Alcohol and Alcoholism Advance Access originally published online on November 22, 2006
Alcohol and Alcoholism 2007 42(1):49-54; doi:10.1093/alcalc/agl093
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
ALCOHOL INTAKE AND INCIDENCE OF CORONARY DISEASE IN AUSTRALIAN ABORIGINES
1 University of Western Australia, School of Medicine and Pharmacology, Royal Perth Hospital Unit Pilbara Region, Western Australia
2 School of Public Health, Curtin University of Technology Pilbara Region, Western Australia
3 School of Population Health, University of Queensland Pilbara Region, Western Australia
4 Puntukurnu Aboriginal Medical Service Pilbara Region, Western Australia
5 Miln Walker and Associates Pty Ltd PO Box 167, Belair, South Australia
*Author to whom correspondence should be addressed at: School of Medicine and Pharmacology, Royal Perth Hospital Unit, Box X2213 GPO, Perth 6847, Australia. Tel.: +61 8 9224 0276; Fax: +61 8 9224 0246; E-mail: vburke{at}cyllene.uwa.edu.au
Received 5 October 2006; in revised form 8 October 2006; in revised form 25 October 2006; accepted 28 October 2006
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONPOPULATION AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Aims: To examine risk for coronary heart disease (CHD) and cardiovascular disease (CVD) in relation to alcohol in a cohort of Australian Aborigines. Methods: In 1988–1989, alcohol intake, drinking pattern, and beverage preference were elicited by interviewer-administered questionnaire in Western Australian Aborigines (258 men and 256 women) and cardiovascular outcomes ascertained through linkage to mortality and hospital admission records to 2002. Results: In proportional hazards models, risk for CHD, relative to lifetime abstainers, was significantly increased in ex-drinkers [Hazard ratio (HR), 2.29; 95% confidence intervals (CI), 1.23–4.27], those drinking 41–60 g/day in men or 21–40 g/day in women (HR 2.80; 95% CI, 1.04–7.53) and those drinking >150 g/day for men or >100 g/day for women (HR, 2.25; 95% CI, 1.03–4.90) with a J-shaped relationship. Low-to-moderate drinkers had lower waist girth, exercised more and had a lower prevalence of overweight and smoking than at-risk drinkers. A preference for wine was associated with lower HR (0.28; 95% CI, 0.10–0.95). With CVD, only ex-drinkers showed significantly increased risk (HR, 1.87; 95% CI, 1.20–2.91). Conclusions: More favourable health-related behaviours in low-to-moderate drinkers suggest that lower risk could be mediated by lifestyle, as proposed in other populations.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONPOPULATION AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Epidemiological studies suggest that light-to-moderate alcohol intake may be protective against coronary heart disease (CHD) (Gronbaek, 2002
), a finding supported by meta-analyses (McClure, 1993; Corrao, et al., 2000). A U-or J-shaped relationship between coronary endpoints and alcohol intake is recognized (Marmot and Brunner, 1991; Murray et al., 2002). However, this association has not been observed in all populations (Hart et al., 1999; Fuchs et al., 2004). Differences between populations may relate to drinking patterns, with binge drinkers at greater risk of coronary events (McElduff and Dobson, 1997), beverage type, with possible benefits of wine-drinking (Klatsky et al., 2003), and to lifestyle factors (Ruidavets et al., 2004). Such associations have not been examined in Australian Aborigines, in whom, relative to the non-Aboriginal population, drinking alcohol at harmful levels is more common (Ministerial Council on Drug Strategy, 2001), and risk of death from cardiovascular disease has increased at least 3-fold (National Health and Medical Research Council, 2000).
Detailed information about patterns of drinking, alcohol intake and beverage preference was recorded in 1988–1989 from a cohort of Aboriginal people in the Kimberley region of Western Australia (Hunter et al., 1991, 1992), along with measurement of blood pressure (BP), blood lipids and anthropometry. Cross-sectional analysis at that time showed a rate of hypertension, defined according to cut-off points of 160/95 mmHg, almost three times that of non-Aboriginal Australians (Smith et al., 1992a). Examination of long-term outcomes in this cohort became possible with the establishment of the Western Australian Data Linkage System in 1995 (Holman et al., 1999), providing access to hospital and death records. We now report associations between alcohol intake, drinking pattern and beverage preference determined in the initial survey and endpoints for CHD and cardiovascular disease (CVD) over a 13 year follow-up period.
|
POPULATION AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONPOPULATION AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Cardiovascular risk factors and lifestyle, including details of usual alcohol consumption, were recorded in 1988–1989 in a cohort of Aborigines, aged between 15 and 88 years, in the Kimberley region of Western Australia (Smith et al., 1992b
). Individuals were randomly selected in an age- and gender-stratified sample from the Community Health Client Register of the Health Department of Western Australia in January 1987 as previously described (Smith et al., 1992a). The Commonwealth Scientific and Industry Research Organisation (CSIRO, Australia) and the Health Department of Western Australia gave ethical approval. All participants provided individual consent (Smith et al., 1992a).
Questionnaires and measurements
Interviewer-administered questionnaires provided information about demographics including usual place of residence classified according to ease of access to alcohol (very-hard-to-get, hard-to-get or easy-to-get). Alcohol intake and preference for beer, wine or spirits were also elicited by an interviewer-administered questionnaire, validated by comparison with sales figures in the community, correlation with measurement of 
glutamyl-transpeptidase (GGT) activity, and other information within the community. Participants were categorized into those who had never drunk alcohol, ex-drinkers, those who drank on most days of the week, those who drank intermittently at intervals of not >2 weeks and episodic drinkers who drank at intervals >2 weeks. The amount of alcohol consumed over 48 h was determined using a contextualized diary of the last two drinking periods and the volume was converted to grams of alcohol per drinking day. Drinkers were then grouped according to alcohol intake: 
40 g/day men, 20 g/day women; 41–60 g/day men, 21–40 g/day women; 61–120 g/day men, 41–80 g/day women; 121–150 g/day men, 81–100 g/day women; >150 g/day men, >100 g/day women. For ex-drinkers, reasons for giving up alcohol were recorded at interview. All interviews (E.H.) and physical examinations (R.S.) were undertaken by the same study personnel.
Smoking habits were recorded and the amount of physical activity was assessed by three questions that asked about the frequency (3 times a week, once or twice a week, about once a month, rarely or never) of exercise.
Blood pressure and anthropometry
BP was measured using a Dinamap oscillometric monitor 86249 (Johnson and Johnson, Sydney) as described previously (Smith et al., 1992a). An initial measurement of BP was taken (
15 min after entry) and 5 min after familiarization, BP was measured in triplicate at 1 min intervals and the mean of the last two was used in analysis. Weight was recorded using a Portable Field Scale Model MPS 120, calibrated daily, and height with a stadiometer (Harpenden Survey Set, CMS Weighing Equipment, London).
Data linkage and diagnosis coding
Participants were linked to hospital, cancer and death records from 1988 to 2000 by the Western Australian Data Linkage Unit. Ethical approval was obtained from the Western Australian Aboriginal Health Information and Ethics Committee, the University of Queensland Health Ethics Committee, and the Kimberley Aboriginal Medical Services Council. While linkage used for all available information, only de-identified data were used subsequently.
Diagnoses were coded from hospital separation records and death records in the period 1988–1999 using the ICD-9-CM (National Coding Centre, 1995) and from 1999 to 2000 using the ICD-10-AM (National Centre for Classification in Health, 2000). Mapping of ICD-10-AM codes to ICD-9-CM codes was carried out using software developed by the National Centre for Classification in Health. Cardiovascular disease (CVD) included the following ICD-9-CM codes: 391, 393–398, 401–404, 410, 411, 414–417, 420–438, 440–444. Coronary heart disease (CHD) comprised the codes 410–414. Diagnosis codes were included if they appeared in any of 21 diagnosis fields.
Statistical methods
CHD endpoints included mortality and a first admission to hospital with CHD; endpoints for CVD combined mortality from CVD and a first admission to hospital with CVD. Time to event was calculated from January 1, 1988 with a censoring date of 31 December 2000. Endpoints were examined using proportional hazards models in Stata (Statcorp, Texas) or SPSS (SPSS, Chicago, Ill). All models were adjusted for age, using a Box–Tidwell transformation (Hosmer and Lemeshow, 2000), and gender, as well as for the type of community classified according to ease of access to alcohol. Generalized linear models, with adjustment for gender, were used in comparing risk factors as continuous variables according to drinking categories. 
2-tests were used for categorical variables that included body mass index (BMI) classified according to the cut-off point of 22 kg/m2, as recommended for Aboriginal people (Daniel et al., 2002). P-values < 0.05 were considered significant.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONPOPULATION AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Characteristics of the population at baseline and lifestyle factors are shown in Tables 1 and 2. During follow-up from 1988 to 1989 to the end of 2002 there were 13 deaths from CHD among women and 11 among men; 55 women and 51 men had a first hospital admission for CHD recorded in this interval. For CVD there were 22 deaths and 97 first hospital admissions among women and 24 deaths and 91 first hospital admissions among men. Follow-up of participants was done for a mean of 11.6 years.
|
|
Among drinkers, only 8 (8%) women and 13 men (8%) had an alcohol intake within the Australian national guidelines of not more than an average of four standard drinks daily for men and two for women (National Health and Medical Research Council, 2001
). Among drinkers, 83% of men drank at least 60 g/day of alcohol while, among women, 82% exceeded 40 g/day. Alcohol intake ranged from 7 to 266 g/day in women and from 7 to 647 g/day in men; 76% of women and 84% of men preferred beer; only 1 woman and 5 men chose low-alcohol beer. Wine was preferred by 17% of women and 14% of men, and spirits by 6% of women and 2% of men.
Coronary heart disease
As shown in Table 3, in models adjusted for age, gender and type of community, risk of mortality or morbidity for CHD was significantly greater among ex-drinkers (P = 0.032) and in the highest category of alcohol intake (>150 g/day for men, >100 g/day for women), relative to lifetime abstainers (P = 0.045). Further adjustment for the classical risk factors of cholesterol, blood pressure, and waist girth showed significantly increased risk in ex-drinkers (P = 0.009), those drinking up to 60 g/day in men and 40 g/day in women (P = 0.041) and in the highest drinking category (P = 0.042). With the inclusion of smoking habits and exercise in the models, statistically increased risk persisted in ex-drinkers (P = 0.011) and those drinking up to 60 g/day in men and 40 g/day in women (P = 0.027). While ratios are not statistically significant for all categories, there is, overall, a pattern consistent with a J-shaped relationship between category of alcohol intake and risk of cardiovascular disease.
|
Cardiovascular disease
Table 4 shows the hazard ratios (HRs) and 95% confidence intervals (CI) with CVD as the endpoint. Relative to lifetime abstainers, the only category showing a statistically significant increase in risk was the ex-drinkers. No pattern suggesting lower risk in light-to-moderate drinkers was seen with CVD as the endpoint.
|
Associations with beverage type and drinking pattern
In models adjusted for age, gender and community type, excluding non-drinkers, and including amount of alcohol intake, pattern of drinking was not a significant predictor of CHD endpoints. The HR for intermittent and episodic drinkers relative to daily drinkers was 1.09 (95% CI, 0.51–2.34).
There was, however, a statistically significant association with beverage preference, independent of drinking pattern and amount of alcohol consumption. The few participants who preferred spirits were excluded from these models because of problems with convergence if they were included. With a preference for beer as the reference category, the HR associated with a preference for wine-drinking was 0.28 (95% CI, 0.10–0.95). The inverse association of CHD with wine-drinking remained statistically significant after adjustment for cholesterol, BP, waist girth, smoking habits and exercise [ relative risk (RR), 0.22; 95% CI, 0.05–0.99]. With CVD as the outcome, there were no statistically significant associations with drinking pattern or beverage preference.
Lifestyle variables associated with category of alcohol intake and beverage preference
Participants were categorized as lifetime abstainers, ex-drinkers, ‘safe’ drinkers who did not exceed guidelines for alcohol consumption and ‘at-risk’ drinkers who exceeded these guidelines. There was a significant between-group difference for waist girth, adjusted for gender (P = 0.002) with greatest mean values in the abstainers (93.1; 95% CI, 90.4–95.8) and ex-drinkers (94.6; 95% CI, 91.9–97.2) with a mean of 84.4 (95% CI, 78.2–90.5) in ‘safe drinkers’ and 89.8 (95% CI, 87.9–91.8) in ‘at-risk’ drinkers. BMI exceeded 22 kg/m2 in 89 abstainers (74%), 84 ex-drinkers (74%), 11 ‘safe’ drinkers (48%), and 143 ‘at-risk’ drinkers (60%) (P = 0.003). Among abstainers 11 (9%) played sport at least weekly, compared with 5 ex-drinkers (5%), 5 ‘safe’ drinkers (23%) and 36 ‘at-risk’ drinkers (16%) (P = 0.007). Current smokers comprised 14 abstainers (12%), 33 ex-drinkers (31%), 8 ‘safe’ drinkers (35%), and 127 ‘at-risk’ drinkers (60%) (P < 0.001).
Preference for wine or beer was not significantly related to waist girth (mean 89.3 cm; 95% CI, 84.3–94.3 in wine drinkers; 88.6 cm; 95% CI, 86.6–90.5 in beer drinkers; P = 0.925), in the prevalence of BMI >22 kg/m2 (59 versus 58%, respectively, P = 0.886), exercise at least once a week (13 versus 16%, P = 0.649), or smoking habits (63% current smokers versus 56%, P = 0.417). No dietary data were available at the time of these analyses.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONPOPULATION AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This analysis has confirmed the higher risk of both CHD and CVD in ex-drinkers while, for CHD, the pattern of risk is consistent with a J-shaped association with alcohol consumption, the lowest risk being in the small group of participants drinking at low-to-moderate levels. We also found a significantly lower risk of CHD in wine drinkers, independent of drinking pattern and the amount of alcohol consumed.
Our finding of significantly increased risk for both CHD and CVD in ex-drinkers is consistent with previous studies (Wannamathee and Shaper, 1997) and is likely to be explained, as suggested, by pre-existing illness; 31% of ex-drinkers in this Aboriginal cohort indicated that they stopped drinking for health reasons. Overall there was a pattern of increased risk for CHD associated with drinking at higher levels with the lowest risk in the group drinking at ‘safe’ levels (National Health and Medical Research Council, 2001), consistent with a J-shaped relationship as reported in other populations (Corrao et al., 2000). However, drinking behaviour in this Aboriginal population is very different from that in other populations in which such a relationship has been reported. Australian Aborigines are less likely to drink alcohol than non-Aboriginal Australians but among those who do drink alcohol, consumption at hazardous levels is common (McClennan and Madden, 2003). Among our cohort, there was a preponderance of drinkers who consumed very high levels of alcohol; an intake of >60 standard drinks per drinking day was recorded among a small group of extreme binge drinkers. Numbers were small in some categories of lower alcohol intake. For these reasons, comparison with other populations must be interpreted cautiously.
Not only does the pattern of alcohol intake among Aborigines differ from that in other populations where alcohol intake and risk of CHD has been examined, other lifestyle behaviours differ widely from those in non-Aboriginal Australians. Lifestyle diseases are common, with a rapidly increasing rate of obesity and diabetes, associated with replacement of a traditional diet by one high in fat and low in fibre, and lack of physical activity (McClennan and Madden, 2003). In addition, an estimated 51% of Aborigines over the age of 18 years are smokers compared with 24% of non-Aboriginal Australians (McClennan and Madden, 2003). Risk is increased further by the prevalence of individuals with multiple risk factors in the Aboriginal population (McClennan and Madden, 2003).
Lifestyle characteristics, rather than the effects of the amount of alcohol consumed, have been suggested to explain the lower risk of CHD among individuals with low-to-moderate alcohol intake (Herberth et al., 1988; Fuchs et al., 2004; Ruidavets et al., 2004), and may be confounders in population studies that examine associations between alcohol intake and CHD (Fuchs et al., 2004). In our study, those drinking at ‘safe’ levels had lower waist circumference, fewer were overweight, more participants played sport at least weekly and the prevalence of smoking was almost half that seen in ‘at-risk’ drinkers. However, the small number of participants in the ‘safe-drinking’ group limits the validity of such inferences.
A preference for wine has also been reported to be associated with lower risk of mortality, possibly explained by drinking pattern, constituents of wine other than alcohol, or lifestyle characteristics (Burke et al., 1995; Tjonneland et al., 1999; Klatsky et al., 2003; Ruidavets et al., 2004). In our Aboriginal cohort there were no statistically significant associations between lifestyle variables and a preference for wine but associations with other lifestyle factors such as diet cannot be excluded. Our observation of a lower risk of CHD in Aboriginal wine preferrers is unexplained.
Drinking pattern may also influence risk for CHD (Tolstrup et al., 2004) and a greater risk of CHD in binge drinkers has been reported (Murray et al., 2002). In our Aboriginal cohort drinking pattern did not significantly predict risk of CHD or CVD independent of amount of alcohol consumed. The inclusion of beverage preference in these models did not affect interpretation.
With CVD as the endpoint we found no indication of a J-shaped association with alcohol intake. Only ex-drinkers had a statistically significant increased risk of CVD. Klatsky et al. (1990) reported differences in associations between alcohol consumption and diagnoses included within CVD. In that study, alcohol intake was associated with higher risk of mortality from hypertension and stroke, but lower risk from CHD, while ex-drinkers had higher rates for CVD mortality, consistent with our findings.
Our study utilizes a unique dataset that links use of alcohol to CHD and CVD in the long-term among Aboriginal Australians. Although the length of follow-up is a strength of the study, this also leads to limitations. All data documenting lifestyle and alcohol intake were obtained up to 13 years before the present analysis. Misclassification arising from changes in alcohol use over that period could obscure relationships with endpoints. Errors in documenting alcohol intake, beverage preference, and patterns of drinking would also affect interpretation of models. While all data were obtained by interviewers experienced in communicating with Aboriginal Australians and reported alcohol intakes were validated with reference to sales of alcohol in the community and GGT activity, there remains the potential for misclassification. This may be relevant to the lack of a statistically significant association with drinking pattern in our analysis, for example. While the number of events is relatively small, no other set of data links long-term outcomes with detailed information about alcohol consumption in individual Australian Aborigines.
In this Aboriginal cohort, a group with greater risk of CHD and higher rates of hazardous drinking than non-Aboriginal Australians, risk for CHD was increased in ex-drinkers and showed a pattern consistent overall with a J-shaped association with alcohol intake. Associations between drinking within national guidelines and less adverse lifestyle choices suggest that these associations may be mediated by health-related behaviours, as suggested in other populations. However, the findings do not exclude a cardioprotective role for alcohol.
|
ACKNOWLEDGEMENTS |
|---|
This study was supported by an Australian Research Council Discovery Project Grant. We thank the staff of the Western Australian Data Linkage Unit, particularly Carol Garfield and Diana Rosman, for their contribution to this project. We also thank the Western Australian Department of Health for access to information on hospital admissions and the Registrar of Births, Deaths and Marriages for access to death registrations.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONPOPULATION AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Burke V., Puddey I. B., Beilin L. J. (1995) Mortality associated with wines, beers, and spirits. British Medical Journal 311:1166.[ISI][Medline]
Corrao G., Rubbiati L., Bagnardi V., et al. (2000) Alcohol and coronary heart disease: a meta-analysis. Addiction 95:1505–1523.[CrossRef][ISI][Medline]
Daniel M., Rowley K. G., McDermott R., O'Dea K. (2002) Diabetes and impaired glucose tolerance in Aboriginal Australians: prevalence and risk. Diabetes Research and Clinical Practice 57:23–33.[CrossRef][ISI][Medline]
Fuchs F. D., Chambless L. E., Folsom A. R., Eigenbrodt M. L., Duncan B. B., Gilbert A., Szklo M. (2004) Association between alcoholic beverage consumption and incidence of coronary heart disease in whites and blacks. American Journal of Epidemiology 160:466–474.
Gronbaek M. (2002) Alcohol, type of alcohol, and all-cause and coronary heart disease mortality. Annals of the New York Academy of Science 957:16–20.
Hart C. L., Smith D. G., Hole D. J., Hawthorne V. M. (1999) Alcohol consumption and mortality from all causes, coronary heart disease and stroke: results from a prospective cohort study of Scottish men with 21 years of follow-up. British Medical Journal 318:1725–1729.
Herberth B., Didelot-Barthemlemy L., Lemoine A., Le Devhat C. (1988) Dietary behaviour of French men according to alcohol drinking pattern. Journal of Studies on Alcohol 49:268–272.[ISI][Medline]
Holman C. D. J., Bass A. J., Rouse I. R., Hobbs M. S. T. (1999) Population-based linkage of health records in Western Australia: development of a health services research linked database. Australian and New Zealand Journal of Public Health 23:453–459.[ISI][Medline]
Hosmer D. W. and Lemeshow S. (2000) Applied Logistic Regression(John Wiley and Sons Inc, New York).
Hunter E., Hall W., Spargo R. (1991) The Distribution and Correlates of Alcohol Consumption in a Remote Aboriginal Population(Monograph No 12, National Drug and Alcohol Research Centre, Sydney).
Hunter E., Hall W., Spargo R. (1992) Patterns of alcohol consumption in the Kimberley Aboriginal population. Medical Journal of Australia 156:764–768.[ISI][Medline]
Klatsky A. L., Armstrong M. A., Friedman G. D. (1990) Risk of cardiovascular mortality in alcohol drinkers, ex-drinkers and non-drinkers. American Journal of Cardiology 66:1237–1242.[CrossRef][ISI][Medline]
Klatsky A. L., Friedman G. D., Armstrong M. A., Kipp H. (2003) Wine, liquor, beer, and mortality. American Journal of Epidemiology 158:585–595.
Marmot M. and Brunner E. (1991) Alcohol and cardiovascular disease: the status of the U-shaped curve. British Medical Journal 303:565–568.[ISI][Medline]
McClennan W. and Madden R. (2003) The health and welfare of Australia's Aboriginal and Torres Strait islander Peoples(Australian Bureau of Statistics, Canberra).
McClure M. (1993) Demonstration of deductive meta-analysis: ethanol intake and risk of myocardial infarction. Epidemiologic Reviews 15:328–351.
McElduff P. and Dobson A. (1997) How much alcohol and how often? Population based case–control study of alcohol consumption and risk of major coronary event. British Medical Journal 314:1159–1164.
Ministerial Council on Drug Strategy. (2001) Alcohol in Australia; Issues and Strategies(Commonwealth of Australia, Canberra).
Murray R. P., Connett J. E., Tyas S. L., et al. (2002) Alcohol volume, drinking pattern, and cardiovascular disease morbidity and mortality: is there a U-shaped function? American Journal of Epidemiology 155:242–8.
National Centre for Classification in Health. (2000) The International Statistical Classification of Diseases and Related Health Problems 10th revision Australian Modification (ICD-9-AM).
National Coding Centre. (1995) The Australian Version of the International Classification of Diseases, 9th revision Clinical Modification (ICD-9-CM) First edn, National Coding Centre.
National Health and Medical Research Council. (2000) Nutrition in Aboriginal and Torres Strait Islander Peoples(National Health and Medical Research Council, Canberra).
National Health and Medical Research Council. (2001) Australian Alcohol Guidelines: Health Risks and Benefits(National Health and Medical Research Council, Canberra).
Ruidavets J.-B., Bataille V., Dallongeville J., et al. (2004) Alcohol intake and diet in France, the prominent role of lifestyle. European Heart Journal 25:1153–1162.
Smith R. M., Spargo R. M., Hunter E. M., et al. (1992a) Prevalence of hypertension in Kimberley aborigines and its relationship to ischaemic heart disease. An age-stratified random survey. Medical Journal of Australia 156:557–562.[ISI][Medline]
Smith R. M., Spargo R. M., King R. A., et al. (1992b) Risk factors for hypertension in Kimberley aborigines. Medical Journal of Australia 156:562–566.[ISI][Medline]
Tjonneland A., Gronbaek M., Stripp C., Overvad K. (1999) Wine intake and diet in a random sample of 48763 men and women. American Journal of Clinical Nutrition 69:49–54.
Tolstrup J. S., Jensen M. K., Tjonneland A., et al. (2004) Drinking pattern and mortality in middle-aged men and women. Addiction 99:323–330.[CrossRef][ISI][Medline]
Wannamathee S. G. and Shaper A. G. (1997) Lifelong teetotallers, ex-drinkers and drinkers: mortality and the incidence of coronary heart disease events in milddle0aged British men. International Journal of Epidemiology 26:523–531.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||