Tobacco Documents Online

-7- but there was a draconian penalty---death. However all juries violated the rules, and no jury would convict anyone accused of the crime. Similarly if a group of assessors are asked an inappropriate question, or suspect that their answer may lead to stupid dracon- ian measures, they may well decide not to state the true opinion. Thus :'.f the question is merely "is this chemical a carcinoaen" with no question of potency, I believe it is now inappropriate. If it is coupled with a draconian action of a complete unquestioning ban if the answer is yes, many scientists will refuse to play the game„ We can enter this diagram at any point; but usually it will be the decision maker who asks for an assessment. We must also be aware that if well done, a risk assessment may illuminate several different decisions. For example, the risk assessment for benzene may apply either in the workplace, or for the environment. F,rrors often arise from not realizing that there are dif- ferent constituencies in risk decisions and each constituency must be satisfied. It is an error to calculate the average risk to society, show that it is small and ignore it in spite of the large risk to an individual segment. I illustrate this by cal- culating the risk of being killed by a bear in Glacial National Park„ According to the figures available to me, and as a risk asses-sor I,accept this data, 2 people were killed in 1967; 1 person in 1975; and 3 people in 1980. That amounts to 6 people over about 15 years, or 0.4 per year. The Park Service informs

- 8 - me that the total number of park visitors is 1,500,000/yr; about 18,000backpack cam.pers/yr (25,000 permit nights); and about 1000 em- ployees of park concessions. Of the 6 people killed, all were backpackers, and 4 were concession employees.* 'T'he risk averaged over the whole U.S. population is: 5 x 220,000,000 - 2 x 10-9 per year. I do not expect the President of the U.S. to lose sleep over it. It is a de minimis risk. 'The risk averaged over park visitors is: 17 x 1,500,000 = 3 x 10-7 per year; still de minimis by my criterion, but on the border of worth bothering about and indeed the park service have done little. 'rhe risk as a fraction of backpack campers is: 1-5 x 18 000 = 2 x- 10-5 per year. r This is a voluntary risk worth reducing, but probably acceptable and, indeed, my wife and I accepted it a few summers ago (but we did take care; put bells on our packs and hung the food on poles). The risk as a fraction of park concession employees is 1 x l~- = 3 x 10-4 per year. This is now large, as large as the risk of driving a car to get N to Glacier Park in the first place, and probably worth addressing ~ by the representatives of this smaller group. ~ But this analysis leads, if we are not careful, to another ~ trap., Why do you include in the denominator those who are not ~ ~ ~ * This is based on oral statements only.

-9- expos,e:d? Because we may know no better. But we should go on reducing the size of the denominator until one reaches a logical trap. We should include in the denominator all those "at risk" and not only those who were killed.c We get 6/6 or a risk of unity when we identify the group at risk as the group actually sleeping in the: path of the bears and subsequently mauled. The denominator must be chosen so that the risk calculation has predictive power. To restrict the denominator too much takes away that power. Bross, of the Rosswell Memorial Hospital in Buffalo, fell into this trap when discussing leukemia incidence. He restricted the group at risk to those who already had health problems which are precursers of leukemia. As shown by Bond6 his calculations have no predictive power. Another error is to assume that a risk benefit analysis should be constant for all time. This omits the possibility of techno- logy i.mprovement. It is clear that the public wants risk to be reduced, and will not stand for additional risks being introduced to soc:iety--unless there is an obvious compensating risk reduction. Therefore we must constantly search for ways of improving technology. I illustrate this by two examples. From the earliest days it was known that x-rays cause skin cancer and in the 1920's it was found that radiation causes leu- kemia and in the 1930's and 1940's it was found that radiation causes ~ a host of other cancers too. ~ 6V. Bond, BNL report. 0)

-10- In the 1920's physicians pointed out the huge benefits of rap9.d diagnosis by use of x-rays, and said that the benefits out- weicrht any conceivable risk due to the x-rays. Of course they were: right by a reasonable standard. This risk benefit compari- son was particularly simple because the items were measured in the same units. X-rays save lives and these can be compared dire:ctly with lives at risk from the radiation induced cancer. But the overall comparison of risk and benefit and the assurance that benefit exceeds risk is only the first question in a. risk benefit approach. in the 1920's it was pointed out that. the same benefit of rapid diagnosis by use of x-rays can be achieved with less risk: more sensitive x-ray film; image intensifiers; shielding against stray x-rays; and indeed it was not very expensive to do so. But these risks have only slowly been reduced, from many rads per x-ray in 1920 to 1 rad in 1950 to 5 millirads today. Why was the reduction slow? Rachel Carson, in her much misquoted book, Silent Spring,? agreed that pesticides were important and that the benefits of their use outweigh the risks. Again, if we think carefully, we can put at least some of the benefits in the same units as the risks; fewer pests means more food; more food means better nutri- tion; better nutrition reduces health risks. But Rachel Carson insisted it should be possible to have the same benefits with less risk ;by more careful use. ;[-iow do we ensure that we constantly address this question? ?Carson, R., Silent Spring, Houghton Mifflin, Boston, 1962.

I be].ieve that we must have a continuing decision process. The proce>.cture in each of the above examples was to show that benefit > risk and the decision therefore was that the action should pro- ceed. But in all cases of new technologies it seems appropriate to look further and ask whether we can reduce the risk with a modest cost. In setting the air quality standards for moving sources (auto- mobiles), Congress demanded a technology forcing approach. The standa~rds were'set progressively lower in the future and it was up to the automobile industry to find the technology to meet them. The i.ndustr ,y screamed, but a Committee of the National Academy of Sciences chaired by Dr. Edward Ginzton of Varian Associates agreed that it was technically possible to'meet this standard, and they are being met. is this technology forcing procedure the best way for techno- logy improvement? Congress, after this success, thinks-'so. If we feel otherwise we must find a better way. I throw on the table for consideration the following idea. When any new product or process is accepted, there must be a fraction (1 to 10%) of the profits spent on a study of the alternatives to the actual action; other pesticides; less use (including how to restrain overzealous salesm,en); better x-ray machines--and less x-rays, including how to prevent,x-rays being taken merely to protect against liability in a negligence suit. This money could be paid by industry to NAS or a foundation to do this study.

-12- I now bring to your attention four separate court and regu- latory actions that recognize the importance of risk assessment and its more general extension, risk analysis. 1. The Court of Appeals,8 in the case where FDA questioned the safety of Monsanto's plastic bottles made with an acrylonitrile polymer, stated that the administrator of FDA can ignore de minimis risks, notwithstanding the superficial rigidity of the Delaney Clause. 2. The U.S. Supreme Court decided in the benzene case9 that the Secretary of Labor for OSHA had to find (technical/legal term) that -there is a significant health risk before proceeding and by quoting the one and only risk assessment before the court implied approval of risk assessment as a procedure for deciding whether a risk is insignificant or.not. Chief Justice Burger stated that "inherent in this statutory scheme is the authority to refrain from regulation of insignificant or de minimis risks." Although the Court quoted my testimony in this case, they failed to quote my successful effort at risk reduction. At the public hearing, the administrative law judge, and half the audi- ence, were smoking and the room was filled with haze. I objected to being exposed to such dangerous carcinogens in the workplace. Now the hearing room is covered with numerous, large "No Smoking" signs. 8Monsanto,' et al., v. Kennedy, FDAO, 613 F 2nd 947 (DC Circuit)), 1979. 9American Petroleum Institute v. OSHA, et al., 581 F 2nd (DC Cir- cuit), 1978; 100 S. Ct. 2844 (1980).

-13- 3. Now we have a consent order10 between FDA and several cosmetic manufacturers. In this "FDA agrees to propose the utili- zation, of scientifically accepted procedures of risk assessment and to ra.ise the issue as to whether, in the view of the procedures, hair dyes containing 4-MMPD present a generally recognized risk to human health." 4. The FDA made a similar decision in deciding that the risk of ls:a.d acetate hair dyes (the Grecian formula) is insignificant. As I look at a 5th case, the cotton dust case, it appears to be different. It has been argued that the Supreme Court decision is a repudiation of risk analysis. But we must note that "exposure to cotton dust represents a significant health hazard to employees,"ll The health hazard at current levels, acute byssinosis although re- versible, is directly observable, unlike the effects of benzene at low doses which is merely inferred by extrapolation. I calculate with a very rough first guess using the numbers in reference 11 that the cotton dust standard proposed by OSHA saves cases of byssinosis (brown lung disease) and hence premature death at a cost of between $50,000 and $500,000 per case--less than the $1 million per case often spent to reduce occupational disease, and $10,000,000 to re- duce occupational death and much less than the $240,000,000 per cal- culated case I found in my calculations in the benzene case. Of course, we.would like to understand the connection (if any) between 10Carson Products, et al., v. DHHS, et al., Civil Action No. CV 980-i'1., U.S. District Court for the Southern District of Georgia, Sept. 1980. 11101. Supreme Court Reporter, p. 2478, American Textile Manufacturers ~ v. Donovan and National Cotton Council of America v. Donovan, 43 Fed. ~ Rea., 27350, col. 1. ®

-14- acute byssinosis and life shortening effects. Data on this is scarce and represents a challenge for risk analysts. My cheerful tentative conclusion is that the Supreme Court be- have:d with commonsense in both decisions. If OSHA had perfomed a reasonable risk-benefit analysis it is not obvious that they would not have reached the same decision, or decided to lower the standard for cotton dust even lower than 200 yg/m3 until they matched $1,000,000 to $1.0,000,000 per life--unless of course they found other, more seriaus secondary effects on the economy. The point here is that, as far as I can tell, there was no widely accepted risk analysis presented in this case; and OSHA's decision to take prompt action to tighten the standard is then not unreasonable. I now come to several errors that arise from inadequate com- parisons of risks. For cancer risk calculations, Crouch and 1 12 have been emphasizing the importance of including uncertainties in a risk calculation. I illustrate this by going over an argument I had with EPA's director of water quality standards at the Toxico- logy ;C'orum in February 1981. I objected to his setting of a stand- ard of 100 ppb for all halomethanes; drinking 2 1/2 liters of water containing chloroform every day for life can be shown to give a risk of 4 x 10-5 per year or 3 x 10-4 per life. This is based on severalassumtions; that the dose response relationship is linear; that men in their lifetimes have as great a risk of cancer ~ as animals in their lifetimes, when exposed daily to the pollutant ~ ~ Edmntnd A.C. Crouch and Richard WIlson, "Interspecies Comparison ~ of Carcinocaenic Potency," Journal of Toxic. and Environ. Health, ~j 1979. Edmjand A.C. Crouch and Richard Wilson, Journal of Risk OA Analysis, 1, 1 (1981) .

with the same fraction of body weight. EPA have the same standard for all halomethanes. Yet bromo- methanes are more toxic and more mutagenic; by skin painting tests more carcinogenic. The exact amount is uncertain, but it is hard to see how the risk could fail to be greater at equivalent exposures, and therefore the standard should be lower.l3 Similarly, tribhlo- rethelene and perchlorethelene are less toxic, and less carcinogenic (if carcinogenic at all) and the allowable exposure might reasonably be greater--or at least it should be explained why it is not. We can reintroduce the zero risk error in a new form if we demand too low a risk, and demand too conservative a way of ca1- culat~~eng it. FDA has proposed14 in a discussion of accidental food additives such as DES, that'risks less than 10-6 per lifetime (1.4 x 108 per year) be ignored as de minimis and by implication other risks should be regulated. This is a very low number. Moreover there are very conservative rules proposed for calculation. By simple extension of these rules, I would find that many munici- pal drinking water systems in the U.S. have risks of 10-4 per year. It is clear that the FDA rule cannot be applied to all situations, and 9.t: therefore needs more discussion to show just where it should be applied (if at all). At the present time, most agencies are inconsistent without saying so. In the next two figures I show how this comes about. The risk, ~ according to our approach, has 4 factors: ~ ~ ~ 14 Q1 ~ ~ W ~ ~

CA+G (95th percentiley POTENCY Species 'Conversion ~'~octor