It’s not surprising that this isn’t published in a journal. The authors set out to prove that vaccines cause autism as opposed to properly conducting a study.
One major red flag is that Andrew Wakefield is one of the authors. Wakefield published a paper in 1998 that led to much of the antivax hysteria. It was determined to be fraudulent and it was discovered that he had been on the payroll of an attorney who was planning to sue manufacturers, which was a clear conflict of interest. It was found that in that role he had earned at least £435,643 at the time, which adjusting for inflation, is equivalent to about £1,000,000 today, or $1,334,695 US. This is quite ironic when the antivax community often chants “follow the money” or “$cience.” The BMJ has a concise summary of the problems in his “study” as well as an entire series about it. Further, he was in the process of trying to patent his own vaccine.
His suggestion that vaccines caused autism, “followed by a campaign led by Wakefield, caused immunisation rates to slump from 92% to 78.9%,” He eventually lost his medical license over the publication and the unethical behavior behind it.
On major flaw is that they treat the studies that they cite as being equal. They are counting case reports as having equal validity to controlled cohort studies and discounts the highest quality evidence in the large population studies. They dismiss these by claiming that they “were undermined by absence of a genuinely unvaccinated control group.” This is highly flawed because unvaccinated cohorts are rate in more developed countries. The correct epidemiological response is to use dose-response, timing, within-sibling, and discordant twin designs which they do not use at all. In essence, they are rejecting anything that doesn’t match their bias.
The authors also use the vaccine package insert as evidence of adverse events and a causal pathway to autism. However, FDA regulation requires manufacturers to “report any side effects that occurred in the vaccine group, even if the number of occurrences was similar to or the same as those in the placebo group. All of these side effects are then listed in the package insert.” The authors are suggesting that the vaccines are causal, which technically is a categorical error.
They also rely on an ecological fallacy, as shown in Figure 2, reproduced below.
Correlation is not causation. A classic example of the flaws in this kind of analysis is plotting ice cream consumption against drowning deaths. In this case, the obvious confounding variable is that people tend to consumer more ice cream when it is hot outside in the summer, which is also when they are more likely to go swimming because lakes and rivers are not frozen. In this case, weather is a major confounding variable.
Another textbook example of a confounding variable is a cigarette lighter. Once could easily say that cigarette lighters cause cancer because cancer rates are higher among those that carry cigarette lighters. Hulscher et al. completely ignore the confounding variables in autism.
One of the big items that they ignore is the role of genetics in autism. In a study of over two million subjects, it was found that genetics may play a role about 80% of the time. The authors completely downplay this as being important.
They use case reports as evidence. These are considered the weakest evidence in studies because they do not establish causation and are subject to recall bias, but the authors frame it as mechanistic proof.
Typically, a systematic review is used for this kind of analysis, which might be why they dodged the term and suggest a comprehensive examination. A systemic review typically includes the search strategy, inclusion and exclusion criteria, an assessment of quality, risk-of-bias analysis, and data extraction methods. Without disclosing those, this is a narrative argument, not a systematic review.
One of the most unfounded claims about autism rates is that “overall, this represents a 32,000% increase since 1970.” This is grossly misleading. They admit “ADDM data collection began in 2000; earlier prevalence figures are based on independent studies.” They fail to disclose that these earlier studies had far narrower diagnostic criteria for autism, which didn’t include many separate diagnoses that are now included in autism spectrum disorder. Those earlier studies also did not screen entire population and did not include nonverbal or institutionalized children because of the narrower diagnostic criteria. The percentage increase they claim is not at all valid.
They also claim “12 studies comparing routinely immunized versus completely unvaccinated children or young adults consistently demonstrated superior overall health outcomes among the unvaccinated.” These studies use chiropractic patients, homeschool groups, or other populations for convenience, and are not representative samples. This is clearly selection bias. They rely on surveys of parents, not medical records. In these groups, parents are more likely to try to make their children sound healthier because of their own antivaccine bias. These results aren’t found in high quality studies.
They regularly use terms and phrases such as iatrogenic exposures, post-encephalitic state, susceptible children, and urgent public health priority, which are not science based, but advocacy for their antivaccine position.
Further, they use speculation about mechanisms without any causal evidence and fail to address the more likely explanations of increases in ASD due to broader diagnostic criteria and definitions in the DSM, increased screening, greater awareness by parents and providers, earlier identification, better survival rates of preterm infants, changes in parental ages, and environmental exposures. A legitimate review would address all of these.
This is antivax rhetoric, not science.
