A High Prevalence of Autism Spectrum Disorder in Preschool Children in an Immigrant, Multiethnic Population in Sweden: Challenges for Health Care

The study area in this study consists of one district of Gothenburg, with ~ 13,200 inhabitants of whom ~ 1350 were children aged 0–5 years at the end of 2018. The district is one of the most multiethnic and socioeconomically disadvantaged areas in Gothenburg. In the study area, 89.6% of the population has foreign background. The most common birth countries are Iraq (10.3%), Somalia (5.5%), Syria (5.4%), and Bosnia-Hercegovina (5.1%). The area has high prevalence rates of ill health, high levels of unemployment and low average income (Statistik och Analys Göteborgs stad 2018). Sweden has a very high Human Development Index (HDI) (United Nations Development Programme 2018), but in the study area ~ 56% of the foreign-born population has a medium/low/unknown HDI (Statistik och Analys Göteborgs stad 2018).

Sweden has complementary health care for all children. The local Child Health Center (CHC) in the study area provides a developmental surveillance program for children between 0–6 years and reaches about 99% of all children (Central Barnhälsovård Västra Götalandsregionen 2018). The public health services in the city of Gothenburg have implemented a screening programme, to enable early detection of ASD. All 30-month-old children are screened for ASD at the CHC with a combination of instruments: the parental questionnaire Modified Checklist for Autism in Toddlers—Revised with Follow-up (M-CHAT-R/F) and the clinical observation of the child, the Joint Attention-Observation (JA-OBS) (Nygren et al. 2012b). The methods are also used if there is suspicion of ASD prior to routine screening.

The target population for this study consisted of all children born during the period of January 1, 2013 to December 31, 2016, and registered at the CHC in the study area on December 31, 2018 (n = 902, 454 males, 448 females).

Since 2013, there is a local collaborative assessment and intervention program with a multidisciplinary team for families with young children with neurodevelopmental problems in the study area. This enables both assessment and interventions close to the family’s local setting, in collaboration with the health care nurse at the CHC, the preschool staff and other social services. The team is located in a family center, which includes the CHC, an open preschool,¹ the maternity health care and parental support provided by the social welfare. The goals for the local program are to provide assessment and individually planned interventions to children with neurodevelopmental disorders in their own neighborhood, to increase the accessibility and continuity in the health care provided, as well as parents’ involvement in the interventions.

Children identified at risk for ASD or other neurodevelopmental disorders at the CHC are referred to the local multidisciplinary team for further assessment and interventions. The multidisciplinary team consists of a pediatrician/child psychiatrist, a developmental psychologist, a special education teacher, a speech and language pathologist, a child specialist nurse, and a social worker. A professional interpreter was needed in about 60% of the assessments. The multidisciplinary assessment for all children included: (1) A physical developmental examination, hearing test, blood tests, including SNP array, and other relevant investigations according to the pediatrician’s assessment; (2) A parental interview regarding the child’s developmental and medical history, current clinical symptoms and social situation; (3) The ESSENCE-Q-REV (Gillberg 2010); (4) The Wechsler Preschool and Primary Scale of Intelligence, Fourth Edition (WPPSI-IV) (Wechsler 2012) for the children over 2:6 years, and for younger children Merrill-Palmer—Revised Scales of Development (MP-R) (Roid and Sampers 2004); (5) The Autism Diagnostic Observation Schedule Second Edition (ADOS-2) (Lord et al. 2012); (6) The Vineland Adaptive Behavior Scales Second Edition (VABS-II) (Sparrow et al. 2005); (7) An observation of the child in his or her preschool.

The ASD diagnosis was based on all obtained information from the assessments, and distinguished in accordance with the criteria of the DSM-5 (299.00, Autism Spectrum Disorder) (American Psychiatric Association 2013).

The results are presented descriptively, counting the number of children, and by presenting means, standard deviations, and proportions.

The prevalence of ASD was estimated at 3.66% (33/902 children) for children aged 2–5 years on the census day. For the males, the prevalence of ASD was estimated at 5.29% and for the females 2.01%. The prevalence rates and number of children diagnosed with ASD in different age groups are displayed in Table 2.

In the current study, the estimated prevalence of ASD among children aged 2–5 years was 3.66%, and this rate is likely to be regarded as a minimum. After December 31, 2018, additional children in the age group had been diagnosed with ASD and the real prevalence is likely to be more than 4.32%. This is higher than the general prevalence reported in many international epidemiological studies (Lyall et al. 2017), as well as in relation to the prevalence of 1.03% among children 2–5 years in the whole city of Gothenburg (unpublished data 2018). However, the results accord well with those of earlier studies from immigrant populations (Abdullahi et al. 2019a; Crafa and Warfa 2015; Fairthorne et al. 2017; Kawa et al. 2017; Magnusson et al. 2012).

The fact that we found a high prevalence of ASD in this sample could be related to several factors. First of all, a universal health care system that reaches all children is important for early identification and diagnosis of ASD. This basic condition was present in the current study, with the developmental surveillance program including autism screening for all children at the age of 30 months at the local CHC. The screening methods, the combination of the M-CHAT-R/F and the JA-OBS, together with the nurses’ level of competence and experience played an important role in identifying the children in need of further assessment. This is in accordance with the findings of Barbaro and Halder (2016) that suggest a combination of instruments, that incorporates parental report and observation components, for early identification of ASD in low resource settings. An earlier population study from Gothenburg has shown a high positive predictive value (90%) for the combination of instruments, and the study confirmed that the screening programme identified ASD in children from immigrant families as well as in children from Swedish families (Nygren et al. 2012b). An indication of a higher prevalence of ASD in children from immigrant families was also reported (Nygren et al. 2012a). We suggest that the combination of methods for early detection of ASD is necessary in areas with a high proportion of immigrant families and in areas with low SES.

Secondly, access to a comprehensive multidisciplinary assessment for diagnosis of ASD is essential (Gillberg 2010). Earlier studies have proposed that immigrant families have more difficulties accessing health care and other services for their children (Carlsson et al. 2016; Lemay et al. 2018; Thomas et al. 2012). In the present study, the local collaborative assessment and intervention program co-located with the CHC, close to the family’s setting, enabled assessment without geographical, organizational or linguistic obstacles.

We argue that the methods for early identification of ASD and access to a comprehensive multidisciplinary assessment are prerequisites for diagnosing ASD in children in the study area. These basic conditions act as confounders in this study, and contribute to the validity of the results.

The reported higher prevalence of ASD in immigrant populations is probably, like ASD itself, multifactorial. Different factors appear to explain the high prevalence of ASD in these populations, such as biological/genetic, non-genetic, as well as epigenetic factors. Thus, higher rate of ASD in immigrant populations are reported across many countries (Abdullahi et al. 2019b; Bolton et al. 2014; Fairthorne et al. 2017; Magnusson et al. 2012), and the risk for ASD with comorbid ID was highest if maternal country of birth was in the African or Asian continents (Abdullahi et al. 2018). In our study, 90% of the children had a mother born in the African or Asian continents, and 64.5% of the children had comorbid ID. A recent study by Abdullahi et al. (2019a) has shown that more severe presentations of ID in children with ASD in immigrant groups allow them to be referred earlier for assessment. This is also likely to reflect the high prevalence of ASD in our group. Crafa and Warfa (2015) have suggested that ethnicity and biological risk factors alone cannot explain the difference rates of ASD found in children born to migrant mothers. Migrations theories of ASD have proposed that environmental factors associated with parents’ migration process contribute to the development of ASD in their offspring, particularly for those with comorbid ID (Magnusson et al. 2012). In our study group, three children (9.7%) were born in the year before the migration, and two of them had ID. Due to limited material, we cannot draw any conclusion from this finding. Migration is often a stressful life event, and there are stressors related to maternal migration that can influence the quality of pre- and perinatal periods. The stressors can have the potential to cause epigenetic changes during the pregnancy and delivery with the associated risk of ASD (Crafa and Warfa 2015).

In a systematic review of epidemiological surveys of ASD, Elsabbagh et al. (2012) have found that, in particular, in low and middle-income countries, there are no data available on the prevalence of ASD. Thus, future studies may demonstrate if the increased risk for ASD in children of immigrant women born outside Europe, Australia or the United States is due to factors affected by the migration itself or may be explained by differences in the prevalence of ASD in different populations.

The exact etiology of ASD remains unclear, and there is lack of consistency in terms of association between specific risk factors and ASD. Genetics has a key role in the etiology, together with environmental factors during the child’s early development. In our study group, an array of risk factors was observed, including maternal immigrant status, clustering of pregnancy and birth complications, heredity, and genetic factors. An increasing number of studies highlight maternal migration as a risk factor for ASD (Crafa and Warfa 2015; Ng et al. 2017), and this was also the most common risk factor in our study group. The link between ASD and migration is still unclear. Several other hypotheses have been proposed concerning the role of migration in ASD, but have not yet been thoroughly tested. These include for example maternal vitamin D deficiency (Mazahery et al. 2016), and maternal stress (Crafa and Warfa 2015). Magnusson et al. (2012) have found that the risk of low-function ASD appeared to be highest when the migration occurred the year prior to birth. This suggests that environmental factors such as a stressful maternal migrations process can affect the neural development and can be considered as a risk factor.

Other known risk factors for ASD are suboptimal pregnancy and birth conditions (Coleman and Gillberg 2012; Guinchat et al. 2012; Ng et al. 2017). In our study group, various prenatal and perinatal risk factors were identified, in addition to maternal migration status, but we cannot evaluate their significance or their potential causal mechanisms, due to the small study group.

ASD has strong genetic influences and twin studies have shown that the heritability is high (Sandin et al. 2014; Tick et al. 2016). In our study group, 29% of the children had a first/second degree relative with ASD or strong suspicion of ASD, and 58.1% had a relative affected by ASD and/or other neurodevelopmental problems. Bolton et al. (2014) have demonstrated that children in an African cohort showed a higher heritability with 36.9% having a positive family history of ASD, compared to 26.3% in the Irish cohort with an ASD diagnosis. Consanguinity has also been considered as a risk factor for ASD, and seems to be more frequent in some cultures (Mamidala et al. 2015). In the present study, an increased degree of homozygosity was found in three children (9.7%).

Discrepancies across studies can partly be explained by heterogeneity of study design and methodological limitations. There has also been variation in the definition of outcomes and variation in the immigration profiles of different countries. This contributes to difficulties in comparing results from different studies in a reliable way. The present study contributes to an in-depth picture of a smaller group of children with ASD in an immigrant population. It shows that there are several risk factors associated with ASD besides the immigrant status of the mother. More epidemiological studies are needed to understand the risk factors in immigrant populations, and possible interactions between these, and smaller studies with more detailed data can be a complement to the epidemiological studies.

Coexisting conditions in ASD are very prevalent (Gillberg and Billstedt 2000; Soke et al. 2018), and accumulating studies propose that ASD shares considerable etiology with other neurodevelopmental disorders (Lundstrom et al. 2015). In our study, all children had at least one coexisting condition and 64.5% of the children had ID. In a Swedish study, Kantzer et al. (2013) have reported that 36% of the preschool children with ASD had ID. Bolton et al. (2014) have also found a higher proportion of children with ASD and moderate or severe ID (72.2%) in an African group compared to an Irish group (16.3%). The risk for ASD with ID seems to be higher when families migrated from regions with a low HDI or low-income country (Abdullahi et al. 2019a, b; Esler et al. 2017). In our study, 45.2% of the mothers were born in these regions. This finding raises important questions about potential biological and/or environmental factors that may contribute to the pattern of greater intellectual impairment among children of immigrant parents from regions with low HDI. Several other hypotheses have also been conceptualized to understand the higher frequency of ASD with ID among children of immigrant families, such as vitamin D deficiency, more socioeconomic disadvantages, reduced level of maternal care during the obstetric period in children with an immigrant history, as well as cultural bias in diagnostic procedures related to language barriers and lack of cultural knowledge (Fernell et al. 2015; Haglund and Kallen 2011; Langridge et al. 2013; Magnusson et al. 2012). Abdullahi et al. (2019a) have proposed that children of mothers from foreign-born backgrounds are more likely to be diagnosed younger and to present more severe traits of ASD and associated ID. However, children with ASD and normal IQ may be underdiagnosed in immigrant populations (Fairthorne et al. 2017; Magnusson et al. 2012). In the present target population, there are probably also children with undiagnosed ASD and AIF whose difficulties will become more apparent during the first school years.

It is imperative that health care professionals gain knowledge from current research on early symptoms in ASD, treatment approaches and intervention opportunities. At the same time, more knowledge is needed on how cultural values in different populations might influence different views of children’s development in general, as well as ASD.

The families in our study lived in low resource settings. Most of the families had several social stress factors, such as unemployment, separated parents, mental or physical health problems of parents, and in several families, there were siblings with neurodevelopmental disorders. Many families had economic problems and insufficient housing, and some of the families also lacked permanent residency. In addition, all children were affected by co-existing conditions. The co-existing conditions provide an increased need for support from health services and affect the life of the whole family (Petrou et al. 2018; Posserud et al. 2018).

New models are needed in health care to increase accessibility to care, as well as to meet the many and individual needs of young children with ASD and their families especially in immigrant communities. These models need to counteract the gaps in health care. We propose one geographical and organizational unit, with a multi-professional team, for both assessment and interventions for young children with ASD. Treatment approaches to reduce the impact of co-existing conditions whilst providing interventions for ASD are needed, but it is also necessary to take into account the family’s social situation. Vivanti et al. (2018, 2014) have emphasized the importance of understanding and taking into account the child’s characteristics as well as the family’s priorities/expectations, needs, and life situation in the selection of treatment objectives and treatment strategies. The importance of individually tailored interventions pursuant to the child’s needs as well as the parent’s resources and life situation, has been highlighted in a study by Nilses et al. (2019) from our research group.

All the children included in the study have been assessed by the same team, consisting of medical, psychological, educational and speech therapy professionals with considerable experience of young children with neurodevelopmental disorders, and consistent diagnostic criteria and diagnostic instruments were used throughout.

Nevertheless, the results of the study must be considered within the context of several study limitations. The prevalence of 3.66% ASD is likely a minimum rate for the area, and the real prevalence is more likely to be higher than 4.32%.

The surveillance area included in this study was a relatively small district of Gothenburg, and the findings only represent this area. With a small number of participants, the results cannot necessarily be generalized to the overall population of immigrant children, and the group is too small for statistical calculations. In addition, research conducted with larger samples and within a wider geographic area is needed.

Data collection methods used in this project relied on the review of records, and results are limited to the information in those records. Certain data were not available, such as maternal vitamin D status, smoking, toxins or illnesses during pregnancy. Therefore, there may be more prenatal and perinatal risk factors than those described in the result.

The strengths of our study include its population-based design. The universal health care system, and the screening programme for ASD in the immigrant area makes it likely that a large proportion of individuals in the community have been identified and diagnosed. There was no dropout in the study. All parents who have been asked have consented to their children’s participation.

Another strength of the study is the focus on an understudied group of children, young children with ASD in an immigrant and low SES community, who are otherwise often excluded from research studies in general.