Statement by the University of Rochester Research Team
Studying the Effects of Methylmercury
Read Before the Senate Subcommittee on Clean Air, Wetlands, Private Property and Nuclear Safety,
Committee on Environment and Public Works
October 1, 1998

Thank you for the opportunity to present the views of our research group which has been studying the health of methylmercury (MeHg) exposure for over 25 years. Mercury is a natural element in the earth's crust, active chemically, and excess exposure to most forms is toxic to the human nervous system. Methylmercury is especially dangerous. The following summary presents our view on the effects of MeHg exposure from fish consumption on child development.

Mercury Poisonings

In the 1950's industrial pollution in Japan resulted in high levels of MeHg in ocean fish and several thousand cases of human poisoning from consuming the contaminated fish. The exact level of exposure was never determined, but it was thought to be very large. During that epidemic some pregnant women who were exposed had few or minimal symptoms of poisoning, but their babies were born with brain damage and many had cerebral palsy, mental retardation, and seizures. This suggested that MeHg crosses the placenta from the mother to the fetus and that the developing nervous system was especially sensitive to its destructive effects.

In the early 1970's another epidemic of MeHg poisoning occurred in Iraq where people ate seed grain coated with a MeHg fungicide. Our research team studied the children of about 80 women who were pregnant during this outbreak and consumed varying amounts of MeHg. We measured the mothers' exposure, examined the children, and concluded that there was a possibility that exposures as low as 10 ppm in maternal hair might be associated with adverse effects on the fetus. This value is 10 times the average in US, but a value seen in fish eating populations.

Mercury found naturally in the environment

In aquatic environments bacteria can convert inorganic mercury to MeHg. Once MeHg enters the food chain, it is bioaccumulated, and all fish have small amounts in their flesh. Predatory fish or mammals such as whales have the largest amounts. Most commercial oceanic fish in the US has < 0.5 ppm MeHg, but some freshwater fish have 2-3 ppm. Fish in Japan had up to 40 ppm.

People who consume fish are exposed to MeHg, and regular fish consumption can lead to hair mercury levels of 10 ppm or higher. The average hair mercury level in the US is < 1 ppm. If MeHg does affect the developing brain at such low levels, mothers who consume large amounts of fish would be exposing their babies to this risk.

Since the toxic effects of MeHg from fish consumption were not scientifically proven, we decided to investigate the question directly. We initiated the study of a sentinel population for the US in 1987.

The Seychelles Child Development Study (SCDS)

The SCDS is a collaborative study carried on by researchers at the University of Rochester Medical Center in Rochester, NY and the Ministries of Health and Education in the Republic of the Seychelles. Funding has come from the National Institute of Environmental Health Sciences, the Food and Drug Administration, and the governments of Seychelles and Sweden. The study has been in progress for over 10 years. The Republic of the Seychelles is an island nation in the Indian Ocean off the East Coast of Africa.

Our study was designed to determine whether prenatal exposure to MeHg from consumption of a fish diet is associated with developmental effects. The team's original hypothesis was that MeHg at levels achieved by regular maternal consumption of fish would be associated with adverse effects on child development.

The Seychelles was chosen partly because they have high levels of fish consumption. The Seychellois usually eat fish twice a day. The average mercury content of fish in Seychelles is 0.3 ppm and that is similar to ocean fish purchased commercially in the United States. There is no mercury pollution in Seychelles and many things that complicate studies of low level exposure are not present. Health care is free, universal and readily available. Prenatal care is nearly 100%, the birth rate is high, and the general health of mothers and children is good. Education is free, universal, and starts at 3˙7E/˙7E years of age. There is limited emigration and both the people and the government are cooperative and supportive.

The study design was carefully planned since detecting the lowest effect level of any toxin requires looking for very subtle differences between children with no or very low exposures and those with higher ones. These differences are often detectable only statistically. To minimize any possibility of bias, a number of decisions were made before the study began. First, the study would be double blind. Neither the clinical team nor the families know the level of any child's exposure. Second, children with a known cause of developmental delay such as meningitis, very low birth weight, or brain trauma were excluded. Third, the test battery included tests previously reported to show associations with MeHg exposure, tests used with other toxic exposures, and tests that might detect subtle changes. Fourth, all testing was performed in age windows to minimize the effect of age on test interpretation. Fifth, multiple complicating factors (covariates) were studied. Sixth, the data analysis plan was determined before the data were collected to minimizes the possibility that the data will simply be analyzed until one finds the expected effect.

The SCDS main study involves over 700 mothers and children who were enrolled during by me during the year I lived on Mahe. They have been evaluated regularly for over 5 years. Prenatal exposure was measured in mothers' hair levels during pregnancy, and postnatal exposure was measured in the children's hair at 5˙7E/˙7E years of age. Both ranged from 1 to 25 ppm. Evaluations have been completed at 6, 19, 29, and 66 months of age. The children's homes were evaluated when they were about 4 years old. Presently evaluations are being done at 96 months of age (8 years).

The results of the SCDS so far indicate no adverse developmental effects from prenatal MeHg exposure in the range commonly achieved by consuming large amounts of fish. Through 66 months of age we have examined the association between 36 primary test outcomes and the children's prenatal mercury exposure. Only one test showed an association with higher mercury exposure, and we are unsure if this effect is adverse. The test was a subjective evaluation of the children's activity level during the evaluation, and in males the activity level declined as MeHg exposure increased. Additionally, a number of secondary endpoints have been examined and no adverse associations with MeHg have been found.

We have also examined the association between the children's postnatal MeHg exposure and the 6 primary test outcomes at 5˙7E/˙7E years of age. With several of these outcomes the children did slightly better as the MeHg increased. This cannot be due to the MeHg, so we believe that the MeHg level may simply be a marker for fish consumption at these low exposures. This may confirm our understanding that fish contains nutrients such as omega-3 fatty acids that are important for brain development.

Our interpretation of the findings

Our studies in Iraq raised the possibility that MeHg exposure from eating fish might adversely affect development, but we do not believe the SCDS has demonstrated an adverse association through the first 5.5 years of life in this population. We consider the Seychelles an appropriate sentinel population for the US since they consume large amounts of fish, the MeHg content of the fish is similar to that of commercially available fish in the US, and the health and welfare of the people are similar.

Fish is an important source of protein in many countries, and large numbers of mothers around the world rely on it for proper nutrition. The mothers' nutrition is very important to the baby's wellbeing. Not one person of any age has been reported with MeHg poisoning from eating fish since the poisonings in Japan during the 1950's and 60's.

The nutrients that fish contains may be important for brain development. For older individuals, fish appears to have cardiac benefits and mental health benefits. Fish consumption is increasing in developed countries including the United States. We believe it would be unwise to limit fish consumption without convincing scientific evidence that exposure at the levels seen with fish consumption is harmful.

The SCDS is continuing, and as the children get older increasingly specific tests can be performed. We are presently completing evaluations at 8 years of age and planning more at 12 years of age. Findings will be reported as they are available.


Because of the public health importance of the question being studied by the SCDS, the potential exists for differing opinions of scientific findings to become highly politicized. The SCDS has received only one published criticism (JAMA, 280:737, 1998), but other points have been raised at conferences. These questions are addressed here individually.

Why did the SCDS measure mercury in the hair rather than in the blood? Hair mercury was used because it is the standard measure used in nearly all other studies of this question. It was also chosen because blood tests are unnecessarily invasive, reflect only recent exposure rather than exposure over time, and can fluctuate widely depending upon recent meals.

Did the SCDS use subjects whose mercury values were too low to detect an association?

The study's goal was to see if the children of women who consume fish regularly were at risk for adverse developmental effects from MeHg. Women in Seychelles eat fish daily and represent a sentinel population with MeHg levels 10 times higher than US women. Their children are more likely to show adverse effects if they are present. These children show no adverse effects through 5˙7E/˙7E years of age suggesting that eating ocean fish when there is no local pollution is safe.

Did the SCDS use the best tests available to detect developmental problems? The SCDS used the same developmental and psychological tests used in most other developmental studies. These tests are deemed to be excellent measures for determining development at the ages studied. As the children become older, additional tests with more specificity are being used.

Did the SCDS find expected associations between developmental problems and birth weight, socioeconomic factors, and other covariates? The study was not designed to examine such relationships. Some children with such problems (i.e. head trauma, very low birth weight, etc.) were excluded from the study because they are so frequently associated with developmental problems. However, many expected relationships were found.

Did the removal of statistical outliers in the analysis bias the study? No. It is standard practice to remove statistical outliers, which are values that are inconsistent with the statistical model employed to analyze the data. Every statistical analysis depends on a model, and every statistical model makes assumptions about the statistical (distributional) properties of the data that must be satisfied if the results of the analysis are to be correct. Sound statistical practice requires that the necessary assumptions be checked as part of any statistical analysis. Examination of outliers constitutes one of these checks. Statistical outliers are defined by the difference between the actual test score for a child and the value predicted by the statistical model. Small numbers of such outliers occurred in test scores for children with widely varying MeHg exposures. In fact the results of the analysis were examined both before as well as after the removal of outliers, and for analyses in the main study they had little effect.

What about the Faroe Islands study where prenatal MeHg exposure was reported to adversely affect developmental outcomes? Questions about the measure of exposure, concomitant exposures, and the statistical analysis have been raised about the Faroes study. Exposure was measured in both umbilical cord blood and maternal hair. Associations with cord blood mercury levels were reported, but these are difficult to interpret since levels vary with recent meals and their relationship to exposure during the earlier parts of pregnancy is unknown. The main source of MeHg was consuming whale meat and blubber and they also contain high levels of PCBs, inorganic mercury, and other toxic compounds. In addition, the authors have not reported the details of the statistical analysis that led to their conclusions.

_ Are the children in Seychelles too developmentally robust to find the effects of MeHg if they are present? Children in Seychelles tested similar to US children on nearly all measures apart from motor skills where they were more advanced. There is no reason to think that they are too robust to show the effects of MeHg if any are present.

_ Are children in Seychelles exposed to PCBs or other toxins? Sea mammals are not consumed in Seychelles and measured PCBs in the children's blood were low.

Should data from the Seychelles be considered interim? No. Among developmental studies, a five˙2Dyear follow˙2Dup is considered quite good, and adequate to identify most toxic exposures.