In this posting I will analyze the data in a different way. I will show that while autism rates vary widely from region to region, and are typically well correlated with degree of urbanization, the same cannot be said of mental retardation.
As usual, the analysis will be as non-technical as possible. The reader is encouraged to download the data, verify my claims, and inform me of any errors in the comments section of this post.
The importance of this analysis rests in testing the hypothesis that neurodevelopmental disorders may be correlated to factors such as environmental pollution, including exposure to environmental lead and mercury.
To carry out the analysis, I will use epilepsy as a baseline. This is because, as I've noted, diagnoses of epilepsy appear to be stable in the CDDS data. Their growth pattern closely matches population growth in the state of California. This suggests that diagnoses of epilepsy are more objective and thus not subject to error such as 'broadening criteria' or 'shifting misdiagnoses'. So a couple basic assumptions of this analysis are the following: (1) Epilepsy is not undergoing an epidemic, per CDDS data; (2) Epilepsy is uniformly diagnosed across regions, and differences in awareness in relation to epilepsy are negligible.
Allow me to define a couple of terms before proceeding:
Autism Index.- Number of autistic clients divided by number of clients with epilepsy.
MR Index.- Number of clients with severe or profound mental retardation divided by number of clients with epilepsy.
I will use these two indexes as an indication of the prevalence of autism and mental retardation in a Regional Center. (It would be much more difficult to gather data on the population served by each Regional Center, and that's why I use this method).
Locations & Map
The CDDS has a Directory of Regional Centers which gives the address of each Regional Center. Additionally, they provide a map of the areas served by each regional center.
I have compiled data on Autism Index and MR Index for all regional centers. This data is calculated from numbers published in the CDDS Quarterly Client Characteristics Report for December, 2005.
|Regional Center||Autism Index||MR Index|
|Frank D. Lanterman||1.37||0.83|
|South Central LA||0.81||0.80|
You will notice that the range of variance of MR Index is 0.45-0.85 (a factor of 1.89), whereas the range of variance of Autism Index is 0.32-1.61 (a factor of 5.03).
The difference in autism rates between Central Valley and Westside is remarkable. To put it in perspective, this gap is equivalent to the 'autism epidemic' in U.S. schools between 1995 and 2003. Interestingly, the MR Indexes in Westide and Central Valley are very similar.
Is there a correlation between autism and MR?
Some areas with a low Autism Index also have a low MR Index. This is the case of Alta California, Redwood Coast and Far Northern. But the pattern breaks if you look at Golden Gate, Inland, Kern and North Bay, all centers with a low Autism Index but moderate to high MR Index.
Some centers with a high Autism Index also have a relatively high MR Index. This is the case of East LA, Frank D. Lanterman and Harbor. But again, the pattern breaks if you look at North LA or Westside, which have the two highest Autism Indexes, but appear to have below-average MR Indexes.
In general, it's hard to discern a pattern of correlation between Autism Index and MR Index. I'll leave it for the reader to graph the table and look for a pattern.
Shouldn't we expect to see an inverse correlation? To some extent, yes. In the time-based state-wide data, you will see an inverse correlation, but it varies little (about -0.2% MR Index variation for every 1% increase in number of autistic clients). So this shouldn't throw off the data too much. You do see some of it in the two Regional Centers with the highest Autism Indexes: Westside and North LA. But there appear to be various random and conflicting factors pulling the numbers in different directions, such as socio-economics and cultural differences (e.g. how severe a condition must be to use the services of a Regional Center for something such as autism or MR, compared to something more objectively diagnosed such as epilepsy).
Is there a correlation between MR and degree of urbanization?
In the MR Index range of 0.4-0.59, we find Alta California (Sacramento area), Redwood Coast, Tri-Counties (Santa Barbara) and Far Northern. One is a densely populated area, one is a moderately populated area, and two are sparsely populated. This can easily be explained as a difference in awareness/culture/resources. If an urbanization-dependent environmental factor is at play, one would have to wonder why it's not in effect in Sacramento.
In the MR Index range of 0.6-0.69 range we find Central Valley, East Bay, North LA, San Andreas, Valley Mountain and West Side. Central Valley (Fresno) and San Andreas (Santa Cruz) are moderately populated areas. The same is likely true of Valley Mountain. East Bay (Silicon Valley) is heavily populated as are Westside (Culver City in LA) and North LA.
In the MR Index range of 0.7-0.85 and above, we find most of the Regional Centers. They tend to be heavily populated, but we find a couple of anomalies. For example, Inland appears to be very sparsely populated. Kern would also appear to be much less densely populated that the nearby LA area, where many locations have a lower MR Index.
In general, there's no clear-cut correlation between degree of urbanization and MR Index. Differences are small and may be easily explained by various random regional factors. Most of the Regional Centers have an MR Index in the range of 0.7-0.85, a very small range of variance.
[Note: For an updated regional analysis, see Regional Differences and Quarterly Growth Due to Two Factors.]
If we compare Autism Index in Westside (LA area) vs. Central Valley (Fresno), we see that there's a difference in prevalence of about 500%. But we can also conclude from the data that autistics in Westside have a much lower prevalence of epilepsy and MR than do those in Central Valley. In other words, the groups are not equivalent. If we assume that MR and epilepsy are good equivalence baselines, given that MR Indexes are the same in both areas, I'd like to point out, alarmingly, that Central Valley (Fresno) has "missed" about 80% of its "train wrecks" compared to Westside (using the wording of a known foe of autistics everywhere).
What causes the groups not to be equivalent? The main cause, evidently, is that autism is a spectrum and there is no objective medical test for autism. Other possibilities: Awareness, misdiagnoses, financial resources, psychiatrist and pediatrician culture, etc.
At this point it should be clear to the reader that autism diagnostic frequency is region-dependent, but that the diagnostic threshold is not equivalent across regions. This in itself does not prove that an environmental factor is not involved at all. MR is fairly stable in relation to epilepsy, but this could be explained by pointing out that an environmental factor might increase the prevalence of all 3 conditions (epilepsy, MR and autism), and act in conjunction with autism criteria discrepancies.
I propose we can discount this hypothesis, however, based on the stability of diagnoses of epilepsy in the state-wide time-based data. Furthermore, one Italian study has shown that status epilepticus is not correlated to degree of urbanization [ref].
Showing that MR varies little from region to region and is not correlated in any significant way with degree of urbanization completely undermines the observation that autism is. That is, if actual autism incidence depended on degree of urbanization, you would expect MR and epilepsy to have a clear dependency on degree of urbanization as well.
In other words, once we take into account the fact that it is the definition of autism which changes from region to region, and not the actual incidence of equally severe cases, then the correlation to degree of urbanization is better explained (or only explained) by other types of factors: Better knowledge, more financial resources, etc.
While it is possible to pretty much discount any urbanization-dependent environmental factors with this analysis, we have not ruled out random region-dependent factors that may affect prevalence or incidence, such as socio-economics, ethnicity and genetic homogeneity. They would appear to have little effect, however. Factors that are not region-dependent are not within the scope of this analysis but they have been addressed before.