THE LIBRARY OF THE
UNIVERSITY OF
NORTH CAROLINA
AT CHAPEL HILL
THE COLLECTION OF
NORTH CAROLINIANA
C614.1
N8?v5
1979/31
_
UNIVERSITY OF N.C AT CHAPEL HILL
lllillill
00034018704
FOR USE ONLY IN
NORTH CAROLINA COLLECTION
I
Digitized by the Internet Archive
in 2009 with funding from
Ensuring Democracy through Digital Access (NC-LSTA)
http://www.archive.org/details/northcarolinavit197902nort
LEADING CAUSES
OF MORTALITY
North Carolina
Vital Statistics
1981—Volume 2
LEADING CAUSES
OF MORTALITY
North Carolina
Vital Statistics
1981—Volume 2
A A
State Center For Health Statistics
EC. Department of Human Resources
DiTiason of Health Semes
STATE OF NORTH CAROLINA
James B. Hunt, Jr., Governor
DEPARTMENT OF HUMAN RESOURCES
Sarah T. Morrow, M.D., M.P.H., Secretary
DIVISION OF HEALTH SERVICES
Ronald H. Levine, M.D., M.P.H.
State Health Director
Margaret Woodcock, M.A., M.P.A.
Asst. Director for Management Services
STATE CENTER FOR HEALTH STATISTICS
Charles J. Rothwell, M.B.A., M.S., Director
September 1983
TABLE OF CONTENTS
PREFACE
I. INTRODUCTION
A. Purpose and Organization
B. Overview of Mortality in North Carolina
II. TECHNICAL NOTES
A. Changes in the Cause-of-Death Classification System . .
B. Changes in Population Bases Due to the 1980 Census . .
C. Mortality Rates
Computation of Mortality Rates
Interpretation of Mortality Rates
Flagging Biased Rates
D. Procedure for Geographic Clustering of Counties ....
III. NORTH CAROLINA'S POPULATION AND HEALTH CARE RESOURCES ... 3-
IV. GENERAL MORTALITY IN NORTH CAROLINA 4-
V. MAJOR CARDIOVASCULAR DISEASE MORTALITY 5-
A. Heart Disease
Acute Myocardial Infarction
Other Ischemic Heart Disease
B. Hypertension With or Without Renal Disease
C. Cerebrovascular Disease
D. Atherosclerosis
VI. CANCER MORTALITY 6-1
A. Total Cancer 6-3
B. Cancer of the Stomach 6-23
C. Cancer of the Colon, Rectum, and Anus 6-31
D. Cancer of the Pancreas 6-41
E. Cancer of the Trachea, Bronchus, and Lung 6-49
F. Cancer of the Female Breast 6-59
G. Cancer of the Cervix Uteri 6-67
H. Cancer of the Ovary and Other Uterine Adnexa 6-75
I. Cancer of the Prostate 6-83
J. Leukemia 6-91
TABLE OF CONTENTS (CONTINUED)
Page
VII. OTHER SELECTED CAUSES OF MORTALITY 7-1
A. Diabetes Mellitus 7-3
B. Influenza and Pneumonia 7-9
C. Chronic Obstructive Pulmonary Disease 7-19
D. Chronic Liver Disease and Cirrhosis 7-29
E. Nephritis, Nephrotic Syndrome, and Nephrosis 7-37
VIII. MAJOR EXTERNAL CAUSES OF MORTALITY 8-1
A. Accidents 8-3
Motor Vehicle Accidents 8-5
Accidents Excluding Motor Vehicles 8-15
B. Suicide 8-25
C. Homicide 8-35
IX. INFANT MORTALITY 9-1
X. MULTIPLE CONDITIONS PRESENT AT DEATH 10-1
lv
CD
O•—
•
UJ
Q£
11 O
ex
0^
(X £
0^
CD
to
PREFACE
The State Center for Health Statistics has produced triennially a major
publication describing North Carolina's mortality experience for a five-year
period. Such a publication was due for 1975-79 data, but the change in 1979
from the Eighth to the Ninth Revision of the International Classification of
Diseases (ICD) made it undesirable to combine pre-1979 cause-of-death data
with data for 1979 and later years. Changes in the cause-of-death
classification were described in detail in the abbreviated 1979 edition
Leading Causes of Mortality . The standard format is resumed in this 1981
_
edition, with three-year rates (1979-81) replacing five-year rates. The delay
in publishing this volume has been largely due to a delay in obtaining the
1981 population projections based on the 1980 census, which are used in the
denominators of the death rates.
This edition of Leading Causes of Mortality is the third major
publication of its type on North Carolina's mortality experience. This volume
includes statistical tables, maps, and graphs, as well as cause-specific
discussions of trends, geographic patterns, risk factors, and recent research.
An overview of the mortality experience in North Carolina is also presented.
The tables in this report provide selected 1981 mortality statistics for
counties, Department of Human Resources regions. Health Service Areas, and the
state (see maps on preceding pages). More than a dozen of North Carolina's
leading causes of mortality are depicted in these tables; in addition, various
cancer sites and total infant mortality are included.
Similar to the last major publication, multiple conditions present at
death are included in this volume. This section highlights the 1982
publication Multiple Conditions Present At Death: A Special Study and presents
the data for 1981 deaths.
A separate technical notes section has been included in this publication.
This section is necessary not only to discuss appropriate interpretation (s) of
mortality rates and changes in both the ICD codes and population bases, but
also to introduce two new statistical procedures used in this volume—Flagging
Biased Rates and Geographic Clustering of Counties. Both procedures are
discussed in detail in Section II and where relevant in the other sections.
Special thanks go to Mr. Shannon Hallman of the Health Services Research
Center, University of North Carolina at Chapel Hill for preparing Section
III—North Carolina's Population and Health Care Resources. For the first
time data have been included on nurse practitioners and physician assistants
and on skilled nursing and intermediate care beds.
If there are any questions concerning this publication, please contact:
Delton Atkinson
State Center for Health Statistics
Division of Health Services
P. 0. Box 2091
Raleigh, North Carolina 27602
(919) 733-4728
I. Introduction
PURPOSE AND ORGANIZATION
Although North Carolina has traditionally experienced low death rates,
our status among states has worsened. In 1960, only seven of the 50 states
had death rates lower than North Carolina, and this increased to 14 in 1970
(1). In 1981, 20 states had lower rates and three had the same rate (2).
Furthermore, while many of the state's cause-specific rates are still below
United States levels, North Carolina's age-adjusted rates are usually higher.
For example, cerebrovascular disease, motor vehicle accidents, all other
accidents and adverse effects, pneumonia and influenza, and nephritis,
nephrotic syndrome, and nephrosis are causes for which North Carolina's 1981
adjusted rates were at least 10 percent above the corresponding United States
rates.
These facts represent a serious challenge to health officials throughout
the state to (i) identify high-risk areas in order to strategically apply
medical and other health services, and (ii) isolate those determinants of
mortality that permit intervention. The data in this volume are intended to
aid in these areas of investigation by providing information against which
North Carolina's county and regional mortality may be assessed in the present
and evaluated in the future.
Ten sections comprise this edition of Leading Causes of Mortality . This
section presents an overview of mortality experience in North Carolina as well
as the purpose and organization of the document. Section II provides
information concerning the calculation, interpretation, and appropriate use of
adjusted and unadjusted rates. In past editions of this volume, readers were
1-3
cautioned about using adjusted rates with a small number of deaths in the
numerator. For this and subsequent editions, problem rates have been
"flagged" in the tables with an asterisk. The procedures used to determine
which rates to flag are described in detail in Section II. Also, a new
procedure used to depict geographic clustering of counties is discussed here.
In Sections III-X, maps and tabular data describe North Carolina's recent
experience with respect to population and selected health care resources,
general mortality, cause-specific mortality (underlying causes), infant
mortality, and multiple conditions present at death. Although the 1976
edition of this volume reported that some 3-year rates may have substantial
random fluctuations, changes in coding of death certificates in 1979
necessitate the use of these rather than 5-year rates. The 1982 edition of
Leading Causes of Mortality will show 4-year rates, and the 1983 and
subsequent editions will again show 5-year rates.
Table 1 on page 1-8 describes the selected cause-of-death categories in
terms of codes from the ninth revision of the International
Classification of Diseases (3). Altogether, the causes selected for
examination in this report accounted for 87 percent of all North Carolina
deaths during 1981.
Description Of Maps
This volume contains 81 computer-produced maps which depict data for the
state's 100 counties. The maps on pages v and vi identify the four regions of
the Department of Human Resources (DHR) and the state's six Health Service
Areas (HSA's)
.
In Section III, five maps portray each county's relative status among the
100 counties with respect to population per primary care physician; population
1-4
per primary care physician, physician assistant, and nurse practitioner;
population per registered nurse; population per short-term general hospital
bed; and population 75+ per skilled nursing and intermediate care beds. These
data together with information concerning county populations and county
mortality levels should aid in the determination of areas most in need of
increased health resources.
In Sections IV-X, the mgps depicting geographical patterns in mortality
should aid in the resolution of specific types of health promotion and health
care needs. Two to four maps are generally shown for each cause-of-death
category. Two maps depict the 1979-81 unadjusted death rate and the 1979-81
age-race-sex-adjusted death rate for each of the 100 counties. These two maps
show six levels of death rates, where level one is the lowest rate interval
and level six the highest . These maps should be viewed with extreme caution
for causes where the number of deaths per county is small, since in these
cases the rates may be very unstable. Users of the maps should note that the
interval values indicated by the map legends are not necessarily continuous
but reflect the actual range of values for each interval.
A clustering routine from the Statistical Analysis System (SAS) was used
to group counties that are "most like each other" with respect to their
unadjusted rate and then their adjusted rate (4). This procedure may result
in very large or very small groups depending upon how counties differ from one
another. The other two maps for a particular cause may depict "spatial"
clustering of two or more adjacent counties with high unadjusted and/or
adjusted rates using a new procedure described in Section II. These maps are
presented for a given cause only when a statistically significant spatial
cluster is found to exist. Unlike the two full-page maps, these maps are
presented in reduced form as part of the narrative and show one to three
1-5
interval levels of death rates, depending on the number of significant spatial
clusters found, along with the significance level (i.e., p-value)
.
Description Of Tables
Sections IV-X contain tables that summarize the recent mortality
experience of counties, six Health Service Areas, four Department of Human
Resources regions, and the state. Except in the case of infant deaths
(Section IX) , a table corresponding to each cause-of-death category includes
the following items of information:
(1) the number of resident deaths occurring during 1981;
(2) the 1981 death rate;
(3) the number of resident deaths occurring during 1979-81;
(4) the 1979-81 average annual death rate;
(5) the 1979-81 average annual age-race-sex-adjusted death rate computed
by the direct method (5).
The formulas for calculating single- and multi-year rates are found in
Section II. In this report, general mortality rates (all causes combined) are
expressed as deaths per 1,000 population while cause-specific rates are
expressed as deaths per 100,000 population. The infant death rates of Table
28, Section IX, are computed as the number of infant deaths per 1,000 live
births.
Multiple Conditions Present At Death
Since 1975, multiple conditions present at death have been coded such
that statistics are available for all conditions present at death as reported
by the certifier. Three diseases in particular—hypertension,
atherosclerosis, and diabetes—are considered associated conditions far more
often than they are considered an underlying cause; hence, no maps are
1-6
displayed in Sections V and VII showing these deaths as an underlying cause.
For the geographic clustering routine maps, only the 1979-81 age-race-sex-adjusted
county rates for these conditions are displayed in Section X. Two of
the three clustering maps displaying unadjusted rates showed patterns similar
to those for adjusted rates, and thus are not presented, while the third
(atherosclerosis) was not statistically significant.
Tables in Section X show deaths crosstabulated by underlying cause and
associated mentioned conditions, death frequencies for selected pairs of
mentioned conditions, and death rates based on mentioned conditions rather
than underlying cause. Rates are deaths with the condition mentioned per
100,000 population.
1-7
OVERVIEW OF MORTALITY IN NORTH CAROLINA
Mortality in North Carolina has exhibited a general downward trend in
this century. Numerous factors have contributed to this decline including
changes in lifestyles, environment, age-race-sex distribution, risk factors,
and the medical care system. In Sections IV through X of this report, these
factors and others are considered in discussing each cause of death. This
overview examines four general determinants of mortality, highlights some of
the major mortality findings, and describes some of the risk factors which
impact on a number of different causes. Further, this section looks at
premature mortality in North Carolina using a concept called "years of life
lost" which emphasizes mortality in the younger age groups (6).
Determinants of Mortality
A broad view of mortality determinants recognizes that medical care is
only one aspect of health maintenance and that many health problems "arise
from causes embedded in the social fabric of the nation as a whole" (7)
.
Accordingly, environment, lifestyle, biology, and medical care must all be
considered as determinants of health.
The natural and manmade physical environments are commonly recognized as
having an important impact on health. For example, water mineral content and
elevation have been cited as influencing the incidence of cardiovascular
disease (8) . Of particular concern recently have been the health consequences
of manmade environmental pollution. The proliferation of various chemical
compounds is a central element of a high-technology, growth-oriented society.
Ill health due to environmental pollution therefore can result from the
1-9
economic environment. Stricter governmental controls may be required in this
area for those businesses that will not voluntarily weigh social costs into
their economic calculations.
Yet economic growth bears on health in ways other than the negative
externalities of environmental pollution. Economic growth means jobs and
income, both of which are central to individual and family well-being.
Unemployment, economic poverty, and their social concomitants are inimical to
mental and physical health. The poor, controlling a lesser share of economic
and social resources, are subject to greater physical vulnerability to
infectious agents, more economic uncertainty and associated stress, and more
hazardous environments in homes and workplaces. They have fewer cushioning
devices such as vacation or travel (9) . Thus controlled growth can have
substantial health benefits if it is broadly based and environmentally sound.
Lifestyle refers to those decisions by individuals that affect their
health and over which they have some control. The degree of individual
control over one's own decisions is a key question to answer in an evaluation
of changing lifestyles as a health improvement strategy. Individuals'
decisions are conditioned to a large extent by their position in the economic
and social hierarchy. Self-destructive and violent behavior, for example, may
reflect alienation due to a marginal position in the society. But policies to
educate individuals about their health are much less complex and easier to
sell politically than those aimed at modifying basic social and economic
determinants of lifestyle and health. "Blaming the victim" by keeping the
problem at the individual level may obscure the origins of disease in the
socioeconomic environment.
This is not meant to suggest that health education does not have its
place. Certain population groups are more likely to engage in lifestyles that
1-10
increase mortality, and education programs that target these groups are an
essential, though short-run, complement to policies oriented toward the
environmental factors that condition lifestyle. Education about nutrition is
likely to have a high health payoff among the poor and less educated, but only
if they have enough money to buy proper food and the facilities to properly
prepare it.
A 1977 Rhode Island telephone survey indicated that the 30-39 year age
group could be targeted as high-risk, based on health perception, weight,
smoking, drinking, sleep habits, and stress. Education efforts aimed at
relieving stress, which is much less publicized than other lifestyle factors
and may affect smoking, drinking, and other behaviors, may be more productive.
As shown in a recent North Carolina study (10) , income, education, and urban
or rural residence may be important indicators and determinants of lifestyle,
and successful education programs must consider variation along such
dimensions. Targeting specific groups is likely to be more successful than
generalized education or media campaigns.
Biological factors are very pervasive determinants of mortality. The
age, race, and sex of an individual are biologically determined and mortality
rates vary consistently along these dimensions. Health is strongly tied to
aging and the life cycle, some diseases that vary by race are thought to be
genetically linked, and biological factors account in part for males and
females having a different incidence of many diseases, with females living
longer on the average. Some health consequences of age, race, and sex are,
however, not biological in origin. Social stratification is in large part
based on age, race, and sex differences, with the old, the nonwhite, and the
female generally being lower on the social and economic hierarchy. Male
mortality is higher partially as a result of aggressive, achievement-oriented
1-11
behavior that accompanies higher status positions (11), while higher nonwhite
mortality is due in part to a lower position in the economic hierarchy (12).
Many causes of disease are biological in origin. In North Carolina an
estimated 150,000 persons are suffering from serious genetic disorders as
evidenced by physical defects, mental retardation, and other health problems
(13) . Forty percent of all children admitted to inpatient pediatric care are
there because of genetic disorders (13). About one-half of all visual
impairment difficulties are caused by genetic factors. Birth defects, of
which about 80 percent are thought to be genetic in origin, are a leading
cause of infant mortality. Some persons have a greater biological
susceptibility to disease. Cancer, for example, may result from problems of
the immunological system, in combination with the presence of external agents.
Many parts of the medical care system function by reacting to health
problems caused elsewhere, though in attempting to restore individuals to a
full and productive life, this system becomes an important determinant of
mortality levels. Disease prevention is also within the purview of the
medical care system, as through innoculation against infectious diseases or
patient education concerning the negative health consequences of certain
lifestyles. Medical care personnel may also be involved in attacking certain
environmental and biological causes of disease, though this type of activity
has traditionally been carried out in the government or public sector.
McKeown and Brown (14) present evidence suggesting that medical practice
in the first half of the 19th century had very little to do with the large
decline in mortality that took place in Western societies, but rather that
transportation improvements, changes in the economic system that assured a
more continuous and nutritious food supply, and improved sanitation practices
urged on by reformers in the cities were responsible. After the practice of
1-12
antisepsis became widespread late in the 19th century, medical care became a
much more positive factor in reducing mortality. During the first half of
this century, the health and average life span of Americans improved
considerably, due substantially to efforts in the medical sector to reduce
infections and acute nutritional diseases. Major gains were observed in
infant and maternal mortality.
Medical care, however, may sometimes have negative health consequences.
It has been estimated that infections acquired inside the hospital strike 1.5
million of the 33 million Americans hospitalized each year, adding to hospital
costs by increasing lengths of stay and causing an additional 15,000 patient
deaths (15). Inappropriate or unnecessary treatment may also increase
mortality as well as health care costs. Risks, however, are always present
even in proper medical treatment, and in most cases they are far outweighed by
the potential benefits.
In summary , a complete program to improve health status and reduce
mortality must include environmental, lifestyle, biological, and medical care
strategies. Too much emphasis in one area may, however, involve substantial
opportunity costs due to neglect of other areas. For example, expenditures
for genetic research, for environmental protection, or to improve substandard
housing and below-poverty-level incomes could potentially have higher long run
health payoffs than would the same amount expended just for medical care. The
rise of heart disease and cancer as major killers is closely linked to changes
in the economic environment, and medical treatment or attempts to modify
individual behaviors are not by themselves likely to control these diseases.
Sedentary occupations, increasing incomes and therefore food consumption
(particularly of foods high in animal fat) , and drinking and smoking
associated with stresses placed on the family and other primary support
1-13
institutions in a highly mobile society all underlie the rise in heart
disease. The spiral of post-WWII economic production and increasing use of
automobiles have led to more and more carcinogens in the air, water, and food.
Cancer control is such a difficult policy area precisely because cancer
prevention will require fundamental changes in the physical and therefore
economic environment. Policies aimed at improving the medical treatment of
cancer victims are probably easier to implement, but they do not deal with the
basic problem. Successful strategies to deal with these and other leading
causes of mortality must modify the basic causes of our modern ills.
1-14
Risk Factors
Risk factors particular to each cause of death are discussed in separate
sections of this volume, but there are some factors common to a number of
different causes of death and these factors are summarized here.
Two of the most pervasive factors contributing to U.S. mortality from
various causes are high blood pressure and cigarette smoking . Higher levels
of blood pressure have been shown to be associated with death from all
cardiovascular diseases, diabetes mellitus, and cirrhosis of the liver
(17,18). Most cases of high blood pressure are amendable to treatment;
however many people either are not aware that they have high blood pressure or
will not maintain the proper weight, diet, and medication regimen to control
it.
Cigarette smoking contributes to death from a large number of causes (17-
22) . A recent report by the U.S. Surgeon General (22) concludes that smoking
is a major cause of lung cancer and of cancers of the larynx, oral cavity, and
esophagus; that it is a contributory factor in the development of cancers of
the bladder, pancreas, and kidney; and that approximately 30 percent of all
cancer deaths are attributable to cigarette smoking. It has also been
reported as a contributor in the development of chronic bronchitis and
emphysema, pulmonary heart disease, myocardial infarction, aortic aneursym,
and a wide variety of other vascular diseases. Smoking has important
interactions with other factors. "It is now clear that smoking may act
synergistically with ionizing radiation or asbestos to produce cancer of the
upper respiratory and digestive tracts, with oral contraceptives to produce
myocardial infarction, and probably with several dietary factors to produce
both cancer and vascular disease". It may be that the low tar, low nicotine
cigarette is less liable to cause lung cancer and chronic bronchitis, but no
1-15
less liable to produce vascular disease because of its carbon monoxide
content. A substantial reduction in smoking -related disease and death is
likely to come about only if a substantial number of smokers quit smoking.
(19)
Diet also has an important impact or certain causes of mortality. Over-eating
may lead to obesity , which contributes to high blood pressure,
diabetes, and cardiovascular disease. Diabetes is itself a risk factor for
stroke and other cardiovascular diseases. In addition, the content of our
moderr diet has important consequences for mortality. "Compared with our
ancestors' diet, that consumed by present-day western populations is higher in
intake of energy, of protein (especially animal protein), and of fat
(especially animal fat), but lower in intake of fibre-containing cereal foods;
this diet is associated with high rates of morbidity and mortality from
degenerative diseases" (23). In particular, decreased intake of animal fat
and protein, cholesterol, salt, sugar, and alcohol is recommended . Eyer (18)
suggests, however, that reduction of animal fat in the diet may just
redistribute cardiovascular deaths from coronary heart disease to stroke, and
that reducing the total level of cardiovascular disease depends on reducing
smoking and high blood pressure.
Heavy alcohol consumption carries a risk of premature death greatly
exceeding normal expectancy (24,25). "Vtfiile the lifestyle typical of many
heavy drinkers contributes to this risk, the effects of alcohol per se account
for a substantial part of the excess mortality" (24). In two Chicago studies,
heavy drinkers had higher mortality from all causes, cardiovascular diseases,
coronary heart disease, and sudden death, than could be entirely explained by
their blood pressure, smoking, and weight (25). The mortality experience of
1-16
moderate drinkers, however, does not seem to differ notably from that of life-long
abstainers (24).
Mental disease may also increase mortality risk. One study (26) found
that schizophrenic, manic, and depressive patients had a significant increase
in mortality risk over that experienced by the general population of which the
study group was a part. Mortality of a control group did not differ
significantly from that of the general population. Better diagnosis and
treatment of persons with mental diseases could help reduce the level of
mortality.
Social class has a very strong impact on mortality (12,20,27). "Social
class gradients of mortality and life expectancy have been observed for
centuries, and a vast body of evidence has shown consistently that those in
the lower classes have higher mortality, morbidity, and disability rates"
(12). The higher morbidity rates in lower status groups is an important
observation since it indicates that their excess mortality is not merely
attributable to a higher case fatality rate but is accompanied also by a
higher prevalence of morbidity. Differences between white and nonwhite
mortality rates can be attributed largely to social class differences.
Persons in lower socioeconomic groups live in a more toxic, hazardous, and
non-hygienic environment resulting in a broad array of disease consequences.
Low education contributes to poor health practices and low income affects many
aspects of health, including proper nutrition. Higher stress levels and
differences in coping with stress also contribute to higher mortality among
the poor (9, 12). Lower class persons generally receive less adequate medical
care, though this probably does not account for a major portion of the social
class differences in morbidity and mortality (12). Syme and Berkman (12)
conclude that to reduce excess mortality among the poor " new
1-17
approaches must be explored emphasizing the primary prevention of disease in
addition to those approaches that merely focus on treatment of the sick. It
is clear also that such preventive approaches must involve community and
environmental interventions rather than one-on-one preventive encounters."
The male sex is another important factor associated with higher
mortality. Males have higher mortality than females beginning at conception
and this differential continues for every age group. A high ratio of males to
females at conception declines to about 104:100 at birth until by age 70
females outnumber males approximately 3:2. In the United States, the male
age-adjusted death rate is about 50 percent higher than that for females (28).
Diabetes is the only major killer for which higher death rates are reported
for women than men, a result due entirely to higher female rates at older
ages. While there is a significant biological component contributing to
higher male mortality (29), and in particular female hormones may lower the
risk of coronary heart disease, Waldron (11) and others (12) have concluded
that sex differences in behavior are a more important cause of higher male
mortality than are inherent sex differences in physiology. "Each of the
factors which we have identified as a major contributor to men's excess
mortality involves a behavior which is more socially acceptable for males than
for females, for example, aggressive competitiveness, working at physically
hazardous jobs, drinking alcohol and, especially in the early part of the
century, smoking cigarettes. The sex differential in smoking and alcohol
consumption seems also to be linked to underlying attitudes, such as
rebelliousness and achievement striving, which are fostered to a greater
extent in males" (11). Waldron estimates that one-third of the difference
between male and female death rates may be due to men's higher cigarette
1-18
smoking, with the major contribution via increased coronary heart disese, lung
cancer, and emphysema.
If in fact much of the sex difference in mortality is not due to
biological factors, then substantial reductions in men's excess mortality can
be achieved by cultural and behavorial changes. Social and lifestyle changes
may also help to reduce female mortality. With the transition earlier in this
century from infectious to degenerative diseases as the major causes of death,
lifestyle became more important in affecting mortality experience, and the
difference between male and female mortality rates increased steadily. In the
1970's, however, this increase slowed and female mortality relative to male
mortality actually worsened for several age groups and for several leading
causes (30). This may be associated with increased smoking (19) and the
adoption of other "male" behaviors by women as job participation and mobility
increase and traditional roles are modified.
A number of risk factors have been reviewed that bear on many causes of
death, and efforts to reduce excess mortality must involve consideration of
these important precursors.
1-19
North Carolina Mortality Highlights
• Life expectancy at birth has continued to increase reaching 70.0
years for white males, 78.5 years for white females, 63.6 for
r.onwhite males, and 73.4 for nonwhite females in 1980. However,
since 1960 life expectancy at birth has increased by about 3 years
for white males and females, 4 years for nonwhite males, and 7 years
for nonwhite females. Cveral 1 , the median age at death in 1981 was
70.4 years, up from 66.4 in 1970 and 28.1 in 1914 when N.C. deaths
were first centrally recorded.
• A total of 49,212 North Carolinians died in 1981. By race and sex,
the deaths per 1000 population were slightly higher for nonwhites
than whites (by 4%), but significant! y higher for males than females
(by 34%).
• Infant mortality declined in the past 10 years reaching 13.2 deaths
per 1000 live births in 1981, from 24.1 in 1970. But despite this
dramatic improvement, North Carolina's rate continues to rank among
the highest in the nation; only 7 states had a higher rate in 1981.
• Age-adjusted mortality rates show wide gaps in the overall risk of
mortality for North Carolina males versus females and nonwhites
versus whites. In the 1979-81 period, the male rate for both races
was nearly double the female rate, and the nonwhite rate for both
sexes remained more than 40 percent above the white rate.
• Comparison of the 1981 N.C. age-race-sex-specific mortal ity rates
with provisional rates for the U.S. during 1981 reveals the
following excesses in North Carolina:
White males: Ages 55-64 (15% above U.S. rate)
White females: Ages 5-14 (23% above U.S. rate)
1-20
Nonwhite males: Ages 35-44 (31%), 45-54 (30%), 55-64 (18%),
65-74 (10%), 75-84 (14%)
Nonwhite females: Ages 0-4 (10%), 45-54 (14%)
Even after adjusting for age, race, and sex the eastern counties,
especially the northeastern counties, seem to have consistently
higher mortality rates. Based on adjusted rates, statistically
significant spatial clusters exist in eastern North Carolina for
total mortality, heart disease, stroke, motor vehicle accidents,
infant mortality, total cancer, lung cancer, and cancer of the
colon, rectum, and anus.
The highest overall age-adjusted mortality rate for nonwhites is in
the Southern Piedmont HSA while the highest age-adjusted rate for
whites is in the Cardinal and Eastern HSAs. Further, race and sex
ratios show that race differences in age-adjusted rates are greater
in the western part of the state and sex differences are greater in
the east.
Based on the number of North Carolina deaths, the 10 leading causes
in 1981 were (in descending order) heart disease, cancer,
cerebrovascular disease, accidents, pneumonia/influenza, chronic
obstructive lung disease, diabetes, suicide, chronic liver
disease/cirrhosis, and homicide. However, based on the number of
years of life lost prematurely for each cause, the 10 leading causes
were heart disease, cancer, accidents, suicide, homicide,
cerebrovascular disease, chronic liver disease/cirrhosis,
pneumonia/influenza, chronic obstructive lung disease, and diabetes.
Although age-adjusted homicide rates have declined for each race-sex
group, this rate remains substantially higher for nonwhite males.
1-21
In 1981 about 40 percent of all homicide victims were nonwhite
males; yet nonwhite males represented less than 12 percent of the
total population.
Nonwhite males have a greater risk of death from motor vehicle
accidents at every age interval except the 15-24 age group where
white males have the highest rate.
Comparing the U.S. and N.C. unadjusted rates, mortality conditions
in N.C. do not appear to be greater except in the case of
cerebrovascular disease, nephritis/nephrosis, and accidents.
However, adjusting for the state's more youthful age structure,
mortality conditions in N.C. appear much less favorable with only a
slight advantage in the case of cancer, chronic obstructive lung
disease, and chronic liver disease/cirrhosis. Based on 1979 age-adjusted
rates, even those advantages appear limited to females,
with N.C. males of both races experiencing above-average risk in
nearly every cause category.
Cancer mortality appears to be rising faster in N.C. than in the
U.S. In the last two years, the state's age-adjusted cancer death
rate has risen about 2.5% (vs. 0.9% nationwide) with notable
increases for cancer of the pancreas (19%) and cancer of the cervix
uteri (26%).
The former downward trend for cervical cancer appears to have
stalled in recent years. Based on age-specific number of deaths
from cervical cancer, there has been considerable random fluctuation
with a sharp nonwhite drop in 1979 involving all age groups. Only
for white women 55-64 has any consistent pattern of increase
recently occurred.
1-22
•
Both the state and the U.S. have recently experienced increases in
pneumonia/influenza and chronic obstructive lung disease death
rates. N.C.'s decreases in cardiovascular and accident mortality
also reflect national trends, although 1979 to 1981 reductions in
the risks of motor vehicle and other accidents were greater in the
U.S. than in North Carolina.
1982 PROVISIONAL DATA
Provisional North Carolina data show that the unadjusted rate for
all deaths declined from 325.6 deaths per 100,000 population in 1981
to 807.0 in 1982. Cancer and chronic obstructive pulmonary disease
appear to be the only causes with a rate increase in 1982.
1-23
Premature Mortality in North Carolina
Since deaths were first centrally recorded in North Carolina, the leading
causes of mortality have been ranked primarily according to number of deaths.
North Carolina deaths in 1981 have been ranked in Table A for each race-sex
group based on this traditional method. As shown, heart disease, cancer, and
stroke (cerebrovascular disease) are the leading causes of death for each
group, but the rankings of the other causes, as well as the size of the rates
for the same cause, vary among the four groups.
Rankings based only on number of deaths, however, do not necessarily
indicate where medical and public health intervention strategies can most
effectively be employed. Since death is postponable but not preventable,
age at death is a very important factor to consider. Prevention of a death
that would otherwise occur early in life could be assigned higher priority
than prevention of a death later in life. A convenient method of ranking
causes of death that incorporates age at death is by "years of life lost" (6).
If the average life expectancy at birth for white males, for example, is 70
years, a death at age 65 would mean 5 years of life lost (on the average),
while a death at age 40 would mean 30 life-years lost. An infant death
results ir. 70 years of life lost, whereas, deaths at age 70 and over do not
contribute to "life-years lost" for white males. Based on 1979-80 life tables
for North Carolina (16), the life expectancies used here to calculate years of
life lost were 70 for white males, 79 for white females, 64 for ronwhite
males, and 73 for nonwhite females. For each death in a given cause group,
age at death was subtracted from the appropriate life expectancy and all of
these life-years lost were then summed up for that race-sex group, with deaths
over the specified ages not counted. A rate of years of life lost per 100,000
1-24
population was then calculated for ranking the causes of death, so that
comparisons can be made across race-sex groups.
A more accurate method of calculating years of life lost may be to use an
average expected remaining years of life for each age of death rather than an
average life expectancy at birth. This method was considered too complicated
for the present application, and probably would not result in rankings
substantially different from those in Table B. It would place somewhat less
emphasis on infant mortality since more years of life lost would be generated
for the causes listed in Table B.
Table B displays the leading causes of death ranked according to years of
life lost per 100,000 population. Heart disease and cancer are still very
important causes of death from this perspective, but other causes become much
more prominent than before. Motor vehicle accidents account for more years of
life lost for nonwhite males than any other cause, and the general seriousness
of motor vehicle accidents as a life and health hazard is clearly shown in
this table. Cancer becomes a much more important cause of death relative to
heart disease from this perspective, since cancer decedents are about ten
years younger on the average than are heart disease decedents. In fact,
cancer accounts for more years of life lost for white females than any other
cause. Infant deaths in 1981, about 8% of which are counted under causes
(such as accidents) shown in Table B, accounted for 1322 years of life lost
per 100,000 total population (compare column 1 of Table B) and therefore
should be a key target for prevention.
1-25
a ~
(N m oo n
r~ i1 oo <N <-t
r- 10 vo en m
H8 oO
C/l O
I 3*8" Is
3 j S o
CM (o CM vo o r»
11
id §
•d -3
1
I 1
>^SoNcuHvohci«roncNMr-tH
OrHv^rooino
-t o o co r~ n
H io r- »
vo «0 en in « <"> o *o r^ oi m fM PI ^ H ri H
1-26
* m in ^» n"> en ci >-t #-i
t— m t (N in
o\ t— 10 in r-i
a
I
& i
.o-\l*a*^ivoocoora\-^o\oomi.n-cioo
1-27
REFERENCES FOR SECTION I
1. United States Department of Health, Education, and Welfare. Public
tteal th Service, National Vital Statistics Division. Vital Statistics of
the United States, Volume II-Mortality (Part A) . I960 and 1970,
Washington, D.C.
2. United States Department of Health and Human Services, Public Health
Service, National Center for Health Statistics. "Annual Summary of
Births, Deaths, Marriages, and Divorces: United States, 1981," Monthl
y
Vital Statistics Report . (PHS) 83-1120, Volume 30, Number 13,
Hyattsville, Maryland, December 20, 1982.
3. World Health Organization. International Classification of Diseases :
Manual of the International Statistical Classification of Diseases" ,
Injuries, and Causes of Death , Volumes I and II. Geneva, Switzerland,
1978.
4. SAS Institute Inc. SAS User's Guide: Statistics, 1982 Edition , Cluster
Procedure. Cary, N.C. , 1982.
5. Grove, R.D. ; Hetzel , A.M. Vital Statistics Rates in the United States :
1940-1960 . Public Health Service Publication Number TsTf, DTs.
Government Printing Office, Washington, D.C, 1968.
6. Weinman, J.C. Statistical Notes for Health Planners: Mortality . Public
Health Service, Health Resources Administration, (HRA) 77-1237, U.S.
Government Printing Office, Rockville, Maryland, February 1977.
7. Lalondt, M.A. A New Perspective on the Health of Canadians . Government
of Canada, Ottawa, 1974.
R. North Carolina Department of Human Resources, Division of Health
Services, Public Health Statistics Branch. "Associations Between
Mortality and Various Social, Economic, and Environmental Factors in
North Carolina," PHSB Studies . No. 3, Raleigh, April 1977.
9. Milio, N. "An Ecological Approach to Health Planning for Illness
Prevention," American Journal of Health Planning . Number 2, 1977.
10. North Carolina Department of Human Resources, Division of Health
Services, Public Health Statistics Branch. "Health Characteristics of
Adults in North Carolina," PHSB Studies . Number 11, Raleigh, N.C, July
1978.
11. Waldror , I. "Why Do Women Live Longer Than Men?," Social Science and
Medicine. Volume 10, 1976, pp. 349-362.
1-28
12. Syme, S.L. ; Berkman, L.F. "Social Class, Susceptibility, and Sickness,"
American Journal_ of Epidemiology . Volume 104, Number 1, 1976, pp. 1-8.
13. North Carolina Department of Human Resources, Division of Health
Services, Public Health Statistics. North Carolina Vital Statistics :
Quarterly Provisional Report . January-March, 1978. Raleigh, N.C.
14. McKeown, T. ; Brown, R.G. "Medical Evidence Related to English Population
Changes in the Eighteenth Century," Population Studies . Number 9, 1955,
pp. 119-141.
15. Newsweek , June 19, 1978; see also Cassel, E. Human Ecology and Public
Health . Kibourne and Smillie Editors, 1969, p. 353.
16. Office of State Budget and Management, Research Section. "Life Table For
The Total Population, and White Male, White Female, Other Than White
Male, and Other Than White Female Populations, North Carolina, 1979-80."
Raleigh, N.C., 1981.
17. Paffenbarger, R.S.; Brand, R.J.; Sholtz, R.I.; Jung, D.L. "Energy
Expenditure, Cigarette Smoking, and Blood Pressure Level as Related to
Death from Specific Diseases," American Journal of Epidemiology . Volume
108, Number 1, 1978, pp. 12-18.
18. Eyer, J. "A Diet/Stress Interaction Hypothesis of Coronary Heart Disease
EDidemiology," International Journal of Health Services . Volume 9,
Number 1, 1979, pp. 161-168.
19. Doll, R. "The Smoking Induced Epidemic," Canadian Journal of Public
Health . Volume 72, November/December 1981, pp. 372-381.
20. Curtiss, J.; Grahn, R.B.; Grahn, D. "Population Characteristics and
Environmental Factors That Influence Level and Cause of Mortality: A
Review," Journal of Environmental Pathology and Toxicology . Volume 4,
Number 2, 1980, pp. 47l-Sll.
21. Doll, R. ; Peto, R. "The Causes of Cancer: Quantitative Estimates of
Avoidable Risks of Cancer in the U.S. Today," Journal of the National
Cancer Institute, 1981 (in press)
.
22. U.S. Department of Health and Human Services, Office on Smoking and
Health. The Health Consequences of Smoking: Cancer, a Report of the
Surgeon General . Rockville, Md., 1982.
23. Walker, A.R. "Dietary Goals, Sensible Eating, and Nutrition in the
Future," South African Medical Journal . Volume 57, Number 4, 1980, pp.
471-511.
24. Schmidt, W. ; Popham, R.E. "Heavy Alcohol Consumption and Physican Health
Problems: A Review of the Epidemiological Evidence," Drug and Alcohol
Dependence . Volume 1, Number 1, 1975, pp. 27-50.
1-29
25. Dyer, A.R.; et.al. "Alcohol Consumption, Cardiovascular Risk Factors,
and Mortality in two Chicago Epidemiologic Studies," Circulation .
Volume 56, Number 6, December 1977, pp. 1067-1074.
26. Tsuang, M.T. ; Woolson, R.F. "Mortality in Patients With Schizophrenia,
Mania, Depression, and Surgical Conditions: A Comparison with General
Population Mortality," British Journal of Psychiatry . Volume 130,
1977, pp. 162-166.
27. Egbuonu, L. "Child Health and Social Status," Pediatrics . Volume 69,
Number 5, 1982, pp. 550-557.
28. Nathansen, C.A. "Sex, Illness, and Medical Care," Social Science and
Medicine . Volume 11, 1977, pp. 13-15.
29. Madigan, F. "Are Sex Mortality Differentials Biologically Caused?"
Milbank Memorial Fund Quarterly . Volume 35, 1957, pp. 202-223.
30. Verbrugge, L.M. "Recent Trends in Sex Mortality Differentials in the
United States," Women and Health. Volume 5, Number 3, 1980, pp. 17-37.
1-30
II. Technical Notes
CHANGES IN THE CAUSE-OF-DEATH CLASSIFICATION SYSTEM
About every 10 years since 1900, what is now the International
Classification of Diseases (ICD) has been revised in order to incorporate the
latest state-of-the-art in disease classification. Each revision produces
some break in the comparability of cause-of-death statistics, and the Ninth
Revision—first applied to 1979 deaths—is no exception. The most serious
breaks in comparability occurred with the following causes:
pneumonia/influenza, chronic obstructive pulmonary disease, hypertension,
atherosclerosis, other ischemic heart disease, and nephritis/nephrosis.
In order to interpret mortality trends, the National Center for Health
Statistics provided national estimates of the discontinuity associated with
the change from the 8th to the 9th ICD, commonly referred to as "comparability
ratios." These ratios for selected causes of death are found in the 1979
Leading Causes of Mortality . Also in the 1979 volume are a description of the
method used to construct comparability ratios and an evaluation of these
ratios using North Carolina data.
Throughout this volume, comparability ratios are applied to pre-1979
rates for each cause listed above in order to make comparisons with 1979-81
rates. Yet, because these ratios are national estimates and are based on 1976
deaths, the reader is cautioned that the magnitude of the error associated
with the corrected pre-1979 rates is unknown. Thus, for some causes and/or
counties any revised pre-1979 rate could be significantly different from the
"true" rate, thereby causing substantial changes in the trend of rates over
time.
2-3
CHANGES IN POPULATION BASES DUE TO THE 1980 CENSUS
As described in the previous section, the change in the disease
classification system from 1978 to 1979 was one source of discontinuity in
mortality rates over time. A second source was a change in population
estimates.
Every ten years (years ending in zero) the United States Bureau of the
Census has conducted a count of the United States population, which includes a
profile of the population's characteristics such as age, race, and sex. After
a census year, estimates of population size and composition are based on
knowledge of the population in the census year plus subsequent events such as
births, deaths, and migration. In the later years of a decade, these
estimates may depart considerably from the "true" population values if the
assumptions about population change differ significantly from what actually
occurred. Since almost all mortality rates in SCHS publications are based on
population counts in the denominator, changes in estimates of population such
^s might result from a new census could cause substantial changes in the trend
of rates over time.
To complicate the problem even further, it is generally acknowledged that
there was an undercount of the population of North Carolina in the 1970
census, particularly among nonwhites. Thus projections for the latter
seventies tended to be low, and rates based on these population projections
were correspondingly high. The 1980 Leading Causes of Mortality provided a
detailed assessment of the effects of using the 1980 census figures versus
1980 population projections as the denominators of rates.
2-4
In this volume the reader should note that all pre-1980 rates have been
corrected for changes in population bases in order to interpret trends. Tlnese
corrected rates may differ from the rates shown in previous SCHS publications.
2-5
MORTALITY RATES
Computation of Mortality Rates
In this report, total death rates are expressed as resident deaths per
1,000 population while all cause-specific mortality rates are expressed as
resident deaths per 100,000 population. All rates in Tables 1-27 use total
population in the denominator, except rates for the sex-specific cancer sites,
which use male or female population in the denominator. The infant mortality
rates of Table 28 and Figure 28 are computed as the number of resident deaths
under one year of aqe per 1,000 resident live births. Population bases for
the rates of this report were provided by the Office of State Budget and
Management in the Governor's Office.
Vital events in this report are allocated to place of residence. For
ri^aths of inmates of long-term institutions (mental, penal, old age, orphan,
nursing home, rest home, etc.) , the institution is considered the usual
residence provided the decedent had resided in the institution at least one
year. College students and military personnel are considered residents of the
college or military community.
The following definitions apply to the rates of Tables 1-27:
Unadjusted Annual Death Rate : the annual death rate computed as resident
deaths per 1,000 or 100,000 population. These rates permit the user knowledge
of an area's status with respect to the observed incidence of mortality during
the given year.
Unadjusted 3-Year Death Rate : the average annual death rate computed as
average resident deaths per 1,000 or 100,000 average population. These rates
permit the user knowledge of an area's status with respect to the observed
incidence of mortality during the 3-year period. These rates are depicted in
the Series A maps (e.g., Figures 2. A, 3. A, etc.).
2-6
Adjusted 3-Year Death Rate : the averaqe annual aqe- race-sex-ad justed rate,
computed by the direct method. Also expressed as deaths per 1,000 or 100,000
population, these rates are those which would be expected if the average
annual age, race, and sex composition of each county's population were the
same as that estimated for the state. These rates are free of differing
effects of age, race, and sex and thus permit the user knowledge of an area's
status with respect to other determinants of mortality during the 3-year
period. However, the user should be cautious when comparing an adjusted death
rate with an unadjusted death rate. Also, adjusted rates for different time
periods cannot be directly compared unless they were adjusted by the same
standard population. The adjusted rates are depicted in the Series B maps
(e.g., Figures 2.B, 3.B, etc.). These adjusted rates are based on a breakout
of ten age groups for four race-sex groups, for a total of 40 age-race-sex
"cells." In 1978 and earlier editions of Leading Causes of Mortality ,
adjusted rates were based on 18 age categories (72 cells) , and the change here
to a smaller number of age categories is in keeping with the procedures of the
National Center for Health Statistics.
Interpretation of Mortality Rates
To assess an area's relative mortality conditions during a three-year
period, both the unadjusted rate and the adjusted rate can be compared to the
state rate for a particular cause of death. Then, provided the area's
unadjusted rate represents a relatively stable situation, viz., the rate has
not fluctuated widely in recent years, the following alternative diagnoses
will apply:
Relative Status of
Unadjusted
Rate
Low
Low
Adjusted
Rate
Low
High
Diagnosis
Low mortality is not due to age,
race, and sex factors; other
mortality conditions are
favorable.
Low mortality is due to favorable
age, race, and sex factors; other
mortality conditions are
unfavorable.
2-7
High
Hiqh
Low
Hiqh
High mortality is due to
unfavorable age, race, and sex
factors; other mortality
conditions are favorable.
High mortality is not due to age,
race, and sex factors; other
mortality conditions are
unfavorable.
In using adjusted rates, it is important that the user understand the
reason for adjustment. The following hypothetical example illustrates. Here,
A & B stand for population subgroups, e.g., whites and nonwhites, males and
females, etc.
County
Death
Population Deaths Rate*
State
Death
Population Deaths Rate*
Population A
Population B
Total
300,000
10,000
310,000
1,000 333.3
10 100.0
1,010 325.8
500,000
800,000
1,300,000
2,500 500.0
1,000 125.0
3,500 269.2
*Deaths per 100,000 population.
Compared to the state, county subgroups A & B both have lower rates, but
the county as a whole (A & B combined) has a higher rate. This seemingly
paradoxical situation results from two factors: different proportions of A &
B in the county vs. the state population and wide differences between the
rates for A vs. B. In this example, it is true that the county has the higher
total rate—BUT ONLY BECAUSE THE POPULATION CONTAINS A HIGH PROPORTION OF
SUBGROUP A.
Certainly, it is important for program planners to know that the county
has a high rate relative to the state rate; this information is needed in
determining manpower and facility needs, etc. But in assessing "risk," the
researcher needs to adjust for confounding factors such as age or race. To do
this, one multiplies each population-specific county rate by the corresponding
2-1
"standard" population, sums these results, and divides by the total
"standard." In the above example, using the state population as the standard,
the computation is:
( 1000 x 500,000) + ( 10 x 800,000)
300,000 10,000
= 0.001897 or 189.7 per
1,300,000 100,000 population
Thus, as rates specific for A & B imply, the county's rate is lower than
the state rate after adjustment for the factor represented by A & B.
Caution !
In assessing the relative mortality conditions of a county, one should be
particularly aware of rates based on a small number of deaths since, in such
cases, random fluctuation in the rate may render rate comparisons risky. The
reader should read very carefully the next section entitled Flagging Biased
Rates.
Flagging Biased Rates
This section discusses the problem of unstable mortality rates and then
goes on to describe a method of flagging rates with a large bias in a given
year. The reason for flagging an unstable or possibly biased rate is to
provide the reader with a criterion for placing confidence in statements,
tables, maps, and other interpretive displays of the mortality data. Not all
calculated rates are an accurate portrayal of the underlying or "true" force
of mortality, and the intent of this section is to caution the reader against
uncritically using the rates in this volume. In the maps in particular, the
reader should be alert to unusually high rates that may result from very small
numerators or from the type of bias described below.
Any rate with a small number of events in the numerator will be unstable,
with possibly large random fluctuations from year to year that do not comprise
2-9
a significant trend. Such a rate is said to have a large standard error . A
useful rule is that any rate based on fewer than 20 events (deaths in this
volume) in the numerator may have a 95 percent confidence interval that is
wider than the rate itself (1). For example, in an area with 20 deaths out of
20,000 population, the measured rate would be 100 deaths per 100,000
population. But due to variability over time in rates with small numerators,
one could say with 95 percent certainty only that the underlying or "true"
rate is between approximately 50 and 150 per 100,000. Many of the rates in
this volume have numerators smaller than 20 and thus a large standard error,
and therefore conclusions about trends based on these rates should be made
very cautiously, or not at all.
The age- race-sex-ad justed rates in this volume, described in the previous
section, are likewise subject to random variability over time, and those rates
involving a small number of deaths will be the most unstable. Computation of
a standard error for these rates is more complex than for the unadjusted
rates, and depends on the sizes of both the numerators and denominators in
each of the age-race-sex-specific cells. Counties with a small number of
persons in one or more of the 40 age-race-sex groups are, however, most likely
to have adjusted rates with a large standard error, and the rates will
therefore show substantial random fluctuation over time. Table C lists 13
counties whose adjusted rates are likely to have a large standard error due to
10 or fewer persons in one or more of the age-race-sex groups in 1980. All of
these counties had less than 1000 nonwhites in 1980 and it is generally the
nonwhite, age 85-and-over race-age group that has the small number of persons.
The adjusted rates in these counties in particular may change substantially
from one year to the next just due to random error.
2-10
TABLE C
Counties with Small Age-Race-Sex Fbpu]atior.
Groups that are Hke]y to have Adjusted Death
Rates with Large Standard Errors
Alleghany Haywood
Ashe Macon
Avery Madison
Cherokee Mitchell
Q ay Watauga
Dare Yancey
Graham
In addition to the problem of potential random variation in rates due to
a large standard error, there is the problem of bias in an adjusted rate in a
given year due to one or more deaths in a small population group. For
example, it has happened that one death occurs in an age-race-sex group where
the estimated population is only ore person, and this results in a rate of
100,000 deaths per 100,000 population. Of course this is not the "true" rate
in that population group, and the "expected" number of deaths per year is much
less than one, but such an occurrerce can result in an extremely biased
adjusted rate. Applying the rate of 100,000 to the appropriate age-race-sex
cell of the standard population, by the direct method of adjustment, will
result in an expectation of 100 percent of that group dying and thus a very
large adjusted rate. In the past no systematic method existed for "flagging"
these biased rates in Leading Causes of Mortality .
For the first time in this volume, rates with a possibly large bias of
this sort are flagged with an asterisk. Rates with a large standard error
have not yet been flagged, though counties likely to have such rates are
identified in Table C. Factors considered in flagging the 1979-81 age-race-sex-
adjusted rates were: 1) the ratio of the number of expected deaths, in any
one of the 40 age-race-sex cells of the standard population to the total
2-11
expected deaths and 2) the size of the denominator of the eel ] -specific rates.
If applying the ce] 1 -specif ic rates of a county to the standard population
results in more than 25 percent of the total expected deaths being generated
by ore cell, then that adjusted rate is flagged if the denominator of the
rate is also less than 30 (or an average of 10 per year). Ihese criteria
result in only 23 out of a total of 2500 county rates in this volume, or about
ore percent, being flagged.
The 25 percent criterion for one cell was higher than what would be
expected for any of the causes. Ir. fact, this could probably be reduced
somewhat. For 1980 North Carolina total deaths, the highest percent of deaths
accounted for by any ore of the 40 age- race-sex cells was 9.7 by white females
age 75-84. Nonwhite males age 25-14 accounted for 17.5 percent of all 1980
homicide deaths. Ihe cut-off of fewer than 30 persons ir the denominator (an
average of 10 per year for 3 years) was a somewhat arbitrary choice, but a
populatior limit was desirable in order to reduce the chances of flagging an
adjusted rate where a high age- race-sex-specific cell rate was due to an
actual epidemic or disaster. Ihus one death in the problem cell of a flagged
rate would result in a eel 1 -specif ic rate of at least 1/29, or 3448 out of
100,000. For North Carolina as a VNhole in 1979-81, only the total heart
disease rate for the age 85+ group was higher than this, and most of the age-race-
sex-sp..-ci fie rates are very much lower.
We experimented with other combinations of the two criteria, and the
r umbers of rates flagged for six different combirations are shown below.
Doubling the denominator cutoff to ^0 and lowering the percentage of expected
deaths from 25 to 20 triples the rumb^r of flagged rates to 70. While only 23
rates are flagged in this volume, it is felt that most of the flagrant biases
2-12
have been identified. Work will continue to evaluate and enhance this
methodology.
Number of Flagged Rates Based On
Denominator Size and Percent of
Total Expected Deaths
Denominator
Size Less
Than
30
45
60
Percent of Total Expected Deaths
Accounted for by One Cell Greater
Than
20% 50%
31
51
70
23
38
51
In addition to putting asterisks by those 1979-81 adjusted rates with a
large bias due to an extremely high cell-specific rate, the rates have been
recomputed by substituting the appropriate North Carolina age-race-sex-specific
rate in the problem cell. Table D compares the flagged rates with
the recomputed rates. In a very extreme case, like the Clay County homicide
rate, substituting the N.C. age-race-sex-specific rate produces a much more
reasonable adjusted rate by reducing the rate from 126.4 to 7.0, the latter
being much more in line with the state homicide rate for 1979-81 of 11.1. In
a few cases, substituting the state rate makes the county's adjusted rate much
smaller than the state figure, which also illustrates the volatility of death
rates based on just a few events. The reader who must use one of these
flagged rates is advised to use the rate in Table D with the substitution
instead, unless this latter rate is clearly out of line with the North
Carolina rate in which case the rate should be treated as missing data.
2-13
u
PROCEDURE FOR GEOGRAPHIC CLUSTERING OF COUNTIES
In previous editions of this report, county rates for each leading cause
of mortality have been mapped. As described in the Introduction of this
volume, these maps showed counties grouped together (usually in six groups)
based on similar rates—in other words, a numerical clustering of rates. From
these maps, one might imply "spatial" clustering if counties with high rates
happened to be contiguous to each other. However, under this procedure there
is no way to know whether this clustering is a "true" geographic cluster or an
artifact of the SAS clustering procedure (2).
In 1980 the State Center for Health Statistics in conjunction with the
UNC Department of Biostatistics developed a new statistical procedure for
detecting and characterizing spatial clusters. This procedure determines
statistically whether counties with high rates are clustering on some
geographic basis. Only a brief summary of the mechanics of the procedure is
presented in this section. For a detailed description, interested readers are
advised to read "Searching for Hierarchical Clusters of Disease: Spatial
Patterns of Sudden Infant Death Syndrome" in Social Science Medicine , Vol.
15D, pp. 287-293 or contact the State Center for Health Statistics.
Briefly the mechanics of this procedure are as follows:
(1) Counties are listed in rank order of their unadjusted or adjusted
rates (highest rates first) for each cause of death.
(2) For each county listed, cumulative counts are obtained of the number
of counties and the number of adjacencies involved between the
county and those counties with higher rates. Two counties are
2-15
considered adjacent if they have a common nonsalt-water boundary.
For some "pairs" of counties (in particular, counties which seem to
touch only at one point on a map) , it was difficult to decide
whether these counties are adjacent. These "troublesome pairs of
counties" in North Carolina are discussed in the above paper as are
the decisions made with regard to these pairs.
(3) Finally, a reference table developed in the above paper is used to
obtain for the cumulative number of high-rate counties a
significance level based on the number of adjacencies observed.
While a few adjacencies might be expected to occur randomly, a
relatively large number of adjacencies will have a low probability
of random occurrence. In this report, the high-rate counties are
said to cluster geographically if the probability of such random
occurrence is .05 or less; maps for a given disease are included in
the text only if statistically significant geographic clusters were
found to exist.
This procedure can result in one or more groups of contiguous high-rate
counties. This concept is referred to as regional ization, defined here as any
area of the state consisting of two or more contiguous counties being
identified as a high-risk area. For some diseases, no regional ization of this
type exists. For some others it does, and for yet others two or more regions
exist. In the latter case, a disease may regionalize with high rates in one
part of the state and with other tiers or levels of high rates in some other
parts of the state. In this report each map shows the multi-tiers or levels
via different shading schemes.
2-16
This method is a simple and inexpensive approach for recognizing
mortality patterns of varying intensity in geographical areas and for testing
for clusters. Further refinements of the procedure are in the planning stage.
2-17
REFERENCES FOR SECTION II
1. Kleinman, J.C. "Infant Mortality," Statistical Notes for Health
Planners . Number 2, National Center for Health Statistics, Rockville,
Maryland, July 1976.
2. SAS Institute Inc. SAS User's Guide: Statistics, 1982 Edition , Cluster
Procedure. Cary, N. C. , 1982.
2-18
North Carolina's
Population and
Health Care Resources
NORTH CAROLINA'S POPULATION AND HEALTH CARE RESOURCES
As stated ir the 1976 edition of this volume (1):
Equal access to health care is a basic right of all
citizens, but to date, this right has been
precluded in North Carol ina by such interrelated
deterrents as poverty, rural ity, and
maldistribution of health care resources.
The ensuing years have brought progress but the old problems still persist.
Heart disease, career, stroke, and accidents continue to be the leading causes
of death. Although declining, North Carolina's infant mortality rate
cortinues to be ranked amorg the highest in the 50 states. Compounding these
problems is a rapidly aging North Carolina population which has served to
minimize decreases in mortality.
To combat these problems, significant strides have been made to improve
the availability of health care resources. The numbers of physicians, nurses,
ar d hospital ard nursing home beds have increased. Services of primary care
physiciars have been augmented with the ir.creased use of physiciar assistants
and rurse pr actitior.ers. But despite these ard other recent improvements,
North Carolina still ranks behind most other states in the availability of
most types of health manpower and facilities (2).
From this assessment, ore fact appears to be quite evident. That is,
neither the problems nor the resources are evenly distributed in North
Carolina. Too oftrj r the problems occur where the resources are scarce. While
this sectior will rot undertake a detailed analysis of health care resources
ir each county, it will examine the levels of selected health manpower and
facilities across the state. Hopefully, these data will aid health officials
and others in planning for improved health conditions throughout the state.
3-3
County Populations
Underscoring the state's need to improve the availability and
distribution of health care resources, increasing size and changing
composition of North Carolina's population are altering the disease and
mortality proneness of various geographical areas. Between 1976 and 1981,
North Carolina's population increased an estimated o.5 percent, with nonwhites
experiencing an 8.6 percent increase. Of prime importance here is the rapidly
riging North Carolina population. During the past 5 years, persons 65 and over
increased by 17.8 percent, a pace faster than that for the United States as a
whole. While the state's population as a whole is projected to increase only
7 percent by 1986, the state's elderly population is expected to increase
another 17.8 percent. Further, persons 75 and older are projected to increase
by 27 percent in the next five years, a pace faster than that of any other age
group.
While detailed analysis of each county's population is beyond the scope
of this volume, county populations have been compared to the state population
with respect to four demographic characteristics—percent rural, percent over
64 years of age, percent nonwhite, and percent with income below the poverty
level. In the 1976 edition of this volume, 31 counties were found to be above
the state rate on all four characteristics. When the same comparison was done
with 1980 data, 30 counties were above the state, including 26 of those in
1976. As before, these demographically high-risk counties comprise fairly
well-defined geographical pockets in central and eastern North Carolina:
3-4
Anson
other eastern counties—Chowan, Pasquotank, Dare, and New Hanover—which have
relatively low ratios (Level 1). Physicians in these latter counties may well
be providing primary care services to surrounding counties.
Among the counties with level-four to level-six ratios, four—Bertie,
Camden, Gates, and Perquimans—have both a demographically high-risk
population (see page 3-4) and high mortality (See Figure 2.A) . Two of these
counties (Camden and Perquimans) have only one primary care physician and one
(Gates) has none.
County Population Per Primary Care Physician, Physician Assistant, and Nurse
Practitioner
In North Carolina there has been a recent increase in the use of
physician assistants and nurse practitioners. Variously referred to as health
practitioners, physician extenders, or mid-level practitioners, these
professionals can perform certain defined medical tasks previously done by
physicians; however, they can perform these tasks only under a physician's
supervision. Yet, their impact on the level of primary care services
available in North Carolina requires that these professionals be considered
for the first time in this volume.
During 1981 a total of 762 physician assistants and nurse practitioners
provided primary care services in 87 counties. Because of the number and
distribution of these professionals, it is desirable to show their effect on
the potential case load of primary care physicians. One approach involves
treating these practitioners as some proportion of a fulltime primary care
physician, thereby reflecting the proportion of physician's work that the
practitioner may perform. A recent study suggested 63 percent as a "good
middle-of-the-road estimate" (3). Using this percent, every 100 physician
3-6
assistants and nurse practitioners will be equivalent (ir terms of work load)
to 63 full time physicians. The number of practitiorers obtained using this
procedure will hereafter be referred to as "physician equivalents." For North
Carol ina , the population-per-pr imary-care-physician-and-physiciar -equivalent
ratio is 1570.
figure l.B shows the county population per primary care physician and
physician equivalent in 1981. The use of this ratio changes substantially the
picture of primary care services available in some counties, especially a few
of the high-ratio counties identified in Figure l.A. For example, in Greene
County, the popul ation-per-physician (1 physician) ratio is 16,078 but the
popul ation-per-physician/physician-equivalent (1 physician, 4 nurse
practitioners, and 2 physician assistants) ratio is 3,364, a 79.1 percent
decrease in potential case load. Ir: Hoke County the respective ratios are
7,067 (3 physicians) and 4,336 (3 physicians and 3 physician assistants).
Gates was the only county without a primary care physician and thus had no
physiciar assistants or nurse practitioners. Cotrties with high ratios
(Levels 5 and 6) in both Figures l.A and l.B should be priority areas for
recruitment of primary care services. Among these counties, four—Bertie,
Camder , Gates, and Perquimans—appear in particular need of additional
services because they have demographically high-risk populations (see page 3-
4), have high mortality rates (Levels 4-6), and have popul at ion/physician-equivalent
ratios over 2.5 times the North Carolina ratio of 1,570.
County Fbpulation Per Registered Nurse
The ratios of Figure l.C are the county populations per registered nurse
working in the county and licensed to practice in North Carolina as of March
1, 1981. As expected, counties with high population/nurse ratios are largely
3-7
those with high population/physician ratios (Figure l.A) . Q,e exception is
Graham County with a very high population/nurse ratio but a relatively low
population/physician ratio.
As discussed for physicians, county needs for registered r.urses must take
into account the characteristics of county populations. In Figure l.C, the 11
counties with population/nurse ratios at levels 4-6 include six counties
—
Currituck, Franklin, Hyde, Northampton, Pamlico, and Warren—which are
demographicall y high-risk according to criteria described on page 3-4 and are
unusual ly mortal ity-prone (Levels 4-6) as depicted in Figure 2. A.
County Population Per Hospital Bed
Short-term general hospitals provide a broad range of medical and
surgical services to patients whose average hospital stay is Jess than 30
days. The ratios of Figure I.Dare the county populations per patient bed in
short-term general hospitals licensed by the North Carolina Division of
Facility Services as of December 31, 1981. As explained for physicians,
federal hospitals are not included but county populations include members of
the armed forces; thus, ratios are biased upward for counties with large
military populations (i.e., Craven, Cumberland, Q-.slow, and Wayne counties).
As show, in Figure l.D, seventeen North Carolina counties have no short-term
general hospitals (level six) and another ten have high population/bed
ratios (Levels 4 and 5). Among these counties, three (Camden, Gates, and
Perquimans) are also demographicall y high-risk according to the criteria given
or page 3-4, have high mortality rates as show, in Figure 2. A, and have high
populatior-per-physician/physician-equivalent ratios. The popul ation-per-bed
ratio for North Carolina is 270.
3-8
Even more so than in the case of physicians, hospitals in certain larger
counties will provide services to the residents of surrounding counties.
Therefore a high ratio for a county does not necessarily mean that its
residents are not receiving hospital services. It is not feasible for every
N.C. county to have a hospital, and regional ization of some services is indeed
desirable.
County Population 75+ per Skilled Nursing and Intermediate Care Beds
Two levels of nursing homes exist in North Carolina—skilled nursing
facilities and intermediate care facilities. Skilled nursing facilities
provide continuous medical supervision to patients for management of chronic
conditions. Intermediate care facilities provide health- related care and
services to individuals who do not require the degree of care or treatment
which a hospital or skilled nursing facility is designed to provide. Many
nursing homes offer both levels of care. (4)
The growth of nursing homes and nursing home beds over the past decade
has been tremendous. In 1970 there were 108 homes containing 6,809 beds for
skilled and intermediate care. By 1981 the number of facilities had increased
to 205 and the number of beds to 19,824, a near doubling of facilities and
tripling of beds. The number of primary users of these beds has increased
rapidly also. In this volume, "users" are defined as persons over age 74 due
to the fact that this age group comprised 71 percent of all admissions to
nursing homes in 1980-81. Between 1970 and 1981, the 75-and-older population
increased 61 percent and, as noted previously, is projected to increase
another 27 percent by 1986 (5). Yet, while the elderly population continues
to grow, the facility and bed growths have essentially stopped with the
current moratorium on the approval of new nursing home beds.
3-9
Figure l.E shows the distribution of population-per-skilled-and-intprmedi
a te-ca re-bed ratios for each county. As exemplified in previous
imps, an uneven distribution exists among counties with the same eastern high-risk
counties having the fewest beds. Several contiguous counties in the
northeast corner of the state have no beds, but in the midst of these
counties, two—Chowan (4.7) and Pasquotank (5.4) —have exceedingly low ratios
in contrast to the state as a whole (11.0 persons 75+ per bed). This suggests
that facilities in these two counties may be serving persons from surrounding
counties. In fact, among the 21 counties with no beds, almost all have a
neighboring county with a low ratio (i.e., less than 10.0).
3-10
CE
CC
_l
~ZL GZ
_J
CO
LU
CO
CC
Q
LU
CC
LU
h-
CO
i—
i
CD
LU
CC
UJ
CC
CO
u.
EGO
3-15
ac
CC
CO
CD
LU
0Q
a:
Q_
CO o
oc
CD
cc
cc
CD
DZ
CO
_l
ce
-Z. CE
CO
LU
CQ
CD
CO
CC
cc
CE
C_J
LU
CE
i—
i
Q
LU
11
CC
LU
~3
CD
CO
_J
REFERENCES FOR SECTION III
1. North Caroling Department of Human Resources, Division of Health
Services, State Center for Health Statistics. Leading Causes of
Mortality - Volume II . Raleigh, December 1977.
2. North Carolina Department of Human Resources, Division of Health
Services, State Center for Health Statistics. N.C. Health Statistics
Pocket Guide . Raleigh, February, 1982.
3. Perry, H.B.; Breitner, B. Physician Assistants: Their Contribution to
Health Care . Human Sciences Press, Inc., New York, 1982.
4. North Carolina Department of Human Resources, Division of Facility
Services, State Health Planning Section. State Medical Facilities Plan
1983 . Raleigh, January 1983.
5. Office of State Budget and Management, Research and Planning Section.
Update: North Carolina Population Projections . Raleigh, N. C, July,
1981.
3-21
IV. General Mortality
in North Carolina
GENERAL MORTALITY IN NORTH CAROLINA
During 1981 a total of 49,212 North Carolinians died. This number
represents an annual death rate of 8.3 resident deaths per 1,000 population,
5.7 percent below the rate of 8.8 in 1970 but almost equal to the corrected
(for reasons described in Section II) rate of 8.2 in 1976. The provisional
1981 rate for the United States was higher at 8.7 (1).
One confounding factor in making comparisons of mortality rates is that
age structure, which has an important effect on mortality, may vary among
geographic areas and over time. Compared to the nation, North Carolina has a
younger but more rapidly aging population. Thus, adjusting the North Carolina
and United States mortality rates for age, a higher mortality rate is obtained
for N.C. (6.1) than the U.S. (5.7) for 1981, illustrating that N.C. 's lower
unadjusted rate is due to its favorable age structure (i.e., more people at
the young low-risk ages) . The "bottoming-out" of the unadjusted rates in the
seventies suggests also that the N.C. population is aging.
While the N.C. trend for unadjusted rates indicates virtually no change
in mortality, examination of adjusted rates shows a different pattern. From
the 1974-78 to the 1979-81 period (periods in which comparable age-adjusted
rates were readily available), the risk of death for North Carolinians
actually declined by 7.9 percent from 668.2 to 615.6 per 100,000 population.
Yet, within the above general comparisons of mortality, significant
subgroup differences are concealed, especially between N.C. and the U.S. For
example, while the unadjusted U.S. rate was higher than that for North
Carolina, each of the ten-year age groups between ages 35 and 64 in North
4-3
Carolina had age-specific rates at least 10 percent higher than comparable
United States groups. The other North Carolina age groups with higher rates
were the 0-4 age group (8.3% higher), the 5-14 group (1.6% higher), and the
65-74 group (4.4% higher). The 15-24, 25-34 and 85+ N.C. age groups had lower
rates than their U.S. counterparts.
As observed with age, general comparisons can mask variations by race and
sex. Looking at North Carolina deaths in the 1979-81 period, the nonwhite
rate (840.5 per 100,000 population) exceeded the white rate (808.8) by only
3.9 percent, but the male rate (948.4) exceeded the female rate (692.0) by
37.1 percent. Yet, even those comparisons mask the exceedingly high overall
mortality rate for nonwhite males. Adjusting the race-sex-specific rates for
age using a relatively young 1940 U.S. population as the standard, the
adjusted rate was lower than the unadjusted rate for each race-sex group
except nonwhite males, who had a higher death rate after adjustment (1010.8
vs. 1126.0). This higher rate implies that nonwhite males are dying at
younger ages than any other race-sex group, a fact also revealed through the
age-specific rates. With the exception of the 15-24 and 85+ age groups,
nonwhite males had the highest age-specific mortality rates of the four race-sex
groups. In each of the ten-year age groups between ages 25 and 54, the
rates for nonwhite males were at least twice those for any other group. In
Sections V through X, the varying risks of mortality by age, race, and sex
groups will be discussed for each cause.
Geographic Patterns
The 1979-81 unadjusted mortality rates for counties, depicted in Figure
2. A, ranged from 3.8 per 1,000 population to 12.8, with a N.C. rate of 8.3 for
the period. In general, Figure 2. A shows several scattered groups of high-
4-4
rate counties, with the northeast seemingly having the largest cluster. In
most cases, these high rates reflect unfavorable mortality conditions other
than age, race, and sex since among the 23 counties with level-five or higher
unadjusted rates (Figure 2. A) , 18 had level-four or higher adjusted rates
(Figure 2.B) . In addition, Figure 2.B shows a large eastern band of
contiguous, high-rate counties (levels 5-6) extending from Virginia to South
Carolina and eastward to the Atlantic Ocean.
Based on procedures described in Section II of this volume, the
clustering of high unadjusted county rates in the northeast was found to be
statistically significant (see map below) . In fact, two separate northeastern
clusters were found to exist: one group involving seven counties centered
around Halifax and Warren counties, and a second group with high but lesser
rates consisting of Beaufort, Hyde, and Tyrrell counties. These 10 counties
are among the top 12 counties experiencing the highest mortality rates in the
1979-81 period. A third distinct cluster exists in the west with Polk, which
had the highest of all county rates, as the base.
GENERAL MORTALITY
MORTALITY RRTES
PER 1,000 POPULATION
10.6 - 12.8 (p <.001)
ES 9.7 - 10.5 (p=.043) D 0.0 - 9.6
NORTH CAROLINA
RESIDENT DATA
1979 - 1981
4-5
The clustering of high adjusted rates in the east was also found to be
statistically significant (see map below) . Unlike the unadjusted rates, no
significant clusters of high adjusted rates exist in the west. Thus, eastern
North Carolina is experiencing high levels of general mortality that cannot be
explained on the basis of age, race, and sex alone. In these cases, high
mortality may be due to socioeconomic and resource factors (see discussion of
health care resources in Section III) or to some other set of local conditions
that invite intervention. The cause-specific data in the sections to follow
will aid counties in identifying their particular kinds of mortality
proneness.
GENERAL MORTALITY
RGE-RACE-SEX ADJUSTED
MORTALITY RATES
PER 1.000 POPULATION
B 9.0 - 9.9 (p <.001)
fg 8.7 - 8.9 (p <-001)
0.0 - 8.6
NORTH CAROLINA
RESIDENT DATA
1979 - 1981
4-6
table: 2
mortality statistics for 19 8 1
north carolina residents
total deaths (per loco population)
GEOGRAPHICAL
AREA
f'ORTH CAROLINA
REGIONS
DHP
DHR
WESTERN
N. CENTRAL
DHR III S. CENTRAL
DHR IV EASTERN
HSA
HSA
HSA
HSA
HSA
HSA
I WESTERN
II PIEDMONT
III S. PIEDMONT
IV CAPITAL
V CARDINAL
VI EASTERN
COUNTIES
1 ALAMANCE
2 ALEXANDER
3 ALLEGHANY
4 ANSON
5 ASHE
6 AVERY
7 BEAUFORT
8 BERTIE
9 BLADEN
10 BRUNSWICK
11 BUNCOMBE
12 BURKE
13 CABARRUS
14 CALDWELL
15 CAMDEN
16 CARTERET
17 CASWELL
18 CATAWBA
19 CHATHAM
20 CHEROKEE
21 CHOWAN
22 CLAY
23 CLEVELAND
24 COLUMBUS
25 CRAVEN
26 CUMBERLAND
27 CURRITUCK
26 DARE
29 DAVIDSON
30 DAVIE
31 OUFLIf>'
32 DURHA"
33 EDGECOMBE
34 FORSYTH
35 FRANKLIN
36 GASTON
37 GATES
3B GRAHA*
39 GRANVILLE
40 GREENF
41 GUILFORD
* SEE SECTION II
NUMBER
MORTALITY STATISTICS FOR 1961, NORTH CAROLINA RESIDENTS
TOTAL DEATHS (PER 1000 POPULATION) CONT'D.
COUNTIES
(CONT'O)
42 HALIFAX
43 HARNETT
44 HAYWOOD
45 HENDERSON
46 HERTFORD
47 HOKE
4e HYCE
49 IREDELL
50 JACKSON
51 JOHNSTON
*2 JONES
53 LEE
54 LENOIP
55 LINCOLN
56 *CDOWFLL
57 MACON
5P "ADISON
59 MARTI"
60 MECKLENBURG
61 MITCHELL
62 MONTGOMERY
63 MOORE
64 r.'ASH
65 NEW HANOVER
66 NORTHAMPTON
67 ONSLOW
68 ORANGE
69 PAMLICO
70 PASCUOTANK
71 PENDER
72 PERCUIMANS
73 PERSON
74 PITT
75 POLK
76 RANDOLPH
77 RICHMOND
78 POPESCM
79 ROCKINGHAM
80 ROWAN
81 RUTHERFORD
82 SAMPSON
83 SCOTLAND
84 STANLY
85 STOKES
86 SURRY
87 SWAIN
«8 TRANSYLVANIA
89 TYRRELL
90 UNION
91 VANCE
92 WAKE
=>3 WARREN
04 WASHINGTON
95 WATAUGA
96 WAYNE
97 WILKES
98 WILSON
99 YADKIN
100 YANCEY
NUMBER
>-
CE
h-az
CD
cr
az
i_u
LU
CD
> in m
GOD
4-9
>-
en
i—
cc
CD
cn
cc
LU
CD
LU
cc
CD
3 *-
REFERENCE FOR SECTION IV
United States Department of Health and Human Services, Public Health
Service, National Center for Health Statistics. "Annual Summary of
Births, Deaths, Carriages, and Divorces: United States 1981, "Monthly
Vital Statistics Report . (PHS) 83-1120, Volume 30, Number 13,
Hyattsville, Maryland, December 20, 1982.
4-13
V. Major Cardiovascular
Disease Mortality
HEART DISEASE
Over 17 percent of deaths among North Carolina residents in 1981 resulted
from heart disease, which has been the leading cause of death for the state
and the nation for almost 50 years (1) . Diseases of the heart accounted for
18,270 deaths in the state in 1981, a rate of 306.5 per 100,000 population.
While the 1981 provisional rate for the United States was higher at 330.6 (2),
North Carolina's age-adjusted rate of 211.7 was 7.8 percent higher than the
nation's (196.4).
Almost 72 percent of the state's heart disease deaths were from ischemic
heart disease in 1981. Of these deaths, acute myocardial infarction (heart
attacks) accounted for 8,091 deaths while the other forms of ischemic heart
disease accounted for 5,036 deaths.
The death rate from diseases of the heart has declined in North Carolina,
with the 1979-1981 age-adjusted rate of 211.9 representing an 8.4 percent
decrease over the 1974-78 rate of 231.4. The largest decrease was among white
males for whom the 1979-1981 age-adjusted rate was down over 11 percent at
301.2 as opposed to 338.9 for the 1974-1978 period. Age-adjusted rates for
the other race-sex groups also declined and were as follows for 1979-1981
(with percent reduction): 128.0 for white females (6.4%), 333.3 for nonwhite
males (2.6%), and 182.9 for nonwhite females (5.7%). Age-adjusted death rates
decreased for both acute myocardial infarction and other types of ischemic
heart disease for all four race-sex groups.
5-3
Risk Factors
A number of factors are associated with an increased risk of heart
disease. Among these are hypertension, atherosclerosis, high cholesterol,
diabetes, cigarette smoking, increasing age, sex (men have higher mortality
rates though the male and female rates tend to converge around the age of
menopause), and a sedentary lifestyle (1,3-5). Although white males had the
largest decrease in death rates from acute myocardial infarction between the
1974-1978 and 1979-1981 periods, they still die of heart attacks more than
twice as often as they die of other ischemic heart disease. For males of all
races, heart attacks caused more deaths than other forms of ischemic heart
disease up to the 85-and-over age group. The same was true for white females,
whereas nonwhite females 75-84 had a higher rate of death from other forms of
ischemic heart disease than from heart attacks.
North Carolina's decrease in heart disease death rates is consistent with
declines nationwide (1,2). While it is difficult to pinpoint specific reasons
for these decreases, awareness of the risk factors associated with heart
disease has been heightened in the medical community and general population.
As a result, people seem to be more conscious of diet and exercise, and the
medical community has responded with earlier diagnosis and treatment of
hypertension and other medically associated risk factors (1,4,5).
Geographic Patterns
Figure 3.A depicts mortality rates for diseases of the heart for 1979-
1981 and shows several contiguous, high-rate counties in the northeast,
southcentral , and western portions the state. Adjusting for age, race, and
sex, however, changes the distribution of heart disease mortality among North
Carolina residents. As shown in Figure 3.B, the high-rate counties (Levels 4-
5-4
6) exist almost exclusively in the east.
Based on the procedures described in Section II of this report, the map
below reveals that of the 13 counties with the highest unadjusted rates, a
statistically significant geographic cluster exists in the northeast.
DISEASES OF THE HERRT
MORTALITY RATES
PER 100,000 POPULATION
13 395.2 - 487.0 (p° .021)
0.0 - 395.1
NORTH CAROLINA
RESIDENT DATA
1979 - 1981
Adjusting for age, race, and sex, the clustering properties of county
rates are stronger and are largely eastern (see map below) . The rare and
rather large spatial grouping of 15 counties extends from Tyrrell to Halifax
and south to Bladen. The interpretation here is that these contiguous
counties are all experiencing higher levels of heart disease mortality than
can be explained on the basis of age, race, and sex and that some regional
factor is in operation.
5-5
DISEASES OF THE HEART
AGE-RACE-SEX ADJUSTED
MORTALITY RATES
PER 100,000 POPULATION
Eg 329.4 - 421.6 (p <.001)
0.0 - 329.3
NORTH CAROLINA
RESIDENT DATA
1979 - 1981
The two major forms of heart disease (acute myocardial infarction and
other types of ischemic heart disease) are examined separately in the
following sections.
5-6
GEOGRAPHICAL
ARE*
T A B L E 3
MORTALITY STATISTICS FOR 1961
NORTH CAROLINA RESIDENTS
DISEASES OF HEART
NUMBER DEATH
OF DEATHS RATE*
1979-61 1979-81
NUMBER
MORTALITY STATISTICS FOR 1961. NORTH CAROLINA RESIDENTS
DISEASES OF HEART CONT'D.
COUNTIES
ICONT'D)
42 HALIFAX
43 HARNETT
44 HAYWOOD
45 HENDERSON
46 HERTFORD
47 HOKE
48 HYDE
49 IREDELL
50 JACKSON
51 JOHNSTON
52 JONES
53 LEE
54 LENOI"
55 LINCOLN
56 MCDOWELL
57 MACON
58 MADISON
59 MARTIN
60 MECKLENBURG
61 MITCHELL
62 MONTGOMERY
63 MOORE
64 NASH
65 NEW HANOVER
66 NORTHAMPTON
67 ONSLOW
68 ORANGE
69 PAMLICO
70 PASOUOTANK
71 PENDER
72 PEROUIMANS
73 PERSON
74 PITT
75 POLK
76 RANDOLPH
77 RICHMOND
78 ROBESON
79 ROCKINGHAM
80 ROWAN
81 RUTHERFORD
82 SAMPSON
83 SCOTLAND
84 STANLY
85 STOKES
86 SURRY
87 SWAIN
88 TRANSYLVANIA
89 TYRRELL
90 UNION
91 VANCE
92 WAKE
93 WARREN
94 WASHINGTON
95 WATAUGA
96 WAYNE
97 WILKES
96 WILSON
99 YADKIN
100 YANCEY
NUMBER
az
i_u
LU
CO
LU
CO
CC
LU
CO
i—
i
Q
cr
cn
LU
CC
IS)
J IT)
!>
CE
_l
ACUTE MYOCARDIAL INFARCTION
An estimated 3,400 Americans suffer heart attacks each day. In 1981 over
292,000 of these attacks in the United States were fatal, a provisional rate
of 127.5 deaths per 100,000 population (2). North Carolina's 1981 rate was
higher at 135.7. Both the unadjusted and age-adjusted rates for acute
myocardial infarction exceeded the corresponding United States rates.
A total of 8,091 fatal heart attacks occurred to North Carolinians in
1981, and over 60 percent of those decedents were males. As shown by age-adjusted
rates for the 1979-81 period, males had a much greater risk than
females. White males had the highest rate (164.4), followed by nonwhite males
(134.8), nonwhite females (61.7), and white females (56.9). The risk of death
from acute myocardial infarction declined from the 1974-78 to the 1979-81
period for each race-sex group with white males experiencing the largest
percent decline (19%)
.
Risk Factors
As is the case with heart disease in general, the risk of fatal heart
attacks is correlated with high cholesterol, cigarette smoking, sex (males
have higher rates of heart attacks), and low levels of exercise (1,3-6).
Also, hypertension and atherosclerosis are known to be related conditions
(3,4) as indicated by the high number of mentions of these two conditions on
death certificates that also mention acute myocardial infarction (see Table 31
in Section X) . Improvements in diagnosis and treatment of these conditions as
well as decreased exposure to other risk factors and improved medical services
(i.e., establishment of coronary care and intensive care units and better
5-13
training among emergency medical services personnel) have contributed to
declines in fatal heart attacks (1,4).
Geographic Patterns
As shown in Figure 4. A, high rates (Levels 4-6) are dispersed throughout
the state with three noticeable areas of high rates—the northeast, southeast,
and southcentral areas of the state. Low rates are found in the northcentral
and western portions. After adjusting for age, race, and sex, the clusters of
counties with high rates persist, most notably in the southcentral and
southeastern portions of the state. The western counties continue to have
relatively low rates.
Procedures described in Section II were used to test statistically
whether the above contiguous areas with high unadjusted rates cluster
geographically. As shown in the map below, four significant geographic
clusters of high mortality from acute myocardial infarction exist; however,
these clusters are not very strong, and when adjustments are made for age,
race, and sex, no statistically significant clusters remain.
ACUTE MYOCARDIAL INFARCTION
MORTALITY RRTES
PER 100.000 POPULATION
B 179.8 - 263.4 (p-.043)
0.0 - 179.7
NORTH CAROLINA
RESIDENT DATA
1979 - 1981
5-14
GEOGRAPHICAL
AREA
T A B L E it
MORTALITY STATISTICS FOR 1961
NORTH CAROLINA RESIDENTS
ACUTE MYOCARDIAL INFARCTION
NUMBER DEATH NUMBER DEATH
OF DEATHS RATE* OF DEATHS RATE*
1981 1961 1979-81 1979-81
NORTH CAROLINA
REGIONS
8091 135. 74 24177 136.81
ADJUSTED
DEATH RATE*
1979-81
136.61
DHR I WESTERN
MORTALITY STATISTICS FOR 1981. NORTH CAROLINA RESIDENTS
ACUTE MYOCARDIAL INFARCTION CONT'D.
COUNTIES
CD
CC
az
Q
az
az
o>-
UJ
o
a:
a:
<3-
UJ
7 in
o
cc
az
Q
DC
a: oo>-
en
UJ
cc
=3
ismn
5-19
OTHER ISCHEMIC HEART DISEASE
During 1981, almost 97 percent of the 5,036 North Carolina deaths due to
ischemic heart disease other than acute myocardial infarction involved chronic
forms of ischemic heart disease. The overall death rate was 84.5 per 100,000
population, up 4.4 percent from the 1979 rate of 80.9. Caution must be
exercised in discussing any pre-1979 rates for other ischemic heart disease
because of changes in the ICD codes. As described in Section II, these
changes made it necessary to correct the 1974-78 rates to make comparisons
with 1979-81. A close examination of the 1979-81 age-adjusted rates and the
corrected 1974-78 rates reveals a decline of 15 percent in the overall rate.
Each of the four race-sex groups experienced a decline, the largest among
nonwhite females (33%) and the smallest among white males (3.4%). The 1979-81
age-adjusted rates were as follows: 78.2 for white males, 33.8 for white
females, 83.5 for nonwhite males, and 45.0 for nonwhite females. Along with
the reduction in rates, the gaps in risk declined between nonwhite and white
males (from a nonwhite/white ratio of 1.32 to 1.07) and between nonwhite and
white females (from a ratio of 1.59 to 1.33).
Geographic Patterns
As exhibited in Figure 5. A, the 1979-81 rates appear to be somewhat
higher in the northeast, but no strong patterns are evident. Wilkes, Polk,
and Johnston counties had the highest unadjusted mortality rates for other
ischemic heart disease in the 1979-81 period. Adjusting for age, race, and
sex, Figure 5.B shows an apparent band of high-rate counties (Levels 4-6) in
5-21
the eastern portion of the state and comparatively low rates in the higher
elevations of the west.
The eastern cluster of high adjusted rates was found to be statistically
significant based on procedures described in Section II. Among the 18
counties with the highest rates, the map below shows a significant geographic
cluster of seven counties from Pitt to Cumberland. As other high-rate
counties are included (up to the highest 35 counties) , most are consistently
adjacent to this central cluster of seven counties in the east, making an even
stronger cluster of eastern counties with high adjusted mortality rates from
other ischemic heart disease. Only five of the 35 high-rate counties are in
the west.
OTHER ISCHEMIC HEART DISERSES
AGE-RACE-SEX ADJUSTED
MORTALITY RATES
PER 100,000 POPULATION
103.1 - 186.8 (p=.0075)
3 9944..22 - 110011..99 <(cpc==..0022))
[7] 87.8 - 94.0 (p <.01)
NORTH CAROLINA
RESIDENT DATA
1979 - 1981
0.0 - 87.
"
5-22
TABLE 5
MORTALITY STATISTICS FOR 1961
NORTH CAROLINA RESIDENTS
OTHER FORMS OF ISCHEMIC HEART DISEASE
GEOGRAPHICAL
MORTALITY STATISTICS FOR 1981. NORTH CAROLINA RESIDENTS
OTHER FORMS OF ISCHEMIC HEART DISEASE CONT'O.
LU
CO
cn
LU
CO
I—
I
Q
cc
CE
CJ
LU
zn
CD
CO
az
LU
en
in
LU
SQD
5-25
LU
CO
cn
LU
CO
cc
cc
LU
CJ
CJ
CO
cc
LU
o
cn
in
CC
ld cn oo oo
EQD
5-27
HYPERTENSION
Hypertension, or high blood pressure, is relatively common in the United
States today. It was estimated in the 1971-74 period that nearly 1 out of 5
Americans ages 18-74 had elevated blood pressure (7). This estimate does not
include persons who had their hypertension under control through the use of
medication or diet, nor does it include persons with borderline hypertension
or institutionalized persons with hypertension. In each of the years 1976
through 1979 it was estimated through the North Carolina Citizen Survey that
close to 20 percent of North Carolina adults had been diagnosed by a doctor as
having hypertension (8) . Hypertension is more prevalent among blacks and more
frequently found among older people than young people. At the younger ages,
the prevalence of hypertension is higher among men, whereas at older ages it
is higher for women, with the crossover occurring at earlier ages for blacks
(7).
In 1970 the corrected (for reasons described in Section II) North
Carolina death rate for hypertension as the underlying cause was 6.2 per
100,000 population, while in 1981 this rate was 4.0, indicating a substantial
decrease over the decade. Part of this decrease could, however, reflect
changes in death certificate coding practices.
The following table compares the age-adjusted underlying cause
hypertension death rate for four race-sex groups between the pariods 1974-78
and 1979-81. Tha earlier rate has been adjusted for both changes in the ICD
and changes in population bases to correct for underestimates. These rates
also show a decline over time, but what stands out are the differences among
5-29
1974-78
death certificate is unknown. Of these 4,310 deaths with mention of
hypertension, 60.1 percent had heart disease as the underlying cause of death,
17.3 percent had cerebrovascular disease as the underlying cause, and 3.3
percent had diabetes. Around 15 percent of all deaths attributed to heart
disease, cerebrovascular disease, and diabetes had a mention of hypertension.
Nonwhites had roughly twice as many death certificate mentions of hypertension
per 100,000 population as whites.
Risk Factors
People are often unaware that they have high blood pressure because there
are no distinctive physical or psychological symptoms. The cause of the most
common form of hypertension is basically unknown, though smoking, high sodium
diets, family history, obesity, age, physical inactivity, and stress are known
to be contributing factors. Blacks are more prone to hypertension than
whites, but diet, stress, and other factors may underlie this racial
association. It has been demonstrated that treatment and control of existing
high blood pressure reduces the risk of subsequent stroke as well as heart and
renal failure (1)
.
The risk of hypertension is also associated with certain characteristics
of a modern society. "The average blood pressure trend by age in different
societies is roughly proportional to their degree of modern social
transformation, in particular to the elevation of disruptive migration, labor
market competition/unemployment, and overwork characteristic of capitalist
modernization. Members of untouched primitive societies or traditional
hierarchical agricultural civilizations have blood pressures at low levels
through the life span; those in developed societies have sharply rising
average blood pressure trends by age which diverge from primitive levels
5-31
during adolescence or labor market entry ages" (9). Within the constraints of
a modern society. North Carolinians must concentrate on lifestyle and medical
strategies for reducing hypertension.
Geographic Patterns
No attempt is made here to assess county-by-county geographic variations
in the underlying hypertension death rate because the number of such deaths in
the 1979-81 period averages only 7.1 per county and at least 20 deaths are
needed to produce a reasonably stable rate. Data from the section on
"Multiple Conditions Present at Death" do suggest that the eastern region of
the state has a substantially higher rate of hypertension mentions than other
areas of the state. Since this region has a large nonwhite population, and
the prevalence of hypertension is higher among nonwhites, this rate is
expected to be somewhat higher. But even the race-specific rates in the
eastern region are considerably higher than for North Carolina as a whole, and
so the higher overall rates appear to be due also to a higher risk of
hypertension in eastern North Carolina, though it is possible that death
certificate recording practices contribute to this difference.
In the section of this volume on "Multiple Conditions Present at Death,"
there are county maps showing 1979-81 hypertension mentions per 100,000
population, with an associated analysis. An average of 127 hypertension
mentions per county in the 1979-81 period makes these rates much more stable
than those for hypertension as the underlying cause, though differences in
physician coding practices may complicate this county-by-county assessment.
5-32
TABLE 6
MORTALITY STATISTICS FOR 1981
NORTH CAROLINA RESIDENTS
HYPERTENSION WITH OR WITHOUT RENAL DISEASE
GEOGRAPHICAL
MORTALITY STATISTICS FOR 1981. NORTH CAROLINA RESIDENTS
HYPERTENSION WITH OR WITHOUT RENAL OISEASE CGNT'O.
COUNTIES
(CONT'D)
42 HALIFAX
43 HARNETT
44 HAYWOOD
45 HENDERSON
46 HERTFORD
47 HOKE
46 HYDE
49 IREDELL
50 JACKSON
51 JOHNSTON
52 JONES
53 LFE
54 LENOIR
55 LINCOLN
56 MCDOWELL
57 MACON
56 MADISON
59 MARTIN
60 MECKLENBURG
61 MITCHELL
62 MONTGOMERY
63 MOORE
64 NASH
65 NEW HANOVER
66 NORTHAMPTON
67 ONSLOW
66 ORANGE
69 PAMLICO
70 PASOUOTANK
71 PENDER
72 PERQUIMANS
73 PERSON
74 PITT
75 POLK
76 RANDOLPH
77 RICHMOND
76 ROBESON
79 ROCKINGHAM
80 ROWAN
61 RUTHERFORD
8? SAMPSON
83 SCOTLAND
64 STANLY
65 STOKES
66 SURRY
87 SWAIN
68 TRANSYLVANIA
89 TYRRELL
90 UNION
91 VANCE
92 WAKE
93 WARREN
94 WASHINGTON
95 WATAUGA
96 WAYNE
97 WILKES
96 WILSON
99 YADKIN
100 YANCEY
NUMBER
CEREBROVASCULAR DISEASE
Cerebrovascular disease, or stroke, claimed the lives of 4,717 North
Carolinians during 1981. Tt ranked as the thir^ leadinq cause of death in
North Carolina with -in overall rate of 79.1 per 100,000 population. The North
Carolina unadjusted rate was 10 percent higher thin the U.S. rate, but its
age-adjusted rate was 27 percent higher which indicates that North Carolina
would have a considerably higher stroke death rate if its ^ge structure were
1 ike that of the U.S.
The state's 1981 cerebrovascular mortality rate was the lowest since
194°). Reductions in the rates have occurred for all four rac-^-sex groups.
Nonwhite males and females, who had the highest rates during the 1979-31
period, experienced th° smallest, percent reductions in rates from the 1974-78
period. The 1979-H1 adjusted rates for the race-sex qroups were 50.1 for
white males, 39.5 for white females, 98.2 for nonwhite males, and 66.3 for
nonwhite females. Tn this period, nonwhite m^les had the highest risk of
mortality from stokes for all ages under 85. For th 3 85-and-over aqo group,
white females and white males had the highest rates.
Risk Factors
The risk factors associated with cerebrovascular disease are essentially
the s~me as those for heart disease: high cholesterol, high blood pressure,
atherosclerosis, heavy smoking, diabetes, lack of exercise, stress, and age
(1,3, A, 6). Consistent with higher rates of hypertension and atherosclerosis
mortality for nonwhites, nonwhites suffer a higher incidence of stroke
mortality and at a mu^h earlier ag-^ (10).
5-35
Factors possibly associated with reduced cerebrovascular and other
cardiovascular mortality are improved patient and physician education a.nd
increased availability of medical services including rescue squads, coronary
car-"1 units, and physician supply (4).
Geographic Patterns
Figure 7. A shows a scattering of counties with relatively high unadjusted
rates (Leve's 4-6) and several pockets of high-rate counties, primarily in the
east. After adjusting for age, race, and sex (Figure 7.B) , even more counties
appear in the high-rate group. Further, these high-rate counties are
concentrated in both the central and eastern portions of the state, although
th^re is an apparent band of inordinately high rates (Levels 5-6) extending
from Halifax down to New Hanover County and across to Richmond.
Based on procedures described in Section II, the eastern pockets of high
unadjusted county rates were found to cluster geographically (see unadjusted
map on next page). Spatial clustering continues to prevail in the east after
adjusting for age, race, and sex (see adjusted map on next page). Thus, it is
:1< ar that some eastern North Carolina counties are experiencing inordinately
high cerebrovascular mortality that cannot be explained by age, race, and sex
factors.
5-36
CERbBRUVrr CULAR Dl! EASE
MURrnu ry r.rtes
PER 100.000 PuPULflTluN
116.2 - 176.4 (p=.003)
!S 107.2 - 115.3 (p=.007)
fj 0.0 - 107.1
NORTH CAROLINA
RESIDENT DATA
1979 - 1981
CEREBROVASCULAR DISEASE
AGE-RACE-SEX ADJOSTED
MORTALITY RATES
PER 100.000 POPULATION
110.2 - 132.9 (p=.033)
gj 100.5 - 106.5 (p=.0025)
91.5 - 100.0 (p=.001)
fj 0.0 - 91.4
NORTH CAROLINA
RESIDENT DATA
1979 - 1981
5-37
GEOGRAPHICAL
AREA
T A B L t 7
MORTALITY STATISTICS FOR 19ei
NORTH CAROLINA RESIOENTS
CEREBROVASCULAR DISEASE
NUMBER DEATH
OF DEATHS RATE*
1979-81 1979-61
NUMBER
MORTALITY STATISTICS FOR 1981, NORTH CAROLINA RESIDENTS
CEREBROVASCULAR DISEASE CONT'D,
COUNTIES
ICONT'D)
42 HALIFAX
43 HARNETT
44 HAYWOOD
4b HENDERSON
46 HERTFORD
47 HOKE
46 HYDE
49 IREDELL
50 JACKSON
51 JOHNSTON
52 JONES
53 LEE
54 LENOIR
55 LINCOLN
56 MCDOWELL
57 MACON
58 MADISON
59 MARTIN
60 MECKLENBURG
61 MITCHELL
62 MONTGOMERY
63 MOORE
64 NASH
65 NEW HANOVER
66 NORTHAMPTON
67 ONSLOW
68 ORANGE
69 PAMLICO
70 PASQUOTANK
71 PENDER
72 PEROUIMANS
73 PERSON
74 PITT
75 POLK
76 RANDOLPH
77 RICHMOND
78 ROBESON
79 ROCKINGHAM
80 ROWAN
81 RUTHERFORD
82 SAMPSON
83 SCOTLAND
84 STANLY
8b STOKES
86 SURRY
87 SWAIN
88 TRANSYLVANIA
89 TYRRELL
90 UNION
91 VANCE
92 WAKE
93 WARREN
94 WASHINGTON
95 WATAUGA
96 WAYNE
97 WILKES
98 WILSON
99 YADKIN
100 YANCEY
NUMBER
CO
CE
UJ
CO
I—
I
Q
CC
en
CJ
CO
CE >o
cc
CD
UJ
CC
UJ
CJ
cr
CD
EED
5-41
UJ
CO
ac
LU
CO
i—
i
Q
CC
cc
CJ>
CO
cr >o
CC
CO
LU
CC
LU
CJ
CD
O
LxJ
CC
ZD
CD
J 00
ATHEROSCLEROSIS
Atherosclerosis (called arteriosclerosis in this publication before 1979)
is a general term covering a number of diseases of the blood vessels and is
often called "hardening of the arteries." Like hypertension, atherosclerosis
is listed as a contributing cause on the death certificate far more often than
it is assigned as the underlying cause of death. In particular,
atherosclerosis contributes to deaths from heart disease, stroke, and
diabetes. It was estimated that in 1970, 87 percent of the more than 1
million deaths in the United States due to heart and blood vessel diseases
were attributable to atherosclerosis and its sequelae (1).
In 1970 the North Carolina death rate for atherosclerosis as the
underlying cause was 12.7 per 100,000 population, while in 1981 this rate was
9.7, indicating a 24 percent decline over the decade. Part of this change,
however, could be due to changes in death certificate coding practices. Also,
the implementation of a computerized system for assigning underlying cause in
1975 resulted in some deaths being assigned to ischemic heart disease that
would have formerly been assigned to atherosclerosis (11).
Comparisons of the age-adjusted atherosclerosis death rates for the four
race-sex groups between the periods 1974-78 and 1979-81 (the 1974-78 rates
were corrected for changes in the ICD and in population bases) shows a decline
over time for each group. White males had the largest percent decline
(30.4%), followed by white females (22.7%), nonwhite females (16.4%), and
nonwhite males (9.4%). Noteworthy are the sex and race differentials. The
white male rate (5.5) was 25 percent higher than the white female rate (4.4)
5-45
in 1979-81 and the nonwhite male rate (8.4) was 38 percent higher than the
nonwhite female rate (6.1). Nonwhites had higher atherosclerosis (underlying
cause) death rates than whites, 53 percent higher for males and 39 percent
higher for females.
"The (underlying cause) death rate for atherosclerosis in the 1979-81
period showed a sharp increase with age for all four race-sex groups to a high
in the 85-and-over age group of 555 for white males, 547 for white females,
341 for nonwhite males, and 362 for nonwhite females. The higher overall rate
for males is due particularly to higher rates in the 35-74 age groups. The
higher overall rate for nonwhites is due to higher rates in the under-85 age
groups.
In 1981 there were 580 deaths with atherosclerosis as the underlying
cause, but over 14,700 death certificates had a mention of atherosclerosis.
This suggests that 30 percent of all North Carolina deaths in 1981 involved
atherosclerosis, though it is not known just how completely this condition is
recorded on the death certificate. Of these 14,723 deaths with mention of
atherosclerosis, 65 percent had heart disease as the underlying cause of death
and 14 percent had cerebrovascular disease listed as the underlying cause.
Around 50 percent of the deaths with heart disease or cerebrovascular disease
as the underlying cause had a mention of atherosclerosis. In 1978, white
males had the highest rate of mentions of atherosclerosis per 100,000
population (366) , with white females and nonwhite males close together at a
lower level (about 300), and nonwhite females much lower still (257). The
reasons for the difference between this race-sex pattern and that based on
underlying cause are not clear, though the rates for mentions of
atherosclerosis are based on about 25 times as many deaths as the rates using
underlying cause.
5-46
Risk Factors
Three major risk-treatable factors contributing to atherosclerosis have
been identified: elevated blood cholesterol levels, high blood pressure, and
cigarette smoking. A number of additional factors are also recognized, such
as diabetes, physical inactivity, obesity, age, male sex, and certain
personality types, that is, "type A" or coronary-prone behavior (1). While
older males generally have the highest atherosclerosis death rates, women and
young people are also at risk of this disease. The theory that
atherosclerosis begins during childhood and continues to develop through
adulthood has begun to receive considerable attention (2).
Increasing epidemiological evidence supports the hypothesis that higher
concentrations of high-density lipoproteins (HDL's) may be a protective factor
in the development of atherosclerosis. "Levels of HDL have been correlated
positively with exercise and moderate ingestion of alcohol and inversely
related to obesity, smoking, poor control of diabetes, and the use of
progest in-containing contraceptives" (1). The cause and effect of the inverse
relationship between HDL and atherosclerosis, however, remain unclear.
Geographic Patterns
Upon examining geographic variation in the atherosclerosis (underlying
cause) death rate using the clustering technique described in Section II, a
significant cluster of counties with high adjusted rates in southcentral North
Carolina was revealed. Richmond, Moore, Scotland, Hoke, Robeson, and
Cumberland counties, all contiguous, had some of the highest 1979-81 age- race-sex-
adjusted rates in the state. Nearby counties that also had high adjusted
rates are Rowan, Union, and Brunswick. This apparent pattern should be
considered very cautiously since the numerators of some of these rates are
5-47
very small. Consequently, it was decided to show the two major maps on
atherosclerosis using mentioned conditions rather than underlying cause data.
These maps, along with an associated analysis, are presented in the Multiple
Conditions Section (Section X) and show 1979-81 atherosclerosis mentions per
100,000 population.
ATHEROSCLEROSIS
AGE-RACE-SEX ADJUSTED
MORTALITY RATES
PER 100,000 POPULATION
NORTH CAROLINA
RESIDENT DATA
1979 - 1981
B| 15.6 - 33.8 (p=.007)
0.0 - 15.5
5-48
TABLE 8
MORTALITY STATISTICS FOR 1981
NORTH CAROLINA RESIDENTS
ATHEROSCLEROSIS
GEOGRAPHICAL
AREA
NORTH CAROLINA
REGIONS
NUMBER
MORTALITY STATISTICS FOR 1981, NORTH CAROLINA RESIDENTS
ATHEROSCLEROSIS CONT'D.
COUNTIES
(CONT'D)
42 HALIFAX
43 HARNETT
44 HAYWOOD
45 HENDERSON
46 HERTFORD
47 HOKE
48 HYDE
49 IREDELL
50 JACKSON
51 JOHNSTON
52 JONES
53 LEE
54 LENOIR
55 LINCOLN
56 MCDOWELL
57 MACON
56 MADISON
59 MARTIN
60 MECKLENBURG
61 MITCHELL
62 MONTGOMERY
63 MOORE
64 NASH
65 NEW HANOVER
66 NORTHAMPTON
67 ONSLOW
68 ORANGE
69 PAMLICO
70 PASQUOTANK
71 PENOER
72 PER8UIMANS
73 PERSON
74 PITT
75 POLK
76 RANDOLPH
77 RICHMOND
76 ROBESON
79 ROCKINGHAM
60 ROWAN
81 RUTHERFORD
82 SAMPSON
83 SCOTLAND
84 STANLY
85 STOKES
86 SURRY
87 SWAIN
88 TRANSYLVANIA
89 TYRRELL
90 UNION
91 VANCE
92 WAKE
93 WARREN
94 WASHINGTON
95 WATAUGA
96 WAYNE
97 WILKES
96 WILSON
99 YADKIN
100 YANCEY
NUMBER
REFERENCES FOR SECTION V
1. Levy, R.I.; Moskowitz, J. "Cardiovascular Research: Decades of
Progress, A Decade of Promise," Science . Volume 217, July 9, 1982, pp.
121-129.
2. United States Department of Health and Human Services, Public Health
Service, National Center for Health Statistics. "Annual Summary of
Births, Deaths, Marriages, and Divorces: United States 1981," Monthly
Vital Statistics Report . (PHS) 83-1120, Volume 30, Number 13,
Hyattsville, Maryland, December 20, 1982.
3. United States Department of Health and Human Services, Public Health
Service, National Center for Health Statistics. "Dietary Intake and
Cardiovascular Risk Factors, Part I - Blood Pressure Correlates: United
States 1971-75," Vital and Health Statistics . Series 11, Number 226,
DHHS Pub. No. (PHS) 83-1676, U.S. Government Printing Office, Washington,
D.C., February 1983.
4. Tyroler, H.A.; Haynes, S. "Community Cardiovascular Surveillance
Program: Phase II - Protocol ," Unpublished Paper, Department of
Epidemiology, School of Public Health, UNC-Chapel Hill, N.C., April 1982.
5. Paffenbarger, R.S.; Brand, R.J.; Sholtz, R.I.; Jung, D.L. "Energy
Expenditure, Cigarette Smoking, and Blood Pressure Level As Related to
Death from Specific Diseases," American Journal of Epidemiology . Volume
108, Number 1, 1978, pp. 12-18.
6. Laporte, R.E. ; Cresanta, J.L. ; Kuller, L. H. "The Relation of Alcohol to
Coronary Heart Disease and Mortality: Implications for Public Policy,"
Journal of Public Health Policy . September 1980, pp. 198-223.
7. Fingerhut, L.A.; Wilson, R.W. ; Feldman, J.J. "Health and Disease in the
United States," Annual Review of Public Health 1980 . Number 1, pp. 1-36.
8. North Carolina Department of Health and Human Resources, Division of
Health Services, State Center for Health Statistics. "Update on Health
Characteristics of Adult Residents of North Carolina's Health Service
Areas," SCHS Studies . Number 19, November, 1980.
9. Eyer, J. "A Diet/Stress Interaction Hypothesis of Coronary Heart Disease
Epidemiology," International Journal of Health Services . Volume 9,
Number 1, 1979, pp. 161-168.
10. North Carolina Department of Human Resources, Division of Health
Services, State Center for Health Statistics. "Wide Gaps in Mortality
Risk: Comparisons Among Race-Sex Groups Across Time and Space
Dimensions," SCHS Studies. Number 18, Raleigh, August, 1980.
5-51
11. U.S. Department of Health and Human Services, Public Health Service,
Office of Health Research, Statistics, and Technology, National Center
for Health Statistics. Vital Statistics: Instructions for Classifying
Multiple Causes of Death, 1981 , Part 2b' Hyattsville, Maryland, November
1980.
5-52
VI. Cancer Mortality
CANCER
Cancer is the secord most frequent cause of death in the united States,
and unlike mortality from other major diseases, cancer mortality continues to
increase. As a result, beginning in 1977, North Carolina's cancer death toll
has exceeded the toll from heart attack and in 1981 was 24 percent higher.
But despite appearances to the contrary, cancer is not entirely a product
of modern times. It has been reported that, even by the 7th century, the
origin of its name—"cancer," the crab—had al ready been lost in antiquity.
In fact, based on findings of large tumorous messes extruding from the
petrified bones of unearthed dinosaur remains, cancer seems to have predated
man himsel f . (1)
Still, there continues to accumulate an overwhelming body of statistical
evidence that modern-day 1 ifestyl es and technologies are exacerbating the
cancer problem. While the exact patho-physiological mechanisms of the disease
remain largely unknown, a number of external chemical and physical agents
known as carcinogens have been implicated.
In the meantime, North Carolina's health community is treating cancer as
a major public health issue as evidenced by not only advanced research and
treatment in our medical institutions but also the Division of Health
Services' involvement in a variety of prevention, diagrosis/treatment , and
epidemiologic research efforts:
• Available in most local health departments and in 25 comprehensive
screening centers across the state, cancer screening services consist
of Pap smears, blood tests, urinalysis, X-ray, and patient histories.
Educational information, e.g., how to conduct breast self-examination,
is also provided.
6-3
• Ihe Mult Health Section administers a financial aid program for
cancer diagnosis and treatment of lower-income patients not receiving
financial aid from any other source.
• To aid in evaluating treatment practices, the Adult Health Section
maintains the North Carolina Cancer Registry which collects diagnosis,
treatment, and follow-up information on cancer patients treated in
participating hospitals. Unfortunately, hospital participation is or,
a voluntary basis.
• The Epidemiology Section is developing an environmental epidemiology
program that conducts investigations of possible carcinogens in high-risk
areas of the state.
e The State Center for Health Statistics reports county, regional,
state, and nationwide statistical information gleaned from death
certificates, hospital discharge records, household surveys, and
epidemiologic investigations.
Clinical arid Epidemiologic Research
Recent discoveries have spurred hope that cancer researchers may soon
find ways to prevent or effectively treat the dread disease. Cne discovery is
that normal human cells contain genes that have potential oncogenic activity.
The hope is that, by learning what makes such genes trigger cancer,
researchers can find ways to prevent it or at least to identify and stop the
process once it's begun. Cooper (2) suggests, however, that, ". . . it is
likely that activation of these genes is only one of several events involved
in carcinogenesis." Further, Rubin (3) warns that explanations for the origin
of cancer have been varied and plentiful in this century and the rush to
accept the "oncogene" explanation is "at best premature."
A discovery about 10 years ago also provides impetus for current cancer
research—that of natural killer (NK) cells that have the spontaneous ability
to destroy tumor cells as well as cells infected with viruses. Since
reactivity of those cells can be rapidly augmented by interferon, it is
conceivable that through immunoprevent ion or immunotherapy, NK cells might
6-4
eventually be harnessed as a viable cancer fighter. Herberman notes, however,
that there remain several practical problems in conducting experiments to
substantiate a major role for NK cells in resistance against tumor growth. (4)
Those problems notwithstanding, several sponsors including the American
Cancer Society (ACS) continue to support interferon research in this country.
In addition to those particular studies, the ACS funds research in the broad
areas of genetic engineering, man-made antibodies, the mechanisms of
carcinogensis, and chemopr event ion, among others. As a result, $62.7 million
or 31 percent of the ACS's 1981/82 budget was expended for research. (5)
Concurrently, it has been estimated that 70-90 percent of human cancers
are likely spawned by environmental factors (6), and both human and animal
studies continue to contribute evidence involving the following categories of
agents (7):
- Consumables: tobacco, alcohol, food including coffee, saccharin, and
other nor.nutr ients, drugs such as antineoplastic agents and the
estrogens, and cosmetics including hair dye, alcoholic mouthwash, and
tal cum powder ;
- Contaminants: asbestos, compounds of certain metals, various organic
chemicals, various radionuclides, wood dusts, air, water, man-made
food additives, biologic agents, and physical agents including
ultraviolet and ionizing radiation.
Numerous epidemiologic studies have ascribed carcinogenic qualities to
the above agents while producing equivocal results regarding some of those as
well as other agents in the environment. In order to come to grips with the
resulting confusion and conflicting information, the American Cancer Society
is undertaking a six-year comprehensive survey of more than 1 million
Americans who will be studied for factors related to personal habits,
lifestyle, and environmental conditions. The society will correlate these
factors with disease incidence and report findings late in the decade. (8)
6-5
In the meantime, the work of epidemiologists has been graphically
described by Shod el 1 (1): "Epidemiologists are like bookies of disease,
stalking the globe to determine point-spreads on which groups of people are
most likely to get which disease." As a result of this work, a world cancer
map is emerging. The patterns reveal that all peoples are susceptible to
cancer but with widely varying propensities for specific types. In the U