Pregnancy outcomes in asylum seekers in the North of the Netherlands: a retrospective documentary analysis

Abstract Background With more than 20,000 asylum seekers arriving every year, healthcare for this population has become an important issue. Pregnant asylum seekers seem to be at risk of poor pregnancy outcomes. This study aimed to assess the difference in pregnancy outcomes between asylum seekers and the local Dutch population and to identify potential substandard factors of care. Methods Using a retrospective study design we compared pregnancy outcomes of asylum-seeking and Dutch women who gave birth in a northern region of the Netherlands between January 2012 and December 2016. The following data were compared: perinatal mortality, maternal mortality, gestational age at delivery, preterm delivery, birth weight, small for gestational age children, APGAR score, intrauterine foetal death, mode of delivery and the need for pain medication. Cases of perinatal mortality in asylum seekers were reviewed for potential substandard factors. Results A total of 344 Asylum-seeking women and 2323 Dutch women were included. Asylum seekers had a higher rate of perinatal mortality (3.2% vs. 0.6%, p = 0.000) including a higher rate of intrauterine foetal death (2.3% vs. 0.2%, p = 0.000), higher gestational age at birth (39 + 4 vs. 38 + 6 weeks, p = 0.000), labour was less often induced (36.9 vs. 43.8, p = 0.016), postnatal hospitalization was longer (2.24 vs. 1.72 days p = 0.006) and they received more opioid analgesics (27.3% vs. 22%, p = 0.029). Babies born from asylum-seeking women had lower birth weights (3265 vs. 3385 g, p = 0.000) and were more often small for gestational age (13.9% vs. 8.4%, p = 0.002). Multivariate analysis showed that the increased risk of perinatal mortality in asylum-seeking women was independent of parity, birth weight and gestational age at birth. Review of the perinatal mortality cases in asylum seekers revealed possible substandard factors, such as late initiation of antenatal care, missed appointments because of transportation problems, not recognising alarm symptoms, not knowing who to contact and transfer to other locations during pregnancy. Conclusion Pregnant asylum seekers have an increased risk of adverse pregnancy outcomes. More research is needed to identify which specific risk factors are involved in poor perinatal outcomes in asylum seekers and to identify strategies to improve perinatal care for this group of vulnerable women.


Background
With more than 20,000 asylum seekers arriving in the Netherlands per year, health care for this vulnerable population has become an important point of interest [1][2][3]. Asylum seekers are refugees whose request for sanctuary has not been processed yet by the country they seek refuge in [4]. Of all asylum seekers, 25% are women of reproductive age [5]. Research suggests that asylum-seeking women are disproportionately affected by health and social problems as compared to men, presumably because they are more vulnerable to physical assault and sexual harassment and they often feel their experiences and fears are not taken seriously [6].
A significant number of asylum-seeking women are or become pregnant during the time they seek refuge. Often, they arrive from countries with high rates of infectious disease, a poor health care system and have been persecuted, tortured or raped [7,8]. During their flight circumstances are often primitive and dangerous. Once they arrive in the Netherlands, asylum seekers have little money, lack purpose in daily life and struggle with worries about the asylum procedure and the situation of family back home [7][8][9]. Research suggests that due to both the continuation of preexisting health problems [10,11] and the result of economic hardship and social deprivation once seeking residence [12,13], asylum seekers have poorer physical and mental health compared to local populations [7,14,15]. Access to health care services is difficult for asylum seekers because of a lack of culturally appropriate information and a limited understanding of the Dutch health care system [9,16].
Literature suggests that pregnancy outcomes are worse in asylum seekers, with higher perinatal and maternal mortality as compared to autochthonous populations [15,[17][18][19][20]. Studies in other countries show that refugee women have more complications such as low birth weight, low APGAR scores, preterm labour, anaemia, excessive bleeding during delivery and an increased incidence of Caesarean sections and admission of their child to the neonatal intensive care unit [11,17,[20][21][22][23][24]. However, research on pregnancy outcomes on asylum seekers (i.e. those refugees whose request for sanctuary has not been processed yet) specifically is limited [25,26]. In the Netherlands asylum-seeking women show a 10-fold increase in maternal mortality, twice as much perinatal mortality and an increased risk of maternal morbidity, including a higher prevalence of uterine rupture, eclampsia, major obstetric haemorrhage and intensive care unit admission during pregnancy [27][28][29][30]. Additional perinatal outcomes in asylum seekers in the Netherlands have not been assessed. This study aims to assess the current difference in perinatal outcomes between asylum seekers and the local Dutch population in an area in the North of the Netherlands with a high density of asylum seekers. Also, cases of perinatal mortality in asylum seekers were reviewed. This information may help to identify specific areas of interest in pregnancy care for asylum-seeking women and act as guidance for health care providers to meet the maternity care needs of this vulnerable population.

Study design
We performed a cross-sectional database study using one midwifery practice and hospital databases to assess maternal and perinatal outcomes of asylum seekers and the local Dutch population in the North of the Netherlands.

Setting
In the Netherlands, asylum seekers live in asylum seekers centres during the processing of their request for sanctuary. In the North of the Netherlands, there are two major asylum seekers centres, Ter Apel and Musselkanaal. Ter Apel is the only central location of the Central Asylum Seeker Organization (COA; Centraal Orgaan Asielzoekers) in the Netherlands where asylum seekers who enter the country are accommodated at first instance. This resulted in a relatively high density of pregnant asylum-seeking women in the area.

Maternity care in the Netherlands
In the Netherlands, low-risk pregnancies are followed up by primary care midwives and family doctors (for nonpregnancy related complaints) [31]. Primary care consists of monthly visits, a check every two weeks in the last phase of the pregnancy, and then every week [32]. Secondary and tertiary care could only be accessed with referral and includes hospital specialist care.

Data collection
The data was retrieved from the primary and secondary care practices which provide the major part of pregnancy and delivery care to asylum-seeking women in Ter Apel and Musselkanaal. Databases were combined as described by Perined [33] to remove duplicate data of patients who were referred from primary to secondary care during pregnancy or delivery. Duplicate cases were identified by matching all cases on the mother's date of birth, due date, duration of pregnancy and the country of origin of the mother.

Study population
Asylum seekers who lived in Ter Apel or Musselkanaal and gave birth between January 2012 and December 2016 under the supervision of midwives from midwifery practice New Life or gynaecologists of the Refaja hospital were included in the study. Patients who were transferred to an asylum seeker centre elsewhere in the Netherlands before birth were excluded. The reference population consisted of the local Dutch population that gave birth under the supervision of the same care practices during the same time frame.

Outcomes
Demographic factors were reported, including age, the number of adolescent pregnancies (19 years old or younger), country of origin, parity and uncertainty of due date. Parity was divided into 3 categories: nulliparous women, low multiparous women (1-3 previous deliveries) and grand multipara (> 3 previous deliveries). The outcome measures of this study were perinatal mortality (defined as death between 22 weeks of pregnancy and 7 days postpartum), maternal mortality, gestational age at delivery, preterm delivery (defined as delivery before 37 weeks of gestation), birth weight, small for gestational age children (SGA; defined as weight below the 10th percentile), APGAR score after 5 min, intrauterine foetal death (IUFD), start of labour (spontaneous, induction or primary caesarean section), mode of delivery (spontaneous, vacuum/forceps assisted delivery or secondary caesarean section) and pain medication (opioid and epidural analgesia). Also, the Adverse Outcome Index-5 (AOI-5) was calculated. The AOI-5 was designed to measure the magnitude of 5 adverse events that occurred during or around the delivery process [34]. It consists of perinatal mortality between a gestational age of 32 weeks and 7 days postpartum, neonatal intensive care unit (NICU) admission above 37 weeks, APGAR score lower than 7 after 5 min, postpartum haemorrhage and third-or fourth-degree perineal laceration. The AOI is defined as the number of women with one or more adverse outcomes during birth as a proportion of all deliveries. All cases of perinatal mortality in asylum seekers were reviewed, aiming to find potential substandard factors. Information from the different patient files was analysed using a structured approach. The checklist included the age of the mother, gestational age at birth, a case description and the results additional examinations like autopsy and amniocentesis. After a review of the patient files, the potential substandard factors were identified.

Statistical analysis
All numerical values were tested for normality using Shapiro-Wilk's Test. Since there were no normally distributed values, values were presented using the median and range. Categorical values were compared using Chi-square or Fishers Exact test. The Mann-Whitney U test was used to compare non-normally distributed and ordinal values. A logistic regression was performed to test for confounders for perinatal mortality. First, a univariate analysis was performed on possible confounding variables. All variables that showed a significant effect were included in a multivariate model. For the multivariate model a penalised likelihood logistic regression was used to reduce the chance of bias due to the low prevalence of perinatal mortality in our population. A value of p < 0.05 was considered significant.

Ethical considerations
As this is an anonymous retrospective database study there were no specific ethical issues to be considered. By law, this study does not fall under the Medical Research Involving Human Subjects Act in the Netherlands.

Study population
Data of 2028 Dutch and 285 asylum-seeking women were included from the hospital database and 868 Dutch and 485 asylum-seeking women from the midwifery practice database. 216 (45%) pregnant women were transferred to an asylum-seeking centre elsewhere in the country before giving birth. After removing duplicates our study population included 2665 women: 344 asylum seekers and 2323 Dutch women.  Table 1 shows the demographic variables of both groups. Asylum seekers were younger (p = 0.000), had more adolescent pregnancies (p = 0.000) and there were more grand multipara women (p = 0.000) as compared to the control group. Most asylum seekers came from Syria (n = 75, 21.8%) and Eritrea (n = 65, 18.9%). Other countries were divided into categories based on geographical location.

Pregnancy outcomes
Asylum seekers showed a higher rate of perinatal mortality (p = 0.000) including a higher rate of intrauterine foetal death (2.3% vs. 0.2%, p = 0.000), had a higher gestational age at delivery (p = 0.000), labour was less often induced (p = 0.016) and they more often received opioid analgesics (p = 0.029) as compared to Dutch women ( Table 2). There was no significant difference in the frequency of epidural analgesia and APGAR scores after 5 min. There were no cases of maternal mortality. Babies born from asylum-seeking women had lower birth weights (p = 0.000), were more likely to be small for gestational age (p = 0.002) and there were more uncertain due dates in the asylum-seeking population (p = 0.000) ( Table 2). After removal of cases with an uncertain due date, there was still no difference in prematurity between the two groups (p = 0.459). The adverse outcome index showed no difference between groups (p = 0.529) ( Table 3). There were no significant differences in pregnancy outcomes between the different countries of origin except for parity (p = 0.001), gestational age at delivery (p = 0.034), the number of women with an uncertain due date (p = 0.036) and the use of an epidural (p = 0.005). Notable was that there were more grand multipara pregnancies in the Middle Eastern (10.7%) and the Eastern European (9.3%) groups and that the use of epidural anaesthesia during delivery was lower in the Eritrean group (see Appendix Table 6). Parity, gestational age at birth and birth weight showed a significant relation to perinatal mortality in univariate regression (Table 4) and were therefore included in a multivariate model (Table 5). After correction for these variables, asylum seekers were 7.2 times more likely to experience perinatal mortality as compared to Dutch women.

Review of perinatal mortality cases
Of the eleven cases of perinatal mortality in asylum seekers, ten were intrauterine deaths and one child died within 24 h post-partum. The IUFD's were diagnosed at the gestational age of 23 + 4, 23 + 6, 24, 25, 30 + 2, 34, 34 + 4, 34 + 6, 36 + 3 and 37 + 3 weeks, respectively. All women had their first antenatal care appointment after a gestational age of thirteen weeks, with an average of 22 + 3 weeks (n = 9, 2 unknown). Three women had no antenatal check-ups at all before an IUFD was discovered at respectively 33 + 2, 34 + 5 and 37 + 4 weeks. Another three women had one or more documented missed antenatal care appointments. Two women missed appointments because they were transferred to a different centre. None of the eleven women took the recommended dosage of folic acid during pregnancy; nine women did not take folic acid at all. Two women had a recorded history of mental health problems and in two cases there was substance abuse during pregnancy. In one of the cases, there was a multiple pregnancy with twin to twin transfusion syndrome. Intrauterine growth restriction was recorded in three cases and one woman developed preeclampsia. Further review of these cases revealed that in six cases there was a delay in seeking care when a woman experienced alarming symptoms: four women felt reduced foetal movement, for two and three days respectively, three weeks and a month before visiting a midwife. One of them did not know who to call during the weekend. One case of neonatal mortality within 24 h post-partum involved a woman who was losing green fluid with reduced foetal movement for three days, and an emergency caesarean section was performed because of signs of foetal distress. The child was born with APGAR scores of one after one, five and ten minutes and was anaemic. In six cases obduction was performed, which revealed two lightweight placentas showing maternal vascular malperfusion and one child had a trisomy 21.

Discussion
This study aimed to assess the difference in maternal and perinatal outcomes between asylum seekers and the local Dutch population in the North of the Netherlands and identify potential substandard factors in the care for asylum seekers. In this study perinatal mortality was higher in asylum seekers, birth weight and APGAR scores were lower and postnatal hospitalization was longer compared to Dutch women. Labour in asylum seekers was less often induced, opioid analgesics were administered more often and there were more adolescent pregnancies. There was no difference in preterm birth rate and the mode of delivery, nor in the adverse outcome index. No cases of maternal mortality were recorded in this study. Overall the findings of our study were in line with previous research [15, 18-21, 23, 25, 30-32]. We found that, even after correcting for confounders, perinatal mortality was higher in asylum seekers. Review of these cases revealed possible substandard factors causing a delay in the first two phases of the Three Delays Model (deciding to seek care and reaching the healthcare facility [35]), consisting of late initiation of antenatal care, missed appointments because of problems with transportation, not recognizing alarm symptoms, not knowing who to contact and transfer during pregnancy. Our findings reenforce those from previous studies and also identify additional delays in the third step of the model (receiving adequate care), such as a language barrier, fear of mistreatment, shame and non-availability of a female doctor [11,13,15,23,[36][37][38]. All these factors may contribute to the limited use of antenatal care in asylum-seeking women [13,18,20,[39][40][41]. Poor attendance to antenatal care has been associated with poor pregnancy outcomes [42,43]. Previous studies showed that a lack of antenatal care results in lower folic acid intake [44], which has been described to increase the risk of low birth weight [45]. This may have played a role in our population as we indeed found lower birth weight and a higher prevalence of SGA children in asylum seekers. Our study was not powered to detect a difference in maternal mortality since the incidence of maternal mortality in the Netherlands is 7 deaths per 100,000 live births [46]. Previous studies suggest that maternal mortality is higher among asylum seekers [13,15,27,28].
We found that the use of opioid analgesics was higher in asylum-seeking women as compared to the Dutch group. A potential reason for this may be because coaching these women can be a bigger challenge for a caregiver due to a language barrier and cultural differences. However, there was no difference in the use of epidural anaesthesia. In our study, the rate of labour induction was higher in Dutch women. Previous research showed conflicting results about the difference in labour induction between groups [39,47]. The option of labour induction after 41 weeks of gestation is discussed with patients in the Netherlands. It is possible that asylumseeking women might not know of the possibility, due to less antenatal care visits and a language barrier. In our study, postnatal hospital stay was significantly longer in asylum-seeking mothers as opposed to other studies [20,48]. A lack of facilities and social support at home could contribute to this.  This study did not observe differences in the incidence of preterm birth, low APGAR scores, adverse outcome index, NICU admission, perineum laceration, postpartum haemorrhage, mode of delivery and the rate of epidural analgesia. Previous studies showed similar results, except for a higher rate of epidural analgesia use during labour in local populations and lower APGAR scores in asylum seekers [17,20,23,39,47,49].
Finally, ours and other studies showed that asylumseeking women were on average younger and had a higher parity rate [17,20,23,39,47,48]. The higher parity rate in asylum seekers could be attributed to cultural differences and little control over family planning decisions, including access to contraceptives [18]. Other studies showed that grand multipara women had a higher incidence of maternal morbidity and therefore poorer perinatal outcomes [50] however in our study, parity and age showed no relation to perinatal mortality in multivariate analysis. An uncertain estimated date of delivery was more common in asylum seekers because of a lack of ultrasounds in early pregnancy. Therefore, the rate of premature children and SGA in this study could be underestimated. We found no significant differences in pregnancy outcomes between the different countries of origin in the asylum-seeking group.

Strengths and limitations
This was the first study comparing a wide range of pregnancy outcomes in asylum seekers and Dutch women. This study included asylum seekers from different countries of origin, while previous studies often included asylum seekers from one specific background. Our sample size (n = 2665) was large compared to previous studies whose sample sizes were all smaller than 1500 women with only two previous studies exceeding 1000 participants.
A language barrier plays a role with most pregnant asylum seekers, however, because of the retrospective character of the study there was not sufficient information to which extent this language barrier played a role and if and how often official translator services were used. Maternal mortality was also not assessed in this study. Our group was too small to assess the difference in maternal mortality and maternal morbidity between groups. For this study, we only included data from one hospital and one midwife practice. However, these facilities have vast experience in caring for asylum seeking women since the asylum-seeking centre in Ter Apel is the largest centre in the Netherlands. Also, the control group consisted of women from the northern Netherlands which was a region where a relatively large proportion of the population had a low socioeconomic status. Therefore, the control group might not have been representative of the general Dutch population. Differences in outcomes between the general Dutch population and asylum seekers may be even larger.

Implications for care providers
This study highlights the importance of improving care for pregnant asylum seekers. Extra attention should be paid to asylum-seeking women during pregnancy by health care providers with the ultimate goal to achieve equity in health. Our study identifies possible substandard factors of the current care system which could facilitate the development of effective health care interventions. Alternative forms of antenatal care for asylumseeking women targeting the identified substandard factors should be developed. In the North of the Netherlands, we are currently working with a group antenatal care program specifically for asylum-seeking women.
In this study 45% of the asylum seekers were transferred to another centre during pregnancy causing discontinuation of antenatal care. Transfer between asylum-seeking centres during pregnancy should be minimized to reduce suboptimal care for an already vulnerable population.

Further research
To provide more data about perinatal outcomes in asylum seekers, larger prospective multicentre studies should be conducted. Comparing the difference in perinatal outcome between different countries of origin might give insight in which women within the asylum-seeking population are extra vulnerable. Since antenatal care use is limited in asylum seekers, alternative forms of antenatal care and its effect on pregnancy outcomes should be studied. We are currently studying whether group antenatal care as compared to standard antenatal care in the Netherlands improves pregnancy outcomes and satisfaction with care in asylum seekers. Also, psychosocial factors and the incidence of mental health problems in asylum seekers should be studied.

Conclusion
Perinatal outcomes in asylum seekers appear to be worse compared to Dutch women. Extra attention should be paid to pregnant asylum seekers to make sure quality maternity care is provided. This study highlights that reducing disparities in pregnancy outcomes between asylum seekers and Dutch women should be an important public health goal in the Netherlands. Further large-scale research should be conducted to improve antenatal care for pregnant asylum seekers and to identify specific risk factors for poor perinatal outcomes in asylum seekers.   Alternative forms of antenatal care and its effect on pregnancy outcomes in asylum seekers need to be studied.