Is the Stroke Belt Disappearing?
An Analysis of Racial, Temporal, and Age Effects
Background and Purpose The stroke risk among white residents of the coastal plain of North Carolina, South Carolina, and Georgia (the “Stroke Belt”) has been reported to be between 1.3 and 2.0 times the national average. In this study we examined (1) whether a similar excess risk exists for blacks in this region, (2) whether this regional excess stroke risk has decreased over time, and (3) whether the regional excess risk is consistent across ages from 45 to more than 85 years.
Methods Using data from the Compressed Mortality File, we estimated the annual relative stroke mortality risk for black and white men and women in a region of 153 coastal plain counties and compared these rates to those for the remainder of the United States.
Results The relative geographic excess risk of stroke death was similar for black residents and white residents of the Stroke Belt for both men and women. Despite the decline in stroke mortality, the relative increased risk of stroke death in the region has remained constant from 1968 to 1991; however, the pattern of excess risk across age differed significantly between race/sex groups.
Conclusions These data show that the Stroke Belt continues to exist for blacks and whites and for men and women. Although the specific causes of the Stroke Belt remain unknown, the public health impact is staggering, with a greater than 40% excess risk of stroke mortality and more than 1200 excess stroke deaths annually.
Since the mid-1960s the literature has clearly documented a “Stroke Belt” or clustering of high stroke mortality rates in the southeastern United States. This region is an ill-defined area in the southeastern United States but is generally described as the coastal plain region of North Carolina, South Carolina, and southern Georgia. An early report noted that for white men in the period 1949 to 1951 the national average age-adjusted mortality rate was 70.4/100 000, whereas rates in the Stroke Belt states ranged from 109.1/100 000 to 128.2/100 000 population—1.5 to 1.8 times the national average.1 A recent publication by Lanska2 documented that this elevation of stroke mortality rates in the southeastern region of the United States began at least as early as the first half of this century (1930s) and has persisted to the present period. The Nationwide Mortality Study reported that US stroke mortality rates were heterogeneous between eight communities, three of which were in the coastal areas of North Carolina, South Carolina, and Georgia (NC-SC-GA) and five of which were in the non–Stroke Belt areas.3 For example, the three NC-SC-GA areas were reported to have an age-adjusted stroke mortality ratio 2.2 times greater for men and 1.6 times greater for white women compared with the other five regions. The heterogeneity of stroke mortality rates within North Carolina was described by Heyman and coworkers,4 who found stroke mortality rates for white men to be approximately 1.4 to 1.7 times greater in the coastal and plains (eastern) regions of North Carolina than in the mountain or Piedmont regions. Differences for white women and blacks were less dramatic. A recent report detailing the geographic variation in stroke mortality provided stroke mortality maps by State Economic Areas and reported the continued existence of the NC-SC-GA Stroke Belt in the coastal plain region and the relative increase of stroke along the Mississippi and Ohio rivers.5 As recently as 1993 persistence of relatively high stroke mortality rates in the southeastern United States (including the NC-SC-GA belt) was confirmed.2 6 7 Finally, a review of Medicare records showed an excess of stroke hospitalizations and mortality, but not an excess of case fatality, in the southeastern United States (including NC-SC-GA).8
However, there are several significant shortcomings in the current description of the Stroke Belt. First, because of the sparsity of blacks in many regions of the United States, the reports documenting the existence of the Stroke Belt have focused almost exclusively on geographic variations of stroke mortality among the white population. To our knowledge, only one previous report focusing on statewide stroke mortality rates has documented the excess black mortality in the southeastern United States2 Our study confirms this finding and extends it to more finely demarcated geographic regions, which potentially offers greater resolution. Second, although numerous reports have described the persistence of areas of high stroke mortality in the southeastern United States,2 5 6 7 8 to our knowledge no report has directly described the increased relative risk of stroke death in the southeastern United States and tracked the degree of increase over time. Finally, there have been no efforts to describe potential differences in the magnitude among residents of different ages. These three shortcomings are the focus of this report.
Subjects and Methods
Data in the Compressed Mortality File9 from the National Center for Health Statistics were used as the foundation for this report. This file contains the number of deaths by cause of death and population for each county in the United States for each year between 1968 and 1991, for men and women, and for three ethnic groups (black, white, and other). Stroke deaths were considered as International Classification of Diseases codes 430 to 438.
For the purposes of this study, 153 contiguous counties within approximately 100 miles of the Atlantic Ocean or within approximately 150 miles of the southern border of Georgia were arbitrarily defined as the Stroke Belt (Fig 1⇓). These counties were not chosen by a data-based evaluation of the Compressed Mortality File but rather by an attempt to reflect the “coastal plain” region of the states. While there are no definitive boundaries, the choice of these counties is generally supported by the stroke maps presented by Wing and colleagues.5
For both the Stroke Belt region and for the remaining area of the 50 states and the District of Columbia (the non–Stroke Belt region), annual age-specific stroke mortality rates were calculated for white men, white women, black men, and black women for ages 45 to 54, 55 to 64, 65 to 74, 75 to 84, and 85 years and older. In addition, an age-adjusted (to the 1990 US population) overall annual stroke mortality rate was calculated for the four race/sex groups. The comparison of stroke mortality rates for the Stroke Belt and non–Stroke Belt regions was characterized as the ratio of these rates. These ratios will be referred to as the “relative risk” in the Stroke Belt region. While this is commonly used terminology,10 the reader should note that technically these are “rate ratio” estimates. As the focus of this article is the geographic differences in stroke mortality rates, we define “excess mortality” as the extent the ratio of the stroke mortality rate for a race/sex group in the Stroke Belt relative to the non–Stroke Belt residents of the same race/sex group exceeds 1.0. For example, if for black women the ratio of stroke mortality rates for residents of the Stroke Belt relative to nonresidents was 1.5, this would represent a 50% excess mortality associated with the geographic area. For presentation purposes, the average of the ratios across 4-year intervals was calculated for selected figures.
Population estimates for blacks and whites in the Stroke Belt and non–Stroke Belt regions are shown in the Table⇓ by 10-year age groups. There were more than 500 000 white men and white women and more than 200 000 black men and black women in the Stroke Belt region. The total Stroke Belt black and white population of 1 818 626 persons aged 45 and older was larger than the population of 39 of the 50 states and represented 2.4% of the total US population of 75 524 017 black and white individuals aged 45 and older. This population is of sufficient size to provide stable estimates of stroke mortality rates. For example, there were 4255 stroke deaths in the Stroke Belt region in 1990. The population for the remainder of the United States is based on national populations including the non–Stroke Belt counties of NC-SC-GA, the remaining 47 states, and Washington, DC.
The annual age-adjusted stroke mortality rates between 1968 and 1991 are shown in Fig 2⇓. In all four race/sex groups there has been a dramatic decline in stroke mortality over this period for both the Stroke Belt and the non–Stroke Belt regions. Despite this decline in stroke mortality, the higher stroke mortality remained in the Stroke Belt for all race/sex groups. As has been previously reported, mortality rates were highest in black men, followed by black women, white men, and white women.11 This race/sex ordering was nearly completely consistent in both the Stroke Belt and non–Stroke Belt regions (except for 2 years when white men and black women cross over).
The ratio of stroke mortality in the Stroke Belt compared with the non–Stroke Belt region is shown in Fig 3⇓. The excess mortality in the Stroke Belt was slightly greater in black men (ranging from 1.43 to 1.50 across the 4-year periods) than in white men and black women (ranging from 1.33 to 1.43 and 1.32 to 1.42, respectively) and was lowest in white women (ranging from 1.27 to 1.33). A small decrease was observed in the ratio of stroke mortality rates for white men, but there was no apparent decrease in the ratio of stroke mortality rates for any of the other three race/sex groups.
Fig 4⇓ shows the Stroke Belt to non–Stroke Belt stroke mortality ratio by year and 10-year age group for each of the four race/sex groups. For white women, the excess stroke mortality was reasonably constant both over the years 1968-1971 to 1988-1991 and across the age groups from 45-54 years to age 85 years and older. For white men, there was a slight decrease in the mortality ratio for each of the four younger age groups (45 to 54, 55 to 64, 65 to 74, and 75 to 84 years), but no such decreases were observed for the oldest age group (85 years and older). Dramatic differences were observed in the mortality ratio across age for black men and women. During the period 1968 to 1971, the stroke mortality rate for young black residents of the Stroke Belt was nearly twice that of their non–Stroke Belt counterparts. During this same period the stroke mortality rate in older blacks was similar to that of their non–Stroke Belt counterparts. However, over time the relative risk of the younger blacks has decreased, while the relative risk of the older blacks in the Stroke Belt has increased. The product of the decreasing relative risk of young blacks and increasing relative risk of older blacks resulted in the stable relative risk over time shown in Fig 3⇑. However, the relative risks of blacks within the age groups has changed dramatically and now is more similar to the relatively constant relative risk across ages observed for whites.
Over the past 30 years there has been a dramatic decline in the United States in stroke mortality for blacks and whites and for men and women. Despite this remarkable decline, the increased relative risk of stroke death in the coastal plain of the Southeastern United States appears to have persisted in approximately the same magnitude in the four race/sex groups from 1968 to 1991.
The societal impact of the Stroke Belt is staggering. Based on data from the Compressed Mortality File, there were 4255 stroke deaths in the 153 Stroke Belt counties in 1990 and an age-adjusted stroke mortality rate of 262/100 000. This represents a 43% excess over the age-adjusted stroke death rate of 183/100 000 in the remainder of the United States. If the stroke mortality rate for the remainder of the United States is applied to the 1.8 million residents of the Stroke Belt aged 45 years and older, the expected number of strokes in the region would be approximately 2972 (rather than the 4255 observed). Thus, there are 1283 additional stroke deaths each year in the Stroke Belt. The fiscal impact of this number of deaths above the expected rate is tremendous.
For both men and women, the stroke mortality is notably higher for blacks than whites.11 To our knowledge, the existence of a higher black stroke mortality rate in the southeastern United States has been documented in a single publication,2 which used the states as the unit of analysis (grouping the high- and low-mortality regions within states). The excess mortality in the Stroke Belt region not only exists for blacks but appears marginally larger for both black men and black women relative to their white counterparts. In addition, for both blacks and whites, the relative risk of stroke death in the Stroke Belt was relatively constant over time.
The pattern of excess mortality across age differed by race/sex groups. For white women, the excess mortality in the Stroke Belt was relatively constant across both time and age groups. In white men there was a small decrease in the relative magnitude of the excess stroke mortality observed only in the younger age groups (45 to 54, 55 to 64, and 65 to 74 years). The pattern was different for black men and women. During 1968 to 1971 the excess black stroke mortality in the Stroke Belt was a result of remarkably elevated risk (approximately 2.0 times) among the young black residents (aged 45 to 54, 55 to 64, and 65 to 74 years) of the Stroke Belt compared with their non–Stroke Belt counterparts. In the older age groups (75 to 84 years and 85 years and older) there was little difference in stroke mortality. Over time the relative excess stroke risk in young blacks has decreased, while an increase in relative risk has been observed among the older blacks. This shift has made the pattern of excess Stroke Belt mortality across ages for blacks more similar to that of whites. This may reflect a “cohort” effect or a change in the exposure and risks of the populations.
Despite the existence of the Stroke Belt since at least 1939,2 little information is available on factors underlying this excess mortality, and the specific etiology of these differences remains a mystery. We speculate that the most likely explanation is that geographic variations in risk factor prevalence (eg, hypertension, diabetes, and smoking) cause the Stroke Belt to exist by either increasing incidence or stroke severity (and thus case fatality). Another potential explanation could be the regional differences in coding of death certificates. While it is possible that coding differences may explain the differences in stroke mortality, it is more difficult to assume the sustained differential coding patterns over fixed portions of a contiguous three-state area that would be required to cause this large of an effect. It is also possible that regional differences in stroke case fatality attributable to differences in the quality of the healthcare delivery system underlie the Stroke Belt. However, again it is difficult to believe that differences exist in a large contiguous area without “pockets” of adequate healthcare delivery (particularly in areas of high average socioeconomic status, such as the Southern Pines resort/retirement area in eastern North Carolina or Hilton Head Island near Charleston, SC). The differences are not easily attributed to differences in socioeconomic status, since the Appalachian Mountain region of North Carolina is a nearby region of very low average socioeconomic status with low stroke mortality rates. The differences are not attributable to race or sex, as demonstrated by the fact that similar differences in stroke mortality rate are present across race/sex groups. Therefore, we argue that the most plausible remaining hypothesis for underlying factors is differences in risk factors resulting from socioeconomic disparity, differential behavior, or exposure to environmental factors.
Although many publications (including our own) have described the Stroke Belt, remarkably little data are available to address the potential underlying causes of the NC-SC-GA stroke mortality cluster. Fundamental to the lack of data is the difficulty of collecting community-based characterizations of nonfatal risk factors and comorbid diseases, which are not routinely reported in the United States. The National Health and Nutrition Examination Survey (NHANES)12 is a large population-based study begun to address exactly this shortcoming. In each of three waves (NHANES I, NHANES II, and NHANES III), a two-stage random sample of selected communities in the United States has been chosen and a broad risk factor characterization of the population performed. Selection of 81 counties serves as the first stage of the sampling for NHANES III. Unfortunately, none of these 81 counties are in the Stroke Belt (coastal regions of NC-SC-GA). There have been selected attempts to address risk factor differences that underlie the existence of the Stroke Belt with the use of national data.2 8 Unfortunately, most reports (and most data sets) describe the stroke rates at the state level. However, there are large variations of stroke mortality within states,5 and these attempts have pooled these likely different populations. This problem is furthered by the observation that the larger cities in North Carolina (eg, Charlotte, Greensboro, Raleigh, Winston-Salem) and South Carolina (eg, Columbia) are too far west and the larger cities in Georgia (eg, Atlanta) are too far north to be part of the Stroke Belt region. Hence, even in the southeastern United States, statistics on a state basis are dominated by populations not in the Stroke Belt itself. Other potential sources of information on risk factor data are two large population-based cohort studies (Atherosclerosis Risk in Communities [ARIC]13 and the Cardiovascular Health Study [CHS]14 ), which do have two clinics in the Southeast (Forsyth County, NC, and Jackson, Miss); however, neither of these communities is in the Stroke Belt area. Although some studies are in the process of collecting epidemiological descriptions of risk factors in the Stroke Belt region (eg, the Charleston Heart Study15 ), these lack data from counties with a low incidence of stroke. Hence, no national data and no data from large epidemiologically based studies are available to address risk factor differences that may explain the existence of the Stroke Belt.
In conclusion, the Stroke Belt continues to persist despite the remarkable nationwide decline in stroke mortality. The relative magnitude of the Stroke Belt is similar across race/sex groups, with only slight increases in the relative risk for blacks and for men. The age-specific pattern differs between the race/sex groups and has changed with time. However, in recent years the regional pattern for blacks has become more similar to that of whites. The public health impact of the Stroke Belt is staggering, with more than 1200 excess stroke deaths annually.
- Received October 24, 1994.
- Revision received February 24, 1995.
- Accepted March 2, 1995.
- Copyright © 1995 by American Heart Association
Borhani NO. Changes and geographic distribution of mortality from cerebrovascular disease. Am J Public Health. 1965;55:673-681.
Lanska DJ. Geographic distribution of stroke mortality in the United States: 1939-1941 to 1979-1981. Neurology. 1993;43:1839-1851.
Kuller L, Anderson H, Peterson D, Cassel J, Spiers P, Curry H, Paegel B, Saslaw M, Sisk C, Wilber J, Millward D, Winkelstein W, Lilienfeld A, Seltser R. Nationwide Cerebrovascular Disease Morbidity Study. Stroke. 1970;1:86-99.
Heyman A, Tyroler HA, Cassel JC, O’Fallon WM, Davis L, Muchbaier L. Geographic differences in mortality from stroke in North Carolina, I: analysis of death certificates. Stroke. 1976;7:41-45.
Wing S, Casper M, Davis WB, Pellom A, Riggan W, Tyroler HA. Stroke mortality maps: United States whites aged 35-74 years, 1962-1982. Stroke. 1988;19:1507-1513.
Lanska DJ, Kryscio R. Geographic distribution of hospitalization rates, case fatality, and mortality from stroke in the United States. Neurology. 1994;44:1541-1550.
Public Use Data Tape Documentation. Compressed Mortality File 1968-91. Hyattsville, Md: US Dept of Health and Human Services, Public Health Service, Centers for Disease Control and Prevention, National Center for Health Statistics; 1993.
Kelsey JL, Thompson WD, Evans AS. Methods in Observational Epidemiology. New York, NY: Oxford University Press; 1986.
Howard G, Anderson R, Sorlie P, Andrews V, Backlund E, Burke GL. Ethnic differences in stroke mortality between non-Hispanic whites, Hispanic whites, and blacks: the National Longitudinal Mortality Study. Stroke. 1994;25:2120-2125.
Plan and Operation of the Third National Health and Nutrition Examination Survey, 1988-94. Series 1: Programs and Collections Procedures, No. 32. Hyattsville, Md: US Dept of Health and Human Services, Public Health Service, Centers for Disease Control and Prevention, National Center for Health Statistics; 1994.
ARIC Investigators. The Atherosclerosis Risk in Communities (ARIC) study: design and objectives. Am J Epidemiol. 1989;129:687-702.
Fried LP, Borhani NO, Enright P, Furberg CD, Gardin JM, Kronmal RA, Kuller HL, Manolio TA, Mittelmark MB, Newman A, O’Leary DH, Psaty B, Rautaharju P, Tracy RP, Weiler PG, for the Cardiovascular Health Study Research Group (CHS). The Cardiovascular Health Study: design and rationale. Ann Epidemiol. 1991;1:263-276.
Keil JE, Southern SE, Knapp RG, Gazes PC. Serum cholesterol risk factor for coronary disease mortality in younger and older blacks and whites: the Charleston Heart Study, 1960-1988. Ann Epidemiol. 1992;2:93-99.