Assessing and Managing Methylmercury Risks Associated With Power Plant Mercury Emissions in the United States

Risk Management Strategies

As a widely recognized developmental toxicant, mercury has been the target of regulation in the United States since the passage of the Clean Air Act in 1970. Mercury-containing pesticides have been banned; disposal of mercury-containing waste is strictly controlled; mercury mining has ceased; and mercury emissions resulting from municipal, medical, or hazardous waste incineration are significantly restricted. Recently promulgated regulations will restrict mercury emissions from coal-fired power plants in the future, although technologies currently in place to control other pollutants also reduce mercury emissions to varying degrees.

There are 2 basic risk management strategies that have been initiated or are under consideration to reduce potential risks from methylmercury exposure. One approach relies on defining exposure limits that are considered unlikely to be associated with adverse effects and then limiting the consumption of certain fish by children and pregnant women. Such exposure limits are meant to limit exposure once environmental contamination has occurred. Another approach requires the implementation of particular technologies that have been determined to reduce mercury emissions and is thereby meant to prevent contaminant emissions before they occur.

Health-Based Exposure Limits

Regulatory agencies and scientific organizations in the United States and Europe have identified quantitative exposure levels for methylmercury -- based partly on science but mostly on policy -- that are considered to be limits on safety. Such limits are goals that, if exceeded, may warrant actions to reduce exposure, although exceeding a limit does not imply lack of safety. Most exposure limits for methylmercury are advisory levels and not regulatory requirements. Table 2 shows the exposure levels considered protective by different organizations. All of the protective exposure levels identified in Table 2 are based on methylmercury's ability to produce developmental neurotoxicity. Most were based on the data from the Seychelles study described above, although some also considered the Faroe Islands and New Zealand data. Some derived a "benchmark dose" from the dose-response relationships for methylmercury exposure and developmental neurotoxicity, that is, a statistical lower confidence limit on the methylmercury concentration in maternal blood that resulted in a 5% or 10% increased incidence of neuropsychological impairment in infants born of women who had consumed fish and pilot whale meat. In other cases, a no-observed-adverse-effect level (NOAEL) was identified, that is, the median maternal hair concentration from the highest exposure group in the Seychelles Islands study (which, as noted above, found no significant positive association between exposure and abnormality). The benchmark dose or NOAEL, expressed as concentrations of mercury in blood or hair, was then converted to the dose of methylmercury from fish that would produce that blood or hair concentration. Finally, that dose was divided by an "uncertainty factor" to obtain a dose considered to be without deleterious effects by accounting for the possibility that some people may be more sensitive to methylmercury toxicity than others. The resulting dose is considered the amount of methylmercury that can be consumed daily without producing developmental neurotoxicity even in the most sensitive children. The EPA calls such limits reference doses (RfDs); the Agency for Toxic Substances and Disease Registry calls them chronic minimal risk levels (MRLs); and others refer to them as tolerable daily intakes (TDIs).

The fact that the exposure limits calculated by different organizations are similar suggests that the results of the different studies upon which they are based are comparable, but they are not. The Faroe Islands study was probably confounded by PCBs in breast milk and found slightly decreasing performance on neurodevelopmental tests among children correlated with increasing cord blood mercury levels. The Seychelles study found slightly improving performance on neurodevelopmental tests with increasing maternal hair mercury. The results of the Seychelles study do not appear to be confounded by other exposures and are therefore more reliable as the basis for setting an exposure limit than the results of the Faroe Islands study. Because there were no adverse neurodevelopmental effects associated with methylmercury exposure in the Seychelles study, however, the benchmark dose derived from the Seychelles data takes as its 5% or 10% response level the statistical upper bound on zero response, that is, the largest response that could have been missed, had it existed. In other words, to be health-protective, the benchmark dose is based on the assumption that the response occurred but was too small to be detected. That assumption is intentionally biased by the policy need and statutory mandate to protect public health.

Despite the health-protective nature of MRLs, RfDs, and TDIs, their meaning is often misinterpreted. MRLs, RfDs, and TDIs are derived on the basis of the most sensitive type of toxicity; in the case of methylmercury, developmental neurotoxicity is the effect that has been observed at levels of exposure lower than those that cause other health effects. Exposure limits based on developmental neurotoxicity are thus expected to protect the most sensitive individuals, children. Exposure limits incorporate uncertainty factors so that even the most sensitive children are protected. Nonetheless, especially in the case of methylmercury, their meaning has been misrepresented.

For example, Mahaffey and coworkers^[50] concluded that more than 300,000 newborns per year in the United States are at increased risk for adverse neurodevelopmental effects from prenatal exposure to methylmercury. That conclusion is based on the fact that 7.8% of 1709 women ages 16-49 surveyed in 1999-2000^[51] had mercury levels exceeding 5.8 mcg/L of blood. That concentration is the benchmark dose of 58 mcg methylmercury per liter of maternal blood from the Faroe Islands study divided by an uncertainty factor of 10.^[8] There were 4,058,814 births in the United States in 2000; 7.8% of 4,058,814 is 316,587. There are a number of problems with that conclusion. First, mercury body burdens increase with age, and the 40- to 49-year-old women in the survey had twice the blood mercury levels as the younger women. Far fewer women over 40 give birth than do younger women. The average age for giving birth in the United States is 27; more than half of all births occur to women in their 20s.^[52] It is therefore likely that far fewer than 7.8% of the women who bore children in 2000 had mercury levels exceeding 5.8 mcg/L of blood. More recent Nano Chemical Systems Holdings (NCSH) data report 5.7% of women ages 16-49 with methylmercury levels exceeding 5.8 mcg/L of blood.^[20]

Even more importantly, exceeding 5.8 mcg methylmercury per liter of blood does not necessarily mean that a fetus is at increased risk. The derivation of an RfD is based on many policy choices in order to serve regulatory purposes and has little scientific basis. According to the EPA, RfDs incorporate a number of uncertainty factors and are defined as "an estimate of an exposure . . . that is likely to be without an appreciable risk of adverse effects over a lifetime"; therefore, "exceeding the RfD is not a statement of risk.^[53]" In other words, although exposures at or below an RfD are unlikely to pose a risk, the converse -- that exposures exceeding an RfD are likely to pose a risk -- is not the case. The number of children "at risk" is determined by the dose-response relationship, not by the number of people whose doses or blood mercury levels exceed the RfD at a single point in time. Meanwhile, none of the US women or children tested in the NCSH study had methylmercury doses exceeding the EPA's benchmark dose for developmental neurotoxicity (58 mcg/L, the lower confidence limit on the dose estimated to produce a 5% response as determined with the Faroe data, assuming no confounding by PCBs).

Other reports claim that more than 600,000 children are at risk on the basis of the estimate of 300,000 described above adjusted upward to account for the approximately 2-fold difference in mercury blood concentrations in the umbilical cord compared with that in the mother. Some argue that the adjustment constitutes double-counting because the EPA's RfD calculation already takes that difference into account implicitly through the use of the uncertainty factor used to account for intraindividual pharmacokinetic differences.^[54] EPA scientists who are responsible for determining the RfD state that they did not take the 2-fold difference into account explicitly when choosing the uncertainty factor although future evaluations will do so, and they hope to use probabilistic methods to choose an RfD in the future instead of relying on uncertainty factors.

The US National Academy of Sciences (NAS) report on methylmercury also offers an estimate of the number of children who are at risk for adverse neurodevelopmental effects from prenatal exposure to methylmercury.^[8] The study authors assumed that children born to women who consume the most fish would be highly exposed to methylmercury and thus at risk. In 1989/1990, 30.5% of adolescent girls and women ages 15-44 surveyed reported consuming fish regularly. Those in the 95th percentile of that group (918,172) were thought to have consumed 100 g of fish or more daily. Because the birth rate during that period was 65.6 per 1000, about 60,000 (65.6 x 918) children were considered to have been born to high fish consumers and were therefore at risk. Because fish consumption alone was considered a surrogate for exposure to methylmercury, for which there was no direct measurement, this classification of risk is arbitrary and unreliable. The levels of methylmercury in fish vary by several orders of magnitude; assuming simply that fish consumption is equivalent to methylmercury exposure is not supportable. In response to an FDA request for clarification, the chairman of the NAS Committee on the Toxicological Effects of Mercury stated that the term "at risk" in the NAS report merely meant "above the RfD" and implied neither harm to those exposed above the RfD nor an absence of harm to those exposed below the RfD.^[55]

Not all limits on methylmercury exposure involve identifying specific concentrations in food that are considered safe (or unsafe). The FDA has worked with EPA to develop advice for consumers in regard to methylmercury and fish consumption. That advice warns pregnant women not to eat shark, swordfish, king mackerel, or tilefish because they may contain high levels of methylmercury.^[56] To avoid losing the benefits of consuming fish, however, pregnant women are advised to eat up to 12 oz (2 average meals) a week of a variety of fish and shellfish that are lower in mercury, such as shrimp, canned light tuna, salmon, pollock, and catfish.

Another example of nonquantitative exposure limits are fish advisories targeted at specific bodies of water. Each state, tribe, or territory establishes its own criteria for issuing fish advisories. An advisory may completely ban eating fish or shellfish or it may recommend consumption limits that are considered safe (eg, numbers of fish meals per specified time period). It is typically more restrictive toward pregnant women, nursing mothers, and young children. In 2003, 21 states had issued statewide advisories for mercury in freshwater lakes and/or rivers; 11 states had statewide advisories for mercury in their coastal waters; and Hawaii had a statewide advisory for mercury in marine fish.^[57] There were also 2 tribal statewide advisories put into effect for mercury in freshwater and marine fish (including lobster) by the Micmac Tribe of Maine.

Some advocates of mercury control point to the increasing numbers of fish consumption advisories due to mercury contamination as evidence that mercury from power plants has local impacts.^[58] The acreage of freshwater lakes under fish consumption advisories for mercury increased 4 times between 1993 and 2002, although that increase is likely to be due to greater awareness, more extensive testing, and a decrease in the mercury RfD, not to an actual increase in mercury contamination,^[59] and no relationship to the proximity of power plants has been investigated.

Technology-Based Exposure Limits

The Clean Air Act Amendments of 1990 required the EPA to investigate by 1993 the emissions of hazardous air pollutants from power plants relying on fossil fuel-fired steam generation for the purpose of deciding whether regulating those emissions was necessary or appropriate. The EPA was also required to investigate specifically mercury emissions from a variety of sources by 1994, in terms of their impact on health and the environment and the availability and cost of control technologies. In 1998, the EPA released a report finding that mercury was the hazardous air pollutant of greatest public health concern emitted from fossil fuel-fired power plants, but did not address whether such power plants should be regulated. Environmental groups then sued the EPA, which ultimately agreed to investigate mercury emissions and control technologies further, to make a decision about the need for regulation, and if needed, to promulgate such regulation (see Natural Resources Defense Council, Inc. v Environmental Protection Agency, No. 92-1415 [D.C. Cir. Jan. 13,1999]). In December 2000, the EPA announced that regulating mercury emissions from oil- and coal-fired electric power plants was indeed necessary and appropriate but did not issue a regulation.

One of the first steps in regulating a source of hazardous air pollutants is to identify an effective technology to control emissions from a category of pollutant sources, in this case, coal-fired electric utilities. The type of technology and extent of emissions control required depend on identifying the maximum achievable control technology (MACT), which in turn cannot be less stringent than the average emissions limitation achieved by the best performing, top 12% of existing sources for which information is available. In the case of coal combustion, identifying MACT is complicated and, therefore, contentious. One approach, based on a limited number of emissions tests from 80 plants, suggests that a 90% reduction in mercury emissions is achievable at some facilities. However, because basing regulation on the best performing 12% of sources does not account for the wide variety of processes and coal types in use, another approach divides plants into categories according to process and to coal type and chemistry, proposing different MACT standards for different categories on the basis of the amount of mercury released per unit of energy created. The effectiveness of control devices at removing mercury depends, to a large extent, on the levels of chlorine in the coal, which vary substantially depending on its source and control the types of mercury compounds in the flue gas.^[60] Some mercury compounds are captured more effectively than others.

According to the EPA and the courts, MACT requirements are expected to be "achievable under the most adverse circumstances which can reasonably be expected to recur" (64 Fed. Reg. 31898, 31915 [June 14, 1999]]. As yet, there is no single technology that can reliably control mercury emissions from all types of coal-fired power plants.^[61] Mercury reductions at all existing coal-fired power plants, including the "best" performing units, result from control equipment that was installed to reduce the emissions of other pollutants. New coal-fired power plants are subject to stringent regulation under a number of Clean Air Act provisions, including new source performance standards. These requirements cause new plants to install high-efficiency particulate removal devices, scrubbing systems, and devices that reduce nitrogen oxides. Additional control strategies that specifically address mercury, such as activated carbon injection, are under development and not yet commercially available.

In December 2003, the EPA proposed the first rule to control mercury emissions from power plants. The rule proposed 2 approaches to limiting mercury emissions, one based on MACT and the other setting a cap on mercury emissions to be achieved with market-based emissions trading. The Clean Air Mercury Rule was finalized in March 2005 and derives its authority from section 111(d) of the Clean Air Act, so is based on the trading approach and not the MACT approach. The rule involves a first phase reflecting the mercury reductions expected as cobenefits to limiting sulfur dioxide and nitrogen oxides, with a cap of 38 tons per year starting in 2010, and a second phase with an additional emissions cap of 15 tons per year starting in 2018. The rule is predicted to achieve a 70% reduction in mercury emissions when fully implemented. The emissions trading or "cap-and-trade" system for controlling mercury and other pollutants is based on the EPA's successful Acid Rain Program. The Acid Rain Program produced reductions in air pollutants faster and at far lower costs than anticipated, achieving more than 50% of the air-pollution reduction attained by the entire EPA with less than one tenth of 1% of agency staff.^[62] The EPA's experience indicates that trading programs, such as the one proposed for mercury, can be cost-effective and health-protective.

Mercury control has also been pursued through legislation. The Bush Administration proposed Clear Skies legislation that would control mercury through emissions trading. The EPA predicted that Clear Skies legislation would lead to a 15% to 60% reduction in mercury deposition in many areas of the United States. Congress has taken no action on Clear Skies legislation to date. Both the regulatory and legislative approaches are intended to limit nitrogen oxides and sulfur dioxide in addition to mercury.

Under the new cap-and-trade approach, the EPA would allocate to each state a mercury budget, or specified amounts of emission "allowances" for mercury. The states decide how to meet the emissions cap (subject to EPA approval), and if a cap-and-trade approach is chosen, those allowances would be allocated to utilities, which would trade them. A utility must hold sufficient allowances to cover its emissions each year, so the limited number of allowances ensures that the required reductions are achieved. The mandatory emissions caps, coupled with significant automatic penalties for noncompliance and stringent emissions monitoring and reporting requirements, are meant to ensure that regulatory goals are achieved and sustained. The flexibility of allowance trading creates financial incentives for utilities to look for new and low-cost ways to reduce emissions and improve the effectiveness of pollution control equipment.

One of the criticisms of the emissions trading approach is that, by using each state as the unit within which trading is made, "hot spot" issues are not addressed, that is, special circumstances involving local mercury deposition from a power plant into nearby water bodies, from which people may catch and eat fish, are ignored. Whether the hot spots criticism is valid is as yet unclear, given the evidence discussed earlier suggesting that connections between local emissions, local deposition, and freshwater fish contamination are poorly understood and appear to be highly site-specific. A recent analysis by the Environmental Law Institute concluded that data from the current major air-pollution emissions trading programs indicate that trading has not created hot spots and, in promoting reductions at the largest power plants, has actually smoothed out pollution emissions instead of concentrating them.^[63] In any case, the Rule is being litigated and its future is uncertain.

Effectiveness

Various proposals to limit mercury emissions from US coal-fired power plants have been made, each of which will reduce emissions. The proposed approaches, timing, cost, and extent of reductions differ, but all would eventually reduce emissions in some way. Whether those reductions in mercury emissions ultimately will have an impact on fish contamination, human exposure, and public health is far less certain, although such uncertainty should not be used as an excuse to avoid reducing emissions cost-effectively.

About 40% of the mercury in the global environment comes from natural sources, such as oceans and volcanoes, and another 40% comes from non-US human activities. US electric utilities that burn coal are estimated to produce about 40% of the mercury emitted in the United States from human activities, but account for only 1% of the global mercury attributable to human sources. Asia accounts for nearly half of the mercury emitted globally from human activities, and China's coal-fired power plants alone represent about 22% of those emissions.^[3,4] Asia's emissions are expected to continue to increase substantially as its need for electricity and its reliance on coal in the absence of environmental controls continue to increase. China alone currently plans to build coal-fired power plants at the rate of 1 per month.^[64]

Meanwhile, limited studies discussed above suggest that the source of methylmercury in tuna and other predatory ocean fish is likely to be primarily natural, not human in origin. Ocean fish account for almost all of most people's exposure to methylmercury. Some studies predict that reducing mercury emissions from US power plants by 50% could reduce methylmercury levels in ocean fish by about 1.5%. On the other hand, preliminary studies suggest that reducing mercury emissions from US power plants may reduce mercury contamination in locally caught fish, depending on a variety of circumstances, which would reduce exposure to people for whom locally caught fish comprise a significant proportion of their diets. A multiscale deposition modeling analysis conducted by the Electric Power Research Institute suggests that any potential local impacts of mercury control would be limited, however.^[65,66] That study characterized mercury deposition both with and without contributions from coal-fired electric utilities and found that for 0.4% of US land area, utilities contribute more than 50% of mercury deposition ("utility-dominated"), whereas for 99.6% of US land area they contribute less than 50% ("non-utility-dominated"). More research is needed to clarify the local impact of mercury emissions.

According to a recent report by the Northeast States for Coordinated Air Use Management (NESCAUM),^[67] a nonprofit association of state air-quality regulators, "Given the global nature of the problem, a significant reduction in US power sector mercury will be insufficient by itself to adequately address mercury contamination of fish and the resulting adverse health impacts." In the absence of international efforts to control mercury emissions, further mercury controls in the United States alone will have only a minor impact on our methylmercury exposure.

© 2006  Medscape

Cite this: Assessing and Managing Methylmercury Risks Associated With Power Plant Mercury Emissions in the United States - Medscape - Mar 09, 2006.