University of Southern California Dissertations and Theses

Determinants Of Neonatal And Postneonatal Mortality: The Case Of Malawi, 1992

(USC Thesis Other)

Determinants Of Neonatal And Postneonatal Mortality: The Case Of Malawi, 1992

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI
films the text directly from the original or copy submitted. Thus, some
thesis and dissertation copies are in typewriter face, while others may be
from any type of computer printer.
The quality of this reproduction is dependent upon the quality of the
copy submitted. Broken or indistinct print, colored or poor quality
illustrations and photographs, print bleedthrough, substandard margins,
and improper alignment can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete
manuscript and there are missing pages, these will be noted. Also, if
unauthorized copyright material had to be removed, a note will indicate
the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by
sectioning the original, beginning at the upper left-hand comer and
continuing from left to right in equal sections with small overlaps. Each
original is also photographed in one exposure and is included in reduced
form at the back of the book.
Photographs included in the original manuscript have been reproduced
xerographically in this copy. Higher quality 6” x 9” black and white
photographic prints are available for any photographs or illustrations
appearing in this copy for an additional charge. Contact UMI directly to
order.
UMI
A Bell & Howell Information Company
300 North Zeeb Road, Ann Arbor MI 48106-1346 USA
313/761-4700 800/521-0600

DETERMINANTS OF NEONATAL AND POSTNEONATAL
MORTALITY: THE CASE OF MALAWI, 1992
by
Ladislas Ritchken Silino Mpando
A Thesis Presented to
The FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF SCIENCE
(SOCIOLOGY - APPLIED DEMOGRAPHY)
December, 1994
UMI Number: 1378427
UMI Microform 1378427
Copyright 1996, by UMI Company. All rights reserved.
This microform edition is protected against unauthorized
copying under Title 17, United States Code.
UMI
300 North Zeeb Road
Ann Arbor, MI 48103
UNIVERSITY O F SOUTHERN CALIFORNIA
T H E G R A D U A T E SC H O O L
U N IV E R S IT Y PA R K
LO S A N G E L E S . C A L IF O R N IA 9 0 0 0 7
This thesis, written by
Jk A > ^ J . f c J . ^ . 9 . &>. ^ ^ 0
i
under the direction of h.i>. Thesis Committee,
and approved by all its members, has been pre
sented to and accepted by the Dean of The
Graduate School, in partial fulfillment of the
requirements for the degree of
Master of Science, Sociology (Applied Demography)
Dean
D ate November 30, 1994
THESIS COMMITTEE
Chairman
TABLE OF CONTENTS
CHAPTER PAGE
I INTRODUCTION 1
1.1 Country Background and Setting 4
II THEORY AND LITERATURE REVIEW 8
2.1 Theoretical Framework and
Analytical Methods 8
2.2 Literature Review 11
III SOURCE AND QUALITY OF DATA 21
3.1 Source of Data 21
3.2 Data Quality and Evaluation 22
3.3 Hypotheses, Variables and their Measurement 25
3.3.1 Socioeconomic Variables 26
3.3.2 Demographic Variables 29
3.3.3 Health and Nutrition Variables 33
IV DESCRIPTIVE ANALYSIS: LEVELS OF EARLY
CHILDHOOD MORTALITY 35
4.1 Early Childhood Mortality Differentials by
Selected Socioeconomic Characteristics 36
4.2 Early Childhood Mortality Differentials by
Selected Characteristics 40
V MULTIVARIATE ANALYSIS: EMPIRICAL RESULTS 43
5.1 Neonatal Mortality 45
5.2 Postneonatal Mortality 51
VI SUMMARY AND CONCLUSIONS 53
REFERENCES 57
ii
LIST OF TABLES
TABLE
1 Salient Socioeconomic and Demographic Indicators
for Malawi, 1966-1992
2 Childhood Mortality Rates (per 1000 Live Births)
by 5-year Period Preceding the MDHS: Malawi, 1992
3 Childhood Mortality Rates (per 1000 Live Births)
by Socioeconomic and Demographic Characteristics:
Malawi, 1992
4 Effects of Socioeconomic and Demographic Character
istics on Neonatal and Postneonatal Mortality:
Logistic and Proportional Hazards Estimates of
Relative Risks
PAGE
5
35
36
48
iii
ABSTRACT:
This study uses the 1992 Malawi Demographic and Health
Survey data to examine the determinants of neonatal and
postneonatal mortality in Malawi. An explicit assumption here
is that determinants of neonatal mortality are different from
those of postneonatal mortality.
The study used SAS software package to perform logistic
regressions for neonatal mortality and proportional hazards
regressions for postneonatal mortality. It focused on the
children who were born during the 5-year period preceding the
survey.
The study concluded that demographic factors (size of the
child at its birth and age of the mother at the birth of the
child were important determinants of deaths among children who
were less than 1 month old. With respect to postneonatal
mortality, the study showed that age of mother at birth,
household economic status, length of preceding birth
intervals, birth order of the child and duration of
breastfeeding emerged as powerful determinants.
CHAPTER 1
INTRODUCTION
This study focuses on identifying the determinants of the
high levels of deaths among children who were not yet 1 month
old (neonatal mortality) as well as the deaths among the
children who had survived the first month of life but failed
to celebrate their first birth day (postneonatal mortality) in
Malawi in 1992. The 1992 Malawi Demographic and Health Survey
(MDHS) show that while neonatal mortality showed encouraging
trends of decline during the 15-year period preceding the
survey, postneonatal mortality revealed increasing trends
during the same period. The other indicators of childhood
mortality, such as infant mortality (probability of a child
dying before the first birth day), or child mortality (the
probability of dying between the first and fifth birth day),
or under-five mortality (probability of dying between birth
and fifth birth day) also indicated similar trends. The
levels of childhood mortality in Malawi were among the highest
in Africa. In this study, early childhood mortality refers
to mortality of infants who have not yet reached their first
birth day.
However, global level, the world has experienced some
substantial declines in childhood mortality since mid 1940's.
These substantial declines in both infant and child mortality
1
which have been observed especially in most developing
countries since World War II appear to have resulted from
improvements in peoples standards of living as well as from
national and international public health activity (Hill and
Pebley, 1989). The speed at which these mortality declines
have taken place has, however, varied by region and country,
most likely depending on levels of socioeconomic. For
instance, while child mortality in Africa declined from 230
deaths per 1000 live births during the period 1960-65 to 200
two decades later (a decline of 13 percent) in Africa,
corresponding declines in Latin America were from only about
102 to 40 deaths per 1000 live births during the same period
(a decline of 59 percent).
In Africa from 1965 to 1985, child mortality rates
declined from 345 to 285 deaths per 1000 live births (a
decline of 17 percent) in Malawi. In Zimbabwe and Uganda
respectively mortality declines were from 155 to 95 (a decline
of 39 percent) and from 195 to 185 deaths per 1000 live births
(a decline of 5 percent) (Ewbank, 1993; Gribble, 1993). With
deteriorating economic conditions in many developing countries
aggravated by natural disasters such as droughts, chloroguine
resistant malaria, etc.. child mortality may fail to go down
further despite the recorded economic development that had
started taking place after the 1940's. According to the World
2
Health Organization (WHO), under-five mortality reflects the
level of socioeconomic development of a nation while the
infant mortality is used as an indicator of the level of
health facilities (WHO, 1981 cited in Gbesemete and Jonsson,
1993). In general, it is known that mortality varies by
socioeconomic and demographic factors but the factors
influencing mortality levels may vary by country.
The child mortality declines in Malawi, the study area of
this paper, were very modest during the 1965-92 period despite
the level of economic development the nation had experienced.
For instance, the 1992 MDHS results show that one in every
four children failed to survive to their fifth birth day. The
infant mortality rate remained virtually constant during the
15-year period preceding the survey. While neonatal
mortality declined dramatically by 34 percent, postneonatal
mortality rose by 26 percent, thus somehow offsetting the
neonatal decline.
Like many other developing countries, Malawi is currently
committed to reducing its childhood mortality as a step
towards achieving the goal of 'health for all by the year
2000'. In order for the Government of Malawi to be able to
draw more feasible strategies to improve child survival,
research exploring the possible determinants of childhood
3
mortality needs to be undertaken. An understanding of the
socioeconomic and demographic factors influencing the levels
and trends childhood would therefore be crucial in order to
develop plausible policies and programs for childhood
mortality reduction.
The health planner in Malawi not only needs to identify
these vulnerable groups in the population which have higher
mortality levels, but if policies are to have an impact on
lowering childhood mortality levels, he/she also needs to seek
the factors which lead to such differentials. It is,
therefore, hoped that the results of this study will be useful
for the policymakers in Malawi to devise pertinent strategies
to lower the soaring childhood mortality rates.
It is, therefore, the purpose of this study to explore
and establish determinants of early childhood mortality in
Malawi. While similar studies may have been conducted in
other countries, factors affecting childhood mortality vary
from country to country. However, such studies have guided in
the selection of variables in this study.
1.1 Country Background and Setting
Malawi is a landlocked country in South-eastern Africa.
It is 118,500 km2 in area of which 94,300 km2 is land area.
4
About 56 percent of the area is arable. The country is divided
into three administrative regions, namely Northern, Central
and Southern. The 1987 Malawi Population and Housing Census
results show that population was unevenly distributed in
Malawi. About 11 percent of the total population lived in the
Northern Region, while 39 and 50 percent lived in the Central
and Southern Regions respectively. The population density in
Northern, Central and Southern Regions were 34, 87 and 125
persons per km2 respectively (NSO, 1991).
Malawi is predominantly an agricultural country and
agricultural produce accounts for 90 percent of its total
exports (tobacco, tea and cotton) (NSO, 1994). Malawi's per
capita GNP was estimated at US $230 in 1992 (PRB, 1994).
Table 1 shows a selection of some salient demographic
indicators for Malawi:
Table 1: Salient Socioeconomic and Demographic
Indicators for Malawi, 1966-1992
Indicator 1966
CENSUS
1977 1987
MDHS
1992
Population (million) 4.04 5.55 7.99
—
Population under 1 yr(%)
— —
3.7 4.2
Population under 5 yr (%)
- -
17.4 16.3
Literacy Rate (%)
Male
- -
52.4
-
Female - -
31.6 -
Dependency Ratio 0.92 0.96 1.00 1.05
5
Table 1 (Continued)
Indicator 1966
CENSUS
1977 1987
MDHS
1992
Access to Piped Water (%)
- -
37.0 25.2
Access to Toilet Facility
- -
67.4 72.3
Intercensal Growth Rate(%)
-
2.9 3.2
-
Percent Urban 5.0 8.5 10.7
-
Percent of Women in
Childbearing Age 47. 6 45. 1 44.2 41. 1
Mean Age at Marriage
(Females)
-
17.7 17. 1 17.6
Contraceptive Prevalence
Rate (%)
- - - 13.0
Crude Birth Rate
-
48.3 41.2 42.9
Total Fertility Rate - 7.6 7.6 6.7
Crude Death Rate
-
25.0 22.9
-
Infant Mortality Rate - 165.0 151.0 134.3
Under-5 Mortality - 321.0 286.0 233.8
Life Expectancy at Birth:
Males 39.2 41.4
Females
—
42.4 44. 6
—
Source: Malawi Demographic and Health Survey Report, 1992
- = not estimated/available
The Government of Malawi has in the recent past viewed
its fertility and mortality levels as too high. This has led
to the preparation of a national population policy which aims
6
at achieving lower population growth rates that are compatible
with the attainment of the country's social and economic
objectives in addition to reducing morbidity and mortality
among mothers and children (NSO, 1994).
Health services in Malawi are provided by the Ministries
of Health and Local Government and non-governmental organiza
tions (NGOs). The Ministry of Health is responsible for
planning and developing health policies and for providing care
in all government hospitals while the Ministry of Local
Government is in charge of health care delivery at district
level or below. NGOs provide services to both hospitals and
smaller medical units.
This thesis is divided into six chapters: Chapter I is
the Introduction and includes a statement or a brief
description of the research question. Chapter II is a review
of literature and also describes the study methodology as well
as the theoretical framework on which the study hinges.
Chapter III discusses data sources and data quality as well
as variables hypothesized to influence early child mortality
in Malawi. Chapter IV describes levels of early child
mortality by selected socioeconomic and demographic variables.
Chapter V presents the results of the multivariate statistics
as well as the discussions of the results and Chapter VI is a
summary of the study and policy recommendations.
7
CHAPTER II
THEORY AND LITERATURE REVIEW
2.l Theoretical Framework and Analytical Methods
The theory for this study comes from Mosley/Chen
analytical framework for the study of determinants of child
survival in developing countries, with some modifications
based on the limitations of the available MDHS data. This is
a proximate determinants approach and parallels that used by
Davis and Blake in developing an analytical framework for
fertility research (Davis and Blake cited in Mosley/Chen,
1984). Explicit in the Mosley/Chen framework is the premise
that there must exist a set of proximate, or intermediate
determinants that directly influence the chances of survival
(Mosley/Chen, 1984) . These proximate determinants are grouped
into five categories as follows: maternal factors,
environmental factors, nutritional deficiency, injury and
personal illness control (Mosley/Chen, 1984).
The factors hypothesized to influence early childhood
mortality in this study will fall into some of these
categories.
The unit of analysis in this study will be the individual
child. The sample of individual children consists of all
8
those who were aged less than 60 months at the time of the
survey; in other words, all children born during the period
January, 1987 and the day of the interview.
In order to investigate the influence of each of the
explanatory variables on the dependent variable, the study
employs binary logistic regression for neonatal mortality.
Logistic regression is an appropriate method for multivariate
analysis when the dependent variable is dichotomous. Because
our dependent variable for neonatal mortality analysis is
whether the child died or survived within the first month of
life, and thus as a probability it must range between 0 and 1,
the logit transformation is used. The independent variables
are then regressed on survival status as in the following
model:
logit (z) = In {Z / (1—Z) } = a + 61X1 + B2X2 +...+ BkXk
where Z= the logit transformation of survivorship status
(that is the log odds of surviving),
Bi = regression constants to be estimated,
and X's = independent variables (Aldrich and Nelson, 1984).
For postneonatal mortality, however, Cox's proportional
hazards models in preference to the usual ordinary logistic
regression models. The proportional hazards models, unlike
the other models, are able to handle truncated or censored
9
observations. Additionally, they allow the inclusion of
variables whose values change over time (Allison, 1984).
In using the ordinary logistic regression models, some
information is lost since all censored observations would be
excluded from analysis. This may substantially reduce the
sample size and consequently the estimates obtained may be
severely biased.
In this study, duration of breastfeeding pose the
problems of censoring, hence the decision to use proportional
hazards models.
The proportional hazards models take the general form
(Allison, 1984):
logeh(t) =a(t) +Eb1X, + EcjYj (t) ..... equation 1
where X5 represents a set of covariates for each
individual whose characteristics are constant over time; Yj
represents a set of time-varying covariates b , - and Cj in
equation 1 are sets of regression-like coefficients indicating
the effects of the independent variables in shifting upward or
downward, a time-varying baseline function a(t) in equation
1. These coefficients are estimated by partial likelihood
method. In this study of postneonatal mortality, the PHREG
computer programs in SAS will be used. Factors affecting
neonatal mortality will be modeled separately (using logistic
10
regression) from those affecting the risk of dying during the
postneonatal period (using proportional hazards procedures).
2.2 Literature Review
Although declines in childhood mortality have occurred in
most developing countries over the last five decades, some
regions and countries have enjoyed more substantial declines
than others DaVanzo, 1988; Hill and Pebley, 1989). Factors
which influence mortality vary from country to country. This
fact coupled with the desire by each individual government to
attain the goal of achieving 'health for all by the year 2000'
has instigated numerous studies on the determinants of health.
A notable research by Mosley/Chen (1984) culminated in the
development of an analytical framework for the study of
determinants of child survival in developing countries. This
approach has gained popularity among many researchers, and has
extensively been used in many studies, albeit with modifica
tions to suit the available data.
The following factors have received attention in previous
studies:
(a) Parental Education
Considerable research has demonstrated an inverse
relationship between maternal education and childhood
mortality. The nature of this relationship is still not
11
understood (Bicego and Boerma, 1993). For example, in their
study of 'Child Mortality Differentials in Sudan', Farah and
Preston (1982) found that an increase in maternal education by
5 years was associated with an 18 percent reduction in child
mortality. Based on their study of 'Household Income and
Child Survival in Egypt', Casterline et al (1989), however,
found that maternal education did not have any significant
effects on child survival . This finding applied both with
and without controls for household income. Similarly, in
their research of 'Socioeconomic, Demographic and
Environmental Determinants of Infant Mortality in Nepal',
Gubhaju et al (1991) found that maternal education was not an
important determinant of infant mortality in Nepal. Several
hypotheses have been advanced to explain this association.
Maternal education, it is postulated, inculcates modern
knowledge and practices, improves the effectiveness of family
health behavior and enables the mother to take necessary
measures to promote child health, including effective use of
modern health services (Chen, 1983).
In their study of maternal education and child survival
where they used data from 17 developing countries, Bicego and
Boerma (1993) controlled for the household economic status and
found that the education-mortality relationship tended to be
more pronounced at older as opposed to younger ages of
12
childhood. Consistent results were also obtained in the study
of ' Effects of Community Factors on Infant and Child Mortality
in rural Bangladesh' in which Al-Kabir (1984) found that
education and the other factors related to education (such as
hygiene and nutrition) were more important determinants of
child than infant mortality.
While extensive research has been done on the maternal
education-child mortality association, not as much has been
done on paternal education-child mortality relationship. The
few studies that have been conducted have had mixed results.
For instance, Farah and Preston (1982) found that although
father's education had a significant impact on child mortality
in Sudan, the effect was only a third as large as that by
maternal education. The effect of husband's education on
childhood mortality in Urban Brazil was about half that of
mother's education (Merrick, 1985). These findings suggest
that mother's education is more important than father's
education on reduction of mortality, presumably because in
most developing societies childrearing or child care is the
mother's direct responsibility. However, in societies where
the majority of adult females are illiterate, education of
fathers may be very important (Pant, 1991). Other researches
have ignored the paternal education variable on the grounds
that it is often highly correlated with their wives' education
13
(Merrick, 1985). However, in the societies where education
among women is very low, the inclusion of husband's education
may turn out to be a very significant determinant on child
mortality, especially when the husband's income is excluded
(which often it is because of data problems). Education of
father can be considered as a proxy for the economic status of
the household because educated fathers are likely to earn more
than uneducated fathers and are thus able to provide better
services and care to their children.
(b) Place of Residence
Most researches classify a place of residence as urban or
rural, but the definitions of urban vary from one country to
another. A place of residence per se is not a determinant of
early childhood mortality but rather acts as a proxy for other
variables that are themselves determinants like availability
of health services, transportation, etc. Because of such
urban-rural differences, some researches in the past have
concluded that the main determinants of infant mortality in
urban areas are socioeconomic factors and are demographic in
rural areas (Kim, 1991; Gubhaju et al, 1991) .
(c) Source of Drinking Water and Type of Sanitation
A number of studies have underscored the role of safe
drinking water (piped water) and sanitation for the
14
improvement of the survival chances of children (Merrick,
1985; DaVanzo, 1984; Gubhaju et al, 1991; Pant, 1991). For
instance, in her study of Child Mortality Determinants in
Malaysia, DaVanzo (1984) found that the absence of modern
toilet sanitation and piped water was strongly associated with
mortality for babies who breastfed little or not at all.
However, the presence of these facilities made no significant
difference for the mortality of babies who were fully
breastfeeding without food supplementation. Presumably, the
reason was that babies who did not breastfeed usually had
other foods mixed with water which could be contaminated. In
the same study, it was also found that the beneficial
influence of breastfeeding was greatest when water and
sanitation were poor. In Sri Lanka, it was found that effects
of unclean water were felt more after the neonatal period than
during the neonatal period (Meegama, 1980). This suggested
that contaminated water had much greater impact on mortality
after the neonatal stage unlike during the first few weeks of
life when breastfeeding was common and where food requiring
the use of water was not given to the baby. It is often cited
in literature that one of the main causes of death during the
childhood period is diarrheal diseases, including
gastroenteritis. These diseases are transmitted through food
and water (IRD/Westinghouse, 1987; Meegama, 1980). In their
study of Determinants of Infant Mortality in Nepal, Gubhaju et
15
al (1991) found that the net effect of the probability of
dying during infancy of children born in households which used
drinking water from a river or lake was 44 percent higher than
the probability of dying of children born in households which
used piped/tube-well drinking water as their source.
Likewise, the net effect of probability of dying of children
belonging to households which did not have their own toilet
facility was 64 percent higher than the probability of dying
of children belonging to households which had their own toilet
facility.
(d) Index of Economic Status
Since data on income is usually not collected in many
surveys, several proxies for household income have been used
in many researches. For example, Bicego and Boerma (1993)
used household possessions (radio and television) to construct
an index of household economic status. They also considered
the type of materials used in floor construction of houses as
well as ownership of motorized means of transportation in the
construction of this index. Similarly, Farah and Preston
(1982) used the same method to construct an index that was to
be used as a proxy for household income. They found that
child mortality in Sudan was higher among children who were
living in traditionally constructed houses (those with mud
floors and grass thatched roofs) than their counterparts
16
living in modern houses (those with modern floor finish and
with iron sheets roofing). However, possession of modern
goods was negatively but weakly and insignificantly associated
with mortality.
(e) Age of Mother at Birth of Child
A large body of literature exists on the influence of
mothers' age at birth of their children. Most of them agree
that infants born to teenage mothers are more likely to die,
especially in the neonatal period, than those born to older
mothers (age 20-34 years). For instance, Gubhaju et al (1991)
found that in Nepal children born to mothers aged less than 20
years had a 47 percent higher risk of dying in infancy
compared to those born to mothers aged 20-2 9 years. In
Malaysia, DaVanzo (1984) also found that babies born to very
young mothers (less than 19 years) were much more likely to
die in the first month of life, and babies born to mothers
older than 40 years were also more likely to die in infancy.
Consistent results were also obtained by Meegama (1980) and
Al-Kabir (1984). The studies also found that in general first
births and births of birth order higher than 5 face an
elevated risk of dying during infancy than the other births.
In Nepal where childbearing starts at age 16 years, Gubhaju et
al (1991) found that first births were, as predicted, at a
higher risk of dying. They attributed this finding to the
17
fact that first order births occurred to women when they were
aged less than 20 years old. The first order births were also
more likely to be of low birth weight.
(f) Duration of Preceding Birth Intervals
Studies have also concluded that an infant's chances of
survival in the first year of life depend on the duration of
the previous birth intervals. In Philippines for example, the
chances of an infant dying were found to be 39 percent lower
when women spaced their children more than 24 months (East-
West Center, 1993). There are four explanations for this
effect. First, women who bear children rapidly do not have
adequate time to recover from the demands of a prior pregnancy
and breastfeeding, and may become nutritionally and physically
exhausted (or maternal depletion syndrome as it is sometimes
called). This may cause the birth of premature, underweight
infants and may also result in inadequate breast milk. A
second explanation based on sibling rests on the premise that
children close in age are placed in competition with each
other for the same maternal and familial resources. The
younger child is likely to suffer because the usually limited
resources may have to be spread more thinly among the
siblings. Third, larger numbers of children within the
household may facilitate not only the spread of infectious
diseases, such as measles, but also the severity of the
18
infection (Koenig et al, 1990; IRD/Westinghouse, 1987).
Fourth, short preceding birth intervals coupled with young age
of mother at birth of her child, or short previous birth
intervals coupled with survival status of preceding child,
have elevated the risks of children death particularly in the
first month of life (Gubhaju et al, 1991).
(g) Duration of Breastfeeding
A good deal of clinical and epidemiological research has
demonstrated the role of breastfeeding on child survival.
While researchers generally agree on the recognition of the
fact that breastfeeding provides the optimal form of infant
nutrition, there has been debate on the advantages of
prolonged breastfeeding without supplementation. Some have
argued that the relative advantages from both the immune and
nutritional potential of the mother's breast milk decrease
rapidly after six months (Palloni and Tienda, 1986). However,
there is general consensus that infant mortality is highest
when the infants are not breastfed at all compared to that
when they are breastfed with or without supplementation
(IRD/Westinghouse, 1987). Many studies have indicated that
breastfed infants experience lower mortality risks than
artificially fed infants in the developing world (Knodel and
Kintner, 1977). Consistent results were also obtained in
Malaysia (DaVanzo, 1984).
19
(h) Use of Tetanus Toxoid
Epidemiological studies have demonstrated that neonatal
tetanus is responsible for a large number of deaths among
infants unless their mothers are immunized against the disease
during pregnancy (IRD/Westinghouse, 1987). Several studies
have also shown that the incidence of neonatal tetanus is much
lower among hospital deliveries than among other births
(Ewbank and Gribble, 1993) . For example, in Kenya neonatal
tetanus mortality among infants whose mothers had not received
any tetanus toxoid injection were about three times as likely
to die of tetanus than their counterparts whose mothers had
received two doses of tetanus toxoid injection and had
delivered in hospitals. Studies of antenatal immunization
against tetanus have shown that two injections of tetanus
toxoid early in pregnancy are 95 percent effective in
preventing neonatal tetanus in the child born of that
pregnancy, but the effect of the injections declines over time
to 40 percent effective four or five years later (Koenig, 1992
cited in Ewbank and Gribble, 1993).
In Malawi, immunization of pregnant women with tetanus
toxoid began in 1984. Coverage increased from about 27
percent in 1985 to more than 60 percent in 1988 (Ewbank and
Gribble, 1993), and further increased to about 75 percent in
1992 (NSO, 1994). This immunization programme is now an
important component of antenatal care in Malawi.
20
CHAPTER III
SOURCE AND QUALITY OF DATA
This chapter describes the source of the data used in
this study. It gives details on survey methodology as well as
an evaluation of the quality of the data. The chapter further
describes the variables used in the study and how they are
measured. Hypotheses regarding the relationship between the
variables and early childhood mortality are also included
here.
3.1 Source of Data
The data used in this study were collected in the 1992
Malawi Demographic and Health Survey (MDHS). The survey was
a nationally representative sample survey designed to provide
information on levels and trends of fertility, early childhood
mortality and morbidity, family planning knowledge and use,
and maternal and child health. The sample comprised 5323
households from which 4849 eligible women age 15-49 and
eligible men age 20-54 were interviewed.
The MDHS sample of households was selected in two stages:
first, 225 enumeration areas (EAs) were selected from the
1987 census frame of 8652 EAs with probability proportional to
population size. This sampling frame of census EAs was
stratified by urban and rural areas within each of the three
21
administrative regions into which Malawi is divided. The
sample excluded households from Mozambican war refugee camps,
and institutions such as army barracks, police camps,
hospitals, etc.
In the second stage, a systematic sample of households
was then selected from the listing of the households with
sampling interval from each EA being proportional to its size
based on the results of the household listing operation. The
selected households formed the MDHS sample within which all
eligible women were interviewed. Further, a sub-sample of
households was drawn within which all eligible men and women
were interviewed (NSO, 1994) .
The health services availability questionnaire was
administered in each EA during the household listing
operation. A responsible respondent was then interviewed in
each EA. Thus the information on availability of health
services was collected at EA level.
3.2 Data Quality and Evaluation
The age ratio analysis of the MDHS done in this study
shows that there was no serious misreporting or misstatement
of the age of females up to age 44. However, some women in
age group 45-49 appear to have reported their ages to be 50-
54. Further, only 0.1 percent of all the women did not state
their age.
22
The early childhood mortality rates used in this study
were computed from data on births and deaths of children
described in the individual survey respondents' birth
histories. Unreported birth dates are a potential problem in
this type of data, but were uncommon in the MDHS; only 0.1
percent of births lacked a year of birth. A further 3.7
percent of the births had a year of birth but lacked a month
of birth, thus necessitating imputation of a birth date.
Furthermore, only 0.2 percent of deaths recorded in the birth
histories lacked an age at death. In this case of missing age
at death, a 'hot deck' procedure was used to impute this
information (NSO, 1994).
Misreporting of age at death will bias estimates of the
age pattern of mortality if the next result of the
misreporting is the transference of deaths between age
segments for which rates are calculated. In the MDHS there
was little heaping on the month of death of 12 months, and due
to strong emphasis during training and filed supervision, very
few deaths were reported to have occurred at age one year,
making any adjustment in infant mortality unnecessary (NSO,
1994). Thus while some researchers who are confronted with
this problem re-define infant mortality as mortality from
birth to age 18 months (Retherford et al, 1989) , or assume
that half of the deaths at age 12 months actually occurred
23
before age 12 months (Pant, 1991; Gubhaju et al, 1991)
adoption of either strategy was uncalled-for in the MDHS.
However, substantial heaping occurred on the duration of
breastfeeding, especially at 6, 12, 18 and 24 months.
A serious weakness with the birth history data is the
omission of children born during the reference period and to
mothers who were dead at the time of the interview. It is,
nevertheless, difficult to estimate the magnitude of this
error.
For all the socioeconomic variables used in this study,
less than 0.1 percent constituted the 'not stated' or
'missing' cases. Information on children's birth weight is
likely to be accurate because interviewers were instructed to
copy down the birth weight from the child's immunization card.
However, for almost 65 percent of the births, the birth was
either not weighed or the birth weight was missing. Thus the
use of this variable would substantially reduce the sample
size. Nevertheless, mothers were also asked for their own
subjective assessment of whether the baby was 'very large',
'larger than average', 'average', 'smaller than average' or
'very small' at birth. Almost 98 percent of all the births
fell in one of these categories. This variable is used as
proxy for birth weight in this study since weight and size at
birth of the child are likely to be highly correlated.
24
3.3 Hypotheses, Variables and their Measurement
The measures of early childhood mortality that are used
in this study are the probability of dying during the first
month of life, or neonatal mortality and the probability of
dying after the first month of life but before exact age of 12
months, or postneonatal mortality. Infant mortality, which is
the probability of dying before the first birth day, is the
sum of neonatal and postneonatal mortality and is thus not
considered as a separate dependent variable. It is crucial to
consider neonatal and postneonatal mortality separately
because as many researches have suggested, factors that affect
the chances of survival for the child transmit their effects
with different forces depending on the age of the child or its
stage of growth. For example, poor maternal nutrition may be
an important determinant for the health of the baby immediate
ly after birth, but once the baby has survived a certain
duration the factor loses its importance. Kim (1991)
concluded that postneonatal mortality is, in general, due to
exogenous factors whereas neonatal mortality is more
influenced by endogenous factors.
The following factors are posited in this study as having
the stated direction of influence on child survival.
25
3.3.1 Socioeconomic Variables
(a) Maternal Education
Past researches have concluded that maternal
education has the effect of reducing childhood mortality rates
although there is debate regarding the mechanism by which this
actually happens. Some have noted the role of education in
influencing a population's use of available health and medical
services, practice of personal hygiene,..etc. It is also
contended that an educated mother is likely to be more
receptive to medical, sanitary and better child care practice
that would eventually lead to improving the child's chances of
survival. Thus the educational level of a mother is
particularly vital to the health of her infants. For these
reasons, a negative relationship is posited between maternal
educational level and early childhood mortality.
This variable is measured by the number of years spent in
school, excluding repeated years. Note that the primary
school curriculum in Malawi lasts for 8 years whereas the
secondary school period is 4 years.
(b) Paternal Education
Paternal education has often been omitted for study in
many researches on the grounds that fathers do not directly
interact with infants. However, in the studies where this
26
variable was not omitted, mixed results have been obtained:
some suggesting that it has an influence on a child's health
in some societies and others suggesting that it does not.
It is argued here that especially in the cases where the
mothers are not educated, education of the father will be
important in the health care for the children. An educated
husband is more likely to encourage his uneducated wife on
issues of hygiene, sanitation and use of medical services, if
needed, than an uneducated husband married to an uneducated
wife. Additionally, education is likely to correlate strongly
with ones income, hence an educated husband is more likely to
earn higher income part of which can be used to purchase
health services in times of need. If the wife's educational
level is also high, it is likely that the husband's income
will not be as important as if the wife had no income to
purchase medical services. It is thus hypothesized in this
study that paternal education and early childhood mortality
will co-vary negatively.
This variable is measured in the same way as maternal
education.
(c) Availability of Piped Water and Sanitation
Since water-borne diseases, such as diarrhea, can be
transmitted to human beings through drinking water, and also
27
that microbes can be transmitted through poor sanitary
facilities, it is hypothesized that availability of adequate
safe drinking water as well as proper toilet facilities would
help curb down the spread of some of the water-borne diseases
which could eventually lead to death.
In Malawi, like many countries, piped water is tantamount
to safe drinking water because it is chemically treated. In
this study, this variable will be treated as a dummy variable
where drinking water takes the value of 1 if piped and a value
of 0 if not piped. Similarly, the availability of a toilet
facility will be considered as a dichotomous dummy variable
where availability of a toilet facility will be coded as 1 and
0 if the household does not have a toilet facility.
(d) Index of Household Economic Status
The ability of a household to purchase medical services,
nutritious food, etc depends largely on the income levels of
the household. Households with higher incomes would find it
easier to purchase these services than those with lower or no
income.
Information on household income was not collected during
the MDHS. However, a variable will be created on the basis of
the availability of durable assets such as cars, radios,
motorcycles, etc and the types of materials used in the
28
construction of a house. This variable will then be used as a
proxy for household income since households with high incomes
are more likely to be able to afford iron sheets for roofing
and cement or any modern floor finish for their houses and are
also likely to possess such goods as cars, motorcycles,
radios, etc. Two points will assigned for a house with
corrugated iron sheets or asbestos roofing and one point will
be assigned for a cement floor as opposed to the traditional
earth/mud floor. Two points will also be assigned for
availability of electricity in the house, or possession of
radio, motorcycle; and three points will be assigned for
possession of a car. A single point will be assigned for each
of paraffin lamp, a bicycle and an ox-cart. Thus the index
runs from the worst value of 0 to the best value of 16.
Households with an index of 5 points or less are arbitrarily
considered to be 'below average' while households with
indices of at least 6 points are considered to be 'above
average'. Bicego and Boerma (1993) and Farah and Preston
(1982) created similar variables in this way.
3.3.2 Demographic Variables
(a) Age of Mother at Birth
Some studies have shown that children born to mothers in
either very young (less than 20 years) or very old (more than
40 years) reproductive groups are more likely to die in
29
infancy. It is often cited in literature that very young
mothers are usually biologically, socially and economically
ill-prepared for childbearing and are maternally immature
(IRD/Westinghouse, 1987).
A U-shaped relationship between age of mother at birth of
a child and mortality in the first year of life has been found
in literature. There is no obvious reason to suspect that the
same relationship does not hold in Malawi, hence this study
predicts a similar type of relationship as that obtained in
other studies.
(b) Birth Order of Child
As a mother's parity increases, the children's
competition for often limited resources, such as food, medical
services or even mother's attention increases. Those who are
young are usually at a disadvantage and this makes them more
vulnerable to malnutrition-related diseases such as
kwashiorkor, and eventually their chances of survival are
greatly reduced. It is hypothesized in this study that high
birth order children, say 5 or higher, are more likely to die
in infancy than lower order births.
In Malawi where up to 75 percent of the women marry
before age 20, first born children may also experience higher
risks of dying since it is likely that these children are born
30
when the mothers are still young. The median age at marriage
for females is 18 years (NSO, 1994).
(c) Birth Intervals
Several researches have suggested that short birth
intervals (less than 24 months) have detrimental effects on
the infant's survival because of the infants' competition for
maternal care and nutrition. Although both children are at a
disadvantage the preceding child may be at a greater
disadvantage because the mother would probably put more
maternal attention and care to the closing child. A negative
association between birth intervals and early childhood
mortality is envisaged in this study.
As several definitions of birth intervals exist in
literature, the present study adopts the definition of a birth
interval as the interval between two consecutive births
(Koenig et al, 1990).
(d) Sex of Child
The sex of each child is unequivocally defined as male or
female. It is cited in literature that gender reflects
inherent genetic factors which favor girls better chances of
survival than boys (DaVanzo, 1983). For the same reasons, it
is hypothesized that early childhood mortality among male
infants will be higher than for their female counterparts.
31
(e) Size of Child at birth
It is assumed that the size of the child at birth is
highly correlated with its birth weight. Since high birth
weights are generally associated with good maternal nutrition
during pregnancy and use of antenatal clinics which would
improve the infant's chances of survival, the children who
were assessed by their mothers as large at birth will also
have improved chances of survival. A negative association is
therefore posited between size of child at birth and early
childhood mortality. Put differently, the larger the size of
the child at its birth the greater are the chances of
survival. Note that birth weights will not be used in this
study because of the enormity of missing and 'not stated'
cases.
The variable 'size at birth ' will be a trichotomous
dummy variable with categories 'very small', 'small' and
'average or larger'. The choice of these categories was
guided by the frequency distribution of the variable.
32
3.3.3 Health and Nutrition Variables
(a) Duration of Breastfeeding
After birth, breastfeeding is the optimal form of infant
nutrition and provides immunization against infections for at
least six months (Knodel and Kintner, 1977). When
breastfeeding is supplemented there is a likelihood of the
infant's exposure to water-borne diseases introduced when the
food supplements are mixed with water that may itself be
contaminated. Thus it is also important to consider the
interaction of the breastfeeding variable with the water
and/or sanitation variables.
It is envisaged that infants who are breastfed longer
would experience lower mortality than those who are breastfed
for a shorter period of time, all things being equal.
(b) Use of Tetanus Toxoid
All children were assigned to either use or non-use
categories based on their mothers' responses to a question
whether they received any injection to prevent tetanus in her
child during pregnancy.
It is hypothesized that mortality would be higher among
children whose mothers who had not received any tetanus-toxoid
injection than those whose mothers had.
33
(c) Use of Antenatal Services
All children were assigned to either use or non-use
categories based on whether their mothers responded that
during that pregnancy they had received antenatal care from a
health professional (ie doctor, nurse/midwife, or medical
assistant) or not.
It is hypothesized that the group of children whose
mothers had not attended antenatal clinic would experience
higher mortality than those whose mothers had. This is so in
view of the fact that mothers who never attended antenatal
clinic when they were pregnant with the child in guest ion were
very likely to have no medical assistance at delivery of the
child. If assistance at delivery was offered by a non-trained
birth attendant or fellow women, there would be a very high
likelihood of using unclean instruments to cut the umbilical
cord, and this may give rise to neonatal tetanus. Thus there
could also be interaction effects between use of tetanus
toxoid and use of antenatal services.
34
CHAPTER IV
DESCRIPTIVE ANALYSIS: LEVELS
OF EARLY CHILDHOOD MORTALITY
A preliminary analysis of the MDHS data shows that levels
of childhood mortality in Malawi are still very high although
the under-five mortality declined by an insignificant 9
percent from 258 deaths per 1000 live births during the period
1978-1982 to 234 during the 1988-92 period. Infant mortality
remained virtually constant at about 135 deaths per 1000 live
births during the same period (see Table 2) . Table 2 also
shows that neonatal mortality declined considerably by 34
percent from 62 to 41 neonatal deaths per 1000 live births at
the expense of postneonatal mortality. It would appear that
health programs in Malawi concentrated more on antenatal than
postnatal care, or perhaps that use of health services
decreased after birth of the child.
Table 2: Childhood Mortality Rates (per 1000 Live
Births) by 5-year Period Preceding the MDHS/
Malawi 1992
Approximate
Calendar
Period
Neonatal
Mortality
Postneo
natal
Mortality
Infant
Mortality
Under-five
Mortality
1988-1992 40.8 93.5 134.3 233.8
1983-1987 57.5 80.0 137.5 246.3
1978-1982 62.1 74.3 136.4 258.0
1983-1992 48.8 86.9 135.7 239.5
Source: 1992 Malawi Demographic and Health Survey Report
35
4.1 Early Childhood Mortality Differentials by Selected
Socioeconomic Characteristics
A preliminary analysis of the 1992 MDHS data also shows,
as expected, a variation in early childhood mortality by
socioeconomic characteristics (see Table 3) . It should,
however, be noted that this level of analysis does not lead to
a statistical conclusion regarding the strength and
significance of the impact of each of the hypothesized
explanatory variables on the risk of death during the neonatal
or postneonatal period. In order to establish the
significance and importance of each factor, multivariate
statistical techniques will be used.
Table 3 : Childhood Mortality Rates (per 1000 Live
Births) by Socioeconomic and Demographic
Characteristics, Malawi; 1992
Character
istic
NNMR PNNMR IMR CMR U5MR
Sex of Child:
Male
Female
49.8
47.9
91.2
82.5
141.0
130.4
125.9
114.4
249.1
229.8
Type of Residence:
Rural
Urban
48.6
50.9
89.4
67.3
138.0
118.1
122 .9
98.9
243.9
205.4
Region of Residence:
Northern
Central
Southern
47.7
44.3
53 .1
73.0
85.9
91.2
120.7
130.2
144.3
92.3
151.0
100.1
201.9
261. 6
230.0
36
Table 3 (continued)
Character
istic
NNMR PNNMR IMR CMR U5MR
Mother's Education:
No Education
Primary, Std 1-4
Primary, Std 5-8
Secondary & over
51.5
49.7
43.1
(35.1)
91.3
85.9
80.4
(61.2)
142.8 130.8 254.9
135.6 124.9 243.6
123.6 97.1 208.7
(96.3) (34 .2) (127.3)
Mother's Age at
Birth of Child:
<20
20-29
30-39
40-49
79.1
43.5
38.3
(38.2)
100.2
81.9
85.0
(91.4)
179.3 143.9
125.5 117.2
123.3 111.2
(129.6)(92.1)
297.4
228.0
220.8
(209.7)
Birth Order of Child
1
2-3
4-6
7 or Higher
76.8
47.5
39.6
40.0
98. 3
86.5
78.9
89.7
175.1
134.0
118.5
129.6
126.8
132.5
110.8
109.7
279.7
248.4
216.1
225.1
Previous Birth
Interval:
<2 yrs
2-3 yrs
3 yrs or more
65.9
35.1
31.1
116. 0
75.5
65.0
181.9
110.6
96.1
161.4
108.5
76.6
313.9
207.1
165.3
37
Table 3 (continued)
Character
istic
NNMR PNNMR IMR CMR U5MR
Size of Child
at Birth:
Very Small
Small
Average or Large
Medical Maternity
Care:
No Antenatal/
Delivery
Either Antenatal
or Delivery
Both Antenatal
and Delivery
(188.1) (128.9) (317.0) (102.2) (386.8)
42.7 (108.5)(151.2)(107.6)(242.5)
30.2 91.1 121.4 106.2 214.7
(68.0)(127.0)(195.0)(193.3)(350.6)
42.3 111.6 154.0 (116.3) 252.4
35.7 81.7 117.4 93.9 200.3
All Malawi 48.8 86.9 135.7 120.1 239.5
Source: 1992 Malawi Demographic and Health Survey Report
( ) Based on 250-500 cases
NNMR= Neonatal Mortality Rate
PNNMR= Post Nepnatal Mortality Rate
IMR = Infant Mortality Rate
CMR= Child Mortality Rate
U5MR= Under five Mortality Rate
(a) Maternal Education
Table 3 shows that early childhood mortality rates vary
among infants born to mothers with different educational
levels. Children born to mothers with at least secondary
education face the least risk of dying at any stage of child
38
growth and those born to mothers without any education
experience the highest probability of dying within five years
of life. For example, neonatal mortality among uneducated
mothers is about one and half times as high as that for
mothers with at least secondary education. Similarly, children
born to uneducated mothers are two times as likely to die
before age 5 than those born to mothers with at least
secondary education. Although confounding factors have not
been controlled for in this preliminary analysis, the results
appear to confirm what many other researches have concluded
regarding the inverse relationship between maternal education
and childhood mortality.
(b) Type of Residence
The levels of neonatal mortality in rural and urban areas
are almost the same. However, postneonatal mortality in rural
areas is about one and a third times as high as that in urban
areas. While one out of every four children die before age 5
in rural areas, one out of every five die before that age in
urban areas. The differences in the rural-urban mortality
rates may be due different environments and circumstances the
children are brought up.
(c) Region of Residence
Although on average, Northern Region residents live
39
farther away from a hospital (median distance, 30 km) as
compared to those in the Centre (24 km) and South (18 km)
(NSO, 1994) childhood mortality at all stages of growth of the
child are lowest in the Northern Region. This suggests that
other factors apart from the proximity of a hospital have
important influences on the child's health.
4.2 Early Childhood Mortality Differentials by Selected
Demographic Characteristics
(a) Sex of Child
Table 3 further shows that the level of mortality among
male children is consistently higher than that among their
female counterparts. Neonatal mortality among male infants is
only 4 percent higher than among female infants. After the
neonatal period, male children mortality is about 8 percent
higher than that for females. While other studies have shown
that female mortality increases and becomes higher after the
postneonatal period because of preferential treatment boys get
from their parents, the situation in Malawi is different. It
would, therefore, appear that sex preference is not an obvious
issue in Malawi.
(b) Mother's Age at Birth of Child
Results of our preliminary analysis shows that childhood
mortality levels at all ages of the child are extremely high
among children born to teenage mothers (less than 20 years).
40
For instance, babies born to teenage mothers face two times
the risk of dying during neonatal period that for infants born
to mothers aged 3 0-39 years at their birth. The Malawi data
confirm the earlier findings from other studies about the
disadvantage of giving birth at very early ages. However, our
results do not verify the findings from other studies that
child bearing at later ages puts the children at an elevated
risk of dying, especially in infancy. This could be so due to
the fact that the number of cases of mothers in the age group
40-49 from which the mortality rates are estimated was small
(less than 500) and so the estimates may be biased.
(c) Birth Order of Child
The data show that childhood mortality among first born
children is much higher than for the other children and is
lowest for those children who are of birth order between 4 and
6. First births are about 92 percent more likely to die in
the neonatal period than children of birth order 4 or higher.
This is somehow consistent with the results from other studies
which have attributed this to the fact that in societies where
age at marriage is very low and where contraception use is
low, childbearing is likely to start at an early age and the
babies may be premature with low weights at birth. However,
in the case of Malawi the vulnerability of first births
diminishes considerably after the neonatal period, so that
41
during the later period the first births are only 15 percent
more likely to die than children of birth order 4 or higher.
(d) Previous Birth Intervals
Table 3 also shows that children born within 2 years of
each other have about two times the risk of dying during
neonatal period than children born more than two years after
the preceding sibling. The mortality rates are best when the
duration of the preceding birth interval is at least 3 years.
The analysis shows that birth spacing of more than 3 years
would reduce infant mortality by half and consequently under-
five mortality by the same amount (Table 3).
(e) Size of Child at Birth
Our preliminary analysis of the Malawi data shows that
children whose sizes at birth were described as 'very small'
were more than six times as likely to die during neonatal
period as those who were considered by their mothers as
'average' or 'large'. Further, about 40 percent of these
children fail to survive to their fifth birth day. However,
the enormity of this relationship may be exaggerated by the
small number of cases from which the different mortality
estimates are made.
42
(f) Medical Maternity Care
The beneficial use of antenatal and delivery services for
children's survival chances is shown in Table 3. It is
observed that mortality among children whose mothers had
received no antenatal care nor had medically supervised
delivery were about two times as vulnerable to death during
neonatal period as those whose mothers had received antenatal
care and were assisted by medical staff at the time of their
delivery.
CHAPTER V
MULTIVARIATE ANALYSIS: EMPIRICAL RESULTS
This chapter discusses the results of the multivariate
analysis, emphasizing the statistically significant
determinants on neonatal and postneonatal mortality.
Preliminary analysis of the data had indicated that
childhood mortality levels differed not only by characteristic
but also by categories of the same characteristic. However,
this level of analysis was not able to tell if such observed
differences were due merely to chance factors or if they were
indeed statistically significant. Using binary logistic
regression for neonatal mortality and Cox proportional
hazards regression for postneonatal mortality, the study
43
analyses the main effects of the hypothesized factors as well
as the interaction effects as a result of the interplay
between some of the factors on the risks of death of a child
during the neonatal period as well as the postneonatal period.
The choice of the variables included in the multivariate
analysis was based on the results of the preliminary
descriptive analysis. In either models, the unit of analysis
is the individual child born during the 5- year period
preceding the date of the survey, October, 1992.
For the logistic regression , the dependent variable is
a dichotomous dummy variable denoting whether or not the child
died during the neonatal period. In the case of the
proportional hazards model, the exposure period would consist
of the time from second month of birth up including all those
children who survived the postneonatal period, but excluding
all those children who could not possibly face the risk of
dying during the postneonatal period. These included all
those babies who died during the neonatal period, or those
babies who were aged under one month at the time of the
survey. Since virtually all the babies were breastfed for at
least one month (or if they died they were breastfed until
death) there would be no variation on neonatal mortality as a
result of the effect of breastfeeding. This variable is thus
omitted in the logistic modelling. However, duration of
44
breastfeeding is considered as a time-varying covariate since
the health status or mortality risks of the babies might
depend not only on whether or not they were breastfed, but on
the duration of breastfeeding as well. Hence the use of Cox
proportional hazards model for the analysis of postneonatal
mortality. With the exception of the breastfeeding variable,
both models evaluate the impact of the same types of variables
(see Table 6.1) . It should, however, be mentioned that due to
a large number of missing cases for weight of the child at
birth, size of the child at birth is used instead. Zero order
correlation analysis (not presented here) shows that the size
and weight of the child at its birth are very significantly
correlated (r = -0.59, a =0.0001).
5.1 Neonatal Mortality:
Table 4 displays a list of socioeconomic and demographic
variables that are analyzed for their effects on the chances
of survival during the neonatal period. It is discernible
from the results that size of the child at its birth (or birth
weight) and the age of the mother at birth of the child
emerged as the most influential determinants of child survival
during the neonatal period. Babies who were 'small' were
about 38 percent less likely to survive than those who were
of 'average' size. Further, those babies who were 'large'
enjoyed a 26 percent higher chances of surviving than those in
45
the reference category. Similarly, babies born to mothers who
were aged under 20 were about 42 percent less likely to
survive during their first month of life than those born to
mothers who were aged 20-29 years. Although babies born to
mothers aged over 30 were also less likely to survive the
neonatal period, the difference is not statistically
significant. Put differently, there is no sufficient evidence
to conclude that babies born mothers aged 30 years or more are
less likely to survive during the neonatal period than those
born to mothers aged 20-29 years.
The results also indicate that maternal education,
whether or not the mother received at least one dose of
tetanus toxoid injection to prevent the child from getting
tetanus and the interaction effects of the birth order of the
child and the age of the mother at its birth were also
moderately significant. Babies born to mothers with primary
school education were about 39 percent more likely to survive
their first month of life than those born to mothers with no
education. Likewise, children born to mothers with at least
secondary school education were about 87 percent more likely
to survive during neonatal period than those born to mothers
in the reference group. Similarly, babies born to mothers who
had received at least one dose of tetanus toxoid injection
during pregnancy were 52 percent more likely to survive their
46
first month of life than those babies born to mothers who had
not received the injection.
It is further noted that first born children born to
teenage mothers were 46 percent less likely to survive during
neonatal period than second or third born babies born to
middle aged mothers, aged 20-29 years. This would seem to
suggest that the risk of dying of first born babies is
aggravated if the mother is aged less than 20. However, it is
not conclusive if babies of birth order 4, 5 or 6 or higher
and born to mothers aged 30 years or older are less likely to
survive the neonatal period or not since the results are not
statistically significant. Furthermore, there is no
sufficient evidence to conclude that sex of the child,
paternal education, whether or not the child comes from a
household with a high income level, or whether the household
in which the babies were born had piped water for drinking or
not, or had a toilet facility or not, had significant
influence on the survival prospects of the baby. Similarly,
there is no sufficient evidence to conclude that birth order
of babies, or the period of time between the birth of the
child and the birth of a prior child, or whether or not the
mother sought antenatal care services from a health
professional, or a traditional birth attendant (TBA), or she
did not seek any services from anybody, had any effect of the
47
survival of the baby during the neonatal period. The
interaction effects of birth interval and age of mother at
birth were evaluated and were found to be not statistically
significant at 0.05 level.
Table 4: Effects of Socioeconomic and Demographic on
Neonatal and Postneonatal Mortality: Logistic
and Proportional Hazards Estimates of
Relative Risks
LOGISTIC
MODEL
PROPORTIONAL
HAZARDS MODEL
Characteristic
Parameter
Estimate
Odds
Ratio
Parameter
Estimate
Relative
Risk
Sex:
(Female)
Male
0.000
-0.021
1.000
0. 980
0.000
-0.034
1.000
0.966
Size at Birth:
Small
(Average) |
Large
| -0.977***
0.000
0.233
0.377
1.000
1.260
-0.010
0.000
-0.009
0.990
1.000
0.991
Maternal Education
(None)
Primary
Secondary/Higher
0.000
0.332**
0.626*
1.000
1.393
1.871
0.000
-0.049
-0.158*
1.000
0.952
0.854
Paternal Education
(None)
Primary
Secondary/Higher
0.000
0.001
0.123
1.000
1.001
1. 130
0.000
-0.062
-0.007
1.000
0.940
0.993
Age of Mother:
Less than 20
(20-29) |
30-39
40+
-0.878***
0.000
0.258
-0.480
0.415
1.000
1.294
0.619
0.001
0.000
-0.096*
0.301***
1.001
1.000
0.909
1.351
Household Economic
Status:
(Below Average) |
Above Average
0.000
0.008
1.000
1.008
0.000
-0.177***
1.000
0.838
48
Table 4 (Continued)
LOGISTIC
MODEL
PROPORTIONAL
HAZARDS MODEL
Characteristic
Parameter
Estimate
Odds
Ratio
Parameter
Estimate
Relative
Risk
Preceding Birth
Intervals (months)
Less than 24
(24-47)
48+
-0.034
0.000
-0.598
0.966
1.000
0.550
0.320***
0.000
-0.392***
1.378
1.000
0.676
Birth Order:
1
(2-3)
4-5
6+
-0.346
0.000
0.051
-0.465
0.707
1.000
1.053
0.628
0.395***
0.000
-0.214**
-0.186
1.484
1.000
0.807
0.831
Source of
Antenatal Care:
(None)
TBA
Health Prof.
0.000
-0.361
0.110
1.000
0.697
1. 116
0.000
0. 157
-0.072
1.000
1.170
0.931
Source of Drink
ing Water:
Piped
(Other)
0.091
0.000
1.095
1.000
0.023
0.000
1.023
1.000
Availability of
Toilet:
Yes
(No)
0.322
0.000
1.379
1.000
0.048
0.000
1.049
1.000
49
Table 4 (Continued)
LOGISTIC
MODEL
PROPORTIONAL
HAZARDS MODEL
Characteristic
Parameter
Estimate
Odds
Ratio
Parameter
Estimate
Relative
Risk
Interactions:
Birth Intervals
& Age of Mother
at Birth:
BI < 24; Age <20
(BI24 36; Age20 29)
BI36_48; Age30_39
BI48+ ; Age40+
-0.302
0.000
0.177
-0.245
0.740
1.000
1.194
0.783
1.173***
0.000
-0.024
0.165
3.231
1.000
0.976
1.180
Birth Order and
Age of Mother
at Birth:
BOl ; Age <20
(B02 3; Age20 29)
B04_5; Age30_39
B06+ ; Age40+
-0.788*
0.000
0.167
-0.716
0.455
1.000
1.181
0.489
0.085
0.000
0. 116
0.022
1.089
1.000
1.123
1.022
Breastfeeding -4.113*** 0.016
Use of Tetanus
Toxoid
No |
(Yes)
0.418*
0.000
1.519
1.000
0.150**
0.000
1. 162
1.000
N
Model Chi-Square
4631
90.3
4228
3812.3
Note: *** p < 0.01 , ** p < 0.05 , * P < 0.10
BOl = Birth Order 1; B02-3 = Birth Order 2 or 3
B04-5 = Birth Order 4 or 5; B06+ = Birth Order 6+,
BI < 24 = Birth Interval less than 24 months,
BI24_36 = Birth interval between 24 and 36 months,
BI36_48 = Birth Interval between 36 and 48 months,
BI48+ = Birth Interval of 48 months or more.
Reference categories are in parentheses
50
5.2 Postneonatal Mortality:
Table 4 also shows the proportional hazards model
estimates for breastfeeding status as well as the same
different socioeconomic and demographic factors as those
analyzed by logistic modelling in the previous section. The
results show that age of mother at the birth of the child,
birth order of the child, duration of breastfeeding and the
interaction effects of birth interval and age of mother at
birth all have statistically significant effects on
determining the hazard of death during the postneonatal
period. Maternal education continues to have moderately
significant effect on postneonatal mortality, while paternal
education hardly has any influence on the mortality of the
postneonates. Whether the mother had received at least one
dose of tetanus toxoid injection during pregnancy, like in the
case of neonatal mortality, has a significant effect on the
survival of the child in the first year of life. The
following variables do not have any significant effects on the
prospects of child survival during the postneonatal period:
sex of child, size of child at birth, paternal education,
availability of toilet facility or piped water, or whether or
not the mother sought antenatal care from a health
professional, a traditional birth attendant or from no one.
The interaction effects of birth order of the child and age of
mother at birth, which were moderately significant in the
51
logistic model for neonatal mortality, are no longer important
in the case of postneonatal mortality.
It is noted that the hazard of babies born in households
with low income levels were about 16 percent larger than those
born in households with higher incomes. In other words, the
hazard of the children born in wealthy households was about 84
percent that of those born in unwealthy ones. It is also
clear from the table that the risk of dying for babies born to
mothers who were aged at least 40 years was 35 percent higher
than the risk of those born to mothers aged 20-29 years.
Similarly, children born within 24 months of the birth of
their immediate elder sibling faced a hazard which was 38
percent larger than those who were born between 24 and 48
months after the birth of their immediate elder sibling. For
those born after more than 48 months after the birth of their
elder sibling, their hazard was about 68 percent of the risk
of those in the reference category. In the case of the babies
who were of birth order 1, their hazard was about 48 percent
larger than the second or third born children (Table 4) .
Further, babies born less than 24 months after the birth of
the prior child and born to mothers aged less than 20 faced
three times the risk of dying during postneonatal period than
those who were born between 24 and 48 months after the birth
of a prior child and to mothers aged 20-29 years.
52
CHAPTER VI
SUMMARY AND CONCLUSIONS:
The high levels of neonatal and postneonatal mortality,
in particular, and under-five mortality in general, need cause
great concern to the health policy maker in Malawi. While
neonatal mortality showed signs of decline during the 15-year
period preceding the date of the survey, this appears to have
been so at the expense of postneonatal mortality which showed
signs of increasing trends during the same period (Table 2).
It is likely that the health programmes which were undertaken
in Malawi targeted a set of variables that had an impact only
on reducing neonatal mortality and not postneonatal mortality.
The net effect was, therefore, a constant level of infant
mortality over time. This study, therefore, aimed at
examining a wide range of factors which could possibly
influence the survival chances of children during neonatal and
postneonatal periods separately. The choice of such factors
was guided mainly by literature. Due to data limitations,
however, the effects of factors such as birth weight and
household income level could not be directly evaluated. The
use of variable size of child at birth as a proxy for birth
weight is justified since the two variables are very highly
correlated. While information on birth weight was collected
in the survey, such information was missing on about 65
53
percent of the children. Hence regression estimates based on
only 35 percent of the observations may not be reliable.
Similarly, the use of the variable constructed by considering
the availability of durable goods in the households, such as
radios, cars, etc is also justified since it is likely that
household with high incomes would be able to purchase these
goods.
Among the determinants of neonatal mortality, size of the
child at birth (birth weight) and age of mother at the time of
birth of the child emerged as the most dominant. Other
factors which were also of some importance were maternal
education, and whether or not the mother had received at least
one dose of tetanus toxoid injection during pregnancy. From
these results, the policy recommendations are clear. First,
in order to further lower the already declining neonatal
mortality level, use of basic, maternal and child health
services which are provided under the umbrella of Maternal and
Child Health Programmes need to be not only universalized but
be intensified as well. The quality of such services need
also to be critically evaluated. With regard to age of mother
at birth of a child, it is clear that childbearing during
teenage is extremely risky for the survival of the child.
Since 50 percent of the females get married by age 18, efforts
need to be made to encourage childbearing among Malawian
females to later ages. This could notably be achieved by
54
encouraging women to use family planning methods. Family
planning methods themselves would have to be universally
available and accessible to all those who need them.
The results of the study also suggest significant
interaction effects of birth order of the child and age of
mother at birth of the child. In particular, first born
children to teenage mothers face a further risk of dying
during neonatal period. Although it would take a considerably
long time to get the results, education of females need to be
improved. Indirectly, by keeping the woman longer in school
preferably up to secondary school level, her exposure period
to childbearing would be reduced and this means she may have
fewer children in her life. Additionally, information,
education and communication (IEC) programmes should be
accorded top priority in Malawi.
With respect to postneonatal mortality, age of mother at
birth of her child, household economic status, length of
preceding birth intervals, breastfeeding status, and birth
order of the child are the dominant determinants of its
survival during the postneonatal period. The interaction
effects of birth intervals and age of mother at birth of the
child are also extremely important. Through the same IEC
programmes suggested to further curb down mortality among
55
neonates, the importance of breastfeeding and longer birth
intervals needs to be stressed. Longer birth intervals, of
course, can be achieved by both increased duration of
breastfeeding as well as use of family planning methods.
56
REFERENCES :
Aldrich, John H. and Forrest D. Nelson. 1984. Linear
Probability. Logit, and Probit Models. Sage
Publications. Beverly Hills.
Allison, P.D. 1984. Event History Analysis: Regression for
Longitudinal Event Data. Sage Publications, Beverly
Hills.
Al-Kabir, Ahmed. 1984. Effects of Community Factors on
Infant and Child Mortality in Rural Bangladesh. WFS
Scientific Reports No. 56, Voorburg, Netherlands:
International Statistical Institute
Bicego, G. T. and J. Ties Boerma. 1993. "Maternal Education
and Child Survival: A Comparative Study of Survey Data
from 17 Countries" in Social Science and Medicine, vol 36
No. 9: 1207-1227, Pergamon Press Ltd
Casterline, J. B; E.C. Crooksey and Abdel Fattah E. Ismail.
1989. "Household Income and Child Survival in Egypt" in
Demography. vol 26 No. 1: 15-32
DaVanzo, J. 1988. "A Household Survey of Child Mortality
Determinants in Malaysia" in Child Survival: Strategies
for Research, ed. W. Henry Mosley and Lincoln C. Chen,
Supplement to Population and Development Review,
vol 10: 307-321
DaVanzo, J. 1988. "Infant Mortality and Socioeconomic
Development: Evidence from Malaysian Household Data" in
Demography. vol 25 No. 4: 581-595, Population
Association of America
East-West Center. 1993. Changes in Fertility Patterns Can
Improve Child Survival in Southeast Asia. Asia-Pacific
Population and Policy, Honolulu, Hawaii
Farah, Abdul-Aziz and Samuel H. Preston. 1982. "Child
Mortality Differentials in Sudan" in Population and
Development Review, vol 8 No. 2: 365-383
Gbesemete, Kwame P. and Dich Jonsson. 1993. "A Comparison of
Empirical Models on Determinants of Infant Mortality: A
Cross-National Study on Africa" in Health Policy, vol
24:155-174
57
Gubhaju, B; K. Streatfield and Abdul K. Majumder. 1991.
"Socioeconomic, Demographic and Environmental
Determinants of Infant Mortality in Nepal" in Journal of
Biosocial Science, vol 23: 425-435
Hill, Althea. 1993. "Trends in Childhood Mortality" in
Demographic Change in Sub-Saharan Africa. National
Academy Press, Washington D.C.
Hill, K. and A.R. Pebley. 1989. "Child Mortality in the
Developing World" in Population and Development Review,
vol 15, No. 4: 657-687
IRD/Westinghouse. 1987. Child Survival: Risks and the Road to
Health. Demographic Data for Development Project,
Washington, D.C.
Kim, Tai-Hun. 1988. "Changing Determinants of Infant and
Child Mortality: On the Basis of the Korean Experience,
1955-1973" in Journal of Biosocial Science vol 20
No. 3 :345-355
Knodel, J. and H. Kintner. 1977. "The Impact of
Breastfeeding Patterns on the Biometric Analysis of
Infant Mortality" in Demography. vol 14, No. 4: 391-409
Meegama, S. A. 1980. Socio-Economic Determinants of Infant
and Child Mortality in Sri Lanka: An Analysis of Post-War
Experience. WFS Scientific Reports No. 8, Voorburg,
Netherlands: International Statistical Institute
Merrick, T. W. 1985. "The Effect of Piped Water on Early
Childhood Mortality in Urban Brazil, 1970 to 1976" in
Demography. vol 22, No. 1: 1-24; Population Association
of America
Mosley, H.W. and L. C. Chen (ed.). 1984. Child Survival:
Strategies for Research. Supplement to Population and
Development Review, vol 10
National Statistical Office. 1994. 1992 Malawi Demographic
and Health Survey Report. Macro International Inc,
Maryland
National Statistical Office. 1984. Malawi Population Census
1977: Analytical Report, vol II. Government Printer,
Zomba, Malawi
Palloni, Alberto and Marta Tienda. 1986. "The Effects of
Breastfeeding and Pace of Childbearing on Mortality at
Early Ages" in Demography. vol 23, No. 1: 31-51
58
Pant, Prakash Dev. 1991. "Effect of Education and Household
Characteristics on Infant and Child Mortality in Urban
Nepal" in Journal of Biosocial Science. vol 23,
No. 3 :437-443
Population Reference Bureau. 1994. World Population Data
Sheet: PRB, Washington, D.C
Retherford, R.D., M.K. Choe, S. Thapa, and B.B.Gubhaju. 1989.
"To What Extent Does Breastfeeding Explain Birth
Interval Effects on Early Childhood Mortality?" in
Demography. vol 26, No. 3 :439-450
Tekce, B. and F. C. Shorter. 1989. "Determinants of Child
Mortality: A Study of Squatter Settlements in Jordan" in
Child Survival: Strategies for Research, ed. W. Henry
Mosley and Lincoln C. Chen, Supplement to Population and
Development Review, vol 10: 257-279
The Cebu Study Team. 1993. "Underlying and Proximate
Determinants of Child Health: The Cebu Longitudinal
Health and Nutritional Study" in American Journal of
Epidemiology, vol 133, No.2: 185-201
United Nations. 1985. Socioeconomic Differentials in Child
Mortality in Developing Countries. U.N, New York
59

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Health And Well-Being By Marital Status: The Effect Of Age, Sex, Income And Social Support. Study Of 1985 Gss Data

PDF

Cross-Sectional And Longitudinal Effects Of Parental Divorce On Parent-Adult Child In-Person Contact

PDF

An analysis of nonresponse in a sample of Americans 70 years of age and older in the longitudinal study on aging 1984-1990

PDF

Determinants Of Intercounty Migration: California, 1970-1973

PDF

Differential Fertility Behavior And Values In Rural And Semi-Urban Costa Rica

PDF

The Construction And Empirical Test Of A Theory Based On Selected Variables In Small-Group Interaction

PDF

Referential Dissociation And Response To Stress

PDF

Employment Distribution, Income, And City Size: A Statistical Analysis

PDF

Transformational Processing Of Sentences Containing Adjectival Modifiers

PDF

An Empirical Examination Of The Relationship Of Vertical Occupational Mobility And Horizontal Residential Mobility

PDF

Dissonance in the population -environment movement over the politics of immigration: Shifting paradigms of discourse vis -a -vis individual rights and societal goals

PDF

Social Components Of Housing Cost In The Western Metropolis

PDF

Ecology Of Negro Communities In Los Angeles County: 1940-1959

PDF

Prediction Of Overt Behaviors In Hospitalized Psychiatric Patients

PDF

The Effects Of Social Security Programs On Fertility Levels

PDF

Interbirth interval analysis of fertility: The case of Kenya in 1993

PDF

Economic Differentiation And Social Organization Of Standard Metropolitanareas In The United States: 1950

PDF

Status Consistency Among The Clergy

PDF

The Magnitude And Rate Of Decline In Alpha-Fetoprotein Levels In Children With Unresectable Or Metastatic Hepatoblastoma, Predict Outcome

PDF

Consensus Of Role Perceptions In A Welfare Planning Council

Asset Metadata

Creator Mpando, Ladislas Ritchken Silino (author)

Core Title Determinants Of Neonatal And Postneonatal Mortality: The Case Of Malawi, 1992

Degree Master of Science

Degree Program Sociology - Applied Demography

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag OAI-PMH Harvest,sociology, demography

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Heer, David M. (committee chair), Biblarz, Timothy J. (committee member), Van Arsdol, Maurice D., Jr. (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c18-9185

Unique identifier UC11357647

Identifier 1378427.pdf (filename),usctheses-c18-9185 (legacy record id)

Legacy Identifier 1378427-0.pdf

Dmrecord 9185

Document Type Thesis

Rights Mpando, Ladislas Ritchken Silino

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA