Note: For purposes of this post, the term ‘low functioning autism’ will refer to autism that co-occurs with mental retardation, or an IQ evaluated at less than 70. The validity of this accepted nomenclature is not the topic of this post.
The prevalence of low functioning autism among 2-9 year-old Amish children in Lancaster County, Pennsylvania, has a lower bound of 14.2 in 10,000. This is in line with the prevalence of reported autism with mental retardation in the California DDS system (at most 13.1 in 10,000) for the same age cohort. Additionally, it is not too far away from the prevalence of low functioning autism as ascertained by more methodologically rigorous epidemiological studies.
While no assertion can be made about the prevalence of high functioning autism among the Amish, the existence of an “Amish anomaly” is unsupported and appears increasingly unlikely.
Dan Olmsted had written a well-known series of articles where he claimed (1) that Amish children are never vaccinated [source] or that only a small percentage are [source], and (2) that the Amish either don’t have autism or that autism is exceedingly rare among the Amish.
These claims were subsequently scrutinized by Prometheus, Autism News Beat and Lisa Jo Rudy. They investigated the matter and found that both claims are most likely mistaken. Autistic children do exist among the Amish, and the Amish do indeed vaccinate. The Amish of Illinois appear to vaccinate at a rate of 90%. Lisa Jo Rudy’s source provided a figure of 70% for the Amish in Lancaster County.
Apologetic responses by Dan Olmsted and Mark Blaxill (PartI and Part II) followed. The gist of the responses is that Dan Olmsted was not actually negligent in his reporting. Olmsted claims he did attempt to contact The Clinic For Special Children without much success. Additionally, Olmsted and Blaxill claim that while the Amish do vaccinate, they do so at a lower rate than that of surrounding communities, and therefore, they argue, the finding of an “Amish anomaly” is still relevant to the hypothesis that vaccination is an autism risk factor. (The responses also appear to indicate that those who scrutinize Olmsted’s reporting are “wackos” and say “some stupid things” — I’m guessing this was done to discourage further criticism).
Notwithstanding these responses, it is clear that the credibility of Olmsted’s reporting suffered a substantial blow. Take, for example, one of his most popular findings: That the few autistic children he came across happened to be ones who were vaccinated. This finding is rendered completely inconsequential by the mere fact that most Amish children are vaccinated at least once.
What I really want to address in this post, nevertheless, is the persistent claim that autism is exceedingly rare among the Amish, with a focus on low functioning autism.
Why low functioning?
Dan Olmsted was not likely looking for high functioning children, and even if he was, it is improbable he had the means to properly ascertain the prevalence of high functioning autism. The expertise required for such an endeavor is non-trivial. A methodology that consists of “asking around” is not likely to produce an accurate result.
Furthermore, Dr. Strauss of The Clinic For Special Children had told Autism News Beat that they do not screen for ASDs, nor do they see high functioning autistic children at their clinic.
Strauss says he doesn’t see “idiopathic autism” at the clinic – children with average or above average IQs who display autistic behavior. “My personal experience is we don’t see a lot of Amish children with idiopathic autism. It doesn’t mean they don’t exist, only that we aren’t seeing them at the clinic.”
Strauss says a child in the general population is more likely to have autism detected early and to receive a diagnosis than an Amish child. “An Amish child may not be referred to an MD or psychologist because the child is managed in the community, where they have special teachers,” he says. “We know autism when we see it, but we don’t go actively into the Amish community and screen for ASD.”
With all due respect to Dr. Strauss, I think a clarification is in order before we proceed. I believe he mixed up terminology in his statement to ANB. Idiopathic autism simply means that no cause has been determined for a particular child’s autism. There may or may not be mental retardation in idiopathic autism. The same is true of non-idiopathic autism.
It seems plausible that The Clinic For Special Children generally only sees autistic children who are low functioning. The important thing is that this is the case of the children in a study we will discuss shortly.
Another important clarification is that when I talk about low functioning autism I am not referring to Autistic Disorder vs. “full syndrome autism” vs. PDD-NOS vs. Asperger’s Syndrome. This appears to be a common point of confusion. There may or may not be mental retardation in Autistic Disorder. It is not a requirement in the diagnosis.
Minimum Prevalence Calculation
I am not going to count autistic children Dan Olmsted claims to have found. This is because it is unclear how many exactly he found where, given his reporting methodology, nor is it clear what the characteristics of the children are. Furthermore, I personally don’t consider his reporting reliable.
Thankfully, there is a peer-reviewed study we can use: Strauss et al. (2006). This is a study of Old Order Amish children from Lancaster County. In Table 1 of the paper we see they look at 9 patients, ages 2 to 9, 100% of whom have mental retardation, and 67% of whom have pervasive developmental delay or autism. In other words, this study identifies 6 low functioning autistic children. Let’s consider this the minimum number of autistic children, ages 2 to 9, among the Amish of Lancaster County. At this point we just don’t know if The Clinic For Special Children has come across other low functioning autistic children.
Also of note is that the children “had no distinguishing physical features, and growth trajectories were normal, although all patients had relatively large heads and diminished or absent deep-tendon reflexes.” This tells me that there was nothing noticeably odd about these autistic children that would lead someone to believe they are different to non-Amish autistic children. The fact that a genetic etiology was determined in these cases by top geneticists is of no consequence to a prevalence calculation in my view.
The Amish population of Lancaster County is about 25,200 (Raber’s 2004 Almanac via this web page). Their life expectancy is similar to that of the US general population. However, they have high population growth, with their population doubling every 20 years. This is similar to population doubling in India, which I will use as a reference to calculate the number of 2 to 9 year olds among the Pennsylvania Amish. According to the CIA World Factbook, 31.5% of the Indian population is 0 to 14 years of age. Extrapolating from this, I will estimate that 16.8% of the Lancaster County Amish are 2 to 9. In other words, there are 4,234 such children.
Since we know there are at least 6 low functioning autistic children in this community, the absolute minimum prevalence of low functioning autism for the cohort is 14.17 in 10,000.
No methodologically sound exhaustive autism screening has been carried out in the Amish population. We only know about cases that have been reported. This is why I thought a comparison with California reporting would be fair. It may not be precise, since the ascertainment methodologies are not the same, but I don’t believe it can be called an unfair comparison. Further, I do not believe the well known limitations of California DDS data make the comparison unfair either.
As of December 2007, there are a total of 17,029 children 2 to 9 in California classified as autistic and who have sought DDS services [source]. Most of them do not have mental retardation. We don’t know exactly how many have mental retardation, but we can determine a ceiling on that. For this, I counted how many autistics total have mental retardation, plus – to be fair – how many have “unknown” mental retardation status. This came to 32.67% of all autistics in the system. Since this proportion is known to have dropped considerably as more young children have been diagnosed in recent times, it follows that 32.67% is an upper bound percentage when it comes to children. In other words, there should be at most 5,564 known low functioning autistic 2-9 year olds in the California system.
Since the 2-9 population of California is 4,248,774 (2006), the prevalence of reported low functioning autism for the cohort is at most 13.1 in 10,000.
Comparison to Epidemiology
I often discuss how the prevalence of already recognized autism, and the prevalence of screened autism (or true prevalence) are two very different things. In particular, when autism is thoroughly screened in populations with intellectual disabilities, a lot of autism which wasn’t previously thought to be there is suddenly found (e.g. Shah et al. 1982; La Malfa et al. 2004; Matson et al. 2007).
Epidemiological studies are not necessarily perfect in this regard, but their methodology goes considerably beyond any work currently available as far as the Amish population is concerned. Let’s take Yeargin-Allsopp et al. (2003), for example.
Cases were identified through screening and abstracting records at multiple medical and educational sources, with case status determined by expert review.
The prevalence of low functioning autism from this Atlanta study was 23.12 in 10,000.
There are two UK studies by Chakrabarti & Fombonne (2001; 2005) ascertaining the 4-6 prevalence of PDD. The derived prevalences of PDD with mental retardation from these studies are 16.15 and 17.5 in 10,000, respectively.
Finally, we have a Chinese study, Zhang & Ji (2001), with a finding of 11 in 10,000 autistic children, all of whom were “intellectually disabled.”