Curse you, Mark and David Geier.
I’m getting tired of having to subject my scientific and critical thinking skills to the assaults on science and reason that you routinely publish in dubious journals to use as weapons in your apparently never-ending crusade to extract as much money as possible out of vaccine manufacturers and the Vaccine Injury Compensation Program. Dissecting your pseudoscientific claims causes me pain, not so much that I’m driven to take a hiatus from blogging, as Matt was by Kent Hovind’s creationism, but almost.
I had hoped to let this cup pass, given how much I’ve written on the topic lately, but, thanks to mentions by Clark, RW, and Tara, I find myself feeling obligated to comment.
Yes, Mark and David Geier, that father-son team of dumpster-diving pseudoscientists and promoters of chelation and chemical castration as treatments for autism dedicated to “proving” that mercury in vaccines causes autism, no matter how much they need to torture the data to do it, are at it again.
This time around, they’ve published an allegedly scientific paper entitled Early Downward Trends in Neurodevelopmental Disorders Following Removal of Thimerosal- Containing Vaccines, published (where else?) in that repository of dubious medical science, The Journal of American Physicians and Surgeons. This paper purports to show that autism rates have started to dip since early 2003, when the last lots of thimerosal-containing childhood vaccines expired, the point after which, with the exception of some flu vaccines, for all intents and purposes no child in the U.S. has received thimerosal-containing vaccines. If true, such an observation would represent compelling evidence for the thimerosal-autism hypothesis. Not surprisingly, loony sources like WorldNut Daily and Vox Day (to whom I administered a blog slapdown a few months ago when he uncritically parroted Dan Olmsted’s credulous reporting on a poorly documented population that is claimed not to have been vaccinated and to have very low rates of autism) have trumpeted the Geiers’ study as “proof” that mercury in vaccines causes autism).
It’s no such thing.
What it is, in actuality, is yet another example of the Geiers mining the Vaccine Adverse Events Reporting System (VAERS) database for purposes for which it was never designed and misusing the California Department of Developmental Services (CDDS) database similarly. First, off, as I pointed out extensively before, the VAERS database is designed only as an early warning system for reporting adverse events thought to be due to vaccines. It is not designed to track the incidence or prevalence of vaccine complications. One reason is that anyone can make entries into it, not just medical professionals, and the results are only checked in the most perfunctory way. For example, as Jim Laidler described before, it takes entering something like a claim that a vaccine turned one into The Incredible Hulk in order to get the staff there to question the entry. Worse, as I described before, the database has been corrupted by litigation, with a dramatic increase of entries linked to litigation claiming that thimerosal caused the plaintiff’s child’s autism. As for appropriate uses of the VAERS database, here’s what it says right on the VAERS information page:
VAERS data are derived from a passive surveillance system and represent unverified [emphasis mine] reports of health events, both minor and serious, that occur after vaccination.
And:
While some events reported to VAERS are truly caused by vaccines, others may be related to an underlying disease or condition, to drugs being taken concurrently, or may occur by chance shortly after a vaccine was administered.
And:
Therefore, VAERS collects data on any adverse event following vaccination, be it coincidental or truly caused by a vaccine. The report of an adverse event to VAERS is not documentation that a vaccine caused the event.
The bottom line is that the VAERS database is not a reliable source to determine autism prevalence or incidence, period. It’s too prone to being affected by scares, like the recent hype over mercury supposedly causing autism, and even outright manipulation, and its entries aren’t even examined by medical professionals in sufficient detail to determine whether adverse events reported were or could be reasonably inferred to have resulted from vaccines. Of course, that never stopped the Geiers from diving into that database again and again and again, all without acknowledging the shortcomings of this database. And it actually wouldn’t be all too surprising if reports of autism attributed to vaccines in the VAERS database did indeed fall after 2003, given that it was widely publicized that thimerosal was removed from nearly all childhood vaccines around that time, making it less likely after then that parents would attribute their children’s autism to vaccines.
But what about the CDDS? It could be a source of determining whether autism rates are indeed rising, at least in California. There’s just one problem. To do that properly, one has to look at new diagnoses. However, the CDDS database reports total cases (both new and old) every month. As Citizen Cain pointed out while criticizing David Kirby’s misinterpretation of the California numbers last year:
Rather, the California Department of Developmental Services reports on the total number of people with autism registered to receive services. Kirby derives an estimate of children who enter the system in 2004, for instance, by subtracting the total number of cases within the system at end of 2003 from the total number of cases in the system at the end of 2004. Kirby calls the change in total cases “new autism diagnoses,” which it is not.
However, as has been pointed out by Interverbal and Citizen Cain, the CDDS does record some other data that might be useful, such as age. Thus, although the purpose of this database is not to measure the incidence of developmental conditions like autism and mental retardation, in the case of autism one could get perhaps get a rough idea of the incidence of autism by looking at the cohort from 3-5 years of age over time, as this is the age range at which autism is most commonly first diagnosed. Of course, a number of potential sources of bias and error need to be kept in mind, as Interverbal described in detail, but one might get some useful information if one did the proper analysis and qualified it with the limitations of the database, a rough estimate of autism rates over time. Too bad that is not what the Geiers did.
So what did the Geiers do? I quote their paper:
The total new number of autism reports received by the CDDS from 36 consecutive reporting quarters (from that starting on January 24, 1994, through that ending on January 6, 2003), and for 15 consecutive reporting quarters (from that starting on January 3, 2002, through that ending on October 4, 2005) were analyzed.
Bzzzt! Wrong answer. (In fact, if you look at the graph on Figure 3, the Geiers even make the mistake of labeling what they are plotting as new cases of autism.) In fact, when correctly analyzed, the California data show new cases of autism in the 3-5 year old cohort continuing to rise.
Even if the data from these databases were more useful for determining the incidence of autism, the way the Geiers did their analysis is utterly laughable. Here’s what they did. (Any statisticians, please chime in here.) Basically, they plotted the data as cases against time period by quarter and then used a statistics package to do linear regression of the data from 1994 to the end of 2002 and then to do it again from the beginning of 2002 until October 2005. They then compared the slopes of the two lines derived from linear regression for each graph and concluded that the slope of the line after 2002 had changed from positive (increasing) to negative (decreasing).
Geez, couldn’t these guys at least have bothered to hire a biostatistician? (Certainly no statistician is listed in the authors, just Mark and David Geier.) If they had, they would have found out that doing interrupted time series analysis is fraught with difficulty, and that the appropriate way to look at such effects over time is not by simplistically comparing slopes before and after a time point, but rather by using one of these techniques, depending upon the specific question and data. Also, choosing the end of 2002 is a bit of a stretch; after all, autism is most commonly diagnosed between ages 3-5, which means that around now is the period in which we should start to see a dramatic drop in autism rates if mercury in vaccines is truly the cause of autism. But it’s still even worse than that, so bad that it’s funny. Basically, what the Geiers have done is to take contiguous data and make up trendlines to “prove a point,” even if the data don’t support it, and it’s hard not to note that the r-squared value of the latter data is very much lower than the r-squared for the earlier data, meaning a much poorer fit to the linear regression. Indeed, Bartholemew Cubbins has made a useful video to show you why this is an invalid statistical technique, and Autism Diva has described it. In addition, they use the Kruskal-Wallis statistic to compare trend lines, but the K-W statistic is generally not used for that purpose. Rather, it is a nonparametric (to be used for non-Gaussian distributions) to compare the medians of three or more unpaired groups, not two slopes, one before and one after a time point. I may not be a statistician, but even I know that. This is freshman-level statistics.
Finally, let’s take a look at the journal in which this “study” was published, namely The Journal of American Physicians and Surgeons (JAPS). Those of you out there who are physicians or scientists, I encourage you to browse its back issues on its website to see what kind of articles this journal publishes. For example, in the very same issue as the Geiers’ paper, there are three articles casting doubt on the concept of the “shaken baby syndrome.” (Blaming “shaken baby syndrome” on a vaccine reaction is how baby killer Alan Yurko got off, and he was aided and abetted by Dr. Mohammed Al-Bayati, he of the HIV/AIDS denialist fame.) Indeed, there seem to be many such articles scattered among the issues available on the JAPS website. There’s also a “case study” of treating Herpes Zoster with vitamin C. In other issues, JAPS has published articles supporting a link between abortion and breast cancer; an article on the “medical dangers of homosexuality” (which also claimed that homosexuals who tried to change to heterosexuality “often succeed in doing so”); and an antifluoridation article; an anti-illegal immigrant screed referring to their “stealthy assault on medicine,” blaming them for bringing disease to the U.S., and demanding that we close our borders; and an article decrying evidence-based medicine.
The journal is a rag. Given the low scientific quality of the articles it publishes, it’s hard for me to conclude anything other than that its claim of “peer review” is a sham. Indeed, I recently became aware that JAPS is, as Terry Kopel described it:
The Journal of American Physicians and Surgeons seems to be little more than a conservative publication gussied up with a medical spin. A look at the references in the illegal-alien report, written by Madeleine Pelner Cosman–a “medical lawyer” whose previous claim to fame appears to be a book on medieval cooking but who has also written an article for a group called Jews For The Preservation of Firearms Ownership–is chock full of hardline conservative cites, including books by Michelle Malkin and former WND writer (and Slantie winner) Jon Dougherty and articles by Phyllis Schlafly and Tom DeWeese.
Indeed, JAPS is so dubious that it isn’t even indexed in the Medline database. Even the speculative and often dubious vanity journal Medical Hypotheses is so indexed. (How it managed that, I still can’t figure out.) Just from perusing its back issues, one should be able to see that the group that publishes JAPS, the Association of American Physicians and Surgeons has an agenda, and that that agenda is anti-homosexual, anti-abortion, and against mandatory vaccination.
No wonder Vox Day is so quick to cite a study from this journal so uncritically.
The bottom line is that this study by the Geiers does not show what it claims to show because (1) it uses reporting databases for purposes for which they were not designed and for which the agencies maintaining them don’t recommend; (2) one of these databases consists mainly of unverified reports of adverse events or diseases that the parent or doctor attribute to vaccines, whether correctly or incorrectly; (3) the authors use such mind-bogglingly simplistic statistical techniques to analyze the data that no conclusion can be made, certainly not the one they claim to make; (4) the investigators, Mark and David Geier, have a track record of abusing science and medicine, even to the point of proposing chemical castration as a treatment for autism in order to “increase the efficacy of chelation therapy”; and (5) the journal in which the findings are published is run by an organization that is expressly antivaccination. Could mercury from thimerosal in vaccines be a major cause of autism? I suppose it’s possible, but certainly this study doesn’t support such a link. Nor do any of the studies by the Geiers, most of which are published in JAPS and other not-so-reputable journals.
Given the fast pace of developments on mercury/autism front, as I was writing this I had to wonder what was happening to produce so much bullshit at such a rapid rate over the last couple of weeks. Then I noticed something. David Kirby’s Evidence of Harm, the book published about a year ago that fanned the flames of this trumped up controversy, ultimately leading to Robert F. Kennedy, Jr.’s mendacious Salon.com article in June, has just come out in paperback. Could all this new activity by the Geiers and RFK, Jr. have anything to do with this?
Nahhh. Couldn’t be.
Hopefully, I won’t have to blog on this topic again for a while. I feel dirty after having had to dive into the bowels of the Geiers’ bad science and quack treatments for autism and RFK, Jr.’s distortions and logical fallacies so many times in a mere week and a half. Given so much bad science, I fear the impact on my science and critical thinking skills if I have to read any more of the Geiers’ papers without at least a rest to let my brain recover. (Even worse, writing this rebuttal distracted me from the Oscars.)
Time for a shower.
Some previous posts on this topic:
NOTE: Due to a serious spam attack on this article, trackbacks and comments have been turned off.
