Articles
Reductions in child mortality levels and inequalities in
Lancet 2007; 369: 850–55
See Comment page 804
Institute for Population and
Social Research, Mahidol
University, Thailand
(P Vapattanawong PhD);
School of Population Health,
University of Queensland,
Brisbane, Australia
(M C Hogan AB, E Gakidou PhD,
T Vos PhD, Prof A D Lopez PhD,
S S Lim PhD); Setting Priorities
using Information on Cost-
Effectiveness (SPICE) Project
Ministry of Public Health,
Thailand (M C Hogan, S S Lim);
International Health Policy
Program, Thailand
(P Hanvoravongchai MD);
Faculty of Medicine,
Chulalongkorn University,
Thailand (P Hanvoravongchai);
and Harvard Initiative for
Global Health, Cambridge, MA,
USA (E Gakidou)
Correspondence to:
Margaret C Hogan, Setting
Priorities using Information on
Cost-Effectiveness (SPICE)
Project, Department of Medical
Services, Ministry of Public
Health, Nonthaburi,
Thailand 11000
m.hogan@sph.uq.edu.au
850
Thailand: analysis of two censuses
Patama Vapattanawong, Margaret C Hogan, Piya Hanvoravongchai, Emmanuela Gakidou, Theo Vos, Alan D Lopez, Stephen S Lim
Summary
Background Thailand’s progress in reducing the under-five mortality rate (U5MR) puts the country on track to achieve
the fourth Millennium Development Goal (MDG). Whether this success has been accompanied by a widening or
narrowing of the child mortality gap between the poorest and richest populations is unknown. We aimed to measure
changes in child-mortality inequalities by household-level socioeconomic strata of the Thai population between 1990
and 2000.
Methods We measured changes in the distribution of the U5MR by economic strata using data from the 1990 and
2000 censuses. Economic status was measured using household assets and characteristics. The U5MR was estimated
using the Trussell version of the Brass indirect method.
Findings Average household economic status improved and inequalities declined between the two censuses. There
were substantially larger reductions in U5MR in the poorer segments of the population. Excess child mortality risk
between the poorest and richest quintile decreased by 55% (95% CI 39% to 68%). The concentration index, measured
using percentiles of economic status, in 1990 was −0∙20 (−0∙23 to −0∙18), whereas in 2000 it had dropped to −0∙12
(−0∙15 to −0∙08), a 43% (22% to 63%) reduction.
Interpretation These findings draw attention to the feasibility of incorporating equity measurement into census data.
Thailand has achieved both an impressive average decrease in U5MR and substantial reductions in U5MR inequality
over a 10 year period. Contributing factors include overall economic growth and poverty reduction, improved insurance
coverage, and a scaling-up and more equitable distribution of primary health-care infrastructure and intervention
coverage. Understanding the factors that have led to Thailand’s success could help inform countries struggling to
meet the fourth MDG and reduce inequality.
Introduction mortality. Although Thailand might be, on average, set to
Over the past five decades there have been substantial meet the MDGs for child health, relatively deprived
improvements in the health of Thai children, with segments of the population might be lagging behind.
reductions in the under-five mortality rate (U5MR) from We aimed to measure changes in child mortality
above 160 per thousand in the 1950s and 60s to below inequalities by household-level socioeconomic strata of
40 per thousand by 1990.1,2 A further 24% reduction from the Thai population using data from the 1990 and 2000
1990 to 20002 puts Thailand well on track to achieve the population censuses obtained from the National Statistics
Millennium Development Goal (MDG) of a two-thirds Office, Thailand.
reduction in U5MR between 1990 and 2015.3 This is in
stark contrast to many other countries, particularly those Methods
in sub-Saharan Africa and southern Asia where varied Population data
progress towards this target has led to much doubt as to The 1990 and 2000 Thai population censuses were full
whether the goal can be achieved.4−6 enumerations of the Thai population. For both censuses,
As the MDGs focus on average levels, one unanswered a random 20% sample of households completed a more
question is whether the success in reducing the average detailed questionnaire on household characteristics and
U5MR has been accompanied by a reduction in the the analysis conducted here was limited to this sample
disparity in child health between subgroups of the Thai (table 1).
population. This notion is in line with global calls for
routine monitoring of equity in health outcomes.7−9 Many Sample size
other countries provide examples of increasing life 1990
expectancy and decreasing child mortality while Number of mothers aged 15−49 years 1 996 409
inequalities between the rich and the poor remain10 or Number of households 2 432 276
worsen.11 Although Thailand has experienced substantial 2000
economic growth over the past two decades (with some Number of mothers aged 15−49 years 2 362 523
interruption due to the 1997 economic crisis), income Number of households 3 066 316
inequality persists at a relatively high level.12 A critical
Table 1: Sample sizes for the 20% sample of the 1990 and 2000 censuses
question is how this has affected the distribution of child
www.thelancet.com Vol 369 March 10, 2007

The data from the Thai census have been used in other
studies;2,13 completeness is very high and quality is
regarded to be good.14 The housing and population
censuses can be linked through the use of several
identifying variables. There was a problem with duplicate
identifiers in the 2000 census, which affected fewer than
1% of observations in most provinces, except for three
provinces (Chachoengsao, Phattalung, and Sakeao) with
rates of duplication exceeding 10%. While some of these
observations were simply duplicate data, which could be
dropped without loss of information, a large proportion
held unique information that was incorrectly coded.
Observations that could not be uniquely identified were
excluded from the analysis.
Procedures
We measured economic status using a method developed
by Ferguson and colleagues.15 As economic status is a
latent variable that cannot be directly observed, we
estimated it using information about predictors of
economic status—such as age and education of the
household head—and indicators of economic status,
primarily consumer goods (such as televisions, cars, and
radios) as well as household services (such as source of
drinking water or type of toilet). Only indicators that
behaved like normal goods—ie, probability of ownership
increased with economic status—were included;
normality was assessed using education level as a proxy.
The 1990 and 2000 censuses included identical predictor
variables, but the indicators varied between the two
surveys. The exact list of included predictors and
indicators is available in the webpanel.
Statistical analysis
A detailed discussion of the statistical methods applied to
estimate economic status can be found elsewhere15 and its
application seen in other studies.16−18 Briefly, a hierarchical
probit model, with a Bayesian adjustment, was used to
calculate a continuous measure of economic status for
each household. Linear regression of the 1990 and 2000
model outputs was used to place the economic status
estimates from both censuses on the same scale. We
assessed agreement between the average economic status
rank at the provincial level (73 provinces in 1990 and 76 in
2000) and gross provincial product (GPP) rank, available
from the Thai National Economic and Social Development
Board,19 using Spearman’s rank correlation.
In both censuses, data on the self-reported number of
children ever born and children surviving were used to
estimate U5MR, applying the Trussell version of the
Brass indirect method.20 The Coale and Demeny “West
family” regional model life tables21 were used to convert
probabilities of dying to mortality levels. U5MR estimates
were derived from women aged 30−34 years, the age-
group believed to give the most robust estimate.20,22
Standard errors in U5MR were estimated using non-
parametric bootstrap sampling (1000 samples).
www.thelancet.com Vol 369 March 10, 2007
Articles
1990 census
0·08
0·06
0·04
0·02
Fraction of households
0
0 2 4 6
2000 census
0·08
0·06
0·04
0·02
0
0 2 4 6
Economic status
Figure 1: Distribution of household economic status, 1990 and 2000
Households were ranked according to economic status
and grouped into strata. The Brass technique was applied
to estimate child mortality in each stratum to determine
the distribution of U5MR by economic status strata.
Concentration curves and concentration indices were
estimated by percentiles.23−25 Analyses were undertaken
with STATA 9.2 (Stata Corporation, Texas USA) and
Microsoft Excel (Microsoft Corporation).
Role of the funding source
The funding sources had no role in the study design, See Online for webpanel
data collection, data analysis, data interpretation, or
writing of this report. The corresponding author had full
access to all data in the study and had final responsibility
for the decision to submit for publication.
Results
Economic status at the provincial level was highly
correlated with gross provincial product; Spearman’s
rank correlation coefficient was 0∙87 (p<0∙0001) in 1990
and 0∙78 (p<0∙0001) in 2000. Mean economic status in
Thailand increased from 1990 to 2000 and was
accompanied by a narrowing of the distribution at the
household level (coefficient of variation 0∙40 in 1990 to
0∙29 in 2000), as shown in figure 1.
Average U5MR was 27∙4 (95% CI 26∙8 to 28∙0) per
1000 livebirths in the 1990 census and 18∙7 (95% CI
18∙2 to 19∙3) per 1000 in the 2000 census, a reduction in
deaths in children younger than 5 years of 32% (95% CI
29% to 34%). U5MR estimates from the 1990 census
corresponded to an estimated reference year of 1984,
whereas estimates from the 2000 census corresponded to
an estimated reference year of 1994.
In the 1990 census the U5MR in the poorest quintile
was triple that seen in the rich and dropped steadily with
increasing economic status (table 2). Over the 10 year
851
 Articles

1990 census 2000 census 1990–2000

Number of births U5MR (95% CI) Number of births U5MR (95% CI) Percentage decrease (95%CI)
Quintile 1 (poorest) 193 735 40·8 (39·5–42·2) 174 226 23·0 (22·0–24·0) 44% (40–47)
Quintile 2 162 895 32·3 (30·8–33·7) 142 740 19·2 (18·1–20·2) 41% (36–45)
Quintile 3 153 870 24·9 (23·5–26·2) 124 648 19·3 (17·9–20·7) 22% (15–30)
Quintile 4 149 675 19·1 (17·8–20·4) 120 879 14·7 (13·5–15·9) 23% (14–31)
Quintile 5 (richest) 147 334 14·8 (13·5–16·2) 96 942 12·9 (11·5–14·4) 13% (0–25)
Rate ratio (Q1/Q5) .. 2·8 (2·5–3·0) .. 1·8 (1·6–2·0) 55%* (39–68)
Absolute difference (Q1–Q5) .. 26·0 (24·1–27·9) .. 10·1 (8·3–11·9) 61% (54–68)

*Percentage change in excess risk (rate ratio –1).

Table 2: U5MR per 1000 population, by quintiles of household economic status

significant progress, a clear gradient across quintiles still

60 existed in 2000 with the richest quintile having a U5MR
around two-thirds the mean level of the bottom four
50
quintiles. There is also a clear gradient in fertility across
quintiles in both years.
40
Figure 2 plots the mean score for 100 strata of
U5MR (per 1000)

30 economic status against observed U5MR within those

strata for 1990 and 2000. As expected, this analysis
20 reveals larger inequalities than the quintile analysis and
shows that the reduction in child-mortality inequality
10 1990 was accompanied by a positive shift in mean household
2000 economic status as well as a weakening of the relation
0 between relatively low economic status and high child
0 0·5 1 1·5 2 2·5 3 3·5 4 4·5
mortality.
Economic status score
Concentration curves (figure 3) reinforce the finding
Figure 2: U5MR by percentiles of household economic status, 1990 and 2000 census of shrinking U5MR inequality between the two
censuses. The concentration index in 1990 was −0∙20
(95% CI −0∙23 to −0∙18), whereas in 2000 it had
1·0 1990 census dropped to −0∙12 (−0∙15 to −0∙08), a 43% (22% to 63%)
2000 census reduction.
Equality
Cumulative % of deaths under 5 years

0·8
0·6
0·4
0·2
0
0 0·2
Cumulative % of births, ranked by economic status
Figure 3: Concentration curves of the U5MR, 1990 and 2000 census
period, substantially larger reductions in U5MR occurred
in the poorer quintiles, such that the poorer three
quintiles had similar rates in the 2000 census and the
excess child-mortality risk between the poorest and
richest decreased by more than half. This decrease in the
poor−rich gap over the 10 year period was evident from
both relative and absolute measures. Despite the
Discussion
Between 1990 and 2000, in addition to successfully
reducing the average level of under-five mortality by
about 30%, Thailand approximately halved inequality
between the poorest and the richest populations. This
remarkable reduction in child-mortality inequality
across economic strata is shown by all three measures
used in this analysis—the rate ratio, the absolute
difference, and the concentration index. The 55%
reduction in the excess child-mortality risk between the
0·4 0·6 0·8 1·0 poorest and richest quintiles compares favourably with
reductions documented in other countries, such as
Turkey (1993−98; 55% reduction in excess child-
mortality risk between poorest and richest quintile),
Egypt (1995−2000; 31%), Tanzania (1996−99; 59%),
Guatemala (1995−98; 28%), and Colombia (1995−2000;
22%).8 What has contributed to the substantial reduction
in child-mortality inequality in Thailand?
The positive shift in household economic status that
we observed between 1990 and 2000 seems to be a major
contributing factor. Other sources confirm substantial
economic growth over this period and show that the

852 www.thelancet.com Vol 369 March 10, 2007


proportion of the population living in poverty decreased
substantially.26−28 World Bank estimates of the proportion
of the population living on less than US$1 a day decreased
from 18% in 1988 to 6% in 1992 to below 2% in 2000.29
Gross domestic product per capita in current international
dollars (purchasing power parity adjusted) grew at an
average rate of 3% between 1990 and 2000, rising to
US$6402.30
Improvement in household economic status, however,
is just one of a range of inter-related factors contributing
to the closing mortality gap between socioeconomic
groups. The Thai census does not allow for direct
measurement of other explanatory factors, but existing
evidence suggests potential key determinants.
A range of pro-poor health insurance schemes
implemented in Thailand since the 1970s has improved
health service coverage for children. In 1975, user charges
were waived for low-income families through the
government medical welfare scheme.31 In 1983, a
subsidised voluntary health-insurance scheme targeting
mostly near-poor rural families and covering basic
treatment, along with mother and child health services,
was piloted and subsequently expanded.32,33 In 1993, the
government medical welfare scheme was extended to
include all children younger than 12 years, the elderly,
and the disabled.31 It was not until 2001, however, that the
entire adult population was insured through the universal
coverage scheme.
In addition to improved financial accessibility for the
more vulnerable segments of the population, health
infrastructure and services were scaled up from the late
1970s, particularly primary health care targeting the
poorer, rural population.34,35 This scale-up was informed
by pilot projects in the 1960s36,37 and was funded in part by
a reallocation of resources away from urban hospital
development.34 By 1989, there was at least one primary
health-care centre for each rural village35,36 with services
focused on basic medical care, health education, family
planning, maternal and child health, sanitation, local
infectious disease control, and provision of essential
drugs.36,37 The number of community hospitals doubled
between the early 1980s and mid-1990s, improving the
links between primary health care and secondary referral
hospitals.35,38 Increased production, financial incentives,
and educational strategies led to a more equitable
allocation of medical doctors in rural areas in the 1980s,
although this was partly reversed by an internal brain
drain towards urban private hospitals in the early to mid-
1990s.34
The combination of improved capacity and financial
accessibility led to increased health-care use and
intervention coverage. There was a quadrupling in the
number of outpatient visits to health centres and
community health posts,38 skilled birth attendance rose
from 66% in 198739 to 95% in 1999,40 and vaccination
coverage (diphtheria-pertussis-tetanus triple vaccine,
measles vaccine, and oral poliomyelitis vaccine in
www.thelancet.com Vol 369 March 10, 2007
Articles
children younger than 1 year, and tetanus toxoid vaccine
in pregnant women) increased from between 20% and
40% in the early 1980s to more than 90% in the 1990s.38
Regular growth monitoring by health-centre workers has
been undertaken for children younger than 5 years in
each village since 1981.36 The total fertility rate dropped
from 2∙41 in 1985−90, to 2∙10 in 1990−95, to 1∙95 in
1995−2000,41 largely due to wide-scale availability and use
of family planning.42
The process, context, and actors that have shaped these
reforms have been covered in detail elsewhere.34,43,44
Changes were not the result of an overall strategy but
more due to pressure from different sources that resulted
in multiple incremental changes.44 One of the more
influential groups was the Rural Doctors’ Society, a group
of medical doctors with a strong focus on equity, which
played an active role in not only rural-health development
but also a range of other health issues.34,44
Despite the substantial reductions in child-mortality
inequality, it is important to recognise that gaps still exist
between economic strata of the Thai population and that
the absolute level of child mortality is still high in the
very poor (around 40 per 1000 in the poorest 1%). In
specific subgroups, such as hill tribe or migrant
populations, these rates may be even higher. Additionally,
the effect of the economic crisis in the late 1990s and the
Thai government’s introduction in 2001 of a universal
health-care coverage scheme45 have not been captured in
this analysis. These points emphasise the need for
ongoing, timely, and consistent measurement of
inequalities to assess the effect of health policies. As
shown here, routine information systems can be used to
allow more sensitive estimation of health disparities than
the usual quintile aggregation.8,46,47 A remaining challenge
for Thailand is to incorporate a way of monitoring adult
health inequalities.
Several limitations need to be taken into account when
interpreting the results. First, duplicate identifiers in the
2000 census required the exclusion of mothers who could
not be uniquely matched with their households. The
proportion of duplicate records, however, was very small.
Second, another analysis2 suggests that the 1990 census
underestimated child mortality relative to other sources
in Thailand. This could lead to inaccurate estimation of
child-mortality inequality, especially if the under-
estimation in the 1990 census, or possible recall bias in
either census, is differential across economic strata.
Third, whereas the estimates of standard error derived
from the non-parametric bootstrap method take into
account sampling error, they do not capture modelling
error due to the Brass indirect technique. To our
knowledge, no method has, as yet, been developed to
quantify this error and given the widespread use of the
technique, this is a clear priority for future research. A
final limitation is time inconsistency. The U5MR
estimates are derived from periods before the time of the
census, whereas economic status is from the year of the
853
 Articles
census. The long-term nature of the economic status
indicator helps to avoid this problem, but given the Thai
experience of rapid economic growth and the 1997 crisis,28
the economic-status measurement in 2000 might not
accurately reflect income in the previous decade. Other
data suggest, however, that there was little movement
between socioeconomic classes over this period.48
At the halfway mark of the MDG for achieving a two-
thirds reduction in child mortality, many countries are
unlikely to reach this target. In Thailand, there is an
encouraging trend that suggests this MDG will be
achieved and that the successful average improvement in
child survival has been accompanied by a substantial
decrease in economic inequality in U5MR. Our findings
clearly draw attention to the importance of including an
equity dimension in health-outcome monitoring and
show how this can be done in a data-limited setting using
routine, population-based data sources. Although we
have provided insight into possible contributing factors,
further research will allow a better understanding of how
Thailand has achieved this remarkable success. This will
help inform other countries to attain the MDG target and
reduce health inequalities in parallel.
Contributors
P Vapattanawong did exploratory research, compiled the data, and
undertook the analysis. M C Hogan did the analysis, drafted the initial
manuscript, and coordinated the research. P Hanvoravongchai
contributed to the analysis and policy discussions. E Gakidou
contributed to the measurement of economic status and interpretation
of inequalities. T Vos contributed to planning and interpretation of
results. A D Lopez generated the initial idea for the study and
contributed to methods and conclusions. S S Lim contributed to the
analysis and drafting of the initial manuscript and coordinated the
research. Subsequent revisions were undertaken by M C Hogan and
S S Lim with input from all authors.
Conflict of interest statement
We declare that we have no conflict of interest.
Acknowledgments
We thank Ajay Tandon for input and advice on measurement of
economic status; Kenneth Hill for technical advice on the measurement
of under-five mortality; Inez Mikkelsen-Lopez for research assistance;
and Jesse Abbott-Klafter for assistance with non-parametric bootstrap
sampling. This research was conducted under the Setting Priorities
using Information on Cost-Effectiveness (SPICE) project and supported
by an international collaborative research grant from the Wellcome
Trust, UK (071842/Z/03/Z) and the National Health and Medical
Research Council of Australia (301199). M C Hogan and E Gakidou
received research support from the National Institute of Aging (PO1 AG
17625-01). E Gakidou was also supported by the Grand Challenges in
For more on the Grand
Global Health Initiative (GC13).
Challenges in Global Health
Initiative see http://www. References
grandchallengesgh.org 1 Ahmad OB, Lopez AD, Inoue M. The decline in child mortality: a
reappraisal. Bull World Health Organ 2000; 78: 1175–91.
2 Hill K, Vapattanawong P, Prasartkul P, Porapakkham Y, Lim SS,
Lopez AD. Epidemiologic transition interrupted: a reassessment of
mortality trends in Thailand, 1980–2000. Int J Epidemiol 2006;
published online Dec 20. DOI:10.1093/ije/dy1257.
3 United Nations General Assembly. United Nations Millenium
Declaration, 2000.
4 Evans DB, Adam T, Tan-Torres Edejer T, et al. Time to reassess
strategies for improving health in developing countries. BMJ 2005;
331: 1133–36.
5 World Health Organization. Health and the millennium
development goals. Geneva: WHO, 2005.
854
6 United Nations. The Millennium Development Goals report 2005.
New York: UN, 2005.
7 Gwatkin DR. How much would poor people gain from faster
progress towards the Millennium Development Goals for health?
Lancet 2005; 365: 813–17.
8 Moser KA, Leon DA, Gwatkin DR. 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: 1180–82.
9 Bellagio Study Group on Child Survival. Knowledge into action for
child survival. Lancet 2003; 362: 323–27.
10 World Health Organization. The world health report 2000—Health
systems: improving performance. Geneva, Switzerland, 2000.
11 Wang L. Determinants of child mortality in LDCs: empirical
findings from demographic and health surveys. Health Policy 2003;
65: 277–99.
12 Balisacan AM, Ducanes GM. Inequality in Asia: a synthesis of
recent research on the levels, trends, effects and determinants of
inequality in its different dimensions. London: The Inter-Regional
Inequality Facility, 2006.
13 Bohning D, Ayuthya RS. Analysis of geographical heterogeneity in
live-birth ratio in Thailand. J Epidemiol Biostat 1999; 4: 115–22.
14 Suwee W, Santipaporn S. Utilization of the 2000 population and
housing census of Thailand. 20th Population Census Conference
2002, Ulaanbaatar, Mongolia.
15 Ferguson BD, Tandon A, Gakidou EE, Murray CJL. Estimating
permanent income using indicator variables. In: Murray CJL,
Evans DB, eds. Health systems performance assessment: debates,
methods and empiricism. Geneva: WHO, 2003.
16 Hosseinpoor AR, Mohammad K, Majdzadeh R, et al.
Socioeconomic inequality in infant mortality in Iran and across its
provinces. Bull World Health Organ 2005; 83: 837–44.
17 Pongou R, Salomon J, Ezzati M. Health impacts of macroeconomic
crises and policies: determinants of variation in childhood
malnutrition trends in Cameroon. Int J Epidemiol 2006; 35: 648–56.
18 Gakidou E, Lozano R, González-Pier E, et al. Assessing the effect of
the 2001–06 Mexican health reform: an interim report card. Lancet
2006; 368: 1920–35.
19 Thai National Economic and Development Board. Gross provincial
product, 2004 edn. http://www.nesdb.go.th/econSocial/macro/gpp_
data/index.html (accessed Feb 7, 2006).
20 United Nations. Manual X: indirect techniques for demographic
estimation. New York: UN, 1983.
21 Coale AJ, Demeny P. Regional model life tables and stable
populations, 2nd edn. Princeton, New Jersey: Princeton University
Press, 1983.
22 Hill K, Pande R, Mahy M, Jones G. Chapter two: data sources and
estimation methods—trends in child mortality in the developing
world, 1960–96: UNICEF.
23 World Bank. Concentration curves. http://siteresources.worldbank.
org/INTPAH/ Resources/Publications/Quantitative-Techniques/
health_eq_tn06.pdf (accessed May 10, 2006).
24 World Bank. Spreadsheet for computing concentration index.
http://siteresources. worldbank.org/INTPAH/Resources/
Publications/Quantitative-Techniques/ concentration_index.xls
(accessed May 10, 2006).
25 World Bank. The concentration index. http://siteresources.
worldbank.org/ INTPAH/Resources/Publications/Quantitative-
Techniques/health_eq_tn07.pdf (accessed May 10, 2006).
26 Supakankunti S. Future prospects of voluntary health insurance in
Thailand. Health Policy Plan 2000; 15: 85–94.
27 Noree T, Chokchaichan H, Mongkolporn V. Thailand’s Country
paper—abundant for the few, shortage for the majority: the
inequitable distribution of doctors in Thailand. Nonthaburi,
Thailand: International Health Policy Program & Asia Pacific
Action Alliance on Human Resources for Health (AAAH), 2005.
28 Tangcharoensathien V, Harnvoravongchai P, Pitayarangsarit S,
Kasemsup V. Health impacts of rapid economic changes in
Thailand. Soc Sci Med 2000; 51: 789–807.
29 World Bank. PovcalNet. World Bank. http://iresearch.worldbank.
org/ PovcalNet/jsp/index.jsp (accessed July 14, 2006).
30 United Nations Development Program. Human development
report. New York: Oxford University Press, 2002.
www.thelancet.com Vol 369 March 10, 2007

31 Pannarunothai S. Medical welfare scheme: financing and targeting
the poor. In: Pramualratana P, Wibulpolprasert S, eds. Health
insurance systems in Thailand. Nonthaburi, Thailand: Health
Systems Research Institute, 2002.
32 Srithamrongsawat S. The health card scheme: a subsidized
voluntary health insurance scheme. In: Pramualratana P,
Wibulpolprasert S, eds. Health insurance systems in Thailand.
Nonthaburi, Thailand: Health Systems Research Institute, 2002.
33 Pannarunothai S, Srithamrongsawat S, Kongpan M, Thumvanna P.
Financing reforms for the Thai health card scheme.
Health Policy Plan 2000; 15: 303–11.
34 Wibulpolprasert S, Pengpaibon P. Integrated strategies to tackle the
inequitable distribution of doctors in Thailand: four decades of
experience. Hum Resour Health 2003; 1: 12.
35 Kachondham Y, Chunharas S. At the crossroads: challenges for
Thailand’s health development. Health Policy Plan 1993; 8: 208–16.
36 Wibulpolprasert S. Community financing: Thailand’s experience.
Health Policy Plan 1991; 6: 354–60.
37 Nitayarumphong S. Evolution of primary health care in Thailand:
what policies worked? Health Policy Plan 1990; 5: 246–54.
38 Wibulpolprasert S, ed. Thailand health profile 2001–04. Nonthaburi,
Thailand: Bureau of Policy and Strategy, Ministry of Public Health,
2005.
39 Chayovan N, Kamnuansilpa P, Knodel J. Thailand—demographic
and health survey 1987: final report. Columbia, MD: Institute for
Resource Development/ Westinghouse, Institute of Population
Studies, Chulalongkorn University, Bangkok, Thailand, 1987.
40 World Health Organization, SEARO. Country health profile:
Thailand. http://searo.who.int/EN/Section313/Section1525_6897.
htm (accessed July 17, 2006).
www.thelancet.com Vol 369 March 10, 2007
Articles
41 United Nations Population Division. World population prospects:
the 2004 revision population database. http://esa.un.org/unpp/
p2k0data.asp (accessed July 20, 2006).
42 United Nations Department of Economic and Social Affairs,
Population Division. World Contraceptive Use. http://www.un.org/
esa/population/publications/ contraceptive2003/About_WCU2003.
htm (accessed July 20, 2006).
43 Tangcharoensathien V, Wibulpolprasert S, Nitayaramphong S.
Knowledge-based changes to the Thai health system.
Bull World Health Organ 2004; 82: 750–56.
44 Green A. Reforming the health sector in Thailand: the role of policy
actors on the policy stage. Int J Health Plann Manage 2000; 15:
39–59.
45 Suraratdecha C, Saithanu S, Tangcharoensathien V. Is universal
coverage a solution for disparities in health care? Findings from
three low-income provinces of Thailand. Health Policy 2005; 73:
272–84.
46 Schellenberg JA, Victora CG, Mushi A, et al. Inequities among the
very poor: health care for children in rural southern Tanzania.
Lancet 2003; 361: 561–66.
47 Wagstaff A. Socioeconomic inequalities in child mortality:
comparisons across nine developing countries.
Bull World Health Organ 2000; 78: 19–29.
48 Fofack H, Zuefack A. Dynamics of income inequality in Thailand:
evidence from household pseudo-panel data. Washington DC: The
World Bank, 1999.
855