Do social factors and country of origin contribute towards explaining a “Latina paradox” among immigrant women giving birth in Germany?

Immigrants tend to have a lower socio-economic status and poorer health than the majority population of the country they migrated to [1]. However, in the US, Hispanic women have been shown to benefit from a specific form of the “healthy migrant effect” and to have better birth outcomes than white women despite their lower socioeconomic status [2]. This phenomenon, often called the “Latina Paradox”, was found for preterm birth, low birth weights (LBW) and small-for-gestational-age (SGA) when comparing native and immigrant population groups [3‐5]. So far, the “Latina Paradox” has mostly been framed as being associated with origin from a particular country (Mexico) or region (Latin America). Yet “country of origin” is a proxy for several concepts that need to be disentangled:

Social factors include religion, culture, and related behaviours. Such factors vary between countries of origin [4, 6]. Immigrants bring along traditions, values and behaviour from their country of origin, which may positively (or negatively) affect perinatal outcomes [4, 8]. Such cultural factors, however, are likely to change in the course of time spent in the target country and over generations. They may be adapted to the culture of the new country, along with other health behaviours, which could influence birth outcomes [9].

Several studies have sought to explore a potential Latina Paradox in Europe, looking at the relationship between birth outcomes and country of origin. For example, in Belgium, Jacquemyn et al. (2012) and Racape et al. (2016) found that being of Turkish or North African origin had a protective influence on preterm birth and low birth weight, respectively. [10, 11]. In The Netherlands Schulpen et al. (2005) reported higher death rates for newborns of immigrant women (Antillan, Turkish and Moroccan) compared to the Dutch population [12]. In Sweden, a large birth register study including women of many different origins produced less clear results: women from Chile and Syria were the only non-Swedes who showed better results than the native population in terms of LBW and preterm birth, leading the authors to conclude that the healthy migrant effect is ethnic- and outcome-specific [13]. Finally, in Germany, a recent study by David et al. (2017), which used the same data set that we will analyse, showed a lower frequency of preterm birth among immigrant women [14]. However, similar to the other studies mentioned above, the authors did not investigate this phenomenon in detail.

To better disentangle the role of the different sets of influencing factors including the role of the cultural factors and country of origin in explaining the Latina paradox, we take two steps. First, we investigate whether a Latina paradox regarding two birth outcomes commonly examined in this context, preterm birth and SGA, can be observed in immigrant women stemming from countries or regions other than Mexico or Latin America (here Turkey, Lebanon, EU15, EU27, other Europe, Middle East (excluding Turkey and Lebanon), North Africa, Sub-Saharan Africa, Far East, Latin America and Caribbean and North America), and in a target country different from the US (here Germany). We then examine whether the effect differs between countries of origin after adjusting for socioeconomic status in the target country. If this is the case, non-social components may play a role in explaining the Latina paradox.

Second, we assess whether the Latina paradox is (at least partly) explained by social factors which non-immigrant women are not exposed to, e. g. religion and culture of the country of origin. Effects of social factors linked to the country of origin are more likely to be causally associated with the birth outcomes if they vary with degree of acculturation, after adjusting for socioeconomic status. If social factors play a role in the Latina paradox, evidence that non-social differences matter as well will become stronger when – after adjusting for socioeconomic status, acculturation and religion – differences in outcomes remain.

8157 women delivered in the three hospitals in the period of data collection. 7100 women participated in the study (response rate 89.6%) [17]. Complete data for all variables relevant for the analysis was available for 6413 out of the 7100 women. Of these women, 2552 were 1st generation immigrants (subgroups: 561 from Turkey and 317 from Lebanon); 885 were 2nd generation women. The proportion of women with low education was higher in the immigrant groups (1st generation: 25.0%; 2nd generation: 9.4%) than the non-immigrant group (3.3%), while the proportion of women with high income was lower in the immigrant groups (1st generation: 11.2%; 2nd generation: 10.6%) compared to non-immigrant women (37.4%). First generation women had a lower diabetes prevalence (0.6%) than non-immigrant women (1.1%), and diabetes prevalence was particularly low among women from Turkey (0.3%) and Lebanon (0.0%). The majority of non-immigrant women stated they followed no religion (49.6%) or had no or little affinity to religion (19.7%) whereas many immigrant women reported a high affinity to religion (1st generation: 41.1%; 2nd generation: 45.8%) (see Table 1). 14.3% of the 1st generation migrants had a low acculturation level vs. 2.6% of the 2nd generation women.

A table of countries of origin shows the distribution of 1st generation immigrants grouped in EU15, EU27, other Europe, Middle East (excluding Turkey and Lebanon), North Africa, Sub-Saharan Africa, Far East, Latin America and Caribbean and North America (see Additional file 1). Overall 9.8% of the births were preterm, with 10.6% in non-immigrant women. The lowest proportion of preterm births was in women from Turkey (5.3%) (see Table 1).

Additional file 2 shows the relation of the country of origin to the birth outcomes using Chi Square Tests. For preterm birth the regions, Sub-Saharan Africa, Latin America and Caribbean and the country Turkey were significant (see Additional file 2). Regarding SGA only Lebanon had a statistically significant association with the outcome.

Table 2 and Additional file 3 show the influence of the country of origin on preterm birth, while controlling for medical variables, age and smoking (model 1), socioeconomic variables (model 2), as well as affinity to religion and acculturation (model 3). In all models, the association between a Turkish origin and preterm birth was statistically significant (model 3, Turkish origin: aOR 0.49, 95%CI: 0.33–0.73). Women with a high acculturation (Turkish origin: aOR 1.62, 95%CI: 1.01–2.95 had a higher chance to deliver prematurely compared to women with a low acculturation (see model 3 in Table 2). The association between affinity to religion and preterm birth was not statistically significant. The regions of origin Latin America and Sub Saharan Africa showed statistically significant associations with preterm birth for all three models (Additional file 3, model 3, Latin American origin: aOR 2.17 95%CI: 1.03–5.57; model 3, Sub Saharan Africa origin: model 3, aOR 1.77, 95%CI: 1.05–2.99).

13.7% of all newborns were SGA newborns. Women from Lebanon had the highest proportion; they delivered 20.8% SGA newborns (see Table 1). Table 3 shows the influence of the country of origin on SGA, controlling for medical variables, age and smoking (model 1), socioeconomic (model 2) variables, as well as affinity to religion and acculturation (model 3). In all three models the association between Lebanese origin and SGA was statistically significant (aOR: 1.72, 95%CI: 1.27–2.34) (see model 3 Table 3). The association between acculturation, affinity to religion, migration status (except Lebanese origin) and SGA was not statistically significant in any of the models.

We found a Latina paradox for women of Turkish origin in Germany regarding preterm birth. The proportion of preterm births among Turkish women was half that of non-immigrant women and regression analyses confirmed that this protective effect persisted even after adjusting for a range of control variables. Our findings also confirm the hypothesis that some social factors are at play in explaining the Latina Paradox for Turkish women in Germany. A high acculturation level, i.e. a high affinity with the German culture and a less strong affinity with the country of origin, increases indeed the chance of preterm birth, erasing some of the protective effect that a Turkish origin may provide. In other words, low acculturation can be considered as a social factor which contributes to explaining the epidemiological paradox regarding preterm birth in our sample. There is no direct comparison in the literature. Several studies have used length of stay in the target country as a proxy variable for acculturation: immigrant women who had resided in Canada for fewer than 5 years had lower preterm risk than non-immigrant women, while those with ≥15 years of stay were at higher risk [38]. Latina women born abroad had better preterm birth outcomes than Latina women born in the US and non-Latina women [39], suggesting that the process of acculturation or integration to the new country attenuates the protective influence from which 1st generation immigrants benefit. More research is needed to identify which specific items of acculturation are responsible for the differences in preterm birth rates.

We also explored the role of affinity to religion as another social factor. First generation women more often reported a medium or high affinity to religion compared to non-immigrant and 2nd generation women. However, regression analyses showed no statistically significant association between affinity to religion and preterm birth. We do not find the protective effect of religiosity on perinatal outcomes that has been shown for Latin American women in the US, where, for example, specific religious groups may promote positive health behaviours during pregnancy (e.g. no smoking, no alcohol) [3]. When trying to test the hypothesis that other religions may have a positive effect on perinatal outcomes through social support in the community, [3] we found no evidence indicating that high affinity to religion tends to lead to favourable health behaviour during pregnancy.

Still with regard to preterm birth, we found that women from Latin America and Sub Saharan Africa have higher odds of having a preterm birth than German women. Although a study which investigated factors for preterm birth in Germany found no significant association between a Latin American or Sub Saharan African origin and the outcome [40], evidence from other countries has shown on different occasions that immigrant women from Sub Saharan Africa tend to have poorer birth outcomes than the native population (see for example [41]). We found no equivalent in the European literature of the relatively poor birth outcomes of Latin American women in our cohort.

In line with some of the existing literature, we found no epidemiological paradox with regard to SGA [38, 39]. On the contrary, descriptive statistics showed that SGA was more frequent among 1st generation women compared to non-immigrant women, and that 1st generation immigrants from Lebanon had SGA rates about 1.5 times those of non-immigrant women. This resonates with evidence from the US where women of a Middle Eastern origin were found to have higher SGA rates than non-immigrant women [5]. After adjusting for socioeconomic status, acculturation and affinity to religion, Lebanon as a country of origin remained the only significant factor influencing SGA rates. Higher SGA rates in women of Lebanese origin may therefore at least partly be explained by non-social factors; one hypothesis being that standard SGA calculation based on birth-weight tables for Germans may not be an adequate measure for newborns from other ethnicities, with different average body types [14].

There is evidence that a social factor, namely the level of acculturation of pregnant women with a migration background, contributes to explaining the epidemiological paradox regarding preterm birth. Furthermore, independently of social factors, Turkey as country of origin has a protective influence on preterm birth whereas Lebanon as country of origin has a negative influence on SGA. Hence, the direction of the observed effect varies depending on the country and the outcome. This suggests that non-social factors that relate to the country of origin contribute to explaining differences in birth outcomes. Service providers therefore need to take a more differentiated view of the potential risks immigrant women face than what might be implied by a broad term like “Latina paradox” or “healthy migrant effect”.