At midnight, the Environmental Health Perspectives (EHP) posted an “in-press” paper on autism and pesticides, slated for full publication in the next few weeks. Janie F. Shelton, Estella M. Geraghty, Daniel J. Tancredi, Lora D. Delwiche, Rebecca J. Schmidt, Beate Ritz, Robin L. Hansen, and Irva Hertz-Picciotto, “Neurodevelopmental disorders and prenatal residential Proximity to Agricultural pesticides: the CHARGE Study,” Envt’l Health Persp. (advanced publication: June 23, 2014).
The paper was embargoed until midnight, but the principal investigator, Prof. Irva Hertz-Picciotto, violated that embargo by talking about the study’s results in a YouTube video, posted two weeks ago. See “Selective Leaking — Breaking Ingelfinger’s Rule” (June 20, 2014).
The paper is already attracting media attention. Predictably, the coverage trades on inaccurate and misleading terms, such as “links” and “increased risks.” See, e.g., Agence France-Presse, “Study finds link between pesticides and autism,” (Yahoo news story claiming “link” in headline, but in text, noting that the study findings “do not show cause-and-effect.”); Arielle Duhaime-Ross (The Verge), “Study further confirms link between autism and pesticide exposure: Living near farms and fields can put a foetus at risk,” (June 23, 2014 12:01 am) (filed one minute after the embargo was officially lifted, and declaring that “neurotoxins, which include everything from pesticides, to mercury and diesel, are thought to alter brain development in foetuses. Now, a new study further confirms this link by showing that pregnant women who live within a mile of farms and fields where pesticides are employed see their risk of having a child with autism increase by 60 percent — and that risk actually doubles if the exposure occurs in the third trimester”); Zoë Schlanger (Newseek), “Autism Risk Much Higher for Children of Pregnant Women Living Near Agricultural Pesticide Areas” (June 23, 2014).
There are few more incendiary issues than autism or brain damage and environmental exposures. The media is unlikely to look very critically at this paper. News reports talk of “links” and “increased risks,” but they do not look at methodological problems and limitations. They should.
The media should also look at conflicts of interest (COIs). Well, in an ideal world, the media and everyone else would stop trying to use COIs as a proxy for interpreting study validity. The reality, however, is that much of the media treats corporate financial interests as sufficient reason to discount or disregard a study. If the media want to avoid being hoisted with their own hypocritical petard, they will look closely at the undisclosed COIs in this new paper by Shelton, et al.
First, they will note that the authors disclose that they have no COIs:
“Competing financial interests: The authors have no competing financial interests.”
Second, the media will note that EHP provides explicit instructions to authors on COI disclosures:
“Competing Financial Interests
EHP has a policy of full disclosure. Authors must declare all actual or potential competing financial interests involving people or organizations that might reasonably be perceived as relevant. Disclosure of competing interests does not imply that the information in the article is questionable or that conclusions are biased. Decisions to publish or reject an article will not be based solely on a declaration of a competing interest.
***
Employment of any author by a for-profit or nonprofit foundation or advocacy group or work as a consultant also must be indicated on the CFID form.”
EHP Instructions to authors (2013).
Third, the media will ask whether the COI disclosure (“none”) was proper. The study is one in a series of papers that comes out of research funded by the federal government, THE CHARGE STUDY: CHILDHOOD AUTISM RISKS FROM GENETICS AND THE ENVIRONMENT (2R01ES015359-06). Journalists may want to look, in the first instance, to the principal investigator, Irva Hertz-Picciotto. Hertz-Picciotto is an epidemiologist at the University of California, Davis, where she is the chief of the Division of Environmental and Occupational Health, Department of Public Health Sciences.
Fourth, the media may want to ask whether Dr. Hertz-Picciotto’s COI disclosure complied with the journal’s requirements. Recall that EHP requires authors to disclose work or consultancy for a “nonprofit foundation or advocacy group… .” Dr. Hertz-Picciotto sits on the advisory board of Autism Speaks, an advocacy group. More telling, Hertz-Picciotto also serves on the advisory board of the radically anti-chemical Healthy Child, Healthy World organization, located in California (12100 Wilshire Blvd. Suite 800, Los Angeles CA 90025). According to its website, Healthy Child Healthy World is a California non-profit corporation that advocates to:
“ • Demand corporate accountability
• Engage communities for collective action
• Support safer chemicals and products
• Influence legislative and regulatory reform.”
Both organizations would seem to come under the EHP COI disclosure policy, but these memberships are not disclosed in the on-line article. Certainly, these affiliations are every bit as potentially enlightening about the principal investigator’s motivations and methodological choices as corporate sponsorship. Of course, it is possible that Dr. Hertz-Picciotto made these disclosures, but the EHP editors chose not to make them public. If so, shame on the editors.
Most important, the media should provide critical review of the substance of the Shelton paper, and certainly more than sound bites on COIs or “links.” For one thing, even a quick review shows that there are four exposure periods (pre-conception, and three trimesters of pregnancy), two outcome variables (autism spectrum disorder and developmental delay), five exposure substances, and three exposure proximities, for 120 comparisons. The statistical analysis in the paper uses an alpha of 0.05, which provides a study-wise Type I error rate, and cannot be used to evaluate any one of the 120 comparisons. The paper’s use of “statistical significance” terminology should be taken with a grain of salt.[1] For another thing, many of the risk factors identified in other studies are not addressed here. See, e.g., Xin Zhang, Cong-Chao Lv, Jiang Tian, Ru-Juan Miao, Wei Xi, Irva Hertz-Picciotto, and Lihong Qi, “Prenatal and Perinatal Risk Factors for Autism in China,” 40 J. Autism Dev. Disord. 1311 (2010) (“In the adjusted analysis, nine risk factors showed significant association with autism: maternal second-hand smoke exposure, maternal chronic or acute medical conditions unrelated to pregnancy, maternal unhappy emotional state, gestational complications, edema, abnormal gestational age (<35 or >42 weeks), nuchal cord, gravidity >1, and advanced paternal age at delivery (>30 year-old)). Ultimately, a more demanding inquiry may be required to investigate the extent to which anti-pesticide advocacy groups have actually created an apparent increase in autism rates by informational, political, and environmental campaigns.
[1] See United States v. Harkonen, No. C 08–00164 MHP, 2010 WL 2985257, at *1 (N.D. Cal. July 27, 2010), aff’d, 510 F. App’x 633, 636 (9th Cir. Mar. 4, 2013)(affirming wire fraud conviction for author of press release who failed to disclose that endpoint was not prespecified, and failed to adjust for multiple comparisons), cert. denied, ___ U.S. ___ (Dec. 16, 2013).
