Statistics: Putting Methylmercury in Seafood Into Context

Statistics are estimates used to collect, analyze and interpret data but there are many limitations to what statistics can tell you, especially about the potential relationship between a specific substance and a certain health outcome.

When it comes to statistics about methylmercury levels in seafood, there are four groups of people who need to pay attention to this issue: 1) women who could become pregnant, 2) women who are pregnant, 3) nursing mothers, and 4) young children. With these individuals in mind, there are a number of important questions that are regularly considered by the federal government and can be used to help to put the statistics about methylmercury into proper perspective. These questions are:

What is the amount of methylmercury found in seafood?
What is the amount of seafood eaten by the four groups of people mentioned above?
How much methylmercury is actually absorbed or gets into the body? (Keep in mind that some of the methylmercury in the seafood may be excreted out of the body or not absorbed at all.)
How much of the absorbed methylmercury gets into blood?
If the methylmercury is absorbed, how much is metabolized and then excreted out of the body?
What other tissues or organs take up methylmercury or metabolize it to be excreted?
In the case of a pregnant woman, how much methylmercury is transferred through the placenta as opposed to being metabolized by and excreted from the mother’s body?
How much of the methylmercury is absorbed by the fetus?
And after, does the amount of methylmercury absorbed by the fetus have any effect on the developing nervous system?
In addition, is the effect big enough to negatively affect the baby’s or young child’s nervous system and mental abilities (IQ or intelligence quotient)?

Based on addressing these questions, the following provides additional background on some of the statistics associated with methylmercury in seafood.

1. The Number of U.S. Women with Unsafe Levels of Mercury in Their Blood

In describing the potential risk of methylmercury levels in seafood, a statistic that is widely used is that 8 percent of women of childbearing age may have unsafe levels of this substance in their blood. This statistic is based on the National Health and Nutrition Examination Survey (NHANES) 1999-2000 -- a large voluntary survey of more than 15,000 Americans that is carried out by the federal government. In NHANES, volunteers are asked questions about what foods they eat, the quantities consumed and how often they eat those foods. In addition, doctors and nurses perform physical examinations while dentists examine the volunteers’ teeth. Moreover, phlebotomists draw samples of blood, which is analyzed for many substances including total cholesterol, high- and low-density cholesterol, and numerous other compounds.

To put the meaning of the 8 percent statistic into perspective, it is important to understand how the estimate was calculated. Using the ongoing NHANES, during 1999-2000, the blood samples of 705 children and 1709 women were analyzed for methylmercury. This resulted in three major findings:

The measures of methylmercury exposure in women of childbearing age and young children generally fell below levels of concern;
None of the women tested had blood methylmercury concentrations at the “benchmark dose lower level” (BMDL) for methylmercury set by EPA, which is the lowest amount at which an effect may occur; and
At the highest percentiles, an estimated 7.8 percent of women had concentrations of methylmercury in their blood higher than EPA’s “reference dose,” which is based on applying a ten-fold safety factor to the BMDL.

In reporting on the 8 percent statistic, some stakeholders have equated EPA’s “reference dose” with the definition of the government’s “safe” level for methylmercury in fish. To address this confusion, EPA published “Reference Dose: Description and Use in Health Risk Assessments,” where the agency cautions the public not to confuse the “reference dose” with the term “safety factor.” Although the “reference dose” is often cited as the amount of methylmercury at which an effect may occur, the appropriate measure is the BMDL.

Since CDC published the 1999-2000 NHANES study, the agency conducted a second survey of the blood levels of more than 3,600 women of childbearing age and 1,500 children aged one to five years and found that all had methylmercury levels significantly below the threshold for any known risk. Published in the November 5, 2004 issue of the Morbidity and Mortality Weekly Report, the updated CDC study found that methylmercury concentrations in the blood of American women had declined over a four-year period (1999-2002). According to the new findings, the number of women with methylmercury levels at or slightly above the reference dose had dropped to 6 percent.

2. The Number of Newborns Born At Risk from Methylmercury

Other statistics used to estimate the potential risks associated with methylmercury involve the number of infants and young children who may have unsafe levels of methylmercury in their systems. While these statistics are widely quoted, they are theoretical estimates that have not been validated by peer-reviewed research studies.

The first statistic is that 60,000 newborns annually are born at risk because of exposure to methylmercury in utero. This statistic originated in 2000 with the National Research Council’s report, Toxicological Effects of Methylmercury. However, in its report, the NRC committee said “ The committee does not believe it is possible to estimate a meaningful number of children that might be affected within the ‘at risk’ population” and further stated “The number should not be interpreted as an estimate of the annual number of cases of adverse neuro-developmental effects.”

A second statistic is that 300,000 newborns each year are born at risk due to exposure to methylmercury in utero. This statistic is a calculation based on multiplying the number of births each year in the United States (4 million) by 8 percent -- the number of women who had concentrations of methylmercury in their blood higher than EPA’s “reference dose.” The number of newborns that may have been exposed in utero and that may be at risk due to methylmercury concentrations higher than those considered to be without increased risk of neurodevelopmental effects are therefore extrapolations. Since the underlying premise of this statistic is that the “reference dose” represents the “safe” level for methylmercury in the blood, this statistic has the same limitations as the 8 percent figure quoted above.

The final statistic is that 630,000 babies may be at risk. This statistic recalculates the 300,000 estimate based on a study that notes possible differences between mothers’ and babies’ cord-blood methylmercury levels. But according to the EPA, this difference in mothers’ and babies’ blood levels was incorporated into the calculation for the “reference dose.” Further, this statistic has the same limitations as the 8 percent figure quoted above.

3. Understanding Statistical Significance

In addition to understanding the genesis of the statistics used to describe the potential risk of any contaminant or “hazard,” it is also important to understand what statistics can and cannot do.

When conducting both observational and experimental research, scientists use statistical measures to convey the possible existence and strength of relationships between a substance (such as methylmercury levels in seafood) and a particular health outcome. Statistics do not provide information about a cause-and-effect relationship. Statistical significance in one study is only part of the picture; to get the whole picture, one must consider the entire body of the literature, or in other words, what other research studies reveal on the subject. Moreover, statistical significance does not necessarily mean that the relationship is biologically important.

[top]