Medical Hypotheses—“just make shit up; we’ll publish it”

Orac was kind enough to pollute my inbox with the latest idiocy from the journal that has never met a crank it didn’t like. As Orac says, “Medical Hypotheses [is] the journal where the editors encourage the authors to make shit up.”

Before I tell you about the latest “hypothesis”, let me give you an idea of what kind of thinking goes into this publication. The latest issue has an editorial that argues that it is the “maverick” scientist who makes the real scientific breakthroughs, and that teamwork is only for the “modestly talented”.

Not that there’s anything inherently wrong with being somewhat unconventional, but “maverick” implies something specific: “a lone dissenter, as an intellectual, an artist, or a politician, who takes an independent stand apart from his or her associates.”

Science works because individuals don’t get to just “make shit up”. If I have what I think is a crazy but great new idea, I have to test it, but even more important, others have to test it too and verify my findings. There is no “maverick” science, as ultimately, ideas must be tested in the wider community to be legitimized.

Mavericks have made interesting discoveries, but never in a vacuum. The Galileos of history worked in context, and they were not brilliant because of their “maverick-ness”, but because they were right.

So, moving on to their latest “hypothesis”

Autism spectrum disorders are complicated and multifactorial. We don’t know much about cause. We do know, by epidemiologic studies that autism and vaccines are not causally associated. We also have no reason to suspect an infectious agent or, say, aliens.

But that doesn’t stop a maverick.

A recent issue of Medical Hypotheses includes an article that tries to link autism to Lyme disease. Why pick those two diseases? Who knows. Perhaps because both attract a lot of cult medicine.

So let’s examine this “hypothesis”.

The paper doesn’t start well, as it begs the question immediately. For example, it simply assumes there is an autism “epidemic” that is “spreading” while evidence has shown this epidemic to be largely artifact. They also note that “[a]n association between Lyme disease (LYD) and other tick-borne infections (TBI) during fetal development and in infancy with autism, autism spectrum disorders (ASD) and autistic symptoms has been noted by numerous clinicians and parents.” I’m not sure what that means. If they are going to base a whole paper on a particular assertion/observation, they probably should cite some data. Just sayin’.

In the introduction, they also make a rather odd statement:

Since environment changes faster than genes, the rapidly emerging epidemic and geographical spread of ASD suggests significant environmental contributors, that may include infections.

W. T. F. First, is there a “geographic spread” of ASD? OK, let’s just assume there is (after all, the authors did). If there is, does this geographic distribution match the distribution of Lyme infections? For instance, is ASD more common in New England and Long Island, where Lyme is endemic? Is it less common in the Southwest, where Lyme is vanishingly rare? Their odd statement about genes and environment is not some grand discovery. Finding environmental factors through epidemiologic studies is pretty, well, standard. I wonder if they did that? I wonder if they even read the extant literature? Apparently, their “epidemiology” was limited to directly comparing incidence of autism and of Lyme disease on a state by state basis, without clear definitions or the use of standard statistics. Here is their blindingly stupid observation:

In a geostatistical review of CDC and IDEA statistics 10 out of the top 15 states overlap for the incidence of autism and LYD (MN, ME, MA, MD, CT, WI, RI, NJ, PA, VA).

The science of statistical analysis was developed to do away with this type of primitive observation.

It’s difficult to overstate how bad this paper is. First, it relies on the journal’s stated preference for maverick (i.e. way the hell out there) ideas. It starts with a hypothesis whose implausibility
matched only by it’s weak observational foundation. It then goes on to support this hypothesis with case reports and other weak “evidence”.

Ideally, an unusual hypothesis like this serves to inspire work on a new idea (contrary to the “maverick theory”). For this to happen, the hypothesis must be based on a plausible idea (it will have to stand up to a Bayesian analysis at some point). This paper is not such a jumping-off point, unless you are jumping off a cliff into a pool of fantasy.

Medical Hypotheses is truly the journal where one can “just make shit up.”


Robert C. Bransfield, Jeffrey S. Wulfman, William T. Harvey,
Anju I. Usman. The association between tick-borne infections,
Lyme borreliosis and autism spectrum disorders. Medical Hypotheses (2008) 70, 967-974.