Alcohol consumption and cognitive performance in a random sample of Australian soldiers who served in the second world war

Introduction

Epidemiological evidence of the differential age related risks and possible benefits of drinking has recently prompted reassessment of sensible levels of drinking and their appropriateness for different age groups.1 2 The potential effects of alcohol consumption on cognitive performance in elderly people are of particular interest in this regard.3

The Australian National Health and Medical Research Council set its recommendations for responsible alcohol consumption on the basis of a wide range of consequences of excessive drinking among younger adults rather than elderly people.4 However, the national health survey has shown that 8.5% of Australian men aged 65-74 years who consumed alcohol drank at the hazardous level (40-60 g per day) and 5.7% at the harmful level (60 g per day).5 We examined the association between the average daily alcohol consumption in 1982 of a random sample of Australian male veterans of the second world war and performance in a range of intellectual tests and the degree of brain atrophy on computed tomography nine years later.

Subjects and methods

This study arose from a project on morbidity among war veterans which began in 1982 but did not include assessment of cognitive performance at that time.6 Participants were chosen randomly from former members of the Australian army who had fought in the second world war and were living in Sydney.6 The response rate was 87%. Survivors were traced in 1991 for the second phase of the study.

Self reported average daily alcohol intake was assessed in both 1982 and 1991 by the quantity-frequency method.7 In 1991, 18 standard neuropsychological tests were used to assess several aspects of cognitive performance (see table 1).8 9 10 11 12 13 14 15 16 17 18

Non-contrast computed tomography was performed on 201 veterans in 1991 using a Siemens Somatom DR3 scanner (Germany) at a slice thickness of 8 mm. Cortical, sylvian, and vermian atrophy were recorded as none, slight, moderate, or severe by a radiologist (SCK) who was unaware of subjects' alcohol consumption or neuropsychological test scores.

As the distribution of alcohol consumption was strongly positively skewed, correlations with the neuropsychological tests were assessed by both the Spearman rank coefficient (ρ) on the original values of alcohol intake and the Pearson product moment coefficient (r) applied to log transformed values. The possibility of a non-linear association between alcohol consumption and results of the cognitive function tests was examined by fitting quadratic curves. Contingency tables and the χ² test were used to examine the significance of differences between percentages. The t test for related samples was used to assess the significance of change in average daily consumption between 1982 and 1991.

To estimate the power of the sample we regarded an r of at least 0.2 as the minimum effect size we would wish to identify. At this effect size and with significance set at 0.05 the power of the sample of 209 would be at least 0.8.19

Results

Of the 342 veterans studied in 1982, 96 had died by 1991, 10 had moved away from Sydney, eight could not be located or were too ill to attend, and 19 refused, leaving 209 consenting participants. In 1982 the average daily consumption of the 178 veterans (85%) who drank alcohol at least once a week ranged from 2 g to 129 g ethanol (mean 38.2 g); 60% (106/178) drank at the safe level, 19% (34/178) at the hazardous level, and 21% (38/178) at the harmful level, the proportions drinking at the hazardous and harmful levels being appreciably higher than those in the Australian male population aged 55 to 74 (9.8% and 7.7% respectively).5 The average period over which they had drunk alcohol ranged from 11 to 56 years (mean 44 (SD 5.6) years).

No significant correlation was found between average daily alcohol consumption in 1982 and any of the cognitive performance measures in 1991 (table 1), and in no case was there evidence of a significant non-linear association. There was no correlation between age and alcohol consumption (ρ=-0.05, P=0.39).

The proportion of drinkers fell from 85% in 1982 to 67% in 1991 (McNemar χ²=29.76, P<0.0001). Among those who drank in 1982 and were still drinking in 1991, average daily consumption fell from 43.7 g to 30.5 g (paired t=6.04, P<0.001). However, among drinkers the correlation between consumption in 1982 and 1991 was moderately strong (r=0.58, P<0.0001), and in 1991, 65% (91/140) drank at the safe level, 26% (37/140) at the hazardous level, and 9% (12/140) at the harmful level. There was no association, linear or non-linear, between alcohol consumption in 1991 and any of the measures of cognitive performance.

The proportions of veterans with moderate or severe cortical, sylvian, or vermian atrophy on computed tomography in 1991 did not differ significantly between those who were non-drinkers and drinkers in 1982; neither was there any difference in relation to alcohol consumption (table 2).

Our failure to find an association between alcohol consumption and cognitive performance or brain atrophy in these men could be a result of differential mortality or loss to follow up. However, the proportion of men who died or were not included in the 1991 survey did not differ significantly according to alcohol consumption in 1982, whether or not stratified by age.

Discussion

Several recent studies of elderly social drinkers have found no association or only weak associations between cognitive performance and alcohol consumption measured either concurrently or retrospectively.20 21 22 23 24 25 Furthermore, any correlations found have tended to disappear when other factors were controlled statistically. We know of only one other prospective study of the effects of alcohol consumption on cognitive performance in elderly people living in the community.26

The elderly veterans we studied were originally recruited in 1982 for another purpose that did not entail measuring cognitive performance, so we could not use cognitive decline between the two studies as an outcome variable. Instead we considered whether apparently high lifelong alcohol consumption during adulthood among a comparatively large number of men in our sample was predictive of diminished cognitive performance in old age. It is commonly believed that many Australian veterans of the second world war have continued to drink comparatively heavily during the 50 years since the war. This continued heavy drinking might be the result of habits acquired during the war, mateship, a desire to suppress memories of wartime experiences, or the atmosphere of the ubiquitous clubs of the Returned Serviceman's League. Compared with Australian men of the same age, many of the veterans in our study had sustained a high rate of alcohol consumption into old age. If alcohol consumption were clearly associated with cognitive performance then the relation would be expected in this group of men. However, we found no association between consumption in 1982 and brain atrophy on computed tomography or several aspects of cognitive functioning almost a decade later in a group whose cognitive performance was within the normal range for people of their age. There is no basis here for predicting future cognitive performance from level of alcohol consumption in older men, even though many had apparently experienced a lifetime of persistent, moderately heavy consumption.