Geographic Differences in Mortality From Stroke in
North Carolina
41
1. Analysis of Death Certificates
ALBERT HEYMAN, M.D, HERMAN A. TYROLER, M.D., JOHN C. CASSEL, M.D., M.P.H.,
W. MICHAEL O'FALLON, PH.D., LINDA DAVIS, B.A., AND LAWRENCE MUHLBAIER, A.B.
Downloaded from http://stroke.ahajournals.org/ by guest on June 16, 2017
SUMMARY Analysis of death certification in North Carolina for
a three-year period, 1969 through 1971, showed regional differences
in mortality rates from stroke in white men, with the highest rates in
the Plains (tobacco growing and farming) area and the lowest rates in
the Mountain region. These geographic differences in death rates
were observed in all but the youngest age decade and also in the
various types of stroke, i.e., hemorrhagic and occlusive cerebrovascular diseases. This regional variation in mortality, however, was
not present in white women or blacks. The prevalence at death of
heart disease, hypertension and diabetes also was higher in the
Plains than in the Mountain region, suggesting that the observed
geographic variation of stroke mortality is related to one or
more of these major risk factors. It is concluded that the
geographic differences in stroke mortality, which had been reported during previous decades, are real and are not due to
variations in death certification, errors in diagnosis, or other
explanations that might artificially produce inaccuracies in
vital statistics.
MORTALITY from cerebrovascular disease in white men
has been found to vary significantly from region to region in
the United States, with the highest death rates appearing in
the South Atlantic and the lowest in the Rocky Mountain
areas. 13 Within the southern states, there are also
differences in the frequency of death from stroke, with
highest rates in the Plains areas of the Carolinas and
Georgia.4'5 The geographic variations in stroke mortality
within these states appear to be greater than those observed
between regional areas of the country.3 The factors responsible for these geographic differences within or between states
are not known, but there is evidence in certain regions of the
country to suggest an association between death from stroke
and the frequency of hypertension.3' "•7 The marked
differences in mortality rates from stroke observed in North
Carolina within a distance of some 200 miles suggest that
there may be within this state regional variations in dietary,
socioeconomic or environmental determinants which
seriously affect the development or outcome of vascular disease.
This study was undertaken to determine whether the
regional differences in stroke mortality rates within North
Carolina, which had been reported during previous decades,
continue to exist and whether they hold for different age
groups and for various types of cerebrovascular disease.
This investigation also attempts to identify possible regional
differences in the risk factors associated with stroke, which
could account for the differences in mortality from this illness. The present report describes our analysis of death certification based on North Carolina and national sources of
vital statistics. Subsequent reports will present the results of
our review of hospital records of deaths occurring in the high
and low areas of stroke mortality within North Carolina.
Methods
In order to verify prior reports of geographic variation in
stroke mortality in North Carolina, detailed analyses were
made of death certificate data provided by the North
Carolina State Board of Health and the National Center for
Health Statistics (NCHS). Thefilesobtained from the State
Board of Health were searched for data regarding deaths
from stroke in the calendar years 1969 through 1971 in white
men age 35 to 69 who were residents of the state. These
deaths consisted only of those in which the underlying cause
was coded by the state nosologists as cerebrovascular disease of any type, i.e., rubrics 430 to 438 in the eighth revision
of the International Classification of Disease (ICD).
Calculations were made of the age-specific and direct ageadjusted mortality rates for stroke for each of the three
years from 1969 through 1971. The appropriate population
tables from the 1960 and 1970 U.S. Census of the 100 counties in the state were used for this purpose. A straight-line interpolation was employed to estimate population size for
determining death rates in 1969 and 1971. Rates were
calculated for each of the four major geographic regions of
the state, i.e., the Mountains, Piedmont, Plains and Coastal
areas (fig. 1), selected according to their geographic
characteristics and to state economic areas as defined by the
Bureau of Census.8 These economic areas consist of subdivisions of the state containing groups of contiguous counties which have similar social and economic characteristics.
In addition to the State Board of Health files in which
stroke was coded as the underlying cause of death, independent data were obtained from the 1969 NCHS Multiple
Cause-Coded death file for North Carolina. (This file was
made available to us by Dr. George Meyers, Director of the
Duke University Center for Demographic Studies.) From
these files, age-specific and age-adjusted mortality rates
were calculated for the selected group of white men
described above, using all 1969 death certificates in which
stroke was listed as being in any way contributory. Mortality rates were calculated for each of the following types of
stroke: (1) cerebral hemorrhage (eighth revision ICD 430.0
to 431.9), (2) ischemic, thrombotic and embolic stroke
From the Center for Cerebrovascular Research, Duke University Medical
Center, Durham, North Carolina 27710; and the Department of
Epidemiology, University of North Carolina School of Public Health, Chapel
Hill, North Carolina 27514.
This study was supported by research grant NS-06233 from the National
Institute of Neurological and Communicative Disorders and Stroke to the
Duke University Medical Center.
STROKE
42
(432.0 to 435.9), and (3) vague ill-defined stroke (436.0 to
438.9).
Rates were calculated also for all entries in the death certificates of ischemic and other types of heart disease,
hypertension and diabetes — four illnesses considered to be
the major precursors of stroke.
Results
Downloaded from http://stroke.ahajournals.org/ by guest on June 16, 2017
Table 1 summarizes the age-adjusted annual mortality
rates for all types of stroke (ICD rubrics 430 to 438) in white
men ages 35 to 69 for the three years from 1969 through
1971 within each of the four defined regions of North
Carolina. These rates are based on data obtained from death
certificate records of the North Carolina State Board of
Health and include only those deaths in which stroke was
coded as the underlying cause. There is a gradient from the
western to the eastern parts of the state, with the lowest rates
in the Mountain area and the highest rates in the Plains.
The four regions of the state were compared in respect to
these rates through the use of a mortality ratio, i.e., the
death rate of each region divided by that of the Mountain
region. If this ratio is significantly greater than one, the mortality rate in the Mountain region is significantly less than
that of the other region. The mortality ratios for Plains versus Mountains ranged from 1.69 in 1971 to 1.75 in 1969. On
the bases of thesefiguresindicating a geographic variation in
stroke mortality, further analyses were made, using data
from the NCHS 1969 Multiple Cause-Coded files to verify
and elucidate this phenomenon.
Table 2 shows the age-specific mortality rates in 1969 for
all strokes (ICD 430 to 438) entered in the Multiple CauseCoded files as either the underlying or secondary cause of
death. Significant differences in the death rates were again
noted among the four regions of the state, with the lowest
rates in the western sections and the highest rates in the
eastern areas. The mortality rate from stroke (considering
all entries) in the Mountain region was 118/100,000; the
rates in the Piedmont, Plains and Coastal regions were 140,
200 and 157/100,000, respectively. The mortality ratio was
again highest for Plains versus Mountain regions and was
calculated to be 1.69 for all entries of stroke. The mortality
ratio for these two areas was slightly higher when stroke was
certified as the underlying cause of death than when this
diagnosis was listed as a secondary cause (1.80 versus 1.52,
respectively). Comparison of these figures with those ob-
VOL. 7, No. 1, JANUARY-FEBRUARY
1976
tained from the North Carolina Board of Health (table 1)
leads to the conclusion that the geographic variation in
stroke mortality is not an artifact caused by regional
differences in the place of entry of stroke diagnoses in death
certificates.
As expected, mortality rates in the entire state were
highest among the older age groups. Among men 65 to 69
years of age, the death rate from stroke entered as underlying and secondary causes was 660/100,000, a figure considerably higher than those of 269, 83 and 21/100,000 noted
in the three younger age decades, respectively. Age-specific
rates were generally lowest in the Mountain region and
highest in the Plains, except for men in the 35 to 44-year age
decade, in whom the number of stroke deaths was too small
to provide reliable rates. It is also possible that deaths in this
younger age group were caused by a particular type of
stroke (e.g., congenital aneurysm or arteriovenous malformation) which would not be influenced by the same regional
determinants as for arteriosclerotic or hypertensive cerebrovascular diseases.
Mortality rates were calculated for the three major types
of stroke, i.e., cerebral hemorrhage (430 to 431), occlusive
disease (432 to 435) and ill-defined disorders (436 to 438),
and are shown in table 3 for each of the four regions of the
state. The regional differences in death rates for each type of
stroke follow the same pattern as those for all strokes combined. The mortality ratio for the Plains versus Mountain
regions for hemorrhagic stroke and for occlusive disease
listed as underlying causes of death was approximately 1.6, a
value lower than that for ill-defined cerebrovascular disorders (2.19). When stroke is entered as a secondary cause of
death, the mortality ratio for each type of stroke is again
highest in the Plains versus Mountain regions than for any
of the other geographic areas.
The regional differences in stroke mortality observed in
white men are not as clearly defined in white women or in
blacks (fig. 2). In each region of the state, the mortality from
stroke among white women was somewhat less than that for
white men, but no significant geographic differences in mortality were noted. Although stroke mortality among blacks
560,
480400-
g^SJ Mountain
I I Piedmont
•
Plains
II Coastal
9. 320Mamtoins
White Males
FIGURE 1 Slate economic areas (I to 11) ofNorth Carolina based
on 1960 report of U.S. Bureau of the Census. Letters A to F indicate standard metropolitan areas. The heavy black lines separate
the four major geographic regions of the state.
White Females
Black Moles
Black Females
FIGURE 2 Age-adjusted mortality rates (per 100,000) for stroke
(ICD 430 to 438) in North Carolina in 1969 according to region of
stale and sex-race groups (all entries). The numbers within each bar
indicate the number of stroke deaths for that region.
GEOGRAPHIC DIFFERENCES IN STROKE MORTALITY///eyma/i et al.
43
TABLE 1 Annual Age-Adjusted Mortality Rates {per 100,000) From Stroke {ICD 430 to 438) Coded as Underlying Cause in While Men {Age 35 to 69) in North Carolina According to Region and Year of Death
Entin; state
Rate
No.
Mountains
Rate
Stroke deaths
Ratio*
632
96.2
85
73.7
1.00
Stroke deaths
Ratio*
602
90.4
74
63.3
1.00
Stroke deaths
Ratio*
611
90.6
'iedmont
Rate
No.
73.1
1.00
85
No.
1969
290
1970
303
1971
281
Plains
No.
Rate
No.
Coastal
Rate
87.7
1.19
193
129.0
1.75f
64
103.9
1.41
90.2
1.43|
164
108.6
1.72f
61
97.9
82.4
1.13
189
123.9
1.69f
56
1.55t
88.6
1.21
*Ratio = rate for each region divided by rate for Mountain area.
fRatio significantly greater than one (P = <0.05).
Downloaded from http://stroke.ahajournals.org/ by guest on June 16, 2017
is considerably higher than among white men or white
women, no regional differences in rates are seen. There are,
however, relatively few blacks living in the Mountain region
of the state.
Table 4 shows the prevalence at death in the various
regions in North Carolina of four illnesses considered to be
important precursors or risk factors for stroke, i.e.,
hypertension, diabetes, ischemic heart disease and other
types of heart disease. The rates for each of these disorders
were generally lowest in the Mountain region and invariably
highest in the Plains area of the state. These differences were
present regardless of whether the deaths were certified as
related in any way to stroke or caused by other illnesses. The
ratios for hypertension and diabetes in the Plains versus
Mountain area among stroke-related deaths were 1.23 and
1.54, respectively. The comparable ratios for ischemic heart
disease and other heart disease were slightly higher, i.e., 1.73
and 2.23, respectively. The fact that these four risk factors
are higher in the Plains than in other sections of the state
suggest that the excess of stroke deaths in the Plains area
may be related to one or more of these illnesses.
Discussion
These analyses of the 1969 to 1971 vital statistics in North
Carolina suggest that mortality from stroke in white men is
different in the four major geographic-economic regions of
the state. In each of these three years, death rates from this
illness were highest in the Plains and lowest in the Mountain
region. A west-to-east gradient for mortality from this disease was also observed, with the death rates in the Piedmont
TABLE 2 Age-Specific and Age-Adjusted Mortality Rales {per 100,000) for Stroke {ICD 4SO to 438) in While
Men in North Carolina in 1969 According to Region of State and Entry of Stroke in Death Certificate
Age
Entire state
Rate
No.
Mountains
Rate
No.
Stroke Coded as Underlying Cause of Death
5
29
12.5
35-44
9
114
53.0
45-54
36
174.7
275
55-64
32
214
410.9
65-69
82
632
96.2
Total
—
Age-adjusted rate
Ratio*
—
Stroke Coded (as Secondary Cause of Death
8.6
4
20
35 - 44
8
64
29.7
45-54
27
94.0
55-64
148
15
249.6
130
65-69
54
362
55.1
Total
—
Age-adjusted rate
—
Ratio*
Piedmont
Rate
No.
14.5
26.9
132.7
317.3
78.0
70.9
1.00
12
50
139
94
295
11.6
23.9
99.5
148.8
51.4
47.6
1.00
10
31
68
58
167
26.2
50.8
232.2
466.1
129.4
118.5
1.00
22
81
207
1.52
462
10.0
45.0
174.6
367.2
87.8
S9.2
1.26
8.4
27.9
85.4
226.6
49.7
50.5
1.06
Plains
No.
Rate
8
43
14.2
84.5
223.2
523.1
123.7
127.7
1.80f
4
12
20
28
64
6
18
41
43
108
10.7
35.4
114.4
374.9
69.9
72.5
1.52f
—
7
12
14
33
14
61
121
103
299
24.9
119.9
337.7
897.9
193.6
200.2
1.69f
4
19
32
42
97
80
60
191
No.
Coastal
Rate
18.3
12
134.7
568.1
104.4
103.8
1.46
—
35.7
80.8
284.0
53.9
53.6
1.13
All Entries of Stroke
35-44
45-54
55-64
65-69
Total
Age-adjusted rate
Ratio*
49
178
423
344
994
21.1
82.7
268.7
660.5
151.3
—
—
9
17
63
47
136
•Ratio = rate for each region divided by rate for Mountain area.
tRatio significantly greater than one (P = <0.05).
18.4
72.8
260.0
593.8
137.4
139.7
1.18
18.3
96.9
215.6
852.1
158.2
157.4
1.33
44
STROKE
VOL. 7, No. 1, JANUARY-FEBRUARY 1976
TABLE 3 Age-Adjusted Mortality Rates (per 100,000) From Stroke in 1969 According to Entry on Death Certificate, Region of State and Mortality Ratio Between Regions (White Men Age 35 to 69)
Entire sitate
Rate
No.
Stroke Entered as Underlying Caust
222
33.8
Hemorrhagic disease
(ICD 430-431)
—
Ratio*
171
Occlusive disease
26.0
(ICD 432-435)
—
Ratio*
239
Ill-defined disease
36.4
(ICD 436-438)
96.2
632
12.0
Coastal
Rate
68
45.1
19
31.0
24
1.00
20.2
78
1.12
23.7
49
1.61
33.0
20
1.11
32.3
27
1.00
22.6
113
1.17
34.2
74
1.63
49.6
25
1.60
40.5
82
1.00
70.9
295
1.51
89.2
191
2.19f
127.7
64
1.79
103.8
1.00
1.26
1.46
i.sot
Downloaded from http://stroke.ahajournals.org/ by guest on June 16, 2017
36
10.8
26
17.2
9
14.6
9.8
41
1.53
12.4
23
2.44
15.5
8
2.06
13.0
35
1.00
30.8
90
1.27
27.3
59
1.59
39.7
16
1.33
26.0
54
1.00
47.6
167
0.89
50.5
108
1.29
72.5
33
0.85
53.6
8
7.1
11
1.00
—
55.1
No.
31.3
—
30.4
Plains
Rate
104
—
12.6
No.
28.0
—
Stroke Entered as Secondary Cause
79
Hemorrhagic disease
(ICD 430-431)
Ratio*
83
Occlusive disease
(ICD 432-435)
Ratio*
200
Ill-defined disease
(ICD 436-438)
Ratio*
362
All stroke
(ICD 430-438)
Ratio*
Piedmont
Rate
No.
31
—
Ratio*
All stroke
(ICD 430-438)
Ratio*
Mountains
Rate
No.
1.00
—
1.06
1.13
1.52f
•Ratio = rate for each region divided by rate for Mountain area.
fRatio significantly greater than one (P = <0.05).
TABLE 4 Age-Adjusted Prevalence Rates (per 100,000) of Risk Factors of Stroke at Time of Death in 1969 According to Entry of Stroke on Death Certificate, Region of State and Ratio Between Regions (White Men Age 35
to 69, All Entries)
No.
Piedmont
Rate
No.
Rate
No.
Coastal
Rate
63
19.0
41
27.3
14
22.7
9.1
34
0.86
10.3
21
1.23
14.1
7
1.03
11.3
1.00
33.1
126
1.13
38.2
85
1.54
57.3
27
1.24
43.7
1.00
11.1
55
1.15
16.6
37
24.9
11
1.32
17.9
Mountains
Rate
No.
Stroke Mentioned on Death Certificate
25
Hypertension
'(ICD 400^04)
Ratio*
11
Diabetes
(ICD 250)
Ratio*
Ischemic heart disease
39
(ICD 410-414)
Ratio*
13
Other heart disease
(ICD 390-398, 420-429)
Ratio*
All Other Deaths
Hypertension
(ICD 400-404)
Ratio*
Diabetes
(ICD 250)
Ratio*
Ischemic heart disease
(ICD 410-414)
Ratio*
Other heart disease
(ICD 390-398, 420-429)
Ratio*
22.0
1.00
Plains
1.73f
2.23
1.49
1.00
1.61
40
34.6
125
37.8
74
49.5
29
47.3
70
1.00
60.8
198
1.09
59.8
144
1.43
95.9
32
1.37
51.8
583
1.00
517.2
1,896
0.98
571.6
998
663.5
401
0.85
651.1
214
1.00
189.9
784
l.llf
236.2
444
l.28f
293.9
148
1.26f
240.4
1.00
•Ratio = rate for each region divided by rate for Mountain area.
fRatio significantly greater than one (P = <0.05).
1.24f
1.58f
1.55|
1.27f
GEOGRAPHIC DIFFERENCES IN STROKE MORTALITY/Heyman
Downloaded from http://stroke.ahajournals.org/ by guest on June 16, 2017
region being slightly higher than those in the Mountains but
lower than those in the Plains and Coastal regions. The findings in the present study confirm previous reports which
showed differences in the years 1950 to 1959* and 1959 to
1961.5 Those reports, however, were based only on entries of
stroke as the underlying cause of death, and it seemed possible that regional differences in certification practices could
account for thesefindings.In the present study, the regional
differences in mortality were present when stroke was certified as either the underlying or secondary cause of death,
and it seems unlikely that the variation in mortality is an artifact related to the place of entry of stroke diagnoses in the
death certificate. Regional differences in mortality were also
present for each age decade except for those in the 35 to 44
age range. As stated earlier, death from stroke in this age
group may be caused by congenital vascular abnormalities
of the brain such as berry aneurysm, rather than by atherosclerosis or hypertensive disease. The fact that these geographic variations exist only for white men, and not for
white women or blacks, also was noted in other areas of the
country by Kuller and his associates.9 This observation
suggests that the geographic determinants of stroke mortality within or between states are sex-race related.
The mortality ratios between the four regional areas of
the state were essentially the same for hemorrhagic stroke,
occlusive vascular disease (thrombosis or embolism), and illdefined types of disease. This finding suggests that hypertension may not be the major factor accounting for the geographic pattern in stroke mortality, since high levels of
blood pressure would be more likely to produce higher rates
of cerebral hemorrhage. In some geographic areas of the
country, mortality from stroke does not correspond to mortality from ischemic heart disease.6 In the Framingham
study,10 however, persons with coronary heart disease (even
in the absence of hypertension) have an excess risk of
cerebral infarction. In such cases, strokes may be due to impaired circulation caused by poor cardiac output, to cerebral
embolism due to mural thrombi and arrhythmias, or simply
to the presence of diffuse atherosclerotic lesions in the
cerebral and coronary arteries.
In addition to these known risk factors, it has been
suggested that the observed geographic patterns in stroke
mortality may be caused by differences in case-fatality rates
et al.
45
from this illness. Studies by Kuller,9 however, have shown
that the case fatality from stroke in one of the high stroke
mortality areas of North Carolina did not differ significantly
from that in low mortality areas elsewhere in the country.
A number of cultural and environmental factors have
been proposed by Sauer11 as contributing to geographic
differences in stroke mortality. Included among them are
personality patterns, smoking habits, occupational and
socioeconomic factors, hardness of drinking water, physical
exercise, diet and intake of salt and trace elements. Studies
of several of these variables in the high and low stroke mortality areas of the country are now being carried out, and it
is hoped that some of the causative factors responsible for
the excess stroke mortality will be identified.
Acknowledgments
We are indebted to Dr. James Palmershein and Mrs. Clara Tessenear for
their assistance in analyzing the death certificate data provided by the North
Carolina State Board of Health.
References
1. Borhani NO: Changes and geographic distribution of mortality from
cerebrovascular disease. Am J Public Health 55: 673-681, 1965
2. Kuller LH, Bolker A, Saslow MS, et al: Nationwide cerebrovascular disease mortality study: I. Methods and analysis of death certificates. Am J
Epidemiol 90: 536-544, 1969
3. Nefzger MD, Acheson RM, Heyman A: Mortality from stroke among
U.S. veterans in Georgia and five western states: I. Study plan and death
rates. J Chron Dis 26: 393-404, 1973
4. Sauer HI, Payne AH, Council CR, et al: Cardiovascular disease patterns
in Georgia and North Carolina. Public Health Rep 81: 455-465, 1966
5. Tyroler HA, Smith HL: IV. Epidemiology and planning for the North
Carolina Regional Medical Program. Am J Public Health 58: 1058-1067,
1968
6. Stallones RA: Epidemiology of cerebrovascular disease — a review. J
Chron Dis 18: 859-872, 1965
7. Acheson RM: Mortality from cerebrovascular accidents in the United
States, Section III. Cerebrovascular disease epidemiology: A workshop.
Public Health Monograph 76, Washington, D.C., U.S. Government
Printing Office, 1966
8. U.S. Bureau of the Census. U.S. Census of Population: 1960. Selected
area reports. State economic areas. Final report PC(3) — 1A.
Washington, D.C., U.S. Government Printing Office, 1963
9. Kuller L, Anderson H, Peterson D, et al: Nationwide cerebrovascular
disease morbidity study. Stroke 1: 86-99 (Mar-Apr) 1970
10. Wolf PA, Kannel WB, McNamara PM, et al: The role of impaired cardiac function in atherothrombotic brain infarction: The Framingham
study. Am J Public Health 63: 52-58 (Jan) 1973
11. Sauer HI: Geographic differences in the risk of premature death.
Business and Government Rev (May-June) 1970
Geographic differences in mortality from stroke in North Caroline. 1. Analysis of death
certificates.
A Heyman, H A Tyroler, J C Cassel, W M O'Fallon, L Davis and L Muhlbaier
Stroke. 1976;7:41-45
doi: 10.1161/01.STR.7.1.41
Downloaded from http://stroke.ahajournals.org/ by guest on June 16, 2017
Stroke is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231
Copyright © 1976 American Heart Association, Inc. All rights reserved.
Print ISSN: 0039-2499. Online ISSN: 1524-4628
The online version of this article, along with updated information and services, is located on the
World Wide Web at:
http://stroke.ahajournals.org/content/7/1/41
Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published in
Stroke can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office.
Once the online version of the published article for which permission is being requested is located, click Request
Permissions in the middle column of the Web page under Services. Further information about this process is
available in the Permissions and Rights Question and Answer document.
Reprints: Information about reprints can be found online at:
http://www.lww.com/reprints
Subscriptions: Information about subscribing to Stroke is online at:
http://stroke.ahajournals.org//subscriptions/