Socioeconomic inequalities in stillbirth rates in Europe: measuring the gap using routine data from the Euro-Peristat Project

Socioeconomic disadvantage, as measured by low levels of mothers’ or fathers’ education, occupational status or income, is associated with raised risks of stillbirth even in countries with universal insurance coverage and generous welfare provisions [1–6]. The hypothesised mechanisms are multiple and interrelated and include smoking, poor diet and other unhealthy behaviours, higher stress, less social support and depression, teenage motherhood, unplanned pregnancies, a higher prevalence of chronic health conditions, such as hypertension, diabetes, or obesity as well as poor access to antenatal care and receipt of suboptimal care [1, 7–9].

Many questions remain surrounding the causes of stillbirth and effective approaches to prevention. A high proportion of stillbirths, up to 30 %, remain unexplained even with complete clinical and autopsy records [10, 11]. Further, most stillbirths in high income countries occur before the onset of labour [12], providing fewer obvious possibilities for medical intervention than, for instance, cases of intra-partum, neonatal or infant death. Investigating the associations between socioeconomic position and stillbirth risk is an important component of efforts to advance knowledge about the aetiology and prevention of stillbirth.

A focus on social inequalities also emphasises the distal determinants of stillbirth risk which are accumulated over the life course and relate to parental health status, behaviours and knowledge preceding the pregnancy. Stillbirths may also perpetuate social differences in health by creating disease, in particular, maternal depression [13], and have been associated with long term risks for the mother [14]. Improving pregnancy outcomes by reducing stillbirths is therefore a component of broader national strategies to interrupt the transgenerational cycle of ill-health.

Comparing outcomes between social groups illustrates the poorer health associated with socioeconomic disadvantage and the outcomes that could be achieved in the absence of these inequalities. Having reliable indicators for measuring social inequalities in stillbirth enables target-setting for health policies, performance benchmarking and monitoring of trends over time. In this study, we sought to assess the magnitude of social inequalities in stillbirth rates in European countries using indicators generated from routine monitoring systems.

Despite the differences in health policies, educational policies and reporting systems across Europe, the available data from routine systems about stillbirths by socioeconomic status pointed to widespread and consistent socioeconomic inequalities in the stillbirth rate. Over three quarters of 29 participating European countries provided data on stillbirths by either mothers’ educational levels or parents’ occupation. While data about mothers’ educational levels were more widely available and comparable than data about mothers’ or fathers’ occupations, some of the larger countries, notably Germany and England and Wales combined only had data by parents’ occupations. Among countries with data, the median RR for stillbirth was 1.9 for women with primary or lower secondary education and 1.4 for those the intermediate category of higher secondary education compared to women with post-secondary education.. Median RR in the lowest occupational categories compared to highest were 1.6 and 1.4, respectively, for mothers’ and fathers’ occupational groups. These differences had substantial impacts at a population-level: if all women faced the stillbirth risks of the most educated, the number of stillbirths would be 25 % lower.

The main strength of our study is the ability to present data from all over Europe collected for the same year using a standardised instrument. We also requested data about mothers’ educational levels using local classifications and mapped these to the ISCED-97 classification to ensure consistency in coding and then checked our coded data with the data providers in each country. Our study also has several limitations. Because we used aggregated data, we were unable to explore the contribution of other demographic or behavioural factors, such as maternal age, parity, migrant status, smoking or body mass index to the higher risks associated with our socioeconomic indicators. Further, even though Euro-Peristat uses a common inclusion threshold of 22 weeks of gestation for births and deaths, not all countries are able to provide these data and even when they can, practices related to recording of early stillbirths differ [19]. While it is likely that these differences relate primarily to legislation and recording practices and not women’s socioeconomic circumstances, social inequalities may be more acute for early stillbirths and underreporting might thus affect the comparability between countries [5]. For countries where TOP cannot be differentiated from spontaneous stillbirths, the social gradient may be attenuated as TOP reflect the prevalence of anomalies for which the social gradient is less consistent [23, 24] as well as access to prenatal screening, which may have an inverse gradient in some countries [25]. Finally, some countries had missing data on socioeconomic factors which could bias the estimates. This bias likely leads to underestimates of the effects as missing data were more frequent among stillbirths and may also be more frequent for socially disadvantaged and migrant women.[26]

Because there are no universal measures of social disadvantage, researchers use a wide variety of different indicators: occupation, education, income, other measures of wealth, housing conditions, lack of access to health care, in particular prenatal care, and others. We used two indicators selected in a Delphi process based on assessments of importance and feasibility. The definition of these indicators, level of education and occupational group, were based on common classifications agreed upon by international organizations. However, categories may not reflect similar constructs in all countries or be understood in the same manner by responders or coders.

For mothers’ educational level, comparability issues related to the limit for differentiating between lower and higher secondary schooling, to the classification of vocational tracts and to the inclusion of a category for no or limited schooling, which may be relevant for some migrant women There are also real differences in educational attainment across Europe. Similarly broad variations in maternal education have been documented in comparative studies based on birth cohorts across Europe [27, 28]. However, despite the differences in the distribution of maternal educational attainment a negative social gradient was observed in most countries when this indicator was used.

For mothers’ and father’ occupation, questions of comparability were more complex as it was not possible to describe original data. Rules for recording and classifying occupations for parents who are not in paid employment or seeking work at the time their babies are born and lone mothers registering births without the involvement of the babies’ fathers were not clear. Compared to mothers’ educational level, there was higher variation in the distribution of categories between countries, RR tended to be lower and did not always follow a linear gradient. Furthermore, while the ISCO classification appeared to be widely used, there are questions about whether a status based classification of occupations is most relevant. Some research suggests that a classification based on employment relations (i.e. employers, employees or self-employed) provides more useful insights into the nature of social inequalities and their potential impact of health [29, 30] and these principles have been used for the development of a European Classification [31]. Given these conceptual and practical difficulties, one way of improving comparisons across countries would be to promote the recording of mothers’ educational level in routine systems either directly or through data linkage.

Studies from other high-income countries have documented elevated risks of similar magnitude for women with low socioeconomic position. For instance, using Canadian data, Auger et al. reported relative risks of 1.6 and 1.3 for low and intermediate educational levels compared with women with the highest educational level [5]. In a large US case control study, the unadjusted OR for educational level were 1.5 and 1.4, respectively for women with primary or some secondary and completed secondary compared to women with a post-secondary education [32]. The reasons for these elevated risks are likely multiple and interconnected. One UK study investigating stillbirths by area-based deprivation scores found that deprivation gaps existed for all causes, except for mechanical events, with the widest gap for stillbirths due to antepartum haemorrhages [3]. The study in Canada found differences throughout the gestational age spectrum, although they were more marked at earlier gestational ages [5].

The finding that social factors are not restricted to specific causes or gestational age groups is not surprising. Many risk factors, including smoking, diet and healthcare factors, affect a number of pregnancy complications such as extremely preterm delivery, growth restriction and congenital anomalies which raise stillbirth risk and are more common among less advantaged groups in the population. Mothers in these groups are more likely to have high body mass indices (BMI) and to smoke during pregnancy and these are important risk factors for stillbirth [33]. Other risk factors, such as migrant status, also affect a wide range of reproductive health outcomes [34, 35]. Social differences in screening and termination of pregnancy may also play a role if socially disadvantaged women are less likely to terminate pregnancies with lethal anomalies, either because of lack of access to screening or differences in attitudes to pregnancy terminations [24, 25, 36]. Antenatal detection of growth restriction may play a role in preventing stillbirth and the extent to which social position affects access to screening should also be considered [37]. The question of how much of the social gradient is explained by these behavioural and healthcare factors, especially those that are potentially modifiable, is an important area for further research.

While we found consistent associations between risks of stillbirth and social factors in European countries, the magnitude of the social gradient varied. Some of this variation could be related to small sample sizes in some countries as well as to the non-comparability of socioeconomic classifications. However, studies comparing the Nordic countries have found differences in the social gradient of stillbirth risk even within these relatively homogenous societies [2]. Other studies of socioeconomic inequalities in Europe have uncovered differences in the inequality gradient for overall mortality rates which are partially correlated with differences in the prevalence of smoking and overweight in the population [38]. Other European studies have also suggested that healthcare can contribute to differences in outcomes, for instance in cancer mortality, which may reflect social inequalities in access and quality of care [39]. In the perinatal field, this is an important area for further investigation and could be a powerful tool for identifying population-based risk factors and health care policies that contribute to stillbirth etiology and thereby to prevention of stillbirth for the benefit of all women.

This study documented widespread social inequalities in stillbirth risk in Europe, revealing the extent to which stillbirth can be a consequence of social disadvantage. While more research is needed to better understand differences in the magnitude of these differences between countries and how to harmonise use of socioeconomic categories, available data in routine systems can be used to set goals for the future and to monitor changes over time.

Data used in this analysis can be requested from the Euro-Peristat coordination team through the project’s website (www.europeristat.com).