So, I'm in one of those Facebook/internet arguments with someone regarding a recent report on a study linking HFCS to Autism...
Looking at the attached Provisional PDF in the second link doesn't lead me to any deeper understanding of the "study".
Since I'm not familiar with this particular study, and I'm not well versed in the science here (as it relates to Autism), I'm wondering if anybody in here is. I'm not looking for skeptical points simply to argue online as much as I am hoping to direct someone I know with one or two reasonable comments regarding how they should look at the comments made online by the authors of this study.
Based on a conglomeration of sources, the investigators [Default et al (2012)
] have pieced together a model of how the combination of mercury exposure—chiefly from high-fructose corn syrup (HFCS)—and insufficient dietary intake of magnesium could, via interaction with a certain gene (PON1), impair nervous system development and lead to autism. However, the investigators have no direct evidence that the model is correct—that is, no one has tested it—so, at this point, it's just a hypothesis.
The investigators' main evidence of a link between HFCS consumption and autism is an ecologic correlation
* between the increase in HFCS consumption and the number of children enrolled in special education classes in the US with autism diagnoses. This association has all the usual limitations of correlation as evidence for causation, plus the additional limitations of ecologic evidence (notably, the ecologic fallacy), plus the novel problem of using enrolment in special education as treatment for autism as a surrogate for incidence of autism. The investigators' claim that HFCS is the primary source of mercury exposure in the US is based on a comparison with Italy, where, the investigators claim: (1) HFCS consumption and autism incidence have remained constant, and (2) other than HFCS, levels of sources of mercury exposure are similar to the US. However, this is just another ecologic comparison, with no connection to exposure or disease diagnosis at the individual level, exacerbated by a plethora of uncontrolled-for differences between countries.
The investigators had previously published a model positing a relationship between mercury from HFCS and autism. In the present paper, they augment their previous model with additional non-mercury mechanisms by which HFCS could influence the development of autism. This gives me the impression (right or wrong) that they have been engaged in searching out additional evidence for a hypothesis to which they were already ideologically committed. The paper does not present a systematic review of the literature on these proposed mechanisms; if there is contrary evidence in the literature—and there usually is—the paper makes no mention of it.
Indeed, the status of mercury itself as a cause of autism is unclear. A recent review
concluded that there is no connection between autism and mercury from vaccines, and that evidence for mercury from other sources is ambiguous.
In conclusion, the paper presents a hypothesis in the form of a model, with no direct evidential support. Rather, the evidence on which the model was based was pieced together from a wide variety of sources, which the authors make no attempt to systematically evaluate—if there is another side to the story (and there usually is), the paper makes no mention of it. Furthermore, the epidemiological evidence offered in support of the model is ecological data, which is very poor evidence for causation.
*An ecologic correlation is a correlation between two variables measured at the group level. Claiming that an ecologic correlation applies to the individuals comprising the groups is known as the ecologic fallacy
. For further explanation, see my post here