## Abstract

Some, but not all, observational studies have suggested that moderate levels of alcohol intake may be associated with improved cognitive function and reduced risk of cognitive decline and dementia. The authors of this 1996–2002 study used data from the Women’s Health Initiative Memory Study of postmenopausal combination hormone therapy to assess cross-sectional and prospective associations of self-reported alcohol intake with cognitive function. Across 39 US academic medical centers, 4,461 community-dwelling women aged 65–79 years were followed an average of 4.2 years with annual Modified Mini-Mental State Examinations and standardized protocols for detecting mild cognitive impairment and probable dementia. Compared with no intake, intake of ≥1 drink per day was associated with higher baseline Modified Mini-Mental State Examination scores (p < 0.001) and a covariate-adjusted odds ratio of 0.40 (95% confidence interval: 0.28, 0.99) for significant declines in cognitive function. Associations with incident probable dementia and mild cognitive impairment were of similar magnitude but were not statistically significant after covariate adjustment. Associations with intakes of <1 drink per day were intermediate. Moderate levels of alcohol intake may be associated with better cognition and reduced risk of significant cognitive decline; however, confounding associations with unmeasured factors cannot be ruled out.

Received for publication January 31, 2004; accepted for publication August 31, 2004.

Recent longitudinal community cohort studies suggest that moderate alcohol intake may benefit cognition. For example, the Epidemiology of Vascular Aging Study (participants aged 59–71 years; 59 percent female) found that mean scores on the Mini-Mental State Examination (1), a measure of global cognitive function, were higher among women reporting moderate levels of alcohol intake compared with no intake (2). The Atherosclerosis Risk in Communities Study (subjects aged 45–69 years; 55 percent female) found that current drinkers, most of whom drank lightly to moderately, had higher cognitive scores than nondrinkers on three cognitive tests, with a clear dose-response association on a word fluency test and an inverted U-shaped response on delayed word recall and digit symbol substitution tests (3). These findings generally agree with those from other cohorts of women (4–6).

Evidence is growing that moderate levels of alcohol intake are associated with a reduced risk of dementia. A case-control study conducted within the Cardiovascular Health Study (participants aged 65–97 years; 60 percent female) found that the odds ratio for incident dementia was 0.37 (95 percent confidence interval (CI): 0.23, 0.60) for subjects reporting consumption of 1–6 drinks per week compared with nondrinkers; for those reporting 7–13 drinks per week, the odds ratio was 0.64 (95 percent CI: 0.36, 1.13) (7). Similarly, the Rotterdam Study (participants aged 60–90 years; 52 percent female) reported hazard ratios for dementia (relative to nondrinkers) of 0.82 (95 percent CI: 0.56, 1.22) for <1 drink per week, 0.75 (95 percent CI: 0.51, 1.11) for 1–6 drinks per week, and 0.58 (95 percent CI: 0.38, 0.90) for 1–3 drinks per day (8). A positive relation has not been detected consistently, however (9–12). For example, the Established Populations for Epidemiologic Studies of the Elderly program (subjects aged ≥65 years; 56 percent female) reported an odds ratio of 1.1 (95 percent CI: 0.8, 1.5) for Alzheimer’s disease comparing consumption of 1 ounce (28.3 g) of alcohol per day with no consumption (9).

We used data from a large, randomized clinical trial of postmenopausal combination estrogen and progestin therapy, which involved annual assessment of cognition and dementia status over an average of 4.2 years of follow-up, to address several goals. We examined associations of cross-sectional global cognitive function with clinically significant decreases in global cognitive function and with the incidence of probable dementia and mild cognitive impairment. In this paper, results from analyses with and without control for demographic/socioeconomic factors and cardiovascular risk factors are presented. Because both alcohol intake and hormone therapy may influence cognition by altering levels of lipoproteins and/or coagulation and have been reported to interact with respect to their association with cognitive function (13), we also assessed whether the influence of combination hormone therapy on cognition varied according to alcohol intake.

## MATERIALS AND METHODS

The Women’s Health Initiative Memory Study (WHIMS) (14) is an ancillary study to the Women’s Health Initiative trials of hormone therapy (15), two large, randomized, double-blind, placebo-controlled, clinical trials of hormone-related outcomes. WHIMS was conducted to assess the relative effect of 0.625 mg/day of conjugated equine estrogen alone or in combination with 2.5 mg/day of continuous medroxyprogesterone acetate on the incidence of dementia and global cognitive functioning in postmenopausal women. Following discovery of an unfavorable risk-to-benefit ratio of its noncognitive endpoints (16), administration of study drugs in the Women’s Health Initiative conjugated equine estrogens plus medroxyprogesterone acetate (E + P) trial was discontinued (July 2002). This decision also ended the E + P trial of WHIMS, which subsequently reported an adverse association between E + P and both dementia and cognitive function (17, 18). We analyzed data from this trial.

The study design, eligibility criteria, and recruitment procedures of the Women’s Health Initiative E + P trial have been described elsewhere (15). Beginning in 1996, women aged 50–79 years were enrolled if they had an intact uterus and were postmenopausal. Women were excluded if there were concerns about competing risks, safety, or adherence/retention (including an assessment of alcoholism by staff). Women were assigned randomly with equal probability to take one daily tablet that contained either 0.625 mg of conjugated equine estrogen with 2.5 mg of medroxyprogesterone acetate (PREMPRO; Wyeth Pharmaceuticals, Collegeville, Pennsylvania) or a matching placebo.

Participants in the WHIMS E + P trial were recruited between May 1996 and December 1999 from women in the Women’s Health Initiative E + P trial who were at least 65 years of age and free of dementia, as ascertained by the WHIMS protocol (14). Written informed consent was obtained. The National Institutes of Health (Bethesda, Maryland) and institutional review boards approved the protocol and consent forms.

Baseline alcohol intake was assessed by using the Women’s Health Initiative food frequency questionnaire (19), which queried intake (beer, wine, and liquor separately) over the past 3 months. Alcohol intakes were grouped as none, <1 drink per day, and ≥1 drink per day. Self-reported alcohol intake has correlations with other measures of alcohol intake of about r = 0.6–0.7 (20). The correlation of self-reported intake with the underlying “true” intake may be roughly estimated as the square root of this correlation, for example, approximately r = 0.80 (21). In a separate questionnaire, women who reported no current intake and responded “yes” to the question, “During your entire life, have you had at least 12 drinks of any kind of alcoholic beverage?” were asked whether intake was stopped for “health problems” or “nonhealth problems.” No information was collected to estimate the extent of lifetime intake.

Global cognitive function was measured with the Modified Mini-Mental State Examination (3MSE) (22) at baseline and annually for up to 6 years. The average time between the first and last examinations (4.2 years) was similar across reported alcohol intakes (p = 0.24). The 3MSE consists of 15 items that sum to 0–100; higher scores reflect better cognitive functioning. Test items measure temporal and spatial orientation, immediate and delayed recall, executive function (control and management of other cognitive processes), naming, verbal fluency, abstract reasoning, praxis, writing, and visuo-constructional abilities. The 3MSE has good reliability, sensitivity, and specificity for detecting cognitive impairment and dementia (23).In our analyses, we assessed associations with baseline 3MSE and with the occurrence, at any time during follow-up, of a decline from baseline of 8 units (2 standard errors (SEs)), which corresponds to a clinically significant decline (18).

3MSE assessments were administered by trained and certified technicians who were masked to other outcomes. Participants who scored below preestablished cutpoints were scheduled for a more extensive neurocognitive assessment and neuropsychiatric examination to determine the presence or absence of probable dementia and mild cognitive impairment, which was centrally adjudicated (14). Regardless of adjudicated dementia status, women continued to be scheduled for their annual 3MSE assessments.

Baseline demographic and clinical factors were collected via self-report and standardized assessments. Factors included in our analyses were those found to be significantly related to the 3MSE (18, 24) and that were expected to be associated with alcohol intake: age; years since menopause; education; ethnicity; family income; use of tobacco; body mass index; history of hypertension, cardiovascular disease, or diabetes; use of 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase inhibitors (statin therapy) or aspirin; prior use of hormone therapy; and intervention assignment.

Differences in baseline demographic and clinical factors among women grouped by alcohol intake were assessed by using chi-square tests and analyses of variance. Cross-sectional analyses associations between the 3MSE and alcohol intake were examined by using analyses of variance and covariance. Women were grouped according to whether they were measured to have had a drop of 8 or more 3MSE units from baseline at any time during follow-up; logistic regression (in which treatment assignment was included as a covariate) was used to examine the relation between this occurrence and baseline alcohol intake, with and without adjustment for demographic/socioeconomic, lifestyle, and clinical factors.

Associations between alcohol intake and incidence of probable dementia and combined probable dementia/mild cognitive impairment were examined with proportional hazards regression and Kaplan-Meier plots. In these analyses, women were censored when they were lost to follow-up or at the trial’s termination.

## RESULTS

At baseline, of the 4,461 women in the study, 1,345 (30.2 percent) reported no alcohol intake, 2,500 (56.0 percent) reported intake of <1 drink per day, and 616 (13.8 percent) reported intake of ≥1 drink per day. Of the women reporting no current alcohol intake, 758 (56.4 percent) reported drinking at least 12 drinks in the past, 111 (8.3 percent) reported having quit drinking because of health problems, and 84 (6.2 percent) reported quitting because of nonhealth problems. Within the group of women reporting ≥1 drink per day, consumption was distributed as 394 (64.0 percent) at 1 per day, 186 (30.2 percent) at 2–3 per day, 21 (3.4 percent) at 4–5 per day, and 15 (2.4 percent) at ≥6 per day. Tables 1 and 2 summarize the distributions of several demographic/socioeconomic, lifestyle, and clinical characteristics of women grouped by level of alcohol intake. Greater alcohol intake was reported by women who were more educated, Caucasian, or currently smoking and those with a higher family income; lower body mass index; no history of hypertension, cardiovascular disease, or diabetes; or no current use of 3-hydroxy-3-methylglutaryl-coenzyme A reductase inhibitors. Women reporting intake of ≥1 drink per day were more likely to have used hormone therapy in the past.

As displayed in table 3, the proportion of nondrinkers scoring below the WHIMS cutpoints for dementia screening at baseline (8.7 percent) was more than double that of women reporting ≥1 drink per day (3.2 percent), with the proportion of occasional drinkers in between (4.9 percent): overall p < 0.001. Baseline mean 3MSE score increased across the three levels of alcohol intake in a graded fashion: for women reporting no alcohol intake, 94.65 (SE, 0.11); for women reporting intake of <1 drink per day, 95.78 (SE, 0.08); and for women reporting ≥1 drink per day, 96.55 (SE, 0.16). These differences remained statistically significant after adjustment for demographic, socioeconomic, lifestyle, and clinical characteristics (p < 0.001). The magnitudes of these differences were attenuated, but remained highly significant, in the full covariate model.

For women reporting no current intake, mean 3MSE scores at baseline were similar for those reporting no prior use compared with prior use: 94.86 (SE, 0.19) versus 94.50 (SE, 0.16), p = 0.49. This difference remained nonsignificant in the full covariate model.

Clinically significant declines in 3MSE scores from baseline occurred for 5.7 percent of women reporting no alcohol intake, 3.3 percent of women reporting <1 drink per day, and 2.3 percent of women reporting ≥1 drink per day. These results are based on the 4,372 (98.0 percent) women who returned for at least one follow-up visit; follow-up was similar among women grouped by baseline alcohol intake (p = 0.30). The numbers of women seen at annual visits 1–6 were 4,282, 4,109, 4,079, 3,683, 1,732, and 84, respectively.

As shown in table 4, relative to those for women reporting no intake, the odds ratios for these declines were 0.56 (95 percent CI: 0.41, 0.77) for women reporting <1 drink per day and 0.40 (95 percent CI: 0.22, 0.72) for women reporting ≥1 drink per day. Full covariate adjustment attenuated these odds ratios to 0.69 (95 percent CI: 0.49, 0.97) and 0.53 (95 percent CI: 0.28, 0.99), respectively; however, the overall differences remained statistically significant (p = 0.042).

Figure 1 depicts the distribution of time from randomization until incidence of probable dementia (n = 61) according to the WHIMS adjudication criteria. The incidence of dementia was inversely ordered according to reported alcohol intake; however, differences among the three groups did not reach statistical significance (table 5: p = 0.09). The fitted hazard ratio for probable dementia, relative to nondrinkers, was 0.64 (SE, 0.17) for women reporting <1 drink per day and 0.41 (SE, 0.20) for women reporting ≥1 drink per day. Differences of marginal statistical significance remained after adjustment for both socioeconomic status/lifestyle and clinical characteristics. Including baseline 3MSE score as a covariate essentially eliminated these differences (p = 0.75).

Figure 2 portrays the time to development of either probable dementia or mild cognitive impairment (whichever appeared first, n = 170). The incidence of this combined endpoint was inversely (and significantly) ordered with respect to alcohol intake (p < 0.001). The relative hazards for the pooled endpoint of mild cognitive impairment or probable dementia were of similar magnitude to those of probable dementia alone: 0.54 (SE, 0.09) for women reporting <1 drink per day and 0.47 (SE, 0.07) for women reporting ≥1 drink per day. Differences were moderated with adjustment for socioeconomic status/lifestyle characteristics (p = 0.13) and were essentially eliminated with adjustment for baseline 3MSE score (p = 0.81).

Alcohol intake did not seem to materially influence the negative effect of E + P on global cognition and probable dementia. The mean difference between 3MSE scores among women assigned to E + P versus placebo, adjusted for baseline 3MSE score, was –0.40 (SE, 0.15) for nondrinkers, –0.22 (SE, 0.11) for women reporting <1 drink per day, and –0.00 (SE, 0.21) for women reporting ≥1 drink per day (interaction p = 0.29). The relative hazard ratios for probable dementia (comparing E + P with placebo assignment) among these groups were 1.70 (SE, 0.70), 2.22 (SE, 0.85), and 4.34 (SE, 4.85), respectively (p = 0.65).

## DISCUSSION

### Cross-sectional associations

WHIMS results are in agreement with other reports of a salutatory effect of moderate alcohol intake on cognition and the risk of dementia. Self-report of moderate alcohol intake was associated with better global cognitive function at baseline. Nondrinkers were more than twice as likely as women reporting ≥1 drink per day to score below the age-adjusted cutpoint for 3MSE screening. Mean baseline 3MSE scores increased in a graded fashion across the three levels of alcohol intake, and differences in these groups remained statistically significant after adjustment for demographic and socioeconomic factors, smoking, and clinical characteristics. The Epidemiology of Vascular Aging Study found a twofold increased odds ratio for high cognitive performance on the Mini-Mental State Examination for women who consumed ≥2 drinks per day compared with nondrinkers (2). Positive associations between moderate levels of alcohol intake and measures of global cognition among women have been reported in other studies (3–5, 25–27);however, no relation has been found in some (12, 28, 29), and at least one has reported a negative association (30).

### Associations with changes in global cognition

Clinically significant declines in global cognition over an average of 4.2 years of follow-up were less common among women who reported alcohol intake at baseline. These differences persisted after statistical adjustment for both demographic/social and clinical factors.

Our findings align with those from other reports. Galanis et al. (31) described a U-shaped relation between alcohol intake and subsequent cognitive decline, with reduced risk for alcohol consumption of ≤2 ounces (56.6 g) per day. Similar U-shaped associations between alcohol intake and subsequent cognition among men were reported by others (4, 9, 32). Leroi et al. (33) found that alcohol intake was associated with less decline among women, but not men. Such findings have not been universal, however. Launer et al. (34) found no protective relation between intake and cognitive decline after adjustment for socioeconomic status and clinical factors, and Edelstein et al. (10) found that alcohol intake was associated with a significant decline in several measures of cognitive function. Peele and Brodsky (35) provide a comprehensive review of research in this area.

### Associations with incident dementia and mild cognitive impairment

We found that moderate alcohol intake was associated with an approximately 50 percent reduced risk of combined probable dementia and/or mild cognitive impairment (p < 0.001). The estimated risk of probable dementia alone was also reduced by about 50 percent; however, this relation did not reach statistical significance (p = 0.11). Differences appeared to emerge 2–3 years following baseline. Similar to our results for cross-sectional cognition and cognitive decline, the association with the combined endpoint could not be accounted for by differences between drinkers and nondrinkers with respect to clinical characteristics. However, it was moderated (and no longer reached statistical significance) after adjustment for demographic and socioeconomic factors. Adjustment for baseline 3MSE score almost completely eliminated this association.

Our estimates closely parallel those of the Cardiovascular Health Study, which reported relative odds of dementia (compared with those for nondrinkers) of 0.65 (<1 drink per week), 0.46 (1–6 drinks per week), 0.69 (7–13 drinks per week), and 1.22 (≥14 drinks per week) (7). Similarly, Ruitenberg et al. (8) found a J-shaped relation between alcohol consumption and risk of dementia. Truelsen et al. (36) reported no overall protective effect of alcohol intake on risk of dementia but found that wine intake (relative to no wine intake) was associated with reduced odds of dementia by factors of 0.43 (monthly), 0.33 (weekly), and 0.57 (daily). Earlier meta-analyses found no association between alcohol intake and risk of Alzheimer’s disease (37). Hebert et al. (38) found a slight, but nonsignificant, increase in incident Alzheimer’s disease associated with moderate alcohol consumption.

### Interaction between E + P therapy and alcohol intake

Tivis et al. (13) reported a benefit of combined hormone therapy on visuo-spatial processing in postmenopausal women (average age: 56 years). This benefit was limited to women who reported low- to mid-moderate alcohol intake (0.01–0.49 ounces (0.28–13.87 g) per day) and apparently reversed among those whose intakes were higher (interaction p = 0.02). We found no evidence of an interaction between assignment to hormone therapy and alcohol intake on global cognition or on the incidence of mild cognitive impairment and dementia.

### Influence of errors in self-report and measurement

It is of concern that self-reported alcohol intake is fallible and that reporting errors may vary by underlying level of cognition. We cannot rule out the possibility that our findings derived from a tendency for women with lower cognitive function to underreport drinking; however, two features of our data argue against this interpretation. We found an association between self-reported alcohol intake and global cognitive function for women who scored above our at-risk screening cutpoints (and thus had little evidence of impairment): of the 1,061 women scoring between the screening cutpoint and 94, 63.2 percent reported alcohol intake compared with 73.4 percent of the 3,132 women scoring 95 or higher (p < 0.001 without covariate adjustment; p = 0.01 with full covariate adjustment). In addition, the magnitude of the associations we described would require unreasonably high levels of underreporting to be spurious. At baseline, odds of alcohol intake decreased by a factor of 0.69 (95 percent CI: 0.64, 0.75) for every 5-unit difference in 3MSE score. Thus, to account fully for this relation, an approximately 30 percent underreporting of any intake for every 5-unit difference would be required, that is, a very large, unidirectional, and systematic difference.

### Potential mechanisms

Moderate alcohol intake may be inversely related to ischemic stroke (39, 40) and to white matter disease and infarcts (41), which may explain decreased rates of vascular dementia (8). A positive association of moderate alcohol drinking with cognitive function may be mediated through vascular factors since the effects of alcohol are stronger on vascular dementia than on Alzheimer’s disease (8). Moderate doses of alcohol may increase prostacyclin concentrations, reduce generation of thromboxane A2, and inhibit platelet function (42–44). They may increase plasma levels of endogenous tissue-type plasminogen activator, a serine protease that regulates intravascular fibrinolysis (45), and fibrinolytic activity while decreasing plasma fibrinogen levels (46). Alcohol is related to increased levels of high density lipoprotein (HDL) cholesterol, its subfractions HDL2 and HDL3, and its associated apolipoproteins A-I and A-II (47–51). The association with HDL cholesterol may account for half of the reduction in coronary events associated with moderate alcohol consumption (52).

In carriers of the apolipoprotein E epsilon-4 genotype (APOE4), plaque formation is increased because of oxidation of apolipoprotein E and binding to beta-amyloid (53). Alcohol may suppress this binding because of its antioxidant effects. APOE4 is associated with lower HDL and higher low density lipoprotein cholesterol; therefore, moderate alcohol consumption may protect cognitive function through its lipid effects in APOE4 carriers. The National Heart, Blood, and Lung Institute twin study found that the protective effect of light alcohol drinking was stronger for APOE4 carriers (54). However, the Epidemiology of Vascular Aging Study found that alcohol drinking was associated with a decreased risk of cognitive decline in those without APOE4, and an opposite association was found in APOE4 carriers (55). The Cardiovascular Health Study found a similar relation for dementia (7).

Alcohol may affect cognition through a release of acetylcholine in the hippocampus. Deficits in the cholinergic system have been linked to problems with attention, learning, and memory (56). Low doses of alcohol may cause a delayed stimulation of hippocampal acetylcholine release in rats (57).

A protective effect of alcohol on cognitive function in moderate drinkers may be due to a relatively poor health status among abstainers or because cognitive status influences alcohol consumption.

### Limitations

Our study has several potential weaknesses. Frank differences exist between women grouped by reported alcohol intake, which may be associated with social status and access to health care. Including socioeconomic status markers as covariates in our analyses could not account for associations with global cognition but did weaken associations with dementia and mild cognitive impairment, which became not statistically significant: it is possible that other unmeasured socioeconomic status and clinical factors may more fully account for these differences. It is also possible that a woman’s cognitive state may affect whether she chooses to drink; women with lower levels of cognitive function may choose to consume less alcohol, reversing the causal pathway. The compressed range of cognition levels in our cohort may have hampered our description of associations.

The associations we have described bear some resemblance to the reported beneficial effects of hormone therapy on cardiovascular disease and dementia prior to clinical trials. Use of hormone therapy was also known to vary among women according to social class, access to care, and overall health status. Although such factors may not fully account for the associations we have described, we cannot dismiss this possibility; recent reinterpretation of effects attributed to hormone therapy from observational studies heightens the caution with which we interpret our findings. The converse also cannot be dismissed: alcohol intake may have confounded some of the benefits ascribed to hormone therapy. It may be that only a randomized clinical trial will clarify whether moderate alcohol intake truly is protective against cognitive decline.

## ACKNOWLEDGMENTS

WHIMS was funded by Wyeth Pharmaceuticals, Collegeville, Pennsylvania. The Women’s Health Initiative is funded by the National Heart, Lung, and Blood Institute of the National Institutes of Health, US Department of Health and Human Services. Wyeth Pharmaceuticals provided the study drug and the placebo to the Women’s Health Initiative trial.

FIGURE 1. Time from randomization until classification of probable dementia by self-reported alcohol intake, Women’s Health Initiative Memory Study, United States, 1996–2002.

FIGURE 1. Time from randomization until classification of probable dementia by self-reported alcohol intake, Women’s Health Initiative Memory Study, United States, 1996–2002.

FIGURE 2. Time until classification of probable dementia or mild cognitive impairment by self-reported alcohol intake, Women’s Health Initiative Memory Study, United States, 1996–2002.

FIGURE 2. Time until classification of probable dementia or mild cognitive impairment by self-reported alcohol intake, Women’s Health Initiative Memory Study, United States, 1996–2002.

TABLE 1.

Demographic, socioeconomic status, and lifestyle characteristics of participants in the Women’s Health Initiative Memory Study, by level of alcohol intake, United States, 1996–2002

 Variable No alcohol intake <1 drink per day ≥1 drink per day p value* (n = 1,345) (n = 2,500) (n = 616) Frequency % Frequency % Frequency % Age (years) 65–69 603 28.9 1,196 57.4 286 13.7 0.47 70–74 488 30.9 871 55.1 222 14.0 ≥75 254 32.0 433 54.5 108 13.6 No. of years since menopause 10–14 199 30.3 372 56.6 86 13.1 0.06 15–24 727 28.7 1,442 56.9 364 14.4 ≥25 304 34.0 475 53.1 116 13.0 Education High school but <4 years of college 532 30.5 969 55.6 243 13.9 ≥4 years of college 346 23.3 865 58.3 272 18.3 Ethnicity American Indian/Alaskan Native 4 40.0 5 50.0 1 10.0 <0.001 Asian/Pacific Islander 59 66.3 25 28.1 5 5.6 Black/African American 106 51.0 98 47.1 4 1.9 Hispanic/Latino 48 49.5 43 44.3 6 6.2 White, non-Hispanic 1,102 27.6 2,296 57.6 590 14.8 Other 25 41.7 29 48.3 6 10.0 Family income ($/year) <19,999 406 42.2 478 49.7 77 8.0 <0.001 20,000–34,999 419 32.2 735 56.4 148 11.4 35,000–49,999 230 25.2 538 59.0 144 15.8 50,000–74,999 148 23.0 362 56.1 135 20.9 ≥75,000 98 20.0 298 60.8 94 19.2 Smoking status Never 877 38.0 1,240 53.7 194 19.1 <0.001 Former 394 22.1 1,049 58.8 341 19.2 Current 60 19.7 176 57.9 68 22.4  Variable No alcohol intake <1 drink per day ≥1 drink per day p value* (n = 1,345) (n = 2,500) (n = 616) Frequency % Frequency % Frequency % Age (years) 65–69 603 28.9 1,196 57.4 286 13.7 0.47 70–74 488 30.9 871 55.1 222 14.0 ≥75 254 32.0 433 54.5 108 13.6 No. of years since menopause 10–14 199 30.3 372 56.6 86 13.1 0.06 15–24 727 28.7 1,442 56.9 364 14.4 ≥25 304 34.0 475 53.1 116 13.0 Education High school but <4 years of college 532 30.5 969 55.6 243 13.9 ≥4 years of college 346 23.3 865 58.3 272 18.3 Ethnicity American Indian/Alaskan Native 4 40.0 5 50.0 1 10.0 <0.001 Asian/Pacific Islander 59 66.3 25 28.1 5 5.6 Black/African American 106 51.0 98 47.1 4 1.9 Hispanic/Latino 48 49.5 43 44.3 6 6.2 White, non-Hispanic 1,102 27.6 2,296 57.6 590 14.8 Other 25 41.7 29 48.3 6 10.0 Family income ($/year) <19,999 406 42.2 478 49.7 77 8.0 <0.001 20,000–34,999 419 32.2 735 56.4 148 11.4 35,000–49,999 230 25.2 538 59.0 144 15.8 50,000–74,999 148 23.0 362 56.1 135 20.9 ≥75,000 98 20.0 298 60.8 94 19.2 Smoking status Never 877 38.0 1,240 53.7 194 19.1 <0.001 Former 394 22.1 1,049 58.8 341 19.2 Current 60 19.7 176 57.9 68 22.4

* Results from chi-square tests.

TABLE 2.

Association of alcohol intake with clinical characteristics and therapy among participants in the Women’s Health Initiative Memory Study, by level of alcohol intake, United States, 1996–2002

 Variable No alcohol intake <1 drink per day ≥1 drink per day p value* (n = 1,345) (n = 2,500) (n = 616) Frequency % Frequency % Frequency % Body mass index (kg/m2) <25 410 28.4 760 52.7 271 18.8 <0.001 25–29 440 27.8 925 58.4 219 13.8 30–34 287 31.1 542 58.7 95 10.3 ≥35 201 41.4 258 53.1 27 5.6 Hypertension status None 642 27.2 1,372 58.2 342 14.5 <0.001 Current/controlled† 241 36.5 343 51.9 77 11.7 Current/uncontrolled 462 32.0 785 54.4 197 13.6 Prior cardiovascular disease No 1,190 29.3 2,293 56.5 577 14.2 0.001 History of stroke 25 38.5 35 53.8 5 7.7 History of other cardiovascular disease‡ 130 38.7 172 51.2 34 10.1 Diabetes No 1,193 28.7 2,360 56.8 604 14.5 <0.001 Yes 151 50.5 137 45.8 11 3.7 Use of HMG-CoA§ reductase inhibitors No 1,186 29.6 2,248 56.2 569 14.2 0.03 Yes 167 34.2 270 55.2 52 10.7 Regular use of aspirin No 968 30.6 1,774 56.0 423 13.4 0.33 Yes 377 29.1 724 55.9 193 14.9 Prior use of hormone therapy No 1,072 30.8 1,957 56.3 446 12.8 0.001 Yes 273 27.7 543 55.1 170 17.2 Intervention assignment E + P§ 664 30.2 1,238 56.3 297 13.5 0.84 Placebo 681 30.1 1,262 55.8 319 14.1
 Variable No alcohol intake <1 drink per day ≥1 drink per day p value* (n = 1,345) (n = 2,500) (n = 616) Frequency % Frequency % Frequency % Body mass index (kg/m2) <25 410 28.4 760 52.7 271 18.8 <0.001 25–29 440 27.8 925 58.4 219 13.8 30–34 287 31.1 542 58.7 95 10.3 ≥35 201 41.4 258 53.1 27 5.6 Hypertension status None 642 27.2 1,372 58.2 342 14.5 <0.001 Current/controlled† 241 36.5 343 51.9 77 11.7 Current/uncontrolled 462 32.0 785 54.4 197 13.6 Prior cardiovascular disease No 1,190 29.3 2,293 56.5 577 14.2 0.001 History of stroke 25 38.5 35 53.8 5 7.7 History of other cardiovascular disease‡ 130 38.7 172 51.2 34 10.1 Diabetes No 1,193 28.7 2,360 56.8 604 14.5 <0.001 Yes 151 50.5 137 45.8 11 3.7 Use of HMG-CoA§ reductase inhibitors No 1,186 29.6 2,248 56.2 569 14.2 0.03 Yes 167 34.2 270 55.2 52 10.7 Regular use of aspirin No 968 30.6 1,774 56.0 423 13.4 0.33 Yes 377 29.1 724 55.9 193 14.9 Prior use of hormone therapy No 1,072 30.8 1,957 56.3 446 12.8 0.001 Yes 273 27.7 543 55.1 170 17.2 Intervention assignment E + P§ 664 30.2 1,238 56.3 297 13.5 0.84 Placebo 681 30.1 1,262 55.8 319 14.1

* Results from chi-square tests.

† Measured as <140/90 mmHg.

‡ Other cardiovascular disease defined as myocardial infarction, angina, percutaneous coronary transluminal angioplasty, or coronary artery bypass graft.

§ HMG-CoA, 3-hydroxy-3-methylglutaryl-coenzyme A; E + P, conjugated equine estrogens plus medroxyprogesterone acetate.

TABLE 3.

Relation between level of alcohol intake and baseline 3MSE* score for participants in the Women’s Health Initiative Memory Study, United States, 1996–2002

 Variable No alcohol intake <1 drink per day ≥1 drink per day p value Baseline 3MSE level (frequency (%)) 95–100 834 (62.2) 1,805 (72.3) 493 (80.2) <0.001† Above the cutpoint‡ to 94 390 (29.1) 569 (22.8) 102 (16.6) Cutpoint or below 117 (8.7) 123 (4.9) 20 (3.2) 3MSE total score at enrollment in the Women’s Health Initiative (mean (standard error)) 94.65 (0.11) 95.78 (0.08) 96.55 (0.16) <0.001§ Adjusting for socioeconomic status/lifestyle¶ (mean (standard error)) 95.24 (0.10) 95.64 (0.07) 96.03 (0.15) <0.001§ Adjusting for clinical characteristics# (mean (standard error)) 94.73 (0.11) 95.76 (0.08) 96.50 (0.16) <0.001§ Adjusting for all covariates (mean (standard error)) 95.28 (0.10) 95.63 (0.07) 96.00 (0.15) <0.001§
 Variable No alcohol intake <1 drink per day ≥1 drink per day p value Baseline 3MSE level (frequency (%)) 95–100 834 (62.2) 1,805 (72.3) 493 (80.2) <0.001† Above the cutpoint‡ to 94 390 (29.1) 569 (22.8) 102 (16.6) Cutpoint or below 117 (8.7) 123 (4.9) 20 (3.2) 3MSE total score at enrollment in the Women’s Health Initiative (mean (standard error)) 94.65 (0.11) 95.78 (0.08) 96.55 (0.16) <0.001§ Adjusting for socioeconomic status/lifestyle¶ (mean (standard error)) 95.24 (0.10) 95.64 (0.07) 96.03 (0.15) <0.001§ Adjusting for clinical characteristics# (mean (standard error)) 94.73 (0.11) 95.76 (0.08) 96.50 (0.16) <0.001§ Adjusting for all covariates (mean (standard error)) 95.28 (0.10) 95.63 (0.07) 96.00 (0.15) <0.001§

* 3MSE, Modified Mini-Mental State Examination.

† Results from a chi-square test.

‡ Screening cutpoint is ≤80 for women with 0–8 years of formal education and ≤88 for women with ≥9 years of formal education.

§ Results from analysis of covariance.

¶ Age, no. of years since menopause, education, ethnicity, family income, and smoking status.

# Body mass index, hypertension status, prior cardiovascular disease, diabetes, prior hormone therapy, statin use, and aspirin use.

TABLE 4.

Odds ratios for occurrence of clinically significant declines of 8 or more units in 3MSE* score from baseline, after adjustment for hormone therapy assignment, for several models, Women’s Health Initiative Memory Study, United States, 1996–2002

 Variable Relative to no alcohol intake p value (all pairwise differences)† <1 drink per day ≥1 drink per day Odds ratio 95% CI* Odds ratio 95% CI No additional covariates 0.56 0.41, 0.77 0.40 0.22, 0.72 <0.001 Adjusting for socioeconomic status/lifestyle‡ 0.67 0.48, 0.95 0.51 0.27, 0.94 0.025 Adjusting for clinical characteristics§ 0.58 0.42, 0.80 0.41 0.23, 0.74 <0.001 Adjusting for all covariates 0.69 0.49, 0.97 0.53 0.28, 0.99 0.042
 Variable Relative to no alcohol intake p value (all pairwise differences)† <1 drink per day ≥1 drink per day Odds ratio 95% CI* Odds ratio 95% CI No additional covariates 0.56 0.41, 0.77 0.40 0.22, 0.72 <0.001 Adjusting for socioeconomic status/lifestyle‡ 0.67 0.48, 0.95 0.51 0.27, 0.94 0.025 Adjusting for clinical characteristics§ 0.58 0.42, 0.80 0.41 0.23, 0.74 <0.001 Adjusting for all covariates 0.69 0.49, 0.97 0.53 0.28, 0.99 0.042

* 3MSE, Modified Mini-Mental State Examination; CI, confidence interval.

† Results from analysis of covariance.

‡ Age, no. of years since menopause, education, ethnicity, family income, and smoking status.

§ Body mass index, hypertension status, prior cardiovascular disease, diabetes, prior hormone therapy, statin use, and aspirin use.

TABLE 5.

Relative hazard ratios for probable dementia by baseline alcohol intake, compared with no intake, for several models after adjustment for hormone therapy assignment, Women’s Health Initiative Memory Study, United States, 1996–2002

 Variable Relative to no alcohol intake p value (all pairwise differences) <1 drink per day ≥1 drink per day Hazard ratio 95% CI* Hazard ratio 95% CI Probable dementia No additional covariates 0.64 0.38, 1.08 0.41 0.16, 1.06 0.09 Adjusting for socioeconomic status† 0.78 0.43, 1.42 0.30 0.08, 1.04 0.10 Adjusting for clinical characteristics‡ 0.65 0.38, 1.10 0.39 0.15, 1.02 0.08 Adjusting for all covariates 0.76 0.41, 1.40 0.28 0.08, 0.99 0.08 Adjusting for baseline 3MSE* score 1.14 0.65, 2.01 0.81 0.30, 2.18 0.75 Mild cognitive impairment or probable dementia No additional covariates 0.54 0.39, 0.74 0.47 0.28, 0.79 <0.001 Adjusting for socioeconomic status† 0.76 0.53, 1.08 0.57 0.32, 1.07 0.13 Adjusting for clinical characteristics‡ 0.59 0.43, 0.81 0.44 0.26, 0.78 0.001 Adjusting for all covariates 0.81 0.56, 1.16 0.57 0.31, 1.09 0.18 Adjusting for baseline 3MSE score 0.89 0.64, 1.25 0.92 0.53, 1.59 0.81
 Variable Relative to no alcohol intake p value (all pairwise differences) <1 drink per day ≥1 drink per day Hazard ratio 95% CI* Hazard ratio 95% CI Probable dementia No additional covariates 0.64 0.38, 1.08 0.41 0.16, 1.06 0.09 Adjusting for socioeconomic status† 0.78 0.43, 1.42 0.30 0.08, 1.04 0.10 Adjusting for clinical characteristics‡ 0.65 0.38, 1.10 0.39 0.15, 1.02 0.08 Adjusting for all covariates 0.76 0.41, 1.40 0.28 0.08, 0.99 0.08 Adjusting for baseline 3MSE* score 1.14 0.65, 2.01 0.81 0.30, 2.18 0.75 Mild cognitive impairment or probable dementia No additional covariates 0.54 0.39, 0.74 0.47 0.28, 0.79 <0.001 Adjusting for socioeconomic status† 0.76 0.53, 1.08 0.57 0.32, 1.07 0.13 Adjusting for clinical characteristics‡ 0.59 0.43, 0.81 0.44 0.26, 0.78 0.001 Adjusting for all covariates 0.81 0.56, 1.16 0.57 0.31, 1.09 0.18 Adjusting for baseline 3MSE score 0.89 0.64, 1.25 0.92 0.53, 1.59 0.81

* CI, confidence interval; 3MSE, Modified Mini-Mental State Examination.

† Age, no. of years since menopause, education, ethnicity, family income, and smoking status.

‡ Body mass index, hypertension status, prior cardiovascular disease, diabetes, statin use, aspirin use, and prior hormone therapy.

Correspondence to Dr. Mark A. Espeland, Department of Public Health Sciences, Wake Forest University School of Medicine, Medical Center Boulevard, Winston-Salem, NC 27157 (e-mail: mespelan@wfubmc.edu).

## References

1.
Folstein MF, Folstein SE, McHugh PR. “Mini-mental state,” a practical method for grading the cognitive state of patients for the clinician.
J Psychiatr Res

1975
;
12
:
189–
98.
2.
Dufouil C, Ducimetiere P, Alperovitch A. Sex differences in the association between alcohol consumption and cognitive performance.
Am J Epidemiol

1997
;
146
:
405–
12.
3.
Cerhan JR, Folsom AR, Mortimer JA, et al. Correlates of cognitive function in middle-aged adults.
Gerontology

1998
;
44
:
95–
105.
4.
Elias PK, Elias MF, D’Agostino RB, et al. Alcohol consumption and cognitive performance in the Framingham Heart Study.
Am J Epidemiol

1999
;
150
:
580–
9.
5.
Cervilla JA, Prince M, Mann A. Smoking, drinking, and incident cognitive impairment: a cohort community based study included in the Gospel Oak project.
J Neurol Neurosurg Psychiatry

2000
;
68
:
622–
6.
6.
Bond GE, Burr R, Rice MM, et al. Alcohol, aging, and cognitive performance: a cross-cultural comparison.
J Aging Health

2003
;
15
:
371–
90.
7.
Mukamal KJ, Kuller LH, Fitzpatrick AL, et al. Prospective study of alcohol consumption and risk of dementia in older adults.
JAMA

2003
;
289
:
1405–
13.
8.
Ruitenberg A, van Sweiten JC, Witteman JCM, et al. Alcohol consumption and risk of dementia: the Rotterdam Study.
Lancet

2002
;
359
:
281–
6.
9.
Hebert LE, Scherr PA, Beckett LA, et al. Relation of smoking and low-to-moderate alcohol consumption to change in cognitive function: a longitudinal study in a defined community of older persons.
Am J Epidemiol

1993
;
137
:
881–
91.
10.
Edelstein SL, Kritz-Silverstein D, Barrett-Connor E. Prospective association of smoking and alcohol use with cognitive function in an elderly cohort.
J Womens Health

1998
;
7
:
1271–
81.
11.
Reid MC, Maceiejewshi PK, Hawkins KA, et al. Relationship between alcohol consumption and Folstein mini-mental status examination scores among older cognitively impaired adults.
J Geriatr Psychiatry Neurol

2002
;
15
:
31–
7.
12.
Schinka JA, Belanger H, Mortimer JA, et al. Effects of the use of alcohol and cigarettes on cognition in elderly African American adults.
J Int Neuropsychol Soc

2003
;
9
:
690–
7.
13.
Tivis LJ, Green MD, Nixon SJ, et al. Alcohol, estrogen replacement therapy, and visuospatial processes in postmenopausal women.
Alcohol Clin Exp Res

2003
;
27
:
1055–
63.
14.
Shumaker SA, Reboussin BA, Espeland MA, et al. The Women’s Health Initiative Memory Study: a trial of the effect of estrogen therapy in preventing and slowing the progression of dementia.
Control Clin Trials

1998
;
19
:
604–
21.
15.
Design of the Women’s Health Initiative clinical trial and observational study. The Women’s Health Initiative Study Group.
Control Clin Trials

1998
;
19
:
61–
109.
16.
Rossouw JE, Anderson GL, Prentice RL, et al. Risks and benefits of estrogen plus progestin in healthy postmenopausal women: principal results from the Women’s Health Initiative randomized controlled trial.
JAMA

2002
;
288
:
321–
33.
17.
Shumaker S, Legault C, Rapp S, et al. The effects of estrogen plus progestin on the incidence of dementia and mild cognitive impairment in postmenopausal women: the Women’s Health Initiative Memory Study.
JAMA

2003
;
289
:
2651–
62.
18.
Rapp SR, Espeland MA, Shumaker SA, et al. Effect of estrogen plus progestin on global cognitive function in postmenopausal women: Women’s Health Initiative Memory Study: a randomized controlled trial.
JAMA

2003
;
20
:
2663–
72.
19.
Patterson RE, Kristal AR, Tinker LF, et al. Measurement characteristics of the Women’s Health Initiative food frequency questionnaire.
Ann Epidemiol

1999
;
9
:
178–
87.
20.
Fuenekes GI, van’t Veer P, van Staveren WA, et al. Alcohol intake assessment: the sober facts.
Am J Epidemiol

1999
;
150
:
105–
12.
21.
Gustafson P. Measurement error and misclassification in statistics and epidemiology: impacts and Bayesian adjustments. Boca Raton, FL: Chapman and Hall/CRC, 2003.
22.
Teng EL, Chui HC. The Modified Mini-Mental State (3MS) examination.
J Clin Psychiatry

1987
;
48
:
314–
18.
23.
Bland RC, Newman SC. Mild dementia or cognitive impairment: the Modified Mini-Mental State examination (3MS) as a screen for dementia.
Can J Psychiatry

2001
;
46
:
506–
10.
24.
Rapp SR, Espeland MA, Hogan P, et al. Baseline experience with the Modified Mini-Mental State Exam: The Women’s Health Initiative Memory Study (WHIMS).
Aging Ment Health

2003
;
7
:
217–
23.
25.
Seux BL, Thijs L, Forette F, et al. Correlates of cognitive status of old patients with isolated systolic hypertension: the Syst-Eur Vascular Dementia Project.
J Hypertens

1998
;
16
:
963–
9.
26.
Zuccala G, Onder G, Pedone C, et al. Dose-related impact of alcohol consumption on cognitive function in advanced age: results of a multicenter study.
Alcohol Clin Exp Res

2001
;
25
:
1743–
8.
27.
Bond GE, Burr R, McCurry SM, et al. Alcohol, aging, and cognitive performance in a cohort of Japanese Americans aged 65 and older: the Kame project.
Int Psychogeriatr

2001
;
13
:
207–
23.
28.
Schinka JA, Vanderploeg RD, Rogish M, et al. Effects of alcohol and cigarette use on cognition in middle-aged adults.
J Int Neuropsychol Soc

2002
;
8
:
683–
90.
29.
Kalmijn S, van Boxtel MP, Verschuren WM, et al. Cigarette smoking and alcohol consumption in relation to cognitive performance in middle age.
Am J Epidemiol

2002
;
156
:
936–
44.
30.
Huadong Z, Juan D, Jingcheng L, et al. Study of the relationship between cigarette smoking, alcohol drinking, and cognitive impairment among elderly people in China.
Age Aging

2003
;
32
:
205–
10.
31.
Galanis DJ, Joseph C, Masaki KH, et al. A longitudinal study of drinking and cognitive performance in elderly Japanese American men: The Honolulu-Asia Aging Study.
Am J Public Health

2000
;
90
:
1254–
9.
32.
Christian JC, Reed T, Carmelli D, et al. Self-reported alcohol consumption and cognition in aging twins.
J Stud Alcohol

1995
;
56
:
414–
16.
33.
Leroi I, Sheppard JM, Lyketsos CG. Cognitive function after 11.5 years of alcohol use: relation to alcohol use.
Am J Epidemiol

2002
;
156
:
747–
52.
34.
Launer LJ, Feskens EJ, Kalmijn S, et al. Smoking, drinking, and thinking: the Zutphen Elderly Study.
Am J Epidemiol

1996
;
143
:
219–
27.
35.
Peele S, Brodsky A. Exploring psychological benefits associated with moderate alcohol use: a necessary corrective to assessments of drinking outcomes?
Drug Alcohol Depend

2000
;
60
:
221–
47.
36.
Truelsen T, Thudium D, Gronbaek M. Amount and type of alcohol and risk of dementia. The Copenhagen City Heart Study.
Neurology

2002
;
59
:
1313–
19.
37.
Graves AB, van Duijn CM, Chandra V, et al. Alcohol and tobacco consumption as risk factors for Alzheimer’s disease: a collaborative re-analysis of case-control studies.
Int J Epidemiol

1991
;
20
:
S48–
57.
38.
Hebert LE, Scherr PA, Beckett LA, et al. Relation of smoking and alcohol consumption to incident Alzheimer’s disease.
Am J Epidemiol

1992
;
135
:
347–
55.
39.
Palomaki H, Kaste M. Regular light-to-moderate intake of alcohol and the risk of ischemic stroke: is there a beneficial effect?
Stroke

1993
;
24
:
1828–
32.
40.
Camargo CA Jr. Moderate alcohol consumption and stroke. The epidemiologic evidence.
Stroke

1989
;
20
:
1611–
26.
41.
Mukamal KJ, Longstreth WT Jr, Mittleman MA, et al. Alcohol consumption and subclinical findings on magnetic resonance imaging of the brain in older adults: the Cardiovascular Health Study.
Stroke

2001
;
32
:
1939–
46.
42.
Ladolfi R, Steiner M. Ethanol raises prostacyclin level in vivo and in vitro.
Blood

1984
;
64
:
679–
82.
43.
Jakubowski JA, Vaillancourt R, Deykin D. Interaction of ethanol, prostacyclin, and aspirin in determining human platelet reactivity in vitro.
Arteriosclerosis

1988
;
8
:
436–
41.
44.
Rubin R, Rand ML. Alcohol and platelet function.
Alcohol Clin Exp Res

1994
;
18
:
105–
10.
45.
Ridker PM, Vaughan DE, Stampfer MJ, et al. Association of moderate alcohol consumption and plasma concentration of endogenous tissue-type plasminogen activator.
JAMA

1994
;
272
:
929–
33.
46.
Meade TW, Chakrabarti R, Haines AP, et al. Characteristics affecting fibrinolytic activity and plasma fibrinogen concentrations.
BMJ

1979
;
1
:
153–
6.
47.
Gaziano JM, Buring JE, Breslow JL, et al. Moderate alcohol intake, increased levels of high-density lipoprotein and its subfractions, and decreased risk of myocardial infarction.
N Engl J Med

1993
;
329
:
1829–
34.
48.
Castelli WP, Doyle JT, Gordon T, et al. Alcohol and blood lipids. The Cooperative Lipoprotein Phenotyping Study.
Lancet

1977
;
2
:
153–
5.
49.
Ernst N, Fisher M, Smith W, et al. The association of plasma high-density lipoprotein cholesterol with dietary intake and alcohol consumption. The Lipids Research Clinics Prevalence Study.
Circulation

1980
;
62
:
IV41–
52.
50.
Hartung GH, Foreyt JP, Mitchell RE, et al. Effect of alcohol intake on high-density lipoprotein cholesterol levels in runners and inactive men.
JAMA

1983
;
249
:
747–
50.
51.
Camargo CA Jr, Williams PT, Vranizan KM, et al. The effect of moderate alcohol intake on serum apolipoproteins A-I and A-II. A controlled study.
JAMA

1985
;
253
:
2854–
7.
52.
Langer RD, Criqui MH, Reed DM. Lipoproteins and blood pressure as biologic pathways for effect of moderate alcohol consumption on coronary heart disease.
Circulation

1992
;
85
:
910–
15.
53.
Strittmatter WJ, Weisgraber KH, Huang DY, et al. Binding of human apolipoprotein E to synthetic amyloid beta peptide: isoform-specific effects and implications for late-onset Alzheimer disease.
Proc Natl Acad Sci U S A

1993
;
90
:
8098–
102.
54.
Carmelli D, Swan GE, Reed T, et al. The effect of apolipoprotein E epsilon4 in the relationships of smoking and drinking to cognitive function.
Neuroepidemiology

1999
;
18
:
125–
33.
55.
Dufouil C, Tzourio C, Brayne C, et al. Influence of apolipoprotein E genotype on the risk of cognitive deterioration in moderate drinkers and smokers.
Epidemiology

2000
;
11
:
280–
4.
56.
Perry E, Walker M, Grace J, et al. Acetylcholine in mind: a neurotransmitter correlate of consciousness?
Trends Neurosci

1999
;
22
:
273–
80.
57.
Henn C, Loffelholz K, Klein J. Stimulatory and inhibitory effects of ethanol on hippocampal acetylcholine release.
Naunyn Schmiedebergs Arch Pharmacol

1998
;
357
:
640–
7.