1 Family Medicine Stockholm, Karolinska Institute, Huddinge and 2 Department of Clinical Medicine, Family Medicine Research Centre, Örebro University, Örebro, Sweden
(Received 5 October 2002; first review notified 3 January 2003; in revised form 23 June 2003; accepted 9 July 2003)
* Author to whom correspondence should be addressed at: Family Medicine Stockholm, Karolinska Institutet, Alfred Nobels alle 12, S-141 83 Huddinge, Sweden. E-mail: holger.theobald{at}klinvet.ki.se
![]() |
ABSTRACT |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
![]() |
INTRODUCTION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Self-reported, subjective health has been shown to be a strong predictor of both coronary heart disease and all-cause mortality (Mossey and Shapiro, 1982; Kaplan and Camacho, 1983
; Idler and Angel, 1990
). It has recently been shown that there is a J-shaped association between self-reported health status and alcohol consumption (Poikolainen et al., 1996
). That is, individuals with the best self-reported health status were moderate consumers; abstainers had a self-reported health status between those of moderate and high consumers; and high consumers had the lowest health status. In two recent studies (Grønbaek et al., 1999
; Poikolainen and Vartiainen, 1999
), it was found that, after controlling for different confounders, a moderate intake of wine was related to a good, self-perceived health. For beer and spirits, no such association was found. In both studies, the variables were obtained from self-administered questionnaires.
The aim of the present study was to investigate the influences of different types of alcoholic beverages on the self-reported health status in a representative sample of the Swedish population. In contrast to the above-mentioned studies, the data were collected at face-to-face interviews; the response rate was higher and the effect of non-response was also considered.
![]() |
SUBJECTS AND METHODS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Outcome variable
Information about the dependent variable, the self-reported health status, was given by the individuals in answer to the question How would you describe your general health? There were five possible responses: very good, good, moderate, bad or very bad. Those who answered that their health status was very good or good were considered to have good, self-reported health status; the others were considered to have poor, self-reported health status.
Independent variables
Age was analysed according to the age groups 1624, 2534, 3544, 4554, 5564, 6574 and 7584 years. The questionnaire consisted of several items, some of which were used in the different statistical analyses. The educational attainment was classified in three groups: elementary school (9 years), completed up to 2 years of high school (1011 years) and completed more than 2 years of high school or university studies (>11 years). Body mass index (BMI) was classified into four groups: underweight (BMI
18.5), normal weight (>18.5 but
25), overweight (>25 but
30) and obesity (>30). Marital status comprised two groups: single and married/cohabiting. The smoking habits were categorized into three groups: current smokers, former smokers and never a smoker. The level of leisure-time, physical activity was dichotomized in the analysis into being physically inactive or occasionally active versus regular physical activity at least once a week.
Alcohol consumption was assessed by questions about the extent of use: the amount and frequency of intake of ordinary beer, strong beer, red or white wine, fortified wine and hard liquor during the last month. Thus, for each of the alcoholic beverages mentioned in our study, the individuals were asked to tell the quantity they consumed each day of the week in the form of glasses, bottles, cans or, if they could, specify the volume of the beverage consumed (for example, the phrasing could be as follows: How much fortified wine did you drink last Friday, how many glasses? How many bottles?). They were also asked to specify the quantity they consumed each time they consumed alcoholic beverages. The individuals who claimed that their intake of alcohol during the last month differed from their regular intake were asked about that and the answers were used for the calculations. A few individuals (less than 100) refused to answer questions concerning alcohol intake and the questions about general health. Thus, the small differences in the number of individuals in some of the calculations reflect missing data. By multiplying the frequency of intake by the amount of alcohol in grams for each beverage, the alcohol consumption in g/week for each beverage could be calculated. We also analysed the frequency of drinking different beverages, that is for each alcoholic beverage we analysed the number of individuals who consumed the specified beverage during different specified time periods. We decided to categorize the individuals into four groups according to alcohol intake and the influence of that intake on self-reported health status. Individuals who consumed less than 150 g/week constituted the reference group. The cut-off limit for high alcohol consumption, 150 g/week, was chosen in accordance with the recommendations made by the Swedish National Food Administration (Sandström et al., 1996) as 150 g alcohol per week (or 21 g/day) is considered to be the upper level for consumption which has no harmful effects on the individuals health. We do not think that it is meaningful to have a lower or more narrow reference category than 1149 g as negative effects on health can probably only be expected with higher consumption. Further, we also found that a narrower reference category did not change the results. As most participants consumed several types of beverage, an analysis of an eventual interaction between the different alcoholic beverages was performed. This was done through inclusion of a new variable in the logistic regression model. This variable was the product of the two alcoholic beverage type variables which we tested. We did not find any significant interactions of combined drinking (data not shown). Thus, we believe it correct to use our statistical model to correct for the use of multiple beverage types in order to estimate the effect of one single type of beverage.
The study was approved by the local Ethics Committee at Huddinge University Hospital.
Statistical analysis
Using logistic regression, crude and adjusted OR with 95% CI were estimated for the relationship between exposure (intake of wine, beer and spirits) and poor, self-reported health status. The analyses were performed by the SAS data package, PROC LOGISTIC (SAS Institute, Cary, NC, USA).
![]() |
RESULTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
|
|
Wine and beer were consumed more often than were the other beverages, as shown by the weekly consumption (Table 4). The relationship between intake of wine and poor self-reported health status showed the following OR (adjusted for age, sex and total alcohol consumption): 1.74 (CI 1.551.96) for those consuming 0 g/week, 1.00 (reference) for those consuming 1149 g/week, 1.37 (CI 0.742.54) for those consuming 150289 g/week and 0.61 (CI 0.066.01) for those consuming >=290 g/week.
|
|
![]() |
DISCUSSION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
It could be argued that some objective measure of physical health instead of self-reported health status could have strengthened the results. In the SALLS there are no such objective measures of physical health. However, as self-reported health status has been shown to be a strong predictor of both coronary heart disease and all-cause mortality (Mossey and Shapiro, 1982; Kaplan and Camacho, 1983
; Idler and Angel, 1990
, Sundquist and Johansson, 1997
), we believe it appropriate to use it in our study.
In our study the response rate was high (80%). A drop-out analysis, in another part of the SALLS, has shown that the majority of the non-participating individuals have similar, relative, death rates as do the participants and thus should have similar, self-reported health status, as this parameter mirrors mortality. Therefore it is possible that our results could be generalized to the whole population. We also found that poor, self-reported health status was associated with increased age, low educational level, high BMI, smoking and low physical activity. These results are in accordance with the results of other studies (Grønbaek et al., 1999; Johansson and Sundquist, 1999
), which suggests that our results are reliable. Furthermore, it has been shown in some studies that more reliable results concerning the alcohol intake are obtained if individuals are asked about their intakes over longer periods of time (O'Hare, 1991
; Duffy and Alanko, 1992
). In our study the alcohol intake was assessed by asking about the intake in the last month, and individuals who claimed that their intake of alcohol during the last month differed from their regular intake were asked about their usual intake. This may give better estimates of alcohol intake than did the study by Grønbaeck et al. (1999)
, in which the intake was based on the last, non-weekend day.
Our results were adjusted for total alcohol consumption, age and sex. Adjusting the results for other confounders, such as smoking, educational level, BMI or level of physical, activity did not change the results. This is in accordance with results from other studies (Grønbaek et al., 1999; Poikolainen and Vartiainen, 1999
). However, in those studies the results were also adjusted for confounders such as chronic disease, presence of social networks and marital status. Those confounders, also, did not change the results. One important lifestyle factor which we, as well as others, could not adjust for was dietary habits; we have no data on this. It has been shown that drinkers of alcoholic beverages who have increasing alcohol consumption tend to consume diets higher in fat and lower in dietary fibre. Adjusting for those variables did not change the effect of alcohol on total, coronary artery disease (Stampfer et al.,1988
; Rimm et al., 1991
). It has also been shown that wine consumers have more healthy dietary habits than do consumers of other beverages (Tjønneland et al., 1999
). However, there are no data on the consequences of that for cardiovascular diseases or self-reported health status.
Many individuals in our study consumed several types of beverage; which is the usual drinking pattern. To analyse the effects of a single type of beverage, we used a statistical model for the adjustment of multiple beverage drinking. There were no interactions of combined drinking, but the use of a statistical model for correction of multiple beverage drinking obviously does not give as reliable data as that obtained from individuals drinking only one type of alcoholic beverage. However, the latter situation is very uncommon in normal everyday life.
In contrast to the two other studies dealing with self-reported health status and the effects of different alcoholic beverages (Grønbaek et al., 1999; Poikolainen and Vartiainen, 1999
), we found a beneficial effect of wine but not of fortified wine. In the other studies, the different types of alcoholic beverage were not subgrouped, as in our study, and it is possible that fortified wine could also have been without effect in those studies, if it had been analysed. The mechanisms behind the positive effects of wine on self-reported health status have not been identified. It has been suggested that the contents of anti-oxidants such as flavonoids and phenolic compounds in wine may have cardioprotective effects (Das et al., 1999
). However, it is not known whether these substances also have positive effects on the self-reported health status, but theoretically it is possible. If there is an effect, it is unlikely that a common substance in wine and fortified wine has favourable effects. However, it has been shown that the content of polyphenols differs in different types of wines, and that the bio-availability varies (Scalbert and Williamson, 2000
), so it is still possible that some substance in wine may have positive effects.
The U-shaped association between self-reported health status and amount of wine consumed does not support a clear-cut, pharmacological effect of some substance in wine. If a pharmacological effect is to be expected from a substance, an optimal dosage and a regular intake of the substance should produce the effect. In our study only a few individuals consumed alcoholic beverages regularly, and beer, which had no effect, was consumed more often than the other beverages. This also argues against the pharmacological action of wine.
There is some evidence that moderate wine drinkers have a healthier lifestyle than those who prefer other alcoholic beverages (Klatsky et al., 1990; Klatsky and Armstrong, 1993
). It is likely that lifestyle factors are important constituents of the favourable effects of wine on self-reported health status. The factors reflecting lifestyle that we examined were not found to be confounders of the positive effect of wine on self-reported health status. Dietary habits are probably of importance, but they could not be analysed as we lacked the data. When we examined the occurrence of smoking and physical activity (important lifestyle and risk factors for ill-health), it was found that only non-users of alcoholic beverages and users of fortified wine differed from the other beverage users, with regard to the percentage of individuals who were smokers and engaged in physical activity. Our study was not designed to evaluate the importance of lifestyle factors in connection with the consumption of different beverages for self-reported health status. Thus it is possible that some of the lifestyle factors that we analysed and other, less important ones, may have synergistic effects on self-reported health status; this may explain the positive effects of wine. Studies designed to resolve these issues are needed.
In conclusion, we found that the moderate consumption of wine is associated with positive effects on self-reported health status. Constituents of the beverage that have favourable effects and/or lifestyle factors are potential, underlying causes.
![]() |
Acknowledgements |
---|
![]() |
REFERENCES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Duffy J. and Alanko T. (1992) Self-reported consumption measures in sample surveys: a simulation study of alcohol consumption. Journal of Official Statistics 8, 327350.
Grønbaek M., Mortensen E. L., Mygind K., Andersen A. T., Becker U., Gluud C. and Sorensen T. I. (1999) Beer, wine, spirits and subjective health. Journal of Epidemiology and Community Health 53, 721729.[Abstract]
Idler E. L. and Angel R. J. (1990) Self-rated health and mortality in the NHANES-I: Epidemiologic Follow-up Study. American Journal of Public Health 80, 446452.[Abstract]
Johansson S. E. and Sundquist J. (1999) Change in lifestyle factors and their influence on health status and all-cause mortality. International Journal of Epidemiology 28, 10731080.[Abstract]
Kaplan G. A. and Camacho T. (1983) Perceived health and mortality: a nine-year follow-up of the human population laboratory cohort. American Journal of Epidemiology 117, 292304.[Abstract]
Klatsky A. L. and Armstrong M. A. (1993) Alcoholic beverage choice and risk of coronary artery disease mortality: Do red-wine drinkers fare best? American Journal of Cardiology 71, 467469.[ISI][Medline]
Klatsky A. L., Armstrong M. A. and Kipp H. (1990) Correlates of alcoholic beverage preference: Traits of persons who choose wine, liquor or beer. British Journal of Addiction 85, 12791289.[ISI][Medline]
Mossey J. M. and Shapiro E. (1982) Self-rated health: A predictor of mortality among the elderly. American Journal of Public Health 72, 800808.[Abstract]
O'Hare T. (1991) Measuring alcohol consumption: A comparison of the retrospective diary and the quantity-frequency methods in a college drinking survey. Journal of Studies on Alcohol 52, 500502.[ISI][Medline]
Poikolainen K. (1995) Alcohol and mortality: A review. Journal of Clinical Epidemiology 48, 455465.[CrossRef][ISI][Medline]
Poikolainen K. and Vartiainen E. (1999) Wine and good subjective health. American Journal of Epidemiology 150, 4750.[Abstract]
Poikolainen K., Vartiainen E. and Korhonen H. J. (1996) Alcohol intake and subjective health. American Journal of Epidemiology 144, 346350.[Abstract]
Rimm E. B., Giovannucci E. L., Willnett W. C., Colditz G. A., Ascherio A., Rosner B. and Stampfer M. J. (1991) Prospective study of alcohol consumption and risk of coronary disease in men. Lancet 338, 464468.[ISI][Medline]
Sandström B. M., Lyhne N., Pedersen J. I., Aro A., Thorsdottir I. and Becker W. (1996) Nordic Nutrition Recommendations 1996; Scandinavian Journal of Nutrition 40, 161165.
Scalbert A. and Williamson G. (2000) Dietary intake and bioavailability of polyphenols. Journal of Nutrition 130, 2073S2085S.
Stampfer M. J., Colditz G. A., Willett W. C., Speizer F. E. and Hennekens C. H. (1988) A prospective study of moderate alcohol consumption and the risk of coronary disease and stroke in women. New England Journal of Medicine 319, 267273.[Abstract]
Statistics Sweden (1996) The Swedish Survey of Living Conditions. Design and Method. Appendix no. 16. Statistics Sweden, Stockholm.
Sundquist J. and Johansson S. E. (1997) Self reported poor health and low educational level predictor for mortality: a population based follow up study of 39,156 people in Sweden. Journal of Epidemiology and Community Health 51, 3540.[Abstract]
Theobald H., Bygren L. O., Carstensen J. and Engfeldt P. (2000) A moderate intake of wine is associated with reduced total mortality and reduced mortality from cardiovascular disease. Journal of Studies on Alcohol 61, 652656.[ISI][Medline]
Tjønneland A., Grønbeck M., Stripp C. and Overvad K. (1999) Wine intake and diet in a random sample of 48 763 Danish men and women. American Journal of Clinical Nutrition 69, 4954.
Wärneryd B. (1990) Levnadsförhållanden. Återintervjustudie i undersökningen av levnadsförhållanden (ULF) 1989 [Living conditions. Re-interview in ULF 1989] [in Swedish], Appendix no. 12. Statistics Sweden, Stockholm.