Fable: The EPA Reference Dose includes a 10-fold “safety margin”

Facts: This slightly different approach to spinning an extra factor of 10 into the EPA’s definition of safe exposure levels is based on a serious conceptual error.

When the EPA set the Reference Dose, it began with a level of mercury in blood that had a statistically significant adverse effect on children’s cognitive development, in the best-designed epidemiological study. That “benchmark” blood level was 58 micrograms per liter. So, from the outset, 58 micrograms per liter was known to be a harmful level, not a no-effect level, a “level that approaches risk” or any other industry distortion. EPA then applied a 10-fold “uncertainty factor” to calculate the Reference Level of 5.8 micrograms of mercury per liter of blood, from which it then derived the Reference Dose of 0.1 micrograms of methylmercury in the diet per kilogram of body weight per day.

An “uncertainty factor” is not a “safety margin.” An uncertainty factor is applied to account for unavoidable scientific uncertainties. One major source of uncertainty in this case is the lack of any obvious “threshold” below which exposure to methylmercury has no adverse effects; in fact, the level of mercury in blood at which no adverse effects occur has not been determined. A second source of uncertainty is the well known variability of individuals’ sensitivity to toxic effects.

People vary in their response to mercury in the diet in two critical ways: First, the blood level that results from a given dose in the diet varies, and second, individual sensitivity to the toxic effects of mercury at any given blood level varies.

An uncertainty factor in a safety standard seeks to account for known human variability, so that sensitive individuals, as well as the average person, are adequately protected. In this case, EPA used empirical data to model the relationship between dietary mercury intake and blood mercury; that evidence showed at least three-fold variation in dietary doses associated with an average blood mercury level of 58 micrograms per liter. EPA then applied an additional factor of three to take into account likely (but unmeasurable) variation among individuals in sensitivity to toxic effects. The combined effect of applying two factors of 3 was a reduction of the RfD by a factor of 9, rounded off to 10 for simplicity’s sake.

Thus, there is no “10-fold safety margin.” Starting with a blood mercury level that had clear, measurable adverse effects, 58 micrograms per liter, EPA divided it by 3, because some people develop a blood level of 58 micrograms per liter from 1/3 of the dietary intake that produces this blood level in the average person. EPA then divided by 3 again, because some people are likely to experience adverse effects at blood mercury levels 1/3 as high as those that affect the average person. By including both of these variables, EPA tried to set a reference dose that would be “safe” for people at the sensitive end of the normal human distribution, not just for average individuals. One size does not fit all, and whatever “safety margin” may be built into the reference dose is sure to be different for different people. Some experts in mercury toxicity believe that an RfD at least 50 percent lower would be fully justified by the available evidence.

In fact, two studies published in 2008, one in New York City and one in Boston, have measured adverse effects of low-level mercury exposure on the developing brains of babies whose mothers ate fish just twice a week or less, and whose blood mercury levels—although significantly higher than average—were close to or even below the EPA Reference Level. This evidence strongly suggests that even ordinary exposure to mercury from fish poses a risk to the developing brain, and that there is no actual margin of safety at all built into the RfD.

The bottom line: The RfD, the current definition of safe exposure, was based on evidence that was available when it was set, in 1999. While the RfD defines “reasonable certainty of no harm,” some uncertainty still remains within that “reasonable certainty.” The RfD protects different individuals to different degrees, and for some, the margin between the RfD and a potentially harmful exposure is undoubtedly quite narrow. As future research provides better data, the RfD is likely to be revised.