The European Journal of Public Health Advance Access originally published online on July 13, 2005
The European Journal of Public Health 2005 15(5):498-503; doi:10.1093/eurpub/cki019
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Health Inequalities |
Changing social variations in self-assessed health in times of transition? The Baltic States 1994–1999
Christiaan W.S. Monden** Department of Sociology, Tilburg University, Tilburg, The Netherlands
Correspondence: Department of Sociology, Tilburg University, PO Box 90153, 5000 LE Tilburg, The Netherlands, tel: +31 13 4662119, fax: +31 13 4668068, e-mail: c.w.s.monden{at}uvt.nl
Received January 9, 2004, accepted May 11, 2004
 |
Abstract |
|---|
 Top Abstract MethodsResultsDiscussionReferences |
|---|
Background: Dramatic social changes took place in the Baltic States (Estonia, Latvia and Lithuania) in the 1990s. This study investigates the extent to which social variations in self-assessed health changed during that period. Methods: Norbalt Living Conditions Survey I (1994) and II (1999) random population-based samples in Estonia, Latvia and Lithuania were analysed. Associations of self-assessed health with six social dimensions (education, economic activity, car ownership, number of rooms, ethnicity and residence) were studied for males and females aged 25–74 years (n = 16 970). Results: Substantial and significant associations with poor health were found for education, economic activity, car ownership and, to a lesser extent, number of rooms. Ethnic differences were found only among women in Estonia. By and large, social variations in health were comparable for most indicators between the three countries. Differences in self-assessed health were stable between 1994 and 1999, except for the relatively worse position of the economically non-active in 1999. Conclusions: Substantial social inequalities in self-assessed poor health exist in the Baltic States. Despite dramatic social changes taking place, social variations in self-assessed health have been rather stable in the second half of the 1990s. The economically non-active seem to have become more disadvantaged.
Keywords: Baltic States, Estonia, Latvia, Lithuania, self-assessed health, social inequalities, socio-economic indicators
The three Baltic States of Estonia, Latvia and Lithuania experienced dramatic changes in the 1990s. Since 1991, when they regained independence from the Soviet Union after a period of almost 50 years of communist rule, the three countries have been in a transition to democratic market-oriented societies. The social and economic reforms of the 1990s have had a profound impact on almost all aspects of life in the Baltic States.1–4 For instance, the labour market, pension system and health-care system all have changed significantly during the last decade. Income inequality has increased substantially. In 1987/1988 the gini-coefficient was 23 in all three countries, and rose to 31 in Latvia, 35 in Estonia and even 37 in Lithuania in 1993/1995.5 In the mid and late 1990s, government income transfers (pensions, child support, etc.) were reformed and/or could not keep up with inflation rates of >10% up until 1997. Economic growth was low or negative in the early 1990s. From 1995 (1996 for Latvia) to 1998, economic growth was high, but the Russian financial crisis slowed it down and led to negative growth in Estonia and Lithuania in 1999.
In all countries for which empirical research is available, substantial differences in health status and mortality rates have been found between socio-economic groups.6 A growing number of studies report on social inequalities in health in central and eastern European societies.7,8 Only a few studies have documented socio-economic health differences in the Baltic Republics,9–11 and for self-assessed health no comparisons over time are available. A study of mortality differences by education in Estonia showed much bigger differences between educational groups in 2000 compared with 1989.12 The huge changes that have occurred in the Baltic States give rise to the question of the extent to which social variations in health have developed in the transition period. The aim of this study was therefore to determine whether there were changes in the magnitude of social variations in self-assessed health in the Baltic States between 1994 and 1999.
|
Methods |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
In 1994 and 1999, the Norwegian Institute for Applied Social Science (Fafo) conducted nationally representative living conditions surveys in the Baltic States (Norbalt I and II) in cooperation with the Statistical Office of Estonia and University of Tartu in Estonia, Central Statistical Bureau of Latvia in Latvia, and the Ministry of Labour and Social Security and the Department of Statistics of the Government of the Republic of Lithuania in Lithuania.13,14 Population registers were used to draw a sample of addresses. In each household, one random respondent was interviewed. In both years and all three countries the non-response due to no contact or refusal was <10%.14 The non-response due to frame imperfections (vacant or non-existing buildings and dwellings used for other purposed than housing) was 2–6%. Face-to-face interviews were held with 20 503 individuals of 18 years and older. Respondents younger than 25 (2223) and older than 74 years of age (1040) were excluded from the analyses. The lower age limit was chosen to make sure the majority of the respondents has finished full time education. Applying a lower age limit might bias the results because the final educational level is not yet known for some respondents. The upper age limit should prevent biased results due to selective mortality. The original sample included respondents up to 98 years of age. After deleting cases with missing information on self-assessed health or control variables (age, marital status, ethnicity and residence), 16 970 cases were left for analysis.
A single item question was applied for self-assessed health: ‘How would you characterize your health in general? Very good, good, fair (average, moderate), poor or very poor.’ This frequently used question has been shown to be a good indicator of general physical health.15,16 It predicts mortality and correlates highly with more objective health measures. In the analyses, self-assessed health was dichotomized into ‘poor health’ (0 = no, 1 = yes). The first three categories, ‘very good’, ‘good’ and ‘average’, were contrasted with the last two answer categories, ‘poor’ or ‘very poor’.
Several indicators of socio-economic position were available in both survey years for all three countries: education, employment status, possession of a car and number of rooms in the dwelling. Education was divided in three levels: primary, secondary and tertiary education. The last category included university and tertiary higher vocational educational. Specialized secondary education after secondary school, a remnant of the Soviet educational system, was grouped with secondary education to make 1994 and 1999 comparable. Employment status was defined according to ILO standards. Three categories were distinguished: employed, unemployed (looking for a job) and non-active. Unfortunately, income information was not available in the 1994 survey. Two other indicators of material welfare were used instead. In both years, respondents were asked whether their household owned a car. Car ownership has been used before as an indicator of socio-economic status.17,18 Another indicator for socio-economic position is the number of rooms in the dwelling (excluding kitchen and bathrooms). Note, however, that after independence many expropriated apartments and houses were returned to former owners or their children. This means that the number of rooms does not necessarily reflect income. In the social stratification of Baltic societies, ethnicity might also play an important role.9 Estonia and Latvia in particular have large ethnic minorities, mostly Russians. In each country, titular and non-titular ethnic groups were distinguished. The samples were too small to differentiate between more ethnic groups. The present study also addressed differences between rural and urban respondents, because previous research on poverty in Latvia19 suggested that there are large differences in living conditions and socio-economic chances between rural and urban areas. Villages and towns with <50 000 inhabitants were categorized as rural. Age and marital status were used as control variables in the analyses. Age was divided in 10 5-year age groups, and four types of marital status were distinguished: never-married, married, divorced and widowed. Logistic regression was used to obtain odds ratios (ORs) and confidence intervals (CIs) (using Stata 8).20
Differences between years were investigated with interaction terms for social indicators and year.21 Likewise, interaction terms with gender were used to test whether social variations in health were equal for men and women. Three-way interactions were used to test whether changes between 1994 and 1999 differ significantly by gender.
|
Results |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
Table 1 presents the distribution of self-assessed health and social dimensions. In 1994, a relatively small number of respondents (25%) considered their health as (very) good. Compared with western countries, the percentage of respondents reporting less than good health was high.21,22 In 1994 and 1999, Latvia had a significantly higher percentage of respondents reporting poor health than Estonia and Lithuania. The latter two showed a significant decrease in the number or respondents who reported poor health between 1994 and 1999. There was no significant change in Latvia.
|
2-test for differences between survey yearIn all three countries, women assessed their health significantly more often as poor than men. The gender difference is approximately equal in Lithuania (age-adjusted OR for men 0.59; 95% CI 0.49–0.71) and Latvia (OR 0.59; 95% CI 0.51–0.68), but significantly smaller in Estonia (OR 0.77; 95% CI 0.68–0.87). These differences did not change between 1994 and 1999.
The educational distributions were very similar in the three countries. For Latvia and Lithuania no changes were observed over time. In Estonia, respondents with primary education seem to be underrepresented in 1999. The number of employed people increased in Estonia and Latvia, whereas in Lithuania there were more unemployed and non-active people in 1999 than in 1994. The increasing number of car owners reflects the strong economic development in the Baltic States. Also, the percentage of respondents who live in a house with less than three rooms decreased between 1994 and 1999. In both cases, Latvia lags somewhat behind in terms of development.
Estonia and Latvia have large ethnic minorities (34% and 44%, respectively) consisting mostly of Russians. Poles and Russians make up most of Lithuania's relatively small ethnic minority (18%).
Tables 2
–4 present the association of self-assessed poor health with socio-economic variables separately for the three countries. The last row of each reports the P value of the interaction terms for survey year and sex. All models include age, sex, marital status and survey year.
|
|
|
Estonia
Table 2 shows the association of socio-economic variables with self-assessed health for Estonia. In the first model, the socio-economic variables are not mutually adjusted. In the second model, which includes all variables, people with primary education were more than twice as likely to report poor health (OR 2.39; 95% CI 1.87–3.05) than those with tertiary education. This association did not differ between men and women (P = 0.35), nor did the association interact with survey year (P = 0.763 for primary education), suggesting stable educational health differences between 1994 and 1999.
With regard to employment status, the results show that unemployed and non-active respondents were more likely to report poor health than employed respondents. No difference between men and women was found. The interaction of labour market status and survey year was positive and significant (P = 0.044 for unemployed, and P = 0.037 for non-active). The health difference between the employed on the one hand and unemployed and non-active on the other was larger in 1999 than in 1994. The three-way interaction showed that this development was equal for men and women.
Respondents from households that do not own a car were more likely to report poor health. No significant interactions with survey year or sex were found.
Living in a household with less than three rooms was associated with poor self-assessed health (OR 1.27; 95% CI 1.11–1.46). This association, too, was equal for men and women and did not differ between 1994 and 1999.
Finally, ethnicity and urbanization were used to explore social variations in health. The interaction term for ethnicity reveals important gender differences. Non-Estonian women more often reported poor health than Estonian women (OR 1.48; 95% CI 1.23–1.77), whereas there were no differences among men (OR 0.92; 95% CI 0.73–1.18). The two-way interaction term with survey year suggests a decrease in ethnic differences, but was not significant. With regard to urbanization, the results show that respondents in rural areas more often reported poor health. This held equally true for men and women and for the two survey years.
Latvia
For the most part, the results for Latvia (table 3) correspond well with those for Estonia. However, some differences can be observed. First of all, in Latvia there has not been a decrease in the number of respondents reporting poor health. Respondents with primary education, the unemployed and non-active, as well as respondents who do not own a car or live in a house with less than three rooms, were more likely to report poor health. In contrast to Estonia, the association of health with the number of rooms, urbanization and ethnicity were not significant in the adjusted model. No significant interactions with survey year were found. A borderline significant interaction between being unemployed and survey year was found (P = 0.066), suggesting that health differences between the unemployed and employed were larger in 1999 compared with 1994. Otherwise, the results indicate stable social variations in health. Some sex differences were found. Educational differences were stronger for women than for men (P = 0.075 for primary education, and P = 0.033 for secondary education), and the difference between employed and non-active respondents was more pronounced among men (P = 0.039).
Lithuania
Table 4 shows that social variations in health in Lithuania are quite similar to those in Estonia and Latvia, although they resemble the Latvian case more closely. Significant associations with self-assessed health were found for education, labour market position and car ownership. The association for being non-active was stronger in 1999 than in 1994 (P < 0.001), and was stronger for men (P = 0.048). Ethnicity was not associated with self-assessed health. With regard to urbanization and number of rooms, significant interactions with survey year were found. In the 1994 survey, number of rooms was associated with health (OR 1.58; 95% CI 1.20–2.09), but this was not the case in 1999. For urbanization the opposite was found; in 1999 respondents from rural areas were less likely to report poor health (OR 0.61; 95% CI 0.46–0.82), but no association was found for 1994.
|
Discussion |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
The aim of this paper was to describe social variations in self-assessed health in the Baltic States and to determine whether these inequalities changed between 1994 and 1999. As for all modern societies, substantial social variations in self-assessed health were found. People with higher socio-economic positions reported better health than those with lower positions. This was true for education, employment status, car ownership and, to a lesser degree, number of rooms. The differences found for education correspond well with those typically found in western European countries.9,10,22 Differences with regard to ethnicity appeared to be rather small and insignificant, except for women in Estonia. Contrary to the expectation, no substantial differences between people in urban and rural areas were found. Only in Estonia were respondents in rural areas more likely to report poor health. Finally, women reported poor health more often than men. The association between socio-economic indicators and health did not differ systematically between men and women.
For all three countries, social variations in self-assessed health seem to have been rather stable in the second half of the 1990s despite ongoing social changes. In Estonia and Lithuania, the health differences between employed people on the one hand and unemployed and non-active on the other were larger in 1999 than in 1994. The reason for this might be a stronger health selection on the labour market and/or a deterioration of the health of the unemployed. The non-active often depend on government income transfers and public services. With ongoing liberalization and increasing prices (falling real incomes), these income transfer and subsidy programmes may lose part of their cushioning effects. The group of non-active people is heterogeneous. Future research should look at more detailed categories of people who are unable to work, retired people and homemakers. Although the speed and nature of the reforms and changes has differed between the three countries, there do not seem to be structural differences regarding the developments in social variations in self-assessed health.
Some possible limitations of this study need to be mentioned. This study concerns self-assessed health, and the results cannot simply be generalized to other health indicators. For instance, it is possible that there has been a rise in socio-economic differences with regard to mental health, mortality or morbidity from specific conditions.
Car ownership and housing size might not be clear-cut indicators of only material wealth.18 More people in the countryside than in cities have a car simply because they need transport. Car ownership might signal something other than just wealth. Therefore, interactions of urbanization and car ownership were tested, but they were not significant. People with a car are healthier in both the countryside and cities. The same holds true for the number of rooms. Houses in the countryside often have more rooms than apartments in the cities, but they lack other conveniences. The number of rooms does not necessarily reflect wealth. Also, the interaction term between urbanization and number of rooms was not significant.
There seems to be an underrepresentation of lower educated in the 1999 Estonian sample. This does not affect the ORs for educational differences as long as the underrepresentation is not selective on health. There are no indications that such a systematic underrepresentation exists. Moreover, the educational distribution for the complete Estonian 1999 and 1994 samples, that is, without the 25–74 years age selection, do correspond very well. It is surprising that no urban–rural differences were found, given the lower level of services and material welfare in the countryside. Larger samples would enable a more detailed analysis of regional differences.
This study is the first to report on social variations in self-assessed health over time in the Baltic States. However, the time span studied was short. Leinsalu et al.12 showed a very substantial increase in educational difference in mortality for Estonia between 1989 and 2000. Whether this increase also holds true for self-assessed health is unclear, although the hypothesis seems obvious. Unfortunately, there do not appear to be any comparable data on self-assessed health from before 1990 in the Baltic States. More importantly, it is as yet unclear whether the transition caused an abrupt increase in socio-economic health differences or whether these differences have been widening ever since the start of the transition. The impact on social inequalities might have been most severe in the first and toughest years of independence. This is an important issue for further investigation.
The accession of Estonia, Latvia and Lithuania to the European Union in May 2004 marked an important new phase, but does not mean that the time of social reforms is over. Moreover, the period of changes in the 1990s might have had lasting effects on the population's health. Therefore, it is important to continue to monitor health and social variations in health in the Baltic States. Next to social variations in morbidity, it is important to describe and monitor social variations in mortality in all three States. In particular, the health differences by economic activity and income require more thorough examination.
Key points
|
|
Acknowledgments |
|---|
The Fafo Institute (Olso, Norway) kindly made the Norbalt I and II surveys available for analysis. Norbalt I and II were supported by the Norwegian Ministry of Foreign Affairs, the Nordic Council of Ministers and the Norwegian Research Council.
|
References |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
1 UNDP. Living standards, education reform and participation. HDR 1997. Riga: UNDP, 1997.
2 UNDP. Living standards and choices. HDR 1997. Vilnius: UNDP, 1997.
3 UNDP. Human development and conflict in simultaneously occurring processes of modernisation and post-modernisation. HDR 1999. Tallinn: UNDP, 1999.
4 Ministry of Economy. Economic development of Latvia 1998. Riga: Ministry of Economy, 1998.
5 Milanovic B. Income, inequality, and poverty during the transition from planned to market economy. Washington: The World Bank, 1998.
6 Mackenbach JP, Kunst AE, Cavelaars EJM, et al. Socioeconomic inequalities in morbidity and mortality in western Europe. Lancet 1997;349:1655–9.[CrossRef][ISI][Medline]
7 Bobak M, Pikhart H, Rose R, et al. Socioeconomic factors, material inequalities, and perceived control in self-reported health: Cross-sectional data from seven post-communist countries. Soc Sci Med 2000;51:1343–50.[CrossRef][ISI][Medline]
8 Carlson P. Self-perceived health in east and west Europe: another European health divide. Soc Sci Med 1998;46:1355–66.[Medline]
9 Leinsalu M. Social variation in self-rated health in Estonia: a cross-sectional study. Soc Sci Med 2002;55:847–61.[CrossRef][ISI][Medline]
10 Monden C. Socio-economic health inequalities in Latvia: A cross sectional study. Scand J Public Health 2004; in press.
11 Kalediene R, Petrauskiene J. Inequalities in life expectancy in Lithuania by level of education. Scand J Public Health 2000;28:4–9.[ISI][Medline]
12 Leinsalu M, Vagero D, Kunst AE. Estonia 1989–2000: enormous increase in mortality differences by education. Int J Epidemiol 2003;32:1081–7.
13 Central Statistical Bureau of Latvia. Living conditions survey 1999. Riga: Central Statistical Bureau of Latvia, 2000.
14 Aasland A, Tyldum G. The Norbalt project: comparative studies of living conditions in the three Baltic States. Soc Indic Res 2002;58:177–89.[CrossRef]
15 Idler EL, Benyamini Y. Self-rated health and mortality: a review of twenty-seven community studies. J Health Soc Behav 1997;38:21–37.[CrossRef][ISI][Medline]
16 Manderbacka K. Questions on survey questions on health. Stockholm: Swedish Institute for Social Research, 1998.
17 Goldblatt P. Mortality and alternative social classifications. In Goldblatt P, editor. Longitudinal study. Mortality and social organisation. OPCS, Seris LS, 6. London: HMSO, 1990: 000–000.
18 Macintyre S, Ellaway A, Der G, et al. Do housing tenure and car access predict health because they are simply markers of income or self-esteem? A Scottish study. J Epidemiol Community Health 1998;52:657–664.[Abstract]
19 Grassmann F. Who and where are the poor in Latvia? Riga: UNDP, 1998.
20 Stata Corporation. Stata reference manual: release 8. College Station, TX: Stata Press.
21 Silventoinen K, Lahelma E. Health inequalities by education and age in four Nordic countries, 1986 and 1994. J Epidemiol Community Health 2002;56:253–8.
22 Cavelaars AEJM, Kunst AE, Geurts JJM. Differences in self-reported morbidity by educational level: a comparison of 11 western European countries. J Epidemiol Community Health 1998;52:219–27.[Abstract]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||