Education modifies the effect of alcohol on memory impairment: the third national health and nutrition examination survey.

We examined whether the relation of alcohol consumption to prevalence of verbal memory impairment was modified by education among 4,804 elderly subjects in the Third National Health and Nutrition Examination Survey. Verbal memory was assessed using delayed recall, with impairment defined as a combined score <4. Alcohol consumption over the previous month prior to the interview was assessed using a food frequency questionnaire. Prevalence of verbal memory impairment decreased from 11.3 to 7.2, 5.7, 5.1 and 4.4% in increasing categories of alcohol consumption (none, 1-4, 5-14, 15-30 and >30 drinks per month) in men, and from 7.2 to 3.5 and 2.8% (for none, 1-14, and >14 per month) in women, respectively. Adjusting for age, race, and other factors, prevalence ratios of verbal memory impairment decreased with each increasing alcohol category, but the effect was attenuated when further adjusted for education. There was a much stronger protection from alcohol among subjects with more education: prevalence ratios were reduced from 1.0 to 0.2 to 0.1 for non-drinkers, 1-14, and >14 drinks/month, respectively (p for trend = 0.007). Our results suggest that alcohol intake is associated with a greater decrease in the prevalence of verbal memory impairment among more educated subjects than among those with less education, possibly related to differences in drinking patterns.