Underestimating arsenic's risk: the latest science supports tighter standards. (The Arsenic Controversy).
But toxicity is only one hazard of arsenic; another is its carcinogenic properties. Scientific evidence of a link between cancer and arsenic dates back to the late nineteenth century when researchers found a connection between regular use of the medicinal Fowler's Solution (one percent arsenate) and skin cancer. In the following decades, similar connections were found among people who had regular exposure to arsenic-based pesticides and to the fumes produced during metal smelting.
More recently, researchers have given considerable attention to incidence of cancer among users of water supplies with high concentrations of arsenic. Earlier this year, the New York Times reported on a five-year study in Chile that showed some 700 people in excess of the background rate died from cancer that was linked to arsenic in drinking water at concentrations of 500 [micro]g/L. In Bangladesh, 30 million people are exposed to similar levels of arsenic in their drinking water, and thousands of Bangladeshis have died from secondary effects of cancerous skin lesions.
Those studies involved arsenic levels that are significantly higher than what the U.S. Environmental Protection Agency (EPA) allows. But, concerning low-level exposure, scientists traditionally have believed that arsenic has few adverse effects. Standard experiments supported that belief: Test animals, particularly rats and mice, did not show any unusual incidence of cancer when they ingested arsenic concentrations proportional to concentrations ingested by the average American. Therefore, researchers have long believed that people do not develop cancer from low-level arsenic exposure.
However, that belief is now in question. A number of recent studies have indicated a potential link between low-level arsenic ingestion by humans and cancers of the bladder, kidney, and lung. What is more, Australian researchers have announced that they had induced mice to develop cancer using small doses of arsenic. Though there is still debate within the scientific community over those studies, evidence is now accruing that such a link exists.
That leads us to a crucial question: Given the recent scientific findings, should the United States lower the allowable limit for arsenic in drinking water below the 50 [micro]g/L level? As often happens, the whole world is watching us as we discuss what to do.
ESTIMATING LOW-DOSE RISK
The fundamental issue in trying to answer that question is how to estimate arsenic risk at lower levels from the generally accepted measurements at higher exposure levels.
Some scientists and physicians argue that a linear dose-response model is appropriate. Under such a model, researchers would assume that the rate of cancer resulting from exposure to 50 [micro]g/L of arsenic would be approximately 10 percent of the rate that results from 500 [micro]g/L. But other scientists believe such a linear projection is inappropriate; they claim that there exists an exposure threshold below which no effect will be seen. According to their model, even if exposure to 500 [micro]g/L of arsenic produces a certain incidence of cancer, it is possible that a 50-[micro]g/L exposure would have no effect at all on cancer rates.
Linearity A half-century ago, Sir Richard Doll presented his multi-stage theory of cancer. Doll noted that most cancers caused by an external agent could not be distinguished from those that occur naturally and may be fundamentally indistinguishable. He thus reasoned that the external agent affected one stage in the cancer development in the same way as the natural processes do.
Twenty-five years later, a group of scientists -- including one of Doll's collaborators, Sir Richard Peto -- pointed out that Doll's theory is very general. If an agent increases the probability of any step in the cancer formation process in the same way as natural processes, then almost any biological dose-response relationship becomes linear at low doses.
That latter argument forms the basis for EPA's assumption of low-dose linearity as a default for all carcinogens. If one uses the default dose-response and starts from the data from Chile where there is a 10-percent increased mortality risk (mostly a lung cancer risk) for people who drink water containing 500 [micro]g/L of arsenic, the default assumption leads to a one-percent projected risk for people drinking water with 50 [micro]g/L of arsenic.
Non-linearity Some biologists and toxicologists insist that there is a threshold below which there is no effect, or at least that risk is below what is projected under the linear model. Until recently, animal experiments supported their beliefs: Low levels of arsenic did not cause cancer in animals and, it seemed reasonable to assume, was unlikely to cause cancer in people. But recent studies show that a metabolite of arsenic, DiMethylArsenic Acid (DMAA) does cause bladder cancer in rodents. What is more, a growing number of arsenic researchers appear to be changing their view about the usefulness of animal testing in determining human risk from low-level exposure to arsenic.
Until more research is done on arsenic exposure, I, as a risk assessor, would use a bimodal distribution for discussion of the probability of health effects corresponding to the differences in risk estimated from the two approaches. Moreover, I would carry the calculation through on both approaches and emphasize the difference to the final decision maker.
COSTS OF COMPLIANCE
Critics of EPA's effort to lower the allowable limit on arsenic have charged that the infrastructure costs that would result from tightening the allowable limit would outweigh the monetized benefits. However, the costs of any action to meet a regulation are far more uncertain, and often much lower, than opponents of the regulation tend to claim.
The most well known example of such an over-projection was the actual cost of reducing the occupational exposure to vinyl chloride in the polyvinyichioride (PVC) industry. In 1972, when the carcinogenicity of the monomer vinyl chloride became apparent, estimates for reducing the exposure were high. However, a temporary downturn in demand enabled the plants to be modified without disruption to overall supply. Manufacturers reduced worker exposure to the compound by sealing the equipment to stop fugitive emissions. When that was done, the manufacturers experienced a net saving of material and, as a result, a saving of money.
Implementation Those who are critical of EPA's new 10-[micro]g/L standard assume that compliance requires infrastructure modification so that, at all times, the concentration of arsenic in all the water is below the permitted level. But that may not be necessary for meeting public health goals based on chronic -- as distinct from acute -- effects of arsenic. What is important is that the total amount of the chemical ingested over a long period -- perhaps five years -- is below the permitted level. That goal allows considerable flexibility in the employment of arsenic-removing equipment.
For example, the water supply of a large community might use filtration and the mixing of high- and low-arsenic-content water to meet the average required level. The Los Angeles water district uses the Los Angeles Aqueduct, which holds water from Mono Lake and Lake Crawley in the eastern Sierras. The water contains arsenic at an average level of 23 [micro]g/L. But careful management, filtering, and mixing of water from other sources over the past 10 years brought that level down to only 2 [micro]g/L in most locations.
Of course, smaller water systems may not be able to employ such methods to lower their arsenic concentrations. But those systems -- with perhaps 1,000 to 5,000 users -- should be allowed other options. Because only 0.1 percent of water entering a house is used for drinking, a utility could supply low-arsenic drinking water separately from the water used for such activities as washing. In my parents house in a London suburb (in 1935), we had separate taps for drinking water and non-drinking water. Small systems could adopt such a structure, or could distribute bottled drinking water to customers.
Another option for those systems is a two-step regulatory system -- a mandatory 50-[micro]g/L standard and an advisory 10-[micro]g/L standard for water systems with users who have active participation in the systems' decision-making. For those systems, all users would be informed of the calculated effects of arsenic exposure on health, and the costs of lowering arsenic concentrations in their water. They would then be able, through ordinary democratic procedures, to participate in the decision of whether to adopt the 10-[micro]g/L standard for their system or to maintain the 50-[micro]g/L standard.
Wells Among the strongest critics of the new standard are developers in the western states. Increasingly, they are dependent upon wells to supply water to their developments, instead of the surface waters that were almost universal in the past. Wells often contain arsenic and thus raise exposure problems that surface waters do not.
The U.S. Geological Survey (USGS) has indicated that it may be able to help drillers find aquifers that contain low levels of arsenic, at a cost that is quite inexpensive compared to the cost of removing arsenic from existing water supplies. That suggests that developers' fears about the cost of the new standard may be misplaced; the cost could be considerably lower than that proposed by the water industry and lower than that proposed by EPA.
The cancer risk of arsenic in drinking water at a 500-[micro]g/L lifetime exposure is approximately the same as the lifetime cancer risk of a heavy cigarette smoker. According to linear dose response, the risk at 50 [micro]g/L is only 10 percent of the risk to smokers. Opponents of a further tightening of the allowable limit might suggest that the risk from arsenic at the 50-[micro]g/L level is insignificant. But such an argument has never succeeded in public discussions of other agents and chemical pollutants.
Emerson once said that excessive consistency is the hobgoblin of small minds. Nonetheless, a regulatory agency such as EPA should have a very clear reason for any lack of consistency in its regulations involving acceptable risk. Unfortunately, over the 30 years of EPA's existence, observers have seen that clarity is not one of the agency's virtues.
For instance, EPA has strict regulations regarding concentrations of thrichloroethylene (TCE) in water supplies. If EPA wanted to set its allowable arsenic level at the same degree of risk as its allowable TCE level, then the agency would be trying to impose a .005-[micro]g/L limit -- an obviously unattainable goal. If the cost-benefit procedure used by EPA in its recent rulemaking on arsenic were to be applied to regulation of TCE or chromates, then TCE regulation would be far less severe than what it is today.
There would be a similar inconsistency with EPA's regulations concerning high-level nuclear waste, such as that which is to be stored at the Yucca Mountain facility. The Yucca Mountain regulations are intended to protect human health for as long as the waste poses a threat. But, unlike nuclear waste that eventually decays, arsenic remains carcinogenic forever. Arsenic rules that are consistent with the Yucca Mountain rules would have to protect human health forever, which means that most drinking water and farm fields that, at one time, were treated with arsenic would today be out of compliance. So, are EPA's regulations for Yucca Mountain excessively restrictive, or should the agency tighten its regulations concerning arsenic? Or is there some clear scientific reason for why arsenic risk should be considered more lightly than risk from nuclear waste?
What, then, do we make of the public opinions on acceptable risk that are implicit in such recent movies A Civil Action and Erin Brokovich? It seems to me that such movies, as representations of public opinion, cannot be ignored. Painful though it may be to revisit past decisions, the United States is ill-served if one does not learn from them. Of course it was not as expensive to regulate TCE as it will be to regulate arsenic, and even less expensive to continue to regulate TCE. Society may therefore wish to retain the existing TCE regulations, but that should be done after careful examination of the new perspective that, hopefully, regulators and the public will gain from the arsenic debate.
One of the important features of the 10-[micro]g/L arsenic rule that EPA wants to implement is that it resulted from the first use of a cost-benefit analysis for a drinking water pollutant. In conducting that analysis, EPA deserves great praise.
The agency was unusually thorough in its study; regulators presented their proposal at scientific assemblies and other public meetings, and provoked discussion. EPA received over 1,000 public comments, and -- although, as is the agency's bad habit, it never formally responded to or acknowledged the comments -- it seems clear that many, if not all, of them were considered seriously.
Of course, critics of the new arsenic rule have offered thoughtful criticisms of how EPA carried out its analysis, and have offered alternate, logical procedures for cost-benefit calculation. It is of vital importance that the new EPA committees discuss such criticisms and alternative procedures. In doing so, committee members will become fully aware of the precedents that they are establishing for the future, and they will set those precedents with their eyes open.
* "Chronic Arsenic Poisoning: History, Study and Remediation" website, hosted by Harvard University's School of government. Found on the Web at: http://phys4.harvard.edu/~wi1son/arsenic_project_main.html.
* "Discounting of Long Term Costs: What Would Future Generations Prefer Us to Do?" by A. Rabl. Ecological Economics, Vol. 17(1996).
* "Evaluating Health Effects of Societal Decisions and Programs," by H. Raiffa, W. Schwartz, and M. Weinstein. In Decision Making in the Environmental Protection Agency, Vol. 2b.Washington, D.C.: National Academy of Sciences, 1977.
* "Lung and Kidney Cancer Mortality Associated with Arsenic in Drinking Water in Cordoba, Argentina," by C. Hopenhavyn-Rich et al. International Journal of Epidemiology, August 1998.
* "Marked Increase in Bladder and Lung Cancer Mortality in a Region of Northern Chile due to Arsenic in Drinking Water," by A. H. Smith et al. Amen can Journal of Epidemiology, Vol. 147, No. 7 (April 1998).
RELATED ARTICLE: Americans and Cost Discounting
In conducting a benefit analysis of a proposed new rule, should society I discount the value of lives saved by the rule in the far future? In Bangladesh, the most important consequence of arsenic exposure is keratoses, which arise within a few years of exposure. Keratoses often leads to gangrene and the eventual amputation of limbs. But in the United States and for the longer term, cancers are the most likely consequence of arsenic exposure. Those cancers will occur some two or three decades after exposure, so, in conducting a cost- benefit calculation of tighter limits on arsenic, should one discount the cost of those future cancer cases?
Jason Burnett and Robert Hahn, in their article previous to mine, argue that one should employ such discounting, using the standard seven-percent discount rate used for money. Such a discount would, in effect, reduce the risk by a factor of between five and 10. Most economists agree with such a move, and merely argue about the expected valuation in the future. But a few others have disagreed with discounting.
The issue of whether or not -- and how -- to discount should have been discussed more thoroughly as part of EPA's rulemaking on the "cost of a statistical life." It was not well defined what "discounting" means, where in the calculation it should be applied, and whether EPA calculations already include some form of discounting. It would be logically superior to assign a cost to Years of Life Lost, rather than lives per se, which would take into account, in some part, the fact that the death of a 30-year-old is more significant than the death of an 80-year-old.
But it is abundantly clear that American society, as a whole, does not agree with discounting. For example, with a seven-percent discount rate on lives lost in the future, little money should be spent on toxic waste disposal. Indeed, with even a 0.1-percent discount rate, the United States is spending far too much on consideration of disposal of high-level nuclear waste. Yet the American public continues to offer strong support for toxic waste cleanup efforts, despite the costs and the supposedly questionable benefits.
To discount lives in a risk calculation is a policy decision that appears to conflict with the will of the American people. EPA utilization of such discounting would create profound precedents and should not be done lightly.
Richard Wilson is the Mallinckrodt Research Professor of Physics at Harvard University. He is also an affiliate of the Harvard Center for Risk Analysis, and the Program on Science and International Affairs at Harvard's Kennedy School of Government. He and co-author Edmund A.C. Crouch have recently finished work on the second edition of their book Risk/Benefit Analysis: Nuclear, Chemical and other Risks. Wilson can be contacted by E-mail at email@example.com.
|Printer friendly Cite/link Email Feedback|
|Date:||Sep 22, 2001|
|Previous Article:||A costly benefit: Economic analysis does not support EPA's new arsenic rule. (The Arsenic Controversy).|
|Next Article:||Arsenic and old facts. (Commentary).|