|
|
PARTICIPANTS AND METHODS |
|---|
Baseline Examinations in 1974 and Follow-Up Examination in 1986
The cohort and examinations have been described earlier (18,19). Initially,
3490 healthy men, mostly businessmen and executivesborn in 1919–1934, had
participated in structured health check-ups
during the 1960s and early 1970s at the Institute of
Occupational Health in Helsinki. These health examinations—usually paid
by their companies—were specifically directed to healthy
men in leading positions in various private companies; consequently,
the participants can be assumed to have similar social
status and wealth in their adult life, and also during the
follow-up. The men were evaluated with questionnaires and clinical
and laboratory examinations in 1974. As 68 men had died
by 1974 and 2721 men responded to the 1974 query, the response rate
was 80%. Of these 2721 men, 2468 (91%) reported their beveragepreference. Thus,
overall we have data of the beverage preference of
72% of the original background population (Figure 1). In the
questionnaires, participants were asked about their weekly alcohol
consumption, and types of beverages consumed were reported by
2468 men. The participants were in positions of responsibility, and
those with known alcoholism or psychiatric disturbances were
excluded from the study at baseline. These exclusions are liable
to diminish the possible effect of problem drinking on the
results in this cohort. The pattern of drinking—for example,
binge drinking—was not identified. Of the men, 131
did not use alcohol at all (abstinence group). In this study, one
unit of alcohol ("restaurant unit": a bottle of beer, a glass
of wine, one drink of spirits) was calculated to contain 14
g of alcohol. Alcohol preference was defined simply as the beverage
that the man reported to consume the most, whatever the
difference to other beverages. With this definition, 694 men
preferred beer, 251 wine, and 937 spirits ("single preference," although
most men consumed all beverages). In addition, 455 men
reported similar amounts of two or three beverages. In the analyses
we compared the three single preference groups unless otherwise
stated.
Figure 1. Flow chart of the study
At baseline in 1974, all the men were professionally active. Most
of the men had various cardiovascular risk factors but only
115 (5%) had a known history of cardiovascular disease. Their
alcohol preferences did not differ from the rest of the cohort.
Weight and height were measured, and body mass index (BMI)
was calculated as weight (kg) divided by height (m) squared. Some
(1657) of the men also reported their weight at 25 years of
age, which was used to calculate midlife weight gain (weight in
1974 minus reported weight at 25 years of age). In 1974 the men
were also asked questions on how they rated their present health
and physical fitness on a 5-step scale ("very good," "good,"
"fair," "poor," "very poor").
In 1985–1986, 1369 men were reassessed with questionnaires and laboratory examinations. This survey was only directed to those men who were clinically healthy in 1974 and had full data of baseline risk factors available. The survey included a question about alcohol consumption. Serum gamma-glutamyl transferase (gamma-GT) activity was measured in 203 men in conjunction with a metabolic study (20). Thus, we could compare reported alcohol consumption with a biochemical marker of alcohol (21) in a small sample of the cohort.
The 2000 Survey and Quality of Life
In 2000, a questionnaire was mailed to all survivors (n =
1781) and remailed once
for nonrespondents; 1515 men (85.1%) responded (Figure
1). Baseline alcohol consumption was not different between respondents
and nonrespondents (p =.32).
Of the respondents, 1127
had also reported their baseline beverage preference. The questionnaire
included items on demographic variables and lifestyle such
as regular exercise and smoking status. The question on alcohol
consumption was similar to that in the earlier surveys of
1974 and 1985–1986. In addition, the Finnish version of
the RAND-36-Item Health Survey 1.0 (practically identical to
the Short Form [SF]-36 health survey; 22,23) was embedded in
the questionnaire. The eight RAND-36 scales (physical function, role
physical, bodily pain, general health, vitality, social function,
role emotional, and mental health) measure health-related quality
of life, and the RAND-36 as a mailed questionnaire has been
validated in the Finnish general population (22).
Mortality Follow-Up
Total mortality of the study population through December 31, 2002,
was retrieved from the National Population Information System,
which is a registry of all Finnish citizens. The register's assessment
of vital status is reliable for people who permanently reside
in Finland (over 95% of the present cohort), irrespective of
whether they die in Finland or abroad. Moreover, the assessment of
vital status is also quite reliable for Finnish citizens living
permanently abroad. Causes of death through December 31,
1999 were determined from the countrywide computerized Cause of
Death Register at Statistics Finland in which trained nosologists code
causes of death. The causes were categorized into 3 groups: cardiovascular,
cancer, or other causes.
Statistical Methods
NCSS statistical software (www.ncss.com) was
used for the analyses. In
the analyses, alcohol consumption was categorized as described above.
Student's ttests,
nonparametric tests, and analyses of
covariance (ANCOVA) were used where appropriate to compare continuous
variables, chi-square and trend tests were used to compare
proportions, and Spearman rank coefficients were used to
assess correlations. Differences in survival curves were analyzed
with the log rank test. Relative risks (RR) with their 95%
confidence intervals (CI) for mortality associated with alcoholic
beverage preference at baseline were calculated using Cox's
proportional hazards regression. Other risk factors were adjusted
for in respective models. The eight RAND-36 scales were
calculated from questionnaires (22,23).
In statistical analyses,
two-tailed tests were used and p values
<.05 were taken as
statistically significant.
|
|
RESULTS |
|---|
Baseline
Baseline characteristics of the cohort according to baseline alcoholic
beverage preference are shown in Table
1. Data from men
reporting no alcohol intake (abstinence group, n =
131) or no single
preference (n = 455) are shown
for comparison. Of the
men, 87% were categorized to consume alcohol moderately (mean
consumption less than 3 drinks/d, 42 g/d), and only 13% reported
high consumption (mean 5 drinks/d, 70 g/d); these proportions were
not different between the preference groups. The men preferring spirits
differed from the other two groups by having a higher BMI
and serum triglycerides, also smoking was more frequent among
this group. Total alcohol consumption was not significantly different
between the groups. Self-report of subjective health and
physical fitness in the three preference groups in 1974 showed
that wine preference was associated with better profiles at
baseline (Figure 2).
Table 1. Age-Adjusted Baseline Characteristics in 1974 According to Alcoholic Beverage Preference.
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Note: Variables are age-adjusted.
Figure 2. Distribution of subjective physical fitness and subjective general health in 1974 according to the alcoholic beverage preference group. Numbers refer to the 5-step scale (1 = "very good," 2 = "good," 3 = "fair," 4 = "poor," 5 = "very poor") and p values indicate the difference between the preference groups
Alcohol Consumption During Follow-Up
Reported alcohol consumptions in 1974 and 2000 were significantly correlated
(r = 0.53, p <.0001).
However, average total alcohol consumption
had decreased in all preference groups in 2000 as compared
to baseline. Reported average consumptions of alcohol among
surviving beer (n = 356), wine
(n = 127), and spirits drinkers
(n = 404) in 2000 were 130.8,
149.4, and 136.2 g/wk, respectively
(p =.46). The alcoholic
beverage preference seemed to
track well over the decades and was preserved in survivors in
2000 (Table 2). The 1974 data were essentially similar,
albeit smaller when only
the 2000 survivors were included (data not shown).Whereas
beer drinkers had only cut their beer intake, wine
drinkers tended to have increased beer consumption, and baseline
spirit preferers had reduced their intake of spirits and
increased their wine intake. It can be noted that wine drinkers had
the lowest overall alcohol consumption in 1974 but the highest in
2000.
Table 2. Average Weekly Consumption of Various Alcoholic Beverages in 1974 and 2000 According to Baseline Alcohol Preference.
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||
In 1985, reported alcohol consumption was significantly associated with
gamma-GT activity in a sample of the cohort (r =
0.41, p <.0001).
The correlation was significant in all preference groups.
Mortality
During the 29-year follow-up, 814 men (33.0% of the initial 1974
cohort) died. Among those who had reported a single preference in
1974 (n = 1882), there were
644 deaths (34.2%). Mortality was
highest in the group preferring spirits (n =
346, 36.9%), and 224
(32.3%) and 74 (29.5%) in the groups preferring beer and
wine, respectively. Unadjusted mortalities between the three groups
were significantly different (p =.03);
survival curves are shown
in Figure 3.
Analysis of causes of death (available up
to 1999) revealed that the survival advantage in the group preferring
wine was due to fewer cardiovascular deaths. Cardiovascular disease
was the cause of death in 6.8% in the wine preference group
versus 12.4% in the other two groups combined (RR 0.54; 95%
CI, 0.33–0.89; p =.01).
Total mortality was further analyzed
with multivariate analyses using proportional hazards regression.
In these analyses, beer and wine preference were compared
with spirits preference (Table 3). Wine preference was
associated with the lowest 29-year mortality irrespective of
covariates, and in the fully adjusted model (age, baseline cardiovascular
risk factors, and alcohol consumption as covariates) reduction
in mortality was 34% (p =.03).
Beer preference wasnot associated with significantly lower mortality as compared to
spirits preference.
Figure 3. Unadjusted survival curves of the baseline alcoholic
beverage preference groups during the 29-year follow-up showing best survival
for wine drinkers, intermediate for beer drinkers, and worst survival for
spirits drinkers. p =.03
(log rank test) between the groups
Table 3. Proportional Hazards Analysis of 29-Year Mortality According to Alcoholic Beverage Preference.
|
||||||||||||||||||||
Notes: Model A = No adjustments; Model B = adjusted for age and baseline smoking; Model C = adjusted for baseline age, smoking, alcohol consumption, body mass index, serum cholesterol, and systolic blood pressure.
*Covariates simultaneously in the Cox's proportional hazards' model.
We also reran multivariate analyses in which the group of no single
preference and the abstainers' group were included. In the
fully adjusted model (with spirits preference as the reference group),
the benefit in mortality remained statistically significant in
favor of men preferring wine (RR 0.65; 95% CI, 0.44–0.96; p=.03).
RRs for beer drinkers (0.91; 95% CI, 0.70–1.18), for
men with no single preference (1.12; 95% CI, 0.89–1.40), and
for abstainers (1.21; 95% CI, 0.79–1.86) were not statistically
significant. If men with no single preference were
the reference group, the RR for wine drinkers was even lower
and highly statistically significant in the fully adjusted model
(0.59; 95% CI, 0.40–0.86; p =.007).
Finally, we compared mortality in men who reported no wine consumption (n = 1034) to those with various levels of weekly wine consumption: 1–3 drinks (n = 936), 4–7 drinks (n = 348), and more than 7 drinks (n = 150). In the fully adjusted model, a U-shaped relationship was seen, with the group consuming 4–7drinks having the lowest mortality (1–3 drinks: RR 0.99; 95% CI, 0.73–1.19; 4–7 drinks: RR 0.68; 95% CI, 0.48–0.96; more than 7 drinks: RR 0.79; 95% CI, 0.49–1.30).
2000 Survey and Quality of Life
In the 2000 questionnaire survey, there were 1127 men who had reported
their baseline alcoholic beverage preference (439 beer, 155
wine, and 533 spirits). Response rates were 93%, 88%, and 90%
in the beer, wine, and spirits preference groups, respectively. Over
90% were retired but 98% were still home dwelling. These characteristics
were not different between the preference groups. The
men preferring spirits continued to have higher BMI (26.2 kg/m2)
than the other groups (25.6 and 25.4 kg/m2 in
beer and wine drinkers,
respectively; p =.007),
and smoking was significantly more
prevalent among them (p =.03).
In turn, the men with preference for
wine at baseline reported to exercise regularly more often in
2000 than did the men in other groups (p =.03).
Results of the health-related quality of life assessed with the RAND-36 questionnaire and adjusted for age and baseline smoking are shown in Figure 4. The men with wine preference at baseline had consistently the best scores on all eight RAND-36 scales, although the scores were statistically significantly higher only in the scales of general health (p =.007) and mental health (p =.01).
Figure 4. Baseline alcoholic beverage preference and the eight
scales of health-related quality of life RAND-36 in old age in 2000. Scales are
adjusted for age and smoking status at baseline. p values
indicate difference between the preference groups. A difference of at least 3
points is considered to be clinically important (23)
|
|
DISCUSSION |
|---|
In our male cohort with similar socioeconomic status in adult life,
preference of wine was associated with decreased mortality when
compared with preference for beer or spirits over a follow-up of
29 years. This association persisted after various baselinecovariates—including
the total amount of alcohol consumed and
important cardiovascular risk factors—were adjusted for.
Wine preference was also associated with better scores in
two scales (general health and mental health) of the
RAND-36-assessedhealth-related quality of life in old age, but already at
baseline wine preferers
seemed to have better subjective health. In contrast,spirits preference was
associated with worse risk profile at baseline
and worse prognosis of survival and quality of life during
follow-up. The results for beer drinkers were between those
for wine and spirits preferers, but the differences were not
statistically significant.
The present results are in accordance with the meta-analysis that showed an inverse association between wine consumption, and to a lesser extent beer consumption, and vascular risk (2). After that meta-analysis, a large study of almost 130,000 adults showed lower mortality risk associated with wine drinking (4). Our finding of better quality of life related to mental health in wine drinkers in old age is also supported by some earlier data (6–9). However, the question of cause and effect and the role of confounders—for example, social factors and cardiovascular risk factors—associated with wine preference has remained. Several studies have further considered the possibility that wine drinkers represent a special trait with overall healthy lifestyle and attitude toward life from a young age (11–16). A recent study from Denmark showed that customers buying wine at the supermarket made more healthy choices of other food items than did people who bought beer (24).
Our study has some strengths for studying the specific effects of alcoholic beverages. Because social class has an obvious influence on alcohol consumption, type of beverage, and alcohol-related mortality (25), our study with all men from the highest socioeconomic class in their adult life represents a cleaner test of the hypothesis in an area where social class may have an important confounding effect. We also measured several cardiovascular risk factors (including serum cholesterol) that were included in the multifactorial models, and we had information on alcoholic beverage preference after baseline during the follow-up. Moreover, our cohort mainly consisted of individuals who consume various alcoholic beverages ("mixed drinkers"). While this diminishes differences betweenbeverage types, it also circumvents possible confounders associated with a uniform pattern. In addition, we related alcoholic beverage preference to the quality of life in old age when the baseline differences in alcohol preference still prevailed.
The results were based on a relatively large cohort of men with a substantial number of mortality endpoints during the follow-up of almost 30 years. The mortality verification by national registers was reliable. With knowledge of the Finnish culture, we also believe that the reporting of alcohol consumption was reliable in this cohort of Finnish businessmen and executives. This belief is further supported by the consistent pattern of reported consumption over the decades. The tracking of the beverage preference from 1974 through 2000 was good, and the reported alcohol intake was significantly correlated with several risk factors known to be associated with alcohol consumption (18). Furthermore, the serum concentration of gamma-GT [a biochemical indicator of alcohol consumption (21)] was available in a small subgroup, and it was related to reported alcohol consumption.
Our study has also some important limitations. This cohort consisted mainly of low to moderate middle-aged drinkers and included only a limited number of heavy drinkers, probably because the men were professionally active and healthy at baseline in 1974. It is also a limitation that we do not have information on drinking pattern and behavior; for example, binge drinking may have negative effects (5) and associate more with one type of beverage than with another. It is speculative, however, that binge drinking would have been more frequent in any particular preference group of our study, because most men were mixed drinkers. Also, as our study population was a selection of men from the highest social class, the extrapolation of the results to the general population (and especially to women) needs to be done cautiously. Furthermore, the men were "self-selected" to participate in health check-ups during the 1960s, and were thus probably health-conscious, but this hardly affects the present study on beverage preference. In contrast, as already stated, homogeneity would also be a strength, as the confounding effects of social status on alcohol consumption and quality of life could be minimized. Although we can assume that adult life social status and wealth (also during follow-up) were similar in our cohort, we were not able to adjust for differences in personal characteristics and early-life phenomena between the beverage preference groups. These can naturally induce residual confounding to the results, although data from the Whitehall II study of British civil servants indicated that adult socioeconomic status is a more important predictor of morbidity, attributable to, e.g., coronary heart disease, than is social status earlier in life (26). The high social status and homogeneity of the cohort has been taken for granted on the basis of work status; in the present database, we do not have data on childhood home circumstances or on type and length of education until in the 2002 survey (our unpublished results). At that time, when a third of the baseline cohort had died, surviving baseline spirits preferers had had less higher education (65%) than did the wine (77%) or beer drinkers (73%). In contrast, the education of the men with no single preference (76%) was similar to that of wine drinkers; still, their mortality was different (see Results). Average years ofeducation were 12.8, 14.4, 14.1, 13.9, and 11.8 in the spirits, wine, beer, no preference, and abstainer groups, respectively (global p value <.0001). Also, the distribution of father's occupation differed marginally significantly (p =.05) between the preference groups, wine preferers tending to originate from families with higher social status (our unpublished results).
The profile of quality of life could not be assessed in 1974 as the RAND-36 did not exist at that time. The available data in Figure 2 nevertheless indicate that, in midlife, wine preference (as compared to other preference groups) is associated with better subjective health. Subjective general health has been found to associate best with a global measure of health-related quality of life (27). However, a question may arise whether these differences were a cause or consequence of wine consumption.
Conclusion
In this group of men from the highest social class in their adult
life, the long-term overall health effects associated with
wine preference were clearly beneficial when compared to other
types of alcohol. Despite a conservative analysis including adjustment
for cardiovascular risk factors and the amount of alcohol
consumed, wine drinkers had a lower overall mortality and
a better quality of life in old age. The results may thus support
the idea that moderate consumption of wine has specific, beneficial
effects on health. However, only a full life-course study
(or a randomized trial) accounting for also pre-midlife characteristics
could finally decide whether it is truly the characteristics
of the alcoholic beverages—and not the characteristics
of the people who make the choices—that account
for these differences.
|
|
A |
|---|
This work was supported by The Academy of Finland, the Päivikki and
Sakari Sohlberg Foundation, the Helsinki University Central Hospital,
and the Finnish Foundation for Cardiovascular Research.
Author contributions are as follows: Design of the experiment (T. E. S., A. Y. S., T. A. M.), collection of data (T. E. S., V. V. S., T. A. M., R. S. T.), analysis of data (A. Y. S., T. E. S., K. P., R. S. T.), writing and revising the manuscript (T. E. S., A. Y. S., V. V. S., K. P., T. A. M., R. S. T.).
|
|
F |
|---|
Decision Editor: Darryl Wieland, PhD, MPH
Received January 24, 2006
Accepted May 19, 2006
|
|
References |
|---|
-
Room R, Graham K, Rehm J, Jernigan D, Monteiro M. Drinking and its burden in a global perspective: policy considerations and options. Eur Addict Res. 2003;9:165-175.[Medline]
-
Di Castelnuovo A, Rotondo S, Iacoviello L, Donati MB, de Gaetano G. Meta-analysis of wine and beer consumption in relation to vascular risk. Circulation. 2002;105:2836-2844.
-
Rimm EB, Stampfer MJ. Wine, beer, and spirits. Are they really horses of a different color? Circulation. 2002;105:2806-2807.
-
Klatsky AL, Friedman GD, Armstrong MA, Kipp H. Wine, liquor, beer, and mortality. Am J Epidemiol. 2003;158:585-595.
[Abstract/Free Full Text] -
Mukamal KJ, Conigrave KM, Mittleman MA, et al. Roles of drinking pattern and type of alcohol consumed in coronary heart disease in men. N Engl J Med. 2003;348:109-118.
[Abstract/Free Full Text] -
Letenneur L, Larrieu S, Barberger-Gateau P. Alcohol and tobacco consumption as risk factors of dementia: a review of epidemiological studies. Biomed Pharmacother. 2004;58:95-99.[Medline]
-
Mukamal KJ, Kuller LH, Fitzpatrick AL, Longstreth WT, Jr, Mittleman MA, Siscovick DS. Prospective study of alcohol consumption and risk of dementia in older adults. JAMA. 2003;289:1405-1413.
[Abstract/Free Full Text] -
Anttila T, Helkala EL, Viitanen M, et al. Alcohol drinking in middle age and subsequent risk of mild cognitive impairment and dementia in old age: a prospective population based study. BMJ. 2004;329:539-545.
[Abstract/Free Full Text] -
Lindeman RD, Wayne SJ, Baumgartner RN, Garry PJ. Cognitive function in drinkers compared to abstainers in the New Mexico Elder Health Survey. J Gerontol A Biol Sci Med Sci. 2005;60A:1065-1070.
[Abstract/Free Full Text] -
de Lange DW, van de Wiel A. Drink to prevent: review on the cardioprotective mechanisms of alcohol and red wine polyphenols. Semin Vasc Med.2004;4:173-186.[Medline]
-
Gronbaek M, Mortensen EL, Mygind K, et al. Beer, wine, spirits and subjective health. J Epidemiol Community Health. 1999;53:721-724.[Abstract]
-
Mortensen EL, Jensen HH, Sanders SA, Reinisch JM. Better psychological functioning and higher social status may largely explain the apparent health benefits of wine. Arch Intern Med. 2001;161:1844-1848.
[Abstract/Free Full Text] -
Barefoot JC, Gronbaek M, Feaganes JR, McPherson RS, Williams RB, Siegler IC. Alcoholic beverage preference, diet, and health habits in the UNC Alumni Heart Study. Am J Clin Nutr. 2002;76:466-472.
[Abstract/Free Full Text] -
Collie ME, Higgins JC. Hope for hops? Arch Intern Med. 2002;162:364-365.
[Free Full Text] -
McCann SE, Sempos C, Freudenheim JL, et al. Alcoholic beverage preference and characteristics of drinkers and nondrinkers in western New York (United States). Nutr Metab Cardiovasc Dis. 2003;13:2-11.[Medline]
-
Paschall M, Lipton RI. Wine preference and related health determinants in a U.S. national sample of young adults. Drug Alcohol Depend. 2005;78:339-344.[Medline]
-
Westin S. Study confounded by lack of correction for social class. BMJ. 1995;311:1167.
[Free Full Text] -
Strandberg AY, Strandberg TE, Salomaa VV, Pitkala K, Miettinen TA. Alcohol consumption, 29-y total mortality, and quality of life in men in old age.Am J Clin Nutr. 2004;80:1366-1371.
[Abstract/Free Full Text] -
Strandberg TE, Salomaa VV, Vanhanen HT, Naukkarinen VA, Sarna SJ, Miettinen TA. Mortality in participants and non-participants of a multifactorial prevention study of cardiovascular diseases. A 28-year follow-up of the Helsinki Businessmen Study. Br Heart J. 1995;74:449-454.
[Abstract/Free Full Text] -
Strandberg TE, Salomaa V, Vanhanen H, Miettinen TA. Associations of fasting blood glucose with cholesterol absorption and synthesis in nondiabetic middle-aged men. Diabetes. 1996;45:755-761.[Abstract]
-
Whitfield JB. Gamma glutamyl transferase. Crit Rev Clin Lab Sci. 2001;38:263-355.[Medline]
-
Aalto AM, Aro AR, Teperi J. RAND-36 as a measure of Health-Related Quality of Life. Reliability, construct validity, and reference values in the Finnish general population. Helsinki, Finland: Stakes, Research Reports; No. 101, 1999.
-
Hays RD, Morales LS. The RAND-36 measure of health-related quality of life. Ann Med. 2001;33:350-357.[Medline]
-
Johansen D, Friis K, Skovenborg E, Groenbaek M. Food buying habits of people who buy wine or beer: cross sectional study. BMJ. 2006;332:519-522.
[Abstract/Free Full Text] -
Mäkelä P, Keskimäki IT, Koskinen S. What underlies the high alcohol related mortality of the disadvantaged: high mortality or poor survival? J Epidemiol Community Health. 2002;57:981-986.
-
Marmot M, Shipley M, Brunner E, Hemingway H. Relative contribution of early life and adult socio-economic factors to adult morbidity in the Whitehall II study. J Epidemiol Community Health. 2001;55:301-307.
[Abstract/Free Full Text] -
Shmueli A. The SF-36 profile and health-related quality of life: an interpretative analysis. Qual Life Res. 1998;7:187-195.[Medline]