Interpretation of differences in stillbirth risk
The present study is the first in Germany to use nationwide perinatal data for the investigation of variations in the risk of stillbirth in relation to maternal migrant background.
Using the German perinatal database, we found relative differences in the risk of stillbirth for women of different regions of origin compared with women from Germany. Women from the Middle East and North Africa (including immigrant women from Turkey) showed the highest risk. After adjustment for maternal characteristics and obstetric factors, the higher risk of stillbirth among women from the Middle East and North Africa was partly attenuated for example when stratified for mother's occupation. However the overall risk remained stable for other characteristics. In contrast to prior studies we did not find any considerable differences between women originating from foreign regions of origin compared to women from Germany in terms of obstetric factors that might explain the higher stillbirth rates in some migrants groups, e.g. in timing and number of antenatal care visits as well as preterm birth and low birthweight. As the factors analysed did not confound or modify the association between migrant background and increased risk of stillbirth in the majority of stratified analyses, migrant status may have an independent effect on the risk of stillbirth. As migrant background might be a proxy for other underlying factors such as cultural factors, social deprivation, access barriers etc., further studies should focus on these factors.
The increased relative risk of stillbirth among women with migrant background was slightly mitigated after stratification for socio-demographic factors, such as single status, low social status, young age of mother and high parity (Tables 2) but it was still elevated compared with women from Germany. These results are in line with a Danish study, where a higher risk of stillbirth for Turkish and Pakistani women was not associated with their higher parity, lower educational level and lower household income compared to the Danish women .
As the variables available for SES were insufficient in this study we have to interpret our results very carefully. In addition, the population groups in each band are very heterogeneous e.g. in terms of socioeconomic status, making it difficult to examine associations between migrant background, SES and risk of stillbirth. Compared to other regions the Middle East and North Africa had the highest proportion of women with low SES. These women also experienced a higher risk of stillbirth. This might point to the importance of socioeconomic factors e.g. education and income as well as socioeconomic-related factors e.g. integration level and working possibilities of migrants when looking at stillbirth risks. This is in line with a study from the Netherlands, where higher infant mortality rates among Turkish migrants compared to women from the Netherlands were partly explained by socioeconomic and demographic factors . Consequently, women from the Middle East and North Africa in Germany comprise an important target group for future prevention strategies in the field of pregnancy care.
Several studies have reported an insufficient and late use of antenatal care among migrant women compared to non-migrant women [21–23, 33]. Causal factors identified include language barriers, lack of knowledge concerning antenatal care, psychosocial factors and lower education. Our findings point in a different direction: we did not find considerable differences in the number and timing of antenatal care between women with and without migrant backgrounds, both with livebirths and stillbirths. The majority of women in all groups attended antenatal care before gestational week 9 and recorded eight to eleven visits. This indicates that access to health care services is not worse for the majority of migrant women in Germany, at least during pregnancy, and the majority seems to be informed about the appropriate timing and number of antenatal visits. What we found was an excess risk of stillbirth for women from the Mediterranean countries and Asia amongst those who attended antenatal care prior to gestational week 9 and for women from the Middle East and North Africa, who attended antenatal care very early and received a sufficient number of visits.
In contrast, previous studies based on regional data of the German perinatal database found a less frequent use of antenatal care, in particular for single women with migrant backgrounds , women from Eastern Europe and Mediterranean countries [27, 30] and a later use of antenatal care among women from the Middle East , from Eastern Europe and from Mediterranean countries . In addition, regional differences in the analysis of BQS data indicate that our findings may vary between federal states as well as urban and rural areas within single federal states in Germany. Comparisons of results between the previous German studies and our study are difficult because of the variations in defined number, timing of antenatal care and differences in the studied migrant groups. Nevertheless, we cannot rule out the possibility of variations in use and timing of antenatal care according to particular countries of origin and regional levels in Germany. Hence, further analyses of risk differences in stillbirth and its determinants on regional level and for precisely defined migrant groups are necessary.
In our study, women with migrant background with only a few or no antenatal visits, or who attended very late, showed a lower risk of stillbirth compared with women without migrant background with similar utilization patterns. Women with migrant background give birth in younger years and have a higher parity compared with women without migrant background (Table 1 and 2). It can be assumed that the women with migrant backgrounds are a selection of young women who already gave birth to several children without severe complications and altogether show a low risk profile.
Several studies from Europe and the USA showed a higher maternal risk of preterm birth and low birthweight for migrant women [13–18]. In contrast, there are several studies showing a lower risk for migrants despite risk factors such as higher parity and lower socioeconomic status [11, 27, 28]. In line with these findings we did not find differences in preterm birth and low birthweight prevalence between women of different regions of origin and women from Germany. When looking at the risk of stillbirth adjusted for birthweight, women from Eastern Europe and from the Middle East and North Africa experienced a significantly higher risk. Our findings also suggest that birthweight relative to the gestational age seems to have an influence on the differences in risk of stillbirth: a low birthweight combined with a regular gestational age ("small or light for dates") contributed to a higher risk of stillbirth, especially for women from the Middle East and North Africa compared to women from Germany. This might be an indication for a higher stillbirth risk among these women associated with intrauterine growth retardation (IUGR) or fetal malnutrition. Detecting IUGR is one of the main aims of routine ultrasound screening in antenatal care. A study found that the sensitivity of routine ultrasound examinations in Germany was extremely low; merely 30% of cases with intrauterine growth retardation were diagnosed antenatally . Further research is needed to validate our indicative findings regarding a higher risk of stillbirth in association with fetal malnutrition or IUGR association among women from the Middle East and North Africa, and to examine whether differences in sensitivity of antenatal diagnostic investigation between women with and without migrant background exist.
Higher risks of stillbirth in some migrant groups might be associated with a more unfavourable use of antenatal care compared to women without migrant background, but it might also be assumed that there are differences in the quality of antenatal care. In this context, studies found that 25-30% of perinatal deaths are due to factors of suboptimal care [36, 37]. Among the commonly found explanatory factors for this are the missed or delayed diagnosis of intrauterine growth retardation and an inadequate management of this condition, as well as the patient's non-compliance or delayed attendance to antenatal care [36, 38]. A Europe-wide study ("Perinatal Death Audit Study") found that migrant women have a higher risk for perinatal mortality due to suboptimal factors during antenatal care . Some of these factors seem to be associated with poor language skills leading to miscommunication between migrants and health providers, as well as a lack of knowledge and attention to cultural-related issues among maternity care professionals . Researchers in Germany have found differences in the quality of antenatal care between women of German and non-German nationality . The authors used the quality of documentation in the maternity record as a proxy indicator for the quality of care. Completeness of anamnestic data and documentation of consultation were lower among foreign women than among women without migrant background. The documentation of prenatal diagnoses (e.g. IUGR) is thus likely to be lower among women with migrant background and may result in insufficient surveillance of intrauterine growth.
Stillbirth and low birthweight relative to the gestational age can also be associated with congenital anomalies. We were not able to show considerable differences in the rate of congenital anomalies or modifications of risk of stillbirth adjusted for anomalies. However, several studies showed a higher perinatal mortality among newborns of women from non-western countries in the Netherlands and among Pakistani women in Great Britain, associated with a higher rate of congenital anomalies in these migrant groups [19, 20, 42]. A Norwegian study found that nearly 30% of infant deaths among Pakistani migrants were due to the high percentage of consanguine marriages in this group , with consanguinity as a risk factor e.g. for congenital anomalies. The prevalence of consanguine marriages is high among Turkish migrants, the largest migrant group by nationality in Germany. An association with higher rates of stillbirth might be probable [43, 44]. However, this remains an assumption because the German perinatal database does not contain valid information on the parents' familial relationship.
Strengths and Limitations
The strengths of this study are the large number of cases, its representativeness, the topicality and the completeness of data. We used population based and Germany-wide data. The data used were collected in a systematic way so we could confidently rule out selection bias. Furthermore, the study is the first to analyse relative differences in stillbirth rate and risk of stillbirth between several migrant groups in Germany.
The study, however, has its limitations. Although completeness of data is very high, the quality of some data items is sub-optimal. Jahn and Berle  were able to show that (obstetric) risk factors in the perinatal database are not documented as well as in the maternity log. Thus, there seems to be a loss of information in the data transmission process. "Occupation of the mother" showed a high non-response, and information on the father's occupation and income is lacking. Although smoking had a high non-response, it was included in the analyses because there was no evidence for a differential proportion of missings among women with and without migrant background. We are also not sure about the completeness of data concerning pre-eclampsia. While the eclampsia rate of 0.1% is similar to prior reported incidences for Germany [46, 47], the pre-eclampsia rate seems to be too low, especially among migrant women. Prior studies reported pre-eclampsia incidences of 5-10% for Germany . Eclampsia and pre-eclampsia occur during pregnancy and are mostly diagnosed and treated outpatiently by a gynaecologist. A possible reason for the inadequate reporting of these in the BQS data might therefore be that information from the maternal card about outpatient diagnoses and treatments are insufficiently transferred to the hospital data and subsequently to the BQS. Birth related factors showed a high completeness.
One of the major limitations in the conduction of migration- and socioeconomic-sensitive research in the field of obstetrics in Germany is the lack of appropriate variables of migrant status and socioeconomic status collected in the perinatal databases. As there is no clear definition of "region of origin" in the instructions for completing the perinatal sheet in German hospitals, we cannot be sure about how the variable is collected in hospitals. Hence, our data set might include more than one migrant generation without opportunity for differentiation. Further, the combination of several countries in one group is unsatisfactory. Each group includes different national, ethnic and cultural populations that are not at all homogenous. This makes the interpretation of our results more difficult. While it can be assumed that women of Turkish origin represent the largest proportion in the group "Middle East/North Africa", we cannot rule out that our conclusions do not apply to the other migrant groups included. In addition, the group "Asia" is not further defined, making it difficult to draw conclusions about this group from the results presented in this paper. There might be huge differences in the group of Asian migrants especially regarding stillbirth rates as Asia includes some of the most deprived areas of the world such as Central Asia with high stillbirth rates, and some of the lowest stillbirth rates, e.g. in Japan. Furthermore, the indicator of socioeconomic status in the database is not optimal as it only provides information about the mother's occupation. Further information, e.g. concerning the maternal education level as well as the father's education level and occupation, would be necessary to describe and study the effect of SES on birth outcomes among migrants in a more appropriate manner. As the perinatal database is the only available routine database for perinatal research in Germany, the implementation of appropriate variables for migrant status (at least country of birth of father and mother as well as country of birth of mother's parents and nationality) and the socioeconomic status comprises a major challenge for the future. The perinatal database was mainly developed for, and is still an instrument for quality control in German hospitals. Hence its use for routine health reporting or health research is limited. A change of data collection and implementation of new variables is highly dependent on the backup from politics in each federal state in Germany, and would thus take a very long time. Nevertheless, there are already successfully working birth registries implemented in single regions in Germany, e.g. the Mainz birth defect monitoring system (called the ''Mainz Model"), which is based on an expanded perinatal sheet focussing mainly on anomalies. These kind of approaches should be promoted and expanded to other federal states in Germany.