Headache, cerebrovascular
`symptoms, and stroke
`The Atherosclerosis Risk in Communities Study
`
`P.E. Stang, PhD; A.P. Carson, MSPH; K.M. Rose, PhD; J. Mo, MD, PhD; S.A. Ephross, PhD;
`E. Shahar, MD, MPH; and M. Szklo, MD, DrPH
`
`Abstract—Objective: To evaluate the occurrence of stroke/TIA symptoms and ischemic stroke events among those with a
`lifetime history of migraine or other headaches with some migraine features in a biracial cohort of older adults. Methods:
`Participants were 12,750 African-American and white men and women from the Atherosclerosis Risk in Communities
`Study (1993 to 1995). The participants were queried about their lifetime headache history and characterized using
`modified International Headache Society diagnostic criteria. Stroke/TIA symptoms were classified using a computerized
`diagnostic algorithm, and ischemic stroke events were identified and validated using medical records. Multivariate logistic
`regression was used to assess the relationship between headache types and stroke/TIA symptoms and ischemic stroke
`events. Results: Migraine with aura was strongly associated with stroke symptoms (odds ratio [OR] 5.46, 95% CI: 3.64 to
`8.18), TIA symptoms (OR 4.28, 95% CI: 3.02 to 6.08), and verified ischemic stroke events (OR 2.81, 95% CI: 1.60 to 4.92).
`Similarly, other headaches with aura were significantly associated with stroke symptoms (OR 3.68, 95% CI: 2.26 to 5.99)
`and TIA symptoms (OR 4.53, 95% CI: 3.08 to 6.67). In contrast, the associations for migraine without aura and other
`headaches without aura were not as consistent or robust. Conclusions: Migraines and other headaches, particularly those
`accompanied by aura, were associated with an increased occurrence of stroke/TIA symptoms and ischemic stroke events.
`NEUROLOGY 2005;64:1573–1577
`
`Migraine and stroke share many features, including
`a vasospastic component, regional decreases in cere-
`bral blood flow, platelet aggregation, and focal neu-
`rologic and ophthalmologic signs and symptoms.1-3
`Persons with a migraine history may experience a
`stroke not occurring during a migraine attack or a
`“migrainous” stroke. Although initial symptoms of
`both types of stroke may be similar and the clinical
`diagnosis only apparent over time, most strokes ex-
`perienced by migraineurs are not related to a partic-
`ular migraine attack because the incidence of
`migrainous stroke is very low.4-6
`Studies of younger women suggest an association
`between migraine and stroke,7-12 with some reporting
`a higher risk when migraines are accompanied by
`aura.10,11 However, in older populations, findings are
`not consistent. Case series and case-control studies
`reported no association between migraine and
`
`stroke13,14 except in the absence of other risk fac-
`tors.15 In contrast, cohort studies reported an in-
`creased risk of stroke among migraineurs.16-18
`As migraine prevalence decreases with age, while
`stroke risk increases,
`it is important to know
`whether migraine confers risk in older populations
`as it may be a marker of more substantive systemic
`disease. Thus, this study evaluated the occurrence of
`stroke/TIA symptoms and ischemic stroke in those
`with a lifetime history of migraine or other head-
`aches, by aura status, in a biracial cohort of older
`adults and assessed whether that relationship was
`affected by cardiovascular disease (CVD) risk factors.
`
`Methods. Study population. The study population consisted of
`participants from the Atherosclerosis Risk in Communities (ARIC)
`Study, a prospective study designed to investigate the etiology and
`natural history of atherosclerosis.19 Probability samples of men
`and women ages 45 to 64 years were recruited from four commu-
`
`Editorial, see page 1496
`
`From the Department of Epidemiology (Drs. Stang, Rose, and Ephross, and A.P. Carson), University of North Carolina at Chapel Hill, Chapel Hill, NC;
`College of Health Sciences (Dr. Stang), West Chester University, West Chester, PA; Safety Evaluation and Epidemiology (Dr. Mo), Pfizer, Inc., New York,
`NY; GlaxoSmithKline Inc. (Dr. Ephross), Research Triangle Park, NC; Division of Epidemiology (Dr. Shahar), University of Minnesota, Minneapolis, MN;
`and Department of Epidemiology (Dr. Szklo), Johns Hopkins University, Baltimore, MD.
`Supported by NHLBI contracts N01-HC-55015, N01-HC-55016, N01-HC-55019, N01-HC-55020, N01-HC-55021, and N01-HC-55022 and by NIH, NHLBI
`NRSA grant 5T32HL07055 as well as a grant by GlaxoSmithKline.
`Dr. Rose was a paid consultant for GlaxoSmithKline in 2001. Dr. Stang is a paid consultant for GlaxoSmithKline (honoraria in excess of $10,000),
`Bristol-Myers Squibb, Pfizer, MedImmune, AstraZeneca, Allergan, Procter and Gamble, and Schering-Plough. Dr. Mo is employed by Pfizer. Dr. Ephross is
`employed by GlaxoSmithKline and receives stock options.
`Received October 7, 2004. Accepted in final form December 29, 2004.
`Address correspondence and reprint requests to Dr. Kathryn M. Rose, Cardiovascular Disease Program, Department of Epidemiology, University of North
`Carolina at Chapel Hill, Bank of America Center, Suite 306, 137 East Franklin Street, Chapel Hill, NC 27514; e-mail: kathryn_rose@unc.edu
`
`Copyright © 2005 by AAN Enterprises, Inc. 1573
`
`1
`
`EX2159
`Eli Lilly & Co. v. Teva Pharms. Int'l GMBH
`IPR2018-01425
`
`

`

`nities in the United States: Forsyth County, NC; Jackson, MS;
`Washington County, MD; suburbs of Minneapolis, MN. African
`Americans were exclusively sampled in Jackson and oversampled
`in Forsyth County so race-specific estimates would be possible. A
`detailed account of the design and procedures of the ARIC Study
`has been published.19
`The baseline examination, conducted during 1987 through
`1989, included 15,792 participants. The current study was re-
`stricted to those who participated in the third follow-up examina-
`tion (1993 to 1995), when a lifetime history of headaches was
`ascertained (N ⫽ 12,887). The following participants were ex-
`cluded from the current study: participants who were not white or
`African American (n ⫽ 38); African Americans residing in Minne-
`apolis, MN, and Washington County, MD (n ⫽ 42); and persons
`with missing headache information (n ⫽ 57). After these exclu-
`sions, 12,750 persons remained eligible for inclusion in this study.
`Assessment of headache history. At the third clinic examina-
`tion, trained interviewers asked participants about their lifetime
`history of headaches lasting 4 or more hours. Lifetime history of
`migraine headaches was defined using a modification of the 1988
`International Headache Society (IHS) diagnostic criteria. The
`modified IHS criteria for the classification of migraine headaches
`in the ARIC Study included 1) headache lasting at least 4 hours;
`2) headache accompanied by throbbing, pulsation, or pounding or
`was unilateral; 3) headaches occurring with nausea, vomiting, or
`sensitivity to light or sound; 4) lifetime history of 1 or more years
`of such headaches meeting the previous criteria. An affirmative
`response to a question about the occurrence of spots, jagged lines,
`or heat waves in one or both eyes was used to differentiate mi-
`graine with aura from migraine without aura. Headaches lasting
`at least 4 hours but not satisfying all the migraine criteria, re-
`ferred to as “other headaches,” were also evaluated. These head-
`aches were further classified by the presence or absence of aura
`symptoms as previously described.
`Assessment of stroke and TIA symptoms. To assess symptoms
`of stroke or TIA, participants were asked by a trained interviewer
`at each clinic examination whether they had experienced the sud-
`den onset of six neurologic symptoms: loss of vision, double vision,
`speech dysfunction, weakness or paralysis, numbness or tingling,
`and dizziness. For each of the six symptoms, duration and con-
`comitant symptoms were assessed. Participants who reported dou-
`ble vision, numbness, and dizziness were asked additional
`questions to ascertain a possible noncerebrovascular cause. A
`computerized diagnostic algorithm on stroke and TIA symptoms
`was used to simulate clinical reasoning used to differentiate vas-
`cular from other events. This diagnostic algorithm has previously
`been used in the Asymptomatic Carotid Atherosclerosis Study and
`has been evaluated as a tool for detection of stroke or TIA.20,21 The
`algorithm’s agreement with the diagnosis of TIA or stroke was
`80%, with a reported sensitivity of 88% and specificity of 72%.21
`Dizziness alone, numbness or paralysis in one body part alone,
`or double vision that did not disappear on closing one eye was not
`sufficient to classify a participant as experiencing stroke or TIA
`symptoms according to the diagnostic algorithm. Neurologic symp-
`toms with a sudden onset and duration of at least 30 seconds were
`used to classify participants as having experienced stroke or TIA
`symptoms. TIA symptoms were differentiated from stroke symp-
`toms based on self-report of the duration of the longest episode.
`Symptoms that resolved within 24 hours were classified as TIA
`symptoms and those that persisted longer were classified as
`stroke symptoms. Any persons exhibiting both stroke and TIA
`symptoms during clinic examinations 1 to 3 were classified as
`having stroke symptoms only.
`Assessment of ischemic stroke. ARIC study participants re-
`ceived annual telephone calls from trained interviewers to ascer-
`tain any hospitalizations during the previous year and changes in
`health status. Local hospitals provided lists of cardiovascular dis-
`ease discharges, which were monitored for the presence of ARIC
`participants. Cases were eligible for possible validation for stroke
`if any of the following occurred: 1) medical records contained dis-
`charge diagnostic codes suggestive of cerebrovascular events (In-
`ternational Classification of Disease, 9th Revision, Clinical
`Modification codes 430 to 438); 2) cerebrovascular keywords were
`present in discharge summary or nurse notes; 3) diagnostic CT or
`MRI with cerebrovascular findings; or 4) admission to neurologic
`intensive care unit.22 The medical records of participants qualify-
`ing for possible validation were obtained, sent to a central loca-
`
`1574 NEUROLOGY 64 May (1 of 2) 2005
`
`tion, and abstracted by a trained nurse. A computer algorithm and
`an independent physician reviewer were used to determine the
`final diagnosis for each participant. A second physician-reviewer
`adjudicated conflicting diagnoses between the
`computer-
`determined diagnosis and the initial physician-reviewer. Due to
`the small number of hemorrhagic strokes, only ischemic strokes
`were evaluated.
`Verified ischemic stroke included all stroke events occurring
`between 1987 and December 2001, regardless of its occurrence
`before or after the assessment of lifetime headache history at the
`third clinic examination. Incident ischemic stroke, a subset of
`verified ischemic stroke, was restricted to those events determined
`to have occurred after headache onset among participants without
`a history of ischemic stroke at baseline. Participants were not
`directly queried about when their headaches first began, so a
`question about the number of years with headache was used to
`determine whether a participant’s headache history potentially
`predated their participation in the ARIC study. Thus, verified
`ischemic strokes included all strokes occurring during the ARIC
`Study (1987 to 2001), whereas incident ischemic stroke events
`were a subset of verified ischemic stroke limited to those reported
`to have first occurred after the onset of the headaches among
`participants with no history of stroke at baseline.
`Statistical analysis. The sociodemographic variables included
`in this analysis were age, sex, and a combined race and ARIC field
`center variable. Additional covariates included diabetes status,
`smoking status (current, former, never), cigarette smoking pack-
`years, parental history of migraine, regular use of aspirin or non-
`steroidal antiinflammatory drugs (NSAIDs), use of hypertension
`medications, systolic blood pressure, and total serum cholesterol.
`Smoking status, cigarette smoking pack-years, parental history of
`migraine, and regular use of aspirin and NSAIDs were self-reported.
`A standard sphygmomanometer was used to take blood pressure in
`the seated position after 5 minutes of rest, based on the average of
`the second and third measurements. Use of antihyptertensive medi-
`cations was determined by evaluating the medication bottles that the
`participant brought to the clinical examination. Diabetes mellitus
`was defined by a fasting blood glucose level ⱖ126 mg/dL, a nonfast-
`ing blood glucose level of ⱖ200 mg/dL, self-report of physician diag-
`nosis, or use of hypoglycemic medications. All the variables were
`assessed at the third clinical examination with the exception of ciga-
`rette smoking pack-years, which was obtained at the first ARIC
`clinical examination (1987 to 1989).
`Multivariate logistic regression was used to evaluate the relation-
`ship between headache history and stroke symptoms, TIA symp-
`toms, verified ischemic stroke, and incident ischemic stroke. The
`models were initially adjusted for age, sex, and race/center. Addi-
`tional adjustment for parental history of migraine, smoking status,
`cigarette smoking pack-years, diabetes status, regular use of aspirin,
`regular use of NSAIDs, use of hypertension medications, systolic
`blood pressure, and total serum cholesterol level was performed in
`subsequent analyses. Interaction terms for age, race, sex, and head-
`ache history were also included in subsequent analyses but were
`ultimately excluded due to small strata and low power to detect
`effect modification. All statistical analyses were performed using
`SAS version 8.2 software (SAS Institute, Inc., Cary, NC).
`
`Results. At the third clinic examination (1993 to 1995),
`the mean age of the study population was 60 years, with
`73% of the population age 55 years and older. The popula-
`tion was 77% white and 56% female, and 62% had at least
`a high school education. Three percent of participants had
`a history of migraine with aura, 5% had a history of mi-
`graine without aura, 2% had a history of other headaches
`with aura, and 12% had a history of other headaches with-
`out aura. Additional details pertaining to the sociodemo-
`graphic characteristics of ARIC participants by headache
`status have been published.23
`Table 1 presents the age-adjusted prevalences or means
`for demographic characteristics and selected risk factors
`by headache status. Women comprised a large proportion
`of all headache groups, with the difference being most
`pronounced for migraine. Some of the risk factors also
`differed by headache type with higher proportions of med-
`
`2
`
`

`

`Table 1 Percentages (means) for demographic characteristics and selected risk factors by headache status in the Atherosclerosis Risk in
`Communities Study, 1993–1995
`
`Migraine with
`aura, n ⫽ 345
`
`Migraine
`without aura,
`n ⫽ 670
`
`Other headache
`with aura,
`n ⫽ 243
`
`Other headache
`without aura,
`n ⫽ 1446
`
`No
`headache,
`n ⫽ 10,046
`
`58.5
`15.1
`83.8
`40.1
`14.4
`10.8
`41.1
`29.1
`35.0
`124.3
`216.5
`
`58.3
`8.7
`80.5
`28.7
`14.2
`9.4
`33.8
`26.6
`24.6
`121.5
`211.5
`
`59.3
`25.5
`72.8
`22.6
`20.6
`15.6
`40.3
`28.3
`39.2
`127.0
`204.4
`
`58.5
`16.7
`63.7
`15.0
`16.1
`9.3
`32.6
`21.2
`27.9
`124.1
`209.1
`
`60.4
`24.9
`51.5
`8.1
`18.3
`13.5
`27.6
`15.9
`31.9
`124.8
`206.9
`
`Mean age, y
`% African American
`% Female
`% Parental history of migraines
`% Current smoker*
`% Diabetic*
`% Regular aspirin users*
`% Regular NSAID users*
`% Use hypertension medications*
`Systolic blood pressure (mean)*
`Total cholesterol (mean)*
`
`* Age adjusted.
`
`NSAID ⫽ nonsteroidal antiinflammatory drug.
`
`ication use (aspirin, NSAIDS, and antihypertensives)
`among participants with headaches with aura vs those
`with headaches without aura.
`The crude prevalence of stroke symptoms, TIA symp-
`toms, verified ischemic stroke, and incident ischemic
`stroke was 3.6%, 4.8%, 3.0%, and 2.8%. The age, sex, and
`race-adjusted prevalence of verified ischemic stroke was
`7.0% for migraine with aura, 2.1% for migraine without
`aura, 7.1% for other headaches with aura, 3.3% for other
`headaches without aura, and 2.7% for no headaches.
`Among whites, TIA symptoms were more prevalent in
`women than in men (6.1% vs 4.0%). In contrast, among
`both whites and African Americans, verified ischemic
`stroke was more prevalent in men than in women. Stroke
`symptoms, TIA symptoms, and verified ischemic stroke
`were more prevalent among those with diabetes, hyperten-
`sion, and higher total serum cholesterol as well as among
`current smokers (data not shown).
`In age-, sex-, and race/center-adjusted multivariate
`models, migraine with aura and other headaches with
`aura were strongly associated with stroke symptoms and
`TIA symptoms (table 2). Migraine without aura and other
`headaches without aura were associated with both stroke
`and TIA symptoms; however, the magnitude of these asso-
`ciations tended to be modest and some were not statisti-
`cally significant. No significant
`interaction between
`headache status and the included covariates was found.
`In contrast, migraine without aura and other headaches
`without aura were not associated with verified ischemic
`stroke, whereas migraine with aura and other headache
`with aura were significantly associated with verified ische-
`mic stroke (table 3). These associations were not modified
`by the selected covariates nor did they change substan-
`tially after additionally controlling for other covariates (di-
`abetes,
`regular aspirin use,
`regular NSAID use,
`hypertension medication use, systolic blood pressure, pack
`years of smoking, current smoking status, parental history
`of migraine, and total serum cholesterol). When we re-
`stricted verified ischemic strokes to those determined to be
`incident events (i.e., no history of stroke before the base-
`
`line examination and the first stroke occurred after head-
`ache onset), results did not substantively change.
`
`Discussion. Migraines and other headaches, in
`particular those accompanied by aura, were associ-
`ated with an increased occurrence of stroke symp-
`toms, TIA symptoms, verified ischemic stroke, and
`incident ischemic stroke in this middle-aged and
`older study population. Previous studies of migraine
`and stroke have suggested an increased risk of
`stroke for migraineurs, particularly those with aura;
`however, these studies were limited to younger
`
`Table 2 Multivariate odds ratios (95% CI) for the association of
`headache history with stroke and TIA symptoms, the
`Atherosclerosis Risk in Communities Study, 1993–1995
`
`Stroke
`symptoms,
`n ⫽ 12,065
`
`TIA
`symptoms,
`n ⫽ 12,213
`
`Model 1*
`6.06 (4.18–8.78) 5.06 (3.69–6.94)
`Migraine with aura
`2.35 (1.62–3.41) 1.43 (1.01–2.04)
`Migraine without aura
`4.17 (2.62–6.63) 4.66 (3.20–6.78)
`Other headache with aura
`Other headache without aura 1.33 (0.97–1.83) 1.41 (1.10–1.83)
`Model 2†
`5.46 (3.64–8.18) 4.28 (3.02–6.08)
`Migraine with aura
`2.45 (1.66–3.60) 1.35 (0.93–1.96)
`Migraine without aura
`3.68 (2.26–5.99) 4.53 (3.08–6.67)
`Other headache with aura
`Other headache without aura 1.39 (1.00–1.92) 1.35 (1.03–1.76)
`
`* Adjusted for age, sex, and race/center.
`† Adjusted for age, sex, race/center, hypertension medication use,
`regular aspirin use, regular nonsteroidal antiinflammatory
`drug use, systolic blood pressure, diabetes, parental history of
`migraines, smoking status, pack-years of smoking, and total
`cholesterol level.
`
`May (1 of 2) 2005 NEUROLOGY 64 1575
`
`3
`
`

`

`Table 3 Multivariate odds ratios (95% CI) for the association of headache history with verified ischemic stroke and incident ischemic
`stroke, the Atherosclerosis Risk in Communities Study, 1993–1995
`
`Verified ischemic stroke,*
`n ⫽ 12,681
`
`Incident ischemic stroke,†
`n ⫽ 11,447
`
`Model 1‡
`Migraine with aura
`Migraine without aura
`Other headache with aura
`Other headache without aura
`Model 2§
`Migraine with aura
`Migraine without aura
`Other headache with aura
`Other headache without aura
`
`2.68 (1.58–4.57)
`0.79 (0.40–1.55)
`2.58 (1.47–4.54)
`1.25 (0.88–1.77)
`
`2.81 (1.60–4.92)
`0.82 (0.39–1.69)
`1.84 (0.99–3.45)
`1.39 (0.96–2.01)
`
`1.84 (0.89–3.82)
`0.75 (0.33–1.71)
`2.91 (1.39–6.11)
`0.98 (0.57–1.70)
`
`2.07 (0.96–4.44)
`0.86 (0.37–2.00)
`2.40 (1.11–5.19)
`1.05 (0.58–1.91)
`
`* Includes all verified stroke events occurring after the baseline examination.
`† Includes a subset of verified stroke events restricted to first stroke events and strokes occurring after the onset of headaches.
`‡ Adjusted for age, gender, and race/center (373 verified ischemic strokes and 314 incident ischemic strokes).
`§ Adjusted for age, gender, race/center, hypertension medication use, regular aspirin use, regular nonsteroidal antiinflammatory drug
`use, systolic blood pressure, diabetes, parental history of migraines, smoking status, pack-years of smoking, and total cholesterol level
`(358 verified ischemic strokes and 301 incident ischemic strokes).
`
`women.7-11 In the studies that examined the relation-
`ship in older adults,13-18 findings have been inconsistent
`with the one study that ascertained aura status report-
`ing a stronger association between migraine with aura
`and cerebrovascular events.15 These inconsistent find-
`ings may be due to several methodologic variations,
`such as reliance on coded or recorded information con-
`cerning possible effect modifiers and confounders, use
`of self-reported stroke outcomes, not using symptom-
`based questions to assess headache history, and/or the
`use of a general question about a history of physician-
`diagnosed migraine. In this study, headache status
`was ascertained using a standardized symptom-based
`questionnaire and outcomes and other covariates were
`assessed using standard research protocols.
`The presence of aura was strongly associated with
`the occurrence of stroke/TIA symptoms and ischemic
`stroke events for both migraines and other headaches.
`This finding is consistent with previous studies8-11 of
`migraine reported in younger, predominantly female
`populations. One study15 has postulated that the rela-
`tionship between migraine with aura and ischemic
`stroke may be due to an interaction between migraine
`and other stroke risk factors. We assessed effect modi-
`fication of the migraine-stroke association by age and
`other stroke risk factors but did not find meaningful
`variations. However, our power to detect interactions
`was limited, so we cannot exclude the possibility that
`such variations may exist. Migraine without aura and
`other headaches without aura had significant but more
`modest associations with stroke and TIA symptoms but
`no significant association with verified ischemic stroke.
`There are several possible limitations to our
`study. Because we assessed lifetime history of head-
`ache, recall error may have occurred. Migraine, espe-
`cially with aura, and cerebrovascular events share
`many clinical features that may make it difficult for
`a participant to discriminate in the recall of epi-
`1576 NEUROLOGY 64 May (1 of 2) 2005
`
`sodes. For example, TIAs, especially those involving
`the posterior circulation, are often accompanied by
`headache.24 Similarly, because our aura question
`was framed within the context of headaches, it may
`be difficult for a participant to distinguish individual
`symptoms of migraine or its accompaniments as be-
`ing part of a headache syndrome because some
`symptoms may occur independent of headache. This
`would not be a rare occurrence as others have shown
`that migraineous visual accompaniments occur in 1.2%
`of the population, typically after age 50 years.25 Be-
`cause the strength of the association between stroke
`and TIA symptoms and headaches with auras is
`strong, these data suggest that some fraction of stroke
`and TIA symptoms may indeed represent migraine
`phenomena, whereas the reverse is also possible. In
`this study, aura was defined purely as a visual phe-
`nomenon, so we were unable to identify additional par-
`ticipants with nonvisual aura symptoms
`(i.e.,
`sensorimotor). This may have affected our observed as-
`sociations with cerebrovascular symptoms. This mis-
`classification probably would have attenuated the
`association between aura and cerebrovascular symp-
`toms. The strong association found for other headaches
`with aura and cerebrovascular events may also have
`been due to a propensity for these participants to re-
`port aura symptoms associated with their usual head-
`aches when these symptoms may actually have
`occurred independent of their usual headaches but as
`part of a cerebrovascular event.
`It is also possible that participants who experienced
`a cerebrovascular event that was preceded by head-
`ache or headache symptoms were more likely to report
`headache. However, this is unlikely, as one would have
`expected to see a higher rate of migraine or headache
`in general in this analysis than what has been re-
`ported. In addition, headache history was not obtained
`until the third clinic examination, so survival bias is a
`
`4
`
`

`

`possibility because 19% of participants from the first
`clinic examination who subsequently died or did not
`attend the third clinical examination were not included
`in this study. This may have caused an underestima-
`tion of the true association between migraine and
`stroke. Of the four cerebrovascular outcomes evalu-
`ated, incident stroke was the only one for which the
`exposure, by definition, had to occur before the out-
`come and is the outcome for which the association with
`migraine was weakest. For both verified ischemic
`stroke and incident stroke, the association with head-
`ache types was consistent and of the same magnitude;
`however, the temporal requirement for headache, cou-
`pled with the relatively small number of outcomes,
`may have selected participants in whom competing
`risks for these events attenuated the impact of head-
`ache type found for the other outcomes.
`The association between migraine and stroke was
`slightly attenuated after adjusting for risk factors.
`This suggests that the strength of the association
`between aura and cerebrovascular outcomes may be
`independent of classic risk factors and may act
`through a separate mechanism. However, because
`risk factor status was obtained at the same time as
`migraine history, we cannot infer that risk factor
`status at that time reflects cumulative history.
`It has been suggested that migraine may be part of
`a more generalized vasospastic syndrome because of its
`possible association with Raynaud phenomenon26,27
`and Prinzmetal angina.28 This explanation is plausible,
`because migraine in general and aura specifically are
`believed to be manifestations of the spreading depres-
`sion of Leao,29 a spreading wave of depolarization
`across the cortex that is accompanied by decreases in
`blood flow.30 The decreased blood flow in combination
`with increased platelet aggregability may increase the
`likelihood of stroke in the migraineur. Thus, mi-
`graineurs may have an increased vascular instability
`that persists long after the classic symptoms of mi-
`graine have ended. Although the current study cannot
`prove or disprove this theory, it does add additional
`evidence to the predictive value of headache and aura
`for cerebrovascular symptoms and ischemic stroke in
`an older population with increased risk of cerebrovas-
`cular events. Because aura represents a distinct clini-
`cal and physiologic finding, its strong association with
`cerebrovascular outcomes can potentially be related to
`either the shared common mechanism of reduced blood
`flow or diagnostic misclassification. Aura itself may be
`such a striking clinical finding that the patient is less
`likely to have difficulty recalling its occurrence, which
`results in higher specificity and sensitivity of aura
`symptoms. In addition, the underlying physiology of
`aura, apparently directly related to changes in blood
`flow, would appear to increase the risk of cerebrovascu-
`lar events in those patients most sensitive to these
`effects.
`Further research that focuses on the potential
`mechanisms underlying the complex relationship be-
`tween headaches, particularly when accompanied by
`aura, and cerebrovascular outcomes is warranted.
`
`Acknowledgment
`The authors thank the staff and participants in the ARIC Study
`for their contributions.
`References
`1. Kalendovsky Z, Austin JH. ‘Complicated migraine’: its association with
`increased platelet aggregability and abnormal plasma coagulation fac-
`tors. Headache 1975;15:18–35.
`2. Fitscha P, Sinzinger H. Platelet hyperactivity during acute migraine
`attack. Vasa 1984;13:224–227.
`3. Lauritzen M, Olesen J. Regional cerebral blood flow during migraine
`attacks by Xenon-133 inhalation and emission tomography. Brain 1984;
`107:447–461.
`4. Broderick JP, Swanson JW. Migraine-related strokes. Clinical profile
`and prognosis in 20 patients. Arch Neurol 1987;44:868–871.
`5. Heinrich JB, Sandercock PAG, Warlow CP, Jones LN. Stroke and mi-
`graine in the Oxfordshire Community Stroke Project. J Neurol 1986;
`233:257–262.
`6. Sochurkova D, Moreau T, Lemesle M, et al. Migraine history and
`migraine-induced stroke in the Dijon Stroke Registry. Neuroepidemiol-
`ogy 1999;18:85–91.
`7. Collaborative Group for the Study of Stroke in Young Women. Oral
`contraceptives and stroke in young women. JAMA 1975;231:718–722.
`8. Tzourio C, Kittner SJ, Bousser MG, Alperovitch A. Migraine and stroke
`in young women. Cephalalgia 2000;20(3):190–199.
`9. Tzourio C, Tehindrazanarivelo A, Iglesias S, et al. Case-control study of mi-
`graine and risk of ischaemic stroke in young women. BMJ 1995;310:830–833.
`10. Carolei A, Marini C, De Matteis G. History of migraine and risk of cerebral
`ischaemia in young adults. The Italian National Research Council Study
`Group on Stroke in the Young. Lancet 1996;347:1503–1506.
`11. Chang CL, Donaghy M, Poulter N. Migraine and stroke in young wom-
`en: case-control study. The World Health Organisation Collaborative
`Study of Cardiovascular Disease and Steroid Hormone Contraception.
`BMJ 1999;318:13–18.
`12. Kurth T, Slomke M, Kase C, Cook N, Gaziano M, Buring J. Headache,
`migraine, and the risk of stroke in women. Program and abstracts of
`the American Academy of Neurology 56th Annual Meeting, April 24–
`May 1, 2004, San Francisco, CA, abstract S24.001.
`13. Mosek A, Marom R, Korczyn AD, Bornstein N. A history of migraine is
`not a risk factor to develop an ischemic stroke in the elderly. Headache
`2001;41:399–401.
`14. Tzourio C, Iglesias S, Hubert JB, et al. Migraine and risk of ischaemic
`stroke: a case-control study. BMJ 1993;307:289–292.
`15. Henrich JB, Horwitz RI. A controlled study of ischemic stroke risk in
`migraine patients. J Clin Epidemiol 1989;42:773–780.
`16. Merikangas KR, Fenton BT, Cheng SH, Stolar MJ, Risch N. Association
`between migraine and stroke in a large-scale epidemiological study of
`the United States. Arch Neurol 1997;54:362–368.
`17. Buring JE, Hebert P, Romero J, et al. Migraine and subsequent risk of
`stroke in the Physicians’ Health Study. Arch Neurol 1995;52:129–134.
`18. Mitchell P, Wang JJ, Currie J, Cumming RG, Smith W. Prevalence and
`vascular associations with migraine in older Australians. Aust N Z
`J Med 1998;28:627–632.
`19. The ARIC Investigators. The Atherosclerosis Risk in Communities (ARIC)
`Study: design and objectives. Am J Epidemiol 1989;129:687–702.
`20. The Asymptomatic Carotid Atherosclerosis Study Group. Study design
`for randomized prospective trial of carotid endarterectomy for asymp-
`tomatic atherosclerosis. The Asymptomatic Carotid Atherosclerosis
`Study Group. Stroke 1989;20:844–849.
`21. Karanjia PN, Nelson JJ, Lefkowitz DS, et al. Validation of the ACAS
`TIA/stroke algorithm. Neurology 1997;48:346–351.
`22. Rosamond WD, Folsom AR, Chambless LE, Wang C, McGovern PG, et
`al. Stroke incidence and survival among middle-aged adults: 9 year
`follow-up of the Atherosclerosis Risk in Communities (ARIC) cohort.
`Stroke 1999;30:736–743.
`23. Carson AP, Rose KM, Sanford CP, et al. Lifetime prevalence of mi-
`graine and other headaches lasting 4 or more hours: the Atherosclerosis
`Risk in Communities (ARIC) study. Headache 2004;44:20–28.
`24. Vestergaard K, Andersen G, Nielsen MI, Jensen TS. Headache in
`stroke. Stroke 1993;24:1621–1624.
`25. Wijman CAC, Wolf PA, Kase CS, et al. Migraineous visual accompani-
`ments are not rare in late life: The Framingham Study. Stroke 1998;29:
`1539–1543.
`26. Stang P, Sternfeld B, Sidney S. Migraine headache in a pre-paid health
`plan: ascertainment, demographics, physiological and behavioral fac-
`tors. Headache 1996;36:69–76.
`27. Zahavi I, Chagnac A, Hering R, et al. Prevalence of Raynaud’s phenom-
`enon in patients with migraine. Arch Intern Med 1984;144:742–744.
`28. Miller D, Waters DD, Warnica W, et al. Is variant angina the coronary
`manifestation of a generalized vasospastic disorder? N Engl J Med
`1981;304:763–766.
`29. Leao AA. Spreading depression of activity in cerebral cortex. J Neuro-
`physiol 1944;7:379–390.
`30. Woods RP, Iacoboni M, Mazziotta JC. Brief report: bilateral sprea