The Fallacy of Thimerosal Removal & Autism Increase: A Failure of Science, A Bigger Failure to Children Worldwide - AGE OF AUTISM
By Jake Crosby
Readers of Age of Autism are aware of the term “the big hungry lie” coined by regular contributor J.B. Handley, used to describe the tactics of the CDC and the drug industry’s attempts to disassociate autism from vaccines in any way, shape, or form.
Perhaps the biggest lie of all is the one that has been repeated all too long, that after thimerosal was reduced or eliminated from vaccines, autism rates continued to go up. There have been multiple instances of this claim and each time it has been proven false, right up to the recent lie that after thimerosal was removed from vaccines in 2001, autism rates continue to increase. These two claims, the first that thimerosal was removed from vaccines, and the second that autism rates have not gone down as a result, continue to be used to justify the injection of thimerosal into pregnant women and children with flu shots. The claims have also been used to justify the immunization of children in developing countries with vaccines preserved with thimerosal. Sadly, neither claim is any more truthful than previous equally erroneous claims, the earliest of which originated from Scandinavia, then spread to Canada and most recently came out of California.
The paper by Stehr-Green et al., for example, purported to study autism rates in Sweden after thimerosal removal in 1993, but only hospitalizations in relation to autism were analyzed. Anyone remotely familiar with autism knows that it is not the kind of condition for which one would typically go to a hospital for treatment.
In the same study were also analyses of autism rates in Denmark, which were even more flawed. Many here remember the infamous Danish studies published in 2003, which served as the primary basis for the IOM’s predetermined conclusion in 2004, that autism rates shot up after thimerosal removal in 1992. In reality what happened was the Danish were worried there was a connection between thimerosal and autism, and right after thimerosal was eliminated from all their vaccines, they rapidly changed their registration program to include a lot more children. Such an interpretation of these studies -- designed by the CDC, and conducted by Statens Serum Institut, the largest vaccine-manufacturer in Denmark -- that autism rates skyrocketed after thimerosal removal, can be regarded as little more than propaganda.
When SafeMinds reanalyzed the data of the latest Denmark study, Hviid et al., by applying the same standards of higher case ascertainment to children born before 1992, they found a prevalence of 1 in 500, compared to a prevalence of 1 in 1,500 ten years later, a 66% drop. Unfortunately, this would not be the last time the CDC would design such self-contradicting studies.
In Canada, Eric Fombonne, a psychiatrist with ties to Sanofi-Pasteur, who is not even an epidemiologist, conducted his own combined thimerosal-MMR study on a school district in Montreal, and it was a total failure. His claim that autism rates went up after thimerosal was removed rested entirely on the Kindergarten cohort, for which enrollment was optional, so only about half the kids out of the total enrolled. However, all the children with autism enrolled because the school provides many services to autistic children. In fact, the school district Fombonne studied has an autism center for excellence, and even draws children with autism from other districts. Had enrollment been mandatory, the estimated prevalence would have dropped by one-half, indicating a decrease rather than an increase. According to biochemist Dr. Paul G. King, this is what is called “negative enrollment bias.” Furthermore, there was also thimerosal exposure during the years where exposure was labeled “nil.”
His MMR data were no more reliable. Instead of using local MMR immunization rates to compare to autism rates, he used immunization data from Quebec City, 145 miles away. Even though the Cochrane Collaboration had this to say about his previous MMR study from 2001, "The number and possible impact of biases was so high that interpretation of the results was difficult," and even though the collaboration included a person who also acted as a legal consultant to MMR manufacturers, such discrediting apparently has not stopped Fombonne from doing more completely flawed, post-marketing research.
But back to thimerosal, because just two years ago, Robert Schechter and Judith Grether of the California Department of Public Health accessed the records of the California Database for Evaluation and Research (CDER) of the California Department of Developmental Services in children ages 3-5. The purpose was to see if autism rates had declined after the supposed removal of thimerosal from vaccines. According to Schecter and Grether's analysis, they hadn’t. Using their interpretation, the two researchers determined that thimerosal must not be a primary cause of autism, in a study published in the Archives of General Psychiatry entitled “Continuing Increases in Autism Reported to California’s Developmental Services System: Mercury in Retrograde.” However, their own errors, it now appears, contradicted their conclusions.
3 Year Olds, A Reliable Age For The Final Cohort?
The very last cohort the study looks at are 3 years olds, which already is not a sufficient age group to base any conclusions from, as they would be hopelessly premature. This is especially relevant to point out because Schechter and Grether suggest that the first full thimerosal-free year was 2003, yet this was the final birth year fully studied, and children born during this year would have been diagnosed the earliest for reasons that will be explained later. Moreover, the claim that children received no more exposure to thimerosal after 2002 is not true either. Expiration dates on many of the vaccines that contained thimerosal were well after 2003. Furthermore, unpublished statistics show that there were sharp increases in thimerosal exposure from flu shots given to infants and pregnant women, while other sources of mercury exposure further confound the study’s conclusions.
Even if the premise for this study had been correct, that thimerosal was removed in 2002, it is still inherently flawed by the fact that its conclusions that thimerosal and autism are unrelated are based on one birth year of very young children (three year olds born in 2003), which is only the beginning of when autism cases start getting filed into the DDS CDER archives. Drawing any conclusions from this age group alone would be a false hope.
Perhaps recognizing this problem, the authors then proceed to combine cases designated as three years of age with those ages four and five, but this only adds to the problem, creating a simple ascertainment bias that would make the designated 3 year old age group, by virtue of having less time to enter the system, smaller than the group labeled as four years old, which would in turn be smaller than the age group labeled as 5 years of age. This is especially important because older kids would have been more likely to receive greater quantities of thimerosal than younger kids. Yet they are all combined into one whole age group from which to draw conclusions based on prevalence.
Mercury in Retrograde?
This relates primarily to the less quantifiable problems in relation to autism rates, which were issues with elemental mercury exposure from amalgam fillings, methylmercury exposure from coal-burning facilities, the remaining childhood vaccines preserved with thimerosal that were not taken off the shelves, and for that matter thimerosal exposure from the flu shot while exposure from other vaccines was being reduced.
Complicating this further is the fact that there seems to be no consensus on when the first year routinely recommended vaccines truly contained no thimerosal preservative. Schechter and Grether said the preservative was all gone from vaccines in the middle of 2002, citing the IOM Report. However, all the IOM said was that the ACIP gave an “expressed preference” that all thimerosal be removed by 2002, hardly reflective of actual thimerosal content in vaccines. The FDA said the last lots preserved in thimerosal expired at the beginning of 2003, but the Council of State Governments said it was early 2004, while parents have found vaccines on the shelves of doctor’s offices with expiration dates that surpass all these years. So no one really knows.
On top of all this, the parallel increase in uptake of flu shots, like those during pregnancy, may contribute to earlier onset autism, due to earlier exposure, and therefore contribute to the disorder being diagnosed earlier. Unfortunately, there is no available immunization data for prenatal flu shots.
What is available, however, is the immunization data for the rate of postnatal flu shots among children ages 6-23 months of age, as reflected among clients enrolled in the Northern California Kaiser HMO, which jumped from 5% during the 2001-2002 influenza season, to 45%, in the following season. By winter of 2004-2005, 57% of 6-23 month olds were getting flu shots, practically all of which were preserved with thimerosal. Data for pregnant women are not available, but they were also a target group for flu vaccines in the same period.
That is the final and main problem I found with this paper, which has been used to support the untrue claim that autism rates have continued to go up. First, the claim that prevalence of autism was counted in 3-5 year old children is misleading. Schecter and Grether’s estimates of age rely on subtracting the year of birth of the child from the year the child is currently enrolled as an active client, but that does not mean the child really is that age For example, I was born in 1988, so by Schechter and Grether’s counts I would be 22, but I’m actually 21. So many children are getting counted as older than they really were. Many children labeled as four years old are actually three and many children labeled as five years old are actually four, and roughly half the five year olds are not included but actually classified in with children six or older. Children in the studied age group could be as old as five, but that’s not the same as including the entire five year old age group.
So the Schecter and Grether study only fully accounts for 3 and 4 year olds; the only other study I can recall which looks at children that young is the infamous Verstraeten study. This is key, especially since during the years it looked at the rates after thimerosal "removal" (which is also dubious), the California Department of Developmental Services’ regional centers made changes that may artificially skew autism clients, especially those in the youngest age groups, towards an upward trend.
According to a CDDS report “Controlling Regional Center Costs:”
“Responding to this concern (increasing autism rates), the Legislature enacted a requirement for the Department to develop evaluation and diagnostic procedures for the diagnosis of ASD and to develop a training program for regional center clinical staff in the utilization of the diagnostic procedures. These procedures were published in 2002.”
So, now the youngest possible autism age groups in the CDER archives of the CDDS, 3 and 4 year olds, the only ones Schechter and Grether fully account for, are heavily biased. Not only would developing procedures for diagnosis skew autism figures towards the youngest children, but also since they are done at reporting centers, children may now enter the DDS system as soon as they are diagnosed. This will mean the proportion of younger children to older children with autism will shoot way up, and that is exactly what we see in the Jaunuary 2009 Hertz-Picciotto study published in Epidemiology, where starting in 2002, new cases of autism in the CDDS of children ages 0-4 go up linearly but the rate for children ages 5-9 starts to flatline. It also means that, amidst all this, the increasing proportion of total numbers of new cases, assuming the autism rate were to continue to remain the same, must also balance out with the total number of cases leaving the system as well as the decreasing proportion of older children entering the system. So changes in the youngest clients would still be reflected indirectly in total new autism cases, as they are all part of the same system. As a result, looking solely at autism rates in the youngest children does not give the full picture.
What does, however, are the results obtained by Dr. Mark Geier, a fellow of the American College of Epidemiology, and his son David Geier, when they analyzed both the Vaccine Adverse Event Reporting System of autism-related adverse events and the California Department of Developmental Services data of total new autism cases and found a decrease in both, according to a study entitled “Early Downward Trends in Neurodevelopmental Disorders Following Removal of Thimerosal-Containing Vaccines,” published in the Journal of American Physicians and Surgeons.
Not only that, the CDDS graph of autism (page 12) from this study provides further evidence for early diagnosis bias in the Schechter and Grether paper. An increase in the proportion of younger children who enter the system, namely three, four and some five year olds who are the youngest clients enrolled in the CDDS CDER database, while it may positively affect the reporting system overall, would still mean that there would have to be a corresponding decrease of older children with autism being enrolled to the system; that would be offset by an acceleration of younger cases if rates were to remain stagnant, and as a result cause a stagnation in the number of new cases. New autism cases would also have to offset the increasing numbers of older cases reported from previous quarters leaving the system, if they do, then the autism rate has not gone down.
However, this is not the case based on the fact that the increase in prevalence among 3-5 year olds from 1995 to 2007 in Schechter and Grether is a straight line maintained at a constant rate, not an accelerating curve as would be expected with younger children entering the system, likely due to thimerosal removal. Further evidence for this can be seen from Figure 3 of Geier et al., showing the scatter plot of total new autism cases entered to the CDDS, which shows a downwards trend beginning in early 2002.
To be fair, however, the Geiers' study drew some criticism. Critics claimed the decrease was the result of state law that went into effect in mid 2003. However, the change in trend noted by the Geiers began in early 2002.
The role of changing methods in California
Those who use the California DDS data to exonerate thimerosal's role in autism will point to changes in the state law in 2003 that raised demands of those being served by CDDS, saying that clients must show “substantial disability” in three or more areas of life because the state was facing a budget crisis, and that therefore, the California database was if anything, taking fewer cases than it would have.
This, however, is unlikely to have made an impact on cases of full-blown autism admitted to the system, the only autism spectrum disorder for which this new law applied. Rick Rollins, autism parent and cofounder of the UC Davis MIND Institute, says “children with full syndrome autism generally fail in at least 3 and as many as 6 of the areas of 'major life activities' as defined above, therefore one would expect that autism would be the least impacted of all the categories by the new, additional requirements for eligibility.”
Furthermore, if one were to believe these changes have a significant impact on the reporting of full-blown autism at all, one would expect a disproportionately lower growth in full-blown autism as compared with Aspergers and PDD-NOS, since the “substantial disability” criteria only applies to classic autism between the years of 2002 and 2007. However, growth for autism compared with other ASDs for which “substantial disability” criteria does not apply increased at approximately the same rate during this period. (page 27): See HERE.
Then there are the changes that took place in the CDDS in 2002 that culminated in the emergence of early diagnosis and evaluation policies for ASDs that had not previously existed, that would positively skew the numbers of autism spectrum disorders enrolled in the database, especially in the youngest children. This is far more likely to bias the database in a positive direction than requiring extra proof of “substantial disability” in cases with classic autism, which is already a substantial disability.
While I have previously speculated that increased exposure to thimerosal from flu shots plays a role, the results are likely to be primarily due to drastic administrative changes as stated before in reference to the 2007 Report “Controlling Regional Center Costs.” The budget crisis that had been going on during this time period, if anything, caused the Department to divert more funds to early diagnosis and intervention programs for autism, given that autism has increased at a much higher rate than the other disabilities the system keeps track of. This change to the CDDS database came as a result of a state law passed the year before in 2001, the most widely cited year for alleged thimerosal removal.
Even slight fluctuations in the average age of diagnosis alone can have a major impact on autism rates in young ages. In Denmark, for example, in a study published in the Archives of Pediatrics and Adolescent Medicine entitled “Autism Prevalence Trends Over Time in Denmark: Changes in Prevalence and Age at Diagnosis,” a drop in the average age of autism diagnosis from 5.1 to 4.7 was attributable to a 37% autism increase in 3 year olds while the drop in the average age of ASD diagnosis from 5.9 to 5.3 as attributable to a growth of 66% in 3 year olds. Even modest shifts in the average age of diagnosis can have a huge impact on autism in the youngest age groups. So the change in age of diagnosis definitely would have impacted the studied age groups of 3, 4 and some 5 year olds. (19047542[PMID]) See HERE.
However, it should be noted that one of the study’s authors, Poul Thorsen, had a hand in one of the previous studies from Denmark “exempting” thimerosal, and another equally flawed study attempting to do the same with the MMR vaccine. He is currently under criminal investigation, facing possible charges of fraud and forgery.
Erasing the trail on the California autism data
Just as egregious was when the CDDS changed its reporting mechanisms in 2008 in a big way, which would include many more cases, one week before the Schecter and Grether study was published. This meant that the database from there on out would be unusable to track the autism rates to determine if there would be a decline any time soon. This had a profound impact on the monitoring of autism cohort systems in California, ultimately leading up to their closure for autism surveillance, This occurred one week before the publication of this premature and totally biased study looking at the autism rates in the CDDS. Such a sequence of events raises considerable doubt about the integrity of the research.
This was not unlike when the CDC blocked off all access to the Vaccine Safety Datalink Project after December 2000, after which, presumably, the thimerosal-phase out began. A fact worth noting is that these results are also consistent with the words of an anonymous CDC monitor who was quoted in “Evidence of Harm” by David Kirby as saying that the autism rate in the Vaccine Safety Datalink was going down during thimerosal reduction.
Even before this study, however, the California Department of Health has proven it is not credible, especially its immunization branch. The CDC funds it, and when this study in California was being done, Robert Davis was head of the Immunization Safety Office; he also helped Epidemic Intelligence Surveillance officer Thomas Verstraeten eliminate the relative risks with each draft of his study.
Robert Schechter, the lead author, is merely the successor of Loring Dales who did the glaringly flawed study from 2001 that tried to clear the MMR vaccine in a similar fashion. This study was later criticized for not having sufficient power to detect an association if one were to exist, according to a later study in 2002 by Madsen et al. which also attempted to exonerate the MMR, but omitted many children who received the MMR vaccine and developed autism because they were too young to be diagnosed. The 2001 California study also included Natalie Smith, then head of immunization in California, in its list of coauthors, as well as an attendee to the illegal Simpsonwood Meeting in June 2000 where officials discussed bringing down statistical connections between thimerosal and neurodevelopmental disorders while hiding data from the public.
The second author of the California thimerosal study, Judith Grether, prior to joining the California Department of Public Health was an epidemiologist for the March of Dimes, a charity founded on the premise of developing an effective polio vaccine. She coauthored a paper in 2002 with Lisa Croen of the Health Management Organization, Kaiser Permanente, to argue against a real rise in autism. Croen and Grether later retracted their findings after having their errors were pointed out to them by a research team led by Mark Blaxill, along with fellow autism father and professor of neurosurgery Dr. David Baskin of the Baylor College of Medicine and McGill epidemiologist Professor Walter Spitzer.
Ultimately, affiliations and prior discrediting on autism research makes it not surprising that the coauthors, Schechter and Grether, ignored a major artificial bias that would turn a decrease, especially in the youngest cases, post-reduction of thimerosal into an increase, not unlike the Denmark studies, the Swedish data, or Eric Fombonne’s “study.” The public’s knowledge of the thimerosal-autism link has been greatly skewed ever since.
The Media and the CDC’s Disinformation Campaign
Purveyors of spreading this misinformation include Eric Fombonne, who wrote a complementary article to this study entitled “Thimerosal Disappears But Autism Remains.”
Another familiar misinformant is Arthur Allen, author of “Vaccine: The Controversial Story of Medicine’s Greatest Lifesaver,” having written on his blog, "The most convincing evidence comes from California, where the number of 3-to-5 year old children diagnosed with autism has doubled over the last five years, although children now being diagnosed with autism received little or no thimerosal-containing vaccines."
In The Los Angeles Times, Michael Fumento, a freelance writer noted for his strong industry connections, just wrote in a February 5, 2010 piece, “Anti-vaccinationists initially claimed California autism cases dropped. False. The ‘data do not show any recent decrease in autism in California’ despite the discontinuation of thimerosal use, the state's Department of Developmental Services found in 2008.”
Meanwhile, Gardiner Harris writes in The New York Times, “Because of concerns over the preservative, vaccine makers in 2001 largely eliminated thimerosal from routinely administered childhood vaccines.
But this change has had no apparent impact on childhood autism rates.”
USA Today also repeats this inaccuracy, claiming “autism rates continued to rise after thimerosl was removed from virtually all child vaccines in 2001,” even linking to Schechter and Grether’s completely flawed study.
This myth has even trickled down to academia and is currently being taught in universities. According to my own textbook, “Human Genetics” by Ricki Lewis, “Scientific evidence does not support a link to the mercury compound once used in vaccines-autism has increased since that ingredient has been removed.”
Even worse, the California Department of Public Health study is widely cited to claim that thimerosal is safe, and therefore fine to leave at preservative levels in seasonal flu shots routinely recommended for pregnant woman and children, despite the fact that the thimerosal exceeds EPA safety limits.
In fact, last January the CDC launched a press release to encourage pregnant women and children to get the multi-dose H1N1 vaccine that also contains the preservative. The last section is titled, “Research Shows No Link Between Thimerosal and Autism.” The last sentence of this reads, “In fact, sadly, autism rates have actually gone up since thimerosal was taken out of childhood vaccines in 2001, providing further evidence that thimerosal-containing vaccines are not related to autism.”
What is truly sad is that this big hungry lie continues to be repeated in order to justify the population-wide poisoning of countless infants and fetuses.
Jake Crosby is a college student at Brandeis University who is double-majoring in History and Health: Science, Society and Social Policy, and a contributing editor to Age of Autism.