American Journal of Epidemiology Copyright © 1992 by The Johns Hopkins University School of Hygiene and Public Health All rights reserved
Vol. 135, No. 4 Printed in U.S.A.
Relation of Smoking and Alcohol Consumption to Incident Alzheimer's Disease
Liesi E. Hebert,1 Paul A. Scherr,2 Laurel A. Beckett,3 4 H. Harris Funkenstein,5 Marilyn S. Albert,6 Marilyn J. Chown," and Denis A. Evans7
The authors examined the effects of smoking and alcohol use in a prospective community-based study of incident Alzheimer's disease. Two in-home interviews of the total elderly population of East Boston, Massachusetts, conducted in 1982 and 1985 were used to sample individuals for clinical evaluation for Alzheimer's disease. A total of 513 persons underwent detailed clinical evaluation including neurologic, neuropsychologic, and psychiatric evaluation to diagnose Alzheimer's disease. In weighted logistic regression controlled for age, sex, and education, the estimated odds ratio of Alzheimer's disease was 0.7 (95% confidence interval 0.3-1.4) for ever smokers compared with never smokers. For 40 pack-years of smoking, the odds ratio of Alzheimer's disease was 0.8 (95% confidence interval 0.6-1.1). Consumption of 1 oz (30 ml) of alcohol per day was associated with an odds ratio of 1.1 (95% confidence interval 0.8-1.5). These results suggest that recent mild-to-moderate consumption of alcohol is not substantially related to incidence of Alzheimer's disease and that smoking does not increase risk of the disease. Am J Epidemiol 1992;135:347-55. aged; alcohol drinking; Alzheimer's disease; epidemiology; incidence; risk factors; smoking
Alzheimer's disease is a condition that has enormous impact on the quality of life of the elderly. With the changing age distribution in developed countries, the proportion of the population that is elderly will continue to increase well into the next century (1). The occurrence of Alzheimer's disease will also increase since it is strongly related to
age. Thus far, there has been little success in identifying preventable factors related to Alzheimer's disease (2). Smoking deserves consideration as a factor contributing to Alzheimer's disease because it can affect central nervous system function and because a substantial proportion of the population is exposed to cigarette smoke. The results of nine
Received for publication February 14,1990, and in final form October 15, 1991. 1 Department of Health Policy and Management, Harvard School of Public Health, Boston, MA. 2 Aging Branch, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control, Atlanta, GA. 3 Channing Laboratory, Department of Medicine, Brigham and Women's Hospital, and Department of Medicine, Harvard Medical School, Boston, MA. 4 Department of Biostatistics, Harvard School of Public Health, Boston, MA. 5 Division of Neurology, Department of Medicine, Bngham and Women's Hospital, and Department of Medicine, Harvard Medical School, Boston, MA. 6 Departments of Psychiatry and Neurology, Massachusetts General Hospital, and Department of Psychiatry, Harvard Medical School, Boston, MA.
'Center for Research on Health and Aging, RushPresbyterian-St. Luke's Medical Center, Chicago, IL. Reprint requests to Dr. Denis A. Evans, Center for Research on Health and Aging, Rush-Presbyterian-St. Luke's Medical Center, 1653 West Congress Parkway, Chicago, IL 60612. This study was supported by grant AG05362, cooperative agreement 06789, and contracts NO1-AG-1-2106 and NO1-AG-O-2107 from the National Institute on Aging; National- Research Service Award training grant T32ESO7O69 from the National Institute of Environmental Health Sciences; and the CUNFO data management and analysis system supported by CRC grant M01 RR 0263506. The support and cooperation of the staff of the East Boston Neighborhood Health Center are gratefully acknowledged.
347
348
Hebertetal.
case-control studies (3-11) of the relation of smoking to Alzheimer's disease have varied widely. Seven studies (3-6, 8, 9) found no relation. One study (7) found a significant positive association between smoking and Alzheimer's disease while, in contrast, one (11) suggested an inverse relation. A meta analysis (12) that included data from seven (3-9) of these studies also suggested that smokers were at lower risk of Alzheimer's disease than were nonsmokers. Abuse of alcohol has been identified clinically as a cause of dementia (13, 14). The relation of moderate alcohol consumption to the development of dementia is unclear. Five of the case-control studies of Alzheimer's disease cited above obtained information about alcohol consumption. No significant relation between alcohol use and Alzheimer's disease was found in any of the five (3-5, 8, 9). A meta analysis (12) of data from case-control studies, including these five studies, also found no association. No prospective cohort studies have examined the relation of either smoking or alcohol consumption to Alzheimer's disease. This paper reports the relation of lifetime smoking history and recent alcohol consumption to the subsequent onset of Alzheimer's disease in a prospective, population-based study. METHODS Study population
The study was conducted in East Boston, Massachusetts, a geographically defined, urban, working-class community that is one of four centers of the Established Populations for Epidemiologic Study of the Elderly Program sponsored by the National Institute on Aging. Beginning in 1982, noninstitutionalized individuals aged 65 years or more were interviewed in their homes. A total of 3,809 individuals, 85 percent of the ageeligible residents, participated in the study (15). The interview included a structured test of immediate memory that was used to define the eligible population. In this test, 2,136 individuals recalled at least four of six items correctly, a score consistent with good
performance. The group scoring well in 1982 was known to contain few individuals with Alzheimer's disease. From previously reported clinical evaluations of a sample of this group, 97 percent were estimated to be free of prevalent Alzheimer's disease (16, 17). This group was followed for the occurrence of new cases of Alzheimer's disease. Three years later, the survivors were interviewed in their homes with 93 percent of them participating. Among the people who scored well in 1982, 241 died in the 3-year interval; 1,649 were retested successfully in 1985, and information for 247 was unavailable or incomplete. After testing in 1985, the retested individuals were classified into three groups according to change in memory test score from time one to time two. One group contained people with no decline in immediate memory score. A second contained people with substantial decline, whose immediate memory score decreased two or more points and who also scored two or less on the delayed memory test. A third group contained all people with performance intermediate between the first and second groups. There were 691 people in the group with no decline, 705 in the group with intermediate decline, and 253 in the group with substantial decline. Change in memory was not used as a screening test, but rather as a stratification tool; individuals from all three groups were examined. All individuals who demonstrated a substantial decline were invited to participate in the detailed clinical evaluation in order to maximize the number of affected individuals. The sample from the group with no decline was frequency matched to the age and sex distribution of the poor memory group to ensure that, after evaluation, there would be both unaffected and affected individuals of similar age. The group with intermediate decline was randomly sampled to determine the occurrence of Alzheimer's disease in that group. To increase the sample size, data from a follow-up of a subset of participants in a previously reported study (16) were included in the analyses. The subset consisted of participants who performed well on the mem-
Smoking, Alcohol Consumption, and Alzheimer's Disease
ory test at the initial home interview and who received memory testing at the second home interview. These participants were eligible for a subsequent clinical evaluation for incident disease. In this way, a consistent set of entry criteria was used for all study subjects. Of the 690 people selected from both sources, 32 died before they could be evaluated for Alzheimer's disease following the second home interview. Of the remainder, 513 (78 percent) were evaluated for incident Alzheimer's disease. The average interval from first home interview, when the population was largely disease free, to clinical evaluation for incident Alzheimer's disease was 4.7 years. Outcome
People selected for detailed clinical evaluation received neuropsychologic tests and a comprehensive neurologic examination (16). Diagnosis of Alzheimer's disease was made within the study by a neurologist and neuropsychologist. Using a standardized, structured examination and a classification system that conforms to the National Institute of Neurologic and Communicative Disorders and Stroke/Alzheimer's Disease and Related Disorders Association (18) criteria for the disease, subjects were classified as having probable Alzheimer's disease, possible Alzheimer's disease, or no evidence of the disease. The classification scheme permitted more than one disease to be coded as present simultaneously. Thus, subjects diagnosed as having Alzheimer's disease also may have had other, coexisting illnesses responsible for decline in cognition. For primary analyses, those with possible Alzheimer's disease were grouped with those with no evidence of disease. In a second set of analyses, those with possible Alzheimer's disease were omitted from the analyses. Predictors
Information on smoking habits and use of alcohol was obtained as part of the 1982 interview. Smoking was analyzed in four ways: smoking status (ever versus never
349
smoker), packs of cigarettes smoked daily, number of years of smoking, and pack-years. Alcohol information was analyzed in four ways. 1) People who drank any alcohol in the past year were compared with nondrinkers. 2 and 3) The number of ounces of alcohol consumed (19) was used as a continuous variable both with (method 2) and without (method 3) an indicator for no alcohol consumption in the year. 4) Three categories of consumers (12
149 211 151
22.5 8.3 4.7
42 24 10
1 0.4 0.3
263
13.9
50
1
85 80 80
2.3 11.7 5.4
5 12 6
0.2 0.9 0.4
132 248 57 71
14.3 7.3 10.2 14.5
23 31 9 13
1 0.6 0.8 1.0
Smoking Never Ever (pack-years) >0-75
Years of formal schooling 0-7
>8
Sample total
(n)
Incident probable Alzheimer's disease %
n
1 5 1 7 8 9 15 18 12
10 19 13 13 12 9
None 0.5 oz/day None 0.5 oz/day None 0.5 oz/day
36 80 43 35 78 41 61 44
1.3 3.8 0.9 18.3 7.3 25.8 23.6 15.1 20.1
None 0.5 oz/day None 0.5 oz/day
42 68 39 89 180 88
17.8 24.2 24.5 13.2 3.7 8.3
90
* Percentage is weighted to represent the proportion in the entire eligible population, so percentages do not exactly match the sample numbers. t Quantity consumed in the month preceding the 1982 interview; 0.5 fluid oz = 15 ml.
352
Hebert et al.
TABLE 5. Incident probable Alzheimer's disease risk estimates from a weighted logistic regression model with smoking included as ever versus never: data from a stratified random sample of 513 individuals aged 65 years or older in East Boston, Massachusetts, who performed well on memory tests in 1982 and were retested in 1985 Odds ratio
Coefficient
95% confidence interval
Smoked cigarettes
-0.428
0.3
0.7
0.3-1.4
No alcohol in year*
0.375
0.3
1.5
0.7-3.0
0.124
0.4
1.1
0.8-1.5
3.3
2.0-5.2
1.2
0.5-2.8
0.5
0.3-0.7
Ounces of alcohol/
dayt Age Odds ratio for 10 years
0.118
Sex (M/F)
0.165
Education Odds ratio for 4 years of schooling
-0.170
Vol. 135, No. 4 Printed in U.S.A.
Relation of Smoking and Alcohol Consumption to Incident Alzheimer's Disease
Liesi E. Hebert,1 Paul A. Scherr,2 Laurel A. Beckett,3 4 H. Harris Funkenstein,5 Marilyn S. Albert,6 Marilyn J. Chown," and Denis A. Evans7
The authors examined the effects of smoking and alcohol use in a prospective community-based study of incident Alzheimer's disease. Two in-home interviews of the total elderly population of East Boston, Massachusetts, conducted in 1982 and 1985 were used to sample individuals for clinical evaluation for Alzheimer's disease. A total of 513 persons underwent detailed clinical evaluation including neurologic, neuropsychologic, and psychiatric evaluation to diagnose Alzheimer's disease. In weighted logistic regression controlled for age, sex, and education, the estimated odds ratio of Alzheimer's disease was 0.7 (95% confidence interval 0.3-1.4) for ever smokers compared with never smokers. For 40 pack-years of smoking, the odds ratio of Alzheimer's disease was 0.8 (95% confidence interval 0.6-1.1). Consumption of 1 oz (30 ml) of alcohol per day was associated with an odds ratio of 1.1 (95% confidence interval 0.8-1.5). These results suggest that recent mild-to-moderate consumption of alcohol is not substantially related to incidence of Alzheimer's disease and that smoking does not increase risk of the disease. Am J Epidemiol 1992;135:347-55. aged; alcohol drinking; Alzheimer's disease; epidemiology; incidence; risk factors; smoking
Alzheimer's disease is a condition that has enormous impact on the quality of life of the elderly. With the changing age distribution in developed countries, the proportion of the population that is elderly will continue to increase well into the next century (1). The occurrence of Alzheimer's disease will also increase since it is strongly related to
age. Thus far, there has been little success in identifying preventable factors related to Alzheimer's disease (2). Smoking deserves consideration as a factor contributing to Alzheimer's disease because it can affect central nervous system function and because a substantial proportion of the population is exposed to cigarette smoke. The results of nine
Received for publication February 14,1990, and in final form October 15, 1991. 1 Department of Health Policy and Management, Harvard School of Public Health, Boston, MA. 2 Aging Branch, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control, Atlanta, GA. 3 Channing Laboratory, Department of Medicine, Brigham and Women's Hospital, and Department of Medicine, Harvard Medical School, Boston, MA. 4 Department of Biostatistics, Harvard School of Public Health, Boston, MA. 5 Division of Neurology, Department of Medicine, Bngham and Women's Hospital, and Department of Medicine, Harvard Medical School, Boston, MA. 6 Departments of Psychiatry and Neurology, Massachusetts General Hospital, and Department of Psychiatry, Harvard Medical School, Boston, MA.
'Center for Research on Health and Aging, RushPresbyterian-St. Luke's Medical Center, Chicago, IL. Reprint requests to Dr. Denis A. Evans, Center for Research on Health and Aging, Rush-Presbyterian-St. Luke's Medical Center, 1653 West Congress Parkway, Chicago, IL 60612. This study was supported by grant AG05362, cooperative agreement 06789, and contracts NO1-AG-1-2106 and NO1-AG-O-2107 from the National Institute on Aging; National- Research Service Award training grant T32ESO7O69 from the National Institute of Environmental Health Sciences; and the CUNFO data management and analysis system supported by CRC grant M01 RR 0263506. The support and cooperation of the staff of the East Boston Neighborhood Health Center are gratefully acknowledged.
347
348
Hebertetal.
case-control studies (3-11) of the relation of smoking to Alzheimer's disease have varied widely. Seven studies (3-6, 8, 9) found no relation. One study (7) found a significant positive association between smoking and Alzheimer's disease while, in contrast, one (11) suggested an inverse relation. A meta analysis (12) that included data from seven (3-9) of these studies also suggested that smokers were at lower risk of Alzheimer's disease than were nonsmokers. Abuse of alcohol has been identified clinically as a cause of dementia (13, 14). The relation of moderate alcohol consumption to the development of dementia is unclear. Five of the case-control studies of Alzheimer's disease cited above obtained information about alcohol consumption. No significant relation between alcohol use and Alzheimer's disease was found in any of the five (3-5, 8, 9). A meta analysis (12) of data from case-control studies, including these five studies, also found no association. No prospective cohort studies have examined the relation of either smoking or alcohol consumption to Alzheimer's disease. This paper reports the relation of lifetime smoking history and recent alcohol consumption to the subsequent onset of Alzheimer's disease in a prospective, population-based study. METHODS Study population
The study was conducted in East Boston, Massachusetts, a geographically defined, urban, working-class community that is one of four centers of the Established Populations for Epidemiologic Study of the Elderly Program sponsored by the National Institute on Aging. Beginning in 1982, noninstitutionalized individuals aged 65 years or more were interviewed in their homes. A total of 3,809 individuals, 85 percent of the ageeligible residents, participated in the study (15). The interview included a structured test of immediate memory that was used to define the eligible population. In this test, 2,136 individuals recalled at least four of six items correctly, a score consistent with good
performance. The group scoring well in 1982 was known to contain few individuals with Alzheimer's disease. From previously reported clinical evaluations of a sample of this group, 97 percent were estimated to be free of prevalent Alzheimer's disease (16, 17). This group was followed for the occurrence of new cases of Alzheimer's disease. Three years later, the survivors were interviewed in their homes with 93 percent of them participating. Among the people who scored well in 1982, 241 died in the 3-year interval; 1,649 were retested successfully in 1985, and information for 247 was unavailable or incomplete. After testing in 1985, the retested individuals were classified into three groups according to change in memory test score from time one to time two. One group contained people with no decline in immediate memory score. A second contained people with substantial decline, whose immediate memory score decreased two or more points and who also scored two or less on the delayed memory test. A third group contained all people with performance intermediate between the first and second groups. There were 691 people in the group with no decline, 705 in the group with intermediate decline, and 253 in the group with substantial decline. Change in memory was not used as a screening test, but rather as a stratification tool; individuals from all three groups were examined. All individuals who demonstrated a substantial decline were invited to participate in the detailed clinical evaluation in order to maximize the number of affected individuals. The sample from the group with no decline was frequency matched to the age and sex distribution of the poor memory group to ensure that, after evaluation, there would be both unaffected and affected individuals of similar age. The group with intermediate decline was randomly sampled to determine the occurrence of Alzheimer's disease in that group. To increase the sample size, data from a follow-up of a subset of participants in a previously reported study (16) were included in the analyses. The subset consisted of participants who performed well on the mem-
Smoking, Alcohol Consumption, and Alzheimer's Disease
ory test at the initial home interview and who received memory testing at the second home interview. These participants were eligible for a subsequent clinical evaluation for incident disease. In this way, a consistent set of entry criteria was used for all study subjects. Of the 690 people selected from both sources, 32 died before they could be evaluated for Alzheimer's disease following the second home interview. Of the remainder, 513 (78 percent) were evaluated for incident Alzheimer's disease. The average interval from first home interview, when the population was largely disease free, to clinical evaluation for incident Alzheimer's disease was 4.7 years. Outcome
People selected for detailed clinical evaluation received neuropsychologic tests and a comprehensive neurologic examination (16). Diagnosis of Alzheimer's disease was made within the study by a neurologist and neuropsychologist. Using a standardized, structured examination and a classification system that conforms to the National Institute of Neurologic and Communicative Disorders and Stroke/Alzheimer's Disease and Related Disorders Association (18) criteria for the disease, subjects were classified as having probable Alzheimer's disease, possible Alzheimer's disease, or no evidence of the disease. The classification scheme permitted more than one disease to be coded as present simultaneously. Thus, subjects diagnosed as having Alzheimer's disease also may have had other, coexisting illnesses responsible for decline in cognition. For primary analyses, those with possible Alzheimer's disease were grouped with those with no evidence of disease. In a second set of analyses, those with possible Alzheimer's disease were omitted from the analyses. Predictors
Information on smoking habits and use of alcohol was obtained as part of the 1982 interview. Smoking was analyzed in four ways: smoking status (ever versus never
349
smoker), packs of cigarettes smoked daily, number of years of smoking, and pack-years. Alcohol information was analyzed in four ways. 1) People who drank any alcohol in the past year were compared with nondrinkers. 2 and 3) The number of ounces of alcohol consumed (19) was used as a continuous variable both with (method 2) and without (method 3) an indicator for no alcohol consumption in the year. 4) Three categories of consumers (12
149 211 151
22.5 8.3 4.7
42 24 10
1 0.4 0.3
263
13.9
50
1
85 80 80
2.3 11.7 5.4
5 12 6
0.2 0.9 0.4
132 248 57 71
14.3 7.3 10.2 14.5
23 31 9 13
1 0.6 0.8 1.0
Smoking Never Ever (pack-years) >0-75
Years of formal schooling 0-7
>8
Sample total
(n)
Incident probable Alzheimer's disease %
n
1 5 1 7 8 9 15 18 12
10 19 13 13 12 9
None 0.5 oz/day None 0.5 oz/day None 0.5 oz/day
36 80 43 35 78 41 61 44
1.3 3.8 0.9 18.3 7.3 25.8 23.6 15.1 20.1
None 0.5 oz/day None 0.5 oz/day
42 68 39 89 180 88
17.8 24.2 24.5 13.2 3.7 8.3
90
* Percentage is weighted to represent the proportion in the entire eligible population, so percentages do not exactly match the sample numbers. t Quantity consumed in the month preceding the 1982 interview; 0.5 fluid oz = 15 ml.
352
Hebert et al.
TABLE 5. Incident probable Alzheimer's disease risk estimates from a weighted logistic regression model with smoking included as ever versus never: data from a stratified random sample of 513 individuals aged 65 years or older in East Boston, Massachusetts, who performed well on memory tests in 1982 and were retested in 1985 Odds ratio
Coefficient
95% confidence interval
Smoked cigarettes
-0.428
0.3
0.7
0.3-1.4
No alcohol in year*
0.375
0.3
1.5
0.7-3.0
0.124
0.4
1.1
0.8-1.5
3.3
2.0-5.2
1.2
0.5-2.8
0.5
0.3-0.7
Ounces of alcohol/
dayt Age Odds ratio for 10 years
0.118
Sex (M/F)
0.165
Education Odds ratio for 4 years of schooling
-0.170
