I guess Barack Obama’s mad hypnotic powers worked.
One non-political thing that this election has reminded me of is that when you’ve been blogging as long as I have (nearly four years now–almost as long as a Presidential term!–assuming you’re good and have found a niche in the blogosphere, you can become one of the “go-to” bloggers for certain subjects. Even though I’ve taken on the pseudonym (and, some might say, the persona) of a cranky talking computer with a bad attitude that looked like a cheap Plexiglas box of multicolored blinking lights and was featured in a 30 year old British science fiction series that was known for its fantastic plots but BBC-level low budget sets recycled from old Doctor Who episodes, in person I’m nowhere near as arrogant as my namesake, who was known for his amazing feats of condescension towards his human owners. On the other hand, I do realize that I’m quite a good blogger and writer. I also realize that I’ve occupied a certain niche in the medical and skeptical blogosphere that’s remained fairly unique even now, covering medicine, science, skepticism, and even Holocaust denial.
Into this rather strange mix that, oddly enough, has held together over the years, is an interest in the cult of antivaccinationism (these days led by that ex-Playmate, former gross-out comedienne Jenny McCarthy) that believes that somehow, someway, mercury in vaccines causes autism or vaccines themselves cause autism. This belief persists despite virtually no scientific support and against multiple lines of evidence, including large epidemiological studies, that have looked for and utterly failed to find a detectable correlation between either mercury in vaccines and autism or between vaccines in general and autism. An unexpected development of my interest in antivaccine lunacy and the quackery that flows from that lunacy has been an interest in the science of autism and what causes the condition. Consequently, one of the topics for which I’ve become a “go-to” blogger is dubious autism science.
So it came as no surprise that a number of my readers have e-mailed me about a story about a rather odd little study about autism. Actually, it was a bit of a surprise when the first e-mail arrived, mainly because I hadn’t known that this study was in the pipeline or that it had been published, but soon I became aware that David Kirby was using it as “exoneration” and my usual sources started to weigh in. To some extent I was blindsided on this one, but I quick. So what is this story that’s buried my e-mail in box under, oh, around a dozen e-mails or so? (Hey, this isn’t Pharyngula, you know. I don’t get hundreds of e-mails about anything.)
There were stories in USA Today and MSNBC about a study that concluded that rain causes autism.
OK, that’s not really what the study claims to have found. In fact, the authors were pretty conservative about drawing inferences. However, it did claim to find a correlation between precipitation rates and autism prevalence on the Pacific Coast. The study, published in the November issue of the Archives of Pediatrics and Adolescent Medicine by investigators whose lead author is Dr. Michael Waldman of Cornell University and with collaborations with investigators at Purdue University, and the Children’s Hospital of Philadelphia (Paul Offit must be having kiniptions over this) and entitled Autism Prevalence and Precipitation Rates in California, Oregon, and Washington Counties. This study purports to provide evidence that autism prevalence is associated with precipitation. It seems to show just such a correlation, but there are a number of reasons to be very skeptical of the conclusions being drawn from this study, not the least of which are the uses to which mercury militia apologist puts it.
Besides looking at this study as a case study for the truism that correlation does not equal causation, something I want to scream at the top of my lungs when I read a study like this. The other thing that I always look at (and you should too) is the underlying hypothesis of the study:
Similarly, in the recent CDC study of autism prevalence mentioned previously, the highest autism prevalence among the 14 states studied was found in New Jersey, the second-most-northern state in the study, whereas the lowest autism prevalence was found in Alabama, the most southern state in the study.
This pattern suggests a hypothesis: namely, that there is an environmental trigger for autism among genetically vulnerable children that is correlated with bad weather, possibly because the environmental trigger is associated with indoor activities. We explore this hypothesis by empirically investigating the association between autism prevalence and precipitation rates.
This is a tad disingenuous. In reality, this study is an outgrowth of the infamous “television causes autism” study, which started with a dubious hypothesis and ran with it right into the ground. Such a hypothesis is so dubious on a strictly scientific basis, based on what we know about the biology of autism, that it would take an incredible amount of evidence to validate it. The current study resurrects some of the more pithy comments directed at the television study about the strange things that can happen when economists and management faculty apply their methodology to epidemiology and medicine. More importantly, when looking at this study, I ask why: Why do the investigators think that precipitation levels would correlate with autism prevalence? One potential reason is that increased precipitation leads to more time indoors, which could correlate with decreased vitamin D levels, which could be a “trigger” for autism. Fair enough, as far as it goes, but there are quite a few links in the proposed chain of causation, and the more links there are the more tenuous the link between the first link and the last link in the chain. Moreover, there may be other common factors that confound this proposed correlation by being associated with populations living in areas with more precipitation.
But let’s get to the study itself.
One thing leaped out at me right away, and that’s how the authors correlated precipitation with autism levels. They did not look at raw average precipitation rates. Rather, they invented a variable to correlate to autism rates:
In our tests, we used a relative precipitation variable, which we defined as the difference between the annual precipitation received in a county and the mean precipitation for all counties in the sample. The National Climatic Data Center records daily precipitation at more than 8000 weather stations in the United States. To calculate precipitation in a specific county in a specific year, we first calculated the mean across all weather stations in the county for each day of the year. We added the resulting values from all the days in the year to get the total precipitation, and then calculated mean annual precipitation by county from 1987 through 2001 and for each 3-year interval when the 1987 through 1999 birth cohorts were younger than 3 years. To find relative precipitation levels, we then subtracted the mean annual precipitation level for the counties and years in our sample. The 1987-2001 period spans the dates when children who were school aged in 2005 were younger than 3 years, the time during which autism symptoms emerge and any putative, postnatal factor would be present.
Note that the authors did not correlate autism prevalence directly with raw mean precipitations but instead used a “relative precipitation variable.” When I see something like that, I know right away that there was no correlation between raw mean precipitation levels and autism. If there had been, you can rest assured that the authors would not have bothered to go to the trouble to do this little bit of mathematical legerdemain–excuse me, I mean “transformation.” This is a general principle of epidemiological studies: If there’s a correlation with raw data, then there’s no reason to do any sort of mathematical adjustment. Don’t get me wrong, though. Such transformations are sometimes valid from a statistical standpoint, but such calculated measurements always give me pause when I see them–especially so when no explanation is given for why they are being used or why the authors consider them more valid or useful than using just the untransformed data for their calculations. Such data massaging always sends up red flags.
There are also multiple other flaws with this study:
Remember the study about industrial mercury I blogged about? You know, the one in which it was claimed that proximity to mercury-emitting power plants correlated with autism prevalence? Remember how I emphasized how important it was to control for urbanicity when looking at autism diagnoses because autism diagnoses correlate with more highly urban areas? Guess what? The authors of the current study, although they tried to correlate for household income, didn’t even attempt to control for urbanicity. That alone makes this study highly suspect, at least to me.
Another problem with this study is that it examines only the Pacific Coast, specifically California, Oregon, and Washington. There is no indication that the observations made in this study are generalizable. Sure, you can argue that the study presents a provocative finding, and the authors certainly state that it needs to be followed up on, but it’s still not a comprehensive sampling. Moreover, rainfall can vary considerably from year to year, but autism rates do not. Prevalence may be increasing, largely because of widened diagnostic criteria and diagnostic substitution, but it’s not an increase superimposed on a curve that widely varies from year to year. How can such a hypothesis account for the fact that rainfall in general is not increasing in most places but autism prevalence is?
Of course, what would be interesting about this study if it holds up is what links high levels of rainfall to increased autism prevalence. One potential factor that could link the two is, as the authors mention, vitamin D. However, there is a paucity of data linking any sort of vitamin D deficiency to autism; a lot more data would be needed. Another potential factor is being indoors and exposure to “indoor toxins.” However, there is also a paucity of evidence that anything associated with being indoors correlates with autism. Unfortunately, the authors read far too much into their data:
This study empirically examines the hypothesis that there is an environmental trigger for autism among genetically vulnerable children that is positively associated with precipitation. If there is such an environmental trigger, then the prevalence of autism should be higher in counties that receive abundant precipitation, and especially for birth cohorts that receive above-average precipitation, relative to the county’s average amount, when the cohort was younger than 3 years.
Our results support this hypothesis.
No they do not. I guess that’s what you get when you have economists doing epidemiology about medicine, you get these sorts of “fast and loose” statements about “genetically vulnerable children” when the data didn’t even address the question of genetic susceptibility. It didn’t even have the methodology to do so. Now, It’s possible there may be a genetic susceptibility to autism that is triggered by an environmental factor or factors, but nothing–I repeat, nothing–in this study supports that hypothesis. Measures of genetic susceptibility were not even a part of the study–or even looked at! To use the words “genetic susceptibility” in the conclusions and to say that this study somehow supports an interaction of genetic susceptibility and environmental factors is just plain incorrect.
The problem with this study, unfortunately, is not just how weak it is, how dubious the hypothesis under study is, or how the authors clearly don’t understand the rudiments of genetic susceptibility. Those are all problems, but the big problem with this study is that it’s custom made for cranks. In fact, autism crank numero uno David Kirby has already weighed in on–where else?–The Huffington Post:
A new study out of Cornell University says that children growing up in the rainiest or snowiest areas of the country seem to have a higher risk for autism than children living in drier climates.
The authors estimated that removing precipitation as a factor in autism would slice the prevalence of the disorder by 33% to 43%.
Among the possible explanations given were: A lack of vitamin D from a sun-deprived life under the clouds, an increased amount of time spent indoors amid toxic household chemicals, or the presence of dangerous neurotoxins in the precipitation itself, which in turn might trigger a genetic predisposition to ASD.
One of the most omnipresent, growing (and obvious) air-borne neurotoxins in the world to consider, of course, is mercury.
The desperation is palpable here. As I’ve pointed out before on numerous occasions, epidemiological study after epidemiological study have failed to find even a whisper of a correlation between mercury an autism. Antivaccinationists realize this, too, which is why they have started to grasp at environmental sources of mercury in addition to mercury from vaccines, even to the point of toutting lame studies, such as those of Raymond Palmer. Oh, David Kirby is glib, and he applies his usual twisting of logic, language, and science to try to argue that this study supports his previous fearmongering about mercury, but, let’s face it, these days Kirby is a pathetic character, grasping at any mercury-coated straw he can find in search of vindication of his crankery.
Once again, this study tells me one thing, and should tell you the same: Correlation does not equal causation. If you want to know just how dubious this story is, you should be aware that an editorial was published with the study entitled Precipitation and autism: Do these results warrant publication? In the editorial, Dr. Noel Weiss does backflips with logic trying to say that, yes, this study does warrant publication, all the while slamming it as tentative and not particularly convincing. If I were Dr. Weiss, I’d be embarrassed.
It’s also depressing to see such dubious studies being provided to cranks like David Kirby as a weapon to use against reason.