Background and Purpose We sought to determine whether the “Stroke Belt” has continued to shift and to assess variation in geographic patterns by age, sex, and race.
Methods Mortality data for Health Service Areas for 1988 to 1992 were used for analyses of geographic mortality patterns for stroke by race, sex, and age (50, 70, and 90 years).
Results In 1988 to 1992, considerable geographic variation in stroke mortality was demonstrated for each sex/race group. In black and white women and men, previously described high mortality in the southeastern United States persisted. Mortality rates were generally higher in the South than in the North and in the East than in the West. Compared with data from 1962 to 1988, there was a continuation of the previously described westward shift of high-rate areas to the Mississippi River valley, a trend more marked at age 50 years than at 70 or 90 years. Although rates in the Pacific region were low overall, a surprising area of high rates was seen in southern California among women at all three ages examined.
Conclusions In whites, rapid declines in stroke mortality in the Southeast have left West South Central states with relatively high mortality rates; this trend may continue as younger cohorts age. However, rates in the Southeast also remain high, especially for blacks.
For at least 50 years, stroke death rates have been higher in the southeast region of the United States than in other US regions.1 2 Recently, the definition and usefulness of the term “Stroke Belt” have been debated as changes in geographic mortality patterns have been noted.3 4 5 Analysis of Medicare hospitalization data indicated that stroke hospitalization rates are also higher in the Southeast than in other regions.6 However, little difference in case-fatality rates was found among regions, suggesting that variation in stroke incidence and/or recurrence might be the most important determinants of variation in mortality.6 An analysis of a national cohort study demonstrated higher stroke incidence in the southeastern United States than in other regions.7 This suggests that intervention to lower stroke risk factors should be targeted to areas with high mortality rates.8 The National Heart, Lung, and Blood Institute recently began a Stroke Belt Initiative for the prevention of stroke mortality in southeastern states with high stroke mortality rates. The failure of stroke mortality to continue its long-term decline between 1992 and 1993 signals the importance of continued detailed monitoring of mortality patterns and trends.9 Although a number of descriptive analyses have described geographic patterns of stroke mortality,1 2 recent innovations in computer mapping techniques and the availability of mortality data for HSAs through 1992 allowed new analyses to more adequately describe geographic mortality patterns by race, sex, and age and to contrast current patterns with those described previously for earlier periods.10 Examination of age-specific patterns may suggest cohort effects and presage patterns in decades to come.
The numbers of deaths during 1988 to 1992 among residents of the 50 US states and the District of Columbia for which stroke was indicated to be the underlying cause of death (ICD-9 categories 430 through 438) were summarized by sex, race (black and white), age, and place of residence.11 12 Age-specific rates were computed by dividing each of these counts by the similarly stratified 1990 census counts, which were multiplied by 5 to correspond to the 5 years of stroke deaths. Directly age-adjusted rates were also computed for each combination of sex, race, and place of residence, with the 1940 total US population used as the standard.
Place of residence, originally reported at the level of county or similar administrative unit, was assigned to one of 805 HSAs. These are defined as aggregations of counties that are relatively self-contained with respect to location of hospital care.13 We examine the effects of region, as well as the local HSA effects, in this report. Each HSA was assigned to the state where it was located or, for 77 of the 805 HSAs that include counties in two states, to the state where the majority of its population lived. Regions were defined as census divisions, except that the largest divisions were subdivided to create 12 regions for whites and 10 regions for blacks, plus Alaska and Hawaii (see “Appendix” for region definitions).
To examine the contribution of local and regional effects on the overall geographic patterns of stroke mortality, we analyzed the underlying data using a mixed effects log-linear cubic spline model for the age-specific rates.14 That is, the logarithms of the age-specific rates were modeled as a function of age, and these regression age effects for the HSAs within a region were assumed to vary randomly around the overall regional age effects. This analytical approach was developed for a project to map death rates for all leading causes of death in the United States.10 15 The age-specific rates for stroke deaths were found to be well explained by the cubic form of the model.
We computed predicted age-specific rates for each HSA and for each region from this model using SAS PROC MIXED.16 The predicted HSA rates were further smoothed by a nonparametric smoothing algorithm, weighted by the inverse of the rates’ standard errors.17 The purpose of this additional step was to bring the picture of broad geographic patterns into better focus by moderating extreme rates in HSAs with few deaths. The resulting smoothed age-specific maps can point to different contributions by age group to the summary age-adjusted rate patterns.
All maps were produced with the use of Atlas GIS software.18 19 The maps were color-coded according to the percentiles of the rate distribution. The age-specific rate categories each represent approximately 20% of the 798 rates, excluding Alaska and Hawaii. The age-adjusted rate categories include, respectively, the lowest 10%, next 10%, 20%, 20%, 20%, 10%, and highest 10% of the 805 HSAs. For blacks, more than 25% of HSAs had no deaths due to stroke. These HSAs were all assigned to the lowest color category on the age-adjusted maps; consequently, the second to lowest category is unused for blacks. Areas on the age-adjusted rate maps that have unreliable rates (defined as having a coefficient of variation of at least 23%) are hatched. These rates have a large standard error because they are based on sparse data, typically fewer than 20 deaths, and therefore should be interpreted with caution. Similarly, regional rates for blacks that are predicted to be near zero (Fig 4⇓) are based on few deaths among sparse populations. These rates may be interpreted more as predicting almost no stroke deaths in these areas rather than as predicting the death rate had the population been larger. Further details of data collection and analytical methods are available in the recently published Atlas of United States Mortality.10
In 1988 to 1992, US age-adjusted stroke death rates per 100 000 for all ages were as follows: black men, 56.6; black women, 43.4; white men, 28.1; white women, 23.8. Statistically significant deviations from the overall US age-adjusted rates were seen for HSAs in most regions of the country for each sex and race. Furthermore, the range of age-adjusted rates was large. For example, in white women the 90th percentile (30.9) was 1.6 times higher than the 10th percentile (19.7) among age- adjusted rates for HSAs; in white men this ratio (38.4/22.7) was 1.7. These ratios could not be meaningfully applied for blacks because of the large number of HSAs with small black populations and unstable rates.
Fig 1⇓ shows the maps of age-adjusted death rates by sex and race. Rates in HSAs along the southern Atlantic coast from North Carolina to Georgia were particularly high for all four groups; another cluster of high-rate HSAs is seen along the lower Mississippi River for white men and women, for black men, and, to a lesser extent, for black women. Scattered areas have higher rates in the Pacific coastal states for both white and black women. Rates are generally low in the Mountain and West North Central states (see “Appendix”).
Fig 2⇓ shows maps of smoothed predicted stroke death rates per 100 000 by HSA among black women and white women for three ages: 50, 70, and 90 years. Corresponding maps for men resembled those shown here. As was noted on the age-adjusted rate maps, rates were relatively high in the Carolinas and Georgia for both white and black women for each age group. The contributions of each age group to the elevated age-adjusted rates along the southern Mississippi River can also be seen for both racial groups. Among white women, the cluster of relatively high rates moves more northward and becomes less concentrated with increasing age. Surprisingly, rates are high in southern California for black women and along the entire Pacific coast for white women.
A similar pattern can be seen for men by examining the predicted regional rate plots (Figs 3⇓ and 4⇓). Rates are highest among black men for the South Atlantic–South region. Among white men, either this region or the East South Central region has the highest rates. For ages 50 and 70 years, white male regional rates are generally higher than the corresponding female rates, but female rates in a number of regions surpassed the male rates for age 90. Fewer regional differences are seen between black men and women. Greater regional differences were seen with increasing age, particularly among women.
In 1988 to 1992, considerable geographic variation in stroke mortality was demonstrated for each sex/race group. In black women and men, previously described high mortality in the southeastern United States persisted. Mortality rates were generally higher in the South than the North and the East than the West. Compared with data from 1962 to 1988,2 there was a continuation of the previously described westward shift of high-rate areas to the Mississippi River valley. Although rates in the Pacific region were low overall, a surprising area of high rates was seen in southern California among women at all three ages examined. This might reflect both migration in the 1940s and 1950s from the Southeast and South Central states and continuing migration, and exposure to local conditions favoring relatively high stroke rates. In 1979 to 1981, stroke death rates for all California residents were higher than among persons born in California, consistent with an adverse effect on migration among blacks and whites.20 Obvious racial contrasts include high rates for the entire state of Florida in blacks compared with low rates in all of Florida except the Panhandle in whites, likely as a result of migration of affluent, elderly whites to southern Florida from northern states.20 In addition, areas of high stroke rates are more densely concentrated in the southeastern states in blacks, with no striking variation by age.
Patterns in white women suggest a Pacific paradox. Stroke and lung cancer rates are relatively high, while ischemic heart disease rates are low in Pacific states.10 21 The pattern was apparent at ages 50, 70, and 90 years among white women but was much less pronounced among white men. A pattern of high stroke mortality but low ischemic heart disease mortality in an affluent country is reminiscent of that seen in Japan. Pacific risk factor and lifestyle patterns as well as the aforementioned adverse effects of migration should be examined in future studies. In white women, the rate of decline in stroke mortality was less in California than in southeastern states in 1979 to 1989 and in 1970 to 1978, leading to a narrowing of the regional difference in mortality over time.22 Preliminary analyses of stroke risk factors by state with the use of data from the Behavioral Risk Factor Surveillance System indicated that California ranked among the top 10 states for history of ever smoking and reported alcohol consumption by white women aged 65 years and over. High alcohol intake might promote stroke while offering protection from ischemic heart disease in older white women, consistent with the observed mortality pattern. For history of hypertension, diabetes, and obesity, California’s rank among states was between 20th and 40th (from highest to lowest prevalence). Migration of affluent, elderly whites to Arizona may help explain low rates in southern Mountain states.20
White women and men show a spread of areas with high stroke mortality from coastal southeastern states to East South Central and West South Central states, including the Mississippi River valley. Compared with data from 1962 to 1988,2 there was a continuation of the previously described westward shift of high-rate areas to the lower Mississippi River valley and now westward into Texas. This pattern was more pronounced in whites aged 50 than 70 and 90 years, which suggests that it will continue as the younger cohorts age. Whether this age effect is due to cohort effects, age-specific exposures, different stroke subtypes and etiology at younger compared with older ages, or other factors is not known. Shifts in the Stroke Belt indicate more rapid declines in areas with the highest mortality in the 1960s to 1980s than in areas in the West South Central region.22 Despite the greater fall of stroke mortality rates in the South Atlantic region, the coastal plain of the Carolinas and Georgia remains in the top decile of mortality among HSAs.
In a recent study in white men aged 45 to 74 years, residence in the Southeast was associated with a 39% increased age-adjusted risk of stroke incidence compared with the Northeast region.7 This excess risk could not be explained by regional differences in multiple stroke risk factors. In white women aged 45 to 74 years, only part of the 59% excess in age-adjusted risk associated with residence in the Southeast compared with the Midwest could be explained by the regional differences in risk factors measured in NHANES I. In blacks, statistically significant regional differences could not be demonstrated, probably because of limited statistical power. However, a strong association of increased stroke risk with residence outside Standard Metropolitan Statistical Areas in blacks was demonstrated that was independent of region or other stroke risk factors. This was the first nationwide study that attempted to explain geographic variation in stroke incidence or mortality with the use of individual-level variables and a prospective cohort study design rather than more limited ecological methods.
A recent analysis of 1989 Medicare data showed high rates of hospitalization for stroke in the Southeast and low rates in Mountain and Northeast states, consistent with the NHEFS data for incidence.6 Incidence is one of the chief determinants of hospitalization rates together with prevalence, comorbidity, and factors related to medical practice and access to care. The lack of variation in case-fatality rates calculated from 15 through 180 days after admission for stroke in Medicare data suggests that variation in stroke mortality is likely related to variation in stroke incidence and/or other factors influencing hospital utilization.6 In whites in 1965, the Nationwide Cerebrovascular Disease Morbidity Study found a higher incidence of hospitalized stroke in three areas with high stroke death rates (Upper Coastal Plain, North Carolina; Pee Dee, South Carolina; Savannah, Ga) than in three low-rate areas (Miami, Fla; Denver, Colo; Flint Hills, Kan).23 Stroke hospital case-fatality rates did not vary consistently among areas. No data on stroke risk factors were available in these studies.
Similarity in the geographic distribution of stroke mortality and of several major stroke risk factors has led to the suggestion that these factors may explain the variation in stroke mortality.1 6 24 25 However, major inconsistencies in the patterns of risk factors and stroke mortality are consistent with NHEFS findings that standard risk factors explain at most only part of the variation in stroke mortality.7 A few analyses have examined the variation in prevalence of hypertension among US regions.26 27 An analysis of NHANES II data (1976 to 1980) found the prevalence of hypertension to be higher in the Southeast (44%) than in all other regions combined (34%) in black women; this difference was not found for other sex/race groups.26 Furthermore, hypertension control was lower for black women in the Southeast than in other regions. In NHANES I, mean systolic blood pressure was slightly higher in the Southeast than in other regions for each sex/race group both for the NHEFS cohort and for the entire NHANES I sample.7 28 An analysis of NHIS data for 1983 to 1987 found higher prevalence of self-reported hypertension in the Southeast region than in other regions in both blacks and whites of both sexes, consistent with earlier reports.27 29
Consistent with the stroke mortality pattern, the prevalence of cigarette smoking was highest in the Southeast and lowest in the Northeast region in white men and highest in the Southeast and lowest in the Midwest in white women in the NHEFS cohort and in previous reports of NHIS and other data.7 27 30 However, this pattern was not seen for blacks. The prevalence of self-reported heart disease was also higher in the Southeast than in other regions among white men and white women.27 Prevalence of self-reported diabetes was reported to be highest in southeastern and East North Central states,31 consistent with mortality findings for white men. The prevalence of diabetes at age 35 to 54 years was the only stroke risk factor consistently highest in Savannah, Ga, intermediate in Hagerstown, Md, and lowest in Pueblo, Colo, duplicating the gradient of stroke mortality rates in the Three-Area Epidemiologic Study.32 Ecological studies have suggested a relationship of low socioeconomic status and high stroke mortality.33 34
It has been suggested that the diet in the Southeast may resemble the diet associated with high stroke mortality in Japan in being lower in animal protein and higher in grain-derived complex carbohydrates and sodium than diets of other US regions.1 Could some similarities to the Japanese diet or lifestyle have been adopted in the Pacific region, contributing to relatively high stroke mortality but low ischemic heart disease mortality? The southeastern diet may also be lower in potassium and calcium, which also have been suggested as nutrients inversely associated with stroke occurrence. These patterns might be especially pronounced in the nonmetropolitan areas of the Southeast and among blacks.
Higher stroke mortality in the Southeast in whites and in blacks could also be related to a lower access to and utilization of preventive health services in these areas that is not fully controlled for by baseline risk factor levels. For example, a longer duration of undetected and/or uncontrolled hypertension in blacks in the Southeast may not be fully reflected in baseline hypertension prevalence or blood pressure differences. Use of preventive services tended to be lower in the Southeast region in national surveys in 1973, 1982, and 1990.35 36 37
A source of possible bias was misclassification due to inaccuracy of diagnoses coded on death certificates, as discussed at length elsewhere.25 38 39 However, the overwhelming body of evidence is against geographic patterns of mortality being purely artifactual.1 2 3 4 5 20 21 22 23 24 25 32 33 34 38 39 40 Also of concern is the migration of healthy elderly whites from the Northeast to Florida and Arizona.20 This also would tend to produce lowering of rates for the Southeast and elevation of rates for the Northeast. Immigration of Hispanic whites to Florida from Cuba and to southern California from Mexico might be expected to lower overall stroke rates for whites. For those younger than 65 years, stroke death rates are similar in Hispanic and non-Hispanic whites; in those older than 65 years, stroke death rates are lower in Hispanics.41 Although diabetes mortality was elevated in areas with high proportions of persons of Hispanic origin among whites in South Texas and New Mexico, it was not elevated in California or Florida.10 Thus, elevated rates in California are not likely due to immigration of Hispanics.
Rapid declines in stroke mortality in the Southeast have left West South Central states with relatively high mortality rates; this trend may continue as younger cohorts age. However, rates remain high in the Southeast, especially among blacks. Continued monitoring of stroke mortality, morbidity, and risk factors within and among regions is needed to explain these ongoing trends. Special population-based registers and surveys as well as analyses of administrative databases will be of value.
Selected Abbreviations and Acronyms
|HSA||=||Health Service Area|
|ICD-9||=||International Classification of Diseases, 9th Revision|
|NHANES I||=||National Health and Nutrition Examination Survey|
|NHEFS||=||NHANES I Epidemiologic Study|
|NHIS||=||National Health Interview Survey|
The following states are included in each region (Fig 5⇓):
(1) New England: Maine, New Hampshire, Vermont, Massachusetts, Connecticut, Rhode Island
(2) Middle Atlantic: New York, Pennsylvania, New Jersey
(3) South Atlantic–North: Delaware, Maryland, Virginia, West Virginia, District of Columbia
(4) South Atlantic–South: North Carolina, South Carolina, Georgia, Florida
(5) East South Central: Kentucky, Tennessee, Mississippi, Alabama
(6) East North Central: Wisconsin, Michigan, Illinois, Indiana, Ohio
(7) West North Central–North: North Dakota, South Dakota, Minnesota
(8) West North Central–South: Nebraska, Iowa, Kansas, Missouri
(9) West South Central: Texas, Oklahoma, Arkansas, Louisiana
(10) Mountain-North: Idaho, Montana, Wyoming
(11) Mountain-South: Nevada, Utah, Colorado, Arizona, New Mexico
(12) Pacific: Washington, Oregon, California
Because of sparse populations, the West North Central and Mountain regions were not further subdivided for blacks, and Alaska and Hawaii data were not modeled.
Reprint requests to Linda W. Pickle, PhD, Office of Research and Methodology, National Center for Health Statistics, Centers for Disease Control and Prevention, 6525 Belcrest Rd, Room 915, Hyattsville, MD 20782.
- Received January 27, 1997.
- Revision received May 21, 1997.
- Accepted May 21, 1997.
- Copyright © 1997 by American Heart Association
Lanska DJ. Geographic distribution of stroke mortality in the United States: 1939-1941 to 1979-1981. Neurology. 1993;43:1839-1851.
Casper ML, Wing S, Anda RF, Knowles M, Pollard RA. The shifting Stroke Belt: changes in geographic pattern of stroke mortality in the United States, 1962 to 1988. Stroke. 1995;26:755-760.
Howard G, Evans GW, Pearce K, Howard VJ, Bell RA, Mayer EJ, Burke GL. Is the Stroke Belt disappearing? Stroke. 1995;26:1153-1158.
Howard G, Howard VJ. The end of the Stroke Belt? It may be too early to declare victory. Stroke. 1995;26:1150-1152.
Lanska DJ, Kuller LH. The geography of stroke mortality in the United States and the concept of a Stroke Belt. Stroke. 1995;26:1145-1149.
Lanska DJ, Kryscio R. Geographic distribution of hospitalization rates, case fatality, and mortality from stroke in the United States. Neurology. 1994;44:1541-1550.
Gillum RF, Ingram DI. The relation between residence in the Southeast region of the United States and stroke incidence: the NHANES I Epidemiologic Follow-up Study. Am J Epidemiol. 1996;144:665-673.
National Heart, Lung, and Blood Institute. Stroke Belt Projects: Heart Memo 1995: Fall. Bethesda, Md: US Dept of Health and Human Services, Public Health Service, National Institutes of Health; 1995:4.
Gillum RF, Sempos CT. The end of the long-term decline in stroke mortality in the United States. Stroke. 1997;28:1527-1529.
Pickle LW, Mungiole M, Jones GK, White AA. Atlas of United States Mortality. Hyattsville, Md: National Center for Health Statistics; 1996.
National Center for Health Statistics. Vital Statistics of the United States, 1990, Vol II: Mortality, Part A. Washington, DC: Public Health Service; 1994.
World Health Organization. Manual of the International Statistical Classification of Diseases, Injuries, and Causes of Death, Based on the Recommendations of the Ninth Revision Conference, 1975. Geneva, Switzerland: World Health Organization; 1977.
Makuc DM, Haglund B, Ingram DD, Kleinman JC, Feldman JJ. Health Service Areas for the United States. National Center for Health Statistics. Vital Health Stat 2 (112); 1991.
Wahba G. Spline Models for Observational Data. Philadelphia, Pa: Society for Industrial and Applied Mathematics; 1990.
Pickle LW, Mungiole M, White AA, Jones GK. Analysis of mapped mortality data by mixed effects models. In: Proceedings of the Biometrics Section of the 1996 American Statistical Association meeting; Chicago, Ill. 1997:227-232.
Latour D, Latour K, Wolfinger RD. Getting Started With PROC MIXED. Cary, NC: SAS Institute Inc; 1994.
Mungiole M, Pickle LW, Simonson KH, White AA. Application of a weighted headbanging algorithm to mortality data maps. In: Proceedings of the Statistical Graphics Section of the 1996 American Statistical Association meeting; Chicago, Ill. 1997:45-49.
Strategic Mapping, Inc. Atlas GIS for Windows. Santa Clara, Calif: Strategic Mapping Inc; 1994.
Use of geographic information systems in epidemiology (GIS-Epi). Epidemiol Bull. 1996;17:1-6.
Lanska DJ, Peterson PM. Effects of interstate migration on the geographic distribution of stroke mortality in the United States. Stroke. 1995;26:554-561.
Lanska DJ, Peterson MS. Geographic variation in the decline of stroke mortality in the United States. Stroke. 1995;26:1159-1165.
Kuller L, Anderson H, Peterson D, Cassel J. Nationwide cerebrovascular disease morbidity study. Stroke. 1970;1:86-99.
Gillum RF. Cerebrovascular disease morbidity in the United States, 1970-1983: age, sex, region, and vascular surgery. Stroke. 1986;17:656-661.
Gillum RF. Stroke in blacks. Stroke. 1988;19:1-9.
Roccella EJ, Lenfant C. Regional and racial differences among stroke victims in the United States. Clin Cardiol. 1989;12(suppl 4):18-22.
Roberts J, Maurer K. Blood pressure levels of persons 6-74 years, United States, 1971-74. National Center for Health Statistics. Vital Health Stat 11 (203); 1977. Washington, DC: US Government Printing Office, Health Resources Administration; US Dept of Health, Education, and Welfare publication HRA 78-1648.
Wilder CS. Prevalence of chronic circulatory conditions, United States, 1972. National Center for Health Statistics. Vital Health Stat 10 (94); 1974.
CDC. Surveillance for selected tobacco-use behaviors: United States, 1990-1994. MMWR CDC Surveill Summ. 1994;43(SS-3):1-43.
CDC. Regional variation in diabetes mellitus prevalence: United States, 1988 and 1989. MMWR CDC Surveill Summ. 1990;39:805-809.
Stolley PD, Kuller LH, Nefzger MD, Tonascia S, Lilienfeld AM, Miller GD, Diamond EL. Three-Area Epidemiological Study of geographical differences in stroke mortality, II: results. Stroke. 1977;8:551-557.
Casper M, Wing S, Strogatz D. Variation in the magnitude of black-white differences in stroke mortality by community occupational structure. J Epidemiol Commun Health. 1991;45:302-306.
Bloom B. Use of selected preventive care procedures, United States, 1982. National Center for Health Statistics. Vital Health Stat 10 (157); 1986.
Moss AJ, Wilder MH. Use of selected medical procedures asso-ciated with preventive care, United States, 1973. National Center for Health Statistics. Vital Health Stat 10 (110); 1977.
Plant A, Schoenborn CA. Health promotion and disease prevention, United States, 1990. National Center for Health Statistics. Vital Health Stat 10 (185); 1993.
Lanska DJ, Peterson MS. Geographic variation in reporting of stroke deaths to underlying or contributing causes in the United States. Stroke.. 1995;26:1999-2003.
Cohen BB, Kleinman JC. Death Rates From Ischemic Heart and Other Related Diseases by Health Service Area, 1968-72: Statistical Notes for Health Planners No. 10. Hyattsville, Md: National Center for Health Statistics; 1979. US Dept of Health, Education, and Welfare publication PHS 80-1237.
Gillum RF, Wilson JB. The burden of stroke and its sequelae. Dis Management Health Outcomes.. 1997;1:84-94.
Gillum RF. Epidemiology of stroke in Hispanic Americans. Stroke. 1995;26:1707-1712.