Shifting age-parity distribution of births and the decrease in infant mortality.
- [show abstract] [hide abstract]
ABSTRACT: In 5 years the maternal and infant care projects of the Children's Bureau have had significant impact. They have extended comprehensive maternal health services to significant numbers of women of low socioeconomic status, apparently have catalyzed a reduction in infant and perinatal mortality, and have demonstrated models for maternal and infant care to medical students and other members of the health team. The projects have also revealed problems in the administration of health services by political subdivisions, difficulties in evaluation of service programs, and a sincere concern about maintenance of standards of care. (C) 1970 The American College of Obstetricians and GynecologistsObstetrics and Gynecology 02/1970; 35(1):114-9. · 4.80 Impact Factor
- Health services reports 11/1973; 88(8):733-742.
- American Journal of Public Health 09/1972; 62(8):1130-4. · 3.93 Impact Factor
Distribution of Births
and the Decrease in
NAOMI M. MORRIS, MD, MPH
J. RICHARD UDRY, PhD
CHARLES L. CHASE, MSPH
Possible causes for a decline in the infant mortality rate
during the late 1960s are proposed and examined
The infant mortality rate in the United States has
declined since 1960. From 1960 to 1965 the decline was
negligible and was a cause for alarm. Most of the decline has
occurred since 1965. During the period 1965 to 1970 the
rate dropped from about 25 to about 20 deaths per 1,000
live births.; Data through 1972 indicate that it is continuing
to fall at the same rate.
mortality rate is related to the falling birth rate observed
during the late 1960s. The decline in the overall infant
mortality rate could have been caused by births proportion-
ately shifting from high mortality risk age and birth order
cells to lower risk cells.* This explanation thus attributes
The authors are with the Department of Maternal and
Child Health, School of Public Health, University of North
Carolina, Chapel Hill, North Carolina 27514. This article
was prepared for presentation to the Biostatistics Section at
the 100th Annual Meeting of the American Public Health
1972, and was revised December, 1974. The research was
supported by Grant MC-R-370029-06-0, from the Research
Division, Maternal and Child Health Services, HSA, Depart-
ment of Health, Education, and Welfare.
*Defining average risk as a rate equal to the observed
overall infant mortality rate for the year plus or minus 10
per cent, using data from the 1960 Live Birth Cohort
Study, high risk age-birth order cells are seen to be any
births to girls under 15; parities 2 and over born to mothers
15 to 19; parities 4 and over, mothers, 20 to 24 and 45 to
Jersey, November 15,
individual decisions and behavior). We will show that the
changing age-birth order distribution of births since 1965
does account for about 27 per cent of the reduction in the
United States infant mortality rate that has occurred.
Another possible explanation concerns the advent of
better maternity and infant care services for high risk
In 1965 the first Maternity and Infant Care
Projects (MIC), funded by the U.S. Children's Bureau, were
operating. By 1970 more than 125,000 women per year
(counting new maternity admissions only) were being
served by such projects. The effect would presumably be
the reduction of the risk of infant mortality within high
risk categories to something on the order of risk experi-
enced by low risk categories.
Nicholas Wright has speculated concerning the first
explanation.' Reviewing developments of the 1960s, he has
stated, "With 1960 birth order-specific infant mortality
rate to "family planning."
49; parities 5 and over, mothers 25 to 29; parities 6 and
over, mothers 30 to 44. Low risk age-birth order cells are
parities 1 and 2, born to mothers 20 to 24; parities 1 to 4,
born to mothers 25 to 29; parities 2 to 4, born to mothers
30 to 34; and parities 2 to 5, born to mothers 35 to 39.'
National Natality Survey and the National Infant Mortality
Survey of 1964-1966, but inasmuch as these data were
confined to legitimate, single births we prefer to use the
more representative data from the 1960 Live Birth Cohort
Study for the calculations in this paper.' *2
is supported by data from the
rates held constant, the birth order distribution in 1968 on
the whole favored a lower overall infant mortality rate than
in 1960." However, he thought that this effect would be
cancelled out by the changing distribution of maternal age
toward lower maternal ages between 1960 and 1968.
Data supporting the impact of MIC services on infant
mortality are equivocal.3 Wholey and Silver estimated in
1966 that MIC projects could reduce infant mortality in
high risk patients to a level closer to the national average, or
from about 41 per 1,000 live births to about 24 to 29.4
This estimate is apparently supported by some data such as
mortality (which usually constitutes approximately 90 per
cent of infant mortality) for project patients 15 years old
or younger was 19
nonproject patients it was 37.5 The socioeconomic status
of the two groups was not matched, however, leaving some
reservation as to whether the difference in mortality rates is
prenatal and infant
provided by the MIC projects.
In discussing only pregnant women and their babies,
we are not giving credit for high risk births averted by the
family planning services offered by MIC projects. These
services, made available particularly to medically indigent
women, have been an important contribution of the MIC
projects. Without any measure, however, of exactly how
many high risk births have been prevented through these
services, we cannot estimate the potential effect they may
have exerted on the national infant mortality rate. We can
show that the maximum contribution of MIC prenatal and
infant care services to the reduction in national infant
mortality rate is quite modest.
project,in which neonatal
births, while for
Age-birth order-specific infant mortality rates (rates for
each age-birth order cell) for births occurring in the United
States during 19601 *
distributions of births for each of the succeeding years,
yielding estimated numbers of deaths in each cell for those
succeeding years under an assumption of constant age-birth
order mortality rates within each cell. The sum of the
estimated deaths for each year is used to estimate an infant
mortality rate for that year, and the estimated rate is then
plotted on a graph with the actual rate for the same year. A
least squares regression line is computed for each of the two
sets of rates for all of the years for which data are available.
These are plotted and the slopes are examined.
This calculation assumes that other characteristics of
the mother count for little, compared to age and parity, in
determining the probabilities of infant death. This assump-
tion does some violence to current knowledge, but is
necessary to be able to estimate the effects on infant
mortality of changing age-parity distributions from vital
to age-birth order
*Data from United States 1960 Live Birth Cohort
Study. The study is described in Reference 6.
obtained in MIC projects on the infant mortality rate, we
compare the actual rates with estimates of the rates under
the assumption of no MIC projects. To do this we add to
the numbers of infant deaths for each year more deaths
which represent those assumed to be prevented by MIC
projects. The highest risk identifiable groups in the United
States have infant mortality rates which are around 40 per
1,000 live births, while the overall U.S. rate is about half
that. We give the benefit of the doubt to the MIC projects,
and assume that they have reduced the rate in the patients
they have served by about 20 per 1,000 during the
the effect of prenatal and infant care
Figure1 shows the declining trend in the infant
mortality rate for 1960-1972, the years for which we have
age-birth order distributions for births available to us. The
solid line is the linear fit of the triangle points beginning
with 1965. The circle points represent the estimated infant
mortality rate assuming constant age-birth order-specific
mortality rates since 1960 and the broken line is the linear
fit of those points beginning with 1965. The estimated rates
have been adjusted upward by the constant 1.3 so that the
estimate for 1960 coincides with the actual rate (26.0). The
estimate was 24.7 for 1960, the difference being due to
incomplete matching of birth and death records in the
Cohort Study from which the age-birth order-specific
rates came. The
represents the part of the decline which was attributable to
the changing age-birth order distribution from high risk to
low risk categories. The remainder of the decline-the
difference between the broken line and the solid line-is
attributable to reduction of age-birth order-specific mortal-
ity rates. Numerically, the slope of the broken line is -0.24
and the slope of the solid line is -0.89. About 27 per cent
slope of the broken line
0 ESTIMATED MORTALITY RATE
A ACTUAL INFANT MORTALITY RATE
ESTIMATED b- -. 24
AJPH APRIL, 1975, Vol. 65, No. 4
TABLE 1-Number of New Maternity Patients Served by MIC
Estimated Numbers of
Prevented by MIC Projects, United States, 1965-1973
Dr. Donald A. Trauger, Chief, Health Programs,
Research Branch, Division of Research, U.S. Children's Bureau,
SRS, Department of Health, Education, and Welfare, and Promoting
the Health of Mothers and Children, FY 1973, U.S. Department of
Health, Education, and Welfare Publication No. (HSA) 74-5002,
Health Services Administration, Maternal and Child Health Service,
Rockville, MD, 1973.
A ACTUAL INFANT Mo1tTAUIVY RATE
0 ESTIMATED R.ATE AFTER
RECEIVING EFFECT OF MIC
ACTUAL b *-.87
of the decline in overall rates is thus attributable to the
shifts in age-birth order birth distributions.
Table 1 shows the numbers of new maternity patients
which were served by MIC projects in the United States
during the years 1965-1973. Under the assumption that
these patients would have contributed 20 additional deaths
enrollment in the MIC projects had there been no MIC
projects, the last column shows the deaths that are added to
the actual deaths in order to estimate the hypothetical
infant mortality rate in the absence of MIC projects. These
estimates are shown in Figure 2 as the circle points, and the
broken line is the regression line associated with them. The
births during the year following
triangle points again represent the actual rates for these
years; the solid line is the regression line from them. The
slope of the solid line is -0.87 and the slope of the broken
line is -0.78, indicating that about one-tenth of the decline
in infant mortality over these years might (given our
assumption) be attributed to the prenatal and infant care
services of the MIC projects.
Whereas shifts in birth order distributions for births
since 1960 that would favor lower infant mortality rates
were thought previously to be cancelled out by shifts in
maternal age distributions that would raise the mortality
rate, our data show that the net effect actually accounts for
about 27 per cent of the decline in infant mortality rates.*
Because we do not know the age-birth order distribution
for the infant deaths that did not occur to patients of MIC
projects we cannot make statements about an MIC project
effect that is additive with the effects due to shifting
age-birth order distributions. In fact, it is likely that some
of the improving statistics for MIC project deliveries are
related to lower initial risk as a result of more favorable
age-birth order distributions.
However, 73 per cent of the reduction in infant
changes1. This percentage must be due to
age-birth order-specific declines in infant mortality rate.
Under the most generous assumptions, only a small fraction
of the decline could possibly be attributed to the success of
MIC projects alone in delivering high quality services to
high risk mothers and infants. Nearly three-fourths of the
total reduction in infant mortality rate since 1965 remains
to be explained.
for by age-birth order
Twenty-seven per cent of the reduction in the U.S.
infant mortality rate for 1965-1972 is attributable solely
to shifts in age of mother and birth order of infant.
Individual "family planning" has thus made a greater
contribution to the reduction of infant mortality than has
been previously realized.
1. Wright, N. H. Some Estimates of the Potential Reduc-
tions in the United States Infant Mortality Rate by
Family Planning. Am. J. Public Health 62:1130-1134,
*Since this paper was originally presented, Gendell
and Hellegers have published data on changing perinatal
mortality rates in Baltimore 1961-1966.8 Their findings
for perinatal mortality in this 5-year period of falling birth
rates in Baltimore were similar to ours for U.S. infant
mortality 1965-1972, namely that ".
drop" was attributed to age-birth order shifts.
.. one-quarter of the
2. Infant Mortality Rates: Relationships with Mother's
Reproductive History, United States. U.S. Department
of Health, Education, and Welfare, Vital and Health
Statistics Series 22, No. 15, 1973.
3. Pearse, W. H. The Maternity and Infant Care Program.
Obstet. Gynecol. 35:114-119, 1970.
4. Wholey, J. S., and Silver, G. A. Maternal and Child
Health Care Programs.
U.S. Department of Health,
Education, and Welfare, 1966.
5. Zackler, J., Andelman, S. L., and Bauer, F. The Young
Gynecol. 103:305, 1969.
Risk. Am. J.
6. Chase, H. A Study of Infant Mortality from Linked
Records: Method of Study and Registration Aspects.
National Center for Health Statistics, Series 20, No. 7,
7. Vital Statistices of the United States-Natality Volumes
for 1960-1968. U.S. Department of Health, Education,
and Welfare. Annual Summaries, Vital Statistics of the
United States, 1972.
8. Gendell, M., and Hellegers, A. E. The Influence of the
Changes in Maternal Age, Birth Order, and Color on the
Health Serv. Rep. 88:733-742, 1973.
CALL FOR NOMINATIONS FOR SECTION ELECTIONS
Suggestions are being sought by each of the Section Nominating Committees of the American
Public Health Association in order to develop a slate of candidates for Section elections to be held
later this year. The nominations should be submitted as soon as possible, since the Section
Nominating Committees must submit their slate of nominees to APHA headquarters by April 15,
Each Section has at least one vacancy occurring annually for a Section officer (i.e., chairman,
vice-chairman, secretary, or secretary-elect), two seats on the Section Council, and at least one seat on
the Governing Council.
Each Section must provide, in writing, the following information on each nominee for each
office: full name, academic degree, current position, address, and current and past APHA activities.
This information should be sent to the respective Nominating Committee Chairpersons, listed below,
in care of APHA, 1015 Eighteenth Street, NW, Washington, DC 20036.
Chairpersons of Section Nominating Committees are: Community Health Planning, Patricia
Evans, MPH; Dental Health, Martha Fales, PhD; Environment, William V. Hickey; Epidemiology, E.
Russell Alexander, MD; Food and Nutrition, Joan Carter, MPH; Health Administration, Franklin D.
Yoder, MD, MPH; Injury Control and Emergency Health Services, to be appointed; Laboratory,
Elizabeth D. Robinton, PhD; Maternal and Child Health, Fred Seligman, MD, MPH; Medical Care,
Jerry A. Solon, PhD; Mental Health, Anna Barker; New Professionals, Dorothy Bellas; Occupational
Health and Safety, Paul H. Witt, MA; Podiatric Health, Theodore Clarke, DPM; Public Health
Education, Caroline S. MacColl, MSPH; Public Health Nursing, Edith Wright, RN, MPH; Radiological
Health, Robert C. Will, MS; School Health, Emily Hammond, MEd; Social Work, Robert W.
Lindstrom, MSW; Statistics, Earl S. Pollack, ScD; Veterinary Public Health, to be appointed.
Section election deadlines are as follows:
April 15-Section Nominations Committees must submit nominations slates in writing
August 1-Nominations by petition due to Section Nominating Chairpersons
August 8-Nominations by petition due at APHA headquarters
August 15-Ballot is mailed to each Section member
September 15-Ballot is due at APHA headquarters
Section nominations slates will be published in the July issue of the American Journal of Public
Health. Members will have the opportunity to add names to slates by the petitioning process.
AJPH APRIL, 1975, Vol. 65, No. 4