European Journal of Public Health, Vol. 20, No. 6, 640–646
ß The Author 2009. Published by Oxford University Press on behalf of the European Public Health Association. All rights reserved.
doi:10.1093/eurpub/ckp200 Advance Access published on 15 December 2009
................................................................................................
Educational health inequalities in former
Yugoslavia: evidence from the South-East
European Social Survey Project
Terje Andreas Eikemo1,2, Martijn Huisman3, Francesca Perlman4,
Kristen Ringdal5
Background: An important gap in our knowledge of social inequalities in health is the former
Yugoslavia, a region of culturally and historically diverse countries, with recent conflict. The aim of
the present paper is to investigate relative and absolute inequalities in self-assessed health in former
Yugoslavia (Bosnia-Herzegovina, Croatia, Kosovo, Macedonia, Montenegro, Slovenia and Serbia) by sex
and education. Methods: The data source is the South-East European Social Survey Project fielded in
December 2003 to Winter 2004, covering the former Yugoslavia with a total sample of 18 481 respondents. Data from Slovenia were obtained from the 2004-wave of the European Social Survey. The health
outcome variables were self-reported general health (SRH) and limiting longstanding illness (LLI).
Results: Both absolute and relative educational health inequalities were present throughout the
former Yugoslavia to a larger or lesser extent, although odds ratios (ORs) for LLI and SRH were not
significant for Montenegrin women [LLI OR = 1.12, 95% confidence interval (CI): 0.92–1.37; SRH
OR = 1.16, 95% CI: 0.96–1.40] and with respect to the reporting of LLI among Slovenian men
(OR = 1.16, 95% CI: 0.96–1.44). Overall, Montenegro held the best position. Conclusions: The
prevalence of poor health and the degree of relative inequality in self-assessed health in the former
Yugoslavian countries were similar in order to one another, and to other East European countries
during the same period. Influences on subjective health require further elucidation. Further research
should study a wider range of health outcomes using larger survey samples and a wider range of
cultural and other predictor variables.
Keywords: education, former Yugoslavia, health inequalities, morbidity
................................................................................................
Introduction
ocial inequalities in health continue to be a key public
problem in advanced countries, including
European countries.1 Not only are socio-economic inequalities
in morbidity and mortality reported in many European
countries,2 they are in fact found to be substantial in all
countries with available data.3 However, we do not yet have
the full picture, as many countries remain unexplored,
even within Europe. Countries for which data have been
lacking until now are the countries that were part of the
former Yugoslavia (Bosnia-Herzegovina, Croatia, Kosovo,
Macedonia, Montenegro, Slovenia and Serbia).
Socio-economic health inequalities are almost universal4,5
and, considering also the former Yugoslavia’s history of
cultural diversity, violent conflicts and rapid economic,
political and social change,4–6 we would expect that this
will also be the case in former Yugoslavian countries.
Shealth
1 Department of Public Health, Erasmus MC, University Medical
Centre, Rotterdam, The Netherlands
2 SINTEF Technology and Society, Department of Health Services
Research, Trondheim, Norway
3 Interdisciplinary Center for Psychiatric Epidemiology, University
Medical Center Groningen, University of Groningen, Groningen,
The Netherlands
4 London School of Hygiene and Tropical Medicine, London, UK
5 Norwegian University of Science and Technology (NTNU),
Department of Sociology and Political Science, Trondheim,
Norway
Correspondence: Terje Andreas Eikemo, PO Box 2040, 3000 CA
Rotterdam, The Netherlands, tel: +47-99034077, Fax: +47 93270800,
e-mail: [email protected]
These countries have experienced several transitions: from a
communist to a more democratic political system, from a
planned to a market economy. Several of these countries
have also experienced war as a result of their cession from
Yugoslavia in the early 1990s, and more recently Serbia itself
was bombed by NATO as a result of the ongoing conflict in
Kosovo. These transitions and conflicts undoubtedly brought
very stressful experiences with health consequences for a large
number of people.6–8 Moreover, times of war have many other
destabilizing influences on daily life. In addition, economic
experiences have varied, with Croatia, Slovenia and Serbia
amongst the more wealthy countries in the Western Balkans,
and Kosovo, Montenegro, Macedonia and Bosnia-Herzegovina
among the poorer.
At the same time, predicting the magnitude of inequalities
in self-assessed health in these countries is difficult. Previous
experiences with cross-national comparisons of inequalities in
self-assessed health have shown that there can already be substantial differences between countries that are more homogeneous in terms of economic, political and religious history
than the Yugoslavian countries,9 demonstrating that the
patterns of cross-national differences in inequalities in health
are highly complex.
To fill the gap in the current knowledge base of
socio-economic inequalities in health in Europe, the present
study aimed to investigate (relative and absolute) self-assessed
health inequalities in former Yugoslavia by educational
attainment.
Methods
The data source is the South-East European Social Survey
Project (SEESSP) based on face-to-face interviews fielded in
December 2003 to Winter 2004. The surveys covered former
Health inequalities in former Yugoslavia
Yugoslavia with the exception of Slovenia with a total sample
of 18 481 respondents (data from Slovenia were obtained from
the 2004-wave of the European Social Survey, asking the same
health questions as in the SEESSP). The SEESSP surveys
measured socio-demographic and attitudinal variables related
to a large number of subjects, from attitudes towards ethnic
relations to questions about gender roles, and two questions
about subjective health. The field organizations that were
chosen were the most experienced and most used by international organizations and the SEESSP is the only available
survey covering this region. Since no address registers were
available, we applied a multistage cluster sampling design for
each country aiming to obtain self-weighting representative
samples. In addition, large ethnic minorities were
overrepresented and some smaller minorities were covered in
special samples. The former were down-weighted and the latter
excluded from the representative samples. In each country
400–500 random geographical starting points were chosen
from municipality maps (stratified by urban and rural areas),
and clusters of 4–15 (20–25 in Kosovo) respondents were
interviewed for each starting point. Within households, the
respondent was selected from household members aged 18
and above, using the ‘nearest birthday’ method. Respondents
were asked for their verbal consent to be interviewed and were
told that participation was voluntary. The overall refusal rate
was reported to be around 30% for all the countries, although
exact refusal rates were not registered. In Bosnia-Herzegovina,
the basic representative sample was actually three separate
samples from different sampling frames: one from the predominantly Croat municipalities, one from the predominantly
Bosniak municipalities of the Federation and one from
municipalities in Republic Srpska. The selection of separate
samples from different parts of Bosnia-Herzegovina was
considered the best way to ensure the robustness of comparisons within the country and to obtain the best possible representative samples of the major ethnic groups. Within the
predominantly Bosniak, Croat and Serb municipalities, all of
the large and medium sized municipalities were included, and
a random sample of the smallest municipalities was included,
for a total of 114 municipalities (including Brcko) from
all cantons.7 This sample of municipalities was agreed upon
by two of the largest survey sampling organizations
in Bosnia-Herzegovina, Mareco Index Bosnia and PULS.
Within the selected municipalities, samples of households
were selected proportional to estimated municipality size, so
the unweighted sample is representative of the distribution of
persons across municipalities within each of the three major
sample regions. This principle of proportional selection was
also applied in the other countries.
In Macedonia and Kosovo, there is a very high degree of
ethnic residential segregation. For this reason, it was possible
to select stratified, multi-stage cluster samples of the areas
where the minorities reside, and thus obtain approximately
representative samples of adults of the largest within-country
minority. Including these samples in country-wide analyses
improves the standard errors of estimates of statistics for
ethnic Albanians in Macedonia and for ethnic Serbs in Kosovo.
In countries where recent censuses had been carried out,
that is, Croatia, Serbia and Macedonia, adjustments weights
by post-stratification by sex and age were applied to further
reduce sampling bias. In Bosnia-Herzegovina, no census has
been carried out since before the war, and there are no reliable
population statistics as basis for post-stratification weights. In
Montenegro, the results from the 2001 census were published
too late for post-stratification weights to be computed before
the data were finalized.
The overall sampling design is more thoroughly described
elsewhere10 and the sampling designs for Bosnia-Herzegovina
641
and Kosovo are also extensively described in two different
articles.7,8
The health outcome variables were self-reported general
health (SRH) and limiting longstanding illness (LLI). SRH is
a valuable indicator of people’s health status, since it predicts
mortality consistently in many countries, with worsening
subjective health associated with progressively higher
mortality, chronic disease and behavioural risk factors.11–13 It
is also a stable measure, with good test–retest reliability, and
consistent reporting.12 SRH was measured by a question asking
‘How is your health in general?’. Eligible responses were ‘very
good’, ‘good’, ‘fair’, ‘bad’ and ‘very bad’. We dichotomized
the variable into ‘very good or good’ health versus ‘less
than good’ health (‘fair’, ‘bad’ and ‘very bad’). As for LLI,
people were asked if they were hampered in daily activities
in any way by any longstanding illness or disability, infirmity
or mental health problem. Eligible responses were ‘yes a lot’,
‘yes to some extent’ and ‘no’. We dichotomized this variable
into ‘yes’ (regardless of whether to some extent or a lot)
and ‘no’.
Education is a widely used indicator of socio-economic
position within the social sciences. It is well adapted to our
study because it avoids interpretation problems with respect to
health-related social mobility later in life and to the widespread
social mobility that followed the end of communism. The
association between socio-economic position and poor
health is well established and education has additional
specific influences through increasing knowledge and skills
that may affect cognitive function, make individuals more
receptive to health education messages or more able to communicate with and access health services.14
However, one limitation of level of education is that educational systems may vary considerably between countries, which
hamper interpretation of observed patterns of educational
inequalities across countries. Therefore, our measure of
education was based on a variable describing full-time
education in years (see country statistics, table 1). In comparative studies, it is important to take into account the extent of
variations of reported years of education in different countries.
We did this by applying a total impact measure of education.
First, for each country separately, we standardized the
continuous variables of educational attainment, such that the
national average was equal to 0 and the standard deviation
equal to 1 year of education. Second, we multiplied this
variable by 1, so that higher values corresponded with
lower educational levels. Next, the standardized variable was
introduced as an independent variable in a logistic regression
analysis, controlled for age, with health variables as the
dependent variable. Finally, odds ratios (ORs) were
computed as the antilogarithm of the estimated logistic
regression coefficients. The ORs represent our relative health
inequality measure and should be interpreted as the health
difference between people with average years of education
and those whose number of year of education is one
standard deviation below the national average.
Presentations of relative health inequalities may be
misleading if absolute measures are not presented additionally.15,16 Both relative and absolute inequality measures are
meaningful for monitoring health inequalities, provided that
they are interpreted while taking into account the overall levels
of the health outcomes.3 Thus, absolute differences in health
were calculated as the age-adjusted rate difference (RD)
between the higher and lower education group using the
median of the total impact measure for men and women
separately within each country as cut-off point. The median,
minimum and maximum values of the total impact measure
are reported in table 1.
642
European Journal of Public Health
Results
The prevalence in men of LLI (table 2) varied between 23%
(Macedonia and Montenegro) and 33% (Slovenia), and of fair
or worse health (table 3) between 30% and 41 (Slovenia and
Kosovo). These figures for women were 30% and 34% and
40% and 51%.
Tables 2 and 3 further demonstrate that health inequalities
were present in men and women in all countries, although
ORs for LLI and SRH were not significant for Montenegrin
women (LLI OR = 1.12, 95% CI: 0.92–1.37; SRH OR = 1.16,
95% CI: 0.96–1.40) and with respect to the reporting of LLI
among Slovenian men (OR = 1.16, 95% CI: 0.96–1.44).
The pattern of inequalities in LLI and SRH across
the countries is illustrated in figure 1. ORs and rate differences
in LLI seemed to be smaller than elsewhere in
Bosnia-Herzegovina, Slovenia (in particular among men) and
Montenegro (in particular among women).
The results of absolute (RDs) and relative (ORs) inequalities
in SRH showed the following. Among men, smallest ORs were
observed in Montenegro and Kosovo, while Slovenia and
Croatia exhibited the largest. Smallest RDs were found in
Table 1 Country statistics (N = 17 359)a
Country/region
Bosnia-H
Croatia
Kosovo
Macedonia
Montenegro
Serbia
Slovenia
Sample
size (N)b
Average years of education
(standard deviation)
Median (minimum and maximum)
values of total impact measurec
Men
Men
Women
Men
2519
895
1027
1031
773
1522
514
2627
1221
1167
1169
705
1566
623
11.83
11.70
10.95
10.95
11.54
10.93
11.64
Women
(2.88)
(3.94)
(3.50)
(3.64)
(3.38)
(3.83)
(3.49)
10.73
10.82
8.14
9.37
10.87
10.39
10.88
(3.63)
(4.23)
(4.21)
(3.86)
(3.34)
(4.36)
(3.57)
Women
0.06
0.08
0.30
0.29
0.14
0.02
0.18
(
(
(
(
(
(
(
4.22–4.11)
5.16–2.97)
2.87–3.13)
3.58–3.01)
3.68–3.12)
3.41–2.85)
3.26–3.05)
0.35
0.28
0.03
0.35
0.34
0.37
0.03
(
(
(
(
(
(
(
3.10–2.95)
3.35–2.56)
3.29–1.93)
3.79–2.43)
2.73–3.25)
3.12–2.38)
3.40–2.21)
a: All people are aged 25 or more
b: After listwise deletion
c: Standard deviation = 1, mean = 0
Table 2 ORs (95% CI), prevalence rates and rate differences (95% CI) of limiting longstanding illness according to education in
former Yugoslavia (N = 17 359)a
Country/region
Bosnia-H
Croatia
Kosovo
Macedonia
Montenegro
Serbia
Slovenia
Men
Women
Prev.
RD (95% CI)
OR (95% CI)
Prev.
RD (95% CI)
OR (95% CI)
29.3
28.3
25.2
23.0
23.9
25.8
33.2
7.6
17.6
10.1
12.2
6.9
13.7
0.4
1.34
1.63
1.45
1.65
1.31
1.54
1.18
31.7
28.9
37.1
30.3
32.3
33.2
34.2
10.3
13.5
19.9
18.2
5.4
11.6
10.4
1.37
1.66
1.54
1.65
1.12
1.63
1.33
(7.4–7.9)
(16.4–8.7)
(9.5–10.7)
(11.5–12.9)
(6.5–7.4)
(13.1–14.4)
(0.4–0.5)
(1.21–1.47)
(1.38–1.93)
(1.24–1.70)
(1.41–1.93)
(1.10–1.57)
(1.36–1.75)
(0.96–1.44)
(9.9–10.7)
(12.7–14.2)
(18.8–21.1)
(17.2–19.3)
(5.0–5.8)
(11.1–12.2)
(9.6–11.2)
(1.24–1.52)
(1.41–1.95)
(1.33–1.79)
(1.43–1.91)
(0.92–1.37)
(1.43–1.87)
(1.08–1.65)
a: All people are aged 25 or more. Prev = age-adjusted prevalence of ill health. RD = age-adjusted rate difference (percentage)
between high and low socio-economic group
Correlation tests (Pearson’s r). Significant associations boldfaced.
RD OR: 0.93 (men), 0.75 (women)
RD Prev: 0.50 (men), 0.21 (women)
OR Prev: 0.58 (men), 0.19 (women)
Table 3 ORs (95% CI), prevalence rates and rate differences (95% CI) of poor self-assessed health according to education in
former Yugoslavia (N = 17 359)a
Country/region
Bosnia-H
Croatia
Kosovo
Macedonia
Montenegro
Serbia
Slovenia
Men
Women
Prev.
RD (95% CI)
OR (95% CI)
Prev.
RD (95% CI)
OR (95% CI)
36.4
39.3
41.8
34.1
30.6
36.4
41.2
8.9
20.6
10.5
16.2
8.7
13.7
19.0
1.49
1.80
1.24
1.50
1.27
1.38
1.77
42.6
42.8
56.3
44.8
40.8
47.8
49.1
17.4
18.9
15.4
13.2
6.5
18.3
10.7
1.66
1.81
1.40
1.49
1.16
1.79
1.49
(8.6–9.3)
(19.2–21.9)
(9.9–11.1)
(15.2–17.1)
(8.1–9.3)
(13.0–14.4)
(17.4–20.6)
(1.34–1.65)
(1.52–2.14)
(1.07–1.44)
(1.30–1.73)
(1.07–1.50)
(1.23–1.56)
(1.42–2.20)
(16.8–18.1)
(17.9–19.9)
(14.5–16.3)
(12.5–13.9)
(6.1–7.0)
(17.4–19.2)
(9.9–11.6)
(1.49–1.85)
(1.54–2.14)
(1.22–1.61)
(1.30–1.71)
(0.96–1.40)
(1.56–2.05)
(1.21–1.83)
a: All people are aged 25 or more. Prev = age-adjusted prevalence of ill health. RD = age-adjusted rate difference (percentage)
between high and low socio-economic group
Correlation tests (Pearson’s r). Significant associations boldfaced.
RD OR: 0.64 (men), 0.90 (women)
RD Prev: 0.26 (men), 0.17 (women)
OR Prev: 0.05 (men), 0.40 (women)
643
Health inequalities in former Yugoslavia
Montenegro, Bosnia-Herzegovina and Kosovo, while they were
clearly largest in Slovenia and Croatia. Among women,
Montenegro had by far smallest ORs and RDs, while Croatia
and Serbia demonstrated the largest ORs and RDs.
The ranking of countries according to absolute and relative
inequalities were largely in agreement and strongly correlated
(table 2). There were no significant associations between RDs
and prevalence rates, nor between ORs and prevalence rates
(table 3).
Discussion
Summary of findings
As expected, educational health inequalities were present
throughout the former Yugoslavia. There were differences
between countries in the magnitude of these inequalities,
where Montenegro seemed overall to be holding the best
position, and Croatia the most unfavourable one. It should
be noted, however, that Croatia had large, but not largest
OR of LLI among men, nor largest RD of LLI among
women. Before these results will be discussed several study
limitations should be addressed.
Limitations of the study
First, the sample size of Slovenia is smaller for this country as
compared to the other countries, while the sample for
Bosnia-Herzegovina is much larger (see table 1 for an
overview). However, the confidence intervals are not wider
than those observed for the other regions except Croatia and
Bosnia-Herzegovina. Furthermore, the data were not collected
A
from the South-East European Social Survey Project, but from
the European Social Survey. However, exactly the same
questions about health were asked (the LLI and SRH
measures of SEESSP were actually based on the ESS) in both
surveys at the same time. Moreover, the design and procedure of
the SEESSP were similar to those of the ESS, but the results for
Slovenia should for the above reason be interpreted with care.
Second, due to the cross-sectional character of the study, it
is not possible to fully elucidate the inequality effects of health
over the life course.17
Third, absolute estimates are always problematic when
self-reported survey data are used from several different
countries. The prevalence level typically varies a lot (table 1)
and the reasons are manifold. However, we have analysed
neighbouring countries and their similarities should not be
overlooked.
Interpretation of key findings
One of the main pictures emerging from this study is unexpectedly small inequalities in SRH and LLI as compared to
previous comparative European studies. For example, ORs
for morbidity according to education ranged between 1.5
and 2.5 in the 1997 study by Mackenbach et al.18 and
between 1.2 and 1.7 (relative index of inequality) 10 years
later by the same research group.5 In our study, ORs were
never larger than 1.8. This comparison to previous studies
raises the key question why inequalities are not larger in
former Yugoslavia. A similar finding was presented in a
previous study, in which class-related health inequalities
were not observed larger in Eastern Europe as compared to
B
Limiting longstanding illness among men
2.2
Limiting longstanding illness among women
2.2
Odds ratio
Odds ratio
Macedonia
2
Croatia
Croatia
2
Macedonia
Serbia
1.8
Serbia
Kosovo
Bosnia-H
Montenegro
1.6
1.6
1.4
Bosnia-H
1.4
Montenegro
Slovenia
1.2
Kosovo
1.8
Slovenia
1.2
1
1
Absolute difference (%)
0.8
0
5
10
15
20
25
Poor general health among men
C
Absolute difference (%)
0.8
0
10
15
20
25
Poor general health among women
D
2.2
5
2.2
Odds ratio
Slovenia
Odds ratio
Croatia
2
1.8
Bosnia-H
Croatia
Bosnia-H
Slovenia Macedonia
Macedonia
1.8
1.6
Serbia
2
Kosovo
Serbia
Kosovo
1.6
Montenegro
1.4
1.4
Montenegro
1.2
1.2
1
1
Absolute difference (%)
0.8
0
5
10
15
20
25
Absolute difference (%)
0.8
5
10
15
20
Figure 1 ORs (95% CI) (y-axis) and absolute differences (95% CI) of ill health (x-axis) by education in former Yugoslavia
25
644
European Journal of Public Health
the West.15 It was suggested that mortality selection and
negative experiences of the communist regimes, which
eventually lead to lower expectations to health and
healthcare, causing less complaining of ill health among
people with low education, could explain the comparatively
low inequalities in the East. Whilst less complaining about ill
health amongst the less educated could be an explanation for
the low inequalities in our study, it is also possible that the
better educated are less likely to report good health, especially
since the prevalence of reported ill health is higher than many
surveys in Western Europe (and more similar to Russia)—
whether this relates disease prevalence or cultural influences
on reporting is unclear. Furthermore, former communist
countries emphasized an egalitarian ideology in the past,
which might be still persistent in the social structure. These
explanations could also account for the comparatively low
inequalities found in the former Yugoslavian countries.
However, there are also some unexpected findings from
within the studied region. It might have been expected that
educational inequalities would be larger, and self-assessed
health worse, in areas with mixed religious affiliation, greater
political instability and low GDP per capita (i.e. Kosovo,
Montenegro and Bosnia-Herzegovina), but this was not what
our findings demonstrated. Whilst the generally high
prevalence of LLI and poor SRH in Kosovo could be
expected in a poor country with ongoing unrest, the similar
finding in the relatively wealthy and peaceful Slovenia was
surprising. The relatively good health and narrower inequalities
in the comparatively poor Montenegro, and the intermediate
findings in previously conflict-ridden Bosnia-Herzegovina are
also not in agreement with this expectation. However,
numerous studies have demonstrated that socio-economic
inequalities in health may not always be smaller in affluent
and even egalitarian societies,4,5,15,19 which may be deeply
rooted in structural inequalities exemplified by social class,
education and income differences.
Overlapping confidence intervals added to uncertainties in
the pattern of inequalities, with the exception of Montenegro,
where relative health inequalities were clearly lower than in
some of the other countries.
Differences in reporting subjective health, especially in
relation to education and culture, could account for some of
the inconsistencies. First, people with higher education may
emphasize different dimensions when rating their health.20
For example, if the lower educated included the stressors of
daily life in their rating, this might weaken the association
between subjective health and subsequent mortality
compared with higher educated individuals. Recently, several
studies have assessed how education moderates the association
of SRH with mortality, which may be regarded as partially
testing this possibility.20–24 Results were mixed, with some
studies reporting no socio-economic differences in the
predictive power of self-rated health on mortality21,22,24 and
others reporting that the predictive power is stronger in higher
socio-economic groups,20,23 and one reporting that the
predictive power is greater in the lower socio-economic
groups.25 Also in cases where the ecological approach has
been used to compare the magnitude of socio-economic
inequalities in SRH with the magnitude of inequalities in mortality, results have been inconsistent.26–28 However, education,
in contrast with other determinants, predicted self-rated health
and mortality to a similar degree within a Russian study,29
whilst results of this single study suggest a degree of stability;
overall it is possible that SRH may give different estimates of
socio-economic inequalities compared with other health
outcomes.
Second, reporting variations may also result from the wide
cultural and ethnic diversity in countries of the former
Yugoslavia, although within-country variations in ethnic composition make this difficult to study. There are international
variations in the typical value for SRH,12 and whilst this is
partly related to life expectancy, there may be other
influences. There has been little data on the comparability
across cultures, but a study in Finland and Italy showed that
although SRH is a useful summary of physical health, it may
be sensitive to cultural environment.30 Thus, health expectations may vary according to culture (the greater levels of
inequality in the relatively wealthy Slovenia could be related
to higher expectations among the higher educated). Direct
cultural comparisons of SRH outcomes in general should
therefore be made with caution. It should be noted, however,
that this issue is of greater concern with respect to the
(age-adjusted) prevalence of ill health, as compared to odds
ratio measures of inequality that are reported here.
The prevalence and socio-economic distribution of other
causes of ill health may also vary between countries. Evidence
from the EUROTHINE study, a large European-wide study on
the magnitude of socio-economic inequalities in health and
health-related behaviours, suggests that despite universal
inequalities in objective health across Europe, some causes
may be more important in different countries. For instance,
the social patterning of health behaviours varies between
European countries,31 and several of these explain an
important part of variations in health inequalities across
Europe.2,5,32 Future studies aiming to unravel the causes of
cross-country variations in health inequalities in former
Yugoslavian countries have the hard task of considering a
multitude of factors, ranging from (but not limited to)
welfare systems to smoking, and from income inequality
to access to healthcare.
The prevalence of LLI and SRH and the magnitude of
inequalities in LLI reported in this study for the former
Yugoslavian countries correspond to those that have been
reported before for other Central Eastern European
countries. The prevalence was broadly similar to that in the
Czech Republic, Estonia, Hungary, Poland and Slovakia, which
were reported in the study of Eikemo et al.,4 which used exactly
the same measures of health and education, collected at
approximately the same time (2002–2004). Regarding the
size of the relative inequalities, the ORs of LLI and poor
SRH in the former Yugoslavian countries typically varied
between 1.2 and 1.7 (1.8 for SRH), similar in order to other
Eastern European countries,4 although wide confidence
intervals in this study limit exact comparisons. Ranking
countries when differences between point estimates are small
leads to missing the big picture that there may not be
meaningful differences in the magnitude of inequalities
between countries in the first place. We therefore performed
sensitivity analyses to formally test whether health varies by
education between the countries, by fitting interaction terms
between countries and education (with respect to both health
indicators) in pooled analyses. These results showed that the
interaction terms added significantly to the models only
containing education and single countries (controlled for
age), suggesting that there are indeed meaningful differences
between the countries. Next, we tested the significance of each
interaction term (in several steps using different reference
countries). The picture that emerged is that although the interaction terms contribute to the models as a whole, differences
between pair of countries are not always significant
(Supplementary tables SA and SB). More solid conclusions
about cross-country differences can be drawn when point
estimates are further apart, when patterns of inequalities
across countries are consistent across several health
outcomes, and/or when patterns are consistent in terms of
absolute and relative inequalities.
Health inequalities in former Yugoslavia
Given the unique experience of war in some of the former
Yugoslavian countries we should have indeed expected that
inequalities in these countries would be similar, or larger,
but definitely not clearly smaller, than in other Central
Eastern European countries. However, the survey took place
some years after the conflict ended, and we do not know how
this might have affected the results. Whilst the adverse consequences to self-rated health33 and psychological34 well-being
can persist for many years after war, the amount of research of
this type is very limited. In summary, these results fill up the
gap in the current knowledge about the magnitude of health
inequalities in European countries.
Conclusion
Educational health inequalities were present throughout the
former Yugoslavia (except among Montenegrin women). The
countries with the most unfavourable histories of conflict, and/
or the largest income inequalities, and/or the largest religious heterogeneity, i.e. Kosovo, Montenegro and
Bosnia-Herzegovina, did not exhibit consistently large
inequalities as compared to the other countries from former
Yugoslavia. Further research should study a wider range of
health outcomes using larger survey samples and a wider
range of cultural and other predictor variables to fill further
gaps in the knowledge base of socio-economic inequalities in
health in Europe. Efforts should be made to include these
countries in broader European overviews of inequalities
in health.
645
References
1
Siegrist J, Marmot M. Social inequalities in health. New evidence and policy
implications. Oxford: Oxford University Press, 2006.
2
Huisman M, Kunst AE, Bopp M, et al. Educational inequalities in
cause-specific mortality in middle-aged and older men and women in eight
western European populations. Lancet 2005;365:493–500.
3
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;6:15.
4
Eikemo TA, Huisman M, Bambra C, Kunst AE. Health Inequalities
according to educational level under different welfare regimes. Sociol Health
Illn 2008;30:565–82.
5
Mackenbach JP, Stirbu I, Roskam A-JR, et al. Socioeconomic Inequalities in
Health in 22 European Countries. N Engl J Med 2008;358:2468–81.
6
Pevalin DJ, Robson K. Social determinants of health inequalities in Bosnia
and Herzegovina. Public Health 2007;121:588–95.
7
Ringdal GI, Ringdal K, Simkus A. War Experiences and war-related distress
in Bosnia and Herzegovina eight years after war. Croat Med J 2008;49:75–86.
8
Ringdal GI, Ringdal K, Simkus A. War-related distress among Kosovar
Albanians. J Loss Trauma 2008;13:59–71.
9
Cavelaars A, Kunst AE, Geurts JJM, et al. Differences in self reported
morbidity by educational level: a comparison of 11 Western European
countries. J Epidemiol Community Health 1998;52:219–27.
10 Simkus A. Guest editor’s introduction: The SEESSP project, Determinants of
Social Attitudes in the Western Balkans. Int J Sociol 2008;37:3–14.
11 Davey Smith G, Blane D, Bartley M. Explanations for socio-economic
differentials in mortality: evidence from Britain and elsewhere. Eur
J Public Health 1994;4:131–44.
Supplementary data
12 Idler EL, Benyamini Y. Self-rated health and mortality: a review of
twenty-seven community studies. J Health Soc Behav 1997;38:21–37.
Supplementary data are available at EURPUB online.
13 Lahelma E, Manderbacka K, Rahkonen O, Karisto A. Comparisons of
inequalities in health—evidence from National Surveys in Finland, Norway
and Sweden. Soc Sci Med 1994;38:517–24.
Funding
The study was supported by a Wellcome Trust Intermediate
Clinical Fellowship.
Conflicts of interest: None declared.
Key points
Substantial socio-economic inequalities in morbidity
and mortality have been reported for all European
countries with available data.
The size of inequalities in morbidity varies internationally but the situation in many countries remains
unexplored, even in Europe
Until now, data from former Yugoslavian countries
have remained unexplored. Mapping and explaining
health inequalities in this region is particularly
important given its recent history of cultural
diversity, violent conflicts and rapid economic,
political and social change.
Educational health inequalities were present throughout
the former Yugoslavia. Inequalities seemed to be
generally large in Croatia and small in Montenegro.
Educational inequalities in health were not larger, and
self-rated health was not worse, in parts of former
Yugoslavia with mixed religious affiliation, greater
political instability and low GDP per capita.
Also in times of great political transitions, conflicts
and war, educational attainment is of considerable
benefit to people’s health.
14 Galobardes B, Shaw M, Lawlor DA, et al. Indicators of socioeconomic
position (part 1). J Epidemiol Community Health 2006;60:7–12.
15 Eikemo TA, Kunst AE, Judge K, Mackenbach JP. Class related health
inequalities are not larger in the East: a comparison of 4 European regions
using the new European Socio-Economic Classification. J Epidemiol
Community Health 2008;62:1072–8.
16 Vagero D, Erikson R. Socioeconomic inequalities in morbidity and mortality
in western Europe. Lancet 1997;350:516.
17 Kuh D, Ben-Shlomo Y, Lynch J, et al. Life course epidemiology. J Epidemiol
Community Health 2003;57:778–83.
18 Mackenbach JP, Kunst A, Cavelaars A, et al. Health EWGiSIi. Socioeconomic
inequalities in morbidity and mortality in western Europe. Lancet
1997;349:1655–9.
19 Eikemo TA, Bambra C, Joyce K, Dahl E. Welfare state regimes and income
related health inequalities. A comparison of 23 European countries. Eur
J Public Health 2008;18:593–9.
20 Huisman M, Van Lenthe F, Mackenbach JP. The predictive ability of
self-assessed health for mortality in different educational groups. Int
J Epidemiol 2007;36:1207–13.
21 Van Doorslaer E, Gerdtham UG. Does inequality in self-assessed health
predict inequality in survival by income? Evidence from Swedish data.
Soc Sci Med 2003;57:1621–9.
22 McFadden E, Luben R, Bingham S, et al. Does the association between
self-rated health and mortality vary by social class? Soc Sci Med
2009;68:275–80.
23 Dowd JB, Zajacova A. Does the predictive power of self-rated health for
subsequent mortality risk vary by socioeconomic status in the US? Int
J Epidemiol 2007;36:1214–21.
24 Burstrom B, Fredlund P. Self rated health: is it as good a predictor of
subsequent mortality among adults in lower as well as in higher social
classes? J Epidemiol Community Health 2001;55:836–40.
646
European Journal of Public Health
25 Singh-Manoux A, Dugravot A, Shipley MJ, et al. The association between
self-rated health and mortality in different socioeconomic groups in the
GAZEL cohort study. Int J Epidemiol 2007;36:1222–8.
26 Huisman M, Van Lenthe F, Mackenbach JP. Author’s response. Int J
Epidemiol 2008;37:1437–8.
27 Singh-Manoux A, Dugravot A, Shipley M, et al. Author’s response. Int J
Epidemiol 2008;37:1439–40.
28 Subramanian SV, Ertel K. Is the use of self-rated health measures to assess
health inequalities misleading? Int J Epidemiol 2008;37:1436–40.
29 Perlman F, Bobak M. Determinants of self rated health and mortality in
Russia—are they the same? Int J Equity Health 2008;7:19.
30 Jylha M, Guralnik JM, Ferrucci L, et al. Is self-rated health comparable across
cultures and genders? J Gerontol Ser B: Psychol Sci Soc Sci May
1998;53:S144–52.
31 Eurothine Final Report. Available at http://mgzlx4.erasmusmc.nl/eurothine
(Accessed 9 March 2009).
32 Mackenbach JP, Huisman M, Andersen O, et al. Inequalities in lung cancer
mortality by the educational level in 10 European populations. Eur J Cancer
2004;40:126–35.
33 Stessman J, Cohen A, Hammerman-Rozenberg R, et al. Holocaust survivors
in old age: the Jerusalem Longitudinal Study. J Am Geriatr Soc
2008;56:470–7.
34 Hauff E, Vaglum P. Organised violence and the stress of exile. Predictors of
mental health in a community cohort of Vietnamese refugees three years after
resettlement. Br J Psychiatry 1995;166:360–7.
Received 9 March 2009, accepted 6 November 2009