CHAPTER 7

EARLY CHILDHOOD MORTALITY

George Kichamu

7.1 Background and Assessment of Data Quality

This chapter presents information on mortality among children under five years of age in Kenya.
Specifically, estimates are presented on levels, trends and differentials in neonatal, postneonatal, infant, child,
and under-five mortality. This information is useful to both the demographic assessment of the population
and the evaluation of health policies and programmes. Estimates of infant and child mortality may be used
as inputs into population projections, particularly if the level of adult mortality is known from another source
or can be inferred with reasonable confidence. Information on mortality of children also serves the needs of
organisations which provide health services by identifying sectors of the population which are at high
mortality risk.

The risk of death during childhood varies by age of the child, typically being highest immediately
following birth and decreasing as the child gets older. The pace at which mortality declines with increasing
age will thus ultimately determine the proportion surviving to specific ages. If mortality declines very rapidly
after birth as in most industrialised countries, then infant and under-five mortality rates will be relatively low.
In these settings the highest proportion of under-five deaths occur in the first few days of life and are related
to genetic, maternal, and perinatal factors. If, however, as in less-developed countries, environmental and
nutritional risks to survival persist through infancy and afterwards, then those risks will be reflected in an
attenuated age pattern and an overall higher level of mortality under age five. In this chapter, age-specific
mortality rates are defined as follows:

• Neonatal mortality (NN): the probability of dying within the first month of life,

• Postneonatal mortality (PNN) : the arithmetic difference between infant and neonatal
mortality,

• Infant mortality (1q0): the probability of dying between birth and the first birthday,

• Child mortality (4q1): the probability of dying between exact age one and the fifth birthday,

• Under-five mortality (5q0): the probability of dying between birth and the fifth birthday.

All rates are expressed as deaths per 1,000 live births, except child mortality, which is expressed as deaths
per 1,000 children surviving to the first birthday.

The mortality rates presented in this chapter are calculated from information drawn from the
questions asked in the birth history section of the women's questionnaire. Preceding the birth history, probing
questions are posed on the aggregate childbearing experience of respondents (i.e., the number of sons and
daughters who live with the mother, the number who live elsewhere, and the number who have died). In the
birth history, for each live birth, information is collected on sex, month and year of birth, survivorship status,
and current age, or if the child had died, the age at death.

89
 The quality of mortality estimates calculated from retrospective birth histories depends on the
completeness with which births and deaths are reported and recorded. Potentially the most serious data
quality problem is the selective omission from the birth histories of births that did not survive, which can lead
to underestimation of mortality rates. Other potential problems include displacement of birth dates, which
may cause a distortion of mortality trends, and misreporting of the age at death, which may distort the age
pattern of mortality.

When selective omission of childhood deaths occurs, it is usually most severe for deaths in early
infancy. If early neonatal deaths are selectively underreported, the result is an unusually low ratio of deaths
under seven days to all neonatal deaths and an unusually low ratio of neonatal to infant deaths.
Underreporting of early infant deaths is most commonly observed for births that occurred longer before the
survey; hence, it is useful to examine the ratios over time.

Inspection of these ratios (shown in Appendix Tables C.5. and C.6) indicates that no significant
numbers of early infant deaths were omitted in the 1998 KDHS. First, the proportion of neonatal deaths
occurring in the first week of life is high, 74 percent. Further, this proportion is roughly constant over the
20 years before the survey (between 68 and 77 percent). Second, the proportion of infant deaths occurring
during the first month of life is entirely plausible in level (42 percent), and is stable over the 20 years before
the survey (varying between 39 and 46 percent). This inspection of the mortality data indicates there is no
evidence of selective underreporting of deaths or misreporting of age at death.

It is important to recognise that any method of measuring childhood mortality that relies on mothers’
reports (e.g., birth histories) rests on the assumption that adult female mortality is not very high or, if it is
high, there is little or no correlation between the mortality risks of mothers and their children. In countries
with high rates of adult female mortality, these assumptions will seldom hold and the resulting childhood
mortality rates will be underestimated to some degree.

7.2 Levels and Trends in Early Childhood Mortality

Table 7.1 presents childhood mortality rates for the periods 0-4, 5-9, and 10-14 years before the
survey. Under-five mortality for the period 0-4 years before the survey (approximately 1994-1998) is 112
deaths per 1,000 births. This means that, currently, 1 in 9 Kenyan children does not live to the fifth birthday.
For every two deaths that occur in the first 12 months of life, approximately one occurs in the ensuing four
years. Current infant mortality stands at 74 deaths per 1,000 births and child mortality (1-4 years) at 41 per
1,000. In the first 12 months, there is roughly one neonatal death for every two postneonatal deaths.
Mortality during the neonatal and postneonatal periods is 28 and 45 deaths per 1,000, respectively.

Table 7.1 Rates of early childhood mortality

Neonatal, postneonatal, infant child, and under-five mortality by five-year periods

preceding the survey, Kenya 1998
__________________________________________________________________
Years Neonatal Postneonatal Infant Child Under-five
preceding mortality mortality mortality mortality mortality
survey (NN) (PNN) (1q0) (4q1) (5q0)
__________________________________________________________________
0-4 28.4 45.3 73.7 40.8 111.5
5-9 25.5 42.1 67.7 33.5 98.9
10-14 28.8 33.2 61.9 29.5 89.6

90
 The 1998 KDHS data, in combination with similarly collected data from the 1993 KDHS, provide
evidence of a worsening mortality situation in the 1990s. Figure 7.1 shows infant and under-five mortality
rates calculated for each of the three five-year periods before the 1993 KDHS and 1998 KDHS. Both infant
and under-five rates exhibit slow but steady improvement (decline) through the late 1980s, after which the
rates plateau and then begin to rise. The increases become rather pronounced during the period between the
early and mid-1990s, and have occurred at all ages. While neonatal mortality has not changed substantially,
postneonatal mortality (1-11 months) and child mortality (12-59 months) have both risen by more than one-
third during the period between the mid-1980s and the mid-1990s.

Figure 7.1
Trends in Infant and Under-five Mortality
1993 KDHS and 1998 KDHS
Deaths/1000 Live Births
120

100 Under-five Mortality

80

60
Infant Mortality

40

20

0
1980 1985 1990 1995 2000
Calendar Year
1993 KDHS Infant 1998 KDHS Infant
1993 KDHS Under-five 1998 KDHS Under-five

7.3 Socioeconomic Differentials in Early Childhood Mortality

Many behaviours that determine survival of young children, such as use of health services, are
themselves preconditioned by social and economic factors. Differentials in infant and child mortality by
urban-rural residence, province, and mother’s level of education are presented in Table 7.2. The mortality
estimates are calculated for a 10-year period before the survey so that the rates for each category are based
on sufficient number of cases to support reliable statistical estimation. 1

Under-five mortality is 23 percent higher in rural areas (109 per 1,000) than in urban areas (88 per
1,000) (Figure 7.2); the urban-rural difference is especially pronounced during the first year of life. The risk
of children dying varies widely across provinces. Under-five mortality is highest in Nyanza Province, where
1 in 5 children dies before the fifth birthday (199 per 1,000), and lowest in Central Province where 1 in 30
children under age five dies (34 per 1,000). The provincial differentials seen here are consistent with those
observed in the 1993 KDHS data.

1
Later, a three-year reference period is used for examination of mortality differentials by use of maternity services
and mother’s assessment of the child’s size at birth. This is necessary since the relevant data are drawn from the child
health section of the questionnaire, which is referenced to the period covering the 3 years before the survey.

91
 Table 7.2 Neonatal, postneonatal, infant child, and under-five mortality by socioeconomic
characteristics

Neonatal, postneonatal, infant child, and under-five mortality for the ten-year period
preceding the survey, by socioeconomic characteristics, Kenya 1998
_________________________________________________________________________
Post-
Neonatal neonatal Infant Child Under-five
Socioeconomic mortality mortality mortality mortality mortality
characteristic (NN) (PNN) (1q0) (4q1) (5q0)
_________________________________________________________________________
Residence
Urban 20.3 35.1 55.4 34.8 88.3
Rural 28.4 45.4 73.8 37.6 108.6
Province
Nairobi 19.5 21.6 41.1 26.1 66.1
Central 17.7 9.7 27.3 6.3 33.5
Coast 27.7 42.1 69.8 27.9 95.8
Eastern 22.6 30.5 53.1 26.1 77.8
Nyanza
38.1 97.3 135.3 73.4 198.8
Rift Valley
28.3 22.0 50.3 18.5 67.8
Western
20.1 43.8 63.9 62.5 122.5
Education
No education 27.5 54.7 82.2 43.9 122.5
Primary incomplete 31.7 59.7 91.4 51.5 138.1
Primary complete 29.7 31.7 61.4 27.2 86.9
Secondary+ 16.7 23.4 40.0 20.7 59.9

Total 27.0 43.7 70.7 37.1 105.2

Figure 7.2
Under-five Mortality by Background Characteristics

RESIDENCE
Rural 109
Urban 88

PROVINCE
Nyanza 199
Western 122
Coast 96
Eastern 78
Rift valley 68
Nairobi 66
Central 34

MOTHER'S EDUCATION
No education 123
Primary incomplete 138
Primary complete 87
Scondary+ 60

0 50 100 150 200

Deaths per 1,000 Live Births

KDHS 1998

92
 There is, as expected, a strong link between a mother’s level of education and her children’s survival
chances. This relationship is, however, not a simple one, because mortality does not decrease uniformly with
increasing level of education. Children of women with a primary incomplete education have slightly higher
mortality at each age segment (under-five mortality: 138 per 1,000) than children of women with no education
(123 per 1,000). Children of women with a completed primary education and those with some secondary
schooling have much lower under-five mortality risks (87 and 60 per 1,000, respectively). This inverted U-
shape pattern of the education-mortality relationship was also observed in the 1993 KDHS data, and can
potentially be explained by the fact that at least two parts of the country, Rift Valley and Coast provinces,
exhibit mortality levels that are lower than would be expected based on their levels of maternal education.
This example underscores the potential problem of interpreting bivariate associations without consideration
of the more complex multifactorial reality. In this case, an in-depth analysis of provincial variation in the
education-mortality relationship would provide greater insight than is offered here.

7.4 Biodemographic Differentials in Early Childhood Mortality

The relationship between early childhood mortality and various demographic variables is examined
in Table 7.3. Male children experience slightly higher mortality than their female counterparts. Under-five
mortality rates for males and females are 108 and 103 deaths per 1,000 births, respectively. The excess
mortality among male children is largely due to higher risk during the first month of life, a universal pattern
explained by sex-linked heritable factors.

The relationship between childhood mortality and mother’s age at birth shows the expected U-shaped
pattern, with children of the youngest and the oldest women experiencing the highest risk of death. A similar,
but less pronounced, pattern occurs regarding birth order of the child. Generally, first-order births and high-
order births (7+) are observed to have higher mortality levels than births of order 2-6. An exception to this
pattern is seen in the neonatal period, during which first births exhibit relatively low risk. This may be due
to a favourable change in the maternal age profile for first births, with a smaller proportion of first births
occurring among very young women.

A marked relationship exists between the length of the preceding birth interval and the risk of death
in early childhood. The KDHS data show that short birth intervals significantly reduce a child’s chances of
survival. The risk associated with rapidly-paced childbearing is especially grave during the neonatal period
when a child born after a short interval (<24 months) carries a risk of dying (42 per 1,000) more than three
times that of a child born after 4 or more years (13 per 1,000). These findings underscore the potential for
mortality reduction that could result from increased birth spacing in Kenya.

Maternal care during pregnancy and delivery has a significant bearing on the health of both mother
and child and thus on the risk of early childhood mortality. The KDHS data show that children born to
women who obtained both antenatal and delivery care from medically trained persons during pregnancy have
considerably lower mortality than children whose mothers received only antenatal or delivery care, or
received neither.

A child’s size at birth is an important indicator of potential risk of dying during infancy, particularly
during the first months of life. In the KDHS, mothers were asked whether the referenced child was very
small, small, average size, large, or very large at birth. This type of subjective assessment has been shown
to correlate closely with actual birth weight. Newborns perceived by their mothers to be very small or small
are much more likely to die in the first year than those perceived as average or larger in size, with most of
the excess mortality occurring during the first month.

93
 Table 7.3 Neonatal, postneonatal, infant child, and under-five mortality by biodemographic
characteristics
Neonatal, postneonatal, infant child, and under-five mortality for the ten-year period preceding
the survey, by selected biodemographic characteristics, Kenya 1998
___________________________________________________________________________
Post-
Neonatal neonatal Infant Child Under-five
Biodemographic mortality mortality mortality mortality mortality
characteristic (NN) (PNN) ( 1 q0 ) (4 q 1 ) ( 5 q0 )
___________________________________________________________________________
Sex of child
Male 29.6 44.9 74.5 35.9 107.8
Female 24.3 42.4 66.8 38.4 102.6

Age of mother at birth

< 20 36.8 60.2 97.0 48.4 140.7
20-29 21.5 37.1 58.6 32.3 89.1
30-39 30.9 43.5 74.5 37.0 108.7
40-49 39.4 69.8 109.2 68.7 170.5

Birth order
1 22.1 40.6 62.7 34.0 94.7
2-3 24.2 38.5 62.7 31.9 92.6
4-6 26.6 40.6 67.3 37.2 101.9
7+ 41.1 65.6 106.7 53.9 154.9

Previous birth interval

< 2 years 41.5 60.7 102.2 53.6 150.3
2-3 years 26.7 41.1 67.9 33.7 99.3
4 or more years 13.1 31.2 44.3 24.7 67.9

Medical maternity care

Received both ANC and DS 23.9 23.2 47.1 NA NA
Received no ANC or DS or
received only ANC or DS 32.9 59.3 92.2 NA NA
1
Size at birth
Small or very small 51.8 45.3 97.2 NA NA
Average or larger 21.3 42.9 64.2 NA NA

Total 27.0 43.7 70.7 37.1 105.2

____________________________________________________________________________
ANC = Antenatal care
DS = Delivery services (from medical personnel)
NA = Not applicable

7.5 High-Risk Fertility Behaviour

Numerous studies have demonstrated a strong relationship between a mother’s pattern of fertility and
her children’s survival chances (Boerma and Bicego, 1992; United Nations, 1994). The results presented in
the previous section support this. Typically, infants and young children have a higher risk of dying if they
are born to very young mothers or older mothers, if they are born after a short interval, or if their mothers
have already had many children. In the following analysis, mothers are classified as “too young” if they are
less than 18 years old at the time of birth, and “too old” if they are age 35 years or more at the time of birth.
A “short” birth interval is defined as one less than 24 months, and a “high-order” birth is one occurring after
three or more previous births (i.e., birth order 4 or higher). Births are also cross-classified by combinations
of these characteristics. Thus, a birth may have from zero to three potentially high-risk characteristics. While
first births are often considered high risk, they are not an avoidable risk in the same sense as the other factors
and are thus treated separately in this analysis.

94
 Column one of Table 7.4 shows the percentage of births five years before the survey that fall into
various risk categories. Fifty-six percent of births are in at least one high-risk category, with about 18 percent
having multiple high-risk characteristics. Risk ratios are presented in column two; the risk ratio is the ratio
of the proportion of children in a particular risk category who have died to the proportion in the specified
reference category who have died. Births in the reference category are those who are not in any high-risk
category. The first finding of interest is the high risk associated with births to very young mothers (<20
years), both as a single high-risk factor and in combination with a short birth interval. Second, high birth
order (>3) is an even more important risk factor not only because it is associated with high risk—alone and
in combination with short birth interval and older age (35+ years) at birth—but because a very high
proportion of all births are of order 4 and above (41 percent).2 The latter finding points to the importance of
limiting family size to enhance the health and survival prospects of children.

The third column of Table 7.4 shows the distribution of currently married, nonsterilised women by
the risk category into which a currently conceived birth would fall. A comparison of this percentage with the
distribution of actual births in the last five years indicates that without fertility control, the percentage of
births falling into multiple high-risk categories that contain high birth order (>3) would rise significantly.
Further, the percentage of births with multiple high-risk characteristics would rise from 18 to 41 percent.

2
This includes birth order >3 with mother’s age at birth 35+, birth order >3 with birth interval < 24 months, and birth
order >3 with mother’s age at birth 35+ and birth interval < 24 months.

95
Table 7.4 High-risk fertility behaviour

Percent distribution of children born in the five years preceding the survey by category
of elevated risk of dying, and the percent distribution of currently married women at
risk of conceiving a child with an elevated risk of dying, by category of increased risk,
Kenya 1998
______________________________________________________________________
Births in 5 years
preceding the survey
Percentage
_____________________
of currently
Percentage married
a
Risk category of births Risk ratio women
______________________________________________________________________
b
Not in any high-risk category 24.8 1.00 24.5

Unavoidable risk category

(First births) 19.4 0.88 5.2

Single high-risk category

Mother's age <18 5.9 2.63 0.3
Mother's age >34 0.2 0.00 2.9
Birth interval <24 months 8.0 1.06 9.4
Birth order >3 23.2 1.41 17.0

Subtotal 37.3 1.52 29.6

Multiple high-risk category

c
Age <18 & birth interval <24 months 0.5 2.57 0.1
Age >34 & birth interval <24 months 0.0 0.00 0.2
Age >34 & birth order >3 9.2 1.39 25.6
Age >34 & birth interval <24 months
& birth order >3 1.3 3.77 4.1
Birth interval <24 months
& birth order >3 7.3 2.17 10.6

Subtotal 18.4 1.90 40.7

In any high-risk category 55.7 1.65 70.3

Total 100.0 - 100.0

Number of births 5,558 - 4,834
______________________________________________________________________
Note: Risk ratio is the ratio of the proportion dead of births in a specific high-risk
category to the proportion dead of births not in any high-risk category.
a
Women were assigned to risk categories according to the status they would have at the
birth of a child, if the child were conceived at the time of the survey: age less than 17
years and 3 months, age older than 34 years and 2 months, latest birth less than 15
months ago, and latest birth of order 3 or higher.
b
Includes sterilised women
c
Includes the combined categories Age <18 and birth order >3.

96