Amount and type of alcohol and risk of dementia

Subjects and methods.

The CCHS has previously been described in detail.25 In brief, the study was initiated in 1976; 19,698 persons living in central Copenhagen (Nørrebro and Østerbro) were invited to participate. The subjects were re-examined for the third study examination in 1991 to 1994, and those aged 65 years or more were invited to undergo the Mini-Mental State Examination (MMSE).26 There were 2,784 persons who at this examination were aged 65 years or older (figure). Seven eligible participants died and 832 did not respond to the invitation, giving a response rate of 70.0% (1,945/2,777). In 53 subjects the MMSE could not be evaluated because the person did not want to participate (21) or owing to severely impaired sight or hearing loss, or because the person was aphasic (32). Thus, 1,892 (68.1%) of the originally invited subjects were tested at the third CCHS examination, and of these, 277 (14.6%) were interviewed in their homes.

• Download figure
• Open in new tab
• Download powerpoint

Figure. Flow chart for dementia examination in the Copenhagen City Heart Study, 1991–1994.

A total of 261 subjects had a MMSE score of 24 or less. They were all invited to a further detailed interview examining whether they had dementia. Fifty-one did not wish to continue, leaving 210 (80.5%) for the final dementia examinations. These included blood samples, the Cambridge Examination for Mental Disorders of the Elderly,27 the Hachinski Ischemia Score,28 the Hamilton Depression Scale,29 and the global deterioration scale for assessment of primary degenerative dementia.30 Dementia was diagnosed according to the American Psychiatric Association’s criteria in the Diagnostic and Statistical Manual of Mental Disorders (DSM-III-R).31 AD was diagnosed according to the guidelines from the working group of the National Institute of Neurologic and Communicative Disorders and Stroke and the AD and Related Disorders Association.32 A diagnosis of VaD was given when the Hachinski Ischemia Score was ≥7 and according to the DSM-III-R criteria.31 The examiners for possible dementia were blinded to the answers the subjects gave at the first health examination.

Results.

A total of 83 cases with dementia were diagnosed; the remaining 1,626 were classified as controls. Out of those with dementia, 40 (48.2%) had AD, 15 (18.1%) VaD, 11 (13.3%) dementia from other causes, and in 17 (20.5%) the cause was unknown. There were a total of 1,062 women and 647 men, and 45 (4.2%) women and 38 (5.9%) men were demented.

Men had a higher weekly alcohol intake than women (table 1). One-third of the women consumed less than one unit of alcohol per week, as compared to only 7.4% of men reporting a low intake (p < 0.01). Especially for beer, men had a more frequent intake than women, of whom more than half reported never or hardly ever drinking beer (p < 0.01), and more men than women were drinking spirits (p < 0.01), whereas there was no significant difference for the intake of wine (p = 0.28). Cases were more often living alone, had fewer years of education, and, especially for women, there were few in the high income group.

Discussion.

In this study, monthly intake of beer was associated with a significantly higher risk of dementia, whereas monthly and weekly intake of wine was associated with significantly lower risk of dementia. We failed to detect a significant association between weekly alcohol intake and frequency of spirits consumption on risk of dementia.

Case-control studies of the relation between alcohol intake and risk of dementia are potentially susceptible to recall bias, as is also the case when the information is obtained from close relatives. Prospective studies may present a better opportunity to examine the association between alcohol and type of alcohol consumption and risk of dementia. By conducting the case-control nested in a cohort study, we were able to specifically identify the demented subjects, while avoiding recall bias by using data collected 15 years before dementia was diagnosed.

Two longitudinal studies from the Honolulu Heart Program and the Zutphen Elderly Study found that subjects with a moderate intake of alcohol had a better cognitive status than abstainers and heavy drinkers.13,14⇓ In both studies the endpoint was cognitive function, which differs considerably from the diagnosis of dementia. For example, in our study, 210 had an MMSE score of 24 or less, but only 83 were eventually diagnosed with dementia. Subjects with a low MMSE score but who were not diagnosed with dementia were found to have other psychiatric conditions, such as depression and premorbid mental impairment. Others were normal. The MMSE test has a score range from 0 to 30; we used a cutoff point of 24, which has been reported to provide a good compromise between sensitivity and specificity, with values around 90% for both.36 The optimal cutoff score is likely to vary between studies depending on the educational level of the tested subjects.36,37⇓ Further examination of subjects with low MMSE scores increased the specificity, whereas some mildly demented subjects may not have been detected in the current study. This misclassification will most likely attenuate the associations examined.

The time lag between the first and the third study examination ensures that institutionalization of demented persons, which has been suggested as a potential bias in studies of wine intake and risk of dementia,38 is an unlikely explanation for the results in the current study. Nevertheless, as in other studies, nonparticipation in the examinations could introduce bias. The different associations between beer, wine, and spirits found in our study indirectly suggest that that has not occurred. If nonparticipation plays a role explaining our results, it would mean that subjects who drank wine and were demented did not participate in the third examination, whereas those who drank beer and spirits and were demented did participate. To further examine the possible effect of nonparticipation, we analyzed the differences between nonparticipants vs participants. There were no differences with regards to age, male to female ratio, or weekly or daily intake of beer and spirits. Weekly or daily intake of wine was marginally higher in participants, which could be expected as previous studies from the CCHS have shown that mortality is lower for people drinking wine and drinking moderately.39 This result suggests that an alternative hypothesis for the results, that people drinking wine have died before the third study examination, can be excluded. Based on these observations, nonparticipation seems an unlikely explanation for the results, although it should be noted that information on drinking habits is limited to those who participated in the first study examination.

In the French PAQUID study the authors found that intake of wine was associated with a lower risk of dementia.23 The follow-up period was limited to 3 years and it is possible that subjects who later were found to have dementia already had changed their drinking habits as a result of early cognitive deterioration. Dementia often has an insidious onset where the deterioration advances over several years. The olfactory system may be affected at an early stage,40 which can change dietary habits, and would affect the answers a person gives at a health examination. In this study the participants were asked about their type of alcohol intake years before the interview, and few if any demented subjects at the first study examination would have survived the time to the third study examination. Thus, the responses included in this study are likely to be unbiased with regard to possible dementia.

The weekly intake of alcohol was divided into five categories and none of them was significant in analyses adjusting for potential confounders. The rationale was to examine whether there was an indication of a J-shaped relationship between alcohol intake and risk of dementia. The current results do not support such a relation and it should be noted that although a high intake of alcohol was not significantly associated with increased risk of dementia this could merely be the result of low power to detect a significant difference rather than an indication of no relation. In fact, the point estimates suggest that there may be a higher risk of dementia in subjects drinking 15 units or more per week. Similar reservations should be directed to the results for intake of beer and spirits. For example, in the analyses of beer, only a monthly intake was significant, but the estimates for both weekly and daily consumption indicated a doubling of the risk of dementia. Also, the categorization of alcohol frequency intake raises the issue of multiple comparison as each consumption group was compared to the referent category. A formal test for multiple comparisons was not made as the current results mainly relate to the pattern between alcohol intake and risk of dementia rather than between specific intake frequencies and risk of dementia.

The main argument for a possible relation between wine and dementia is through an effect of flavonoids and their antioxidation activity,21 and not because of the alcohol content. A previous study showed that the inverse association between wine consumption attenuated after adjustment for flavonoids suggesting that it was an intermediate variable between wine and dementia.15 Flavonoids are also found in other food items such as tea, fruits, and vegetables, and we had no data on the participants’ dietary habits. Previous studies from a large Danish cohort have shown that wine drinkers have a healthier diet than that of subjects drinking beer or spirits.41 They had a higher intake of fruit, fish, vegetables, and salad, tended not to use fats on bread, and used olive oil for cooking more frequently. To our knowledge, no studies have supported a strong effect of diet on risk of dementia,42 but it remains a hypothetical bias. Also, subjects who frequently drink beer often have low intake of especially B vitamins, where especially thiamine deficiency is associated with an increased risk of dementia. However, although beer and spirits drinkers had a more unhealthy diet as compared to wine drinkers in the Danish study, there were only minor differences between the two groups. Should diet alone explain the results in the current study, a substantial difference in dietary habits between beer and spirits drinkers would be expected. We have no satisfactory explanation for the increased risk of dementia associated with intake of beer, whether it is directly related to the consumption or some unmeasured bias.

As our aim of this study was to assess the relation between intake and type of alcohol and risk of dementia, we included all types of dementia. The pathology of the different types of dementia varies considerably, and although we used standard criteria to differentiate among subjects with AD, VaD, and dementia from other causes, it should be emphasized that these diagnostic criteria allow only a diagnosis of probable AD. It is known that a proportion of patients diagnosed with VaD using these criteria have additional Alzheimer changes at autopsy. The results of our subanalyses including only AD were suggestive of no difference as compared to all dementias combined, but they were not conclusive.

Drinking patterns are likely to differ for the intake of beer, wine, and spirits. Whereas wine consumption often is associated with meals and a moderate intake, beer and spirits may more often be consumed at other times during the day or on certain days of the week. It is possible that a sudden high intake of alcohol may have especially deleterious effects on cognitive function. Unfortunately, we had no information on drinking patterns, but a recent study from Copenhagen suggested that binge vs steady drinking of beer and wine was not a strong confounder.43 It could also be hypothesized that the association for wine could be due to no intake of beer or spirits. However, our study population was characterized by a considerable overlap of intake of beer, wine, and spirits, and absence of intake of one type of alcohol is an unlikely explanation for the results.

In this study, a moderate intake of wine was inversely associated with risk of dementia, whereas there was no association between intake of spirits or weekly alcohol consumption. It is emphasized that these results do not suggest that the population should start drinking wine or increase wine consumption. Especially in the elderly, alcohol intake may be deleterious owing to the normal loss of lean body mass in which to distribute water-soluble alcohol and the risk of interactions between medication and alcohol.44 The results suggest, however, that there might be substances in wine that can reduce the occurrence of dementia.