Tobacco Documents Online

ENVIRONMENTAL TOBACCO SMOKE Chair: Ernst L. Wynder American Health Foundation, NY DR. WYNDER: Good afternoon. I am Ernst Wynder. I am delighted that you got me out of New York to see this beautiful environment, and I really enjoy to be with you again. I was reflecting on my own long career in epidemiology, and I was thinking that if my first thought hadn't been smoking and lung cancer where we published in 1950, and showed that active smoking was, indeed, the major cause of cancer of the lung and if, instead I had, let us say, picked the relationship of diet and colon cancer or something else with a difficult epidemiological interpretation, I might never have become an epidemiologist. Perhaps I might have become a psychiatrist. This is only a joke to the extent that some of you know psychiatrists. (Laughter.) DR. WYNDER: So, right off the bat, I would like to distinguish between macro and micro differences and in my first slide I want to show that the field of macro differences, indeed, is not over, and in fact, to me Japan always has been a gold mine in epidemiology. Currently Dr. Blackburn and I are involved in a nutrition intervention study on breast cancer where we place women who already have breast cancer on a 15 percent fat diet after a standard therapy and the control eat what we normally eat. If you cannot find an explanation for difference of four-fold or postmenopausally eight-fold, it is very difficult to find them for differences with a relative risk of 1.4. The same applies to cancer of the prostate, and I am amazed with all the discussion we have on prostate cancer in this country we overlook. A marvelous lead right here in Japan, ten-fold differences. You might, of course, say that Japanese are genetically immune against prostate cancer, but of course, fortunately for epidemiology, if not for the population that immigrates, they come to our country, and within one or two generations they have our rate of prostate cancer, and what few people have noticed also a major difference in lung cancer between us and the Japanese males, even though the Japanese males for a long time have smoked at least as much as males in this country. So, we are currently engaged in a study to explain this. We are looking for difference in smoking habits, with difference in onset of smoking, for differences in tea consumption, difference in use of charcoal filters, and for differences in diet. The point I want to make is that there are considerable differences between the Japanese and ourselves that continue to be a major lead for epidemiologists. When we compare micro and macro exposure we, also, have got to be aware that against macro exposure we have no macro defenses. It was clear to me in 1950, that if you actively smoke a cigarette you destroy your ciliary defenses; you destroy the mucous-producing glands and the tobacco has a chance to be absorbed by the bronchial epithelium. We need to recognize that we, in our anatomy have remarkable macro defenses and some micro defenses which would not do well in massive exposures but which might do well against micro exposures. My discussion here in introducing the theme really is to intxoduce to you the problems that we face in the epidemiology of weak association, and as the previous speakers have well pointed out, we must balance what we 307

01 find in animals to what we experience in humans. I am very grateful to the National Cancer Institute that over the years they have permitted us to develop the American Health Foundation that today employs a staff of some 240 that range from molecular biology to health promotion. All of our division recognize their limitations, but they are fortunately, not the same. When we look at the epidemiology of weak association, and I will just discuss it in general terms, there are four problem areas. One of them is how do I select my case and my control: At the American Health Foundation we have never chosen a control that was not the patient him or herself. In many studies where they do not have enough patients that are still alive, investigators choose surrogates. I don't know whether any surrogate knows as much about me than I know about myself. Just the other day to show to one of our young interviewers I attended one of the interviews at Memorial Hospital. It was a bladder cancer patient, and we went through occupational history, and we asked, "Have you been exposed to a toxic environment Mr. Smith?" And he said, "No," he really didn't recollect that he was exposed to anything that he could think about, whereupon his son said, "But, father, you know you had this great exposure to diesel exhaust when you drove the truck." So, the son had a different view about his father's exposure than the father has himself. There are biases and confounders and subgroup analysis. I will briefly allude to several of these. What is generally not recognized, that when you deal with tobacco, tobacco is associated with many different variables, including academic status. If today we go into a high school, and we correlate academic standing with who smokes, we find a negative correlation. We, also, find a close correlation to alcohol consumption, sexual activity and absenteeism. Here we are showing, which is less apparent, that smokers generally have a higher fat consumption, a higher meat consumption and a higher serum cholesterol level because they have a higher fat intake. On the other hand, smokers have a lower intake of fresh fruits and vegetables. This all means that if you are living with a smoker that the fat consumption of that household would be different than if you lived in a family that had no smoker in the household. You have clearly got to think about fat as aconfounderto tobacco consumption. An aside, in all the studies now in the literature where they show that fruit and vegetable consumption seems to reduce the risk among smokers to cancer of the lung, I am not sure whether it is due to fruits and vegetables or whether it is due to higher intake of ineatconsumption because our own data suggest that dietary fat is an ubiquitous tumor enhancerfor most carcinogenic exposures. Now, if you look at the sex ratio of lung cancer, squamous lung cancer has a sex ratio of 1.7, largely because in general men used to smoke more than women, but in spite of the fact that all types of lung cancers are caused by active smoking, you will see for adenocarcinoma which is the next highest in frequency a sex ratio of .8. So, in spite of greater consumption of tobacco by men, women have relatively more adenocarcinoma. So, you say to yourself what is the reason, and incidentally, what is quite new, a similar sex ratio applies to small cell and large cell on the basis of our own data, in line with those from the SEER data. As to why adenocarcinoma relatively speaking ought to be more common in women than in men, even though the men smoke more, you must think about the endocrine system. You could not think that is solely by diet even though in this experiment by John Weisburger is in line with the fact that I said that dietary fat is a ubiquitous tumor enhancer for carcinogens. Dr. Hoffmann has shown the same finding for NNK. 308

On a high corn oil diet the tumor yield whether mammary tumor or lung tumors is much higher than on a low corn oil diet. We must think more and more to all types of carcinogenesis in humans, that diet has a broad effect on cancer risk and therefore the Japanese data are so very interesting because they used to consume only 15 to 20 percent of their calories from fat whereas we have consumed 40 percent. Coming back to lung cancer in non-smokers, you will see how rarely we see lung cancer and small cell and squamous cell and somewhat more large cell and more adeno, and a relatively uncommon type of alveolar cell cancer of the lung. Adenocarcinoma is of particular interest to us. It is a peripheral cancer, and we are studying to what extent other risk factors than smoking could be a factor. Now, needless to say, if you think about something women have that we don't have, one factor certainly relates to reproductive and hormonal variables. It is rather interesting in my own mind that it took us unti11985, that we even included the reproductive history into our lung cancer questionnaire because we had not thought about it. As all of us know, you never make a discovery unless you think about it. DR. WYNDER: The point is we have shown that full-term pregnancy, hormonal replacement short cycles and even menopause suggest that female hormones may be a risk factor for adenocarcinoma of the lung, together with low body weight. I don't know quite how low body weight increases your risk for such types of lung cancer, particularly in women, but it is likely to be related to some endocrine factors. Here is certainly a factor to adenocarcinoma of the lung, namely the reproductive system hormonal factors that we have not seriously considered heretofore. I believe the same hormonal factor applies to adenocarcionoma of the lung in men because as all of you know, we all have estradiol, as well as estrogen receptors in our lung. If you look at squamous and adeno and lunch cancer, you see that over the years the percent of non- smokers in our own series has decreased in squamous from a very low level of about 3 percent to less than 1 percent, but adenocarcinoma has, also, decreased from 10 percent to 5 percent. Thus, non-smokers with lung cancer in our population is decreasing. It is not increasing. So, we are not looking for some new variable that has appeared in recent years. We fmd the same trend for glandular lung cancer in women, from a high 85 percent as I reported in 1950, to about 6 percent today. A big problem in interview studies is bias because we depend on what the individual tells us. We have four kinds of biases, recollection, denial, wish, social desirability. This bias is different in the case than in the controls because the case who has cancer looks differently at a certain question than the control in all of these biases. I will explain some of them to you. Now, here is a whole list of biases. I wrote a paper on this some years ago, and one of my favorites is the travel time bias. If you have a summer home in the seaside, and the average it takes you is 2 hours and 15 minutes, and you speak to somebody who has got a home there and you say, "George, you have this wonderful home in Southampton, how long does it take you?" He says, "One hour and 35 minutes." That was sometime around midnight when there was no one else on the road. (Laughter.) DR. WYNDER: This is the bias of the homeowner, but now, the problem becomes worse. You speak to a fellow who could never afford a home in Southampton, and you say, "Would you like to have a home in Southampton?" He says, "Never, because it would take me 4 hours to get there." (Laughter.) 309

DR. WYNDER: So, we have different kinds of biases. Everybody knows about food and obesity, asbestos and lung cancer, drinking and cirrhosis, sexual exposure and cervix cancer patients, delay in diagnosis for a cancer patient and most recently we have another, variable fat intake and breast cancer patients. The New York Times, my favorite scientific paper always reports about dietary fat and breast cancer. Now, a patient comes in, and you say, "You have cancer of the breast?" "Yes." "Do you eat much fat?" "Doctor, one thing I never do, eat fat." And then the husband says to the wife, "Why don't you tell the doctor the truth?" As to cookie consumption in children, how many children will tell you the truth how many cookies they consumed last week? We also have biases as to preferring publishing positive scientific reports. Next, we have a bias divided for investigators and interviewees. What I am discussing here applies to any study, not the one we are discussing particularly today. We have an investigator bias. Those of you who are part of industry, of course, know you have a significant bias. Of course, it is difficult for you to say, "My product is harmful," and get job advancement in your particular industry. You have a bias we all recognize. But we, in academia, have, also, a bias because we rather present a positive than the negative finding. Next comes an interviewer bias that I alluded to already before. Investigators in New Zealand asked people, "How tall are you, and how much do you weigh?" If I really had been prepared for this question, I would have gotten your weight and height yesterday on a questionnaire and would now, weigh you and measure your height. What this study showed that we, as men overestimate our height significantly. (Laughter.) DR. WYNDER: And those of you who are women underestimate your weight significantly, and now, I am sure if Alvin would put this through meta analysis you could do 1,000 of these kind of studies, and you would come out with the same thing as we have recently shown in the American Health Foundation. My colleagues didn't know I wanted to do that. We asked them how much they weighed, and two weeks later we measured them. It is a bias that is inherent in our nature. If I am 5' 10-1/2" tall, I like to say that I am 5' 11" and obviously I always like to say that I am thinner than I actually am. This kind of bias also applied for occupational studies. In our studies on diesel exhaust when we classified by occupation and adjusted for smoking which you always must do because obviously smoke is a major risk for lung cancer, we find no difference. But when we ask self-reported exposure while not significant we get an odds ratio of about 1.3 because obviously some of our patients are on Workman's Compensation. All self-reported exposure are likely to suffer from this type of bias. Here is another bias. My friend, George Blackburn referred to me a study by Mertz on nutrition. They measured how much people have eaten. They then did surveys and the average individual underreported their caloric intake by 20 percent. This seemingly is part of human nature. If you go home next week and your wife asks you how much you have eaten of this marvelous food they serve you here, you say, "Well, I hardly have consumed anything except salad and iced tea." (Laughter.) 310

DR. WYNDER: I noticed how many beers you drank last night. But this bias is even greater if you ask about fat intake where a bias of some 30 percent has been reported. The point I am making here, whatever it is, if it relates to human nature you have an inherent bias that is difficult to measure and therefore, very difficult to control. Here is another example that relates to exposure to previous smoking. It is a very important bias because it deals with the denial social desirability bias. This study is part of the Monica study in Germany where they interviewed men in 1984 and 1985. Dr. Heller reported that among those in 1984-85 who declared themselves to be ex-smokers, 4 years later 17 percent say that they never smoked. Now, this is a very significant denial bin much larger than estimated in the EPA report. Why is this so important for lung cancer? It is important for lung cancer because even if you gave up smoking 20 years ago, you never reach the level of those who have never smoked, and that is one of the amazing things to me about the effect of active smoking on lung cancer development. Even at 30 years you still have a relative risk in our own data and the data from the American Cancer Society and data from Doll and Hill somewhere between 1.5 and 2.5. In this case measuring cotinine levels obviously does not help. It is much easier, of course, of course, to deny something that happened 20 or 30 years ago than what happened a week ago. My fmal slide relates to the criteria ofjudgment, so important in reaching a causative conclusion. In 1964, when the first Surgeon General's report was published, there was a well-written segment entitled Criteria of Judgment based upon the criteria of Bradford Hill, written by Cochzan. Even though we had a relative risk of up to 40 to 1 for active smoking, they examined a whole series of criteria before concluding that this was a causative association. Today we often have odds ratios in many different fields of 1.2 or 1.4 and even forgetting about faults in methodology. We rarely consider the criteria of judgment as we recently did when somebody suggested that alcohol increases the risk of cancer of the breast, and we went through the same criteria of judgment and we came to the conclusion it did not, and we did the same in a study on saccharin and bladder cancer. These criteria must be considered. Now, let me state very clearly a true 1.3 relative risk involving many people is a major public health problem. All I am saying here, a low magnitude of effect makes it more difficult to make sure it is a causative association. In respect to ETS we have got to ask, "How consistent are the studies with squamous against adeno; how consistent are they with home exposure or work exposure?" All these have to be pulled together, and you have to have an internal, as well as external consistencies of the data. Time trends, we have gotto ask ourselves, are these cancers in non-smokers increasing andif not, certainly nothing new has happened. Clearly it is biologically possible for inhaled substances to induce lung cancer. You inhale diesel exhaust, pollutants on Times Square, tobacco components. But you have to prove it in terms of epidemiological factors in an unbiased fashion. You have to look at confounders, biases, and here is another interesting, factor. With today's computers you push a button, and all of a sudden, you have 15 outcomes and you choose a couple that seem to relate to the theory that you want to establish. I have seen papers showing odd ratios of .6 and .8 that are totally ignored and then the 1.2, that emphasized. Refer to all of them. What I have tried is to just introduce the field of the epidemiology of weak association. I have not specifically commented on the ETS, or this will be done by my colleagues. I have discussed some significant problems ## in epidemiology of weak associations, particularly when the proof of the association depends largely on the reply that the patient or surrogate gives to a specific answer which in certain instances is very difficult to give and thus difficult to evaluate. Thank you very much. (Applause.) 311

DISCUSSION DR. WYNDER: Is there any particular question, maybe one question that you have relative to the epidemiology of weak association? DR. SHUBIK: Your slides are very interesting, Ernst. I wonder if you thought that the decrease in adenocarcinomas in non-smokers could be due to the decrease in environmental tobacco smoke? DR. WYNDER: I wouldn't want to comment on that, but whenever you have a decrease in non-smokers in itself something must changed in the exogenuous or endogenuous environment. On the other hand, there has been a relative increase in adenocarcinoma in men. Here our suggestion is either as Dr. Hoffmann believes there may be an increase in nitrosamines that can be endogenously react with pulmonary tissue or as the airborne theory holds that as tar yield in cigarettes went down you inhale more deeply. Thirty years ago you inhaled with a 30 cc's per puff volume; today it is over 50 in order for you to get enough nicotine, and the more deeply you inhale, the more nicotine effect you get, and the more deeply inhaled, the more our tertiary or secondary bronchi are affected by tobacco smoke, and they may have a lesser defense against inhalants than the major bronchi have. If there is one final question? DR. HALL: I may have misunderstood, Dr. Wynder. Your discussion just now related to the rate in smokers, did it not? DR. WYNDER: The slide that you referred to suggests that in our own data the percent of non-smokers with adeno and squamous lung cancer is decreasing over the years. In women you could expect it because the pool of non-smokers is shrinking. Thus, there are now more smokers, and therefore, lung cancer occurred years ago in a larger pool of non-smokers. That pool has now decreased. But, I believe, as I pointed out particularly in women, endocrine factors play a tumor enhancing role which in turn may be modulated by nutritional factors. Having said that now the debate will start, and I will act as a neutral observer. What science is made of, you discuss two sides of an issue and now or sometime in the future, the true facts will prevail. This makes science so much more attractive than politics. With this as a preamble, I am glad to have my colleague from Yale, Dr. Stolwijk who has been part of the Advisory Board to the EPA who will now give you some evidence as to why the EPA report is the way it turned out to be. 312

EPA SCIENCE ADVISORY BOARD EXPERIENCE Jan Stolwijk Yale University, CT DR. STOLWIJK: I appreciate the opportunity to be a part of this discussion. The title is the -Experience of the EPA Science Advisory Board. I have been on it for about 6 years, and I am now off it and now a consultant to it because one rotates off, I gather, and I should have known that there were unusual problems in the EPA Science Advisory Board because after I got on I encountered a final report on the mission and functioning of the EPA Science Advisory Board, a report to the board from the SAB Subcommittee on the Mission and Functioning of the SAB. So, that tells you that there was some soul searching going on within the SAB as to what its mission was, what its function was and how it should function. I suspect that I am expected to discuss the report on secondary tobacco smoke, passive smoking, and that requires that there be a little history. I will try to fill in a little history of what happened. The SAB, as you know, has a number of subcommittees, and some of the subcommittees are created by the Executive Board, and some of the subcommittees are created by Congress as a part of a statutory requirement. There is such a committee in the SAB which is called the Committee on Indoor Air Quality and Total Human Exposure. That was instituted under the SARA(?) legislation in 1988, and it was about indoor air quality and total human exposure because that was mentioned rather elaborately in one of the sections of that law. The committee responds to requests from EPA, from one or more of the divisions or parts of EPA to review things that EPA has done, that its contractors have done in the pursuit of the mission of the agency. In this particular case, there was a report generated by the staff which was on the respiratory effects of passive smoking, lung cancer and other disorders which had been generated at the request of the Indoor Air Office which, also, was established under that same SARA legislation. That office does not do any regulation. It has no regulatory authority. It has an advisory and educational function. It is supposed to educate the public and the professions and anybody else who has responsibility in how to improve indoor air quality or maintain it and what can be done about it and what specific policies might be adopted by people who have the power and the authority to adopt such policies in companies or for buildings or whatever. The report that was generated was generated with EPA at the request of that Indoor Air Office which since has been combined with the Radon Office and now lives in the Office of Indoor Air Quality and Radon in the Air and Radiation Division. When the report was generated and submitted to SAB for its review, it got a review that was fairly, well, it got a 2-day review, plus the time that each individual member could spend reading the report and making notes as to what sorts of things ought to be done. There was a 2-day hearing where the committee actually got together, and there was, also, an opportunity for the public and other interested parties to submit their opinions and their ideas, and it was one of the more elaborate hearings that the Science Advisory Board has had, there being an almost equal number of people who were interested in restricting smoking as much as possible and people who were interested in not restricting smoking, in fact, if it could be helped. 313

Both of these parties argued in favor of or against the report as it then stood on the basis of technical strengths or technical defects, and as a result the committee drafted the response which in the Science Advisory Board goes to the Executive Board of the Science Advisory Board and then gets returned to the Administrator of EPA, and it finds its way down to the office that requested the study to be done. When the committee looked at this, and the committee, by the way, consists or contains within it three or four people who probably would be considered epidemiologists. There would be three or four people who would be exposure experts. There would be three or four people who know about the engineering in buildings and things of that nature, and there were a number of consultants brought on board so that there was a quite a list of people who actually participated in this and a large number of people, also, wrote pieces of the response from the SAB. I could read you, the questions from the EPA come in the form of specific questions to which they specifically would like an answer. We are, also, at liberty to advise them about other things that are not contained in the questions. The first question was (I will take it here from a piece of the executive summary) has EPA met the requirements stated in its carcinogen guidelines for characterizing ETS in category A, that is to say, is the evidence sufficient to conclude that ETS is causally associated with lung cancer. The committee says that it concurs with the judgment of EPA that environmental tobacco smoke should be classified as a class A carcinogen but notes that it had some difficulty with the guidelines for carcinogen risk assessment as they currently are formulated, and we elaborated on that, and mostly because the guidelines actually do not foresee the kind of situation, the kind of conditions that we had to discuss here. The guidelines are, and I can say it probably in this gathering, the guidelines were developed by toxicologists for toxicolgists and did not particularly consider epidemiology, except as an adjunct at the very end of the assessment. So, there was not very much in the guidelines as to how epidemiology should be interpreted, and it was mostly seen as sort of an adjunct for the demonstration of the carcinogenicity in humans but most of the, if you look at the column inches in the guidelines that have to do with animal toxicology, you will find that there is about 100 column inches there, and if you look at the column inches that deal with epidemiology, you will fmd about six column inches. So, you see the difference of emphasis that actually was given to the two. Based on the evidence that was available that there was very little in the way of animal carcinogenicity data and it wasn't in the form that was anticipated in the guidelines and there was in the guidelines, also, not really a way of dealing with complex mixtures. Guidelines anticipate single substances. That is because toxicological assessments preferably are made on single substances and so that is what the guidelines actually refer to. These were the difficulties that we had with declaring tobacco smoke as a class A carcinogen, that if you tried to follow the guidelines, it was literally not possible to do so. EPA had never actually, to my knowledge, classified mainstream tobacco smoke as a carcinogen. It had never come up as a need to do that, and it never had been done. The other question is, the second question that we specifically were asked, "Is spousal smoking a proper measure of ETS exposure to assess lung cancer risk?" This is clearly one question that if you listened to our Chair's presentation is a very difficult question. In fact, any exposure that occurred in the past, and it has a very much delayed effect, a 20-year delayed effect, any human exposure of that type is an almost impossible thing to do with the kind of accuracy that one would like to have. One gets ingenious as to finding ways of doing it, and spousal smoking is one of the ways, one of the shortcuts that you have for finding a special class of people that have at least some kind of class of exposure to environmental tobacco smoke and not having smoked themselves, you 314

know, and so, it is one of the techniques that one finds to actually detect and describe exposures that occur over a very long period of time, and getting those exposures down in questionnaires which is the usual way it has to be done if it goes over a 20-or-30-year period, I think that Ernst Wynder has very eloquently given you all the difficulties and all the trials and tribulations that you have. In fact, I have to give a paper and some kind of consultation at WHO in a couple of weeks, and my assigned task is how do you assess exposures for long delayed effects, and the substance of my presentation will be that you basically cannot do it, and you have to try to find special circumstances where the difficulty of doing this is somewhat diminished, but it always remains difficult. I am currently involved in one place where actually you can do it somewhat. I am finishing up an epidemiologic study of residential radon exposures. So, we had 1000 cases, lung cancer cases. We have 1000 controls obtained in the best possible way. So, we don't have selection biases, etc., and what we are now doing is we are measuring the radon levels in all the houses they have ever lived in which altogether amount to about 6,000 houses, and so, we will have for them a known exposure, if we can believe that the exposure in the house remained the same over the period between 1940, and now, and sometimes we cannot find the houses because they have been torn down or there are other problems, but at least here is one case where you can actually make a measurement because there is a physical presence that still exists, and you can actually make a measurement. However, we also have to assess in these same people what their smoking history was and what their secondary tobacco exposure was, and there we fall into exactly the same difficulties and of course, errors in the smoking history when you have a very strong component. I concur with Ernst Wynder's finding about the less and less never smokers. In the 5000 people that I looked at in Connecticut that had lung cancer, I found that 6 percent had never smoked after you really investigated. So, 94 percent had smoked at some time or another, and that means that since that is a very strong and by far the strongest effect and the radon exposures are relatively small, you are again into the kind of problem of a very weak association with very low exposures that is very ubiquitous and very widespread across the population, and these studies are very expensive and very difficult to do and always subject to considerable criticism. Anybody can find things to be critical about in that kind of study. It is a very difficult thing to do. We were, also, asked whether there was a difference in the studies in the relative risk observed between studies in the US and those overseas; are those differences of concern, and if so, to what degree, and again, we would have to say that they are, indeed, because for some reason not only the Germans but most of the people on the European continent do not get as much lung cancer as we would predict from the amount of smoking they do, and the same, again is true in Japan, also. We don't know why that is, and it makes it very difficult to transport risks that have been measured in one culture into another culture, and transportation is always a very dangerous and difficult thing to do, and you undertake some risks if you do that. The use of ineta analysis, is meta analysis an appropriate tool to use in the document and has it been applied correctly? Have the epidemiologic studies been properly evaluated and combined using this technique? The kind of answer that we have to give to that is that that is a general term, and it works to the extent that studies that you are trying to do a meta analysis on, in fact, lend themselves to it. There has to be similarity in the structure of the data that has been collected. There has to be similarity in a whole lot of decisions that were made when the studies were actually done and to the extent that these similarities do not exist you are on dangerous ground applying meta analysis to that. 315

We warned them against doing meta analyses on dissimilar studies and that considerably more work needed to be done on this. We basically reviewed the draft that we were given and congratulated the staff that had worked up the thing that had come up with the document that they had. The final document was a very different document from the first version of it. There was very considerable tightening up of the logic of the approach. There was a reversal in the sense of the justification of the decision that it is a class A human carcinogen. There was much more of it laid on the fact that mainstream tobacco smoke was clearly one of the best-demonstrated carcinogens and that the similarities between mainstream smoke and sidestream and environmental tobacco smoke, the similarities were much greater than the dissimilarities. There was much more in common than what they didn't have in common. So, the Science Advisory Board advised the EPA to rely more heavily on the similarity and on the much more firm proof of mainstream tobacco smoke. There was a similar piece of advice given that they had in the first draft not adequately developed the evidence and presented the evidence that existed for the effects of environmental tobacco smoke in children. There was a very considerable and large effort, I think, also, a greater attempt to do qualitative interpretation and I think that indicated that there is one area of the effect of environmental tobacco smoke that has been documented because of the fact that children don't yet smoke and because there are fewer confusing and confounding situations there. There is a higher sensitivity in children for acute respiratory disease so that it is more easily demonstrated there and it has been demonstrated on the basis of a large number of different and differently designed studies, and so, we suggested strongly that they rely more on that particular element than on the spousal smoking studies. Nevertheless, it clearly was necessary to deal with the spousal smoking studies, and the spousal smoking studies as Dr. Wynder has indicated have a great deal of trouble in them, and there was a tendency or a desire on the part of EPA to do what I call the body count, in other words to actively attribute a certain number of lung cancer deaths to this kind of exposure, and I think in the discussion of the matter that took place with the staff and with the committee we tried to be as discouraging of that particular effort as we knew how because we felt it would weaken what the message was that they had. It would end up in controversy that would probably not be based on the strongest evidence that actually was listed in the overall report, and in fact, they did modify the calculations, but they persisted in leaving in the estimates of what the attribution was. There are going to be, I think in the next few weeks congressional hearings where Mort Lipman will be testifying before, I think, an Agriculture Subcommittee and, also, before the Energy and Environmental Subcommittee about that particular document and the designation which in the meantime has been done by the Administrator for EPA in the designation of environmental tobacco smoke as a class A carcinogen, and I think that is a short synopsis of the experience we had going through this particular review. Whenever the Science Advisory Board is reviewing a document like this which is a document with very many pages and an enormous amount of content, there is the traditional desire to rewrite the document, but clearly a committee doesn't have the resources or the charge or anything else to rewrite the document and redo it. I think in this particular case if you would look at the two versions that actually have been circulated, we feel that the second version is considerably improved over the first version, and I think EPA shares that attitude, also. So, that is one of the consequences of that review that has taken place. It is, also, I might add there is another thing. If you, as a committee, get asked to review an effort, a document of this type and the document, of necessity is based on all the information and has to be based on all the information that is available, there is a great recognition that there probably will not be additional data 316