Original papers

Environ Engg and Policy 1 (1998) 3–9 Q Springer-Verlag 1998

Risk/benefit analysis of prohibition of the mercury electrode process in caustic soda production Junko Nakanishi, Tosihiro Oka, Masashi Gamo

3 Abstract This study attempts a risk-benefit analysis of the Japanese government decision of the prohibition of the mercury electrode process used for production of caustic soda in 1973. The reduction in human health risk and the costs incurred due to the decision are evaluated, on the basis of which an implicit value of its benefit-risk (BRR) ratio is estimated. Human health risk is evaluated in terms of loss of life expectancy, which enables us to quantify and compare human health risks, irrespective of cancer or noncancer toxic risks. The value of BRR of the decision is calculated to be 570 million yen, approximately U.S. $5.7 million, per life-year saved. Comparing the BRR values of a couple of other policies enacted or proposed in Japan, ranging from 24 to 45 million yen per life-year saved, its value is very high. This result implies that the decision in study is not cost-effective.

well as those put into practice in the past. In addition, accumulated estimations of the implicit BRR values of past policy decisions may be useful for making future policy decisions. In conducting risk-benefit analyses, it is critical to define endpoints for the evaluation of risk. In our studies, human death is chosen as the endpoint for human health risks; loss of life expectancy (LLE) is used as a measure for evaluating human health risks, using the human mortality endpoint. The LLE enables us to quantify and compare health risks irrespective of the carcinogenic or noncarcinogenic nature of toxic risks [Gamo et al. (1995)]. This paper provides the results of a risk-benefit analysis of the government decision to prohibit the mercury electrode process in caustic soda production in 1973. In the study, the reduction in human health risk due to the prohibition of the mercury process and the costs incurred by the decision are evaluated. 1 Since the mid 1960’s, the Japanese government has inIntroduction In making policy decisions on environmental issues, risk- troduced a succession of strict measures to reduce the benefit analysis is an important emerging tool. Risk-ben- toxic environmental effects of mercury following the ocefit analysis is a variation of cost-benefit analysis that ex- currence of two outbreaks of Minamata Disease (MD) amines trade-offs between benefits and risks. Using this due to industrial effluents from the acetaldehyde productechnique, policy makers can judge whether a policy is tion processes carried out in Minamata (Kumamoto Precost-effective by comparing the value of its benefit-risk fecture) and in Niigata. In the beginning, these measures ratio (BRR) with those of other policies. Such other poli- focused on the regulation of organic mercury, because cies may include alternatives to the policy in question as organic mercury was believed to be dangerous, while inorganic mercury was thought to be not so dangerous. However, the situation drastically changed in 1973 followReceived: 29 Oktober 1997 ing reports of the possible occurrence of a third and fourth outbreak of MD in areas where the caustic soda J. Nakanishi production process was the only source of mercury. The Institute of Environmental Science and Technology, Yokohama reports led to panic reactions such as consumers’ nationNational University, 79–7 Tokiwadai, Hodogaya-ku, Yokohama wide boycott of fish or fear of mercury poisoning. In the 240–8501, Japan same year, the government enacted the policy that the mercury electrode process used in caustic soda producT. Oka tion should be replaced by nonmercury processes by the Fukui Prefectural University and the Japan Science and end of fiscal 1977. At that time, 95% of caustic soda in Technology Corporation Japan was produced using the mercury electrode process. M. Gamo Finally, by September 1986, the caustic soda plants using National Institute for Resources and Environment and the mercury electrode process were completely replaced by Japan Science and Technology Corporation nonmercury plants. In the meantime, some plants changed their production process twice, from the mercuCorrespondence to: J. Nakanishi ry process to a diaphragm process, and then from the diaphragm to an ion-exchange membrane process, due to This work has been supported by CREST (Core Research for the poor quality of caustic soda produced by the diaphEvolutional Science and Technology) of the Japan Science and ragm process. Technology Corporation (JST). Authors are very grateful to The occurrence of the fourth outbreak of Minamata Dr. S. Nakai for fruitful discussion regarding epidemiological studies. Disease was suspected to originate from the Tokuyama

Environ Engg and Policy 1 (1998)

4

district facing Tokuyama Bay located at the northwest of the Seto Inland Sea. There were two factories with caustic soda production plants in Tokuyama which used the mercury electrode process to produce about one-tenth of the total caustic soda production in Japan and discharged their effluents into the bay. According to a large-scale survey of mercury pollution, the mercury level in eatable portions of fish for five species caught from Tokuyama Bay was over 0.4 mg/kg of their total weight, the criterion of mercury for edible fish, and sediment of the bay was also contaminated with mercury in large areas. In addition, the fact that more than 80% of the mercury found in the fish was methyl mercury, and the factories were discharging inorganic mercury, surprised the academic community and terrified the public. 1 Due to the high mercury levels, fishing in Tokuyama Bay was closed from 1976 to 1983, though it was partially reopened in 1979. Meanwhile, dredging of the sediment of the bay contaminated with mercury was carried out. The epidemiological survey of 618 residents in the Tokuyama district conducted by the Yamaguchi Prefectural Office and Medical Department of Yamaguchi University demonstrated that residents with a mercury content in their hair over 10 mg/kg constituted 4.2% of the 618 residents examined, and those with mercury content over 30 mg/kg constituted 0.6%. In addition, relationships between fish consumption and mercury content in hair, and between fish consumption and age-adjusted incidence of paresthesia, were observed [Nose et al. (1975)], [Nakanishi, Ukita (1989)]. However, the government officially denied the occurrence of MD there, because in those days the government strongly advocated that only patients contracting not only paresthesia but also constriction of the visual field and ataxia should be recognized as MD. It is known that as methylmercury exposure levels increase, first paresthesia followed by more severe symptoms such as ataxia and the constriction of the visual field occur. However, the government gradually changed the definition of MD. The incident at Tokuyama Bay was a typical case of inorganic mercury being discharged and then transformed into methylmercury in the environment, which then accumulated in fish and the human body. Hence, this is different from Minamata, Kumamoto Prefecture, where methylmercury was discharged directly from industry into the environment. The evidence that inorganic mercury is transformed into methylmercury by chemical and biological reactions in the environment has been accumulated. However, the methylmercury levels in fish measured are much greater than those predicted on the basis of reaction rates of a broad spectrum of chemical and biological methylation reactions. (Nakamura, 1994).

2 Risk estimation 2.1 Assumptions regarding caustic soda (NaOH) production In this section, incremental human health risk that a continuance in use of the mercury electrode method would pose is estimated. In estimating the incremental risk, the following assumptions are made. A total of four million tons of caustic soda per year is produced in chloralkali plants located in ten areas throughout Japan. Each plant is roughly identical to the others with respect to its production capacity and process employed. In other words, four hundred thousand tonnes of caustic soda are produced annually in each area. All mercury discharged in one area into the environment ultimately flows into a nearby bay. A total of ten bays throughout Japan are considered. The geographical and biological conditions of each of the ten bays is represented by that of Tokuyama Bay in this study. Six grams of mercury are consumed (lost) per ton of caustic soda produced. Twenty percent of the mercury consumed is discharged into the environment, and thus 0.48 tons of mercury are discharged into each bay annually. Most of the 80% of mercury that is consumed is retained in process plants, and some is retained in wastes such as brine mud. Assumptions made for this study are summarized in Table 1, along with the data from Tokuyama Bay in the latter 1960’s [Nakanishi et al. (1989)].

2.2 Estimation of mercury level in sediment from the bay A weighted average mercury level in the sediment (C mg/ kg of sediment, dry basis) is estimated using (1): 63.9

Cp1/63.9

#

0

63.9

CdSp1/63.9

#

aDS P0.25dS,

(1)

0

where a is a constant specific for Tokuyama Bay, which is determined by its geometric or hydrological conditions, S is the contaminated area in km 2 with a mercury level more than C (mg/kg), and D is the amount of mercury discharged by the chloralkali plants. (1) is derived from the relationship between the level of heavy metals in sediment and the amount of the heavy metals discharged into the bay, which Nishimura and Kumagai (1974) formulated on the basis of results obtained from the research on heavy metal pollution in several bays. The value of a is determined to be 2.34 using the data of mercury levels in the sediment from Tokuyama Bay and the amount of mercury discharged into the bay in the latter 1960’s. Given that the total area of Tokuyama Bay is 63.9 km 2 and that D is 0.48 tons, an estimated weighted average mercury sediment level at 0.53 mg/kg is obtained as shown in Table 1.

2.3 Estimation of mercury levels in fish 1 It is often important to distinguish organic mercury from inorganic mercury. However, in this paper, ‘mercury’ is used to mean “total mercury”, which indicates the sum of inorganic mercury and organic mercury.

The mercury level in fish is estimated using (2) derived from the results reported by Nakanishi et al. (1989) regarding the relationship between mercury level in fish and the sediment from Tokuyama Bay.

J. Nakanishi et al.: Analysis of prohibition of the mercury electrode

Table 1. Mercury data

Table 2. Daily intake of Methylmercury

Assumptions in this study nation NaOH production 4000 (1000 tonnes) Hg discharged 4.8 (tonnes/year) Hg in sediment (mg/kg) Hg in gilthead (mg/kg) Hg in sardine (mg/kg)

Tokuyama Bay in the 1960’

Group

Consumption of fish from the bay (g/day)

Background MeHg intake (mg/day)

Increment of MeHg intake (mg/day)

400

320 97 (mean) 0

32 9.7 9.7

8.7 2 0

one bay baseline 400 0.48

0

3.62

1 2 3

0.53

0.03

4

Mettgpmethylmercury

0.24

0.2

0.6

5

1.5 [Shirai (1988)]. In addition, it is assumed that people 0.06 0.05 0.13 consume an equal amount of giltheads and sardines and that 80% of the mercury in fish is methylmercury. The daily intake of methylmercury is calculated using mercury levels in fish from the bay of 0.24 mg/kg for gilthead and 0.06 mg/kg for sardine, while background daily inFor gilthead (Kurodai in Japanese) yp0.2 c0.08 C, For sardine yp0.05c0.02 C, (2) take of methylmercury is calculated using mercury levels of 0.2 mg/kg for gilthead and 0.05 mg/kg for sardine. where C and y are the mercury level in the sediment Background methylmercury intake and increment of me(mg/kg, dry weight basis) and in the fish (mg/kg, wet thylmercury intake due to mercury pollution are shown weight basis), respectively. Gilthead is representative of in Table 2. the fish species with a high mercury level, and sardine is representative of the fish species with a low mercury lev2.5 el. The calculated averages of mercury level are 0.24 mg/ Probability of methylmercury poisoning kg for gilthead and 0.06 mg/kg for sardine, when the avNext, to represent numerically the human health effects erage mercury level in sediment is 0.53 mg/kg. These recaused by exposure to methylmercury, the relationship sults are shown in Table 1. between the methylmercury daily intake and the probability of paresthesia reported by Nordberg and Strangert (1976) is applied to different groups of the total popula2.4 tion. The equation by Nordberg and Strangert was obMethylmercury intake by residents The residents of the study area are classified into three tained by assuming individual variability with regard to groups in terms of their fish consumption habits. The two parameters: the log-normal distribution of the first group are heavy fish eaters, while the general popu- threshold body burden for symptoms of paresthesia, and lation is divided into the second group of residents, those the normal distribution of half-life of methylmercury in consuming only fish caught in the bay contaminated with the human body. The possible increase in cases of paresmercury, and the third group of residents, those not con- thesia due to increased concentrations of mercury in fish suming any fish from the bay. The number of the first is calculated. The figures thus calculated are regarded as group of 300 in each area is determined on the basis of incidence of paresthesia per year. This method of estimaabout 10% of the population of families in fish-related tion is quite conservative and is discussed in more detail business such as fishermen, fish mongers, and sushiin section 5. The most conservative estimate of the procooks in the Tokuyama district. It is assumed that people jected increase in cases of paresthesia due to mercury rein this group consume 320 grams of fish caught from the lease through caustic soda production nationwide is bay per day [Futatsuka (1979)]. The number of the sec40.6 cases per year, as shown in Table 3. ond group is determined by amount of fish caught in Tokuyama Bay and it is assumed to be 133 thousand in 3 each area. Furthermore, it is assumed regarding general Estimation of LLE population that individual variability regarding fish con- In this section, the amount of loss of life expectancy sumption follows a log-normal distribution, with the geo- (LLE) that would have been avoided by the prohibition metric mean of 97 grams per day [Ministry of Health and of the mercury electrode process is estimated. To calculate the LLE, we assume that contracting paresthesia at a Welfare (1997)] and the geometric standard deviation of

Table 3. Human health risk in Japan (the most conservative estimates)

Population (thousand) Probability of Paresthesia Paresthesia Risk (cases/year) LLE (life-year)

Group 1

Group 2

Total

3 6.40!10 P3 1.92 3.6

1330 2.91!10 P5 38.7 71.6

1333 40.6 75.2

Environ Engg and Policy 1 (1998)

6

certain age would increase the annual death rate from that of the control population at that age by 27.17% for male and 19.71% for female uniformly irrespective of age. This is based on the standardized mortality ratio (SMR) values for MD patients estimated in the epidemiological study by Kinjo et al. (1991). In this study, the SMR values are used to estimate the increase in death ratio due to MD for all age classes, because the data concerning the increase in death rate by age are not available to us. Kinjo et al. examined the mortalities of MD patients by year, age, sex, and cause of death, and compared them with those of the other people living in the same area. Significant increases were observed for total deaths, and SMRs of MD patients for total deaths were estimated to be 1.27 for male and 1.20 for female. The LLE due to increases in death rates is calculated by the following procedure. Let d(t) represent the death rate at the age of t, the survival rate until the age of t, s(t), has to meet the equations,

4 Evaluation of benefits 4.1 Estimation of expenditures

In this section, the reduction in benefit occurring from prohibition of the mercury electrode process is evaluated. If caustic soda produced by the mercury electrode process and that produced by nonmercury processes are identical in terms of quality of the products and the running costs of the process, the reduction in benefit occurring from the prohibition is represented by the costs incurred from the replacement of plant processes from the mercury electrode process to nonmercury processes. Here, the differences in quality and running costs between the three processes, the mercury electrode, the diaphragm, and the ion-exchange membrane, are regarded to be negligible. According to a survey by the Japan Soda Industry Association, the total expenditures incurred in replacing the s(0)p1, mercury electrode process with nonmercury processes s(t)p[1–d(t–1)]s (t–1), for tp1,2,..., T (3) amounted to 334.2 billion yen from fiscal 1973 through where T represents the maximum age of survival (we as- fiscal 1988. Of this, 287.3 billion yen was spent to change sumed Tp109 for male and Tp111 for female). The life the mercury electrode process to the diaphragm process expectancy at the age of k is equal to L(k), which meets: or the ion-exchange membrane process, and 46.7 billion yen was spent for the second replacement, from the T–1 1 diaphragm process to the ion-exchange membrane procS(t)cS(tc1) 2 ipk (4) ess [Japan Soda Industry Association (1982)]. L(k)p (kp0,1,..., T–1). The yearly expenditures for replacement are estimated S(k) based on data published by the Japan Soda Industry AsIf the above assumption regarding the increase of sociation; the expenditures aggregated for several years death rates due to the development of paresthesia is giv- are equally distributed for each year. The expenditures en, the death rates from the development of paresthesia thus estimated for each year from 1973 to 1988 are adat the age of k can be expressed as m d(t),( tpk,kc1,...,T), justed in terms of 1989 yen value and are shown in cowhere m is the SMR value. The survival rates after the lumn B in Table 4. Annual worth of the stream of the exdevelopment are calculated by substituting m d(t) for m penditures from 1973 to 1988 is calculated to be 19.8 bild(t–1) in (3). Using these survival rates, we can calculate lion yen, assuming that social discount rate is 5% and the the life expectancy after paresthesia development from duration of the effects of the whole stream of the expen(4). Let L(t) denote the life expectancy at the age of t ditures is 33 years. This is consistent with the assumption with paresthesia development at the age of t, and let p(t) about life-span of plants given below. denote the proportion of the population of age t in the whole population, the LLE from one case of paresthesia 4.2 is equal to: Estimation of cumulative reduction in mercury discharged T First, the amount of caustic soda produced using the new A p(t)[L(t)–Lb(t)]. processes that replaced the mercury electrode process is tp0 estimated for each year. Next, the reduction in mercury discharged into the environment resulting from the procWe used the above values of mortality due to paresthesia, the population by age in 1990, and the life-table of ess change for each year is estimated assuming a reducJapan in 1990 to obtain the value of 2.24 years as LLE for tion of 1.2 grams for every one tonne of caustic soda promen and the value of 1.49 years for women. The average duced by the new processes. The figures for the consecutive years from 1973 to 1986 are shown in column C in of these two values weighted by the populations of both Table 4. Moreover, it is assumed that the reduction in sexes is 1.85 years. This means that LLE is induced by mercury discharged brought about by replacement of the one case of paresthesia. If we assume the estimation of mercury process is valid for twenty years, the life-span of 40.6 cases of paresthesia avoided by the prohibition of the mercury process per year, the avoided LLE would be the newly built caustic soda production plants. Then, the sum of the cumulative reductions in mercury discharged 75.1 years per annum. for the respective years is calculated and shown in column D. Regarding the quality of the environment as a kind of resource, the amount of mercury discharged is treated as if it were an expenditure. That is to say, the annual worth of the sum of the valid reduction in mercu-

A 3

4

J. Nakanishi et al.: Analysis of prohibition of the mercury electrode

exposed to a given amount of mercury for one year would develop paresthesia with the probability expected by Nordberg’s formula, which were formulated assuming the long-term exposure to methylmercury; (2) all residents in the polluted districts are replaced every year, and the same number of residents would develop paresthesia every year as in the first year. The assumption (1) is not wrong in principle, considering that the half-life of methylmercury in the human body is less than half a year. The assumption (2), however, may differ from reality, in 4.3 that almost the same residents continue to live in the Estimation of the BRR value As shown in Table 1, the replacement of the mercury same areas for many years. If the ratio of migration of electrode process by the nonmercury processes brought the residents is 10%, the incidence of paresthesia per year about 4.8 tonnes of cumulative reduction in mercury dis- is estimated to be about a tenth of those calculated by charged per year. For this purpose, an expenditure of making the two assumptions stated above. This ambigui43.0 billion yen, 8950 yen per gram multiplied by ty stems from lack of knowledge of the relationship be4.8 tonnes, was incurred. And the reduction in human tween the severity of paresthesia and the duration of exhealth risk of 75.1 life-years was brought about. In conposure to methylmercury; this lack of knowledge stems clusion, the BRR value of the prohibition of the mercury not only from insufficient research but also from the electrode process for the most conservative risk estimacomplexity of the mechanisms of noncarcinogenic toxic tion scenario is about 570 million yen per life-year, apeffects. proximately U.S. $5.7 million per life-year saved, which is We believe and assume that a no-effect threshold can calculated by 43.0 billion yen divided by 75.1 life-year. be determined for noncarcinogenic toxic effects. Therefore, the probability of a noncarcinogenic disease just reflects individual variability in threshold, and in other 4.4 words, reflects a portion of the population who are sensiEvaluation of the policy Let us compare the BRR value of the policy in this study tive to toxic substances. Noncarcinogenic toxic effects are with those of other policies. Previously, we estimated the not stochastic, while carcinogenic effects are controlled BRR values for two cancer death reducing policies and by easily understood stochastic mechanisms. In evaluatone pesticide regulation. The former consists of the poli- ing population risk, it is important to take into account cy of replacement of chlorination by ozonation in the individual variability. We consider three parameters in process of water purification [Nakanishi (1989)] and the individual variability: fish consumption, threshold level of policy of reduction in benzene content in gasoline from paresthesia, and half-life of methyl mercury in the human the current level to 1% in volume [Nakanishi (1995)], body. and the pesticide case concerns replacement of chlordane To examine the validity of the paresthesia developby chlorpyrifos [Oka et al. (1997)]. The estimated BRR ment evaluation method used in this study, we estimate ranges from 300 million yen to 570 million yen per cancer the magnitude of risk in the latter 1960’s based on the death reduced. According to the results by Gamo et al. data shown in Table 1 and compare the estimates with (1996), that the magnitude of risk of a cancer death is the observed facts. The most conservative estimation inidentical to that of the risk of 12.6 years in terms of LLE, dicates that about 149 residents will develop paresthesia the BRR values are calculated to be between 24 and per year in the Tokuyama district. On the other hand, ac45 million yen per life-year saved for the above three pol- cording to results of medical examination of 732 residents icies. in the polluted areas of the Tokuyama district, 48 resiTengs et al. (1995) analyzed the cost per year of life dents developed tactile paresthesia and 417 residents desaved for 587 life-saving interventions. According to their veloped the subjective symptom of numbness [Nose et al. results, it ranges from saving (the minus cost) to more (1975)]. Considering the number of residents with numbthan U.S. $10 billion per life-year saved. Among 587 life- ness in a control area, 67 residents with numbness are atsaving interventions, the median cost per life-year saved tributed to methylmercury pollution. If we assume that for toxin control interventions in the area of environthe incidence of tactile paresthesia in the population of mental protection issues is U.S. $4.2 million. The BRR the control area is zero (though information concerning value with the policy in this study is much higher than the incidence of tactile paresthesia in the control area is the three policies formerly studied by us, and even a little not available), patients with paresthesia consisting of higher than the median values for cases in the United numbness and tactile paresthesia would number about States studied by Tengs et al. 100. Taking into account the discussion at the head of this section, we consider the estimates to fit the facts. There are two problems in using the SMR values of 5 the MD patients as substitutes of the increase in mortaliDiscussion As mentioned in section 2.4, in evaluating the probability ty rate, which is uniform for all age classes. First, our asof paresthesia, the most conservative assumptions are sumption does not take into account the possible differmade, which will produce greater risks than the most ence in percentage increase in death rate among ages. probable risks; the following are assumed: (1) residents Second, the more important problem may be that paresry discharged from 1973 to 2005 is calculated under the social discount rate of 5% and the period of 33 years. The figure thus calculated is 2216 kg/year. The annual worth of expenditures is divided by the annual worth of the cumulative reduction in mercury discharged. This is calculated to be about 8950 yen per one gram/year of cumulative reduction in mercury discharged, which roughly equals $90 per gram per year.

7

Environ Engg and Policy 1 (1998)

8

thesia in general caused by methylmercury poisoning is not the same as MD. In the study of Kinjo et al. ‘MD patients’ means patients officially designated as MD. To be officially designated, they have to meet some criteria, because to be designated as MD implies the qualification for compensation. A less serious part of paresthesia, as a result, may not be included in MD. Consequently, to use the SMR for MD patients may have brought about an overestimation of the risk of paresthesia. On the other hand, Kinjo et al. obtained the values of SMR from the data on the deaths from 1973 to 1987 in spite of the fact that MD occurred since 1956, because the official designation of MD patients started in 1970. This means that the deaths of a significant segment of MD patients were not taken into account in the estimation of SMR. In this sense, the SMRs obtained in their study may represent those with paresthesia in general caused by methylmercury poisoning. By using LLE we could evaluate the risk of paresthesia on the same ground as that of cancer. We regard the increase in mortality due to paresthesia as reflecting both lethal and nonlethal cases of paresthesia, i.e., nonlethal paresthesia also contributes to the increase in mortality. The greater the pain, fear, and medical costs of the disease, the greater the mortality of a disease. Nevertheless, if the magnitude of these elements are not expected to be proportional to the mortality, we ought not neglect them. Though we do not have any idea at the moment about how to deal with differences in pain and fear between cancer and paresthesia, we can mention here how to deal with difference in medical costs. The medical cost of MD is estimated to be 1.8 million yen/year per case of patient, while that of cancer is estimated to be 1.7 million yen/year per patient [Ministry of Health and Welfare (1990a) (1990b)]. Because one case of paresthesia is estimated to cause LLE of 1.85 years, we can consider 973 thousand yen to be the cost for one year of LLE in the case of paresthesia. Similarly, with cancer, 135 thousand yen are thought to be the cost for one year of LLE. If we are allowed to regard these costs as being removed along with risk-reduction, these costs should be subtracted from the cost to reduce the risk in calculating BRR. The medical costs per year of LLE are, however, very small compared to the estimated costs of risk-reduction per year of LLE. The step of evaluating the SMR value for patients with methylmercury poisoning has uncertainties, with the greatest influence on the value of LLE. Therefore, a sensitivity analysis with regard to the SMR values is conducted. Regarding Kinjo’s data, a 95% confidence limit for the LLE value for a case of paresthesia consists of an upper limit, 2.87 years, and a lower limit, 0.80 years. The results indicate that the 95% confidence intervals of the BRR value is 370 million to 1320 million yen per life-year saved. As shown in Table 4, we assume that the effects due to the replacement of the mercury electrode process by the nonmercury processes are valid for twenty years, corresponding to the lifespan of the equipment in a factory from the engineering perspective. This assumes that, in absence of the government’s prohibition of the mercury

Table 4. Costs and effects due to replacement of the Mercury electrode process A

B

C

D

Year

Expenditures in 1989 yen (billion)

Reduction in Mercury discharged (kg)

Cumulative reduction in Mercury discharged (kg/year)

1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

68.66 57.44 55.59 51.83 51.18 1.24 1.19 1.12 1.12 4.82 15.15 15.60 23.40 20.09 14.33 1.60 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

0 790 742 304 48 43 59 15 283 48 125 403 518 134 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

0 790 1532 1836 1884 1928 1986 2002 2285 2333 2458 2861 3379 3514 3514 3514 3514 3514 3514 3514 3514 2724 1982 1677 1629 1586 1527 1512 1229 1181 1056 653 134

Annual Worth

19.8

2216

electrode process, they would have been phased out at the end of their design lives. We think this assumption is appropriate because the ion-exchange membrane process is as efficient as the mercury electrode process. In section 4.4, we conclude that the policy in this study is not cost-effective on the basis of the high BRR value of the policy. However, this criterion for policy evaluation has a critical drawback in that it considers only effects on human health but not the nonhuman adverse effects. In Japan, in those days, both the government and the public were not concerned about the nonhuman adverse effects resulting from mercury exposure. In addition, no nonhuman adverse effects in Tokuyama Bay have been reported. Therefore, we evaluate the policy only from the perspective of human health adverse effects. However, we are sure that this policy would contribute, to some extent, towards the mitigation of nonhuman adverse effects due to mercury exposure.

J. Nakanishi et al.: Analysis of prohibition of the mercury electrode

Generally speaking, the environmental program, which aims to enhance human health directly, can contribute, to some extent, towards reducing adverse effects on the ecosystem. Nevertheless, in many cases we measure the value of the benefits of the policy only from the perspective of human health risk reduction, which in turn underestimates the benefits of the policy. Therefore, we think that it is inappropriate to transfer the benefits of risk reduction of other sectors, such as medicine or occupation, into environmental protection. In this study, then, we compare the values of BRR among the environmental policies. However, there remains the problem that in our analysis, we neglect the variances, among environmental policies, of the effects of ecological risk reduction, even though the extent of the subordinate effects varies among environmental policies. Finally, we would like to mention the distributive problem of human health risk. This analysis is a good tool for the establishment of the policy to save more lives with limited resources. However, it is true that this analysis does not consider risk distribution among people in the nation. Even though a policy would be evaluated as cost-effective because of low value of its BRR, if a small portion of the population would be exposed to serious risk, the policy should be evaluated to be inappropriate from the perspective of the equity. In this case, individual risk is not so great: the worst estimates of paresthesia probability are small compared to the background incidence of paresthesia of 6.3% (WHO, 1990). Because the distributive problem of this policy is not serious, the efficient utilization of scarce social resources is important in this issue.

References

1. Japan Soda Industry Association (1982) One hundred year’s history of caustic soda industry in Japan. 2. Futatsuka S (1979) Epidemiological study on Minamata disease In: Arima S (ed) Minamata disease. Seirinsya, Tokyo, pp 95–119 3. Gamo M, Oka T, Nakanishi J (1995) Method evaluating population risks from chemical exposure: a case study concerning prohibition of chlordane use in Japan. Reg Toxicol Pharmacol 21 : 151–157 4. Gamo M, Oka T, Nakanishi J (1996) Estimation of the loss of life expectancy from cancer risk due to exposure to carcinogens using life table. Environ Sci 9 : 1–8 5. Harada M (1997) Neurotoxicity of methylmercury; Minamata and the Amazon. In: Yasui M et al. (eds) Mineral and metal neurotoxicology. CRC Press, Boca Raton, New York, London and Tokyo, pp 177–188 6. Kinjo Y, Nakano A, Sakamoto M, Futatsuka M, Kato H (1991) Clarification of mortality patterns among Minamata disease patients. Environ Sciences 1 : 73–88 7. Nakamura K (1992) Mercury pollution in marine environment and its microbial degradation. Environ Sci 5 : 1–14 8. Ministry of Health and Welfare (Statistics and Information Department, minister’s secretariat) (1990a) Estimates of national medical care expenditures 1989. Health and Welfare Statistics Association, Tokyo 9. Ministry of Health and Welfare (Statistics and Information Department, minister’s secretariat) (1990b) Patient survey 1990. Health and Welfare Statistics Association, Tokyo 10. Ministry of Health and Welfare (Community Health, Health Promotion and Nutrition Division) (1997) Nutritional survey 1997. Daiichi Shuppan, Tokyo 11. Nakanishi H, Ukita M, Sekine, M, Murakami S (1989) Mercury pollution in Tokuyama bay. Hydrobiologia 176/ 177 : 197–211 12. Nakanishi J (1989) Troubles in drinking water. Iwanami Publishing, Tokyo 13. Nakanishi J (1995) On environmental risk. Iwanami Pub6 lishing, Tokyo, pp 159–176 Conclusion 14. Nishimura H, Kumagai M (1974) Pollution in the Seto inThe results of a risk-benefit analysis of the Japanese govland sea-pollution by heavy metals. Kagaku 44 : 103–109 ernment decision of the prohibition of the mercury elec- 15. Nordberg GF, Strangert P (1976) Estimations of a dose-response curve for long-term exposure to methylmercuric trode process in 1973 indicates that its implicit value of compounds in human beings taking into account variability BRR is estimated to be 570 million in 1989 yen, U.S. of critical organ concentration and biological half-time. In: $5.7 million, value per life-year saved. This value is comNordberg GF (ed) Effects and dose-response relationships pared with the BRR values of three other policies ranging of toxic metals. Elsevier Scientific Publishing Company, from 24 to 45 million yen per life-year; the policy of reAmsterdam, pp 273–282 placement of chlorination by ozonation in the process of 16. Nose Y, Furuno J, Miyoshi T (1975) Epidemiological studies on the “Minamata disease” in Tokuyama. Ann Rep Soc Yawater purification; the policy of reduction in benzene maguchi Ind Health 20 : 7–31 content in gasoline from the current level to 1% in volume; and the policy of replacement of chlordane by chlo- 17. Oka T, Gamo M, Nakanishi J (1997) The cost per life-saved in the prohibition of chlordane in Japan: an estimation rpyrifos. The policy of the prohibition of the mercury based on risk assessment regarding the cancer risk and electrode process is evaluated to be not cost-effective. noncancer risk. Jap J Risk Anal 8 : 174–186 This study demonstrates that a measure of the loss of 18. Shirai F (1988) Studies on the relationship between the life expectancy (LLE) is useful for evaluating human characteristics of daily food consumption and the concentration of mercury in urine. Jpn J Hyg 43 : 923 health risks. Using the LLE, we are successful in quantitatively comparing the effectiveness of the policy for reduc- 19. Tengs TO, Adams ME, Pliskin JS, Safran DG, Siegel JE, Weinstein MC, Graham JD (1995) Five-hundred life-saving ing noncancer toxic risks such as regulation on mercury interventions and their cost-effectiveness. Risk Anal with that of the policy for reducing cancer risks such as 15 : 369–390 regulation on benzene. 20. WHO ICPS (1990) Methylmercury, Environmental Health Criteria 101, pp 76–99

9