An Assessment of the Quality of Neonatal Mortality Data Margaret Mambori Mwaila Q50/76335/2009 University of NAIROBI Library 0435762 0 A Project submitted to the Population Studies and Research Institute in Partial Fulfilment for the award of a Degree of Master of Arts at the University of Nairobi
36
Embed
An Assessment of the Quality of Neonatal Mortality Data
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
An Assessment of the Quality of Neonatal MortalityData
Margaret Mambori Mwaila Q50/76335/2009
University of NAIROBI Library
0435762 0
A Project submitted to the Population Studies and Research Institute in Partial Fulfilment for the award
of a Degree of Master of Arts at the University ofNairobi
DECLARATION
This research proposal is my original work and to the best of my knowledge has not been presented for a degree in any other university:
Name
Margaret M Mwaila
Signature Date
This research proposal has been submitted for examination with our approval as University Supervisors:
Supervisors Signature Date
Dr Alfred A Otieno
Mr Ben Obonyo Jarabi oU fj
DEDICATION
To: Karanja, Nyambura and MutwaWithout your unwavering support, this far I would not have come. Love.
ACKNOWLEDGEMENT
Foremost, I take this opportunity to sincerely thank my supervisors, Dr Alfred A Otieno and Mr
Ben O Jarabi for their assiduous effort in supporting and guiding my thought process and all the
work put forth in this study.
Second, all my lecturers through whose hands, I have been remodelled to view development
from the population and social perspective. My gratitude goes to Prof. John Oucho, Dr Lawrence
Ikamari (for the stewardship of the Institute and wise counsel), Dr Kimani Murungaru, Dr
Wanjiru Gichuhi (for the headstart in reviewing literature), Dr Anne Khasakhala, Mr George
Odipo and Mr Andrew Mutuku. Through the guidance of Prof. Oucho, the class was able to
work on a model national policy and programme - to be published. The work earned the class a
presentation at the UNFPA office in Gigiri.
This report will be incomplete without my acknowledging my colleagues - the Class of 2009!
There were moments of stress, but never falling short of comics and opportunities for a good
laugh! Victor Achieng will forever remain etched on our minds for providing unlimited reading
resources (5-10 pages of references even for a two-page writeup!); our very own ‘beloved Prof
Ohon for the sweeping statements - e.g. “the isomorphism of population landscapes, contours
and topologies....”. But that was not all about Ohon, he earned the title ‘Prof from being too
philosophical. Wait until he makes a presentation! Of course, Annette Aduda who ensured there
was a pinch of laughter even in the most absurd of situations and a good story to accompany her
reason for being late for anything. Gilbert Omedi and Sarafina Wanja (watu wa Uthiru) were
always very patient and willing to assist - I am ever so grateful for the support they gave me.
Oh! You could always tell when Inviolata Njeri was stressed (especially during exams) - she
would dance! Joseph Murage provided the IT support and fatherly care. Bernard Nyauchi and
Victor Esendi were always at hand to help. You could easily miss out Benjamin Kitavi - always
so calm and collected. Once he intimated to me, “Na Wanja anapenda kuongea” (Wanja likes
talking). And to complete the picture, I would like to thank Florence Rwamba, Faith Gichanga
and Lucy Njue.
IV
ABSTRACT
Although there has been a reduction in childhood mortality globally, sub Saharan Africa has
realized a very slow improvement. Neonatal mortality in Kenya has shown very little
improvement over the 20 year period. Over the years, health statistics measurement is gaining
currency in terms of reporting returns on investments, impact of interventions and planning.
Major challenges with regard to data quality still remain: first, lack of complete and function
vital registration system which ensures up to date collection of data. Second, due to cultural
attachments and ignorance of legal requirements, early neonatal deaths may go unreported. Total
omission of births and deaths of neonates is definite given the positive correlation between death
of a mother and her child which leads to underestimation of child mortality. Age heaping at day
7 and 14 are commonplace. This is especially important as it distorts estimates of early neonate
deaths. Generally, errors in non-reporting of neonate deaths have serious ramifications with
regard to socio-economic and development planning.
The main objective of this study is to assess the quality of neonatal mortality data from all
KDHS data (1989 - 2008-09) and document impacts and improvements realised. This study will
focus on errors in reporting neonate births and deaths, omissions of neonatal deaths,
displacement of events in time and documentation of trends over the study period.
The sample size comprises 39,560 respondents experiencing 28,654 births and 830 neonate
deaths. The information is compiled from responses from women or reproductive age
experiencing an event over a five-year period prior to a KDHS. Use was made of descriptive
statistics to show associations and differentials.
Using data from five year periods preceding a subsequent survey, this study found that
completeness of information on neonate births and deaths has been vibrant (over 95% reporting).
However, births are more likely to be reported than deaths. Sex ratios at death seem to be over
reported especially for male neonates. Heaping at days 7 and 14 is evident while deaths are
generally misreported.
To achieve MDG 4 and indeed to ensure quality neonate data, Kenya needs to invest most
appropriately in quality data collection systems in order to inform interventions aimed at
reducing reporting error.
V
TABLE OF CONTENTS
DeclarationDedication
AcknowledgementsAbstract
List of Tables ...List of Figures...1. Introduction
1.1 Background1.2 Problem Statement1.3 Research Question
1.4 Objectives of Study
1.5 Study Rationale
1.6 Scope and Limitations of the Study2. Literature Review
2.1 Introduction
2.2 The Case of Neonatal Mortality in Kenya2.3 Data Quality Challenges ...
2.4 Age Pattern of Neonatal Deaths ...3. Methodology
3.1 Introduction
3.2 Data Source
3.3 Methods of Data Analysis
3.4 Description of Variables ...3.5 Displacement of Events in Time ...
3.6 Misreported Age at Death4. Data Quality Assessment
4.1 Introduction
4.2 Completeness of Information on Births and Deaths4.3 Omissions of Deaths
4.4 Displacement of Events in Time ...4.5 Misreported Age at Death
Pageiiiii
ivv
viiiix
12
3
3
34
6
7
79
10
10
10
11
13
15
16 16 17 20
21
VI
5. Data Quality Differentials
5.1 Introduction ... ... ... ... ... ... ... ... 235.2 Completeness of Information Regional Differentials ... ... ... 23
5.3 Ethnicity and Sex Ratio Differentials ... ... ... ... ... 23
Neonatal deaths refer to deaths of babies within the first 28 days after a live birth. Neonatal
mortality is the death of a product of a live birth within 28 days of life. Early neonatal
mortality, on the other hand, refers to death of newborns aged 0 to 6 days among born alive
infants (KNBS and ICF Macro, 2010; WHO, 2006; CDC, 2002).
Globally, more than 4 million neonates die out of the 130 million babies born annually. 99
percent of these deaths occur in developing countries. On average, developing countries
record a neonate mortality rate of 39 deaths per 1000 live births (WHO, 2006). Neonatal
deaths account for over 40 percent of all childhood deaths, globally, yet historically, neonatal
mortality has received limited attention (Oestergaard, et al, 2011). “Neonatal mortality is
increasingly important because the proportion of under-five deaths that occur during the
neonatal period is increasing as under-five mortality declines’' (UNICEF, 2011).
In Kenya, most of the deaths are experienced within less than 24 hours upon a live birth.
KNBS (2010) observes that “high maternal and neonatal deaths occur during the first 48
hours after delivery” also observed by UNDP (2010). A neonatal and perinatal mortality
study done in Kilifi district in the Coast Province of Kenya using 2004-05 Demographic
Surveillance Systems (DSS) data found that neonatal death was the highest contributor to
infant deaths (65 percent) (Bauni, et al, 2005). A situation acknowledged by the United
Nations Millennium Development Goals (MDGs) report of 2010 for Kenya.
Mortality is an important demographic variable required for development planning.
Specifically, child mortality is an important indicator of the impact of child survival
interventions and broadly the socio-economic development of a country (UNICEF, 2011). In
recent times, there has been a growing need for implementers and donors to show returns on
investments, track progress/improvements and overall planning. All these require quality
health statistics, therefore the need to set up health information management and vital
registration systems.
Data quality, is an integral component for ensuring proper estimates of mortality are made.
Data quality can almost be assured in countries with functional vital registration systems.
However, developing countries rely more on indirect methods to derive mortality estimates
(US Census Bureau, 2011; WHO, 2005). “The reliability of the neonatal mortality estimates
depends on accuracy and completeness of reporting and recording of births and deaths.
Underreporting and misclassification are common, especially for deaths occurring early on in
life” (WHO, 2006). Under reporting, especially in countries recording high out of hospital
childbirths, has been noted (Global Health Action, 2011).
African population censuses and surveys are bedevilled by age measurement, not just in
terms of vagueness or digit preference, but also due to “complete ignorance” (Magadi, 1990).
Magadi (1990) further notes that age misreporting and variations in coverage distort the
relation between two age distributions leading to biased estimates.
In her study to estimate the age distribution, census coverage and death registration
completeness in Kenya, Magadi (1990) conceded that data quality is influenced by
demographic, environmental, socio-economic and cultural factors. And errors in data affect
“demographic estimates derived from use of direct or indirect estimation techniques”. The
derivation of estimates using wrong data causes problems when used to model intervention
because such interventions may fail to yield impact. Additionally, Wamai (2004) noted that
poor data quality leads to poor estimates that affect socio-economic planning, health
interventions (with impact on fertility and mortality, sex ratio, planning, etc), in effect a
country’s development.
1.2 Problem Statement
In Kenya, neonate deaths make a significant contribution to infant mortality (60%) compared
to 40 percent globally (UNICEF, 2009; WHO, 2006; WHO, 2009; KNBS and ICF Macro,
2010; UN, 2010; Oestergaard, et al, 2011). In spite of improvements in child health, early
neonatal mortality has been rising over the years contributing to more than three quarters
(82%) of all deaths experienced at infancy in Kenya (KNBS and ICF Macro, 2010). Yet a lot
of emphasis has been laid on studies and programming around infant and child mortality with
little or no consideration of neonate deaths (Oestergaard, et al, 2011).
Functional vital registration systems and Health Management Information Systems (HMIS)
are the best sources of data needed for direct estimations of neonatal mortality. However, due
to their non-existence or dysfunction over the years, the demographic and health surveys are
becoming the major source of data for estimation of vital events in developing countries
(Lancet, 2010; Oestergaard, et al, 2011). Additionally, while the UN interagency Group for
Child mortality, makes annual estimates of infant and under five mortality, it has never done
for neonatal mortality (Oestergaard, et al, 2011) therefore, the need to check the quality of
such data. Despite numerous studies indicating uncertainty with regard to data quality, little
attempt has been made at assessing the data to detect and innovate means of reducing such
errors. It then follows that the quality of neonatal mortality data cannot be assured. This
study attempts to assess data quality of neonatal data collected over five DHS’ (1989 till
2009).
1.3 Research Question
What is the extent of errors in neonatal mortality data and what are the implications of such
error in estimating neonatal mortality.
1.4 Objectives of the Study
This study aims at:
■ Assessing the quality of all KDHS data (1989 - 2008-09) and documenting
improvements realised in neonatal mortality data quality
Specifically, the study will:
a. Assess completeness of information on neonate births and deaths
b. Assess the extent of omissions of neonatal deaths
c. Assess the extent of displacement of events in time
d. Assess the extent of misreporting of age at death
e. Establish completeness of information on neonate births and deaths and
f. Sex ratio at birth and ethnic group differentials
1.5 Study Rationale
As global infant and under-five mortality begin to decline, neonatal mortality emerges as an
increasingly crucial component of overall under-five mortality thus, receiving additional
attention. Therefore, information on neonatal mortality at international level is in great
demand (WHO, 2006). Again, government, policy makers and programme implementers
need reliable data to model interventions and make crucial investment decisions.
Hall (2005) notes a number of challenges developing countries face with regard to making
accurate measurement of neonatal mortality. Among these are: incomplete or non-vibrant to
non-existent vital registration system, home deliveries not captured by the health
information system, lack of common understanding and policies on death classification
between neonatal and stillbirth. The extreme can be misreporting by healthcare providers - a
public relations gimmick to attract clientele.
It is important to carry out a data quality check to ensure that errors are not the culprit in the
“unstable” trend that is emerging in Kenya’s neonatal mortality. While studying the stall in
mortality decline in Ghana, Johnson, et al (2005) noted the importance of assessing data
quality before concluding on the validity of mortality rates - the report noted: data changes
over time and sampling variability are important aspects requiring quality checks. Period
mortality rates (such as those collected by DHS’) can be affected by the following reporting
errors: incomplete information on the date of birth or death, omissions of births and deaths,
displacement of events in time and age at death misreporting. Sullivan (2008), while
assessing child mortality estimates and credibility of mortality declines using 22 DHS’,
noted that the most important quality errors of DHS birth histories are: those in recorded
dates of birth, underreported deaths, sampling errors and misreporting of age at death.
This study lays emphasis on the importance of data quality in coming up with good estimates
in order to model impactful interventions aimed at reducing neonate and childhood mortality.
Such a study has not been carried out in Kenya before.
1.6 Scope and Limitations of Study
This study limits its scope to the data quality assessment in neonatal mortality in Kenya.
Given that Kenya lacks a functional vital registration system, reporting on deaths of children,
especially those who die at infancy can be riddled with “socio-cultural beliefs and taboos”
(Muganzi in Oucho, et al, 2000; Warnai, 2004; Magadi, 1990). Further, Muganzi brought to
the fore the retrospective nature of the DHS which opens it to age heaping and under
reporting (especially in the case of maternal and neonate death). Such retrospective nature
also makes the DHS susceptible to misreporting, especially for neonatal mortality given
memory recall lapse (Brass, 1968; Kpedekpo, 1977 in Kenyi, 1993). For these reasons,
Muganzi in Oucho, et al (2000) cautioned researchers to bear such errors in mind during data
interpretation and estimates derivation.
The KDHS is a sample survey whose results are generalised to the whole population. A case
is noted of North Eastern region where data was collected only in Garissa town, whose
population can hardly offer a true reflection of the greater pastoralist population of the region
yet that was the case (coverage error). Until 2003, earlier DHS’ have excluded the North
Eastern province despite its unique characteristics.
Use of secondary data is usually a limitation in itself given that the data cannot be specific to
any other study apart from the one data was collected for. Elowever, this study takes
advantage of the consistency of the KDHS data to document data quality errors and show
improvements of the same over the years.
2. LITERATURE REVIEW
2.1 Introduction
This chapter presents the desk review on neonatal mortality especially, its concomitant data
quality issues. Apart from briefly providing the topology of neonatal mortality in Kenya, this
chapter discusses data quality challenges and the age structure of neonate deaths.
According to the 2010 Millennium Development Goals (MDGs) Progress Report, the highest
rates of child mortality were experienced in sub-Saharan Africa. In 2008, the region
contributed 33 out of the 34 countries with high under-five mortality rate (over 100 deaths
per 1000 live births).
UNICEF (2011) notes a global increase of 10 percent in neonatal deaths between 1990 and
2010. The risk of neonatal death is six times higher in developing countries compared to
developed countries with sub Saharan Africa bearing the highest risk and making the least
progress (UNICEF, 2011). While sub-Saharan Africa bears the highest burden with 41
neonatal deaths per 1000 live births (WHO, 2006), a developed country such as Sweden has
a neonatal mortality of 2-3 per 1000 live births. As Kenya grapples with 82 percent early
neonate deaths (KNBS, 2010), Singapore boasts of 0 deaths of neonates (Global Health
Action 2011). Developing countries are home to 99 percent of the four million neonatal
deaths experienced annually at the global level. Moreover, of these, three million are early
neonate deaths (WHO, 2006; Lancet, 2005).
The WHO defines a live birth as
the complete expulsion or extraction from its mother o f a product o f conception, irrespective o f the duration o f the pregnancy, which, after such separation, breathes or shows any other evidence o f life — e.g. beating o f the heart, pulsation o f the umbilical cord or definite movement o f voluntary muscles - whether or not the umbilical cord has been cut or the placenta is attached. Each product o f such a birth is considered live born
2.2 The Case of Neonatal Mortality in Kenya
Neonatal deaths have not significantly dropped in the past 20 years. Figure 1 below traces the
trend in neonate mortality over the 20 year period (KDHS history). In 1989, there were 204
neonate deaths. These reduced to 157 by 1993 and further to 98 by 1998. However, the 2003
KDHS observed an increase to 196 which reduced to 175 by 2008-09 - realising only an 11
percent reduction. Using a line of best fit, it can be observed that neonate deaths reduced
from 204 to 175 over the 20 year period. The trend is unpredictable as it is unstable.
Figure 2.1: Trends in Neonate Deaths, 1989-2008-09
2.3 Data Quality Challenges
One important objective of the recently launched National Road Map for Maternal and
Newborn Health (MNH) MoPHS (2010) is “to strengthen data management and utilisation
for improved MNH”. This will help improve reporting of neonatal mortality thereby improve
data quality.
Numerous studies across the world have noted challenges in data quality. UNICEF (2011)
notes a profound challenge with regard to developing countries’ lack of complete vital
registration systems and health management information systems - both of which are
valuable in generating accurate child mortality estimates. Magadi (1990) pointed out that
African population censuses and surveys are bedevilled by errors due to illiteracy. Among
the problems noted were lack of records (such as birth and death certificates) and societal
attachments to such events. Kichamu (1986), in his thesis on Mortality Estimation in Kenya
with a special Case Study of Vital Registration in Central Province, observed that censuses
and vital registration data reported errors and biases. For example, he noted that it is common
practice to underreport dead children and infants, include stillbirths in number of dead
children (also noted by WHO, 2006), underreport children ever born and ages of children and
mothers. Magadi (1990) also conceded that age reporting, especially age heaping or digit
preference and systematic transfer of ages across age boundaries are common sources of
errors. Hill and Choi (2006), in their study of Neonatal Mortality in the Developing World,
conceded that there are two types of errors that could have profound impacts on early
neonatal and overall neonatal mortality measurements: first, “omissions of live births that die
in the first few days of life” and second, heaping of deaths at day seven.
Omission of child deaths in the birth history is a serious response error for the calculation of
childhood mortality rates. Such error may be deliberate or due to problems of recall. This
problem is common in neonate deaths especially early neonatal deaths. It leads to under
estimation of neonate mortality and distorts age patterns of mortality (Curtis, 1995).
Curtis (1995) pointed to two critical structural biases in neonatal mortality data collection:
one, selection bias where only surviving women are surveyed. This locks out child mortality
experiences of women who died during the study period. As noted by Curtis (1995), this bias
reduces estimation of neonate, infant and child mortality, especially given the premise that
death of a mother increases the risk of child death. The magnitude of selection bias can be
deepened by HIV/AIDS deaths and longer periods between surveys. According to WHO
(2006) in high mortality countries, maternal deaths lead to 5 percent of neonate death under
reporting. Second, demographic and health surveys target women up to 49 years at the time
of the survey. This means that neonate deaths are reported only for women aged 44 and
below at the 5 year study period. On a similar point, Garenne (2003), while reviewing survey
data to study sex ratios in African populations noted that due to targeting of women of
reproductive age (15-49 years), there tends to be over-representation of young women - a
bias in itself.
Hall (2005) noted the following challenges that developing countries face in making accurate
measurements of neonatal mortality: incomplete to non-existence thereof of vital registration
systems, misunderstanding and misclassification of death between stillbirth and neonate (also
noted by Kichamu (1986); Aleshina, et al (2003); Sullivan, (2008) and WHO (2006)) and
deliberate misreporting of death by healthcare workers (also noted by Mwale (undated) while
analysing DHS data in Malawi). Ignorance of the legal requirements for reporting births and
deaths may also lead to underreporting.
2.4 Age Pattern of Neonate Deaths
Neonatal mortality records sharp declines during early neonate and neonate stages then
continue to decline gradually through late infancy and early childhood. Underreporting of
early childhood deaths in retrospective years leads to abnormally low neonate to infant
mortality ratio. In a study by Curtis (1995), Kenya recorded an unusually low neonate to
infant mortality ratio since 1989. Table 2.1 below adds credence to the above observation.
Table 2.1: Age/Sex Pattern of Neonate Deaths, 1989 - 2008-09
Age at Death (Days)
1989 1993 1998 2003 2008-09Sex of child Sex of child Sex of child Sex of child Sex of child
TotalMale Female Total Male Female Total Male Female Total Male Female Total Male Female
Completeness of neonate births and deaths information was reported to be vibrant with all
period studies reporting over 90 percent completeness. Reporting of month and year for
births was the most complete and even indicated an improving trend. However,
completeness of month and year for neonate deaths showed an unstable trend.
Sex ratios at birth were normal - within the modelled threshold. However, while it is true
that more neonate males die compared to females, the ratios were rather high, e.g. 1.54 in
2008-09 - an indication of over-reporting of male neonate deaths. The sex ratio at birth and
ethnicity differential agreed with Garenne’s hypothesis that generally, Bantu ethnic group
has low sex ratios at birth.
Heaping at days that are multiples of seven is evident across the study period. The most
commonly reported occurring on day seven. Throughout the study period, the threshold of a
ratio of 1 was not achieved.
Displacement of events in time was also noted during the study period. The ratio of 100
could not be achieved.
While the quality of neonatal mortality data shows improvement, errors still exist. However,
inaccuracies in this report are not severe enough to warrant invalidity. It is also important to
note that in Kenya the DHS provides the most reliable periodic data for estimations
compared to the Health Information management and Vital Registration Systems.
6.2 Recommendations
National governments and international health bodies need to invest in improved methods
for data collection and measurement of neonatal deaths. For all data sources (both routine -
e.g. HMIS and Vital Registration and non routine - e.g. surveys), there is need to identify
and train qualified staff on data collection, analysis, supervision and reporting so as to
ensure quality.
Reducing the health questions asked at the cut-off period will also improve misreporting and
displacement of events in time as proposed by Sullivan, et al (1990).
25
References
Aleshina, Nadezdha, Gerry Redmond (2003), ‘How High is Infant Mortality Rate in Central and Eastern Europe and the CIS?’. Innocenti Working Paper No. 95. Florence: UNICEF Innocenti Research Centre
Bauni E K, Gatakaa H, Williams T N, Nokes D J, Tsofa B K, Scott J A, 2005. Perinatal and Neonatal Mortality among the Mijikenda Community of Kilifi District
Becker, S and El Daw Suliman, 2005. Have Infant and Child Mortality Increased in West Africa? Evaluation of Evidence from Demographic and Health Surveys, John Hopkins University
Centers for Disease Control, 2002. National Vital Statistics System. Available at Anderson RN, Smith B. Deaths: leading causes for 2002. National Vital Statistics Reports 53(17). http://www.cdc.gov/nchs/deaths.htm
Curtis, Sian L. 1995. Assessment of the Quality of Data Used for Direct Estimation of Infant and Child Mortality in DHS-II Surveys. Occasional Papers No.3. Calverton, Maryland: Macro International Inc.
Economic Policy, Planning and Statistics Office (EPPSO), SPC and Macro International Inc. 2007. Republic of the Marshall Islands Demographic and Health Survey 2007
Global Health Action 2011, 4: 5724 - DOI: 10.3402/gha.v4i0.5724 3
Garenne, M. 2003. Sex Ratios at Birth in African Populations: A Review of Survey Data. Detroit, Michigan: Wayne State University Press
Hall, S. 2005. Neonatal Mortality in Developing Countries: What can we learn from DHS Data?
Hill, K. and Choi, Y. 2006. Neonatal Mortality in Developing World. http://www.demographic-research.org/Volumes/Vol 14/18/DOI: 10.4054/DemRes.2006.14.18
James, W. 1987. The Human Sex Ratio
Johnson, Kiersten, Shea Rutstein and Pav Govindasamy. 2005. The Stall in Mortality Decline in Ghana: Further Analysis of Demographic and Health Surveys Data. Calverton, Maryland, USA: ORC Macro
JK Rajaratnam et al. ’s Neonatal, postneonatal, childhood, and under-5 mortality for 187 countries, 1970-2010: a systematic analysis of progress towards Millennium Development Goal 4 and their Supplementary webappendix. (The article and appendix are accessible through free registration with The Lancet.)Kenya National Bureau of Statistics (KNBS), and ICF Macro, 2010. The Kenya Demographic and Health Survey, 2008-09. Claverton, Maryland: KNBS and ICF Macro
Kenyi, 1993. Correlates of Neonatal Mortality in Kenya: A Look at the Kenya DHS 1989 at National Level. Unpublished Masters Lhesis
Kichamu, 1986. Mortality Estimation in Kenya with Special Case Study of Vital Registration in Central Province. Unpublished Masters Lhesis
Lancet. 2005. 4 million neonatal deaths: when? Where? Why?2005 Mar 5-l;365(9462):891-900.
Lhe Lancet, Volume 375, Issue 9730, Pages 1988 - 2008, 5 June 2010 Doi:10.1016/S0140-6736(10)60703-9: Neonatal, postneonatal, childhood, and under-5 mortality for 187 countries, 1970—2010: a systematic analysis of progress towards Millennium Development Goal 4
Magadi, 1990. Estimation of Age Distributions, Census Coverage and Death Registration Completeness in Kenya. Unpublished Masters Lhesis
Ministry of Health, 2007. National Reproductive Health Policy: Enhancing Reproductive Health of all Kenyans
Misati, J, 2003. Determinants of Child Survival in Kenya: A Comparative Study. Unpublished Maters Thesis
MOPHS, 2010. National Road Map for Accelerating the Attainment of MDGs related to Maternal and Newborn Health in Kenya
Mwale, M W, (undated). Infant and Child Mortality in Malawi
Oestergaard MZ, Inoue M, Yoshida S, Mahanani WR, Gore FM, et al. (2011) Neonatal Mortality Levels for 193 Countries in 2009 with Trends since 1990: A Systematic Analysis of Progress, Projections, and Priorities. PLoS Med 8(8): el 001080. doi: 10.1371 /journal.pmed. 1001080
Oucho, J O, ABC Ocholla-Ayayo, Elias H O Ayiemba, L Odhiambo, Omwando, 2000. Population and Development in Kenya: Essays in Honour of S H Ominde. School of Journalism Press: Nairobi
Pullum, Thomas W. 2008. An Assessment of the Quality of Data on Health and Nutrition in the DHS Surveys, 1993-2003. Methodological Reports No.6. Calverton, Maryland, USA: Macro International Inc.
Research Institute of Obstetrics and Pediatrics [Kyrgyz Republic] and Macro International Inc. 1998. Kyrgyz Republic Demographic and Health Survey, 1997. Calverton, Maryland: Research Institute of Obstetrics and Pediatrics, Ministry of Health of the Kyrgyz Republic and Macro International Inc.
Sullivan, Jeremiah M., George T. Bicego and Shea Oscar Rutstein, Assessment of the Quality of the Data Used for the Direct Estimation of Infant and Child Mortality in the Demographic and Health Surveys in An Assessment of DHS-I Data Quality, DHS
2 7
Methodological Reports 1, Institute for Resource Development/Macro Systems, Inc., Columbia, Maryland, December 1990
Sullivan, Jeremiah M., 2008. An Assessment of the Credibility of Child Mortality Declines Estimated from DHS Mortality Rates (working paper for UNICEF)
UNDP, 2010: Millennium Development Goals Report for Kenya http://www.ke.undp.org/index.php/mdgs/goal-4-reduce-child-mortality
UNICEF, WHO, World Bank, UNDESA/Population Division, 2011. Levels and Trends in Child Mortality: Report of 2011: Estimates Developed by UN Inter-Agency Group for Child Mortality Estimation
United Nations, 2010. The Millennium Development Goals Report 2010. New York: UN
UNICEF, 2009. State of the World’s Children. Celebrating 20 years of the Convention of the Rights of the Child. New York: UNICEF
UNICEF, 2008. Infant Mortality Rates Still High. New York: UNICEF
US Census Bureau, 2011. A Presentation on Mortality Estimations
Wamai, 2004. Detection and Correction of Age Errors: A Case Study of 1989 and 1999 Kenya Census Data. Unpublished Masters Thesis
WHO, 2009. Neonatal mortality, risk factors and causes: a prospective population-based cohort study in urban Pakistan. Bulletin of the World Health Organization 2009;87:130-138. doi: 10.2471/BLT.08.050963
WHO, 2005. Bulletin of the World Health Organisation
WHO, 2006. Neonatal and Perinatal Mortality. Country, Region and Global Estimates. Geneva: WHO
World Health Organisation, 2006. The Global Health Report, 2005