Autism in Somali Immigrant Communities: Epidemic, Coincidence or Genetic Pattern?

EXECUTIVE SUMMARY: Health-care and special-education professionals in the city of Minneapolis, Minnesota are noticing what they think is an unusually high proportion of severely autistic children within Minneapolis's Somali immigrant community. The Minnesota Department of Health is looking into it, trying to determine whether there really is a higher rate of autism prevalence among Somalis in Minneapolis.

A similar study was done last year in Sweden, looking at autism prevalence among Somali immigrants in Stockholm. While that study did find an elevated rate of diagnosed ASDs among Somali children, the group of Somali children with autism diagnoses was so small (seventeen individuals) that the finding of increased autism prevalence should not be taken very seriously.

Both the Swedish study and the planned Minnesota study involve looking at special-education statistics for their estimates of how many children --- both in the general population and within the Somali immigrant population --- have autism diagnoses. There are several problems with these data, particularly in Minneapolis, whose autism programs are open to all Minnesotans and thus can be expected to draw people from all over the state.
_____________________________________________

Via the What Sorts of People? blog, I found this New York Times article about a possible "autism cluster" in the Somali community of Minneapolis, Minnesota.

What is an autism cluster?

A "cluster" is an unusual aggregation, real or perceived, of health events that are grouped together in time and space and that is reported to a public health department (CDC 1990). Several breakthroughs and triumphs in infectious disease control have resulted from the epidemiologic evaluation of clusters of cases.
...
Investigations of noninfectious disease clusters have also resulted in notable examples of breakthroughs linking a particular health effect to an exposure, such as angiosarcoma among vinyl chloride workers (Waxweiler et al. 1976), neurotoxicity and infertility among kepone workers (Cannon et al. 1978), dermatitis and skin cancer among people wearing radioactively contaminated gold rings (Baptiste et al. 1984), adenocarcinoma of the vagina and maternal consumption of diethylstilbestrol (Herbst et al. 1971) and phocomelia and consumption of thalidomide (McBride 1961).

Whenever a cluster of a noninfectious disease occurs, questions arise as to what could have caused it:

While there is little research on autism clusters, reports of cancer clusters are so common that health agencies across the country respond to more than 1,000 inquiries about suspected ones each year. A vast majority prove unfounded, and even when one is confirmed, the cause is seldom ascertained, as it was for Kaposi's sarcoma among gay men and mesothelioma among asbestos workers [or any of the conditions listed in the above quotation].

It is "extraordinarily difficult" to separate chance clusters from those in which everyone was exposed to the same carcinogen, said Dr. Michael J. Thun, the American Cancer Society's vice president for epidemiology.

Since the cause of autism is unknown, the authorities in Minnesota say it is hard to know even what to investigate.

According to the "Autism and the Somali Community" page on the Minnesota Department of Health website, the MDH is currently analyzing data on Somali children in the Minneapolis Autism Program, and will release a report on its findings on April 1*.

Until then, we've still got a similar study of the prevalence of autism in the children of Somali immigrants in Stockholm, Sweden. The methods this study's authors (Dr. Martina Barnevik-Olsson, Dr. Christopher Gillberg** and Dr. Elisabeth Fernell) used were quite similar: as the MDH plans to do, Olsson, Gillberg and Fernell reviewed records of all Somali children (either Somali-born, or children of Somali-born parent/s) diagnosed with an autism spectrum disorder and enrolled in autism support programs, determining the prevalence of autism both within the Somali community and in the larger city, and also trying to identify any common factors in the children's medical histories.

The Swedish study did find an elevated prevalence rate among the Somalis of Stockholm: Somali children were about three or four times as likely as non-Somali children to have a diagnosis of autism or PDD-NOS. (Prevalence rates for these groups were 0.7% and 0.19%, respectively).

Unfortunately, the group of Somali children diagnosed with an ASD was made up of only seventeen people. Those seventeen, as a percentage of the 2,437 Somali children living in Stockholm, represented a bigger chunk of their demographic than the 484 autistic children not of Somali descent did, but with smaller numbers percentages get less reliable.

The study's authors acknowledge this, and also one other potential weakness of their study's design (one the Minneapolis study will probably share):

One limitation of this study is the retrospective collection of data, relying on clinical notes that had not been intended for research. Only the autistic disorder or PDDNOS diagnoses for the Somali children were checked and reconfirmed. The data are therefore not complete and must be seen as preliminary.

There were some interesting things they found out, however preliminary and unconfirmed:

Our clinical impression is that this group of Somali children constitutes a rather homogeneous group in the autism spectrum with regard to four characteristics: (1) the age at which the developmental deviation was noted (12-24 mo.); (2) the presence of a definite intellectual impairment in all 17 individuals, mostly of a moderate to severe degree; (3) the fact that motor function was not impaired, except in one child; and (4) the fact that the activity level was exceptionally raised in the vast majority of the children.
...
All individuals had learning disability. This is a developmental disorder in which genetic background factors have a prominent role. Recessive inheritance has been especially demonstrated in severe learning disability. In many parts of the world, especially the Islamic countries[***], marriages between close relatives are common. In a previous study from Stockholm county, covering a population with a high rate of non-European/non-North American immigrants, the prevalence of severe learning disability was found to be 3.7 and 5.9 per 1,000 respectively, in the European and in the non-European population.

The clinical profile shared by the seventeen Somali children in the Swedish study --- particularly the predominance of learning disability and intellectual impairment --- also seems to show up in the Minnesotan children.

From a MinnPost.com article:

About a quarter of all autism [sic] children who attend autism classrooms for students functioning too low to be mainstreamed in regular schoolrooms are Somali. Special education specialists said that indicates that the degree of autism Somali children are developing is on the severe end of the autism spectrum.

"I'm not seeing Aspergers syndrome and the full spectrum of autism in Somali children. It is the more classic forms of autism in general; it is the more severe forms of autism that we're seeing in our Somali babies that are born here," said Anne Harrington, early childhood special education coordinator for the Minneapolis district and a specialist on the topic.

And from the Times story:

In the last decade, [Harrington] said, "we've begun seeing a tremendous number of kids born here who have the most severe forms of autism."
...
"They had classic symptoms," [pediatrician Dr. Daniel S. McLellan] said. "Really impaired language, didn't watch faces, didn't make eye contact, didn't communicate with gestures, just lost in their own worlds. Nobody would mistake it for anything else."

While the Minneapolis Somali community is likely to be larger than its Stockholm counterpart (a conservative estimate from the State of Minnesota puts the number of Somalis living in the state at 14,000-15,000, with most of them living in Minneapolis), there are some other problems likely to emerge with the special-education statistics.

From a news story published last September by the Simons Foundation Autism Research Initiative:

Epidemiologists are generally skeptical of disease clusters, and this one is no exception.

"Those numbers [referring to an earlier finding that Somali children make up 6% of Minneapolis schoolchildren and 17% of its special-ed students designated autistic] are strikingly different, but it's not really an appropriate comparison," says Judy Punyko, an epidemiologist at the Minnesota Department of Health.

The Minneapolis statistics include only children from certain age groups and only those attending public schools, Punyko notes. What's more, Minneapolis autism programs have an open-enrollment policy, meaning they often accept children from other school districts, potentially skewing the prevalence numbers for the district. "I want to get the real numbers and understand these numbers before we move on," Punyko says.

In August, Punyko created a study group of 12 experts --- including epidemiologists, physicians, school administrators and special education teachers --- to compare the autism prevalence across all Somali children in the city with age-matched controls.

Even then, she adds, educational data may be incomplete or inaccurate. The 13 special-education categories reported to the state and federal government are intended to help provide a child with the best available educational services. If a child has two conditions, such as autism and developmental delay, they can only be assigned to one primary category; Minnesota does not require a medical diagnosis of autism to be included in the category. Finally, some schools, especially those in the poorer districts, often overlook milder forms of autism.

So, like the Swedish researchers, the Minneapolis researchers will have to deal with data that probably weren't collected with the same level of rigor they'd normally use to collect epidemiological data. I also suspect that those poorer school districts probably house a lot of Somali immigrant families, which, given the tendency of those districts not to notice "milder" autism, would make the above-cited observation about Somali "clusters" having a disproportionate share of severely autistic children a self-fulfilling prophecy.

*Yes, April Fool's Day. I don't think there's much of a chance of this being a big practical joke, but I am always a bit leery of papers published on April Fool's Day.

**Gillberg's name was already familiar to me, as he's done a lot in the field of autism research, but one thing I hadn't known before Googling him was that he'd proposed a set of diagnostic criteria for Asperger syndrome. They're not the criteria the DSM currently uses, but apparently they're quite faithful to Asperger's original descriptions.

***I'm going to interject here that "the Islamic countries" is a very unhelpful geographic category. Is he aware of just how many countries, in how many different parts of the world, have a lot of Muslims in them? It makes about as much sense as talking about "the Christian countries" --- you could be speaking of Europe, Australia, the Phillippines, Central America, South America, North America, and also large swathes of south and central Africa! Similarly, "the Islamic countries" include huge areas of Asia (all of the Middle East, and much of south and central Asia), Africa, parts of Europe, and Indonesia.

Barnevik-Olsson M, Gillberg C, & Fernell E (2008). Prevalence of autism in children born to Somali parents living in Sweden: a brief report. Developmental medicine and child neurology, 50 (8), 598-601 PMID: 18754897