Globally, more than 13 million infants (>10% of all births) are born preterm (<37 weeks' gestational age) every year.¹ With improvements in neonatal care, the survival rate of infants at highest risk for developmental impairment, those born very preterm (<32 weeks gestational age) continues to increase. In Germany, 1% to 2% of all infants are born very preterm.¹ However, increased survival has not resulted in equally improved life chances and quality of life.² Children born very preterm today are still growing up with a highly increased risk for a cluster of neurodevelopmental and mental health problems. These include persistent difficulties with attention, emotions, social behaviour, cognition, language, sensory, and motor skills.³ The early difficulties of children born very preterm cascade into lifelong increased risks for mental health problems and academic underachievement.⁴ In addition, studies unequivocally show that developmental outcomes of children born very preterm vary according to sociocultural contexts such as family education, income, and ethnicity.^{4, 5} However, research has neglected the role of immigrant background and language barriers for the development of children born very preterm. The few studies that do exist have focused on associations of very preterm birth and immigrant background with short-term outcomes in infancy.⁶ One study from the Netherlands found that multilingualism was associated with low cognitive and verbal outcomes of children born very preterm at 2 years and 5 years of age.⁷ In a French very preterm cohort, non-European-born immigrant status was associated with increased risk for behavioural, psychosocial, and multidomain impairments at age 5 years 6 months.⁸

In 2020, there were 281 million immigrants worldwide, an increase of 62% from 2000.⁹ Europe accounts for the largest rise in immigrants;⁹ social, educational, and health care systems are struggling to cater to immigrants' needs. In the health sector, systemic language barriers pose serious challenges to equity and equality of care.¹⁰ In Germany, almost 42% of children live in families with at least one immigrant parent,¹¹ and they more often struggle with socioeconomic inequalities than children without immigrant parents (hereafter referred to as ‘German children’).¹² At the same time, immigrant children are at higher risk for behavioural and emotional problems than their German peers.¹³ However, such findings are confounded by intersectionality of inequalities such as low education that pose barriers to immigrant families' access to health care.^{14, 15}

As a result of migration, linguistic diversity is also growing exponentially worldwide. Immigrants' proficiency in a country's official majority language correlates with educational attainment,¹⁶ health care utilization,^{17, 18} and overall health.¹⁹ Speakers of other languages than their host countries' official majority languages experience different levels of barriers, depending on the degree of similarity with their first language. For example, some languages are relatively mutually intelligible within their language families, such as Danish, Swedish, and Norwegian,²⁰ making them easily attainable for their speakers. Accordingly, some immigrant groups have advantages over others, because of linguistic similarities of their first language to the host society's official majority language(s). To operationalize such variations in language barriers, we assess the role of linguistic distance between children's first language and the official language of the host country. High linguistic distance (i.e. greater distance between two languages) has been associated with poor health²¹ and low educational attainment.²² In Germany, linguistic distance may indicate the level of similarities and differences between Turkish, Arabic, Italian, English, and German. Children whose first language is Turkish, for example, need to overcome a higher linguistic distance to the majority language German (linguistic distance = 99.77 points) than children whose first language is Italian in Germany (linguistic distance = 86.30 points) (Figure 1).

Very preterm birth, immigrant status, and language barriers in the form of linguistic distance may all be independently associated with high risk for behavioural and emotional problems in childhood. Language barriers and intersectional inequalities of immigrant children born very preterm have been neglected by research. Accordingly, the aim of this study is to assess whether behavioural and emotional problems of 5- to 6-year-old preschool children born very preterm across Germany are associated with immigrant background and linguistic distance of their first language to the host country's official language, German.

We tested the following hypotheses: (1) immigrant status and linguistic distance of children's first language to German are both associated with behavioural and emotional problems of children born very preterm; (2) linguistic distance is independently associated with behavioural and emotional problems of children born very preterm, after adjusting for the effects of immigrant status, maternal level of education, maternal age, infant sex, gestational age, multiple birth status, intraventricular haemorrhage (IVH), and bronchopulmonary dysplasia (BPD), and accounting for the nestedness of data within birth hospitals.

METHOD

Data are from the national multicentre observational German Neonatal Network (GNN) cohort study, including all births between 22 + 0 weeks and 31 + 6 weeks gestational age from 1st April 2009 until 31st December 2016 in 68 German neonatal intensive care units.²³

RESULTS

Table 1 outlines descriptive information for the cohort of children followed up at 5 to 6 years comparing German (n = 2591) versus foreign-born mothers (n = 629). On average, foreign-born mothers were older and less likely to have a high-school degree than German mothers. As expected, most of the children born to German mothers spoke German as their first language (median linguistic distance = 0.00), while there was large variation in linguistic distance scores among children of foreign-born mothers. Among the diverse language groups represented in the sample, the largest were Russian, Turkish, English, and Arabic. It is important to note that 6.9% of children with German mothers had other first languages than German, indicating substantial linguistic diversity because of immigrant heritage. Children of foreign-born mothers were older and they had lower IQ scores than children of German mothers.

Unadjusted models showed that higher linguistic distance was associated with higher SDQ total scores (Table 2, Model 1). Mothers' foreign country of birth (i.e. immigrant status) was not associated with higher SDQ total scores (coefficient = 0.440; 95% confidence interval [CI] = −0.010, 0.890). Linguistic distance remained significant (coefficient = 0.01; 95% CI = 0.01, 0.02) when both linguistic distance and immigrant status were included simultaneously in one model (Model 2). Random effects indicated substantial variation in the association of linguistic distance with SDQ total scores according to birth hospitals (coefficient = 0.63; 95% CI = 0.39, 1.00). Finally, in Model 3 linguistic distance from children's first language to German was independently associated with SDQ total scores (coefficient = 0.01; 95% CI = 0.00, 0.02) after adjusting for confounders (i.e. maternal level of education, maternal age, infant sex, gestational age at birth, multiple birth status, IVH, BPD, and age at assessment). As before, random effects indicated variation in the association between children's linguistic distance and SDQ total scores according to birth hospitals. In this model, if all other factors were held stable, a 10-point higher linguistic distance would result in a 0.10-point higher SDQ total score. Accordingly, for example, the SDQ total problem score would be on average 1 full point higher among Turkish-speaking immigrant children living in Germany (linguistic distance = 99.77, n = 96) compared with German children. Figure 2 shows unadjusted effects for selected language groups in the sample.

Analyses were repeated within the subgroup of children with foreign-born mothers to assess the specific size of the association between linguistic distance and the SDQ total scores of immigrant children born very preterm (Table 3). As in the full sample, the fixed effect of children's linguistic distances to German on SDQ total scores remained stable after adjusting for confounders (coefficient = 0.01; 95% CI = 0.00, 0.02).

We carried out a sensitivity analysis, additionally including a fixed effect of children's IQ scores as confounder in the fully adjusted mixed-effects model. Because of missing IQ data for some children the sample size was lower (n = 2657 for the fully adjusted model). While IQ was associated with children's SDQ scores (coefficient = −0.07; 95% CI = −0.09, −0.06), the effect of linguistic distance was attenuated (coefficient = 0.01; 95% CI = −0.00, 0.01).

DISCUSSION

This study confirms an association of linguistic distance between the first language of children born very preterm and their host country's official majority language with higher behavioural and emotional problems at 5 years of age.²⁷ Our findings from the large German-wide longitudinal GNN cohort study of children born very preterm suggest that there is no direct association of immigrant status with behavioural and emotional problems, but that instead language barriers are an explanatory factor. Importantly, the contribution of linguistic distance to explaining the behaviour of children born very preterm remained stable even after adjusting for social, biological, and perinatal clinical factors such as IVH and BPD. This points to a fundamental role of language barriers for the development of immigrant children born very preterm.

It is important to note that the size of the fixed effect of linguistic distance (i.e. coefficient = 0.01; CI = 0.00, 0.02) in our models may seem small in comparison to other variables' effects. However, the mixed-model coefficients represent the change in the mean response (SDQ total score) associated with a 1-unit change in that term. Accordingly, for example, if all other factors in the model would be held stable, the SDQ total problem score would be on average more than 1 full point higher among Arabic-speaking immigrant children living in Germany compared with non-immigrant German children. Considering the SD of mean SDQ total scores, one may argue whether one-fifth SD difference is meaningful. However, response differences on one Likert-scaled SDQ item may indicate meaningful subclinical differences in children's everyday behaviours. In reality, children's language barriers are associated with other risks such as low maternal education or low IQ, which for an individual child present clusters of multiple risks. While researchers must rely on clearly defined hypothesis testing, it is also important to raise awareness for this intersectionality of risks. The small but significant association between linguistic distance and SDQ total scores provides novel pointers for specific prevention studies taking language barriers into account.

Immigrant status alone, operationalized as a binary category in German versus foreign-born mothers, was not associated with children's behaviour and emotions. This supports assessing systemic variables such as language barriers when studying immigrant children's development.^{17, 18, 21} There are wide variations between immigrants' experiences depending on a range of intersectional factors including their heritage languages, countries of origin, reasons for migration, legal status, ethnicity, educational qualifications, and economic resources.¹⁴ The traditionally used oversimplification of the variable ‘immigrant’ is often masking these heterogeneous conditions and intersectional inequalities shaping immigrants' lived experiences. The operationalization of language barriers via linguistic distance helps explain important aspects of this complex situation.²² Researchers from other geographic regions may consider implementing linguistic distance as a screening tool to assess whether these findings from Germany are replicated. Based on the standardized operationalization of the linguistic distance score variable, we believe that the current findings are not specific for German or Germany, but that language barriers in the form of linguistic distance are a universal, global phenomenon.

As expected, social, biological, and perinatal clinical factors made important contributions to children's behavioural and socio-emotional problems. There is substantial overlap among these vulnerabilities, creating additive risks for certain individuals. In particular when working with immigrant groups, it is important to stress the contribution of the mother's level of education to their children's behavioural and emotional problems.¹³ Foreign-born mothers in the current sample were less likely than German mothers to have completed high school. If all other factors in the model were held stable, children of mothers with a high-school degree (i.e. German Abitur) would have on average 1.28 lower points on the SDQ total score compared with children of mothers who did not complete high school. The sensitivity analysis including a fixed effect of children's IQ scores as additional confounder in a subsample showed that IQ was associated with children's SDQ scores while the effect of linguistic distance was attenuated. This points to the well-known association between systemic inequalities, language barriers, biases in cognitive assessment, and perceived behaviour problems children from immigrant families are facing.¹⁵

The random effects included in our nested models showed small but stable variations in the association between children's linguistic distance and SDQ total scores according to birth hospitals. Considering regional and organizational differences between hospitals' patient populations (e.g. percentage of immigrants), catchment areas (e.g. rural vs urban, socioeconomic distribution), and individual treatment experiences, this is not surprising. Nevertheless, the current data suggest that treatment disparities between hospitals in Germany may not be characterized by as severe inequalities as reported by US-based studies.²⁸