Original PDF Flash format national-vital-statistics-reports;-vol-47-no.-3-(10/7/98)  


National Vital Statistics Reports; Vol 47 No. 3 (10/7/98)

National Vital Statistics Reports
From the CENTERS FOR DISEASE CONTROL AND PREVENTION
National Center for Health Statistics
National Vital Statistics System
Volume 47, Number 3
October 7, 1998
Age Standardization of Death Rates:
Implementation of the Year 2000 Standard
by Robert N. Anderson, Ph.D., and Harry M. Rosenberg, Ph.D.
Abstract
comparisons of U.S. mortality given the aging of the U.S. population (1).
The crude death rate for the United States rose from 852.2 per 100,000
This report discusses the rationale for and implications of the
population to 880.0 during 1979–95. This increase in the crude death
implementation of a new population standard for the age standardiza-
rate was due to the increasing proportion of the U.S. population in older
tion (age adjustment) of death rates. The new standard is based on
age groups that have higher death rates. Age standardization, often
the year 2000 population and beginning with data year 1999, will
called ‘‘age adjustment,’’ is one of the key tools used to control for the
replace the existing standard based on the 1940 population. This
changing age distribution of the population, and thereby to make
report also includes a technical discussion of direct and indirect
meaningful comparisons of vital rates over time and between groups.
standardization and statistical variability in age-adjusted death rates.
In contrast to the rising crude death rate, the age-adjusted death rate
Currently, at least three different standards are used among Depart-
for the United States dropped from 577.0 per 100,000 U.S. standard
ment of Health and Human Services agencies. Implementation of the
population to 503.9 during 1979–95. This age-adjusted comparison is
year 2000 standard will reduce confusion among data users and the
free from the confounding effect of changing age distribution and
burden on State and local agencies. Use of the year 2000 standard
therefore, better reflects the trend in U.S. mortality. To use age
will also result in age-adjusted death rates that are substantially larger
adjustment requires a ‘‘standard population,’’ which is a set of arbitrary
than those based on the 1940 standard. Further, the new standard will
population weights (see ‘ Methods’’).
affect trends in age-adjusted death rates for certain causes of death
Since 1943 NCHS and the States have used a standard based on
and will narrow race differentials in age-adjusted death rates. Although
the 1940 U.S. population termed the ‘‘U.S. standard million population’’
age standardization is an important and useful tool, it has some
for age-adjusting rates. Although the 1940 standard is widely used, at
limitations. As a result the examination of age-adjusted death rates
least three different standards are currently used by Federal and State
should be the beginning of an analysis strategy.
agencies (2,3,4).
It has been recognized that the use of a single age-adjustment
Introduction
standard by Federal agencies would help to alleviate confusion and
misunderstanding among data users and the media. Multiple standards
The purpose of this report is to provide the rationale for and the
also create burdens for the States, who attempt to make their data
implications of implementing a new population standard for age-
consistent with Federal statistics. In recent years the 1940 standard has
adjusting death rates. Based on the year 2000 population, the new
been perceived as outdated and incompatible with the current and
standard replaces the existing 1940 standard million population that
‘‘older’ age structure of the population. NCHS sponsored two national
has been used for over 50 years. The change will be implemented by
workshops (1991, 1997) to examine these issues. Participants of the
the National Center for Health Statistics (NCHS), effective with deaths
occurring in 1999. This report also includes a technical discussion of
direct and indirect standardization and statistical variability in age-
adjusted death rates (see ‘ Technical notes’’).
Acknowledgments
The crude death rate is a widely used measure of mortality.
This report was prepared in the Division of Vital Statistics. Lester R. Curtin
However, crude death rates are influenced by the age composition of
and Van L. Parsons of the Office of Research and Methodology and Jeffrey D.
the population. As such, comparisons of crude death rates over time
Maurer of the Mortality Statistics Branch contributed to the ‘ Technical
or between groups may be misleading if the populations being com-
notes.’’ This report was edited by Patricia Keaton-Williams and typeset by
Jacqueline M. Davis of the Publications Branch, Division of Data Services.
pared differ in age composition. This is relevant, for example, in trend
U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICES
Centers for Disease Control and Prevention
National Center for Health Statistics
CENTERS FOR DISEASE CONTROL
AND PREVENTION
2
National Vital Statistics Report, Vol. 47, No. 3, October 7, 1998
first workshop on age adjustment examined technical issues and prob-
size. Consequently, to compare differences in mortality among groups
lems related to the calculation and interpretation of age-adjusted death
or across time periods, the number of events must be related to the
rates. Participants included representatives from NCHS, other compo-
size of the ‘‘population at risk’’ of experiencing the event. In this way,
nents of the Centers for Disease Control and Prevention, the National
one can compare the relative risk of death between groups or time
Institutes of Health, the National Academy of Sciences, State health
periods. The most informative method of making comparisons of
departments, and academia (5). Recommendations were made to
mortality risk between groups is to examine differences in age-specific
continue the use of the 1940 standard by NCHS, to encourage other
death rates. The age-specific death rate is defined as the number of
Federal and State governmental agencies to use this standard when
deaths occurring in a specified age group divided by the midyear
publishing official mortality statistics, and to have NCHS study issues
population of that age group, usually expressed per 1,000 or 100,000
that might lead to the introduction of a new or additional standard by
population. Age-specific death rates allow one to compare mortality
the year 2000 (6).
risk among groups or over time specific for a particular age group.
The second workshop focused on policy issues related to a
Although effective in eliminating the effect of age composition,
coordinated approach to age standardization within the Department of
age-specific comparisons can be cumbersome, because they require
Health and Human Services (DHHS) (7). Workshop participants con-
a relatively large number of comparisons, one for each age group.
cluded that although there were no compelling technical reasons to
The crude death rate is a summary measure or average defined
change population standards, the public health community might be
as the total number of deaths divided by the total midyear population
better served by a new, uniform, and more contemporary standard.
and is often expressed per 1,000 or 100,000 population. Although it is
Additionally, workshop participants recommended that the new standard
the simplest way to express relative mortality risk, the crude death rate
for age-adjusting rates be based on the year 2000 U.S. population (see
is often inadequate because many health outcomes such as death vary
‘‘Technical notes’ for the recommendations of the second workshop).
substantially by age. Because the risk of dying is much greater at older
These recommendations were subsequently approved as policy by the
than at younger ages, populations with older age distributions tend to
Secretary, DHHS.
have higher crude death rates than younger populations. Table A illus-
trates a comparison between two hypothetical groups of the same
Methods
population size, 10,000 persons, but with different age compositions and
different age-specific death rates. Group A has proportionately more
Data
elderly persons; sixty percent are aged 65 years and over (column 2)
compared with 10 percent for group B (column 6). Further age-specific
Mortality data in this report are from annual statistical files of the
death rates for each age group in group A are lower than those for group
National Vital Statistics System, which is a compilation of statistics
B. Group A has a crude death rate of 50 deaths per 1,000 population
from all death certificates filed in the 50 States and the District of
(column 3 all ages), and group B has a crude death rate of 40 deaths
Columbia (8). The projected population age distribution for the year
per 1,000 population (column 7 all ages).
2000 standard was prepared by the U.S. Bureau of the Census (9)
At first glance the relative risk of mortality appears to be greater
and converted by NCHS to a standard million population by dividing
for group A than for group B. However, close examination shows that
the age-specific populations by the total population and multiplying by
the substantially older age distribution of group A resulted in the higher
1 million (see ‘ Technical notes’’).
crude death. An examination of the age-specific death rates in table A
shows that death rates increase sharply with age, and that the age-
Death rates
specific rates of group B are higher (column 7) than those of group A
The burden of disease in a population is typically denoted by the
(column 3) at every age. Contrary to what the comparison of crude death
total number of health events (such as deaths). However, the absolute
rates revealed, the mortality risk was higher for each age group in group
number of events is seldom useful for making comparisons between
B than in group A. Thus, to make meaningful comparisons of mortality
groups or examining changes over time, because it depends largely
risk between the two groups, the effect of variation in the age distribution
on population size. That is, a large population tends to generate more
between groups (or time periods) must be taken into account.
health events than a smaller population simply because of its larger
Table A. Group comparison of crude and age-adjusted death rates
Group A
Group B
Weighted
Weighted
Standard
Standard
Deaths
Population
Rate1
rate2
Deaths
Population
Rate1
rate2
population
weight3
Age
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
All ages . . . . . . . . . . . . . . . . . . .
500
10,000
50
. . .
400
10,000
40
. . .
10,000
1.0
0–24 years . . . . . . . . . . . . . . . . .
20
1,000
20
6
180
6,000
30
9
3,000
0.3
25–64 years . . . . . . . . . . . . . . . . .
120
3,000
40
12
150
3,000
50
15
3,000
0.3
65 years and over . . . . . . . . . . . . .
360
6,000
60
24
70
1,000
70
28
4,000
0.4
Age-adjusted death rate1 . . . . . . . . .
. . .
. . .
. . .
42
. . .
. . .
. . .
52
. . .
. . .
. . . Category not applicable.
1Rate per 1,000 population.
2The weighted rate is calculated by multiplying the age-specific rate by the standard weight.
3The standard weight for each age group is calculated by dividing the standard population at each age by the total standard population.
National Vital Statistics Report, Vol. 47, No. 3, October 7, 1998
3
To overcome the effect of population age composition on com-
Second, age standardization may mask important information if the
parisons of crude death rates, as well as the unwieldiness of multiple
age-specific rates in the populations being compared do not have a
comparisons of age-specific death rates, a summary measure of mor-
consistent relationship (13, 16). This problem arises in cancer mortality.
tality risk that controls for variation in age distributions was needed. The
Table B shows 1979–95 cancer death rates for three broad age groups
age-adjusted death rate is such a summary measure. Age-adjusted
and the age-adjusted death rate. The trend in the age-specific death
rates were first used in 1841 for the analysis of mortality data (10). The
rate for the youngest age group (0–24 years) decreased by 33 percent
age-adjusted death rate is defined as the death rate that would occur
during this period, while the rate for the oldest age group (65 years and
if the observed age-specific death rates were present in a population
over) increased by 15 percent. In contrast, the age-adjusted death rates
with an age distribution equal to that of a standard population. The
(based on the 1940 standard) changed very little. Thus, the trend in the
age-adjusted death rate is typically computed by the method of direct
age-adjusted death rate for cancer does not reflect the complexities in
standardization. Indirect standardization may also be used to control for
the underlying age-specific rates. As averages, age-adjusted rates, like
differences in age composition (see ‘ Technical notes’’). However, direct
other averages, may lose information shown in their components,
standardization is the most widely used method. The age-adjusted
especially when age-specific rates reflect divergent trends over time.
death rate computed by the direct method is a weighted average of the
More often, however, age-specific rates move roughly in parallel. Thus,
age-specific death rates. The weights represent standard population
age-adjusted death rates are a widely accepted and useful convention
proportions by age and are applied to the age-specific death rates of
for analyzing trends (17–19). Age-adjusted death rates are also highly
each comparison group or time period (11). In table A the standard
effective for making comparisons among population groups (18) and
population of 10,000 is shown in column 9 and as relative weights
among geographical areas (20) because age distribution often varies
summing to 1.0 in column 10. The age-adjusted death rate is calculated
substantially between such comparison groups. In sum, thorough mor-
by multiplying each age-specific rate (columns 3 and 7) by the standard
tality analyses should include examination of age-adjusted rates as well
weight (column 10) and summing the weighted age-specific death rates
as age-specific rates. In cases where age standardization may mask
(see ‘ Technical notes’’). Because each group or time period shares a
important age-specific trends or differences, presentation of age-
common age distribution represented by the age-specific standard
adjusted rates should be supplemented with age-specific rates.
population weights, the effects of variation in age distribution are
eliminated. In table A the age-adjusted death rate of group A is 42
Effects of changing to the year 2000 standard
(column 4) compared with 52 for group B (column 8), reflecting the effect
Changing from the 1940 standard population to the year 2000
of the lower age-specific rates at each age in group A. Thus, group B
standard will affect the magnitude of age-adjusted death rates, and in
has a higher relative mortality risk measured by the age-adjusted death
some cases, trends in mortality. This is because the age structures of
rate than group A, despite having a lower crude death rate.
the 1940 and year 2000 populations differ. From 1940 to year 2000,
Age adjustment by the direct method requires a standard age
the U.S. population ‘‘aged’’ considerably. This occurred for two
distribution or ‘‘standard population.’’ Selection of an appropriate stan-
reasons: Fertility declined and age-specific death rates declined,
dard population is to some extent arbitrary because no ‘‘correct’’ stan-
particularly among the elderly population, resulting in greater survival
dard population exist, although there are statistical reasons to guide the
at older ages. Figure 1 shows population pyramids for the 1940 U.S.
selection of a standard (12–14). The principal guidance in the statistical
literature is that the standard population selected should not be con-
Table B. Age-specific, crude, and age-adjusted death
sidered ‘‘abnormal’’ relative to the populations being studied. That is,
rates for cancer: United States, 1979–95
the standard population should reflect a reasonable age distribution
[Age-specific rates per 100,000 population in specified age group; age-adjusted rates
(15). The selection of the standard population will not substantially affect
per 100,000 standard population (1940); cancer includes Ninth Revision, International
comparisons among groups or time periods if the age-specific rates in
Classification of Diseases, 1975 categories 140–208, including malignant neoplasms
the populations being compared have a roughly consistent relationship.
of lymphatic and hematopoietic tissues]
That is, the relative differences are constant from one age group to the
Age-specific rates
next. If the age-specific rates are not consistent, comparisons will be
0–24
25–64
65 years
Age-adjusted
dependent on the standard selected. The most commonly used stan-
Year
years
years
and over
rate
dard in the United States is based on the age distribution of the 1940
U.S. population (8). This standard is expressed in terms of a ‘‘standard
1979. . . . . . . . . . . .
5.1
142.7
986.6
130.8
1980. . . . . . . . . . . .
5.2
142.9
1,011.3
132.8
million’’ (i.e., the relative age distribution of the 1940 population of the
1981. . . . . . . . . . . .
4.8
139.4
1,008.6
131.7
United States totaling 1 million) in 10-year age groups.
1982. . . . . . . . . . . .
5.0
138.4
1,023.9
132.8
Two important caveats apply when age adjusting rates. First,
1983. . . . . . . . . . . .
4.8
136.6
1,034.0
133.1
1984. . . . . . . . . . . .
4.4
136.2
1,045.2
134.1
considered alone, the age-adjusted death rate does not reflect the
1985. . . . . . . . . . . .
4.4
134.6
1,051.1
134.4
mortality risk of a ‘‘real’’ population. The average risk of mortality of a
1986. . . . . . . . . . . .
4.3
131.0
1,062.4
134.2
real population is represented by the crude death rate. The numerical
1987. . . . . . . . . . . .
4.1
128.8
1,067.3
134.0
1988. . . . . . . . . . . .
4.0
127.0
1,076.3
134.0
value of an age-adjusted death rate depends on the standard used and,
1989. . . . . . . . . . . .
4.0
124.5
1,095.9
134.5
therefore, is not meaningful by itself. Age-adjusted death rates are
1990. . . . . . . . . . . .
3.9
123.4
1,111.3
135.0
appropriate only when comparing groups or examining trends across
1991. . . . . . . . . . . .
3.9
121.3
1,117.3
134.5
1992. . . . . . . . . . . .
3.8
118.6
1,121.8
133.1
multiple time periods (13). A comparison of age-adjusted death rates
1993. . . . . . . . . . . .
3.7
117.1
1,133.7
132.6
among groups or time periods does reflect differentials in the average
1994. . . . . . . . . . . .
3.6
115.6
1,134.5
131.5
risk of mortality.
1995. . . . . . . . . . . .
3.4
113.7
1,136.6
129.9
4
National Vital Statistics Report, Vol. 47, No. 3, October 7, 1998
Figure 1. Population pyramids for the 1940 and 2000 U.S. populations expressed as a percent of total population
population and the projected year 2000 U.S. population. The 1940
the year 2000 standard was 918.5, nearly double that based on the
population is more tapered, having a wider base and narrowed tip.
1940 standard and much closer in magnitude to the crude death rate.
The year 2000 population shows a higher concentration of population
Thus, the age-adjusted death rate based on the year 2000 standard
in the middle and older age groups, such as between 35 to 45 years
much more closely reflects the observed average risk of mortality in
of age and 65 years of age and over. The proportion of the population
1995—represented by the crude death rate—than the age-adjusted
for these age groups increased from 0.139 to 0.163 and 0.068 to
rate based on the 1940 population. The age-adjusted rate based on
0.124, respectively (table C). Thus, the population aged 65 years and
the year 2000 standard is larger because the year 2000 population
over essentially doubled during this period. Because the standard
standard, which has an older age structure, gives more weight than
populations serve as the weights for calculating age-adjusted rates,
the 1940 standard to death rates at the older ages where mortality is
the differences in the age structure of the populations between 1940
higher.
and year 2000 translate directly into a change in the weights used for
Although many studies have emphasized correctly that the choice
age standardization. Table C shows the 1940 and year 2000 standard
of standard makes relatively little difference in terms of the relative trend
populations with their corresponding age-specific weights. The differ-
ence in age distribution between the 1940 and year 2000 standards
Table C. The 1940 and year 2000 U.S. standard
has implications for the presentation and interpretation of mortality
populations
statistics regarding age-adjusted death rates.
1940
2000
Magnitude of the age-adjusted death rate
Age
Number
Weight
Number
Weight
Change in the population standard from 1940 to the year 2000
All ages . . . . . . . . . .
1,000,000
1.000000
1,000,000
1.000000
will affect the magnitude of the age-adjusted death rate for the United
Under 1 year. . . . . . .
15,343
0.015343
13,818
0.013818
1–4 years. . . . . . . . .
64,718
0.064718
55,317
0.055317
States. The rate based on the year 2000 standard will be much larger
5–14 years . . . . . . . .
170,355
0.170355
145,565
0.145565
than that based on the 1940 standard. As noted earlier, the magnitude
15–24 years . . . . . . .
181,677
0.181677
138,646
0.138646
of the age-adjusted rate is largely dependent on the age distribution of
25–34 years . . . . . . .
162,066
0.162066
135,573
0.135573
35–44 years . . . . . . .
139,237
0.139237
162,613
0.162613
the applied standard. Figure 2 shows the trend in mortality in terms of
45–54 years . . . . . . .
117,811
0.117811
134,834
0.134834
the crude death rate along with age-adjusted death rates based on
55–64 years . . . . . . .
80,294
0.080294
87,247
0.087247
the 1940 and the year 2000 standards. The 1995 age-adjusted death
65–74 years . . . . . . .
48,426
0.048426
66,037
0.066037
75–84 years . . . . . . .
17,303
0.017303
44,842
0.044842
rate based on the 1940 standard was 503.9 deaths per 100,000
85 years and over. . . .
2,770
0.002770
15,508
0.015508
standard population. The age-adjusted death rate for 1995 based on
National Vital Statistics Report, Vol. 47, No. 3, October 7, 1998
5
Figure 2. Crude and age-adjusted death rates based on the 1940 and 2000 standard populations: United States, 1979–95
(13,21), the choice can make a difference in some cases, when age-
standard population using the 1940 standard but is 63.9 using the
specific rates trace divergent trends, or when the age structure of the
year 2000 standard, a 2.4-fold difference. Large differences also occur
alternative standard populations differ. The crude death rate changed
for heart disease, malignant neoplasms (cancer), chronic obstructive
very little from 1979 to 1995, from a rate of 852.2 per 100,000 population
pulmonary disease, pneumonia and influenza, diabetes, nephritis
to 880.0, an increase of 3.3 percent. In contrast, during this period the
(kidney disease), septicemia, Alzheimer’s disease, and atheroscle-
age-adjusted death rate based on the 1940 standard decreased by
rosis. Age-specific death rates for all of these causes of death are
12.6 percent, from 577.1 per 100,000 standard population to 503.9,
higher in older age groups, and, as a result, these causes are more
while the rate based on the year 2000 standard decreased by 9.2 per-
affected by the larger weights of the year 2000 standard.
cent from 1011.1 to 918.5. Tracing a similar pattern over time (figure 2),
In contrast for those causes where risk is more uniform among the
the decline using the year 2000 standard is attenuated compared with
age groups, the differences in rates based on the two standards are
the decline using the 1940 standard because age-specific declines have
much smaller. These causes include accidents, Human immunodefi-
been smaller at the older ages, to which the year 2000 standard gives
ciency virus (HIV) infection, suicide, chronic liver disease, and homicide,
proportionately more weight than the 1940 standard. Nevertheless, the
which are more concentrated in the younger and middle-age groups and
trend lines in figure 2 for the age-adjusted rates based on the year 2000
consequently are much less affected by the disparity in weights between
standard and the 1940 standard are roughly parallel, showing that the
the two population standards.
decrease in the age-adjusted death rate from 1979 to 1995 is similar
Choice of the age standard does affect trends in some of the
regardless of the standard used.
leading causes of death. The effect is least when changes in age-
specific rates are parallel and is greater when age-specific trends
Leading causes of death
diverge over time. For most of the leading causes, trends in age-
adjusted death rates are virtually parallel regardless of the standard.
Changing to the year 2000 standard affects age-adjusted death
Thus, trends for heart disease, stroke, diabetes, HIV infection, suicide,
rates for specific causes of death largely in terms of the magnitude of
chronic liver disease, homicide, and atherosclerosis are approximately
the rate and much less in terms of the trend. However, the effect
the same using the year 2000 standard and the 1940 standard. For
varies greatly among the leading causes of death. Table D shows
example, for heart disease the age-adjusted death rate based on the
trends from 1979 to 1995 in age-adjusted death rates using the 1940
1940 standard declined by 30 percent from 1979 to 1995 and by
and year 2000 standards for each of the 15 leading causes of death
26 percent based on the year 2000 standard. The difference reflects the
in the United States in 1995. For those causes where risk increases
greater emphasis that the year 2000 standard weights give to the less
sharply with age, chronic diseases in particular, the change in
rapid decline in the heart disease death rates at the older ages than
magnitude is up threefold. For cerebrovascular diseases (stroke), for
at the younger ages. Specifically, age-specific death rates for heart
example, the age-adjusted death rate is 26.7 deaths per 100,000
disease among those aged 25–64 years declined by 43 percent, while
6
Table D. Age-adjusted death rates and percent change based on the 1940 and year 2000 standard populations for 15 leading causes of death:
United States, 1979–95
National
[Age-adjusted rates are per 100,000 standard population. The asterisks preceding the categories indicate that they are not part of the Ninth Revision, International Classification of Diseases, 1975; the categories were added by
the National Center for Health Statistics in 1987]
V
ital
Cause of death (Based on the Ninth
Percent
Statistics
Revision, International Classification of
change
Diseases, 1975) and year
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1979–95
All causes
Report,
1940 . . . . . . . . . . . . . . . . . . . . . . .
577.1
585.1
568.6
554.7
552.5
548.1
548.8
544.8
539.2
539.9
528.0
520.2
513.8
504.5
513.3
507.4
503.9
–12.7
2000 . . . . . . . . . . . . . . . . . . . . . . .
1,011.1
1,039.1
1,007.2
985.0
989.8
982.5
988.1
978.6
970.0
975.7
950.6
938.7
925.5
910.9
931.5
920.2
918.5
–9.2
Diseases of heart
V
(390–398,402,404–429)
ol.
1940 . . . . . . . . . . . . . . . . . . . . . .
199.6
202.0
195.3
190.9
189.6
184.3
181.4
176.0
170.8
167.7
157.5
152.0
148.2
144.3
145.3
140.3
138.3
–30.7
47,
2000 . . . . . . . . . . . . . . . . . . . . . . .
401.7
412.1
397.0
389.0
388.8
378.8
374.9
365.1
355.9
352.5
332.1
321.8
313.8
306.1
310.0
299.7
296.3
–26.2
No.
Malignant neoplasms, including
neoplasms of lymphatic and
3,
hematopoietic tissues (140–208)
October
1940 . . . . . . . . . . . . . . . . . . . . . . .
130.8
132.8
131.7
132.8
133.1
134.1
134.4
134.2
134.0
134.0
134.5
135.0
134.5
133.1
132.6
131.5
129.9
–0.7
2000 . . . . . . . . . . . . . . . . . . . . . . .
204.0
207.9
206.4
208.3
209.1
210.8
211.3
211.5
211.7
212.5
214.2
216.0
215.8
214.3
214.6
213.1
211.7
3.8
Cerebrovascular diseases (430–438)
7,
1940 . . . . . . . . . . . . . . . . . . . . . . .
41.6
40.8
38.2
35.9
34.5
33.6
32.5
31.1
30.5
30.0
28.3
27.7
26.8
26.2
26.5
26.5
26.7
–35.8
1998
2000 . . . . . . . . . . . . . . . . . . . . . . .
97.3
85.3
89.7
84.4
81.3
78.9
76.6
73.3
71.8
70.8
67.1
65.5
63.4
62.1
63.2
63.3
63.9
–34.3
Chronic obstructive pulmonary diseases
and allied conditions (490–496)
1940 . . . . . . . . . . . . . . . . . . . . . . .
14.6
15.9
16.3
16.2
17.5
17.8
18.8
18.9
18.9
19.6
19.6
19.7
20.1
19.8
21.4
21.0
20.8
42.9
2000 . . . . . . . . . . . . . . . . . . . . . . .
25.5
28.3
29.0
29.0
31.6
32.4
34.5
34.8
35.0
36.5
36.6
37.2
38.0
37.9
40.9
40.6
40.5
58.7
Accidents and adverse effects (E800–E949)
1940 . . . . . . . . . . . . . . . . . . . . . . .
42.8
42.3
39.7
36.6
35.3
35.1
34.8
35.2
34.7
35.0
33.9
32.5
31.0
29.4
30.3
30.3
30.5
–28.8
2000 . . . . . . . . . . . . . . . . . . . . . . .
47.9
47.7
44.7
41.4
40.3
40.1
39.9
40.0
39.6
40.2
39.0
37.5
36.0
34.6
35.7
35.7
36.0
–24.8
Pneumonia and influenza (480–487)
1940 . . . . . . . . . . . . . . . . . . . . . . .
11.2
12.9
12.3
10.9
11.9
12.2
13.5
13.6
13.2
14.3
13.8
14.0
13.4
12.7
13.5
13.0
12.9
15.6
2000 . . . . . . . . . . . . . . . . . . . . . . .
26.1
31.4
30.0
26.5
29.7
30.6
34.5
34.8
33.8
37.3
35.9
36.8
34.9
33.1
35.2
33.9
33.8
29.4
Diabetes mellitus (250)
1940 . . . . . . . . . . . . . . . . . . . . . . .
9.8
10.1
9.8
9.6
9.9
9.5
9.7
9.7
9.9
10.2
11.6
11.7
11.8
11.8
12.5
12.9
13.3
36.1
2000 . . . . . . . . . . . . . . . . . . . . . . .
17.5
18.1
17.6
17.2
17.6
17.2
17.4
17.2
17.5
18.0
20.5
20.7
20.7
20.8
22.0
22.7
23.4
33.8
Human immunodeficiency virus infection
(*042–*044)
1940 . . . . . . . . . . . . . . . . . . . . . . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
5.5
6.7
8.7
9.8
11.3
12.6
13.8
15.4
15.6
184.2
2000 . . . . . . . . . . . . . . . . . . . . . . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
5.6
6.9
9.0
10.2
11.8
13.2
14.5
16.2
16.4
189.8
Suicide (E950–E959)
1940 . . . . . . . . . . . . . . . . . . . . . . .
11.7
11.4
11.5
11.6
11.4
11.7
11.6
11.9
11.7
11.5
11.3
11.5
11.4
11.1
11.3
11.2
11.2
–3.9
2000 . . . . . . . . . . . . . . . . . . . . . . .
12.6
12.2
12.3
12.5
12.4
12.6
12.5
13.0
12.8
12.5
12.3
12.5
12.3
12.1
12.2
12.0
12.0
–4.6
Chronic liver disease and cirrhosis (571)
1940 . . . . . . . . . . . . . . . . . . . . . . .
12.1
12.2
11.4
10.6
10.2
10.0
9.7
9.3
9.2
9.1
9.0
8.6
8.3
8.0
7.9
7.8
7.6
–37.0
2000 . . . . . . . . . . . . . . . . . . . . . . .
14.8
15.1
14.2
13.2
12.8
12.7
12.3
11.8
11.7
11.6
11.6
11.1
10.7
10.5
10.3
10.2
10.0
–32.8
Nephritis, nephrotic syndrome, and
nephrosis (580–589)
1940 . . . . . . . . . . . . . . . . . . . . . . .
4.4
4.5
4.5
4.5
4.6
4.8
4.9
4.9
4.8
4.8
4.5
4.3
4.3
4.3
4.5
4.3
4.3
–1.1
2000 . . . . . . . . . . . . . . . . . . . . . . .
8.6
9.1
9.1
9.3
9.6
10.0
10.4
10.4
10.4
10.4
9.6
9.3
9.3
9.4
9.7
9.4
9.5
10.0
Homicide and legal intervention
(E960–E978)
1940 . . . . . . . . . . . . . . . . . . . . . . .
10.2
10.8
10.4
9.7
8.6
8.4
8.3
9.0
8.6
9.0
9.4
10.2
10.9
10.5
10.7
10.2
9.4
–7.8
2000 . . . . . . . . . . . . . . . . . . . . . . .
9.9
10.5
10.1
9.4
8.4
8.1
8.0
8.6
8.3
8.5
8.8
9.5
10.1
9.6
9.8
9.3
8.5
–13.6
See footnotes at end of table.
Table D. Age-adjusted death rates and percent change based on the 1940 and year 2000 standard populations for 15 leading causes of death:
United States, 1979–95—Con.
[Age-adjusted rates are per 100,000 standard population. The asterisks preceding the categories indicate that they are not part of the Ninth Revision, International Classification of Diseases, 1975; the categories were added by
the National Center for Health Statistics in 1987]
Cause of death (Based on the Ninth
Percent
Revision, International Classification of
change
Diseases, 1975) and year
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1979–95
Septicemia (038)
1940 . . . . . . . . . . . . . . . . . . . . . . .
2.3
2.6
2.9
3.0
3.4
3.7
4.1
4.4
4.5
4.6
4.2
4.1
4.1
4.0
4.1
4.0
4.1
76.0
2000 . . . . . . . . . . . . . . . . . . . . . . .
4.3
5.0
5.4
5.9
6.7
7.4
8.3
9.0
9.3
9.7
8.8
8.6
8.6
8.4
8.6
8.3
8.4
94.7
Alzheimer’s disease (331.0)
1940 . . . . . . . . . . . . . . . . . . . . . . .
0.3
0.4
0.5
0.7
1.0
1.3
1.6
1.8
2.0
2.1
2.2
2.2
2.1
2.1
2.3
2.5
2.7
980.0
2000 . . . . . . . . . . . . . . . . . . . . . . .
0.4
0.7
0.9
1.3
2.2
3.1
4.1
4.6
5.5
5.8
6.1
6.4
6.3
6.3
7.2
7.8
8.4
1,862.8
Atherosclerosis (440)
1940 . . . . . . . . . . . . . . . . . . . . . . .
5.7
5.7
5.2
4.9
4.7
4.2
4.0
3.7
3.6
3.5
3.0
2.8
2.6
2.4
2.4
2.3
2.3
–59.4
2000 . . . . . . . . . . . . . . . . . . . . . . .
17.9
18.0
16.5
15.3
14.6
13.2
12.6
11.7
11.2
10.9
9.3
8.5
7.9
7.5
7.4
7.2
6.9
–61.5
. . . Category not applicable.
National
V
ital
Statistics
Report,
V
ol.
47,
No.
3,
October
7,
1998
7
8
Table E. Age-specific death rates and percent change for 15 leading causes of death by three broad age categories: United States, 1979–95
National
[Rates are per 100,000 population in specified age group. The asterisks preceding the categories indicate that they are not part of the Ninth Revision, International Classification of Diseases, 1975. The categories were added
by the National Center for Health Statistics in 1987]
Cause of death (Based on the Ninth
Percent
V
ital
Revision, International Classification
change
of Diseases, 1975) and age
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1979–95
Statistics
All causes
0–24 years . . . . . . . . . . . . . . . . . . .
121.0
121.0
114.0
109.6
104.5
103.2
103.0
104.4
102.1
103.1
101.9
99.5
97.1
91.3
91.3
88.1
84.9
–29.8
Report,
25–64 years . . . . . . . . . . . . . . . . . .
500.2
498.0
481.9
462.7
453.7
445.7
441.4
434.1
426.9
423.2
414.1
406.2
400.7
394.7
400.1
398.6
697.3
39.4
65 years and over . . . . . . . . . . . . . . . 5,060.5
5,252.0
5,117.3
5,056.4
5,134.5
5,118.7
5,174.9
5,130.2
5,095.8
5,146.3
5,012.3
4,963.2
4,924.0
4,880.6
5,047.7
5,014.1
5,052.8
–0.2
Diseases of heart (390–398,402,404–429)
V
0–24 years . . . . . . . . . . . . . . . . . . .
2.5
2.9
2.7
2.7
2.9
3.0
2.9
3.0
2.9
2.9
2.5
2.5
2.5
2.4
2.4
2.5
2.4
–5.1
ol.
25–64 years . . . . . . . . . . . . . . . . . .
155.2
151.9
146.0
139.6
135.7
130.0
125.6
119.5
113.7
109.0
101.6
96.9
94.0
91.9
91.8
89.4
88.4
–43.1
47,
65 years and over . . . . . . . . . . . . . . . 2,256.1
2,330.4
2,253.5
2,227.2
2,240.1
2,193.3
2,182.1
2,133.9
2,089.3
2,083.2
1,968.6
1,914.0
1,881.0
1,844.5
1,891.0
1,840.7
1,835.3
–18.6
No.
Malignant neoplasms, including neoplasms of
lymphatic and hematopoietic tissues
3,
(140–208)
October
0–24 years . . . . . . . . . . . . . . . . . . .
5.1
5.2
4.8
5.0
4.8
4.4
4.4
4.3
4.1
4.0
4.0
3.9
3.8
3.8
3.7
3.6
3.4
–33.1
25–64 years . . . . . . . . . . . . . . . . . .
142.7
142.9
139.4
138.4
136.6
136.2
134.6
131.0
128.8
127.0
124.5
123.4
121.3
118.6
117.1
115.6
113.7
–20.4
65 years and over . . . . . . . . . . . . . . .
986.6
1011.3
1008.6
1023.9
1034.0
1045.2
1051.1
1062.4
1067.3
1076.2
1095.8
1111.3
1117.3
1121.8
1133.7
1134.5
1136.6
15.2
7,
Cerebrovascular diseases (430–438)
1998
0–24 years . . . . . . . . . . . . . . . . . . .
0.7
0.8
0.7
0.6
0.7
0.7
0.6
0.5
0.5
0.6
0.5
0.6
0.6
0.5
0.6
0.5
0.6
–22.6
25–64 years . . . . . . . . . . . . . . . . . .
22.6
21.5
20.6
19.2
18.4
17.9
17.0
16.4
16.0
15.5
14.7
14.3
13.8
13.7
13.6
13.8
13.8
–39.1
65 years and over . . . . . . . . . . . . . . .
576.6
573.1
534.7
506.5
489.7
477.0
465.7
445.9
438.1
434.1
412.8
403.5
394.1
388.5
401.4
405.2
413.8
–28.2
Chronic obstructive pulmonary diseases
and allied conditions (490–496)
0–24 years . . . . . . . . . . . . . . . . . . .
0.3
0.4
0.4
0.4
0.4
0.4
0.4
0.4
0.5
0.4
0.4
0.4
0.5
0.4
0.5
0.5
0.5
48.6
25–64 years . . . . . . . . . . . . . . . . . .
10.8
11.3
11.3
10.7
11.4
11.3
11.6
11.3
11.1
11.2
11.1
10.7
10.7
10.2
10.7
10.5
10.3
–4.4
65 years and over . . . . . . . . . . . . . . .
152.2
170.6
175.8
177.0
193.4
199.5
213.4
216.0
217.5
227.8
228.0
234.1
240.6
242.2
263.7
262.5
263.9
73.4
Accidents and adverse effects (E800–E949)
0–24 years . . . . . . . . . . . . . . . . . . .
39.1
38.3
34.9
32.2
30.4
30.1
29.7
30.8
29.5
29.3
27.1
25.7
24.7
22.1
22.6
22.3
21.8
–44.1
25–64 years . . . . . . . . . . . . . . . . . .
42.1
41.8
40.3
36.7
35.5
35.1
34.9
34.9
34.7
35.1
34.6
33.3
31.2
30.4
31.5
31.5
32.2
–23.5
65 years and over . . . . . . . . . . . . . . .
95.6
97.2
90.5
86.0
87.2
87.5
87.9
86.6
87.2
89.6
87.5
84.3
83.3
82.5
84.8
85.4
86.8
–9.3
Pneumonia and influenza (480–487)
0–24 years . . . . . . . . . . . . . . . . . . .
2.1
1.9
1.6
1.5
1.5
1.4
1.4
1.4
1.4
1.3
1.4
1.3
1.3
1.2
1.2
1.1
1.1
–49.2
25–64 years . . . . . . . . . . . . . . . . . .
6.1
6.8
6.5
5.7
5.8
5.9
6.3
6.4
6.1
6.5
6.3
6.2
6.0
5.4
5.9
5.7
5.6
–8.2
65 years and over . . . . . . . . . . . . . . .
145.6
178.1
171.8
153.2
174.7
182.0
207.0
209.6
204.4
226.9
219.7
226.8
217.2
209.1
225.3
219.4
221.6
52.1
Diabetes mellitus (250)
0–24 years . . . . . . . . . . . . . . . . . . .
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
–29.7
25–64 years . . . . . . . . . . . . . . . . . .
8.5
8.8
8.4
8.2
8.4
7.9
8.1
8.1
8.2
8.6
9.4
9.5
9.5
9.6
10.0
10.4
10.8
26.5
65 years and over . . . . . . . . . . . . . . .
95.3
98.7
96.0
94.1
96.6
95.0
96.0
94.0
95.8
98.1
113.4
114.3
115.0
115.7
123.6
128.5
132.6
39.1
Human immunodeficiency virus
infection (*042–*044)
0–24 years . . . . . . . . . . . . . . . . . . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
0.8
0.9
1.0
0.9
1.1
1.0
1.1
1.2
1.2
–
25–64 years . . . . . . . . . . . . . . . . . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
10.2
12.5
16.5
18.8
21.7
24.5
26.9
30.1
30.4
–
65 years and over . . . . . . . . . . . . . . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
1.1
1.2
1.3
1.4
1.7
1.9
2.0
2.1
2.3
–
Suicide (E950–E959)
0–24 years . . . . . . . . . . . . . . . . . . .
5.7
5.7
5.7
5.6
5.5
5.7
5.9
5.9
5.7
5.7
5.6
5.6
5.5
5.4
5.6
5.7
5.5
–4.8
25–64 years . . . . . . . . . . . . . . . . . .
16.2
15.8
16.2
16.2
15.8
15.9
15.5
16.0
15.7
15.2
15.0
15.3
15.1
14.8
14.9
14.8
14.8
–8.4
65 years and over . . . . . . . . . . . . . . .
18.7
17.8
17.1
18.4
19.3
19.8
20.4
21.6
21.8
21.1
20.3
20.6
19.7
19.1
18.9
18.1
18.1
–3.5
Chronic liver disease and cirrhosis (571)
0–24 years . . . . . . . . . . . . . . . . . . .
0.2
0.2
0.2
0.2
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
–57.0
25–64 years . . . . . . . . . . . . . . . . . .
19.5
19.5
17.7
16.1
15.5
14.9
14.3
13.5
13.3
13.2
12.8
12.1
11.6
11.3
11.2
11.2
11.0
–43.6
65 years and over . . . . . . . . . . . . . . .
35.6
37.3
37.0
35.4
34.7
35.6
34.5
34.1
33.2
33.0
34.3
33.5
32.8
32.4
31.5
31.2
30.5
–14.4
See footnotes at end of table.
Table E. Age-specific death rates and percent change for 15 leading causes of death by three broad age categories: United States, 1979–95—Con.
[Rates are per 100,000 population in specified age group. The asterisks preceding the categories indicate that they are not part of the Ninth Revision, International Classification of Diseases, 1975. The categories were added
by the National Center for Health Statistics in 1987]
Cause of death (Based on the Ninth
Percent
Revision, International Classification
change
of Diseases, 1975) and age
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1979–95
Nephritis, nephrotic syndrome, and
nephrosis (580–589)
0–24 years . . . . . . . . . . . . . . . . . . .
0.5
0.4
0.5
0.4
0.4
0.5
0.4
0.4
0.4
0.4
0.3
0.3
0.3
0.3
0.3
0.3
0.3
–41.7
25–64 years . . . . . . . . . . . . . . . . . .
3.2
3.2
3.0
3.0
2.9
2.9
3.0
2.9
2.8
2.9
2.6
2.5
2.4
2.4
2.5
2.3
2.4
–24.2
65 years and over . . . . . . . . . . . . . . .
47.6
50.8
51.8
53.6
55.8
58.5
61.2
61.5
61.6
61.3
57.1
55.7
56.6
58.0
60.2
59.3
60.2
26.4
Homicide and legal intervention (E960–E978)
0–24 years . . . . . . . . . . . . . . . . . . .
7.5
8.1
7.7
7.2
6.4
6.3
6.3
7.2
7.0
7.6
8.2
9.4
10.4
10.2
10.7
10.2
9.2
22.7
25–64 years . . . . . . . . . . . . . . . . . .
13.4
14.2
13.8
12.7
11.4
11.0
10.9
11.5
10.9
11.1
11.3
11.8
12.1
11.4
11.2
10.6
9.7
–27.5
65 years and over . . . . . . . . . . . . . . .
5.2
5.6
5.0
4.9
4.5
4.3
4.3
4.5
4.6
4.4
4.2
4.0
4.1
3.8
3.7
3.5
3.2
–38.6
Septicemia (038)
0–24 years . . . . . . . . . . . . . . . . . . .
0.5
0.5
0.5
0.5
0.5
0.6
0.6
0.6
0.6
0.5
0.6
0.5
0.5
0.5
0.5
0.5
0.5
–1.5
25–64 years . . . . . . . . . . . . . . . . . .
1.8
2.0
2.2
2.2
2.4
2.6
2.8
2.9
2.9
2.9
2.7
2.6
2.6
2.6
2.5
2.6
2.7
50.0
65 years and over . . . . . . . . . . . . . . .
22.7
26.8
28.8
31.9
37.2
41.2
47.2
51.1
53.6
56.0
50.3
49.4
50.0
49.2
51.4
49.6
50.4
122.3
Alzheimer’s disease (331.0)
0–24 years . . . . . . . . . . . . . . . . . . .
–
–
–
0.0
–
0.0
0.0
–
–
–
–
0.0
–
–
0.0
0.0
0.0
–
25–64 years . . . . . . . . . . . . . . . . . .
0.2
0.3
0.3
0.3
0.3
0.3
0.4
0.4
0.4
0.4
0.3
0.3
0.3
0.2
0.3
0.3
0.3
44.9
65 years and over . . . . . . . . . . . . . . .
2.6
4.1
5.6
8.4
14.8
21.0
27.5
31.0
36.7
39.3
41.5
43.1
43.4
43.8
50.1
54.9
60.3
2221.6
Atherosclerosis (440)
0–24 years . . . . . . . . . . . . . . . . . . .
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
–49.7
25–64 years . . . . . . . . . . . . . . . . . .
1.2
1.3
1.2
1.1
1.1
1.0
1.0
0.9
0.9
0.8
0.7
0.7
0.7
0.6
0.6
0.6
0.6
–48.6
65 years and over . . . . . . . . . . . . . . .
109.5
109.9
101.9
95.3
91.7
83.7
80.1
74.3
72.1
69.9
60.0
55.2
52.2
49.6
50.2
49.2
47.4
–56.7
. . . Category not applicable.
0.0 Quantity more than zero but less than 0.5.
National
– Quantity zero.
V
ital
Statistics
Report,
V
ol.
47,
No.
3,
October
7,
1998
9
10 National Vital Statistics Report, Vol. 47, No. 3, October 7, 1998
the decline among those older than 65 years was only 19 percent
the important age-specific differences in the mortality race ratio. To
(table E).
better understand race differentials in mortality, it is essential to
For other leading causes of death, trends using the two different
augment analyses of age-adjusted death rates with analyses of age-
standards are less consistent. The previously described case of cancer
specific rates.
(table B) is illustrative because of the clear pattern of divergent age-
While the magnitude of the mortality race ratio is affected by the
specific trends. Based on either standard, the trend in age-adjusted
change in standard, the trend in the ratio over time is not seriously
death rate for cancer increased gradually from 1979 reaching a peak
affected. Figure 3 shows the trend in the mortality race ratio for 1979–95
in 1990 and declining steadily thereafter (table D). However, based on
based on the 1940 and year 2000 standards. The trends in the mortality
the 1940 standard, the 1995 rate is 0.7 percent below the rate for 1979;
race ratio based on both standards are nearly parallel. Thus, regardless
while using the year 2000 standard, the 1995 rate is 3.8 percent above
of the standard used, the widening or narrowing of the race gap in
that for 1979. The relatively higher 1995 rate based on the year 2000
mortality will be approximately the same even if the magnitude of the
standard reflects the greater emphasis that the year 2000 standard
gap itself is different.
gives to increases in age-specific death rates at the older ages than the
Another widely used measure of mortality risk is the expectation
decreases at the younger ages; while the 1940 standard gives more
of life at birth, which is derived from life tables. Like the age-adjusted
emphasis to the decreases at the younger ages.
death rate, the life expectancy measure is standardized so that com-
parisons over time or between groups are not affected by the actual age
Race differences in mortality
distributions of the respective populations. However, unlike the age-
The year 2000 standard has implications for race and ethnic
adjusted death rate, life table measures are entirely free of assumptions
differentials in mortality. In particular, the difference between mortality
about the structure of the populations being compared. Instead, life
for the black and white populations will be affected as will that for the
tables generate their own ‘‘life table population’’ and thus, are not
Hispanic and non-Hispanic populations. One way of showing the
weighted by an arbitrary, externally imposed standard population. As a
differential in mortality between population groups is the ‘‘mortality
result comparisons of life expectancy at birth are unaffected by the
race ratio,’’ which is the ratio of the age-adjusted death rate for one
change from the 1940 to the year 2000 standard.
group (e.g., the black population) to that of another group (the white
Discussion
population). The mortality race ratio for the black and white popula-
tions in 1995 is reduced from 1.6 using the 1940 standard to 1.4 using
Participants of two national workshops reviewed the technical
the year 2000 standard (see table F). Using the 1940 standard, the
and policy issues associated with alternative population standards.
black population has an age-adjusted death rate that is 60 percent
Participants of the second workshop recommended changing to the
higher than that for the white population (18). In contrast, the year
year 2000 standard from the 1940 standard, which has been used for
2000 standard results in a rate for the black population that is only
over 50 years. They also recommended that all health agencies use
40 percent higher. The explanation for the narrowing of the differential
the year 2000 standard for routine presentation of mortality statistics.
lies in the age-specific death rates and the population structure of the
These recommendations will be become policy of DHHS, effective
two race groups. Table F shows age-specific death rates by race for
September 1998.
three broad age groups. The mortality ratio is highest for the youngest
The adoption of a single standard will reduce confusion among
age group (0–24 years), where the black population has double the
data users and will reduce the burden on State and local agencies, who
mortality of the white population. For the oldest age group (65 years
now must produce multiple data series to be consistent with the rates
and over), however, the mortality ratio is 1.1, denoting only 10 percent
based on different standards used by DHHS agencies. A new standard
higher mortality between the elderly black and white populations. The
means, however, that mortality time series at all geographical levels
reduction in the overall (all ages combined) mortality ratio from the
must be recomputed using the new standard. Further, long-range goal
1940 to the year 2000 standard reflects the greater weight that the
setting efforts such as ‘‘Healthy People’’ must recalibrate their health
year 2000 standard gives to the older population, where race
goals measured in terms of age-adjusted rates.
differentials in mortality are smaller. Because the age-specific rates in
Age-adjusted death rates calculated before implementation of the
the black and white populations being compared do not have a
year 2000 standard will not be comparable to rates using the new
consistent relationship, the single ratio of age-adjusted rates masks
standard. Comparisons of age-adjusted death rates based on different
standards can lead to erroneous conclusions regarding trends in health
Table F. Age-specific and age-adjusted death rates by
and mortality. Use of the year 2000 standard will result in age-adjusted
race: United States, 1995
death rates that are often substantially larger than those based on the
1940 standard. The new standard will affect trends in age-adjusted
[Age-adjusted death rates are per 100,000 standard population. Age-specific rates
are per 100,000 population in specified age group]
death rates for certain causes of death and will narrow race differentials
in age-adjusted death rates. However, use of the year 2000 standard
White
Black
Rate
death rate
death rate
Ratio
will result in race differentials in mortality that more closely approximate
those of the ‘‘real’’ population than mortality race differentials based on
Age-adjusted rates
the 1940 standard. These effects will require explanation to data users
1940 standard . . . . . . . . . . . . . .
476.9
765.7
1.6
2000 standard . . . . . . . . . . . . . .
890.0
1,224.5
1.4
and the media.
The two workshops underscored the strengths and weaknesses of
Age-specific rates
0–24 years . . . . . . . . . . . . . . . .
73.0
149.1
2.0
age-adjusted death rates. Although age standardization is an important
25–64 years . . . . . . . . . . . . . . .
365.4
691.1
1.9
and useful tool, some of its limitations become apparent when changing
65 years and over . . . . . . . . . . . .
5,049.3
5,679.2
1.1
the population standard. The numerical value of the age-adjusted rate
National Vital Statistics Report, Vol. 47, No. 3, October 7, 1998
11
Figure 3. Mortality race ratio based on the 1940 and 2000 standard populations: United States, 1979–95
depends on the standard used and is meaningful by itself only for
8.
National Center for Health Statistics. Technical appendix. Vital statistics
comparing groups or trends. Further, age standardization is less useful
of the United States, 1994, vol II, mortality, part A. Washington: Public
when age-specific rates in the populations being compared do not have
Health Service. 1998. Available at http:// www.cdc.gov/nchswww
a consistent relationship. Finally, because age-adjusted death rates are
9.
Day, JC. U.S. Bureau of the Census. Population projections of the
averages, they represent merely the beginning of an analysis strategy
United States by age, sex, race, and Hispanic origin: 1995 to 2050.
that should proceed to age-specific analyses and then to examination
Current population reports; series P-25, no 1130. U.S. Government
of additional sociodemographic, temporal, and geographical variables.
Printing Office. Washington, DC. 1996.
10.
Neison FGP. On a method recently proposed for conducting inquiries
into the comparative sanatory condition of various districts. Journal of
References
the Royal Statistical Society of London 7: 40–68. 1844.
1.
Treas JT. Older Americans in the 1990s and beyond. Population Bulletin
11.
Shryock HS, Siegel JS. The methods and materials of demography.
50(2). 1995.
San Diego: Academic Press. 1976.
2.
Sondik EJ. Use of age-adjusted disease rates for cancer. In: Feinleib M,
12.
Curtin LR. A short history of standardization for vital events. In: Feinleib
Zarate AO, eds. Reconsidering age adjustment procedures: Workshop
M, Zarate AO, eds. Reconsidering age adjustment procedures: Work-
proceedings. National Center for Health Statistics. Vital Health Stat
shop proceedings. National Center for Health Statistics. Vital Health
4(29). 1992.
Stat 4(29). 1992.
3.
Ries LAG, Miller BA, Hankey BF, Kosary CL, Harras A, Edwards BK, et
13.
Curtin LR, Klein RJ. Direct standardization (age-adjusted death rates).
al., eds. SEER Cancer Statistics Review, 1973–1991. Bethesda,
National Center for Health Statistics. Healthy People 2000: Statistical
Maryland: National Cancer Institute. 1994.
notes, no 6—revised. 1995.
4.
Centers for Disease Control and Prevention. Trends in ischemic heart
14.
Curtin LR, Maurer J, Rosenberg HM. On the selection of alternative
disease deaths—United States, 1990–1994. Morbidity and Mortality
standards for the age-adjusted death rate: Proceedings of the social
Weekly Report, 46(7): 146–150. 1997.
statistics sections, American Statistical Association. 218–23. 1980.
5.
Feinleib M, Zarate AO, eds. Reconsidering age adjustment procedures:
15.
Workshop proceedings. National Center for Health Statistics. Vital
Wolfenden HH. On the methods of comparing the mortalities of two or
Health Stat 4(29). 1992.
more communities and the standardization of death rates. Journal of the
6.
Feinleib M. Summary and recommendations. In: Feinleib M, Zarate AO,
Royal Statistical Society. 86:399–411. 1923.
eds. Reconsidering age adjustment procedures: Workshop proceed-
16.
Fleiss, JL. Statistical methods for rates and proportions. New York: John
ings. National Center for Health Statistics. Vital Health Stat 4(29). 1992.
Wiley and Sons. 1973.
7.
Anderson RN. Report of the second workshop on age adjustment. In
17.
National Center for Health Statistics. Vital Statistics of the United
preparation.
States, vol II, mortality. Washington: Public Health Service. 1992.
12 National Vital Statistics Report, Vol. 47, No. 3, October 7, 1998
18.
National Center for Health Statistics. Health United States 1996–97 and
23.
Fay MP, Feuer EJ. A semi-parametric estimate of extra-Poisson
Injury Chartbook. Hyattsville, Maryland. 1997.
variation for vital rates. Statistics in Medicine 16:2389–2401. 1997.
19.
Anderson RN, Kochanek KD, Murphy SL. Report of final mortality
24.
Brillinger DR. The natural variability of vital rates and associated
statistics, 1995. Monthly vital statistics report; vol 45 no 11, supp 2.
statistics. Biometrics 42:693–734. 1986.
Hyattsville, Maryland: National Center for Health Statistics. 1997.
25.
Seal HL. Mortality data and the binomial probability law. Skandinavisk
20.
Pickle LW, Mungiole M, Jones GK, White AA. Atlas of United States
Aktuarietidskrift 33: 188–216. 1949.
Mortality. Hyattsville, Maryland: National Center for Health Statistics.
26.
Arnold SF. Mathematical statistics. Englewood Cliffs, New Jersey:
1996.
Prentice Hall. 1990.
21.
Spiegelman M, Marks HH. Empirical testing of standards for the age
27.
Fay MP, Feuer EJ. Confidence intervals for directly standardized rates:
adjustment of death rates by the direct method. Human Biology
A method based on the gamma distribution. Stat Med 16:791–801.
38:280–92. 1966.
1997.
22.
Chiang CL. Standard error of the age-adjusted death rate. Vital
statistics–Special reports; vol 47 no 9. National Center for Health
Statistics. Washington: Public Health Service. 1961.
National Vital Statistics Report, Vol. 47, No. 3, October 7, 1998
13
Technical notes
where 0 < w < 1 and the w sum to 1. The age-adjusted death rate
si
si
(AADR) is then given by
Year 2000 standard weights
Di
The year 2000 standard million population is constructed from
AADR = ∑ w
= ∑ w
si c P
si c Ri
(3)
i
the projected year 2000 population prepared by the U.S. Bureau of
i
i
the Census (9). The projections shown in this report are from the
middle series calculated based on the July 1, 1994, population
Indirect standardization
estimated from the 1990 Decennial Census. The projected year 2000
age-specific populations, their proportion distribution, and the stan-
Indirect standardization is less commonly used, but is useful
dard million are shown in table I. The standard million is simply
especially when age-specific numbers of deaths are unavailable
calculated by multiplying the proportion distribution rounded to six
(11,13). For indirect standardization, a standard set of age-specific
decimal places by 1,000,000. The standard weights (w ) are equal to
death rates are applied to the overall mortality experience of the
si
the proportion distribution of the standard million.
observed population. This technique yields an ‘‘expected’’ number of
deaths in the observed population, assuming that the age-specific
Direct standardization
death rates of the standard population apply to the observed
population. The indirect standardized death rate (ISDR) is computed
The age-adjusted death rate is most often computed using the
from the expected number of deaths and is given by
direct method as it is the simplest and most straightforward method of
standardization (10). Let
Rs c D
ISDR = ∑
(4)
R
D = the number of deaths in age interval i, and
si c Pi
i
P = the midyear population in age interval i.
i
i
The age-specific death rate (R ) is then given by
i
where R is the crude rate of the standard population, D is the total
s
D
number of deaths in the observed population, R is the age-specific
si
i
R =
death rate in age interval i in the standard population, and P is the
i
P
(1)
i
i
population of age interval i in the observed population. Most often,
however, the ratio of observed deaths to expected deaths is pre-
which is usually expressed per 1,000 or 100,000 population.
sented. This ratio is called the standardized mortality ratio (SMR) and
is given by
The age-adjusted death rate is a weighted average of the
age-specific death rates where the age-specific weights represent the
observed deaths
D
relative age distribution of the standard population. Let
SMR =
=
(5)
expected deaths
∑Rsi c Pi
P = the population in age interval i in the standard population.
si
i
The standard weights (w ) are then given by
si
Psi
Variability
w =
(2)
si
∑Psi
Age-adjusted death rates, with the exception of preliminary
i
estimates, are typically based on complete counts and are not subject
to sampling error. However, mortality data, including age-adjusted
Table I. Projected year 2000 U.S. population and
death rates, may be subject to random variation. That is, the number
proportion distribution by age
of deaths that actually occurred may be considered as one of a large
series of possible results that could have arisen under the same
Proportion
distribution
Standard
circumstances (22). When the number of deaths is small, random
Age
Population
(weights)
million
variation may be relatively large, and thus, caution must be used in
interpreting age-adjusted death rates and other mortality data.
Total . . . . . . . . . . . . . . . . . . . . .
274,634,000
1.000000
1,000,000
Random variation is typically measured in terms of variance or
Under 1 year . . . . . . . . . . . . . . . .
3,795,000
0.013818
13,818
standard error (the square root of the variance). The calculation of the
1–4 years . . . . . . . . . . . . . . . . . .
15,192,000
0.055317
55,317
5–14 years . . . . . . . . . . . . . . . . .
39,977,000
0.145565
145,565
standard error of the age-adjusted death rate is shown below.
15–24 years . . . . . . . . . . . . . . . . .
38,077,000
0.138646
138,646
The age-adjusted death rate is a weighted average of the age-
25–34 years . . . . . . . . . . . . . . . . .
37,233,000
0.135573
135,573
specific death rates (equation 3). Because the age-specific standard
35–44 years . . . . . . . . . . . . . . . . .
44,659,000
0.162613
162,613
45–54 years . . . . . . . . . . . . . . . . .
37,030,000
0.134834
134,834
weights are invariant and the probability of death in one age interval
55–64 years . . . . . . . . . . . . . . . . .
23,961,000
0.087247
87,247
is independent of the probability of death in any other age interval, the
65–74 years . . . . . . . . . . . . . . . . .
18,136,000
0.066037
66,037
variance of the age-adjusted death rate is given by
75–84 years . . . . . . . . . . . . . . . . .
12,315,000
†0.044842
44,842
85 years and over . . . . . . . . . . . . .
4,259,000
0.015508
15,508
var (AADR) =
† Figure is rounded up instead of down to force total to 1.0.
∑w2 var (R)
si
i
(6)
i
14 National Vital Statistics Report, Vol. 47, No. 3, October 7, 1998
To calculate the variance of the age-specific death rate (R ), one
X
i
x2
must make certain assumptions about the process of death. The first
v =
~ Γ ( x2 ,1)
v
v
(11)
assumption is that all persons in age interval i have the same risk of
x
death (homogeneity). This assumption allows for simplicity in the cal-
The lower 100(1–α)-percent confidence limit for x2 / v is given by
culation of the variance and is typically applied although the risk of death
is distinctly heterogeneous within age intervals. Although beyond the
L( x2 ) = Γ –1S D (α / 2)
scope of this report, the extra variation in vital rates due to within-
v
x2 ,1
v
(12)
age-group heterogeneity in the risk of death can be estimated and
applied to statistical tests (23). The second assumption involves the
The upper 100(1–α)-percent confidence limit is given by
underlying distribution used to calculate the variance. The number of
deaths occurring in a population has typically been assumed to follow
U( x2 ) = Γ –11
2
a binomial distribution (22, 24, 25). However, critical assumptions of the
v
(x + k )2
(1 – α / 2)
M
,1
(13)
v + k2
binomial are not very realistic when applied to an open population (24).
M
As a result, variance estimates based on the binomial tend to under-
where k = k = max
(k ) is a continuity correction made
M
ie{1,. . .,I}
i
estimate the variance associated with the death rate. Death in open
necessary by the fact that we are using a continuous distribution to
populations can be alternatively viewed as deriving from a Poisson
estimate confidence limits for a discrete random variable.
distribution. The Poisson is much simpler conceptually and computa-
From equation 3, increasing the number of deaths by 1 in age interval
tionally and provides reasonable, conservative estimates of the variance
i results in a k = w / p increase in the age-adjusted death rate. If k is
i
i
i
i
of the death rate (24). Using the properties of the Poisson distribution,
constant for all age intervals, k = k. However, given that the values for
i
the variance of the age-specific death rate is given by
w and p typically used in calculating age-adjusted death rates are
i
i
D
1
D
R 2
variable across age intervals, it is unclear what value of k is
i
i
i
var(R ) = var (
) =
var (D ) =
=
(7)
appropriate. A conservative upper confidence limit can be obtained by
i
P
2
i
2
D
i
P
P
i
i
i
using the maximum value of k = k (27).
i
M
A close approximation of equation 14 that alleviates the need to
Substituting equation 7 into equation 6, the standard error of the
calculate k is
M
age-adjusted death rate S(AADR) is given by
Œ
)
∑
D
2
U ( x2 = Γ –1Sx2
(1 – α / 2)
w 2 R
v
i
+ 1 ,1
(14)
S(AADR) = √var(AADR) =
si c
(8)
v
i
Di
For x* = cx, it can be shown that L(x*) = c L(x) and U(x*) = c U(x)
(26). Let x* = x2 / v and let c = x / v.
Confidence intervals
Then
)
For an age-adjusted death rate X, let E(X) = x be given by
L( x2v
(15)
equation 3 and var(X) = v be given from equation 8. The age-
L(x) =
x
adjusted death rate is a linear combination of Poisson random
v
variables. However, it is clear that the age-adjusted death rate X is not
a Poisson random variable itself because E(X) is not equal to var(X).
and
Indeed, a linear combination of independent Poisson random vari-
)
ables does not have a simple form (26). However, it can be placed in
U ( x2v
(16)
U(x) =
the more general family of gamma distributions of which the Poisson
x
is a member. Given a gamma distribution with parameters a and b, let
v
X ~ Γ (a, b). Then E(X) = x = a / b and var(X) = v = ab2. Describing
a and b in terms of x and v gives a distribution for X such that
These results can easily be calculated using statistical packages such
as SAS, which have a function to calculate the inverse gamma
v
distribution (or the inverse chi-square distribution, see 27). Table II
X ~ Γ ( x2, )
(9)
v
x
shows a set of factors that may be applied to age-adjusted death
rates to calculate 95-percent confidence intervals. These factors are
A useful property of the gamma distribution is that one can divide
derived from the standard gamma distribution such that for any value
X by b (25) such that
of x2 / v rounded to the nearest integer the lower confidence factor
X ~ Γ
(LCF) is
(a,1)
(10)
b
L( x2 )
This converts the gamma distribution into its standard form, i.e.,
v
LCF( x2 ) =
(17)
where b = 1. This greatly simplifies calculations. Expressing equation
v
x2
9 in its standard form gives
v
National Vital Statistics Report, Vol. 47, No. 3, October 7, 1998
15
The upper confidence factor (UCF) for any integer value of x2 / v is
2. Agencies should implement the new population standard by
calculated in the same way substituting U(x2 / v) into equation 17.
data year 1999.
These factors can then be multiplied by the age-adjusted death rate
3. Agencies should continue to use and publish their standards
(x) such that L(x) = x c LCF and U(x) = x c UCF. As the L(x) and U(x)
until the new standard is officially adopted (beginning with data
based on factors in table II are restricted to integer values of x2 / v,
year 1999). To avoid confusion, agencies implementing the
they will differ slightly from values of L(x) and U(x) calculated in SAS
new standard before data year 1999 should simultaneously
where there is no such restriction. The SAS code used to generate
publish rates adjusted to the old and new standards.
table II is shown at the end of the table.
4. After the implementation date, agencies should use the new
The Poisson distribution and its gamma family members are
standard in all press releases and other communication with
asymmetrical distributions with zero as the lower bound. However, for
the public.
X ~ Γ (a, b ), when a is large, X is approximately normally distributed
–1
5. NCHS will be responsible for selecting a name for the new
and thus, nearly symmetrical. Therefore, when constructing confidence
standard and will determine the number of significant digits.
intervals for the age-adjusted death rate a normal approximation may
6. Agencies should continue to use the current 11 age groups
be applied when x2 / v is large to simplify calculations. In practice,
(less than 1 year, 1–4 years, 5–14 years, 15–24 years, 25–34
95-percent confidence intervals are reasonably symmetrical when x2 / v
years, 35–44 years, 45–54 years, 55–64 years, 65–74 years,
is greater than or equal to 100, although this cutoff point is somewhat
75–84 years, and 85 years and over) for calculating age-
arbitrary. As a result values for x2 / v in table II are limited to integer
adjusted rates using the new standard.
values from 1 to 99. The normal approximations of the 95-percent
7. NCHS will convene an implementation committee that will be
confidence limits are given by L(x) = x – 1.96 √v and U(x) = x + 1.96
responsible for developing a time table and strategies for
√v.
implementation and for commissioning papers to publicize the
change in standard.
Recommendations of the second workshop on age adjustment
8. NCHS will publicize the new standard in NCHS publications,
the Morbidity and Mortality Weekly Report, Public Health
1. The population standard for age-adjusting death rates should be
Reports, and appropriate professional newsletters. Scholarly
changed from the 1940 standard million population to the
papers could also be published in appropriate professional and
projected U.S. 2000 population to be published by the Census
technical journals.
Bureau in the spring of 1998. A single standard should be used
9. NCHS will convene a work group to evaluate the age-
by all agencies for official presentation of data. For special
adjustment standard at least every 10 years.
analyses, alternative standards may be used as appropriate to
the research.
16 National Vital Statistics Report, Vol. 47, No. 3, October 7, 1998
Table II. Lower and upper 95-percent confidence limit factors for age-adjusted death rates based on a gamma
distribution with parameter x2 / v, where x is the age-adjusted death rate and v is the variance of the age-adjusted
death rate
Lower
Upper
Lower
Upper
confidence
confidence
confidence
confidence
x2/v
factor
factor
x2/v
factor
factor
1 . . . . . . . . . . .
0.02532
5.57164
51 . . . . . . . . . .
0.74457
1.31482
2 . . . . . . . . . . .
0.12110
3.61234
52 . . . . . . . . . .
0.74685
1.31137
3 . . . . . . . . . . .
0.20622
2.92242
53 . . . . . . . . . .
0.74907
1.30802
4 . . . . . . . . . . .
0.27247
2.56040
54 . . . . . . . . . .
0.75123
1.30478
5 . . . . . . . . . . .
0.32470
2.33367
55 . . . . . . . . . .
0.75334
1.30164
6 . . . . . . . . . . .
0.36698
2.17658
56 . . . . . . . . . .
0.75539
1.29858
7 . . . . . . . . . . .
0.40205
2.06038
57 . . . . . . . . . .
0.75739
1.29562
8 . . . . . . . . . . .
0.43173
1.97040
58 . . . . . . . . . .
0.75934
1.29273
9 . . . . . . . . . . .
0.45726
1.89831
59 . . . . . . . . . .
0.76125
1.28993
10 . . . . . . . . . . .
0.47954
1.83904
60 . . . . . . . . . .
0.76311
1.28720
11 . . . . . . . . . . .
0.49920
1.78928
61 . . . . . . . . . .
0.76492
1.28454
12 . . . . . . . . . . .
0.51671
1.74680
62 . . . . . . . . . .
0.76669
1.28195
13 . . . . . . . . . . .
0.53246
1.71003
63 . . . . . . . . . .
0.76843
1.27943
14 . . . . . . . . . . .
0.54671
1.67783
64 . . . . . . . . . .
0.77012
1.27698
15 . . . . . . . . . . .
0.55969
1.64935
65 . . . . . . . . . .
0.77178
1.27458
16 . . . . . . . . . . .
0.57159
1.62394
66 . . . . . . . . . .
0.77340
1.27225
17 . . . . . . . . . . .
0.58254
1.60110
67 . . . . . . . . . .
0.77499
1.26996
18 . . . . . . . . . . .
0.59266
1.58043
68 . . . . . . . . . .
0.77654
1.26774
19 . . . . . . . . . . .
0.60207
1.56162
69 . . . . . . . . . .
0.77806
1.26556
20 . . . . . . . . . . .
0.61083
1.54442
70 . . . . . . . . . .
0.77955
1.26344
21 . . . . . . . . . . .
0.61902
1.52861
71 . . . . . . . . . .
0.78101
1.26136
22 . . . . . . . . . . .
0.62669
1.51401
72 . . . . . . . . . .
0.78244
1.25933
23 . . . . . . . . . . .
0.63391
1.50049
73 . . . . . . . . . .
0.78384
1.25735
24 . . . . . . . . . . .
0.64072
1.48792
74 . . . . . . . . . .
0.78522
1.25541
25 . . . . . . . . . . .
0.64715
1.47620
75 . . . . . . . . . .
0.78656
1.25351
26 . . . . . . . . . . .
0.65323
1.46523
76 . . . . . . . . . .
0.78789
1.25165
27 . . . . . . . . . . .
0.65901
1.45495
77 . . . . . . . . . .
0.78918
1.24983
28 . . . . . . . . . . .
0.66449
1.44528
78 . . . . . . . . . .
0.79046
1.24805
29 . . . . . . . . . . .
0.66972
1.43617
79 . . . . . . . . . .
0.79171
1.24630
30 . . . . . . . . . . .
0.67470
1.42756
80 . . . . . . . . . .
0.79294
1.24459
31 . . . . . . . . . . .
0.67945
1.41942
81 . . . . . . . . . .
0.79414
1.24291
32 . . . . . . . . . . .
0.68400
1.41170
82 . . . . . . . . . .
0.79533
1.24126
33 . . . . . . . . . . .
0.68835
1.40437
83 . . . . . . . . . .
0.79649
1.23965
34 . . . . . . . . . . .
0.69253
1.39740
84 . . . . . . . . . .
0.79764
1.23807
35 . . . . . . . . . . .
0.69654
1.39076
85 . . . . . . . . . .
0.79876
1.23652
36 . . . . . . . . . . .
0.70039
1.38442
86 . . . . . . . . . .
0.79987
1.23499
37 . . . . . . . . . . .
0.70409
1.37837
87 . . . . . . . . . .
0.80096
1.23350
38 . . . . . . . . . . .
0.70766
1.37258
88 . . . . . . . . . .
0.80203
1.23203
39 . . . . . . . . . . .
0.71110
1.36703
89 . . . . . . . . . .
0.80308
1.23059
40 . . . . . . . . . . .
0.71441
1.36172
90 . . . . . . . . . .
0.80412
1.22917
41 . . . . . . . . . . .
0.71762
1.35661
91 . . . . . . . . . .
0.80514
1.22778
42 . . . . . . . . . . .
0.72071
1.35171
92 . . . . . . . . . .
0.80614
1.22641
43 . . . . . . . . . . .
0.72370
1.34699
93 . . . . . . . . . .
0.80713
1.22507
44 . . . . . . . . . . .
0.72660
1.34245
94 . . . . . . . . . .
0.80810
1.22375
45 . . . . . . . . . . .
0.72941
1.33808
95 . . . . . . . . . .
0.80906
1.22245
46 . . . . . . . . . . .
0.73213
1.33386
96 . . . . . . . . . .
0.81000
1.22117
47 . . . . . . . . . . .
0.73476
1.32979
97 . . . . . . . . . .
0.81093
1.21992
48 . . . . . . . . . . .
0.73732
1.32585
98 . . . . . . . . . .
0.81185
1.21868
49 . . . . . . . . . . .
0.73981
1.32205
99 . . . . . . . . . .
0.81275
1.21746
50 . . . . . . . . . . .
0.74222
1.31838
* Program to compute 100(1- alpha)-percent confidence limit factors;
* for a gamma or Poisson-distributed variable with parameter n;
Percent let alpha=0.05; * For 95-percent confidence limit factors;
data CI;
alo = &alpha/2;
ahi = 1-&alpha/2;
do n = 1 to 99;
LCF = gaminv ( alo, n)/n;
UCF = gaminv (ahi, n+1)/n;
output;
end;
proc print data=CI;
var n LCF UCF;
run;
20 National Vital Statistics Report, Vol. 47, No. 3, October 7, 1998
Suggested citation
Contents
Anderson RN, Rosenberg HM. Age Standardization of Death Rates:
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Implementation of the Year 2000 Standard. National vital statistics reports;
vol 47 no. 3. Hyattsville, Maryland: National Center for Health Statistics.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1998.
Methods. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Death rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Effects of changing to the year 2000 standard . . . . . . . . . 3
National Center for Health Statistics
Magnitude of the age-adjusted death rate . . . . . . . . . . . 4
Director, Edward J. Sondik, Ph.D.
Leading causes of death. . . . . . . . . . . . . . . . . . . . . . . . 5
Deputy Director, Jack R. Anderson
Race differences in mortality . . . . . . . . . . . . . . . . . . . . 10
Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Division of Vital Statistics
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Director, Mary Anne Freedman
Technical notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
To receive this publication regularly, contact the National Center
for Health Statistics by calling 301-436-8500
E-mail: nchsquery@cdc.gov
Internet: www.cdc.gov/nchswww/
Copyright information
All material appearing in this report is in the public domain and may be
reproduced or copied without permission; citation as to source, however,
is appreciated.
U.S. DEPARTMENT OF
FIRST CLASS MAIL
HEALTH & HUMAN SERVICES
POSTAGE & FEES PAID
Centers for Disease Control and Prevention
CDC/NCHS
National Center for Health Statistics
PERMIT NO. G-284
6525 Belcrest Road
Hyattsville, Maryland 20782-2003
DHHS Publication No. (PHS) 99–1120
8-0711 (10/98)
OFFICIAL BUSINESS
PENALTY FOR PRIVATE USE, $300

Document Outline