Alcohol consumption and the prevalence of metabolic syndrome: A meta-analysis of observational studies

Atherosclerosis 204 (2009) 624–635

Contents lists available at ScienceDirect

Atherosclerosis journal homepage: www.elsevier.com/locate/atherosclerosis

Alcohol consumption and the prevalence of metabolic syndrome: A meta-analysis of observational studies Alkerwi Ala’a a,∗ , Boutsen Michel b , Vaillant Michel a , Barre Jessica a , Lair Marie-Lise a , Albert Adelin c , Guillaume Michèle c , Dramaix Michèle b a b c

Centre de Recherche Public Santé, Centre d’Etudes en Santé, Grand-Duchy of Luxembourg, 1A rue Thomas Edison, L-1445 Strassen, Luxembourg Département de Biostatistique, Ecole de Santé Publique, Université Libre de Bruxelles, Belgium Ecole de Santé Publique, Université de Liège, Belgium

a r t i c l e

i n f o

Article history: Received 30 May 2008 Received in revised form 16 September 2008 Accepted 29 October 2008 Available online 11 November 2008 Keywords: Metabolic syndrome Alcohol consumption Meta-analysis Observational studies Random effects Gender-specific analysis

a b s t r a c t Background: In the past two decades, the metabolic syndrome has given rise to much clinical and research interest. The broad overlap of alcohol consumption with different components of metabolic syndrome makes alcohol–metabolic syndrome relationship a controversial topic. Objectives: To support the evidence available about the relationship between alcohol consumption and metabolic syndrome as a comprehensive clinical entity, as well as to identify the gender-specific dose–response, by performing a meta-analysis based on information from published data. Methods: Manual and computer searches in different bibliographic databases were performed to identify the relevant scientific publications, on the relation between alcohol consumption and metabolic syndrome. Alcohol intake was converted into a same unit (g/day) and then categorized using standard classification in order to provide relevant comparisons. Fixed and random effects models were used to aggregate individual odds ratios and to derive pooled estimates and 95% confidence intervals. Results: Fourteen relevant publications were identified on the relation between alcohol consumption and the prevalence of metabolic syndrome. 7 studies were included in the meta-analysis. The results showed that alcohol consumption of less than 40 g/day in men and 20 g/day in women significantly reduced the prevalence of metabolic syndrome. Conclusion: “Responsible alcohol intake” appears to be associated with a reduced prevalence of metabolic syndrome. Favorable metabolic effect seemed to be restricted to alcohol consumption of less than 20 g/day among women, and of less than 40 g/day among men. These findings support the actual recommendations regarding alcohol consumption among apparently healthy people. © 2008 Elsevier Ireland Ltd. All rights reserved.

1. Introduction Metabolic syndrome, the combination of major cardiovascular risk factors-obesity, hypertension, dyslipidemia and hyperglycaemia, is receiving increased attention from physicians and public health decision-makers, due to its association with adverse cardiovascular outcomes. This clustering has been recognised as a strong indicator of increased risk of cardiovascular morbidity and mortality [1–12]. The aetiology of the metabolic syndrome is complex, determined by the interplay of both genetic and environmental factors [13–15]. For preventive purposes, knowledge about widespread and potentially modifiable behaviour is important.

∗ Corresponding author. Tel.: +352 26 970 743; fax: +352 26 970 719. E-mail address: [email protected] (A. Alkerwi). 0021-9150/$ – see front matter © 2008 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.atherosclerosis.2008.10.036

Alcohol consumption, a lifestyle factor and one of the most prevalent habits worldwide, has been suggested to be relevant with respect to the risk of type 2 diabetes, hypertension and obesity, the key parts of the “metabolic syndrome”. The relation of alcohol consumption to each component of (MS) is multifaceted: although alcohol intake has been associated with increased risk of hypertension [16,17], regular moderate alcohol consumption is associated with a lower risk of diabetes mellitus [18], perhaps through improved insulin sensitivity [19–21]. The relationship between alcohol consumption and insulin resistance shows a U-shaped curve: insulin resistance is minimal in individuals with regular light-to-moderate alcohol consumption and increased in both heavy drinkers and abstaining subjects [22]. In addition, alcohol consumption is associated with higher HDLcholesterol [23,24] and high triglycerides [25]. It has been found to be both inversely and directly related to increased risk of obesity [26–28].

A. Alkerwi et al. / Atherosclerosis 204 (2009) 624–635

Literature review provides that moderate alcohol intake lowers risk of coronary heart disease [29,30], stroke [31,32], atherosclerosis [33] and death [34]. While the association between light–moderate drinking and reduced risk of cardiovascular disease mortality is well established, epidemiological evidence about the alcohol–(MS) relation is still sparse and controversial. Over the past decade, many observational epidemiologic studies have examined the role of alcohol as both a risk factor and a potential protective factor against the development of (MS). In some studies, moderate alcohol consumption has been linked to lower prevalence of (MS) [35], whereas heavy and early lifetime drinking increases the risk of developing (MS) [36]. Other cross-sectional studies examining the relation of lifestyle to (MS), have shown no preventive effect of alcohol on the occurrence of the (MS) [20,37,38]. The objective of the present study was to perform a systematic review of the published observational studies by using quantitative meta-analytic methods [39,40]. By this approach, we wish to obtain insight regarding the relationship between alcohol consumption and the prevalence of (MS), in term of gender, which may be overlooked by individual studies because of relatively small sample size and insufficient statistical power [41,42]. In the present study, dose–response alcohol–metabolic syndrome relationship has been investigated. The (MS) serves as a useful clinical tool to raise awareness among health professionals and help in identifying high-risk individuals. Due to its increasing clinical and public health importance, research examining risks or benefits of alcohol consumption in patients with (MS) is needed to assist clinicians in advising their patients and politicians developing appropriate strategies of prevention. 2. Materials and methods 2.1. Search strategy A comprehensive literature search of the MEDLINE and EMBASE computerized databases was performed (for studies from January 1998 through May 2007) using the Medical Subject Headings (MeSH), according to the strategy illustrated in Table 1. A master’s level medical librarian (CD) collaborated to design the literature search strategies. In Medline database, the MeSH term “metabolic syndrome” was recognised only after 2001. The MeSH term “Insulin resistance syndrome” was used for the period 1998–2000. A supplementary search of the web-site of the Alcohol Beverage Medical Research Foundation (ABMRF) that tracks alcohol-related research has been done. A manual search was also conducted using references cited in published original and relevant review articles. In addition, the “related links” of each relevant article, identified by computerized search process, were reviewed to identify any studies that had not

Table 1 Strategy for MEDLINE. 1998–2001

2002–2007

Insulin resistance syndrome AND Cross-sectional studies AND Alcohols OR Ethanol OR Drinking behaviour OR Alcohol drinking OR Alcoholic beverages OR Life style OR Health behaviour

Metabolic syndrome AND Cross-sectional studies AND Alcohols OR Ethanol OR Drinking behaviour OR Alcohol drinking OR Alcoholic beverages OR Life style OR Health behaviour

625

been found in the databases search. These steps allowed identifying longitudinal studies that were also included for their baseline or incidence data. Attempts have been made to contact the authors to obtain certain non-published information. 2.2. Study selection Consistent with standard meta-analysis techniques, four selection criteria were defined in relation to date, language, setting, and population; we included studies published since 1998 (year corresponding to the first official definition of the (MS) by the WHO [43]). We selected studies published in English, French, German or Spanish, to avoid possible bias by the use of stringent linguistic criteria. Selection was limited to community-based studies, excluding those relied on hospitalized patients. We restricted the review to studies of healthy adults; research devoted to addicts, alcoholics’ people was excluded, particularly to avoid the confounding bias of alcoholism-associated complication. Thorough attempts were made to include only-abstracts articles. Epidemiological studies that addressed the relationship between alcohol consumption and the prevalence of (MS) were retrieved and reviewed for predefined eligibility criteria. To be included in the meta-analysis, a study had to meet the following criteria: 1. observational designed (cross-sectional or baseline longitudinal studies); 2. healthy population community-based; 3. a clear definition of the endpoint which corresponds to the prevalence of (MS), calculated according to one of the definitions (OMS [43], NCEP ATP III [44], IDF [45]); 4. risk estimate of occurrence of (MS) should be reported; 5. self-reported alcohol consumption by using questionnaire; 6. quantified self-alcohol consumption. All potentially relevant titles and abstracts, with regard to the study objective, were selected for full-text examination. The eligibility of all full-text articles were investigated by using piloted, standardised forms. The references in relevant publications were checked for additional studies of interest. Original peer-reviewed publications on observational cross-sectional, baseline cohort studies on the relationship between alcohol consumption and the prevalence of (MS) were included. Randomized controlled trials, editorials, letters to the editors, and review articles were not included. The search of conference proceedings and query of experts did not identify additional articles. 2.3. Data extraction As described by Stroup et al. [40], two investigators independently used a standardised data-collection form to assess the eligibility for inclusion. Information has been tabulated on: article’s first author’s name, year, study characteristics and design, definition used to identify the (MS) and its frequency, type of alcohol consumption, crude or adjusted risk estimate (relative risk (RR), or odds ratio) for (MS), and potential confounder factors used in the analyses. Studies computing odds ratios comparing the different types of alcohol consumption were retained for analysis. For studies that reported results from various covariates analyses, the estimates based on the model that included the most potential confounders were abstracted. When there were discrepancies between investigators for inclusion or exclusion criteria, other investigators conducted additional evaluation of the study, then disagreements were resolved by discussion in meeting.

626

A. Alkerwi et al. / Atherosclerosis 204 (2009) 624–635

2.4. Alcohol data conversion Measurement of alcohol consumption varied among studies. It was based on the frequency of alcohol consumption per day [46], or on the amount of alcohol in grams per week [47] or per month [48]. Several studies reported alcohol data by using a standard unit of alcohol measurement, called “drinks”, which even differed from country to other. It is calculated either as drinks per day [49–52], or drinks per month [53]. The definition of standard drink was notified whenever it was mentioned in the publication (Table 2). To facilitate direct comparison between studies, alcohol consumption data were first converted into a same unit (g/day) [31]. If a paper did not give data about the quantity of alcohol equivalent to a standard drink, the average alcohol consumption was assumed to be 12 g in studies carried out in US, 10 g in studies carried out in U.K. and Europe and 21.2 g in Japan [18,30]. This approach helped to identify the quantity of alcohol consumed by each category of drinkers and to make the comparison between studies possible. In case that alcohol data was reported as drinks/month [53], we first converted drinks into grams, then divided it by 30 in order to get the quantity consumed in grams per day, or by 7 to get the daily dose which was given in grams per week (Table 2). 2.5. Statistical methods 2.5.1. Effect measures The definition of each type of alcohol consumption (light, moderate and heavy) differed substantially among studies and was reported as categorical data (classes corresponding to ranges of alcohol consumption). Therefore, to get meaningful results in terms of risk assessment, it was important to harmonise the categories of alcohol consumption, before starting the statistical analysis. Categorical levels for alcohol quantities consumed per day were selected to be consistent with the report of WHO which suggests different categories of average volume of alcohol consumption by age and sex, constructed in a way that the risk of many chronic diseases were about the same for both men and women in the same drinking category [54]. This classification of alcohol drinking linked to health outcomes, was based on disease-specific meta-analyses. It aimed at assessment of alcohol impact by comparative quantification of health risk and burden of diseases due to alcohol use [55,56]. The drinking categories were defined into four groups: • Abstainers (reference): a person who drinks 0 g/day of alcohol. • Category I (responsible drinking): for females 0.1–19.99 g pure alcohol daily; for males 0.1–39.99 g pure alcohol daily; • Category II (hazardous drinking): for females 20–39.99 g pure alcohol daily, for males 40–59.99 g pure alcohol daily; and • Category III (harmful drinking): for females 40 g or more pure alcohol daily, for males 60 g or more pure alcohol daily. We assigned the level of alcohol consumption from each study to these groups based on the midpoint of their alcohol consumption classes. This categorization of alcohol drinking makes possible the comparison of heterogeneous classification of alcohol intake and at the same time, allows inclusion of data from studies in which precise information on levels of alcohol consumption were not available. When the upper bound of the highest category was not indicated, we used the range of previous category. The odds ratio (OR) was used as a measure of effect with its 95% confidence interval (95%CI) between alcohol consumption and risk of (MS). In some studies, the average alcohol consumption from more than 1 category fell into the same group of alcohol consumption. These were pooled to obtain a unique estimate per group of alcohol consumption for each study [31]. When the confidence

interval of the log (OR) was not symmetrical, the largest value for the SE has been taken into account to be introduced in the metaanalysis. In most of included studies, the abstainers were used as the reference group, except in Rosell study [57], where the light alcohol intake category was considered as reference. A sensitivity analysis carried out without the Rosell study showed no significant difference between the abstainers and light consumers in the pooled estimates (data not shown). Therefore, we simply inverted the odds ratio and 95% IC in the meta-analysis of Category I versus Abstainers. For the other meta-analyses, we included the published OR’s. Distinct analyses were carried out for men and women, aimed at comparing each alcohol consumption category with the reference category independently. 2.5.2. Assessment of publication bias and heterogeneity Publication bias was examined through the use of a funnel plot [58]. Funnel plot asymmetry was further tested by using Egger’s method [59]. The Cochran’s Q-test of heterogeneity was performed to detect non-homogeneity between ORs of individual studies [60]). Higgins & Thompson’s I2 [61] was also used due to inherent limitations of the Cochran’s Q in detecting true heterogeneity. Sensitivity analyses were performed to assess whether there was potential heterogeneity sources and studies more influencing the results in the analyses. Potentially influencing studies were therefore removed from the analyses and results compared. 2.5.3. Choice of model Distinct analyses based on gender were carried out for each comparison of alcohol consumption categories versus the reference category (abstainers). Depending on a significant or not significant ORs homogeneity test or large value of I2 (see assessment of heterogeneity), fixed or random effects (DerSimonian and Laird) model were respectively used to aggregate data from studies and produce the pooled estimates. Results were displayed as forest plots. 2.5.4. Software All data were recorded in Microsoft Excel. The analyses were performed by using Review Manager Version 4.2.8 (Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2003). 3. Results 3.1. Study selection Following the search strategy, 41 articles were identified as potentially relevant articles. The selection of studies was initially based on reading the titles and abstracts. 18 articles were rejected for not matching the inclusion criteria. Of 23 remaining articles, one article was excluded because the study concerned the severely obese subjects [62]. This option enabled us to avoid bias related to differences other than alcohol consumption between individuals. 7 papers were eliminated because of insufficient reported data either about the quantity of alcohol intake [63], or the (MS) [64,65,20]. In case of multiple reports publication [66,67], only the most recent and relevant article was included [26] (Fig. 1). In all, 14 observational studies met the eligibility criteria, only 7 of them that reported separate results by gender were selected for the metaanalysis (5 studies [57,35,26,49,38] presented results for both sexes, one study [50] for male only et one [68] for female only). The characteristics of the included observational studies, that assessed the association between alcohol consumption and the

Table 2 Characteristics of included studies published after 1998 that examined the relation between alcohol consumption and the prevalence of metabolic syndrome.

1

3

Study name, country

Number of participants (sample size)

Study design and setting

Age range at baseline (years)

Definition used to identify metabolic syndrome

Confounders adjusted for

Categories of alcohol consumption for each study

Converted categories of alcohol consumption into g/day

Prevalence of metabolic syndrome (%)

Measures of association: OR (95%CI)

Relevant findings

Zhu, 2004 [49]

NHANES III, USA (1988–1994)

11,239 subjects

Cross-sectional

>20 years

NCEP ATP III

Age, race, education, income, menopause and other lifestyle modifiable factors

Never: 0, L-M: ≤2 drinks/day, H: ≥5 drinks/day

(1) 0 g/day, (2) ≤24 g/day, (3) ≥60 g/day

M: 23.00

1.00, 0.89 (0.64–1.23), 1.27 (0.89–1.81)

F: 21.90

1.00, 0.76 (0.61–0.95), 0.62 (0.44–0.87)

J-shaped effect of alcohol. Increased risk with heavy drinking in men. Moderate alcohol consumption was not significantly related to the risk of metabolic syndrome in men Moderate alcohol intake was associated with lower odd ratio in women (protective effect)

M: 31.63

1.00, 0.68 (0.36–1.28), 0.72 (0.50–1.03), 0.66 (0.44–0.99), 0.80 (0.55–1.16)

F: 30.25

1.00, 0.88 (0.43–1.34), 0.47 (0.33–0.66), 0.47 (0.30–0.74), 0.39 (0.21–0.74) 0.79 (0.41–1.51), 1, 0.95 (0.65)

Djoussé, 2004 [35]

Rosell, 2003 [57]

NHLBI Family Heart Study, USA (1993–1994)

Stockholm, Sweden (1997–1999)

4510 subjects

4232 subjects

Cross-sectional

Cross-sectional

25–92 years

60-year-old men and women

NCEP ATP III

EGIR

Age, education, current smoking, physical activity, CHD, energy intake, dietary cholesterol, use of multivitamins, fibre, percent energy from polyunsaturated fatty acid, saturated fatty acids.

Smoking, education, immigration, employment, physical activity, vegetable intake.

Never = 0, 0.1–2.5 g/day, 2.6–12.0 g/day, 12.1–24.0 g/day, >24.0 g/day

Never: 0, L: 1–10 g/day, M: >10–30 g/day, H: >30 g/day

No conversion

No conversion

M: 19

F: 13

1.23 (0.80–1.89), 1, 0.84 (0.50–1.40)

U-shaped relation in men, Dose–response relation in women. Moderate alcohol consumption is associated with lower prevalence of MS, among men and women, irrespective of the type of beverage consumed.

627

Compared with low alcohol drinkers, moderate wine drinkers exhibited a more favorable metabolic effect. MS is more common in non-drinkers women, and lower in wine drinkers.

A. Alkerwi et al. / Atherosclerosis 204 (2009) 624–635

2

Authors and year of publication (Ref. no.)

628

Table 2 (Continued ) Study name, country

Number of participants (sample size)

Study design and setting

Age range at baseline (years)

Definition used to identify metabolic syndrome

Confounders adjusted for

Categories of alcohol consumption for each study

Converted categories of alcohol consumption into g/day

Prevalence of metabolic syndrome (%)

Measures of association: OR (95%CI)

Relevant findings

4

Wannamethee, 2006 [50]

British Regional Heart Study, UK (1978–2000)

3501 males

Cross-sectional data from a longitudinal cohort study

60–79 years

NCEP ATP III

None or occasional ≤ 1 drink/day, L = 1–2 drink/day, M = 3–4 drink/day, H = 5 drink/day (1 drink = 10 g)

(1) 0 g/d, (2) 10–20 g/day, (3) 30–40 g/day, (4) 50 g/day

Total: 25.60

1, 0.96 (0.78–1.18), 0.82 (0.57–1.20), 0.85 (0.48–1.50)

No statistical association was found between alcohol intake and metabolic syndrome.

5

Yoon, 2004 [26]

Korean National Health and Nutrition Examination Survey (KNHNES), Korea, 1998

7962 subjects

Cross-sectional

≥20 years

NCEP ATP III

Age, body mass index, physical activity, smoking, total fat and carbohydrate intake. Age, BMI, education level, income, marital status, smoking, exercise and percentage of energy from fat.

None, L: 1.0–14.9 g/day, M: 15.0–29.9 g/day, H: ≥30 g/day, very H: ≥80 g/day

No conversion

Male: 20.80

1, 0.71 (0.53–0.95), 0.88 (0.63–1.24), 0.87 (0.63–1.19), 1.07 (0.71–1.63)

Dose–response relation was found between alcohol consumption and the odds ratio for MS. Light alcohol consumption (1–15 g/day) is negatively associated with MS and might have a favorable effect.

Female: 26.90

1, 0.80 (0.65–0.98), 0.86 (0.53–1.40), 1.55 (0.77–3.13) 1, 0.65 (0.54–0.79), 0.34 (0.26–0.47)

6

Freiberg, 2004 [53]

NHANES III, USA, 1988–1994

8125 subjects

Cross-sectional

>20 years

NCEP ATP III

Age, sex, race/ethnicity, education, annual income, smoking, physical activity, percentage of carbohydrates intake.

<1 drink/month, 1–19 drinks/month, ≥20 drinks/month

<12 g/month = <0.4 g/day), 12–228 g/month = 0.4–7.6 g/day, ≥240 g/month = ≥8 g/day

No presented data

Low prevalence of MS among drinkers compared to non-drinkers. Inverse relation of alcohol consumption and the MS among drinkers de ≥ 20 drinks/month, and stronger with beer and wine beverages. Mild to moderate alcohol consumption is associated with a lower prevalence of the metabolic syndrome, especially in white individuals.

A. Alkerwi et al. / Atherosclerosis 204 (2009) 624–635

Authors and year of publication (Ref. no.)

7

Gigleux, 2006 [69]

Quebec Cardiovascular Study, Canada, 1985–1998

1966 males

Populationbased, cohort study (baseline)

35–64 years

Modified NCEP ATP III

Not mentioned

8

Santos, 2007 [38]

Porto, Portugal, 1999–2003

2164 subjects

Populationbased study, cross-sectional

19–92 years

NCEP ATP III

Age, education, total physical activity, smoking, total ethanol intake

Urashima, 2005 [48]

Tokyo, Japan, 2000–2004

22,892 subjects

Check-up visit in health centre

20–93 years

No conversion

Total: 29

Relative risk, 1, 0.72 (0.55–0.95), 0.67 (0.52–0.89), 0.57 (0.43–0.75)

Consumption of 1 drink or more daily (>15 g/day) had significant cardio protective effects in men.

No conversion

Male: 17.1

1, 0.85 (0.37–1.96), 0.93 (0.45–1.93), 1.56 (0.82–2.96)

No significant association found in male and female

Female: 22.8

1, 1.16 (0.82–1.64), 1.22 (0.85–1.76), 1.79 (0.91–3.51) 1, 1.31 (1.2–1.44), 1.94 (1.66–2.89)

Significant increased in the risk of metabolic syndrome in those drinking more than 150 g/week (>21.5 g/day)

Japanese MS definition

Age and gender

(1) 0–150 g/week, (2) 151–450 g/week, (3) 451 g/week (1 drink = 21.2 g)

(1) 0–21.4 g/day, (2) 21.5–64.2 g/day, (3) 64.4 g/day

4.8

American NCEP

Age and gender

(1) 0–21.4 g/day, (2) 21.5–64.2 g/day, (3) 64.4 g/day (1) 0 g/day, (2) 12.6–25.2 g/day, (3) ≥37.8 g/day

5.3

1, 1.20 (1.05–1.38), 1.69 (1.36–2.10)

Age, race, education, BMI, physical activity, smoking, carbohydrate intake, crude fibre and total fat intake Age, bone density, perimenopausal status, family history of hypertension and inability to exercise

(1) 0–150 g/week, (2) 151–450 g/week, (3) 451 g/week (1 drink = 21.2 g) (1) No drink/day, (2) 1–2 drinks/day, (3) ≥3 drinks/day (1 drink = 12.6 g)

No presented data

(Multivariable RRs), (1) 1.36 (1.11–1.66), (2) 1 (Referent), (3) 1.24 (0.83–1.85)

No alcohol intake is associated with an increased risk for the metabolic syndrome.

Non, L: ≤83 g/week, M: 84–167 g/week, (H) ≥168 g/week (1 drink = 12 g)

(1) 0 g/day, (2) ≤11.8 g/day, (3) 12–23.8 g/day, (4) ≥24 g/day

No presented data

1, 0.71 (0.63–0.81), 0.78 (0.65–0.93), High = no data

Low and moderate alcohol consumption were negatively associated with MS. Compared to non-drinkers, women drink ≥83–167 g/week have lower risk.

10

Carnethon, 2004 [51]

CARDIA study, USA, 1985–2001

4192 subjects

Communitybased, cohort study after 13.6 years of follow-up

18–30 years

NCEP ATP III

11

Lidfeldt, 2003 [68]

WHILA Study, south Sweden, 1995–2000

10,766 females

Community based

50–59 years

Adapted WHO definition

A. Alkerwi et al. / Atherosclerosis 204 (2009) 624–635

9

1st quartile: <1.3 g/day, 2nd quartile: 1.3–5.4 g/day, 3rd quartile: 5.5–15.1 g/day, 4th quartile: ≥15.2 g/day Non, L: 0.1–10 g/day, M: 10–29 g/day, H: ≥ 30 g/day

629

630

Table 2 (Continued ) Study name, country

Number of participants (sample size)

Study design and setting

Age range at baseline (years)

Definition used to identify metabolic syndrome

Confounders adjusted for

Categories of alcohol consumption for each study

Converted categories of alcohol consumption into g/day

Prevalence of metabolic syndrome (%)

Measures of association: OR (95%CI)

Relevant findings

12

Villegas, 2004 [52]

Cork and Kerry Diabetes and heart Disease Study, South Ireland

1018 subjects

Cross-sectional

50–96 years

WHO definition

Never, L: 0.55–0.99 unit/day, M: 1–2.99 unit/day, H: ≥3 units/day (1 drink = 12 g)

0 g/day, 6.6–11.8 g/day, 12–35.9 g/day, ≥36 g/day

Total: 21

Occasional = 1 (referent), 1.14 (0.71–1.83), 0.89 (0.46–1.71), 1.62 (0.77–3.42), 1.51 (0.64–3.56)

No significant association between current alcohol consumption and the prevalence of metabolic syndrome.

13

Wilsgaard, 2007 [46]

Tromso Study, Norway, 1979–2001

17,014 subjects

Population based longitudinal study

Age, sex, smoking, socioeconomic status, physical activity, pre-existing cardiovascular disease. Age, coffee consumption, smoking, education, leisure time physical activity and baseline examination

0–1 time/month, 2 times/month, 3–5 times/month, >6 times/month

Non-convertible data

No presented data

Male hazard ratio, 1.00, 1.18 (0.97–1.44), 0.98 (0.82–1.19), =0.95 (0.97–1.13)

Alcohol intake is inversely associated with metabolic syndrome in women but not in men. Low or no intake of alcohol is associated with an increased risk of metabolic syndrome.

14

Lee, 2006 [47]

Seoul and Kyung-gi, Republic of Korea, 2001

4341 subjects

Health check-up in hospital

NCEP ATP III

42.3 ± 10.4

NCEP ATP III

Age, sex, BMI,

L: low alcohol consumption, M: moderate alcohol consumption, H: heavy alcohol consumption, MS: metabolic syndrome.

(1) No drink, (2) >0– < 200 g/week, (3) ≥200– < 400 g/week, (4) ≥400 g/week

No presented 0 g/day, (2) data >0–28.5 g/day, (3) ≥28.5–<57.1 g/day

Female hazard ratio, 0–1 = 1.00, 2 = 0.80 (0.66–0.97), 3–5 = 0.72 (0.59–0.88), >6 = 0.72 (0.58–0.89) Risk ratio, no 1, 0.8 (0.6–1.2), 0.8 (0.5–1.6), 0.8 (0.5–2.2)

No significant association between alcohol consumption and metabolic syndrome after age and sex adjustment.

A. Alkerwi et al. / Atherosclerosis 204 (2009) 624–635

Authors and year of publication (Ref. no.)

A. Alkerwi et al. / Atherosclerosis 204 (2009) 624–635

631

Fig. 1. Flow chart of articles inclusion in meta-analysis.

prevalence of (MS), are summarized in Table 2. The studies covered different populations, where the majority came from North America, Europe and from eastern countries, with sample sizes ranged from 1018 in an Irish study [52] to 22,892 subjects in a Japanese study [48]. Men were the majority or sole participants in several studies. The Swedish WHILA Study was the only female community-based study. 8 studies had a cross-sectional design, 4 had cohort longitudinal design, whereas 2 were based on persons recruited during the check-up visits to health care centres. The (MS) was defined by WHO criteria in 2 studies, by NCEPATP III criteria in 11 studies, and only one study by EGIR criteria. There was a wide range of (MS) prevalence estimation, varying from less than 5.5% into Japanese population to more than 31% among the Americans. The included studies differed considerably with respect to the number of confounders adjusted ranging from 6 to 12 confounders including age, gender, education, smoking, physical activity, energy intake and other lifestyle factors. 3.2. Studies overview A general overview from 14 observational studies about alcohol–(MS) relationship has shown controversial reported results. Some studies have not found a statistically significant association between alcohol consumption and the prevalence of (MS) [50,38,51,52,47] while others have reported a positive association, showing J-shaped [47,49] or U-shaped relationship [35]. Compared to abstainers, light-to-moderate alcohol consumers (average 15–36 g/day) have mostly a favorable metabolic effect [26,69], among white individuals [53]. This protective effect was indirectly

explained by a wide range of confounding factors such as age, body mass, smoking, type of alcohol, exercise, and socio-economic status. Statistically significant gender differences existed in many articles [49,35,57]. In a Swedish female community-based study, low and moderate alcohol consumption was negatively associated with (MS). Women drinking 12–24 g/day have lower risk compared to non-drinkers [68]. Similar results have been found into Norwegian Tromso study [46]. In USA, the prevalence of (MS) varied from 23 to 31.63% in men and from 21.9 to 30.25% in female. In Europe, this prevalence seemed to be low in both sex; 19% in male versus 13% in female (Sweden), 17.1% in male versus 22.8% in female (Portugal). The lowest prevalence of (MS) was recorded in Japan, 4.8% versus 5.3 in male and female, respectively. 3.3. Data synthesis Two distinct sub-groups analyses were carried out in terms of gender. For men, the database comprised the studies from Zhu et al., Djousse et al., Rosell et al., Wannamethee et al., Yoon et al. and Santos et al. for a total of 17,202 subjects. The women part of the database comprised the studies from Zhu et al., Djousse et al., Rosell et al., Santos et al., Lidfeldt et al. and Yoon et al. for a total of 22,233 subjects. 3.4. Statistical analyses 7 studies [49,35,57,50,26,38,68], were included in the statistical pooling, which presented data for male and female separately. The

632

A. Alkerwi et al. / Atherosclerosis 204 (2009) 624–635

Fig. 2. Odds ratios for metabolic syndrome in men comparing category 1 of alcohol intake versus no-drinkers by using a fixed effect model.

Fig. 3. Odds ratios for metabolic syndrome in women comparing category 1 of alcohol intake versus no-drinkers.

other 7 studies could not be included because of missing data in terms of gender.

1.09]) (Fig. 4) or in women, although the OR was lower in women (OR = 0.81, 95%CI = [0.57; 1.14]) (Fig. 5).

3.5. Category 1 versus non-drinkers meta-analysis

3.7. Category 3 versus non-drinkers meta-analysis

In men, the fixed effect model showed a significant relationship of the (MS) and drinking 0.1–39.9 g/day of alcohol. A protective effect could be attributed to alcohol consumption (OR = 0.84, 95%CI = [0.75; 0.94]) (Fig. 2). Due to the heterogeneity, a random effect model was used in women (Fig. 3). The odds ratio of (MS) was 0.75 (95%CI = [0.64; 0.89]).

There was no data available for evaluating the odds of (MS) due alcohol consumption in the third category for women. In men, the two studies evaluated with a random effect model gave a non-significant result with an OR = 0.99 (95%CI = [0.71; 1.38]) (Fig. 6). 3.8. Sensitivity analysis

3.6. Category 2 versus non-drinkers meta-analysis The association between the MS and hazardous drinking of alcohol was not significant either in men (OR = 0.95, 95%CI = [0.83;

A sensitivity analysis was carried out without the Rosell study. There were no significant changes in the pooled estimates when this study was not accounted in the meta-analysis.

Fig. 4. Odds ratios for metabolic syndrome in men comparing category 2 of alcohol intake versus no-drinkers by using a fixed effect model.

A. Alkerwi et al. / Atherosclerosis 204 (2009) 624–635

633

Fig. 5. Odds ratios for metabolic syndrome in women comparing category 2 of alcohol intake versus no-drinkers.

Fig. 6. Odds ratios for metabolic syndrome in men comparing category 3 of alcohol intake versus no-drinkers.

For the category of alcohol consumption 0.1–19.9 g/day in women, heterogeneity was mainly due to Djoussé study. Successive excluding and reintroducing of the others studies led to little variation of the odds ratios in this category. Following the same process for the analysis of category 2 of alcohol consumption (20–39.9 g/day) in women heterogeneity remained. 3.9. Publication bias The funnel plots did not show marked asymmetry and all Egger’s tests were not significant. 4. Conclusion and discussion The present study attempted to address the cardio-metabolic effects of alcohol consumption by pooling the observational, crosssectional evidence in relation to the prevalence of (MS). To the best of our knowledge, this is the first systematic overview of observational studies that has investigated alcohol–(MS) association and the dose–response in terms of gender, by suggesting a new method to harmonise the alcohol data. Literature review showed that the relation between alcohol consumption and (MS) is not consistent. Some studies have shown positive association, whereas others have observed a negative association or no relation at all. This controversy could be related to the complex mechanistic relation between alcohol consumption and each component of MS. While mild to moderate alcohol consumption has a favorable influence on lipids metabolism, abdominal obesity and glucose regulation [18], on the other hand, alcohol consumption causes hypertension [16], and hypertriglyceridemia [25], constituting alcohol-related metabolic syndrome. Our meta-analyses results demonstrate that people who drink alcohol beverages have a lower prevalence of the (MS) as compared with the current non-drinkers. The inverse relation of alcohol consumption and the (MS) was especially noticeable in men who consume <40 g/day and in women who consume <20 g/day.

We are conscious that meta-analysis of observational studies present particular challenges because of inherent biases and differences in study design [40], yet, we hope that this research may provide a tool for helping to understand the alcohol–MS relationship and particularly gender-specific dose–response. As with other meta-analyses of published studies, the present analysis has various limitations and strong points. First, its quality is inherited from the validity of the original included studies. Most studies did not take into account different drinking patterns, for example drinking with/without meals, binge drinking, and lifetime or occasional. In the majority of studies, lifetime abstainers and former drinkers were combined into one category “non-drinkers”, leading thus to limited information about risk analyses for these two groups separately. Combining abstainers and former drinkers into currently “non-drinkers” could induce bias, and veil the real effect of non-alcohol use. Some of these abstainers might have give up drinking for health reasons. The U- or J-shaped association between alcohol consumption and mortality from cardiovascular disease has been suggested due to the inclusion of ex-drinkers in the reference group of abstainers [31]. Most studies did not analyse the effect of beverage type (wine, beer, spirits, flavoured mixed drink), and race (white or black) on the risk of (MS). Therefore, we cannot answer whether different types of alcoholic beverages are equivalent in their ability to protect against (MS), or if a specific race might offer a greater protection. There are a number of confounding variables that may have a considerable effect on the pattern and type of alcohol consumption, such as socio-economic status, education, occupation, diet, physical activity and exercise, which were not taken in consideration in all the studies. So far, uncontrolled confounding by other risk factors can be reasonably excluded because the great majority of studies were adjusted for most potential confounders. To reduce a potential bias, we abstracted the inclusive model and carried out sex-specific subgroup meta-analyses of only relevant and appropriate data. Second, the data on alcohol consumption are based on selfdeclaration with the possibility of misclassification of exposure due to under-reporting. However, epidemiological community-based

634

A. Alkerwi et al. / Atherosclerosis 204 (2009) 624–635

studies have shown that the reliability of self-reported alcohol consumption is good [70,71]. An advantage could reside, in our opinion, in the fact that we included studies published in countries with widespread alcohol acceptance and nearly similar socio-cultural habits regarding this issue, which might reduce the potential bias related to self-reported alcohol intake. Third, publication bias of non-significant association was not encountered in our meta-analysis, since the funnel plots did not suggest a positive-results publication bias. Nevertheless, an availability bias cannot be excluded in spite of our intensive efforts to get individual data, most recent peer-reviewed, and other than English published papers. Fourth, the quality of alcohol research relies primarily on the accuracy of assessment of alcohol consumption. To date, there is no existing consensus for measuring alcohol consumption and for defining a “standard drink”. The major difficulty in our metaanalysis was related to the heterogeneous categorizations of alcohol consumption among studies. Lack of a standardised measure of alcohol use complicated the interpretation of finding across studies. Therefore, we first quantified the daily alcohol intake by converting different units of alcohol consumption into grams per day. Then we matched each reported category with the four groups of alcoholhealth risk assessment, adopted by the WHO, in order to get a meaningful conclusion. By this paper, we tried to shed light on an important methodological issue that created difficulty in making conclusion out of published studies. Our alcohol consumption quantification method and the construct validity of alcohol categorization may lack precision, but in our opinion sufficient to evaluate alcohol dose–response. To our knowledge, no other investigators have suggested such a method to overcome the problematic of alcohol intake assessment. Fifth, in our meta-analysis, the definition of (MS) should not provide unstable estimates, because the NCEP-ATP III definition was used in mostly all included studies. Sixth, despite the intensive effort to search and include relevant literatures, the availability of articles was relatively limited, notably for the category III (harmful drinking), leading to underpowered inference about the effect of harmful drinking on MS from this meta-analysis. Finally, given the cross-sectional design, we cannot draw any causal inferences regarding the association of alcohol consumption with the (MS). However, the observed alcohol–(MS) association suggests that “responsible drinking” of up to 19.99 g/day in women and up to 39.99 g/day in men placed the individual at lower risk of having the (MS). Without a large-scale long-term intervention trail, we cannot confirm that the adoption of such behaviours by individuals will result in a lowering of the risk of developing a metabolic syndrome. For the time being, such trials seem to be unfeasible for several reasons, including ethical concerns [72]. Nevertheless, our results lead us to objective that this may be the case and thus have a value for public health recommendations. It is a compelling evidence to support the concept that daily alcohol intake in moderation may be part of a healthy lifestyle. Although this finding may be important for persons who regularly consume alcoholic beverage with moderation, no generalisation about the benefits of a change in lifestyle should be drawn. From a public health point of view, we should not recommend current non-consumers to drink alcohol and need not discourage regular moderate consumers from consuming alcohol. Overall, the effect of alcohol intake should be considered in context with drinking patterns and the presence of other cardiovascular risk factors. Any potential benefit of alcohol consumption has to be balanced with alcohol’s well-known social and health adverse effects (cancers, accidents) [73]. In summary, concerning the (MS), as a clinical entity having essential atherosclerotic consequences, “responsible alcohol

intake” appears to be associated with a reduced prevalence of (MS). Favorable metabolic effect seemed to be restricted to alcohol consumption of less than 20 g/day among women, and of less than 40 g/day among men. Acknowledgment We express our thanks to Coralie Dessenne, a medical librarian at the Centre d’Etudes en Santé, who provides an important help to design the literature search plan. Contributors: AAlkerwi conceived the design, conducted data acquisition, contributed to the data interpretation, wrote the first draft of the manuscript and critically revised the manuscript. MB performed the statistical meta-analysis. MV assisted in the statistical analyses and data interpretation. JB helped in data retrieving and data capturing. MD provides input into the design and analysis, reviewed the global work and critically revised the manuscript. MLL, AAlbert and MG reviewed the global work and critically revised the manuscript. All authors read and gave their final approval of the submitted manuscript. References [1] Gami AS, Witt BJ, Howard DE, et al. Metabolic syndrome and risk of incident cardiovascular events and death: a systematic review and meta-analysis of longitudinal studies. J Am Coll Cardiol 2007;49(4):403–14. [2] Galassi A, Reynolds K, He J. Metabolic syndrome and risk of cardiovascular disease: a meta-analysis. Am J Med 2006;119(10):812–9. [3] Girman CJ, Rhodes T, Mercuri M, et al. The metabolic syndrome and risk of major coronary events in the Scandinavian Simvastatin Survival Study (4S) and the Air Force/Texas Coronary Atherosclerosis Prevention Study (AFCAPS/TexCAPS). Am J Cardiol 2004;93(2):136–41. [4] Lakka HM, Laaksonen DE, Lakka TA, et al. The metabolic syndrome and total and cardiovascular disease mortality in middle-aged men. JAMA 2002;288(21):2709–16. [5] Stern MP, Williams K, Gonzalez-Villalpando C, Hunt KJ, Haffner SM. Does the metabolic syndrome improve identification of individuals at risk of type 2 diabetes and/or cardiovascular disease? Diabetes Care 2004;27(11):2676–81. [6] Malik S, Wong ND, Franklin SS, et al. Impact of the metabolic syndrome on mortality from coronary heart disease, cardiovascular disease, and all causes in United States adults. Circulation 2004;110(10):1245–50. [7] Isomaa B, Almgren P, Tuomi T, et al. Cardiovascular morbidity and mortality associated with the metabolic syndrome. Diabetes Care 2001;24(4): 683–9. [8] Lorenzo C, Williams K, Hunt KJ, Haffner SM. The National Cholesterol Education Program - Adult Treatment Panel III, International Diabetes Federation, and World Health Organization definitions of the metabolic syndrome as predictors of incident cardiovascular disease and diabetes. Diabetes Care 2007;30(1):8–13. [9] Bonora E. The metabolic syndrome and cardiovascular disease. Ann Med 2006;38(1):64–80. [10] Wilson PW, D’Agostino RB, Parise H, Sullivan L, Meigs JB. Metabolic syndrome as a precursor of cardiovascular disease and type 2 diabetes mellitus. Circulation 2005;112(20):3066–72. [11] Lau DC, Yan H, Dhillon B. Metabolic syndrome: a marker of patients at high cardiovascular risk. Can J Cardiol 2006;22(Suppl B):85B–90B. [12] He Y, Jiang B, Wang J, et al. Prevalence of the metabolic syndrome and its relation to cardiovascular disease in an elderly Chinese population. J Am Coll Cardiol 2006;47(8):1588–94. [13] Ford ES, Kohl III HW, Mokdad AH, Ajani UA. Sedentary behavior, physical activity, and the metabolic syndrome among U.S. adults. Obes Res 2005;13(3):608–14. [14] Liese AD, Mayer-Davis EJ, Haffner SM. Development of the multiple metabolic syndrome: an epidemiologic perspective. Epidemiol Rev 1998;20(2):157–72. [15] Sookoian SC, Gonzalez C, Pirola CJ. Meta-analysis on the G-308A tumor necrosis factor alpha gene variant and phenotypes associated with the metabolic syndrome. Obes Res 2005;13(12):2122–31. [16] Xin X, He J, Frontini MG, Ogden LG, Motsamai OI, Whelton PK. Effects of alcohol reduction on blood pressure: a meta-analysis of randomized controlled trials. Hypertension 2001;38(5):1112–7. [17] Nakanishi N, Makino K, Nishina K, Suzuki K, Tatara K. Relationship of light to moderate alcohol consumption and risk of hypertension in Japanese male office workers. Alcohol Clin Exp Res 2002;26(7):988–94. [18] Koppes LL, Dekker JM, Hendriks HF, Bouter LM, Heine RJ. Moderate alcohol consumption lowers the risk of type 2 diabetes: a meta-analysis of prospective observational studies. Diabetes Care 2005;28(3):719–25. [19] Bell RA, Mayer-Davis EJ, Martin MA, D’Agostino Jr RB, Haffner SM. Associations between alcohol consumption and insulin sensitivity and cardiovascular dis-

A. Alkerwi et al. / Atherosclerosis 204 (2009) 624–635

[20]

[21]

[22] [23] [24]

[25]

[26]

[27]

[28] [29] [30]

[31] [32] [33]

[34] [35]

[36]

[37] [38] [39] [40]

[41]

[42] [43]

[44]

[45] [46] [47]

ease risk factors: the Insulin Resistance and Atherosclerosis Study. Diabetes Care 2000;23(11):1630–6. Goude D, Fagerberg B, Hulthe J. Alcohol consumption, the metabolic syndrome and insulin resistance in 58-year-old clinically healthy men (AIR study). Clin Sci (Lond) 2002;102(3):345–52. Kiechl S, Willeit J, Poewe W, et al. Insulin sensitivity and regular alcohol consumption: large, prospective, cross sectional population study (Bruneck study). BMJ 1996;313(7064):1040–4. Magis DC JB, Scheen AJ. Alcohol, Sensibilité à l’insuline et diabète sucré. Rev Med Lieg 2003;58:501–7. Hulley SB, Gordon S. Alcohol and high-density lipoprotein cholesterol: causal inference from diverse study designs. Circulation 1981;64(3 Pt 2):57–63. van Tol A, van der Gaag MS, Scheek LM, van Gent T, Hendriks HF. Changes in postprandial lipoproteins of low and high density caused by moderate alcohol consumption with dinner. Atherosclerosis 1998;141(Suppl 1):S101–3. Kato I, Kiyohara Y, Kubo M, et al. Insulin-mediated effects of alcohol intake on serum lipid levels in a general population: the Hisayama Study. J Clin Epidemiol 2003;56(2):196–204. Yoon YS, Oh SW, Baik HW, Park HS, Kim WY. Alcohol consumption and the metabolic syndrome in Korean adults: the 1998 Korean National Health and Nutrition Examination Survey. Am J Clin Nutr 2004;80(1):217–24. Nanchahal K, Ashton WD, Wood DA. Alcohol consumption, metabolic cardiovascular risk factors and hypertension in women. Int J Epidemiol 2000;29(1):57–64. Wannamethee SG, Shaper AG. Alcohol, body weight, and weight gain in middleaged men. Am J Clin Nutr 2003;77(5):1312–7. Ajani UA, Gaziano JM, Lotufo PA, et al. Alcohol consumption and risk of coronary heart disease by diabetes status. Circulation 2000;102(5):500–5. Rimm EB, Williams P, Fosher K, Criqui M, Stampfer MJ. Moderate alcohol intake and lower risk of coronary heart disease: meta-analysis of effects on lipids and haemostatic factors. BMJ 1999;319(7224):1523–8. Reynolds K, Lewis B, Nolen JD, Kinney GL, Sathya B, He J. Alcohol consumption and risk of stroke: a meta-analysis. JAMA 2003;289(5):579–88. Berger K, Ajani UA, Kase CS, et al. Light-to-moderate alcohol consumption and risk of stroke among U.S. male physicians. N Engl J Med 1999;341(21):1557–64. Kiechl S, Willeit J, Rungger G, Egger G, Oberhollenzer F, Bonora E. Alcohol consumption and atherosclerosis: what is the relation? Prospective results from the Bruneck Study. Stroke 1998;29(5):900–7. Agarwal DP. Cardioprotective effects of light–moderate consumption of alcohol: a review of putative mechanisms. Alcohol Alcohol 2002;37(5):409–15. Djousse L, Arnett DK, Eckfeldt JH, Province MA, Singer MR, Ellison RC. Alcohol consumption and metabolic syndrome: does the type of beverage matter? Obes Res 2004;12(9):1375–85. Fan AZ, Russell M, Dorn J, et al. Lifetime alcohol drinking pattern is related to the prevalence of metabolic syndrome. The Western New York Health Study (WNYHS). Eur J Epidemiol 2006;21(2):129–38. Kawada T, Okada K. The metabolic syndrome: prevalence and associated lifestyles in Japanese workingmen. J CardioMetabol Synd 2006;1(5):313–7. Santos AC, Ebrahim S, Barros H. Alcohol intake, smoking, sleeping hours, physical activity and the metabolic syndrome. Prev Med 2007;44(4):328–34. Botella JGH. Doing and reporting a meta-analysis. Int J Clin Health Psych 2006;6(2):425–40. Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA 2000;283(15):2008–12. Corrao G, Bagnardi V, Zambon A, Arico S. Exploring the dose–response relationship between alcohol consumption and the risk of several alcohol-related conditions: a meta-analysis. Addiction 1999;94(10):1551–73. Corrao G, Rubbiati L, Bagnardi V, Zambon A, Poikolainen K. Alcohol and coronary heart disease: a meta-analysis. Addiction 2000;95(10):1505–23. Alberti KG, Zimmet PZ. Definition, diagnosis and classification of diabetes mellitus and its complications. Part 1. Diagnosis and classification of diabetes mellitus provisional report of a WHO consultation. Diabet Med 1998;15(7):539–53. Executive Summary of The Third Report of The National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, And Treatment of High Blood Cholesterol In Adults (Adult Treatment Panel III). JAMA 2001;285(19):2486–97. Alberti KG, Zimmet P, Shaw J. The metabolic syndrome—a new worldwide definition. Lancet 2005;366(9491):1059–62. Wilsgaard T, Jacobsen BK. Lifestyle factors and incident metabolic syndrome. The Tromso Study 1979–2001. Diabetes Res Clin Pract; 2007. Lee WY, Jung CH, Park JS, Rhee EJ, Kim SW. Effects of smoking, alcohol, exercise, education, and family history on the metabolic syndrome as defined by the ATP III. Diabetes Res Clin Pract 2005;67(1):70–7.

635

[48] Urashima M, Takashi W, Tsutomu F, et al. Prevalence of metabolic syndrome in a 22,892 Japanese population and its association with life style. JMAJ 2005;49(9):441–50. [49] Zhu S, St-Onge MP, Heshka S, Heymsfield SB. Lifestyle behaviors associated with lower risk of having the metabolic syndrome. Metabolism 2004;53(11):1503–11. [50] Wannamethee SG, Shaper AG, Whincup PH. Modifiable lifestyle factors and the metabolic syndrome in older men: Effects of lifestyle changes. J Am Geriatr Soc 2006;54(12):1909–14. [51] Carnethon MR, Loria CM, Hill JO, Sidney S, Savage PJ, Liu K. Risk factors for the metabolic syndrome: the Coronary Artery Risk Development in Young Adults (CARDIA) study, 1985–2001. Diabetes Care 2004;27(11):2707–15. [52] Villegas R, Creagh D, Hinchion R, O’Halloran D, Perry IJ. Prevalence and lifestyle determinants of the metabolic syndrome. Ir Med J 2004;97(10):300–3. [53] Freiberg MS, Cabral HJ, Heeren TC, Vasan RS, Curtis Ellison R. Alcohol consumption and the prevalence of the Metabolic Syndrome in the US.: a cross-sectional analysis of data from the Third National Health and Nutrition Examination Survey. Diabetes Care 2004;27(12):2954–9. [54] Rehm J, Klotsche J, Patra J. Comparative quantification of alcohol exposure as risk factor for global burden of disease. Int J Methods Psychiatric Res 2007;16(2):66–76. [55] Rehm J, Room R, Monteiro M, et al. Alcohol. In: Ezzati M, Lopez AD, Rodgers A, Murray CJ, editors. Comparative quantification of health risks: Global and regional burden of disease due to selected major risk factors. Geneva: WHO; 2004. p. 959–1108. [56] Rehm J, Room R, Monteiro M, et al. Alcohol as a risk factor for global burden of disease. Eur Addict Res 2003;9(4):157–64. [57] Rosell M, De Faire U, Hellenius ML. Low prevalence of the metabolic syndrome in wine drinkers—is it the alcohol beverage or the lifestyle? Eur J Clin Nutr 2003;57(2):227–34. [58] Sterne JA, Egger M. Funnel plots for detecting bias in meta-analysis: guidelines on choice of axis. J Clin Epidemiol 2001;54(10):1046–55. [59] Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997;315(7109):629–34. [60] Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ 2003;327(7414):557–60. [61] Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med 2002;21(11):1539–58. [62] Dixon JB, Dixon ME, O’Brien PE. Alcohol consumption in the severely obese: relationship with the metabolic syndrome. Obes Res 2002;10(4):245–52. [63] Liu J, Young TK, Zinman B, Harris SB, Connelly PW, Hanley AJ. Lifestyle variables, non-traditional cardiovascular risk factors, and the metabolic syndrome in an Aboriginal Canadian population. Obesity (Silver Spring) 2006;14(3): 500–8. [64] Flanagan DE, Moore VM, Godsland IF, Cockington RA, Robinson JS, Phillips DI. Alcohol consumption and insulin resistance in young adults. Eur J Clin Invest 2000;30(4):297–301. [65] Beulens JW, van Beers RM, Stolk RP, Schaafsma G, Hendriks HF. The effect of moderate alcohol consumption on fat distribution and adipocytokines. Obesity (Silver Spring) 2006;14(1):60–6. [66] Park HS, Oh SW, Cho SI, Choi WH, Kim YS. The metabolic syndrome and associated lifestyle factors among South Korean adults. Int J Epidemiol 2004;33(2):328–36. [67] Park YW, Zhu S, Palaniappan L, Heshka S, Carnethon MR, Heymsfield SB. The metabolic syndrome: prevalence and associated risk factor findings in the US population from the Third National Health and Nutrition Examination Survey, 1988–1994. Arch Intern Med 2003;163(4):427–36. [68] Lidfeldt J, Nyberg P, Nerbrand C, Samsioe G, Schersten B, Agardh CD. Sociodemographic and psychosocial factors are associated with features of the metabolic syndrome. The Women’s Health in the Lund Area (WHILA) study. Diabetes Obes Metab 2003;5(2):106–12. [69] Gigleux I, Gagnon J, St-Pierre A, et al. Moderate alcohol consumption is more cardioprotective in men with the metabolic syndrome. J Nutr 2006;136(12):3027–32. [70] Williams GD, Aitken SS, Malin H. Reliability of self-reported alcohol consumption in a general population survey. J Stud Alcohol 1985;46(3):223–7. [71] Giovannucci E, Colditz G, Stampfer MJ, et al. The assessment of alcohol consumption by a simple self-administered questionnaire. Am J Epidemiol 1991;133(8):810–7. [72] Di Castelnuovo A, Rotondo S, Iacoviello L, Donati MB, De Gaetano G. Metaanalysis of wine and beer consumption in relation to vascular risk. Circulation 2002;105(24):2836–44. [73] Burger M, Mensink G, Bronstrup A, Thierfelder W, Pietrzik K. Alcohol consumption and its relation to cardiovascular risk factors in Germany. Eur J Clin Nutr 2004;58(4):605–14.