|
|
| |||||||||||||||||||||||||||||||||||||||
Background Differences between blacks and whites have been reported in the incidence of several forms of cardiovascular disease, including hypertension and stroke. We examined racial differences in the incidence of cardiac arrest in a large urban population and in subsequent survival.
Methods We collected data on all nontraumatic, out-of-hospital cardiac arrests in Chicago from January 1, 1987, through December 31, 1988, and compared the incidence and survival rates for blacks and whites. We examined the association between survival and race and seven other known risk factors by logistic-regression analysis. We computed incidence rates by coupling our data with U.S. Census population data.
Results Our study population comprised 6451 patients: 3207 whites, 2910 blacks, and 334 persons of other races. The incidence of cardiac arrest was significantly higher for blacks than for whites in every age group. The survival rate after cardiac arrest was 2.6 percent in whites, as compared with 0.8 percent in blacks (P<0.001). Blacks were significantly less likely to have a witnessed cardiac arrest, bystander-initiated cardiopulmonary resuscitation, or a "favorable" initial rhythm or to be admitted to the hospital. When they were admitted, blacks were half as likely to survive. The association between race and survival persisted even when other recognized risk factors were taken into account. We did not find important differences between blacks and whites in the response times of the emergency medical services.
Conclusions The black community in our study was at higher risk for cardiac arrest and subsequent death than the white community, even after we controlled for other variables.
There is remarkably little information about racial differences in the risk of cardiac arrest. Gillum concluded that for blacks "data are completely lacking on trends in sudden death"3. Studies from Nashville, New Orleans, and South Carolina all report higher rates of cardiac arrest in blacks than in whites4,5,6. A recent retrospective, multivariate analysis found no significant racial differences9. The study was inconclusive, however, in that it contained relatively small numbers of blacks, provided no information on the population served by the emergency medical services (EMS) system or the incidence of cardiac arrest, and reported an unusually high rate of ventricular fibrillation.
Risk factors for survival after cardiac arrest have been identified, including age, initial cardiac rhythm, whether there were witnesses to the cardiac arrest, whether cardiopulmonary resuscitation (CPR) was initiated by a bystander, time to treatment, location, and, in one study, socioeconomic status10,11,12,13,14,15. However, race has not been identified as an important risk factor. This may be due to the small numbers of nonwhite patients in previous studies.
The CPR Chicago Project was created to gather data on out-of-hospital cardiac arrests in Chicago. We have previously reported a high incidence of cardiac arrest in Chicago and a low overall survival as compared with other studies16,17. This report examines racial differences in the incidence of cardiac arrest and of survival after cardiac arrest and assesses whether these differences persist after previously recognized risk factors are taken into account.
Methods
The EMS System in Chicago
The study was conducted in Chicago, which has approximately 2.7 million inhabitants in an area of 590 km2 (228 mi2). The EMS system in Chicago responds to more than 300,000 calls per year and is activated by dialing 911 on the telephone. It is a single-tiered system with 55 two-person units providing advanced life support 24 hours a day, and it transports all patients with cardiac arrest to 1 of 46 hospitals. The protocols used to treat cardiac arrest follow the recommendations of the American Heart Association.
Data Collection
The methods of data collection have been described elsewhere16. Data were prospectively collected for all victims of nontraumatic cardiac arrest in whom paramedics attempted resuscitation from January 1, 1987, through December 31, 1988. Data from multiple sources (dispatchers, reports and questionnaires completed by paramedics, and hospital records) were combined when there were complete data. Data included age, sex, race, geographic location of arrest, patient's address, whether the arrest was witnessed, pertinent intervals of time, cardiac rhythms, type of therapy administered, whether a bystander initiated CPR, and hospital data (including hospital discharge or death). The research protocol was approved by the EMS system and the review boards of all 46 hospitals.
Our use of terms complies with the recommendations of the Utstein II International Consensus Conference18. "Survival" denotes survival to discharge from the hospital. "Witnessed arrests" are those seen or heard by a witness before the arrival of the paramedics. "Paramedic-witnessed arrests" are those that occurred after the arrival of the paramedics. "Bystander CPR" denotes CPR attempts made by a bystander before the arrival of the paramedics. "Rhythm" refers to the first documented cardiac rhythm after the arrest. The interval from the call to the arrival of the ambulance begins with the time of the 911 call to the fire department and ends with the moment the paramedics radioed their arrival on the scene. The interval from the call to defibrillation begins with the time of the 911 call and ends with the first defibrillation (an interval usually estimated by the paramedics after arrival at the hospital).
Criteria for Inclusion and Exclusion
The policy of the EMS system excluded from attempts at resuscitation only patients who had been decapitated, had rigor mortis, or had signs of physical decomposition. All patients in whom resuscitation was attempted were taken to a hospital; none were pronounced dead at the scene. Patients under 18 years old and those with obvious noncardiac causes of arrest were excluded from the analysis18.
Population Statistics
The population statistics used to calculate incidence rates in Chicago during the study period were estimated by interpolation from the 1980 and 1990 U.S. Census data19,20. Since January 1, 1988, the midpoint of our study, was 7 1/2 years after the 1980 census, weights of 0.25 for 1980 data and 0.75 for 1990 data were chosen for linear interpolation.
Cases and Variables Used in the Analysis
We excluded 499 cases from the analysis because of incomplete data. The analysis was performed for the 6451 cases with complete data and included the following variables: age (18 to 111 years), interval from the 911 call to the arrival of the ambulance (1 to 22 minutes), sex, race (white, black, other, or data missing), rhythm (asystole, idioventricular, ventricular fibrillation or ventricular tachycardia, or other), bystander-initiated CPR (yes, no, or data missing), witnessed arrest (yes, no, or data missing), paramedic-witnessed arrest (yes or no), hospital admission (yes or no), and survival (yes or no).
Statistical Analysis
The statistical analysis included means, frequencies, contingency tables, and t-tests. All tests were two-sided, with a 5 percent level of significance unless otherwise specified. Statistical Analysis System (SAS Institute, Cary, N.C.) and Generalized Linear Interactive Models (GLIM, Royal Statistical Society, London) software were used for the basic and the multivariate analyses.
Age stratification was used in the analysis of incidence. This was necessary owing to differing age distributions between groups in Chicago, where generally blacks are younger than whites and men younger than women. We followed the U.S. Census in stratifying age according to five-year intervals for each race-sex group. Incidence was calculated by dividing the number of cases of cardiac arrest by the estimated population in the subgroup defined according to age, sex, and race. Thus, we estimated the annual rate of cardiac arrest in each subgroup.
To determine the relative risk for blacks as compared with whites, a linear model for the log incidence rate was fitted to the data21. It included terms for each age group, race, and sex. We also studied the interaction of these variables and included an interaction term for race and the median age of each age group. The deviance was used as a measure of goodness of fit.
To determine the relative effect of risk factors on survival, we estimated the association of eight predictor variables and their first-order interactions on survival with use of logistic regression22. The predictor variables included race, sex, age, paramedic-witnessed cardiac arrest, bystander-initiated CPR, bystander-witnessed cardiac arrest, initial cardiac rhythm, and EMS response time. Sex is a commonly used predictor of mortality, and the six variables other than race were included in the model because of their documented effect on the survival rate10,11,12,13. Although some factors proved not to be significant, we kept them all in the model to permit comparison with previously published work. The outcome variables were successful hospital admission and survival to discharge from the hospital. The results of the analyses are presented as odds ratios with 95 percent confidence intervals. The odds multiplier for each variable is the likelihood of survival given the more favorable risk category divided by the likelihood of survival given the less favorable category. To check the fit of the model with the actual data, we tested how well the model predicted the observed outcome for the specific groups determined according to race, sex, and age22. All P values are two-tailed.
Results
The study population comprised 6451 patients, of which 3207 (50 percent) were white, 2910 (45 percent) were black, and 334 (5 percent) were from other races; 3664 (57 percent) were men, and 2787 (43 percent) were women. There were 114 survivors (2 percent), 439 patients who were admitted but died in the hospital (7 percent), and 5898 patients who were pronounced dead in emergency departments (91 percent). Overall, 2972 of the arrests were witnessed (46 percent), 1381 patients received CPR from a bystander (21 percent), and 1394 were noted to have ventricular fibrillation or ventricular tachycardia as the initial cardiac rhythm (22 percent). The mean age was 67.4 years. Women, on average, were five years older than men. The annual incidence of cardiac arrest for persons over the age of 17 was 167 per 100,000.
Incidence
In all age groups, both black men and black women had higher rates of cardiac arrest than their white counterparts (Figure 1). In each racial group, men had higher incidence rates than women. The relative risks indicate that the incidence was higher among blacks than among whites (Figure 2). This difference in relative risk ranged from 239 percent in the youngest age group to 128 percent in the oldest age group.
|
|
The overall survival rate for blacks was 0.8 percent (24 of 2910), and for whites it was 2.6 percent (84 of 3207) (chi-square = 28.3, 1 df; P<0.001) (Table 1). The higher survival rate for whites was similar among men and women. The chances of admission to the hospital were likewise lower for blacks (183 of 2910, or 6.3 percent) than for whites (336 of 3207, or 10.5 percent) (chi-square = 34.5, 1 df; P<0.001) (Table 1). In the subgroup of patients admitted to the hospital, 13 percent of the black patients (24 of 183) survived to discharge, as compared with 25 percent of the white patients (84 of 336) (chi-square = 10.2, 1 df; P = 0.001) (Table 2).
|
|
|
Cardiac arrests were witnessed at a significantly higher rate among whites than among blacks (49 percent vs. 42 percent) (chi-square = 25.7, 1 df; P<0.001) (Table 1). There was no difference between whites and blacks in the rate of paramedic-witnessed cardiac arrest (Table 1). Whites were more likely to receive CPR from a bystander than blacks (25 percent vs. 18 percent) (chi-square = 45.5, 1 df; P<0.001) (Table 1).
Initial Cardiac Rhythm
Rates of ventricular fibrillation or ventricular tachycardia were higher among whites (26 percent) than blacks (17 percent) (chi-square = 70.2, 1 df; P<0.001) (Table 4). Interestingly, most of this difference was due to the presence of a higher percentage of patients with this cardiac rhythm among white men (31 percent), whereas black men, white women, and black women had rates of 19 percent, 19 percent, and 15 percent, respectively. The incidence of asystole was highest among black women (63 percent), followed by black men (61 percent), white women (57 percent), and white men (50 percent).
|
The mean time from the 911 call to the arrival of the ambulance was measured in whole minutes only, and the data were not distributed normally. Although blacks and whites had the same median interval (six minutes), a nonparametric (Mann-Whitney) test showed the distributions to be significantly different because the median interval for whites was shifted slightly toward shorter values (Figure 3). The mean interval was 18 seconds longer for blacks than for whites. No difference was detected in the interval from the 911 call to defibrillation.
|
Logistic-regression analysis confirmed that race was significantly related to the odds of admission to the hospital and of survival to discharge. These results obtained when the other seven predictors and the first-order interactions were included. Within the subgroup of patients who survived to admission, the analysis of survival to discharge revealed similar results.
The results of the logistic-regression analysis are shown as odds ratios with 95 percent confidence intervals (Figure 4). In considering the influence of race as compared with that of the continuous variables of age and the interval to the arrival of the ambulance, we can express the effect of being black in terms of an equivalent change in these variables for a comparable white person. If all the other factors are held constant, it would require a difference of 69 years of age or of 8.3 minutes in the time from the 911 call to the response in order to equalize the odds for blacks and whites. Finally, we examined the fit between our model and the actual data. There was only one standardized residual over 2, indicating that there was no lack of fit for the model.
|
In our analysis of racial differences in cardiac arrest, there were two major findings. First, the incidence of cardiac arrest was significantly higher among blacks in every age group than among whites. Second, the survival rate after an out-of-hospital cardiac arrest among blacks was only 31 percent of that among whites. Even among patients admitted to the hospital, survival for blacks was 52 percent of that for whites. The association between survival and race was as important as the association between survival and age, sex, bystander-initiated CPR, paramedic-witnessed arrest, and the interval from the call to the arrival of the ambulance -- all previously reported as important variables.
These results may be viewed from another perspective. Although we report survival rates, it would be equally valid to report mortality rates. Given our data, the mortality from cardiac arrest for whites would be 97 percent, whereas for blacks it would be 99 percent. This perspective highlights the fact that in Chicago nearly every victim of cardiac arrest dies, and only a slight (2 percent) difference separates whites from blacks. Part of the reason for this low overall survival rate is that we included patients who would not be considered for resuscitation in some other studies of CPR.
The quality of EMS services does not appear to explain the lower survival rates among blacks. In the distribution of intervals from the 911 call to the arrival of the ambulance, our data reveal a small shift toward shorter intervals for whites than for blacks. However, the difference in the mean time interval was only 18 seconds, which could account for only a fraction of the difference in survival. Given the overall low survival rates and long intervals until defibrillation, the speed with which EMS services reach both blacks and whites needs improvement.
The American Heart Association emphasizes that successful treatment for cardiac arrest depends on community-wide emergency cardiac care (a "chain of survival"); responsibility does not lie with the EMS system alone10,23. Efforts to improve survival in the black community begin with a detailed examination of the chain of survival and its links: early access, early CPR, early defibrillation, and early advanced cardiac life support.
The first link, early access, refers to the critical events of witnessing the arrest, calling 911, dispatching the ambulance, driving the ambulance, and reaching the patient's side. There are important intervals we could not measure. For example, the interval from the patient's collapse to the 911 call is currently unmeasured in all studies. It is unclear why blacks had more unwitnessed cardiac arrests. We lack information on the interval from the paramedics' arrival at the scene to the time they reach the patient's side. Consideration of access is important for all patients in a city with low survival rates, and particular efforts may improve access for blacks.
The second link, early CPR, has been shown repeatedly to be associated with improved survival. Our findings reveal a significant difference between blacks and whites in the rate at which bystanders initiate CPR. New initiatives in education about CPR could be effective in increasing the rate of bystander-initiated CPR in blacks.
The third link, early defibrillation, is the most important for patients in cardiac arrest. Our analysis suggests no racial differences. Obtaining accurate data on the interval to defibrillation is, however, a major methodologic weakness of most studies of cardiac arrest, including ours.
The final link in the chain, the early provision of advanced cardiac life support, includes intubation, intravenous access, epinephrine, and advanced life-support maneuvers. Because of the current limitations of data collection, we could not assess the effect of these interventions. The Utstein II guidelines highlight the need for improved methods and for future research in this area.
Our study is primarily descriptive and raises many unanswered questions about the factors responsible for racial differences in survival. Race may be a marker for other coexisting factors, such as environment, genetics, or reduced access to health care. If reliable data on socioeconomic status were available, one might be able to separate the effect of this variable from that of race. Hallstrom et al. have shown a strong effect of socioeconomic status on survival in a predominantly white population15. They used a novel approach: assigning a surrogate value for socioeconomic status that was derived from property-tax assessments. Unfortunately, their data are inconclusive with regard to race because of small numbers. Future studies should also collect the medical history of each patient to permit a more detailed analysis of the effect of risk factors and coexisting illnesses.
We used Utstein II definitions, which exclude obvious noncardiac causes of arrest, such as trauma18. However, there may be less obvious noncardiac causes of arrest that have not been identified, such as occult overdoses or cocaine-related deaths. In general, these conditions are seen more often in younger than in older persons, whereas the reverse is true for the incidence of cardiac arrest. We wonder whether occult disease could be a source of bias. This would provide an alternative explanation for some of the racial differences observed, which were particularly marked in younger persons.
Given the low survival rates observed, it is logical that improving the links in the chain of survival for all victims will result in a substantial improvement in the survival rate for blacks. Since cardiac arrest is often a late manifestation of longstanding cardiovascular disease, we should extend the chain of survival to include comprehensive health care services. This would entail preventing the risk factors associated with cardiac arrest, with a focus on high-risk subgroups. If problems in the chain of survival are particular to the black community, they should be identified. Primary care providers and black patients should be aware that blacks are at significantly higher risk for cardiac arrest and death than are whites. This information may provide additional motivation for reducing cardiovascular risk factors amenable to change (such as hypertension, diet, smoking, and control of diabetes). Failure to intervene earlier with regard to these factors may be the critical weak link in the chain of survival.
Supported by the American Heart Association of Metropolitan Chicago and the Section of Emergency Medicine at the University of Chicago.
We are indebted to the American Heart Association of Metropolitan Chicago (AHA-MC) and the many people who have collaborated to create the CPR Chicago Project, for their vital contributions; to the paramedics of the Bureau of Emergency Medical Services of the Chicago Fire Department, who face the daily reality of treating cardiac arrest and have made this study possible; to the 46 receiving hospitals of the Chicago Fire Department, for help with medical records; to the Chicago Police Department; to the Project Medical Directors' Consortium of the Metropolitan Chicago Health Care Council; to the CPR Chicago Advisory Committee: Morton Arnsdorf, M.D., Bruce Brundage, M.D., Paramedic Officer Patrick Howe, George Kondos, M.D., Max Koenigsberg, M.D., Deputy Fire Commissioner James Joyce, Paul Meier, Ph.D., T. Paul B. O'Donovan, M.D., Paramedic Officer Anthony Scipione, Jr., Paramedic Officer Timothy Stokes, Chief Paramedic Cortez Trotter, and Stanley Zydlo, M.D.; to the staff members of the AHA-MC and the CPR Chicago Project: Linda Alexander, David Bogolub, Clifford Boyce, Robert Cordt, Shirley Courtney, Dwaynette Jones, Nancie Kent, Bette Naysmith, Diana Neubecker, Mary Pat Ostrander, Ronald Ray, and Kathy Redd; to Bruce Spencer, Ph.D., Alfred Hallstrom, Ph.D., Sue Leurgans, Ph.D., Ronald Thisted, Ph.D., and Carin Olson, M.D., for their thoughtful suggestions; to Miriam Goodman, for assistance with artwork; to the members of the CPR Chicago Scientific Review Committee: Joseph Messer, M.D., Eric Louie, M.D., William Ashley, M.D., Patrick Scanlon, M.D., and Richard Cummins, M.D., M.P.H.; and to James Walter, M.D., and the Section of Emergency Medicine at the University of Chicago for their support, encouragement, and ideas.
Source Information
From the Section of Emergency Medicine, Department of Medicine (L.B.B., B.H.H.), and the Department of Statistics (F.A.W.), University of Chicago, Chicago; the Department of Preventive Medicine, Rush-Presbyterian-St. Luke's Medical Center, Chicago (P.M.M.); the Emergency Department, Resurrection Hospital, Chicago (K.V.R.); the Department of Epidemiology and Statistics, University of Oklahoma, Oklahoma City (D.W.S.); and the Bureau of Emergency Medical Services, Chicago Fire Department, Chicago (J.B.).
Address reprint requests to Dr. Becker at MC 5068, 5841 S. Maryland Ave., University of Chicago Hospitals, Chicago, IL 60637.
References
| |||||||||||||||||||||||||||||||||||||||
Related Letters:
Heart Disease and Race
Lai S., Page J. B., Rodney E., Vasavada B. C., Sacchi T. J., Platt F. W., Becker L. B., Meyer P. M., Whittle J., Conigliaro J., Good C.B.
Extract |
Full Text
N Engl J Med 1994;
330:216-218, Jan 20, 1994.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
Previous
