Monday, September 3, 2012

Writing About Autism and Science? 10 Things to Remember

Emily Willingham

Word on the street -- well, really on a blog from a researcher -- is that the writer of the autism/inflammation New York Times op-ed, Moises Velasquez-Manoff, is working on an "annotated" version of the article that will "back up" his claims." Some annotation would have been useful to begin with; as I noted in my analysis of that op-ed, in many instances, discerning the origin of his information was difficult or impossible. I understand that something that appears originally in print can't have hyperlinks to appropriate references, but the writer certainly could have beefed up in-text citations (author names, journal publications), at least avoiding the criticism that the piece was unsatisfactorily sourced.

Sourcing was not the only issue with that piece, however. Velasquez-Manoff has a book to sell, and that book, like many popular science books, has a narrative that may not necessarily stand up to precise scientific scrutiny, if the NYT piece was an indication (people will be able to determine for themselves on September 4). Indeed, a criticism leveled at many such books is this lack of care in qualifying scientific results and interpreting them with caution and caveats. It's an understandable instinct. Think about the good storytellers you know. How much care do they exercise in avoiding hyperbole or sins of co-mission and omission in telling a "good" story? Some writers of popular science books commit these sins to tell a good story, too. Others do not, yet their work is popular and well received. It is possible to be accurate and cautious and still tell a rousing tale.

The problem with writing about science, though, is that science isn't just a story. It's about facts and open questions, and it's almost never defensible to write as though a door has closed, a box has been checked, or a mystery has been completely solved. We owe it to readers to avoid simplification to the point of a sin of omission and to avoid overinterpreting to the point of hyperbole.

When it comes to writing stories about health and medicine, the stakes climb. With these stories, we're not writing only about scientific findings. What we write is also about people. Many writers seek a narrative hook, a personal story that frames the rest of the piece. I can't even count the number of autism-related stories that open with a (very real) tale of woe featuring an overwhelmed or traumatized parent talking about the grief and horror of having a child with autism. This tactic catches the reader -- and happens to be one that the largest autism nonprofit in the United States also employs -- successfully tugging at heart and purse strings and attracting mouse clicks. But this tugging and this narrative approach are so frequent in such stories as to be near-cliches, and they do few favors for the autistic people these stories are really about.

The presentation of autism as a monster to battle or a stalker out to destroy your life has repercussions that some autistic adults argue go beyond an unfair and painful characterization of what they believe is Who They Are. News stories about autistic people whose parents and caregivers have murdered them often carry a clear attitude of "autism is so hard, no wonder they got killed." When every news story you read describes autism as a horrific affliction and all of those with it as suffering, when mainstream news organizations persist in focusing only on what parents have to say about autism rather than talking to autistic people, when stories focus on preventing autism -- with worms, no less -- autistic people, real, living, breathing people, feel pain and get angry and argue that even if they are nonspeaking, they can be perfectly capable of communicating for themselves.

If you are someone who writes about health and medicine and who covers a story related to neurobiology -- particularly autism -- please consider the following 10 suggestions. They might help you avoid the pitfalls of hyperbole and poor interpretation and causing pain to autistic people.
  1. Interview an autistic person for insight whenever possible. If you need suggestions for leads, feel free to contact me [or any TPGA editor]. If you were writing a piece about any other human condition, would you talk only to parents or relatives of people with that condition if the people who have it could communicate for themselves?
  2. If a researcher claims to have "solved" autism, please exercise healthy skepticism and follow up with someone who doesn't have a dog in the hunt. Of all of the neurobiological conditions, autism may be the most variable. It's extremely unlikely that any one research path or group or hypothesis will explain all autism. Don't ride that wave with them.

  3. Don't generalize. Stick with what the findings say, not what the discussion or the conclusions or the authors or the news releases say. Have an ear for when someone is overgeneralizing. Example generalization: "X causes autism." What causes autism has not been established, and the causes themselves -- and how they work -- are likely going to form a very long list. We are still very early in formulating that list, much less what the items on that list do.
  4. Don't mistake correlation for cause. When a study reports a "link," that term usually means a mathematical relationship: When X was more frequent, autism was more frequent." That doesn't mean that X causes autism. It doesn't even mean that X has anything to do with autism.
  5. Don't overstate the meaning of risk. Risk is a scary word, although we all live with the 100% risk of dying someday, regardless of what other risks we face. When a study result refers to "increased risk," look at the numbers. If they say that the presence of factor X was associated with a relative risk of 2.1, for example, then the population with factor X had twice the autism compared to the group without that factor. If the average risk of having a child with autism in the absence of that factor is 1%, then this particular factor was associated with about a 2% risk. And relative risk applies only for that study -- it does not tell you what the actual risk is.
  6. Keep in mind that even these links don't imply a true causal relationship. They're just math associations. A famous example of how these relationships can end up being misinterpreted is the protein CRP and heart disease. Because of a mathematical association between the presence of this protein and the occurrence of heart disease, researchers thought for a pretty long time that CRP might cause heart disease, and drugs were even targeted to lowering its levels. Turns out, it doesn't cause heart disease, so the drugs were no use. Instead, it's either a side effect of heart disease (reverse causality) or just higher because of some indirect influence. Now, take any recent X factor you've heard is "linked" to or "causes" autism and substitute it and autism into the above story to understand how unpromising correlation can really be.
  7. Be aware of how you write about autism and of the fact that autistic people may read what you're writing. How you describe autism is, for those readers, describing themselves, their very being. Please try to avoid lapsing into the parlance of affliction, suffering, disease, desperation for a cure, war, and despair or comparisons of "low" and "high" function. A good science geek knows that function is often a matter of environment, not a constant measure. Although some autism parents may disagree, one key to making this world a better place for autistic people is for society not to see or treat them as unhearing, nonverbal, illiterate rocking obsessives who don't understand what people are saying about them. Unlike neurodegenerative or fatal diseases, autism is not universally perceived or lived as a negative condition, and it's important to remember that.
  8. If the study in question is about mice, never talk about how the results will lead to a therapy or a cure or write about the mice as though somehow, they are just tiny humans with tails. Mice have misled us before. They are only a way to model what might happen in a mammal sorta kinda related to us. They are not Us, otherwise we'd live in tiny, crowded places, having 10 children at once and ignoring them when they grow fur, and this autism thing wouldn't be an issue.
  9. Don't use phrases like "gene that causes autism" or "gene that is linked to autism" or "faulty gene" or "defective gene." What you really want to say is "gene variant" or "version of the gene." There isn't an "autism" gene; there are gene changes that might be linked to autism.
  10. Also avoid referencing "environmental factors" without providing some specific examples. Those examples should not be "chemicals" or "toxins," which are vague, meaningless, and stupid. Established environmental risk factors for autism include parental age and extreme prematurity. Try those, but handle with care.
Finally, I know deadlines are tight, but never take a paper author's interpretation as The Final Word. Try to find someone not connected with the work and get their comment. Journalism 101, I know, but it's surprising how often articles do not include this kind of balance. By balance, I don't mean "gives the other side." I just mean, "possibly modulates enthusiastic author's overinterpretation or overselling of results and their significance."

Which is where we got started with this most recent brouhaha in the first place.


This essay was previously published at