Content uploaded by Saroj Jayasinghe
Author content
All content in this area was uploaded by Saroj Jayasinghe
Content may be subject to copyright.
The social determinants of childhood mortality in Sri Lanka: time-
trends & comparisons across South Asia
Tanja A.J. Houweling, Saroj Jayasinghe & Tarani Chandola
Department of Epidemiology & Public Health, University College London, London, UK
Received August 23, 2007
Background & objectives: Sri Lanka has been able to achieve low childhood mortality levels at low
cost. However, this achievement may have been at the expense of increasing mortality inequalities
between socio-economic groups. This study addresses the question whether socio-economic mortality
inequalities rise as overall mortality falls by describing socio-economic inequalities in under 5
mortality in Sri Lanka and comparing the magnitude of these inequalities over time and with other
South Asian countries. Further, the role of female autonomy, fertility, malnutrition, and health care
use in explaining the observed patterns in mortality inequality were also examined.
Methods: Time-trends in inequality in under 5 mortality by maternal education were described
using data from the 1987, 1993 and 2000 Sri Lanka Demographic and Health Surveys (DHS). Using
DHS data, the magnitude of these inequalities was compared across 50 low and middle income
countries, and with three South Asian countries in particular. Socio-economic inequalities in
determinants of under 5 mortality were estimated for Sri Lanka over time, and compared with such
inequalities in the other South Asian countries.
Results: Absolute inequalities in under 5 mortality in Sri Lanka were very low internationally, while
relative mortality inequalities were high. The decline in under 5 mortality between the 1987 and
2000 survey in Sri Lanka had been accompanied by rising relative mortality inequalities across
educational groups. High and improving levels of health care use and declining levels of malnutrition
in Sri Lanka ran parallel with high and increasing relative inequalities in undercoverage of health
care and malnutrition.
Interpretation & conclusion: Despite the low overall under 5 mortality levels and absolute mortality
inequalities, Sri Lanka exhibited a clear mortality gradient across educational groups. Further, the
high and rising relative inequalities in under 5 mortality in Sri Lanka showed that the achievement
of low mortality might be at the expense of increasing relative mortality inequalities between socio-
economic groups. Increasing inequalities in malnutrition and undercoverage of health care, perhaps
related to a strong gradient in female autonomy across educational groups, may have contributed to
the rising relative under 5 mortality inequalities in this country.
Key words Asia - child mortality/trends - comparative study - mothers/education - socio-economic factors - Sri Lanka
239
Indian J Med Res 126, October 2007, pp 239-248
Sri Lanka has been able to achieve relatively low
childhood mortality levels at low cost
1
. However, the
achievement of low mortality may have been at the
expense of increasing inequalities between socio-
economic groups
2
. The rate of decline for lower socio-
economic groups may not be as quick as for the better-
off, resulting in higher absolute inequalities in mortality.
In addition, even if absolute mortality inequalities
remain stable or decline, it is possible for relative
mortality inequalities to increase. It is therefore
important to examine whether the low childhood
mortality that Sri Lanka has achieved has been at the
expense of increasing absolute and relative inequalities
in health between socio-economic groups.
Fig. 1 shows a scatter plot of gross domestic product
(GDP) per capita by under 5 mortality, showing a clear
curve-linear association between higher GDP per capita
and lower under 5 mortality levels
3
. Sri Lanka is located
towards the bottom of the graph, distinct from its South
Asian neighbours. However, what remains unknown is
whether Sri Lanka retains this privileged position vis-
à-vis its neighbours and other low and middle income
countries in terms of the absolute and relative mortality
inequalities.
This study was undertaken to answer the question
whether socio-economic mortality inequalities rise as
overall mortality levels fall. It aimed to describe socio-
economic inequalities in under 5 mortality in Sri Lanka
and compare the magnitude of these inequalities over
time and with other South Asian countries. More
specifically, the question if Sri Lanka is performing well
in relation to its time-trends, neighbours and other low-
middle income countries in terms of relative and absolute
socio-economic inequalities in under 5 mortality, was
examined. In order to understand the international
patterns and time-trends in inequality in under 5 mortality,
we explored the role of several mortality determinants.
Specifically, we describe inequalities in maternal and
childhood malnutrition, under-coverage of health care
(i.e., childhood immunization, professional antenatal care
and delivery attendance), fertility, and female autonomy
and compare these across the South Asian countries and
over time for Sri Lanka.
240 INDIAN J MED RES, OCTOBER 2007
Fig. 1. Under 5 mortality (per 1000 live births) by GDP per capita (PPP, constant international 2000 $) for 176 countries, 2005.
Data source: Ref 3. All countries for which data for 2005 were available are included in this figure.
Sri Lanka
Bangladesh
Nepal
India
Material & Methods
Data: The Demographic and Health Surveys (DHS)
were the main data source for this study. The DHS are
nationally representative surveys, for which usually
between 5000 and 10,000 women, aged 15-49 yr, are
interviewed. The surveys include retrospective birth
histories, which provide mortality data for individual
children, and information on socio-economic
characteristics such as maternal education.
Information on under 5 mortality rate stratified by level
of maternal education was obtained, from the DHS
website
4
, for all 57 countries with surveys in 1995 or later.
These include countries in Africa, Asia, Latin America
and North Africa/the Near East. If multiple surveys were
available for the same country, only the most recent data
were obtained. For Sri Lanka the same information was
obtained from the 1987, 1993 and 2000 Sri Lanka DHS
Final Reports
5-7
. For the South Asian countries included
in the study, we used data closest to 2000 as to facilitate
comparisons with the 2000 Sri Lanka data. The under 5
mortality rates obtained are based on information on births
and deaths during the 10 yr prior to the survey.
The three surveys in Sri Lanka do not include the
Eastern Costal belt and the Northern province for security
reasons (approximately 14% of the 1987 population
5
).
In 1993 and 2000 also a few areas in the Amparai district
were excluded for the same reason. This means that the
data presented in this paper cannot necessarily be taken
to be representative for the country as a whole.
Design and analysis: A combined time-trend/cross-country
design was used. First, using cross-country comparisons
across all low and middle income countries with recent
DHS data, Sri Lanka’s international position in terms of
socio-economic inequalities in under 5 mortality was
assessed. This was combined with a more detailed
comparison of Sri Lanka with the three other South Asian
countries in the data set (India, Nepal, Bangladesh).
Secondly, we described trends in inequality in under 5
mortality between the 1987 and 2000 survey for Sri Lanka.
Maternal education, rather than household wealth,
was used to assess socio-economic inequalities in under
5 mortality, as under 5 mortality rates by household
wealth were not readily available for Sri Lanka.
Moreover, the currently available measure of household
wealth may not allow for time-trend analysis
8
. Three
educational categories were used: no education, primary,
and secondary+. For Sri Lanka, also basic rates were
presented using a more detailed categorisation (4 groups).
Two main measures were used to describe and
compare mortality inequalities, i.e., the relative index
of inequality (RII) as a measure of relative inequality
and the slope index of inequality (SII) as a measure of
absolute inequality. Both the RII and the SII capture
the differences in mortality experience between all
educational groups, and can be interpreted as the
(estimated) mortality ratio (difference - in case of SII)
between the lowest and highest educated
9
. This
estimate captures only the linear relationship between
socio-economic position and mortality. As these
measures take the relative size of educational groups
into account, they allow for direct comparisons
between countries and time periods. The RII and SII
were calculated on the basis of aggregate under 5
mortality rates by educational group, using linear
regression analysis with relative group size as
weighting variable. Countries with extreme heaping
(>98% of the population in one educational group)
(five countries) were excluded from the analysis, as
were two outliers (Peru and Philippines, with an RII
of 25 and 81 respectively), leaving us with 50 countries
for the international comparisons. The results obtained
using the RII and SII were compared with results using
the rate ratio and rate difference, comparing the
extreme groups ‘no education’ and ‘secondary+’. The
rate ratio and rate difference do not take group size
into account.
Finally, a first step towards explaining the time-
trends and international variations in the magnitude of
mortality inequality was set by describing inequalities
in mortality determinants for Sri Lanka over time, and
by comparing these inequalities with those in the three
other South Asian countries. For consistency reasons,
all mortality determinants included in this study were
defined in negative terms (e.g., undercoverage rather
than coverage of health care).
Results
Sri Lanka in international perspective: Sri Lanka had
among the lowest absolute inequalities (in terms of the
slope index of inequality) in under 5 mortality across
educational groups among the 50 countries (Fig. 2).
Further, the gap in under 5 mortality levels between the
extreme groups ‘no education’ and ‘some secondary and
higher’ was smallest in Sri Lanka (results not shown).
Relative inequalities in under 5 mortality showed a
more mixed picture. When using an inequality measure
that takes into account the mortality experience in all
educational groups, as well as the relative size of the
HOUWELING et al: SOCIAL DETERMINANTS OF CHILDHOOD MORTALITY 241
242 INDIAN J MED RES, OCTOBER 2007
Fig. 2. Absolute inequality (Slope index of inequality) in under 5 mortality by maternal education by average under 5 mortality (per 1000 live
births) for 50 DHS countries.
Fig. 3. Relative inequalities (Relative index of inequality) in under 5 mortality by maternal education by average under 5 mortality (per 1000
live births) for DHS 50 countries.
Nepal
India
Bangladesh
Sri Lanka
Sri Lanka
India
Nepal
Bangladesh
educational groups (RII), relative inequalities in under
5 mortality in Sri Lanka were high compared to the other
countries (Fig. 3). At the same time, when comparing
only the lowest and highest educated, relative
inequalities in Sri Lanka were only moderately large
(results not shown).
When comparing under 5 mortality levels by
educational group in Sri Lanka in more detail with the
other South Asian countries, the following was
observed. At any given educational level, under 5
mortality in Sri Lanka was substantially lower than in
other South Asian countries (Fig. 4). Under 5 mortality
among children born to women with no education in
Sri Lanka, for example, was only 32 per 1000 compared
to around 120 in the other countries. Absolute
inequalities across educational groups (SII) were
substantially lower in Sri Lanka. Relative inequalities
across educational groups were, however, substantially
higher in Sri Lanka compared to the other South Asian
countries. This is mainly because of the comparatively
high under 5 mortality levels among the primary
educated in Sri Lanka.
Time-trends in under 5 mortality inequality in Sri Lanka:
Total under 5 mortality declined by 51 per cent, from 42
to 21 per 1,000, between the 1987 and 2000 survey
(Table I). All educational groups experienced declines in
under 5 mortality. Absolute mortality declines were
probably strongest among the group with no education.
Hence, the educational mortality gap between the lowest
and highest educated probably declined in absolute terms.
When taking all educational groups into account, relative
inequalities showed an increase over time (from an RII of
2.4 in 1987 to 7.5 in 2000). The increasing RIIs in Sri
Lanka were primarily due to slower under 5 mortality
declines in the ‘primary education’ group. In all periods, a
systematic gradient in mortality, with higher mortality
levels among children of the lower educated, was observed.
In 2000, under 5 mortality among children born to mothers
with primary education or below was more than twice as
high as that in the ‘more than secondary education’ group
(Table I).
Determinants of time and place variations:
Rural residence - The overall percentage of women
living in rural areas was fairly similar across the countries
(Table II). In Sri Lanka, the difference in rural residence
between lower and higher educated women was smaller
than in the other countries. Sri Lanka has seen little
changes over time in this respect, perhaps only a minor
decline in educational differences in rural residence.
Female autonomy - Compared to women in
Bangladesh and Nepal, women in Sri Lanka reported less
problems relating to getting permission to seek care for
themselves and knowing where to go; getting money for
treatment seems more of a problem (Table III). However,
educational inequalities in female autonomy were as high
or higher (in absolute and relative terms) in Sri Lanka
HOUWELING et al: SOCIAL DETERMINANTS OF CHILDHOOD MORTALITY 243
Sri Lanka Bangladesh India Nepal
average under 5 mortality 21 110 101 108
relative index of inequality 7.51 2.66 4.72 3.58
slope index of inequality 32 96 125 118
Fig. 4. Under 5 mortality levels (per 1000 live births) by educational
group for four South Asian countries DHS 1998-2000.
Data Source: Ref. 4, 6.
Table I. Under 5 mortality levels (per 1000 live births) for total
population and by maternal education, Sri Lanka
Under 5 mortality Decline between
(per 1,000 live births) 1987-2000
1987 1993 2000 Absolute %
Total population 42 32 21 21 51
Maternal education
no education 71 62 [32] 40 55
primary 43 42 33 10 22
secondary 41 27 19 22 54
more than secondary 26 20 14 12 45
RII [a] 2.4 7.3 7.5
SII [b] 35 48 32
Figures within parentheses are based on 250-499 births. [a] RII gives
the estimated mortality ratio between the lowest and highest educated
when modelling a linear trend in effect, using 3 educational
categories; [b] SII gives the estimated mortality difference between
the lowest and highest educated when modelling a linear trend in
effect, using 3 educational categories.
Source: Ref. 5-7.
Under 5 mortality
244 INDIAN J MED RES, OCTOBER 2007
Table III. Women's autonomy in four South Asian countries - % of women experiencing problems when needing to get medical advice or
treatment for themselves
Sri Lanka Sri Lanka Sri Lanka Bangladesh India Nepal
1987 1993 2000 2004 [1] 1998/99 2001
Problems with getting permission for treatment (%):
Total na na 5 17 na 17
No education na na 16 19 na 19
Primary na na 11 19 na 13
Secondary or higher na na 3 14 na 8
Rate difference 14 5 11
Rate ratio 6.39 1.36 2.30
Not knowing where to go (%):
Total na na 6 10 na 28
No education na na 20 11 na 32
Primary na na 12 11 na 24
Secondary or higher na na 3 7 na 12
Rate difference 17 3 19
Rate ratio 6.32 1.48 2.56
Problems with getting money for treatment (%):
Total na na 33 14 na 66
No education na na 67 16 na 76
Primary na na 57 15 na 53
Secondary or higher na na 24 10 na 29
Rate difference 43 5 46
Rate ratio 2.74 1.50 2.58
[1] for Bangladesh, 2004 data were used as 2000 DHS data were not readily available for these indicators
na - not available
Source: Ref. 4-7
Table IV. Total fertility rate by educational attainment for four South Asian countries
Sri Lanka Sri Lanka Sri Lanka Bangladesh India Nepal
1987 1993 2000 2004 [1] 1998/99 2001
Total na na 1.9 3.3 2.8 4.1
No education na na 2.4 4.1 na 4.8
Primary na na 2.9 3.3 na 3.2
Secondary or higher na na 1.9 2.4 na 2.2
Rate difference 0.5 1.7 2.6
Rate ratio 1.3 1.7 2.2
na - not available
Source: Ref. 4-7
Table II. Per cent of women living in rural areas, by educational attainment, for four South Asian countries
Sri Lanka Sri Lanka Sri Lanka Bangladesh India Nepal
1987 1993 2000 1999/2000 1998/99 2001
Total 84 81 80 80 74 90
No education 92 89 86 86 86 94
Primary 89 86 86 83 74 88
Secondary or higher 80 78 78 66 52 72
Rate difference 12 11 8 20 34 22
Rate ratio 1.15 1.14 1.10 1.30 1.66 1.30
Source: Ref. 4-7
HOUWELING et al: SOCIAL DETERMINANTS OF CHILDHOOD MORTALITY 245
Table V. Prevalence of undernutrition among mothers and children in four South Asian countries
Sri Lanka Sri Lanka Sri Lanka Bangladesh India Nepal
1987 1993 2000 1999/2000 1998/99 2001
Maternal malnutrition (% with BMI below 18.5)[1]
Total na na 22 45 41 27
No education na na 38 52 47 30
Primary na na 31 49 44 18
Secondary or higher na na 18 30 30 19
Rate difference na na 20 22 16 11
Rate ratio na na 2.07 1.74 1.53 1.59
% of children aged 3-36 months that is stunted [2]
Total 28 21 12 40 46 43
No education 51 41 35 48 55 47
Primary 34 29 19 41 45 35
Secondary or higher 21 16 10 25 31 26
Rate difference 29 24 25 23 24 21
Rate ratio 2.39 2.48 3.50 1.90 1.77 1.81
[1] % of women with body mass index (BMI) below 18.5, for women with births in the last three years preceding the survey. The BMI excludes
pregnant women and those who are less than three months postpartum. [2] height for age below-2 standard deviations from the median
reference population.
na - not available
compared to the other countries. In particular, differences
between the secondary education group and the rest of
the population were large in Sri Lanka.
Fertility - The total fertility rate was lower in Sri
Lanka than in the other South Asian countries (Table
IV). An educational gradient in total fertility was
observed for Sri Lanka, but it was weaker than in Nepal
and Bangladesh. Primary educated mothers, rather than
mothers with no education, had the highest fertility
rate in Sri Lanka, in contrast with Bangladesh and
Nepal.
Malnutrition - Over one in five mothers in Sri Lanka
was undernourished (Table V). The prevalence of
maternal undernourishment in Sri Lanka was somewhat
lower than in Nepal, and about twice as low as in India
and Bangladesh. However, absolute and relative
educational inequalities in Sri Lanka were as large or
larger than in the other countries. In 2000 the country
exhibited a particularly large contrast between the
secondary educated and the rest of the population.
Similarly, the substantially lower levels of chronic
malnutrition among under three year olds in Sri Lanka
went together with a steeper socio-economic gradient
compared to the other countries. Moreover, the strong
overall decline in childhood malnutrition in Sri Lanka
went together with declining absolute inequalities and
increasing relative inequalities in malnutrition. In 2000,
very substantial inequalities between educational groups
were still observed, with 35 per cent of children in the
‘no education’ group being stunted and only 10 per cent
of children in the ‘some secondary education or higher’
group.
Health care use - Sri Lanka has achieved virtually
universal professional antenatal care and nearly
universal professional delivery care (overall 0 and 3%
of births respectively not received these types of care),
strongly contrasting with high levels of under-coverage
in the other South Asian countries (Table VI). The
improvements in coverage of antenatal and delivery
care over time in Sri Lanka, have been accompanied
by declining absolute inequalities. Absolute
inequalities in maternal care in 2000 in Sri Lanka were
(much) smaller than in the other South Asian countries.
Relative inequalities in undercoverage of delivery care
had, however increased over time, and were high from
a South Asian perspective, as remaining
undercoverage became increasingly concentrated
among the lower educated. Similarly, the strong
improvements in full childhood immunization
coverage in Sri Lanka-in absolute terms particularly
among the lower educated-had seemingly gone
together with rising relative inequalities in under-
coverage of immunization. In 2000, Sri Lanka still
exhibited a clear educational gradient in
undercoverage of delivery care and immunization.
Discussion
Sri Lanka is well known for its low mortality levels
at a low level of national per capita income. Our study
showed that the low mortality levels in Sri Lanka were
combined with small absolute inequalities in under 5
mortality between educational groups. These absolute
inequalities were low from an international perspective.
This may be expected, as low overall mortality levels
can only be attained if all major social groups are doing
well
10
. However, at these comparatively low mortality
levels, Sri Lanka’s relative mortality inequalities were
high from an international perspective. Moreover, the
50 per cent fall in under 5 mortality in Sri Lanka has
probably been accompanied by increasing relative
mortality inequalities across educational groups. This
appeared primarily due to a lagging behind of the
primary education group. Our results for Sri Lanka
showed that a very good position in terms of overall
under 5 mortality levels, can be accompanied by high,
and further increasing, relative mortality inequalities.
246 INDIAN J MED RES, OCTOBER 2007
Sri Lanka’s position internationally, and its trends
over time, depend on the specific groups compared
and inequality measures used. There are indications
that the ‘no education’ group in Sri Lanka has seen
comparatively large declines in under 5 mortality. This
is remarkable, as increasingly small groups tend to
become negatively selected, i.e., as having a low
educational attainment becomes more an exception,
it is increasingly associated with characteristics that
instigate high mortality. The experience of this small
group (5% of women in 2000) can, however, not be
taken to be representative for country as a whole. In
contrast, primary educated women constitute a much
larger group (18% of women in 2000
5
). We used
inequality measures that capture the mortality
experience of all educational groups and that take
differences in group size across countries and over time
into account (the RII and SII). These measures can be
less robust when only a small number of socio-
economic categories are used, especially in
combination with a non-linear association between
Table VI. Health care undercoverage, by maternal education in four South Asian countries
Sri Lanka Sri Lanka Sri Lanka Bangladesh India Nepal
1987 1993 2000 1999/2000 1998/99 2001
% undercoverage of antenatal care [1]:
Total 3 0 0 67 35 52
No educaiton 14 0 2 80 52 61
Primary 4 1 0 69 24 36
Secondary or higher 2 0 0 40 10 14
Rate difference 12 0 2 40 41 48
Rate ratio 6.75 - - 2.01 5.10 4.47
% undercoverage of delivery care [2]:
Total 12 6 3 88 58 87
No educaiton 31 21 16 95 76 93
Primary 16 10 7 92 53 82
Secondary or higher 7 3 2 70 26 54
Rate difference 24 18 14 25 50 39
Rate ratio 4.52 6.60 9.20 1.36 2.94 1.73
% undercoverage of full childhood immunization [3]:
Total 33 5 7 nc nc nc
No educaiton 44 11 19 nc nc nc
Primary 41 6 9 nc nc nc
Secondary or higher 28 4 6 nc nc nc
Rate difference 16 7 13
Rate ratio 1.59 2.67 3.06
[1] % birth during 5 yr prior to survey without professional antenatal care; [2] % births during 5 yr prior to survey without professional
delivery care; [3] % of children aged 12-23 months that is not fully immunized (BCG, DPT3, Polio3, measles) among children with a health
card. nc-not comparable with data from Sri Lanka. [4] antenatal and delivery care data for India refer to births in 3 yr prior to the survey
Source: Ref.4-7
HOUWELING et al: SOCIAL DETERMINANTS OF CHILDHOOD MORTALITY 247
socio-economic position and mortality. It is unlikely
that this explains the increasing relative mortality
inequalities in Sri Lanka, as an increase is also
observed when describing mortality inequalities
between two broad educational categories (primary
education or lower vs. secondary education and above).
Secondly, sample sizes in DHS are often too small to
report changes in mortality inequality at conventional
levels of significance
8
. Also in our study, we could
not exclude the possibility that our observations were
due to chance variations. Nevertheless, the systematic
nature of the increase in relative mortality inequalities
over three time periods for Sri Lanka suggested the
increase was real. Although DHS are the best data
source available for a combined cross-country/time-
trend analysis of mortality inequalities in low and
middle income countries
8
, it would be desirable if more
powerful data sources become available for such
research
11
. Finally, the level and trends in relative
mortality inequalities in Sri Lanka presented here
might be underestimated given that a few regions with
security problems in the country were excluded from
the DHS survey. It may be expected that these regions
exhibit lower educational levels, higher mortality
levels and a slower decline in under 5 mortality, but
this would require further research.
The high and increasing relative mortality
inequalities concurrent with declining overall under 5
mortality levels in Sri Lanka were in accordance with
the general tendency of relative inequalities to rise when
overall mortality levels fall
8
. This tendency can probably
be explained by differential improvements in proximate
mortality determinants across socio-economic groups,
with slower and later improvements among more
disadvantaged groups and regions
2,8
.
Levels of health care use are high in Sri Lanka
compared with the other South Asian countries and have
improved over time. Coverage of antenatal care is
virtually universal and coverage of professional delivery
attendance is exceptionally high in all educational
groups
12
. Also full childhood immunization coverage
reaches over 80 per cent in all groups. However, these
high overall levels and the strong improvements in
health care use over time in Sri Lanka, went together
with comparatively high and increasing relative
inequalities in undercoverage. Similarly, the (from a
South Asian perspective) low levels of maternal and
childhood malnutrition in Sri Lanka, were accompanied
by comparatively steep educational gradients. Further,
while childhood malnutrition in Sri Lanka declined,
relative inequalities increased. In contrast, inequalities
in total fertility are comparatively low in Sri Lanka.
The highest fertility level was, however, observed
among the primary educated, which might have
contributed to the relatively high mortality levels in this
group.
Socio-economic inequalities in under 5 mortality
are often partly explained by the fact that lower socio-
economic groups tend to live in rural areas, which
typically exhibit characteristics that instigate high
mortality
8
. Sri Lanka, with its small rural-urban
inequalities in under 5 mortality, was exceptional in
that sense (results not shown). Whereas inequalities
within rural areas between the estates (tea plantations)
and other rural areas were very large and probably
increasing (results not shown), this appeared not to
fully explain the increasing relative mortality
inequalities by educational attainment, given the
relatively small size of the estate population (6-7% of
population). Hence, the high and increasing relative
inequalities in under 5 mortality in Sri Lanka seemed
to be a predominantly socio-economic, rather than a
regional phenomenon.
One of the factors that perhaps partly explains the
high and increasing relative mortality inequalities in
Sri Lanka was the comparatively strong educational
gradient in female autonomy in this country. Earlier
studies have associated Sri Lanka’s low mortality
levels with its good record in terms of female
autonomy
13
. Average levels of female autonomy,
however, hide a strong educational gradient in
autonomy, which may have contributed to the
increasing concentration of malnutrition and under-
coverage of health care among lower educated group
as overall levels improve.
In conclusion, despite the low overall under 5
mortality levels and absolute mortality inequalities, Sri
Lanka exhibited a clear mortality gradient across
educational groups. Under 5 mortality among children
born to mothers with primary education or below was
about twice as high as that among children born to
higher educated mothers. Further, relative inequalities
across all educational groups in Sri Lanka were high
internationally, and had probably increased over time.
This shows that the achievement of low under 5
mortality levels in Sri Lanka may have been at the
expense of increasing relative inequalities between
socio-economic groups. As levels of health care use in
Sri Lanka are high and further increasing,
undercoverage becomes increasingly concentrated in
lower socio-economic groups. Similarly, as overall
levels of malnutrition decline in Sri Lanka, malnutrition
becomes increasingly concentrated in lower socio-
economic groups. The lower levels of female autonomy
among the lower educated compared to the higher
educated in Sri Lanka, might have played a role in the
unequal improvement of these mortality determinants
and the rising mortality inequalities.
References
1. Halstead SB, Walsh JA, Warren KS, editors. Good health at
low cost. New York: Rockefeller Foundation; 1985.
2. Victora CG, Vaughan JP, Barros FC, Silva AC, Tomasi E.
Explaining trends in inequities: evidence from Brazilian child
health studies. Lancet 2000; 356 : 1093-8.
3. World Bank. World development indicators 2006. Washington,
The World Bank; 2006.
4. www.measuredhs.com, accessed on April 19, 2007.
5. Department of Census and Statistics, Ministry of Plan
Implementation, Institute for Resource Development. Sri
Lanka Demographic and Health Survey 1987, Colombo:
Department of Census and Statistics; 1988.
6. Department of Census and Statistics in collaboration with Ministry
of Health NaW. Sri Lanka Demographic and Health Survey 2000.
Colombo: Department of Census and Statistics; 2002.
7. Department of Census and Statistics, Ministry of Finance P,
Ethnic Affairs and National Integration. In collaboration with
Ministry of Health HaSS. Sri Lanka Demographic and Health
Survey 1993. Colombo: Department of Census and Statistics;
1995.
8. Houweling TAJ. Socio-economic inequalities in childhood
mortality in low and middle income countries, PhD thesis:
The Netherlands: ErasmusMC University Medical Center
Rotterdam; 2007.
9. Mackenbach JP, Kunst AE. Measuring the magnitude of socio-
economic inequalities in health: an overview of available
measures illustrated with two examples from Europe. Social
Sci Med 1997; 44 : 757-71.
10. Houweling TAJ, Kunst AE, Huisman M, Mackenbach JP.
Using relative and absolute measures for monitoring health
inequalities: experiences from cross-national analyses on
maternal and child health. Int J Equity Health; 2007 (in press).
11. Korenromp E, Arnold F, Williams B, Nahlen B, Snow R.
Monitoring trends in under-5 mortality rates through national
birth history surveys. Int J Epidemiol 2004; 33 : 1-9.
12. Houweling TAJ, Ronsmans C, Campbell OM, Kunst AE. Huge
poor-rich inequalities in maternity care: an international
comparative study of maternity and child care in developing
countries. Bull World Health Organ 2007; 85 : 745-54.
13. Caldwell JC. Health transition: the cultural, social and
behavioural determinants of health in the Third World. Social
Sci Med 1993; 36 : 125-35.
248 INDIAN J MED RES, OCTOBER 2007
Reprint requests: Dr Tanja A.J. Houweling, Department Epidemiology & Public Health, University College London
1-19 Torrington Place, WCIE 6BT, London, UK
e-mail: tanja.houweling@gmail.com
