Cerebrovascular Disease and Depression Symptoms in the Cardiovascular Health Study
Background and Purpose—Evidence is mounting linking cerebrovascular disease with depressive symptoms in the elderly. Lesions in both white and gray matter have been associated with depressive symptoms and major depression. We sought to investigate the relationship between depressive symptoms and white and gray matter lesions in subjects participating in the Cardiovascular Health Study.
Methods—In a sample of 3660 men and women who underwent a standardized interview, physical examination, and MRI scan, we examined the association between number of white and gray matter lesions and white matter grade (a measure of severity) and reported depressive symptoms using a modified version of the Centers for Epidemiologic Studies Depression (CES-D) scale. We controlled for a variety of demographic and medical variables as well as functional status and Modified Mini-Mental State Examination score.
Results—The number of small (<3 mm) basal ganglia lesions was significantly associated with reported depressive symptoms, but white matter grade was not. In subsequent logistic regression models, number of basal ganglia lesions remained a significant predictor after controlling for non-MRI variables and severity of white matter lesions.
Conclusions—Our findings extend previous reports that linked cerebrovascular changes to depressive symptoms in clinical populations to a large community-based population. This report provides further evidence of the importance of basal ganglia lesions in geriatric depression.
There is growing evidence that cerebrovascular disease, even in the absence of clinical stroke, may be related to development of depression. The term “vascular depression” has been used to describe a specific syndrome of major depression found in individuals with cerebrovascular disease, seen as white and gray matter hyperintensities as found on T2-weighted MRI brain scans.1 2 3 Such patients tend to have as part of their depression significant loss of interest (apathy), cognitive impairment, and psychomotor retardation.4 Most studies that have investigated the relationship between cerebrovascular disease and depression have examined clinical samples of depressed patients and controls, usually elderly, with and without cerebrovascular disease. There is general agreement that older depressed patients tend to have a greater severity of MRI hyperintensities than do age-matched controls and that, among patients with geriatric depression, those with late-onset depression (first depressive episode when older than 50 years) have more cerebrovascular disease than early-onset depressives. Thus far, these findings have been extended to only a limited degree to larger populations of individuals. The Cardiovascular Health Study (CHS) provides an opportunity to examine the role of cerebrovascular pathology and depressive symptoms in a large population-based sample of older adults.
The CHS is a multicenter longitudinal study of risk factors for coronary heart disease and stroke in 5888 men and women aged 65 years and older.5 6 As part of their standardized clinical evaluation, 3657 men and women have undergone brain MRI. A previous report has examined clinical correlates of MRI white matter changes7 but did not examine depressive symptoms. We sought to examine the relationship between MRI vascular changes and depression in the CHS. While the CHS protocol did not include formal psychiatric evaluation, information about depressive symptoms was obtained on subjects with a modified version of the Center for Epidemiologic Studies Depression (CES-D) scale. We hypothesized that subjects with higher CES-D scores would have greater severity of white and gray matter pathology than those in the group with lower CES-D scores. Furthermore, we hypothesized that these differences would persist after controlling for age, sex, race, systemic vascular illness, cognitive status, and impairment in activities of daily living (ADL).
Subjects and Methods
The original CHS cohort consisted of 5201 subjects recruited from a random sample of Health Care Financing Administration Medicare eligibility lists between 1989 and 1990 in 4 communities across the country: Forsyth County, North Carolina; Sacramento County, California; Washington County, Maryland; and Pittsburgh (Allegheny County), Pennsylvania. Between 1992 and 1993, an additional 687 black participants were recruited from the centers in Pennsylvania, California, and North Carolina, bringing the total CHS cohort to 5888 subjects. Details regarding the CHS study design and characteristics of the original CHS cohort have been published previously.5 6 All eligible participants had to be 65 years or older, able to give informed consent, and able to respond to questions without a proxy respondent. Potential respondents who were wheel-chair bound at home, institutionalized, or undergoing cancer treatment at baseline were excluded. Among the eligible participants, 58% agreed to enroll in the study and undergo baseline evaluation. Participants who enrolled in the study were more likely to be younger, more educated, and more likely to be married than nonparticipants.5 6 The present analyses included 3660 subjects who agreed to have an MRI brain scan in years 4, 5, and 6 of the study (see MRI section below).
Depression was assessed with the use of the shortened (10-item) version of the CES-D scale8 administered close to the time of the MRI. As with the full-version CES-D, the shortened version is a self-reported measure of depressive symptoms experienced in the past week. Its 10 items are coded on a scale of 0 to 3 points, for a maximum of 30 points, and focus on mood (5 items), level of irritability (1 item), energy level (2 items), concentration (1 item), and sleep (1 item). Higher score indicates greater depressive symptoms.
Functional and Cognitive Assessment
Functional impairment was assessed by subjects’ self-report of difficulty in tasks of ADL and instrumental activities of daily living (IADL). A modified version of the Health Interview Survey Supplement on Aging questionnaire was used to assess both ADL and IADL.9 The ADL section assessed ability to bathe, dress, and feed and toilet oneself, as well as locomotion and continence.10 The IADL section assessed the ability to use a telephone, shop, prepare food, perform light household work, perform heavy household work, and handle finances.11 The Modified Mini-Mental State Examination (3MS)12 was used to measure cognitive status. The 3MS examination expands on the original Folstein Mini-Mental State Examination (MMSE)13 and evaluates a broader array of cognitive functions as well as a wider range of difficulty levels on an objective scale of 0 to 100. The 3MS includes tests of orientation, registration, attention, calculation, recall, language, and visuospatial skills.
Presence of hypertension was determined by seated resting blood pressure measurements at the clinic visit following a standardized protocol and by ascertaining history of hypertension and antihypertensive medication use. Subjects were classified as hypertensive if systolic blood pressure was ≥160 mm Hg, if diastolic was ≥95 mm Hg, or if they reported a history of hypertension and were on antihypertensive medication. Subjects were classified as borderline hypertensive if systolic blood pressure ranged between 140 and 159 mm Hg or diastolic blood pressure ranged between 90 and 94 mm Hg. Subjects were classified as normotensive if systolic blood pressure was <140 mm Hg, diastolic blood pressure was <90 mm Hg, and there was no reported history of hypertension. Subjects were asked whether they had coronary heart disease or had ever had a myocardial infarction. Medical records were obtained, and the information was evaluated by a physician panel using standard criteria to detect cardiovascular disease events.14 15 Coronary heart disease was included as present if there was reported and confirmed myocardial infarction and/or reported and confirmed angina and/or prior coronary revascularization procedures.16 Age at the time of the MRI was also provided.
As part of the extended CHS follow-up protocol, all members of the cohort were invited to have cranial MRI scans during years 4, 5, or 6 of the study.16 17 Study participants without contraindication to MRI scanning (ie, metallic or electric implants) and who were able to tolerate the procedure underwent the standardized imaging protocol. MRI scans were performed with 1.5-T scanners at 3 field centers and a 0.35-T instrument at the fourth field center using acquisition sequences previously reported.16 The MRI scans were reviewed by trained readers blinded to clinical data. For lesions >3 mm, the κ statistics were 0.59 and 0.63 for interrater and intrarater reliability, respectively.18
A white matter lesion severity score was determined by grading the extent of increased white matter signal intensity on the spin-density images in the periventricular and subcortical white matter area. The grading score was on a 10-point scale from 0 to 9, with a higher score indicating higher white matter grade.18 19
Lesions in cortical areas and in the basal ganglia were also examined. They were included if they were focal, nonmass lesions having a vascular pattern and were hyperintense to gray matter on both spin-density and T2-weighted images.18 Previous CHS publications have termed these lesions “infarctlike lesions.”18 20 Cortical lesions were included if they were located in frontal, parietal, temporal, or occipital cortical areas. Basal ganglia lesions were included in the caudate nucleus, lentiform nucleus, internal capsule, and thalamus.
We used a cross-sectional study design to investigate the relationship between modified CES-D score and MRI lesions. Since the MRI scans were obtained over a 3-year period, we chose the CES-D score and other non-MRI variables obtained closest in time to the MRI scan.
Because of the skewed distribution of modified CES-D scores, we divided the range of scores into quartiles for comparisons. We chose this method instead of performing a mathematical transformation of the depression score so that the results would make sense clinically. Data were analyzed in 2 ways: (1) across all 4 quartiles of modified CES-D score and (2) by comparing highest and lowest quartiles of depression score. The latter comparison was made to compare a group that likely did not have clinically relevant depression symptoms with a group likely to have clinically relevant depression, including major depression.
Quartiles of CES-D were tested as the dependent variable utilizing the extension of the logistic model to a limited number of ordinal categories. In such modeling, it is not necessary to assume that the numbering of these categories implies linearity in their relationship to independent predictors, only ordinality. A parallel lines regression model is based on the cumulative distribution probabilities of the predicted dependent categories. A linear fit to an independent predictor implies a linear relation to the logit, or measure of increased risk for classification from one category to the next highest. Multiple intercepts model overall differences in marginal probability of being in one category as opposed to another. In all logistic models, assumptions of normality for individual predictors are relaxed compared with parametric regression, it being necessary only to assume a multinormal distribution for all predictors fitted simultaneously. Here, the ordinal dependent groups were quartiles of CES-D score, using either all 4 quartiles or just the first and last. In using only the first and last quartiles, the logistic model reduces to the common bivariate logistic function.
The data were first summarized with extended logistic analysis to examine individually a variety of MRI and non-MRI variables in each quartile of modified CES-D score. The next set of extended logistic analyses examined the relationship of white matter grade or lesion number for small (<3 mm) and large (≥3 mm) lesions in the basal ganglia, cortical white matter, and subcortical white matter with quartile of modified CES-D score.
An extended logistic model was then developed, including significant non-MRI variables for each type of MRI lesion and white matter grade. Finally, to assess the effect of basal ganglia lesion controlling for white matter lesion severity, an extended logistic model was developed that included significant non-MRI variables along with separate variables for basal ganglia and white matter lesions. In all logistic models, 2 separate analyses were performed: one across all 4 quartiles of modified CES-D score and the other comparing the lowest and highest quartiles.
The sample consisted of 3660 subjects with complete depression data and an MRI scan. Table 1⇓ shows demographic, medical history, and cognitive and functional status data on these subjects by quartile of modified CES-D score. Quartiles 1 through 4 consisted of 907, 799, 1068, and 880 individuals, respectively. Because the depression scale is discrete rather than continuous in its distribution, constructing precisely equal quartiles was not possible. The cut points we used in constructing our approximate quartiles minimized deviation from the ideal in our sample. Applying the logistic procedure to all 4 quartiles and then to the highest and lowest quartiles demonstrated that higher CES-D score was significantly associated with older age, female sex, nonwhite race, presence of hypertension, presence of coronary heart disease, and higher ADL and IADL disability.
Table 2⇓ lists data and analyses related to white matter lesions. White matter grade, a measure of severity of white matter change on a scale of 0 to 9, was significantly associated with CES-D score quartile. Number of small, large, or combined subcortical white matter lesions was not associated with CES-D score quartile. The number of large cerebral cortical white matter infarcts was associated with CES-D score for all 4 quartiles but not when quartiles 1 and 4 were examined. In results not included in the table, there were only 24 individuals with any small white matter cerebral cortical lesions; this variable was not associated with CES-D quartile.
As shown in Table 2⇑, we further investigated the white matter grade by constructing it as a categorical variable. We aimed to compare the most severe white matter grade with all other grades, and therefore we performed a logistic regression by CES-D quartile on white matter grades 0 to 5 versus grades 6 to 9, the latter containing the top 4%. This strategy again demonstrated that severity of white matter grade was significantly associated with CES-D quartile. The odds ratio of the dichotomized white matter grade variable for the lowest versus highest CES-D scores was 1.973.
Table 3⇓ demonstrates results for the basal ganglia lesions. Number of small basal ganglia lesions was highly associated with CES-D quartile, while number of large basal ganglia lesions was also statistically significant, but less so. Combining number of small and large lesions into a dichotomous “any basal ganglia lesion” variable demonstrated a significant association as well, although not in subsequent logistic regression models.
Results of subsequent logistic regression models are found in Tables 4 through 6⇓⇓⇓. We initially constructed a model examining white matter grade along with the significant non-MRI variables from Table 1⇑. From Table 4⇓, we see that white matter grade was not significantly associated with CES-D quartile, while the non-MRI variables (except age and hypertension) were significant predictors of CES-D quartile when all 4 quartiles were examined, but white matter grade was not significantly associated with CES-D quartile in the model examining quartiles 1 and 4.
In Table 5⇑, we show the results of logistic regression models examining small basal ganglia lesions along with non-MRI variables. Number of small basal ganglia lesions remained a strong predictor of CES-D quartile. As with the white matter grade regression model, the non-MRI variables, with the exception of age and hypertension, remained associated with CES-D quartile.
Table 6⇑ demonstrates a final logistic regression model that examined the effects of both white and gray matter lesions along with non-MRI variables in predicting CES-D quartile. We chose the 2 lesion predictors from previous models, namely, number of small basal ganglia lesions and white matter grade, to include in the model with the other non-MRI variables. As can be seen in the model, number of small basal ganglia lesions remained a significant predictor of CES-D quartile, even after accounting for the effects of white matter grade, age, sex, race, presence of hypertension or coronary heart disease, 3MS score, and ADL and IADL scores. The odds ratio of number of basal ganglia lesions was 1.212 for all CES-D quartiles and was 1.401 when CES-D quartiles 1 and 4 were compared.
The major finding of this study is that lesions in the basal ganglia are associated with greater CES-D scores, even after controlling for the effects of age, sex, race, presence of hypertension and cardiovascular heart disease, functional disability, 3MS score, and white matter lesion severity. Separately, number of gray matter lesions predicted CES-D quartile when controlling for the effects of these non-MRI variables, but white matter grade did not. This report extends previous findings from clinical samples to a larger, multisite community sample. Both subcortical gray and white matter lesions have been associated with depression. These studies generally have used clinical samples and employ a case-control format with elderly depressives with nondepressed age-matched control subjects.
The greatest strength of the study is its large community-based sample. Of necessity, a very large study such as the CHS is required to have enough subjects with vascular lesions to test hypotheses about structural brain changes and depression while controlling for several obvious factors that may lead to depressive symptoms. Even so, it might be argued that other factors (eg, other medical conditions, history of depressive episodes, current treatment for depression) ideally should be included. Since the CHS is primarily a study assessing cardiovascular disease in older adults, neither extensive psychiatric history nor interviews were obtained. We chose to focus on 2 medical conditions, hypertension and cardiovascular heart disease, that may increase the risk of cerebrovascular changes. Not having a more detailed psychiatric history is likely one limitation of the present study. Without a psychiatric interview, we also cannot determine whether these psychiatric symptoms are due to depression or some other medical or neurological condition (eg, frontal lobe syndromes). On the other hand, the robustness of the models we developed that did not include these psychiatric variables clearly demonstrates the importance of the MRI lesion variables.
Two other studies investigating the relationship of MRI vascular changes and depressive symptoms have been undertaken and are currently under review (R. Sato, PhD, et al, unpublished data, 1999, and R. McClelland, MS, et al, unpublished data, 1999). Both studies examined infarctlike lesions >3 mm only and found that an initial association between number of lesions and depressive symptoms lost statistical significance in subsequent regression models that included ADL, IADL, and cognitive function as covariates. We also found a loss of significance for large basal ganglia lesions in multivariate models. Our finding that small basal ganglia lesions remained a significant independent predictor of depression symptoms in regression models has not been reported in studies using the CHS database. Small, hyperintense lesions in subcortical white and gray matter have been shown to be increased in depressives compared with age-matched nondepressed controls.3 One potential limitation of this finding for smaller lesions is that, in the CHS, interreader reliability has been shown to be less for smaller lesions than for larger lesions (κ=0.32 versus 0.78).17
One further limitation of this study may be selection bias within the population of subjects in the CHS. Those who agreed to have an MRI may have less depression than those who refused. However, if individuals who refused MRI also had greater lesions on MRI, the result would be a decreased power to detect a difference among individuals by CES-D quartile, a bias toward the null hypothesis. Our results would be biased against the null hypothesis if depressed MRI refusals had fewer lesions on MRI, an unlikely possibility.
Other potential limitations relate to use of the modified CES-D instrument. While the sensitivity and specificity of this version of the CES-D have not been broadly established, other shorter versions of the CES-D have shown good psychometric properties, some with similar sensitivity and specificity to the full CES-D.21 22 23 Like the original CES-D, the modified version used in the CHS is a self-report of one’s depressive symptoms in the past week, which may be influenced by a variety of factors. The instrument is not a diagnostic tool for depression, although cut point scores on the CES-D have been compared with clinical diagnosis of major depression. For example, in a population of geriatric primary care patients, the optimum cutoff point for the CES-D was found to be 21, yielding a sensitivity of 92% and a specificity of 87% with clinical diagnosis used as a gold standard.24 We divided CES-D scores into quartiles in part to cluster those with higher scores who may have diagnosable depression. Clearly, more validation of this modified instrument is warranted.
Our finding of a significant contribution for small basal ganglia lesions in logistic models and lack of significance for larger basal ganglia infarcts is interesting and may provide insight into the link between vascular changes and development of depressive symptoms. The pathophysiological mechanisms associated with development of small versus large lesions may differ, with small lesions linked to damage to long penetrating arteries that supply subcortical areas and larger lesions associated with damage to larger vessels. In patients with a motor deficit with infarcts in the territory of lenticulostriate branches from the middle cerebral artery, 75% had either hypertension or diabetes, but 35% had an embolic source, from large vessels (28%), and/or from the heart (15%).25 The etiology of lesions seen in depressed individuals may also be multifactorial, but smaller lesions associated with depressive symptoms in the present study may point to an association with risk factors for small-vessel disease (eg, hypertension, diabetes). Whether damage to the small vessels leads to hypoperfusion and the relationship of hypoperfusion and arteriosclerosis of the smaller vessels to lesion development remain unclear.26
Our finding that, compared with number of basal ganglia lesions, white matter lesion severity was not associated with CES-D quartile may be due to several factors. White matter lesions may be less important than basal ganglia lesions in causing depressive symptoms. Thus, previous findings of an association of white matter changes with depression1 may have occurred because the white matter lesions were actually a proxy measure for basal ganglia lesions. Only through large studies such as the CHS can the separate effects of basal ganglia and white matter lesions be examined. Alternatively, white matter lesions may be important, but there may be a threshold effect of severity of white matter lesions that is not as pronounced for basal ganglia lesions. We found evidence for this when we dichotomized white matter grade into the most severe (grades 6 to 9) and less severe cases of white matter pathology. The dichotomous white matter grade variable was a much better predictor of CES-D quartile than the “continuous” measure (grades 0 to 9), indicating that there is a poor linear relationship between depressive symptoms across grades 0 to 9 and supporting a threshold of severity of white matter pathology. One final explanation is that, despite our large sample, we were still underpowered to detect a difference for white matter lesions among the CES-D quartile groups.
Presence of hypertension was significantly associated with CES-D quartile, but hypertension was no longer a significant predictor in logistic models. In analyses not shown, we found that hypertension was correlated with every other variable in the model, and therefore it is not surprising that it lost significance. This finding is important for the debate about “vascular depression.” Lyness et al27 have shown that risk factors for cerebrovascular disease do not themselves predict depressive symptoms or severity of geriatric depression. For vascular depression to become recognized as a subtype of major depression, presence of actual brain damage, documented through neuroimaging or neuropsychological testing, may be warranted.3 Once vascular depression is identified in this manner, further studies of its underlying pathophysiology as well as refinement of a clinical subtype can be undertaken.
Other variables associated with depression scores included female sex, nonwhite race, low functional status, and cognitive impairment. While other studies consistently show that depression is more common in women, ours is the first population study replicating the finding in vascular depression. There has been little research examining race in this largely older age group, and there are clear barriers to participation of older minorities in geriatric depression studies.28 Our finding of a significant association between nonwhite race and depression score adds to that literature. While the finding associating low functional status and cognitive impairment with higher depression scores is not surprising, since this is a cross-sectional study, it is important to note that impairment in function or cognition either may be a consequence of depression or may lead to depression.
In this large population study, the finding of an association between small basal ganglia lesions and depressive symptoms has implications for the treatment and prevention of depression in older individuals. Presence of depressive symptoms in patients with risk factors for cerebrovascular disease may be a sign of “silent cerebral infarction.”29 For these individuals, aggressive treatment of these risk factors may play an important role in the overall management of depressive symptoms. Both primary care and specialty physicians should therefore screen for depression in these patients, and they should be made aware that their efforts to monitor and treat patients with vascular risk factors may play a critical role in the primary prevention of depression in their patients. Future epidemiological studies and clinical drug trials need to address the role of treatment of vascular risk factors in both the prevention and treatment of depression.
Participating Institutions and Principal Staff of the Cardiovascular Health Study
Forsyth County, North Carolina—Bowman Gray School of Medicine of Wake Forest University: Gregory L. Burke, Sharon Jackson, Alan Elster, Walter H. Ettinger, Curt D. Furberg, Gerardo Heiss, Dalane Kitzman, Margie Lamb, David S. Lefkowitz, Mary F. Lyles, Cathy Nunn, Ward Riley, John Chen, Beverly Tucker; Forsyth County, North Carolina—Bowman Gray School of Medicine, ECG Reading Center: Farida Rautaharju, Pentti Rautaharju; Sacramento County, California—University of California, Davis: William Bommer, Charles Bernick, Andrew Duxbury, Mary Haan, Calvin Hirsch, Lawrence Laslett, Marshall Lee, John Robbins, Richard White; Washington County, Maryland—The Johns Hopkins University: M. Jan Busby-Whitehead, Joyce Chabot, George W. Comstock, Adrian Dobs, Linda P. Fried, Joel G. Hill, Steven J. Kittner, Shiriki Kumanyika, David Levine, Joao A. Lima, Neil R. Powe, Thomas R. Price, Jeff Williamson, Moyses Szklo, Melvyn Tockman; MRI Reading Center, Washington County, Maryland—The Johns Hopkins University: R. Nick Bryan, Norman Beauchamp, Carolyn C. Meltzer, Naiyer Iman, Douglas Fellows, Melanie Hawkins, Patrice Holtz, Michael Kraut, Grace Lee, Larry Schertz, Cynthia Quinn, Earl P. Steinberg, Scott Wells, Linda Wilkins, Nancy C. Yue; Allegheny County, Pennsylvania—University of Pittsburgh: Diane G. Ives, Charles A. Jungreis, Laurie Knepper, Lewis H. Kuller, Elaine Meilahn, Peg Meyer, Roberta Moyer, Anne Newman, Richard Schulz, Vivienne E. Smith, Sidney K. Wolfson; Echocardiography Reading Center (Baseline)—University of California, Irvine: Hoda Anton-Culver, Julius M. Gardin, Margaret Knoll, Tom Kurosaki, Nathan Wong; Echocardiography Reading Center (Follow-Up)—Georgetown Medical Center: John Gottdiener, Eva Hausner, Stephen Kraus, Judy Gay, Sue Livengood, Mary Ann Yohe, Retha Webb; Ultrasound Reading Center—Geisinger Medical Center: Daniel H. O’Leary, Joseph F. Polak, Laurie Funk; Central Blood Analysis Laboratory—University of Vermont: Edwin Bovill, Elaine Cornell, Mary Cushman, Russell P. Tracy; Respiratory Sciences—University of Arizona, Tucson: Paul Enright; Coordinating Center—University of Washington, Seattle: Alice Arnold, Annette L. Fitzpatrick, Bonnie K. Lind, Richard A. Kronmal, Bruce M. Psaty, David S. Siscovick, Lynn Shemanski, Will Longstreth, Patricia W. Wahl, David Yanez, Paula Diehr, Maryann McBurnie, Chuck Spiekerman, Scott Emerson, Cathy Tangen, Priscilla Velentgas; National Heart, Lung, and Blood Institute Project Office: Robin Boineau, Teri A. Manolio, Peter J. Savage, Patricia Smith.
This study was supported by contracts N01-HC-85079 and N01-HC-85086 from the National Heart, Lung, and Blood Institute.
- Received June 7, 1999.
- Revision received July 15, 1999.
- Accepted July 15, 1999.
- Copyright © 1999 by American Heart Association
Radloff LS. The CES-D scale: a self-report depression scale for research in the general population. Appl Psychol Meas. 1977;1:385–401.
National Center for Health Statistics, Fitti JE, Kovar MG. The Supplement on Aging to the 1984 National Health Interview Survey. Washington, DC: Public Health Service, US Government Printing Office; 1987. Vital and Health Statistics, Series 1, No. 21. Publication DHHS 87–1323.
Katz S, Fort AB, Moskowitz RW. Studies of illness in the aged: the index of ADL, a standardized measure of biological and psychosocial function. JAMA. 1963;185:914–919.
Teng EL, Chui HC. The Modified Mini-Mental State (3MS) Examination. Clin Psychiatry.. 1987;48:314–318.
Manolio TA, Kronmal RA, Burke GL, Poirier V, O’Leary DH, Gardin JM, Fried LP, Steinberg EP, Bryan RN. Magnetic resonance abnormalities and cardiovascular disease in older adults: the Cardiovascular Health Study. Stroke. 1994;25:318–327.
Bryan RN, Manolio TA, Schertz LD, Jungreis C, Poirier VC, Elster AD, Kronmal RA. A method for using MR to evaluate the effects of cardiovascular disease on the brain: the cardiovascular health study. AJNR Am J Neuroradiol. 1994;15:1625–1633.
Price TR, Manolio TA, Kronmal RA, Kittner SJ, Yue NC, Robbins J, Anton-Culver H, O’Leary DH, for the CHS Collaborative Research Group. Silent brain infarction on magnetic resonance imaging and neurological abnormalities in community-dwelling older adults: the Cardiovascular Health Study. Stroke. 1997;28:1158–1164.
Hatazawa J, Shimosegawa E, Satoh T, Toyoshima H, Okudera T. Subcortical hypoperfusion associated with asymptomatic white matter lesions on magnetic resonance imaging. Stroke. 1997;28:1944–1947.
Fujikawa T, Yamawaki S, Touhouda Y. Incidence of silent cerebral infarction in patients with major depression. Stroke. 1993;24:1631–1634.