Ethnic Disparities in Stroke Recognition in Individuals with Prior Stroke

Reserch Articles Ethnic Disprities in Stroke Recognition in Individuls with Prior Stroke Chrles Ellis, PhD Leonrd E. Egede, MD, MS,c SYNOPSIS Ojective. Studies of stroke wreness suggest tht knowledge of erly wrning signs of stroke is low in high-risk groups. However, little is known out stroke knowledge mong individuls with history of prior stroke who re t significnt risk for recurrent stroke. Methods. Dt from 2,970 dults with history of prior stroke from the 2003 Behviorl Risk Fctor Surveillnce System were exmined. Recognition of the five wrning signs of stroke nd pproprite ction to cll 911 ws compred cross three rcil/ethnic groups: non-hispnic white, non-hispnic lck, nd Hispnic/other. Multiple logistic regression nlyses were used to: (1) determine the ssocition etween rce/ethnicity nd recognition of multiple stroke signs nd pproprite first ction nd (2) identify independent correltes of recognition of multiple stroke signs nd tking pproprite ction to seek tretment mong individuls with prior stroke. Results. Recognition of ll five signs of stroke nd tking pproprite ction to cll 911 ws lowest mong the non-hispnic lck group (22.3%) nd Hispnic/other group (16.7%). In multivrite models, Hispnic/other (odds rtio [OR] 0.42 [0.25, 0.71]), ge 50 64 (OR 0.64 [0.43, 0.97]), ge $65 (OR 0.36 [0.23, 0.55]), nd.high school eduction (OR 1.79 [1.22, 2.63]) emerged s independent correltes of recognition of ll five signs of stroke nd first ction to cll 911. Conclusions. Less thn 35% of people with prior stroke cn distinguish the complex symptom profile of stroke nd tke pproprite ction to cll 911. Trgeted eductionl ctivities tht re sensitive to differences in rce/ethnicity, ge, nd eduction levels re needed for individuls with prior stroke. Deprtment of Rehilittion Sciences, Medicl University of South Crolin, Chrleston, SC Deprtment of Medicine, Center for Helth Disprities Reserch, Medicl University of South Crolin, Chrleston, SC c Rlph H. Johnson Vetern s Affirs Medicl Center, Chrleston, SC Address correspondence to: Leonrd E. Egede, MD, MS, Center for Helth Disprities Reserch, Medicl University of South Crolin Deprtment of Medicine, 135 Rutledge Ave., Rm. 280H, Chrleston, SC 29425; tel. 843-792-2969; fx 843-876-1201; e mil <egedel@musc.edu>. 2008 Assocition of Schools of Pulic Helth 514

Ethnic Disprities in Stroke Recognition 515 Stroke is leding cuse of mortlity nd moridity in the U.S., with estimted nnul direct nd indirect costs estimted t nerly $58 illion. 1 A numer of strtegies re currently in plce to reduce the incidence of stroke nd the ssocited strin on finncil nd helth-cre resources. Presently, ntionwide comprehensive cmpigns re ongoing to increse pulic wreness of stroke, prticulrly erly wrning signs. 2,3 Despite ntionl efforts to improve stroke wreness, the pulic s knowledge of erly wrning signs of stroke remins low. 4 9 Specific fctors tht contriute to poor stroke wreness remin uncler. For instnce, lthough it is generlly expected tht individuls t gretest risk for stroke would exhiit greter stroke wreness, studies of t-risk individuls suggest otherwise. High-risk groups, rcil/ethnic minorities, 4,5,8,10,11 the elderly, 4,5,8,10 nd individuls with risk fctors for stroke 5,8,10 re lest likely to recognize erly wrning signs of stroke nd tke pproprite ction. Individuls with prior stroke re one stroke risk fctor group tht hs received miniml ttention in the stroke wreness literture. Studies of stroke wreness mong individuls with stroke risk fctors trditionlly emphsize hypertension, 4,5,8,10 crdiovsculr disese (CVD), 4,5,8,12 dietes, 4,5,8,12 high cholesterol, 4,8,12,13 smoking history, 4,5,7,8,10,12 nd other stroke risk fctors. Fewer studies hve considered individuls with prior history of stroke 14,15 even though estimtes suggest tht within five yers, 24% of women nd 42% of men with history of prior stroke will hve second stroke. 16 Further, individuls with history of prior stroke typiclly exhiit numer of ssocited fctors (e.g., rce/ethnicity, ge, presence of other stroke risk fctors) tht increse their likelihood of second stroke. Understnding stroke wreness in this high-risk group is criticl to the development of pproprite eductionl strtegies to reduce their likelihood of recurrent stroke. Furthermore, the study of this t-risk group would improve our understnding of independent correltes of stroke wreness nd pproprite first ction tken, therey providing support to current efforts to decrese stroke incidence-ssocited urden. The purpose of this study ws to exmine recognition of the five wrning signs of stroke nd pproprite ction to cll 911 in individuls with prior stroke. We chose to compre stroke wreness cross rcil/ethnic groups of individuls with prior stroke ecuse rcil/ethnic disprities exist in the recognition of erly wrning signs of stroke in the generl popultion, even though rcil/ethnic minorities re t significntly higher risk for stroke. 1 We sought to exmine the influence of rce/ethnicity on stroke wreness in individuls with prior stroke nd to identify independent correltes of recognition of erly wrning signs of stroke nd pproprite ction to cll 911. Our reserch questions were s follows: 1. Are there rcil/ethnic differences in recognition of multiple stroke signs nd tking pproprite ction to cll 911 mong individuls with prior stroke? 2. Are there rcil/ethnic differences in odds of recognition of multiple stroke signs nd tking pproprite ction to cll 911 fter controlling for relevnt covrites mong individuls with prior stroke? 3. Wht re the independent correltes of recognition of multiple stroke signs nd tking pproprite ction to cll 911 mong individuls with prior stroke? We hypothesized tht: (1) individuls with prior stroke would differ in the recognition of multiple stroke signs nd tking pproprite ction to cll 911 y rce/ethnicity; (2) the odds of recognizing the multiple stroke signs nd tking pproprite ction to cll 911 mong individuls with prior stroke would differ y rce/ethnicity fter controlling for relevnt covrites; nd (3) independent correltes of recognition of multiple stroke signs nd tking pproprite ction to cll 911 mong individuls with prior stroke would emerge tht differed y rce/ethnicity. RESEARCH DESIGN AND METHODS Study setting nd smple We nlyzed dt from the 2003 Behviorl Risk Fctor Surveillnce System (BRFSS) survey. The BRFSS is stte-sed, rndom-digit-diled telephone survey of the U.S. popultion $18 yers of ge sponsored y the Centers for Disese Control nd Prevention. 17 The BRFSS uses complex smpling involving strtifiction, clustering, nd multistge smpling to yield ntionlly representtive estimtes. Surveys include core questions sked of ll prticipnts in modules on specific pulic helth topics of interest to stte helth progrms. Our smple included only individuls tht identified themselves s hving hd stroke. Demogrphic nd socioeconomic chrcteristics We creted three ge ctegories: 18 49, 50 64, nd $65 yers. These ge ctegories re used to detect grdients in stroke, which is known to increse with ge. The ge ctegories re defined y the BRFSS. We comined rce nd ethnicity to crete three rcil/ethnic groups: non-hispnic whites (whites), non-hispnic lcks (lcks), nd Hispnic/other. Rce/ethnicity

516 Reserch Articles is sed on self-report using ctegories creted y the BRFSS. Becuse the percentge of individuls who did not clssify themselves s non-hispnic whites (whites) or non-hispnic lcks (lcks) ws pproximtely 5% of the smple, third ctegory (Hispnic/other) ws creted y BRFSS to represent those who identified themselves s Hispnic, Asin, ntive Hwiin/Pcific Islnders, nd Americn Indin/Alsk Ntive. Three levels of eduction,high school grdute, high school grdute, nd.high school grdute were creted; nd four income ctegories,$25,000,,$50,000,,$75,000, nd.$75,000 were creted. We defined mritl sttus s mrried nd not mrried; employment sttus s employed nd unemployed; nd insurnce sttus s insured nd uninsured. We defined perceived helth sttus s excellent/very good/good vs. fir/poor nd identified ll individuls with usul helth-cre provider. We identified presence of stroke comoridities, such s dietes, hypertension, nd CVD. Recognition of stroke wrning signs nd pproprite ction to cll 911 The recognition of erly wrning signs of stroke nd ction to initite tretment ws sed on self-report. Responses were derived from the 2003 BRFSS Hert Attck nd Stroke module. Respondents indicted whether ny of the following wrning signs were n indiction of n imminent stroke: (1) sudden confusion, troule speking or understnding; (2) sudden numness or wekness of the fce, rm, or leg; (3) sudden troule seeing in one or oth eyes; (4) sudden troule wlking, dizziness, loss of lnce or coordintion; nd (5) sudden hedche with no known cuse. Respondents were lso sked, If you thought someone ws hving stroke, wht is the first thing you would do? Respondents chose from list of ctions tht included: (1) tke the ptient to the hospitl, (2) tell them to cll the doctor, (3) cll 911, (4) cll their spouse or fmily memer, or (5) do something else. Dt nlysis STATA Version 8.0 18 ws used for sttisticl nlysis to control for the complex survey design of the 2003 BRFSS nd provide estimtes tht generlize to the U.S. popultion. We performed four types of nlyses. First, we compred demogrphic chrcteristics of prticipnts y rce/ethnicity. Second, we compred recognition of the five individul stroke signs nd pproprite ction to cll 911 y rcil/ethnic group. Third, we rn multiple logistic regression models to exmine the independent effects of rce/ethnicity on recognition of ech of the five individul stroke wrning signs controlling for relevnt covrites. Non-Hispnic white respondents served s the reference group in multivrite models. Covrites were selected sed on clinicl relevnce nd evidence of confounding effect on stroke recognition in prior studies. The covrites used in ll multiple logistic regression models included ge, sex, eduction, income, mritl sttus, employment, insurnce sttus, nd comoridity. Comoridities included dietes, hypertension, nd CVD. Fourth, we rn multiple logistic regression models to identify independent correltes of: (1) recognition of ll five wrning signs collectively nd (2) recognition of the five collective wrning signs nd pproprite ction to cll 911. In multivrite models, reference groups were: rce/ethnicity (non-hispnic white); ge (18 49); sex (men); eduction (,high school grdute); income (,$25,000); mritl sttus (mrried); employment sttus (employed); insurnce sttus (insured); dietes (no); hypertension (no); CVD (no). RESULTS The 2003 BRFSS smple included 2,970 dults with prior stroke. Approximtely 53% of the smple ws $65 yers old, 56.8% were women, 75.2% were white, 35.4% were high school grdutes, 59.3% hd incomes,$25,0000, 50.5% were mrried, nd 18.4% were employed. Aout 90.0% hd helth insurnce, 43.0% reported their helth ws excellent, very good, or good, nd 92.7% hd usul cre provider. Approximtely 27.0% hd dietes, 69.5% hd hypertension, nd 37.2% hd CVD. Tle 1 compres the demogrphic chrcteristics of individuls with prior history of stroke y rce/ethnicity. There were significnt rcil/ethnic differences y ge, eduction, income, mritl sttus, insurnce sttus, nd presence of dietes nd hypertension. Recognition of signs of stroke nd pproprite first ction The mjority of prticipnts with prior stroke who prticipted in the 2003 BRFSS survey recognized individul wrning signs of stroke. Aout 95% recognized sudden confusion, or troule speking or understnding; 96% recognized sudden fcil wekness or numness of the rm or leg; 86% recognized sudden troule seeing in one or oth eyes; 93% recognized sudden troule with wlking, dizziness, or loss of lnce or coordintion; nd 80% recognized sudden hedche with no known cuse s wrning signs of stroke. Approximtely 83% identified clling 911 s the pproprite ction to tke if someone ws hving stroke. The recognition of the five wrning signs of

Ethnic Disprities in Stroke Recognition 517 Tle 1. Demogrphic chrcteristics of individuls with prior stroke y rce/ethnicity All Non-Hispnic white Non-Hispnic lck Hispnic/other n52,970 n52,236 n5413 n5321 Percent Percent Percent Percent P-vlue Age (in yers) 18 49 17.4 15.2 22.0 24.1 50 64 29.7 27.1 37.0 35.8 $65 52.9 57.7 41.0 40.1,0.001 Sex: women 56.8 54.8 62.7 60.6 0.112 Eduction,High school grdute 26.8 24.1 35.5 31.4 High school grdute 35.4 35.0 37.2 35.0.High school grdute 37.8 40.9 27.3 33.6 Income,$25,000 59.3 53.2 77.5 70.6,$50,000 25.9 29.1 17.3 18.7,$75,000 8.5 10.0 2.8 7.9 $$75,000 6.3 7.7 2.5 2.8 0.007,0.001 Mrried 50.5 55.5 28.4 50.3,0.001 Employed 18.4 18.5 13.8 23.5 0.082 Insured 90.3 92.3 82.5 89.0,0.001 Helth sttus: excellent/very good/good 43.0 45.1 39.4 34.9 0.060 Hs usul helth provider: yes 92.7 93.4 90.0 92.1 0.272 Dietes: yes 27.0 24.0 39.1 29.0,0.001 Hypertension: yes 69.5 66.7 80.6 71.9 0.003 Crdiovsculr disese: yes 37.2 37.4 31.8 43.5 0.149 P-vlue is for comprison cross the three rcil/ethnic groups. Sttisticlly significnt t p,0.05 stroke nd pproprite ction to cll 911 ws low cross the three groups. Overll, less thn 32% of respondents recognized ll five wrning signs of stroke nd only 26% recognized ll five wrning signs nd would cll 911 s n pproprite first ction if someone ws hving stroke (Tle 2). Rcil/ethnic differences in recognition of signs of stroke Significnt rcil/ethnic differences were present in recognition of sudden confusion, or troule speking or understnding s wrning sign of stroke. Blck people were lest likely to recognize this symptom (90.9%) compred with white respondents (96.0%) nd Hispnic/others (98.5%). Recognition of the remining four individul wrning signs of stroke ws nonsignificnt nd reltively similr cross groups. Significnt rcil/ethnic differences were present in recognizing ll five stroke wrning signs nd pproprite ction to cll 911. Hispnics/others were lest likely to recognize ll five wrning signs of stroke (19.8%) compred with white respondents (33.9%) nd lck respondents (27.6%). Hispnic/others (16.7%) were lso lest likely to recognize ll five wrning signs nd would cll 911 s the first ction if someone ws hving stroke compred with white (28.9%) nd lck (22.3%) respondents (Tle 2). Odds of recognition of stroke signs y rce/ethnicity Tle 3 shows the odds of recognizing the individul wrning signs of stroke y rce/ethnicity. With white respondents s the reference group nd djusting for relevnt covrites, there were no significnt rcil/ ethnic differences in recognizing ny of the individul stroke wrning signs. Independent correltes of stroke wreness nd pproprite ction Tle 4 shows the independent correltes of (1) recognition of the five collective signs of stroke nd (2) the five collective signs of stroke nd tking pproprite ction to cll 911. Rce/ethnicity, ge, nd eduction emerged s fctors tht independently influence recognition of the five collective signs of stroke nd tking

518 Reserch Articles Tle 2. Recognition of signs of stroke nd pproprite ction y rce/ethnicity in individuls with prior stroke All Non-Hispnic white Non-Hispnic lck Hispnic/other Percent Percent Percent Percent P-vlue Sudden confusion, troule speking or understnding 95.3 96.0 90.9 98.5 0.003 Sudden numness or wekness of the fce, rm, or leg 95.6 95.5 94.6 98.3 0.515 Sudden troule seeing in one or oth eyes 86.1 87.6 80.9 82.8 0.145 Sudden troule wlking, dizziness, loss of lnce or coordintion 93.3 93.6 92.3 92.1 0.820 Sudden hedche with no known cuse 80.0 79.7 79.8 82.7 0.886 Cll 911 s first ction 83.0 83.2 81.5 84.4 0.829 Recognized ll five symptoms 31.1 33.9 27.6 19.8,0.001 Recognized ll five symptoms nd pproprite ction to cll 911 26.4 28.9 22.3 16.7 0.001 P-vlue is for comprison cross the three rcil/ethnic groups. Sttisticlly significnt t p,0.05 pproprite ction to cll 911. With white respondents with prior stroke s the reference group, Hispnic/ other (OR 0.42 [0.26, 0.70]) were less likely to recognize the five collective signs of stroke. Hispnic/others were lso less likely to recognize the five collective signs of stroke nd tke pproprite ction to cll 911 (OR 0.42 [0.26, 0.70]). With individuls ged 18 49 s the reference group, respondents ged 50 64 (OR 0.64 [0.43, 0.97]) were less likely to recognize the five collective signs of stroke nd tke pproprite ction to cll 911. Individuls ged $65 (OR 0.34 [0.22, 0.51]) were less likely to recognize the five collective signs of stroke compred with the reference group (ged 18 49). Respondents ged $65 were lso less likely to recognize the five collective signs of stroke nd tke pproprite ction to cll 911 (OR 0.36 [0.23, 0.55]). Finlly, with individuls hving,high school eduction s the reference group, individuls with.high school eduction (OR 1.70 [1.17, 2.46]) were more likely to recognize the five collective signs of stroke nd lso more likely to recognize the five collective signs of stroke nd tke pproprite ction to cll 911 (OR 1.79 [1.22, 2.63]). DISCUSSION The results of this study demonstrted tht recognition of the individul wrning signs of stroke is high mong Tle 3. Odds of recognition of signs of stroke y rce/ethnicity in individuls with prior stroke Non-Hispnic lck Hispnic/other OR 95% CI OR 95% CI Sudden confusion, troule speking or understnding 0.61 0.21, 1.73 5.70 0.85, 37.99 Sudden numness or wekness of the fce, rm, or leg 0.96 0.25, 3.64 1.74 0.29, 10.40 Sudden troule seeing in one or oth eyes 0.76 0.36, 1.63 0.58 0.21, 1.55 Sudden troule wlking, dizziness, loss of lnce or coordintion 0.79 0.29, 2.16 0.65 0.21, 2.00 Sudden hedche with no known cuse 1.06 0.51, 2.19 1.31 0.59, 2.89 Non-Hispnic white is the reference group. Odds rtio djusted for ge, sex, eduction, income, mritl sttus, employment, insurnce, nd comoridity. Comoridity includes dietes, hypertension, nd crdiovsculr disese. OR 5 odds rtio CI 5 confidence intervl

Ethnic Disprities in Stroke Recognition 519 Tle 4. Independent correltes of recognition of ll five signs of stroke nd tking pproprite ction in individuls with prior stroke All five symptoms All five symptoms nd tking pproprite ction OR 95% CI OR 95% CI Rce/ethnicity Non-Hispnic white (reference) 1.00 1.00 1.00 1.00 Non-Hispnic lck 0.74 0.50, 1.09 0.73 0.49, 1.09 Hispnic/other 0.42 c 0.26, 0.70 0.42 c 0.25, 0.71 Age (in yers) 18 49 (reference) 1.00 1.00 1.00 1.00 50 64 0.68 0.46, 1.03 0.64 c 0.43, 0.97 $65 0.34 c 0.22, 0.51 0.36 c 0.23, 0.55 Sex Men (reference) 1.00 1.00 1.00 1.00 Women 1.32 0.99, 1.74 1.22 0.92, 1.64 Eduction,High school grdute (reference) 1.00 1.00 1.00 1.00 High school grdute 1.16 0.81, 1.67 1.26 0.86, 1.85.High school grdute 1.70 c 1.17, 2.46 1.79 c 1.22, 2.63 Income,$25,000 (reference) 1.00 1.00 1.00 1.00,$50,000 1.26 0.90, 1.76 1.15 0.82, 1.64,$75,000 1.53 0.88, 2.64 1.42 0.81, 1.64 $$75,000 0.75 0.42, 1.35 0.69 0.81, 1.28 Mritl sttus Mrried (reference) 1.00 1.00 1.00 1.00 Not mrried 0.92 0.69, 1.23 0.91 0.68, 1.23 Employment sttus Employed (reference) 1.00 1.00 1.00 1.00 Unemployed 1.08 0.74, 1.59 1.08 0.73, 1.62 Insurnce sttus Insured (reference) 1.00 1.00 1.00 1.00 Uninsured 1.11 0.68, 1.81 0.98 0.59, 1.62 Helth sttus 0.98 c 0.73, 1.33 1.08 0.79, 1.48 Dietes No (reference) 1.00 1.00 1.00 1.00 Yes 0.66 0.48, 0.90 0.70 0.51, 0.97 Hypertension No (reference) 1.00 1.00 1.00 1.00 Yes 1.19 0.87, 1.64 1.17 0.84, 1.62 Crdiovsculr disese No (reference) 1.00 1.00 1.00 1.00 Yes 1.19 0.90, 1.58 1.24 0.93, 1.67 Odds rtio djusted for ge, sex, eduction, income, mritl sttus, employment, insurnce, nd comoridity. Comoridity includes dietes, hypertension, nd crdiovsculr disese. Non-Hispnic white is the reference group. c Sttisticlly significnt OR 5 odds rtio CI 5 confidence intervl

520 Reserch Articles individuls with prior stroke. Similr levels of wreness exist cross rcil/ethnic groups of individuls with history of prior stroke, with the exception of sudden confusion, or troule speking or understnding, which ws lest likely recognized y lck respondents. In contrst, few with history of stroke recognize the five collective wrning signs of stroke nd would cll 911 s first ction for tretment. Poor recognition of the multisymptom stroke profile (ll five wrning signs) nd first ction to seek tretment ws low cross ll respondents, though significntly more prominent in lck nd Hispnic/other respondents. In this study, we exmined recognition of ll five stroke signs collectively to provide more ccurte index of stroke wreness fter stroke. Studies of stroke wreness typiclly ssess recognition of individul wrning signs/symptoms; 4,5,7,8,10 however, mesurement of individul wrning signs of stroke does not provide n ccurte index of wreness of the collective stroke signs tht cn occur. Mny ptients report significnt difficulty in the identifiction of wrning signs of stroke initilly ecuse stroke symptoms often vry in numer nd degree nd cn e hrd to recognize. 19 Our findings indicte stroke wreness is multidimensionl prolem in those individuls t the highest risk for stroke (i.e., those with stroke risk fctors such s history of prior stroke) for two resons. First, the proility of stroke increses exponentilly fter the first occurrence of stroke. Second, this mplified stroke risk nd greter need to understnd wrning signs of stroke is further complicted y fctors relted to rce/ethnicity. Becuse rce/ethnicity is elieved to independently increse risk for stroke, 4,5,8,10,11 improved stroke wreness is prticulrly criticl for rcil/ethnic minority popultions with prior stroke. Our previous investigtions of stroke wreness suggest rce/ethnicity my hve greter negtive influence on stroke wreness thn previously reported in the literture. We exmined rcil/ethnic differences in recognition of stroke wrning signs nd ction to initite tretment fter stroke mong 36,150 veterns in the 2003 BRFSS dt. 20 Recognition of the five individul wrning signs of stroke ws high mong veterns; however, few recognized ll five symptoms collectively or the multisymptom profile of stroke. Poor stroke wreness existed mong veterns despite hving equl ccess to primry nd emergency cre 21 nd more frequent nd equl opportunity for stroke eduction. Given these disprities in stroke wreness in popultions with equl ccess to cre, we might conclude tht rce/ethnicity is n independent fctor tht potentilly mgnifies poor seline stroke wreness. Further, poor stroke wreness ppers to exist mong individuls from rcil/ethnic minority ckgrounds whether they hve history of prior stroke or re sent of stroke history. 4,5,8,10,11 Our secondry findings provide evidence tht rce/ethnicity is n independent predictor of poor stroke wreness. Being Hispnic/other or older thn 65 yers of ge were independent correltes of poor recognition of the five collective wrning signs of stroke nd tking first pproprite ction to cll 911 for tretment. In multivrite models, rce/ethnicity emerged s n independent predictor of poor stroke wreness. Support for this hypothesis is found in n rticle y Pncioli nd collegues, who lso identified rce/ethnicity s predictor of poor stroke wreness. 5 Older ge (.65 yers) ws lso ssocited with poor stroke wreness, s noted in previous studies of stroke wreness. 5,10 Currently, there is less greement regrding primry predictors of poor stroke wreness prticulrly mong individuls t risk for stroke. Weltermnn nd collegues exmined stroke wreness (stroke symptom knowledge) in 93 stroke ptients nd 40 fmily memers nd volunteers nd found tht history of prior stroke, ge.70, nd poor self-reported helth sttus emerged s predictors of poor stroke knowledge. 15 While older ge, 5,10 rce, 5 nd lower levels of eduction 5,10 re frequently ssocited with poor stroke wreness, sex (mle), 5,10 hypertension, smoking sttus, nd self-reported poor helth sttus 10 hve lso een reported s independent fctors ssocited with poor recognition of erly wrning signs of stroke. One significnt predictor of greter stroke wreness did emerge: post-high school eduction. Individuls with.high school eduction were more likely to recognize the five collective wrning signs nd tke ction to cll 911. Improved wreness of erly wrning signs of stroke mong t-risk individuls is criticl to ntionl efforts to reduce stroke incidence, mortlity, nd moridity. People with prior history of stroke re t gret risk for second stroke, nd even though it is generlly expected they would hve greter stroke wreness nd tke pproprite ction, 13 our results do not gree. Therefore, when considering the high likelihood of second stroke nd our oservtions of poor stroke wreness even mong those with prior stroke, current eductionl strtegies re not chieving their desired outcome. In tht regrd, novel eductionl progrms nd strtegies must e considered tht re sensitive to the multidimensionl (i.e., rce, ge, eduction, stroke risk fctors) risk fctor profiles tht mny t-risk individuls exhiit. Improvements in pulic wreness of stroke re

Ethnic Disprities in Stroke Recognition 521 urgently needed ecuse poor wreness of the complex presenttion of stroke (ll five symptoms) hs numer of consequences. First nd foremost, poor recognition of erly wrning signs cn result in delys in seeking proper medicl cre. 22 24 Delys in seeking cre re ssocited with greter stroke severity, higher stroke mortlity, nd greter post-stroke disility. 25 28 The effectiveness of stroke medictions is then decresed when tretment is delyed. Despite significnt improvements in medictions designed to minimize stroke severity nd ssocited long-term disility, ptients who do not immeditely recognize stroke wrning signs nd seek urgent cre t symptom onset re less likely to hve positive stroke outcomes. 29 A numer of strtegies hve een proposed for eductionl progrms designed to improve stroke wreness. These include: (1) emphsize erly recognition of ll five primry wrning signs collectively, s well s ctivtion of the 911 system, 30 (2) reinforce the link etween presence of symptoms nd recognition tht stroke is occurring, 14 nd (3) stress the need to cll 911 to fcilitte erly intervention. Trgeted eductionl progrms must e designed for specific groups t highest risk for stroke. 14 Progrms designed for individuls with prior history of stroke should lso stress the link etween their current risk fctor (prior stroke) nd the likelihood of stroke reoccurrence. Eductionl progrms for ptients with history of prior stroke should include comprehensive nd integrted informtion concerning stroke risk fctors, stroke reoccurrence rtes, nd cuses of future strokes. 31 In our work, stroke wreness mong rcil/ethnic minorities sent of stroke risk fctors, with stroke risk fctors, nd with prior history of stroke is surprisingly similr. These findings suggest tht current strtegies re not fvored y rcil/ethnic minorities, nd tht current levels of stroke wreness re not cceptle s long s stroke continues to occur t sustntilly higher rte mong minority groups. In ddition to generl eduction progrms for the pulic, progrms should e designed specificlly for rcil/ethnic minorities. To support these efforts, future studies re needed to determine optiml eductionl strtegies nd the most pproprite mediums (TV, rdio, nd Internet) for rcil/ethnic minorities. Rcilly/ethniclly sensitive progrms should include the ptients t risk nd their fmily, friends, nd primry cre physicins. Progrms should e offered through community-sed centers providing cre to lrge numers of minority ptients, churches, nd socil orgniztions s pproprite. 32 Support for eductionl progrms should e sought from community leders nd provided in communitysed venues tht re most recognized nd vlued y rcil/ethnic minorities. Limittions The results of this study should e considered in light of the following potentil limittions. First, telephone surveys my yield ised estimtes ecuse of exclusion of households without telephones. However, studies hve estlished the vlidity of the BRFSS telephone survey. 33,34 Second, importnt fctors tht my ffect knowledge of stroke risk fctors were not mesured in this study, including helth litercy, qulity of stroke eduction, culturl ppropriteness of eductionl mterils, nd ttendnce t eductionl progrms. Third, the Hispnic/other group included individuls who clssified themselves s Hispnic, Asin, Ntive Hwiin/Pcific Islnder, nd Americn Indin/ Alsk Ntive. Additionl studies including dequte smples of these ethnic groups for nlyses would provide dditionl insights into stroke wreness for these groups. Additionlly, inclusion of individuls who do not spek English s primry lnguge should e included in future studies, s lck of litercy decreses the likelihood of enefiting from pulic helth cmpigns to improve stroke wreness. These fctors my provide dditionl explntions for the findings of the study nd should e evluted in future studies. Finlly, the use of close-ended questions my hve influenced some responses reltive to more openended questions. 6 CONCLUSION A high percentge of individuls with prior history of stroke recognize individul stroke wrning signs, ut few recognize ll five stroke wrning signs nd would tke pproprite ction to cll 911. The lowest rtes of recognition of the multiple signs ssocited with stroke nd tking pproprite ction exist mong rcil/ethnic minorities: lck respondents nd Hispnic/other respondents. Trgeted interventions re needed to improve stroke wreness mong rcil/ ethnic minorities. REFERENCES 1. Thom T, Hse N, Rosmond W, Howrd VJ, Rumsfeld J, Mnolio T, et l. Hert disese nd stroke sttistics 2006 updte: report from the Americn Hert Assocition Sttistics Committee nd Stroke Sttistics Sucommittee. Circultion 2006;113:e85-151. 2. Ntionl Institutes of Helth (US). Know stroke. Know the signs. Act in time. NIH Puliction No. 02-4872. 2008 [cited 2008 Jn 30]. Aville from: URL: http://www.ninds.nih.gov/disorders/ stroke/knowstroke.htm 3. Ntionl Institutes of Helth (US). Wht you need to know out stroke. NIH Puliction No. 04-5517. 2008 [cited 2008 Jn 30].

522 Reserch Articles Aville from: URL: http://www.ninds.nih.gov/disorders/stroke/ stroke_needtoknow.htm 4. Greenlund KJ, Neff LJ, Zheng ZJ, Keenn NL, Giles WH, Ayl CA, et l. Low pulic recognition of mjor stroke symptoms. Am J Prev Med 2003;25:315-9. 5. Pncioli AM, Broderick J, Kothri R, Brott T, Tuchfrer A, Miller R, et l. Pulic perception of stroke wrning signs nd knowledge of potentil risk fctors. JAMA 1998;279:1288-92. 6. Rowe AK, Frnkel MR, Snders KA. Stroke wreness mong Georgi dults: epidemiology nd considertions regrding mesurement. South Med J 2001;94:613-8. 7. Kothri R, Suereck L, Juch E, Broderick J, Brott T, Khoury J, et l. Ptients wreness of stroke signs, symptoms, nd risk fctors. Stroke 1997;28:1871-5. 8. Schneider AT, Pncioli AM, Khoury JC, Rdemcher E, Tuchfrer A, Miller R, et l. Trends in community knowledge of the wrning signs nd risk fctors for stroke. JAMA 2003;289:343-6. 9. Alkdry MG, Wilson C, Nicholson D. Stroke wreness mong rurl residents: the cse of West Virgini. Soc Work Helth Cre 2005;42:73-92. 10. Reeves MJ, Hogn JG, Rfferty AP. Knowledge of stroke risk fctors nd wrning signs mong Michign dults. Neurology 2002; 59:1547-52. 11. Ferris A, Roertson RM, Funmi R, Mosc L; Americn Hert Assocition; Americn Stroke Assocition. Americn Hert Assocition nd Americn Stroke Assocition ntionl survey of stroke risk wreness mong women. Circultion 2005;111:1321-6. 12. Bldes LL, Oser CS, Dietrich DW, Okon NJ, Rodriguez DV, Burnett AM, et l. Rurl community knowledge of stroke wrning signs nd risk fctors. Prev Chronic Dis 2005;2:A14. 13. Alkdry MG. An ounce of prevention keeps stroke wy: West Virgini study of wreness of stroke risks nd symptoms. West Virgini Pulic Affirs Reporter 2005;22:2-8. 14. Crroll C, Hort J, Fox C, Tere L, Gison J. Stroke in Devon: knowledge ws good, ut ction ws poor. J Neurol Neurosurg Psychitry 2004;75:567-71. 15. Weltermnn BM, Homnn J, Roglewski A, Brch S, Voss S, Ringelstein EB. Stroke knowledge mong stroke support group memers. Stroke 2000;31:1230-3. 16. Ntionl Stroke Assocition. Recovery fter stroke: recurrent stroke. 2006. 17. Centers for Disese Control nd Prevention (US). The Behviorl Risk Fctor Surveillnce System user s guide. Atlnt: Deprtment of Helth nd Humn Services (US); 2003. 18. Stt Corp. Stt Sttisticl Softwre: Relese 8.0. College Sttion (TX): Stt Corp.; 2004. 19. Yoon SS, Byles J. Perceptions of stroke in the generl pulic nd ptients with stroke: qulittive study. BMJ 2002;324:1065-8. 20. Ellis C, Egede L. Rcil-ethnic differences in stroke wreness mong veterns. Presented t the HSR&D Ntionl Meeting; 2007 Fe 21 23; Arlington, Virgini. 21. Wilson NJ, Kizer KW. The VA helth cre system: n unrecognized ntionl sfety net. Helth Aff (Millwood) 1997;16:200-4. 22. Schroeder EB, Rosmond WD, Morris DL, Evenson KR, Hinn AR. Determinnts of use of emergency medicl services in popultion with stroke symptoms: the second Dely in Accessing Stroke Helthcre (DASH II) study. Stroke 2000;31:2591-6. 23. Morris DL, Rosmond WD, Hinn AR, Gorton RA. Time delys in ccessing stroke cre in the emergency deprtment. Acd Emerg Med 1999;6:218-23. 24. Willims JE, Rosmond WD, Morris DL. Stroke symptom ttriution nd time to emergency deprtment rrivl: the dely in ccessing stroke helthcre study. Acd Emerg Med 2000;7:93-6. 25. Lcy CR, Suh DC, Bueno M, Kostis JB. Dely in presenttion nd evlution for cute stroke: Stroke Time Registry for Outcomes Knowledge nd Epidemiology (S.T.R.O.K.E.). Stroke 2001;32: 63-9. 26. Smith MA, Doliszny KM, Shhr E, McGovern PG, Arnett DK, Luepker RV. Delyed hospitl rrivl for cute stroke: the Minnesot Stroke Survey. Ann Intern Med 1998;129:190-6. 27. Morris DL, Rosmond W, Mdden K, Schultz C, Hmilton S. Prehospitl nd emergency deprtment delys fter cute stroke: the Genentech Stroke Presenttion Survey. Stroke 2000;31:2585-90. 28. Moser DK, Kimle LP, Alerts MJ, Alonzo A, Croft JB, Drcup K, et l. Reducing dely in seeking tretment y ptients with cute coronry syndrome nd stroke: scientific sttement from the Americn Hert Assocition Council on crdiovsculr nursing nd stroke council. Circultion 2006;114:168-82. 29. Tissue plsminogen ctivtor for cute ischemic stroke. The Ntionl Institute of Neurologicl Disorders nd Stroke rt-pa Stroke Study Group. N Engl J Med 1995;333:1581-7. 30. Awreness of stroke wrning signs 17 sttes nd the U.S. Virgin Islnds, 2001. MMWR Mor Mortl Wkly Rep 2004;53(17): 359-62. 31. Cpln LR. Prevention of strokes nd recurrent strokes. J Neurol Neurosurg Psychitry 1998;64:716. 32. Scco RL. Preventing stroke mong lcks: the chllenges continue. JAMA 2003;289:3005-7. 33. Bowlin SJ, Morrill BD, Nfziger AN, Lewis C, Person TA. Reliility nd chnges in vlidity of self-reported crdiovsculr disese risk fctors using dul response: the ehviorl risk fctor survey. J Clin Epidemiol 1996;49:511-7. 34. She S, Stein AD, Lntigu R, Bsch CE. Reliility of the ehviorl risk fctor survey in triethnic popultion. Am J Epidemiol 1991;133:489-500.