Chapter 6: A Case Study – Autism in the Minnesotan Somali Population
Using a case study of rising autism in a Minnesotan Somali population, this chapter evaluates how epidemiological autism assessment measures, genetics, as well as social and cultural factors together influence prevalence estimates. While measures to assess autism prevalence such as broad diagnostic criteria and use of educational data are likely to contribute significantly to higher autism levels overall, genetic factors do not appear to play a significant role in the autism prevalence differences in the Minnesotan populations. While studies of Somali culture and autism in African immigrants suggests that culture could play a substantial role in autism diagnosis and services, analysis of autism prevalence in relation to the prevalence of Somalis and African Americans in the United States suggests that cultural differences are most likely to explain observed variation in Minnesotan autism prevalence rates. Ultimately, this case study suggests that cultural influences on autism assessment measures are most likely to contribute to observed racial and ethnic differences in autism prevalence.
Race, Ethnicity, and Autism in Minnesota
In response to parental concerns of rising autism in Somali parents, the Minnesota Department of Health (MDH) measured the administrative prevalence of ASD in Minneapolis. In 2008, the MDH responded to Somali parents’ concern about the large number of Somali Children in the Early Childhood Special Education (ECSE) Citywide ASD Classroom Program by determining the administrative prevalence of autism in 3 and 4 year old preschoolers living in Minneapolis (1). Administrative prevalence was defined as the percentage of 3 to 4 year old preschool children in Minneapolis able to receive ASD related services through the Minnesota Public School system (1).While Asian, black, Hispanic, Native American, and white race were determined through general public school data when identifying cases, Somali origin was determined by language spoken at home (1). Birth certificate data was used to determine the overall number of children in each racial and ethnic group for prevalence calculations (1). Prevalence was determine for all children age 3 and 4 eligible for ASD services in a one year period for three school years (05-06, 06-07, and 07-08)(1).
The MDH found higher administrative ASD prevalence in the Somali population and lower administrative ASD prevalence in Asian and Native American populations in Minneapolis. While administrative ASD prevalence for non-Somali children ranged from 0.21% to 0.72%, Somali ASD prevalence ranged from 0.93% to 1.54% (1). Somali prevalence was 2 to 7 times greater than non-Somali prevalence depending on the model used (1). Differences in Somali and non-Somali ASD prevalence decreased over the three year period, and low Asian and Native American administrative prevalence was also found (1). While significant results regarding ethnic variation in autism prevalence were obtained, various study limitations may account for these differences.
These analyses’ main limitations include errors in prevalence assessment measures, failure to investigate genetic, social, and cultural contribution, and their narrow applicability. As educational data was used to identify autism cases, not all children considered autistic in this study may have a medical diagnosis of autism (1). Misclassification of autistic children and misidentification of Somali origin based on language spoken at home in educational data may also lead to errors in identifying autistic cases (1). Errors in estimating the Somali population size due to an increase in migration of Somalis to Minnesota since the 2000 census may lead to larger Somali prevalence estimates by creating an artificially low denominator for prevalence calculations (1). Additionally, errors in establishing Somali ethnicity could occur as cases were establishing using language and total population ethnicity was established using birth location (1). As Minnesota had no public health autism surveillance system, no baseline ASD rates exists for the population, making assumptions about ASD prevalence change over time impossible (1). Other systematic differences including differences in public school methods used to identify autism cases may influence prevalence if children of a particular racial or ethnic group tend to attend the same schools (1). Genetic and social factors could also contribute to the high administrative prevalence observed in the Minnesotan Somali population. These study limitations will be discussed in further detail subsequently with proposed methods for their amelioration.
Epidemiological Autism Assessment Measures
Autism prevalence is estimated from school, hospital, clinical or specialty service records or from face-to-face interview and diagnosis (23, 32). Although interview and diagnosis is the best method as it confirms consistent evaluation of all subjects, it is impractical to apply to all cases and is often only used as a second stage screening technique (23). Records from health or psychological professionals are the next best option; however, differences in diagnosis across professionals and difficulty obtaining these records due to privacy measures often make this method difficult to practically apply (23). In the U.S., educational data is about autism is readily available because of the individuals with Disabilities Education Act (IDEA), which requires annual reporting of those receiving public special education (4). Although this data is regularly available, it often reports unreliable prevalence.
Previous research suggests that educational data give unreliable autism prevalence measures because of inconsistent definitions of autism. A study by Laidler et al. demonstrated that education-based autism prevalence increases at a constant rate as a birth cohort ages with an exception of ages 11 and 12 (17). As previous research suggests that most autism diagnoses are made before age 8 and that the transition from elementary to middle school should expose autism cases, this evidence suggests that educational data does not report accurate autism prevalence patterns (17, 18). Other studies comparing epidemiological prevalence measures to educational prevalence measures suggests that educational data underestimates prevalence (19, 28). While national IDEA criteria exist for autism, states are free to modify such criteria, making data difficult to compare across states because of the definition’s varying specificity (19, 27). Despite these difficulties with national reliability, all educational data tends to report an increase in autism over time, which some researchers believe is due to increased diagnostic substitution, or classifying diseases that would previously be classified as another disorder as autistic (19, 28).
The lack of complete medical and birth certificate information of autistic patients decreased the reliability of autism prevalence estimates in the MDH report. Included data consisted of Minnesota birth certificates and special education administrative data such as ASD disability category, special education programming, and other demographic variables (1). Due to privacy restrictions the birth certificate and education information could not be linked; this forced the researchers to make assumptions about which children met the criteria to be included in their birth cohort and decreased the reliability of prevalence estimates (1). Classifying children with multiple diagnoses according to the program that best fits their educational needs also decreased the reliability of autism estimates as included cases may not be autistic (1). Finally, autism prevalence estimates may be artificially low because not all autistic children participate in the program from which the educational data was taken (1). As the second chapter demonstrated that differences in screening methods are likely to impact global ASD prevalence, it is logical that using education data alone leads to unreliable autism prevalence estimates. If racial and ethnic differences exist in obtaining autism educational diagnosis or in participating in the program, prevalence estimates will not reflect true racial and ethnic differences in autism prevalence.
The broad Minnesotan criteria defining autism may allow for a stronger influence of culture on prevalence estimates. The national IDEA criteria define autism as a developmental disability causing repetitive behaviors as well as impaired social behavior and verbal and non-verbal communication generally before age 3. While some states narrow this definition to exclude Asperger’s syndrome, the Minnesotan rule includes all forms of autism, Asperger’s syndrome, and other PDDs and does not include an age by which symptoms should be observed (19, 30). As Chapter 2 demonstrated that changing diagnostic criteria have a substantial impact on autistic disorder and ASD prevalence globally, it is likely that broad Minnesotan ASD diagnostic criteria significantly alters prevalence patterns. As different cultures may perceive mild autistic deficits as normal, this broad definition could lead to larger racial and ethnic differences in autism prevalence. Analyzing the educational programs of different autistic individuals in different ethnic groups would shed light on how much cultural interpretation of autistic symptoms influences educational classification. Additionally, these broad diagnostic criteria explain Minnesota’s overall high autism prevalence compared to other states (2, 13, 28).
Future Minnesotan efforts to determine autism prevalence should include both educational and medical data as well as clinician review to increase reliability and reduce the potential for racial and ethnic bias. The CDC’s Autism and Developmental Disabilities Monitoring Network (ADDM) first identifies children through educational and medical records from schools, hospitals, and other similar areas (23). While racial and ethnic disparities may prevent certain groups from accessing particular autism services, pooling data from multiple sources increases the chance that all individuals with autism will be identified regardless of ethnicity. In the second phase, clinician reviewers analyze data to determine whether the given behavior data meets an established autism criteria (23). By using central reviewers with consistent autism diagnostic criteria, the impact of cultural interpretation of autistic symptoms is reduced. Although instituting such a system requires much planning, it is necessary to determine whether observed racial and ethnic differences in autism prevalence are real or artifacts of autism assessment measures.
The lack of evidence supporting variations in autism-associated genes in Somali and other ethnic populations suggests that genetic variation does not explain the observed Minnesotan prevalence patterns. Genetic study analyzing the Y chromosome in a Somali male population found that the 77.6% are part of the E3b1 haplogroup, commonly found in those of Ethiopian, Egyptian, and other East African origin, and that 10.4% are part of the K2 haplogroup, of Eurasian origin (25). Analyses of maternal mitochondrial DNA confirm the relationship of Somali populations with European populations (6). While unique SNPs in cytokine and cytokine receptor genes have been found in Somali populations but not Caucasian populations, Somali populations have not been systematically studied for unique autism-associated gene variants (9). Chapter 3 suggested that while racial and ethnic autism-associated gene variants may exist, it is unlikely that these differences would be substantial enough to contribute to differences in global autism prevalence. Based on these findings, the lack of evidence supporting Somali-specific genetic variants, and the decreasing prevalence gap over time, it is unlikely that genetic factors contribute to high prevalence of autism in Minnesotan Somalis. Future studies examining Somali populations for genetic variants of neurexin (NRXN), GABA a Receptor alpha 4 (GABRA4), GABA A receptor beta 1 (GABRB1), tryptophan hydroxylase 2 (TPH2), and neurofibromin 1 (NF1) as Caucasian and African American populations have showed different variants at those genes (5, 7, 8, 10, 22) If substantial genetic variation is found, neurobiological studies investigating gene products in Somali and other populations should be conducted.
Social and Cultural Factors
It is unlikely that earlier autism diagnosis of Somali children contributes to higher autism prevalence rates. While a study conducted in Denmark by Parner et al. suggests that autism earlier diagnosis leads to higher prevalence American studies have not been able to confirm racial and ethnic variation in the timing of autism diagnosis (1, 18, 21, 27). The MDH reports age of first contact with the public school system, noting that Somali children came into contact with the public school system at slightly older ages than non-Somali children in 2005-06 but not at other ages (1). Although this slight age difference is probably not large enough to account for the observed racial and ethnic variation in autism prevalence, the older age of Somali diagnosis suggests they should have lower autism prevalence instead of a higher prevalence. The contradiction between these claims suggests that ethnic differences in timing of diagnosis do not contribute to the observed Minnesotan prevalence patterns.
Somali culture may lead to a greater sensitivity to the autism diagnosis among the Somali population and well as higher rates of autism diagnosis. Somalia is an Eastern African country with a strong patriarchal family clan structure whose population is 98% Sunni Muslim (12). Civil war beginning in the 1960’s caused about 45% of Somalis to leave the country, with large amount entering Minnesota around 1993 (12, 26). A study analyzed the view of 5 Minnesotan Somali women on disability suggested that Somalis view mental disabilities as more severe than physical, that the family typically takes care of disabled children, and that it is Allah’s will if a child is disabled (12). Another study focusing solely on the views of Minnesotan Somalis on mental illness suggests that Somalis may believe that mental illness is cased by evil spirits (26). While not all Somalis share this view, there is a stigma attached to mental illness (26). A Canadian study of Somali mothers of children with ASD suggests that Somali parents have an increased mistrust of physicians and educators due to cultural barriers and the lack of information on autism’s etiology (15). The perception of few autism cases in Somalia coupled with the negative attitude towards mental illness suggests that Somali parents would be especially sensitive to the autism diagnosis. This increased sensitivity may have prompted the initial MDH report. Cultural differences in acceptable social interactions as well as difficulties learning both English and Somali may lead to increased autism diagnosis through a misinterpretation of culturally normal behavior as autistic symptoms. Researchers should study how Somali parents perceive autism and associated educational services as well as how physicians interpret Somali culture.
Increased numbers of Somalis in high income school districts with well trained special education faculty may lead to increased autism prevalence in Somali populations. Previous studies associated lower autism prevalence with poorer school districts (20). This suggests that low-income school districts may have special education staff that is not as well trained at identifying autism, leading to lower prevalence levels. If Minneapolis Somali populations are clustered in high-income school districts, this may lead to better identification of autism cases. Further research on the quality of the special education staff at schools will high Somali attendance could better determine if this is a valid explanation for high Minnesotan Somali autism prevalence.
While studies have reported that immigrants, particularly African, have higher rates of autism, these differences are likely to be due to culturally factors instead of racial ones. A Swedish study reported a higher prevalence of autistic disorder and PDDNOS in Somali children. As the Somali prevalence matched previous estimates in Swedish populations but the non-Somali estimate was significantly lower, selection bias in non-Somali cases may explain this difference (1, 3). Additional small studies have identified increased autism prevalence in Caribbean, Ugandan, and other populations (11). Immigrant status does not universally confer autism risk, however, as an Israeli study found that Ethiopian immigrants had lower autism prevalence (14). A recent U.K. study further suggests that increased autism risk is due to differences in immigrant culture instead of race (16). The study found that autistic children have a greater chance of having a Caribbean, African, or Asian mother. While the study also found an increased risk for mothers of black origin, this finding was no longer significant when accounting for region of birth, suggesting that race alone is not a risk factor for autism (16). As white immigrants showed no increased risk of autism and are more likely to have a similar culture to U.K. women, culture seems to be the cause of this rise in autism in immigrants (16). As the Somali population in Minnesota consists primarily of immigrants, it seems most likely that their differences in culture account for the observed variation in autism prevalence.
National Estimates of Race, Ethnicity and Autism
To determine whether Somali ethnicity was associated with autism, I studied the relationship between autism prevalence, Somali ethnicity, immigrant status, and African American race on a national level. Autism prevalence values for each state were calculated by the Thoughtful House Center for Children using Child Count data from the Individuals with Disabilities Education Act (4, 31). CDC live birth data were used to calculate overall prevalence (29). I used the integrated public use microdata series (IPUMS-USA) 2005 American Community Survey (ACS) sample to obtain information on Somali heritage and race by state (24). The ACS was designed to replace the census long form and is conducted via mail with follow up via telephone and face-to-face interviews (24). Provided weights adjust for geographic sampling, non response, and other survey specific methods (24). Although demographic information was available by smaller geographical regions than the state, the lack of autism prevalence data for smaller geographical regions prevented a more specific analysis. While race was assessed directly, Somali nationality was determined through the primary ancestry and ethnic origin question and immigrant status was determined by birthplace.
I collected basic descriptive statistics and performed all correlations using Stata 10 software (Stata Corp, College Station, TX). When determining the percentage of individuals of Somali heritage or of a certain racial group in each state, I applied IPUMS-USA person weights using the aweight command in Stata. I correlated percentage of the state population that were of Somali ancestry, African American, Immigrants, and African Immigrants with autism prevalence of 8 year old children in 2005.
While both percentage of population that was of Somali ancestry and African immigrants had significant correlation values, these correlations were fairly weak (R = 0.5041, p = 0.002 and R = 0.3383, p =0.0152)(Figure 6.1). Percentage of the population that are immigrants and African American had not significant low correlation values (R = 0.2280, p = 0.1076 and R = -0.0927, p = 0.5177)(Figure 6.2). The low correlation between Somali ancestry, immigrant status, and African American race suggests that none of these variables influence autism diagnosis enough to drastically impact prevalence levels. The larger and more significant correlation values of Somali ancestry and African immigrant status compared to race suggest that differences in ethnicity and culture contribute more to autism variation than differences in race and skin color alone.
Limitations to these analyses include the use of population level data and educational data to estimate autism prevalence. With population level data, one cannot be sure that co-occurring increases in autism prevalence and a racial or ethnic characteristic suggest increasing numbers of autistic individuals with that racial or ethnic characteristic. Further studies on populations with information on racial, ethnic, and autism status should be conducted to better assess the association between Somali heritage and autism. Additionally, educational data is not the best way to estimate autism prevalence as individuals may be classified as autistic to receive certain educational treatment without actually having the disorder. Clinical confirmation of autism status can avoid this bias.
Close analyses of the increasing Minnesotan prevalence of autism in the Somali population suggests that cultural influence on diagnostic measures in the most likely method by which race and ethnicity contribute to differences in autism prevalence. Education-based prevalence estimates allow room for cultural differences in the perception of autistic symptoms to play a substantial role. Although genetic differences may exist between Somali and American populations, further studies using large amounts of Somali immigrants are needed to confirm such speculative claims. Ultimately, additional research including cultural work groups need to be conducted to determine how specifically culture leads to higher autism prevalence and methods to correct for such cultural differences should be implemented.
1. 2009. Autism Spectrum Disorders Among Preschool Children Participating in the Minneapolis Public Schools Early Childhood Special Education Programs. In M. D. o. Health (ed.). Minnesota Department of Health, St. Paul.
2. Barbaresi, W. J., S. K. Katusic, R. C. Colligan, A. L. Weaver, and S. J. Jacobsen. 2005. The Incidence of Autism in Olmsted County, Minnesota, 1976-1997: Results from a Population-Based Study. Archives of Pediatric Adolescent Medicine 159:37-44.
3. Barnevik-Olsson, M., C. Gillberg, and E. Fernell. 2008. Prevalence of Autism in Children Born to Somali Parents Living in Sweden: A Brief Report. Developmenal Medicine and Child Neurology 50:598-601.
4. Center, D. A. 2010, posting date. 2005 Individuals with Disabilities Education Act (IDEA) Child Count Data. [Online.]
5. Collins, A. L., D. Ma, P. L. Whitehead, E. R. Martin, H. H. WRight, R. K. Abramson, J. P. Hussman, J. L. Haines, M. L. Cuccaro, J. R. Gilbert, and M. A. Pericak-Vance. 2006. Investigation of autism and GABA receptor subunit genes in multiple ethnic groups. Neurogenetics 7:167-174.
6. Comas, D., R. Reynolds, and A. Sajantila. 1999. Analysis of mtDNA HVRII in severa human populations using an immobilised SSO probe hybridization assay. European Journal of Human Genetics 7:459-468.
7. Coon, H., D. Dunn, J. Lainhart, J. Miller, C. Hamil, A. Battaglia, R. Tancredi, M. F. Leppert, R. Weiss, and W. McMahon. 2005. Possible association between autism and variants in the brain-expressed tryptophan hydroxylase gene (TPH2). American Journal of Medical Genetics Part B 135B:42-46.
8. Delorme, R., C. M. Durand, C. Betancur, M. Wagner, S. Ruhrmann, H. J. Grabe, G. Nygren, C. Gillberg, M. Leboyer, T. Bourgeron, P. Courtet, F. Jollant, C. Buresi, J. M. Aubry, P. Baud, G. Bondolfi, G. Bertschy, N. Perroud, and A. Malafosse. 2006. No human tryptophan hydroxylase-2 gene R441H mutation in a large cohort of psychiatric patients and control subjects. Biological Psychiatry 60:202-3.
9. Dhiman, N., I. G. Ovsyannikova, R. A. Vierkant, S. Pankratz, R. M. Jacobson, and G. A. Poland. 2008. Associations between Cytokine/Cytokine Receptor SNPs and Humoral Immunity to Measles, Mumps, and Rubella in a Somali Populations. Tissue Antigens 72:211-220.
10. Feng, J., R. Schroer, J. Yan, W. Song, C. Yang, A. Bockholt, E. H. Cook, C. Skinner, C. E. Schwartz, and S. S. Sommer. 2006. High frequency of neurexin 1beta signal peptide structural variants in patients with autism. Neuroscience Letters 409:10-13.
11. Fombonne, E. 2003, posting date. Epidemiological surveys of autism and other pervasive developmental disorders: an update. Plenum Press. [Online.]
12. Greeson, C. J., P. M. Veach, and B. S. LeRoy. 2001. A Qualitiative Investigation of Somali Immigrant Perceptions of Disability: Implications for Genetic Counseling. Journal of Genetic Counseling 10:359-378.
13. Gurney, J. G., M. S. Fritz, K. K. Ness, P. Sievers, C. J. Newschaffer, and E. G. Shapiro. 2003. Analysis of Prevalence Trends of Autism Spectrum Disorder in Minnesota. Archives of Pediatric Adolescent Medicine 157:622-627.
14. Kamer, A., A. H. Zohar, R. Youngmann, G. W. Diamond, D. Inbar, and Y. Senecky. 2004. A Prevalence Estimate of Pervasive Developmental Disorders Among Immigrants to Israel and Israeli Natives. Social Psychiatry and Psychiatric Epidemiology 39:141-145.
15. Kediye, F., A. Valeo, and R. C. Berman. 2009. Somali-Canadian Mothers' Experiencs in Parenting a Child with Autism Spectrum Disorder. Journal of the Association for Research on Mothering 11:211-223.
16. Keen, D. V., F. D. Reid, and D. Arnone. 2010. Autism, Ethnicity, and Maternal Immigration. The British Journal of Psychiatry
17. Laidler, J. R. 2005. US Department of Education Data on "Autism" are not Reliable for Tracking Autism Prevalence. Pediatrics 116:e120-e124.
18. Mandell, D. S., J. Listerud, S. E. Levy, and J. A. Pinto-Martin. 2002. Race Differences in the Age at Diagnosis Amogn Medicaid-Eligible Children with Autism. Journal of the American Academy of Child Adolescent Psychiatry 41:1447-1453.
19. Newschaffer, C. J., M. D. Falb, and J. G. Gurney. 2005. National Autism Prevalence Trends from United States Special Education Data. Pediatrics 115:e277-e282.
20. Palmer, R. F., S. Blanchard, C. R. Jean, and D. S. Mandell. 2005. School District Resources and Identification of Children with Autistic Disorder. American Journal of Public Health 95:125-130.
21. Parner, E. T., D. E. Schendel, and P. Thorsen. 2008. Autism Prevalence Trends Over Time in Denmark: Changes in Prevalence and Age of Diagnosis. Archives of Pediatric Adolescent Medicine 162:1150-1156.
22. Plank, S. M., S. A. Copeland-Yates, K. Sossey-Alaoui, J. M. Bell, R. J. Schroer, C. Skinner, and R. C. Michaelis. 2001. Lack of Association of the (AAAT)6 Allele of the GXAlu Tetranucleotide Repeat in Intron 27b of the NF1 Gene with Autism. American Journal of Medical Genetics 105:404-405.
23. Rice, C. E., J. Baio, K. Van Naardan Braun, N. Boernberg, and Meaney. 2007. A Public Health Collaboration for the Surveillance of Autism Spectrum Disorders. Paediatric and Perinatal Epidemiology 21:179-190.
24. Ruggles, S., J. T. Alexander, K. Genadek, R. Goeken, M. B. Schroeder, and M. Sobek. 2010. Integrated Public Use Microdata Series: Version 5.0 [Machine-readable database]. University of Minnesota, Minneapolis.
25. Sanchez, J. J., C. Hallenberg, C. Borsting, A. Hernandez, and N. Morling. 2005. High Frequencies of Y Chromosome Lineages Characterized by E3b1, DYS19-11, DYS392-12 in Somali Males. European Journal of Human Genetics 13:856-866.
26. Scuglik, D. L., R. D. Alarcon, A. C. Lapeyre, M. D. Williams, and K. M. Logan. 2007. When Poetry No Longer Rhymes: Mental Health Issues Among Somali Immigrants in the USA. Transcultural Psychiatry 44:581-595.
27. Shattuck, P., M. Durkin, M. Maenner, C. J. Newschaffer, D. S. Mandell, L. Wiggins, L.-C. Lee, C. Rice, E. Giarelli, R. Kirby, J. Baio, J. Pinto-Martin, and C. Cuniff. 2009. Timing in Identification Among Children with an Autism Spectrum Disorder: Findings From a Population-Based Surveillance Study. Journal of the American Academy of Child Adolescent Psychiatry 48:474-483.
28. Shattuck, P. T. 2006. The Contribution of Diagnostic Substitution to the Growing Administrative Prevalence of Autism in US Special Education. Pediatrics 117:1028-1037.
29. Statistics, C. f. D. C.-D. o. V. 2010, posting date. Vital Statistics Data - Birth Data. [Online.]
30. Statutes, R. o. 1999. Autism Spectrum Disorders (ASD). In S. o. Minnesota (ed.), vol. 3525.1325, Minnesota.
31. Thoughtful House Center for Children, G. I. P., FightingAutism. Austin, TX.
32. Williams, J. G., P. T. Higgins, and E. G. Brayne. 2005. Systematic review of prevalence studies of autism spectrum disorders. Archives of Disease in Childhood 91:8.