Search This Blog

Saturday, June 29, 2013

Skepticism About the Pollution Study

At Forbes, Emily Willingham raises additional questions about a recent study purporting to show a possible relationship between autism and air pollution:
So we have five groups consisting of 65 mothers each, all nurses–occupational exposure comes to mind–and an estimate piled on top of an estimate. For example, addresses covered two to three-year periods even though gestation takes nine months:
Children born from 1987 to 1990 were assigned the geographic location of their mother in 1989 (the first year of study). Children born in 1991 or 1992 were assigned the mother’s mailing address in 1991, and births from 1993 to 2002 were assigned the nurses’ addresses, updated every other year, in similar manner.
And the pollution estimates covered even spottier time points:
Hazardous air pollutant concentrations were assessed by the EPA National Air Toxics Assessments in 1990, 1996, 1999, and 2002, which uses an inventory of outdoor sources of air pollution, including both stationary sources (e.g., waste incinerators, small businesses) and 7 mobile sources (e.g., traffic) to estimate average ambient concentrations of pollutants for each Census tract based on dispersion models (United States Environmental Protection Agency 2011).
In other words, pollution estimates were based on an inventory made in each of four years over the 15-year period of births in the study, and that inventory wasn’t a direct measure but based on identified sources of air pollution, which were then used to estimate “ambient concentrations” by census tract based on a “dispersion model.” This is not a tight association, folks. That’s not the fault of the authors or even of the study design, but it’s also a strong reason to avoid getting too excited about the results.

Yet a lingering question is, Why is anyone asking this question in the first place? The scientific method relies on observations that lead to a hypothesis to test. Scientists are not supposed to, a la Baroness Greenfield, point to one thing and point to another and say, “maybe related!” just because, well, why not? In this case, we don’t even have a correlation between, say, increasing air pollution levels and increasing autism rates. And we have a far more compelling explanation for the latter: diagnostic shift and increased recognition and diagnostic capture.