Out-of-Hospital Cardiac Arrest: RacialDifferences in Outcome in Seattle

Martin R. Cowie, MRCP, Carol E. Fahrenbruch, MSPH, Leonard A. Cobb,MD, and Alfred P. Hallstrom, PhD

IntroductionOut-of-hospital sudden cardiac arrest

has been identified as a key area in whichto study the dual problem of the poorerhealth status of minority populations andtheir poorer access to the health care sys-tem.' These issues not only have troubledpoliticians2 but have also become of in-creasing concem to health professionals.3

The purpose of this report is to de-scribe the racial differences in the out-come ofsudden cardiac arrest occurring inthe area served by the Seattle Fire De-partment's emergency medical servicesdivision, and to analyze these differencesin terms of service-related and demo-graphic factors known to influence resus-citation and survival. We are unaware ofprevious studies addressing this issue.Since 1970, the Seattle emergency medi-cal services program has provided rapidresponse to out-of-hospital medical emer-gencies, including cardiac arrest,4,5 andhas enhanced public awareness of cardio-pulmonary resuscitation (CPR) through itscommunity training program.6

Methods

From routine incident reports com-pleted by paramedics, all cases of out-of-hospital cardiac arrest were identified forwhom the Seattle Fire Department's as-sistance was requested from May 30,1984, through July 31, 1986. The SeattleFire Department is the sole agency re-sponsible for delivering out-of-hospital ad-vanced life support in Seattle, a city ofapproximately 500 000 population. Vic-tims with arrests due to trauma, drug over-dose, or other noncardiac conditions suchas massive hemorrhage were excludedfrom analysis. Only victims aged 20 years

or older were considered. The character-istics of the Seattle emergency medicalservices system have been describedpreviously.4-6

Data extracted from the incident re-ports included the age, race, and gender ofthe victim; the victim's initial cardiacrhythm; treatments provided; whether thearrest was witnessed; and whether CPRwas provided prior to the arrival of theemergency services team. Response timesof the fire department units were obtainedfrom the dispatch center recordings andverified by individual case review for aseparate study.4 For patients who wereinitially resuscitated, hospital recordswere reviewed to determine the patient'sstatus at hospital discharge. Patients whowere not able to live independently wereclassified as having major neurologic dis-ability.

Factors known to influence the out-come of resuscitation6 were selected foranalysis in victims who were in arrestwhen the emergency team arrived at thescene of collapse and who received para-medic treatment. Factors included thepresence of ventricular fibrillation, wit-nessed collapse, prompt (bystander) initi-ation of CPR, emergency team responsetimes (including early application of de-fibrillation), younger age, and location ofcollapse (home vs away from home).

The authors are with the Departments of Med-icine and Biostatistics at the University ofWashington and the Harborview Medical Cen-ter in Seattle, Washington.

Requests for reprints should be sent toLeonard A. Cobb, MD, Division of Cardiology,ZA-35, Harborview Medical Center, 325 NinthAve, Seattle, WA 98104.

This paper was accepted January 12,1993.

Editor's Note. See related editorial byBergner (p 939) in this issue.

American Journal of Public Health 955

.,. % I.."...-"",",.",

r, -10 T.".1 I'll

Cowie et aL

Differences of discrete variableswere tested by Fisher's Exact Test or x2

with Yate's correction; continuous vari-ables were tested by the t test or theMann-Whitney rank sum nonparametricstatistic. Racial differences in outcomevariables stratified by age (20 to 61; 62 andolder) were tested with the Mantel-Haen-szel x2 statistic for combining two-by-twotables. The significance of race in the mul-tivariate model was assessed by stepwiselogistic regression after adjusting for theeffects of covariates known to affect out-

comes. Two-tailed tests of significance arereported.

Results

During the 26 months of this case se-

ries, fire department personnel respondedto 1332 victims (1087 Whites, 137 Blacks,and 108 of other ethnic groups or un-

known race) of out-of-hospital cardiac ar-

rest aged 20 or older and ofpresumed car-

diac etiology. Analysis was limited to theWhites and Blacks (n = 1224) because of

the small number and varied ethnic back-ground of the other victims. The ratio ofWhites to Blacks in these cases with car-diac arrest was 7.9:1, compared with10.8:1 in the Seattle population aged 20and over.8 The annual incidence rates forthese cases of cardiac arrest per 1000 pop-ulation aged 20 and over were 1.6 forWhite and 2.1 for Black victims. How-ever, because the Black population wasyounger, the Black incidence rate was ageadjusted to the age structure of the Whitepopulation for an adjusted rate of 3.4 perthousand (P < .001, Whites vs Blacks).The overall survival rates for Blacks andWhites, respectively, were 10.2% (14/137)and 16.7% (182/1087) (P < .07); the pro-portions discharged without major neuro-logic disability were 8% (11/137) and15.2% (165/1087) (P< .04).

Treated Cardiac AtestsOfthe 1224 patients, resuscitative ef-

forts were ceased for 129 White and 12Black victims because of rigor mortis orassessment by other health care profes-sionals not to resuscitate. Thus, 1083 re-suscitatable patients received advancedlife support from fire department para-medics. Ofthese, 977 patients were pulse-less and unconscious when fire depart-ment personnel arrived on the scene and106 developed cardiac arrest after theemergency team arrived. The ratios ofWhites to Blacks in the latter two groupswere 7.4 and 12.3, respectively (P > .20).Tables 1 to 6 all show data restricted topatients who were found in cardiac arrestwhen first examined.

Table 1 shows the outcomes for thesepatients according to racial group andwithin the two age strata. Although weobserved a significantly lower rate of ini-tial resuscitation within the Black popula-tion (17.1% vs 40.7%, P < .0001), thegroups fared somewhat equally once theywere admitted to the hospital after spon-taneous circulation was restored. Overall,the survival rate was lower in Blacks(9.4% vs 17.1%,P < .03). The proportionwho survived without major neurologicdisability was 15.3% for Whites and 6.8%in Blacks (P < .009).

Analysis by race of variables knownto be predictive of outcome (Table 2)showed that Black victims were youngerthan the White ones (P < .001), which,other factors being equal, would tend tofavor the survival of the Black victims.Neitherthe proportion withwitnessed col-lapse nor the emergency team responsetimes were significantly different, al-though witnessed arrest tended to be less

956 American Journal of Public Health July 1993, Vol. 83, No. 7

Race and Cardiac Arrest

frequent in Black victims. Black victimswere less likely to have had CPR initiatedby a bystander prior to the arrival of thefire department personnel (18.1% vs31.7%, P < .003). The proportion ofBlack victims found in ventricular fibril-lation tended to be lower than that ob-served in White victims (40.2% vs 49.1%,respectively) while asystole was the firstrecorded rhythm in a correspondinglyhigher proportion of Blacks (37.6% vs28.4%). Pulseless electrical rhythm (elec-tromechanical dissociation) was presentin approximately equal proportions ofBlacks and Whites (Table 2).

Following adjustment for known co-variates, logistic regression analyses wereconducted to examine the role of race inrelation to initial resuscitation and sur-vival. Covariates included age, gender, re-sponse time of the first emergency teamunit, response time of the first defibrilla-tor-equipped emergency team unit, re-sponse time of the advanced life supportunit, whether the arrest was witnessed,whether a bystander performed CPR, andinitial cardiac rhythm on emergency teamarrival (ventricular fibrillation vs all oth-ers; asystole vs all others). For the logisticregression analysis, the 13 cases of ven-tricular tachycardia (all in Whites) werecombined with the ventricular fibrillationcases. Of the 977 patients, 92.0% (109Blacks and 790 Whites) had knownvaluesfor all covariates and were used in thesemultivariate analyses (Tables 3 and 4).Race was a significant predictor of bothinitial successful resuscitation and sur-vival without major neurologic disability.As shown in Tables 3 and 4, the odds ratiofor White vs Black race was approxi-mately 4 for resuscitation and more than2.5 for hospital discharge without majorneurologic disability. When the 13 pa-tients with ventricular tachycardia wereexcluded from the logistic analyses, nearlyidentical odds ratios were estimated.

Treated Victim with VenticularFibrilation on Emergency TeamAnival

Table 5 shows outcomes by race,stratified by age, for the 422 White and 47Black victims who had ventricular fibril-lation when first examined. The initial re-suscitation rate was significantly less forBlack victims found in ventricular fibril-lation (31.9% vs 60.4%, P < .0008). Al-though survival to hospital discharge andsurvival without neurologic morbiditytended to be lower in Blacks than inWhites (14.9% vs 26.3% for the latter out-come), these differences were not consid-

ered statistically significant (P > .24 and< .09, respectively).

Similar to the findings in the total se-ries, the Black ventricular fibrillation pa-tients were younger on average than theWhite patients (59.2 vs 66.5 years,P < .002), were slightly less likely to havehad a witnessed arrest, and had compara-ble emergency team response times (Ta-ble 6). Automated extemal defibrillatorswere used by first responders in similarproportions of the two groups of ventric-ular fibrillation patients: 49.5% in Whitesand 44.7% for Blacks. In ventricular fibril-lation victims, bystander-initiated CPRwas provided slightly less often to BlacksthantoWhites(28.3%vs40.1%;P > .15),and Black victims were more likely than

White victims to have collapsed at home(76.6% vs 59.5%, P < .04). (The site ofcollapse was known only for the ventric-ular fibrillation victims.)

Again, logistic regression analyseswere performed to examine whether thevictims' race was independently predic-tive of successful initial resuscitation andsurvival without major neurologic disabil-ity. The covariates in the model were lo-cation of collapse (home vs away), patientage, gender, race, witnessed arrest, re-sponse time of the first arriving unit, re-sponse time of the advanced life supportunit, and the response time of the unitwhose personnel delivered the first shock.Three (White) patients who were notshocked could not be included in the mul-

American Journal of Public Health 957July 1993, Vol. 83, No. 7

Cowie et aL

tivariate analysis. All covariates wereknown in 442 of the 469 patients, (46Blacks and 396 Whites, or 94.2%). In thefirst multivariate analysis, following ad-justment for the known covariates, theodds ratio for resuscitation and admissionto hospital was 3.2 (P < .001, 95% confi-dence interval [CI] = 1.58, 6.44), favoringWhite over Black ventricular fibrillationvictims. In a second multivariate analysis,survival to hospital discharge without neu-rologic morbidity had a nonsignificantodds ratio of2.1 (P < .10, 95% CI = 0.84,

5.18), tending to favor Whites overBlacks.

Victims Who Developed CardiacAn-est under Surveilance ofEmergency Team Personnel

Of the 106 patients who developedcardiac arrest after arrival of the emer-gency services personnel, 98 were Whiteand only 8 were Black. Resuscitationrates in this small group were similar buttended to favor the Blacks (87.5% vs

61.2%, P > .20). The rates of survival

without neurologic disability were 33.7%for Whites and 37.5% in Blacks.

DiwussionIn this case series ofpersons assessed

by emergency service personnel becauseof out-of-hospital cardiac arrest, the age-adjusted annual incidence for Blacks wasdouble that of Whites (3.4 vs 1.6 arrestsper 1000 population aged 20 and over).The Black victims were younger and hadsignificantly poorer initial resuscitationand survival rates. Although our observa-tions could possiblybe attributed in part tosampling variability related to the rela-tively small number of Blacks, we think itunlikely that this represents a major short-coming of the study.

The poorer outcomes might be par-tially explained by the fact that, on arrivalof the emergency team, a slightly lowerproportion of the Black victims (40.2% vs49.1%) were found in ventricular fibrilla-tion, the rhythm that carries amuch higherprobability of survival than asystole orelectromechanical dissociation. How-ever, when considering only those victimsfound in ventricular fibrillation, the initialoutcome of resuscitation was still muchpoorer in Black than in White victims. Inthe subset of patients with ventricular fi-brillation, the racial difference in survivalwas less impressive, but the overall rate ofsurvival without neurologic disability stillfavored Whites, albeit only marginally sofrom a statistical perspective.

Previous work has demonstratedfeatures of cardiac arrest in this commu-nity that were associated with outcome,Sbut the Black-White difference is not ex-plained completely by those variables. Itis therefore likely that additional factorsare at work to reduce the probability ofsuccessful resuscitation. The features oftreating cardiac arrest that relate to out-come have been described in predomi-nantly White populations, and it is pos-sible that the relationships between thesevariables and outcome, particularly anyunderlying associationswith health or be-havior, may not be the same for diverseethnic groups. Additionally, the illnessbehavior or "symptom response" of mi-nority populations may differ from that oftheir cocitizens.9"10 Although the num-bers were small, proportionately moreBlacks were in cardiac arrest when firstexamined compared with those who de-veloped arrest while under paramedicsurveillance. That might suggest hesi-tancy or delay among the Blacks beforerequesting aid. Unfortunately, we do not

July 1993, Vol. 83, No. 7958 American Joumal of Public Health

have adequate data from interviews toestimate either the delays between col-lapse and request for assistance or theprevalence of symptoms prior to cardiacarrest.

We found no racially based differ-ences in emergency team response times,in the proportion of cases who receivedadvanced life support, or in outcomewhen paramedics witnessed the arrests.Hence, there were no indications that ser-vice factors might have been responsiblefor the differences in outcome accordingto race.

Socioeconomic differences amongpopulations or population subgroups areknown to be correlated with healthstatus,11-13 and most minority popula-tions within the United States have alower socioeconomic status than theirWhite counterparts.'4 It may be that thepoorer background level ofhealth and thegreater prevalence of concomitant pa-thology' within the Black victims of sud-den cardiac arrest act to reduce the like-lihood of successful resuscitation. Thereis evidence that Blacks with recognizedcoronary artery disease tend to have par-ticularly severe disease.15 Perhaps thosewho fall prey to sudden cardiac arresthave a more extensive degree of coro-nary artery disease or greater myocardialdysfunction, which thereby makes resus-citation less likely. Additionally, it is pos-sible that a consideration of the underly-ing cardiac disorders-for example,noncoronary cardiomyopathies-mightshed light on the lower rate of initial re-suscitation.

What can be done to improve thissituation? Some improvement might be

realized by increasing the uptake of CPRinstruction within the Black population.Minority populations are less knowledge-able with regard to CPR,6 and the presentstudy has shown that Blackvictims ofout-of-hospital sudden death may be lesslikely to have CPR initiated by a by-stander.

Again, there was no evidence herethat service-related factors contributed tothe racial difference in resuscitation. Wesuggest that the underlying socioeco-nomic and health status of the minoritypopulations are factors likely to affect out-come from cardiac arrest, and that furtherstudies to clarify and ultimately correctthis relationship are appropriate. [l

AcknowledgmentsThisworkwas supported in partby a grantfromthe Medic One-Emergency Medical ServicesFoundation.

The authors are grateful for the earlierwork of W. Douglas Weaver and for the assis-tance of Deborah Hill in the preparation of thedatabase from which these cases were drawn.Special thanks go to Deborah Hodges, who ex-pertly typed the manuscript.

References1. Heckler MM. Report of the Secretary's

Task Force on Black and White MinorityHealth. Vol 1. Executive Summary.Washington, DC: US Govt Printing Of-fice; 1985.

2. Johnson J. Health chief vows minoritiesdrive. New York Times, April 25, 1989.

3. Blendon RJ, Aiken LH, Freeman HE, Co-rey CR. Access to medical care for Blackand White Americans. JAMA4. 1989;261:278-281.

4. Weaver WD, Hill D, Fahrenbruch CE, etal. Use of the automatic extemal defibril-lator in the management of out-of-hospital

Race and Cardac Amet

cardiac arrest. N EngI J MedL 1988;319:661-666.

5. Weaver WD, Cobb LA, Hallstrom AP,Fahrenbruch C, Copass MK, Ray R. Fac-tors influencing survival after out-of-hospi-tal cardiac arrest.JAm CoUCardioL 1986;752-757.

6. Cobb LA, Hallstrom AP. Community-based cardiopulmonary resuscitation: whathave we leamed?Ann NYAcad Sci 1982;382:330-341.

7. Litwin PE, Eisenberg MS, Hallstrom AP,Cuimmins RO. The location ofcollapse andits effect on survival from cardiac arrest.Ann EmergMed 1987;16:97-101.

8. Generalpopulation charactenstis; age byrace, Spanishoringb andser forareasandplaces. Washington, DC: US Bureau oftheCensus; 1980:49-52, table 25.

9. Keil JE, Saunders DE, Lackland DT, et al.Acute myocardial infarction: period prev-alence, case fatality, and comparison ofBlack and White cases in urban and ruralareas of South Carolina.Am HeartJ. 1985;109:776-784.

10. Cooper RS, Simmons B, Castaner A,Prasad R, Franklin C, Ferlinz J. Survivalrates and prehospital delay during myocar-dial infarction among Black persons.AmJCardioL 1986;57:208-211.

11. The Black report. In: Townsend P, David-son N, eds. Inequalities in Health. Lon-don, England: Penguin; 1988.

12. Whitehead M. The health divide. In:Townsend P, Davidson N, eds. Inequali-ties in Healtkh London, England: Penguin;1988.

13. WilkinsonRG. Class andHeakh. London,England: Tavistock Publications; 1986.

14. Health Status of Minorities and Low In-come Groups. Washington, DC: US Deptof Health and Human Services, PublicHealth Services; 1985.

15. Simmons BE, Castaner A, Campo A, Fer-linz J, Mar M, Cooper R. Coronary arterydisease in Blacks of lower socioeconomicstatus: angiographic findings from the CookCounty Hospital Registry. Am Heart J.1988;116:90-97.

July 1993, Vol. 83, No. 7 American Journal of Public Health 959