Key words
immigration - perinatal data - women of Turkish origin
Schlüsselwörter
Migration - Perintaldaten - türkeistämmige Frauen
Introduction
In Berlin today more than one out of every four newborns is presumed to have parents
with an immigrant background [1]. The constant flow of refugees from Syria, Afghanistan and various north African
states into Germany since the summer of 2014 has brought attention to the health-related
issues associated with immigration. In recent years various studies have illustrated
significant disparities in perinatal care and outcomes between women with and without
immigrant background: The incidence and prevalence of preterm birth, rates of low
and very low birth weight newborns and the incidence of congenital malformations in
particular seem to be increased [2], [3], [4], [5], [6].
In Germany issues of immigration in obstetrics have long been marginalised by the
scientific community, although in the mid-1960s “health-care provision for migrant/guest
workers” and associated problems became part of day-to-day hospital routine [7], [8], [9], [10], [11], [12], [13]. In a review of the literature on perinatal data of immigrants from 1960 to 1989,
the low number of publications from German-speaking countries and particularly from
the Federal Republic of Germany despite the subjectʼs continuing relevance, is notable.
Older perinatal data analyses from Germany report contradictory findings: Four out
of six studies in the 1970s describe “worse” care in the context of stillbirths; two
however report better care compared to that received by German women [7]. These older data and empiric clichés regarding “peculiarities” of labour in immigrants
were and often still are generalised or passed on without due reflection. However,
to date there are almost no retrospective or prospective longitudinal studies addressing
the question of whether progressive acculturation and integration processes in certain
immigrant groups in Germany have improved or worsened perinatal outcomes in the decades
since the so-called “guest workers” were recruited in the 1960s, nor which factors
affect these outcomes. New studies describing and analysing obstetric care in immigrant
groups should be conducted, however it is also important to research historical developments
as this could shed light on the current situation. This monocentric, retrospective
data analysis of three birth cohorts from the 1970s to the 1990s thus attempted to
answer the following seven research questions, with German women serving as a comparison
in each case:
-
Do immigrant women receive uterine stimulants more often during labour?
-
To what extent did immigrant women receive analgesia during labour?
-
Do immigrant women have instrumental or operative deliveries (VE, forceps, caesarean
section) more often?
-
Do immigrant women have more episiotomies and/or high grade perineal tears?
-
Which group of women has a higher preterm birth rate?
-
Are postpartum haemorrhages > 500 ml more common among immigrant women?
-
How common is rooming-in in the different study groups?
Materials and Methods
An analysis of data taken from the birth registers of the municipal Rudolf-Virchow
Hospital in Berlin-Wedding was conducted for three decades represented by the arbitrarily
chosen years 1974, 1984 and 1994. Data was anonymously transferred from the birth
registers to an Excel table by hand (v. R. R.). A comprehensive plausibility test
was performed on a sample of subjects from each birth year (M. D.). The observation
period was restricted to the above mentioned years since a review of the birth register
revealed almost no immigrant deliveries at the Virchow Hospital in 1964, and extension
to the year 2000 did not seem sensible on consideration of recent changes to the treatment/care
profile of the Charitéʼs Virchow Hospital site: At the end of 1995 the University
Womenʼs Hospital Charlottenburg merged with the Virchow Hospitalʼs obstetrics and
gynaecology department, and in 1998 the Virchow Hospital became part of the Charité
University Hospital, Berlin (Charité – Universitätsmedizin Berlin). As a result there
have been changes in the treatment profile and patient clientele in terms of high
risk pregnancies and births.
Exclusions
For data analysis women were grouped as either primipara or multipara, so that those
with missing information on parity had to be excluded from the study. Group allocation
according to immigrant background was on the basis of a name analysis ([Fig. 1]); thus women whose name information was incomplete or illegible in the birth register
were also excluded, as were women whose names did not clearly fit into any one of
the three subgroups. In addition, we excluded women who had a miscarriage (defined
as delivery of a fetus with no signs of life and birth weight under 500 g).
Fig. 1 Name analysis algorithm.
Group allocation
Since grouping of study subjects on the basis of their own immigration experience
was not possible, a name analysis was used that was based on an algorithm previously
developed for the identification of people of Turkish origin (Spallek et al. 2014
[1]) and adapted to the current study circumstances ([Fig. 1]). First names and surnames that were common in multiple study groups were defined
as so-called “doublets”. Patients whose first and surnames could clearly be assigned
to different groups, and those with incomplete name information were excluded from
the analysis. Name allocation was performed by two independent experts, one of whom
had Turkish immigrant background, and was supervised by a third person.
Study subjects were then grouped as follows: Group 1: German women; group 2: women
of Turkish origin; group 3: “immigrants of other origin”.
Women with immigrant background were subdivided into a relatively homogeneous group
of Turkish origin, and a relatively heterogeneous group of “other” origin, since those
of Turkish origin represented the largest group of immigrants overall in the study
period, as they also do currently in Berlin.
Statistical analysis
The statistical program SPSS 20.0 was used for the analysis. In a first step the immigrant
groups were initially each compared to the group of German women – though not with
each other – using Pearsonʼs χ2 test, since the study parameters were all nominal variables from unrelated samples.
Where expected case numbers were < 5 per cell Fisherʼs exact test was used. Variables
with significant differences on χ2 test were then subjected to a binary logistic regression analysis (backward stepwise).
The following criteria were included in each regression analysis (reference variables
in bold type):
-
Descent/origin (German/Turkish/other)
-
Parity (primiparous/multiparous)
-
Year (1974/1984/1994)
-
Apgar (≥ 7, ≤ 7)
-
Prematurity < 37 + 0 weeks gestational age (GA)
-
Stillbirth > 500 g (miscarriages under 500 g were excluded)
-
Transfer to neonatal intensive care unit
-
Caesarean section rate (regardless of indication)
-
Rates of episiotomy and grade III/IV perineal tears
-
Rate of postpartum haemorrhage (> 500 ml)
-
Local and regional anaesthesia; analgesia in labour
-
Paternal presence at delivery and rooming-in
The likelihood ratio test is applied to determine the regression coefficient when
using the backward stepwise logistic regression. The value 0.10 was set as the exclusion
criterion. The Hosmer-Lemeshow test was used to test the validity of the binary logistic
regression model. The significance level was set at α = 0.05. Model relevance was
determined using R2 tests according to Cox & Snell and Nagelkerke. Odds ratios and 95 % confidence intervals
were determined for each regression analysis. No adjustment for multiple testing was
performed, since the analysis was retrospective.
The study concept of this retrospective analysis was discussed with and approved by
the hospitalʼs institutional board. The study was carried out in accordance with the
principles of good scientific practice assurance of the Charité – Universitätsmedizin
Berlin, and in accordance with data protection requirements.
Results
A total of 5454 perinatal data sets were documented. Seven patients were excluded
due to incomplete name information and a further 298 patients whose names could not
be clearly assigned to a single study group were also excluded (dropout rate: 5.6 %).
The analysis results that follow thus apply to a study collective of 5149 births,
with 1937 births from the year 1974, 1314 from 1984 and 1898 from the year 1994.
Sociodemographic characteristics and perinatal outcome
The percentage of patients with immigrant background was initially considerably lower
than that of German patients (Germans 60.8 % in 1974 and 63.5 % in 1984). This changed
however, and in 1994 immigrants made up well over 50 % of the study population ([Table 1]). While the percentage of mothers over 35 years of age remained constant over the
study period, the percentage of mothers younger than 24 years of age at the time of
childbirth fell in favour of those between 25 and 35 years old ([Figs. 2] and [3]).
Fig. 2 Comparison of the age distribution of Germans, immigrants of Turkish origin and immigrants
of other origin.
Fig. 3 Age distribution over the study period.
Table 1 Percentage of patients according to origin for the three study years: 1 974, 1 984,
1 994.
|
German origin
|
Turkish origin
|
Other origin
|
Total
|
|
Absolute
|
Percentage
|
Absolute
|
Percentage
|
Absolute
|
Percentage
|
Absolute
|
|
1974
|
1 178
|
60.8 %
|
562
|
29.0 %
|
197
|
10.2 %
|
1 937
|
|
1984
|
834
|
63.5 %
|
353
|
26.9 %
|
127
|
9.7 %
|
1 314
|
|
1994
|
729
|
38,4 %
|
683
|
36.0 %
|
486
|
25.6 %
|
1 898
|
|
Total
|
2 741
|
532 %
|
1 598
|
31.0 %
|
810
|
15.7 %
|
–
|
It was not possible to obtain reliable data on obstetric history (abortions, miscarriages,
ectopic pregnancies) or general personal/medical history from the information documented
in the birth registers.
Rates of caesarean section, episiotomy, high grade perineal tears and severe postpartum
haemorrhage (defined as blood loss > 500 ml) were not significantly different between
Germans, Turkish immigrants or “immigrants of other origin”. The rate of preterm birth
(defined as before 37/0 weeks gestation) was highest among German women, independent
of parity, and lowest among immigrants of Turkish origin. The consecutive logistic
regression analysis showed no statistically significant association between immigrant
background and prematurity. Parity had no effect on prematurity either.
Disparities in obstetric care
During the observation period uterine stimulants (labour induction/augmentation agents)
were used more often among primipara than multipara. Among all multipara as a group,
they were used most often in German women (66.6 %); in contrast among all primiparous
women, they were used most often in “immigrants of other origin” (83.6 %). On logistic
regression analysis there was no significant difference between Turkish immigrants
and Germans (p = 0.867). “Immigrants of other origin”, however, received uterine stimulants
significantly more often (OR 1.287, p = 0.008).
Independent of parity, over the course of the observation period (comparison of the
years 1974 vs. 1984 vs. 1994) a distinct decrease in the use of analgesics in labour
was observed, from 84 down to 4.9 %, and from 63.2 down to 1.9 %. In 1974 German women
received analgesics significantly more often than Turkish immigrants (68.1 vs. 56.5 %,
p = 0.001); no significant difference was evident on comparison with “immigrants of
other origin”. In 1994 however, “immigrants of other origin” received analgesics significantly
less frequently than Germans (0 vs. 2.9 %). A similar result was shown for the use
of local and regional anaesthesia (from 53.6 down to 7.5 % in primipara, and from
25.4 down to 1.4 % in multipara), which were used significantly less in both groups
with immigrant background across all three observation years. There were also significant
differences for rooming-in: Over the whole observation period both Turkish immigrants
and “immigrants of other origin” were significantly less likely than Germans to be
accommodated in the same room as their children, with odds ratios of 0.142 and 0.195
respectively (p in both cases < 0.001).
Discussion
Reports on “births among foreign nationals” were published in Germany for the first
time in the mid-1960s, bringing language and communication problems and their possible
adverse effects on labour and obstetric outcomes to the attention of clinicians [11], [12]. Studies from the late 1970s, which were linked to the introduction of national
perinatal data recording, formed the basis of discussions on the subject in the Federal
Republic of Germany well into the 1990s. Based on data from the greater Hannover area,
Oeter et al. (1979), for example, found that social factors such as nationality were
significantly associated with perinatal mortality, children of non-German origin being
negatively affected, and that transfer to childrenʼs hospital services occurred significantly
more often among newborns of foreign/immigrant mothers [14]. This is in complete contrast to the so-called Mexican paradox, which is discussed
in a US American publication: Despite having numerous risk factors (lower education
level, worse medical care, lower acculturation level) immigrant women from Latin America
have more favourable pregnancy and labour outcomes and a lower incidence of perinatal
complications than their “white” US American counterparts [15]. Positive protective influences of the informal, familial network of pregnancy support
were seen as a possible explanation. Whether such influences from acculturation and
integration processes are present/detectable for immigrants in Germany too, has not
been systematically studied to date. When the available data is considered, a “research
paradox” is evident: The number of studies and the state of scientific knowledge remain
unsatisfactory, particularly in Germany, despite the fact that since the mid-1960s
immigration issues have become increasingly prevalent and are increasingly encountered
in the hospital and practice health care sectors. The results of the current study,
which documents a development over three decades, at least provide evidence of disparities
in perinatal care in the past between German women, immigrants of Turkish origin and
of “other” origin. The parameters reflecting direct, personal care and interaction
between medical staff and patients especially show significant differences, suggesting
possible deficits in medical care: Immigrants of Turkish origin received epidural
anaesthesia less often than immigrants of other origin or German patients. Labour
augmentation agents/uterine stimulants were also used less often.
Even though similar results were found by Rizzi et al. [16] for caesarean section rate, and by Rust et al. [17] and Glance et al. [18] for use of epidural anaesthesia in labour in Italy and the USA respectively, the
generalisability of our study results is limited by the retrospective nature of data
collection, and by the fact that sociodemographic information was missing. Accommodation
of mother and child together in hospital, so-called “rooming-in”, was also significantly
less common among immigrants. Similar results can be found in the international literature.
Walsh et al. [19] report that immigrant women received an epidural in labour significantly less frequently.
In 2004 Yoong et al. published a comparative study finding a lower epidural anaesthesia
rate among Kosovo Albanians in Great Britain who had little or no English language
proficiency [23]. Rust et al. [17] analysed the deliveries of almost 30 000 women in the USA and reported an epidural
anaesthesia rate among African American women, Hispanic and Asian women significantly
lower than that for non-Hispanic “white” women. Limiting factors were, however, that
it remained unknown whether epidural anaesthesia was offered by medical staff or requested
by labouring women, and whether logistical or objective reasons could have explained
the findings. In our retrospective analysis there was no difference in caesarean section
rates. This result is however contrary to internationally published evidence. In a
comparison of matched pairs, Rizzo et al. (2004) found that the method of delivery
“planned caesarean” was more common among native Italian women than local immigrants.
A 2009 prospective analysis of birth data from approx. 1800 children born in Austria
only showed a significant difference for one subgroup, namely immigrants of Turkish
origin, who had fewer primary caesarean sections and more vaginal deliveries compared
to non-immigrants [20]. Our data nevertheless agree with other data available from Germany, which have
not shown any difference in caesarean section rates [21].
In summary, this relatively comprehensive, retrospective perinatal data analysis of
three historical birth cohorts has identified disparities that at least confirm that
patients with immigrant background receive different care, and are suggestive of deficits
in care. The study also highlights that the situation has not changed measurably over
the course of three decades despite increasing immigration and acculturation in Germany.
With this background, the question remains as to why no other systematic or prospective
analyses of any kind have yet been published in Germany, for example of the likes
of those published in a neighbouring European country, the Netherlands: E. g. the
Generation R Study from the greater Rotterdam area investigated the effects of ethnicity
on pregnancy, birth and child development in over 6000 women [22].
Although we have only conducted a historical review, with retrospective data analyses,
these data nevertheless provide important insights towards an objective, evidence-based
debate on the obstetric care of immigrant women. No recent data on the subject exist:
To find studies similar to this review an extensive literature search, back to the
1960s and 1970s, is necessary [9], [13].
