Silent Brain Infarction on Magnetic Resonance Imaging and Neurological Abnormalities in Community-Dwelling Older Adults
The Cardiovascular Health Study
Background and Purpose Infarctlike lesions are frequently detected in symptomatic and asymptomatic older persons undergoing cerebral MRI, but their significance in older adults has not been examined. We determined the prevalence of MRI infarcts in a population-based sample of men and women aged ≥65 years and related these findings to demographic, cognitive, and neurological status.
Methods MRI scanning was performed in 3660 Cardiovascular Health Study (CHS) participants after brief neurological examinations and tests of cognitive function. MRIs were read centrally for the presence of an infarct ≥3 mm in diameter or smaller infarctlike lesions.
Results MRI infarcts were detected in 1131 of 3647 participants with readable infarct information (31%) and in 961 of the subgroup of 3397 participants (28%) without known prior stroke (“silent” MRI infarcts). Smaller infarctlike lesions were found in 196 of 2516 participants who had no MRI infarcts ≥3 mm. MRI infarcts were more common in participants who were older, had prior stroke, impaired cognition, visual field deficits, slowed repetitive finger tapping (all P<.0001), weakness on toe and heel walking, and history of memory loss, coma, or migraine headaches. Multivariate analysis in those without prior stroke showed strong associations of silent MRI infarcts with older age, history of migraines, lower digit symbol scores, and more abnormalities on neurological examination.
Conclusions MRI evidence of brain infarction is common in older men and women without a clinical history of stroke. Their strong associations with impaired cognition and neurological deficits suggest that they are neither silent nor innocuous.
With the advent of highly sensitive techniques for brain imaging such as MRI, a wide range of potentially abnormal findings has been reported and noted to increase in frequency with age.1 Such findings include lesions with an appearance typical of infarction; when these appear in persons without a history compatible with stroke, they are often referred to as “silent infarcts.” The strong relationship of these findings with age and other stroke risk factors suggests that they may themselves be risk factors or even manifestations of clinically important cerebrovascular disease.
While numerous reports have documented the frequency of presumed infarctions on CT or MRI in persons with a history of (or at risk for) overt stroke,2 3 4 5 6 few studies have reported prevalence of such lesions in an asymptomatic community-dwelling population.7 In addition, their relationship to abnormalities on neurological examination or cognitive testing has not been defined. Earlier studies often relied on CT, which is less sensitive than MRI in demonstrating small infarcts in the basal ganglia or posterior fossa.8 9 Prior studies also tended to include relatively small numbers of patients10 11 12 and to be limited to one condition, such as hypertension or atrial fibrillation.13 14 15 16 Few have included large samples of clinically normal persons.
The Cardiovascular Health Study (CHS) is a population-based observational study of 5888 men and women aged ≥65 years from four US communities. Cranial MRI was performed 3 years after entry to (1) assess the prevalence of infarcts in those with and without a clinical history of stroke and (2) examine the associations of these lesions with measures of cognition and neurological damage in those without known stroke.
Subjects and Methods
Participants in the CHS were recruited from four counties in the United States (Allegheny County, Pennsylvania; Forsyth County, North Carolina; Sacramento County, California; and Washington County, Maryland). Recruitment was from a random sample of Health Care Financing Administration Medicare eligibility lists from these four counties. Potential participants were excluded if they were institutionalized, currently under treatment for cancer, wheelchair dependent in the house, or expected to move from the area in the next 3 years. The study was approved by an institutional review committee, and subjects gave informed consent. Study design and methods have been published.17 The baseline examinations were conducted between June 1989 and May 1990. Shortly after the study began, cranial MRI was added to improve ascertainment and classification of stroke and other brain abnormalities.18
At entry into the study, participants reported any prior physician-diagnosed vascular events including stroke. Participants with a positive history of stroke and those who had a stroke during the study, but before the MRI was done, were defined for this analysis as having “prior stroke.” Participant self-report of stroke at entry was accepted whether or not medical record or physician confirmation was available. Stroke during the study was ascertained by questions at the annual visit and interim telephone contacts and review of hospitalizations. When notification of a possible stroke was received, medical and interview information was obtained and the occurrence of a stroke determined by the Cerebrovascular Disease Adjudication Committee.19 Strokes detected and confirmed after entry were referred to as “clinically recognized stroke” since adjudication protocols required supporting evidence of stroke such as signs, symptoms, or abnormalities on imaging.
Tests of cognitive and fine motor function included the digit symbol substitution test, the Modified Mini-Mental State Examination,20 and the finger-tapping test.21 The finger-tapping test consists of tapping a computer “mouse” as fast as possible for 15 seconds with each hand and recording the total taps separately for the left and right hands. These tests were repeated annually during the CHS follow-up. For the purposes of this analysis, the data reported were from the most recent test preceding the MRI scan.
Cranial MRI scanning was performed between June 1992 and June 1994 according to a standard protocol.22 The MRI examinations included spin-density, T2- and T1-weighted spin-echo images with 5-mm thickness and zero gap and oriented parallel to the anterior-posterior commissure line. Images were read centrally by neuroradiologists. The design and methodology of the MRI study have been described.22
Infarcts by MRI were defined as lesions with abnormal signal in a vascular distribution and no mass effect. Infarcts of the cortical gray matter and deep nuclear regions and capsule were defined as lesions bright on spin-density and T2-weighted images compared with normal gray matter and isodense or hypodense on T1-weighted images. Infarcts in the white matter were also bright on spin-density and T2-weighted images but in addition were hypointense on T1-weighted images, approximating the intensity of cerebrospinal fluid. The requirement for hyperintensity on spin-density images was intended to distinguish small deep nuclear region infarcts (those in the caudate nucleus, lentiform nucleus, internal capsule, external capsule, extreme capsule, and thalamus) from dilated perivascular spaces. Stroke location was coded as cortical, subcortical, posterior fossa, or some combination of these. In this report, presumed ischemic lesions in any region were classified as “small infarctlike lesions” when <3 mm and as “MRI infarcts” when ≥3 mm. All MRI interpretations were made in the absence of any clinical information, including age.
For clarity of presentation, the term “stroke” is applied here only to self-report of physician diagnosis of stroke at entry or to adjudicated, clinically recognized stroke. MRI infarcts (as defined above) in the absence of either a self-reported physician diagnosis at entry or clinically recognized stroke during the course of the study are referred to as “silent infarcts” to distinguish them from symptomatic and clinically recognized stroke syndromes.
Brief Neurological Examination
At the time of the MRI examination, a short history and neurological examination were done by technicians trained by one of the authors (S.J.K.). Participants were queried regarding prior coma, cancer, brain tumor, brain operation, seizure or convulsion, loss of memory other than for people’s names (“Do you have loss of memory other than for people’s names?”), migraine headaches (“Have you ever had migraine headaches?”), injuries causing loss of consciousness (“Have you ever had an injury that resulted in loss of consciousness [knocked out]?”), and prior physician diagnosis of cerebral palsy or any other neurological illness.
For the neurological examination, participants who could walk 15 feet were asked to do so and observed for evidence of a hemiparetic gait (decreased arm swing and leg stiffness on the same side). They were then asked to walk on the balls of their feet and on their heels (they could be assisted for balance) and observed for weakness or inability to make at least four such steps with each foot. Each participant was scored for weakness on each side (inability to maintain the heel off the floor when walking on balls of the feet or to keep the balls of the feet off the floor when walking on the heels). Balance was tested by having them stand with the feet together and eyes closed and maintain their balance without stepping out for 30 seconds; taking one or more steps constituted inability to maintain balance. Visual fields were tested by confrontation with the use of finger movements in all four quadrants. Repeated (twice) failure to correctly notice finger movements in a field was scored as a left or right visual field deficit. Drift was tested by having the participants hold their arms out with palms up and eyes closed for 10 seconds while being observed for drift downward on one side.
A summary of the neurological examination findings was made by summing the number of abnormalities on both sides, which were then categorized as follows: no abnormalities; 1 or 2 abnormalities; 3 or 4 abnormalities; and 5 or 6 abnormalities (the maximum observed). The choice of categorical intervals was based on the similarity of the prevalence of MRI infarctions in adjacent intervals; for example, prevalence of MRI infarctions in participants with 1 or 2 neurological examination abnormalities was similar.
Significance of differences between participants with and without MRI infarcts or infarctlike lesions was assessed by χ2 tests for proportions and t tests for continuous variables. Multiple logistic regression analysis was used to assess independent correlates of MRI infarct or infarctlike lesions among measures of cognitive function and abnormalities detected on neurological examination. Each of the neurological questionnaire and examination variables and cognitive function scores, as well as age and sex, was allowed to compete for entry in initial stepwise models (P=.05 to enter, P=.10 to drop). We constructed a second model using total number of abnormalities on neurological examination rather than each abnormality (eg, left drift, right heel weakness) individually. Since this second model provided a slightly better fit than models allowing each examination abnormality to enter in stepwise fashion, it was used as the final multivariate model.
Of the 5477 surviving CHS participants, 3660 underwent MRI scanning; information on infarcts was obtained in 3647. Participants who were scanned were significantly younger, had more education and higher incomes, and were more likely never to have smoked and less likely to have had prior cardiovascular disease, hypertension, or diabetes than those who did not undergo scanning.23 One or more MRI infarcts (lesions ≥3 mm) were detected in 1131 of these 3647 participants (31%). MRI infarcts were detected in 170 (68%) of the 250 participants with prior, clinically recognized stroke and 961 (28%) of the 3397 without prior stroke (P<.0001). Diameter of the largest infarct was ≥15 mm in 258 participants (7.1%), 3 to 15 mm in 873 (23.9%), and <3 mm in 196 (5.4%). Approximately one sixth of the 961 participants with MRI infarct but without clinical stroke had infarcts ≥15 mm in diameter. MRI infarcts were slightly more common in men than women, but this difference was not significant (Fig 1⇓). For infarcts with a single location, 8% were cortical, 72% were subcortical, and 7% were in the posterior fossa.
MRI infarcts were associated with increased age in those without prior stroke (76.0 versus 74.6 years; P<.0001) but not significantly so in those with prior stroke (Table 1⇓). Prevalence of MRI infarcts did not differ between black (n=563) and white participants. Those with prior stroke but without MRI infarcts scored lower on tests of cognitive and fine motor function and had more abnormalities on neurological examination than those without prior stroke but with MRI infarct.
Associations between MRI infarcts and neurological findings in those with and without prior stroke were similar but were not significant in those with prior stroke, presumably because of the small sample size (n=250). For this reason, the discussion of associations immediately below will be limited to the group without prior stroke. MRI infarcts were strongly associated with lower digit symbol and Modified Mini-Mental State scores and fewer finger taps (Table 1⇑). MRI infarcts were also associated with visual field deficits, weakness on walking on toes or heels, and history of coma or memory loss other than for people’s names (Table 2⇓). Participants without prior stroke but with visual field loss were more than twice as likely to have MRI infarcts as those without field loss, while those with history of coma were 69% more likely to have MRI infarcts than those without such a history. The total number of neurological examination abnormalities was also strongly related to presence of infarcts by MRI (Fig 2⇓).
One or more small infarctlike lesions (<3 mm) were detected in 196 of the 2516 participants (7.8%) who had no MRI infarcts. The same items from cognitive function testing, neurological questionnaire, and examination were tested for their relation to small infarctlike lesions in these participants (data not shown). Only a history of loss of memory other than for names was associated with the likelihood of finding such small lesions; 27 of 219 (12.3%) with memory loss and no large lesions had these small lesions, and 164 of 2221 (7.4%) without memory loss had small lesions (P=.009).
In multivariate analysis of all participants undergoing MRI, age, prior stroke, history of migraine headaches, digit symbol score, and number of neurological abnormalities were independently associated with presence of MRI infarct (Table 3⇓). A 7-year increment in age (the interquartile range) was associated with a 32% increased prevalence of MRI infarct, and prior stroke was associated with a more than fourfold increase after multivariate adjustment. Exclusion of participants with a prior stroke had little effect on these risk estimates. Digit symbol score retained a strong association, which was significant at P<.001. Additional adjustment for other factors associated with MRI infarcts in multivariate analysis (systolic and diastolic blood pressures, current smoking, and diabetes) did not materially affect these relationships, and other cardiovascular disease risk factors (total cholesterol levels, atrial fibrillation, congestive heart failure, left ventricular hypertrophy by electrocardiogram, and antihypertensive medication use) were not significantly associated with MRI infarcts on multivariate analysis, and we did not adjust for them further. Factors associated with small infarctlike lesions on multivariate analysis in those without MRI infarct included age, prior stroke, and histories of brain operation and memory loss other than for names (data not shown).
MRI-defined infarcts ≥3 mm were detected in 31% of this population-based sample of older adults and in 28% of those without prior reported or clinically recognized stroke. Comparison of MRI findings with autopsy specimens have established the high sensitivity and reliability of MRI in detecting lacunar infarctions.24 25 MRI protocol requirements for gray matter lesions to be bright on spin-density images relative to gray matter and for white matter lesions to be hypointense on T1 images (approximating the intensity of cerebrospinal fluid) were intended to distinguish them from nonnecrotic Virchow-Robin spaces and from white matter disease, respectively.22 If these abnormalities do indeed represent small areas of cerebral infarction, as pathological evidence would suggest, then their high prevalence in clinically “stroke-free” older subjects is of concern.
That these MRI infarcts were not without consequence is suggested by their strong associations with neurological abnormalities and impaired cognitive function, associations that persisted after adjustment for age and prior stroke. Even after exclusion of those with known prior stroke, persons with MRI infarcts had on average a four-point lower digit symbol score and a nearly two-point lower Modified Mini-Mental State score. Conversely, persons without prior stroke but falling in the lowest quintile of digit symbol score were twice as likely to have MRI infarcts, after adjustment for other factors, as those in the highest quintile. Persons with MRI infarcts were also 60% to 100% more likely to have visual field deficits or demonstrable weakness on walking. Such abnormalities strongly suggest that these silent infarcts are not clinically benign.
Comparison With Prior Literature
Previous MRI studies of clinically “normal” subjects have demonstrated prevalences of silent infarcts as low as 11% in 44 neurologically normal caregivers and relatives of Alzheimer’s disease patients (mean age, 68 years) examined by psychiatrists and neurologists.10 A Japanese study in which 246 people “registered for health screening of the brain” with “no history of cerebral disease and who were socially active and neurologically normal”11 demonstrated silent MRI infarcts in 6% of subjects in their forties to 21% of subjects in their sixties. A Swedish study used a population-based random sample of 77 people with a mean age of 65 years12 with no history of brain lesions; all subjects agreed to undergo a transesophageal echocardiogram as well as MRI. Prevalence of MRI infarction in this group was 12%. Participants in the present study had a mean age of 75 years and might be expected to have more infarcts based on their age alone, as discussed below. By selecting a demonstrably “normal” group of subjects or those willing to undergo transesophageal echocardiography, prior studies may have examined highly selected and nongeneralizable population samples.
Many prior studies have focused on the rate of silent infarcts in various disease states. Acute ischemic stroke patients have been found to have unrelated, silent infarcts in 11% to 29%.2 3 4 5 6 Hypertension,13 14 atrial fibrillation,15 16 severity of coronary artery disease,26 and idiopathic dilated cardiomyopathy27 have all been associated with silent infarcts. Several studies have documented silent infarcts in 16% to 21% of patients undergoing carotid endarterectomy for symptomatic carotid disease.28 29 The Asymptomatic Carotid Atherosclerosis Study (ACAS) demonstrated a 15% prevalence in asymptomatic patients with carotid artery atherosclerosis.30 These studies of carotid artery disease and stroke all used CT scans to define unrecognized infarction, which are generally less sensitive than MRI in detecting small infarcts.14 15 Associations of age with silent infarct were strong and consistent in many of these largely clinical series,4 5 6 13 15 16 26 29 similar to the results of the present study.
The finding of a weak association of MRI infarcts with history of migraine headaches is intriguing but of questionable significance, particularly since migraine history is based on self-report rather than physician diagnosis. One would generally expect misclassification of migraine history to lead to an underestimation of an association, if one indeed exists. Previous studies have demonstrated increased prevalence of white matter lesions in patients with migraine31 32 and have reported cases of infarctlike lesions on MRI in migraine patients,33 but we have been unable to find other population-based studies to date that confirm these findings.
Less information is available on the relationship of abnormal neurological findings to MRI infarcts in participants with no history of stroke. The ACAS investigators demonstrated higher prevalence of abnormalities of gait, reflexes, hearing, eye movements, funduscopic examination, and visual fields in subjects with silent infarct detected by CT, but these associations were not significant.30 In the present study, visual fields and gait weakness were similarly associated with silent infarct and remained significant after adjustment for other factors. The brief neurological examination conducted as part of the CHS MRI protocol did not include the other ACAS items.
Silent Infarcts Versus Clinically Recognized Stroke
Silent infarcts differ from clinical strokes in that although there is damage to the brain, it is usually so strategically placed or so small that it does not cause symptoms or signs leading to a diagnosis of stroke.30 Sometimes evidence of unreported damage can be found on neurological examination. In these cases, the associated infarct is more likely larger, in the nondominant hemisphere, and more superficial.4 Demonstrable gait weakness or visual field deficits, which are strongly associated with these lesions, can be major impairments and may contribute to falls and other accidents in the elderly.
The strong associations of clinically recognized stroke with impaired cognitive function and neurological abnormalities in this study provide further support for the contention that recognized strokes are often more severe than asymptomatic infarcts. For each functional score or examination component, persons with prevalent stroke had poorer scores and higher frequency of abnormal findings than those without prevalent stroke, regardless of presence of MRI infarcts. Clinically recognized strokes thus appear to be more strongly related to abnormal functioning in a variety of domains than do MRI infarcts.
Demonstrable Abnormalities in Unrecognized Infarction
Despite the strong association of neurological examination abnormalities and silent infarcts in the present study, most participants without prior stroke and with these lesions (792 of 948, or 84%) had no demonstrable abnormalities on examination. In previous studies the frequency of normal examinations in subjects with these lesions but without clinically recognized prior stroke was 75%4 and 59%.30
The two tests of mental function and a history of memory problems were each associated with MRI infarcts in bivariate analysis, and the digit symbol test was independently associated with these lesions in multivariate analysis. Although causal relationships cannot be established from cross-sectional data, this finding raises the intriguing possibility that investigation of MRI infarcts may lead to greater insight into the etiology of cognitive decline. If impaired cognition and MRI infarcts do not share a direct causal link, it is possible that they are associated with a third factor, such as the demonstrated decrease in regional cerebral blood flow seen in patients with silent infarction, that is causing them both.12 In the study of idiopathic dilated cardiomyopathy mentioned above, significantly worse cognitive performance was found in the patients with myopathy than in the normal control subjects, and cognition was most impaired in patients with cerebral abnormalities on MRI scan. Sulcal widening and severity of white matter disease have been associated with poorer performance on the Mini-Mental State Examination and the digit symbol test in CHS participants.18 23
The lack of associations of small infarctlike lesions <3 mm with cognitive and fine motor function and examination abnormalities may be due to the small number of subjects with these lesions only. The significance of such lesions at present remains unknown. Still, the finding of a strong and independent association of these lesions with a history of memory problems suggests interesting avenues for further investigation of reported memory loss in the elderly.
Limitations of the Present Study
The screening neurological examinations in CHS were conducted by trained technicians and are of unknown accuracy. Prior studies have generally included more comprehensive examinations conducted by neurologists, but this was not feasible in CHS. Assuming that these limitations would lead to decreased sensitivity and increased variability in assessment of neurological findings, the present study likely underestimates both the prevalence of abnormalities and the strength of their associations with unrecognized infarction. Similarly, the somewhat selected nature of the CHS sample and of the subsample undergoing MRI would tend to underestimate the prevalence of MRI abnormalities, as has been reported previously.23 Such biases would not, however, be expected to affect associations of risk factors or examination findings with MRI infarcts.34
Strengths of the Present Study
CHS participants represent a large, population-based sample of community-dwelling older adults from four diverse geographic areas. CHS results are thus more generalizable than previous work, which nearly always used subjects from clinical samples in a single geographic area.
Another strength of this study is that all MRI studies and pre-MRI examinations were performed according to a standard protocol at each field center. All studies were then read in one location by radiologists with subspecialty training at the CHS MRI Reading Center and without access to any clinical information. In the pilot study, intrareader and interreader reliability for ≥3-mm infarcts was high (κ=0.71 and κ=0.78, respectively) but not so for <3-mm infarctlike lesions (intrareader κ=0.71; interreader κ=0.32). For this reason, the protocol was modified after the pilot study to include duplicate reading of infarcts and small infarctlike lesions.
The large number of CHS participants allows the correlation of lesion characteristics such as size with evidence of neurological dysfunction and cognitive function. In addition, other extensive data are available on the CHS participants, including questionnaires, physical examinations, laboratory studies, and noninvasive tests. These data will be valuable in comparing risk factors for silent infarction with those for clinical stroke. The prognostic potential of these MRI findings can be studied with the use of the follow-up data that are being collected on clinical events.
Silent brain infarctions found on MRI scanning were common (28%) in these elderly men and women. They were more common in older persons and in those with poorer performance on mental tests and those with neurological deficits. The strong associations of silent infarcts with impaired cognition and neurological deficits suggest that these lesions are neither truly silent nor innocuous. The high prevalence of silent infarcts in older adults suggests that population-based MRI scanning provides an excellent tool for investigating cerebrovascular disease risk factors in asymptomatic community-dwelling older adults.
This study was supported by contracts NO1-HC-85079, NO1-HC-85080, NO1-HC-85081, NO1-HC-85082, NO1-HC-85083, NO1-HC-85084, NO1-HC-85085, NO1-HC-85086, and NO1-95100 from the National Heart, Lung, and Blood Institute.
- Received March 18, 1997.
- Accepted April 1, 1997.
- Copyright © 1997 by American Heart Association
Brant-Zawadzki M, Fein G, Van Dyke C, Keirman R, Davenport L, De Groot J. MR imaging of the aging brain. AJNR Am J Neuroradiol. 1985;6:675-682.
Kase CS, Wolf PA, Chodosh EH, Zacher HB, Kelly-Hayes M, Kannel WB, D’Agostino RB, Scampini L. Prevalence of silent stroke in patients presenting with initial stroke: the Framingham Study. Stroke. 1989;20:850-852.
Davis PH, Clarke WF, Bendixen BH, Culebras A, Adams HP, Woolson RF, and the TOAST Investigators. Silent stroke in patients enrolled in the TOAST Study. Neurology. 1994;44(suppl 2):A225. Abstract.
Chodosh EH, Foulkes MA, Kase CS, Wolf PA, Mohr JP, Hier DB, Price TR, Furtado JG. Silent stroke in the NINCDS Stroke Data Bank. Neurology. 1988;38:1674-1679.
Jorgensen HS, Nakayama H, Raaschou HO, Gam J, Olsen TS. Silent infarction in acute stroke patients: prevalence, localization, risk factors and clinical significance: the Copenhagen Stroke Study. Stroke. 1994;25:97-104.
Herderschee D, Hidra A, Algna A, Koudstaal PJ, Kappelle LJ, van Gijn J, for the Dutch TIA Trial Study Group. Silent stroke in patients with transient ischemic attack or minor ischemic stroke. Stroke. 1992;23:1220-1226.
Mohr JP, Biller J, Hilal SK, Yuh WTC, Tatemichi TK, Hedges S, Tali E, Nguyen H, Mun I, Adams HP Jr, Grimsman K, Marler JR. Magnetic resonance versus computed tomographic imaging in acute stroke. Stroke. 1995;26:807-812.
Kobayashi S, Okada K, Yamashita K. Incidence of silent lacunar lesion in normal adults and its relation to cerebral blood flow and risk factors. Stroke. 1991;11:1379-1383.
Lindgren A, Roijer A, Rudling O, Norrving B, Larsson EM, Eskilsson J, Wallin L, Olsson B, Johansson BB. Cerebral lesions on magnetic resonance imaging, heart disease, and vascular risk factors in subjects without stroke: a population-based study. Stroke. 1994;25:929-934.
Shimada K, Kawamoto A, Matsubayashi K, Ozawa T. Silent cerebrovascular disease in the elderly: correlation with ambulatory pressure. Hypertension. 1990;16:692-699.
Ikeda T, Gomi T, Kobayashi S, Tsuchiya H. Role of hypertension in asymptomatic cerebral lacunae in the elderly. Hypertension. 1994;23(suppl I):I-259-I-262.
Ezekowitz MD, James KE, Nazarian SM, Davenport J, Broderick JP, Gupta SR, Thadani V, Meyer ML, Bridgers SL. Silent cerebral infarction in patients with nonrheumatic atrial fibrillation: the Veterans Affairs’ Stroke Prevention in Nonrheumatic Atrial Fibrillation Investigators. Circulation. 1995;92:2178-2182.
Manolio TA, Kronmal RA, Burke GL, Poirier V, O’Leary DH, Gardin JM, Fried LP, Steinberg EP, Bryan N, for the Cardiovascular Health Study Collaborative Research Group. 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.
Longstreth WT Jr, Manolio TA, Arnold A, Burke GL, Bryan N, Jungreis CA, Enright PL, O’Leary D, Fried L, for the Cardiovascular Health Study Collaborative Research Group. Clinical correlates of white matter findings on cranial magnetic resonance imaging of 3301 elderly people: the Cardiovascular Health Study. Stroke. 1996;27:1274-1282.
Braffman BH, Zimmerman RA, Trojanowski JQ, Gonatas NK, Hickey WF, Schlaepfer WW, Brain MR. Pathologic correlation with gross and histopathology, I: lacunar infarction and Virchow-Robin spaces. AJNR Am J Neuroradiol. 1988;9:621-628.
DeWitt LD, Kistler JP, Miller DC, Richardsson EP, Buonanno FS. NMR-neuropathologic correlation in stroke. Stroke. 1987;18:342-351.
Schmidt R, Fazekas F, Offenbacher H, Dusleag J, Lechner H. Brain magnetic resonance imaging and neuropsychologic evaluation of patients with idiopathic dilated cardiomyopathy. Stroke. 1991;22:195-199.
Hougaku H, Matsumoto M, Handa N, Maeda H, Itoh T, Tsukamoto Y, Kamada T. Asymptomatic carotid lesions and silent cerebral infarction. Stroke. 1994;25:566-570.
Brott T, Tomsick T, Feinberg W, Johnson C, Biller J, Broderick J, Kelly M, Frey J, Schwartz S, Blum C, Nelson JJ, Chambless L, Toole J, for the Asymptomatic Carotid Atherosclerosis Study Investigators. Baseline silent cerebral infarction in the Asymptomatic Carotid Atherosclerosis Study. Stroke. 1994;25:1122-1129.
Igarashi H, Sakai F, Kan S, Okada J, Tazaki Y. Magnetic resonance imaging of the brain in patients with migraine. Cephalalgia. 1991;11:69-74.
Schlesselman JJ. Case-Control Studies: Design, Conduct, Analysis. New York, NY: Oxford University Press; 1982:128-129.