Article Text

Download PDFPDF

A comparative study of educational inequality in the risk of stillbirth in Denmark, Finland, Norway and Sweden 1981–2000
  1. Ane L Rom1,
  2. Laust H Mortensen1,2,
  3. Sven Cnattingius3,
  4. Annett Arntzen4,
  5. Mika Gissler5,6,
  6. Anne-Marie Nybo Andersen2,7
  1. 1National Institute of Public Health, University of Southern Denmark, Copenhagen, Denmark
  2. 2Division of Epidemiology, Institute of Public Health, University of Southern Denmark, Odense, Denmark
  3. 3Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden
  4. 4Faculty of Social Science, Vestfold University, Tønsberg, Norway
  5. 5National Institute for Health and Welfare, Helsinki, Finland
  6. 6Nordic School of Public Health, Gothenburg, Sweden
  7. 7Department of Public Health, University of Copenhagen, Denmark
  1. Correspondence toProfessor Anne-Marie Nybo Andersen, Department of Public Health, University of Copenhagen, Oster Farimagsgade 5, P.O. Box 2099 DK-1014 Copenhagen, Denmark; amny{at}


Background The stillbirth rates in Denmark, Finland, Norway and Sweden are among the lowest in the world, but socioeconomic disparities in stillbirth still exist. This study examined the educational patterns in the risk of stillbirth in Denmark, Finland, Norway and Sweden from 1981 to 2000.

Methods From the national birth registries, all singleton live births and stillbirths with a gestational age of at least 28 weeks were selected in Denmark (n=1 182 888), Finland (n=419 729), Norway (n=1 006 767) and Sweden (n=1 974 101). The births were linked with individual data on parental socioeconomic factors from various national registers. Linear and logistic regression were used to calculate RR and risk differences for stillbirth according to maternal educational attainment.

Results The risk of stillbirth was lowest in Finland and highest in Denmark. The risk decreased over time in Denmark, Norway and Finland, but remained stable in Sweden. Educational gradients were found in all countries in all time periods under study. In Denmark, the gradient remained stable over time. In Norway the gradient decreased slightly during the 1990s, whereas the gradient increased in Sweden. The gradient in Finland was relatively stable.

Conclusions There were persisting educational inequalities in stillbirth in Denmark, Finland, Norway and Sweden in the 1980s and 1990s. Inequalities were stable or decreasing except in Sweden, where an increase in inequality was observed. This increase was not solely attributable to a decreasing absolute risk of stillbirth as both the relative and absolute measures of inequality increased.

  • Comparative study
  • Nordic countries
  • perinatal epidemiology
  • social epidemiology
  • socioeconomic factors
  • stillbirth

Statistics from

Stillbirth rates have slightly decreased in most western countries during the past few decades and the stillbirth rate in Denmark, Finland, Norway and Sweden are among the lowest in the world.1 Nevertheless, socioeconomic disparities in stillbirth still exist. A recent literature review on social inequality in fetal and perinatal death in the Nordic countries concluded that social differences in stillbirth were present in all studies from Norway and Finland, in the majority of the Danish studies, whereas the findings in the Swedish studies were mixed.2

The Nordic welfare states are relatively similar in social structures, populations and ways of living.3 In addition, antenatal and obstetric care is standardised and free of charge. This relative homogeneity between the countries makes them well suited for comparative studies. However, only few comparative studies of socioeconomic disparity in the risk of stillbirth in the Nordic countries have been made. A review by Cnattingius and Haglund4 reported that the socioeconomic inequalities in late fetal death were smaller in Finland and Sweden compared with Denmark and Norway. Bakketeig et al5 compared Denmark, Norway and Sweden in three different time periods and found similar results. Finally, in a register study by Knudsen and Källén,6 the authors found no clear socioeconomic differences in stillbirth risk between Denmark and Sweden. The conclusions of the aforementioned studies were limited due to the use of secondary data, non-comparable populations, different study periods and different definitions of stillbirth and socioeconomic position.4–6

In the present study, we used a dataset with comparable information on sociodemographic variables and birth data from Denmark, Finland, Norway and Sweden. The aim was to study the social patterns in the risk of stillbirth, by comparing educationally patterned disparities in the risk of stillbirth between the countries and within the countries over time. Such a description might point towards potentials for the prevention of socioeconomic inequality in the risk of stillbirth.

Materials and methods

Data and study population

This study is based on the Nordic Collaborative Project on Health and Social Inequality in Early Life (NorCHASE) database. The database was created to enable comparative studies on social inequality in perinatal outcome and early life, and is based on data from national population-covering registries and censuses in Denmark, Finland, Norway and Sweden. Details of the included registers have been reported elsewhere.7–9 The database contains information on all women who gave birth in Denmark, Norway and Sweden from 1981 to 2000. In accordance with data protection regulations, the available Finnish data comprised data on all stillborn and deceased children and a random 50% sample of all other births in Finland in the period from 1987 to 2000. To account for the sampling design of these data, we used inverse probability weighted regression to analyse the Finnish data. A probability weight of two was assigned to infants from the random sample. All deceased infants who had a 100% probability of being selected were assigned weights of one.

We included all singleton live births and all singleton stillbirths with a gestational age of at least 28 completed gestational weeks during the period 1981 to 2000 in Denmark, Norway and Sweden, and in Finland from 1987 to 2000. A total of 5 010 528 births (Denmark n=1 182 888, Norway n=1 018 572, Sweden n=1 974 134, Finland n=420 633 (not weighted)) were included. In the analysis we excluded births with missing values on maternal age and maternal parity. The main analysis was thus based on a population of 1 182 888 Danish, 419 729 Finnish (not weighted), 1 006 767 Norwegian and 1 974 101 Swedish births.

Definition of outcome and exposure measures

The outcome measure was stillbirth, defined as the birth of a child with a gestational age of 28 or more completed weeks that showed no life signs, which was the stillbirth definition used in Denmark and Sweden during the study period. Finland also registers stillbirths between 22 and 27 weeks and Norway between 12 and 27 weeks, but for comparative reasons these births were not included in our analyses.

Maternal educational attainment was chosen as the indicator of socioeconomic position. To get comparable data we defined the variable as the highest completed education in 2003. Education was categorised according to the international ISCED classification system and converted into three educational groups: less than 10 years (pre-primary, primary and lower secondary), 10–12 years (upper secondary, post-secondary and non-tertiary), 13 or more years (tertiary).10 In Finland, only higher than compulsory education was recorded in the national register. As a consequence, missing values in the Finnish data were coded as an education of 10 years or less. Maternal age (<20 years, 20–34 years, ≥35 years) parity (0, 1+) and year of birth (1981–5, 1986–90, 1991–5, 1996–2000) were chosen as covariates.

Statistical analysis

Due to data protection requirements, all analyses were made for each country separately. Selected characteristics of the study population were described in 5-year intervals. In addition, time trends in the stillbirth rate (proportion of stillbirths per 1000 births) in different categories of the educational attainment, maternal age and parity groups were tabulated.

We used linear and logistic regression to model the association between maternal educational attainment and stillbirth. Simple and multiple regression analyses were made, and the association was reported as RR and risk differences (RD) with 95% CI. Because the risk of stillbirth was low, we used OR estimated with logistic regression to approximate the RR. Time trends in the educational level–stillbirth association were examined across 5-year intervals. In the multiple regression models, we included maternal age and parity in addition to educational attainment. Tests for interaction between year of birth and maternal educational level were examined in a model that included terms for maternal education, period and a period-by-maternal education interaction term. The p values reported in table 1 represent the likelihood of observing the data (or more extreme) if the true value of this interaction term is zero. Significance was based on likelihood ratio tests and the tests were considered statistically significant at p<0.05.

Table 1

Unadjusted risk estimates (RR and RD) of the association between maternal educational attainment and stillbirth: singleton births with a gestational age of 28 weeks or more in Denmark, Finland, Norway and Sweden, 1981–2000

To examine the robustness of our findings, we repeated all analyses including all pregnancies, ie, both single and multiple pregnancies, on primiparous women, on a subcohort restricted to women born in same country as she was giving birth in, and on a subsample of women aged 25–34 years, respectively. All results were presented in 5-year intervals reflecting 1981–5, 1986–90, 1991–5, and 1996–2000.


The distributions of maternal educational attainment, maternal age and parity are presented in 5-year groups for the four countries (table 2). During the study period, the maternal educational attainment increased in all four countries, but with some differences. Denmark had the lowest proportion of women with more than 12 years of education in the first 5-year period and experienced the largest increase in educational attainment. The proportion of Danish women with less than 10 years of education decreased with more than 10 percentage points during the study period. In Sweden, the proportion of women with less than 10 years of education decreased, while the proportion with more than 12 years remained constant. Norway and Finland experienced a stable growth in the proportion of women with more than 12 years of education. The proportion of teenage mothers decreased in all four countries, while the proportion of mothers aged 35 years or more increased during the study period.

Table 2

Selected sociodemographic characteristics of the study population: singleton births with a gestational age of 28 weeks or more in Denmark, Finland, Norway and Sweden, 1981–2000

The stillbirth rate decreased during the study period in Denmark, Finland and Norway (table 3), but in Sweden the rate was relatively stable. The largest decrease was observed in Norway: from 5.2 per 1000 births in 1981–5 to 3.6 per 1000 births in 1996–2000. Compared with the other countries, the lowest rate was observed in Finland from 1987 to 2000. Women between 20 and 34 years of age had the lowest stillbirth rate. Women aged 35 years or more and primiparous women had a higher prevalence of stillbirth compared with younger women and parous women, respectively. The high risk of stillbirth among those with no information on parity in Finland seems to suggests that Finnish recording practices depend on the pregnancy outcome so that parity is not recorded as carefully in pregnancies resulting in a stillbirth when compared with normal pregnancies.

Table 3

Risk of stillbirth in the study populations (n/1000 births): singleton births with a gestational age of 28 weeks or more in Denmark, Finland, Norway and Sweden, 1981–2000

An educational gradient in stillbirth was found in all countries and in all periods under study.

The unadjusted estimates of the RR and RD for the association between maternal educational attainment and stillbirth are shown in table 1. We found a clear educational gradient in stillbirth in all countries in all four periods. The adjusted RR are presented in figure 1 where time trends can be observed. Adjustment for age and parity did not attenuate the estimates, if anything the estimates increased marginally. The only statistically significant changes in the RD/RR over time were observed in Sweden. Judging from the point estimates, the educational differences in the risk of stillbirth tended to decrease over time in Norway. Compared with women with more than 12 years of education, the adjusted RR for women with less than 10 years of education was 1.62 in the first period, 1.71 in 1986–90, and during the 1990s it decreased to 1.38. In Sweden the adjusted RR increased from 1.40 to 2.10 during the period under study, while in Denmark the educational differences were relatively stable with an estimated RR of approximately 1.5. In Finland, the RR of stillbirth increased for women with 10–12 years of education and fell for women with less than 10 years of education compared with women with more than 12 years of education. In the beginning of the period Sweden had the weakest educational gradient compared with the other countries. Nevertheless, Sweden had the strongest gradient by the end of the period (figure 1). As expected, the RD displayed similarly a clear educational gradient during the period in all four countries. In Sweden the difference in risk was 2.4 per 1000 births between women with less than 10 and more than 12 years of education at the end of the study period, compared with almost 1.0 per 1000 in Finland and Norway and 1.6 per 1000 in Denmark (table 1).

Figure 1

Time trends in the adjusted RR of stillbirth according to maternal educational attainment. Singleton births with a gestational age of 28 weeks or more in Denmark, Finland, Norway and Sweden, 1981–2000. For each country the left line represents the adjusted risk estimates for women with less than 10 years of education and the right line represents the risk of women with 10–12 years of education compared with women with more than 12 years of education. Estimates are adjusted for maternal age and parity and shown with 95% confidence limits.

To test if changes in the inclusion criteria affected the results, we conducted a set of sensitivity analyses (data not shown). When multiple pregnancies were included the RR estimates attenuated slightly. Restricting the analysis to only primiparous women in age group 24–35 years and to women who were born in the Nordic countries, respectively, did not change the conclusions from the main analyses. Education was assessed at the end of the period, which means that the education might have had been achieved after childbirth, which introduces the possibility of reverse causation in the form of an effect of stillbirth on educational achievement. To examine this we looked at the Finnish data, in which additional information on education in the year of the birth of the child was available; 81% of the women with a short education, who did not experience a stillbirth, did not increase their level of education during the follow-up. The corresponding number for women who did experience a stillbirth was 84%, but the difference was not statistically significant. This might, however, suggest that there is a small influence of stillbirth on later educational attainment or that there are unobserved common causes of stillbirth and educational attainment. When we compared the two measures of education as predictors of stillbirth, the association is somewhat stronger when the highest completed education was used when compared with education in the year before the birth of the offspring.


Our study shows that an educational gradient in stillbirth existed in Denmark, Finland, Norway and Sweden from 1981 to 2000. In Denmark, the gradient remained constant during the period, in Finland the difference increased between the highest educated and the rest, whereas the difference between women with 10–12 years of education and women with less than 10 years of education diminished. In Norway the gradient decreased during the 1990s. Only in Sweden did changes in the association between maternal education and the risk of stillbirth reach conventional levels of statistical significance. In Sweden the risk of stillbirth related to low maternal education increased during the end of the 1990s. In 1996–2000 the adjusted RR for Swedish women with less than 10 years of education was 2.10 (95% CI 1.78 to 2.48) compared with women with more than 12 years of education.

This large population-based register study included more than five million births in four countries over a 20-year period. Stillbirths are considered 100% ascertained in the Nordic medical birth registries,11–14 and the information on educational level from censuses and educational registers have high ascertainment and reliability.7 The size and completeness of the dataset are major strengths, and the study, to our knowledge, is the first of its kind. We used data from a large number of national registries and the almost complete coverage should preclude selection bias. As no information in this study was collected by self-report, the information was not sensitive to information bias either. Another strength of this study is the presentation of both RR and RD as association measures. The proportion of missing data on covariates was generally low. In Finland, where only higher than compulsory education was recorded in the national register, missing values were coded as an education of 10 years or less (table 2). This category consequently included both women with no education, unknown education, education achieved in other countries and education less than 10 years, which may imply some misclassification, probably of the non-differential type. It has to be acknowledged that our data do not permit us to study the interplay between the timing of education and the timing of pregnancy in any detail.

For comparative reasons we measured maternal education at the end of the study period, which makes reverse causation possible. It is reasonable to assert that a stillbirth will influence life chances and decisions in life. However, results from the subset of women aged 25–34 years, who were likely to have achieved their highest education were only marginally different from the main analysis. Using the Finnish dataset, which had information on educational attainment in the year before the birth of offspring, we examined whether women who experienced a stillbirth had a decreased risk of completing a higher education afterwards, and found that a stillbirth was associated with a statistically non-significant decrease in the risk of increasing one's educational attainment. It is thus not plausible that reverse causation would explain the findings.

Several risk factors for stillbirth have been recognised.15–20 Cnattingius and Stephansson21 recently reviewed the literature on the epidemiology of stillbirth and found that major maternal risk factors of stillbirth include high maternal age, smoking and overweight. The potential confounders included in our study were maternal age and parity.16 20–24 The use of register data limited our opportunity to adjust for other potential confounders or mediating factors. Stephansson et al,20 who investigated the association between maternal occupation and stillbirth, found a substantial socioeconomic gradient in stillbirth even after adjustment for maternal sociodemographic and anthropometrical characteristics, differences in lifestyle and attendance at antenatal care.

Women's knowledge about their reproductive capacity (reproductive failure or success) may influence the lifestyle in a subsequent pregnancy and thereby introduce confounding.25 Sensitivity analyses were therefore made for nulliparous women only, and the estimates did not essentially differ from those obtained in the full cohort. Likewise, the estimates were only slightly attenuated when all births (ie, singleton and multiple births) were included.

The time trend in the educational gradient was different in Sweden compared with the other countries. One hypothetical explanation for the increased inequality could be the increase in the proportion of births from women with migrant background in Sweden during the study period. This explanation was, however, rejected by our sensitivity analyses on a restricted sample of women born in the Nordic countries. We will note, however, that the relationship between socioeconomic position and pregnancy outcomes among minority women is complex and may change over time as new minor groups appear and the socioeconomic position of existing groups changes. This needs to be studied further in its own right.26

Most Nordic studies on single countries have found socioeconomic inequality in the risk of stillbirth.15 16 18 20 27–38 The results of our study add to these findings. The majority of studies were register-based studies. Few did not find a socioeconomic gradient in stillbirth.39–42 Two Danish studies only included 117, respectively, 37 cases of stillbirth.41 42 Ericson and colleagues39 40 analysed the time trends in the effect of socioeconomic factors on stillbirth using a combined social measure based on education, occupation, cohabitation and citizenship and found no differences in the risk of stillbirth in either 1976 or 1981. However, in 1986, they found that the less privileged group had an increased risk of stillbirth compared with the privileged group. In a recent case–control study by Goy et al43 no association between educational attainment and stillbirth was found. Only household income was found to be a statistically significant predictor of stillbirth. A possible explanation could be that the small number of women with a low level of education in the study may have limited its ability to capture differences in stillbirth across education.

Recently, three other comparative studies on time trends in socioeconomic inequalities in reproductive outcomes have been published, using the NorCHASE dataset.8 44 45 These studies showed that educational gradients exist in fetal growth, preterm birth and infant death in Denmark, Finland, Norway and Sweden. These studies did not, however, observe any increased risk of adverse outcomes in Sweden at the end of the study period similar to what was observed in this present study. This seems to suggest that the mechanisms responsible for producing the educational inequalities are different depending on the perinatal outcome.

The explanation of the differences between educational groups and countries, respectively, remains obscure, and the mechanisms behind it are expected to be complex and multifactorial. Attendance to and quality of antenatal and obstetric care,17 20 42 46 and individual health behaviours and lifestyle15 20 47–49 are all factors that are expected to influence the risk of stillbirth. It has been suggested that differences in the risk of stillbirth between socioeconomic groupings may be due to social differences in acting on signals of pathological pregnancy, such as seeking care due to reduced fetal movements.20 Another suggestion may be that low socioeconomic position may affect the quality, quantity or content of medical care.46 Intermediate and high-level white-collar workers in Sweden were found to have more visits to antenatal care compared with women of low socioeconomic position.20 We could not investigate the influence of antenatal care in our study. Stephansson and colleagues20 reported that neither time of registration to antenatal care, nor number of antenatal care visits were associated with stillbirth risk in Sweden. It is, however, possible that midwives and doctors may be more likely to understand worrying signs from women of their own social class than from women of lower classes.20

Even though the Nordic countries have quite similar welfare systems with universal health care and social benefits and a high proportion of women participating in the labour market, differences in the economic development over time were observed.7 In the 1990s, Finland and Sweden were hit by economic recession, which caused unemployment rates to increase. Denmark experienced a prolonged economic recession from the early 1980s to the early 1990s, while Norway has not been subject to similar recessions in this period.7 50 51 The findings from Finland and Sweden in the early and late 1990s partly confirm the hypothesis that economic recession increases the educational gradients in stillbirth.

The Nordic countries are regarded as being homogenous. However, changes in policies and differences in the distribution of potentially mediating risk factors such as prepregnancy overweight and smoking may contribute to the educational differences between countries. In the Nordic countries, as well as in the rest of the world, the prevalence of delayed childbirth and overweight increased during the study period, and the relative importance of these factors is likely to increase.52–55 The prevalence of smoking decreased in all of the four countries in the period under study.56–59 In this study, we could not investigate the educational differences in smoking according to stillbirth, but it is well known that the prevalence of smoking is higher among women with low socioeconomic position compared with women with high socioeconomic position.15 56 57 60 Therefore, smoking is presumably an important contributor to the differences between countries. Similarly, factors related to the organisation and quality of health care possibly play an important role in the differences between countries.

We conclude that educational differences in stillbirth were found in Denmark, Finland, Norway and Sweden from 1981 to 2000, and that the development in the socioeconomic gradients was different between the countries, which may indicate that prevention is possible. Even though the stillbirth rate is generally low in the Nordic countries, the overall findings of an educational gradient, and the development in Sweden especially, is a matter of concern.

This study contributes to the knowledge of socioeconomic inequality in the risk of stillbirth in the Nordic countries and helps to highlight the importance of prevention especially among women of a low socioeconomic position. It is recommended that women with short educations get more attention, support and observation during pregnancy and labour. Determinants at both the individual and the societal levels should be taken into account to reduce the socioeconomic inequality in stillbirth.

What is already known on this subject

An educational gradient in the risk of stillbirth has been observed in many countries, including the Nordic welfare states. We examined how the educational inequality in the risk of stillbirth varied over time in Denmark, Finland, Norway and Sweden from 1981 to 2000.

What this study adds

Educational gradients in the risk of stillbirth were observed in all countries from 1981 to 2000, but there was considerable variation in the magnitude of and time trends in the gradients, which suggests that a potential for prevention exists, particularly among women of low educational attainment.



  • Funding This study was supported by a grant from the NordForsk Research Programme on longitudinal epidemiology, which is supported by the Nordic Council of Ministers.

  • Competing interests None.

  • Ethical approval The project was approved by the Danish Data Protection Agency according to Danish legislation.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.