The European Journal of Public Health Advance Access originally published online on July 19, 2006
The European Journal of Public Health 2007 17(1):8-16; doi:10.1093/eurpub/ckl104
Child and adolescent health |
Major improvements, but persisting inequalities in infant survival in Estonia 1992–2002
Ilona Koupil1, Kaja Rahu2, Mati Rahu2, Helle Karro3 and Denny Vågerö1
1 Centre for Health Equity Studies (CHESS), Stockholm University/Karolinska Institute 106 91 Stockholm, Sweden
2 Department of Epidemiology and Biostatistics, National Institute for Health Development, Hiiu 42, 11619 Tallinn, Estonia and Estonian Centre of Excellence in Behavioural and Health Sciences Tallinn/Tartu, Estonia
3 Department of Obstetrics and Gynaecology, University of Tartu Lossi 36, 51003 Tartu, Estonia
Correspondence: Professor Ilona Koupil, Centre for Health Equity Studies (CHESS), Stockholm University/Karolinska Institute, 106 91 Stockholm, Sweden, tel: +46 8 163952; fax: +46 8 162600, e-mail: ilona.koupil{at}chess.su.se
Received May 7, 2005, accepted March 2, 2006
 |
Abstract |
|---|
 Top Abstract Material and methodsResultsDiscussionReferences |
|---|
Background: Inequality in adult health increased in Estonia during the transition period after 1991. We examined inequality in infant survival from 1992 to 2002. Methods: All 132 854 singleton live births reported to the Estonian Medical Birth Registry in 1992–2001 were linked to the Estonian Mortality Database. The effect of mother's education, nationality, marital status, and place of residence on neonatal (0–27 days) and post-neonatal (28–364 days) death was evaluated in logistic regression with adjustments for maternal age, parity, smoking, sex of the infant, birth weight, and gestational age. Results: Infant mortality decreased substantially. Risk of death in neonatal period was lowest in Tartu, with a decline from 4.9/1000 in infants born in 1992–1996 to 2.1/1000 in those born in 1997–2001. Decline in neonatal mortality in other regions was from 9.2/1000 to 5.1/1000. Persisting regional differences were unexplained by mothers' nationality, education, or marital status, or the infants' length of gestation. Decline in post-neonatal mortality was less marked and although risk differences between different socio-economic groups decreased, mothers' marital status and education in particular remained strongly associated with risk of post-neonatal death [odds ratio for infants born to mothers with basic or lower education compared to university education 3.70 (95% confidence interval 2.34–5.85) in 1992–1996 and 3.56 (2.06–6.14) in 1997–2001]. Conclusions: Infant survival improved appreciably in Estonia after 1991 and risk differences between social groups decreased. The improvements were not accompanied by reduction in the strength of the effects of social characteristics on infant death measured as risk ratios.
Keywords: Estonia, health inequalities, infant death, neonatal death, post-neonatal death, socio-economic factors
Estonia has been undergoing major political, societal, and economic changes after regaining its independence in 1991. A series of political and economic reforms lead to an accession to the European Union in 2004. Together with other transitional economies, Estonia experienced an economic crisis in early 1990s. Gross domestic product (GDP) in Estonia decreased till 62% of the 1989 level in year 1994. Since 1994, Estonia has enjoyed economic growth with an average annual change in GDP of 
4%. However, in 2002, GDP was still below the level from 1989 and the real wages were at 70% of the 1989 level. There was a substantial increase in unemployment and a dramatic increase in economic inequality, reflected in a rise in the Gini coefficient from 0.25 in 1989 to 0.39 in 2001.1
Fundamental health care reforms in Estonia started in 1991 with the introduction of a social health insurance system where all costs were covered by health insurance funds collected from earmarked payroll tax. Subsequent health sector reforms included decentralisation of the former state-controlled health care planning and delivery system in mid-1990s and efforts to strengthen primary care by introducing family physicians with the aim to reduce the use of specialist care and contain health care costs.2,3
At the beginning of 1990s, together with other Baltic countries, Estonia experienced an unprecedented rise in adult mortality. After 1994, adult mortality fell and life expectancy increased again, with yet another slight decline after 1999.4,5 There is persuasive evidence for increasing inequality in adult health in Estonia during the transition period after 1991.6 Between 1989 and 2000, educational differences in total mortality among adults increased for all age groups. Moreover, during the same period, life expectancy at age 25 increased in university educated men and women but declined by more than 2 years in those with lower secondary or lower education.7 A simultaneous widening of ethnic differences in mortality has been documented.8
It is plausible that the profound economic and social changes and the economic crisis of early 1990s in particular, could also affect infant and child health through changes in living conditions of parents or direct effects of the reorganised health services. Routine data for the whole country suggest that in sharp contrast to the trends in adult health, infant mortality was generally declining throughout the period after 1992 (figure 1). There is, however, no information on whether infant survival improved in all social groups or regions of the country and there are reasons for concern about potential differences in infant survival between social groups in Estonia. A strong focus on health care costs containment could have effects on equity in access to care for certain groups of the population. It is also likely that the increasing income inequality and social patterning of unhealthy lifestyles such as smoking or alcohol drinking in the adults could affect the health of infants and young children. This hypothesis is consistent with a previously documented increase in variation in mean birth weight by maternal education during the period from 1992 to 1997.9 The objective of our study was to investigate social and regional differences in infant deaths as well as their changes in Estonia from 1992 to 2002.
|
|
Material and methods |
|---|
|
TopAbstractMaterial and methodsResultsDiscussionReferences |
|---|
The Estonian Medical Birth Registry was established in 1991. Since 1992, it has collected, among other health indicators, data on child's sex, single/multiple birth, weight and length at birth, gestational age, mother's parity, age, smoking habits, marital status, education, and nationality. When a birth registration form is filled at a maternity ward and/or post-natal ward, every newborn is also given a unique personal identification number. The WHO definition of live birth10 has been used in Estonia since 1 January 1992. More than 99% of births are attended by skilled health personnel1 and a vast majority of births occur in hospitals. The quality and completeness of birth registration in Estonia is considered generally good.11 There was no indication that the quality of data in the birth or death registers changed significantly over the study period.
All live births reported to the Estonian Medical Birth Registry in 1992–2001 were linked to the Estonian Mortality Database, using the unique personal identification number. The linkage was successful for 99% of infant deaths. Causes of death, coded according to ICD-9, were available from the Mortality Database and were grouped into ICD-9 chapters for the analysis. The standard definition of neonatal (at age 0–27 days) and post-neonatal (at age 28–364 days) deaths was used. In contrast to the definitions of neonatal and post-neonatal mortality from routine statistics that are based on events recorded during a calendar year, the design of our study is a longitudinal follow-up of children born between 1992 and 2001. Risk of death in neonatal period is calculated per 1000 live births and risk of death in post-neonatal period is calculated per 1000 infants who survived the neonatal period. Information on emigration was not available but it was considered unlikely that emigration during the first year of life could bias our results appreciably.
Maternal marital status was classified as married, cohabiting, and other (single, divorced, or widowed). Maternal education was classified in four categories as follows: basic or less, secondary, higher secondary (special or vocational), and university. Categories of mother's nationality were Estonian and other. Place of residence refers to mother's permanent residence at the time of delivery and is classified into four categories (the university city of Tartu, the capital city of Tallinn, larger cities Narva, Kohtla-Järve, or Pärnu, and the remaining, mainly rural areas). Results were adjusted for maternal age in 5 year categories (<20, 20–24, 25–29, 30–34, and 35+ years), parity (0, 1, 2, 3+ previous live births and stillbirths), infant sex, and gestational age (<33, 33–35, 36–37, 38–39, 40–41, and 42+ weeks). Data on smoking were collected and analysed in three categories: ‘did not smoke during pregnancy’, ‘quit smoking during the first trimester’, and ‘continued to smoke during pregnancy’.
There were 133 734 singleton live births reported to the Estonian Medical Birth Registry in 1992–2001, among whom 1452 infant deaths were recorded up to end-2002 (risk of infant death 10.9/1000 live births). Information on birth weight, gestational age, sex of the newborn, maternal age, parity, or one or more of the social characteristics was missing in 880 births and multivariate analyses were restricted to 132 854 (99.3%) singleton live births with all data available, among whom there were 1443 infant deaths recorded up to end-2002 (risk of infant death 10.9/1000 live births). There were 2820 births with missing information on mother's smoking and analyses with further adjustments for smoking thus had to be restricted to 130 034 (97.2%) singleton live births, among whom 1386 infant deaths were recorded (risk of infant death 10.7/1000 live births).
The associations between mothers' social characteristics and the risk of neonatal and post-neonatal death was estimated as absolute risk differences and by logistic regression analysis. To detect changes in the extent of social variation over time, comparisons were made between births in 1992–1996 and 1997–2001. Analyses were carried out using the STATA statistical package.12
The study was approved by the Tallinn Medical Ethics Committee.
|
Results |
|---|
|
TopAbstractMaterial and methodsResultsDiscussionReferences |
|---|
Among 132 854 singleton live births in 1992–2001, there were 933 neonatal deaths (7.0/1000 live births) and 510 post-neonatal deaths (3.9/1000 infants surviving neonatal period). Conditions originating in perinatal period (ICD-9 codes 760–779) were the most common group of causes of neonatal deaths and congenital anomalies (ICD-9 codes 740–759) were the most common causes of post-neonatal deaths, accounting for 72 and 32% of all deaths in the given age, respectively.
Compared to 1992–1996, there were a considerably higher proportion of births to unmarried mothers, and a slightly higher proportion of births to mothers with basic education, and those living in the main urban areas in 1997–2001 (tables 1 and 2). In each subcategory of each social characteristic, neonatal mortality declined substantially between the two periods studied. The overall decline in neonatal mortality was from 8.9/1000 in infants born in 1992–1996 to 4.8/1000 in those born in 1997–2001. In Tartu, neonatal deaths declined from 4.9/1000 in infants born in 1992–1996 to 2.1/1000 in those born in 1997–2001. The respective decline in neonatal mortality in other regions combined was from 9.2/1000 to 5.1/1000. Substantial regional differences persisted on adjustment for mothers' nationality, education, marital status, parity, and age, sex, and gestational age of the infant (table 2). The decline in neonatal mortality between the two study periods was driven by decreasing number of deaths from conditions originating in perinatal period, a trend that was much more manifest in Tartu, compared to other regions. On the other hand, neonatal mortality from congenital anomalies declined only slightly and did not differ markedly between the regions.
|
|
Mother's education, nationality, and place of residence were all statistically significantly associated with risk of neonatal death in univariate analysis and also after adjustments for maternal age, parity, and infant sex (tables 1 and 2). With full adjustments for other social and birth characteristics, infants born to non-Estonian mothers had significantly higher risk of death in neonatal period in the more recent period (table 2). On further adjustment for length of gestation, the association of mother's marital status, education, and nationality with neonatal death was substantially weakened and only the association of mother's place of residence with neonatal death remained statistically significant (tables 1 and 2).
Maternal marital status and education were strongly related to risk of death in post-neonatal period. While the association of maternal education with post-neonatal death was weakened on adjustments for maternal age, parity, and infant's sex, the associations of mother's marital status and nationality with post-neonatal death became stronger after those adjustments. Multivariate analyses further adjusted for all social factors studied and length of gestation also showed a statistically significantly higher risk of death in post-neonatal period in infants born to mothers of non-Estonian nationality, in mothers with basic or less education and those not married. Place of residence was not statistically significantly associated with mortality in post-neonatal period (tables 3 and 4).
|
|
Decline in post-neonatal mortality was less marked than the decrease in neonatal deaths, although there were marked improvements in post-neonatal mortality from infectious and parasitic diseases (ICD-9 codes 001–139) and from respiratory diseases (ICD-9 codes 460–519). Although risk differences for death in post-neonatal period between infants born to different socio-economic groups generally decreased, mothers' marital status and education in particular remained strongly associated with risk of post-neonatal death (tables 3 and 4). The actual numbers of deaths in specific disease categories are too small to allow for complex multivariate analyses. However, the persisting differences in post-neonatal mortality between social groups seem to be predominantly related to higher risk of death from respiratory diseases, congenital anomalies, and injuries (ICD-9 codes 800–999) among infants born to women with lowest education, by higher mortality from respiratory diseases and unspecified conditions (ICD-9 codes 780–799) in infants of single mothers, and by higher mortality from congenital anomalies among infants of non-Estonian mothers.
The prevalence of smoking in pregnancy increased from 3.5 to 7.7% between the two study periods while the proportion of women who gave up smoking during the first trimester remained at a similar level (2.2 and 2.5%, respectively). There was a higher prevalence of smoking in pregnancy among mothers who were of non-Estonian nationality, were not married, or had lower education in both study periods. We observed a strong effect of smoking on risk of death in post-neonatal period with an OR for smokers compared to non-smokers of 2.01 (1.56–2.59) but no effect on risk of neonatal death, adjusted for all social characteristics, maternal age, parity, and infant sex and gestational age.
Additional adjustments for mother's smoking attenuated slightly the association of maternal marital status, education, and nationality with post-neonatal deaths but did not appreciably affect the social or regional variation in neonatal mortality. The greatest reduction in the odds ratio for post-neonatal death after additional adjustment for smoking was for maternal marital status and education: the fully adjusted odds ratio for mothers with basic compared to university education being reduced from 2.36 (95% confidence interval 1.25–4.44) to 1.96 (1.03–3.73) in infants born in 1997–2001.
|
Discussion |
|---|
|
TopAbstractMaterial and methodsResultsDiscussionReferences |
|---|
Estonia enjoyed substantial improvements in infant survival during the period after 1991, with particularly marked decline in neonatal mortality between 1992 and 1997. There was also a general tendency to reduction in the risk difference between social groups studied, with the exception of nationality. However, the recent improvements were not accompanied by any appreciable reduction in the social inequalities in infant death measured as risk ratios. Mother's marital status and education remain particularly strong determinants of post-neonatal death and there are emerging differences in neonatal deaths by place of residence.
The quality of birth and death registration in Estonia is considered to be adequate. Consistency checks and comparisons of birth and death records did not indicate any problems that could seriously bias the results. A 99% successful linkage of infant deaths to the birth register and a relatively small and stable proportion of missing data also suggest that this study provides valid estimates of social variation in infant deaths. We consider it unlikely that selective emigration affected the results of our study.
The results of our study are consistent with many previous findings in that the social patterns in mortality are different when the neonatal and post-neonatal period are evaluated separately, showing larger social disparities for post-neonatal death than for neonatal death.13–16 This finding of a different social pattern in neonatal and post-neonatal death is usually attributed to different causal fields: congenital malformations accounting for a substantial proportion of neonatal death, and factors related to home circumstances, and parental care influencing the risk of post-neonatal death.15
Consistent with previous findings,17,18 maternal smoking in pregnancy was associated with a higher risk of death in post-neonatal period. Information on maternal smoking in the post-partum period and during their offspring's first year of life was not available in our study. The combined effects of smoking and foetal growth on infant mortality are known to be quite complex,19 and the effect of smoking on foetal growth and infant mortality has been also shown to interact with gestational age.20 In our simplified analysis, we noted that only a part of the social and regional variation in infant deaths appeared to be due to maternal smoking. However, underreporting of smoking in pregnancy has been recently shown to be considerably more common among non-Estonian, less educated, and socially inactive women in Estonia,21 and it is thus very likely that smoking accounts for more of the social variation in infant deaths than shown by our analyses. The generally increasing proportion of women who smoke in pregnancy calls for further study and implementation of intervention strategies embracing all women of childbearing age in Estonia.
In addition to the demonstrated effects of smoking, other upstream and downstream factors involved in generating health inequality in the first year of life need to be elucidated and addressed. As injuries are generally known to cluster in subpopulations characterised by social disadvantage,22 account for a sizeable proportion of the excess deaths among infants of mothers with lowest education in our study, and are largely preventable, parental care and home environment are potential targets for further investigation and intervention.
The recent achievements in reducing neonatal and post-neonatal mortality in Estonia to levels comparable with many other European countries are quite remarkable. These overall improvements in indicators of birth outcome are noteworthy, given the mortality crisis that was affecting the adult population of Estonia at the same time. The decrease in infant mortality in Estonia took place at the time when income inequality was increasing considerably, thus contradicting the findings of positive association of infant mortality with income inequality drawn from cross-sectional comparisons.23 A dramatic decline in birth rates that occurred in the early years of transition in Estonia1 could be in fact a related indirect effect of the economic and social changes. How reproductive behaviour and fertility changed within social groups is not well understood and cannot be formally addressed in our analyses.
The increase in infant health inequality due to disproportionately greater increase in birth weight, reduction of preterm births,9 and improved survival in more advantaged social groups in Estonia after 1992 should be a cause of concern. Despite methodological problems in direct quantification of the contribution of foetal growth to inequality in infant survival, 24,25 we believe that factors affecting growth in utero are likely to account for a substantial proportion of variation in infant mortality by maternal marital status, education and nationality in Estonia. An effective strategy to further reduce infant mortality and its variation should thus include policies that address the socio-economic determinants of foetal growth.
Further exploration of regional inequalities in the quality of neonatal care and implementation of strategies to reduce them are also warranted. The overall decline in perinatal mortality in Estonia since 1990s has been attributed to establishment of modern newly equipped neonatal intensive care units in reorganised perinatal centres and introduction of a neonatal transportation system.26,27 It is plausible that the information and technological advances in obstetric or neonatal care benefit residents in the catchment area of the Tartu University Women's Clinic more than the population living in other regions of the country, especially in case of treating conditions originating in the perinatal period such as complications of pregnancy, labour, and delivery or short gestation. This would be consistent with recent findings from Sweden where mortality risk among extremely preterm infants born at general hospitals was shown to be considerably higher compared to those born at university hospitals.28 An increase in regional inequalities in neonatal care was also found in another transition country, the Czech Republic, in the early 1990s.29
Although the evidence on access to health care services in different population groups and regions in Estonia is limited it suggests that there are important geographic, financial, and information barriers to health care utilisation and in access to specialist care in particular.30,31 It is plausible that while the residents in rural areas may have adequate access to primary care through general practitioners, they may be disadvantaged in access to specialist care with potential effects on perinatal outcomes. Case studies of neonatal deaths and further inquiries into health service use adjusted for health needs would be helpful in addressing these concerns.
The risk of neonatal and post-neonatal death decreased considerably in all social groups in Estonia in the period 1992–2002. Mother's marital status and education remain strong determinants of post-neonatal death and there are emerging regional differences in neonatal mortality. Effective strategies to reduce inequality in infant deaths need to include policies addressing socio-economic determinants of foetal growth and infant survival, parents' smoking habits, and home circumstances, as well as equity in access to perinatal services.
| Key points
|
|
Acknowledgments |
|---|
We would like to thank Mall Leinsalu for comments on an earlier draft. Details of funding: The study was partly supported by the Estonian Ministry of Education and Science (target funding 0192111s02). The Swedish Council for Social Research and Centre for Health Equity Studies contributed to the funding of the analysis presented in this manuscript. IK is currently funded by the Swedish Council for Social Research. Statement of independence of researchers from funders: The funders were not involved in the study design or execution and the views and conclusions expressed in this manuscript are the responsibility of the authors.
|
References |
|---|
|
TopAbstractMaterial and methodsResultsDiscussionReferences |
|---|
1 United Nations Children's Fund, Innocenti Research Center. (2004) Innocenti Social Monitor. Florence, 2004.
2 Koppel A, Meiesaar K, Valtonen H, et al. (2003) Evaluation of primary health care reform in Estonia. Soc Sci Med 56:2461–6.[CrossRef][ISI][Medline]
3 Habicht J and Kunst AE. (2005) Social inequalities in health care services utilisation after eight years of health care reforms: a cross-sectional study of Estonia, 1999. Soc Sci Med 60:777–87.[CrossRef][ISI][Medline]
4 McKee M. (2003) Commentary: winners and losers. Int J Epidemiol 32:1087–8.
5 World Health Organization, Regional Office for Europe. European Health for All data base. Version of June 2004.
6 Leinsalu M. (2004) Troubled transitions. Social variation and long-term trends in health and mortality in Estonia. Health Equity Studies No 2(Almqvist & Wiksell International, Stockholm).
7 Leinsalu M, Vågerö D, Kunst AE. (2003) Estonia 1989–2000: enormous increase in mortality differences by education. Int J Epidemiol 32:1081–7.
8 Leinsalu M, Vågerö D, Kunst AE. (2004) Increasing ethnic differences in mortality in Estonia after the collapse of the Soviet Union. J Epidemiol Commun Health 58:583–9.
9 Koupilova I, Rahu K, Rahu M, et al. (2000) Social determinants of birthweight and length of gestation in Estonia during the transition to democracy. Int J Epidemiol 29:118–24.
10 World Health Organization. (1993) International Statistical Classification of Disease and Related Health Problems. Tenth Revision(World Health Organization, Geneva).
11 Karro H, Rahu M, Gornoi K, Baburin A. (1998) Estonian medical birth registry 1992–1994: association of risk factors with perinatal mortality. Eur J Obstet Gynecol Reprod Biol 80:151–6.[CrossRef][ISI][Medline]
12 StataCorp. (2003) Stata Reference Manual. Release 8.0(Stata Press, College Station, Texas).
13 Cnattingius S and Haglund B. (1992) Socio-economic factors and feto-infant mortality. Scand J Soc Med 20:11–3.[ISI][Medline]
14 Leon DA, Vågerö D, Otterblad Olausson P. (1992) Social class differences in infant mortality in Sweden: comparison with England and Wales. Br Med J 305:687–91.[ISI][Medline]
15 Arntzen A and Andersen AMN. (2004) Social determinants for infant mortality in the Nordic countries, 1980–2001. Scand J Public Health 32:381–9.[CrossRef][ISI][Medline]
16 Arntzen A, Samuelsen SO, Bakketeig LS, Stoltenberg C. (2004) Socioeconomic status and risk of infant death. A population-based study of trends in Norway, 1967–1998. Int J Epidemiol 33:279–88.
17 Tuthill DP, Stewart JH, Coles EC, et al. (1999) Maternal cigarette smoking and pregnancy outcome. Paediatr Perinat Epidemiol 13:245–53.[CrossRef][ISI][Medline]
18 Cnattingius S. (2004) The epidemiology of smoking during pregnancy: smoking prevalence, maternal characteristics, and pregnancy outcomes. Nicotine Tob Res 6:Suppl 2, S125–40.
19 Wilcox AJ. (1993) Birth weight and perinatal mortality: the effect of maternal smoking. Am J Epidemiol 137:1098–104.
20 Ananth CV and Platt RW. (2004) Reexamining the effects of gestational age, fetal growth, and maternal smoking on neonatal mortality. BMC Pregnancy Childbirth 4:22.[CrossRef][Medline]
21 Pärna K, Rahu M, Youngman LD, et al. (2005) Self-reported and serum cotinine-validated smoking in pregnant women in Estonia. Matern Child Health J 9:385–92.[CrossRef][ISI][Medline]
22 Brenner RA, Overpeck MD, Trumble AC, et al. (1999) Deaths attributable to injuries in infants, United States, 1983–1991. Pediatrics 103:968–74.
23 Spencer N. (2004) The effect of income inequality and macro-level social policy on infant mortality and low birthweight in developed countries—a preliminary systematic review. Child: Care, Health and Dev 30:699–709.[CrossRef]
24 Wilcox AJ and Russell IT. (1983) Perinatal mortality: standardizing for birthweight is biased. Am J Epidemiol 118:857–64.
25 Wilcox AJ. (2001) On the importance—and the unimportance—of birthweight. Int J Epidemiol 30:1233–41.
26 Gissler M, Karro H, Tellmann A, Hemminki E. (2000) Births in Finland and Estonia from 1992 to 1996: convergent differences? BJOG 107:179–85.[CrossRef][Medline]
27 Nadisauskiene RJ and Padaiga Z. (2000) Changes in women's health in the Baltic republics of Lithuania, Latvia and Estonia during 1970–1997. Int J Gynaecol Obstet 70:199–206.[CrossRef][Medline]
28 Johansson S, Montgomery SM, Ekbom A, et al. (2004) Preterm delivery, level of care, and infant death in Sweden: a population-based study. Pediatrics 113:1230–5.
29 Koupilova I, McKee M, Holcik J. (1998) Neonatal mortality in the Czech Republic during the transition. Health Policy 46:43–52.[CrossRef][ISI][Medline]
30 Kalda R, Põlluste K, Maaroos HI, Lember M. (2004) Patients' opinions on family doctor accessibility in Estonia. Croat Med J 45:578–81.[ISI][Medline]
31 Rooväli L and Kiivet RA. (2006) Geographical variations in hospital use in Estonia. Health Place 12:195–202.[CrossRef][ISI][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||