Background and Purpose The Canadian Study of Health and Aging (CSHA) was conducted in communities and institutions in 10 Canadian provinces. One objective of the study was to study risk factors for vascular dementia (VaD).
Methods This was a population-based case-control study. It included 129 patients clinically diagnosed with VaD with duration of symptoms no more than 3 years and 535 control subjects, frequency matched by age group, study center, and residence in community or institution, who were clinically confirmed to be cognitively normal. Odds ratios (ORs) were calculated using unconditional logistic regression for potential risk factors for VaD.
Results Risk of VaD was associated with history of arterial hypertension (OR, 2.08; 95% confidence interval, 1.29 to 3.35). Other significantly elevated ORs were seen for history of alcohol abuse (2.45), history of heart condition (1.71), use of aspirin (3.10), and occupational exposure to pesticides and herbicides (2.60), as well as liquid plastic or rubber (2.59). The OR for less than 6 years of education compared with 10 or more years was 4.02.
Conclusions The study confirmed some previously reported risk factors for VaD, such as history of heart disease. Higher levels of education seemed to lower the risk or delay onset of symptoms of VaD. Use of aspirin may be a predictor of survival rather than a risk factor. The occupational associations, particularly with pesticides and fertilizers, need further study.
Dementia is becoming one of the most important disorders in our aging society. Its prevalence in Canada has been estimated at 8% in persons aged ≥65 years and reaches 34.5% among those aged ≥85 years.1
VaD is second to AD as the leading cause of progressive irreversible dementia in Canada,1 being responsible for 24% to 48% of dementing illness in the elderly population. The epidemiology of this condition has been recently reviewed by Hébert and Brayne.2 Its prevalence increases with age and varies greatly from country to country, ranging from 1.2% to 4.2% of those aged ≥65 years. Although strokes are more common in men, the prevalence of VaD is similar in both sexes. The incidence of VaD is about 6 to 10 cases per 1000 persons aged >70 years per year. The mean duration of the disease is about 5 years, and survival is less than for the general population and for those with AD.
Essentially, VaD designates cognitive deterioration caused by ischemia, usually as a result of occlusion of cerebral arteries, but also due to other causes, including hypoperfusion.3 In consequence, VaD represents a heterogeneous group of conditions such as multi-infarct dementia, strategic single-infarct dementia, lacunar dementia, Binswanger’s subcortical encephalopathy, cerebral amyloid angiopathy, leukoaraiosis associated with dementia, hypoperfusion syndromes, and hemorrhagic dementia.
Because VaD is a preventable type of dementia,4 the determination of risk factors for VaD is extremely important. An opportunity to analyze risk factors for VaD was provided by the CSHA, which was designed to estimate the prevalence of different types of dementia and to examine risk factors associated with those diseases. The study was conducted in 18 CSHA study centers distributed across all Canadian provinces, and it was coordinated by the University of Ottawa in collaboration with the Laboratory Centre for Disease Control.
The prevalence rates of VaD observed in this study1 were 7 per 1000 in the community and 115 per 1000 in institutions, for a global rate of 15 per 1000 population aged ≥65 years. There was a linear increase of prevalence rates with age but no difference according to sex.
This article describes the risk factors associated with VaD using a population-based case-control design.
Materials and Methods
The design of the CSHA has been described in detail elsewhere.1 5 The study was peer reviewed by a special committee convened by the National Health Research and Development Program of Health Canada. Ethics reviews were carried out in each institution that took part in conducting the study. Subjects or their proxies signed letters of consent covering their participation in the study and were free to withdraw at any point.
Study subjects were recruited from both the community and institutions on the basis of age-stratified (65 to 74, 75 to 84, and 85+ years) random samples in 36 communities. Subjects were aged ≥65 years as of October 1, 1990, and were fluent in English or French. Data collection took place between February 1991 and May 1992.
Participants living in the community were first screened for cognitive impairment using the 3MS6 7 and were asked to undergo a clinical examination if the screening test indicated cognitive impairment (defined for this study as a score of <78/100) or if they were unable to complete the screening test because of deafness or other impairments. A number of subjects who screened cognitively normal were also examined clinically. All institutionalized participants were examined clinically without first being screened.
The clinical examination has been detailed elsewhere.1 Briefly, a nurse first gathered data by standard methods using the CAMDEX history interview,8 repeated the 3MS, and measured vital signs. Next, a physician recorded a standard history and physical examination. From this, a preliminary diagnosis was made according to standard criteria. The physician also completed the Hachinski Ischemia Score9 (a score of ≥7 was considered indicative of VaD). Dementia and its severity were diagnosed according to the criteria of the Diagnostic and Statistical Manual of Mental Disorders (DSM-III-R).10 AD was diagnosed using the criteria of McKhann et al.11 VaD was diagnosed on clinical grounds and without routine neuroimaging according to the ICD-10 criteria.12 ICD-10 requires, in addition to the general criteria for dementia being met, that the deficits in higher cortical function must be “unevenly distributed,” that there is clinical evidence of focal brain damage, and that there is evidence of cerebrovascular disease judged to be etiologically related to the dementia. Consideration of these features overlaps in many aspects with the Ischemia Score, but the latter did not form part of the diagnostic algorithm. The ICD-10 criteria specify five major subtypes (acute onset, multi-infarct dementia, subcortical vascular dementia, mixed cortical and subcortical vascular dementia, and other or unspecified) that were recorded but, in the absence of routine neuroimaging, have not been further analyzed.
As detailed in a related report,13 the validity of the diagnostic standardization was tested formally in two ways. All 2914 diagnoses were reviewed in a computer-checking procedure against an expert diagnostic algorithm, with concordance in 98% of cases on the “first pass.” As an adjunct to the algorithm checking, the Ischemia Scores of those with ICD-10 VaD diagnoses (n=207) were reviewed; 85% had scores of ≥7. In addition to these tests of internal consistency, in 210 cases, interrater reliability (which, when experts are used, also serves as a form of convergent construct validation) was carried out. The κ value was 0.81 for dementia. This review included 14 VaD cases, with agreement on 10 and reassignment to cognitive impairment without dementia or AD in the remaining cases. Overall agreement remained strong at 0.70.
Subjects scoring 50 or more on the 100-point 3MS received a detailed neuropsychological test battery (described elsewhere).14 This battery took between 2 and 4 hours to complete and included specific tests of memory, abstract thinking, judgment, language, visuospatial function, and attention and concentration. Standard laboratory tests, including random (nonfasting) plasma glucose measurements, were carried out to investigate possibly reversible causes of dementia. After the additional neuropsychological and laboratory data were available, a multidisciplinary case conference was held, at which time a final diagnosis was made using the same criteria described above.
Risk Factor Study
Case subjects were selected for the risk factor study if they were diagnosed with VaD. To reduce the potential for prevalence-incidence bias, only cases of recent onset (ie, onset of symptoms not more than 3 years before the study diagnosis) were included. Those who were diagnosed as cognitively normal in institutions and those who were screened and diagnosed as normal in the community formed the control group for this study. An exception was made to the composition of the case and control groups when we looked at education as a risk factor.
Performance on the screening examination was affected by level of education, and this may have introduced a bias when considering education as a risk factor for VaD. A higher proportion of early cases may have been detected among the less educated (who would be more likely to screen positive) than among the more educated (who would be less likely to screen positive). In addition, the requirement that control subjects screen negative (as well as be clinically diagnosed as cognitively normal) may have created a more highly educated control group. These factors could have accounted for all or part of any elevation in the ORs for those with less education.
To correct this bias, we introduced a sliding cut point on the 3MS according to number of years of education, which gave persons with different levels of education an equal probability of being included as a case subject. This meant that as the level of education decreased, we used lower scores on the 3MS as cut points. We excluded subjects whose 3MS score was above the new cut point according to their level of education. We also added new control subjects if their score was above the new cut point (ie, they now met the criteria of both screening negative and being diagnosed as cognitively normal). The resulting ORs for the different levels of education (Table 2⇓) should give an unbiased estimate of the effect of education as a risk factor for VaD.
The study design included frequency matching of control subjects to case subjects by study center, residence in community or institution, and age group (65 to 74, 75 to 84, and 85+ years). Subsequent experience showed this to be unfeasible, so we controlled for age and residence in all analyses.
It was necessary to have proxy respondents (usually a close relative) complete the risk factor questionnaire for the case subjects; to minimize bias, the questionnaires for the control subjects were also completed by proxies. The questionnaire covered demographic characteristics, occupational and environmental exposures, lifestyle (including smoking, alcohol, and a limited dietary history), as well as family and medical history (including antecedent diseases and medication use). The questionnaire was designed to be completed by the proxies themselves, although in seven centers an interviewer administered it. In addition, data on the following risk factors were obtained during the clinical examination: high blood pressure (systolic >160/≤95, systolic and diastolic >160/>95, normal blood pressure ≤160/≤95), orthostatic hypotension (decrease of 20 mm Hg in systolic pressure from supine to sitting position), history of arterial hypertension, presence of cardiac symptoms, BMI (weight in kilograms/[height in meters]2), current use of aspirin and all nonsteroidal anti-inflammatory drugs, and history of head trauma.
All analyses were conducted controlling for age (as a continuous variable), sex, education (as a continuous variable) except where otherwise specified, and residence in community or institution. ORs were calculated as estimates of the relative risk15 using unconditional logistic regression. Confidence intervals (95%) are shown.
Response rates for participation in the screening interview and the clinical examination have been reported elsewhere.1 Response rates for the risk factor questionnaire for the case subjects were 84% for those in institutions and 90% for those in the community who screened cognitively impaired. The response rate for the control subjects was 89%.
Description of the Study Population
The age/sex distribution of the sample is presented in Table 1⇓. Frequency matching by age group in the study design was effective overall (mean age for cases, 80.4 years; for controls, 79.0 years). However, there was some discrepancy in the proportions of case and control subjects in the two older age groups (75 to 84 and 85+ years), reflecting the difficulty of finding cognitively normal subjects in the oldest age group.
There were more women than men as both case and control subjects. Controls had completed more years of education than cases (the average number of years for cases was 8.7; for controls, 10.6; P<.01). A higher proportion of cases than controls were residents in institutions (55.8% versus 25.6%). Proxies were roughly equally likely to be sons or daughters for both case and control subjects (43.8% and 41.3%, respectively).
Results are presented for those risk factors for which there existed a prior hypothesis, including risk factors for stroke, since stroke and vascular dementia are so closely linked. In addition, results are presented for additional risk factors that were statistically significantly elevated.
The close link of VaD with stroke is demonstrated by the fact that a much larger percentage of case than control subjects had a history of stroke (79.5% versus 6.3%). Persisting focal neurological symptoms were also much more common in cases than controls (50.9% versus 3.5%), and to a lesser extent, the same was true for reversible (8.3% versus 3.1%) and transient (18.0% versus 5.2%) focal neurological symptoms and for history of paralysis (36.0% versus 4.8%).
Risk of VaD was not related to current hypertension or orthostatic hypotension (see Table 2⇓) but was related to history of arterial hypertension, with an OR of 2.08 (95% CI, 1.29 to 3.35). The OR for cardiac symptoms was not significantly high at 1.35. An elevated risk approaching statistical significance was seen for BMI of <20 compared with BMI of 20 to 25, with an OR of 2.02 (95% CI, 0.98 to 4.13). The risk for those with BMI >25 was not high.
Table 2⇑ details the relationship with the other risk factors documented by the questionnaire. History of alcohol abuse (OR, 2.45), history of heart condition (1.71), use of aspirin (3.10), and occupational exposure to pesticides and herbicides (2.60) and liquid plastic or rubber (2.59) were significantly associated with VaD. In addition, subjects with less education were at higher risk of VaD (OR of 4.02 for those with 0 to 6 years compared with those with ≥10 years).
One of the major strengths of the study is that it was population-based, with the control subjects drawn from the same population as the case subjects. This allows generalization of the results to the Canadian population.
Studies that include prevalent cases may be biased by factors associated with survival (Neyman’s bias). These protective factors could be wrongly identified as risk factors because they allow individuals to survive. In this study, we tried to overcome this bias by limiting the analysis to cases with symptoms occurring not more than 3 years before the study diagnosis. However, because survival from VaD is quite low, particularly in individuals at high vascular risk, Neyman’s bias could still be present. This phenomenon could explain the surprisingly high OR associated with use of aspirin, especially since the case subjects had taken aspirin for an average of only 2.3 years (compared with 3.7 years for the control subjects).
This study did not systematically include the use of a CT scan in the diagnosis of VaD, as required by two sets of criteria recently proposed by the NINDS-AIREN group (National Institute of Neurological Disorders and Stroke and the Association Internationale pour la Recherche et l’Enseignement en Neurosciences)16 and the California group.17 Use of CT scans in diagnosing dementia increases the proportion of dementia labeled as mixed or vascular by the visualization of vascular lesions in subjects previously labeled as AD, but it does not have an impact on demented subjects already clinically labeled as having VaD.18 In nondemented subjects, the presence of vascular lesions on the CT scan would not have excluded them from the control group. Consequently, the absence of CT scan in the diagnostic criteria was not associated with misclassification of case and control subjects.
The OR for history of stroke was not presented because stroke is one of the criteria for diagnosing VaD. In addition, the Hachinski Ischemia Score includes an item on the presence of high blood pressure, so the risk of VaD associated with this factor was overestimated.19 However, this scale was not included in the diagnostic algorithm. History, clinical examination, and neuropsychological tests played a much larger role than the Hachinski scale in the clinical decision. Despite these cautions, our study confirms the significant role of high blood pressure observed in previous studies.2 20 21 22 23 Our finding that a history of high blood pressure is a more important risk than the presence of measured hypertension extends an earlier analysis of vascular risk factors in all causes of vascular impairment, including VaD, in which no difference between treatment of vascular risk factors or in vascular risk factor control was observed in patients with VaD.24 Importantly, these data are also consistent with recent observations that midlife but not late-life hypertension is associated with late-life cognitive impairment.25 26 27 28 Indeed, late-life cognitive impairment, including vascular and other dementias, can be associated with normal or low late-life blood pressure.25 26 27 28 29 The presence of past or concurrent heart disease was also in agreement with a previous study.22 However, history of diabetes was not associated with VaD in this study. This contradicts the observations of Meyer et al22 and the well-established link between diabetes and stroke. This could be explained by a survival bias, with diabetic subjects being more prone to death.
In this study, those subjects with higher levels of education were at reduced risk of VaD. This is consistent with the literature on VaD,30 31 as well as for AD5 32 and cognitive decline in general.33 34 35 If it is true that increased education or mental activity leads to greater brain density,32 it could be argued that, as for AD, more damage can occur to the brain before symptoms become obvious. Alternatively, as an indication of socioeconomic status, education might be a surrogate for other factors that could affect the risk of VaD, such as diet.
The OR for smoking and VaD was below 1, whereas Meyer et al22 saw an elevation in risk. Our results could be due to the decreased survival of smokers; ie, those who smoked and were actually at increased risk of stroke and VaD (or other smoking-related diseases) may already have died.
History of alcohol use or abuse was a risk factor for VaD in this study and in the Hisayama cohort in Japan36 but not in the study of Meyer et al.22 Further study is needed to determine whether this is a causal association, since it is potentially preventable.
The association between occupational exposure to pesticides and fertilizers and VaD is interesting because we also saw a somewhat elevated OR for AD5 and because exposure to pesticides has been linked with Parkinson’s disease,37 38 all of which are neurological diseases. A prospective study that collected detailed exposure data would be necessary to shed more light on these associations.
Occupational exposure to liquid plastics or rubbers as a risk factor for VaD should be further considered for biological plausibility before it is discounted as a chance finding or studied in more detail.
This study showed a nearly significant association of low BMI with VaD. Thinness is probably more a consequence rather than a cause of VaD. This association has been observed in many studies, but the exact mechanism is not well understood.39
This study has confirmed some previously observed risk factors for VaD and introduced some new findings. The literature on risk factors for VaD is still sparse, and of course, replication of findings is necessary to infer causality. Further study is merited, particularly of some apparent risk factors that are potentially modifiable and for which preventive measures could be usefully promoted. We are currently following up the participants of the CSHA. One of the objectives of this second study is to conduct a cohort analysis of risk factors for VaD, which should contribute further to our understanding of this problem.
Selected Abbreviations and Acronyms
|BMI||=||body mass index|
|CHSA||=||Canadian Study of Health and Aging|
|ICD-10||=||International Classification of Diseases, 10th revision|
|3MS||=||Modified Mini-Mental State Examination|
The data reported in this article were collected as part of the Canadian Study of Health and Aging. This study was funded by the Seniors Independence Research Program and administered by the National Health Research and Development Program of Health and Welfare Canada (project 6606-3954-MC[S]). The study was coordinated through the University of Ottawa and the Canadian Government’s Laboratory Centre for Disease Control.
Reprint requests to Dr Joan Lindsay, Cancer Bureau, Laboratory Centre for Disease Control, Health Canada, Address Locator 0602E2, Ottawa, ON K1A 0L2, Canada.
- Received September 24, 1996.
- Revision received November 27, 1996.
- Accepted November 27, 1996.
- Copyright © 1997 by American Heart Association
Canadian Study of Health and Aging Working Group. Canadian Study of Health and Aging: study methods and prevalence of dementia. Can Med Assoc J. 1994;150:899-913.
Canadian Study of Health and Aging. The Canadian Study of Health and Aging: risk factors for Alzheimer’s disease in Canada. Neurology. 1994;44:2073-2080.
Hébert R, Bravo G, Girouard D. Validation de l’adaptation française du Modified Mini-Mental State (3MS). Rev Gériatr. 1992;17:443-450.
Roth M, Huppert FA, Tym E, Mountjoy CQ. CAMDEX: the Cambridge Examination for Mental Disorders of the Elderly. Cambridge, UK: Cambridge University Press; 1988.
American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders, 3rd ed, Revised. Washington, DC: American Psychiatric Association; 1987.
McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan EM. 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;34:939-944.
World Health Organization. International Classification of Diseases and Related Health Problems, 10th Revision. Geneva, Switzerland: World Health Organization; 1992.
Schlesselman JJ. Case-Control Studies. New York, NY: Oxford University Press; 1982.
Román GC, Tatemichi TK, Erkinjuntti T, Cummings JL, Masdeu JC, Garcia JH, Amaducci L, Orgogozo J-M, Brun A, Hofman A, Moody DM, O’Brien MD, Yamaguchi T, Grafman J, Drayer BP, Bennett DA, Fisher M, Ogata J, Kokmen E, Bermejo F, Wolf PA, Gorelick PB, Bick KL, Pajeau AK, Bell MA, DeCarli C, Culebras A, Korczyn AD, Bogousslavsky J, Hartmann A, Scheinberg P. Vascular dementia: diagnostic criteria for research studies. Report of the NINDS-AIREN International Workshop. Neurology. 1993;43:250-260.
Chui HC, Victoroff JI, Margolin D, Jagust W, Shankle R, Katzman R. Criteria for the diagnosis of ischemic vascular dementia proposed by the State of California Alzheimer’s Disease Diagnostic and Treatment Centers. Neurology. 1992;42:473-480.
Schoenberg BS. Epidemiology of vascular and multi-infarct dementia. In: Meyer JS, Lechner H, Marshall J, Toole JF, eds. Vascular and Multi-Infarct Dementia. Mount Kisco, NY: Futura Publishing Co; 1988.
Skoog I. Risk factors for vascular dementia: a review. Dementia. 1994;5:137-144.
Ueda K, Kawano H, Hasuo Y, Fujishima M. Prevalence and etiology of dementia in a Japanese community. Stroke. 1992;23:798-803.
Meyer JS, McClintic KL, Rogers RL, Sims P, Mortel KF. Aetiological considerations and risk factors for multi-infarct dementia. J Neurology Neurosurg Psychiatry. 1988;51:1489-1497.
Forette F, Boller F. Hypertension and the risk of dementia in the elderly. Am J Med. 1991;90(suppl 3A):3A-14S-3A-19S.
Rockwood K, Ebly E, Hachinski V, Hogan D. Presence and treatment of vascular risk factors in vascular cognitive impairment. Arch Neurol. In press.
Guo Z, Viitanen M, Fratiglioni L, Winblad B. Low blood pressure and dementia in elderly people: the Kungsholmen project. Br Med J. 1996;312:805-808.
Elias MF, Wolf PA, D’Agostino RB, Cobb J, White LR. Untreated blood pressure level is inversely related to cognitive functioning: the Framingham Study. Am J Epidemiol. 1993;138:353-364.
Rockwood K, Lindsay J, McDowell I. High blood pressure and dementia. Lancet. 1996;348:65. Letter.
Mortel KF, Meyer JS, Herod B, Thornby J. Education and occupation as risk factors for dementia of the Alzheimer and ischemic vascular types. Dementia. 1995;6:55-62.
Katzman R. Education and the prevalence of dementia and Alzheimer’s disease. Neurology. 1993;43:13-20.
Yoshitake T, Kiyohara Y, Kato I, Ohmura T, Iwamoto H, Nakayama K, Ohmori S, Nomiyama K, Kawano H, Ueda D, Sueishi K, Tsuneyoshi M, Fujishima M. Incidence and risk factors of vascular dementia and Alzheimer’s disease in a defined elderly Japanese population: the Hisayama Study. Neurology. 1995;45:1161-1168.
Seidman-Ripley J. Monograph series on aging-related diseases, II: Parkinson’s disease. Chron Dis Can. 1993;14:34-47.
Barbeau A, Roy M, Cloutier T, Plasse L, Paris S. Environmental and genetic factors in the etiology of Parkinson’s disease. Adv Neurol. 1986;45:299-306.