European Journal of Clinical (1997) 51, 326±332 ß 1997 Stockton Press. All rights reserved 0954±3007/97 $12.00

Alcohol beverage drinking, and body mass index in a cross-sectional survey

SMaÈnnistoÈ1, K Uusitalo1, E Roos1, M Fogelholm2 and P Pietinen1

1National Public Health Institute, Department of Nutrition, Helsinki, Finland; and 2the UKK Institute, Tampere, Finland

Objective: The study was carried out to determine the associations of beverage drinking with macronutrients, antioxidants, and body mass index. Setting: Dietary subsample of the 1992 Finmonica cardiovascular risk factor survey in Finland; a cross-sectional study. Subjects: 985 women and 863 men were drawn from the population register in the four monitoring areas. All subjects were 25±64 y of age. Methods: The mailed questionnaire included questions covering socioeconomic factors, physical activity, smoking, and alcohol consumption. The diet was assessed using a three-day food record. Results: The dietary differences between abstainers and alcohol consumers were more signi®cant than between consumers of different alcoholic beverages. Among drinkers, intake as a percentage of energy was higher and carbohydrate intake was lower than among abstainers. Those who preferred wine, however, had the highest vitamin C intake; female wine drinkers also had the highest carotenoid intake. With the exception of those who mainly preferred spirits, alcohol energy was not added to the diet but seemed to substitute food items both in men and women. Despite the similar total daily energy intakes, daily energy expenditure, and physical activity index, male drinkers were leaner than abstainers. In women, the proportion of underreporters of energy intake increased with increasing alcohol consumption, and the association between alcohol and body mass index was similar to that in men after the exclusion of underreporters. Conclusions: Alcohol consumers were leaner than abstainers, and wine drinkers in particular had more antioxidants in their diet. Sponsorship: The Finnish Foundation for Alcohol Studies. Descriptors: alcoholic beverages; alcohol intake; body mass index; diet; energy

Introduction patterns of alcohol consumption (Simpura et al, 1995) have changed markedly during the last twenty years, Finland is The U-shaped association between alcohol consumption still exceptional among Western countries. This was sup- and coronary heart disease in both men and women is quite ported by the ®ndings of Artaud-Wild et al (1993), who well established (Doll et al, 1994; Fuchs et al, 1995). The concluded that the differences in cholesterol and saturated bene®cial effects of alcohol have been associated with fat intakes explained the differences in coronary mortality higher HDL/LDL cholesterol ratios and with the inhibition in 40 countries but not in Finland and France. Further, of platelet aggregation. Moreover, the descending part of because drinking at meals is not common in Finland, the the U-shaped curve may be the result of a higher number of mean number of drinking occasions per week remains far wine drinkers, and the ascending part a result of heavy below most other European countries and is spirit-drinkers (Grùnbñk et al, 1995). In other words, common (Simpura et al, 1995). different effects of alcoholic beverages are possibly not There are also other lifestyle aspects related to alcohol caused by the same main component () but by other consumption. In some studies but not in all, alcohol con- characteristic components (Dorfman et al, 1985). sumers had lower body mass index (BMI) than abstainers, The `French Paradox' describes the phenomenon despite their higher total energy intakes as reviewed by observed in France, where a high intake of Hellerstedt et al (1990) and Prentice (1995). Alcohol has not led to high mortality from coronary heart disease. consumers also smoke more than abstainers (Le Marchand This ®nding has directed attention toward the possible et al, 1988). bene®cial effects of red wine (Renaud & de Lorgeril, We have previously found that middle-aged men did not 1992). Protective components in wine are not well estab- substitute alcohol for food but added alcoholic beverages to lished, but some evidence of the positive effects of ¯avo- the diet, and that alcohol consumption in¯uenced daily noids and phenolics have been presented (Kinsella et al, nutrient intakes only slightly (MaÈnnistoÈ et al, 1996). The 1993). Although the Finnish diet (Pietinen et al, 1996) and energy from alcohol increased body weight less than expected, especially in daily alcohol users. However, although that study of 27 215 men was substantial, it was Correspondence: S MaÈnnistoÈ, National Public Health Institute, Department restricted to smoking men who were mainly spirit drinkers. of Nutrition, Mannerheimintie 166, 00300 Helsinki, Finland. Recieved 16 August 1996; revised 11 December 1996; accepted 24 Thus women, young people, and those who mainly drink January 1997 wine or beer were excluded. Alcohol, diet and BMI S MaÈnnistoÈ et al 327 In this study we assessed whether selected indicators of this paper the term `alcohol consumption' refers to absolute lifestyle (diet, obesity, smoking, physical activity) were alcohol intake, and alcohol consumers are all those subjects related to the type of consumed alcohol among 25±64 y old who consumed alcohol, regardless of beverage type. adults. Because recent dietary studies have suggested that All physical activities, including activity at work (8 h), daily energy intake is often under-reported compared to on the way from home to work (self-reported time), during estimated energy needs computed as a multiple of basal leisure time (self-reported time), during sleep (8 h), and metabolic rate (BMR) (Black et al, 1991), the results both during the rest of the day, were assigned an intensity unit include and exclude underreporters of energy intake. (MET) based on the rate of energy expenditure of each activity (Ainsworth et al, 1993). One MET is de®ned as the energy expenditure for sitting quietly for one hour. The Subjects and methods products of duration (h) and intensity (MET) were totaled The subjects were participants in the dietary subsample of and divided by 24 h to obtain an average daily MET value, the 1992 Finmonica risk factor survey, which followed the that is, the daily physical activity index (Fogelholm et al, WHO Monica protocol (WHO, 1988). A sample including 1994). The estimation of daily energy expenditure was subjects aged 25±64 y, strati®ed by age and sex, were calculated by multiplying the physical activity index with drawn from the national population register in the four BMR. We used FAO/WHO formulae to calculate the BMR Finmonica monitoring areas: the provinces of North Kar- for each individual (FAO/WHO/UNU, 1985). Body mass elia and Kuopio in eastern Finland, the city of Turku and its index (BMI), weight (kg) divided by the square of height rural municipalities in southwest Finland, and the southern (m2), was used as the measure of adiposity, and the waist- cities of Helsinki and Vantaa. All subjects born between the to-hip ratio was used as the measure of adiposity distribu- 1st and 11th day of each month were chosen for the tion. Waist circumference was measured midway between subsample of the dietary survey. Of the eligible subjects the lower rib margin and iliac crest, and hip circumference (2822), 76% came to the local health center, 1861 of whom was measured at the widest circumference over the greater returned an acceptable three-day food record. The records trochanters. were checked by a nutritionist. Those records which had The ratio of reported energy intake to estimated BMR less than three days or inadequate reporting of foods was calculated for all subjects. Those whose ratio was consumed were excluded. Thus, the ®nal response rate in below 1.28, according to the recommendation of FAO/ the dietary survey was 61% of men and 71% of women WHO/UNU (1985) were regarded as having implausibly (Kleemola et al, 1994). Since six women and eight men measured energy intakes for free-living persons, and were were excluded because of missing data on alcohol con- classi®ed as energy underreporters. The detailed analysis of sumption, the ®nal data of this study included 985 women underreporting based on this criteria has been reported and 862 men. Approval for the Finmonica risk factor elsewhere (Hirvonen et al, 1997). survey was given by the Ethics Committee of National The adjusted mean nutrient intakes were obtained by the Public Health Institute. analysis of covariance. The background variables used in The subjects ®lled in a self-administered questionnaire the models were age, area, smoking, education, energy covering questions on health, socioeconomic factors, phy- from food, and alcohol intake. Log-transformation and sical activity, smoking, and alcohol intake. The ®rst alcohol geometric means were used as the values for vitamin C question dealt with alcohol consumption during the pre- and carotenoids since their distributions were skewed to the vious 12 months. The following questions concerned the right. frequency and the average portion of beer, wine and spirits. The consumption of red and white wine was not differ- Results entiated. The subjects were divided into four alcohol consumption categories by ethanol intake (0 g, 0±100 g, Alcohol consumption and subject characteristics 100±280 g and 280 g/week for men, and 0 g, 0±50 g, 50± Of the men, 8% were classi®ed as abstainers and 8% as 100 g and 100 g/week for women). Further, those subjects heavy drinkers, while the percentages for women were 18% who reported that on average at least 50% of their alcohol and 3%, respectively. The mean weekly alcohol consump- intake was beer, wine, or spirits were classi®ed as beer, tion was 101 g (standard error 5.2 g) among men and 25 g wine or spirit consumers, respectively. The remaining (1.7 g) among women (Table 1). Alcohol consumption had subjects were categorized as mixed drinkers. a highly skewed distribution; the median of alcohol intake A three-day food record was used to assess each sub- was much lower than the mean consumption, and the ject's diet and alcohol intake over the following three days subjects in the highest alcohol decile consumed about after his or her visit to a health center. The discussion on half of all reported alcohol (45% for men and 57% the adequate amount of food record days (Thompson & for women). Men consumed four ®fths of the reported Byers, 1994) and the history of the three-day food record in alcohol. this study have been reported elsewhere (Roos et al, 1995). Alcohol consumption was signi®cantly related to age, A blank diary was used for recording the dietary and living area, and smoking among both men and women. alcohol intake, and a 63-page picture booklet was used to Younger people consumed more alcohol than the elderly; assess portion sizes. Individual daily energy and nutrient subjects living in the capital area consumed the most intakes were computed using the food composition data- alcohol; and regular smokers consumed more than twice base of the National Public Health Institute. The nutrient as much alcohol as non-smokers. In addition, also women content of various alcoholic beverages were included in the with more education than six years (62%) consumed twice daily nutrient intake, but alcohol was excluded when the as much alcohol (30 g/d) as others (14 g/d) after adjusting percentage contribution of , carbohydrate, protein, and for age, area, physical activity, smoking, BMI and marital sugar to total energy was calculated. To describe an status (P ˆ 0.0003). addition or a substitution of alcohol to the diet, we used The number of men was the greatest in the beer- and the alcohol consumption indicated by the food record. In spirit-drinking categories, while the majority of women Alcohol, diet and BMI S MaÈnnistoÈ et al 328 Table 1 Characteristics of alcohol consumption in men and women according to the questionnaire

Men Women (n ˆ 862) (n ˆ 985)

Abstainers, % 8 18 Heavy drinkers, %a 8 3 Mean consumption, g/week 101 (5.2)b 25 (1.7)b Median consumption, g/week 52 7 Proportion (%) of total consumption in the highest alcohol decile 45 57 Proportion (%) of those who consumed less than 1 drinkc per week 22 59 Proportion (%) of total consumption by sex 78 22

a At least 280 g/week for men and 190 g/week for women. b Standard error. c One alcohol portion contains about 12 g alcohol.

were beer or wine drinkers (Table 2). Abstainers were intake of men was higher in the questionnaire than in older, had fewer years of formal education, and they the three-day food record, with the exception of mixed smoked less than alcohol consumers. Male abstainers had drinkers. Women, however, reported lower alcohol con- the highest BMI. When the data was analyzed by multiple sumption in the questionnaire compared to the food record. regression (including age, physical activity, education, The underreporting of energy intake increased with smoking, energy from food and alcohol intake), alcohol alcohol consumption in women but not in men. The had still explanatory power on BMI in women (P ˆ 0.05) proportion of underreporters was 40% among female but not in men (P ˆ 0.33); R2 of the total model was 0.22 abstainers but increased to 57% among those who con- both in men and women. The proportions of current sumed at least 100 g of alcohol per week. Among men the smokers were the highest among male beer and spirit proportions were from 42±46% in all alcohol categories. As drinkers, while female beer drinkers clearly smoked the Table 2 shows, male wine drinkers were more likely to most. The differences in the waist-to-hip ratio between underreport their energy intake than those who preferred categories were statistically signi®cant, but not practically other alcoholic beverages. Among women, the proportions relevant. The physical activity index and daily energy were quite similar between the type of alcoholic beverages. expenditures were similar among men in all groups, The energy underreporters consumed more alcohol (108 whereas female spirit drinkers had the highest values and g/week for men and 27 g for women), and they had a wine drinkers had the lowest values. The reported alcohol higher BMI than non-underreporters.

Table 2 Characteristics of subjects by alcohol beverage drinking patterns (mean and standard error)

Alcohol consumers

Abstainers Beer Wine Spirits Mixed P-value

Men N 70 367 75 228 122 Ð Agea 51 (1.3) 43 (0.6) 47 (1.3) 49 (0.7) 45 (1.0) 0.0001 Years of educationb 10 (0.4) 11 (0.2) 13 (0.4) 11 (0.2) 12 (0.3) 0.0001 Current smokers, % 14 36 24 35 31 Ð Energy underreporters, %c 44 40 57 44 44 Ð BMI, kg/m2 28.0 (0.4) 26.1 (0.2) 26.5 (0.4) 26.9 (0.2) 26.1 (0.3) 0.001 Excluding underreportersc 26.9 (0.6) 25.4 (0.2) 25.7 (0.6) 26.4 (0.3) 25.3 (0.4) 0.02 Waist-to-hip ratiob 0.93 (0.008) 0.92 (0.004) 0.92 (0.008) 0.93 (0.005) 0.90 (0.006) 0.007 Physical activity indexb 1.65 (0.03) 1.68 (0.01) 1.65 (0.03) 1.67 (0.02) 1.65 (0.02) 0.89 Daily energy expenditureb, kcald 3005 (69) 3035 (30) 3025 (67) 3086 (39) 3019 (52) 0.75 Excluding underreportersc 3033 (95) 3048 (39) 3094 (101) 3103 (53) 2934 (71) 0.41 Alcohol intake from the food record, g/week 0 122 (9.0) 103 (16.4) 91 (10.8) 94 (13.5) Ð From questionnaire 0 125 (8.7) 114 (18.0) 105 (10.5) 71 (8.8) Ð

Women N 174 248 255 162 146 Ð Agea 51 (0.9) 40 (0.7) 47 (0.7) 48 (0.9) 44 (0.9) 0.0001 Years of educationb 11 (0.2) 12 (0.2) 13 (0.2) 11 (0.2) 12 (0.3) 0.0001 Current smokers, % 5 32 20 19 16 Ð Energy underreporters, %c 40 45 50 49 49 Ð BMI, kg/m2 26.6 (0.4) 25.1 (0.3) 24.9 (0.3) 26.7 (0.4) 25.3 (0.4) 0.0001 Excluding underreporters 26.1 (0.4) 24.0 (0.4) 24.1 (0.4) 25.7 (0.5) 24.5 (0.5) 0.0005 Waist-to-hip ratiob 0.80 (0.005) 0.79 (0.004) 0.78 (0.004) 0.79 (0.005) 0.78 (0.006) 0.008 Physical activity index 1.59 (0.02) 1.59 (0.01) 1.56 (0.01) 1.65 (0.02) 1.60 (0.02) 0.002 Daily energy expenditureb, kcald 2247 (29) 2229 (24) 2181 (23) 2362 (29) 2267 (30) 0.0001 Excluding underreportersc 2278 (37) 2215 (33) 2169 (32) 2400 (41) 2274 (43) 0.0003 Alcohol intake from the food record, g/week 0 54 (5.0) 31 (3.9) 26 (6.6) 30 (5.1) Ð From questionnaire 0 42 (3.5) 33 (4.2) 17 (4.3) 22 (2.4) Ð

a Area adjusted. b Age and area adjusted. c The ratio of reported energy intake to estimated BMR under 1.28 (FAO/WHO/UNU, 1985). d 1 kcal ˆ 4.2 kJ. Alcohol, diet and BMI S MaÈnnistoÈ et al 329

Figure 1 Energy intake from food (white area) and alcohol (shaded Figure 3 Energy intake from food (white area) and alcohol (shaded area), adjusted to age and area, by alcohol consumption among men area), adjusted to age and area, by alcohol beverage drinking among men including underreporters. 1 kcal ˆ 4.2 kJ. including underreporters. 1 kcal ˆ 4.2 kJ.

Figure 2 Energy intake from food (white area) and alcohol (shaded Figure 4 Energy intake from food (white area) and alcohol (shaded area), adjusted to age and area, by alcohol consumption among women area), adjusted to age and area, by alcohol beverage drinking among including underreporters. 1 kcal ˆ 4.2 kJ. women including underreporters. 1 kcal ˆ 4.2 kJ.

Associations between alcohol drinking and diet consumers. The intake of vitamin C (P ˆ 0.02) was also Among men, energy from food decreased with increasing lower in male abstainers. Among women, the percentage of alcohol consumption (g/week), but there were no noticeable energy from fat (P ˆ 0.02) was lower in abstainers, while differences in the total energy intakes between abstainers the intake of carotenoids was the highest in wine drinkers and alcohol consumers, with (P ˆ 0.66) or without and the lowest in beer-drinkers (P < 0.01). Only small (P ˆ 0.44) energy underreporters. This suggests that men differences were found in dietary results when the energy substituted alcohol for food (Figure 1). Among women, underreporters were excluded; the differences between the alcohol consumers had signi®cantly lower total daily percentages of energy from fat (P ˆ 0.09) and from protein energy intake (P ˆ 0.001) and energy intake from food (P ˆ 0.16) leveled-off in men. than abstainers (Figure 2). After the exclusion of under- reporters, however, the total daily energy intakes between Discussion the categories remained similar (P ˆ 0.27). The association between energy intake and alcohol Since the 1960s Finland has changed from a spirit-drinking consumption was the same when alcohol consumers were to a beer-drinking country. In particular, young and well- divided into consumers of different alcoholic beverages. educated people have adopted new consumption patterns Men and women both seemed to substitute alcohol for food, favouring lighter alcoholic beverages (Simpura et al, 1995). except male spirit drinkers (Figures 3 and 4). Among men In 1992 the share of lighter beverages was 70% of the total there were no statistical differences in total daily energy alcohol consumption, and the total alcohol consumption intakes between categories (P ˆ 0.67), but differences sur- was 7.2 litres per capita, calculated as ethanol (The Finnish faced among women (P ˆ 0.01). When the energy under- State Alcohol Company, 1993). The consumption not reporters were excluded, all categories, substituted alcohol included in the of®cial statistics was estimated to be for food, except female spirit drinkers. about 20% of the total alcohol consumption. About 10% The differences in nutrient intakes were larger between of men and 18% of women are abstainers (Simpura et al, abstainers and alcohol consumers than between consumers 1995). In our study the proportion of abstainers was 8% of of different alcoholic beverages (Table 3). Among men, the men and 18% of women. Men and women consumed on percentages of energy from fat (P ˆ 0.04) and protein average 5.7 litres and 1.3 litres of absolute alcohol per year, (P ˆ 0.007) were lower, and energy from carbohydrates respectively. Surveys generally generate low estimates were (P ˆ 0.002) higher among abstainers than alcohol (usually less than a half) of alcohol consumption compared Alcohol, diet and BMI S MaÈnnistoÈ et al

330 Table 3 Adjusted mean nutrient intakesa of women and men by alcohol beverage drinking

Alcohol consumers

Abstainers Beer Wine Spirits Mixed P-value (all)

Men Fat, E%b 33.3 35.3 35.0 36.0 35.4 0.04 Saturated fat, E%b 15.2 15.8 15.8 16.2 15.7 0.37 Protein, E%b 16.8 17.1 17.9 17.9 17.1 0.007 Carbohydrate, E%b 50.4 48.3 48.2 46.7 48.4 0.002 Sugar, E%b 10.5 9.7 10.2 9.8 10.3 0.59 Total carotenoids, mgc 3080 3325 3452 3302 3636 0.64 Vitamin C, mgc 88 112 134 115 113 0.02 Women Fat, E%b 33.5 34.8 34.3 35.4 35.0 0.02 Saturated fat, E%b 15.2 15.6 15.3 15.8 15.6 0.39 Protein, E%b 16.4 16.4 16.6 16.5 16.3 0.90 Carbohydrate, E%b 50.1 48.9 49.1 48.1 48.7 0.09 Sugar, E%b 10.5 11.2 11.0 10.9 11.2 0.65 Total carotenoids, mgc 4115 3630 4508 4008 4177 0.01 Vitamin C, mgc 130 137 142 135 139 0.74

a Values adjusted to age, area, smoking, education, energy from food and alcohol intake. b Alcohol was excluded when the percentage contribution of other energy nutrients to total energy was calculated. c Geometric means are shown. Log transformation was used for signi®cance test.

to of®cial statistics (Simpura et al, 1995). In addition to that wine drinkers also have more antioxidants in their diet. underestimation, a majority of the high consumers of Wine drinkers tend to have higher education than abstainers alcohol do not usually participate in health surveys. In or spirit drinkers. However, the division of subjects accord- our study, quite satisfactory overall response rates of 61% ing to their alcohol beverage drinking is very similar to the of men and 71% of women were achieved. Nevertheless, division by education. The differences observed in diet younger men living in the metropolitan area of Helsinki may, therefore, be more closely associated with educational (Kleemola et al, 1994), and women who consume an levels than with alcohol beverage drinking. above-average amount of alcohol had a lower response Among both men and women, the percentage of energy rate than others (Roos et al, 1996). from fat was signi®cantly higher in alcohol consumers than There is a general agreement that the association in abstainers. Among male drinkers, the nonalcoholic between alcohol intake and mortality is a J- or a U- energy percentage derived from carbohydrates was lower shaped curve, as reviewed by Poikolainen (1995). The while the percentage derived from protein was greater, but bene®cial effects of alcohol relate to the inhibition of the associations with protein and fat leveled-off after the platelet aggregation and to the higher ratio of HDL cho- exclusion of underreporters. Many other studies have lesterol to LDL cholesterol. There is less coherent evidence reported that alcohol consumers have lower carbohydrate of the relationship between types of alcohol beverages and intake than abstainers (Le Marchand et al, 1989; Colditz et mortality from coronary heart disease as reviewed by al, 1991). Colditz et al (1991) suggested that the consump- Rimm et al (1996). However, in the Copenhagen City tion of candy and sugar is inversely related to alcohol Heart Study the descending part of the U-shaped curve intake, raising the possibility that sweet food items and could be explained by wine consumption while diseases, alcohol may be competitors in diet. Our study did not injuries, and of spirit drinkers explained the support this observation because the percentages of energy ascending part (Grùnbñk et al, 1995). Flavonoids, some from sugar were quite similar between alcohol beverage phenolic substances, and tannin in red wine may categories. protect against coronary heart disease, mainly by de- The results on the association between alcohol consump- creasing oxidative tendencies in the metabolism (Kinsella tion and obesity are contradictory. Alcohol is the second et al, 1993). most energy-dense macronutrient and it is metabolized very Little attention has been paid to the associations between effectively. However, many epidemiological studies in types of alcoholic beverages and diet, and the question different populations support the hypothesis that high remains whether wine drinkers have a healthier diet than alcohol intake may be associated with lower BMI (Heller- others. In Western Australia wine was related to healthier stedt, 1990; Prentice, 1995). These discrepancies have been dietary choices among working men (Burke et al, 1995); explained by energy wasting systems as MEOS (Lieber, those who preferred wine consumed more fruit, vegetables, 1991) or futile cycle (Prentice, 1995). Recent Finnish and bread while meat, fried foods, eggs, and salt were laboratory experiments have shown that alcohol is partly associated with a preference for beer. In our study, female metabolized not only in the liver but also in the large bowel wine drinkers had a signi®cantly higher intake of carote- by a bacteriocolonic pathway, in this case all energy from noids, and male wine drinkers had the highest intake of alcohol may not become available for the human body vitamin C of all alcohol consumers. Forman et al (1995) (Salaspuro, 1996). Opposite arguments have also been found that carotenoids were more ef®ciently metabolised presented. No noticed effect of alcohol has been observed with than without alcohol. Further, red wine has been found on thermogenesis and it was carbohydrate that was sub- to increase the antioxidant capacity of serum within a stituted by alcohol in experiences; the carbohydrate de®cit couple of hours after meals (Maxwell et al, 1994). It is can explain the noticed weight loss since glycogen binds possible that pure ethanol and some components of wine three times its weight of water. It can also be that some may prevent coronary heart disease, but our results indicate other unmeasured lifestyle factors related to alcohol con- Alcohol, diet and BMI S MaÈnnistoÈ et al 331 sumption may offset the additional energy from the alcohol Belfrage P, Berg B, Cronholm T, Elmqvist D, HaÈgerstrand I, Johansson B, (Prentice, 1995). Only few studies have reported anything Nilsson-Ehle P, Norden G, SjoÈvall J & Wiebe T (1973): Prolonged administration of ethanol to young, healthy volunteers: effects on about the associations between types of alcoholic beverages biochemical, morphological and neutro-physiological parameters. and body mass index. Beer drinkers were shown to gain Med. Scand. Acta. Suppl. 552, 5±43. weight during a ®ve-week experiment in which they con- Black AE, Goldberg GR, Jebb SA, Livingstone MBE, Cole TJ & Prentice sumed 63 g alcohol per day (Belfrage et al, 1973), and in AM (1991): Critical evaluation of energy intake data using fundamental Finland spirit drinkers were shown to be heavier than other principles of energy physiology: 2. Evaluating the results of published surveys. Eur. J. Clin. Nutr. 45, 583±599. alcohol consumers (MaÈnnistoÈ et al, 1996). Burke V, Puddey IB & Beilin LJ (1995): Mortality associated with wines, In our study, male alcohol consumers had a lower body beers, and spirits. BMJ 311, 1166. mass index than abstainers although there were no differ- Colditz GA, Giovannucci E, Rimm EB, Stampfer MJ, Rosner B, Speizer ences between total energy intakes, physical activity index, FE, Gordis E & Willett WC (1991): Alcohol intake in relation to diet and obesity in women and men. Am. J. Clin. Nutr. 54, 49±55. and daily energy expenditures. Thus, males seemed to Doll R, Peto R, Hall E, Wheatley K & Gray R (1994): Mortality in relation substitute alcohol for foods. The female alcohol consumers to consumption of alcohol: 13 years' observations on male British also had a lower body mass index and substituted alcohol doctors. BMJ 309, 911±918. for foods. Among females, the results including under- Dorfman A, Kimball AW & Friedman LA (1985): Regression modelling reporters showed that total energy intake decreased with of consumption or exposure variables classi®ed by type. Am. J. Epidemiol. 122, 1096±1107. increasing alcohol consumption, but after exclusion of FAO/WHO/UNU (1985): Report of a joint expert consultation. Energy underreporters the differences in the total energy intake and protein requirements. WHO Tech. Rep. Ser. No. 724. WHO: between alcohol beverage categories disappeared. The Geneva. observed lower body mass index may be explained as a Fogelholm M, Kaprio J & Sarna S (1994): Healthy lifestyles of former Finnish world class athletes. Med. Sci. Sports Exerc. 26, 224±229. contribution of the MEOS or other wasting system for Forman MR, Beecher GR, Lanza E, Reichman ME, Graubard BI, Camp- alcoholic energy. Spirit drinkers, however, were the hea- bell WS, Marr T, Yong LC, Judd JT & Taylor PR (1995): Effect of viest among alcohol beverage categories; they tended to alcohol consumption on plasma carotenoid concentrations in premeno- add alcohol to the diet. In Finland, spirit drinkers represent pausal women: a controlled dietary study. Am. J. Clin. Nutr. 62, 131± the traditional way of using alcohol, in which consumption 135. Fuchs CS, Stampfer MJ, Colditz GA, Giovannucci EL, Manson JE, is concentrated on the weekends and at special occasions. Kawachi I, Hunter DJ, Hankinson SE, Hennekens CH, Rosner B, Many working-class men, for instance, have a habit of Speizer FE & Willett WC (1995): Alcohol consumption and mortality buying a `Friday-night bottle' of spirits, which is consumed among women. N. Engl. J. Med. 332, 1245±1250. without food and is mainly used to get drunk. It may be that Gaziano JM & Hennekens C (1995): Royal colleges' advice on alcohol consumption. BMJ 311, 3±4. such drinking habits do not maintain wasting systems for Grùnbñk M, Deis A, Sùrensen TIA, Becker U, Schnohr P & Jensen G alcoholic energy (MaÈnnistoÈ et al, 1996). (1995): Mortality associated with moderate intakes of wine, beer, or The old dietary recommendations either do not address spirits. BMJ 310, 1165±1169. alcohol intake (National Research Council, 1989) or they Hellerstedt WL, Jeffery RW & Murray DM (1990): The association advise limited use (The State Advisory Board on Nutrition, between alcohol intake and adiposity in the general population (review). Am. J. Epidemiol. 132, 594±611. 1987). Because the available epidemiological evidences Hirvonen T, MaÈnnistoÈ S, Roos E & Pietinen P (1997): Increasing clearly indicate that moderate alcohol consumption, in prevalence of underreporting does not necessarily distort dietary sur- comparison with abstinence, lowers the risk of coronary veys. Eur. J. Clin. Nutr. 51, 297±301. heart disease, a new recommendation has been recently Jackson R & Beaglehole R (1995): Alcohol consumption guidelines: relative safety vs absolute risks and bene®ts. Lancet 346, 716. published (Gaziano & Hennekens, 1995). The Royal Col- Kinsella JE, Frankel E, German B & Kanner J (1993): Possible mechan- leges of Physicians, Psychiatrists, and General Practitioners isms for the protective role of antioxidants in wine and plant foods. have determined `sensible limits' of alcohol intake to be 21 Food Technol. 47, 85±89. units a week for men and 14 units for women. Jackson & Kleemola P, Virtanen M & Pietinen P (1994): The 1992 dietary survey of Beaglehole (1995) have also suggested that the subjects Finnish adults. Helsinki: Publications of the National Public Health Institute B2. most likely to bene®t from drinking small amounts of Le Marchand L, Ntilivamunda A, Kolenel LN, Vanderford MK & Lee J alcohol are men over 40 and women over 50 y old who (1988): Relationship of smoking to other life-style factors among are at high risk of coronary heart disease and at low risk of several ethnic groups in Hawaii. Asian-Pac. J. Public Health 2, 120± alcohol-related injuries and diseases. In our study, alcohol 126. Le Marchand L, Kolonel LN, Hankin JH & Yoshizawa CN (1989): consumers were leaner than abstainers, and wine drinkers Relationship of alcohol consumption to diet: a population-based study in particular had the healthiest diet. Alcohol consumption in Hawaii. Am. J. Clin. Nutr. 49, 567±572. itself, or different characteristics of alcohol beverages, may Lieber CS (1991): Perspecives: do alcohol calories count? Am. J. Clin. explain the `French Paradox', but the quality and quantity Nutr. 54, 976±982. of alcohol consumption can also be an indicator of educa- Maxwell S, Cruickshank A & Thorpe G (1994): Red wine and antioxidant activity in serum. Lancet 344, 193±194. tion or lifestyle. MaÈnnistoÈ S, Pietinen P, Haukka J, Ovaskainen M-L, Albanes D & Virtamo J (1996): Reported alcohol intake, diet and body mass index in male AcknowledgementsÐWe are grateful to PaÈivi Kleemola, M.Sc., for verify- smokers. Eur. J. Clin. Nutr. 50, 239±245. ing the food records and analysing the baseline dietary data in the National Research Council (1989): Recommended Dietary Allowances, Finmonica risk factor survey. 10th edn. Washington, DC: National Academy Press. Pietinen P, Vartiainen E, SeppaÈnen R, Aro A & Puska P (1996): Changes in diet in Finland from 1972 to 1992. Impact on coronary heart disease References risk. Prev. Med. 25, 243±250. Poikolainen K (1995): Alcohol and mortality: review. J. Clin. Epidemiol. Ainsworth BE, Haskell WL, Leon AS, Jacobs Jr. DR, Montoye HJ, Sallis 48, 455±465. JF & Paffenbarger Jr. RS (1993): Compendium of physical activities: Prentice AM (1995): Alcohol and obesity. Int. J. Obes. 19, Suppl. 5, S44± classi®cation of energy costs of human physical activities. Med. Sci. S50. Sports Exerc. 25, 71±80. Renaud F & de Logeril M (1992): Wine, alcohol, platelets and the French Artaud-Wild SM, Connor SL, Sexton G & Connor WE (1993): Differences paradox for . Lancet 339, 1523±1526. in coronary mortality can be explained by differences in cholesterol and Rimm EB, Klatsky A, Grobbee D & Stampfer MJ (1996): Review of saturated fat intakes in 40 countries but not in France and Finland. A moderate alcohol consumption and reduced risk of coronary heart paradox. Circulation 88, 2771±2779. disease: is the effect due to beer, wine, or spirits? BMJ 312, 731±736. Alcohol, diet and BMI S MaÈnnistoÈ et al 332 Roos E, Ovaskainen M-L & Pietinen P (1995): Validity and comparison of The Finnish State Alcohol Company (1993): Alcohol statistical yearbook three saturated fat indices. Scan. J. Nutr. 39, 55±59. 1992. Helsinki: Government Printing Of®ce. Roos E, PraÈttaÈlaÈ R, Lahelma E, Kleemola P & Pietinen P (1996): Modern The State Advisory Board on Nutrition (1989): Dietary guidelines and healthy? Socioeconomic differences in the quality of diet. Eur. J. and their scienti®c principles. Helsinki: Government Printing Clin. Nutr. 50, 753±760. Centre. Salaspuro M (1996): Bacteriocolonic pathway for ethanol oxidation: char- Thompson FE & Byers T (1994): Dietary assessment resource manual. J. acteristics and implications (review). Annals of Medicine 28, 195±200. Nutr. 124, Suppl. 2245S±2261S. Simpura J, Paakkanen P & Mustonen H (1995): New beverages, new WHO MONICA Project Principal Investigators (1988): The World Health drinking contexts? Signs of modernization in Finnish drinking habits Organization MONICA project (monitoring trends and determinants of from 1984 to 1992, compared with trends in the European Community. ): a major international collaboration. J. Clin. Addiction 90, 673±683. Epidemiol. 41, 105±114.