Meta-Analysis Comparing Mediterranean to Low-Fat Diets for Modification of Cardiovascular Risk Factors

Article Outline

• Abstract
• Methods
  • Literature Search
  • Study Selection and Quality Assessment
  • Outcomes and Data Extraction
  • Statistical Analysis
• Results
  • Unbalanced Co-interventions among Included Trials
  • Body Weight, Body Mass Index, and Waist Circumference
  • Blood Pressure
  • Lipid Values
  • High-sensitivity C-reactive Protein
  • Fasting Plasma Glucose and Serum Insulin
  • Clinical Outcomes
  • Sensitivity Analyses
• Discussion
• Appendix 1
  • Characteristics of Included Trials
  • Quality of the Trials
• Appendix 2
• References
• Copyright

Abstract 

Background

Evidence from individual trials comparing Mediterranean to low-fat diets to modify cardiovascular risk factors remains preliminary.

Methods

We systematically searched MEDLINE, EMBASE, Biosis, Web of Science, and the Cochrane Central Register of Controlled Trials from their inception until January 2011, as well as contacted experts in the field, to identify randomized controlled trials comparing Mediterranean to low-fat diets in overweight/obese individuals, with a minimum follow-up of 6 months, reporting intention-to-treat data on cardiovascular risk factors. Two authors independently assessed trial eligibility and quality.

Results

We identified 6 trials, including 2650 individuals (50% women) fulfilling our inclusion criteria. Mean age of enrolled patients ranged from 35 to 68 years, mean body mass index from 29 to 35 kg/m². After 2 years of follow-up, individuals assigned to a Mediterranean diet had more favorable changes in weighted mean differences of body weight (−2.2 kg; 95% confidence interval [CI], −3.9 to −0.6), body mass index (−0.6 kg/m²; 95% CI, −1 to −0.1), systolic blood pressure (−1.7 mm Hg; 95% CI, −3.3 to −0.05), diastolic blood pressure (−1.5 mm Hg; 95% CI, −2.1 to −0.8), fasting plasma glucose (−3.8 mg/dL, 95% CI, −7 to −0.6), total cholesterol (−7.4 mg/dL; 95% CI, −10.3 to −4.4), and high-sensitivity C-reactive protein (−1.0 mg/L; 95% CI, −1.5 to −0.5). The observed heterogeneity across individual trials could, by and large, be eliminated by restricting analyses to trials with balanced co-interventions or trials with restriction of daily calorie intake in both diet groups.

Conclusion

Mediterranean diets appear to be more effective than low-fat diets in inducing clinically relevant long-term changes in cardiovascular risk factors and inflammatory markers.

Keywords: Low-fat diet, Mediterranean diet, Meta-analysis

Unhealthy diet and physical inactivity are major risk factors for cardiovascular disease in the US, leading to 400,000 excess deaths in the year 2000.¹ Different dietary approaches for cardiovascular risk factor management have been investigated. Low-fat, high-carbohydrate diets may have benefits such as continuing weight loss for 3 years, prevention of type 2 diabetes, and improved control of hypertension as shown in clinical trials.² However, there is a lack of evidence in terms of a benefit on cardiovascular mortality.³

The traditional Mediterranean, moderate-fat diet is becoming increasingly promoted as a model of healthy eating,⁴, ⁵, ⁶ despite the lack of evidence that this diet leads to sustained weight loss.⁷ It is characterized by a high intake of monounsaturated fat, plant proteins, whole grains, and fish; moderate intake of alcohol, and low consumption of red meat, refined grains, and sweets.⁴ In several cohort studies, Mediterranean diet was associated with a reduced incidence of coronary heart disease and stroke as well as cardiovascular, cancer, and overall mortality.⁸, ⁹, ¹⁰, ¹¹

The goal of this meta-analysis was to summarize the evidence of all randomized controlled trials with a minimum follow-up of 6 months comparing the effects of Mediterranean to low-fat diets on cardiovascular risk factors. We restricted the analysis to overweight/obese individuals with at least one additional cardiovascular risk factor because we did not identify any trial comparing the effects of the 2 diets in normal-weight individuals.

Back to Article Outline

Methods 

Literature Search 

We searched the electronic databases MEDLINE, EMBASE, Biosis, Web of Science (all from their inception to January 2011), and the Cochrane Central Register of Controlled Trials using the terms “diets, fat restricted [Mesh]“ and “Mediterranean diets.” We restricted the search to articles indexed as clinical trial (publication type) and those that included the root random in their titles or abstracts. We also searched reference lists of identified articles, clinical trial registries of ongoing or planned trials, recently published editorials and reviews on the topic, and we contacted experts in the field for further eligible trials. No language restrictions were imposed.

Study Selection and Quality Assessment 

Two authors (KS, AN) independently assessed trial eligibility and quality. Eligible trials had to compare Mediterranean with low-fat diets in either overweight/obese patients with at least one additional cardiovascular risk factor (primary prevention) or patients with established coronary artery disease (secondary prevention); to have a randomized controlled design and a minimum follow-up of 6 months; and to report intention-to-treat data on changes of body weight, blood pressure, and lipid values. We included trials where Mediterranean diets were defined as diets with moderate fat intake (where the main sources of added fat were olive oil and nuts), rich in vegetables, and low in red meat (with poultry and fish replacing beef and lamb).¹² Low-fat diets were defined as diets aiming at an energy intake with ≤30% of calories from fat.¹³ We evaluated the quality of trials according to concealment of treatment allocation; blinding of patients, caregivers, and clinical outcome assessors; full description of losses to follow-up and withdrawals; the proportion of patients with complete clinical follow-up; and trials not stopping early for benefit.¹⁴, ¹⁵

Outcomes and Data Extraction 

Two authors (KS, AN) independently extracted published trial data and additional data provided by the original investigators. We considered the following cardiovascular risk factors baseline and 2 years of follow-up as outcomes of interest: mean differences in body weight, body mass index, waist circumference, systolic and diastolic blood pressure, total high-density lipoprotein (HDL) cholesterol, low-density lipoprotein cholesterol, high-sensitivity C-reactive protein (hs-CRP), fasting plasma glucose, and serum insulin between. In addition, we extracted any clinical outcome data when available.

Statistical Analysis 

We pooled treatment effects and calculated weighted mean differences for all risk factors between patients randomized to Mediterranean and low-fat diets by using a random-effects model.¹⁶ Because we could not obtain standard deviations for the differences of the means of risk factors from one trial,¹⁷ we first calculated the standard errors by dividing the differences of the means by the percentage points of the t-distributions corresponding to the P-values given, and then calculated the standard deviations by multiplying the standard errors by the square roots of the number of observations.

We investigated the presence of publication bias by means of funnel plots.¹⁸ We tested for heterogeneity with the Cochran Q test and measured inconsistency (I²; the percentage of total variance across studies that is due to heterogeneity rather than chance) of treatment effects across all cardiovascular risk factors of interest.¹⁹, ²⁰ We conducted sensitivity analyses to examine treatment effects according to quality components of included trials; primary versus secondary prevention trials, trials with balanced versus trials with unbalanced co-interventions, and trials with restriction of daily calorie intake versus trials without restriction of daily calorie intake. We used Stata 10.1 (StataCorp LP, College Station, Tex) for data analysis.

Back to Article Outline

Results 

Seven trials including a total of 3650 patients fulfilled our inclusion criteria (Figure 1). One of the identified trials was the Indo-Mediterranean Diet Heart Study.²¹ Because the validity of this trial, which includes 1000 patients, has been seriously questioned,²² we did not include the results of this trial in our primary analysis. However, because the paper has never been officially retracted, we conducted a sensitivity analysis including this trial to evaluate whether the results of our meta-analysis would change after its inclusion. We did not include the Lyon Diet Heart Study²³ in our analyses because changes in cardiovascular risk factors were only assessed on a per-protocol, but not on an intention-to-treat, basis.

• 
  • View Large Image
  • Download to PowerPoint

• Figure 1. 

  Trial flow. ITT=intention to treat; RCT=randomized controlled trial.

The relatively small number of included trials precluded a sensitive exploration of publication bias, although the plots of standardized effect against precision for all outcomes did not indicate evidence for such a bias (P >.1).¹⁸

Characteristics and methodological quality of included trials are summarized in Table 1.⁶, ¹³, ¹⁷, ²¹, ²⁴, ²⁵, ²⁶, ²⁷, ²⁸, ²⁹ Follow-up of included trials was 2 years in 4 trials¹⁷, ²⁴, ²⁵, ²⁶ and 4 years in one trial.²⁷ The Predimed trial⁶ is still ongoing and is planned to have a mean follow-up of 6 years; we included 2-year follow-up data of this trial in our meta-analysis except for laboratory analyses, which were measured after only 1 year of follow-up and pooled with the 2-year lipid values of the other trials (for further details of included trials see Appendix 1, online).

Table 1. Characteristics of Included Trials

Study (First Author, Year)	Primary/Secondary Prevention and Inclusion Criteria	Recruitment Period and Place	Follow-up (Years)	Caloric Restriction	Recommendations for Mediterranean Diet	Recommendations for Low-fat Diet	Co-interventions	Reported Concealed Allocation/Blinded Assessors/Loss to Follow-up <10%, % Follow-up
Esposito, 200317	Primary prevention: Obese premenopausal women (20-46 years), <1 hour of physical activity per week.	2/1999 to 2/2002; Outpatient Department of the Division of Metabolic Diseases, Second University of Naples, Italy	2	MD group only (mean caloric intake ≤1300 kcal/day first year, ≤1500 kcal/day year)	50%-60% carbohydrates; 15%-20% proteins; <30% total fat, <10% saturated fat; 10%-15% monounsaturated fat, 5%-8% polyunsaturated fat; 18 g of fiber per 1000 kcal28	<30% of daily caloric intake from fat	MD group: Individualized program to reduce dietary calories, to set personal goals and optimize self-monitoring, monthly sessions with nutritionist and exercise trainer in the first year and offer of behavioral and psychological counseling LF group: None	Yes/Yes (for laboratory values)/Yes, 93%
Esposito, 200424	Primary prevention: Sedentary individuals (<1 hour of physical activity per week) with metabolic syndrome as defined by the Adult Treatment Panel III criteria ⁎	6/2001 to 1/2004; Outpatient Department of the Division of Metabolic Diseases, Second University of Naples, Italy	2	No	50%-60% carbohydrates;15%-20% proteins; <30% total fat, <10% saturated fat; <300 mg of cholesterol per day; ≥250-300 g fruits, ≥125-150 g of vegetables, ≥25-50 g walnuts, encouraged to eat 400 g of whole grain/day, increase of intake of olive oil	50%-60% of calorie intake from carbohydrates, 15%-20% from proteins, <30% from total fat	MD group: Individualized program to reduce dietary calories, to set personal goals and optimize self-monitoring, monthly sessions with nutritionist in the first year LF group: None	No/No/Yes, 91%
Shai, 200825	Primary and secondary (40% of included individuals) prevention: Obese (BMI >27 kg/m2) patients (40-65 years), or type 2 diabetes or coronary heart disease (independent of age and BMI)	7/2005 to 6/2007; Workplace at research center in Dimona, Israel	2	Yes, both groups (mean caloric intake ≤1500 kcal/day for women, ≤1800 kcal/day for men)	≤35% of calories from fat (main source of fat 30-45 g of olive oil and <20 g of nuts/day); diet rich in vegetables, low in red meat (poultry and fish replacing beef and lamb)	30% of calories from fat, 10% from saturated fat, <300 mg of cholesterol/day; low-fat grains, vegetables, fruits and legumes; limit additional fats, sweets and high-fat snacks.	None	Yes/Yes/No, 88%
Tuttle, 200826	Secondary prevention: Confirmed diagnosis of first MI (<6 weeks)	7/2000 to 6/2005; Heart Institute of Spokane, WA, USA	2	No	I ≤200 mg of cholesterol/day; ≤7% from saturated fat calories; increased intake of omega-3 fatty acids (>0.75% of calories) and mono-unsaturated fat (20%-25% of calories); cold-water fish 3-5 times/week and oils from olives, canola, and soybeans	American Heart Association Step II diet: ≤200 mg of cholesterol/day, ≤7% from saturated fat calories	None	Yes/Yes/Yes, 62%
Estruch, 2008(Predimed study)6	Primary prevention: Type 2 diabetes or ≥3 cardiovascular risk factors, 55-80 years	From 10/2003 to 10/2004; Primary care centers of 10 teaching hospitals in 8 Spanish cities	Ongoing, 6 years planned (1-year data used for lab. values, 2-year for other risk factors)	No	Virgin olive oil for cooking and dressings; ≥2 servings/day of vegetables; ≥3 servings/week of fish or seafood; ≥3 servings/week of nuts or seeds; white meat (poultry or rabbit) instead of red or processed meat; ≥7 glasses of wine/week; ≥2 dishes with salsa/week	American Heart Association Dietary Guidelines 200013	MD group: Quarterly dietary group sessions and free provision of virgin olive oil or mixed nuts LF group: None	Yes/No/Yes, 95%
Esposito, 200927	Primary prevention: Newly diagnosed type 2 diabetes, BMI >25 kg/m2, 30-75 years, <1 hour of physical activity per week	1/2004 to 9/2008; Research Center Diabetes Clinic of the Azienda Ospedalaria Universitaria, Naples, Italy	4(2-year data used for this meta-analysis)	Yes, both groups (mean caloric intake ≤1500 kcal/day for women, ≤1800 kcal/day for men)	<50% of calorie intake from carbohydrates, rich in vegetables and whole grains, low in red meat (replaced by poultry and fish), ≤30% from fat (main source 30-50 g olive oil)	<30% of total calorie intake from fat, <10% from saturated fat, rich in whole grains; American Heart Association Dietary Guidelines 200013	None	Yes/Yes (for laboratory values)/Yes, 91%
Singh, 200221	Primary and secondary prevention: ≥1 cardiovascular risk factor or type 2 diabetes or angina pectoris or previous MI	NA	2	No	<30% of total calorie intake from fat, <10% from saturated fat, <300 mg of cholesterol/day; ≥400-500 g of fruits, vegetables and nuts/d; 400-500 g of whole grains, legumes, rice, maize and wheat/d; 3-4 servings of mustard seed or soy bean oil /day	<30% of total calorie intake from fat, <10% from saturated fat, <300 mg of cholesterol/day; (NCEP step I diet)29	None	No/No/Yes, 98%

BMI=body mass index; LF=low-fat diet; MD=Mediterranean diet; MI=myocardial infarction; NA=not available; NCEP=National Cholesterol Education program.

Baseline characteristics of included individuals are summarized in Table 2. Mean age of enrolled patients ranged from 35 to 68 years. Mean body mass index of included subjects ranged from 29 to 35 kg/m². There was only one pure secondary prevention trial,²⁶ and only one more trial included individuals with established cardiovascular disease (40% of included individuals).²⁵ All but one trial directed their interventions to individuals consenting to actively adopt diet changes in free-living individuals.²⁵ In the Daily-Dose Consensus Interferon and Ribavirin: Efficacy of Combined Therapy (DIRECT) trial,²⁵ meals were provided during lunch in the self-service cafeteria of a workplace at a research centre in Israel. Two trials restricted calorie intake in both groups,²⁵, ²⁷ one trial restricted calories only in subjects randomized to a Mediterranean diet;¹⁷ in all other trials, calorie intake was not restricted in either of the 2 groups.

Table 2. Characteristics of Included Individuals at Baseline

Study (First Author, Year)	No. Randomized		Mean Age, Years		Male No (%)		Mean Body Weight, kg		Mean BMI, kg/m2		Hypertension No. (%)		Diabetes No. (%)		Dyslipidemia No. (%)		Current Smokers No. (%)		Coronary Artery Disease No. (%)
	MD	LF	MD	LF	MD	LF	MD	LF	MD	LF	MD	LF	MD	LF	MD	LF	MD	LF	MD	LF
Esposito, 200317	60	60	34	35	0	0	94	95	35	35	0	0	0	0	NA	NA	0	0	0	0
Esposito, 200424	90	90	44	44	49(54)	50(56)	78	77	28	28	NA	NA	NA	NA	NA	NA	0	0	0	0
Shai, 200825	109	104	53	51	89(82)	89(86)	91	91	31	31	37(34)	23(22)	15(14)	12(12)	29(27)	28(27)	16(15)	19(18)	46(42)	38(37)
Tuttle, 200826	51	50	58	58	41(80)	34(68)	90	91	29	30	22(43)	25(50)	10(20)	10(20)	NA	NA	13(25)	15(30)	51(100)	50(100)
Estruch, 2008(Predimed study)6	1223	598	68	68	563(46)	305(51)	75	76	29	30	978(80)	484(81)	599(49)	281(47)	807(66)	395(66)	171(14)	84(14)	0	0
Esposito, 200927	108	107	52(10)	52(11)	54(50)	52(49)	86(10)	86(10)	30(3)	30(4)	26(24)	25(23)	108(100)	107(100)	16(15)	17(16)	23(21)	22(21)	NA	NA
Singh, 200221⁎	499	501	49(10)	48(9)	454(91)	441(88)	66(8)	66(7)	24(3)	24(2)	195(39)	175(35)	95(19)	115(23)	359(72)	371(74)	254(51)	220(44)	289(58)	296(59)

LF=low-fat diet; MD=Mediterranean diet; NA=not available; SD=standard deviation.

Persistence on diet varied between 85% and 95% in subjects assigned to Mediterranean diets and from 78% to 93% in subjects assigned to low-fat diets. Baseline and mean changes in dietary intake between baseline and 2-year follow-up are summarized in Table 3. Baseline values and mean changes in cardiovascular risk factors from baseline to 2-year follow-up are presented in Appendix 2, online.

Table 3. Baseline and Mean Changes in Dietary Intake between Baseline and 2-Year Follow-up

Study (First Author, Year)	Baseline Energy Intake (kcal/day)		Δ-Energy (kcal/day)		Baseline Protein Intake (in % of Total Calories)		Δ-Protein (% of Total Calories)		Baseline Carbo-hydrate Intake (in % of Total Calories)		Δ-Carbo-hydrates (% of Total Calories)		Baseline Fat Intake (in % of Total Calories)		Δ-Fat (% of Total Calories)		Baseline Saturated Fat Intake (in % of Total Calories)		Δ-Saturated Fat (% of Total Calories)
	MD	LF	MD	LF	MD	LF	MD	LF	MD	LF	MD	LF	MD	LF	MD	LF	MD	LF	MD	LF
Esposito, 200317	2070	2100	−420	−110	14	17	3	0.5	58	55	−3	−1	28	28	0	1	12	8	−4	−0.5
Esposito, 200424	2235	2065	−170	−70	14	14	0.3	0.1	57	58	0.7	0.1	29	28	−1	0.4	13	8	−5	0.3
Shai, 200825	NA	NA	−372	−573	18	18	0.4	0.8	52	52	−1.3	−1.1	32	31	1.4	−1.4	10	10	−0.1	−0.1
Tuttle, 200826	1759	1811	62	3	18	18	0	−1	52	53	2	1	30	27	−0.5	2.4	9	8	−0.6	0.1
Estruch, 2008(Predimed study)6	2378	2264	−1	−164	17	17	−0.1⁎	0.2⁎	40	42	−1.1⁎	1⁎	40	39	1.8⁎	−1.1⁎	10	10	−0.6⁎	−0.5⁎
Esposito, 200927	2345	2304	−505	−470	17	16	1.4	1.7	52	52	−8.9	0.4	31	32	NA	NA	10	10	−0.3	−1
Singh, 200221†	2159	2170	−144	−81	15	15	−0.6	0.3	57	57	2.5	−0.9	28	28	−1.5	1.1	13	13	−4.8	−0.4

LF=low-fat diet; MD=Mediterranean diet; NA=not available.

Unbalanced Co-interventions among Included Trials 

In 2 trials,¹⁷, ²⁴ only participants randomized to the Mediterranean, but not the low-fat, group were offered specific individualized programs. In one of these trials,¹⁷ the level of physical activity increased more in the Mediterranean (from 64 to 175 minutes per week) than in the low-fat diet group (from 71 to 102 minutes per week) (P=.009). In the Predimed study,⁶ only participants randomized to the Mediterranean groups received individual motivational interviews and group educational sessions on a quarterly basis, and either 30 g per day of mixed nuts or 1 L of olive oil per week for free. In 3 trials,²⁵, ²⁶, ²⁷ there was no difference in trial design between participants assigned to Mediterranean or low-fat diets.

Body Weight, Body Mass Index, and Waist Circumference 

Body weight, body mass index, and waist circumference decreased more in subjects randomized to Mediterranean diets than in subjects randomized to low-fat diets. After 2 years, the weighted mean difference (WMD) in body weight between subjects randomized to Mediterranean and low-fat diets was −2.2 kg (95% confidence interval [CI], −3.9 to −0.6, P for heterogeneity <.001, I² = 97%), the WMD in body mass index −0.6 kg/m² (95% CI, −1 to −0.1, P for heterogeneity <.001, I²=94%), and the WMD in waist circumference −0.9 cm (95% CI, −2-0.2, P for heterogeneity <.001, I²=92%) (Figure 2).

• 
  • View Large Image
  • Download to PowerPoint

• Figure 2. 

  Mean changes in body weight, body mass index, waist circumference, systolic and diastolic blood pressure after 2 years of follow-up in randomized controlled trials comparing Mediterranean to low-fat diets. BMI=body mass index; CI=confidence interval.

Blood Pressure 

Both systolic and diastolic blood pressure values decreased more favorably in subjects randomized to Mediterranean diets than in subjects randomized to low-fat diets (Figure 2). The WMD for systolic blood pressure was −1.7 mm Hg (95% CI, −3.4 to −0.1, P for heterogeneity <.001, I²=89%), and for diastolic blood pressure −1.5 mm Hg (95% CI, −2.1 to −0.8, P for heterogeneity=.03, I²=60%) (Figure 2).

Lipid Values 

Total cholesterol and triglyceride values changed more favorably in subjects randomized to Mediterranean diets than in subjects randomized to low-fat diets (WMD for total cholesterol −7.4 mg/dL (95% CI, −10.3 to −4.4, P for heterogeneity=.002, I²=73%) (Figure 3). There were no statistically significant differences in low-density lipoprotein cholesterol (WMD −3.3 m g/dL; 95% CI, −7.3-0.6; P for heterogeneity=.3, I²=23%) or HDL cholesterol (WMD 0.9 mg/dL; 95% CI, −1.9-3.8, P for heterogeneity<0.001, I²=99%).

• 
  • View Large Image
  • Download to PowerPoint

• Figure 3. 

  Mean changes in cholesterol values, high-sensitivity C-reactive protein, fasting plasma glucose, and serum insulin after 2 years of follow-up in randomized controlled trials comparing Mediterranean to low-fat diets. *1-year follow-up data. CI=confidence interval; HDL=high-density lipoprotein; LDL=low-density lipoprotein.

High-sensitivity C-reactive Protein 

High-sensitivity C-reactive protein (Hs-CRP) decreased more favorably in subjects randomized to Mediterranean diets than in subjects randomized to low-fat diets. The WMD for hs-CRP was −1.0 mg/L (95% CI, −1.5 to −0.5, P for heterogeneity <.001, I²=82%) (Figure 3).

Fasting Plasma Glucose and Serum Insulin 

Plasma glucose decreased more favorably in subjects randomized to Mediterranean diets than in subjects randomized to low-fat diets (WMD −3.8 mg/dL, 95% CI, −7.0 to −0.6, P for heterogeneity=.18, I²=97%) (Figure 3). There was no significant difference in serum insulin between the 2 groups (WMD −1.1 μU/mL, 95% CI, −2.9 to 0.8, P for heterogeneity <.001, I²=98%).

Clinical Outcomes 

Only one trial reported clinical outcomes.²⁶ There were 3 nonfatal myocardial infarctions and 1 stroke in the low-fat diet group, and 1 nonfatal myocardial infarction and 3 strokes in the Mediterranean diet group. No deaths were reported.

Sensitivity Analyses 

There were no qualitative changes in the point estimates of all analyses when we restricted the analyses to the trials with reported concealed treatment allocation or blinded outcome assessment, or when we included the results of the trial by Singh et al²¹ in the meta-analysis.

Point estimates for most outcomes consistently favored subjects randomized to Mediterranean diets both in primary and secondary prevention trials. Point estimates for differences in HDL cholesterol favored Mediterranean diets only in primary, but not in secondary prevention subjects, but neither of these changes reached statistical significance.

When we compared trials with balanced²⁵, ²⁶, ²⁷ versus unbalanced⁶, ¹⁷, ²⁴ co-interventions, and trials with restriction²⁵, ²⁷ versus no restriction of daily calorie intake in both groups,⁶, ¹⁷, ²⁶ we found no qualitative differences in the point estimates of the mean changes of any of the cardiovascular risk factors. There was no longer evidence of heterogeneity when restricting analyses to trials with balanced co-interventions or to trials with restriction of daily calorie intake in both groups, except for body mass index, waist circumference, and HDL cholesterol.

Back to Article Outline

Discussion 

In this meta-analysis of all available randomized controlled trials comparing Mediterranean with low-fat diets in overweight/obese individuals, most cardiovascular risk factors and vascular inflammatory markers improved more favorably in individuals allocated to a Mediterranean diet. The observed differences for the individual risk factors were modest, but the direction of the changes consistently favored Mediterranean over low-fat diets across outcomes. The observed heterogeneity across individual trials could be eliminated by restricting analyses to trials with balanced trial design or trials with restriction of daily calorie intake in both diet groups. Evidence on clinical outcomes remains inconclusive because only one trial reported clinical events.

The present study has strengths and limitations. We carried out a comprehensive literature search for randomized controlled trials reporting intention-to-treat data in individuals at increased cardiovascular risk, allowing us to assess the impact of the 2 diets on a broad spectrum of cardiovascular risk factors. Although formal testing did not indicate any publication bias, such bias cannot be definitely ruled out due to the relatively small number of trials included and the low power of any test to detect publication bias. Although only 2 of the included trials reported blinded outcome assessment for all outcomes, the quality of the included trials was reasonably good, with all but one trial each reporting concealed treatment allocation and almost complete follow-up of >90%. In addition, the results of our analyses proved to be robust across various sensitivity analyses accounting for differences in trial quality, population studied, and co-interventions.

Our analysis has some limitations. It is based on only 6 trials, with 3 trials published by the same group of authors.¹⁷, ²⁴, ²⁷ We observed significant heterogeneity for most outcomes analyzed. However, when we restricted analyses to trials with balanced co-interventions, most cardiovascular risk factors were modified more favorably in individuals on Mediterranean diets and there was no longer evidence of heterogeneity for cardiovascular risk factors, with the exception of body mass index, waist circumference, and HDL cholesterol. The same was true when restricting analyses to trials with restricted daily calorie intake in both groups.

Only one of the identified trials was a pure secondary prevention trial,²⁶ so our results may be limited to the modification of cardiovascular risk factors in primary prevention. Because sensitivity analysis did not reveal any major differences in changes of cardiovascular risk factors between primary and secondary prevention trials, there is a suggestion that Mediterranean diets are superior to low-fat diets not only in primary, but also in secondary prevention.

None of the included trials addressed participants' quality of life while adhering to the prescribed diet. Thus, we lack information about potential differences in quality of life among participants randomized to Mediterranean or low-fat diets. However, rates of persistence on Mediterranean and low-fat diets were similar, making large differences in quality of life unlikely.

The methodology of our meta-analysis did not allow us to identify any individual component of Mediterranean diets that may be particularly beneficial in modifying cardiovascular risk factors. The results of our meta-analysis imply that heterogeneous patterns of Mediterranean diets are effective in lowering cardiovascular risk focusing on a specific type of a diet as a whole rather than on individual diet components.

All but one trial included in our meta-analysis were conducted in Mediterranean countries.²⁶ This may add to the strengths of our results because individuals randomized to low-fat diets may have followed a Mediterranean-style diet to some extent. On the other hand, it raises the question about the generalizability of our results to non-Mediterranean countries.

None of the included trials was powered to detect any differences in clinical outcomes between the 2 diets. However, the findings of our meta-analysis are supported by many prospective cohort studies demonstrating a beneficial effect of Mediterranean diets on cardiovascular outcomes.³⁰ Evidence from cohort studies studying self-elected eating patterns may, however, be biased by confounding.³¹ In order to rule out potential confounding, there is a need for unbiased evidence from randomized controlled trials demonstrating the benefit of a particular diet on patient-important outcomes. So far, only 2 secondary and no primary prevention trials comparing the effects of Mediterranean with low-fat diets on clinical outcomes have been published. Unfortunately, serious concern has been raised about the integrity of the principal investigator of one of these trials.²² In the other trial, the Lyon Diet Heart Study from 1994,³² the combined primary endpoint of cardiac death and nonfatal myocardial infarction was reduced by an impressive 73% (95% CI, 41%-88%) after a mean follow-up of 27 months. The Lyon Diet Heart Study was stopped early for apparent benefit after only 41 primary outcome events. Early stopping for benefit may result in an overestimate of the net health benefit.¹⁵ In addition, hardly any of the patients in the Lyon Diet Heart Study were on statin therapy at the time. It thus remains unclear whether the benefit of Mediterranean diets persists with statin therapy in patients at high risk of cardiovascular events.

Given the limitations of our meta-analysis and of the 2 secondary prevention trials, more evidence is needed before calling for the implementation of Mediterranean diet in individuals at increased vascular risk from a public health perspective. The observed beneficial results of Mediterranean diets on cardiovascular risk factors and cardiovascular morbidity and mortality should be reproduced in at least one adequately powered cardiovascular disease prevention trial.

In summary, our meta-analysis suggests a favorable effect of Mediterranean, as compared with low-fat, diets on most cardiovascular risk factors and inflammatory markers. Although the observed effects on individual risk factors were modest, the consistent benefit over a broad range of cardiovascular risk factors may eventually lead to a reduction in cardiovascular outcomes.

Back to Article Outline

Appendix 1 

Characteristics of Included Trials 

Characteristics of included trials are summarized in Table 1. One trial included patients with established cardiovascular disease only.²⁶ One trial enrolled subjects who were either obese, or had type 2 diabetes or established cardiovascular disease.²⁵ Four trials were primary prevention trials including either obese, sedentary, postmenopausal women,¹⁷ sedentary subjects with the metabolic syndrome,²⁴ overweight subjects with newly diagnosed type 2 diabetes²⁷ or individuals at high cardiovascular risk (either type 2 diabetes or 3 or more cardiovascular risk factors).⁶ In the latter trial, subjects randomized to a Mediterranean diet were randomized to provision of either a free liter of virgin olive oil per week or provision of free nuts (30g/day). For the purpose of this analysis, we collapsed the 2 groups into one without differentiating between subjects assigned free virgin olive oil or free nuts.

Follow-up of included trials was 2 years in 4 trials,¹⁷, ²⁴, ²⁵, ²⁶ and 4 years in one trial.²⁷ The Predimed trial⁶ is still ongoing and is planned to have a mean follow-up of 6 years; we included 2-year follow-up data of this trial in our meta-analysis, except for laboratory analyses, which were measured after only 1 year of follow-up and pooled with the 2-year lipid values of the other trials.

Quality of the Trials 

Assignment of study participants was concealed in 5 trials,⁶, ¹⁷, ²⁵, ²⁶, ²⁷ and possibly concealed in one trial.²⁴ All trials used an open design. Blinded outcome assessment for all outcomes was reported in 2 trials,²⁵, ²⁶ and only for laboratory analyses in 2 trials.¹⁷, ²⁷ There was no blinded outcome assessment in one trial,²⁴ and one trial⁶ did not mention whether blinded outcome assessment was performed for any of the outcomes. Full description of losses to follow-up and withdrawals was reported in one trial,²⁵ partially reported in 3 trials,¹⁷, ²⁶, ²⁷ and not reported in 2 trials.⁶, ²⁴ Four of the 6 included trials⁶, ¹⁷, ²⁴, ²⁷ had a loss to follow-up <10%. No trial was stopped early for benefit. The 2 reviewers were in full agreement when rating the methodological quality of included trials. In 3 trials⁶, ¹⁷, ²⁴ the method used to account for missing data remained unclear, 2 trials²⁵, ²⁷ used the last value carried forward method, and one trial a multilevel random effects model.²⁶

Back to Article Outline

Appendix 2 

Baseline and Mean Changes (Standard Deviations) in Outcomes between Baseline and 2-year Follow-up

Study	Mediterranean Diet		Low-fat Diet
	Baseline (SD)	Mean change (SD)	Baseline (SD)	Mean change (SD)
Weight (kg)
Esposito, 200317	95(9.4)	−14(8.5)	94(9.2)	−3(8.5)
Esposito, 200424	78(8.0)	−4.0(1.1)	77(8.0)	−1.2(0.6)
Shai, 200825	91(14)	−4.4(6.0)	91(12)	−2.9(4.2)
Tuttle, 200826	90(17)	−0.75(6.5)	91(18)	−0.99(6.0)
Predimed, 20086	75(11)	−0.07(4.2)	76(11)	−0.2(4.4)
Esposito, 200927	86(10)	−4.9(2.5)	86(10)	−3.7(2.1)
Singh, 200221⁎	66(7.5)	−3.5(5.2)	66(7.3)	−0.9(2.7)
Body Mass Index (kg/m2)
Esposito, 200317	35(2.3)	−5.2(8.1)	34.7(2.4)	−1(3.1)
Esposito, 200424	28(3.4)	−1.2(0.3)	28(3.2)	−0.4(0.4)
Shai, 200825	31(4.1)	−1.5(2.2)	31(3.2)	−1(1.4)
Tuttle, 200826	29(5.0)	−0.2(2.1)	30(5.0)	−0.3(1.9)
Predimed, 20086	29(3.3)	−0.03(1.6)	30(3.6)	−0.08(1.8)
Esposito, 200927	30(3.4)	−1.9(0.9)	30(3.6)	−1.1(0.6)
Singh, 200221⁎	24(3.0)	−1.3(2)	24(2.3)	−0.3(1)
Waist circumference (cm)
Esposito, 200317	NA	NA	NA	NA
Esposito, 200424	92(9.0)	−2.0(0.5)	93(10)	0(0.01)
Shai, 200825	106(9.1)	−3.5(5.1)	105(9.2)	−2.8(4.3)
Tuttle, 200826	101(15)	−0.18(6.4)	103(14)	−0.72(7.1)
Predimed, 20086	100(9.5)	−0.83(6.4)	100(10)	−0.85(6.7)
Esposito, 200927	98(10)	−4.4(2.8)	98(10)	−3.3(2.8)
Singh, 200221⁎	NA	NA	NA	NA
Systolic blood pressure (mm Hg)
Esposito, 200317	124(8.5)	−3.0(8.7)	123(7.9)	−1.0(5.3)
Esposito, 200424	134(9.0)	−4.0(2.0)	136(10)	−1.0(1.0)
Shai, 200825	133(14)	−5.5(14.3)	130(13)	−4.3(11.8)
Tuttle, 200826	120(18)	1.6(14.7)	119(15)	4.6(16.1)
Predimed, 20086	152(18)	−3.4(15.9)	151(19)	−1.6(16.4)
Esposito, 200927	139(12)	−4.5(3.7)	140(12)	−4.5(1.7)
Singh, 200221⁎	132(17)	−5.3(10.4)	131(17)	−2.2(5.2)
Diastolic blood pressure (mm Hg)
Esposito, 200317	85(4.7)	−3(8.7)	85(4.9)	−1.3(9)
Esposito, 200424	85(6.0)	−3(1)	86(7.0)	−1(1)
Shai, 200825	81(9.2)	−2.2(9.5)	79(9.1)	−0.9(8.1)
Tuttle, 200826	73(11)	−0.74(11.3)	71(8.0)	1.4(9.1)
Predimed, 20086	85(9.6)	−3.4(8.0)	84(9.8)	−1.9(8.8)
Esposito, 200927	87(8.0)	−3.2(2.8)	86(8.0)	−2.5(2.3)
Singh, 200221⁎	86(10)	−2.7(6.8)	86(9.0)	−1.1(3.5)
Total cholesterol (mg/dL)
Esposito, 200317	197(23)	−4.0(15)	193(23)	0(14)
Esposito, 200424	199(34)	−11(6)	193(32)	−2.0(2.0)
Shai, 200825	213(NA)	−7.9(27)	201(NA)	2.3(31)
Tuttle, 200826	161(36)	−3.7(35)	161(38)	9.7(42)
Predimed, 20086	213(39)	−3.6(34)	212(38)	−1.1(32)
Esposito, 200927	220(35)	−18(12)	217(35)	−7.7(7.7)
Singh, 200221⁎	222(38)	−26(32.8)	223(38)	−7.4(9.3)
Low-density lipoprotein cholesterol (mg/dL)
Esposito, 200317	NA	NA	NA	NA
Esposito, 200424	NA	NA	NA	NA
Shai, 200825	123(34)	−5.6(27)	117(36)	−0.05(31)
Tuttle, 200826	93(28)	−0.45(30)	93(32)	10(32)
Predimed, 20086	131(32)	-3(28)	130(32)	−1.3(29)
Esposito, 200927	NA	NA	NA	NA
Singh, 200221⁎	141(30)	−−24.2(18)	137(26)	−6.1(9.4)
High-density lipoprotein cholesterol (mg/dL)
Esposito, 200317	46(10)	8(27.9)	46(10)	0(26)
Esposito, 200424	41(9.0)	4(2.0)	42(9.0)	1.0(1.0)
Shai, 200825	39(9.4)	7.26(0.79)	39(9.6)	9.14(0.97)
Tuttle, 200826	38(7.0)	0.82(8.2)	36(9.0)	5.6(9.8)
Predimed, 20086	56(13)	0.44(8.9)	55(12)	0.1(8.6)
Esposito, 200927	43(7.7)	4.6(4.6)	43(7.7)	0(0.8)
Singh, 200221⁎	45(10)	1.4(7.0)	44(5.8)	−1.5(1.5)
Triglycerides (mg/dL)
Esposito, 200317	142(44)	−19(70)	142(53)	−8.0(59)
Esposito, 200424	168(57)	−18(8.0)	172(54)	1.0(3.0)
Shai, 200825	174(68)	−21.8(62)	157(62)	−2.7(91)
Tuttle, 200826	143(71)	−19(48)	183(196)	−50(189)
Predimed, 20086	133(66)	−2.0(59)	143(81)	−2.0(62)
Esposito, 200927	168(71)	−42(46)	168(71)	−25(37)
Singh, 200221⁎	163(34)	−44.4(49)	164(25)	−9.7(13)
Highly sensitive C-reactive protein (mg/dL)
Esposito, 200317	3.2(1.5-8.4)†	−1.1(3.2)	3.4(1.4-8.3)†	−0.3(1.8)
Esposito, 200424	2.8(0.7-5.4)†	−1.1(0.4)	2.9(0.5-5.7)†	−0.1(0.3)
Shai, 200825	4.6(3.4)	−0.9(2.4)	3.6(2.9)	−0.5(2.3)
Tuttle, 200826	3.8(3.9)	−1.5(3.8)	4.4(4.6)	−1.4(3.5)
Predimed, 20086	5.2(5.1)	−0.1(3.6)	5.1(3.6)	1.7(3.9)
Esposito, 200927	NA	NA	NA	NA
Singh, 200221⁎	NA	NA	NA	NA
Plasma glucose (mg/dL)
Esposito, 200317	106(14)	−9.0(26)	105(13)	−2.0(11)
Esposito, 200424	113(10)	−8.0(3.0)	114(10)	−2.0(1.5)
Shai, 200825	94(38)	−2.0(26)	87(26)	4.2(15)
Tuttle, 200826	92(43)	−2.7(25)	94(34)	−0.7(18)
Predimed, 20086	119(41)	−3.3(24)	120(38)	−1.2(32)
Esposito, 200927	162(34)	−2.1(1.6)	159(33)	−1.1(1.1)
Singh, 200221⁎	108(25)	−7.7(13)	107(28)	−3.8(10)
Serum insulin (μU/mL)
Esposito, 200317	14(4.0)	−5.0(16)	14(4.0)	−2.0(6.5)
Esposito, 200424	15(6.0)	−4.0(1.9)	16(7.0)	−0.5(1.0)
Shai, 200825	15(8.0)	−0.18(0.5)	13(6.8)	−0.15(0.5)
Tuttle, 200826	10(5.0)	2.2(9.6)	13(8.0)	−0.8(23)
Predimed, 20086	NA	NA	NA	NA
Esposito, 200927	18(7.2)	−1.8(1.8)	19(8.3)	−2.0(1.8)
Singh, 200221⁎	NA	NA	NA	NA

SD=standard deviation. SI conversion factors. To convert plasma glucose values to millimoles per liter, multiply by 0.0555. To convert cholesterol values to millimoles per liter, multiply by 0.0259. To convert triglyceride values to millimoles per liter, multiply by 0.0113. To convert serum insulin values to picomoles per liter, multiply by 6.

Back to Article Outline

References 

1. Mokdad AH, Marks JS, Stroup DF, Gerberding JL. Actual causes of death in the United States, 2000. JAMA. 2004;291(10):1238–1245
   • View In Article
   • CrossRef
2. Avenell A, Broom J, Brown TJ, et al. Systematic review of the long-term effects and economic consequences of treatments for obesity and implications for health improvement. Health Technol Assess. 2004;8(21):iii–iv1-182
   • View In Article
   • MEDLINE
3. Howard BV, Van HL, Hsia J, et al. Low-fat dietary pattern and risk of cardiovascular disease: the Women's Health Initiative Randomized Controlled Dietary Modification Trial. JAMA. 2006;295(6):655–666
   • View In Article
   • CrossRef
4. Willett WC, Sacks F, Trichopoulou A, et al. Mediterranean diet pyramid: a cultural model for healthy eating. Am J Clin Nutr. 1995;61(6 Suppl):1402S–1406S
   • View In Article
   • MEDLINE
5. Serra-Majem L, Roman B, Estruch R. Scientific evidence of interventions using the Mediterranean diet: a systematic review. Nutr Rev. 2006;64(2 Pt 2):S27–S47
   • View In Article
   • MEDLINE
   • CrossRef
6. Estruch R, Martinez-Gonzalez MA, Corella D, et al. Effects of a Mediterranean-style diet on cardiovascular risk factors: a randomized trial. Ann Intern Med. 2006;145(1):1–11
   • View In Article
7. Malik VS, Hu FB. Popular weight-loss diets: from evidence to practice. Nat Clin Pract Cardiovasc Med. 2007;4(1):34–41
   • View In Article
   • CrossRef
8. Fung TT, Rexrode KM, Mantzoros CS, et al. Mediterranean diet and incidence of and mortality from coronary heart disease and stroke in women. Circulation. 2009;119(8):1093–1100
   • View In Article
   • CrossRef
9. Knoops KT, de Groot LC, Kromhout D, et al. Mediterranean diet, lifestyle factors, and 10-year mortality in elderly European men and women: the HALE project. JAMA. 2004;292(12):1433–1439
   • View In Article
   • CrossRef
10. Martinez-Gonzalez MA, de la Fuente-Arrillaga C, Nunez-Cordoba JM, et al. Adherence to Mediterranean diet and risk of developing diabetes: prospective cohort study. BMJ. 2008;336(7657):1348–1351
    • View In Article
    • CrossRef
11. Scarmeas N, Luchsinger JA, Schupf N, et al. Physical activity, diet, and risk of Alzheimer disease. JAMA. 2009;302(6):627–637
    • View In Article
    • CrossRef
12. Willett WC, Skerrett PJ. Eat, Drink, and Be Healthy: the Harvard Medical School Guide to Healthy Eating. New York: Simon & Schuster; 2001;
    • View In Article
13. Krauss RM, Eckel RH, Howard B, et al. AHA Dietary Guidelines: revision 2000: a statement for healthcare professionals from the Nutrition Committee of the American Heart Association. Circulation. 2000;102(18):2284–2299
    • View In Article
14. Juni P, Witschi A, Bloch R, et al. The hazards of scoring the quality of clinical trials for meta-analysis. JAMA. 1999;282(11):1054–1060
    • View In Article
    • MEDLINE
    • CrossRef
15. Bassler D, Briel M, Montori VM, et al. Stopping randomized trials early for benefit and estimation of treatment effects: systematic review and meta-regression analysis. JAMA. 2010;303(12):1180–1187
    • View In Article
    • CrossRef
16. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials. 1986;7(3):177–188
    • View In Article
    • MEDLINE
    • CrossRef
17. Esposito K, Pontillo A, Di Palo C, et al. Effect of weight loss and lifestyle changes on vascular inflammatory markers in obese women: a randomized trial. JAMA. 2003;289(14):1799–1804
    • View In Article
    • MEDLINE
    • CrossRef
18. Sterne JA, Egger M, Smith GD. Systematic reviews in health care: Investigating and dealing with publication and other biases in meta-analysis. BMJ. 2001;323(7304):101–105
    • View In Article
    • MEDLINE
    • CrossRef
19. Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med. 2002;21(11):1539–1558
    • View In Article
    • MEDLINE
    • CrossRef
20. Higgins JP, Thompson SG, Deeks JJ, et al. Measuring inconsistency in meta-analyses. BMJ. 2003;327(7414):557–560
    • View In Article
    • CrossRef
21. Singh RB, Dubnov G, Niaz MA, et al. Effect of an Indo-Mediterranean diet on progression of coronary artery disease in high risk patients (Indo-Mediterranean Diet Heart Study): a randomised single-blind trial. Lancet. 2002;360(9344):1455–1461
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (107 KB)
    • CrossRef
22. Horton R. Expression of concern: Indo-Mediterranean Diet Heart Study. Lancet. 2005;366(9483):354–356
    • View In Article
    • Full Text
    • Full-Text PDF (45 KB)
    • CrossRef
23. De Lorgeril M, Renaud S, Mamelle N, et al. Mediterranean alpha-linolenic acid-rich diet in secondary prevention of coronary heart disease. Lancet. 1994;343:1454–1459
    • View In Article
    • Abstract
    • CrossRef
24. Esposito K, Marfella R, Ciotola M, et al. Effect of a mediterranean-style diet on endothelial dysfunction and markers of vascular inflammation in the metabolic syndrome: a randomized trial. JAMA. 2004;292(12):1440–1446
    • View In Article
    • CrossRef
25. Shai I, Schwarzfuchs D, Henkin Y, et al. Weight loss with a low-carbohydrate, Mediterranean, or low-fat diet. N Engl J Med. 2008;359(3):229–241
    • View In Article
    • CrossRef
26. Tuttle KR, Shuler LA, Packard DP, et al. Comparison of low-fat versus Mediterranean-style dietary intervention after first myocardial infarction (from The Heart Institute of Spokane Diet Intervention and Evaluation Trial). Am J Cardiol. 2008;101(11):1523–1530
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (344 KB)
    • CrossRef
27. Esposito K, Maiorino MI, Ciotola M, et al. Effects of a Mediterranean-style diet on the need for antihyperglycemic drug therapy in patients with newly diagnosed type 2 diabetes: a randomized trial. Ann Intern Med. 2009;151(5):306–314
    • View In Article
28. Robertson RM, Smaha L. Can a Mediterranean-style diet reduce heart disease?. Circulation. 2001;103(13):1821–1822
    • View In Article
29. National Cholesterol Education Program. Second Report of the Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel II). Circulation. 1994;89(3):1333–1445
    • View In Article
    • MEDLINE
30. Sofi F, Cesari F, Abbate R, et al. Adherence to Mediterranean diet and health status: meta-analysis. BMJ. 2008;337:a1344
    • View In Article
    • CrossRef
31. Lawlor DA, Davey SG, Kundu D, et al. Those confounded vitamins: what can we learn from the differences between observational versus randomised trial evidence?. Lancet. 2004;363(9422):1724–1727
    • View In Article
    • Full Text
    • Full-Text PDF (81 KB)
    • CrossRef
32. De Lorgeril M, Salen P, Martin JL, et al. Mediterranean diet, traditional risk factors, and the rate of cardiovascular complications after myocardial infarction: final report of the Lyon Diet Heart Study. Circulation. 1999;99:779–785
    • View In Article
    • MEDLINE