Standardization of rates and ratios - Assignment
1. From the data in the table below, compute for each sex separately (for Rateboro) and for the
United States (both sexes) the following measures. Write your answers (rounded to 4 decimal
places) in the table; show all work for (c) and (d).
a. crude death rates
b. age-specific death rates
c. directly-standardized death rates for Rateboro males and females (separately) using the U.S.
population as a standard.
d. indirectly standardized death rates as in (c).
Population and Deaths in 1980 in Rateboro
Adults by Age and Sex and U.S. Total
(hypothetical data)
Rateboro United States
Males Females Both Sexes
Age Pop. Deaths Rate Pop. Deaths Rate Pop* Deaths* Rate
18-34 900 6 800 1 60,000 90
35-59 800 3 800 5 45,000 270
60-74 300 15 500 10 20,000 600
75 + 200 22 500 38 15,000 1500
Total 2200 46 2600 54 140,000 2460
(*In thousands. Population and deaths for Rateboro are actual figures.)
Direct standardized rate:
Indirect standardized rate:
2. Based on the results for question 1.:
a. Do males or females have a more favorable mortality experience in Rateboro? Cite the rates
or other figures on which you have based your decision.
b. How do you account for the similarity in the crude death rates for Rateboro males and
females?
c. Briefly discuss the reasons for and against (i) rate adjustment, and (ii) direct versus indirect
methods--in these data.
___________________
* Thanks to Barbara Richardson, Ph.D. for the first version of this question.
_____________________________________________________________________________________________
EPID 168 Standardization - Assignment - 149
rev. 9/2/91, 8/4/1999
d. How would you feel about the conclusion, by an experienced epidemiologist, that "the
Rateboro data are generally consistent with the typical finding of a more favorable mortality
experience of U.S. females; the anomolous result for the 35-59 year-old group, with the high
death rate among females (more than 50% greater than the rate for males) is evidence that
the Rateboro environment is more suitable for males in the age range 35-59 than for
females."
3. The following extract from "Breast cancer in women after repeated fluoroscopic examinations
of the chest" (John D. Boice, Jr., and Richard R. Monson, J Natl Cancer Inst 59:823-832, 1977)
describes their adjustment procedure:
"...Expected breast cancer cases were determined with the use of age-calendar year
specific incidence rates of Connecticut (refs), a neighboring State whose cancer registry
has been in existence since 1935. The years at which a woman was at risk for breast
cancer development (i.e., the years after sanitarium admission or fluoroscopy exposure)
were computed separately for each 5-year age group, each 5-year period since start of
observation, and each quinquennium from 1930 to 1970 through 1974 and for the six
month period from January 1975 through June 1975. Multiplication of the age-calendar
year specific WY [women-years] at risk by the corresponding Connecticut incidence rates
determined the number of expected breast cancers."
a. What method of adjustment is being used, direct or indirect?
b. The following tables show hypothetical data from a follow-up study like that done by Boice
and Monson. Why is it not possible to calculate from the information below the number of
breast cancer cases expected for the period 1950-1969 with the method used by Boice and
Monson (as described above)? (Note: this is a "sticky" question. Do not try to calculate or
derive numbers.)
Distribution of Women-Years (WY) among exposed subjects
Period
Age 1950-54 1955-59 1960-64 1965-69
30-34 1900 -- -- --
35-39 1800 1700 -- --
40-44 1700 1600 1500 --
45-49 1600 1500 1400 1300
Average breast cancer incidence rates from the
Connecticut Cancer Registry (1950-1969), by age (rate per 1000 WY)
Age (years) Rate
30-34 .2
35-39 .4
40-44 .8
45-49 1.2
c. What advantage does this adjustment procedure have over simple age adjustment?
_____________________________________________________________________________________________
EPID 168 Standardization - Assignment - 150
rev. 9/2/91, 8/4/1999
4. Tuberculosis (TB) has been called the "captain of all men of death" because of its ability to
decimate populations. Improvements in the physical conditions of life in the present century,
especially nutrition, housing, and the work environment, greatly reduced this scourge even
before the advent of effective chemotherapy for the mycobacterium. The discovery of isoniazid
and its effectiveness in reducing infectiousness led to the application of public health measures
for tracing and treating active cases, thereby effectively controlling TB in the United States and
other developed countries. Indeed, U.S. public health policy has set the elimination of TB by
the year 2010 as a goal.
However, TB incidence in U.S. minority populations has never been reduced to the same extent
as the overall U.S. incidence, and the ratio of TB risk in nonwhites to whites has grown steadily
from about 3 in the mid-1950s to over 5 in the mid-1980s. In 1986, however, the long-term
decline in TB was reversed, with an estimated 9,226 cases in 1985-87 beyond those projected
from the 1981-84 trend. The 25-44 year age group had the largest 1985-87 increase, made up of
a 17% increase among non-Hispanic blacks and 27% among Hispanics. The HIV epidemic has
been implicated in the upswing in tuberculosis; poverty, homeless, and immigration of persons
from higher TB areas may also have a role. [Source: Reider HL, Cauthen GM, et al.
Tuberculosis in the United States. JAMA 1989 (July 21); 262(3):385-389.]
In this question, you are asked to interpret data from three North Carolina counties. The
following tables show the number of TB cases during the period January 1, 1986 to December
31, 1990, the mean population during that time period, and the corresponding U.S. TB rates.
Cases of tuberculosis in three N.C. counties
during January 1, 1986 - December 31, 1990
White White Nonwhite Nonwhite
County males females males females
Johnston 11 8 43 13
Orange 5 3 3 4
Wilson 6 10 51 27
Source: NC TB Control Branch
Mean population sizes of three N.C. counties
during January 1, 1986 - December 31, 1990
White White Nonwhite Nonwhite
County males females males females
Johnston 31,721 33,955 6,910 8,078
Orange 34,542 37,649 7,510 8,753
Wilson 19,844 22,259 10,692 12,788
Source: (Log Into North Carolina [LINC] database)
_____________________________________________________________________________________________
EPID 168 Standardization - Assignment - 151
rev. 9/2/91, 8/4/1999
Mean annual incidence of tuberculosis,
United States, January 1, 1986 to December 31, 1990
White White Nonwhite Nonwhite
males females males females
Cases per 7.4 3.6 39.2 19.8
100,000
Source: Centers of Disease Control, Tuberculosis in the United States
Your interpretation should compare the counties to each other and to the U.S. Is there a
greater-than-expected TB incidence in any of the counties? Is an increase confined to particular
race-sex-groups?
Suggestions:
a. Compute the race-sex-specific TB rates for each county and overall.
b. Compute an SMR comparing each county to the national TB rates.
5. .This question is optional. If you like it, do it; if you don't like it, forget it! Show that:
a. if age-specific rates for group A are all equal and age-specific rates for group B are all equal
(but not equal to those in group A, i.e., rai = ra and rbi = rb for all i), then:
Directly standardized rate for A Crude rate for A
―――――――――――――――――― = ――――――――――
Directly standardized rate for B Crude rate for B
Under what conditions will this ratio equal the ratio of indirect standardized rates?
c. if age-specific rates in groups A and B are not all equal, but for each stratum
rai
–––– = K [Where K is the same for all strata ]
rbi
then SMR (for A using B as the standard) = K
d. If the proportional age distributions in two populations are identical, then direct adjustment,
indirect adjustment, and crude rates are all comparable between the two populations.
6. (Optional) Solve problem #1 using a computer spreadsheet.
_____________________________________________________________________________________________
EPID 168 Standardization - Assignment - 152
rev. 9/2/91, 8/4/1999