Rocket science fuel

A key study on the safety of perchlorate in the water is coming under attack.

The researchers, backed by a grant from the industries that make and use perchlorate, concluded that the infinitesimal amounts in their test had no effect on the healthy adults who signed on for the two-week study. In February, the research team's findings became the linchpin of a national policy on how much perchlorate can be safely consumed. Federal regulators will use the policy to decide whether to limit perchlorate in drinking water, and what the limit should be. But a growing number of scientists see cracks in the foundation of that policy. The government's reliance on the study has come under fire by regulators in at least three states.

The main objection seems to be that the report doesn't reveal all the details about every subject in the study.

The unpublished data, obtained by The Press-Enterprise, show that perchlorate could have inhibited thyroid function in at least two people recruited for the study. The researchers mathematically summarized much of the data in ways that made it impossible to see potential effects, said Michael S. Hutcheson, head of the air and water toxics division for the Massachusetts Department of Environmental Protection.

So what were the findings that got buried in the summary?

The authors didn't report in the journal that a 34-year-old woman had a 39 percent reduction in iodide absorption and a 46-year-old woman had a 36 percent reduction. Three others had increased function, including one whose absorption increased by 39 percent.

My hunch: If you take a bunch of people and monitor their thyroid function for two weeks, you'll find that iodine absorbtion fluctuates. Some will show increased absorbtion by the end of two weeks, and some will show decreased absorbtion. Before we can draw any conclusions about the effect of perchlorate on the volunteers in this study, we should at least have some idea what happens in the absence of any increased dose of perchlorate.

Until we know what "normal" looks like, we can't evaluate what "abnormal" is.