Adherence to the Mediterranean diet by the Greek and Cypriot population: a systematic review

Abstract

Background: The traditional Mediterranean diet is defined as the dietary pattern in the countries of the Mediterranean basin between the 1950s and 1960s, and it is now widely accepted that has a beneficial effect on health. A debate exists from empirical and research data if the traditional Mediterranean diet remains the main dietary pattern of the region or if it has changed overtime. Aims: This systematic review addresses whether the people of Cyprus and Greece still follow the traditional Mediterranean diet or whether the diet has become more ‘Westernised’. Methods: The MEDLINE database was searched using relevant free terms and independently reviewed by two authors. In addition, all reference lists of identified studies were hand-searched to identify additional, relevant studies. Results: The review resulted in 18 research papers that met the inclusion and exclusion criteria and represented 15 independent studies. The main outcome was consistent between studies and indicated moderate adherence of the Greek, and (probably) of the Cypriot, population to the Mediterranean diet. The majority of studies found no statistically significant differences by gender. There was an observed inter-study lower adherence to the Mediterranean diet by the younger population. Few studies addressed intra-study variations by age. Conclusions: This review shows that adherence to the Mediterranean diet is moderate in Greece (and probably also in Cyprus).This suggests a continuing transition from dietary patterns in the 50 s–60 s towards a more Westernized diet.

Introduction

Interest in the Mediterranean diet increased after the widely cited Seven Countries study ¹ suggested a cardio-protective role of the Mediterranean dietary pattern. According to Trichopoulou et al .² ‘The Mediterranean diet could be considered as the dietary pattern found in the olive oil-growing area of the Mediterranean region in the late 1950s and early 1960s, before the invasion of the fast food culture in the area’. The traditional Mediterranean diet is based mainly on plant sources such as vegetables, fruits, whole grains, legumes and seeds. Fish and poultry are consumed weekly and dairy products daily, in low to moderate amounts, with an emphasis on fish. The primary source of fat intake is from olive oil and meals are accompanied by moderate amounts of wine. ²

The Mediterranean diet is associated with reduction in risk of a number of diseases and conditions. It has been found to reduce all-cause mortality, ³ the incidence and mortality from cardiovascular disease and cancer ^2,3 and the incidence of neurodegenerative diseases such as Alzheimer and Parkinson’s disease ³ and of Type 2 diabetes. ⁴

Despite these health benefits, there is debate as to whether adherence to the Mediterranean diet in modern Greek and Cypriot society remains the same or whether people have moved towards a more ‘Westernised/Americanised’ diet. Ecological studies using the food balance sheets of the United Nations’ Food and Agriculture Organization have shown a significant drop in the Mediterranean dietary pattern among the Mediterranean countries over the last decades, with Greece experiencing the biggest drop and Cyprus also having a considerable negative change. ^5–7 We aimed to examine the evidence for the current/recent adherence of the Greek and Cypriot population to the Mediterranean diet, after the significant changes that have appeared over the last decades, including food availability.

Methods

This study aims to systematically review the evidence on adherence of the Greek and Cypriot population to the Mediterranean diet.

Search strategy

A MEDLINE search was conducted, up to 15 July 2013, using the free terms ‘adheren* or prevelan*’(topic) and ‘Mediterranean diet*’(topic) and ‘gre* or cypr*’(topic). The search was conducted independently by two researchers (A.K. and N.E.), and any disagreement was settled after discussion between the two. The papers were reviewed progressively in stages on the basis of the title, abstract and finally full text. One of the authors (A.K.) hand-searched references of all the identified papers ( figure 1 ). The search was not limited to language, date or any other limitations.

Figure 1

MEDLINE database and other search procedures

To be included in the review, the study had to fulfil the following criteria:

• The study was conducted in a Cypriot or Greek population or both nationalities who were permanently residing in Cyprus or Greece or both.

• Adherence to the Mediterranean diet was measured as a primary or secondary outcome of the total study population and/or of the total population of each gender of the study and/or of different age groups.

• A Mediterranean diet score was used.

• A direct method of data collection such as questionnaires, interviews, self-reported individual methods and surveys was used.

• The data were not collected after an intervention; i.e. the data must have originated from an observational cross-sectional or longitudinal study or at baseline of an intervention study (before the intervention commenced).

Studies which were not original (e.g. reviews and meta-analysis) or that used an indirect method of measurement of data collection (e.g. the World Health Organization/Food and Agriculture Organization data or household food surveys) were excluded from the review, as were studies that collected data after an intervention.

When the same study population or a partially overlapping population was described in several papers, then only the study with the highest sample size was included in the review. The author of the study was conducted when unsure about the population used or when it was thought that there might be overlap.

Data extraction

The data extraction process was conducted by A.K. and reviewed by N.E. Information extracted included the author(s), title, publication media, date of publication, study design, year of survey, population characteristics, sample size, age range, Mediterranean diet scoring system used, health status, Mediterranean diet score (Mdiet score) by country and/or gender and/or age group and/or of the total population.

Statistics

Mean score and standard deviation

The total mean score and standard deviation of the Mdiet scores were noted, but if they were reported for subgroups only, the total population’s mean score and standard deviation were calculated using appropriate formulae. In the same way, the mean and standard deviation, for all studies using the same Mdiet score, was calculated and is reported as ‘Total’ (see Supplementary Appendix for equation used).

Percentages of adherence to the Mediterranean diet

There are a number of scoring systems that can be used to measure adherence to the Mediterranean diet. These scoring systems have several differences, e.g. different maximum and minimum possible scores and different definitions of low, medium and high adherence. This makes inter-study comparison difficult. In addition to considering the definitions presented in the original research papers, we calculated a measure of percentage adherence to enable comparison between studies that used different scoring systems. This was defined as the minimum Mdiet score subtracted from the Mdiet score of the total population, then divided by the difference of the maximum and minimum Mdiet score of the particular scoring system and multiplied by 100%. ^8–11 However, for the KIDMED scoring system, the lowest possible score was not zero, but −4. Therefore, the total maximum score was 16, which is the highest possible score (12) minus the lowest possible score (−4).

We defined adherence cut-off points as tertiles. Below the first cut-off point, or 33.33% adherence, participants were considered to have low adherence, between the second and third tertile (33.33–66.67%), they were considered to have moderate adherence. Finally, above the 66.67% cut-off point, participants were considered to have good adherence to the Mediterranean diet.

Results

The initial MEDLINE search resulted in 108 papers. After the three-stage reviewing process, 92 papers were excluded. A hand search of references generated a further two papers. The final total of 18 papers represented 15 independent studies ( figure 1 ).

All 15 included studies were cross-sectional. They were all conducted in Greece apart from one study ^8,9 that had a mixture of Cypriot and Greek participants. The demographic characteristics of the study populations are presented in table 1 and represent a diverse range of groups, such as general adult, paediatric and elderly, school and university students. No study that represented a group of people with a common health background such as cardiovascular problems or diabetes met the criteria for inclusion.

Table 1

Demographic characteristics of the included studies

Author	Year of data collection	Country; area; representative	Sample size	Age range (years)	Health status
Costarelli et al. 12	Not available	Greece; Athens and Dodecanese	359	13–16	General population
Katsiardanis et al. 13	2005–2006	Greece; Valestino	557	>65	General population
Lydakis et al. 14	2011	Greece; Heraklion, Crete	277	12	No clinical history of cardiac (congenital) or renal conditions, diabetes mellitus, receiving immunosuppression or cytotoxic drugs
Chrysohoou et al. 10	2009	Greece; Ikaria island	538	>65	No clinical history of CVD, other atherosclerotic disease and use of diuretic drugs
Farajian et al. 15	2009	Greece; 10 regions; Representative	4786	10–12	General school population
Filippidis et al. 16	2006	Greece; Representative	1005	18–99	General population
Van Diepen et al. 17	2008	Thessaloniki; Greece	85	21.6 ± 3.2 a	Healthy university students
Tyrovolas et al. 8 Tyrovolas et al. 9	2005–2007	Greece and Cyprus; 7 islands in Greece and the Republic of Cyprus; Representative	1190	> 65	No clinical history of CVD, cancer or institutionalized
Arvaniti et al. 18	2005–2006	Greece; Athens	700	10–12	School boys and pre-menstrual girls
Kontogianni et al. 19	2007	Greece; Representative	1305	3–18	General population
Kanoni and Dedoussis 20	Not available	Greece; Athens	782	> 60	General population
Dedoussis et al. 21	Not available	Greece	163	>60	Healthy, non-institutionalized, free of medications and chronic conditions
Panagiotakos et al. 22 Panagiotakos et al. 23 Arvaniti et al. 24	2001–2002	Greece; Attica	3042	18–89	No clinical history of CVD, other atherosclerotic disease, chronic viral disease and surgery the week before the data collection
Psaltopoulou et al. 25	1994–1999	Greece	20 343	20–86	No diagnosis of hypertension; volunteers
Doupis et al. 26	Not available	Greece	832	17–39	General navy recruits

CVD cardiovascular disease.

a: Mean ± standard deviation when range is not available.

The results of the 15 independent cross-sectional studies are shown in table 2 and in figure 2 in a separate graph for each scoring system. The cut-off points used in figure 2 to define low, moderate or high adherence were those defined in the original research papers. Note that different studies used different terms for ‘moderate’ adherence (e.g. average and medium). We use the term ‘moderate’ throughout this review.

Figure 2

Results using different scoring systems with original definitions of adherence levels. (a) KIDMED scoring system. (b) MedDietScore scoring system. (c) MDS scoring system. (d) 44-point scoring system

Table 2

Mdiet score of the included studies

References	Mean Mdiet score of all population	Mean Mdiet score of female population	Mean Mdiet score of male population	P value between females and males	Mean Mdiet score of age group categories (years)	P value between age groups
KIDMED
Costarelli et al. 12	6.3 ± 2.4	6.2 ± 2.5	6.3 ± 2.5	P = DNS	DNS
Lydakis et al. 14	6.6 ± 2.20	6.82 ± 2.17	6.48 ± 2.22	P = 0.197	DNS
Farajian et al. 15	3.65 ± 2.27	3.66 ± 2.24	3.64 ± 2.29	P = 0.86	DNS
Arvaniti et al. 18	4.8 ± 2.0 a	4.8 ± 2.0	4.8 ± 1.9	P = 0.87	DNS
Kontogianni et al. 19	5.13 ± 1.9 a	5.17 ± 1.9 a	5.09 ± 2.0 a		5.4 ± 1.8 (3–12) 4.8 ± 2.1 (13–18)	P = DNS
MedDietScore
Katsiardanis et al. 13	34.7 ± 2.87 a	35.1 ± 2.48	34.1 ± 3.25	P = DNS	DNS
Chrysohoou et al. 10	35 ± 2	35 ± 3	34 ± 2	P = 0.26	DNS
Van Diepen et al. 17	26.1 ± 3.4	DNS	DNS	P = NS	DNS
Tyrovolas et al. 8 and Tyrovolas et al. 9	33.5 ± 4.0	33.7 ± 3.8	33.3 ± 4.3	P = 0.10	DNS
Kanoni and Dedoussis 20	30.0 ± 3.2 a	29.7 ± 3.0	30.7 ± 3.4	P < 0.001	DNS
Dedoussis et al. 21	28.2 ± 3.8 a	27.7 ± 3.4	29.0 ± 4.2	P = SS b	DNS
Panagiotakos et al. 22 –Arvaniti et al. 24	26 ± 3	27.18 ± 3.21	25.46 ± 2.94	P < 0.001	DNS
Doupis et al. 26	24.5 ± 3.68	DNS	DNS		DNS
MDS
Psaltopoulou et al. 25	4.4 ± 1.69 a	4.3 ± 1.67	4.5 ± 1.70	P < 0.001	DNS
44-point score
Filippidis et al. 16	26.59 ± 6.23	26.14 ± 6.17	27.09 ± 6.25	P = 0.017	25.84 ± 6.58 (18–36) 26.92 ± 5.93 (37–56) 27.05 ± 6.08 (57–99)	P = 0.011

Mdiet score, Mediterranean diet score; DNS, data/value was not shown; NS, not statistically significant (but P value not given); SS, statistically significant.

a: Value not available but calculated using the formulas mentioned in the Methods section—Statistics .

b: After age adjustment. P < 0.05 (but P value not given).

The most commonly used scoring system was the MedDietScore used by eight independent studies. ^{8–10,13,17,20–24,26} The point range of the score was 0 to +55 and the cut-off points used to define adherence were the 2nd and 3rd tertiles, that is 33.33% and 66.67% (represented by scores of 18.3 and 36.7) ( figure 2 b). All five paediatric studies ^{12,14,15,18,19} used the KIDMED scoring system that has a possible score range between −4 and +12 and cut-off points at 46.88% and 71.88% (representing scores of +3.5 and +7.5; figure 2 a).The MDS ²⁵ and 44-point scoring system ¹⁶ were less widely used (see figure 2 c and d). Figure 3 presents the results with adherence cut-off points defined as tertiles for all scoring systems.

Figure 3

Percentage adherence to the Mediterranean diet

The results show consistent moderate adherence to the Mediterranean diet, whether the original Mdiet score system and cut-off points are used ( figure 2 ) or whether they were converted to percentages and the 2nd and 3rd tertiles are used as the cut-off points ( figure 3 ), regardless of the study population.

Most studies found no statistically significant difference between genders. ^{8–10,14,15,17,18} Four studies showed statistically significant higher adherence to the Mediterranean diet among males ^16,20,21,25 and only one for females, ^22–24 with four studies not reporting the P value ^12,13,19 or the gender adherence Mdiet scores. ²⁶

Furthermore, although a visual comparison of figure 2 a and b might suggest general lower adherence in the younger population (using the KIDMED score) than the adult population (using the MedDietScore), this difference becomes less apparent in figure 3 . Only two studies investigated adherence by age. One study in the adult population showed a statistically significant increase with age. ¹⁶ One in the paediatric population showed higher adherence among younger children, but no statistical test was reported. ¹⁹

Discussion

The current review has investigated adherence to the Mediterranean diet of two related Greek-speaking Mediterranean populations, namely the Greek and Cypriot populations. All studies show consistent results regarding adherence to the Mediterranean diet: widespread moderate adherence. These results were seen independently, whether the original Mdiet score and cut-off points were used ( figure 2 ) or whether results were converted to percentages and tertiles were used as cut-off points ( figure 3 ). Similar results, i.e. moderate adherence, have been observed when using a Mdiet scoring system in several other countries of the Mediterranean region such as Spain, ²⁷ Italy ^28,29 and France ²¹ but also elsewhere in Europe, such as the Netherlands, ¹⁷ Germany and Poland. ²¹

This review suggests that adherence to the Mediterranean diet has decreased since the 1950s and 1960s, when it was high by definition. ² Almost all studies included in this review were conducted after 2000. This may be because interest has changed from single nutrients to whole diets in the last decade ³⁰ ; with a corresponding development of the various Mdiet scores. The MDS scoring system was developed in the decade between 1990 and 2000, while the rest followed after the year 2000. This finding is backed up by data that used the food balance sheets of the Food and Agriculture Organization of the United Nations of the early 1960s and just before ⁷ or after 2000. ^5,6 These data showed a statistically significant large decrease (56.3%) in the Mediterranean Adequacy Index (MAI) in Mediterranean Europe, including Cyprus and Greece (3.39 ± 1.12 to 1.48 ± 0.39, P = 0.0009, MAI); in contrast, a non-statistically significant increase was observed in Northern countries and a decrease in Eastern European countries. ⁷ While Greece ranked first between 1961 and 1965 using the MAI score, among the 41 countries studied, it ranked only 10th between 2000 and 2003. Cyprus dropped from 20th place to 27th place during the period. ⁵ Although this study, in line with the previous studies, ^5–7 reports a decrease in adherence to the Mediterranean diet, it is not possible to determine with certainty when exactly this happened; i.e. if there was regular decline over time or whether there were periods of rapid change, or alternatively, whether this phenomenon was limited to isolated time intervals. There is insufficient data on this topic between the 1960s and the 2000s to determine this.

Although moderate adherence seems to be the norm in most countries studied, some variation does exist. Van Diepen et al .¹⁷ who compared the Mdiet score of Greek and Dutch university students, found an unexpected higher statistically significant adherence in favour of the non-Mediterranean origin Dutch students (26.1 ± 3.4, 27.5 ± 3.9, P ≤ 0.05, MedDietScore). Dedoussis et al .²¹ who studied elderly subjects from five different countries, showed that Greece had a statistically significant lower Mdiet score when compared with Italy, a higher one than Poland, but no discernable difference from Germany and France. Comparison between Greek and Cypriot populations in the present review was not possible as there was very limited data on Cyprus; only one independent study was found which had a mixture of Cypriot and Greek subjects. ^8,9

The majority of studies included in the current review reported no statistically significant difference between genders ( table 2 ). Only four studies showed a gender difference, of which three independent studies ^20,21,25 reported a significant higher Mdiet score for males and one independent study ^22–24 reported a higher Mdiet score for females. Two studies in the Italian population ^21,28 reported a statistically significant higher score for males, while in the other populations studied, namely the French, German and Polish, there were no statistically significant differences between male and female participants. ²¹

There is a question as to whether adherence to the Mediterranean diet is lower in younger populations. Results from inter-study comparison show a probable lower but small difference in adherence of the younger population against the older population (51.9% vs. 52.5% respectively; figure 3 ). This difference in adherence was more obvious when using the original Mediterranean diet scoring systems ( figure 2 ). This discrepancy may be explained by the considerable different cut-off points used by the MedDietScore (33.33% and 66.67%; figure 2 b) that was used for the study of adult population and the KIDMED score (46.88% and 84.38%; figure 2 a), used for the study of the paediatric population. Only two studies investigated the Mdiet score between various age groups within the same study population. The first study among adults showed similar results to the inter-study comparison, i.e. a statistically significant increase with age ( P = 0.011). ¹⁶ The second study in a paediatric sample reported higher adherence among the younger group (3–12 years) than the older age group (13–18 years). ¹⁹ The heterogeneity of the result of the second study may reflect the fact that he diets of younger aged children are likely to be influenced by other parameters such as food provided by the parents. ³¹

The aforementioned change in the diet of Greek-speaking populations in the South Eastern Mediterranean basin reflects a transition from the traditional Mediterranean diet to a more Westernized diet and may (at least partially) explain the observed deterioration in the health of these populations. Indeed, in recent decades, these populations have experienced an inexorable increase in the incidence of diet-related disease such as obesity, diabetes, cardiovascular disease, some cancers, dyslipidaemia and hypertension. ³² The beneficial association of the Mediterranean diet pattern to health, and reduction in diet-related diseases, is supported by several studies. ^3,4,33 Other contributing non-diet, modified lifestyle-related risk factors may include the high prevalence of smoking ³⁴ and reduced physical activity observed in the Greek population. ³⁵ In an effort to take these factors into account, Polychronopoulos et al .³⁶ have proposed the Mediterranean lifestyle scoring system. This Mediterranean lifestyle scoring system combines the Mediterranean diet, alcohol consumption, physical activity and smoking to create a score with range from 0 to 4. This scoring system is not yet used extensively and even more importantly, its results need to be shown that are related to hard point clinical outcomes such as mortality and morbidity in order to prove its usefulness.

The KIDMED diet score is the Mediterranean diet scoring system of choice for all five studies, ^{12,14,15,18,19} which investigated a paediatric population. In comparison, a number of diet scores were used for the adult population. The most popular of the adult Mediterranean scoring systems was the MedDietScore ( table 2 ). The different type of scoring systems used to measure the adherence to the Mediterranean diet makes inter-study comparison more difficult, ³⁷ and for this reason, we decided to use the percentage of adherence to the Mediterranean diet ( figure 3 ), although we are aware of its limitations. Each Mdiet scoring system has its own advantages and limitations, which are explored elsewhere ^38–40 and are beyond the aims of this study.

In conclusion, this review has brought together evidence on adherence to the Mediterranean diet in Greece and Cyprus. The included studies have been carried out in a diverse range of study populations, and we consider that the moderate adherence observed is likely to be valid picture. Given the well-documented health benefits of the Mediterranean diet, evidence that adherence is following a downward path with time and may be lower in younger populations is worrisome. Public health and education measures need to be implemented if this decrease in adherence to the Mediterranean dietis to be reversed.