A COMPARATIVE STUDY OF UNDER-FIVE CHILD MORTALITY IN
DIFFERENT HOUSING SETTLEMENTS IN SOWETO, SOUTH
AFRICA 2002.
By
EZEKIEL SITIENEI KUTTO
STUDENT NUMBER: 0718065Y
A RESEARCH REPORT SUBMITTED TO THE FACULTY OF
HEALTH SCIENCES, UNIVERSITY OF WITWATERSRAND,
JOHANNESBURG IN PARTIAL FULFILLMENT OF THE
REQUIREMENTS FOR THE AWARD OF MASTER OF SCIENCE
(MEDICINE) IN THE FIELD OF EPIDEMIOLOGY AND
BIOSTATISTICS.
JULY 2008
i
DECLARATION
I, Ezekiel Sitienei Kutto hereby declare that this research report is my own unaided work.
It is being submitted for the degree of Master of Science (Medicine) in the field of
Epidemiology and Biostatistics at the University of Witwatersrand, Johannesburg .It has
not been submitted entirely or partially for any degree or examination at this or any other
university.
Signature Ezekielkutto
25th day of the Month of July year 2008
ii
DEDICATION
I wish to dedicate this research report to my beloved wife Veronica Jepchumba Sitienei
and my two beloved sons Leon and Larry for their endurance and for the moral support
they gave me during the one year study period at the University of Witwatersrand in
Johannesburg, South Africa.
iii
ACKNOWLEDGEMENTS
I would like to express my deepest gratitude to Perinatal HIV Research Unit of the
University of Witwatersrand for providing me with Soweto Household Survey data of
2002 used in this study. I trust that the findings of the study will contribute more
knowledge to the scientific world in regard to under-five mortality in Soweto Townships.
I also wish to extend my sincere gratitude to the funders of Soweto household survey;
United States Agency for International Development (USAID).
I am profoundly grateful to my supervisors; Mr. Edmore Marinda and Helen Struthers
who were available all the time for supervision, encouragement and guidance. I count
myself lucky for having them as my supervisors. I also wish to thank Ronelle Van
Niekerk of Perinatal HIV Research Unit for her support regarding the Soweto household
survey dataset.
Finally but not least I wish to express my sincere heartfelt thanks to my course
coordinator Ronel Kellerman, course administrator Lindy Mataboge, all lecturers of the
school of public health and my student colleagues for their support during my study
period at University of Witwatersrand.
Last but foremost I thank the almighty God for all the blessings.
iv
TABLE OF CONTENTS
Table of contents Page
Declaration????????????????????????????....i
Dedication?????????????????????????????.ii
Acknowledgement??????????????????????????iii
Table of contents??????????????????????????..iv
List of figures???????????????????????????...vii
List of Tables???????????????????????????...viii
Abstract??????????????????????????????.ix
CHAPTER ONE????????????????????????...........1
1.1 Introduction???????????????????????????.1
1.2 Background information on the study area???????????????.3
1.3 Problem statement????????????????????????...5
1.4 Significance of the study??????????????????????.6
1.5 Aims of the study?????????????????????????.7
1.6 Research question?????????????????????????7
1.7 Objectives of the study???????????????????????7
1.8 Literature review?????????????????????????.8
CHAPTER TWO??????????????????????????.16
2.1.0 Study design??????????????????????????16
2.1.1 The design of Soweto Household Survey??????????????...16
v
2.1.2 Questionnaire design and development????????????????.17
2.1.3 Field work??????????????????????????......17
2.1.4 Data entry process???????????????????????......17
2.2 Study population and Sample size??????????????????....17
2.3 Key words????????????????????????????.18
2.4.0 Explanatory Variables??????????????????????....18
2.4.1 Outcome variables???????????????????????......19
2.5 Ethical considerations??????????????????????........19
2.6 Dissemination and utilisation of the findings?????????????........19
2.7.0 Data management?????????????????????????19
2.7.1 Data analysis???????????????????????????20
CHAPTER THREE??????????????????????????..21
3.1 Description of the study population in Soweto Townships????????........21
3.2 Household characteristics in Soweto Townships by housing settlement....??.......22
3.3 Estimated Proportion of under-five mortality in Soweto Townships????.........22
3.4 Distribution of under-five mortality by housing settlement in Soweto
Townships?????????????????????????????......24
3.5 Reported causes of under-five death in Soweto Townships?????....................25
3.6 Under-five mortality rates in the different settlements in Soweto Townships???26
3.7 Levels of differentials in under-five mortality in the different housing
Settlements in Soweto Townships??????????????????....28
3.8 Factors associated with child mortality in Soweto Townships????????....30
vi
CHAPTER FOUR???????????????????????????.33
4.0 Discussion.????????????????????????????...33
4.1 Under-five mortality differentials in Soweto Townships???????????33
4.2 Causes of under-five mortality in Soweto Townships?.??????????....35
4.3 Predictors of under-five mortality in Soweto Townships ??????????...37
4.4 Limitation of the study????????????. ???????????...39
CHAPTER FIVE ???????????????????????????..41
5.1 Conclusions????????????????????????????..41
5.2 Recommendations??????????????????????????42
REFERRENCES????????????????????????????43
APPENDICES????????????????????????????...48
vii
LIST OF FIGURES Page
Figure 1.0: Estimated percentage of under-five mortality in Soweto Townships
according to age category????????????????????? ???.24
Figure 2.0: Distribution of causes of under-five mortality in Soweto Townships?........26
Figure 3.0: Kaplan Meier under-five mortality estimate in Soweto Townships???...28
viii
LIST OF TABLES Page
Table 1.0: Description of the study population in Soweto Townships??????....24
Table 2.0: Distribution of under-five mortality by housing settlements in
Soweto Townships?????????????????........................................25
Table 3.0: Child mortality rates in the different housing settlements in Soweto
Townships??????????????????????????????.27
Table 4.0: Factors associated with under-five mortality in Soweto Townships???..32
ix
ABSTRACT
Background: The study examines the differentials in child mortality existing in the
different settlements in Soweto Townships as at May 2002. It attempts to establish the
association existing between housing settlement and under-five mortality and as well
examine how household characteristics such as building materials, source of drinking
water, sanitation facilities and source of energy is associated with under-five
mortality. Methods: The research comprises secondary data analysis of a household
survey conducted in Soweto in May 2002 by Perinatal HIV Research Unit. The aim
of the study is to describe child mortality and explore its relationship to five different
housing settlements in Soweto Townships. Kaplan Meier curves were fitted to
examine differentials in child mortality in the different housing settlements and Log
rank test was used to compare the survival curves. Logistic regression models were
fitted to establish factors that were associated with under-five mortality in Soweto
Townships as a whole. Results: A total of 2741under-five surviving children and 84
under-five deaths were studied. Informal settlements recorded the highest overall
under-five mortality rate (15.9 per 10000 child years) while private sector housing
settlement recorded the least (3.3 per 10000 child years) of all the housing
settlements. The overall under-five mortality in Soweto Townships was 10.4 per
10000 child years. After controlling for other household characteristics settlement was
significantly associated with high under-five mortality (Council settlement OR 3.3,
P=0.032, 95 CI; 1.113, 10.24, Informal settlement OR 5.10, P=0.005, 95% CI; 1.633,
15.99 and Hostel settlement OR 4.09, P=0.012, 95% CI; 1.357, 12.35). Use of
paraffin and candles were also significantly associated with high under-five mortality
x
(OR 3.4, P-value<0.001, 95% CI; 2.416, 19.15 and OR 7.2 P-value=0.014, 95% CI;
1.25, 8.88 respectively). Conclusions: Private sector housing settlements reported
lower under-five mortality rates in Soweto Townships (3.3 per 10,000 child years)
and was less associated with high mortality in comparison to the other housing
settlements (Informal, council and hostel housing settlements) in Soweto Townships.
Use of paraffin and candles for lighting in households were mainly associated with
high under-five mortality in Soweto Townships in comparison to those households
that used of electricity.
- 1 -
CHAPTER ONE
GENERAL INTRODUCTION
1.1 INTRODUCTION
Under-five mortality rate (U5MR) is one of the most important indicators for assessing
the health status of a community or a country and is a useful prerequisite for planning
health interventions for child survival [1]. This indicator is linked to internationally
recognised goals that countries strive to achieve in an attempt to fulfill general
development standards and children?s rights. Like the infant mortality rate (IMR), the
U5MR provides a key baseline indication of how a country is progressing with its plan to
realise children?s rights, in particular their rights to life, health-care services, nutrition,
education, adequate standard of living, water, social security and protection. [2]
One of the most important Millennium Development goals (MDG) developed by the
United Nations is to reduce under five mortality rate by two thirds between 1990 and
2015. Their rationale is that human development cannot be sustained anywhere unless
children are protected everywhere in the world [3].
High under-five mortality is mainly correlated with inadequate maternal and child health
(MCH) services, insufficient nutrition, poor immunisation coverage, environmental
exposure and other exogenous factors which include those related to housing conditions
[4].
- 2 -
In many low income countries of sub-Saharan Africa, 10-20% of children die before
reaching five years of age compared to high income countries, for example 0.7% in
England and Wales [5].
In South Africa, it is estimated that nearly 100000 children under 5 years of age still die
each year [6]. So to achieve the MDG of under-five mortality in South Africa by 2015,
simply means reducing under-five deaths to lower than 67,000 a year.
HIV/AIDS is the leading cause of under-five deaths accounting for 40.3% in South
Africa. Diarrhoeal disease, lower respiratory tract infections, and malnutrition, when
adjusted for HIV/AIDS co-morbidity, together account for 20.3% of the under-five
deaths [6].
Household characteristics such as building materials, source of drinking water, sanitation
facilities and source of energy reflect economic status and are known to be important
components of child survival. Together with other factors, household characteristics are
thought to have direct effects on child mortality [7].
Besides reporting causes of under-five deaths, this study aims to describe child mortality
and explore the differentials in under-five mortality in five different housing settlements
in Soweto. It specifically aims to examine the impact housing settlement has on under-
five mortality and how household characteristics such as building materials, source of
- 3 -
drinking water, sanitation facilities and source of energy are associated with under-five
mortality.
The findings will hopefully contribute towards evaluating and developing policies
regarding housing and social amenities to reduce child mortality in Soweto Townships
and other Townships across the world.
1.2 BACKGROUND INFORMATION ON THE STUDY AREA
South Western Townships (Soweto) is an urban African Townships with a population
estimated at 1.1 million people in 2002. Soweto is the most populous black urban
residential area in South Africa. It is situated 15km to the south west of Johannesburg,
Gauteng province. Soweto ranks among the poorest settlement in Johannesburg, although
individual Townships tend to have a mix of wealthier and poorer residents. It hosts
complex social networks and a range of socio-economic strata, with a variety of activities
and lifestyles. Residents of Soweto live in dwellings ranging from low income housing
such as single sex hostels and informal shack settlements to formal brick houses owned
by high income earners [8].
With its high unemployment rate, the area houses many gangsters and has been a seedbed
of criminal activity causing mayhem in many parts of Johannesburg [9]
Soweto was founded in the 1950?s when the Africans from the freehold Townships of
Alexandra, Sophia town, Martindale and Newclare were forcibly removed by the then
apartheid government and rehoused in what came to be known as greater Soweto through
- 4 -
an accelerated housing programme for Africans. It was established to house mainly black
labourers, who worked in mines and other industries in the city, away from the city
centre. Soweto then became the major reception area for Africans who moved to live near
the city of Johannesburg [10].
In the year 2000, it was estimated that almost two-fifths of the households were found in
backyard areas with an estimated population of over 200000 people accommodated.
Majority of the backyard dwellings were located behind the stands of the council houses.
Council houses occupy 74% of all the stands and make up to 84% of all the formal
structures [11, 20].
Soweto informal settlements are characterised by poor housing, overcrowding, and little
infrastructure. Dwellings in these settlements are made of corrugated iron sheets and
plastic. The informal settlements are unregulated clusters of shacks that sprout on vacant
land, have been illegally settled and have minimal service provision, without electricity
or running water. It is estimated that about 65,000 Sowetans live in informal settlements
[10, 13]. Hostels are large buildings with units that were built on the outskirts of various
Townships to house migrant workers who have historically lived on the fringes of
Soweto. These hostels initially were for single workers and segregated by sex. Many
thousands of people live in the hostels in very poor conditions [8, 12, and 13].
Another important housing domain in Soweto is the private sector developed houses
which has grown significantly and it is estimated that about 90000 Sowetans now live in
- 5 -
neighborhoods where almost all the homes have been built by private developers and are
characterised by good services and served with electricity [14].
1.3 PROBLEM STATEMENT
Nearly 100 countries are falling short of the goal to reduce child mortality adopted by
world leaders at the United Nations? Millennium Summit in 2000 [14]. One in 12 children
worldwide do not live to age five, despite the availability of proven, low-cost
interventions on reducing child deaths [14]. At the end of the last decade, two thirds of
Africans were living in absolute poverty. More than half still lacked safe water and 70%
did not have proper sanitation. Infant mortality in Africa was 55% higher than the rest of
the world's low income, developing countries [16].
In South Africa, a nation of 45 million people with the most sophisticated infrastructure
on the continent, mortality rates in 2002 were at 59.6 deaths per 1000 live births. This
occurred even after introduction of free health care and improved nutritional programmes
and was largely attributed to pediatric HIV [18]. Antiretroviral therapy was not readily
available to poor populations in Soweto during this study period [17].
It is argued that a safely built environment, including adequate housing conditions, is one
of the most elemental human needs. Nonetheless, around one billion (one-sixth) of the
world?s population currently live in slums and are squatters. Poor settlements often lack
basic health service consequently resulting in high infant and under-five mortality [18].
In South Africa it is estimated that more than eight million people lived in informal
settlements in 2001, and that Johannesburg, which is the biggest urban centre in southern
- 6 -
Africa, had an estimated influx of 20000 new households per month, 90% of them into
Soweto Townships [19]. It is estimated that 35% of households in Soweto live in
settlements characterised by poor sanitation and lack of electricity [19].
Given all these grievances ranging from poor sanitation and poor housing, infant
mortality and under-five mortality is undoubtedly high.
1.4 . SIGNIFICANCE OF THE STUDY
Although a number of studies have been undertaken in Soweto Townships, there are no
well documented studies on the under-five mortality differential in the different housing
settlements within Soweto Townships [20, 22]. Many of these studies focused on other
factors not specifically under-five mortality; for example Ines Ackerl Kristensen in 2004
conducted a study on acute respiratory infections in children less than one year of age in
Soweto, South Africa. She found out that crowding and socio-economic factors such as
the father's education are important determinants for acute respiratory infections [20].
In addition it is still not clear, what other important factors are associated with child
mortality within Soweto Townships and to what extent variations in the housing type and
household characteristic contribute to under-five child death.
Global level data have been published on housing characteristic variation and child
mortality, but information at country level (South Africa) is scarce. There seem to be
lack of studies on the relationship between housing settlement/ household characteristics
and under-five mortality in South Africa [22].
- 7 -
This study intends to utilise child mortality data of various housing settlements in Soweto
collected during Soweto household survey by the Perinatal HIV Research Unit in 2002.
This study therefore intends to determine child mortality rates estimates in the five
different housing settlements in 2002 in order to understand the contribution of housing
settlement to child mortality in Soweto Townships.
1.5 AIMS OF THE STUDY
The aim of the study is to describe child mortality and explore its relationship to five
different housing settlements in urban Soweto Townships in 2002.
1.6 RESEARCH QUESTION
Is housing settlement associated with under-five mortality?
1.7 OBJECTIVES OF THE STUDY
1) To estimate and compare <1yr, 1-5yrs and overall (0-5yrs) child mortality rates in
the different housing settlements in Soweto Townships.
2) To identify and examine housing settlement, household characteristics and other
selected factors associated with child mortality in Soweto Townships.
3) To report causes of under-five death in the different housing settlements in
Soweto Townships.
- 8 -
1.8 LITERATURE REVIEW.
Inadequate and insecure housing is a huge crisis facing the world today. The number of
people living in inadequate housing has increased recently according to the United
Nations Centre for Human Settlements (Habitat). It is estimate that 20 % of the world's
population live in substandard housing with lack of food, little access to clean water,
forced eviction, gender discrimination, poor health, unemployment and low income or no
income. Children are particularly vulnerable to the impacts that these issues have and
child mortality is undoubtedly high in these areas [21].
The situation is worse in sub-Saharan Africa, where 60% of urban housing units are
temporary structures, and about half do not conform to building regulations [22].
Adequate housing is one of the most basic human needs and human right enshrined in
international law. Housing is not just about having a roof on one?s head. It is inextricably
linked to safety and security, access to services, resources and economic opportunities
[23].
Housing type and household characteristic factors are known to be associated with child
mortality in urban and rural areas of many developing countries. Studies conducted in
Malawi demonstrated that children of mothers who lived in households with no toilet
facility or sourced drinking water from a well had a higher risk of dying compared to
children who lived in households with flush toilet and piped water [24].
- 9 -
Another study undertaken in Bangladesh showed that housing conditions and access to
safe drinking water and hygienic toilet facilities are the most critical determinants of
child survival in urban areas [25].
Universally, there is vast literature that has focused on the determinants of under-five and
infant mortality. Most of the studies have shown significant association between under-
five/ infant mortality and household characteristics. For example a study conducted by
Moser et al in 2004 using data from Demographic and Health Survey data from different
countries to examine under-five mortality by household assets and household
characteristics- such as roofing, floor materials, source of drinking water, toilet facilities
and availability of electricity. The study showed increasing levels of under-five mortality
in countries with lower ratios of household assets and inadequate household
characteristics [27]. Furthermore a study done on Global Burden of Disease using
disability-adjusted life-year (DALY) to compare death and disability from various
disorders in developing and developed countries in 1990 demonstrated poor water supply
and sanitation as among the greatest predictors of DALY and attributable to 6.8% of the
world wide DALY [27].
Poor water supply and poor sanitation are features attributed to poor housing areas (slums
and informal settlements) as opposed to developed modern houses often situated in places
where they access safe water and proper sanitation systems. In a study conducted in
Nairobi among slum dwellers, over 50% of the respondents indicated water and
sanitation were the most pressing need [28]
- 10 -
Fotso et al in 2007 published a report on progress towards the child mortality millennium
development goal in urban sub-Saharan Africa. The report focused on overcrowded
slums and shantytowns characterised by poor environmental and sanitation conditions,
poor access to basic amenities and social and health services, and poor livelihood
opportunities that worsen the residents' susceptibility to various health problems. In their
findings (Fotso et al) they established an inter-relation between access to safe water for
drinking and decline in child mortality [29].
The importance of access to safe drinking water on child health, especially in urban areas,
has been documented in many studies since diarrhoea is a major cause of death among
under-five children in sub-Saharan Africa [29, 30]. Migration to urban areas has been the
main fuelling factor of population growth in cities, straining existing water infrastructure
and as a result underprivileged urban populations often pay exorbitant prices for unclean
water, while services to wealthier groups are heavily subsidised [30].
Child mortality differentials according to water supply and sanitation in many urban areas
of developing countries suggest that lack of access to piped water and toilet facilities
reduce significantly the survival chances of under-five children. A study conducted in
Eritrea using data collected by the Demographic and Health Survey (DHS) project in
1995 showed that the effect of household environment (water supply and toilet facility)
remains substantially significant during the post-neonatal and child periods, even after
adjusting for other socioeconomic variables. The study attempted to deal with the
- 11 -
question of whether access to piped water and flush toilet is associated with under-five
mortality in urban areas of Eritrea [31].
Housing settlements are important determinants of health in urban settings as pointed out
by Danielle et al study on Social Determinants Health Urban Populations in the United
States of America (USA). The study recognised that industrial activity can have
significant impact on cities with respect to pollution, and less expensive housing is often
found in areas with less desirable physical environments. More often the underprivileged
and lower income earners live in such environments. The study further concluded that the
place of residence is situated within a particular social setting and that can have
substantial impacts on health in terms of exposure [32].
A substantial body of literature demonstrates that poor housing can contribute to
infectious disease transmission, injuries, asthma symptoms, poisoning and mental health
problems both directly and indirectly. Susan et al reviewed 72 articles selected from 12
electronic databases of US interventions from 1990 to 2001 to evaluate the success of
public health interventions related to housing. In the conclusion of her study she
demonstrated that it is possible to design and carry out interventions that can lead to
improved health by making changes in housing-related conditions [33]. Another study
conducted with an aim of identifying socio-economic factors associated with mortality
among cities in Japan demonstrated a positive correlation between mortality and old
housing(r>0.2) [34].
- 12 -
A similar study conducted among pre-school children in urban Trivandrum city, capital of
Kerala showed high incidence of illness and almost six times greater risk among less than
three years old children living in environmentally deprived areas. Morbidity information
on each child was gathered for one year through weekly visits and recording of symptoms
was done by trained investigators. The study went further to conclude that morbidity
burden in Kerala, for under- three year old children is very high and is directly related to
the quality of the housing environment [35].
The health risks faced by children can be traced back to their homes and schools because
children spend most of their time in these places. The home environment, in particular,
represents an important source of fetal and early childhood exposures to biologic,
chemical, and physical agents.
The knowledge on the relationship between housing and health inequality, particularly
within urban inner-city neighborhoods, has existed for a number of decades. In 1938, the
American Public Health Association (APHA) identified knowledge gaps with respect to
housing and health, in order to understand and evaluate better the relative effects on
humans of the various problems that may exist in housing and its environment.
Measuring the direct impact of housing quality on health is hitherto a challenge. In a
recent study on the impact of housing on health, investigators estimated that indices of
urban residential quality explained up to 25% of the variability in health status in Japan
while housing quality remains an important component of health disparities in America
and round the world [36].
- 13 -
Ignorance and difficult living conditions in the urban slums of developing world are
likely to result in low health care use, hygiene awareness and lack of understanding of the
origin of sickness leading to high morbidity and mortality. Children living under such
conditions are at especially high risk of diseases such as diphtheria, pertussis, tetanus,
measles, poliomyelitis, tuberculosis, injuries, diarrhoeal diseases etcetera.
A cross sectional survey study of 1500 households conducted in slum population of
Dhaka city in Bangladesh reported death rates in households per 1000 children (0?107
months) within the last year from the interview to be 20.5 for boys and 27.0 for girls. The
study noted that even with high vaccination coverage mortality and morbidity among
young children remained alarmingly high indicating socio-economic, environmental and
housing conditions as major drivers of mortality and morbidity [37].
Measles is known to be one of the leading killers amongst under-five children in poor and
the developing countries. Measles is a major cause of child death in refugee camps and in
internally displaced populations, where living conditions are extremely at risk of the
development and spread of the disease. The fatality ratios in children in complex
emergencies have been as high as 20%?30% [38]. Measles disease is known by medical
profession as a mild disease except in populations living in extreme unfavorable
conditions like those found in the developing world and particularly in the slums and
complex emergency settings like the refugee camps [39].
- 14 -
Research conducted among Palestinian refugees in Beirut Lebanon showed an association
between the presence of illness among household members and housing conditions. The
association between housing conditions and the presence of illness among household
members showed a significant positive gradient. Households with five to seven problems
in housing were one and a half times more likely to be ill, and those with eight to fifteen
problems(defined below) were twice more likely to be ill compared to households with
zero to four problems [40]. Problems related to housing were classified as housing
infrastructure services which included drinking water, electrical power, sewage and
garbage disposal, as well as floods due to rainwater and housing conditions; the housing
conditions index was based on items such as household infestation, adequate lighting,
ventilation, heating, the presence of humidity and cracks in walls and ceiling[40].
Another cross-sectional study of 403 families conducted in a squatter settlement of
Karachi over a two week period showed that factors significantly associated with
respiratory infections among under-five included poor housing [41].
Many studies have shown associations between overcrowding at household level and
mortality. Studies conducted in Stockholm, Sweden, to investigate overcrowding and the
risk of measles death indicated a negative association between the overall risk of death
and large household size. The findings of the study concluded that the crowding may
have statistically independent effect to the risk of death from measles [42]. Another study
conducted in Rio de Janeiro, Brazil, through brass methods of indirect estimation showed
- 15 -
residence in a shantytown (favela) as one of the major determinants of mortality among
vulnerable populations in Brazil [43].
- 16 -
CHAPTER TWO
METHODOLOGY
2.1.0 STUDY DESIGN
The study is a cross sectional analytical study from a household survey conducted in
Soweto in May 2002 by the Perinatal HIV Research Unit (University of the
Witwatersrand). The main aim of the primary study was to determine the effects of adult
morbidity and mortality on household welfare and children?s well being. Secondary data
from this survey has been used to investigate factors associated with under-five mortality
in these Townships.
2.1.1 THE DESIGN OF SOWETO HOUSEHOLD SURVEY OF 2002.
The Soweto household survey of 2002 was a cross-sectional household survey that was
stratified and two stage cluster sampling used. A total of 4912 households were surveyed
and detailed information on 22724 individuals was obtained.
The survey process was carried out in two phases. In the first phase 4501 households with
children under the age of 16 years were surveyed, while the second phase comprised 411
households with no children under the age of 16 years. The respondents of the study were
caregivers in the household or head of the household.
Stratification was based on the type of housing; Council houses (these were houses
formally owned by the council, leased and currently owned by the lesser), private sector
houses (privately owned housing estates developed by the formal private construction),
- 17 -
backyard dwellings (rooms built in the backyard of council houses), informal settlements
houses (temporary structures with little infrastructure otherwise called shacks).
2.1.2 QUESTIONNAIRE DESIGN AND DEVELOPMENT.
The instrument used was a modified questionnaire previously used in the Kagera survey
in Tanzania and developed by the World Bank.
2.1.3 FIELD WORK
The data collection process was carried in two phases. The first survey was conducted in
May and June 2002 and the second survey was conducted in October 2002 with the use
of the same instrument.
2.1.4 DATA ENTRY PROCESS
Data was entered in Microsoft Access, relational database software. Quality control
measures were implemented starting from the questionnaire, in the field and during data
entry process.
2.2 STUDY POPULATION AND SAMPLE SIZE
The study population consisted of all children who were five-years or younger between
May 1998 and May 2002.Two thousand eight hundred and twenty five children (2825)
who were 5 years or younger (between 1998 and 2002) from 4501 households were
included into this study.
- 18 -
2.3 KEY WORDS
square4 Housing settlement - refers to the type of house structure and area where people
live.
square4 Under-five mortality Rate (U5MR) - is the probability (expressed as a rate per
1000 live births) of a child born in a specified year dying before reaching the age
of five.
square4 Infant Mortality Rate (IMR) - refers to the number of children dying under one
year of age divided by the number of live births that year. The infant mortality
rate is also called the infant death rate.
2.4.0 EXPLANATORY VARIABLES
The study utilised available variables in Soweto household survey dataset. The
explanatory variables included Housing settlement and household characteristics.
Housing settlement stratified into five strata namely: Hostel settlements, informal
settlements, backyard dwellings, council houses settlement and privately owned houses.
Household characteristic included type of housing building materials, source of drinking
water, toilet facilities, number of living rooms and source of energy for cooking.
2.4.1 OUTCOME VARIABLES
The outcome variable used in the study is child death or mortality. The under-five
mortality rates were estimated within the different housing settlements to ascertain which
settlement had the highest impact on child mortality and compared with the children who
- 19 -
survived during the same period. In addition the infant mortality rate (deaths within the
first year of life), one to five year (1-5yr) child mortality rates and overall (0-5) child
mortality rates were estimated and reported in the different housing settlements.
2.5 ETHICAL CONSIDERATIONS
Permission was obtained from the Perinatal HIV Research Unit (PHRU) of the University
of the Witwatersrand to use the data of the Soweto Household cross-sectional survey of
2002. For the purpose of this research ethical clearance was obtained from the University
of the Witwatersrand committee for Research on Human subjects. Data obtained was
maintained in confidence and used specifically for purposes of this research.
2.6 DISSEMINATION AND UTILISATION OF THE FINDINGS.
The study upon completion will be revised, written up and published. It is expected that
at least an article out of this project will be published in a Southern Africa peer review
journal of Epidemiology.
The findings of the study are also anticipated to be presented to various policy makers on
sectors such as housing, water and sanitation, construction industry, energy etc so as to
guide them in policy decision making process.
2.7.0 DATA MANAGEMENT
A number of variables relevant to the researcher for analysis purposes were selected and
variables that were not useful to the researcher were dropped. Microsoft Access was used
to manage data. Observations that were not meaningful for analysis such as those that
- 20 -
represented refusal, not applicable and don?t know were set to missing. The study at the
end utilized information of a total of 2741 under-five children who were alive and 84
children who died in families living in Soweto Townships as at may 2002.
2.7.1 DATA ANALYSIS
Data analysis was carried out in Intercool Stata (Version 9). The first part of the results
section gives a description of study participants in terms of distribution by housing
settlements, sex and age distribution for under-five children reported to be alive in
Soweto Townships as at May 2002 and under-five children who died in Soweto
Townships between 1998 and 2002. Person time analysis was used to determine child
mortality rates in each of the housing settlement. Log Rank test was used to compare
survival curves between the different housing settlements.
Kaplan Meier Survival curves were used to describe the survival pattern of under-five
children between 1998 and 2002. To investigate the association between child mortality
and housing settlement, univariate and multivariate logistic models were used. Interaction
between housing settlement and household characteristics was investigated as well.
- 21 -
CHAPTER THREE
RESULTS OF THE FINDINGS
3.1 DESCRIPTION OF THE STUDY POPULATION IN SOWETO TOWNSHIPS
The largest age category of the children in this study (Table: 1.0) were aged between
three and five years 1074 (38%) while the least category were aged between one and two
years 524(18.6%). The mean age was 24.9 months (std 15.44).The distribution in terms
of age categories in the various settlements was statistically significantly different (P-
value=0.006, chi2= 27.81). There were about as many boys as girls among children who
were alive at the time of data collection in 2002.There were 1,323 (48.3%) males and
1,418 (51.7%) females. However sex was not recorded for those children who were
reported to have died before their 5th birthday during the survey. A larger number of
children, 64 (22.2%) lived in the Backyard housing settlement and the least 15.7% (443)
lived in private sector housing settlement (Table 1.0). Distribution of these children in
the settlements was statistically significantly different (P-value<0.05, chi2=9.34).
Table 1.0: Description of the study population in Soweto Townships
Variable category Frequency (%)
Housing settlement
Informal 543(19.2%)
Council 618(21.9%)
Backyard 624(22.2%)
Hostels 597(21.1%)
Private sector 443(15.7%)
Total 2825(100%)
Age(Months)
0-12 693(24.5%)
12-24 524(18.6%)
24-36 534(18.9%)
36-60 1074(38.0%)
Total 2825(100%)
Sex(Children alive)
Males 1324(48.3%)
Females 1417(51.7%)
Total 2741(100%)
- 22 -
3.2 HOUSEHOLD CHARACTERICTICS BY HOUSING SETTLEMENT IN SOWETO
TOWNSHIPS.
Under-five mortality have been greatly associated with household characteristics- such as
roofing, floor materials(building materials), source of drinking water, toilet facilities and
availability of electricity [27] .Further studies have shown that sanitation conditions, poor
access to basic amenities and social and health services opportunities worsen the human
susceptibility to various health problems [29]. It is therefore imperatively important to
look at distribution of these characteristics in Soweto Townships.
All the houses in private sector housing settlement were built of brick (100 %; n=443)
while backyard settlements had the least number of houses built of bricks (6.3%; n=122).
Council, Informal and hostel housing settlements had 90.8 % (561), 55% (298) and
87.6% (522) respectively built of bricks. At the same time backyard settlements had the
highest number of houses made of corrugated iron sheets (58.3%; n=493). Informal
settlements had the highest number of housing made of temporary structures (44.8%;
n=231). There was statistically significant differences in the type of building material
used in the settlements (P-value <0.0001)
All the houses (100 %; n=443) in private sector housing settlement had flush toilets.
Over 96% of the council and hostels houses housing settlements also had flush toilets.
Less than (36.8%) of the houses in the Backyard and informal settlement had flush
toilets. Majority of the toilet facilities available in the backyard settlements were the
traditional pit latrine type (84.9%). There were statistically significant differences (p-
value < 0.0001) in the distribution of toilet facilities types among the 5 settlements.
- 23 -
Most of the houses in the private sector had piped water inside the house 423(95.5%) and
the remaining 20(4.5%) had water piped into the yard. Only 47(7.9%) of the backyard
dwellers had piped water in their houses. The majority of dwellers in backyard
settlements 328(79.49%) had their sources of water from public taps. Sixty seven percent
(365) of the informal settlement dwellers had water piped in to the yard. There was
significant difference in the distribution of type of water source in the different
settlements (P-value <0.0001)
With regard to sources of energy, all housing settlements except backyard had over 95%
of the sources of energy either as electricity or solar. Majority of the Backyard dwellers
253(40.7%) used paraffin as source of energy for lighting. Informal settlements had the
highest number of single room houses 374(40.5%) while the private housing sector had
the lowest 4(0.4%). Council, Hostel and Backyard settlements had 8.6%, 21.2% and
29.3% of houses with single rooms.
3.3 ESTIMATED PROPORTION OF UNDER-FIVE MORTALITY IN SOWETO
TOWNSHIPS
The study comprised a total of 2825 children reportedly born in Soweto Townships
between January1998 and May 2002 when the survey was undertaken. Of these 2.97%
(84) were reported to have died before their 5th birthday (Figure 1). The majority of
deaths among these children (77%) happened in the first year of life. The mean age at
death was about 8 months.
- 24 -
Figure 1: Estimated percentage of under-five mortality according to age category
in Soweto Townships from a cross sectional survey in Soweto 2002.
Estimate percentage mortality according to age
category
78%
8%
7% 7%
0-1yrs
1 to <2yrs
>2 to<3yrs
>3yrs
3.4 DISTRIBUTION OF THE UNDER-FIVE MORTALITY BY HOUSING
SETTLEMENT.
From table 3.0 below, Under-five deaths appear to be statistically different among the
housing settlements in Soweto Townships (P-Value<0.05; chi2 =9.43). Most of the
under-five deaths 26.2 %( 22) were reported among hostel dwellers and the least under-
five deaths were reported in privately owned brick houses 4.8% (4).
Similarly, the majority of infant deaths 27.7% (24) were reported also among hostel
dwellers and lowest 3.1% (2) were reported in privately owned houses (Table 2.0)
- 25 -
Table2.0: Distribution of child mortality by housing settlement in Soweto Townships
Housing settlement Infant Deaths 1-5yr Deaths U5Deaths
Hostels 18 (27.7%) 4 (21.1%) 22 (26.2%)
Council house 14 (21.5%) 3 (15.8%) 17 (20.4%)
Backyard 15 (23.1%) 5(26.3%) 20 (23.8%)
Informal settlements 16 (24.6%) 5 (26.3%) 21 (25.0%)
Private sector 2 (3.1%) 2 (10.5%) 4 (4.8%)
Total 65 (100%) 19 (100%) 84 (100%)
3.5. CAUSES OF UNDER-FIVE DEATHS IN SOWETO TOWNSHIPS
Severe diarrhoea was the most reported cause of under-five deaths and was responsible
for 9.1% (6) of all the under-five deaths reported in Soweto Townships (Figure 2.0).
Other causes of death reported included; Pneumonia 7.6% (5), Meningitis 7.6% (5),
Injuries 6.1% (4), Chronic cough 3.0% (2). Aids and Heart disease were each reported to
be responsible for 1.5% (1) deaths each respectively. However the majority of the under-
five causes of deaths were reported as others 50.5% (47).Severe diarrhoea 5(25%) was
the leading reported cause of under-five death among the children who lived in hostel
settlement, while meningitis 1(5%), heart disease 1(5%) and pneumonia 1(5%) were
reported least responsible for the under-five deaths in this settlement. In council housing
settlement pneumonia and injuries were reported as responsible for almost a third of the
under-five deaths. The causes of the other three quarters of under-five deaths were
recorded as others. In the Backyard housing settlement meningitis 3(15%) was the main
reported cause of under-five mortality. Pneumonia 1(5%) and Aids 1(5%) were also
among the recorded causes of under-five deaths in this settlement. Chronic cough
- 26 -
2(15.38%) was the leading reported cause of death among under-five in informal
settlements while pneumonia 1(33.3%) was reported to be responsible for a third of all
the under-five deaths in private sector housing settlement.
Figure 2.0: Distribution of cause of under-five deaths in Soweto Townships
Distribution of causes of Under-five deaths in
Soweto townships
0
10
20
30
40
50
60
70
He
ar
t D
ise
as
e
Inju
ry
Ch
ro
nic
co
ug
h
Se
ve
re
dia
rrh
oe
a
Me
nin
giti
s
Pn
eu
m
on
ia
Aid
s
Ot
he
rs
Causes of deaths
de
at
hs
(%
)
Series1
3.6 MORTALITY RATES IN THE DIFFERENT HOUSING SETTLEMENTS IN
SOWETO TOWNSHIPS
Table 3.0 below shows child mortality rates in the different housing settlements in
Soweto Townships. Informal settlements had the highest level of infant mortality rate
(IMR) and under-five mortality rate (U5MR) while private sector housing settlement had
the least mortality rate of all the housing settlements. The mortality rates reported in the
above housing settlements were slightly higher for the informal and lower for the private
sector housing settlements compared to the child mortality rates reported for Soweto
Townships as a whole. Death rates among children aged 1-5years were higher in the
- 27 -
hostel housing settlements (5.2 per 10,000 child years) compared to all other housing
settlements.
Table 3.0: Child mortality Rates in the different housing settlements in Soweto
Townships
Mortality Rates Housing settlement
IMR(0000) 1-5yrMR (0000) U5MR(0000)
Hostel 2.26
CI (1.31, 3.90)
5.20
CI (2.0, 13.9)
8.90
CI (5.12, 15.83)
Council Houses
1.86
CI (1.05, 3.27)
3.20
CI (1.0, 9.8)
9.40
CI (5.7, 15.6)
Backyard 2.00
CI (1.16, 3.45)
3.40
CI (1.1, 10.4)
9.90
CI (6.3, 16.9)
Informal settlements
3.12
CI (1.94, 5.01)
4.20
CI (1.4, 13.1)
15.90
CI (10.3, 24.7)
Private sector 0.42
CI (0.11, 1.68)
2.80
CI (0.70, 11)
3.30
CI (1.3, 8.9)
Soweto
Townships(overall)
1.97
CI (1.52, 2.56)
3.70
CI (2.2, 6.2)
10.40
CI (8.2, 13.1)
3.7 LEVELS OF DIFERENTIALS IN CHILD MORTALITY IN THE
DIFFERENT HOUSING SETTLEMENTS IN SOWETO TOWNSHIPS.
There were statistical differences in mortality between some housing settlements in
Soweto Townships (Figure 3.0)
- 28 -
Figure 3.0: Kaplan Meier under-five mortality estimate in Soweto Townships
Hostel and Private sector housing settlements showed significant differences in under-
five mortality. The probabilities of survival in hostel settlement for 0-5years, 1-5years
and 0-12months age categories were 97.3%, 97.4% and 98.0% respectively while in the
private sector settlement the probabilities of survival for these age categories were 98.9%,
98.8% and 99.5% respectively. There were significant differences in mortality rates in the
above mentioned groups in these settlements (P-values 0.013, 0.047 and 0.037
respectively). However there were no significant differences in under-five mortality rates
between hostel ((8.9 per 10000child years and the other three housing settlements council
houses, backyard and informal settlements ?which reported under-five mortality rates of
9.4/10000, 9.9/10000 and 15.9/10000 child years respectively) P-values >0.05).
0.95
0.99
0 10 20 30 40 50
Age in Months
Private sector settlement Council settlement
Backyard settlement Informal settlement
Hostel settlement
Survival of under-five Children in Soweto
Townships
P
rob
ability
of
S
u
rvival
0.97
- 29 -
Similarly, informal and Private sector housing settlements showed significant differences
in overall under-five (0-5yr) mortality (P-value=0.003). Interestingly, 1-5yr mortality
(reported as 4.2/10000 and 2.8/10000child years respectively) was not significantly
different (P-value=0.634 respectively). There were no significant differences in under-
five mortality observed in informal settlement in relation to the backyard and council
housing settlements which reported mortality of 9.9/10000 and 9.4/10000 child years
respectively (P-value= 0.865).
Significant differences in overall (0-5) under-five mortality and infant mortality was
observed between Backyard housing settlement which recorded mortality of 9.9/10000
child years and private sector house settlement which recorded mortality of 3.3/10000
child years (P-values=0.042 and 0.027 respectively). However there was no significant
difference among the deaths of those aged between 1-5yrs (P-value=0.877) in these two
settlements. Backyard and informal housing settlements which recorded mortality of
9.9/10000 and 15.9/10000 respectively, showed no significant differences in under-five
mortality (P-value=0.122). The pattern of child mortality in council housing settlement
did not seem to differ with those shown in all the other four housing settlements (Hostel,
Backyard, informal and private sector housing settlements) which recorded under-five
mortality of 8.9/10000, 9.9/10000 and 3.3/10000 child years respectively(P-value=
0.059)
- 30 -
3.8 FACTORS ASSOCIATED WITH CHILD MORTALITY
To examine individual effects of various explanatory variables on under-five mortality in
Soweto Townships, logistic regression analyses was performed separately for each of the
six variables (Univariate regression model) and then all the six variables were fitted in
one model (Multivariate regression model). A number of variables showed significant
association with under-five mortality. These variables include: Housing settlements and
source of energy for lighting (Table: 4.0).
All the categories of housing settlements in the univariate model showed significant
association of under-five mortality compared to private sector housing, while in
multivariate models council housing settlement became non-significantly associated to
under-five mortality (P-value=0.176, 95% CI 0.661, 9.58).
Children who lived in families where candles were used for lighting were significantly
associated with under-five death compared to under-five children who lived in
households where either electricity, gas or solar was used as a form of energy in both
univariate and multivariate models. The odds of under-five deaths was 2.6 times more in
univariate model and 7.2 times in multivariate model in households where candles were
used compared to those household where either electricity, solar or gas was used for
lighting. Households that used paraffin for cooking did not show any evidence of
association with under-five mortality in univariate model (OR 1.78, 95% CI 0.98-3.23, p-
value 0.058) while in multivariate model it was significantly associated with under-five
mortality (P-value 0.014, OR 3.43, CI 1.25, 8.88)
- 31 -
Other household characteristics however did not show any evidence of association with
under-five mortality in both univariate and multivariate models. These include variables
such as source of drinking water, Number of household rooms, Building materials and
type of toilet facilities (See table 4.0 below).
- 32 -
Table 4.0: Factors associated with under-five mortality in Soweto Townships
VARIABLE UNIVARIATE
OR P-value(95% CI)
Housing settlement
Private sector 1 - -
Council ? 3.12 0.043*(1.04, 9.29)
Backyard ? 3.63 0.019*(1.23, 10.71)
Informal sector? 4.41 0.007*(1.50, 12.96)
Hostel ? 4.21 0.009*(1.44, 12.27)
MULTIVARIATE
OR P-value(95% CI)
1 - -
3.37 0.032*(1.113, 10.24)
2.52 0.176 (0.661, 9.58)
5.10 0.005* (1.633, 15.99)
4.09 0.012*(1.357,12.35)
Energy source
Electricity 1 - -
Paraffin 1.78 0.058(0 .981, 3.23)
Candles 2.67 0.004*(1.372, 5.18)
1 - -
3.43 0.014* (1.25, 8.88)
7.02 0.001*(2.416, 19.15)
Building Materials
Brick 1 - -
Temporary/Mud 1.21 0.856(0.161, 9.02)
Corrugated iron 1.11 0.654(0.698, 1.77)
Water source
Piped into dwelling 1 - -
Piped into yard 1.40 0.597(0.701, 1.85)
Public Tap 1.49 0.196(0.814, 2.73)
No of Rooms
Single Room 1 - -
No. of Rooms(>1<=5) 0.720 0.150(0.461, 1.126)
No. of Rooms(>5<8) 0.359 0.092(0.111, 1.180)
Toilet facility type
Flush to sewage system 1 - -
Pit latrine 1.053 0.872(0.564, 1.965)
No facilities/Bush 1.613 0.515(0.382, 6.801)
Note:
*Shows significant p-values.
All the six variables were fitted in both univariate and multivariate models.
However, only significant values were reported in multivariate model.
CHAPTER FOUR
DISCUSSION
The study has examined the differentials existing in under-five mortality in different
housing settlements in Soweto Townships, South Africa during the period between 1998
and 2002. The study identified household characteristics that are associated with child
mortality. It has also determined whether the type of housing settlement impacted on
under-five mortality.
Severe diarrhoea was the most reported cause of under-five death in Soweto Townships
and was responsible for 9.09% of all the deaths in Soweto Townships. It was responsible
for a total 6 under-five deaths in informal, backyard and hostel settlements on equal
ratios. This may be attributed to poor socio-amenities served to these settlements as
opposed to private developed settlement which reported no death as result of diarhoea in
Soweto Townships.
4.1 UNDER-FIVE MORTALITY RATES DIFFERENTIALS
The overall mortality rate in Soweto Townships was 10.4 per 10,000 child years with
private sector housing settlement showing the least under-five mortality rate of 3.3 per
10,000 child years. Although overall mortality rate (0-5 years) was low compared to
mortality figures reported in South Africa in 2001 for the same age group of 71 deaths
per 1000 live births[18], it is likely that under-five deaths in Soweto Townships was
seriously under reported because the household characteristics and housing type in some
of the settlements in Soweto Townships have been found to be associated to under-five
- 34 -
mortality and definitely child mortality figures were undoubtedly expected to be higher
than those reported above .
The findings of the study showed high under-five mortality rates in informal (15.9 per
10,000 child years) and Backyard settlements (9.9 per 10,000 child years) compared to
the private sector housing settlement (3.3 per 10,000 child years). This may be attributed
to wide variations in living conditions as dictated by socio-economic inequalities between
these settlements. Social amenities are limited in less developed (informal and backyard
where less than 36% of the households had flush toilets) settlements making under-five
children more vulnerable to diseases and death. Poor housing type may also be
contributing to high under-five mortality. These findings are consistent with studies
conducted in Ethiopia by Abera Kumie and Yemane Berhane on crowding in a traditional
home. In this study poor housing conditions was found to support the transmission of a
variety of communicable diseases to the level of endemicity causing high childhood
mortality [44].
The Kaplan Meier curves has clearly depicted a probability survival greater than 98% for
under-five children living in private sector housing compared to those in the informal
settlements which reported a probability of survival of less than 95% and backyard
settlements which reported probability of survival of 97% in Soweto Townships. People
in Soweto informal and backyard settlements may be affected by lack of adequate good
housing and proper sanitation (Less than 55% of the houses were built of bricks and less
than 36% of the households had flush toilets) as opposed to the residents of Soweto
private sector housing who were served with good housing and good social amenities
- 35 -
(100% of the households were built of brick and had flush type of toilets). This often has
a huge impact on children living in informal and backyard settlements which often is
home to a wide array of infectious diseases as demonstrated in a study that was
conducted among Preschool Children in Kerala [35].
The same trend has been observed in the under-five mortality rate differential between
the hostel housing settlement and the private sector housing where under-five mortality is
almost four fold lower in the latter. Just like informal and backyard settlements hostels
housing settlements have a spectrum of problems. First, this settlement was built of single
unit houses for migrant workers during apartheid regimes and now housing families with
inherent resource scarcities. Secondly, it is possible that they may not be able to afford
good nutritious food and other social amenities necessary for the good growth and
development of a healthy child which may be the case among the dwellers in the private
housing settlements.
There were no differences in the level of mortality rates between council housing
settlement and private sector housing settlement. This may be attributed to the fact that
this (council housing) settlement just like in the private sector housing had adequate
housing built of bricks (>90%) and over 96% of the households had water piped into the
homesteads.
4.2 CAUSES OF UNDER-FIVE MORTALITY IN SOWETO TOWNSHIPS
The contribution of housing settlement as dictated by environmental factors to under-five
mortality has been seen in Soweto Townships (Table 4.0). Although, most of the verbal
- 36 -
autopsies reported causes of under-five death as ?other?, severe diarrhoea, pneumonia,
injury and tuberculosis commonly associated with poor environmental conditions were
among the reported causes of under-five deaths in Soweto Townships. The classification
of majority of causes of under-five death as ?others? is attributed to the fact that most
symptoms for under-fives are non specific. Recall problems and stigma among the
caregivers during the survey process may also explain why a lot of causes of death
classifications are reported as ?others?.
Severe diarrhoea featured predominantly as a major cause of death in hostel and backyard
settlements. These settlements in Soweto Townships are replete with poor environmental
factors (unclean water, public pit latrines etc) which obviously predispose children to
diarrhoeal diseases and poor health outcomes. Inadequate flush toilets facilities and use
of water from public taps (<36% and <40% respectively) in these settlements predispose
children to infections leading to diarrhoeal diseases. These findings are consistent with
the studies conducted in Kenya, which showed that children living in unsanitary
conditions i.e. slum were more likely to experience diarrhoea than their counterparts in
healthier neighborhoods (UN-HABITAT 2003).
Pneumonia just like diarrhoea is precipitated by poor living conditions as those found in
the backyard, hostel and informal settlements in Soweto Townships. Overcrowding in
hostels and council housing settlements may have contributed to the development of
pneumonia which resulted in deaths among under-five.
- 37 -
4.3 PREDICTORS OF UNDER-FIVE MORTALITY IN SOWETO TOWNSHIPS
The results of univariate logistic regression analysis show that private sector housing
settlement and source of energy for lighting were significantly associated to under-five
mortality in Soweto Townships.
In univariate and multivariate models, the odds of under-five death were significantly
higher for those children who lived in informal housing settlements (Univariate Model;
OR 4.41, P-value 0.007, CI 1.50, 12.96 and Multivariate Model; OR 5.10, P-value 0.005
CI1.633, 15.99) compared to those who lived in private housing settlements. This could
be attributed to housing and social amenity differentials existing between the two housing
settlements.
Residents in private sector housing settlements are well served with good modern
housing, social amenities such as electricity, water and sewage system opposed to
residents in informal settlements. As a result under-five mortality is undoubtedly lower
when compared to informal settlements which are characterized by poor housing and in
adequate social amenities. It is important also to note that chronic cough and tuberculosis
were among the under-five causes of death in the informal settlements which is
precipitated by poor environmental and housing conditions such as those found in
informal settlements. These findings are consistent to findings of the studies conducted
by Peter et al among the Miao in Yunnan, Southwest China in 2001 which demonstrated
important links between child mortality and environmental risk factors [45].
- 38 -
Use of candles as a form of energy, in both univariate and multivariate models has shown
statistically significant association with under-five mortality in Soweto Townships.
The odds of under-five deaths are significantly higher in families where candles are used
as form of lighting relative to those families where electricity, solar or gas is the source of
energy for lighting (Univariate model; OR 2.67, P-value 0.004, CI 1.372, 5.18 and
Multivariate model; OR 7.02, P-value< 0.0001, CI 2.416, 19.15).Use of candles is an
indicator of poverty.
The use of paraffin for lighting which did not show any significant association with
under-five mortality in univariate model, was however significant in multivariate model.
The odds of under-five deaths were 3.4 times more likely in families where paraffin was
used as a source of energy for lighting relative to the use of either electricity, gas or solar
energies (OR 3.43, P-value 0.014, CI 1.25, 8.88)
Just like candles paraffin is a relatively cheap form of energy and when burnt often emit
various pollutants that are harmful particularly to under-five children. Its use undoubtedly
has an impact on the health of under-five children as demonstrated by a study undertaken
by Wichmann and Voyi on Influence of Cooking and Heating Fuel Use on 1?59 Month
Old Mortality in South Africa. The study suggested that exposure to cooking and heating
smoke from polluting fuels is significantly associated with 1?59 month mortality
(RR=1.95; 95% CI=1.04, 3.68).
- 39 -
Other household characteristics such as source of water, type of toilet facilities, number
of household rooms and building materials did not show any significant association with
the under-five mortality in Soweto Townships.
4.4 LIMITATIONS OF THE STUDY
The main limitation of the study is the design of the survey itself. A cross-sectional study
cannot answer the important question of the causal relationship between the variables
(exposure and outcome) because both information on the explanatory and outcome are
obtained at the same time.
The mortality figures reported seem to be way below other figures for South Africa,
which suggest strong under reporting due to recall bias. Most of the causes of under-five
deaths have been reported as others hence limiting the conclusions regarding leading
causes of child deaths in the different housing settlements. In addition causes of death as
well gender were missing for some of the children who died.
The analysis was restricted to variables in the dataset hence it was not possible to analyse
other important variables known to have significant association with child mortality such
as sex, mother?s education level etc.
Important demographic factors such as sex were missing, hence it was difficult for the
researcher to determine the effects of sex on under-five mortality as well as understand
the distribution of under-five deaths by sex in the various housing settlements. Other
- 40 -
variables known to have association with under-five mortality such as socio-economic
status and household size were as well missing.
It was difficult to ascertain whether the housing settlements in which under-five deaths
reported are actually the settlements where the death occurred since information on
change of housing settlement within the last five year period was not provided.
Information regarding the date of death of the child?s mother was not provided and
therefore it was difficult to ascertain who died first.
This report is based on secondary data, thus some of the relevant variables to answer
some of the important questions were not collected such as mother?s educational levels
mother?s marital status, number of persons housed in a household etc.
- 41 -
CHAPTER FIVE
CONCLUSIONS AND RECOMMENDATIONS
5.1 CONCLUSIONS
This study has examined the differentials existing in under-five mortality rates in
different housing settlements in Soweto Townships during a five year period 1998 - 2002.
The results have shown that overall under-five mortality rates in private sector housing
settlements were lower compared to under-five mortality rates in informal, hostels and
backyard housing settlements. No differences exist in under-five mortality rates in private
sector and council housing settlements in Soweto Townships.
The study further revealed that housing settlement and source of energy are the main
factors associated with under-five mortality in Soweto Townships after adjusting for
other household characteristics such as sources of water for drinking, type of toilet
facility and number of house rooms. Under-five children living in hostels, council and
informal settlements are more vulnerable to death compared to under-five children living
in private sector housing settlement in Soweto Townships.
The use of candles and paraffin for lighting increase the risk of death among under-five
children compared to the use of electricity, Gas and solar energy sources in Soweto
Townships. Household characteristics such as toilet type, Number of rooms and Type of
building materials did not show any association with under-five mortality in Soweto
Townships.
- 42 -
5.2 RECOMMENDATIONS
The study has revealed an important link between under-five mortality and housing
settlement. There is need for the quality of housing in Soweto Townships to be improved
to the standards of those found in private sector settlements which have showed lower
under-five mortality compared to the other settlements.
Modern energy sources and technologies such as electricity and solar systems have
proved to play a vital role in under-five mortality risk reduction as demonstrated in Sri
Lanka which has been extraordinarily successful in reducing its under-five mortality rates
in the last half the century that currently stand at 13 per 1000 live births. Studies
conducted in this country showed that those households having no access to electricity,
under-five mortality were 2 times higher than households having access to electricity. It
is therefore important that homes without electricity be provided with electricity to avoid
the use of candles and paraffin for lighting which have been associated with under-five
mortality in Soweto Townships.
More importantly, is to address poverty which seems to be the root cause of development
of deprived settlements all over the world and may be explaining the social economic and
housing type prevailing in Soweto Townships. Indeed, there is also need to explore
hidden environmental and social factors which might also be contributing to the increased
risk of under-five mortality in Soweto Townships.
- 43 -
REFERENCES
1) Luis H, Miguel T and Fernando G, 2006; National and sub-national
under-five mortality profiles in Peru: a basis for informed policy
decisions. BMC Public Health. 6: 173.
2) Schenk K, Murove T, Williamson J, 2006; Protecting children's rights in
the collection of health and welfare data. Health Hum Rights.9(1):80-100.
3) Lawn JE, Costello A, Mwansambo C, Osrin D, 2007; Countdown to
2015: will the Millennium Development Goal for child survival be met?
Arch Dis Child. 92(6):551-6
4) Mrudula K, Geeta K, C Purohit, B Bhavsa, 2000; Comparative Study of
Under-Five Deaths and Family Planning Status in District Ahmedabad.
Indian Journal of Community Medicine. Vol. 25, No. 3
5) Kath A, David A, and Davidson R, 2005; How does progress towards the
child mortality millennium development goal affect inequalities between
the poorest and least poor? Analysis of Demographic and Health Survey
data.BMJ.2005;331(7526):1180?1182.doi:
10.1136/bmj.38659.588125.79
6) Mark T, Mickey C, David S, Debbie B, Michael H, David G,2007;Setting
Priorities in Child Health Research Investments for South Africa. doi:
10.1371/journal.pmed.0040259
7) Kristanto B, 1983; Some socio-economic factors affecting infant and
child mortality with special reference to Indonesia. Majalah Demografi
Indones.10(19):i-19-32.
8) Glenda g, Ronelle N, Helen S,2006;The effects of adult morbidity and
mortality on household welfare and the well-being of children in Soweto.
Vulnerable Children and Youth Studies,1(1): 17?30
9) The making of Soweto. Accessed online on 1st Nov 2007 from;
http://www.joburg.org.za/soweto/overview.stm)
10) Alan G and Owen Crankshaw, 1999;Comparing south African and Latin
American experience: Migration and Housing mobility in Soweto
- 44 -
11) Alan C, Alan G and Alan M, 2000; Backyard Soweto. International
journal of urban and regional Research vol. 24.4
12) Backstory: Touring the real South Africa. Retrieved online on 1st Nov
2007 from; http://www.csmonitor.com/2007/0124/p20s01-litr.html
13) Leah G and Vardar S, 2003; Health, ageing and social differentials: A
case study of Soweto, South Africa. Journal of Cross-Cultural
Gerontology 18: 105?125, 2003
14) Bozzoli et al. Continuity or Rupture: The City, Post-Apartheid - South
Africa. Accessed on 1st November 2007 from:
http://findarticles.com/p/articles/mi_m2267/is_4_65/ai_54098118/pg_5
15) Joan S, 2004; Child Mortality Report .JAMA 2:2329
16) Dorothy E , Solomon R, 1997; Africa in the 21st century-can despair be
turned to hope; BMJ 315:1444-1446.
17) Anupam G, Shabbar J, Stephen K, Debbie B and Michael L, 2006; Rates
and causes of child mortality in an area of high HIV prevalence in rural
South Africa. Tropical Medicine & International Health 11 (12), 1841?
1848.
18) Lee W, Albert I Alon U and Mitermayer G, 2007; Slum health: Diseases
of neglected populations. BMC Int Health Hum Rights 7: 2.
19) Eyob Z, Diane M 2003; Inequities in under-five child malnutrition in
South Africa. Int J Equity Health 2: 7.
20) Ines Ackerl K, 2004; Community study of acute respiratory infections in
children less than one year of age. Danish Medical Bulletin - No. 3. Vol.
51 Page 308.)
21) Risk: Inadequate sanitation. Accessed on 24th October 2007 from;
http://www.viva.org/?page_id=50&risk_id=RJS
- 45 -
22) Population report; Population report; Meeting the Urban Challenge.
Accessed on 24th October 2007 from;
http://www.infoforhealth.org/pr/m16edsum.shtml
23) Bret Thiele, 2002; The Human Right to Adequate Housing: A Tool for
Promoting and Protecting Individual and Community Health. American
Journal of Public Health 712-715 Vol 92, No. 5
24) Macassa G, Ghilagaber G, Bernhardt E, Burstrom B, 2004; Contribution
of household environment factors to urban childhood mortality in
Mozambique. East Afr Med J.81(8):408-14
25) Islam MM, Azad KM. 2007; Rural-urban migration and child survival in
urban Bangladesh: are the urban migrants and poor disadvantaged? J
Biosoc Sci. 1-14 [Epub ahead of print]
26) K Moser, D A Leon, D Gwatkin , 2004; How does progress towards the
child mortality Millennium Development Goal affect inequalities between
the poorest and least poor? Paper presented at the Global Forum for
Health Research, Mexico City, November 2004
27) Murray J, Lopez D, 2007, 1997; Global mortality, disability, and the
contribution of risk factors: Global Burden of Disease Study. Lancet.
349(9063):1436-42
28) Kimani E. and Ngindu M, 2007; Quality of Water the Slum Dwellers
Use: The case of a Kenyan Slum. J Urban Health.
29) Jean F, 2006; Child health inequities in developing countries-
differences across urban and rural areas. Int J Equity Health. 5: 9.
30) Jean-C, Alex E, Nyovani M, and James C, 2007; Progress towards the
child mortality millennium development goal in urban sub-Saharan
Africa: the dynamics of population growth, immunization, and access to
clean water. BMC Public Health. 2007; 7: 218.
31) Woldemicael G, 2000; The effects of water supply and sanitation on
childhood mortality in urban Eritrea. J Biosoc Sci.;32(2):207-27
- 46 -
32) Danielle C, Sandro G, Waleska T, and David V, 2007; Social
Determinants of the Health of Urban Populations: Methodologic
Considerations. J Urban Health.; 84(Suppl 1): 42?53.
33) Susan C, Susan K, Nicholas F,2003;Healthy Housing: A Structured
Review of Published Evaluations of US Interventions to Improve Health
by Modifying Housing in the United States, 1990?2001. Am J Public
Healthr 93(9): 1471?1477
34) Yoshiharu F, Keiko N and Takehito T, 2004; Wide range of
socioeconomic factors associated with mortality among cities in Japan.
Health Promotion International, Vol. 19, No. 2, 177-187.
35) C. Soman, Malathi D, S. Rajasree, V. Raman K and K. Vijayakumar,
1991; High Morbidity and Low Mortality?the Experience of Urban
Preschool Children in Kerala. Journal of Tropical Pediatrics 37(1):17-24;
doi:10.1093/tropej/37.1.17
36) Patrick B, Nick F, Warren G, Bruce L, Rebecca M, and Linda B, 2004;
The Relationship between Housing and Health: Children at Risk.
Environmental Health Perspectives Volume 112 Number 15.
37) Akhtar H, Keramat Ali and Gunnar K, 1999; Determinants of mortality
among children in the urban slums of Dhaka city, Bangladesh. Tropical
Medicine and International Health volume 4 no 11 pp 758?764
38) William J. Moss, 2007; Measles Still Has a Devastating Impact in
Unvaccinated Populations. PLoS Med 4(1): e24
39) Assaad F, 1983; Measles: summary of worldwide impact. Rev Infect Dis
5(3):452-9
40) Rima R, Shiraz H, and Joumana S, 2007; Harboring illnesses: On the
association between disease and living conditions in a Palestinian refugee
camp in Lebanon. Int J Environ Health Res:16(2):99-111.
41) Rennie M; Housing and environmental factors and their effects on the
health of children in the slums of Karachi, Pakistan. Journal of Biosocial
Science (1997), 29: 271-281 Cambridge University Press.
42) Bo B, Finn D and Lars S, 1999; Child Mortality in Stockholm during
1885?1910: The Impact of Household Size and Number of Children in
the Family on the Risk of Death from measles. American Journal of
Epidemiology Vol. 149, No. 12: 1134-1141
- 47 -
43) S Iyer and M Monteiro, 2004; The risk of child and adolescent mortality
among vulnerable populations in Rio de Janeiro, Brazil. Journal of
Biosocial Science (2004), 36: 523-546
44) Abera Kumie1 and Yemane Berhane. Crowding in a traditional rural
housing (?Tukul?) in Ethiopia; Accessed on 7th February 2008 from;
http://www.cih.uib.no/journals/EJHD/ejhd16-n3/ejhdv16no3-page303.PDF
45) Peter F, Nagib A, C?line M, Hui Z and Zeyuan L, 2001; Risk factors and
child mortality among the Miao in Yunnan, Southwest China. Social
Science&Medicine Volume 53, Issue 12.
- 48 -
APPENDICES
APPENDIX ONE: UNIVARIATE MODEL OUTPUTS
(i)Univariate model of under-five mortality and housing settlements
xi:logistic Mort_1 i.newdomain
i.newdomain _Inewdomain_1-5 (naturally coded; _Inewdomain_1 omitted)
Logistic regression Number of obs = 2825
LR chi2(4) = 11.58
Prob > chi2 = 0.0208
Log likelihood = -372.24676 Pseudo R2 = 0.0153
----------------------------------------------------------------------------------------------------
Mort_1 | Odds Ratio Std. Err. z P>|z| [95% Conf. Interval]
-------------+--------------------------------------------------------------------------------------
_Inewdomai~2 | 3.104409 1.736146 2.03 0.043 1.037393 9.28998
_Inewdomai~3 | 3.634106 2.003479 2.34 0.019 1.233477 10.70691
_Inewdomai~4 | 4.41523 2.425604 2.70 0.007 1.504274 12.95924
_Inewdomai~5 | 4.19913 2.29792 2.62 0.009 1.436647 12.2735
-------------------------------------------------------------------------------------------------------
(ii)Univariate model of under-five mortality and source of Energy
xi:logistic Mort_1 i.Energy_source
i.Energy_source _IEnergy_so_1-4 (naturally coded; _IEnergy_so_1 omitted)
Logistic regression Number of obs = 2814
LR chi2(2) = 9.07
Prob > chi2 = 0.0107
Log likelihood = -369.67884 Pseudo R2 = 0.0121
------------------------------------------------------------------------------------------------------
Mort_1 | Odds Ratio Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------------------------------
_IEnergy_s~3 | 1.779288 .5407665 1.90 0.058 .9807234 3.228092
_IEnergy_s~4 | 2.665774 .9032352 2.89 0.004 1.372188 5.178847
--------------------------------------------------------------------------------------------------------
- 49 -
(iii)Univariate model of under-five mortality and type of wall building material
xi:logistic Mort_1 i.Build_materials
i.Build_mater~s _IBuild_mat_1-5 (naturally coded; _IBuild_mat_1 omitted)
Logistic regression Number of obs = 2821
LR chi2(2) = 0.22
Prob > chi2 = 0.8958
Log likelihood = -377.80523 Pseudo R2 = 0.0003
--------------------------------------------------------------------------------------------------
Mort_1 | Odds Ratio Std. Err. z P>|z| [95% Conf. Interval]
-------------+------------------------------------------------------------------------------------
_IBuild_ma~4 | 1.205357 1.237533 0.18 0.856 .1611359 9.016526
_IBuild_ma~5 | 1.112637 .2648084 0.45 0.654 .6978582 1.773945
(iv) Univariate model of under-five mortality and water sources
xi:logistic Mort i.water_source
i.water_source _Iwater_sou_1-3 (naturally coded; _Iwater_sou_1 omitted)
Logistic regression Number of obs = 2807
LR chi2(2) = 1.60
Prob > chi2 = 0.4498
Log likelihood = -373.20727 Pseudo R2 = 0.0021
---------------------------------------------------------------------------------------------------
Mort_1 | Odds Ratio Std. Err. z P>|z| [95% Conf. Interval]
-------------+-------------------------------------------------------------------------------------
_Iwater_so~2 | 1.140214 .2826373 0.53 0.597 .7014391 1.853459
_Iwater_so~3 | 1.489999 .4597948 1.29 0.196 .8137964 2.728073
----------------------------------------------------------------------------------------------------
- 50 -
(v)Univariate model of under-five mortality and Number of household rooms.
xi:logistic Mort_1 i.Rooms_nos
i.Rooms_nos _IRooms_nos_1-47 (naturally coded; _IRooms_nos_1 omitted)
note: _IRooms_nos_4 != 0 predicts failure perfectly
_IRooms_nos_4 dropped and 16 obs not used
note: _IRooms_nos_47 dropped due to collinearity
Logistic regression Number of obs = 2808
LR chi2(2) = 4.54
Prob > chi2 = 0.1035
Log likelihood = -375.25313 Pseudo R2 = 0.0060
------------------------------------------------------------------------------------------------
Mort_1 | Odds Ratio Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------------------------
_IRooms_no~2 | .7199014 .1642441 -1.44 0.150 .4603339 1.125831
_IRooms_no~3 | .359434 .2179684 -1.69 0.092 .1095046 1.179793
--------------------------------------------------------------------------------------------------
(v) Univariate model of under-five mortality and type of toilet facility.
. xi:logistic Mort_1 i.toilet_type
i.toilet_type _Itoilet_ty_1-3 (naturally coded; _Itoilet_ty_1 omitted)
Logistic regression Number of obs = 2778
LR chi2(2) = 0.39
Prob > chi2 = 0.8243
Log likelihood = -365.95225 Pseudo R2 = 0.0005
-----------------------------------------------------------------------------------------------
Mort_1 | Odds Ratio Std. Err. z P>|z| [95% Conf. Interval]
-------------+--------------------------------------------------------------------------------
_Itoilet_t~2 | 1.052819 .3350712 0.16 0.872 .5642211 1.964527
_Itoilet_t~3 | 1.612651 1.184139 0.65 0.515 .3824013 6.800823
-----------------------------------------------------------------------------------------------
- 51 -
APPENDIX TWO: MULTIVARIATE MODEL STATA OUTPUT
xi:logistic Mort_1 i.newdomain i.Energy_source i.Build_materials i.water_source
i.Rooms_nos i.toilet_type
i.newdomain _Inewdomain_1-5 (naturally coded; _Inewdomain_1 omitted)
i.Energy_source _IEnergy_so_1-4 (naturally coded; _IEnergy_so_1 omitted)
i.Build_mater~s _IBuild_mat_1-5 (naturally coded; _IBuild_mat_1 omitted)
i.water_source _Iwater_sou_1-3 (naturally coded; _Iwater_sou_1 omitted)
i.Rooms_nos _IRooms_nos_1-47 (naturally coded; _IRooms_nos_1 omitted)
i.toilet_type _Itoilet_ty_1-3 (naturally coded; _Itoilet_ty_1 omitted)
note: _IRooms_nos_4 != 0 predicts failure perfectly
_IRooms_nos_4 dropped and 16 obs not used
note: _IRooms_nos_47 dropped due to collinearity
Logistic regression Number of obs = 2734
LR chi2(14) = 27.05
Prob > chi2 = 0.0190
Log likelihood = -347.81434 Pseudo R2 = 0.0374
-------------------------------------------------------------------------------------------------------
Mort_1 | Odds Ratio Std. Err. z P>|z| [95% Conf. Interval]
------------------------------------------------------------------------------------------------------
_Inewdomai~2 | 3.365174 1.890126 2.13 0.032 1.098501 10.12381
_Inewdomai~3 | 2.519372 1.648045 1.29 0.176 0.6610907 9.506119
_Inewdomai~4 | 5.10362 2.675241 2.53 0.005 1.6025824 15.89455
_Inewdomai~5 | 4.092971 2.131604 2.33 0.020 1.242855 11.42454
_IEnergy_s~3 | 3.42689 1.66107 2.40 0.014 1.248159 8.865632
_IEnergy_s~4 | 7.021591 3.566894 3.62 0.000 2.406577 19.01456
_IBuild_ma~4 | 0.7552301 0.7976152 -0.27 0.790 0.095303 5.984833
_IBuild_ma~5 | 0.5962384 0.2172533 -1.42 0.156 0.2919207 1.217797
_Iwater_so~2 | 0.7776655 0.2303333 -0.85 0.396 0.4351903 1.389653
_Iwater_so~3 | 0.602572 0.3166111 -0.96 0.335 0.2151591 1.687556
_IRooms_no~2 | 0.8292717 0.2332674 -0.67 0.506 0.4778161 1.439239
_IRooms_no~3 | 0.4712081 0.3078451 -1.15 0.249 0.1309526 1.695553
_Itoilet_t~2 | 0.747281 0.3320463 -0.66 0.512 0.3127963 1.78528
_Itoilet_t~3 | 1.193837 0.9807063 0.22 0.829 0.2386189 5.972899
--------------------------------------------------------------------------------------------------------