Relation of Education and Occupation-based Socioeconomic Status to Incident Alzheimer s Disease

Transcription

1 American Journal of Epidemiology Copyright 2004 by the Johns Hopkins Bloomberg School of Public Health All rights reserved Vol. 159, No. 2 Printed in U.S.A. DOI: /aje/kwh018 Relation of Education and Occupation-based Socioeconomic Status to Incident Alzheimer s Disease Anita Karp 1,2, Ingemar Kåreholt 1, Chengxuan Qiu 1,2, Tom Bellander 3, Bengt Winblad 1,2, and Laura Fratiglioni 1,2 1 Aging Research Center, Division of Geriatric Epidemiology and Medicine, Neurotec, Karolinska Institutet, Stockholm, Sweden. 2 Stockholm Gerontology Research Center, Stockholm, Sweden. 3 Department of Occupational and Environmental Health, Stockholm County Council, and Institute of Environmental Medicine, Karolinska Institutet, Stockholm, Sweden. Received for publication June 7, 2002; accepted for publication July 21, In this study, the authors evaluated whether the association between low educational level and increased risk of Alzheimer s disease (AD) and dementia may be explained by occupation-based socioeconomic status (SES). A cohort of 931 nondemented subjects aged 75 years from the Kungsholmen Project, Stockholm, Sweden, was followed for 3 years between 1987 and A total of 101 incident cases of dementia, 76 involving AD, were detected. Less-educated subjects had an adjusted relative risk of developing AD of 3.4 (95% confidence interval: 2.0, 6.0), and subjects with lower SES had an adjusted relative risk of 1.6 (95% confidence interval: 1.0, 2.5). When both education and SES were introduced into the same model, only education remained significantly associated with AD. Combinations of low education with low or high SES were associated with similar increased risks of AD, but well-educated subjects with low SES were not at high risk. Low SES at 20 years of age, even when SES was high at age 40 or 60 years, was associated with increased risk; however, this increase disappeared when education was entered into the model. In conclusion, the association between low education and increased AD risk was not mediated by adult SES or socioeconomic mobility. This suggests that early life factors may be relevant. aged; Alzheimer disease; dementia; education; social class Abbreviations: CI, confidence interval; SES, socioeconomic status. Several studies have reported an increased risk of dementia and Alzheimer s disease among less-educated persons (1 10). Different mechanisms have been suggested to explain this association. Katzman (5) proposed that education might increase brain reserves by increasing synaptic density in the neocortical association cortex. Stern et al. (6) extended the cognitive reserve hypothesis and took into account the possible beneficial influence of mental activity throughout the entire life span, where occupational attainment along with level of education could influence the risk of Alzheimer s disease. Education might also be a surrogate measure for an intellectual outlook (7), as well as a proxy for intelligence, which in turn could be inversely related to the risk of Alzheimer s disease (8, 9). In addition, the issue of a possible detection bias due to later diagnosis of dementia in highly educated subjects has been raised (4, 10). It is known that education is an indicator of socioeconomic status (SES). Different studies have shown high (11) to moderate (2) correlations between education and occupation-based SES. Nevertheless, these two variables represent different aspects of and periods in the life course and may generate differences in health through various mechanisms (11, 12). Lynch and Kaplan describe education as the transition from a socio-economic position largely received from parents to an achieved socio-economic position as an adult (13, p. 22). Socioeconomic factors acting over the life course have been found to affect health and premature death, as well as risk of cardiovascular disease (14, 15). Cumulative Reprint requests to Anita Karp, Aging Research Center, Karolinska Institutet, Box 6401 (Olivecronas väg 4), S Stockholm, Sweden ( anita.karp@neurotec.ki.se). 175

2 176 Karp et al. economic hardship has been found to lead to poorer cognitive functioning (16). Recently, Turrell et al. (17) examined the relation between socioeconomic mobility and cognitive function among Finish men in late middle age and suggested that all stages of the life course play a role in influencing adult cognitive function. Few studies have reported an association between adult SES based on occupation and the incidence of Alzheimer s disease or dementia. Evans et al. (2) found that all of the investigated markers of SES (education, occupational prestige, and income) predicted the development of Alzheimer s disease. It has been reported that manual work, when it involves production of goods, could increase the risk of clinical Alzheimer s disease or dementia (18). However, these studies focused on principal occupation, not all occupational periods. In addition, the relation of social mobility through different socioeconomic levels over the life course to the risk of dementia in old age has not been studied. The general aim of this study was to determine whether the reported association between low education and increased risk of Alzheimer s disease and dementia could be explained by occupation-based SES. Specifically, we examined how the risk of Alzheimer s disease and dementia varied by different combinations of 1) education and main occupationbased SES and 2) education and lifetime number of years spent in a low occupation-based SES. Since education and occupation-based SES are interrelated, special attention was given to the categories that deviated from the expected pattern, that is, the combinations of high education/low occupation-based SES and low education/high occupationbased SES. Furthermore, we investigated social mobility patterns to explore whether later advancement or setbacks in occupation-based SES may affect the relation between education and Alzheimer s disease incidence. MATERIALS AND METHODS Study population The study population was derived from the Kungsholmen Project, a community-based longitudinal study of aging and dementia in Stockholm, Sweden. All registered inhabitants of the Kungsholmen district of Stockholm who were aged 75 years or older on October 1, 1987, were invited to participate in the project. Among the 1,810 persons who agreed to participate and signed an informed consent form at the baseline examination ( ), 1,473 (361 men and 1,112 women) were diagnosed as being free of dementia. At baseline, detection of prevalent cases of dementia was conducted using a two-phase design. All subjects were given a brief psychometric test as a screening test for dementia. Persons who screened positive (were suspected to have dementia) and a random sample of subjects who screened negative (were not suspected to have dementia) were clinically examined (19). The same diagnostic criteria and procedure for diagnosing dementia were used at baseline and at follow-up, as described below. At the time of the follow-up examination ( ), 150 subjects refused to participate and 22 had moved out of the Stockholm area. Information about lifetime occupational history was unavailable for 370 subjects because of refusal by the informant. Therefore, the current study population consisted of 931 persons. Identification of incident cases of Alzheimer s disease and dementia At follow-up, all subjects were clinically examined by physicians, neuropsychologically assessed by psychologists, and interviewed by nurses. Incident cases of dementia were defined as those that developed during the follow-up period. The diagnosis of clinical dementia was made according to the criteria of the Diagnostic and Statistical Manual of Mental Disorders, Third Edition, Revised (20), using a three-step procedure. First, a preliminary diagnosis was made by the examining physician. Second, all cases were independently reviewed by a specialized clinician, and a second diagnosis was made. If the diagnoses were in agreement, they were accepted as final. In cases of disagreement, a third opinion was obtained, and the concordant diagnosis was accepted. The diagnosis of dementia was made when the subject completely fulfilled the criteria of the Diagnostic and Statistical Manual of Mental Disorders, Third Edition, Revised. The diagnosis of Alzheimer s disease in particular required gradual onset and progressive deterioration of cognitive functioning and exclusion of all other specific causes of dementia. Neuroimaging and autopsy data were not available. Our clinical diagnosis of Alzheimer s disease corresponds to the diagnosis of probable Alzheimer s disease according to the NINCDS-ADRDA [National Institute of Neurological and Communicative Disorders and Stroke Alzheimer s Disease and Related Disorders Association] criteria (21). For the 161 deceased subjects, a preliminary diagnosis was made by a physician who consulted medical records and death certificates, and the diagnosis was then reviewed by a senior clinician. When only discharge diagnoses from hospitals (n = 50) or only death certificates (n = 11) were available, the reported diagnosis was accepted. When we repeated the analyses in the survivors cohort, the results were similar. Measurement of educational attainment Information on total years of formal education was collected at baseline. In this study, educational level was divided into two main categories: 2 7 years (6 years of primary school and, in some cases, 1 extra year of practical vocational training) and 8 years (intermediate and university levels). This dichotomization was based on a previous analysis (4) in which different levels of education were examined in relation to Alzheimer s disease. Persons with an intermediate amount of education (8 10 years of schooling) did not differ from university-educated subjects ( 11 years of schooling) in terms of Alzheimer s disease risk. Data on educational background were missing for three persons. These subjects were omitted from all analyses concerning education.

3 Education, Socioeconomic Status, and Alzheimer s Disease 177 Assessments of occupation-based SES Main occupation-based SES. Specially trained nurses interviewed a relative or another person significant to the subject about the subject s full lifetime work history. The questionnaire was developed by an occupational hygienist. Up to eight occupational periods with specific information regarding time and place were listed. The subject s longestheld job was considered his or her main occupation. Occupations were grouped according to the socioeconomic classification system developed by Statistics Sweden (22). The Swedish socioeconomic classification system primarily contains dimensions of ownership of the means of production and division into blue-collar and white-collar occupations as assessed by normal trade-union affiliation. It also contains aspects of skills in the subdivisions inside bluecollar and white-collar occupational categories. However, customary educational requirements for the respective occupations are used, not information about education at the individual level. Quite a few women (21 percent) were homemakers during their longest occupational period. Information on husband s occupation, which is frequently used to estimate the SES of wives, was not available. Instead of including homemakers in a category of their own, we used their second-longest-held job to estimate their SES. Eighteen subjects had been housewives throughout their working lives and were excluded from the analyses involving main occupation. We first analyzed the relation of SES to incidence of Alzheimer s disease and dementia by using three categories (blue-collar workers, white-collar employees, and selfemployed persons plus those in academic professions). Since the two last categories had similar risks, the two groups were merged. Lifetime occupational SES. Data were gathered about the length of each occupational period over the subject s lifetime, and all occupations were classified as described above, with the exception that no substitutions from other occupational periods could be made for the homemaker work periods. Instead they were classified as nonmanual periods, because 1) these women often held nonmanual jobs in their other occupational periods and 2) they were generally more highly educated than the nonhousewives. Finally, we added up the numbers of years in all periods of manual labor to obtain the sum of years spent in a low occupation-based SES. In the current study, we report results from analyses in which the number of years in a low SES was less than or equal to 25 versus more than 25 (the median amount of time spent in a low SES by the low-ses fraction of the population). Socioeconomic mobility. We computed occupation-based SES at ages 20, 40, and 60 years in order to estimate individual socioeconomic mobility patterns. When information about a particular time period was missing, we substituted an equal number of years from the two closest work periods. Sixty-three persons lacked information about occupational position at 60 years of age; instead, information concerning the occupational period prior to that was chosen to represent it. Covariates We used all covariates with available information from our database that may have acted as confounding factors. As major possible confounders, we considered age, gender, vascular diseases, and alcohol data. Information on age and gender was derived from the Swedish National Population Register. Information on cerebrovascular and heart diseases and diabetes mellitus was derived from the computerized Stockholm Inpatient Registry System, in which all admission and discharge diagnoses from every hospital in Stockholm are recorded. In the current study, all diagnoses made at any time from 1969 to the baseline examination were used. Diseases were diagnosed according to the International Classification of Diseases, Eighth Revision (codes for cerebrovascular disease, codes for coronary disease, code 428 for heart failure, code 427 for arrhythmia, and code 250 for diabetes mellitus). Information on diabetes from the inpatient register was integrated with self-reports of a doctor s diagnosis of diabetes, use of oral hypoglycemic medication or insulin, or a blood glucose level greater than 11 mmol/ liter. Arterial blood pressure was measured at baseline with a mercury sphygmomanometer by trained nurses. If the first reading was abnormal (systolic pressure 160 mmhg or diastolic pressure 95 mmhg), two additional readings were made. The mean of the second and third readings was used for analyses (23). Furthermore, we created a simple additive index ranging from 0 to 5 for the vascular diseases or risk factors described above. Information on alcohol use was collected from relatives, who specified how many glasses of wine, bottles of beer, and/or glasses of liquor the proband consumed weekly. Since consumption of alcohol in this predominantly female elderly population was found to be particularly low (82.4 percent of the persons with available data on alcohol reported that they used no alcohol at all, and only 3.5 percent consumed an amount corresponding to more than five glasses of wine per week), we decided to divide the alcohol data into three categories: no use, any use, and no information. The latter category was included in the analyses because reporting on alcohol use can sometimes be a sensitive matter, and we found it relevant to investigate whether the absence of information was associated with the risk of dementia. Other covariates, which were controlled for in further analyses, were: 1) cognitive status at baseline, measured by the Mini-Mental State Examination (24); 2) social network at baseline, measured with a four-grade index (poor, limited, moderate, and rich social network) as defined in a previous study (25); and 3) engagement in mental and physical activities at baseline, graded in three frequency categories (26). Statistical analysis Logistic regression was used to evaluate the differences in baseline characteristics between participants and dropouts. Cox proportional hazards regression analyses were used to estimate the relative risk of incident Alzheimer s disease and dementia associated with different combinations of educa-

4 178 Karp et al. TABLE 1. Baseline characteristics of the study population (n = 931) by dementia status at follow-up, The Kungsholmen Project, Stockholm, Sweden, Characteristic Nondemented (n = 830) * Data on educational level were missing for three subjects. Dementia status at follow-up Alzheimer s disease (n = 76) All types of dementia (n = 101) No. % No. % No. % Age (years) Sex Male Female Educational level* Elementary/2 7 years Vocational training Intermediate/8 10 years High + university/ 11 years Occupation-based socioeconomic status Blue-collar worker White-collar employee Self-employed/academic profession Lifetime housewife Alcohol consumption No use of alcohol Any use of alcohol Missing data Coronary heart disease Heart failure + arrhythmia High blood pressure Cerebrovascular disease Diabetes mellitus tion and occupation-based SES, as well as socioeconomic mobility. The follow-up time for nondemented persons was estimated from the date of the baseline interview to the date of follow-up examination or death. For the demented persons, half of this time was calculated, since dementia onset was assumed to be the midpoint between baseline and follow-up examination. Age, gender, vascular disease index, and alcohol use were entered into all models. Additional models were used to verify the possible confounding effect of baseline Mini- Mental State Examination score, social network, and leisure activities. We also repeated all analyses after excluding the stroke cases, with the purpose of examining whether stroke might obscure or increase diagnostic difficulties in dementia due to cognitive dysfunction in subjects with previous stroke. The interrelation between education and SES was explored using two stratifications. First, the study population was divided into four strata characterized by 1) high education and high SES; 2) high education and low SES; 3) low education and high SES; and 4) low education and low SES. Second, we used four strata by combining low and high education as defined above, with permanence in low SES for less than or equal to 25 years versus more than 25 years. RESULTS Persons who dropped out of the study were comparable to participants in terms of age (1-year increment: odds ratio = 1.0, 95 percent confidence interval (CI): 0.98, 1.02), gender (female: odds ratio = 0.8, 95 percent CI: 0.6, 1.0), and vascular disease (presence of vascular disease: odds ratio = 0.8, 95 percent CI: 0.6, 1.0). The odds ratio for dropping out with regard to a low level of education was 1.4 (95 percent CI: 1.1, 1.7). However, the occurrence of dementia among

5 Education, Socioeconomic Status, and Alzheimer s Disease 179 TABLE 2. Relative risks of clinically diagnosed Alzheimer s disease or dementia associated with education and occupation-based socioeconomic status (lifetime longest-held job), derived from separate models (models 1 and 2) and from the same model (model 3), The Kungsholmen Project, Stockholm, Sweden, Model 1A Educational level Model 1B All subjects (no.) No. of cases Alzheimer s disease RR*, 95% CI* RR 95% CI * RR, relative risk; CI, confidence interval; SES, socioeconomic status. Relative risks were estimated after adjustment for age and gender. Relative risks were estimated after adjustment for age, gender, vascular diseases index, and alcohol data. No. of cases All types of dementia RR 95% CI RR 95% CI High (>10 years) Intermediate (8 10 years) , , , , 1.8 Low (2 7 years) , , , , 4.0 Educational level Model 2 High (>7 years) Low (2 7 years) , , , , 3.8 Occupation-based SES* Model 3 High Low , , , , 2.3 Low educational level vs. high , , , , 3.7 Low occupation-based SES vs. high , , , , 1.7 the 370 dropouts lacking information on occupational history was similar to that among participants in our study. A total of 101 subjects developed clinically definite dementia during the follow-up period. Among dementia cases of all types, 76 persons were diagnosed with clinically definite dementia of the Alzheimer s type. Table 1 provides characteristics of the participants by dementia status at follow-up. The relative risks for clinical dementia of the Alzheimer s type and of all types in relation to education and occupationbased SES are reported in table 2. Compared with persons with a high level of education ( 11 years), the adjusted relative risks of dementia of the Alzheimer s type and dementia of all types were similar for subjects with an intermediate amount (8 10 years) of education and elevated for subjects with a low level (2 7 years) of education (see model 1A). In model 1B, we merged the high and intermediate levels of education into one reference category. In model 1B and model 2, subjects with either a low level of education or a low occupation-based SES were found to be at higher risk of Alzheimer s disease and dementia, even after adjustment for all covariates examined. However, when both education and SES were introduced into the same model simultaneously (table 2, model 3), only low education remained a significant risk factor for Alzheimer s disease and dementia. These results were similar for men and women, although the relative risks for men were not statistically significant, probably because of less statistical power. Subjects in the low education/low occupation-based SES category were more often female (83 percent) than subjects TABLE 3. Relative risks of clinically diagnosed Alzheimer s disease or dementia associated with combinations of education and occupation-based socioeconomic status (lifetime longest-held job), The Kungsholmen Project, Stockholm, Sweden, All subjects (no.) * RR, relative risk; CI, confidence interval; SES, socioeconomic status. Relative risks were estimated after adjustment for age and gender. Relative risks were estimated after adjustment for age, gender, vascular diseases index, and alcohol data. No. of cases Alzheimer s disease RR*, 95% CI* RR 95% CI No. of cases All types of dementia RR 95% CI RR 95% CI High education/high SES* High education/low SES , , , , 4.4 Low education/high SES , , , , 4.1 Low education/low SES , , , , 4.0

6 180 Karp et al. TABLE 4. Relative risks of clinically diagnosed Alzheimer s disease associated with social mobility (occupation-based socioeconomic status at ages 20, 40, and 60 years), The Kungsholmen Project, Stockholm, Sweden, Occupation-based socioeconomic status Total no. of Model 1* Model 2 No. of cases at ages 20, 40, and 60 years subjects RR 95% CI RR 95% CI High/high/high High/high/low , , 2.6 High/low/high 6 0 High/low/low 12 0 Low/high/high , , 1.8 Low/high/low , , 2.1 Low/low/high , , 2.9 Low/low/low , , 1.7 Educational level (low vs. high) , 6.5 * Relative risks were estimated after adjustment for age, gender, vascular diseases index, and alcohol data. Relative risks were estimated after adjustment for age, gender, vascular diseases index, alcohol data, and education. RR, relative risk; CI, confidence interval. There were no incident cases in this category. in the high education/high occupation-based SES category (71 percent). The number of persons in the high education/ low occupation-based SES group was smaller (n = 36) in comparison with the other groups, and it contained the largest proportion of women (89 percent) and housewives (33 percent). The low education/high occupation-based SES group differed noticeably in that 26 percent of its members were self-employed, whereas only 9 percent of persons in the high education/high occupation-based SES group were selfemployed. Low education in combination with either low or high occupation-based SES was associated with increased risk of Alzheimer s disease and dementia, even after adjustment for the major covariates. The combination of high education and low occupation-based SES was not associated with any increased risk of Alzheimer s disease and dementia (table 3). The stratified analyses of the relations of combinations of education and total years in low occupation-based SES to Alzheimer s disease showed results similar to those of the stratified analysis depicted in table 3. Compared with persons with high education and less than 25 years in a low SES, the subgroup with both low education and 25 or more years in a low SES had an adjusted relative risk of 3.1 (95 percent CI: 1.6, 5.8), and in the subgroup with low education but fewer than 25 years in a low SES, the adjusted relative risk was 3.4 (95 percent CI: 1.9, 6.3). High education in conjunction with 25 or more years in a low SES was not associated with increased risk of Alzheimer s disease. Table 4, model 1, shows the relative risks of Alzheimer s disease in relation to eight social mobility patterns after controlling for age and gender. Having a low occupationbased SES at ages 20, 40, and 60 years resulted in a borderline-significant elevated risk of Alzheimer s disease compared with the reference category of high SES at all three ages. Having a low occupation-based SES at ages 20 and 40 years but not at age 60 years was also associated with significantly elevated risks of Alzheimer s disease. Furthermore, low occupation-based SES at 20 years of age, regardless of whether SES was high at age 40 and/or age 60, was associated with elevated Alzheimer s disease risks, though not statistically significantly. However, after additional adjustment for education, none of the significant associations remained. Twenty-eight percent of the subjects with high occupational SES at ages 20, 40, and 60 years but 91 percent of the subjects with low SES at ages 20, 40, and 60 years had a low educational level. Analyses of all-type dementia showed similar results. Finally, when the baseline cognitive status variables (Mini-Mental State Examination score, social network, and engagement in physical and mental activities) were included in the model as covariates, the relations of SES and education with Alzheimer s disease and dementia did not substantially change. When we repeated all analyses after excluding cases of stroke, the associations between education and Alzheimer s disease and between education and dementia were similar, and all results showed the same patterns as in the main analyses. DISCUSSION The main finding of this study is that a low level of education and a low occupation-based main SES are individually associated with increased risk of Alzheimer s disease and dementia, but only low education remains as a risk factor when both variables are examined simultaneously. A similar pattern was observed when numbers of years of all work periods in the low-ses position were added together and when life-span individual social mobility patterns were studied. In comparison with highly educated subjects who had a high-ses main occupation, a twofold increased risk of Alzheimer s disease was detected, not only among lesseducated subjects with low SES but also among subjects

7 Education, Socioeconomic Status, and Alzheimer s Disease 181 with low education and a high-ses position. In addition, subjects with an initially high occupation-based SES had no increased risk of Alzheimer s disease independently of later SES changes, but those persons who had a low occupationbased SES at a young age and remained in this category for most of their occupational life had a percent increased risk of Alzheimer s disease. However, even in this case, low education explained the associations. Our results are in agreement with previous findings of a positive association between a low level of education and increased risk of Alzheimer s disease and dementia (1 10). Evans et al. (2) reported that each of the socioeconomic measures (education, income, and occupational prestige) predicted Alzheimer s disease risk. However, when they included all three measures in the same model, only education (not income or occupational prestige) retained a significant association with Alzheimer s disease. Recently, investigators from the Nurses Health Study (27) reported a strong association between low education and cognitive decline but little association with other markers of SES. The fact that the association between low education and increased risk of Alzheimer s disease and dementia in general was present independently of main occupation-based SES, and the fact that the association of SES with Alzheimer s disease and dementia was mainly due to SES in early work life, seems to indicate that unknown education-related factors acting during the first two decades of life may be involved in the development of Alzheimer s disease and dementia. This interpretation is in agreement with a number of previous reports. In a recent article, Kaplan et al. (28) concluded that higher childhood socioeconomic position, as well as greater educational attainment, was associated with better cognitive function in adulthood. In a study of education and dementia in an Italian population with a middle-tohigh socioeconomic level, the first decade of life was suggested to be a critical period for development of dementia later in life (29). Hall et al. (30) discussed the possibility that low education by itself is not a major risk factor for Alzheimer s disease but rather a marker for other accompanying deleterious socioeconomic or environmental influences in childhood. Poor-quality childhood or adolescent environments have been suggested to prevent the brain from reaching complete levels of maturation, which in turn may put people at higher risk of Alzheimer s disease (31, 32). Results from the PAQUID Study (33) showed that occupation did not change the risk of Alzheimer s disease, which instead was related to cognitive abilities in childhood and adolescence. The early-life factors in question may be early socioeconomic factors, mental stimulation, or cognitive abilities. The cognitive abilities may be affected by both environmental (34) and genetic influences (9). Since many studies, like ours, lack an independent measure of intelligence, the importance of early-life cognitive abilities is difficult to evaluate. A number of studies have given special attention to the effect of early-life circumstances on health and mortality in general. Early-life SES has been reported to be connected with cardiovascular diseases and general health status later in life (35 37). Disadvantaged childhood living conditions have been hypothesized to trigger unhealthy life trajectories, where increased illness risk is the sum of several healthdamaging factors (38). New results from a random Swedish national sample indicate that occupational class differences in adult mortality depend, to a fairly large extent, on childhood conditions (39). We examined the relation between SES and Alzheimer s disease and dementia from a lifetime perspective, ranging from school to retirement, but unfortunately we lacked specific data on the SES of the family of origin. Thus, we can only suggest the involvement of earlylife factors in the development of Alzheimer s disease and dementia as a hypothesis. An alternative explanation for our findings is that education may be a better indicator of adult lifestyle and habits than occupational SES, and that lifestyle and habits such as alcohol use and smoking may explain the detected association. However, when our data were adjusted for alcohol use, and additionally when current social network index and engagement in physical and mental activities were entered into the model as covariates, we obtained similar results. Smoking was not related to Alzheimer s disease or dementia in our population (40). With regard to cognitive function in late middle age, Turrell et al. (17) reported that disadvantaged socioeconomic circumstances in childhood may, to some extent, be overcome by later upward mobility. However, in our study, low education remained associated with an increased risk of Alzheimer s disease even after adjustment for adult social mobility. Some potential limitations of this study must be considered. Persons who dropped out of the study were less educated than participants. Since low education was related to Alzheimer s disease and dementia, this may have affected our results, but it is more likely to have skewed our findings towards underestimation of the association than towards overestimation. The occupational data were obtained through informants. However, a single open-ended question about occupation (without any specification of time, place, or number of years) had already been posed to the participants during the baseline interview, and it was possible to compare these responses with the informant data regarding the subject s main job. When information from the two different sources was used to estimate SES, agreement was 80 percent. The Swedish system for classification of SES from occupation is well established and has been evaluated (41). However, it remains that occupation-based SES is more difficult to evaluate than education. Estimation of SES from occupation was difficult for the housewives in the cohort (21 percent). Our approximations were likely to create nondifferential misclassification of SES, leading to a dilution of the association. We performed additional analyses (data not shown) treating the housewives as a third occupational category, and this did not affect our results. Finally, we must be cautious in interpreting our findings as reflecting a direct effect of education on Alzheimer s disease risk and not one that is mediated by occupation-based SES. In fact, there is the possibility of unmeasured confounders being responsible for the patterns of associations found in the study. For example, work-related stress may be associated with occupation-based SES but not with education and may increase the risk of Alzheimer s disease and dementia,

8 182 Karp et al. though there is currently no clear evidence in the literature concerning such an association. The main conclusion that can be drawn from this study is that a low level of education is more directly associated with increased Alzheimer s disease incidence and is not mediated by low adult occupation-based SES, regardless of whether the low-ses occupational period was extended over a substantial number of years ( 25 years) and regardless of adult socioeconomic mobility pattern. The results are not inconsistent with the already-suggested hypotheses of cognitive or brain reserves (5, 6). Alternatively, these findings may partly reflect a detection bias by which subjects with a low level of education tend to be clinically diagnosed with Alzheimer s disease and dementia at an earlier point in time (4). However, the possibility that factors related to early-life SES may play a role in the development of Alzheimer s disease and dementia should not be disregarded. Our findings support this hypothesis and stress the importance of collecting data on early life conditions when studying dementia in old age. ACKNOWLEDGMENTS Research grants were received from the Swedish Council for Work Life Research (no. F0150/2000), the Swedish Council for Social Research (no ), The Gamla Tjänarinnor Foundation, and the Solstickan Foundation. The authors thank all members of the Kungsholmen Project study group for their cooperation in data collection and management. REFERENCES 1. Schmand B, Smit J, Lindeboom J, et al. Low education is a genuine risk factor for accelerated memory decline and dementia. J Clin Epidemiol 1997;50: Evans DA, Hebert LE, Beckett LA, et al. Education and other measures of socioeconomic status and risk of incident Alzheimer disease in a defined population of older persons. Arch Neurol 1997;54: Letenneur L, Gilleron V, Commenges D, et al. Are sex and educational level independent predictors of dementia and Alzheimer s disease? Incidence data from the PAQUID Project. J Neurol Neurosurg Psychiatry 1999;66: Qiu C, Bäckman L, Winblad B, et al. The influence of education on clinically diagnosed dementia: incidence and mortality data from the Kungsholmen Project. Arch Neurol 2001;58: Katzman R. Education and the prevalence of dementia and Alzheimer s disease. Neurology 1993;43: Stern Y, Gurland B, Tatemichi TK, et al. Influence of education and occupation on the incidence of Alzheimer s disease. JAMA 1994;271: Friedland RP. Epidemiology, education and the ecology of Alzheimer s disease. Neurology 1993;43: Plassman BL, Welsh KA, Helms M, et al. Intelligence and education as predictors of cognitive state in late life: a 50-year follow-up. Neurology 1995;45: Gatz M, Svedberg P, Pedersen NL, et al. Education and the risk of Alzheimer s disease: findings from the study of dementia in Swedish twins. J Gerontol B Psychol Sci Soc Sci 2001;56: P Fratiglioni L, Viitanen M, Agüero-Torres H, et al. Low education and AD: effects on diagnosis rather than risk. (Abstract). Neurobiol Aging 1998;19(suppl 4): Winkleby MA, Jatulis DE, Frank E, et al. Socioeconomic status and health: how education, income, and occupation contribute to risk factors for cardiovascular disease. Am J Public Health 1992;82: Davey Smith G, Hart C, Hole D, et al. Education and social class: which is the more important indicator of mortality risk? J Epidemiol Community Health 1998;52: Lynch J, Kaplan G. Socioeconomic position. In: Berkman LF, Kawachi I, eds. Social epidemiology. New York, NY: Oxford University Press, 2000: Smith GD, Hart C, Blane D, et al. Lifetime socioeconomic position and mortality: prospective observational study. BMJ 1997; 314: Power C, Manor O, Matthews S. The duration and timing of exposure: effects of socioeconomic environment on adult health. Am J Public Health 1999;89: Lynch JW, Kaplan GA, Shema SJ. Cumulative impact of sustained economic hardship on physical, cognitive, psychological and social functioning. N Engl J Med 1997;337: Turrell G, Lynch JW, Kaplan GA, et al. Socioeconomic position across the lifecourse and cognitive function in late middle age. J Gerontol B Psychol Sci Soc Sci 2002;57:S Qiu C, Karp A, von Strauss E, et al. Lifetime principal occupation and the risk of Alzheimer s disease in the Kungsholmen Project. Am J Ind Med 2003;43: Fratiglioni L, Viitanen M, von Strauss E, et al. Very old women at highest risk of dementia and Alzheimer s disease: incidence data from the Kungsholmen Project, Stockholm. Neurology 1997;48: American Psychiatric Association. Diagnostic and statistical manual of mental disorders, Third Edition, Revised (DSM-III- R). Washington, DC: American Psychiatric Association, 1987: McKhann G, Drachman D, Folstein M, et al. Clinical diagnosis of Alzheimer s disease: report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer s Disease. Neurology 1984; 34: Statistics Sweden. Swedish socioeconomic classification, reports on statistical co-ordination. Stockholm, Sweden: Statistics Sweden, Guo Z, Viitanen M, Winblad B. Low blood pressure and fiveyear mortality in a Stockholm cohort of the very old: possible confounding by cognitive impairment and other factors. Am J Public Health 1997;87: 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: Fratiglioni L, Wang HX, Ericsson K, et al. The influence of social network on the occurrence of dementia: a communitybased longitudinal study. Lancet 2000;355: Wang HX, Karp A, Winblad B, Fratiglioni L. Late-life engagement in social and leisure activities is associated with a decreased risk of dementia: a longitudinal study from the Kungsholmen Project. Am J Epidemiol 2002;155: Lee S, Kawachi I, Berkman L, et al. Education, other socioeconomic indicators, and cognitive function. Am J Epidemiol 2003;157: Kaplan G, Turrell G, Lynch JW, et al. Childhood socioeconomic position and cognitive function in adulthood. Int J Epide-

9 Education, Socioeconomic Status, and Alzheimer s Disease 183 miol 2001;30: De Ronchi D, Fratiglioni L, Rucci P, et al. The effect of education on dementia occurrence in an Italian population with middle to high socioeconomic status. Neurology 1998;50: Hall KS, Gao S, Unverzagt FW, et al. Low education and childhood rural residence: risk for Alzheimer s disease in African Americans. Neurology 2000;54: Moceri VM, Kukull WA, Emanuel I, et al. Early-life risk factors and the development of Alzheimer s disease. Neurology 2000;54: Moceri VM, Kukull WA, Emanual I, et al. Using census data and birth certificates to reconstruct the early-life socioeconomic environment and the relation to the development of Alzheimer s disease. Epidemiology 2001;12: Helmer C, Letenneur L, Rouch I, et al. Occupation during life and risk of dementia in French elderly community residents. J Neurol Neurosurg Psychiatry 2001;71: Del Ser T, Hachinski V, Merskey H, et al. An autopsy-verified study of the effect of education on degenerative dementia. Brain 1999;122: Kaplan GA, Salonen JT. Socioeconomic conditions in childhood and ischaemic heart disease during middle age. BMJ 1990;301: van de Mheen H, Stronks K, Looman CW. Does childhood socioeconomic status influence adult health through behavioural factors? Int J Epidemiol 1998;27: Wamala SP. Socioeconomic status and cardiovascular vulnerability in women: psychosocial, behavioral, and biological mediators. (Doctoral thesis). Stockholm, Sweden: Karolinska Institute, Lundberg O. The impact of childhood living conditions on illness and mortality in adulthood. Soc Sci Med 1993;36: Kåreholt I. The long shadow of socioeconomic conditions in childhood: do they affect class inequalities in mortality? In: Jonsson JO, Mills C, eds. Cradle to grave: life-course change in modern Sweden. Durham, United Kingdom: Sociology Press, 2002: Wang HX, Fratiglioni L, Frisoni GB, et al. Smoking and the occurrence of Alzheimer s disease: cross-sectional and longitudinal data in a population-based study. Am J Epidemiol 1999; 149: Andersson LG, Erickson R, Wärneryd B. To describe the social structure an evaluation of the 1974 suggestion for socioeconomic classification. (In Swedish). Stat Tidskr 1981;19: