By Jim Bowery
Version 20030613
Copyright 2003
The author grants the right to copy without modification.
ABSTRACT
The following essay shows very strong (p<.001) "ecological" support for the hypothesis that autism rates are driven by an interaction between populations. One population is susceptible -- primarily of recessive homozygous character such as people of Finnish ancestry (most probably those possessing the HLA A3 allele known to be involved in austim-associated autoimmune disorders as well as associated with multiple sclerosis which is another neuropathology suspected as an autoimmune disorder that also increases with latitude). A second population is the vector -- recent influxes of immigrants from particularly high population density regions of southern Asia. The result was so strongly supportive of my hypothesis that even I was astounded. Validation this interaction and then discovery of its precise nature must be left to detailed research and analysis. However, at the level of State ecological correlations, this association is the most predictive, not only out of the major hypotheses put forth for autism etiology, such as multiple vaccinations, mercury and the "refrigerator mom", but out of an exhaustive search of thousands of combinations of biologically relevant variables under linear, polynomial and exponential relationships. Furthermore, application of this model to continental Europe helps explain the otherwise mysteriously high rate of autism in Finland's northern provinces. In conclusion, the by-State ecological correlations shown below are compelling that this hypothesis should be given research resources to see if its correlations hold or increases at the county level and if so, then at the individual level.
INTRODUCTION
Having worked in Silicon Valley during the 1990s, I was around a number of families who had autistic children during the profound increase of that disease. In one instance a Palo Alto startup of about 100 people had 3 families diagnosed with at least one severely autistic child during a 3 year period, including the first born son of an associate.
I notified an autism epidemiology specialist in UC Berkeley but his response was that "Yes, we know about such microclusters and we don't know the cause." and he didn't indicate any interest in further probing the microcluster that I had reported.
This got me thinking about possible etiology of autism spectrum disorders.
I came up with a hypothesis that I then tested. Alternative hypotheses such as MMR vaccine and "refrigerator mom" hypotheses were also tested the same way.
In my hypothesis, there are populations from closer to human origins, and therefore having undergone a larger degree of what might be called "ecological annealing", that have developed sexual competition mechanisms involving some sort of parasitic castration of competing males. This may be thought of as a form of evolved biological warfare between males. Autism, affecting primarily male children, is also observed to disturb the most sexually dimorphic brain structure: the amygdala. The amygdala, for instance, shrinks substantially, soon after castration.
Deriving a test of this hypothesis from my direct experiences of families of autists in a microcluster, a report that stated 90% of autists are blood type A and a northern Scandinavian center for the ABO A2 allele, I decided to look at a nationality most remote from human origins, the Finns, and see if there was a synergy between demographies containing a disproportionate number of Finns and demographies containing a disproportionate number of recent Indian immigrants. There had been an enormous influx of new immigrants from India that, more than at prior points in history, contained people from all castes. This conjunction of demographies would be compared, in a simple statistical test, against competing theories, for correlations with autism rates in 2000AD.
I then collected demographic data on a state by state basis from sources such as the US Census, Department of Education and Centers for Disease Control. I then looked at all combinations of 2 demographic variables across a range of such variables to see which correlated most strongly with autism rates. At first I was unable to find demographic data breaking down immigration by origin but preliminary results suggested there was a contribution from immigration, as well as a particular susceptibility among populations high in Finnish ancestry. This encouraged me to continue the line of study and I did finally locate immigration broken down by origin.
THE RESULTS
Of the thousands of 2-variable combinations involving biologically relevant variables, the combination with the highest Pearson correlation with autism (60%) rates was the one I predicted:
Finns Percapita * Immigrants from India Percapita
(Please note that "autism spectrum disorders" is a poorly standardized diagnostic category whose reproducibility may be little better than 60%. Even if one identified the specific pathogenic agent causing autism, to which a specific set of genes were susceptible, and were able to test the entire population, it is quite plausible that present diagnostic standards would be little better than 60% at predicting who would have those factors and who wouldn't.)
Furthermore, both of these demographies, alone have a Pearson correlation of only 42%(+-1%) which is again what one would expect if the conjunction of two variables were required for the etiology of autism.
See this link.
(Oregon and Massachusetts are excluded as data points due to their being the States with the highest and lowest autism percapita rates respectively. Failing to exclude these datapoints creates the impression that the best correlation is with nonWestern immigration to industrial regions, rather than immigration from India per se to regions of Finnish ancestry.)
Adding economic data there was only one combination of variables that exceeded this and it did so by just 1% (r=61%). It is weakly supportive of the "refrigerator mother" hypothesis. It is not strongly supportive due to the fact that while working parents percapita was one of the 2 variables, the other variable was public education expenditure per student which had, by itself, a Pearson correlation of 54% whereas working parents percapita was only 25% -- indicating the vast majority of the variance in autism rates was explained by public education expenditure per student rather than working parents. There are a number of possible explanations for why public education expenditure per student would be correlated with autism percapita, among them the most obvious being simply that a high cost of education is associated with autism spectrum disorders.
See this link.
(For those skeptical that this hypothesis was derived before the fact -- ie: those who suspect this is a post hoc hypothesis resulting from data dredging may wish to examine a control run of the correlation program which used precisely the same mechanisms to rank order correlations except substituting random data for the variables other than autism rates. See this link. Note the bivariate correlations topped out at 54% rendering higher correlations, such as that reported here, unlikely to be due to chance.)
MMR vaccination rates show virtually zero correspondence with autism rates. When viewed in combinations with other demographic variables, it came in combinations far from the top -- far enough from the top that it is plausible that such correlations are due to chance or due solely to the other variable.
Mercury has also been hypothesized as a factor in autism, however data from the Environmental Protection Agency on percapita water-way mercury pollution by State fails to show a significant correlation with autism.
CONCLUSION
Many epidemiological studies start with gross ecological correlations like this so that hypotheses can be tested at a low cost before proceeding to more costly and more detailed studies. Here the next logical step is a by-County ecological correlation to see if the correlation holds up under the more stringent test. If so, it may be appropriate to take remedial public health action at that time but not conclude that the hypothesis put forth here is confirmed. Confirmation would require individual case studies. Nevertheless, with just this low cost study and low degree of ecological resolution it is justifiable for some individuals to take low-cost precautions. For instance, it might be advisable for couples who suspect they have Finnish ancestry to avoid Indian restaurants if they are attempting to conceive or have conceived. This is not a great sacrifice for people given the degree of ecological correlation already shown at the State level.
The important thing is to proceed with the by-County correlation to further confirm the hypothesis. However, the political ramifications of this hypothesis are preventing it from being studied despite the fact that there exists no other hypothesis that enjoys nearly the degree of support of demographic data. It is important to understand that the potential of immigration to contribute to disease in host populations is not limited to historically well-known examples such as Amerindians suffering during the influx of immigrants from Europe, nor the more recent and obvious examples such as Severe Acute Respiratory Syndrome. If it is the case that the odious human costs of autism that have exploded in recent years involves irresponsible imposition of immigration on susceptible populations then it is incumbent on public officials to inquire as to who is responsible for indemnifying the populations affected -- and this is true in advance of any identification of the precise etiology. Immigration and associated policies cannot be imposed with the historic claims of ignorance that accompanied immigration waves prior to epidemiologic understanding. Insurance of the potentially susceptible populations is important and impossible to avert responsibility for. Indemnification of the impacted populations in the absence of prior policy recognizing this risk is equally important and incumbent on all of ethical and moral character.