The European Journal of Public Health Advance Access published online on December 26, 2007
The European Journal of Public Health, doi:10.1093/eurpub/ckm125
Neonatal and postneonatal mortality by maternal education—a population-based study of trends in the Nordic countries, 1981–2000
Annett Arntzen1, Laust Mortensen2, Ole Schnor2, Sven Cnattingius3, Mika Gissler4 and Anne-Marie Nybo Andersen2,5
1 Faculty of Social Science, Vestfold University College, PO Box 2243, N-3303 Tønsberg, Norway
2 National Institute of Public Health, Øster Farimagsgade 5A, DK-1399 Copenhagen K, Denmark
3 Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, PO Box 281 S-17177 Stockholm, Sweden
4 STAKES, National Research and Development Centre for Welfare and Health, PO Box 220, FIN-00531 Helsinki, Finland
5 Department of Epidemiology, University of Southern Denmark, JB Winsløw Vej 9B, DK-500 Odense, Denmark
Correspondence: Annett Arntzen, Faculty of Social Science, Vestfold University College, PO Box 2243, N-3303 Tønsberg, Norway, tel: +47 33031201, fax: +47 33031105, e-mail: annett.arntzen{at}hive.no
Received May 29, 2007, accepted December 3, 2007
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionConclusionSupplementary DataAcknowledgementsReferences |
|---|
Background: This study examined changes in the educational gradients in neonatal and postneonatal mortality over a 20-year period in the four largest Nordic countries. Methods: The study populations were all live-born singleton infants with gestational age of at least 22 weeks from 1981 to 2000 (Finland 1987–2000). Information on births and infant deaths from the Medical Birth Registries was linked to information from census statistics. Numbers of eligible live-births were: Denmark 1 179 831, Finland 834 299 (1987–2000), Norway 1 017 168 and Sweden 1 971 645. Differences in mortality between education groups were estimated as risk differences (RD), relative risks (RR) and index of inequality ratio (RII). Results: Overall, rates of infant mortality were in Denmark 5.9 per 1000 live-births, in Finland 4.2 (1987–2000), in Norway 5.3 and in Sweden 4.7. Overall the mortality decreased in all educational groups, and the educational level increased in the study period. The time-trends differed between neonatal and postneonatal death. For neonatal death, both the absolute and relative educational differences decreased in Finland and Sweden, increased in Denmark, whereas in Norway a decrease in absolute differences and a slight increase in relative differences occurred. For postneonatal death, the relative educational differences increased in all countries, whereas the absolute differences decreased. Conclusions: All educational groups experienced a decline in infant mortality during the period under study. Still, the inverse association between maternal education and RR of postneonatal death has become more pronounced in all Nordic countries.
Keywords: educational level, neonatal mortality, Nordic countries, postneonatal mortality, time-trends
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionSupplementary DataAcknowledgementsReferences |
|---|
Infant mortality is widely used as an indicator of public health, quality of health services, distribution of wealth and the general standard of living in a society.1,2 A consistent long-term investment in social welfare and health care has gone alongside a substantial reduction in infant mortality in the Nordic countries in the 20th century.3
Despite low risk of infant death and relatively small social differences in the Nordic countries, previous studies have observed constant and even increasing social inequalities in infant mortality.3–5 It has been difficult to draw inferences from the earlier comparisons between the countries, since different inclusion criteria as well as measures and classifications of social position were used.6,7 In addition, the social pattern differ between neonatal- and postneonatal death.5
To elucidate any systematic differences between the Nordic countries in the association between socio-demographic factors and infant mortality and changes over time, we have taken advantage of the population-based registers and constructed a dataset for comparative analysis. Due to a unique personal identification number system, it has been possible to link parental census information to data in the Medical Birth Registries. This study represents the largest Nordic study of the association between educational level and infant mortality and it's development in Denmark, Finland, Norway and Sweden over the past two decades.
More specifically, we aimed to assess if the educational gradients in neonatal and, postneonatal mortality were increasing, stable or decreasing in the four Nordic countries between 1981 and 2000.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionSupplementary DataAcknowledgementsReferences |
|---|
Study population
The Medical Birth Registries in the Nordic countries contain information on all live-births and fetal deaths. All live-born singletons from 1981 to 2000 (Finland 1987–2000) with a gestational age of 22 weeks or more were studied. In all, we included from Denmark 1 179 831 infants, from Finland 834 299 infants, from Norway 1 017 168 infants and from Sweden 1 971 645 infants. These registry data were linked to information from national registers and censuses on parental educational levels and other socio-demographic factors. The infants were allocated to four time periods, each of them 5 years.
Variables
Infant (0–364 days) death was divided into neonatal (0–27 days) death and postneonatal (28–364 days) death, which were used as outcome variables. Neonatal and postneonatal mortality rates were expressed as deaths per 1000 live-born infants. The number of infant death was 6956 in Denmark (4330 neonatal and 2626 postneonatal deaths), 3504 in Finland (2339 neonatal and 1165 postneonatal deaths), 5433 in Norway (3118 neonatal and 2315 postneonatal deaths) and 9224 in Sweden (5673 neonatal and 3551 postneonatal deaths).
The independent variables were maternal education, maternal age and parity. Information on educational level was obtained from national educational registries and censuses, and refers to the number of years of education attained by year 2000 by all mothers in the study. The categorization of education was based on the International Standard Classification of Education (ISCED).8 Three groups were distinguished: high education refers to more than 12 years of education (ISCED codes 5–8), medium education refers to 10–12 years education (ISCED codes 3–4) and low education refers to less than 10 years education (ISCED codes 0–2). In Finland, the group with less than 10 years of education also consists of mothers with unknown education, since the basic, compulsory education of 9 years was not registered in the Finnish Education Register. Mother's age at delivery was categorized as younger than 20 years, 20–34 years and 35 years or older. Parity was dichotomized as primiparity and multiparity. (Supplementary data)
Statistical methods
The association between maternal education and the risk of neonatal and postneonatal death was estimated as absolute risk differences (RD) using linear regression, while relative risks (RR) and relative index of inequality (RII) were estimated using log binomial regression.9,10 The RII was calculated by assigning to each educational category a score corresponding to the midpoint of the category's range in the cumulative distribution of education in the population a given year.11 Consequently, all values of the score variable fell within the range of 0 and 1 and, included as a continuous variable in the log-binomial regression. The RII can be interpreted as the RR between the lowest and highest education. Tables 1 and 2 present crude risk measures of the relation between maternal educational level and neonatal and postneonatal mortality, respectively. Tables 3 and 4 also present the adjusted RR, where both the maternal age and parity were included in the model in addition to maternal education. All results are presented together with 95% confidence intervals. SAS statistical software was used in the analyses.
|
|
|
|
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionSupplementary DataAcknowledgementsReferences |
|---|
Characteristics of the study population are described in Supplementary data. The number of births among teenagers decreased markedly during the last decades, especially in Denmark and Norway and the proportion of older (35 years or more) mothers increased in all countries. The educational level increased considerably from the 1981 to 2000 in Denmark, Finland and Norway. The proportion of mothers with at least 13 years of education increased from 28.8% to 31.9% in Denmark, from 38.6% to 44.5% in Finland (1987–2000) and from 30.3% to 38.4% in Norway. The same tendencies were found for the fathers (data not shown). In contrast, there were virtually no changes over time in percentage of mothers and fathers with high education in Sweden.
Figure 1 shows the decrease in infant mortality rates in the period 1981–2000 in Denmark, Finland, Norway and Sweden. From the 1990s, Finland, Sweden and Norway have had fairly similar infant mortality rates while the rate was considerably higher in Denmark.
|
Figure 2 shows the decline in neonatal and postneonatal mortality rates between earliest and latest period under study by maternal education in all four Nordic countries. Infants to mothers with low education generally had the largest decrease in neonatal and postneonatal over the whole study period. The exceptions are neonatal mortality in Denmark and postneonatal mortality in Sweden, where mortality of infants with low educated mothers showed less favourable developments compared with mortality of infants to mothers with high education.
|
For neonatal mortality, both absolute and relative measures showed similar trends (table 1). In Denmark and Finland, the differences by maternal education were largest in 1991–95, whereas in Norway and Sweden the differences were largest in 1986–90. In the last period (1996–2000) both the absolute and relative educational differentials for neonatal mortality were smaller in Finland and in Sweden compared with the earliest period, whereas the differences increased in Denmark. In Norway, a substantial decrease occurred for absolute differences, but the relative differences increased slightly.
For postneonatal mortality, the absolute and relative differences peaked at different periods (table 2). In Denmark and Norway the absolute differences peaked in 1986–90 and in Finland and Sweden the RD were largest in 1991–95. The relative and absolute differences in postneonatal mortality peaked at the same time only in Finland, whereas in Denmark the relative differences were highest in 1991–95. In Norway and Sweden, the relative differences increased continuously throughout the study period. Compared with the earliest period, the relative educational differences in postneonatal mortality had increased in all Nordic countries by the years 1996–2000, whereas the absolute differences decreased in Denmark, Finland and Norway.
We found a persistent inverse association between maternal educational level and RR of neonatal death, and adjusting for maternal age and parity did generally not change these estimates (table 3). However, for mothers with middle education in Denmark 1991–95 and for low educated mothers in Finland in 1996–2000, adjusting for maternal age and parity increased the risk of neonatal mortality to a statistically significant level.
Table 4 shows a persistent inverse association between mothers’ educational level and postneonatal mortality. The RR reveal an increasing inverse association between educational level and postneonatal mortality in Denmark, Norway and Sweden. In Finland the highest risk ratio for low educated mothers was found in the period 1991–95. When maternal age and parity were included in the model, the RR generally decreased, but remained statistically significant. These arrenuation of risk were largely an effect of maternal age (data not shown).
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionSupplementary DataAcknowledgementsReferences |
|---|
The risk of infant death decreased considerably in all educational groups in the Nordic countries from 1981 to 2000. Nevertheless, the inverse association between maternal education and infant death persists throughout the whole period of observation. For neonatal mortality, both absolute and relative differences by maternal education decreased in Finland and Sweden, but increased in Denmark. In Norway there was a decrease in absolute differences in neonatal mortality, and a slight increase in relative differences. For postneonatal mortality, the relative differences by maternal education had increased in all Nordic countries, whereas the absolute differences had decreased. Another important finding was that the reduction in absolute risk from the 1980s to 2000 was highest for infants to low and medium educated mothers, compared with those whose mothers had high education.
The size and completeness of the longitudinal dataset is a major strength in this study. It includes both medical and social information about the total population of live-born infants. Infant mortality is considered to be 100% ascertained in the Nordic Medical Birth Registries.12–15 The information on educational level from Censuses and Educational registers have high ascertainment and high reliability.4 The included variables have been recorded in a consistent way over time by the Medical Birth Registries and official statistics. Thus, it is unlikely that the results can be explained by selection bias, changes in coding or classifications. All updated coding and classification systems have been retrospectively implemented in the national register systems in all study countries for the whole study period.
Acquiring the information of mothers’ education only at one point of time, i.e. in 2000, can be considered as a weakness of this study, as the educational level measured in 2000 may differ from the attained education at the time of giving birth. If the highest educational level was completed after birth was given, then we likely have overestimated the level of education and underestimated the educational differences. It has become more common to complete secondary education or start with higher education later in life and we do not know whether those mothers who continue with their educational career after years of break share the same characteristics with those who continue their educational career on continuous basis. However, in Finland and Denmark we were able to repeat the study by measuring maternal education the year before birth, and we observed only minor changes in the estimates. Therefore, and also because of the increased maternal age, we find it unlikely that our results are seriously biased.
Different measures of inequality complement each other as indicators of inequality trends.16 The RD measures the absolute difference in mortality rates between mothers with low and high education, while the RR measures the relation between two groups. The RII was the only measure used that explicitly allows for the changes in the distribution of educational levels, and measures the difference between subjects at the top and bottom of the educational hierarchy. The overall level of education increased over time, and the number of low educated mothers and their proportions of all mothers have therefore decreased during the study period. Thus, we considered that both absolute and relative measures are important to consider in the context of this study.
Socioeconomic status (SES) can be measured by income, educational level or by the status on the occupational hierarchy (i.e. social class). Our dataset include educational level, income and classifications according to the SES. However, national tax and social benefit system and policies vary, and each county has their own social class schemes. In addition, these variables are mostly only available from Censuses (every 5th or 10th year). In case only one social indicator can be chosen, the level of education is recommended in many circumstances, because it is easy to measure and is considered to be a very important health determinant.16 Previous research has shown that maternal education has a great importance also for pregnancy outcomes.17,18
It is well known that education is an indicator of social, environmental and cultural conditions. Also the acquisition of health related knowledge, optimized use of health services and willingness to invest in human capital differ according to level of education.19 High education may act through the promotion of high self esteem and self efficacy and the ability to solve problems.20 Disadvantageous groups are more exposed to difficult life events or difficult life conditions, and have less social support to limit the impact of such stressors.21–23 Due to the overall increased level of education in the population, there are reasons to believe that mothers with less than 10 years of education has become a more selected group in several ways in the late 1990s compared with one or two decades earlier. These women may have the highest occurrence of negative lifestyle factors associated with infant mortality, such as poor living conditions, poor nutrition, smoking and substance use.
Neonatal deaths have been linked to the quality of obstetric and neonatal care, including availability of experienced staff around the time of birth. Neonatal deaths are attributable to prematurity, birth defects and other diseases and conditions of the newborn. Totally or almost free prenatal, obstetric and infant health care is available for all, and the utilization of these services is very high in the Nordic population. Work-related psychological strain seems to be smaller in countries with highly developed social support systems, and occupations with especially high workload and health hazards have declined during the study period.24,25 Thus, the assumption is close that socioeconomic factors have a less influence on neonatal mortality compared with deaths in later periods, and that high quality obstetric health services and the provision of social welfare services somewhat override social inequalities. However, the obtained differences in neonatal mortality by maternal education (the lowest educated had 19–72% higher risk compared with the highest educated) as well as differences between Denmark and the other Nordic countries indicate that the results may not only be explained by differences in quality of care. Other social differences, including risk of preterm birth and risk factors for preterm birth and neonatal mortality, such as smoking and genital infections may be alternative explanations.24,26,27
Postneonatal mortality have repeatedly been reported to be more closely associated with SES than neonatal mortality.1,5,7,28 The causes of death differ between newborns and older infants. In Norway the social inequality has widened for postneonatal deaths caused by sudden infant death syndrome (SIDS) and infections from the 1970s to the 1990s.29 It is intriguing to consider how the causes of death might differ between the Nordic countries in different education groups.
Structural changes have taken place in all four Nordic countries in the period under study.Sweden and especially Finland faced a serious economic recession in the early 1990s. In this period the unemployment rate in Finland jumped from 3–4% to 17% and in Sweden from 2–3% to 8%. It is likely that this recession affected the everyday life of people.24 The absolute and relative risk differences between mothers with low and high education in both neonatal and postneonatal mortality were especially high in Finland in the period 1991–95. In Sweden the absolute RD in the postneonatal period was also highest in this time period. The unemployment level in Denmark was already relatively high in the early 1980s, rose again in somewhat in the early 1990s, but cannot compare with the sudden and dramatic changes in Finland and Sweden. Although unemployment also increased in Norway, it remained clearly below the levels found in the other Nordic countries. However, the RD in postneonatal mortality between mothers with low and high education was especially high in Denmark and Norway in 1986–90, the same period for the peak of unemployment in these countries.4,25
At the time of economic recession, infant mortality decreased more for infants whose mothers had highest education, whereas the mortality level decreased much less or even increased for mothers with lowest education, resulting in very high educational differences at that period. In our study, however, we cannot address whether these changes in risks were the effect of the economic recession or some other factors.
The structural inequalities between the countries cannot satisfactory explain the pattern of the social gradient in infant deaths from 1981 to 2000, and infant death might not be the best indicator of the general health conditions and the welfare arrangements of a society when the risk of infant death is low. Nevertheless, data from Finland shows that the relationship between educational level and infant mortality might not be independent from structural factors at a societal level. However, the decreased absolute and relative differences in neonatal mortality in Finland and Sweden and the increasing social gradient in postneonatal mortality, foremost pronounced for Norway (measured by RR and RII), implies a need for a broader examination of public health performances for specific marginalized groups. The two decades under study were turbulent as regards to structural changes, economic conditions and the occupational structures, which likely intertwined with the continuing marginalization process. Thereby, postneonatal death might be the most important indicator reflecting individual conditions rather than societal structural phenomena.
|
Conclusion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionSupplementary DataAcknowledgementsReferences |
|---|
In summary, this population-based longitudinal study confirms the known association between education and infant mortality. Also, it underlines the importance of distinguishing between neonatal- and postneonatal mortality and of using different risk measures in social epidemiology. Finally, it shows that important socially patterned inequalities persist even in populations with a very low level of neonatal and postneonatal mortality.
|
Supplementary Data |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionSupplementary DataAcknowledgementsReferences |
|---|
Supplementary Data are available at EURPUB Online.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionSupplementary DataAcknowledgementsReferences |
|---|
The NorChase data collection was funded by the Research Programme in Longitudinal Epidemiology, which is supported by the Nordic Council of Ministers and administered by NordForsk (The Nordic Research Board). We thank Sven Ove Samuelsen, National Institute of Public Health Norway, for help with the statistical analysis.
Conflicts of interest: None declared.
Key points
|
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionSupplementary DataAcknowledgementsReferences |
|---|
1 Antonowsky A, Bernstein J. Social class and infant mortality. Soc Sci Med (1997) 11:453–70.
2 Stockwell EG, Swanson DA, Wicks JW. Temporal variations in the relationship between infant mortality and economic status. Soc Indic Res (1998) 20:217–27.[CrossRef]
3 Health Statistics in the Nordic Countries 2003. Theme Section: Children's Health. Copenhagen: NOSOSCO 73:2005.
4 Arntzen A, Mortenson LH, Cnattingius S, et al. Trends in social indicators in the Nordic countries, 1980–2002. In: Copenhagen: NORCASE - National Institute of Public Health (2006).
5 Arntzen A, Samuelsen SO, Bakketeig LS, Stoltenberg C. Socioeconomic status and risk of infant death. A population-based study of trends in Norway, 1967–1998. Int J Epidemiol (2004) 33:279–88.
6 Arntzen A, Nybo-Andersen AM. The association between social factors and infant mortality in the Nordic countries, 1980–2001. Scand J Public Health (2004) 32:381–9.[CrossRef][Web of Science][Medline]
7 Cnattingius S, Haglund B. Socio-economic factors and feto-infant mortality. Scand J Soc Med (1992) 20:11–3.[Web of Science][Medline]
8 International Standard Classification of Education (ISCED). UNESCO. (1997).
9 Spiegelman D, Hertzmark E. Easy SAS calculations for risk or prevalence ratios and differences. Am J Epidemiol (2005) 162:199–200.
10 Mackenbach JP, Kunst AE. Measuring the magnitude of socio-economic inequalities in health: an overview of available measures illustrated with two examples from Europe. Soc Sci Med (1997) 44:757–71.[CrossRef][Web of Science][Medline]
11 Harper S, Lynch J. Measuring Health inequalities. In: Methods in Social Epidemiology—Oakes JM, Kaufman JS, eds. (2006) San Francisco: Josey-Bass/Wiley.
12 Juel K, Helweg-Larsen K. The Danish registers of causes of death. Dan Med Bull (1999) 46:354–7.[Web of Science][Medline]
13 Gissler M, Teperi J, Hemminki E, Meriläinen J. Data quality after restructuring a nationwide medical birth registry. Scand J Soc Med (1995) 23:75–80.[Web of Science][Medline]
14 Births in Norway through 30 years (1967–1996). Bergen: Medical Birth Registry of Norway. (1997).
15 The Swedish Medical Birth Register - a summary of content and quality. Epidemiologic Center, National Borad of health and Welfare, Stockholm, Sweden.och finns bara online: http://www.sos.se/fulltext/112/2003-112-3/2003-112-3.pdf.
16 Kunst AE, Bos V, Mackenback JP. Guidelines on monitoring socio-economic inequalities in health in European countries (2001) Rotterdam: Erasmus University.
17 Arntzen A, Magnus P, Bakketeig LS. Different effects of maternal and paternal education on early mortality in Norway. Paediatr Perinat Ep (1993) 7:376–86.[CrossRef]
18 Rantakallio P. Social background of mothers who smoked during pregnancy and their influence on the offspring. Soc Sci Med (1979) 13A:423–9.[Web of Science]
19 Davey Smith G, Hart C, Hole D, et al. Education and occupational social class: which is the more important indicator of mortality risks? J Epidemiol Community Health (1998) 52:153–60.[Abstract]
20 Pincus T, Challahan LF. Associations of low educational level and poor health status: behavioural, in addition to demographic and medical, explanations? J Clin Epidemiol (1994) 47:355–61.[CrossRef][Web of Science][Medline]
21 Oliveus G, Ostergren PO, Hanson BS, Lyttkens CH. Parental economic stress: evidence of an overlooked public health risk among Swedish families. Eur J Public Health (2004) 14:354–60.
22 Hoffman S, Hatch MC. Stress, social support and pregnancy outcome: a reassessment based on recent research. Paediatr Perinat Ep (1996) 10:380–405.[CrossRef]
23 Dahl E, Elstad JI. Recent changes in social structure and health inequalities in Norway. Scand J Public Health (2001) 29((Suppl 55):):7–17.
24 Henriksen TB, Hedegaard M, Secher NJ. The relation between psychological job-strain, and pre-term delivery and low birth-weight for gestational age. Int J Epidemiol (1994) 23:764–74.
25 Lahelma E, Lundberg O, Manderbacka K, et al. Health inequalities in the Nordic countries from the 1980s to the 1990s. Scand J Public Health (2001) 9((Suppl 55):):1–5.[Medline]
26 Kramer MS, Goulet L, Lydon J, et al. Socio-economic disparities in preterm birth: causal pathways and mechanisms. Peadiatr Perinat Ep (2001) 15:104–23.[CrossRef]
27 Bjørk C, Mortensen LH, Morgen CS, et al. Socioeconomic inequality in preterm birth. a comparative study of the Nordic Countries from 1981 to 2000. (In press).
28 Olsen O, Madsen M. Effects of maternal education on infant mortality and stillbirths in Denmark. Scand J Public Health (1999) 27:128–36.[CrossRef][Web of Science][Medline]
29 Arntzen A, Samuelsen SO, Daltveit AK, et al. Socioeconomic factors and cause-specific postneonatal death in Norway, 1969–1995. Int J Epidemiol (2006) 35:1083–9.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  E. Hemminki Perinatal health in the Nordic countries -- Current challenges Scand J Public Health, September 1, 2009; 37(7): 671 - 673. [PDF]
|
|
|
|
|
|
M. Gissler, O. Rahkonen, L. Mortensen, A. Arntzen, S. Cnattingius, A.-M. Nybo Andersen, and E. Hemminki Sex differences in child and adolescent mortality in the Nordic countries, 1981--2000 Scand J Public Health, June 1, 2009; 37(4): 340 - 346. [Abstract] [PDF]
|
|
|
|
 |
|
 C. Bille and A.-M. N. Andersen Preconception care BMJ, February 12, 2009; 338(feb12_2): b22 - b22. [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||