Friday, March 24, 2006

Latest on Geier Retraction

I was getting a bit impatient, so I asked Paul Choate of CDDS Data Extraction what the status was on the new data requested by Dr. Geier. I also mentioned to him that there's already a report that analyses new clients vs. drop-outs, but it only goes to 2004. Paul says this is a report by the local state college.

The basic part of the job is done, pending an administrative matter, he tells me. He goes on to explain that cases entering the system are leveling off, which is not surprising.


There is a recent flattening of autism cases, but it’s not necessarily supporting evidence for the downward thimerosal-based trend for at least five reasons.

1. There was a large artificial population spike in 9/2002 due to a system fix, and all caseload records counts grew proportionally, not just autism. Geier did not adjust for that quarter’s change data.

2. There was a change in state law in 8/2003 (CA AB1762, W&IC 4512) where the requirements for DD services was increased from one substantial life-functioning deficit to three substantial deficits. This law was specifically crafted to depress autism caseload growth, and it has had a flattening effect on the higher functioning caseload’s growth, which are Autism, CP, and Epilepsy. Geier did not adjust for that effect.

3. The autism population served by DDS is self-selecting, and has not been shown a constant and direct proportion of the true incidence in the general population.

4. The growth in Autism caseload has been geographically isolated.

5. The diagnostic characteristics of the population have changed dramatically, less MR and younger age among other things.


I wasn't aware of the 2002 spike, but the drop in 2003 coincides with changes in the Lanterman Act, as explained by Interverbal. There's an increase in newly reported clients (true new cases) in 2004. In 2005 Paul appears to be saying there's a "flattening". I interpret this as the population growth leveling off. Basically, what the law changes achieved was a short-lived drop in the number of cases entering the system, but it also resulted in lower caseload growth, perhaps leading to a leveling off of the autistic population served by CDDS in upcoming years. There's no sign of a drop apparently.

In any case, the new data will show that Dr. Geier's paper is unsalvageable, containing large shifting errors. His conclusions are invalid due to the methodological flaws mentioned, such as the cofounding factors due to changes in the law. However, the fact that what he claims are "New Cases" actually are not by itself renders the paper useless.

5 comments:

  1. Oh, and Paul also says the following:

    Our Autism expert, Ron Huff PhD at Alta Regional Center, told me the MIND institute is forming a committee to respond to Geier’s and similar claims. The people at the CHARGE study at UC Davis are also looking at his claims.

    ReplyDelete
  2. 1. There was a large artificial population spike in 9/2002 due to a system fix, and all caseload records counts grew proportionally, not just autism. Geier did not adjust for that quarter’s change data.

    That reminds me of the famous New York City "crime wave" of 1951. Going from 1950 to 1951, there were 200%-2000% increases in reported burglaries, aggravated assaults, robberies, and similar crimes. But the city hadn't become more dangerous; all that had happened was that a new police commissioner had cracked down on sloppy record-keeping.

    The Geiers really should have attempted to control for total caseload, or total non-MR caseload.

    ReplyDelete
  3. The Geiers really should have attempted to control for total caseload, or total non-MR caseload.

    They should've attempted to control for something. They controlled for nothing.

    The non-MR caseload is misleading too. People assume that non-MR caseload should remain constant with 'broadening criteria'. This is not correct because we can assume that more and more in the population with MR are being recognized as also being auistic.

    Another way to think about it is that the proportion of MR (using the standard evaluation method) in additional autistics is closer to that of the general population, but not the same.

    ReplyDelete
  4. Make that MR caseload above.

    ReplyDelete
  5. Thanks for staying on the case Joseph. You have my respect and gratitude, fwiw.

    ReplyDelete