 6)   High MUFA:SFA ratio                   depression, apoE genotype,
                                                                                                                                 7)   Moderate alcohol consumption          and stroke
                                                                                                                                 8)   Low red and processed meat intake
                                                                                                                                 9)   Low dairy product intake
Scarmeas et al,         United States           241/1199                    Mild                      4.5       76.73     M/F    1)   High legume intake                  Age, sex, ethnicity, education,
  2009 (12)                                       (only in those             cognitive                                           2)   High cereal intake                    apoE genotype, total energy,
  (Washington                                     who were                   impairment                                          3)   High fruit intake                     and BMI
  Heights-                                        cognitively                                                                    4)   High vegetable intake
  Inwood                                          normal at                                                                      5)   High fish intake
  Columbia                                        baseline)                                                                      6)   High MUFA:SFA ratio
  Aging                                                                                                                          7)   Moderate alcohol consumption
  Project)                                                                                                                       8)   Low meat intake
                                                                                                                                 9)   Low dairy product intake
Fung et al,             United States           2391/74,886                 CHD                      20         38–63     F      1)   High legume intake                  Age, smoking habit, BMI,
  2009                                          1763/74,886                  Stroke                                              2)   High whole-grain products intake      menopausal status, hormone
  (13) (Nurses’                                                                                                                  3)   High fruit intake                     use, total energy, multivitamin
  Health                                                                                                                         4)   High vegetable intake                 intake, alcohol, family history,
  Study)                                                                                                                         5)   High fish intake                      physical activity, and
                                                                                                                                 6)   High nuts intake                      aspirin use
                                                                                                                                 7)   High MUFA:SFA ratio
                                                                                                                                 8)   Moderate alcohol consumption
                                                                                                                                 9)   Low red and processed meat intake
                                                                                                                                                                                                               UPDATED META-ANALYSIS ON MEDITERRANEAN DIET

Trichopoulou            Greece                  1075/23,349                 Overall                   8.5       20–86     M/F    1)   High legume intake                  Age, sex, education, smoking
  et al, 2009                                                                 mortality                                          2)   High cereal intake                    habit, waist:hip ratio, MET
  (14) (EPIC-                                                                                                                    3)   High fruit/nuts intake                score, total energy, and BMI
  Greece)                                                                                                                        4)   High vegetable intake
                                                                                                                                 5)   High fish intake
                                                                                                                                 6)   High MUFA:SFA ratio
                                                                                                                                 7)   Moderate alcohol consumption
                                                                                                                                 8)   Low meat and poultry intake
                                                                                                                                 9)   Low dairy product intake
                                                                                                                                                                                                (Continued)
                                                                                                                                                                                                               1191

                                                                   Downloaded from ajcn.nutrition.org by guest on February 10, 2015
1192

TABLE 1 (Continued )

Author, year (cohort)           Country         Sample size/total cohort        Outcome         Follow-up      Age      Sex     Components of the adherence score                  Adjustment

                                                       n/total n                                    y           y
Buckland et al,         Spain                   606/40,151                  CHD                    10.4       29–69    M/F     1)   High legume intake                  Age, education, BMI, physical
  2009 (15)                                                                                                                    2)   High cereal intake                    activity, smoking habit,
  (EPIC Spain)                                                                                                                 3)   High fruit/nuts intake                diabetes, hypertension,
                                                                                                                               4)   High vegetable intake                 hyperlipidemia, and total
                                                                                                                               5)   High fish intake                      energy
                                                                                                                               6)   High MUFA:SFA ratio
                                                                                                                               7)   Moderate alcohol consumption
                                                                                                                               8)   Low meat and poultry intake
                                                                                                                               9)   Low dairy product intake
Martı́nez-González     Spain                   100/13,609                  CVD                     4.9        383     M/F     1)   High legume intake                  Age, sex, family history of CVD,
 et al, 2010 (16)                                                                                                              2)   High cereal intake                    total energy, physical activity,
 (SUN Study)                                                                                                                   3)   High fruit/nuts intake                smoking habit, BMI, diabetes,
                                                                                                                               4)   High vegetable intake                 aspirin, hypertension, and
                                                                                                                               5)   High fish intake                      hypercholesterolemia
                                                                                                                               6)   High MUFA:SFA ratio
                                                                                                                                                                                                             SOFI ET AL

                                                                                                                               7)   Moderate alcohol consumption
                                                                                                                               8)   Low meat and poultry intake
                                                                                                                               9)   Low dairy product intake
Buckland et al,         United Kingdom,         449/485,044                 Gastric                 8.9       35–70    M/F     1)   High legume intake                  Age, sex, BMI, education,
  2010 (17)               France, Denmark,                                    adenocarcinoma                                   2)   High cereal intake                    smoking habit, and total
  (Spanish cohort         Sweden, Germany,                                                                                     3)   High fruit/nuts intake                energy
  of the EPIC-Heart       Italy, Spain,                                                                                        4)   High vegetable intake
  Study)                  Netherlands,                                                                                         5)   High fish intake
                          Norway, and                                                                                          6)   High olive oil intake
                          Greece                                                                                               7)   Moderate alcohol consumption
                                                                                                                               8)   Low meat and poultry intake
                                                                                                                               9)   Low dairy product intake
    1
      MUFA, monounsaturated fatty acid; SFA, saturated fatty acid; CHD, coronary heart disease; EPIC, European Prospective Investigation into Cancer and Nutrition; MET, metabolic equivalent tasks; SUN,
Seguimiento Universidad de Navarra; CVD, cardiovascular disease.
    2
      Range (all such values).
    3
      Mean.

                                                                   Downloaded from ajcn.nutrition.org by guest on February 10, 2015
UPDATED META-ANALYSIS ON MEDITERRANEAN DIET                                                                     1193
European countries increased to 9 studies (50% of the total                        from and incidence of neoplastic diseases (Figure 3), a greater
number of studies included).                                                       adherence to the Mediterranean diet still determined a signifi-
   By grouping the studies according to the different clinical                     cant amount of protection, with a similar extent than that of
outcomes, the overall mortality did not change for the number                      previous meta-analysis (RR = 0.94; 95% CI: 0.92, 0.96; P ,
of studies evaluated (ie, 9 studies that included an updated                       0.00001) and again no significant heterogeneity (I2 = 6%; P =
analysis of the Greek cohort of the EPIC (European Prospective                     0.38). The results of the previous meta-analysis were also
Investigation into Cancer and Nutrition) with a total of 514,118                   confirmed for the occurrence of neurodegenerative diseases
subjects and 34,322 deaths, whereas a consistent increase of                       (Figure 4), which extended to mild cognitive impairment, with
studies and subjects analyzed for the other clinical outcomes                      a 13% reduced risk of individuals who most adhered to the
has been reported. Indeed, cardiovascular mortality and in-                        Mediterranean diet (RR = 0.87; 95% CI: 0.81, 0.94; P ,
cidence were consistently augmented by the number of studies                       0.00001) and no significant heterogeneity across studies (I2 =
and subjects analyzed, with a total of 7 cohorts, 534,064                          0%; P = 0.73).
subjects, and 8739 deaths or incident cases. Similarly, the
mortality from and incidence of neoplastic diseases reached
1,006,410 subjects and 11,378 events, whereas the number of                        Publication bias
studies that evaluated the association between adherence to the
Mediterranean diet and neurodegenerative diseases increased                          The funnel plots of effect sizes compared with SEs that were
from 3 to 5 cohorts, with a total of 136,235 subjects and 1074                     performed to investigate possible publication biases were sym-
cases, including an additional clinical outcome such as mild                       metric for all different clinical outcomes, which suggested the
cognitive impairment.                                                              absence of possible publication biases.

                                                                                                                                                                     Downloaded from ajcn.nutrition.org by guest on February 10, 2015
Meta-analysis                                                                      Meta-regression analysis
   Meta-analytic pooling under a random-effects model con-                            Finally, to investigate the effects of various study character-
firmed the already reported significant association between a 2-                   istics on the study estimates of the RRs, we conducted a meta-
point increased adherence to the Mediterranean diet and a reduced                  regression analysis by grouping studies according to some
risk of mortality from all causes (RR = 0.92; 95% CI: 0.90, 0.94;                  characteristics, such as the number of subjects studied, number of
P , 0.00001) (Figure 1) with no evidence of statistical                            cases observed, country of origin of the patients, age of the
heterogeneity across studies (I2 = 33%; P = 0.15). When studies                    patients, and length of follow-up. The sample size of the included
that evaluated different clinical outcomes were grouped, we                        studies was the only contributor to the protection compared with
observed that a 2-point increase of adherence to the Mediter-                      the overall mortality (b = 0.059 6 0.018; P = 0.013) in the
ranean diet still remained associated with a reduced risk of                       present model. By plotting the estimates of associations for
mortality from and incidence of cardiovascular diseases (RR =                      overall mortality with the number of subjects analyzed for each
0.90; 95% CI: 0.87, 0.93; P , 0.00001) (Figure 2), which                           study, it was shown that the average protection compared with
showed no significant heterogeneity across studies (I2 = 35%;                      the overall mortality was reached with studies that investigated
P = 0.15). Likewise, in studies that investigated the mortality                    .2000 subjects (Figure 5).

   FIGURE 1. Forest plot of the association between a 2-point increase of adherence score to the Mediterranean diet and the risk of all-cause mortality. The
center of each square indicates the relative risk of the study, and the horizontal lines indicate 95% CIs. The area of the square is proportional to the amount of
information from the study. The diamond indicates pooled estimates.
1194                                                                    SOFI ET AL

  FIGURE 2. Forest plot of the association between a 2-point increase of adherence score to the Mediterranean diet and the risk of mortality from or
incidence of cardiovascular diseases. The center of each square indicates the relative risk of the study, and the horizontal lines indicate 95% CIs. The area of

                                                                                                                                                                   Downloaded from ajcn.nutrition.org by guest on February 10, 2015
the square is proportional to the amount of information from the study. The diamond indicates pooled estimates. CHD, coronary heart disease.

DISCUSSION                                                                        neurodegenerative diseases. Furthermore, in this updated meta-
   In this updated meta-analysis of prospective studies that in-                  analysis, which included 7 articles published in the past 3 y that
vestigated the association between adherence to the Mediterranean                 could not be considered in the previous meta-analysis, some other
diet and health status, findings were pooled from .2 million                      clinical outcomes, such as stroke and mild cognitive impairment,
subjects and 50,000 deaths or incident cases from any causes and/                 were included, and a meta-regression analysis to investigate the
or from any cardiovascular, neoplastic, and neurodegenerative                     possible role of confounding factors was also conducted.
diseases. We confirmed the results of our previously published                       Several studies over the past decades showed the Mediterra-
meta-analysis, which suggested a significant protection against                   nean diet to be significantly associated with different health
major chronic degenerative diseases for subjects who reported                     outcomes, such as a favorable health status, better biochemical
a greater adherence to the Mediterranean diet (3). Indeed, a 2-                   profile, and quality of life (1–3, 26, 27). Accordingly, over-
point increase of adherence to the Mediterranean diet significantly               whelming evidence on the health effects of such a dietary profile
determined a 8% reduction of death from any causes, a 10% re-                     has been reported in many different study cohorts by stimulating
duction from death and/or incidence of cardio- and cerebrovas-                    both scientific and lay communities to promote this dietary
cular diseases, a 6% reduction from death and/or the incidence                    pattern as the ideal dietary pattern for preventing non-
of neoplastic diseases, and a 13% reduction of the incidence of                   communicable diseases (3, 27).

  FIGURE 3. Forest plot of the association between a 2-point increase of adherence score to the Mediterranean diet and the risk of mortality from or
incidence of neoplastic diseases The center of each square indicates the relative risk of the study, and the horizontal lines indicate 95% CIs. The area of the
square is proportional to the amount of the information from the study. The diamond indicates pooled estimates.
UPDATED META-ANALYSIS ON MEDITERRANEAN DIET                                                                   1195

  FIGURE 4. Forest plot of the association between a 2-point increase of adherence score to the Mediterranean diet and the risk of incidence of
neurodegenerative diseases. The center of each square indicates the relative risk of the study, and the horizontal lines indicate 95% CIs. The area of the
square is proportional to the amount of the information from the study. The diamond indicates pooled estimates.

   In this context, with the aim of identifying a practical tool to              the interest of researchers on this issue rose, and a total of 7 new
estimate how populations follow the guidelines of Mediterranean                  cohort prospective studies were added to the literature (11–17).

                                                                                                                                                                  Downloaded from ajcn.nutrition.org by guest on February 10, 2015
diet, adherence scores were computed (4). Many different scores                  On the other hand, the issue has been shown to be of extreme
were created and tested in different study cohorts, but the most                 interest for the general population as well as for health admin-
important was definitely that created in 1995 and developed in                   istrators and governments for better tailoring their prevention
its final form in 2003 by Trichopoulou et al (6, 18). In their                   strategies and policies.
studies, which investigated the Greek cohort of the EPIC study,                     The present updated meta-analysis has some advantages with
Trichopoulou et al showed, for the first time to our knowledge,                  respect to the previous one: first, it present, to out knowledge, the
a significant association between an increase of such an adher-                  most comprehensive and updated assessment on this issue, with
ence score and the reduction of overall mortality. Subsequently,                 inclusion of new studies and an update of previously reported
many studies that used this kind of epidemiologic approach have                  studies; second, the association between the Mediterranean diet
been released by extending the evidence to many different health                 and health outcomes was extended to other relevant disease states
outcomes (3).                                                                    that were not previously investigated, such as stroke and mild
   The results of the first meta-analysis published in 2008, which               cognitive impairment. These clinical outcomes are of clinical
comprised a total of 12 studies and .1.5 million subjects, were                  relevance for the ever-increasing age of the general population.
that a 2-point increase of score that estimated adherence to the                 Third, it reinforces the findings that a slight increase of the
Mediterranean diet significantly determined protection com-                      adherence score to the Mediterranean diet, as was observed by an
pared with premature death and various diseases. In the past 3 y,                increase of 2 points of the adherence score, is significantly

  FIGURE 5. Meta-regression analysis of the relation between sample sizes of the studies and all-cause mortality. The area of the circle is proportional to the
number of subjects studied in the study; the dashed line indicates the overall estimate of the association for studies that investigated all-cause mortality.
1196                                                                         SOFI ET AL

associated with a reduced risk of mortality and the incidence of                       2. Serra-Majem L, Roman B, Estruch R. Scientific evidence of inter-
main chronic degenerative diseases. Moreover, in the present                              ventions using the Mediterranean diet: a systematic review. Nutr Rev
                                                                                          2006;64:S27–47.
updated meta-analysis, we performed a meta-regression analysis                         3. Sofi F, Cesari F, Abbate R, Gensini GF, Casini A. Adherence to Medi-
to give deeper insight into the nature of the association between                         terranean diet and health status: meta-analysis. BMJ 2008;337:a1344.
the adherence score and health status. The meta-regression                             4. Bach A, Serra-Majem L, Carrasco JL, et al. The use of indexes evalu-
analysis showed that studies with larger study populations were                           ating the adherence to the Mediterranean diet in epidemiological studies:
                                                                                          a review. Public Health Nutr 2006;9:132–46.
more significantly associated with a lesser degree of heterogeneity                    5. Kourlaba G, Panagiotakos DB. Dietary quality indices and human
and a higher estimate of association. In particular, studies with                         health: a review. Maturitas 2009;62:1–8.
populations of .2000 subjects seemed to be more significantly                          6. Trichopoulou A, Costacou T, Bamia C, Tricopoulou D. Adherence to
related to the mean outcome observed in the meta-analysis. This                           a Mediterranean diet and survival in a Greek population. N Engl J Med
                                                                                          2003;348:2599–608.
finding could be easily explained by the increased power of the                        7. Knoops KTB, de Groot LCPGM, Kromhout D, et al. Mediterranean diet,
studies related to the higher number of subjects analyzed.                                lifestyle factors, and 10-year mortality in elderly European men and
   In contrast, although we believe that this updated meta-                               women. JAMA 2004;292:1433–9.
analysis provides useful information, the findings must be                             8. Lagiou P, Trichopoulos D, Sandin S, et al. Mediterranean dietary pattern
                                                                                          and mortality among young women: a cohort study in Sweden. Br J Nutr
interpreted with caution because of some weaknesses. First, the                           2006;96:384–92.
short interval of time that passed from the first meta-analysis and                    9. Mitrou PN, Kipnis V, Thiébaut AC, et al. Mediterranean dietary pattern
the present meta-analysis, as well as the not so high number of                           and prediction of all-cause mortality in a US population. Arch Intern
new studies, did not allow us to be confident that these updated                          Med 2007;167:2461–8.
                                                                                      10. Moher D, Liberati A, Tetzlaff J, Altman DG; PRISMA Group. Preferred
data can be used for all studied outcomes. Second, the use of                             reporting items for systematic reviews and meta-analyses: the PRISMA
meta-analysis technique to evaluate the protection of a dietary                           statement. BMJ 2009;339:b2535.

                                                                                                                                                                       Downloaded from ajcn.nutrition.org by guest on February 10, 2015
pattern, such as the Mediterranean diet, compared with the oc-                        11. Feàrt C, Samieri C, Rondeau V, et al. Adherence to a Mediterranean
currence of several and different diseases can lead to an over-                           diet, cognitive decline, and risk of dementia. JAMA 2009;302:638–48.
                                                                                      12. Scarmeas N, Stern Y, Mayeux R, Manly JJ, Schupf N, Luchsinger JA.
estimation of the results. Third, the transferability of such an                          Mediterranean diet and mild cognitive impairment. Arch Neurol 2009;
adherence score to the general population and the day-to-day                              66:216–25.
clinical practice is not free of practical problems. The adherence                    13. Fung TT, Rexrode KM, Mantzoros CS, Manson JE, Willett WC, Hu FB.
score’s ease of use is low because the method of calculation                              Mediterranean diet and incidence of and mortality from coronary heart
                                                                                          disease and stroke in women. Circulation 2009;119:1093–100.
implies the analysis of raw data that comes from a cohort study,                      14. Trichopoulou A, Bamia C, Trichopoulos D. Anatomy of health effects of
and the real amount of each food category to consume is not                               Mediterranean diet: Greek EPIC prospective cohort study. BMJ 2009;
established. Some relevant differences among the studies are                              338:b2337.
present, and the transferability to the clinical practice is difficult                15. Buckland G, González CA, Agudo A, et al. Adherence to the Medi-
                                                                                          terranean diet and risk of coronary heart disease in the Spanish EPIC
to understand. However, with the use of such meta-analyses, we                            Cohort Study. Am J Epidemiol 2009;170:1518–29.
are able to understand the real estimate of the association between                   16. Martı́nez-González MA, Garcı́a-López M, Bes-Rastrollo M, et al. Medi-
the adherence score to the Mediterranean diet and health status                           terranean diet and the incidence of cardiovascular disease: a Spanish co-
and to stimulate the interest of researchers to make uniform and                          hort. Nutr Metab Cardiovasc Dis (Epub ahead of print 20 January 2010).
                                                                                      17. Buckland G, Agudo A, Luján L, et al. Adherence to a Mediterranean
standardize, if possible, the score and its components.                                   diet and risk of gastric adenocarcinoma within the European Prospective
   In conclusion, we updated the results of our previous meta-                            Investigation into Cancer and Nutrition (EPIC) cohort study. Am J Clin
analysis and were able to confirm the estimates of the association                        Nutr 2010;91:381–90.
of adhering to the Mediterranean diet in terms of protection                          18. Trichopoulou A, Kouris-Blazos A, Wahlqvist ML, et al. Diet and overall
                                                                                          survival in elderly people. BMJ 1995;311:1457–60.
against overall mortality and incidences of various chronic de-                       19. Kouris-Blazos A, Gnardellis C, Wahlqvist ML, Trichopoulos D, Lukito
generative diseases, with an extension of the findings to some                            W, Trichopoulou A. Are the advantages of the Mediterranean diet
other health outcomes such as stroke and mild cognitive im-                               transferable to other populations? A cohort study in Melbourne, Aus-
pairment. Moreover, the results of the meta-regression analysis                           tralia. Br J Nutr 1999;82:57–61.
                                                                                      20. Lasheras C, Fernandez S, Patterson AM. Mediterranean diet and age
indicated that the number of subjects enrolled in a population                            with respect to overall survival in institutionalized, nonsmoking elderly
study is the most important contributor to the statistical model of                       people. Am J Clin Nutr 2000;71:987–92.
association. These results will likely help researchers and health                    21. Trichopoulou A, Orfanos P, Norat T, et al. Modified Mediterranean diet
administrators to better tailor forthcoming studies in the field of                       and survival: EPIC-elderly prospective cohort study. BMJ 2005;330:991.
                                                                                      22. Fung TT, Hu FB, McCullough ML, Newby PK, Willett WC, Holmes
nutrition and public health.                                                              MD. Diet quality is associated with the risk of estrogen receptor-negative
                                                                                          breast cancer in postmenopausal women. J Nutr 2006;136:466–72.
   The authors’ responsibilities were as follows—FS and AC: conception and
                                                                                      23. Benetou V, Trichopoulou A, Orfanos P, et al. Conformity to traditional
design of the study and drafting of the manuscript; FS, RA, GFG, and AC:                  Mediterranean diet and cancer incidence: the Greek EPIC cohort. Br J
analysis and interpretation of the data: FS; RA and GFG: critical revision of             Cancer 2008;99:191–5.
the article for important intellectual content; FS: statistical expertise and guar-   24. Scarmeas N, Stern Y, Tang MX, Mayeux R, Luchsinger JA. Mediter-
antor of the manuscript; and all authors: final approval of the manuscript.               ranean diet and risk for Alzheimer’s disease. Ann Neurol 2006;59:
None of the authors had a conflict of interest.                                           912–21.
                                                                                      25. Gao X, Chen H, Fung TT, et al. Prospective study of dietary pattern and
                                                                                          risk of Parkinson disease. Am J Clin Nutr 2007;86:1486–94.
                                                                                      26. Sanchez-Tainta A, Estruch R, Bullo M, et al. Adherence to a Mediterranean-
REFERENCES                                                                                type diet and reduced prevalence of clustered cardiovascular risk factors
 1. Martinez-Gonzalez MA, Bes-Rastrollo M, Serra-Majem L, Lairon D,                       in a cohort of 3204 high-risk patients. Eur J Cardiovasc Prev Rehabil
    Estruch R, Trichopoulou A. Mediterranean food pattern and the primary                 2008;15:589–93.
    prevention of chronic disease: recent developments. Nutr Rev 2009;67:             27. Sofi F. The Mediterranean diet revisited: evidence of its effectiveness
    S111–6.                                                                               grows. Curr Opin Cardiol 2009;24:442–6.
You can also read