Air Pollution and SIDS

Air Pollution and SIDS
August 1, 1997



A new study issued by the Environmental Protection Agency and the Centers for Disease Control and Prevention concludes that air pollution causes SIDS and other post-neonatal mortality. The study’s timing--it was released to coincide with deliberations over EPA's proposed air quality standards on particulate matter and ozone--weak hypothesis, and even weaker methodology have led some to ask whether its authors haven’t redefined SIDS to mean “science is dead syndrome.”

The EPA/CDC researchers examined 4 million infants born between 1989 and 1991, finding that areas with high levels of particulate air pollution had a 26 percent higher incidence of SIDS than the least polluted areas.

Public health expert Steven J. Milloy, executive director of The Advancement of Sound Science Coalition (TASSC), asks the following questions on his widely acclaimed Junk Science Home Page (http://www.junkscience.com):

  • Why did the EPA/CDC researchers study outdoor air quality and SIDS? If SIDS is caused by something in the air, isn't it likely that indoor air is more relevant? Isn't that where infants spend most of their time?
  • Assuming that the 26 percent increase in SIDS is accurate, isn’t this nevertheless a weak statistical association? The National Cancer Institute, American Cancer Society, and many others have found that weak statistical associations (those that report less than 100 percent increases in risk) are unreliable.
  • Where are the exposure data in the EPA/CDC report? The researchers assumed infants were exposed to particulate matter at levels equal to measurements taken by outdoor monitors. But isn't that somewhat unreliable . . . especially, since infants spend the vast majority of their time inside?
  • Why did the study exclude infants from California, Nebraska, Indiana, Louisiana, New York, Oklahoma, South Dakota, and Washington, and those born in counties with populations less than 100,000?
  • Why did the researchers study only the years 1989 to 1991? Would the statistical association be lost if the analysis were done for 1988 to 1991, or 1989-1992, or some other time period?
  • Isn't it true that no one really knows what causes SIDS? Doesn't that make it difficult to rule out possible confounding factors?