Human exposure to methylmercury (MeHg) occurs primarily through ingestion of seafood and freshwater fish 1. Due to biomagnification in food chains, relatively high MeHg concentrations occur in piscivorous marine species 2 and may exceed the highest recommended limit 3, while smaller non-predatory species, such as herring or sardine, contain concentrations of one-tenth of this limit or even less 3. Methylation, bioaccumulation through food chains, and human intake levels are difficult to model 1. Thus, risk assessment must rely on biomarkers of total human uptakes.

Once absorbed, MeHg acts as a developmental neurotoxicant 4 5 6 7. As the critical effect is considered to be developmental brain toxicity 8 9, MeHg intake by pregnant women is of primary concern 10. In the 1990s, results emerged from three large epidemiologic studies in New Zealand, the Faroe Islands and the Seychelles Islands 6 11 12 13 14. The first two concluded that chronic low-dose prenatal MeHg exposure from maternal consumption of fish was associated with subtle end points of neurotoxicity in children 15. Support for the notion of seafood-mediated MeHg neurotoxicity later emerged also from the Seychelles 16. In further research, Faroes investigators provided extended evidence of a dose–response relationship between prenatal MeHg exposure and lasting neurodevelopmental deficits 15 17 18. Subsequently, epidemiological studies in French Guiana 19 20 21 and in other parts of the world 22 23 24 showed the effects of MeHg on childhood neurodevelopmental disorders. This research has prompted further studies focused on French populations, especially in coastal regions of western France 3 25.

Reducing human exposure to anthropogenic mercury is both a public health priority and an economic challenge, and controversies persist in both research interpretation and policy decisions 10. The consequences of MeHg contamination, similar to those observed for lead (Pb) exposure in children, include a loss in Intelligence Quotient (IQ), with associated lower school performance and educational attainment, thus leading to long-term impacts on societal benefits of pollution abatement 26 27. The economic impacts caused by MeHg on humans have been assessed in the United States, through the studies by Rice et al. 28 29 and publications from the US Environmental Protection Agency (EPA) 30, and the study of Griffiths et al. 31. Although these calculations have been extrapolated to global estimates 32 33, few economic evaluations have been performed in Europe 34 35. To extend and update the international assessments of economic impacts of MeHg exposure and to utilize biomonitoring data, the present study aims to assess the economic consequences of MeHg-associated neurotoxicity, using exposure data from French studies. As biomonitoring results become available from other countries, additional national estimates can be made using the same methodology.

The economic assessment requires toxicological and epidemiological assumptions. Hg concentrations in hair and in umbilical cord blood have been used as biomarkers to ascertain prenatal exposure to MeHg, although both exposure indicators are associated with some imprecision 36. In selecting the dose–response function (DRF), a major difficulty arises when deciding on the MeHg dose metric 15. While a linear model is attractive, it does not provide the best statistical fit to the data 18. Studies that used a log transformed exposure scale assume that each doubling of exposure causes the same deficit. In the absence of a clear threshold, an additional decision has to be made in regard to a toxicological reference value, so that the epidemiological findings are translated into a “cut-off point”, below which only negligible adverse effects exist. Using both a linear and a logarithmic dose–response curve (DRC), we provide estimates of the economic consequences of prenatal MeHg exposure for three different such cut-off points as a basis for development of public policies to prevent MeHg exposure at national and international levels.

As shown in Table 1, the three study populations exhibit different exposure distributions. While the exposure levels are the lowest in the national sample, the distribution is shifted towards higher values in Brittany and, even more so, in Loire Atlantique. Associations between IQ losses and HHg exposures assuming linear and logarithmic relationships are reported for the different distributions and scenarios in Figure 1.

The Table 1 presents the number of children from the 2008 birth cohort exposed to different levels of MeHg based on HHg concentrations in three French population samples. We noted that 58.95% of women of childbearing age corresponded to the national sample, 89% to the Brittany and 84% to the LA sample, respectively.

Table 2 and Table 3 present the IQ losses and the estimates of the economic impact for the linear model, expressed as benefits associated with Hg exposures above the three cut-off points for the 2008 birth cohort assuming exposure distributions based on the three study samples. Due to the differences in exposure, the greatest benefits would be achieved with the Brittany sample, should all values be reduced below 0.58 μg/g. The estimated median annual total benefits were € 1.62 billion, € 0.77 billion, and € 0.005 billion using the national sample, according to the three cut-off points (0.58, 1.0, and 2.5 μg/g). The corresponding values were € 3.02 billion, € 1.75 billion € and € 0.02 billion using the Brittany data, and € 2.51 billion, € 1.31 billion and € 0.07 billion from the Loire-Atlantique exposure distribution. If we had used a rounded cut-off level of exposure at 0.5 μg/g , the total benefits estimated for the national sample would have been about 17% higher than those estimated to be above the cut-off level of 0.58 μg/g.

The Table 2 presents the IQ losses, for the linear model, the upper bound value per segment from the three cut-off points (0.58, 1.0 and 2.5 μg/g) for the 2008 birth cohort assuming exposure distributions based on the three study samples.

Table 3 presents the estimates of the economic impact for the linear model, expressed as individual benefits (B_i), benefits per range (B_[range]) and the total benefits for one year (TB), associated with Hg exposures from the three cut-off points for the 2008 birth cohort assuming exposure distributions based on the three study samples.

Table 4 presents the benefits associated with different levels of Hg exposure reductions by using the logarithm model. For all three samples, we used the same number of children per range for the both models (Table 1). The estimated total benefits were € 5.46 billion, € 9.13 billion, and € 8.17 billion based on the national, Brittany and Loire-Atlantique exposure distributions, respectively, above the 0.58 μg/g cut-off. Thus, the total benefits were estimated to be more than three times higher than those obtained from the linear model (Figure 2).

Table 4 presents the estimates of the economic impact for the logarithm model, expressed as individual benefits (B_i), benefits per range (B_[range]) and the total benefits for one year (TB), associated with Hg exposures from 0.58 μg/g for the 2008 birth cohort assuming exposure distributions based on the three study samples. Losses of IQ are 0.5, 1.5 and 3.5 IQ points for [0.58; 1.16], [1.16; 2.90] and [2.90; Max], respectively.

The aim of this article was to evaluate the economic impacts of neurotoxicity associated with prenatal MeHg exposure in France. Our estimations were carried out assuming a linear and a logarithmic relationship between Hg exposure and IQ losses 6 18, for three hypothetical cut-off points, 0.58 μg/g, 1.0 μg/g and 2.5 μg/g, respectively, based on three different evaluations 1, 36 and 39. In agreement with European 43 recommendations, we do not pretend that any of the cut-off points are completely safe exposure levels, but merely represent hypothetical exposure levels below which adverse effects might be negligible. Three French data sets, i.e., national, Brittany and Loire Atlantique, enabled a characterization of the Hg exposure distribution from exposure biomarker results in women of childbearing age or pregnant women. Data were applied to the 2008 national birth cohort. We calculated the health impact (loss of IQ/individual) and the annual economic impact or benefits of Hg reduction in terms of personal avoided costs and for the entire cohort (Bi and TB, respectively), above each cut-off point.

These results highlight that prenatal MeHg exposure has serious impacts on the life-time productivity and on society due to adverse cognitive and associated economic consequences. Benefits were higher with the logarithmic than the linear model (see Figure 2), as the logarithmic DRF is steeper at low exposure levels that affect a larger proportion of children. The two regional exposure estimates, close to sea coasts, where marine food is more easily available, are higher than the national ones. This finding is in line with the French national statistics for fish consumption, which show that the population residing on the western coast of France (18% of the total population) consumes more of the total amount of fish available in the whole country for numerous species 38 44.

The results show that policies that aim to reduce childhood MeHg exposure would have large-scale social benefits. The focus on a child’s life-time earning loss is similar to the avoidable costs in relation to lead exposure reduction 26. Other costs were ignored, such as direct medical costs linked to treatment or interventions for children with neurodevelopmental disorders. We also neglected indirect costs, such as those related to special education or additional years of schooling for children as a consequence of these disorders, as well as intangible costs. In addition, our study did not consider other avoided direct health care costs, such as those potentially related to the treatment of cardiovascular or neurodegenerative effects of MeHg exposure, which could be important for high fish consumers 5, but would be difficult to estimate.

Several uncertainties hamper accurate impact estimation. One concerns the slope of the dose–response relationship between maternal MeHg exposure and loss of IQ points. Our first estimates used an average mean of 0.465 IQ point loss per μg/g in HHg. The lower boundary (0.295 IQ point loss per μg/g) or the upper one (0.62 IQ point loss per μg/g) can be also used to obtain minimal or conservative estimates. We used both the linear and the log DRF, as the latter scale showed a better fit, suggesting that a doubling of exposure is associated with a loss of 1.5 IQ points 41. This slope is of course also uncertain.

Three large-scale prospective epidemiologic studies investigated children who experienced MeHg exposures in utero at concentrations relevant: the Faroes study 6, the New Zealand study 11 12, and the Seychelles study 13 14. These studies provided evidence of a dose–response relationship between concentrations of MeHg and neuro-developmental disorders among children 9. As also decided by the NRC 15, we relied on the Faroes study as the most extensive data base that was only minimally affected by confounding. This way, we avoided considerations of residual confounding from seafood benefits. Also, in regard to the three cut-off points, recent research 22 24 reflects the occurrence of adverse effects close to the lowest cut-off level. However, it is unclear when effects become negligible, and this uncertainty in particular affects the estimates based on the log curve.

While the use of biomonitoring data from random samples of the general population is an advantage, our calculations were based on sample sizes ranging between 126 and 503 at national and regional levels. The representativeness of the study samples can be challenged, and larger surveys are needed to obtain more precise data, especially in regard to the prevalence of high-level exposures.

Comparison of mean mercury exposure distributions in different countries indicates that, France exhibits Hg concentrations that are greater than in Germany and the US. The consumption of fish in the latter countries is about half that of the French, while countries such as Spain, Sweden and Japan show greater consumption figures 37. Our results, in line with US findings 33 34, document that a reduction of childhood MeHg exposure may have substantial social benefits. The monetary value of the annual health benefits due to prevention of cognitive disorders generated by a 20% reduction exposure to MeHg in the US population was estimated at $_US 170 million/year 29, but this result was certainly underestimated due to underestimation of the hair mercury-IQ dose response slope and the high threshold for neurotoxicity. Using a different approach, Trasande et al. calculated that decreased economic productivity resulting from diminished intelligence over a lifetime results in an aggregate economic cost in each annual birth cohort of $_US 8.7 billion annually (range: $0.7–$13.9 billion, $₂₀₀₀) 40. About 15% of this cost was said to be attributable to mercury emitted from coal-fired power plants in the United States 40, although the basis for these calculations may be challenged. In comparing these figures to the ones calculated for France, note should be taken that annual US birth cohorts are about five times greater. Despite the differences in assumptions and published estimates, the results document that benefits of MeHg exposure control are substantial.

Our paper did not estimate the annual costs of investments in pollution abatement because of the paucity of the available data. The known investment costs for Hg emissions control include data from reduction of mercury usage in the chlorine industry (estimation of €₂₀₀₅ 0.4 billion), measures taken in dentistry (€₁₉₉₇ 0.031 billion), plus expenses for recycling and treatment of mercury releases. These French expenses are total, not annualized. While the utility industry is responsible for a main part of global mercury emissions, its contribution and the costs for abatement vary substantially between countries. In the US, an estimate of $_US750 million per year has been reported for industrial investments needed to obtain a reduction of Hg emissions 35. However, abatement efforts should not be undertaken at a national level alone, and calculations need to consider global expenses. Thus, due to regional and hemispherical air dispersion of Hg, a concerted European policy on the emissions is necessary to obtain significant reductions of exposure levels in Europe. The French Institute for Industrial Risks (INERIS) has put forward two main routes for the reduction of mercury releases in the environment: one is substitution of mercury by non-fossils fuels (wood, biogas, biomass) in oil or coal combustion plants and by process changes in the chlorine industry (change to the membrane cell technology); the second would encourage more effective ways to collect and recycle waste containing mercury in batteries, thermometers, dental amalgams, and energy-saving lamps 45. These costs would have additional socio-economic yields from better control of mercury emissions: job creation and modernization of capital equipment 46.

Methylmercury exposure mainly originates from fish and seafood, which contain essential nutrients that have beneficial effects on brain development 8, 47. For this reason, a reduction in the concentrations of MeHg in fish is a desirable long-term goal rather than a replacement of fish in the diet by other foods. In the interim, the best method of maintaining fish consumption and minimizing Hg exposure is the consumption of fish known to have lower MeHg concentrations 15 and advisories to women about avoiding Hg intake during pregnancy and breastfeeding may be a cost-effective preventive action.

French studies and recommendations from the French Agency for Food, environmental and Occupational Health Safety (ANSES), stress the need for health education regarding fish species consumption in order to protect vulnerable populations. The INRA study provided evidence on the risks and the benefits of fish consumption by pregnant women to guide decision making in order to reduce risks and optimize nutritional benefits in consumers 3 44. Thus, implementation of consumption strategies within populations at risk can be in the form of pertinent dietary recommendations 48.

Other factors also play a role in regard to fish consumption. The Sustainable Development indicators show that, despite the implementation of quotas, overfishing continues: 13% of so-called pelagic fish catches (including tuna) are in excess to the precautionary threshold 49 and may for this reason need to be diminished. In this connection, economic aspects of the fishing industry can also be considered. While tuna is high in MeHg content, it is the most popular of seafood species consumed in France (220,000 tons/y between 2008 and 2010) [56, 57], mainly canned (about 94% of the value of household purchases of tuna), and represents 8-9% of household expenditures for fish purchases, i.e. € 0.56 billion in 2010. For comparison, sardines are cheaper and with low MeHg but consumed less frequently (63,000 tons/y), with household expenditures corresponding to € 0.16 billion in 2010. Despite these high figures, the economic importance of high-mercury species is lower than the benefits calculated in the present study, thus emphasizing the need for abatement.

More extensive human biomonitoring would allow a more precise measurement of exposure and would help elaborate recommendations and information to reduce environmental exposures to MeHg 50. Such studies need to be extended to all of the EU and beyond. However, information alone would not suffice to change dietary habits and taxes and subsidies would be necessary to encourage consumption changes 51. Our results suggest that the benefits of exposure control justify such actions.

Annual benefits of removing Hg exposure can be estimated in the order between € 1 billion and € 9 billion in France. While our results support enhanced public policies for the prevention of MeHg exposure, the economic estimates are highly influenced by uncertainties regarding the dose–response relationship. Benefits might be underestimated because costs linked to all aspects of neurotoxicity and to cardiovascular diseases have not been considered. The data from France support the notion that precautionary measures are called for to minimize exposure to this hazardous pollutant.

ANSES: French Agency for Food, environmental and Occupational Health Safety; HHg: hair mercury; DRF: Dose–response Function; ENNS: Etude Nationale Nutrition Santé; Hg: Mercury; MeHg: Methylmercury; INERIS: French Institute for Industrial Risks; INRA, National Institute of Agricultural Research; INSERM: National Institute of Health and Medical Research; InVS: French Institute for Public Health Surveillance; IQ: Intelligence Quotient; JECFA: Joint FAO/WHO Expert Committee on Food Additives; NRC: U.S. National Research Council; PÉLAGIE: Longitudinal study on pathologies of pregnancy, infertility and childhood; Pb: Lead; RfD: Reference Dose.

The authors declare that they have no competing interests.

CP performed the literature review, drafted the manuscript and carried out the analysis. MB, PG, DZN, NF, SC, ARG and PH contributed substantially to defining the methods of the analysis, interpreting the results of the study and editing the manuscript. All authors read and approved the final version.