[Editor’s Note: The announcement December 27, 2018, by the United States Environmental Protection Agency (EPA) that it is revising cost findings behind its Mercury and Air Toxics Standards (MATS) ignited a firestorm of complaints on the ground that mercury is a known neurotoxin that can cause brain damage in infants and young people. While that is true, it’s also true that the risk depends greatly on the degree of exposure—and there is compelling evidence that mercury emissions from coal-fired power plants pose no significant risk. In his paper The Cost of Good Intentions: The Ethics and Economics of the War on Conventional Energy, published by the Cornwall Alliance in 2011, Dr. Timothy Terrell, an environmental regulatory economist, included an appendix that quantified that risk, rebutting claims by the Evangelical Environmental Network and others that the risk was large and the MATS rule was therefore well justified. We present that appendix here. Its lessons were timely then, and they are timely again now, demonstrating that the EPA’s action is well justified.—E. Calvin Beisner]

Theoretically, some mercury emitted from various human activities, such as those that require burning fossil fuels, could wind up in human blood through the consumption of fish that have accumulated mercury.^[1] This is the source of the Evangelical Environmental Network’s recent objections to coal-fired power plants. The risks, however, are being overstated, as are the damage estimates.

Defining Risk, Risk Dose, and Population at Risk

Part of the misunderstanding of the hazards of mercury may result from the EPA’s methods of establishing the level of mercury in one’s blood above which people are said to be at a risk level providing reason for concern. This level, called the “reference dose,” is what the EPA has determined to be the highest daily dose that the most sensitive people in the population can be exposed to over a lifetime without adverse effects. Since 1995, the EPA has set a limit of 1 mcg (microgram) per kg (kilogram) per day of body weight.^[2] For concentrations of mercury in blood, this is equivalent to a maximum of 5.8 mcg/L of blood, or approximately 5.8 parts per billion (ppb).^[3] It is important to understand how the EPA arrived at this limit.

The EPA benchmark relies on a long-term study in the Faroe Islands,^[4] which was based on evaluation of methylmercury levels in umbilical cord blood and hair samples taken in childhood. This study indicated some effects from methylmercury exposure, which for one of five tests started at 85 (not 5.8) mcg/L of umbilical cord blood. That is an estimate, and as with any statistical work, there is some uncertainty associated with that number. The actual concentration of methylmercury leading to detectable results for that particular test was probably close to 85 mcg/L, lying somewhere between 58 and 112 mcg/L. The EPA chose the lower limit of 58.^[5] Then the EPA divided by 10 to account for remaining uncertainties, obtaining the reference dose of 5.8 mcg/L. In short, the “reference dose” is not necessarily the same as “the level which causes harm to unborn children.” The EEN’s unequivocal statement that “700,000 babies are born with harmful levels of mercury in their blood” is based on a misinterpretation of the statistics.

There are other problems with the 700,000 figure. The 1999–2000 National Health and Nutrition Examination Study (NHANES) indicated that about 7.8 percent of U.S. women of childbearing age (16–49) had levels of mercury exceeding the EPA’s reference dose (5.8 mcg/L, not the benchmark dose, 85 mcg/L, the lowest at which adverse effects were observed).^[6] Fecundity statistics allow us to estimate the babies born to this group at about 317,000 annually, at that time. This number is approximately doubled to account for the fact that umbilical cord blood has about 1.7 to 1.8 times the mercury concentration that the mother’s blood does. This implies that the limit of 5.8 mcg/L applied to the umbilical cord corresponds to only 3.5 mcg/L in the mother’s blood. The number of women who had levels of mercury in their blood exceeding 3.5 mcg/L is about 15.7 percent of the 16–49 age group, and assuming that this group also accounts for 15.7 percent of the births, this group would account for around 630,000 births each year. This helps us understand where the EPA’s occasional use of a larger number^[7] came from and where the EEN obtained its 1 in 6 (or about 700,000) babies figure.

However, as older women tend to have more mercury in their blood than younger women, and also have fewer children,^[8] fewer than 15.7 percent of the women actually giving birth in the U.S. have mercury levels over the 3.5 mcg/L level (see Table 1). The 700,000 or 1 in 6 figure is, for this reason alone, an overestimate. A quick comparison of the percentage of women in various age groups having mercury concentrations over 3.5 mcg/L according to NHANES and the Census data on births to (approximately) those same age groups indicates that the actual number “at risk” based on these data is at least 100,000 fewer than EEN supposes.

Table 1

NHANES age grouping	BTHg concentration > 3.5 mcg/L, percent (NHANES)	Births (2010 Census)	U.S. Census age grouping
16–19	4.9	301,000	15–19
20–29	12.8	1,930,000	20–29
30–39	15.8	1,323,000	30–39
40–49	22.1	131,000	40–44

Sources: Mahaffey et al. (2004), p. 565, Table 3 and Statistical Abstract of the United States, 2012, p. 70, Table 91.

Other factors, added to this age-bias problem, indicate that the overestimate is considerably larger than 100,000. For instance, it is important to recognize that none of the women in the NHANES study had levels of mercury in their blood that were close to the Faroe Islands study’s lower limit of 58 mcg/L. In fact, those who had more than the EPA’s reference dose of mercury in their blood had around one-eighth the mercury required to reach even that conservative lower limit.^[9] Also, more recent data suggest that even the 7.8 percent figure resulting from the NHANES study may be too high. A large nationwide study by the Centers for Disease Control completed in 2004 found that fewer than 6 percent of women of childbearing age had blood levels of mercury exceeding 5.8 mcg/L.^[10] Robert Ferguson contends that since the CDC’s Agency for Toxic Substances and Disease Register’s (ATSDR) “minimum risk level”^[11] is 13.6 mcg/L, not 5.8 mcg/L, the risk may be lower still. Applying the CDC’s maximum, by Ferguson’s estimation, would mean that only 1 in 100 U.S. women in the NHANES survey, not 1 in 6, were in the range of concern.^[12]

Furthermore, the EPA selection of a “benchmark” did not take into account some studies that show that the effects of mercury on children are undetectable. A twelve-year study conducted in the Seychelles Islands found no indication that methylmercury from a high-fish diet caused neurological damage in children.^[13] The Faroe Islands population contained important differences. Faroe Islanders consume pilot whale meat and blubber, known to have higher levels of methylmercury than fish, but also other contaminants like polychlorinated biphenyls (PCBs) and cadmium that have also been linked to neurological disorders. For PCBs, the exposure in the Faroe Islanders was 600 times higher than the EPA’s reference dose. Some scientists believe that PCBs and methylmercury may work together in causing neuro-degenerative symptoms in the brain, though this is not well-established.^[14] The Seychelles study found no PCBs in its subjects. Perhaps more important, pilot whales do not contain as much selenium as ocean fish—and selenium can reduce the adverse effects of mercury.^[15] Seychelles fish diets were more selenium-rich.

These considerations suggest that the EEN’s figure of 700,000 at-risk babies a year grossly overestimates the actual harm to children from mercury. Even the EPA seems now to be using a much lower figure of “more than 300,000.”^[16] But there are other reasons to be skeptical of the alleged connection between mercury from U.S. power plants and health problems.

The Tenuous Tie between Power Plant Emissions and Mercury Exposure

First, only a minority of the mercury deposited in the United States comes from sources inside the U.S. The EPA itself estimated that only 16 percent of the mercury deposited in the continental United States came from the U.S. and Canada.^[17] Location matters: west of the Mississippi, almost all comes from outside the United States. US power plants contribute less than 1 percent of the global atmospheric mercury,^[18] with a huge fraction of mercury produced from natural sources like deep ocean vents, geologic sources (e.g., Yellowstone National Park produces more natural mercury emissions than all eight of Wyoming’s coal-fired power plants), and forest fires. Forest fires in the U.S. emit roughly the same amount of mercury each year as all U.S. power plants.^[19]

Second, methylmercury concentrations in oceanic fish do not appear to have increased over time, even though global mercury emissions have. A study in Hawaii found that yellowfin tuna had the same methylmercury levels in 1998 as they had almost thirty years before,^[20] even though the mercury levels in the atmosphere nearly tripled over that period of time. Non-human sources of mercury were thought likely to be responsible. A similar study found no difference in mercury concentrations in tuna caught between 1878 and 1909 and tuna caught in 1972.^[21] Another study that looked at concentrations of mercury in striped bass in the San Francisco Bay area over the period from 1970 to 2000 found no clear evidence that mercury levels were increasing, despite a general increase in mercury in the environment over that period of time.^[22]

While the pathway to human consumption of mercury is largely oceanic fish, it seems doubtful that U.S. electric utilities—or man-made sources in general—are contributing significant amounts to the levels of mercury in these fish. This leaves freshwater fish. Freshwater, farm-raised fish have low methylmercury levels, because of how they are fed. That leaves wild freshwater fish (which make up only 10 percent of U.S. fish consumption) as possibly significant sources of methylmercury from power plants.^[23] Local “hot spots” of mercury, such as from coal-fired power plants, could cause elevated mercury levels in these fish.^[24] But the links between local coal-fired power plants and adverse effects on human health remain far from certain. Gail Charnley points out,

Most of the mercury emitted from power plants is elemental mercury or is rapidly degraded to elemental mercury, which tends to remain in the atmosphere and be transported away from the source, entering the global mercury cycle. A much smaller proportion remains in a form that is more likely to be deposited close to the source.^[25]

The EEN’s fact sheet did mention that “[a] study in Florida showed a 60% mercury decrease in fish after 10 years of strict regulations of local sources.”^[26] But not all sites showed a decline, and variations in mercury in local areas are not well understood.^[27] Furthermore, the study pointed out that even a total or near-total elimination of man-made sources of mercury might not be sufficient to bring the mercury concentrations in the fish down enough to meet the regulation.^[28]

Mercury reductions in fish may have even less benefit when one considers the impact selenium content has on methylmercury toxicity. Some research indicates that the vast majority of freshwater fish in the continental United States has sufficient selenium content to protect fish consumers against methylmercury. A 2009 EPA-funded study analyzing 40 species of freshwater fish at 137 sites in the western U.S. found that while 56 percent of the fish had quantities of mercury above what has been considered a “safe” threshold, 97.5 percent of the fish had enough selenium to counteract the effects of the mercury.^[29] All but one of the fish in the 468-fish sample that had an insufficient ratio of selenium to mercury were pikeminnows (also called squawfish), which are commonly considered a “trash” fish and are not normally consumed as food.

Substituting Greater Risk for Lesser?

For those who would prefer to err on the side of caution, it is important to realize that even if fish did contain levels of mercury that could cause some harm, this would not necessarily mean that consuming fish would cause net harm to one’s health. Fish have positive health benefits that could offset any existing danger of mercury some might contain. Fish are a fairly low-cost source of protein, and some fish are also important sources of omega-3 fatty acids, which are important to pregnant women and nursing babies as well as to others in the population. While higher-mercury fish species may not be the same as the species high in omega-3,^[30] many consumers are unlikely to observe the finer distinctions in mercury warnings, and may avoid fish altogether.^[31] John Middaugh, state epidemiologist with the Alaska Division of Public Health, told the FDA that abandonment of subsistence fish diets in Alaska since the FDA warned against fish-borne mercury in 2001 may have resulted in major increases in obesity, diabetes, heart disease, and vitamin deficiencies.^[32] Gary Myers, the lead researcher for the aforementioned Seychelles Islands study, testified to the Senate Environment and Public Works Committee that

We do not believe that there is presently good scientific evidence that moderate fish consumption is harmful to the fetus. However, fish is an important source of protein in many countries and large numbers of mothers around the world rely on fish for proper nutrition. Good maternal nutrition is essential to the baby’s health. Additionally, there is increasing evidence that the nutrients in fish are important for brain development and perhaps for cardiac and brain function in older individuals.^[33]

In attempting to reduce a risk that seems exceedingly small for the vast majority of the population, the mercury alarmists may actually be harming human health by backing into the health risks that result from less fish consumption.

Not only the risks but also the damages are exaggerated. The EEN, citing a 2005 study by Trasande, Landrigan, and Schechter, warns that brain damage from mercury emitted by U.S. power plants causes “around $1.3 billion”^[34] in annual losses. However, a 2007 study by Griffiths, McGartland, and Miller, based on EPA’s assumptions, showed that the earlier study overstated losses by well over 600 percent. The 2007 study shows that Cross-State Air Pollution regulation would reduce actual damage by at most $210 million, or, if borne evenly by the 700,000 babies the EEN claims are affected, $300 per person per lifetime. With 80-year life expectancy, that equals $3.75 per person per year (0.009% of 2010 U.S. per capita income). The later study also found that “U.S. EPA assumptions … decrease the estimated impact of U.S. sources (including power plants) by almost 97%.”^[35]

Finally, to be consistent, mercury exposure from coal-fired power plants should be compared with mercury exposure from other sources commonly regarded as reasonable risks. Consider, for example, compact fluorescent light (CFL) bulbs—the spiral-shaped bulbs that have become the iconic energy-conservation technology promoted by so many environmentalists. While a CFL light bulb manufacturer reminds consumers that a coal-fired power plant produces 13.3 mg of mercury required to use an incandescent bulb, and 3.3 mg for a CFL,^[36] the manufacturer fails to mention that only a tiny fraction of that 13.3 mg would be transformed into organic mercury and pass through a long environmental chain into a human body. In addition, the comparison implies that the energy use reduction would occur solely through lower demand on coal-fired power plants, when in fact slightly over half of our electric power in the U.S. comes from other sources like natural gas, nuclear power, and hydro. The CFLs themselves contain an additional 4–5 mg (on average) of mercury—in close proximity to the occupants of the home—some fraction of which would be released if the bulb breaks. The EPA’s brochure on what to do if a CFL bulb breaks (at http://www.epa.gov/cfl/CFL_brochure.pdf) suggests a level of care consistent with handling hazardous waste. Are mercury emissions acceptable when they occur inside our homes, but unacceptable when they occur at a power plant miles away?

[Editor’s note: For full bibliographical information on sources cited by abbreviations in the endnotes, see the list of References in the original paper from which this article is adapted.]

[1] The mercury hazard with which we are generally concerned here is the organic form of mercury known as methylmercury, which can be formed by microbial interactions with mercury. Burning coal does not produce methylmercury directly.

[2] The EPA’s reference dose for environmental exposure is currently the lowest in the world, at 0.3 micrograms per kilogram of body weight per day (mcg/kg/day). The World Health Organization has set the risk level at 0.5 mcg/kg/day, and Japan—where two major mercury-poisoning events occurred in the 1950s and 1960s—set its level at 0.48 mcg/kg/day.

[3] This is analogous to the first 5.8 seconds out of 31.7 years. 5.8 parts per billion and 5.8 mcg/L in blood are not exactly equivalent, because a liter of blood has a mass of 1060 grams, not 1000 grams (as a liter of water does). 5.8 ppb of blood translates to about 6.1 mcg/L. However, the literature on methlymercury in blood tends to use mcg/L and ppb in blood interchangeably. Compare, e.g., Borum et al. (2001), pp. 4–68 and 4–76. We follow Mahaffey’s (2004), EPA’s, and the 1999–2000 NHANES study’s practice of generally using mcg/L as the operative unit.

[4] Murata et al. (2004).

[5] For the other four tests, the probable lower limits were 46, 79, 103, and 104. The test that yielded 46 ppb was not used by the EPA in establishing the reference dose because of problems administering that test. So even 58 ppb (or mcg/L—see note 33 above) is a very conservative estimate of when problems could start to occur, given the much higher numbers other tests yielded.

[6] Mahaffey, Clickner, and Bodurow (2004).

[7] See, e.g., Mahaffey (2004).

[8] U.S. Census data indicate that in 2010, there were about 1,454,000 births to women 30–44 years of age, and 2,231,000 births to women 15–29 years of age. Data were not reported for women over 44. Statistical Abstract of the United States, 2012, p. 70, Table 91. The NHANES study indicates that the percentage of women with mercury content over 3.5 mcg/L is

[9] FDA (2002), p. 5.

[10] Centers for Disease Control (2004).

[11] See http://www.atsdr.cdc.gov/mrls/pdfs/atsdr_mrls_december_2010.pdf.

[12] Ferguson (2004), p. 11.

[13] Myers et al. (2003a).

[14] Bemis and Seegal (1999), but see Schantz, Widholm, and Rice (2003).

[15] Kaneko and Ralston (2007), Kaneko and Bartram (2009), Ralston and Raymond (2010).

[16] See http://www.epa.gov/mercury/exposure.htm.

[17] The “EPA estimated that 144 tons of mercury was deposited in the continental United States in 2001, and that 121 (or 84%) came from sources outside of the United States and Canada.” Griffiths et al. (2007), p. 844. See EPA (2005), also Charnley (2006).

[18] Szwarc (2004), p. 35.

[19] Soon (2011), p. 1.

[20] Kraepiel, Keller, Chin, Malcolm, and Morel (2003).

[21] Miller et al. (1972).

[22] Greenfield et al. (2005).

[23] Heuss (2003), p. 11.

[24] Hammerschmidt and Fitzgerald (2006), White, Keeler, and Landis (2009).

[25] Charnley (2006).

[26] Evangelical Environmental Network (2011), p. 3.

[27] Charnley (2006) argues that

[t]he potential relationship between power plant mercury emissions and methylmercury concentrations in locally caught fish is complex and poorly understood. Conclusions about the effectiveness of limiting local mercury emissions as a means of reducing local freshwater fish methylmercury levels should be postponed until studies of the impact of current efforts to limit emissions become available.

[28] The study noted that “it may not be practical or even theoretically possible to bring [mercury] concentrations below [the regulatory maximum] on the basis of local anthropogenic [mercury] load reductions alone. … [C]oncentrations in these fish would need to drop by a factor of six to reach [the regulatory maximum]. This decrease is roughly equivalent to the estimated increase in mercury loading to the Everglades over pre-industrial loadings…. Assuming that these inferred increases are from both local and larger scale sources, it is plausible that reductions in [mercury] loading [in compliance with the regulation] may significantly reduce fish mercury concentrations, but still be inadequate to reduce concentrations below [the regulatory maximum], even if all anthropogenic contributions to [mercury] deposition are eliminated.” [emphasis added] Atkeson et al. (2003), appendix II, p. 49; also see p. 68.

[29] Peterson et al. (2009).

[30] Mahaffey (2004), p. 7.

[31] In addition, mislabeling or misbranding of fish means that the consumer cannot be certain of the type of fish being consumed. Mahaffey (2004, p. 568) notes that “less than half the fish labeled as red snapper were red snapper.”

[32] Dept of HHS, FDA, CFSAN, Food Advisory Committee—Methylmercury. Transcript of meeting July 24, 2002, available at http://www.fda.gov/OHRMS/DOCKETS/ac/02/transcripts/3872t2.pdf.

[33] Myers et al. (2003b).

[34] Evangelical Environmental Network (2011), p. 3.

[35] Griffiths et al. (2007a), p. 841. Also see subsequent correspondence: Trasande et al. (2007) and Griffiths et al. (2007b)

[36] See http://www.gelighting.com/na/home_lighting/ask_us/faq_compact.htm#what_is_mercury.