Moderate alcohol intake is associated with survival in the elderly:
the Dubbo Study
Leon A Simons, John McCallum, Yechiel Friedlander, Michael Ortiz and
MJA 2000; 173: 121-124
For editorial comment, see Stockwell
More articles on Cardiology and cardiac surgery
Objective: To examine the relationship between
alcohol intake and survival in elderly people.|
Design and setting: A prospective study over 116 months of
non-institutionalised subjects living in Dubbo, a rural town
(population, 34 000) in New South Wales.
Participants: 1235 men and 1570 women aged 60 years and
over who were first examined in 1988-89.
Main outcome measures: All-causes mortality; gross cost
of alcohol per life-year gained.
Results: Death occurred in 450 men and 392 women. Intake of
alcohol was generally moderate (ie, less than 14 drinks/week). Any
intake of alcohol was associated with reduced mortality in men up to 75
years and in women over 64 years. In a proportional hazards model, the
hazard ratio for mortality in men taking any alcohol was 0.63 (95% CI,
0.47-0.84) and in women was 0.75 (95% CI, 0.60-0.94). Cardiovascular
deaths in men were reduced from 20/100 (95% CI, 14-26) to 11/100 (95%
CI, 9-13) and in women from 16/100 (95% CI, 13-19) to 8/100 (95% CI,
6-10). The reduction in mortality occurred in men and women taking
only 1-7 drinks/week -- hazard ratios, 0.68 (95% CI, 0.49-0.94) and
0.78 (95% CI, 0.61-0.99), respectively, with a similar protective
effect from intake of beer or other forms of alcohol. After almost 10
years' follow-up, men taking any alcohol lived on average 7.6 months
longer, and women on average 2.7 months longer, compared with
non-drinkers. The gross cost for alcohol per life-year gained if
consuming 1-7 drinks/week was $5700 in men, and $19 000 in women.
Conclusions: Moderate alcohol intake in the elderly
appears to be associated with significantly longer survival in men
60-74 years and in all elderly women.
The consumption of moderate amounts of alcohol, compared with
abstention or with heavy alcohol intake, appears to be associated
with reduced all-causes mortality in middle-aged
subjects.1-3 This effect may be
partially mediated through a reduced risk of coronary heart disease
(CHD)4 and stroke.5 Some studies
attribute the protection to a specific effect of wine;6,7 other studies
attribute it to any type of alcohol.8|
In elderly people, some of this benefit from moderate alcohol intake
may be negated by mortality from other causes.9 In a prospective study of men
and women aged 65 years and over in the United States (Established
Populations for Epidemiologic Studies of the Elderly), alcohol
intake under 21 drinks/week was associated with a 30%-40% lower
all-causes mortality in two cohorts, but with no influence in a third
cohort.10 In Australian men and
women aged 60 years and over with 77 months' follow-up, the intake of
1-7 drinks/week was associated with a 22%-25% reduction in
all-causes mortality, although this reduction did not achieve
statistical significance.11 We have examined the
relationship between alcohol intake and survival in this Australian
cohort during a more extended follow-up of 116 months. We present the
results and include an economic analysis of moderate alcohol intake.
The Dubbo Study is an ongoing prospective examination of
cardiovascular and other diseases in an elderly Australian cohort
first examined in 1988-89. All non-institutionalised residents of
Dubbo, New South Wales, born before 1930 were eligible;
participation rate was 73% (1235 men and 1570 women). Methods and
measures have already been described in detail.12,13 The
baseline examinations comprised demographic, psychosocial and
standard cardiovascular risk assessments, including examination
of fasting blood samples.
Questions on alcohol usage were those asked in the National Heart
Foundation Risk Factor Prevalence Study,14 and yielded an
approximation of usual alcohol intake coded as zero, 1-7, 8-14, 15-28
and more than 28 drinks/week (referring to a standard drink
containing 10 g of alcohol). Specific intakes of beer, wine and
spirits were not separately sought, but subjects were asked whether
they normally drank beer or not, allowing a separation of drinking
behaviour into beer and "other".
Outcomes from August 1988 to 31 December 1998 were included in the
analysis, a median 116 months' follow-up. Hospitalisation and death
records were monitored continuously, and postal surveys were
conducted every two years to confirm vital status. The survey in 1997
successfully traced more than 98% of surviving participants.
Records were coded according to the International
classification of diseases, ninth revision, clinical
The independent contribution of any risk factor to mortality was
examined in a Cox proportional hazards model. Point estimates and 95%
CIs for the relative hazard of death were calculated from the
regression coefficients (presented as hazard ratio, a measure of
relative risk). The models included categories of alcohol intake as
described above, and, where relevant, a categorical term describing
whether a subject normally drank beer or "other" (ie,
wine/spirits). The proportional hazards model assumes
constant relative hazard over the length of follow-up. This
assumption was confirmed for each model by a plot of log-minus-log
survival, demonstrating parallel curves over all categories of
alcohol intake. Statistical analyses were conducted using SPSS for
A weighted alcohol acquisition cost per week using the midpoint of the
intake ranges was calculated for each alcohol intake stratum and for
each sex using 1999 Dubbo club prices. Individuals were assumed to
remain in their initial consumption strata over the whole time
period. Survival curve data from the Cox model were used to calculate
total expenditure on alcohol, as well as survival benefit for each
stratum of intake. An incremental analysis (difference in
cost/difference in benefit) was conducted using the
no-alcohol-consumption stratum as the reference. This yielded an
estimate of the gross cost per life-year gained. Although the study
collects hospitalisation records, hospitalisation costs were not
available and a net cost per life-year gained could not be estimated.
The study was approved by the institutional ethics committees at St
Vincent's Hospital, Sydney, the University of New South Wales and the
Australian National University. All participants gave informed,
Pattern of alcohol intake
The pattern of alcohol intake and its clinical associations have been
fully documented in an earlier report.11 The pattern of alcohol
intake, by quantity and type, is shown in Box 1. On a day when alcohol was
consumed, 40% of all men and 45% of all women took one or two drinks, 23%
and 7% took three or four drinks, and 15% and 1%, respectively, took
five or more drinks.
Death occurred in 450 men (36%) and 392 women (25%). Consumption of
more than 14 drinks/week was uncommon. Hence, where relevant, the use
of alcohol has been grouped into zero use and any use.
Alcohol intake: Age-specific all-causes mortality by alcohol
intake is presented in Box 2. Alcohol use in men appeared to be
associated with reduced mortality up to age 74 years, but not beyond.
In women, its use was associated with reduced mortality in all groups
older than 64 years.
Predictors of all-causes mortality: The
independent contribution of alcohol to all-causes mortality was
explored in men 60-74 years and in all women in proportional hazards
models which adjusted for the presence of major demographic,
psychosocial and cardiovascular variables at study entry. The
significant predictors of all-causes mortality are summarised in
Box 3. (Alcohol intake was not a significant predictor of mortality in
men aged more than 74 years.) Any alcohol intake was significantly
associated with reduced all-causes mortality in both sexes.
Quantity or type of alcohol intake: The relationship between
quantity or type of alcohol intake and all-causes mortality in the
proportional hazards model is presented in Box 3. The risk of
mortality was significantly reduced at all levels of alcohol intake,
except in women taking 15-28 drinks/week (representing only 3% of
women in the study). A similar degree of reduction in all-causes
mortality was observed at all levels of alcohol intake. The
protection observed was broadly similar in those using beer versus
wine/spirits, although this only reached statistical significance
for beer consumption.
Pattern of alcohol intake: In a subsequent model, the quantity of
alcohol consumed per week was replaced by a variable denoting the
usual number of drinks taken on a given day, a measure of the pattern of
drinking. Using zero intake as the reference group, the hazard ratio
in men with a consumption of one or two drinks on a given day was 0.64 (95%
CI, 0.46-0.89), with three or four drinks 0.68 (95% CI, 0.47-0.98),
and with five or more drinks 0.69 (95% CI, 0.45-1.06). The
corresponding hazard ratios in women were 0.74 (95% CI, 0.59-0.93),
0.68 (95% CI, 0.38-1.22) and 1.38 (95% CI, 0.58-3.29) (there were only
17 women in this group).
Hazard curves: The hazard curves calculated from the
proportional hazards models in men and women are presented in Box 4. By
the end of almost 10 years' follow-up, men taking any alcohol lived on
average 7.6 months longer, and women on average 2.7 months longer,
than their counterparts taking no alcohol.
Specific causes of death: Cardiovascular death (ie, CHD and stroke)
was reduced from 20/100 (95% CI, 14-26) in non-drinkers to 11/100 (95%
CI, 9-13) in men taking any alcohol. The corresponding reduction in
cardiovascular death in women was from 16/100 (95% CI, 13-19) to 8/100
(95% CI, 6-10). Deaths attributed to any cancer were unchanged in men
(6/100 [95% CI, 3-9] versus 7/100 [95% CI, 5-9]) and in women (4/100
[95% CI, 2-6] versus 5/100 [95% CI, 3-7]).
Gross cost for alcohol per life-year gained
From the number of life-years added to survival for each quantity of
intake, we have estimated the gross cost for alcohol per life-year
gained. In men 60-74 years and in women 60 years and over taking 1-28
drinks/week, the respective costs were $13 000 and $31 000 per
life-year gained. Since much of the benefit from alcohol intake was
observed at a moderate intake of only 1-7 drinks/week, the respective
costs at this intake were $5700 and $19 000 per life-year gained.
In this well-defined, community-based sample of rural elderly
Australians, alcohol intake could be described as moderate rather
than heavy.11 Our results confirm that
any intake of alcohol is associated with significantly reduced
all-causes mortality in men 60-74 years and in all elderly women,
consistent with our previous report at 77 months'
significance has now been reached due to the greater number of deaths
and increased statistical power. Absolute death rates in men were
substantially higher than in women, especially in the "young old".
This would account for the greater average survival advantage shown
in Box 4 (7.6 months versus 2.7 months). Equivalent reductions in
mortality occurred at 1-7 drinks/week and at higher intakes. In men
there was no evidence of a differential effect between 1-2 drinks on a
given day and an intake of five or more drinks on a given day.
Not all studies have documented an association between all-causes
mortality and alcohol intake.16,17 The reduction in CHD
and stroke mortality associated with alcohol use we observed is
consistent with findings of previous reports,8,18,19 as well as
those of recent reports in large middle-aged cohorts from the United
States.4,5,20 These data are
gradually causing health authorities to reconsider public policy on
moderate alcohol intake, say 1-7 drinks/week, in the prevention of
future morbidity and mortality.21 Studies in younger
populations indicate a "U"- or "J"-shaped relationship between
alcohol intake and all-causes mortality.2 This has not been a general
finding in the elderly, possibly because these cohorts contain an
excess of "healthy survivors".10 It is surprising that we
could find no relationship between alcohol intake and mortality in
men aged over 74 years, but it is plausible that very elderly men lose
the benefit of alcohol intake because they become subject to
competing causes of mortality.9
There is a potential for misclassification of alcohol intake between
zero and low intake because of under-reporting. This would diminish
any apparent relative benefit of alcohol intake on all-causes
mortality. Hence, our statistically significant findings may
represent a minimum estimate of the benefit of moderate alcohol
intake. We have demonstrated essentially a "threshold effect"
between alcohol intake and all-causes mortality in either sex.
Protection does not improve greatly as alcohol intake increases
further (Box 3). Thus, under-reporting would then have less impact on
our findings. This threshold effect is important, particularly if we
were to move to a public health position of suggesting that abstainers
should begin to imbibe!
The economic findings take no account of any changes in healthcare
costs arising downstream through the benefits or otherwise of
alcohol intake. What we regard as an acceptable cost per life-year
gained is arbitrary, but it is informative to compare our calculated
costs with, for example, recently published gross costs for the use of
simvastatin in patients with established CHD, with survival
increased by around 20%.22 Assuming such patients
receive lifetime therapy with simvastatin from their mid-50s, the
gross cost per life-year gained in the UK population would be £5100 (or
about $13 000). Although one is comparing unrelated "therapy",
moderate alcohol intake, at least in older men, may turn out to be a
popular and cost-effective means of improving survival which does
not require government subsidy!
The relative merits of wine versus other forms of alcohol consumption
remain controversial. Data from France6 and Denmark7 highlight
specific benefits of wine, and this has generated a new research
effort to identify which components of wine, apart from alcohol, may
be the most beneficial. Antioxidants are among the most prominent
suggestions.23 Others consider that
alcohol in any form gives protection against cardiovascular
disease,8 largely through its effect
in raising high density lipoprotein (HDL) cholesterol
levels.11 In Dubbo, the quantity of
alcohol intake was highly correlated with HDL cholesterol (r =
0.32, P < 0.001 and r = 0.23, P <
0.001 in men and women, respectively). Other suggested
mechanisms for cardiovascular protection include favourable
effects of alcohol on thrombotic and fibrinolytic pathways, reduced
insulin resistance and improved endothelial function through
increased nitric oxide production.5 Alcohol may also influence
survival in ways which currently defy measurement: regular alcohol
intake may reflect a special lifestyle integrated with less tangible
Although excess alcohol intake is undoubtedly toxic to the central
nervous system, recent studies suggest that a moderate intake may
reduce the risk of dementia. In a study of elderly French people 65
years and over, appropriately from Bordeaux, the rate of
hospitalisation for dementia in non-drinkers over three years was
4.9/100, but only 3.9/100 in those taking any alcohol (with a more
striking effect on the risk of Alzheimer's disease).24 During 116
months' follow-up in the Dubbo population, the respective rates of
hospitalisation for dementia were 4.3/100 and 2.5/100 (P <
0.01). It is premature to promote the use of alcohol for
prevention of dementia, but the 21st century may witness a completely
new role for alcohol in health.
The Dubbo Study is supported in part by grants from the National Health
and Medical Research Council of Australia, Astra Pharmaceuticals
Pty Ltd, Amrad Pharmaceuticals Pty Ltd, Bristol-Myers Squibb
Australia Pty Ltd, Merck Sharp & Dohme Australia Pty Ltd, Parke Davis
Pty Ltd and Pfizer Pty Ltd. We acknowledge the dedication of the Dubbo
Nurse-Manager Kerrie Pearson, the assistance of Gina Brinsmead in
economic analysis and Helen Adams in preparation of the manuscript.
- Klatsky AL, Armstrong MA, Friedman DD. Alcohol and mortality.
Ann Intern Med 1992; 117: 646-654.
Holman CDJ, English DR, Milne E, Winter MG. Meta-analysis of
alcohol and all-cause mortality: a validation of NHMRC
recommendations. Med J Aust 1996; 164: 141-145.
Thun MJ, Peto R, Lopez AD, et al. Alcohol consumption and mortality
among middle-aged and elderly US adults. N Engl J Med 1997;
Albert CM, Manson JE, Cook NR, et al. Moderate alcohol consumption
and the risk of sudden cardiac death among US male physicians.
Circulation 1999; 100: 944-950.
Berger K, Ajani UA, Kase CS, et al. Light-to-moderate alcohol
consumption and the risk of stroke among US male physicians. N Engl
J Med 1999; 341: 1557-1564.
Renaud S, Geuguen R, Siest G, Salamon R. Wine, beer, and mortality in
middle-aged men from Eastern France. Arch Intern Med 1999;
Gronbaek M, Deis A, Sorensen TIA, et al. Mortality associated with
moderate intakes of wine, beer or spirits. BMJ 1995; 310:
Rimm EB, Klatsky A, Grobbee D, Stampfer MJ. Review of moderate
alcohol consumption and reduced risk of coronary heart disease: is
the effect due to beer, wine or spirits? BMJ 1996; 312:
Van de Water HA, Boshuizen HC. The impact of substitute morbidity
and mortality on public health policies. Leiden: TNO Prevention and
Health, Division of Public Health and Prevention, 1995.
Scherr PA, LaCroix AZ, Wallace RB, et al. Light to moderate alcohol
consumption and mortality in the elderly. J Am Geriatr Soc
1992; 40: 651-657.
Simons LA, Friedlander Y, McCallum J, Simons J. Alcohol intake and
survival in the elderly: a 77 month follow-up in the Dubbo Study.
Aust N Z J Med 1996; 26: 662-670.
Simons LA, McCallum J, Friedlander Y, et al. Dubbo Study of the
elderly: sociological and cardiovascular risk factors at entry.
Aust N Z J Med 1991; 21: 701-709.
Simons LA, McCallum J, Friedlander Y, Simons J. Predictors of
mortality in the prospective Dubbo Study of Australian elderly.
Aust N Z J Med 1996; 26: 40-48.
Risk Factor Prevalence Study Management Committee. Risk Factor
Prevalence Study: Survey No 3 1989. Canberra: National Heart
Foundation Australia and Australian Institute of Health, 1990.
SPSS for Windows NT [computer program], version 9.0. Chicago,
Ill: SPSS Inc, 1999.
Leino EV, Romelsjo A, Shoemaker C, et al. Alcohol consumption and
mortality. II. Studies of male populations. Addiction 1998;
Hart CL, Smith GD, Hole DJ, Hawthorne VM. Alcohol consumption and
mortality from all causes, coronary heart disease, and stroke:
results from a prospective cohort study of Scottish men with 21 years
follow up. BMJ 1999; 318: 1725-1729.
Hennekens CH, Willett W, Rosner B, et al. Effects of beer, wine and
liquor in coronary deaths. JAMA 1979; 242: 1973-1974.
Stampfer MJ, Colditz GA, Willett WC, et al. A prospective study of
moderate alcohol consumption and the risk of coronary disease and
stroke in women. N Engl J Med 1988; 319: 267-273.
Sacco RL, Elkind M, Boden-Albala B, et al. The protective effect of
moderate alcohol consumption on ischemic stroke. JAMA 1999;
Hommel M, Jaillard A. Alcohol for stroke prevention? N Engl J
Med 1999; 341: 1605-1606.
Pickin DM, McCabe CJ, Ramsay LE, et al. Cost effectiveness of
HMG-CoA reductase inhibitor (statin) treatment related to the risk
of coronary heart disease and cost of treatment. Heart 1999;
Frankel EN, Kanner J, Germann JB, et al. Inhibition of oxidation of
human low-density lipoprotein by phenolic substances in red wine.
Lancet 1993; 341: 454-457.
Orgogozo J-M, Dartigues J-F, Lafont S, et al. Wine consumption and
dementia in the elderly: a prospective community study in the
Bordeaux area. Rev Neurol (Paris) 1997; 153: 185-192.
(Received 23 Dec 1999, accepted 10 Apr 2000)
University of New South Wales Lipid Research Department, St
Vincent's Hospital, Sydney, NSW.
Leon A Simons, MD, FRACP, Associate Professor of Medicine;
Judith Simons, MACS, Analyst-Programmer.
Faculty of Health, University of Western Sydney MacArthur, Sydney,
John McCallum, DPhil, Professor and Dean.
Department of Social Medicine, Hebrew University - Hadassah
Hospital, Jerusalem, Israel.
Yechiel Friedlander, PhD, Associate Professor in
Pfizer Pty Ltd, Sydney, NSW.
Michael Ortiz, PhD, Health Outcomes Manager.
Reprints will not be available from the authors.
Professor L A Simons, Lipid Research Department, St Vincent's
Hospital, Darlinghurst, NSW 2010.
|Back to text|
|Back to text|
|3: Proportional hazards model of all-causes
mortality and alcohol intake in elderly subjects
||Hazard ratio (95% CI)
||Men 60-74 years
||Women 60+ years
|Significant predictors of all-causes mortality
|Any alcohol intake
Blood pressure medication
Poor expiratory flow
| 0.75 (0.60-0.94)
| Relationship of quantity or
type of alcohol to all-causes mortality
| The reference category for alcohol
usage was zero intake. Other variables in the models were body mass index,
family history of coronary heart disease (CHD), prevalent CHD, blood pressure,
lipid levels, self-rated health, and physical disability. Poor expiratory
flow refers to peak expiratory flow tertile I.
|Back to text|
|Back to text|